Full Issue PDF · San Francisco General Hospital, San Francisco, CA Roberta Gottlieb, MD, ... SOCIAL MEDIA COORDINATOR Tamika Edaire, BS, American College of Cardiology, Washington,

EDITOR-IN-CHIEF

Douglas L. Mann, MD,Washington University Schoolof Medicine, St. Louis, MO

DEPUTY EDITOR

L. Kristin Newby, MD,Duke Clinical Research Institute,Durham, NC

GUEST EDITOR-IN-CHIEF

Robert Roberts, MD,University of Arizona College of Medicine,Tucson, AZ

GUEST EDITOR

Juan F. Granada, MD,CRF-Skirball Center for Innovation, Orangeburg, NY

STATISTICAL EDITOR

Cindy Green, PhD,Duke University Medical Center, Durham, NC

DIVISIONAL SENIOR DIRECTOR, PUBLISHING

Justine Varieur Turco, MA,American College of Cardiology, Washington, DC

EXECUTIVE MANAGING EDITOR

Monica R. Payne-Emmerson, MS,American College of Cardiology, Washington, DC

MANAGING EDITOR

Kimberly Trevey, BA,American College of Cardiology, Washington, DC

DIRECTOR, PRODUCT MANAGEMENT,

DIGITAL PUBLISHING

Nandhini Kuntipuram, MCA, PMP,American College of Cardiology, Washington, DC

WEB MANAGER, DIGITAL PUBLISHING

Elizabeth Bradtke, BA,American College of Cardiology, Washington, DC

SOCIAL MEDIA COORDINATOR

Tamika Edaire, BS,American College of Cardiology, Washington, DC

EDITORIAL ASSISTANT

Jennifer Rapp, BA,American College of Cardiology, Washington, DC

EDITORS-IN-CHIEF

JACC

Valentin Fuster, MD, PhD,Mount Sinai Health System,New York, NY

JACC: Cardiovascular Interventions

David J. Moliterno, MD,University of Kentucky,Lexington, KY

JACC: Cardiovascular Imaging

Y. Chandrashekhar, MD, DMUniversity of Minnesota/VAMC,Minneapolis, MN

JACC: Heart Failure

Christopher M. O’Connor, MD,Inova Heart and Vascular Institute,Falls Church, VA

JACC: Clinical

Electrophysiology

David J. Wilber, MD,Loyola University Medical Center,Chicago, IL

JACC: Case Reports

Julia Grapsa, MD, PhD,Barts Health NHS Trust,London, UK

JACC: CardioOncology

Bonnie Ky, MD, MSCE,Perelman School of Medicine at theUniversity of Pennsylvania,Philadelphia, PA

ASSOCIATE EDITORS

Brian H. Annex, MD,University of Virginia,Charlottesville, VA

Nanette H. Bishopric, MD,University of Miami Schoolof Medicine, Miami, FL

Nikolaos G. Frangogiannis, MD,Albert Einstein College of Medicine,Bronx, NY

Daniel P. Kelly, MD,Perelman School of Medicine atthe University of Pennsylvania,Philadelphia, PA

Peter Libby, MD,Brigham and Women’s Hospital,Harvard Medical School,Boston, MA

William Robb MacLellan, MD,University of WashingtonSchool of Medicine, Seattle, WA

Geoffrey S. Pitt, ScM, MD, PhD,Weill Cornell Medicine,New York, NY

Eva van Rooij, PhD,Hubrecht Institute Netherlands,Utrecht, the Netherlands

EDITORIAL CONSULTANTS

Mark Anderson, MD, PhD,Johns Hopkins UniversitySchool of Medicine,Baltimore, MD

Themistocles Assimes, MD, PhD,Stanford UniversitySchool of Medicine,Palo Alto, CA

Noel Bairey-Merz, MD,Cedars-Sinai Heart Institute,Los Angeles, CA

Craig Basson, MD,Weill Medical College ofCornell University,Needham, MA

Jeffrey Berger, MD,New York University Schoolof Medicine,New York, NY

Don Bers, PhD,University of California,Davis, CA

Michael Bristow, MD, PhD,University of Colorado AMC,Aurora, CO

Daniel Burkoff, MD, PhD,Cardiovascular ResearchFoundation,Framingham, MA

John Burnett, MD,Mayo Clinic Rochester,Rochester, MN

John Canty, MD,University at Buffalo,Buffalo, NY

Barbara Casadei, MD,University of Oxford,Oxford, UK

Karen Christman, PhD,University of California-San Diego,San Diego, CA

Peter Crawford, MD, PhD,University of Minnesota,Minneapolis, MN

Craig Emter, PhD,University of Missouri,Columbia, MO

Zahi Fayad, PhD,Mount Sinai Medical Center,New York, NY

Glenn Fishman, MD,Mount Sinai School ofMedicine,New York, NY

Peter Ganz, MD,San Francisco General Hospital,San Francisco, CA

Roberta Gottlieb, MD,Cedars-Sinai Medical Center,Los Angeles, CA

Josh Hare, MD,University of Miami MillerSchool of Medicine,Miami, FL

EDITORIAL CONSULTANTS CONTINUED

Ray Hershberger, MD,The Ohio State University,Columbus, OH

Carolyn Ho, MD,Brigham and Women’s Hospital,Boston, MA

Jennifer Ho, MD,Massachusetts General Hospital,Boston, MA

Farouc Jaffer, MD, PhD,Massachusetts General Hospital,Harvard Medical School,Boston, MA

Tim Kamp, MD, PhD,University of Wisconsin,Madison, WI

Walter Koch, PhD,Temple University Schoolof Medicine,Philadelphia, PA

David Lanfear, MD,Henry Ford Hospital,Heart and Vascular Institute,Detroit, MI

Jin-Moo Lee, MD,Washington UniversitySchool of Medicine,St. Louis, MO

Richard Lee, MD,Brigham and Women’s Hospital,Boston, MA

Jonathan Lindner, MD,Oregon Health and ScienceUniversity,Portland, OR

Peter Liu, MD,Institute of Circulatory &Respiratory Health–CanadianInstitutes of Health ResearchUniversity Health Network,Ottawa, Ontario, Canada

Eduardo Marban, MD, PhD,Cedars-Sinai Heart Institute,Los Angeles, CA

Ali Marian, MD,University of Texas Health ScienceCenter–Houston,Houston, TX

Kenneth Margulies, MD,University of PennsylvaniaPerelman School of Medicine,Philadelphia, PA

Peter McCullough, MD, MPH,Baylor Heart and VascularInstitute, Baylor UniversityMedical Center, Baylor Heart andVascular Hospital,Dallas, TX

Timothy McKinsey, MD,University of Colorado,Denver, CO

Javid Moslehi, MD,Vanderbilt School of Medicine,Nashville, TN

Jorge Plutzky, MD,Brigham and Women’s Hospital,Boston, MA

David Port, PhD,University of ColoradoSchool of Medicine,Aurora, CO

Sumanth Prabhu, MD,University of Alabama atBirmingham,Birmingham, AL

Hani Sabbath, PhD,Henry Ford Hospital,Washington UniversitySchool of Medicine,Detroit, MI

Paul Simpson, MD,San Francisco VA MedicalCenter and University of California–San Francisco,San Francisco, CA

Mark Sussman, PhD,San Diego State University,San Diego, CA

Jenny Van Eyk, PhD,California School of HealthSciences,Los Angeles, CA

Richard Vega, MD,Translational Research Institute,Orlando, FL

Xander Wehrens, MD, PhD,Baylor College of Medicine,Houston, TX

Arthur Wilde, MD, PhD,Academic Medical CenterUniversity of Amsterdam,Amsterdam, the Netherlands

Myles Wolf, MD, MMSc,Northwestern UniversityFeinberg School of Medicine,Chicago, IL

Sean Wu, MD, PhD,Stanford Cardiovascular Institute,Stanford, CA

CME/MOC/ECME EDITORS

Amanda Coniglio, MD,Duke University School of Medicine,Durham, NC

Michelle Kelsey, MD,Duke University School of Medicine,Durham, NC

Vishal Rao, MD,Duke University School of Medicine,Durham, NC

SOCIAL MEDIA EDITORS

Reza Ardehali, MD, PhD,UCLA Division of Cardiology,Los Angeles, CA

Meena S. Madhur, MD, PhD,Vanderbilt University School of Medicine,Nashville, TN

ETHICS COMMITTEE

Holly Atkinson, MD,Mount Sinai Health System,New York, NY

Lawrence S. Cohen, MD,Yale University School of Medicine,New Haven, CT

Kim Fox, MD,National Heart and Lung Institute,Imperial College, Royal Brompton Hospital,London, UK

Robert Frye, MD,Mayo Clinic Rochester,Rochester, MN

Philip J. Landrigan, MD,Mount Sinai,New York, NY

Richard L. Popp, MD,Stanford University School of Medicine,Palo Alto, CA

Eric N. Prystowsky, MD,The Care Group, LLC,Indianapolis, IN

James Willerson, MD,Texas Heart Institute and the University ofTexas Health Science Center,Houston, TX

CORRESPONDENCE FOR

AMERICAN COLLEGE OF

CARDIOLOGY

All correspondence for the

College, other than that related to

JACC: Basic to Translational Science

should be sent to Resource Center,

American College of Cardiology,

2400 N Street, NW,

Washington, DC 20037

2020-2021 OFFICERS

Athena Poppas, MD, FACC, President

Dipti Itchhaporia, MD, FACC,Vice President

Howard “Bo” T. Walpole, Jr., MD, MBA, FACC, Treasurer

Daniel M. Philbin, Jr., MD, FACC, Secretary and Board of Governors Chair

Cathleen C. Gates,Acting Chief Executive Officer

2020-2021 PUBLICATIONS AND EDITORIAL COORDINATION COMMITTEE

Viviany R. Taqueti, MD, MPH, FACC,Chair

Rhonda M. Cooper-DeHoff, MD, FACC

John U. Doherty, MD, FACC

Islam Y. Elgendy, MD, FACC

Prasad C. Gunasekaran, MD

Fadi G. Hage, MD, FACC

Fred M. Kusumoto, MD, FACC

Renato D. Lopes, MD, PhD, FACC

Sandra M. Oliver-McNeil, DNP, ACNP-BC

Syed Tanveer Rab, MBBS, MACC

Janice B. Sibley, MS, MA,ACC Executive Vice President, Education and Publishing

Justine Varieur Turco, MA,ACC Divisional Senior Director, Publishing

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

ª 2 0 2 0 T H E A U T H O R S . P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E A M E R I C A N

C O L L E G E O F C A R D I O L O G Y F O U N D A T I O N . T H I S I S A N O P E N A C C E S S A R T I C L E U N D E R

T H E C C B Y - N C - N D L I C E N S E ( h t t p : / / c r e a t i v e c o mm o n s . o r g / l i c e n s e s / b y - n c - n d / 4 . 0 / ) .

PRECLINICAL RESEARCH

Lipoprotein(a) Cellular Uptake Ex Vivoand Hepatic Capture In Vivo Is Insensitiveto PCSK9 Inhibition With Alirocumab
Kévin Chemello, BS,a,* Sandra Beeské, PHD,b,* Thi Thu Trang Tran, PHD,b Valentin Blanchard, BS,a
Elise F. Villard, PHD,b Bruno Poirier, PHD,b Jean-Christophe Le Bail, PHD,b Gihad Dargazanli, PHD,b

Sophie Ho-Van-Guimbal, PHD,b Denis Boulay, PHD,b Olivier Bergis, PHD,b Marie-Pierre Pruniaux, PHD,b

Mikaël Croyal, PHD,c Philip Janiak, PHD,b Etienne Guillot, PHD,b Gilles Lambert, PHDa

ISSN 2452-302X

From the aLaboratoire Inser

France; and the cUniversit

contributed equally to this

Innovation) was funded by

recipients of scholarships fr

V). Drs. Beeské, Tran, Villa

employees of Sanofi. Dr. L

Affiris, and Nyrada Inc. All

disclose.

The authors attest they are

stitutions and Food and Dru

the JACC: Basic to Translati

Manuscript received Decem

VISUAL ABSTRACT

m

é

w

t

om

rd

am

ot

i

g

on

b

Chemello, K. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):549–57.

HIGHLIGHTS

� Modulating LDL receptor expression genetically (in familial hypercholesterolemia) or pharmacologically (using statins or

the PCSK9 inhibitor alirocumab) does not alter the cellular uptake of Lp(a) in primary human lymphocytes.

� Lp(a) hepatic capture is not modulated by PCSK9 inhibition with alirocumab in liver-humanized mice.

� LDLR does not appear to play a significant role in mediating Lp(a) plasma clearance in vivo.

https://doi.org/10.1016/j.jacbts.2020.03.008

UMR 1188 DéTROI, Université de La Réunion, Sainte Clotilde, France; bSanofi R&D, Chilly-Mazarin,

de Nantes, CRNH Ouest, Inra UMR 1280 PhAN, Nantes, France. *Mr. Chemello and Dr. Beeské

ork and are joint first authors. The French National Project CHOPIN (CHolesterol Personalized

he Agence Nationale de la Recherche (ANR-16-RHUS-0007). Mr. Chemello and Mr. Blanchard are

the Région Réunion and the European Union (European Regional Development Fund INTERREG

, Poirier, Le Bail, Dargazanli, Ho-Van-Guimbal, Boulay, Bergis, Pruniaux, Janiak, and Guillot are

bert has received research funding and consulting fees from Amgen, Sanofi-Regeneron, Pfizer,

her authors have reported that they have no relationships relevant to the contents of this paper to

n compliance with human studies committees and animal welfare regulations of the authors’ in-

Administration guidelines, including patient consent where appropriate. For more information, visit

al Science author instructions page.

er 26, 2019; revised manuscript received March 11, 2020, accepted March 11, 2020.


http://basictranslational.onlinejacc.org/content/instructions-authors

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jacbts.2020.03.008&domain=pdf

http://creativecommons.org/licenses/by-nc-nd/4.0/

ABBR EV I A T I ON S

AND ACRONYMS

3D = 3-dimensional

apoB100 = apolipoprotein

B100

AU = arbitrary unit

bodipy = boron

dipyrromethene

BSA = bovine serum albumin

ELISA = enzyme-linked

immunosorbent assay

FCR = fractional catabolic rate

FRG = Fah(L/L)Rag2(L/L)

Il2rg(L/L)

HoFH = homozygous familial

hypercholesterolemia

LC-MS/MS = liquid

chromatography tandem mass

spectrometry

LDL = low-density lipoprotein

LDL-C = low-density

lipoprotein cholesterol

LDLR = low-density

lipoprotein receptor

Lp(a) = lipoprotein(a)

MFI = mean fluorescence

intensity

PBS = phosphate-buffered

saline

PBMC = peripheral blood

mononuclear cell

PCSK9 = proprotein

convertase subtilisin/kexin

type 9

rPCSK9 = recombinant

proprotein convertase

subtilisin/kexin type 9

Chemello et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Lp(a) Clearance and PCSK9 Inhibition J U N E 2 0 2 0 : 5 4 9 – 5 7

550

SUMMARY

Lipoprotein(a) (Lp[a]) is the most common genetically inherited risk factor for cardiovascular disease. Many

aspects of Lp(a) metabolism remain unknown. We assessed the uptake of fluorescent Lp(a) in primary human

lymphocytes as well as Lp(a) hepatic capture in a mouse model in which endogenous hepatocytes have been

ablated and replaced with human ones. Modulation of LDLR expression with the PCSK9 inhibitor alirocumab did

not alter the cellular or the hepatic uptake of Lp(a), demonstrating that the LDL receptor is not a major route

for Lp(a) plasma clearance. These results have clinical implications because they underpin why statins are not

efficient at reducing Lp(a). (J Am Coll Cardiol Basic Trans Science 2020;5:549–57) © 2020 The Authors. Pub-

lished by Elsevier on behalf of the American College of Cardiology Foundation. This is an open access article under

the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

E levated lipoprotein(a) (Lp[a]) is thesingle most common geneticallyinherited risk factor for cardiovascular

disease and calcified aortic valve stenosis (1).Elevated Lp(a) is common; approximately25% of the general population has Lp(a) levelsin the atherogenic range (i.e., above 30 to50 mg/dl or 75 to 125 nmol/l) (2). Lp(a) is alow-density lipoprotein (LDL)-like particlesecreted by the liver. Its major structural dif-ference with LDL is that Lp(a) contains a secondlarge protein, apolipoprotein(a) (apo[a]), boundto the apolipoprotein B100 (apoB100) moiety ofa LDL particle by a single disulfide bond (1).

The liver represents the main route forLp(a) clearance from the circulation, andvarious receptors have been proposed tomediate Lp(a) cellular uptake (3). Given thestructural similarity between LDL and Lp(a),the LDL receptor (LDLR) has received themost attention as a candidate receptor forLp(a). However, statins, which increase

LDLR expression and reduce LDL, do not lower thecirculating levels of Lp(a) in humans (4). On thesepremises, it had not been anticipated that propro-tein convertase subtilisin/kexin type 9 (PCSK9) in-hibitors, which increase the cell surface expressionof LDLR via an inhibition of LDLR intracellulardegradation, would not only lower LDL but alsoreduce Lp(a) plasma levels (5).

This observation has led to a flurry of researchaimed at investigating the roles of PCSK9 and LDLR inLp(a) plasma clearance. Thus, in HepG2 cells andprimary human fibroblasts, PCSK9 was shown toreduce the binding and cellular uptake of Lp(a) viaLDLR (6). LDLR inhibition with PCSK9 or LDLRblockade using antibodies targeting the extracellulardomain of the receptor reduced Lp(a) binding toHepG2 cells (7). These results were confirmed inHuH7 hepatoma cells and primary murine

hepatocytes (8). In contrast, we and others have re-ported no significant role of LDLR in mediating Lp(a)cellular uptake in primary human hepatocytes or infibroblasts and HepG2 cells (9,10).

The incorporation of stable isotopes in apo(a) allowsthe determination of Lp(a) kinetic parameters in vivo,but studies conducted in humans also yielded oppositeconclusions regarding the role of LDLR and the effectsof PCSK9 inhibition on Lp(a) clearance. For instance,the Lp(a) fractional catabolic rate (FCR) was similar incontrol individuals and homozygous familial hyper-cholesterolemia (HoFH) patients who lack LDLRfunction (11). In contrast, the PCSK9 inhibitor alir-ocumab was shown to increase (albeit not signifi-cantly) the FCR of Lp(a) in 1 study (12), whereas thePCSK9 inhibitor evolocumab in monotherapy did notalter Lp(a) FCR. However, combined with a statin,evolocumab did increase Lp(a) FCR in that study (13).We have recently reported that alirocumab does notsignificantly modulate Lp(a) FCR in nonhuman pri-mates (14). Therefore, the role of LDLR in mediatingLp(a) plasma clearance remains a matter of consider-able debate.

Lp(a) is only found in humans, old-world monkeys,and hedgehogs. None of the common animal modelsnaturally presents the Lp(a) trait, which severelycomplicates functional in vivo analysis (2). Using anoriginal mouse model repopulated with humanhepatocytes (15) combined with transilluminationtomography imaging techniques as well as primaryhuman lymphocytes (16,17) and flow cytometry totrack fluorescent lipoproteins, we provide new evi-dence that LDLR is not a significant contributor toLp(a) clearance ex vivo and in vivo.

METHODS

LP(A) AND LDL FLUORESCENT LABELING. Plasmafrom an anonymous male donor with Lp(a) levels>75 nmol/l (with a mean number of 22 kringle IV


J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Chemello et al.J U N E 2 0 2 0 : 5 4 9 – 5 7 Lp(a) Clearance and PCSK9 Inhibition

551

domains determined by liquid chromatography tan-dem mass spectrometry [LC-MS/MS]) was purchasedfrom Bioreclamation IVT (Westbury, New York). Lp(a)was isolated by sequential ultracentrifugation(1.050 < d < 1.125 g/ml) at 40,000 g. Lp(a) fraction wasdialyzed against phosphate-buffered saline (PBS)(137 mmol/l NaCl, 2.7 mmol/l KCl, 8 mmol/lNa2HPO4, and 2 mmol/l KH2PO4) and purified by fastperformance liquid chromatography on a LysineSepharose 4 FF column (GE Healthcare, Velizy-Villacoublay, France). Lp(a) was subsequentlydialyzed against PBS. Native purified human LDLsamples were purchased from Alfa Aesar (Haverhill,Massachusetts). Lp(a) and LDL were fluorescentlylabeled with boron dipyrromethene (bodipy 650/665-X, Thermo Fisher Scientific, Waltham, Massachu-setts), and the nonconjugated dye was removed byextensive dialysis against PBS. The absence of freelabel was checked by high performance liquid chro-matography on Acquity UPLC Columns (200 Ang,1.7 mm, 4.6 mm � 150 mm) from Waters (Saint Quen-tin, France).

PERIPHERAL BLOOD MONONUCLEAR CELL ISOLA-

TION FROM PATIENTS AND HUMAN VOLUNTEERS.

Peripheral blood mononuclear cells (PBMCs) wereisolated using Ficoll Paque Plus (Sigma-Aldrich, StLouis, Missouri) as previously described (16,17) fromhealthy volunteers (12 men and 12 women, age 31 � 7years [range 22 to 58 years]; LDL cholesterol [LDL-C]:2.6 � 0.8 mmol/l [range 1.1 to 4.6 mmol/l], and Lp[a]:28.2 � 4.8 nmol/l [range 6 to 95 nmol/l]) and from 1patient with negative HoFH (a 25-year-old womanwith genetically confirmed compound heterozygotefor LDLR mutations E92X and E387A treated withrosuvastatin 20 mg/d þ ezetimibe 10 mg/d þ lipo-protein apheresis every fortnight, LDL-C: 5.2 mmol/l,Lp(a): 30 nmol/l [on treatment before apheresis]). Theproject was approved by the Human Research EthicsCommittee of the University of Cape Town HealthSciences Faculty. All patients provided writteninformed consent for genetic analysis and furtherresearch. PBMCs were subsequently frozen at �80�Cin RPMI culture medium (Life Technologies, SaintAubin, France) containing 70% fetal calf serum and10% dimethyl sulfoxide until use.

LDLR EXPRESSION, LP(A), AND LDL UPTAKE IN

HUMAN PRIMARY LYMPHOCYTES. Freshly thawedPBMCs were seeded in flat bottom 96-well plates(2.105 cells/well) in RPMI containing 10 mmol/lhdroxy ethyl piperazine ethane sulfonic acid(HEPES), 1 mmol/l sodium pyruvate, and 0.5% fetalcalf serum for 2 h at 37�C. The culture medium was

subsequently supplemented with 0 or 10 mg/mlmevastatin (Sigma-Aldrich) for 24 h. Recombinantgain of function PCSK9-D374Y (0 or 600 ng/ml) (CyclexCo., Nagano, Japan) was added to the medium for thefinal 4 h of the incubation time. In a subset of ex-periments, alirocumab (Sanofi, Chilly-Mazarin,France) was added concomitantly into the wells at afinal concentration of 19.2 mg/ml (16–18).

For cell surface LDLR expression determination,lymphocytes were washed twice in ice-cold PBScontaining 1% bovine serum albumin (BSA) andincubated with an allophycocyanin-conjugatedantibody against the human LDLR (clone 472413)or an immunoglobulin G1 (clone 11711) isotype con-trol (R&D Systems, Lille, France) at 0.625 mg/ml for20 min at room temperature in the dark. Lympho-cytes were then washed twice in ice-cold PBS-1%BSA and once in ice-cold PBS. Cells were analyzedon a Cytoflex flow cytometer (Beckman Coulter,Indianapolis, Indiana). Forward scatter versus sidescatter gates were set to include only viable lym-phocytes. A minimum of 5,000 lymphocytes wasanalyzed using CytExpert software (BeckmanCoulter). The mean fluorescence intensity (MFI) ofcells incubated with the isotype control fluorescentantibody (nonspecific binding) was subtracted fromthe MFI of cells incubated with a specific anti-LDLRfluorescent antibody to determine specific MFIlevels (DMFI) of LDLR cell surface expression(16,17).

For fluorescent LDL and Lp(a) uptake assessment,LDL-bodipy650 or Lp(a)-bodipy650 was added to themedium at a 10-mg/ml final concentration for the final3 h of incubation time. In a subset of experiments, anexcess of unlabeled Lp(a) (200 mg/ml) was added5 min before the addition of fluorescent Lp(a) (9). Inanother subset of experiments, Lp(a) uptake wasperformed in the presence of 0.2 mol/l epsilon ami-nocaproic acid (6). Cells were washed twice in ice-cold PBS-1% BSA, once in ice-cold PBS, and resus-pended in ice-cold PBS supplemented with 0.2% try-pan blue (Sigma-Aldrich) to quench cell surface–bound fluorescent LDL or Lp(a) before flow cytom-etry analysis, exactly as described previously. Back-ground fluorescence was measured in lymphocytesincubated without fluorescent lipoproteins. The MFIof the cells incubated without fluorescent lipopro-teins (autofluorescence) was subtracted from the MFIof cells incubated with fluorescent lipoproteins todetermine the specific MFI levels (DMFI) of LDL andLp(a) uptake in those cells, respectively (16,17). TheDMFI is expressed in arbitrary units (AUs)throughout. All measurements were performed atleast 3 times.



552

In a subset of experiments, primary lymphocyteswere incubated either with 10 mg/ml of native (i.e.,unlabeled) Lp(a) or fluorescent Lp(a)-bodipy asdescribed earlier. Cells were washed intensively, andtheir apo(a) cellular content was measured byenzyme-linked immunosorbent assay (ELISA) usingthe STA-359 ELISA kit (Cell Biolabs, San Diego, Cali-fornia). To ascertain that the integrity of fluorescentLp(a)-bodipy was maintained during uptake experi-ments, Lp(a) diluted in culture medium before andafter incubation with primary PBMCs was subjectedto Western blot analysis for apoB100 under reducingand nonreducing conditions using the AF3260 anti-human apoB100 antibody (Bio-Techne, Rennes,France), as described previously (9,19).

CHARACTERIZATION OF THE CHIMERIC FUMAR-

YLACETOACETATE HYDROLASE (FaH) (-/-)

RECOMBINATION ACTIVATING GENE 2 (Rag2) (-/-)

INTERLEUKIN-2 RECEPTOR GAMMA (Il2rg) (-/-)

(FRG)MOUSE MODEL. In vivo studies were per-formed in agreement with European Union directivesfor the standard of care and use of laboratory animalsand approved by the animal care and use committeeof Sanofi R&D. Chimeric liver-humanized male mice(referred to as Fah[�/�]Rag2 [�/�]Il2rg[�/�] FRGmice) were engineered (15,20,21) and provided byYecuris Corporation (Portland, Oregon). These ani-mals were housed in a pathogen-free facility under astandard 12-h light/12-h dark cycle with free access towater and fed ad libitum a PicoLab high-energy 5LJ5chow diet (Ssniff Spezialdiäten, Soest, Germany) with55%, 20%, and 25% of calories from carbohydrates,proteins, and fats, respectively. The chimera FRGmouse model was initially characterized by assessingthe concentration of human and mouse apoB100,apo(a), and apo(a) kringle IV number using a vali-dated multiplexed assay involving trypsin proteolysisand subsequent analysis of proteotypic peptides(Supplemental Table 1) by LC-MS/MS (14). The limit ofdetection of this assay is 1 nmol/l. Serum lipoproteinswere resolved using a fast performance liquid chro-matography Äkta system (GE Healthcare) andcholesterol measured in the eluted fractions using theAmplex Red Cholesterol Assay Kit (Life Technologies)(14). Serum samples were analyzed for direct LDL-Con a Pentra 400 biochemical analyzer (Horiba ABX,Montpellier, France) using standard colorimetric as-says, for apo(a) using the STA-359 ELISA kit (CellBiolabs) with a limit of detection of 0.1 pmol/l, and forhuman apoB100 using the EA7001-1 ELISA kit(Assaypro, Saint-Charles Missouri). The total PCSK9concentrations were determined using the Quanti-kine SPC900 ELISA (R&D Systems).

LP(A) AND LDL HEPATIC UPTAKE IN FRG MICE. An-imals were prepared for imaging studies by skindepilation of the liver area. During imaging, micewere maintained anesthetized with 2% isoflurane inoxygen, and body temperature was monitored. Afterbaseline imaging capture, mice were injected withalirocumab or immunoglobulin G1 placebo control(Regeneron, Tarrytown, New Jersey) (200 mg/kg,subcutaneously) 18 h before infusion of the Lp(a)-bodipy650 or LDL-bodipy650 tracers (1 mg apoB perkg, intravenously). Repeated fluorescence recordingswere performed 15, 30, and 45 min after fluorescentlipoproteins injections. After a washout period, micewere randomly assigned to a new group for pairedinjections with alirocumab or placebo 18 h beforeinfusion with Lp(a)-bodipy650 or LDL-bodipy650 in acrossover protocol. Repeated fluorescence recordingswere performed. Three-dimensional (3D) fluores-cence imaging was performed using the IVIS Spec-trum CT (PerkinElmer, Villebon sur Yvette, France),allowing fluorescence measurement combined withx-ray imaging (6 transillumination points in the liverarea at Excitation: 640 nm, Emission: 680 nm, proneposition). Living Image 4.5 software (PerkinElmer)was used to reconstruct 3D fluorescent tomographicanalysis for each animal from 2-dimensional opticaland X-ray data; 3D fluorescence volumetric pixelswere quantified inside the region of interest (30 �20 � 20 mm in hepatic area) and expressed inAUs throughout.

STATISTICAL ANALYSES. Statistical analyses wereperformed with Prism 6.01 (GraphPad, La Jolla, Cali-fornia). Data distribution was tested using the D’Ag-ostino-Pearson normality test. Normally distributedvariables are presented as mean � SEM, and non-normally distributed variables are presented as me-dian (25th to 75th percentile). Cell treatment com-parisons among LDLR cell surface expression levelswere assessed by analysis of variance followed by theTukey post hoc test for multiple pairwise compari-sons. Comparisons between groups of mice wereperformed using the Student’s t-test for normallydistributed variables (LDL-C and apoB) or the Mann-Whitney test for non-normally distributed variables(apo[a] and fluorescence volumetric pixels). Correla-tion analyses were performed using the Spearmanrank correlation test. A value of p < 0.05 indicatesstatistical significance.

RESULTS

Primary lymphocytes isolated from a representativecontrol volunteer and an HoFH patient were


FIGURE 1 Lp(a) Cellular Uptake Is Not Modulated by Changes in LDLR Cell

Surface Expression Ex Vivo

Peripheral blood mononuclear cells were plated for 24 h in serum-deprived

medium with or without mevastatin (10 mg/ml) and supplemented or not for

the last 4 h of the incubation with recombinant proprotein convertase sub-

tilisin/kexin type 9 (rPCSK9) (600 ng/ml) with or without alirocumab

(19.2 mg/ml) before flow cytometry analysis. (A) Cell surface low-density

lipoprotein receptor (LDLR) expression, (B) low-density lipoprotein (LDL)–

boron dipyrromethene (bodipy) uptake, and (C) lipoprotein(a) (Lp[a])-bodipy

uptake in primary lymphocytes from a control volunteer and a homozygous

familial hypercholesterolemia (HoFH) patient. Data are expressed in D

mean fluorescence intensity. Histograms represent mean � SEM of a min-

imum of 3 independent experiments performed in duplicates. Comparisons

were made by analysis of variance followed by a Tukey post hoc test.

*p < 0.05. **p < 0.01.


553

incubated sequentially with mevastatin, recombinanthuman PCSK9 (rPCSK9), and the PCSK9 inhibitoralirocumab. Baseline LDLR expression assessed byflow cytometry at the surface of control lymphocyteswas found at DMFI of 104 � 16 AU and at DMFI of 19 �10 AU at the surface of HoFH lymphocytes. Mevasta-tin increased, whereas rPCSK9 reduced LDLR cellsurface expression in lymphocytes from the controldonor. Alirocumab restored LDLR cell surfaceexpression in control lymphocytes treated withrPCSK9. In contrast, neither mevastatin nor rPCSK9significantly modulated LDLR cell surface expressionin HoFH lymphocytes (Figure 1A). We then assessedthe cellular uptake of fluorescent LDL in these cells.Paralleling the levels of LDLR cell surface expression,LDL uptake by control lymphocytes was found atDMFI of 153 � 19 AU and at DMFI of 30 � 15 AU inHoFH lymphocytes. Mevastatin increased, rPCSK9reduced, and alirocumab restored LDL uptake incontrol lymphocytes. In contrast, neither mevastatinnor rPCSK9 significantly altered LDL uptake in HoFHcells (Figure 1B). We next assessed the cellular uptakeof fluorescent Lp(a) in lymphocytes from these in-dividuals. In sharp contrast with LDL uptake, Lp(a)cellular uptake was similar in lymphocytes isolatedfrom the control volunteer (DMFI 208 � 20 AU) andfrom the HoFH patient (DMFI 205 � 29 AU). Mevas-tatin, rPCSK9, and alirocumab treatments did notsignificantly alter Lp(a) cellular uptake in the controland HoFH lymphocytes (Figure 1C). We ascertainedcellular Lp(a) uptake by measuring in parallel thecellular content in apo(a) after incubations withnative Lp(a) or fluorescent Lp(a)-bodipy (Figure 2A).We also ascertained by Western blot under dena-turing and nondenaturing conditions that fluo-rescently labeled Lp(a) particles remained intactduring the incubation process (i.e., that apo[a] andapoB100 proteins remained covalently attached overthe time course of the cellular uptake experiments)(Figure 2B). Finally, to validate the specificity offluorescent Lp(a) cellular uptake in primary lympho-cytes, we verified that Lp(a)-bodipy uptake wasreduced by the addition of 20-fold excess unlabeledLp(a) into the culture medium as well as in the pres-ence of epsilon aminocaproic acid, a lysine analogknown to reduce binding of Lp(a) to cell-surface ly-sines (Figure 2C). Altogether these results demon-strate that unlike LDL uptake, Lp(a) cellular uptake isnot responsive to genetic or pharmacological modu-lations of LDLR cell surface expression in primaryhuman lymphocytes.

In line with these observations, we next investi-gated the relationship between LDLR and Lp(a) bycorrelating LDLR cell surface expression measured in

FIGURE 2 Lp(a) Cellular Uptake Is Not Modulated by Recombinant PCSK9

Peripheral blood mononuclear cells (PBMCs) treated with (solid bars) or without (open bars) recombinant proprotein convertase subtilisin/kexin type 9 (600 ng/ml)

were incubated with 10 mg/ml fluorescent lipoprotein(a) (Lp[a]) or native (unlabeled) Lp(a) for 3 h. (A) Cellular Lp(a) uptake was determined by measuring the content

of apo(a) in the cellular extracts. (B) Lp(a) diluted in culture medium before and after 3 h of incubation with PBMCs was subjected to Western blot analysis for

apolipoprotein B100 (apoB100) under reducing and nonreducing conditions; apoB100 association with apo(a) was evidenced in nonreducing conditions. (C) Lp(a)–

boron dipyrromethene (bodipy) uptake in control lymphocytes was assessed in the presence of a 20-fold excess of unlabeled Lp(a) or in the presence of 0.2 mmol/l

epsilon aminocaproic acid. Comparisons were made by analysis of variance followed by a Tukey post hoc test. *p < 0.05, **p < 0.01 vs. standard conditions.



554

lymphocytes isolated from 24 control volunteers withtheir plasma lipoproteins concentrations. Baselinelevels of LDLR expression measured at the surface oflymphocytes (i.e., without rPCSK9, mevastatin, oralirocumab) significantly and negatively correlatedwith the circulating levels of total cholesterol(Spearman rank correlation coefficient [rs] ¼ �0.35,p ¼ 0.046) and LDL-C (rs ¼ �0.43, p ¼ 0.019)measured in the plasma of these 24 individuals butnot with their circulating levels of Lp(a) (rs ¼ �0.26,p ¼ 0.210), further indicating that LDLR is not a majorphysiological regulator of circulating Lp(a) levels inhumans.

Next, we used the FRG chimeric mouse model inwhich mouse hepatocytes have been ablated andrepopulated with human hepatocytes. We first veri-fied that the lipoprotein profile of FRG mice is similarto that of humans because most of their plasmacholesterol is associated with LDL compared withcontrol wild-type mice in which most of the choles-terol is in high-density lipoproteins (SupplementalFigure 1). In addition, these mice present withdetectable concentrations of human apo(a)/Lp(a)with a mean number of 15.3 kringle IV domainsdetermined by LC-MS/MS. This was ascertained byWestern blot analysis (data not shown). These ani-mals also express human apoB100 and human PCSK9in their plasma. We determined that 72% of their totalapoB100 was human and 28% murine, indicating adegree of hepatic chimerism close to 80% because a

small amount of apoB100 can derive from the intes-tine in rodents (21). We next ascertained that FRGmice responded to alirocumab. Compared with con-trols, alirocumab reduced LDL-C levels (1.86 �0.17 mmol/l vs. 0.93 � 0.11, respectively; p ¼ 0.008),circulating human apoB100 (99 � 11 vs. 64 � 4 mg/dl,p ¼ 0.012), and circulating apo(a)/Lp(a) (1.13 [0.96 to1.43] vs. 0.57 [0.26 to 0.86] nmol/l; p ¼ 0.031; n ¼ 4–5per group). It is noteworthy that human PCSK9plasma levels remained unchanged in immunoglob-ulin G1–treated FRG mice but sharply increased inFGR mice treated with alirocumab (from 99 � 17 to801 � 87 ng/ml [p < 0.001]), indicating the accumu-lation of alirocumab-trapped PCSK9 in the plasma ofthese animals. Chimeric FRG mice treated with alir-ocumab or immunoglobulin G1 control were subse-quently intravenously infused with LDL-bodipy.Fluorescent LDL uptake was monitored by 3D trans-illumination fluorescence tomography for 45 min inthe liver of these animals (Figure 3A). Backgroundfluorescence in the region of interest (liver) at base-line (i.e., before LDL-bodipy infusion) was similarin FRG mice treated with alirocumab or immuno-globulin G1. The fluorescence signal in the region ofinterest increased significantly in the alirocumab andIgG1 treatment groups as soon as 15 min afterLDL-bodipy infusion (Figure 3B). This increase wassignificantly more pronounced in FRG mice treatedwith alirocumab compared with FRG mice treatedwith immunoglobulin G1 at the 30-min time point



FIGURE 3 Alirocumab Increases Fluorescent LDL But Not Fluorescent-Lp(a) Hepatic Uptake In Vivo

After baseline imaging capture, Fah(�/�)Rag2 (�/�)Il2rg(�/�) (FRG) mice treated with alirocumab or immunoglobulin G1 were infused either with low-density

lipoprotein (LDL)–boron dipyrromethene (bodipy) or lipoprotein(a) (Lp[a])-bodipy tracers and recordings of 3-dimensional (3D) transillumination fluorescence to-

mography imaging were performed 15, 30, and 45 min after tracer infusions. Fluorescence volumetric pixels were quantified in the region of interest and expressed in

arbitrary units (AUs). (A) Representative recordings of 3D transillumination fluorescence tomography with fluorescence intensity scale bar. (B) Quantification of LDL-

bodipy hepatic uptake in FRG mice treated with immunoglobulin G1 (plain line, n ¼ 4) or alirocumab (dotted line, n ¼ 6). (C) Quantification of Lp(a)-bodipy hepatic

uptake in FRG mice treated with immunoglobulin G1 (plain line, n ¼ 4) or alirocumab (dotted lines, n ¼ 5). Comparisons between treatments were performed using the

Mann-Whitney test. *p < 0.05 vs. immunoglobulin G1.


555

(1.06 � 0.11 AU vs. 0.56 � 0.12 AU; p ¼ 0.017) as wellas at the 45-min time point (1.14 � 0.14 AU vs. 0.57 �0.09 AU; p ¼ 0.015), demonstrating that alirocumabsignificantly enhanced fluorescent LDL uptake in theliver of these animals (Figure 3B). When FRG micetreated with alirocumab or immunoglobulin G1 wereintravenously infused with Lp(a)-bodipy, the fluo-rescence signal in the hepatic region increased simi-larly 15 min after Lp(a)-bodipy infusions in bothtreatment groups. This increase in fluorescence wasnot significantly different in FRG mice treated withalirocumab compared with FGR mice treated withimmunoglobulin G1 at the 30-min time point (2.18 �0.43 AU vs. 2.05 � 0.49 AU, respectively; p ¼ 0.852)and at the 45-min time point (2.04 � 0.37 AU vs. 1.77� 0.44 AU, respectively; p ¼ 0.639), demonstratingthat alirocumab did not significantly modulate fluo-rescent Lp(a) hepatic uptake in humanized liver FRGmice (Figure 3C). It is noteworthy that fluorescencedensity of the Lp(a)-bodipy tracer was 2.3-fold higherthan that of the LDL-bodipy tracer. Taken together,these results show that Lp(a) hepatic uptake is notresponsive to pharmacological modulation of theLDLR by alirocumab in chimeric liver-humanizedmice.

DISCUSSION

In this study, we showed that modulating LDLRexpression genetically (in HoFH) or pharmacologically

(with statins, rPCSK9, and alirocumab) does not alterthe cellular uptake of Lp(a) in human lymphocytes andthat LDLR expression does not correlate with circu-lating Lp(a). We also showed that Lp(a) hepaticuptake is not modulated by PCSK9 inhibition withalirocumab in liver-humanized mice. These combinedresults indicate that LDLR does not play a significantphysiological role in mediating Lp(a) plasma clearancein vivo.

The cellular experiments of the present studyclearly demonstrate a total absence of change in Lp(a)uptake in primary lymphocytes despite the importantmodulation of cell surface LDLR expression inducedby statins, rPCSK9, and alirocumab treatments. Inaddition, the absence of functional LDLR at the surfaceof HoFH lymphocytes did not impact the ability ofhuman lymphocytes to promote Lp(a) uptake. Theseresults are in line with previous studies conducted inHoFH dermal fibroblasts, human primary hepatocytes,and various cell lines (9,10). Thus, irrespective of thecellular model tested (i.e., dermal fibroblasts, hepa-tocytes, and now primary lymphocytes), Lp(a) cellularuptake is not impacted by PCSK9 inhibitors, mevasta-tin treatment, or the combination of both drugs.However, the present results remain at odds withstudies conducted in mouse primary hepatocytes,human skin fibroblasts, and cell lines by others (6,8).There is certainly an inherent limitation of using pri-mary lymphocytes as a proxy for hepatocytes, but thiscell type is easily accessible, and the LDLR pathway in

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: Sta-

tins reduce LDL-C levels by increasing the expression

of the LDL receptor. Monoclonal antibodies targeting

PCSK9, a novel class of lipid-lowering drugs, also

reduce LDL-C by decreasing the degradation of the

LDL receptor. However, unlike statins, PCSK9 inhibi-

tors also reduce the circulating levels of another class

of atherogenic lipoproteins (i.e., Lp[a]). We now

report that the LDL receptor is not significantly

involved in Lp(a) plasma clearance ex vivo and in vivo.

These results explain why, unlike statins, PCSK9 in-

hibitors reduce Lp(a) plasma levels in dyslipidemic

patients.

TRANSLATIONAL OUTLOOK: Lp(a) is an LDL-like

particle containing a peculiar signature protein,

apo(a). Our study suggests that Lp(a) is not primarily

regulated by its catabolism but rather by the

production of apo(a) in the liver. This underpins the

promising results obtained with novel therapies tar-

geting apo(a) gene expression with antisense oligo-

nucleotides or RNA interference currently under

development.



556

lymphocytes and hepatocytes is similar in that it re-quires the same endocytic machinery, in particular theLDLR adaptor protein 1 (17). Further advocatingagainst a significant role for LDLR in mediating Lp(a)clearance is the absence of significant correlation be-tween the levels of LDLR measured at the surface oflymphocytes and the levels of circulating Lp(a), anobservation that we have alsomade in FH patients (16).The absence of modulation of Lp(a) cellular uptakeobserved here underlines that circulating Lp(a) levelsare primarily regulated at the production rather thanat the catabolism level (1).

The experiments of the present study conducted inchimeric FRG mice also indicate an absence of a sig-nificant role for LDLR in mediating Lp(a) hepatic up-take. These in vivo results are inambiguous in thatthey provide a direct visualization of fluorescenttracer accumulation in the livers of humanized mice,unlike stable isotope studies, which despite theirmerits rely on mathematical modeling and thusindirectly assess Lp(a) kinetic parameters (11–14). Ourresults provide a demonstration of an absence of aneffect of PCSK9 inhibition with alirocumab on physi-ological Lp(a) uptake in human hepatocytes. Indeed,these cells are engrafted in a liver environment andnot coated onto plastic with a collagen matrix, amaterial that has been proposed to nonspecificallybind human apo(a) (8). However, our in vivo studyhas the following limitations: 1) we have only testedFRG mice repopulated with human hepatocytes froma single donor; 2) these animals present with detect-able but low plasma levels of Lp(a); 3) Lp(a) accu-mulation beyond the hepatic region was not assessed;and 4) the rate of chimerism of these mice is not100%. In line with these observations, a recent studyshowed that chimeric FRG mice repopulated withhuman hepatocytes from 2 different donors alsodisplay low Lp(a) plasma levels, albeit on averagetwice higher than those measured in the presentstudy (21). Despite their high cost, FRG mice are apowerful tool to assess human lipoprotein meta-bolism because they display a typical human lipo-protein profile with LDL as the predominantlipoprotein even on a normal chow diet (20). This hasbeen recently ascertained by others on a nonobesediabetic background (21). In that respect, it would beextremely informative to perform similar in vivostudies in double transgenic mice that coexpress hu-man apoB100 and apo(a) (22).

Taken together our ex vivo and in vivo resultsclearly indicate that modulation of LDLR expression

with alirocumab does not alter the cellular nor thehepatic uptake of Lp(a). However, the exact mecha-nisms by which PCSK9 inhibitors reduce Lp(a) remainto be elucidated. In that respect, chimeric FRG micerepopulated with human hepatocytes from donorswith elevated Lp(a) should prove instrumental toinvestigate whether PCSK9 and its inhibitors modu-late apo(a)/Lp(a) production.

ACKNOWLEDGMENTS The authors thank AthanaseBayard, Fabrice Tirode, and Brice Nativel for excel-lent technical assistance.

ADDRESS FOR CORRESPONDENCE: Dr. GillesLambert, Inserm UMR 1188, 2 Rue Maxime Rivière,97490 Sainte Clotilde, France. E-mail: [email protected]. OR Dr. Etienne Guillot,Sanofi R&D Diabetes and Cardiovascular Unit, 1Avenue Pierre Brossolette, 91385 Chilly-Mazarin,France. E-mail: [email protected].

mailto:[email protected]




557

RE F E RENCE S

1. Kronenberg F, Utermann G. Lipoprotein(a):resurrected by genetics. J Intern Med 2013;273:6–30.

2. Tsimikas S, Fazio S, Ferdinand KC, et al. NHLBIWorking Group recommendations to reducelipoprotein(a)-mediated risk of cardiovasculardisease and aortic stenosis. J Am Coll Cardiol2018;71:177–92.

3. McCormick SPA, Schneider WJ. Lipoprotein(a)catabolism: a case of multiple receptors. Pathol-ogy 2019;51:155–64.

4. Khera AV, Everett BM, Caulfield MP, et al.Lipoprotein(a) concentrations, rosuvastatintherapy, and residual vascular risk: an analysisfrom the JUPITER trial. Circulation 2014;129:635–42.

5. Lambert G, Thedrez A, Croyal M, et al. Thecomplexity of lipoprotein (a) lowering by PCSK9monoclonal antibodies. Clin Sci (Lond) 2017;131:261–8.

6. Romagnuolo R, Scipione CA, Boffa MB,Marcovina SM, Seidah NG, Koschinsky ML. Lip-oprotein(a) catabolism is regulated by proproteinconvertase subtilisin/kexin type 9 through the lowdensity lipoprotein receptor. J Biol Chem 2015;290:11649–62.

7. Raal FJ, Giugliano RP, Sabatine MS, et al. PCSK9inhibition-mediated reduction in Lp(a) with evo-locumab: an analysis of 10 clinical trials and theLDL receptor’s role. J Lipid Res 2016;57:1086–96.

8. Romagnuolo R, Scipione CA, Marcovina SM,et al. Roles of the low density lipoprotein receptorand related receptors in inhibition of lipoprotein(a)internalization by proprotein convertase subtilisin/kexin type 9. PLoS One 2017;12:e0180869.Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5531514/. Accessed February 17,2020.

9. Villard EF, Thedrez A, Blankenstein J, et al. PCSK9modulates the secretion but not the cellular uptakeof lipoprotein(a) ex vivo: an effect blunted byalirocumab. J Am Coll Cardiol Basic Trans Science2016;1:419–27.

10. Sharma M, Redpath GM, Williams MJ,McCormick SP. Recycling of apolipoprotein(a) af-ter PlgRKT-mediated endocytosis of lip-oprotein(a). Circ Res 2017;120:1091–102.

11. Rader DJ, Mann WA, Cain W, et al. The lowdensity lipoprotein receptor is not required fornormal catabolism of Lp(a) in humans. J ClinInvest 1995;95:1403–8.

12. Reyes-Soffer G, Pavlyha M, Ngai C, et al. Ef-fects of PCSK9 inhibition with alirocumab on li-poprotein metabolism in healthy humans.Circulation 2017;135:352–62.

13. Watts GF, Chan DC, Somaratne R, et al.Controlled study of the effect of proprotein con-vertase subtilisin-kexin type 9 inhibition withevolocumab on lipoprotein(a) particle kinetics. EurHeart J 2018;39:2577–85.

14. Croyal M, Tran T-T-T, Blanchard RH, et al.PCSK9 inhibition with alirocumab reduces lip-oprotein(a) levels in nonhuman primates bylowering apolipoprotein(a) production rate. ClinSci (Lond) 2018;132:1075–83.

15. Azuma H, Paulk N, Ranade A, et al. Robustexpansion of human hepatocytes in Fah�/�/Rag2�/�/Il2rg�/� mice. Nat Biotechnol 2007;25:903–10.

16. Thedrez A, Blom DJ, Ramin-Mangata S, et al.Homozygous FH patients with identical mutationsvariably express the LDL receptor: implications forthe efficacy of evolocumab. Arterioscler ThrombVasc Biol 2018;38:592–8.

17. Thedrez A, Sjouke B, Passard M, et al. Propro-tein convertase subtilisin kexin type 9 inhibition for

autosomal recessive hypercholesterolemia—briefreport. Arterioscler Thromb Vasc Biol 2016;36:1647–50.

18. Lambert G, Chatelais M, Petrides F, et al.Normalization of low-density lipoprotein receptorexpression in receptor defective homozygous fa-milial hypercholesterolemia by inhibition ofPCSK9 with alirocumab. J Am Coll Cardiol 2014;64:2299–300.

19. Tavori H, Christian D, Minnier J, et al. PCSK9association with lipoprotein(a). Circ Res 2016;119:29–35.

20. Ellis ECS, Nauglers S, Parini P, et al. Mice withchimeric livers are an improved model forhuman lipoprotein metabolism. PLoS One 2013;8:e78550.

21. Minniti ME, Pedrelli M, Vedin L-L, et al. Newinsights from liver-humanized mice on choles-terol lipoprotein metabolism and LXR-agonistpharmacodynamics in humans. Hepatology 2019Nov 30 [Epub ahead of print].

22. Schneider M, Witztum JL, Young SG, et al.High-level lipoprotein [a] expression in trans-genic mice: evidence for oxidizedphospholipids in lipoprotein [a] but not in lowdensity lipoproteins. J Lipid Res 2005;46:769–78.

KEY WORDS lipoprotein(a), liver-humanized mice, low-density lipoproteinreceptor, proprotein convertase subtilisin/kexin type 9

APPENDIX For a supplemental table andfigure, please see the online version of thispaper.

http://refhub.elsevier.com/S2452-302X(20)30124-8/sref1






























https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5531514/

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5531514/







































































ª 2 0 2 0 P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E A M E R I C A N

C O L L E G E O F C A R D I O L O G Y F O UN DA T I O N . T H I S I S A N O P E N A C C E S S A R T I C L E U N D E R


EDITORIAL COMMENT

Lipoprotein(a)An Enigmatic Sheep in the Lipoprotein Herd*

Michael D. Shapiro, DO, MCR,a Sergio Fazio, MD, PHDb

H ypercholesterolemia is the principal riskfactor that drives initiation and develop-ment of atherosclerotic cardiovascular dis-

ease (ASCVD), the leading cause of death anddisability worldwide. Many individuals withhypercholesterolemia do not achieve adequatelow-density lipoprotein-cholesterol (LDL-C) reduc-tion with standard lipid-lowering therapies (e.g., sta-tins) or are unable tolerate them. In 2003, thediscovery of proprotein convertase subtilisin kexintype 9 (PCSK9), a low abundance plasma proteinwith a disproportionately large effect on cholesterolmetabolism and plasma LDL-C concentration, ush-ered in a new era of physiological understandingand therapeutic potential. The development of thera-peutic anti-PCSK9 monoclonal antibodies (e.g.,PCSK9 inhibitors) transformed our ability to managepatients with ASCVD and familial hypercholesterole-mia (FH).

The U.S. Food and Drug Administration initiallyapproved the use of PCSK9 inhibitors based on theirLDL-C lowering efficacy and safety while respective

ISSN 2452-302X

*Editorials published in JACC: Basic to Translational Science reflect the

views of the authors and do not necessarily represent the views of JACC:

Basic to Translational Science or the American College of Cardiology.

From the aCenter for the Prevention of Cardiovascular Disease, Section on

Cardiovascular Medicine, Wake Forest University Baptist Medical Center,

Winston Salem, North Carolina; and the bCenter for Preventive Cardiol-

ogy, Knight Cardiovascular Institute, Oregon Health and Science Uni-

versity, Portland, Oregon. Dr. Shapiro was partially supported by the

National Institutes of Health (NIH) (grant K12HD043488); has been a

member of the Scientific Advisory Board for Esperion, Amgen, and

Regeneron; and has been a consultant for Novartis. Dr. Fazio was

partially supported by NIH (grant R01HL132985); and has been a

consultant for Amgen, Amarin, Kowa, Novo Nordisk, and Esperion.

The authors attest they are in compliance with human studies commit-

tees and animal welfare regulations of the authors’ institutions and Food

and Drug Administration guidelines, including patient consent where

appropriate. For more information, visit the JACC: Basic to Translational

Science author instructions page.

large cardiovascular outcomes trials were ongoing.Both therapeutic antibodies target the same region ofPCSK9 and have similar LDL-C lowering efficacy(w60% reduction in LDL-C) at maximum doses. Theresults of cardiovascular outcome trials have beensimilarly impressive for both evolocumab and alir-ocumab; thus, the PCSK9 inhibitor class is nowendorsed by many international guidelines for use inselect patient populations (1).

One of the interesting and unanticipated facets ofPCSK9 inhibition is its consistent association with thelowering of plasma lipoprotein (a) [Lp(a)] levels. Lp(a)is an enigmatic atherogenic lipoprotein that consistsof an LDL-like particle with a protein constituent[apolipoprotein(a)] covalently bound to its apolipo-protein B moiety. A recent meta-analysis of 27 ran-domized controlled clinical trials that enrolled 11,864subjects demonstrated significant and comparablereductions in Lp(a) with either PCSK9 inhibitortreatment (on average: �21.9%) (2). The mecha-nism(s) that underlie PCSK9 inhibitor associated re-ductions in plasma Lp(a) concentration remainunclear, although several hypotheses have been putforward, including: 1) enhanced Lp(a) clearancethrough the LDL receptor (LDLR) pathway; 2)enhanced Lp(a) clearance via other receptors (LDLR-related protein 1[LRP1], cluster of differentiation 36receptor [CD36], toll-like receptor 2 [TLR2], scavengerreceptor-B1 [SR-B1], and plasminogen receptors); and3) reduction in apolipoprotein (a) production, secre-tion, and/or assembly to form Lp(a) particles.

Of the previously described, the most widely heldview linking PCSK9 inhibition and Lp(a) reductionrelates to enhanced LDLR-mediated clearance. How-ever, the notion that Lp(a) clearance is mediated byLDLR poses several challenges: 1) Lp(a) has poor af-finity for LDLR, far less than that of LDL (3); 2) thecatabolic rate of Lp(a) is similar in subjects with FHand without FH; 3) Lp(a) levels are largely unaffected






J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Shapiro and FazioJ U N E 2 0 2 0 : 5 5 8 – 6 0 Lipoprotein(a)

559

by other therapies that upregulate the LDLR (e.g.,statins, ezetimibe) (4); 4) PCSK9 inhibition in patientswith homozygous FH and null LDLR mutations lowersLp(a) more than does LDL-C levels; 5) similar levels ofLp(a) were observed in carriers versus noncarriers ofloss-of-function mutations in PCSK9 (5,6); and 6)there is no consistent correlation between plasmaPCSK9 and Lp(a) concentrations across epidemiolog-ical studies.

Regardless of mechanism, the epidemiological andgenetic associations of Lp(a) with ASCVD and calcificaortic stenosis drive continued interest in under-standing how PCSK9 inhibition may play a role inreducing the burden of Lp(a) associated disease.Moreover, recent focused subanalyses from theFOURIER (Further Cardiovascular Outcomes ResearchWith PCSK9 Inhibition in Subjects With ElevatedRisk) and ODYSSEY OUTCOMES (Evaluation of Car-diovascular Outcomes After an Acute Coronary Syn-drome During Treatment With Alirocumab) trials lendcredence to the notion that PCSK9 inhibitor�inducedLp(a) reduction may effectively reduce residualASCVD risk (7,8). The findings from these sub-analyses, with respect to PCSK9 inhibitor�associatedLp(a) lowering, are noteworthy and beg the questionas to the potential future role of PCSK9 inhibition inthwarting residual cardiovascular risk in subjectswith established ASCVD and elevated Lp(a), regard-less of LDL-C.

With this as a backdrop, a timely mechanistic studyby Chemello et al. (9) published in this issue of JACC:Basic to Translational Science gets to the heart of thequestion: does the LDLR contribute to Lp(a) clearancefrom plasma? The investigators conducted elegantexperimental work in a murine model in which thehost liver parenchyma was ablated and replaced withhuman hepatocytes under a near-normal architecture(mice with humanized liver). The mice were thentreated with either alirocumab or placebo, and he-patic capture of fluorescent LDL and Lp(a) wasassessed. The investigators found significant plasmaLDL-C and Lp(a) lowering in the animals that receivedalirocumab compared with placebo. However,although alirocumab was associated with a significantincrease in fluorescent LDL uptake by the human livercells, there was no significant impact on fluorescentLp(a) capture by these hepatocytes, thus suggesting adifferential mechanism for the lowering of these 2apolipoprotein B�containing particles from plasma.Similarly, the investigators performed parallel ex-periments evaluating cellular uptake of LDL and Lp(a)in primary lymphocytes isolated from normal sub-jects and from a patient with homozygous FH (absentLDLR function). The lymphocytes were incubated

sequentially with or without mevastatin, recombi-nant PCSK9, or alirocumab. They found that fluores-cent LDL cellular uptake followed the patterns ofLDLR cell surface expression. In contrast, cellularuptake of fluorescent Lp(a) was similar in control andhomozygous FH lymphocytes and was not affected bystatin, PCSK9, or alirocumab treatments. In aggre-gate, these series of observations indicate that theLDLR does not play a major physiological role inclearance of Lp(a) because modulation of LDLRexpression either genetically or pharmacologicallyfailed to materially alter the cellular uptake of Lp(a)ex vivo or hepatic capture in vivo (4). In line withthese experimental findings, the investigators’ pre-vious work suggested that PCSK9 influences apoli-poprotein(a) synthesis and/or its assembly into Lp(a),mechanisms clearly independent of the LDLRpathway (10).

The basic science examined in this study providesmechanistic support for the empirical evidence wehave had for years, namely, that statins lower plasmaLDL-C by upregulating the expression of LDLR onhepatocytes without reduction in plasma Lp(a) con-centration. Nevertheless, the consistent reductions inLp(a) observed in all the PCSK9 inhibitor trials rein-vigorated the debate regarding the relative role ofLDLR in Lp(a) catabolism. However, a series of recentstudies further corroborated the results of the studyexamined here. We previously hypothesized that ifthe LDLR was a major pathway for Lp(a) clearance,then inhibition of PCSK9 should produce propor-tionate reductions in LDL-C and Lp(a) in each subject,with an average approximating the 2:1 ratio (LDL-Cz50% to 60%: Lp(a) z 25% to 30%) seen in largerandomized clinical trials. Results from our recentwork highlighted that a significant proportion of pa-tients actually demonstrate discordant responses inLDL-C and Lp(a) to PCSK9 inhibition, showing robustreductions in LDL-C but minimal or no reduction inLp(a) (11,12). We performed an analysis of the PRO-FICIO (Program to Reduce LDL-C and CardiovascularOutcomes Following Inhibition of PCSK9 in DifferentPopulations) clinical trial program, evaluating 895patients who received evolocumab. Baseline LDL-Cand Lp(a) values were 133 and 46 mg/dl, respec-tively, with average reductions of 63.3% and 29.6%with evolocumab administration, which againconfirmed the expected 2:1 ratio. The study demon-strated moderate correlation (r ¼ 0.37; p < 0.001)between percent LDL-C and Lp(a) reduction. Discor-dance was progressively more prevalent among thosewith higher baseline Lp(a), >10 mg/dl (19.7%),>30 mg/dl (26.5%), and >50 mg/dl (28.6%). Recently,we performed a pooled analysis of 10 randomized

Shapiro and Fazio J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Lipoprotein(a) J U N E 2 0 2 0 : 5 5 8 – 6 0

560

controlled trials from the ODYSSEY Phase III clinicaltrial program, which included patients at high car-diovascular risk and/or with FH. Once again, a highrate of discordance (22%) was observed betweenLDL-C and Lp(a) reduction with alirocumab and wasindependent of FH status (13). Importantly, bothstudies suggested there were other mechanismsand/or pathways beyond LDLR that accounted forreductions in Lp(a) levels induced by PCSK9inhibitors.

Although there is no immediate clinical translationto these findings, they do provide the impetus toidentify other potential mechanisms that govern theinteraction(s) between PCSK9 and Lp(a), and themysteries of Lp(a) assembly, secretion, processing,and clearance. Because PCSK9, and by extension,PCSK9 inhibitors, affect many receptors beyond theLDLR (e.g., APOER2, LRP1, VLDLR, CD36, TLR2,plasminogen receptors), it is conceivable that aPCSK9-controlled Lp(a) receptor may direct exit ofLp(a) from the plasma compartment. The fact that theLp(a) lowering induced by PCSK9 inhibitors is relatedto baseline Lp(a) concentration suggests that Lp(a)clearance may be dependent on apolipoprotein (a)isoform size. The LDL-C and/or Lp(a) discordanceobserved in clinical studies may be due to clearancearbitrated by apolipoprotein (a) isoform size and not

by the apolipoprotein B side of the lipoprotein. In thisscenario, apolipoprotein (a) isoform size caused bygenetic variation in the length of the kringle 4 type 2chain may act as a major determinant of the ability ofLp(a) to clear the circulation via LDLR versus alter-native receptors.

Large gaps remain in our understanding of PCSK9physiology and function and in how antagonism ofPCSK9 induces reduction of plasma Lp(a) levels. Themechanistic study by Chemello et al. (9) providessome clues to the biology of Lp(a) removal from cir-culation, an issue that has remained unresolved sincethe discovery of this unique lipoprotein in 1963.Based on this work and other corroborative evidence,we should move beyond the trending notion thatLp(a) is simply cleared by the LDLR pathway. How-ever, even with this step forward, Lp(a) remains theenigmatic lipoprotein particle that the scientificcommunity strives to figure out.

ADDRESS FOR CORRESPONDENCE: Dr. Michael D.Shapiro, Center for the Prevention of CardiovascularDisease, Section on Cardiovascular Medicine, WakeForest University Baptist Medical Center, MedicalCenter Boulevard, Winston Salem, North Carolina27157. E-mail: [email protected].

RE F E RENCE S

1. Grundy SM, Stone NJ, Bailey AL, et al. 2018AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA guideline on the manage-ment of blood cholesterol: a report of the AmericanCollege of Cardiology/American Heart AssociationTask Force on Clinical Practice Guidelines. J Am CollCardiol 2019;73:e285–350.

2. Cao YX, Liu HH, Li S, Li JJ. A meta-analysis ofthe effect of PCSK9-monoclonal antibodies oncirculating lipoprotein (a) levels. Am J CardiovascDrugs 2019;19:87–97.

3. Raal FJ, Giugliano RP, Sabatine MS, et al. PCSK9inhibition-mediated reduction in Lp (a) with evo-locumab: an analysis of 10 clinical trials and theLDL receptor’s role. J Lipid Res 2016;57:1086–96.

4. Boffa MB, Koschinsky ML. Update on lip-oprotein(a) as a cardiovascular risk factor andmediator. Curr Atheroscler Rep 2013;15:360.

5. Saavedra YG, Dufour R, Davignon J, Baass A.PCSK9 R46L, lower LDL, and cardiovascular dis-ease risk in familial hypercholesterolemia: a cross-sectional cohort study. Arterioscler Thromb VascBiol 2014;34:2700–5.

6. Saavedra YGL, Dufour R, Baass A. Familial hy-percholesterolemia: PCSK9 InsLEU genetic variantand prediabetes/diabetes risk. J Clin Lipidol 2015;9:786–93.e1.

7. O’Donoghue ML, Fazio S, Giugliano RP,et al. Lipoprotein(a), PCSK9 Inhibition, andCardiovascular Risk. Circulation 2019;139:1483–92.

8. Bittner VA, Szarek M, Aylward PE, et al. Effectof alirocumab on lipoprotein(a) and cardiovascularrisk after acute coronary syndrome. J Am CollCardiol 2020;75:133–44.

9. Chemello K, Beeské S, Tran TTT, et al. Lip-oprotein(a) cellular uptake ex vivo and hepaticcapture in vivo is insensitive to PCSK9 inhibitionwith alirocumab. J Am Coll Cardiol Basic TransScience 2020;5:549–57.

10. Villard EF, Thedrez A, Blankenstein J, et al.PCSK9 modulates the secretion but not thecellular uptake of lipoprotein(a) ex vivo: an effectblunted by alirocumab. J Am Coll Cardiol BasicTrans Science 2016;1:419–27.

11. Edmiston JB, Brooks N, Tavori H, et al.Discordant response of low-density lipoproteincholesterol and lipoprotein(a) levels to mono-clonal antibodies targeting proprotein convertasesubtilisin/kexin type 9. J Clin Lipidol 2017;11:667–73.

12. Shapiro MD, Minnier J, Tavori H, et al. Rela-tionship between low-density lipoprotein choles-terol and lipoprotein(a) lowering in response toPCSK9 inhibition with evolocumab. J Am HeartAssoc 2019;8:e010932.

13. Mahmood T, Minnier J, Ito MK, et al. Discor-dant responses of plasma low-density lipoproteincholesterol and lipoprotein(a) to alirocumab: apooled analysis from 10 ODYSSEY Phase 3 studies.Eur J Prev Cardiol 2020 Apr 10 [E-pub ahead ofprint].

KEY WORDS atherosclerotic cardiovasculardisease, hypercholesterolemia, lipoprotein,lipoprotein(a), low-density proprotein convertasesubtilisin kexin type 9





































































In Mice Subjected to Chronic Stress,Exogenous cBIN1 PreservesCalcium-Handling Machineryand Cardiac Function
Yan Liu, MD,a,* Kang Zhou, MD,a,* Jing Li, PHD,a,* Sosse Agvanian, BS,a Ana-Maria Caldaruse, BS,a Seiji Shaw,a
Tara C. Hitzeman, MPH,b Robin M. Shaw, MD, PHD,b TingTing Hong, MD, PHDa,c

VISUAL ABSTRACT

IS

Liu, Y. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):561–78.

SN 2452-302X

HIGHLIGHTS

� T-tubule cBIN1-microdomains are

disrupted in hearts with concentric

hypertrophy.

� cBIN1 replacement therapy rescues

t-tubule microdomains and reduces

concentric hypertrophy in post-ISO hearts

inducing a hyper-efficient phenotype

similar to athletic hearts.

� cBIN1-microdomains organize t-tubule

Cav1.2 and SERCA2a distribution for

improved contractility and lusitropy.

� Exogenous cBIN1 is also effective in

protecting cardiac contractility and

lusitropy in mouse hearts subjected to

pressure overload.






AND ACRONYMS

AAV9 = adeno-associatedvirus9

ANOVA = analysis of variance

AR = adrenergic receptor

ATPase = adenosine

triphosphatase

BW = body weight

CAMKII = Ca2D/calmodulin-

dependent protein kinase

cBIN1 = cardiac bridging

integrator 1

CMV = cytomegalovirus

Di-8-ANNEPs = 4-[2-[6-

(Dioctylamino)-2-naphthalenyl]

ethenyl]-1-(3-sulfopropyl)-

pyridinium, inner salt

EC = excitation contraction

EDV = end diastolic volume

EF = ejection fraction

GFP = green fluorescent protein

HF = heart failure

HW = heart weight

HR = heart rate

HT = heterozygote

ISO = isoproterenol

jSR = junctional sarcoplasmic

reticulum

LSD = least significantdifference

LTCC = voltage-dependent

L-type calcium channel

LV = left ventricular

LW = lung weight

PBS = phosphate-buffered saline

PKA = protein kinase A

PLN = phospholamban

RyR = ryanodine receptor

RWT = relative wall thickness

SD = standard deviation

SEM = standard error of the

mean

SERCA2a = sarcoplasmic

reticulum calcium ATPase

pump 2a

SR = sarcoplasmic reticulum

STORM = stochastic optical

reconstruction microscopy

TAC = transverse aortic

constriction

TEM = transmission electron

microscopy

t-tubule = transverse-tubule

vg = vector genome

WT = wild type

Liu et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Exogenous cBIN1 J U N E 2 0 2 0 : 5 6 1 – 7 8

562

SUMMARY

Fro

vas

Ce

and

(HL

(Dr

hav

Ced

Th

ins

for

Ma

Heart failure is an important, and growing, cause of morbidity and mortality. Half of patients with heart failure

have preserved ejection fraction, for whom therapeutic options are limited. Here we report that cardiac bridging

integrator 1 gene therapy to maintain subcellular membrane compartments within cardiomyocytes can stabilize

intracellular distribution of calcium-handling machinery, preserving diastolic function in hearts stressed by

chronic beta agonist stimulation and pressure overload. This study identifies that maintenance of intracellular

architecture and, in particular, membrane microdomains at t-tubules, is important in the setting of sympathetic

stress. Stabilization of membrane microdomains may be a pathway for future therapeutic development.

(J Am Coll Cardiol Basic Trans Science 2020;5:561–78) ©2020The Authors. Published by Elsevier on behalf of

the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

H eart failure (HF) is a global healthconcern, with an estimated 6.2million people affected in the

United States (1). Approximately 50% of pa-tients with HF have preserved ejection frac-tion (EF) with diastolic failure, for whichthere is lack of effective treatment. Diastolicfailure can result from ventricular remodel-ing and diastolic dysfunction, which canoccur secondary to chronic sympathetic acti-vation (2). Preventing remodeling within in-dividual ventricular myocytes may improveoverall cardiac remodeling and have thera-peutic benefits for failing hearts.

Cardiac transverse tubules (t-tubule) arecritical for the initiation of calcium transientsand maintenance of efficient excitation-contraction (EC) coupling. Pathological t-tu-bule remodeling is a consequence ofb-adrenergic stimulation in HF (3–5).Furthermore, impaired t-tubule micro-domains have been implicated in HF pro-gression (6–9). In fact, t-tubule remodelingcan be the tipping point from hypertrophy tofailure (10). Normal calcium transients (11),

m the aSmidt Heart Institute, Cedars-Sinai Medical Center, Lo

cular Research and Training Institute, University of Utah, Salt

dars-Sinai Medical Center and UCLA, Los Angeles, California.*D

are joint first authors. This study was supported by grants fr

133286 to Dr. Hong; HL152691 to Dr. Shaw, and HL138577 to Dr.

. Hong and Dr. Shaw), and the Department of Defense, Washingt

e reported that they have no relationships relevant to the conte

ars-Sinai Medical Center and the University of Utah.

e authors attest they are in compliance with human studies com

titutions and Food and Drug Administration guidelines, includ

mation, visit the JACC: Basic to Translational Science author inst

nuscript received October 2, 2019; revised manuscript receive

which require L-type calcium channels (LTCCs) to be att-tubule microdomains, are crucial to cardiaccontraction and relaxation. The t-tubule membranescaffolding protein cardiac bridging integrator 1(cBIN1) (12), which facilitates LTCC trafficking (13) andclustering for dyad organization, is also under theregulation of b-adrenergic receptor (AR) signaling (14).Furthermore, cBIN1 is reduced in HF (14–16) and theresultant cBIN1-microdomain disruption impairsnormal stress response, limiting contractility andpromoting arrhythmias. Therapeutic approaches thatpreserve cBIN1-microdomains may benefit stressedhearts by protecting the calcium-handling machinery,slowing HF progression.

In the present study, we explored whether in vivoover-expression of exogenous cBIN1 can limitmyocardial remodeling and dysfunction. Continuousisoproterenol infusion, which causes reducedmyocardial cBIN1 expression and disorganized intra-cellular distribution of calcium-handling proteins, alsoinduces pathological concentric hypertrophy withdiastolic dysfunction. We find that normalization ofcBIN1 through adeno-associated virus 9 (AAV9) medi-ated gene transfer both increases inotropy and pre-serves lusitropy, reducing pathological hypertrophy.

s Angeles, California; bNora Eccles Harrison Cardio-

Lake City, Utah; and the cDepartments of Medicine,

rs. Liu, Zhou, and Li contributed equally to this work

om National Institute of Health, Bethesda, Maryland

Shaw); the American Heart Association, Dallas, Texas

on, DC (Dr. Hong and Dr. Shaw, PR160592). All authors

nts of this paper to disclose. All the work was done at

mittees and animal welfare regulations of the authors’

ing patient consent where appropriate. For more in-

ructions page.

d March 11, 2020, accepted March 11, 2020.



J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Liu et al.J U N E 2 0 2 0 : 5 6 1 – 7 8 Exogenous cBIN1

563

Within cardiomyocytes, we find that exogenous cBIN1preserves the intracellular distribution of LTCCs att-tubules and the localization of the sarcoplasmic re-ticulum (SR) calcium-ATPase 2a (SERCA2a). The pro-tective effects of cBIN1 are both isoform-specific andconfirmed effective in a second model of transverseaortic constriction (TAC)–induced cardiac hypertrophyand HF, indicating that exogenous cBIN1-mediatedpreservation of t-tubule microdomains is a possibletherapeutic approach to improve myocardial functionin hearts under chronic stress.

METHODS

An expanded Methods section is provided in theSupplemental Material.

ANIMAL PROCEDURES. All mouse procedures werereviewed and approved by the Institutional AnimalCare and Use Committee of Cedars-Sinai MedicalCenter. Adult male C57Bl/6 mice (Jackson Laboratory,Sacramento, California) were administered with 3 �1010 vector genome (vg) of AAV9 transducing greenfluorescent protein (GFP) or BIN1 isoforms (Welgen,Inc., Worcester, Massachusetts) via retro-orbital in-jection (17). Three weeks later, mice were implantedsubcutaneously with osmotic mini-pumps releasingphosphate-buffered saline (PBS) or isoproterenol(ISO) (30 mg/kg/day). Fifty-six mice were randomizedinto GFPþPBS, GFPþISO, cBIN1þPBS, or cBIN1þISOgroup (N ¼ 14/group). Another 50 mice were ran-domized into receiving AAV9-GFP, cBIN1, BIN1,BIN1þ17, or BIN1þ13 (N ¼ 10/group) before ISO. AAV9was used because it is a promising gene therapyvehicle and exhibits the highest cardiac tropism (18).The cytomegalovirus (CMV) promoter was used givenits efficiency and safety in cardiac gene transfer (19).AAV9-CMV-GFP was used as the negative control vi-rus because it does not induce cardiomyocyte toxicityand has been successfully used as a negative controlvirus in numerous gene therapy studies with animalmodels of cardiovascular diseases (20). For TACstudy, either adult male cardiac-specific Bin1 hetero-zygotes (HT) (Bin1flox/þ, Myh6-creþ) with their wildtype (WT) (Bin1flox/þ, Myh6-cre -) littermates (12) oradult male C57BL/6 mice (Jackson Laboratory, Sacra-mento, California) were used. All mice were anes-thetized at the age of 8 to 10 weeks and subjected toopen-chest TAC or mock surgery (Sham). TAC wasperformed by tying a 7-0 silk suture against a27-gauge needle between the first and second branchoff the aortic arch. For gene therapy, same as the ISOstudy, mice received retro-orbital injection of 3 � 1010

vg of AAV9 virus transducing cBIN1-V5 or GFP-V5 at3 weeks prior to the onset of TAC.

Echocardiograms were recorded using a Vevo-3100 ultrasound system (Visual Sonics, Toronto,Ontario, Canada) equipped with a 70-MHz trans-ducer. Protein interaction was analyzed usingimmunofluorescent imaging and biochemical coim-munoprecipitation. Peak intensity of Cav1.2 att-tubules is quantified using Image J as previouslyreported (13). Power spectrum analysis wasanalyzed in Matlab using fast Fourier transformconversion (10,13). Intracellular protein distributionwas analyzed using sucrose gradient fractionationusing a previously established method (21). Forcalcium transient measurement, Cal-520-AM (AATBioquest, Sunnyvale, California) was used as previ-ously described (14). Three-dimensional super-res-olution stochastic optical reconstruction microscopy(STORM) images were obtained (14) for nearestneighbor analysis between LTCC–ryanodine receptor2 (RyR) and SERCA2a-cBIN1 molecules.

STATISTICAL ANALYSIS. Data were analyzed usingGraphPad Prism version 7.0 (GraphPad Software, LaJolla, California). All data are presented as mean �SEM or SD as specified. Normality was assessed usingthe Shapiro-Wilk test. Continuous variables werecompared using Student’s t-test/Mann-Whitney Uand 1-way analysis of variance (ANOVA)/Kruskal-Wallis tests. Two-way ANOVA was used to determinedifferences between 2 AAV9 groups with differentdrug infusion, which was then followed by Fisherleast significant difference (LSD) post-hoc adjustmentfor multiple pairwise comparisons. Categorical vari-ables were analyzed using Fisher exact or chi-squaretests. For survival comparison, log-rank test wasused to compare Kaplan-Meier survival curves be-tween groups. Two-sided p values were used andp < 0.05 was considered statistically significant.

RESULTS

EXOGENOUS CBIN1 REDUCES CONCENTRIC HYPERTROPHY

INMOUSEHEARTSAFTER ISO INFUSION. We investigatedthe effect of cBIN1 on myocardial function in animalssubjected to 4 weeks of ISO infusion (Figure 1A). AAV9was used to introduce myocardial expression ofexogenous V5-tagged GFP or cBIN1 (22) 3 weeks priorto the onset of ISO. Anti-V5 labeling identified asimilar percent of myocardial area with detectable V5signal at 7 weeks after AAV9 injection (GFP, 62.4 �10.5%; cBIN1, 57.9 � 8.0%), indicating successfultransduction of exogenous protein in over half of


FIGURE 1 Exogenous cBIN1 Reduces Concentric Hypertrophy in Post-ISO Mouse Hearts

(A) Experimental protocol: 56 mice were randomized into 4 experimental groups: AAV9-GFPþPBS, AAV9-GFPþISO, AAV9-cBIN1þPBS, and AAV9-cBIN1þISO (n ¼ 14/

group). (B)Mouse HW/BW in the 4 groups. (C) Representative images of longitudinal axis view of left ventricles at both end diastolic and end systolic phase at 4 weeks

post-PBS or ISO infusion. Echocardiography analysis of end diastolic volume, LV mass, and relative wall thickness (D), ejection fraction (E), E/e’ (F), stroke volume (G),

and cardiac output (H) is also included. Data are presented as mean � SEM. Two-way ANOVA was used followed by Fisher LSD test for multiple comparison. *p < 0.05,

**p < 0.01, and ***p < 0.001, for PBS vs. ISO comparison within each AAV9 treatment group ##p <0.01 and ###p <0.001 for GFP vs. cBIN1 comparison within each

drug infusion group. AAV9 ¼ adeno-associated virus 9; ANOVA ¼ analysis of variance; BW ¼ body weight; cBIN1 ¼ cardiac bridging integrator 1; GFP ¼ green

fluorescent protein; HW ¼ heart weight; ISO ¼ isoproterenol; LSD ¼ least significant difference; LV ¼ left ventricular; PBS ¼ phosphate-buffered saline; SEM ¼ standard

error of the mean.



564

FIGURE 2 ISO Reduces cBIN1 and Disrupts cBIN1-Microfolds, Which Is Normalized by AAV9-cBIN1

Continued on the next page


565

FIGURE 2 Continu

(A) Western blots of

Quantification in the

10 mm) and power sp

the left (N ¼ 26 to 3

comparison. *p < 0.0

each drug infusion g

Quantitation of the d

sections and 2 to 3 h

GFPþISO vs. cBIN1þGAPDH ¼ glyceralde



566

cardiomyocytes (Supplemental Figure 1). It is possiblethat the remaining almost 40% of negatively stainedcardiomyocytes may express exogenous proteins at alow level below the detection threshold of immuno-fluorescence. In all mice, ISO significantly increasedheart weight to body weight ratio (HW/BW), indi-cating cardiac hypertrophy (Figure 1B). Cardiac ge-ometry and function were assessed usingechocardiography (Figures 1C to 1H). In AAV9-GFP–pretreated animals, ISO induced a significant increasein left ventricular (LV) mass and relative wall thick-ness (RWT), without altering end diastolic volume(EDV), consistent with echocardiography-based clas-sification of concentric hypertrophy (23). In AAV9-cBIN1–pretreated animals, the ISO increase of LVmass was attenuated with a normal RWT andincreased EDV, similar to echocardiography-classified“physiological hypertrophy-like LV remodeling” us-ing a previous reported method (23) (SupplementalFigure 2A). Furthermore, a-smooth muscle actin isincreased in GFPþISO hearts but not in cBIN1þISOhearts (Supplemental Figure 2B), indicating AAV9-cBIN1 limits ISO-induced LV hypertrophy. InAAV9-GFP–pretreated mice, ISO resulted in a smallincrease in EF (p ¼ 0.050 vs. GFPþPBS and p ¼ 0.007vs. cBIN1þPBS group), yet E/e’ strongly increased,indicating the onset of diastolic dysfunction. In micepretreated with AAV9-cBIN1, ISO still increased sys-tolic function and importantly maintained a normalE/e’, indicating positive inotropy with preservedlusitropy. Furthermore, although without bloodpressure measurement, which remains a limitation ofthe current study, ISO significantly increased heartrate (HR) in all animals, indicating that ISO is effec-tive in causing hemodynamic stress. Yet, post-ISO HRwas not different between GFPþISO and cBIN1þISOmice (Supplemental Table 1), confirming that furtherimproved post-ISO cardiac output in AAV9-cBIN1mice was due to muscle efficiency and notincreased rate.

ed

cBIN1 and GAPDH from heart lysates and immunoprecipitated heart lysates

bar graph to the right (N ¼ 6–7 hearts per group). (B) Representative cardio

ectrum (bottom panel) of the corresponding boxed region of interest above.

1 cells from 3 to 4 hearts per group). Data are presented as mean � SEM. T

5 and **p < 0.01 for PBS vs. ISO comparison within each AAV9 treatment gro

roup. (C) Transmission electron microscopy imaging of t-tubule (TT) microfo

egree of contour of TTs from each group is included in the bar graphs to the l

earts from each group). Chi-square test was used to compare TT contour bet

ISO, and cBIN1þPBS vs. other groups. Di-8-ANNEPs ¼ 4-[2-[6-(Dioctylamino)

hyde 3-phosphate dehydrogenase; TTs ¼ t-tubules; other abbreviations as in

CHRONIC ISO-DISRUPTED CBIN1-MICRODOMAINS

CAN BE NORMALIZED BY AAV9-CBIN1. It is wellknown that bothmyocardial inotropy and lusitropy arerelated to cardiomyocyte calcium cycling (24). cBIN1,the structural organizer for dyad microdomains (14),creates t-tubule microfolds to limit extracellular Ca2þ

diffusion (12), facilitates microtubule-dependent for-ward trafficking of LTCCs (13), and clusters LTCCs thatare already delivered to t-tubule membranes. We,therefore, explored how cBIN1-microdomains mayremodel in hypertrophic hearts after chronic ISOinfusion. Western blots of heart lysates indicate thatISO induces a significant reduction in cBIN1 protein,which is normalized by AAV9-cBIN1 (Figure 2A). Notethat immunoprecipitation with anti-BIN1-exon 17antibody followed byWestern blot detectionwith anti-BIN1-exon 13 antibody confirms that the examinedprotein band is the cBIN1 (BIN1þ13þ17) isoform.Gross t-tubule network architecture was also exam-ined in isolated cardiomyocytes labeled with a mem-brane dye, 4-[2-[6-(Dioctylamino)-2-naphthalenyl]ethenyl]-1-(3-sulfopropyl)-pyridinium, inner salt(Di-8-ANNEPs) (Figure 2B). Live-cell imaging followedby power spectrum analysis indicates that overallt-tubule organization (normalized peak powerdensity) remains similar in GFPþISO mice, and isincreased by AAV9-cBIN1. Even though the grosst-tubule network remains organized, using trans-mission electron microscopy imaging, we noted inGFPþISO hearts that there was a reduction in t-tubulemicrofolds, which were preserved in cBIN1þISOhearts (Figure 2C). Quantitation of the degree of con-toured t-tubules using a modified scoring systemestablished previously (12) (1, round or dilatedt-tubule lumen without folds; 2, non-circular con-toured t-tubule lumen without folds; 3, t-tubules with2 to 3 layer of folds; or 4, t-tubules with >3 layer offolds) identifies a significant reduction in t-tubulecontour in GFPþISO hearts, which is normalized incBIN1þISO hearts (p< 0.001, chi-square test). Note the

from GFPþPBS, GFPþISO, cBIN1þPBS, and cBIN1þISO hearts.

myocyte images with Di-8-ANNEPs labeling (top panel) (Scale bar,

Quantification of peak power density is included in the bar graph to

wo-way ANOVA was used followed by Fisher LSD test for multiple

up; #p < 0.05 and ##p < 0.01 for GFP vs. cBIN1 comparison within

lds from myocardial tissue from all 4 groups (scale bar, 1 mm).

eft (N ¼ 232 to 305 TTs from 60 to 100 images of 5 to 6 myocardial

ween groups, p < 0.001 for comparison of GFPþPBS vs. GFPþISO,

-2-naphthalenyl]ethenyl]-1-(3-sulfopropyl)-pyridinium, inner salt;

Figure 1.






FIGURE 3 cBIN1 Increases Cav1.2 Localization to t-Tubules

(A) Western blot of Cav1.2 in heart lysates from GFPþPBS, GFPþISO, cBIN1þPBS, and cBIN1þISO hearts. Quantification (Cav1.2/GAPDH and Cav1.2/Troponin) is

included in the bar graphs to the right (n¼ 5 to 7 hearts per group). (B) Representative confocal images (100�) of anti-Cav1.2 labeling in mouse myocardium from each

group (top 2 panels) (scale bar, 10 mm). The third panel includes power spectrum and the fourth panel includes fluorescence intensity profiles within the boxed areas

along the cardiomyocyte longitudinal axis. Quantification of Cav1.2 peak power density and immunofluorescent intensities at t-tubules in each group is summarized in

the bar graph in C (n ¼ 15 to 32 cell images from 3 to 4 hearts per group). Scale bar: 10 m. (D) Representative calcium transient tracing from each group and

quantification of peak amplitudes (DF/F0) (n ¼ 61 to 88 cells from 6 hearts per group). Data are presented as mean � SEM. Two-way ANOVA was used followed by

Fisher LSD test for multiple comparison. ***p < 0.001 for PBS vs. ISO comparison within each AAV9 treatment group; ###p < 0.001 for GFP vs. cBIN1 comparison within

each drug infusion group. Abbreviations as in Figure 1 and Figure 2.


567



568

exaggerated microfolds (more than 3 layers of folds,score of 4) are found in cBIN1þPBS hearts, a resultof greater than physiological levels of cBIN1. Thesedata indicate that cBIN1 is critical for the formation oft-tubule microfolds, the folds are down-regulatedunder chronic sympathetic overdrive, and thefolds can be restored using cBIN1 exogenous therapy.

We then explored Cav1.2 expression and intracel-lular distribution in cardiomyocytes. In post-ISOhearts, the net myocardial protein expression ofCav1.2 is similar (Figure 3A). However, myocardialtissue immunofluorescent labeling of Cav1.2 revealsthat channel density along t-tubules was significantlyreduced in GFPþISO cardiomyocytes, which wasnormalized by AAV9-cBIN1 (Figures 3B and 3C, powerspectrum and fluorescent profile analysis). These dataare consistent with previous observations of alteredCav1.2 protein distribution despite similar total pro-tein levels (15). With reduced Cav1.2 localization tot-tubules, the peak amplitude of calcium transient(DF/F0) was significantly reduced in GFPþISO car-diomyocytes when compared with that from controlGFPþPBS myocytes (Figure 3D), which was normalizedby exogenous cBIN1. Ryanodine receptor 2 (RyR) totalprotein expression and intracellular distribution werenot different across all groups (SupplementalFigure 3). However, RyRs became hyper-phosphorylated at both protein kinase A (PKA)-dependent S2808 and Ca2þ/calmodulin-dependentprotein kinase II (CAMKII)-dependent S2814, consis-tent with a previous report (25). Together withincreased phosphorylation at T287 in CAMKIId, thesedata indicate that PKA and CAMKII activation-inducedRyR hyperphosphorylation occurs after chronic ISOinfusion. Importantly, AAV9-cBIN1 pretreatment suc-cessfully blunts these pathways and reduces RyRhyperphosphorylation (Supplemental Figure 4).EXOGENOUS CBIN1 IMPROVES SERCA2A DISTRIBUTION

ALONG SR. Cardiac lusitropy is most directly relatedto calcium reuptake via SERCA2a. To our surprise,despite impaired diastolic dysfunction in GFPþISOhearts, total protein expression of SERCA2a wassignificantly increased after ISO infusion (Figures 4Aand 4B). Total protein levels of phospholamban(PLN) and its phosphorylated forms (pS16 and pT17)are not altered (Supplemental Figure 4). Previousstudies indicate that acute ISO-induced PLN phos-phorylation can be normalized after chronic ISOinfusion and even PLN dephosphorylation can occurdue to activation of serine/threonine phosphatasesPP1 and PP2A (26,27). Consistent with these reports,our results indicate that unchanged PLN phosphory-lation after 4 weeks of ISO infusion is a possible netresult of balanced local activation of both kinases and

phosphatases. These data indicate that both SERCA2aprotein and activity are not decreased in post-ISOhearts. Given the effect of cBIN1 on Cav1.2 localiza-tion, we then examined SERCA2a localization.Myocardial tissue sections with SERCA2a labelingwere imaged with spinning-disc confocal microscopyand compared across groups (Figure 4C). In GFPþPBShearts, a subpopulation of SERCA2a was concentratedto the t-tubule/ junctional sarcoplasmic reticulum(jSR) regions, giving rise to an organized distributionwith a major power spectrum peak at 1.8 to 2 mm,corresponding to the full length of a sarcomere. Over-expression of cBIN1 in the cBIN1þPBS hearts furtherincreased SERCA2a signals near t-tubule/jSR. InGFPþISO hearts, intracellular distribution of SERCA2awas disorganized with a significant reduction in peakpower density, which was normalized in cBIN1þISOhearts (quantification in Figure 4D).

Intracellular distribution of Cav1.2 and SERCA2awas further explored using biochemical sucrosegradient-based fractionation of cardiac microsomes(21). As indicated in Supplemental Figure 5A, frac-tion 4 (F4) has the lowest recovery yield whencompared with other fractions. However, even witha low yield, Cav1.2 and cBIN1 are detectable in F4with limited Naþ/Kþ-ATPase and depleted SERCA2a,indicating that F4 is enriched with t-tubule originmicrosomes (Supplemental Figure 5B). Whennormalizing t-tubule protein concentration for F4across all samples (2.5 mg protein loaded per lane),GFPþISO hearts have a significant reduction in bothcBIN1 and Cav1.2 protein per unit t-tubule whencompared with control GFPþPBS hearts, which wasnormalized using AAV9-cBIN1 pretreatment(Figure 5A). These data are consistent with immu-nofluorescent imaging, which identified less t-tu-bule localization of Cav1.2 channels following ISOinfusion and restoration with AAV9-cBIN1. On theother hand, SR proteins are detected only in frac-tions F2 and F3. When normalizing SR proteinconcentration for F2 and F3 (25 mg protein loadedper lane), F3 has relatively more RyR and less PLNthan F2 (Figure 5B), indicating more enrichment ofjSR toward the heavier F3 fraction. Quantification ofSERCA2a expression in F2 and F3 identifies thatwhen compared with AAV9-GFP, AAV9-cBIN1 causesa significant increase in SERCA2a distribution intothe heavier and more jSR-enriched F3, but not thelongitudinal SR-enriched F2 fraction (Figure 5B).Note that ISO alone does increase SERCA2aexpression in F3 in AAV9-GFP mouse hearts, likelydue to an overall increase in total protein expres-sion of SERCA2a in post-ISO hearts (Figure 4A).These data indicate that, in ISO-infused hearts,







FIGURE 4 cBIN1 Organizes Intracellular Distribution of SERCA2a in Post-ISO Hearts

(A)Western blot of SERCA2a in heart lysates fromGFPþPBS, GFPþISO, cBIN1þPBS, and cBIN1þISO hearts. (B)Quantification (SERCA2a/Actin) is included in the bar graph

(n¼ 6 to 8 hearts per group). (C) Representative confocal images of anti-SERCA2a labeling in mouse myocardium from each group (top 2 panels). Scale bar: 10 mm. The

third panel includes the power spectrumof SERCA2a of theboxed area above. (D) Quantification of peak power density of SERCA2a (n¼ 11 to 15 cell images from 3–4hearts

per group). Data are expressed asmean� SEM. Two-wayANOVAwas used followed by Fisher LSD test formultiple comparison. **p<0.01 and ***p, 0.001 for PBS vs. ISO

comparison within each AAV9 treatment group; ### p < 0.001 for GFP vs. cBIN1 comparison within each drug infusion group. Abbreviations as in Figure 1.


569

exogenous cBIN1 can maintain t-tubule micro-domains to localize Cav1.2 and SERCA2a to theirfunctional sites.

Given the reduced Cav1.2 and SERCA2a at thet-tubule/jSR region, we next used STORM imaging toanalyze nanoscale protein-protein co-localization forCav1.2-RyR and SERCA2a-cBIN1 (Figure 6, Videos 1A,

1B, 1C, 1D, 2A, and 2B). Using nearest neighbor anal-ysis, we quantified the distance between the indi-vidual Cav1.2 molecule and its closest RyR molecule.Histogram distribution of distances between Cav1.2-RyR molecules from whole cell images identifies afirst peak near 40 nm in GFPþPBS, GFPþISO, andcBIN1þISO cardiomyocytes, corresponding to dyad

http://jaccbts.acc.org/2020/0238_VID1A.mp4

http://jaccbts.acc.org/2020/0238_VID1B.mp4

http://jaccbts.acc.org/2020/0238_VID1C.mp4

http://jaccbts.acc.org/2020/0238_VID1D.mp4



FIGURE 5 Sucrose Gradient Fractionation of Cardiac Microsomes

(A) Representative Western blots of Cav1.2 and cBIN1 in the F4 (TT) fraction of cardiac microsome from GFPþPBS, GFPþISO, cBIN1þPBS, and cBIN1þISO hearts (2.5 mg

proteins loaded per lane). Quantification is included in the bar graphs (n ¼ 3 hearts per group). (B) Representative Western blots of RyR, PLN, and SERCA2a in the F2

(longitudinal SR enriched) and F3 (jSR enriched) fractions of cardiac microsome from GFPþPBS, GFPþISO, cBIN1þPBS, and cBIN1þISO hearts (25 mg protein loaded per

lane). Quantification of SERCA2a in F2 and F3 is included in the bar graphs to the right (n ¼ 3 hearts per group). Data are expressed as mean � SD. Two-way ANOVA

was used followed by Fisher LSD test for multiple comparison. *p < 0.05, **p <0.01, and ***p < 0.001 for PBS vs. ISO comparison within each AAV9 treatment

group; #p < 0.05 and ##p < 0.01 for GFP vs. cBIN1 comparison within each drug infusion group. jSR ¼ junctional sarcoplasmic reticulum; PLN ¼ phospholamban;

RyR ¼ ryanodine receptor; SD ¼ standard deviation; SERCA2a ¼ sarcoplasmic reticulum calcium ATPase pump 2a; other abbreviations as in Figure 1.



570

couplons. In GFPþISO hearts with preserved systolicfunction, the distribution histogram tends to shift tothe right, yet with a still preserved first peak position(Figure 6B). Interestingly, cBIN1þPBS myocytes havea left-shifted histogram distribution and a signifi-cantly reduced Cav1.2-RyR peak distance, indicating

tightened couplons likely brought closer by exagger-ated cBIN1-microfolds as observed in transmissionelectron microscopy imaging. On the other hand, thedistance between SERCA2a and its nearest neighborcBIN1 at t-tubules has a trend of increase after ISO inAAV9-GFP–pretreated animals (p ¼ 0.063; GFPþPBS


571

vs. GFPþISO), which is significantly decreased inAAV9-cBIN1–pretreated animals (p < 0.001; GFPþISOvs. cBIN1þISO) (Figures 6C to 6D). These data indicatethat cBIN1-microfolds can regulate co-localizationand interaction between EC coupling and calcium-handling proteins.

THE PHENOTYPE OF CBIN1DISO HEARTS IS ISOFORM

SPECIFIC AND UNIQUE TO CBIN1. To further explorewhether the observed phenotype of cBIN1þISO heartsis an isoform-specific effect, we repeated the ISOprotocol in 50 more mice, which were randomized toreceive AAV9 transducing GFP and cBIN1, as well asthe other 3 mouse cardiomyocyte expressing BIN1isoforms including the small BIN1, BIN1þ17, andBIN1þ13. Similarly, 3 weeks after viral administration,mice were subjected to continuous subcutaneous ISOinfusion at 30 mg/kg/day for 4 weeks. The proteinexpression of Cav1.2 and SERCA2a in post-ISO heartswas not significantly different when compared across5 groups of mice transduced with GFP or BIN1 iso-forms (Supplemental Figure 6A). Myocardial tissueimmunofluorescent labeling of Cav1.2 channels re-veals that channel density along t-tubules is signifi-cantly increased only in cBIN1-expressing hearts, butnot the other BIN1 isoforms (Supplemental Figure 6B,quantification in 6D). Immunofluorescent imagingreveals that exogenous cBIN1 introduced by AAV9organizes SERCA2a distribution (SupplementalFigure 6C, quantification in Supplemental Figure6E), consistent with the data from Figure 4.

We next explored the functional consequence ofdifferent AAV9-BIN1 isoform pretreatment usingechocardiography. cBIN1-expressing mice, whencompared with the GFP group, lessened the ISO-induced increase in LV wall thickness, LV mass, andRWT (Figures 7A to 7D, Supplemental Table 2). In allanimals, post-ISO cardiac output is significantlyincreased from its level at baseline, a result from theISO-induced increase in HR (Supplemental Table 2).However, only cBIN1 hearts also have an improvedsystolic function, normalized E/e’, increased strokevolume, and further increased cardiac output whencompared with post-ISO GFP hearts (Figures 7E to 7H).Of note, a partial cBIN1-like effect occurs in micepretreated with BIN1þ17, which significantly reducesLV mass, reduces E/e’, and attenuates ISO-increasedRWT. The observed partial diastolic functionalrescue from BIN1þ17 is consistent with the partialrescue of intracellular distribution of SERCA2ausing immunofluorescent imaging (SupplementalFigure 6C). However, due to the inability of BIN1þ17to increase Cav1.2 distribution at t-tubules, there is

not a positive inotropic effect in AAV9-BIN1þ17–pre-treated hearts following ISO infusion.

THE CARDIAC PROTECTIVE EFFECT OF AAV9-CBIN1

IS CONFIRMED IN TAC-INDUCED HF. Next, wefurther explored the myocardial protective effect ofcBIN1 in a separate mouse model of pressure overloadinduced using TAC. Mice with either genetic defi-ciency of cBIN1 or AAV9-transuced cBIN1 over-expression were tested in this study (Figure 8A). Thedeficiency study involved cardiac specific Bin1 HTmice and WT littermate controls (12), both subjectedto TAC for 8 weeks. The over-expression studyinvolved mice subjected to 8 weeks of TAC with priorinjection of AAV9 transducing cBIN1-V5 or AAV9-GFP-V5, and mice subjected to an open-chest mocksurgery (Sham). Mice were monitored and humanelykilled at 8 weeks post-surgery. The viruses (AAV9-GFP/cBIN1-V5), dosage (3 x 1010 vg), administrationtime (3 weeks prior to surgery), and route (retro-orbital injection) are the same as those used in the ISOstudy. Aortic constriction in all TAC mice wasconfirmed by elevated trans-aortic pressure gradient(Figure 8B), establishing a similar increase in hemo-dynamic afterload in all mice receiving TAC. Kaplan-Meier curves summarizing severe HF-free (EF $35%)survival rates were included in Figure 8C. The rate ofsurvival in Bin1 HT mice is 20.0% (nonsurvival 8 of 10,2 deaths and 6 EF <35%), which is decreased from71.4% (non-survival 4 of 14, 1 death and 3 EF <35%) inWT mice (p ¼ 0.038 using log-rank test). As expected,all Sham mice survived through the entire experi-mental protocol (10 of 10, dotted black line). The rateof survival is decreased to 64.3% in the AAV9-GFPmice (non-survival 5 of 14, 5 EF <35%), which issignificantly improved to 93.7% in the AAV9-cBIN1group (non-survival 1 of 16, 1 EF <35%) (p ¼ 0.020using log-rank test). These data indicate that highercBIN1 protein content in the heart is associated withbetter survival without development of systolic HFfollowing pressure overload.

At 8 weeks after TAC, surviving mice were hu-manely killed and evaluated for ratios of HW/BW andlung weight (LW)/BW (Supplemental Table 3,Figures 8D and 8E). Both HW/BW and LW/BW aresignificantly higher in Bin1 HT mice verses WT mice,indicating worsening of LV hypertrophy and lungedema with BIN1 deficiency. Regarding gene therapy,AAV9-cBIN1 significantly reduces LW/BW from that ofthe control GFP-TAC group to the level of Shamhearts, representing a striking reduction in TAC-induced pulmonary edema. Hypertrophy still occursin the AAV9-cBIN1 hearts, although to a lesser extent.












FIGURE 6 Exogenous cBIN1 Brings Together Cav1.2-RyR and SERCA2a-cBIN1 Molecules in Cardiomyocytes

Super-resolution STORM imaging and nearest neighbor analysis of Cav1.2-RyR (A and B, Videos 1A, 1B, 1C, and 1D) and SERCA2a-cBIN1 (C and D, Videos 2A, and 2B)

molecules in cardiomyocytes isolated from GFPþPBS, cBIN1þPBS, GFPþISO, and cBIN1þISO hearts. (A, C) Top-to-bottom, representative 2-D STORM cell images,

representative 3-D STORM images of couplons, and histogram of nearest neighbor distance distribution obtained from full-cell 3-D STORM images. (B, D) Quantification

of the first peak of nearest neighbor distance distribution histogram using full-cell image analysis (N ¼ 7 to 17 cells from 2–3 animals per group). Data are expressed as

mean � SEM. Two-way ANOVA was used followed by Fisher LSD test for multiple comparison. *p < 0.05 for PBS vs. ISO comparison within each AAV9 treatment

group; #p < 0.05 and ##p < 0.001 for GFP vs. cBIN1 comparison within each drug infusion group. STORM ¼ stochastic optical reconstruction microscopy; other

abbreviations as in Figures 1 and 5.



572

These data indicate that exogenous cBIN1 reducesTAC-induced hypertrophy and prevents deteriorationinto HF. Echocardiography analysis (Figures 8F to 8J,Supplemental Table 4) further indicates that whencompared with WT-TAC mice, there is a significantdecrease in EF and an increase in EDV in Bin1 HT-TAChearts, indicating worsening of dilated cardiomyop-athy with BIN1 deficiency. On the other hand, AAV9-cBIN1 significantly reduces TAC-induced LV dilation

(EDV) and contractile dysfunction (EF), limiting thedevelopment of dilated cardiomyopathy. As a result,AAV9-cBIN1 pretreatment maintains stroke volumeand cardiac output in post-TAC hearts without heartsbeing dilated. Furthermore, tissue doppler identifiedthat the diastolic parameter E/e’ values of both lateraland septal walls are significantly improved in AAV9-cBIN1–pretreated hearts, indicating better diastolicfunction in mice with exogenous cBIN1. These data




http://jaccbts.acc.org/2020/0238_VID1C.mp4

http://jaccbts.acc.org/2020/0238_VID1D.mp4



FIGURE 7 Echocardiography of Post-ISO Hearts Receiving AAV9-GFP, cBIN1, BIN1, BIN1þ17, and BIN1þ13

(A) Representative LV short axis M-mode images from each group at baseline (top) and 4 weeks after ISO treatment (bottom). In all of the 4-week post-ISO images,

papillary muscles are marked by arrows (B to D). Quantitative analysis of LV mass (B), relative wall thickness (C), and ejection fraction (D) from each group (N ¼ 10

mice per group). (E) Representative mitral valve inflow pulsed wave Doppler images (top) and tissue Doppler images of septal mitral valve annulus 4 weeks after ISO

treatment (bottom). (F to H) Quantitative analysis of E/e’ (F), stroke volume (G), and cardiac output (H) from each group (N ¼ 10 mice per group). Data are expressed

as mean � SEM. Two-way ANOVA was used followed by Fisher LSD test for multiple comparison. *p < 0.05, **p < 0.01, and ***p < 0.001 vs. baseline; #p < 0.05,

##p < 0.01, and ###p < 0.001 vs. the GFP group at 4 weeks post-ISO. Abbreviations as in Figure 1.


573

FIGURE 8 cBIN1 Gene Transfer Improves Heart Failure–Free Survival in Post-TAC Mice




574


575

indicate that cBin1 gene therapy preserves myocardialsystolic and diastolic function in stressed hearts andeffectively prevents the development of dilated car-diomyopathy in mouse hearts subjected to pressureoverload.

DISCUSSION

This study indicates a beneficial effect of exogenouscBIN1 in preventing LV hypertrophy and cardiacdysfunction in stressed hearts. In mice subjected tocontinuous ISO infusion, exogenous cBIN1 offers anisoform-specific improvement in cardiac inotropy andlusitropy, limiting the development of LV hypertro-phy. The cardiac protective effect of exogenous cBIN1is further confirmed in mouse hearts subjected topressure overload–induced HF.

Chronically elevated catecholamine levels andactivation of cardiac b-ARs have a critical role in thepathogenesis of HF. Impaired myocardial structureand function have been observed in animals sub-jected to sustained sympathetic activation (28,29).ISO, which is a synthetic catecholamine and non-selective b-AR agonist, has been used in research toinduce the model of LV hypertrophy and dysfunction(30). A high dose of ISO was used here to induce LVconcentric hypertrophy with preserved systolicfunction. Chronic excessive cardiac workload–induced LV hypertrophy is associated with elevatedrisk of cardiovascular events (31), and preventing orreversing ventricular hypertrophy with preservedcardiac diastolic function is crucial to preventing theprogression of stressed hearts to failing hearts. Herewe found that cBIN1 attenuates chronic ISO-inducedhypertrophy and at the same time conveys anisoform-specific improvement in stroke volume andcardiac output in hypertrophic hearts with preservedsystolic function. The increase of LV volume in thecBIN1 hearts is not secondary to pump failure anddilated cardiomyopathy, but rather it reflectsimprovement in myocardial lusitropy (E/e’) with aparallel increase of intrinsic myocardial contractility(inotropy). This phenotype of cBIN1 hearts (Figure 1)

FIGURE 8 Continued

(A) Schematic protocol for the TAC study. (B) Trans-aortic pressure grad

failure–free survival (non-survival is death or EF <35%) in WT vs. Bin1 H

survival comparison. HW/BW (D) and LW/BW (E) in all mice at 8 weeks

surgery. Echocardiography-measured left ventricular ejection fraction (G

are expressed as mean � SEM. Representative E and e’ images in the A

nonparametric Mann-Whitney test used for comparison between WT an

comparison was used for comparison among Sham, AAV9-GFP, and AAV9

and ##p < 0.01 for comparison of AAV9-GFP vs. AAV9-cBIN1. EF ¼ ejec

WT ¼ wild type; other abbreviations as in Figure 1.

is typical of athletic hearts in adaptation endurancetraining as characterized by chamber enlargementand increases of LV volume, stroke volume, and car-diac output (32–34). Aerobic exercise training hasbeen reported to improve myocardial function andinotropic and lusitropic responses in both animalmodels (35,36) and patients with hypertension (37)and diastolic failure (38). Thus, exogenous cBIN1 mayprovide additional exercise-like benefit to patientswith HF, improving exercise capacity and quality oflife.

These post-ISO hearts are at a stage of hypertrophywith preserved systolic function, in which exogenouscBIN1 can effectively translate the increased demandson the heart into a functional effect. As a result, thesefunctional and efficient cBIN1 hearts have limitedhypertrophy development, which will likely preventthe next step of disease progression and HF devel-opment as occurs in the clinical setting. Next, thefunctional protective effect of exogenous cBIN1 inalready decompensated hearts is also observed in amouse model of TAC-induced hypertrophy and HF.Under pressure overload, compensated hypertrophyis an adaptive response. Over time, the adaptiveresponse concedes to cardiac dilatation and theensuing remodeling process becomes maladaptive,leading to worsening HF. We found that the fate ofdilated cardiomyopathy development in pressureoverload–stressed hearts is causally determined bymyocardial content of cBIN1 protein. Followingpressure overload, less cardiac BIN1 in geneticallydeleted Bin1 HT-TAC hearts is associated with moresevere dilated cardiomyopathy, whereas greatercBIN1 with gene transfer improves cardiac systolicand diastolic function, limits HF, and improves HF-free survival (Figure 8). It remains unclear whetherexogenous cBIN1 reduces myocyte death, which alsocontributes to LV dilation in post-TAC hearts. Futurestudies will be necessary to explore the effect ofcBIN1 on myocyte survival in stressed hearts. Never-theless, our data indicate that exogenous cBIN1 notonly limits hypertrophy development in stressedhearts but also prevents myocardial transition from

ient measurement in all mice 5 days post-surgery. (C) Kaplan-Meier survival curves for heart

T mice (left), and AAV9-GFP or cBIN1 pretreated mice (right). Log-rank test was used for

post-TAC. (F) Representative M-mode echocardiography images of all mice 8 weeks after

), end diastolic volume (H), LV mass (I), and E/e’ (J) for all mice at 8 weeks post-TAC. Data

AV9 treatment groups are included (right panel of J). Unpaired Student’s t-test (or

d Bin1 HT. One-way ANOVA or Kruskal-Wallis test followed by Fisher LSD test for multiple

-cBIN1. *p < 0.05, **p < 0.01, and ***p < 0.001 for comparison vs. WT or Sham; #p < 0.05

tion fraction; HT ¼ heterozygote; LW ¼ lung weight; TAC ¼ transverse aortic constriction;



576

hypertrophy to dilated cardiomyopathy and HF inTAC mice.

The mechanism of improvement in cardiacinotropic function by cBIN1 is linked to its known ef-fect in organizing t-tubule microdomains required fordyad organization and efficient EC coupling. cBIN1creates t-tubule microfolds to organize a slow diffu-sion zone trapping extracellular t-tubule lumen ions,attracts LTCCs forward trafficking to t-tubules (13),clusters LTCCs that are already delivered to cell sur-face (39), and recruits RyRs to couple with LTCCs atdyads (14). Here we confirm in vivo that exogenouscBIN1, rather than any other BIN1 isoform, increasesCav1.2 localization to t-tubules (Figure 3,Supplemental Figure 6). These results support thatpreserved cBIN1-microdomain with organized LTCCdistribution is responsible for the observed positiveinotropic effect in sympathetically overdriven cBIN1hearts. Whether cBIN1-microdomain regulates LTCCphosphorylation and its functional response to sym-pathetic stress including a well-established b-subunit-modulated Cav1.2 channel response (40,41) awaitsfuture experimental explorations. Furthermore, RyRis critical to inotropy and hyper-phosphorylated leakyRyR plays a role in HF progression (42). Consistentwith previous reports in an ISO model and human HF(25,42), we found that chronic ISO activates PKA andCAMKII-induced RyR hyperphosphorylation. AAV9-cBIN1 blunts these pathways, normalizing RyR phos-phorylation following chronic sympathetic activationand preventing SR leak.

An additional novel finding from the current studyis that exogenous cBIN1 increases SERCA2a functionthrough organizing its intracellular distribution(Figures 4 to 6). Chronic ISO-induced concentric hy-pertrophy with preserved systolic function is associ-ated with disorganized intracellular distribution ofSERCA2a yet increased overall protein expression. Itis well accepted that SERCA2a activity is decreased inend-stage HF. Our data indicate that, in addition toreduced expression and impaired regulation by PLN,intracellular distribution of SERCA2a may alsocontribute to abnormal SR calcium reuptake activityin HF. Furthermore, as reported in adult rat ventric-ular cardiomyocytes with a-receptor agonistphenylephrine-induced hypertrophy, an adaptive in-crease in SERCA2a protein expression can occur dueto elevated diastolic calcium-induced calcineurin/nuclear factor of activated T-cells activation (43).Thus, increased SERCA2a protein expression here is apossible adaptive response induced by elevated dia-stolic calcium concentration as indicated in elevatedcalcium-dependent phosphorylation at T287 of

CAMKII. Thus, a transient increase in SERCA2a mayoccur at an early stage of all LV hypertrophy withpreserved function. During disease progression, thisadaptive increase in total SERCA2a protein expres-sion will level off and even decrease as occurs in end-stage HF, resulting in severe diastolic and systolicfailure. In cBIN1 hearts, organized SERCA2a along theSR indicates better calcium reuptake, therefore, lessdiastolic calcium overload for hearts still at acompensated stage. These results are consistent witha previous study in a rat model of HF that identifiedthat increased BIN1 expression is associated withSERCA2a expression (44). Future studies exploringcBIN1 regulation of diastolic calcium concentrationand calcineurin/nuclear factor of activated T-cellspathways will be needed to further understand itsrole in regulating SERCA2a expression and activityduring disease progression. Note the effect on SER-CA2a organization is not cBIN1-specific and can bepartially induced by other BIN1 isoforms, particularlyBIN1þ17 (Supplemental Figure 6). This is consistentwith the partial in vivo protective effects fromBIN1þ17 on cardiac hypertrophy and diastolic func-tion (Figure 7). Whether and how BIN1 isoformscooperate to organize SERCA2a distribution in normaland diseased cardiomyocytes require further explo-ration in future studies. Furthermore, through regu-lation of calcium-handling machineries at the SR,including SERCA2a distribution and RyR phosphory-lation, cBIN1 may help maintain normal SR calciumload. As a limitation of the current study, future ex-periments are needed to quantify the effect of cBIN1on SR calcium load, calcium release and reuptakekinetics, and arrhythmogenic spontaneous calciumrelease in chronically stressed hearts.

Nevertheless, the most robust protection of bothinotropy and lusitropy in sympathetic overdrivenhearts is only observed in the cBIN1 group (Figure 7),indicating possible further beneficial effect on lusi-tropy from cBIN1-dependent improvement in LTCClocalization and dyad organization. With isoform-specific improvement in dyad organization, lessorphaned leaky RyRs accumulate outside of dyads(14), limiting calcium leak from the SR and decreasingcytosolic calcium concentration during the diastolicphase. Together with the newly identified role onSERCA2a organization, our data indicate that a cBIN1-microdomain–related regulation offers a uniquebenefit in protecting cardiac lusitropy in addition toits inotropy effect. On the other hand, cBIN1 over-expression may also suppress the pathologicaleffects of ISO stimulation by enhancing the control of

b-AR signaling and the compartmentalization of



PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: The implications

describe the effects of exogenous cBIN1 expression on cardiac

remodeling and function in mouse hearts subjected to chronic

sympathetic overdrive.

TRANSLATIONAL OUTLOOK: cBIN1 replacement therapy

with a positive inotropy and lusitropy effect can be used as a

potential new treatment option for HF.


577

secondary messengers and calcium-handling chan-nels and pumps. Thus, by stabilizing t-tubule micro-domains to regulate all aspects of calcium handling,cBIN1 produces a positive feedforward mechanism forefficient intracellular beat-to-beat calcium cycling. Infuture studies it will be interesting to identifywhether exogenous cBIN1 alters b-AR expression,intracellular distribution, and functional regulationfollowing chronic sympathetic activation.

CONCLUSIONS

We found that over-expression of exogenous cBIN1 isprotective in mouse hearts subjected to chronic b-ARactivation–induced concentric hypertrophy as well aspressure overload–induced hypertrophy and HF. Asthe first proof-of-concept study, there are a few lim-itations to our studies that need to be taken intoconsideration. First, the models used are limited tomouse models subjected to exaggerated stressesincluding a high dose of ISO infusion and severepressure overload induced by aortic constriction.Future experiments in large mammals with commonnatural HF comorbidities, such as hypertension anddiabetes, will be needed. Second, we used an AAV9vector driven by the CMV promoter for gene deliverydue to its consistent transduction efficiency andestablished cardiac tropism. Note the protective ef-fect of exogenous cBIN1 in the current study occurredwhen only 60% of cardiomyocytes were transducedby AAV9 with detectable expression of exogenousprotein. Improving the viral infectivity in car-diomyocytes can additionally help limit or preventISO-induced membrane disruption in all car-diomyocytes, increasing the protective effect on theentire heart. Further experiments using cBin1 pack-aged in AAV9 with an efficient and cardiac-specificpromoter to induce sufficient exogenous proteinexpression in all cardiomyocytes will be neededbefore clinical trials testing the efficacy and efficiencyof cBin1 gene therapy. Third, the current study lacks afull evaluation of hemodynamic parameters such as

blood pressure. Although a similar HR increase in allpost-ISO mice indicates effective and similar hemo-dynamic stress in animals with or without cBIN1,future studies are needed to establish whether cBIN1will impact systemic hemodynamics and blood pres-sure. Finally, future studies are needed to explorehow the cBIN1-microdomain regulates the organiza-tion of intracellular calcium-handling machineries,EC coupling, SR calcium load and release, diastoliccalcium concentration and its downstream calcium-signaling pathways, interplay between signalingpathways of pathological and physiological hyper-trophic remodeling, and molecular transition fromcompensated hypertrophy to decompensatedcardiomyopathy.

ACKNOWLEDGMENTS The authors thank the Elec-tron Imaging Center at California NanoSystemsInstitute of University of California-Los Angeles fortransmission electron microscopy, the Cedars-SinaiMedical Center Histology Core for preparing cry-osections, and Sarcotein Diagnostics for the recom-binant anti-BIN1 exon 13 antibody.

ADDRESS FOR CORRESPONDENCE: Dr. TingTingHong, Smidt Heart Institute, Cedars-Sinai MedicalCenter, 8700 Beverly Boulevard, 1028 Plaza Level,Davis Building, Los Angeles, California 90048.E-mail: [email protected].

RE F E RENCE S

1. Virani SS, Alonso A, Benjamin EJ, et al. HeartDisease and Stroke Statistics-2020 Update:A Report From the American Heart Association.Circulation 2020;141:e139–596.

2. Verloop WL, Beeftink MM, Santema BT, et al.A systematic review concerning the relation be-tween the sympathetic nervous system andheart failure with preserved left ventricularejection fraction. PLoS One 2015;10:e0117332.

3. Nikolaev VO, Moshkov A, Lyon AR, et al.Beta2-adrenergic receptor redistribution inheart failure changes cAMP compartmentation.Science 2010;327:1653–7.

4. Ibrahim M, Gorelik J, Yacoub MH,Terracciano CM. The structure and function ofcardiac t-tubules in health and disease. Proc BiolSci 2011;278:2714–23.

5. Schobesberger S, Wright P, Tokar S, et al. T-tubule remodelling disturbs localized beta2-adrenergic signalling in rat ventricular myocytesduring the progression of heart failure. CardiovascRes 2017;113:770–82.

6. Balycheva M, Faggian G, Glukhov AV, Gorelik J.Microdomain-specific localization of functional ionchannels in cardiomyocytes: an emerging concept





























578

of local regulation and remodelling. Biophys Rev2015;7:43–62.

7. Hong T, Shaw RM. Cardiac t-tubule micro-anatomy and function. Physiol Rev 2017;97:227–52.

8. Glukhov AV, Balycheva M, Sanchez-Alonso JL,et al. Direct evidence for microdomain-specificlocalization and remodeling of functional L-Typecalcium channels in rat and human atrial myo-cytes. Circulation 2015;132:2372–84.

9. Sanchez-Alonso JL, Bhargava A, O’Hara T, et al.Microdomain-specific modulation of L-type cal-cium channels leads to triggered ventriculararrhythmia in heart failure. Circ Res 2016;119:944–55.

10. Wei S, Guo A, Chen B, et al. T-tubule remod-eling during transition from hypertrophy to heartfailure. Circ Res 2010;107:520–31.

11. Bers DM. Cardiac excitation-contractioncoupling. Nature 2002;415:198–205.

12. Hong T, Yang H, Zhang SS, et al. Cardiac BIN1folds T-tubule membrane, controlling ion flux andlimiting arrhythmia. Nat Med 2014;20:624–32.

13. Hong TT, Smyth JW, Gao D, et al. BIN1 localizesthe L-type calcium channel to cardiac T-tubules.PLoS Biol 2010;8:e1000312.

14. Fu Y, Shaw SA, Naami R, et al. Isoproterenolpromotes rapid ryanodine receptor movement toBridging Integrator 1 (BIN1)-organized dyads. Cir-culation 2016;133:388–97.

15. Hong TT, Smyth JW, Chu KY, et al. BIN1 isreduced and Cav1.2 trafficking is impaired in hu-man failing cardiomyocytes. Heart Rhythm 2012;9:812–20.

16. Caldwell JL, Smith CE, Taylor RF, et al.Dependence of cardiac transverse tubules on theBAR domain protein amphiphysin II (BIN-1). CircRes 2014;115:986–96.

17. Basheer WA, Xiao S, Epifantseva I, et al. GJA1-20k arranges actin to guide Cx43 delivery to car-diac intercalated discs. Circ Res 2017;121:1069–80.

18. Zincarelli C, Soltys S, Rengo G, Rabinowitz JE.Analysis of AAV serotypes 1-9 mediated geneexpression and tropism in mice after systemic in-jection. Mol Ther 2008;16:1073–80.

19. Chen BD, He CH, Chen XC, et al. Targetingtransgene to the heart and liver with AAV9 bydifferent promoters. Clin Exp Pharmacol Physiol2015;42:1108–17.

20. Denegri M, Bongianino R, Lodola F, et al.Single delivery of an adeno-associated viralconstruct to transfer the CASQ2 gene to knock-inmice affected by catecholaminergic polymorphicventricular tachycardia is able to cure the diseasefrom birth to advanced age. Circulation 2014;129:2673–81.

21. Amoasii L, Hnia K, Chicanne G, et al. Myotu-bularin and PtdIns3P remodel the sarcoplasmicreticulum in muscle in vivo. J Cell Sci 2013;126:1806–19.

22. Fu Y, Shaw RM. CASAAV technology toexamine regulators of heart failure: cause or ef-fect. Circ Res 2017;120:1846–8.

23. Gaasch WH, Zile MR. Left ventricular structuralremodeling in health and disease: with specialemphasis on volume, mass, and geometry. J AmColl Cardiol 2011;58:1733–40.

24. Periasamy M, Janssen PM. Molecular basis ofdiastolic dysfunction. Heart Fail Clin 2008;4:13–21.

25. Shan J, Kushnir A, Betzenhauser MJ, et al.Phosphorylation of the ryanodine receptor medi-ates the cardiac fight or flight response in mice.J Clin Invest 2010;120:4388–98.

26. Boknik P, Fockenbrock M, Herzig S, et al.Protein phosphatase activity is increased in a ratmodel of long-term beta-adrenergic stimulation.Naunyn Schmiedebergs Arch Pharmacol 2000;362:222–31.

27. Kirchhefer U, Hammer E, Heinick A, et al.Chronic beta-adrenergic stimulation reversesdepressed Ca handling in mice overexpressinginhibitor-2 of protein phosphatase 1. J Mol CellCardiol 2018;125:195–204.

28. Osadchii OE. Cardiac hypertrophy induced bysustained beta-adrenoreceptor activation: patho-physiological aspects. Heart Fail Rev 2007;12:66–86.

29. Soltysinska E, Olesen SP, Osadchii OE.Myocardial structural, contractile and electro-physiological changes in the guinea-pig heartfailure model induced by chronic sympatheticactivation. Exp Physiol 2011;96:647–63.

30. Balakumar P, Singh AP, Singh M. Rodentmodels of heart failure. J Pharmacol ToxicolMethods 2007;56:1–10.

31. Frey N, Katus HA, Olson EN, Hill JA. Hyper-trophy of the heart: a new therapeutic target?Circulation 2004;109:1580–9.

32. Oliveira RS, Ferreira JC, Gomes ER, et al. Car-diac anti-remodelling effect of aerobic training isassociated with a reduction in the calcineurin/NFAT signalling pathway in heart failure mice.J Physiol 2009;587:3899–910.

33. Neves VJ, Fernandes T, Roque FR, Soci UP,Melo SF, de Oliveira EM. Exercise training in hy-pertension: role of microRNAs. World J Cardiol2014;6:713–27.

34. Ooi JY, Bernardo BC, McMullen JR. The ther-apeutic potential of miRNAs regulated in settingsof physiological cardiac hypertrophy. Future MedChem 2014;6:205–22.

35. MacDonnell SM, Kubo H, Crabbe DL, et al.Improved myocardial beta-adrenergic respon-

siveness and signaling with exercisetraining in hypertension. Circulation 2005;111:3420–8.

36. Hoydal MA, Stolen TO, Kettlewell S, et al.Exercise training reverses myocardial dysfunctioninduced by CaMKIIdeltaC overexpression byrestoring Ca2þ homeostasis. J Appl Physiol 2016;121:212–20.

37. Pitsavos C, Chrysohoou C, Koutroumbi M, et al.The impact of moderate aerobic physical trainingon left ventricular mass, exercise capacity andblood pressure response during treadmill testingin borderline and mildly hypertensive males. Hel-lenic J Cardiol 2011;52:6–14.

38. Dieberg G, Ismail H, Giallauria F, Smart NA.Clinical outcomes and cardiovascular responses toexercise training in heart failure patients withpreserved ejection fraction: a systematic reviewand meta-analysis. J Appl Physiol 2015;119:726–33.

39. Fu Y, Hong T. BIN1 regulates dynamic t-tubulemembrane. Biochim Biophys Acta 2016;1863:1839–47.

40. Yang L, Katchman A, Kushner J, et al.Cardiac CaV1.2 channels require beta subunitsfor beta-adrenergic-mediated modulationbut not trafficking. J Clin Invest 2019;129:647–58.

41. Miriyala J, Nguyen T, Yue DT, Colecraft HM.Role of CaVbeta subunits, and lack of func-tional reserve, in protein kinase A modulationof cardiac CaV1.2 channels. Circ Res 2008;102:e54–64.

42. Marx SO, Reiken S, Hisamatsu Y, et al. PKAphosphorylation dissociates FKBP12.6 from thecalcium release channel (ryanodine receptor):defective regulation in failing hearts. Cell 2000;101:365–76.

43. Anwar A, Taimor G, Korkusuz H, et al.PKC-independent signal transduction path-ways increase SERCA2 expression in adult ratcardiomyocytes. J Mol Cell Cardiol 2005;39:911–9.

44. Lyon AR, Nikolaev VO, Miragoli M, et al.Plasticity of surface structures and beta(2)-adrenergic receptor localization in failingventricular cardiomyocytes during recoveryfrom heart failure. Circ Heart Fail 2012;5:357–65.

KEY WORDS diastolic dysfunction, heartfailure, pressure overload, sympatheticoverdrive, t-tubule

APPENDIX For an expanded methods sectionas well as supplemental figures, videos, andtables, please see the online version of thispaper.










































































































































































EDITORIAL COMMENT

Cardiac BIN1 Replacement TherapyAmeliorates Inotropy and Lusitropyin Heart Failure by RegulatingCalcium Handling*
Jessica Gambardella, PHD,a,b Xu Jun Wang, MD,a,c John Ferrara, BS,a Marco Bruno Morelli, PHD,a,c
Gaetano Santulli, MD, PHDa,b,c,d

B ridging integrator 1 (BIN1) is a scaffold proteinbelonging to the superfamily of proteins con-taining the BAR domain, first identified by its

common occurrence in vertebrate Bin1 and Amphi-physins and in the Saccharomyces cerevisiae Rvs pro-teins. Specifically, BIN1 has its BAR domain within itsN-terminus, encoded by exons 1 to 9. Architecturally,BIN1 forms a homodimer through the interaction be-tween helixes of the BAR domain from each mono-mer; such interaction produces a peculiar spatialdisposition with a 6-helix bundle formation, confer-ring an elongated banana shape to BIN1 dimers. TheBIN1 gene is composed of 20 exons, the alternativesplicing of which generates tissue-specific BIN1 iso-forms with different functions. Among more than 10

ISSN 2452-302X




From the aDepartment of Medicine, Wilf Family Cardiovascular Research

Institute, Albert Einstein College of Medicine, New York, New York;bInternational Translational Research and Medical Education Con-

sortium, Naples, Italy; cDepartment of Molecular Pharmacology, Einstein-

Mount Sinai Diabetes Research Center, Fleischer Institute for Diabetes

and Metabolism, Albert Einstein College of Medicine, New York, New

York; and the dDepartment of Advanced Biomedical Sciences, “Federico

II” University, Naples, Italy. Dr. Santulli’s lab is supported in part by the

National Institutes of Health (R01-DK123259, R01-HL146691, R01-

DK033823, P30-DK020541, and R00-DK107895 to Dr. Santulli) and by

the American Heart Association (AHA-20POST35211151 to Dr. Gambar-

della). The authors have reported that they have no relationships rele-

vant to the contents of this paper to disclose.






isoforms, there are 2 cardiac-specific spliced variants:BIN1þ13 and Bin1þ13þ17, also known as cardiac-BIN1(cBIN1). BIN1þ13 is the most abundant isoform inthe heart and seems to be implicated in cell prolifera-tion. cBIN1 is the isoform localized on cardiac trans-verse tubules (T-tubules), where it is indispensablein packaging T-tubule membrane microfolds (1),exploiting its membrane-curving abilities. In additionto enabling correct invagination of the T-tubules,cBIN-1 mediates the formation of functional com-plexes between the L-type calcium channel (LTCC,also known as dihydropyridine channel) and theintracellular calcium release channels ryanodine re-ceptor type 2 (RyR2); these functional complexes areknown as LTCC-RyR2 dyads. As a direct consequence,cBIN1 has a pivotal role in ensuring a proper contrac-tion of cardiomyocytes and has been proposed as apotential therapeutic target in heart failure (HF).Indeed, levels of cBIN1 have been shown to bereduced in animal models of HF, as well as in humanbiopsy samples from patients with end-stagecardiomyopathy.

In this issue of JACC: Basic to Translational Science,Liu et al. (2) evaluate the effects of cBIN1 replacementtherapy in murine HF models obtained by chronicexposure to isoproterenol (ISO) or by pressure over-load (transverse aortic constriction [TAC]). In boththe ISO and TAC models of HF, normalization of cBIN1levels through adeno-associated virus 9 (AAV9)–mediated gene therapy was shown to effectivelyrestore T-tubule/c-BIN1 microfolds and the distribu-tion of calcium regulatory proteins. This molecularremodeling was associated with an improvement ofsystolic and diastolic function in the ISO and TACmodels. A key finding is that only BIN1-13þ17 (cBIN1)






Gambardella et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

cBIN1 Improves Myocardial Function in HF J U N E 2 0 2 0 : 5 7 9 – 8 1

580

administration, rather than other isoforms of cBIN,increases LTCC localization to T-tubules and affectsT-tubule remodeling. An additional novel finding inthe study by Liu et al. is that exogenous cBIN1 in-creases SERCA2a function through organizing itsintracellular distribution, although this was shown tobe not specific for the cBIN1 isoform. Before discus-sing this paper further, it is useful to digress for amoment and review the role of cBIN and calciumhandling in the normal and failing heart.

cBIN1 AND CARDIAC CALCIUM HANDLING

After the depolarization of the sarcolemma mem-brane, a relatively small amount of calcium entersinto cardiomyocytes via the LTCC. Calcium entry thenelicits a massive calcium release from the sarco-plasmic reticulum (SR) via RyR2 channels (3,4). Thismechanism, referred to as calcium-induced calciumrelease, activates the contractile machinery (3). Theefficacy of calcium release from the SR essentiallydepends on the proximity of LTCC channels on theplasma membranes and RyR2 channels on the SR.This feature is one of the main reasons why theplasma membrane of cardiac cells is organized in T-tubules, membrane invaginations that ensure a pre-cise closeness between LTCC and RyR2 channels. Liuet al. (2) did not actually measure the effects of cBIN1replacement on SR calcium load, calcium leak, orcalcium reuptake. Nevertheless, cBIN1 seems tostructurally and functionally contribute to LTCC andRyR2 channel coupling, at least via 3 mechanisms: 1)through its banana-like shape, cBIN1 produces thefitting membrane invagination in the T-tubule; 2)cBIN1 can act as an anchor protein for LTCCs, regu-lating their trafficking and clustering at the dyadlevel; 3) cBIN1 recruits the other major dyadic protein,RyR2, creating the perfect molecular bridge between2 calcium channels on the 2 compartments, plasmamembrane and SR (3).

Another important aspect is that the cBIN1-mediated bridge is not a static assemblage; rather, itis a dynamic spatial and temporal organizationcapable of responding to stress, for example, bysupporting the increased cardiac output demand.Indeed, the recruitment of RyR2 by cBIN1 is phos-phorylation dependent: this selective action onSer2808-phosphorylated RyR2 allows the dyads toinclude the channels with increased calcium sensi-tivity and open probability, thereby supporting thehigher cardiac demand during stress (4). Further-more, the cBIN1 dyad complexes can be rapidly ar-ranged in response to different stimuli. For instance,the activation of b-adrenergic receptors by

catecholamines during stress is known to induce arapid cBIN1 redistribution, probably acting on theaccumulation of BIN1-binding phospholipids. Givenits crucial role in ensuring the structural and elec-trical stability of cardiac dyads, it is not surprisingthat a compromised cBIN1 function contributes tocontractile pump failure and/or arrhythmias, furtherunderscoring the fundamental importance of thismolecule both as a disease marker and therapeutictarget.

cBIN1 IN THE PATHOGENESIS OF HF

AND ARRHYTHMIAS

T-tubule remodeling with the consequent dyaduncoupling and altered cardiac calcium transients is awell-established hallmark of HF development andprogression (1,3). The alterations of cBIN1 observed inHF provide another proof of concept of the functionalcontribution of T-tubule dysregulation to HF patho-physiology. The first evidence supporting the role ofcBIN1 in HF came from the observation of a significantdown-regulation of BIN1 in human failing car-diomyocytes, both at the messenger ribonucleic acidand protein levels; BIN1 knockdown in murine andzebrafish cardiomyocytes triggers a significantimpairment in calcium transients and a severe con-tractile dysfunction. These findings prompted similarresearch in other animal models, confirmingdecreased cBIN1 and T-tubule density in sheepmodels of HF induced by right ventricular tachypac-ing or ascending aortic coarctation. In both models,the alterations in cBIN1 and T-tubule density corre-late with amplitude reduction and higher heteroge-neity of systolic calcium transients.

Whereas the evidence described comes mostlyfrom observational studies, what happens when BIN1is actively removed in the heart? Cardiac-specificablation of the BIN1 gene is sufficient to generateper se a model of HF in aged mice, causing dilatedcardiomyopathy with an approximately 45% reduc-tion in ejection fraction beginning at approximately 8to 10 months of age. Interestingly, although youngeranimals are protected from HF, they rapidly develop afrank dilated cardiomyopathy in response to pressureoverload. At the molecular level, Bin1 cardiac defi-ciency induces mislocalization of voltage-dependentcalcium channels Cav1.2 and interstitial fibrosis.Additionally, the gene encoding for BIN1 seems to behaploinsufficient, as heterozygous mice exhibit anintermediate phenotype. The loss of cBIN1 stronglyaffects electrical coupling in cardiomyocytes,increasing the risk of arrhythmogenesis. Actually,when cBIN1 is not available, the formation of leaky

FIGURE 1 Main Molecular Mechanisms Underlying the Importance of Cardiac BIN1

in HF

The banana-shaped structure of the homodimer formed by 2 BAR domains in BIN1 is

responsible for the structural and functional stability of cardiac dyads, therefore playing

a crucial role in the regulation of excitation-contraction coupling. BIN1 ¼ bridging inte-

grator; HF ¼ heart failure; T-tubule ¼ transverse tubule.

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Gambardella et al.J U N E 2 0 2 0 : 5 7 9 – 8 1 cBIN1 Improves Myocardial Function in HF

581

RyR2 clusters occurs outside the dyads, leading toarrhythmias (1). Moreover, the reduction of cBIN1 asoccurs in cardiomyopathies leads to free diffusion oflocal extracellular calcium and potassium ions,thereby prolonging the action potential duration andincreasing susceptibility to cardiac arrhythmias.

As mentioned, cBIN1 levels seem to reflect cardiaccalcium-induced calcium release efficiency and theresultant pumping capacity of the heart. The patho-genic mechanisms that emerged in preclinical modelshave been recently corroborated in humans, demon-strating that reduced plasma BIN1 levels can predictthe incidence of arrhythmias, making cBIN1 a noveland promising biomarker of cardiac (dys)function.Nikolova et al. (5) developed a cBIN1 score based oncirculating levels of cBIN1 detected in HF with pre-served ejection fraction (HFpEF), healthy controlparticipants, and a control participant who had riskfactors for developing HF. The cBIN1 score wascomputed as the natural log of the inverse ofmeasured cBIN1 plasma concentration. The authorsreported that the cBIN1 score was similar amonghealthy participants, as well as among patients withrisk factors but not HF, whereas the score is signifi-cantly higher in HFpEF patients. Kaplan-Meier anal-ysis of 1-year cardiovascular hospitalizations adjustedfor age, sex, body mass index, and N-terminal pro–B-type natriuretic peptide levels showed that patientswith HFpEF with a cBIN1 score $1.80 had a 3.8-foldgreater risk for hospitalizations compared with thosewith scores of <1.80. In this context, cBIN1 not onlyhas a predictive value for HFpEF diagnosis but also aprognostic value for the hospitalization of patientsduring the first year of follow-up.

The report by Liu et al. (2), extends our knowledgewith respect to the molecular mechanisms (Figure 1)linking cBIN1 to left ventricular hypertrophy and

diastolic dysfunction. Whether the murine modelsused in this study are relevant to patients with HFpEFis unknown and will require further validation indifferent models of HFpEF. Nonetheless, thesestudies raise the interesting possibility that normal-izing myocardial cBIN1 levels may improve outcomesin HF with a reduced and preserved EF.

ADDRESS FOR CORRESPONDENCE: Dr. JessicaGambardella, Department of Medicine, Wilf FamilyCardiovascular Research Institute, Albert EinsteinCollege of Medicine, 1300 Morris Park Avenue, SuiteG35, New York, New York 10461. E-mail:[email protected].

RE F E RENCE S

1. Hong T, Yang H, Zhang SS, et al. Cardiac BIN1folds T-tubule membrane, controlling ion flux andlimiting arrhythmia. Nat Med 2014;20:624–32.

2. Liu Y, Zhou K, Li J, et al. In mice subjected tochronic stress, exogenous cBIN1 preserves calciumhandling machinery and cardiac function. J AmColl Cardiol Basic Trans Science 2020;5:561–78.

3. Gambardella J, Trimarco B, Iaccarino G,Santulli G. New insights in cardiac calcium

handling and excitation-contraction coupling. AdvExp Med Biol 2018;1067:373–85.

4. Santulli G, Xie W, Reiken SR, Marks AR. Mito-chondrial calcium overload is a key determinant inheart failure. Proc Natl Acad Sci U S A 2015;112:11389–94.

5. Nikolova AP, Hitzeman TC, Baum R, et al. As-sociation of a novel diagnostic biomarker, theplasma cardiac bridging integrator 1 score, with

heart failure with preserved ejection fraction andcardiovascular hospitalization. JAMA Cardiol 2018;3:1206–10.

KEY WORDS arrhythmias, calcium handling,cBIN1, diastolic dysfunction, dyads, heart failure,HFpEF, pressure overload, RyR, sympatheticoverdrive, t-tubule
























ª 2 0 2 0 T H E A U T H O R S . P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E AM E R I C A N




Sex-Specific Effects of the Nlrp3Inflammasome on Atherogenesis inLDL Receptor-Deficient Mice
Shuang Chen, MD, PHD,a Janet L. Markman, PHD,a Kenichi Shimada, PHD,a Timothy R. Crother, PHD,a
Malcolm Lane, BS,a Amanda Abolhesn, MS,a Prediman K. Shah, MD,a,b,* Moshe Arditi, MDa,b,*

VISUAL ABSTRACT

IS

Chen, S. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):582–98.

HIGHLIGHTS

� In this study we observed sex-specific effects of the NLRP3 inflammasome on atherogenesis in LDLR-deficient mice, with

NLRP3 inflammasome playing a more prominent role in atherosclerosis in female mice than in males.

� Sex hormones may be involved in NLRP3 inflammasome–mediated atherogenesis and may underlie differential responses

to anti-NLRP3 therapy between males and females.

� Testosterone may play an inhibitory role by blocking NLRP3 inflammasome and inflammation in atherogenesis,

whereas female sex hormones may promote NLRP3 inflammasome–mediated atherosclerosis.

� The results of the present study may help design future clinical trials, with the objective to personalize

cardiovascular care for men and women.

SN 2452-302X https://doi.org/10.1016/j.jacbts.2020.03.016




R E V I A T I O N S

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Chen et al.J U N E 2 0 2 0 : 5 8 2 – 9 8 Role of Sex in IL-b and Atherosclerosis

583

SUMMARYAB B

AND ACRONYM S

ACVD = atherosclerotic

cardiovascular disease

BM = bone marrow

CAS = castration

ER = estrogen receptor

HFD = high-fat diet

FLICA = fluorescent labeled

inhibitors of caspases

IL = interleukin

NLRP3 = nucleotide-binding

Fro

Ce

Ce

thi

Dr

pu

thi

Th

ins

vis

Ma

In the Ldlr-/- mouse model of atherosclerosis, female Nlrp3-/- bone marrow chimera and Nlrp3-/- mice developed

significantly smaller lesions in the aortic sinus and decreased lipid content in aorta en face, but a similar

protection was not observed in males. Ovariectomized female mice lost protection from atherosclerosis in the

setting of NLRP3 deficiency, whereas atherosclerosis showed a greater dependency on NLRP3 in castrated

males. Thus, castration increased the dependency of atherosclerosis on the NLRP3 inflammasome, suggesting

that testosterone may block inflammation in atherogenesis. Conversely, ovariectomy reduced the dependency

on NLRP3 inflammasome components for atherogenesis, suggesting that estrogen may promote

inflammasome-mediated atherosclerosis. (J Am Coll Cardiol Basic Trans Science 2020;5:582–98)

© 2020 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an

open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
in and leucine-rich repeat
ining (NLR) protein3

doma

conta

OVX = ovariectomy

A therosclerotic cardiovascular disease (ACVD)is the leading cause of morbidity and mortal-ity globally in men and women (1). To date,

most clinical studies on ACVD have primarilyincluded men, and the knowledge about ACVD inwomen has been largely based on extrapolation.Although more men than women die from ACVD,and men develop disease at a younger age (40 to 60years of age) (2,3), women have higher mortalitytrends in ACVD (4,5), and experience more complica-tions, such as bleeding and coronary vascular injury(6). Plaque erosion, the cause of coronary thrombosisand acute myocardial infarction, occurs at a higherfrequency in women than in men (7,8). Recent evi-dence highlighted ACVD risk factors exclusive towomen (9), including common disorders of preg-nancy, such as gestational hypertension and diabetes,and frequently occurring endocrine disorders inwomen of reproductive age (e.g., polycystic ovarysyndrome and early menopause) (10,11) caused byhormonal dysregulation. In addition, women withautoimmune disease are at an increased risk ofdeveloping ACVD (12).

Inflammation contributes to all stages of athero-sclerosis, from plaque formation to instability and

m the aDepartment of Biomedical Sciences, Infectious and Immunolog

nter, Los Angeles, California; and the bDivision of Cardiology, Department

nter, Smidt Heart Institute, Cedars-Sinai Medical Center, Los Angeles, Cal

s work and are joint corresponding authors. This work was supported by

. Arditi) and HL111483 (to Dr. Chen). The funders had no role in stud

blish, or preparation of the manuscript. All authors have reported that th

s paper to disclose.

e authors attest they are in compliance with human studies committe

titutions and Food and Drug Administration guidelines, including patien

it the JACC: Basic to Translational Science author instructions page.

nuscript received January 30, 2020; revised manuscript received March 1

final plaque rupture (13). Multiple studies havehighlighted the prominent role of the nucleotide-binding domain and leucine-rich repeat (NLR) pyrindomain containing protein3 (NLRP3) inflammasomeand interleukin (IL)-1 cytokines in atherogenesis(14-17), and IL-1a and IL-1b have been observed inhuman atherosclerotic plaques (18). However, therole of the NLRP3 inflammasome pathway in diet-induced acceleration of atherosclerosis is stillcontroversial, with 2 main groups reporting con-trasting results in experimental mouse models.Although Duewell et al. (19) demonstrated a proa-therogenic role for the NLRP3 inflammasome acti-vation in response to cholesterol in Ldlr-/- mice,Menu et al. (20) reported no differences in athero-sclerosis progression in mice with genetic deletion ofkey inflammasome components. The latter studyused ApoE -/- mice and 8-fold higher cholesterol inthe diet compared with the former study (21). How-ever, another key difference between these 2experimental studies with that whereas Duewellet al. (19) clearly described the use of female mice,Menu et al. (20) did not state the sex of the miceused. Emerging evidence has shown that estrogencan act as an inflammatory protective factor to

ic Diseases Research Center, Cedars-Sinai Medical

of Medicine, Oppenheimer Atherosclerosis Research

ifornia. *Drs. Shah and Arditi contributed equally to

National Institutes of Health Grants HL66436-05 (to

y design, data collection and analysis, decision to

ey have no relationships relevant to the contents of

es and animal welfare regulations of the authors’

t consent where appropriate. For more information,

1, 2020, accepted March 11, 2020.



Chen et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Role of Sex in IL-b and Atherosclerosis J U N E 2 0 2 0 : 5 8 2 – 9 8

584

suppress NLRP3-mediated neuroinflammation in thehippocampus (22,23). However, the relationship be-tween NLRP3 and estrogen in ACVD has not beenelucidated.

Several studies strongly suggest that the key dif-ferences in the immune-inflammatory processes andresulting inflammatory infiltrate between men andwomen with ACVD may be driven by sex hormones(24). Current dogma holds that estrogen has anti-inflammatory effects, whereas testosterone pro-motes inflammation (24). Indeed, the finding that theincidence of ACVD increases in women as estrogendeclines with age and following menopause could beinterpreted to indicate a protective role for estrogenin the heart (24). However, in clinical studieshormone-replacement therapy has failed to decreaseACVD events (25-27), emphasizing the complexity ofthe relationship between vascular biology and estro-gen hormones. Indeed, the role of estrogen signalingon expression of IL-1b seems to differ depending oncell type (22,28-30). Similarly, although in generaltestosterone is believed to promote innate immunecell activation and production of proinflammatorycytokines, there are many conflicting studies (31).Many studies now suggest that testosterone inhibitsatherosclerosis (32-36), and that testosterone defi-ciency increases the risk of atherosclerotic events (37-39). The reason for these conflicting findings may bethat the understanding of the effect of sex hormoneson immune cells is derived mainly from cell cultureand animal studies of normal, healthy cells, ratherthan disease contexts. Importantly, mechanisticstudies examining sex differences in inflammationduring atherosclerosis have, for the most part, not yetbeen conducted.

CANTOS (Canakinumab Anti-inflammatoryThrombosis Outcome Study) recently demonstratedmodest but significant therapeutic benefit of treat-ment with a monoclonal antibody targeting only IL-1b(canakinumab) in patients with previous myocardialinfarction (40). A secondary analysis showed thatsubgroups of women and men achieved similar clin-ical efficacy with canakinumab (41), despite only 26%of the participants being female, indicating that asmaller sample size was needed for females to ach-ieve the same clinical benefit as males. These resultssuggest a sex-specific difference in the therapeuticresponses to IL-1b inhibition, where females may bemore responsive than males. Although the results ofthe CANTOS trial are a milestone in cardiovascularmedicine, the safety concerns and potentially pro-hibitive cost make it unlikely that canakinumab willultimately be used for secondary prevention.

Therefore, finding ways to identify subsets of pa-tients who derive maximum benefits from canakinu-mab (or other anti-inflammatory agents) is critical.Here, we investigated the role of sex in NLRP3inflammasome–mediated inflammation in athero-sclerosis as a first step toward identifying these pa-tient subsets.

METHODS

ANIMAL STUDIES. All animal experiments were per-formed according to the guidelines and approvedprotocols (Protocol #8299) of the Cedars-Sinai Medi-cal Center Institutional Animal Care and Use Com-mittee. Cedars-Sinai Medical Center is fullyaccredited by the Association for Assessment andAccreditation of Laboratory Animal Care and abidesby all applicable laws governing the use of laboratoryanimals. Laboratory animals are maintained inaccordance with the applicable portions of the AnimalWelfare Act and the guidelines prescribed in the U.S.Department of Health and Human Services publica-tion, Guide for the Care and Use of Labora-tory Animals.

MICE. All mice were on the C57BL/6 background forthese studies. Both male and female Nlrp3�/�Ldlr�/�

and Ldlr�/� mice were used (42). For bone marrow(BM) transplantation, BM from wild-type, and Nlrp3-/-

mice was transplanted into irradiated Ldlr-/- mice.After recovery (6 weeks), chimeric mice were placedon a high-fat diet (HFD) containing 0.15% cholesterol(Harlan Teklad) for 12 weeks (42,43).

CASTRATION. One week before HFD, Nlrp3�/�Ldlr�/�

and Ldlr�/� male mice of 8 weeks of age underwentcastration (CAS) or sham-surgery. Mice were main-tained on inhalation anesthesia (1.5% isoflurane) vianose-cone. Before the start of the surgery, carprofen(5 mg/kg body weight) was administered subcuta-neously. The area between the penis and the anuswas shaved and cleaned with betadine followed byalcohol to disinfect the scrotum. The area betweenthe penis and the anus was lifted to make a small 1-mm horizontal cut. To remove the testes, a small 1-mm cut into the inner skin membrane enclosing thetesticles was made and the testicles were exterior-ized. Testicular arteries were tied off using resorb-able Vicryl sutures before removing the testes. Oncethe testes were removed, the wound was sealedwith 2 nylon sutures. Following spontaneousmovement, buprenorphine (0.5 mg/kg body weight)and 300 ml of warm saline were administered sub-cutaneously. In sham-operated mice, both the skinand inner skin membrane between the penis and

TABLE 1 Total Cholesterol Level, Lipoprotein Profile and Triglyceride

Concentrations in Serum (mg/dl) of Mice

Groups TC HDL LDL TG

Ldlr-/- M 1,041 � 90 51 � 16.1 118 � 15.3 148 � 17.6

Nlrp3-/- Ldlr-/- M 1,100 � 58 49 � 13.5 109 � 16.9 129 � 17.1

Ldlr-/- F 987 � 82 59 � 16.8 102 � 19.1 135 � 19.5

Nlrp3-/- Ldlr-/- F 1,029 � 59 62 � 17.1 115 � 16.9 141 � 21.5

Values mean � SEM.

F ¼ female; HDL ¼ high-density lipoprotein; LDL ¼ low-density lipoprotein; M ¼ male;TC ¼ total cholesterol; TG ¼ triglyceride.


585

anus were incised. The testes were drawn out andplaced back and the wound was sealed with inter-rupted nylon sutures.

OVARIECTOMY. One week before HFD, Nlrp3�/�

Ldlr�/� and Ldlr�/� female mice of 8 weeks of ageunderwent ovariectomy (OVX) or sham-surgery. Micewere maintained on inhalation anesthesia (1.5% iso-flurane) via nose-cone. Before the start of the surgery,carprofen (5 mg/kg body weight) was administeredsubcutaneously. The area below the ribs was shavedand cleaned with betadine followed by alcohol. Thisarea was then lifted with forceps to make a small 2-cmhorizontal cut. Resorbable Vicryl sutures were used toclamp the horn beneath the ovary and each ovary wasremoved using forceps and scissors. The uterinehorns were then placed back into the body and theperitoneal cavity was closed using interruptedresorbable Vicryl sutures. The skin was closed withinterrupted nylon sutures. Following spontaneousmovement, buprenorphine (0.5 mg/kg body weight)and 300 ml of warm saline were administered subcu-taneously. In sham-operated mice, both the skin andinner skin membrane were incised. The ovaries wereexternalized and returned to the abdominal cavityand the wound was sealed with interruptednylon sutures.

ASSESSMENT OF ATHEROSCLEROTIC LESIONS IN

THE AORTA AND AORTIC SINUS. The aortas weredissected and the adherent (adventitial) fat wasgently removed. Whole aortas were opened longitu-dinally from the aortic arch to the iliac bifurcation,mounted en face, and stained for lipids with oil red O.Hearts were embedded in optimum cutting tempera-ture compound (OCT) (Tissue-Tek, Sakura, Torrance,California) and serial 7-mm-thick cryosections fromthe aortic sinus were mounted and stained with oilred O and hematoxylin. Six frozen aortic root crosssections for oil red O stain or hematoxylin werecaptured with BZ-X710 microscope (Keyence, Itasca,Illinois) digital camera. Image analysis was performedby a trained observer blinded to the genotype of themice. Lesion areas were quantified with image anal-ysis software using a BZ-X710 microscope (Keyence).The lesion area in the aorta en face preparations wasexpressed as a percent of the aortic surface area, aspreviously reported (44). Necrotic core was measuredby hematoxylin-eosin staining and quantified withimage analysis software.

MESO SCALE DISCOVERY. IL-1b in mouse plasmasamples was measured using the U-PLEX Mouse IL-1bAssay (Meso Scale Diagnostics, Rockville, Maryland)

per the manufacturer’s instructions. The sampleswere read and analyzed by Meso Scale DiscoveryQuickPlex SQ120 instrumentation and Workbench 4.0Software (Meso Scale Diagnostics).

CYTOKINE ASSAY. IL-18 in mouse plasma sampleswas measured using enzyme-linked immunosorbentassay kits for murine IL-18 (Abcam, Cambridge, Mas-sachusetts). Cell culture supernatants were assayedusing commercially available enzyme-linked immu-nosorbent assay kits for murine IL-1b and tumor ne-crosis factor-a (eBioscience, San Diego, California)according to the manufacturer’s instructions.

IMMUNOFLUORESCENCE STAINING AND IMAGE

ACQUISITION. For immunohistochemical staining offrozen sections, fixing and antigen blocking wereperformed using immunoglobulin from the species ofthe secondary antibodies. Next, the sections wereincubated with primary antibodies overnight at 4�C,followed by incubation with the appropriate second-ary antibodies conjugated with fluorescent dyes. Forassessment of macrophage content, cells weredetected using anti-MOMA-2 antibody and for coloc-alization staining, nuclei were counterstained withDAPI. Caspase-1 activity was detected by fluorescentlabeled inhibitors of caspases (FLICA) staining. Im-ages (3 sections per animal) were captured using theBZ-9000 microscope (Keyence) and analyzed by BZanalyzer software.

SERUM LIPID PROFILES. Mice were sacrificed andsera from mice were obtained at the end of experi-ments and after an overnight fast. Total cholesterolconcentrations and lipid profiles were determined induplicate by using a colorimetric assay (infinitycholesterol reagent, Sigma Diagnostics, St. Louis,Missouri) as described earlier (43).

STATISTICAL ANALYSIS. Results are reported asmean � SEM. All data were analyzed with theGraphPad Prism statistical software version 7

FIGURE 1 NLRP3 Deficiency Reduces Diet-Induced Atherosclerosis Development in Female But Not Male Mice

(A) Representative oil red O staining of aortic sinus plaque in Ldlr-/- and Nlrp3-/-Ldlr-/- mice (n ¼ 12). (B) Quantification of area of aortic sinus

plaques. (C) Quantification of lipid content of aortic sinus. (D) Representative oil red O staining of aortic en face Ldlr-/- and Nlrp3-/-Ldlr-/- mice

(n ¼ 12). (E) Quantification of aortic lesion coverage. Data are presented as mean value � standard error of the mean. Statistical significance

was determined using Student’s t-test.



586

FIGURE 2 Comparison of Necrotic Core and Macrophage Content in Diet-Induced Atherosclerosis Lesion of Female Versus Male Mice



587

FIGURE 3 IL-1b and IL-18 Secretion Are Higher in Female Ldlr -/- Mice Compared With Males

(A) Female and male Ldlr�/� mice were fed 12 weeks high-fat diet, plasma concentrations of IL-1b were measured by Meso Scale Discovery.

(B) Plasma concentrations of IL-18 were measured by enzyme-linked immunosorbent assay. (C and D) Peritoneal macrophages were isolated

from female and male Ldlr�/� mice after 12 weeks high-fat diet. Four-hour lipopolysaccharide-primed peritoneal macrophages from female or

male Ldlr -/- mice were stimulated with 5 mM ATP (1 h). IL-1b and TNF-a concentrations in the culture supernatant were determined by

enzyme-linked immunosorbent assay. Vehicle: 0.01% dimethyl sulfoxide in PRMI1640 medium. All data are mean � SEM and represen-

tative of 3 independent experiments in triplicate. Statistical significance was determined using Student’s t-test. IL ¼ interleukin;

LPS ¼ lipopolysaccharide; TNF ¼ tumor necrosis factor.



588

(GraphPad Software, Inc., San Diego, California).Statistical differences between 2 groups wereassessed using a 2-sided Student’s t-test. Values ofp < 0.05 were considered significant.

FIGURE 2 Continued

(A) Necrotic core area (hematoxylin-eosin) in male aortic root (n ¼ 10 to

(n¼ 10 to 12). (C)Macrophage content in male aortic root by MOMA-2 stai

female aortic root by MOMA-2 and DAPI (for nucleus) staining (n ¼ 8 to

determined using Student’s t-test.

SAMPLE SIZE AND POWER CALCULATIONS. We fol-lowed the American Heart Association scientificstatement on the recommendation of design andexecution and reporting of animal atherosclerosis

12). (B) Necrotic core area (hematoxylin-eosin) in female aortic root

ning and DAPI (for nucleus) (n ¼ 8 to 11). (D)Macrophage content in

10). Data are presented as mean � SEM. Statistical significance was



Donor (BM) to Recipient TC HDL LDL TG

WT to Ldlr-/- M 987 � 78 59 � 15.7 110 � 19.2 134 � 28

Nlrp3-/- to Ldlr-/- M 901 � 48 62 � 12.5 104 � 21 128 � 20.1

WT to Ldlr-/- F 1,012 � 77 52 � 15.3 112 � 17.8 150 � 13.2

Nlrp3-/- to Ldlr-/- F 978 � 81 61 � 17.4 117 � 13.9 136 � 16.4

Values are mean � SEM.

BM ¼ bone marrow; WT ¼ wild-type; other abbreviations as in Table 1.


589

studies as published in 2017 by the American HeartAssociation Council on Arteriosclerosis, Thrombosisand Vascular Biology and Council on Basic Cardio-vascular Sciences (45). The sample sizes needed ineach experimental group were based on 80% powerand 2-sided tests for 5% level of significance. Basedon our prior experience, given expected experimentalvariance within a treatment group of up to 20%, wecalculated that a minimum of 8 to 10 animals pergroup was required. To account for any additionalvariability or mortality during the experiments, wecalculated a priori that 10 to 12 mice were used in eachgroup. Both male and female mice were used andanalyzed separately because this was the main focusof this study.

RESULTS

NLRP3 INFLAMMASOME PLAYS A GREATER ROLE IN

HFD-INDUCED ATHEROSCLEROSIS IN FEMALE

COMPARED WITH MALE LDLR-/- MICE. To assess sexdifferences in inflammasome-mediated accelerationof atherosclerosis, we generated Nlrp3-/-Ldlr-/- miceand fed them an HFD for 12 weeks. Despite similarblood cholesterol levels and lipid profile (Table 1), infemale mice, NLRP3 deficiency resulted in a 30%decrease in plaque size in the aortic root comparedwith Ldlr-/- alone, whereas this difference was notsignificant in males (Figures 1A and 1B). DKO femalesalso exhibited 32% less lipid content in aortic rootplaque (Figure 1C) and 38% less lipid coverage inaortic en face compared with control animals,whereas in males there was no significant differencebetween DKO and control animals (Figures 1D and 1E).In HFD-fed male mice, NLRP3 deficiency did notaffect necrotic core size (Figure 2A) or macrophagecontent (Figure 2C), whereas these parameters weresignificantly reduced in NLRP3-deficient femalescompared with control animals (Figures 2B and 2D).Lipid accumulation and cell death within the lesionscontribute to activation of inflammatory cells thatrelease proinflammatory and proatherogenic media-tors into the serum (46,47). We measured IL-1b andIL-18 in the serum and found significantly higherconcentrations of IL-1b and IL-18 in female Ldlr�/�

mice compared with male mice (Figures 3A and 3B).Additionally, peritoneal macrophages were isolatedfrom female and male Ldlr�/� mice after 12 weeksHFD. Cells were pretreated with lipopolysaccharideand then stimulated with NLRP3 activator ATP. Fe-male macrophages secreted significantly more IL-1b,but not tumor necrosis factor-a, compared with malemacrophages (Figures 3C and 3D).

NLRP3 ACTION IN HEMATOPOIETIC CELLS MODULATES

ATHEROSCLEROSIS IN FEMALES. The previously dis-cussed results suggest that the NLRP3 inflammasomeplays a greater role in lesion development in femalemice compared with male mice. We next used a BMchimera (donor / irradiated recipient: wild-type/Ldlr-/-; Nlrp3-/-/Ldlr-/-) approach to deter-mine the role of NLRP3 in BM-derived cells in the sexdifference in inflammasome-mediated acceleration ofatherosclerosis. After 12 weeks on HFD, despitesimilar blood cholesterol levels and lipid profile(Table 2), female recipients of Nlrp3-/- BM developedsignificantly smaller lesions in the aortic sinus andlower lipid content in aortic root (Figures 4A to 4C)and less aorta en face lipid coverage (Figures 4Dand 4E) than did female recipients of wild-type BM.However, this protection was not observed in males(Figures 4A to 4E). These data confirmed our previousresults and further suggest that the NLRP3 inflam-masome in hematopoietic cells plays a greater role inHFD-mediated atherosclerosis in females than inmales in the Ldlr-/- model.

SEX HORMONES MODULATE NLRP3-MEDIATED

ATHEROSCLEROSIS. To determine whether the dif-ferences we observed between sexes were mediatedby sex hormones, we performed CAS in male Ldlr-/-

and Nlrp3-/-Ldlr-/- mice; sham-surgeries in maleLdlr-/- and Nlrp3-/-Ldlr-/- mice were done to controland rule out any effects that may be caused by thesurgery itself (48). After 1 week of recovery, all micewere fed HFD for 12 weeks before sacrifice. As ex-pected based on our previous data, in sham-operatedmale mice, there was no difference in aortic lesionsize between genotypes (Supplemental Figure 1A),but in the CAS group, DKO mice now showed signifi-cant protection (Figure 5A). Similar protection in CASDKO mice was observed in terms of lipid content inthe aortic sinus (Figure 5B), and in aortic en face(Figure 5C), but not in sham-operated male DKO mice


FIGURE 4 Nlrp3 Deficiency in Bone Marrow Cells Reduces Diet-Induced Atherosclerosis Development in Female But Not Male Mice

All mice were on Ldlr-/- background. Irradiated Ldlr-/- mice received wild-type or Nlrp3-/- bone marrow cells. After 8 weeks reconstitution, the mice were

fed high-fat diet for 12 weeks. (A) Representative oil red O staining of aortic sinus plaque in mice (n ¼ 12). (B) Quantification of area of aortic sinus plaques.

(C) Quantification of lipid content of aortic sinus. (D) Representative oil red O staining of aortic en face (n ¼ 12). (E) Quantification of aortic lesion coverage.

Data are presented as mean � SEM. Statistical significance was determined using Student’s t-test. WT ¼ wild-type.



590

FIGURE 5 Comparison of Aortic Sinus Lesion Size in Nlrp3 Ldlr DKO in Male CAS and Female OVX Mice

Oil red O staining of aortic root (A), aortic root lipid content (B), and aortic en face (C) in male CAS mice (n ¼ 8 to 11). Oil red O staining of

aortic root (D), aortic root lipid content (E), and aortic en face (F) in female OVX mice (n ¼ 8 to 10). Data are presented as mean � SEM.

Statistical significance was determined using Student’s t-test. CAS ¼ castration; OVX ¼ ovariectomy.


591



Groups (Males) TC HDL LDL TG

Ldlr-/- sham 942 � 89 51 � 16.3 121 � 16.4 141 � 17.6

Nlrp3-/- Ldlr-/- sham 1,001 � 77 62 � 17.6 116 � 18.2 136 � 15.1

Ldlr-/- CAS 923 � 85 64 � 15.5 109 � 20 125 � 19.1

Nlrp3-/- Ldlr-/- CAS 988 � 78 59 � 18.2 117 � 16.9 131 � 22.3

Values are mean � SEM.

CAS ¼ castration; other abbreviations as in Table 1.

TABLE 4

Groups

Ldlr-/- sha

Nlrp3-/- Ld

Ldlr-/- CAS

Nlrp3-/- Ld

Values are m

OVX ¼ ov



592

(Supplemental Figures 1B and 1C), suggesting thattestosterone may suppress the role of the inflamma-some in atherosclerosis development. The bloodcholesterol levels and lipid profiles were similar ineach group (Table 5), and so was their weight gain(Table 4). We next performed sham-surgery or OVX infemale Ldlr-/- and Nlrp3-/-Ldlr-/- mice. As expected,sham DKO female mice had smaller aortic root lesionssize (Supplemental Figure 1D), lipid content(Supplemental Figure 1E), and lipid content in aorticen face (Supplemental Figure 1F) than sham Ldlr-/-

mice. However, these differences were lost in theOVX group (Figures 5D to 5F) supporting a role forestrogen or progesterone in the effect of the inflam-masome. Notably, the blood cholesterol levels andlipid profiles were similar in each group (Table 5), andso was their weight gain (Table 4). Taken together,these data suggest that testosterone suppressesinflammasome-mediated atherosclerosis, whereasestrogen or progesterone promotes inflammasome-mediated atherosclerosis development.

TESTOSTERONE LIMITS MACROPHAGE INFLAMMASOME

ACTIVITY IN ATHEROSCLEROSIS LESION. Abnormalinflammasome activation and the subsequent in-crease in circulating IL-1b and IL-18 correlate withenhanced macrophage recruitment to lesions (49).Therefore, we next measured caspase-1 activity, aread-out of inflammasome activation, in lesion mac-rophages in aortic roots using FLICA. Plaque macro-phages from Ldlr-/- male mice had significantly lesscaspase-1 activity compared with those from Ldlr-/-

Weight Gain of Sham and Surgery Mice

(Males) Weight Gain (g) Groups (Females) Weight Gain (g)

m 13.2 � 6.6 Ldlr-/- sham 16.8 � 8.2

lr-/- sham 15.4 � 4.7 Nlrp3-/- Ldlr-/- sham 13.2 � 5.3

14.3 � 4.2 Ldlr-/- OVX 14.6 � 6.2

lr-/- CAS 13.6 � 5.6 Nlrp3-/- Ldlr-/- OVX 12.7 � 5.9

ean � SEM.

ariectomy; other abbreviations as in Table 3.

female mice (Figures 6A and 6B). Interestingly, CASsignificantly increased FLICA positivity in the aorticroots compared with sham-operated males(Figures 6C and 6D); however, OVX did not alter FLICApositivity in aortic roots compared with sham-operated females (Supplemental Figure 2). Thesefindings indicate that at baseline, female mice havemore NLRP3 inflammasome activation in plaquemacrophages than males, which may drive theenhanced macrophage accumulation. Additionally, inagreement with the effect on plaque formation, lossof testosterone caused by CAS exacerbates inflam-masome activity.

DISCUSSION

Inflammation plays an important role in atherogen-esis, plaque rupture, and subsequent thrombosisleading to acute ischemic syndromes (13). Recentstudies have suggested key roles for the proin-flammatory cytokines IL-1b and IL-1a in atheroscle-rosis (46,50-53). Notably, genetic deficiency of IL-1a,even when restricted to BM-derived cells, mitigatesatherosclerotic burden in a mouse model (52), andthis protective effect is even more pronounced whencombined with depletion of IL-1b (52).

The NLRP3 inflammasome is a multicomponentcomplex that tightly regulates the maturation andsecretion of IL-1b, IL-1a, and IL-18 (54,55). Given thatthe NLRP3 inflammasome regulates multiple cyto-kines, some researchers have suggested that targetingNLRP3 or caspase-1 may yield better outcomes in in-flammatory disease than targeting the IL-1b alone orother IL-1 cytokines in isolation (16). In this study, wefound that Nlrp3 deficiency decreased lesion devel-opment and aortic lipid accumulation in HFD-fedLdlr�/� female mice, but although the trend wasevident in male mice, this protection was not signif-icant, suggesting that female mice may have greatersensitivity to NLRP3 inflammasome compared withmale mice, whereas in both genders NLRP3 plays arole. Furthermore, we showed that this protectionwas related to inflammasome activity in hematopoi-etic cells, because BM chimera females that receivedNLRP3-deficient BM showed a similar reduction inatherosclerosis. Interestingly, this difference was loston OVX, suggesting a role for estrogen and/or pro-gesterone in the effect. In contrast, in male Ldlr-deficient mice, CAS conferred significant protectionfrom lesion development and lipid accumulation.Taken together, these data suggest that sex hormonesplay a role in inflammasome-mediated atherogenesisand thus may influence the response to inhibitors ofIL-1b, IL-1a, and IL-18.








Groups (Females) TC HDL LDL TG

Ldlr-/- sham 965 � 71 58 � 15.4 104 � 15.4 145 � 17.6

Nlrp3-/- Ldlr-/- sham 1,011 � 69 63 � 18.1 117 � 16.7 139 � 15.1

Ldlr-/- OVX 914 � 78 60 � 16.6 111 � 19.8 141 � 19.6

Nlrp3-/- Ldlr-/- OVX 960 � 88 55 � 17.1 102 � 17.9 136 � 19.3

Values mean � SEM.

Abbreviations as in Tables 1 and 4.


593

A role for sex in modulating atherogenesis is sup-ported by early work in mouse models, whichdemonstrated that atherosclerotic lesions in the aorticroot were larger in female mice (56-58), although thisobservation has not been a consistent finding acrossstudies (59). The general understanding of how sexhormones influence the immune system is that estro-gens have immune-enhancing effects, whereas pro-gesterone and androgens, such as testosterone anddihydrotestosterone, exert mainly immunosuppres-sive effects (60). Consistent with this paradigm, es-trogen markedly enhances lipopolysaccharide-induced IL-1b promotor activity in the murine macro-phage RAW cell line (28,29), and macrophages fromfemale rats secrete more IL-1b and IL-6 in response tolipopolysaccharide (28). Females typically develop amore vigorous innate and adaptive immune responseto antigen challenges (61,62), which can acceleratepathogen clearance but can also lead to increasedimmune-related pathology, such as autoimmune orinflammatory diseases (12,63). Androgens exert anoverall inhibitory effect on Th1 differentiation (64) andsuppress inflammatory immune cells, such as den-dritic cells and macrophages (65). Of interest, CASchanged the protection provided by NLRP3 deficiencyon atherosclerosis and under these conditions NLRP3deficiency provided protection from atherosclerosis. Itis possible that testosterone acts like an independentproatherogenic factor, which in the mouse modelmasks the otherwise NLRP3 protective function.Alternatively, there could be a direct connection be-tween testosterone and NLRP3 inflammasome activa-tion, which needs to be further studied in the future,because our current data cannot rule out thispossibility.

However, contradictions to this dogma have beendemonstrated, revealing complexities that challengethis paradigm. For example, androgen and estrogensignaling have been shown to enhance alternativemacrophage polarization (66). Notably, the regulatoryeffect of estrogen and estrogen receptor (ER)signaling on the NLRP3 inflammasome seems to becontext dependent. For example, in hepatocellularand endometrial carcinoma cells, estrogen upregu-lates the NLRP3 inflammasome via ERb (30,67). But inthe brain and in fibroblast-like synoviocytes, estrogeninhibits activation of the NLRP3 inflammasome (22).ERa and ERb are NOD-like receptors (NLRs) tran-scription regulation factors, because they both regu-late NLR expression and promote inflammasomecolocalization, and a selective ERa antagonist signif-icantly inhibits NLRP3 expression and inflammasomeactivity (68). Furthermore, the role of inflammasome

activation in inflammatory pathways may also becontext-specific, because in contrast to our findings,gene expression studies suggest that the inflamma-some plays a more central role in abdominal aorticaneurysms in males than in females (69). Our findingof a role for female sex hormones in drivinginflammasome-mediated atherogenesis in mice couldbe interpreted as contradictory to clinical studies,because menopause (a state of estrogen deficiency) isassociated with higher ACVD risk in human females.However, the effect of estrogen deficiency oninflammasome activity in human subjects is un-known, and the enhanced risk of ACVD in post-menopausal females may also be related toadvancing age and age-related alterations in inflam-matory responses that are independent of inflamma-some activity.

Despite the well-established link between inflam-mation and atherosclerosis, clinical data demon-strating a direct benefit of targeting inflammation hadbeen absent until the CANTOS trial showed the po-tential for using anti-inflammatory therapy (anti-IL-1b), confirming that IL1b is an important potentialtherapeutic target for human atherosclerosis andrelated complications (40). However, for thisapproach to be clinically useful, it is critical to iden-tify subsets of patients who will derive maximumbenefits from canakinumab (or other anti-inflammatory agents) and are at low risk for seriousinfection. In the CANTOS trial, risk reduction withanti-IL1b therapy was observed in both men andwomen; however, women formed only 26% of thecohort, indicating that women may be more respon-sive than men to this therapy (41). Furthermore, thefact that males in the CANTOS trial demonstratedbenefit from the anti-IL1b therapy does conflict withthe findings reported here, because NLRP3 inhibitioncould affect several pathways beyond IL-1b, includingIL-1a and IL-18 as stated previously. Indeed, there arestill important gaps in the understanding of the roleof the NLRP3 inflammasome and caspase-1 in

FIGURE 6 Comparison of Active Caspase-1 Macrophages in Diet-Induced Atherosclerosis Lesion

(A) Representative images for caspase-1 positivity in lesion macrophages of male versus female Ldlr-/- mice. Caspase-1 activity was assessed

by fluorescent-labeled inhibitors of caspases (green) in macrophages (MOMA-2) (red) in atherosclerotic lesions of Ldlr�/� mice fed high-fat

diet for 12 weeks. (B) Quantification of active caspase-1þ cells in lesion macrophages (n ¼ 10). (C) Representative images for caspase-1

positivity in lesion macrophages of male sham versus CAS Ldlr -/- mice (n ¼ 9 to 11). (D) Quantification of active caspase-1þ cells in lesion

macrophages. Data are presented as mean value � standard error of the mean. Statistical significance was determined using Student’s t-

test. FLICA ¼ fluorescent labeled inhibitors of caspases; other abbreviation as in Figure 5.



594


595

atherosclerosis. Of interest, a recent study linkedincreased NLRP3 expression in human carotid pla-ques to pathological features, such as vascularinflammation, plaque composition, and vulnerability(16). The authors highlighted the importance ofNLRP3 inflammasome and caspase-1-driven IL-1a andIL-1b production in atherosclerotic carotid plaques,supporting the view that release of both of these IL-1isoforms is determined by the NLRP3-caspase-1pathway in atherosclerotic plaques (16).

A novel, common, and powerful cardiovascular riskfactor has recently emerged: clonal hematopoiesis ofindeterminate potential, which arises from somaticmutations in hematopoietic stem cells (17). Studieshave shown that individuals who acquire somaticclonal hematopoiesis of indeterminate potential mu-tations with age have a 40% increase in cardiovascularrisk, independent of traditional risk factors (70). Mostcases of clonal hematopoiesis of indeterminate po-tential are caused by mutations in only a handful ofgenes, including TET2 (17,70-72). Ldlr-/- mice engi-neered to bear the TET2 loss of function that is similarto clonal hematopoiesis and increased cardiovasculardisease risk in humans (17) had activated NLRP3inflammasome in myeloid cells, enhanced IL-1b pro-duction, and developed accelerated atherosclerosis(73,74). Even though male-to-female mice compari-sons were not reported, it is of interest that, in both ofthese experimental studies the recipient mice thatdeveloped increased NLRP3-induced acceleratedatherosclerosis were female (73,74).

Women have higher death rates followingmyocardial infarction than men (75,76), because ofdifferences in the pathogenesis of atherosclerosis,differential efficacy of drugs (77,78), and becausevascular assist devices may fit men better thanwomen (79,80). As Clayton and Tannenbaum haveargued (81), failure to analyze for male and femaleclinical trial participants separately may maskimportant differences in the effects of interventions,toxicity, symptoms, or adverse effects. To date, mostclinical studies in this area contain a majority of malesubjects, and no trial that we are aware of pre-specified analysis of differences by sex. Therefore,understanding how the immune response differs inmen and women in the context of atherosclerosis mayimprove the treatment of cardiovascular disease.

In summary, there is a need for additional inves-tigation on the role of estrogens, progesterone, and

testosterone on the NLRP3 inflammasome and IL-1bas well as IL-1a signaling in atherogenesis as it relatesto the biologic mechanisms underlying pathophysio-logical processes in males and females. Our data addweight and a sense of urgency to the efforts of Rob-inet et al. and the Council on Arteriosclerosis,Thrombosis and Vascular Biology to encourage pre-clinical arterial pathology researchers to consider sexas a biologic variable when designing and reportingexperiments (1), which will improve the design ofclinical trials, and help optimize cardiovascular carefor men and women.

CONCLUSIONS

The present study suggests that loss of NLRP3inflammasome components leads to more significantreductions in atherosclerotic plaque size and lipidcontent in female mice than in male mice. Further-more, CAS increases dependency on NLRP3 inflam-masome components for atherogenesis, and increasesinflammasome activity, suggesting that testosteroneplays an inhibitory role, blocking inflammation inatherogenesis. OVX reduces dependency of athero-genesis on NLRP3 inflammasome components, sug-gesting that female sex hormones sensitizeinflammation in atherogenesis. Our data providebiologic insights into the clinical merit of anti-NLRP3-directed therapies, and the biologic mechanisms un-derlying pathophysiological processes in malesversus females as they pertain to atherosclerosis andthe NLRP3 inflammasome, which could help informthe design of future clinical trials.

ACKNOWLEDGMENTS The authors thank WenxuanZhang, Ganghua Huang, and P. Sun for excellenttechnical assistance.

ADDRESS FOR CORRESPONDENCE: Dr. Moshe Arditi,Department of Biomedical Sciences and Cedars-SinaiSmidt Heart Institute, Infectious and ImmunologicalDiseases Research Center, Cedars-Sinai Medical Cen-ter, 8700 Beverly Boulevard, Room 4221, Los Angeles,California 90048. E-mail: [email protected]. ORDr. Prediman K. Shah, Oppenheimer AtherosclerosisResearch Center and Atherosclerosis Prevention andTreatment Center, Smidt Heart Center, Cedars SinaiMedical Center, 8700 Beverly Boulevard, Los Angeles,California 90048. E-mail: [email protected].




PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: The

CANTOS trial suggested that IL-1b-directed therapy with

a neutralizing monoclonal antibody moderately reduces

recurrent ischemic events and cardiovascular death

among patients with coronary artery disease and elevated

C-reactive protein. Therefore, IL1b, which is under the

control of the NLRP3 inflammasome and caspase-1, is an

important potential therapeutic target for human

atherosclerosis and related complications. However,

NLRP3 inflammasome also controls secretion of IL-1a and

IL-18, inflammatory cytokines that have also been

implicated in development of atherosclerosis, and some

researchers have suggested that targeting NLRP3-

caspase-1 may yield better outcomes than targeting the

IL-1b alone or IL-1 isoforms in isolation. Herein, we show

that sex hormones may be involved in NLRP3

inflammasome–mediated atherogenesis and may lead to

differential responses to anti-NLRP3 therapy between

males and females. In a mouse model of atherosclerosis,

females with global Nlrp3 deletion or those receiving

Nlrp3 -/- BM developed significantly fewer lesions in the

aortic sinus and decreased lipid content in aorta, but

Nlrp3 deficiency did not confer similar protection in

males. Ovariectomized female mice lost protection

mediated by NLRP3 deficiency, whereas castrated males

showed stronger correlations between NLRP3 inflamma-

some and atherosclerosis. Overall, the findings of present

study suggest that testosterone may play an inhibitory

role by blocking NLRP3 inflammasome and

inflammation in atherogenesis, whereas female sex hor-

mones may promote NLRP3 inflammasome–mediated

atherosclerosis.

TRANSLATIONAL OUTLOOK: The specific role and

underlying mechanisms of inflammasome activation and

inflammation in atherogenesis are topics of active

research. The role of the NLRP3 inflammasome

pathway in diet-induced atherosclerosis is still

controversial, and the impact of sex hormones has not

been explored. In this study we observed sex-specific

effects of the NLRP3 inflammasome on atherogenesis

in LDLR-deficient mice, with NLRP3 inflammasome

playing a more prominent role in atherosclerosis in

female mice than in males. The CANTOS study

demonstrated modest therapeutic benefit of a mono-

clonal antibody targeting IL-1b (canakinumab) in male

and female patients with previous myocardial infarc-

tion, indicating that IL-1b is an important therapeutic

target. However, the NLRP3 inflammasome controls

not only IL-1b secretion, but also IL-1a and IL-18,

leading some researchers to advocate that targeting

NLRP3-caspase-1 may yield better outcomes. Further-

more, a secondary analysis of the CANTOS trial

revealed that whereas women and men showed similar

clinical efficacy with canakinumab, only 26% of the

participants were female, suggesting that a smaller

sample size was needed for females to achieve the

same clinical benefit. Therefore, finding ways to iden-

tify subsets of patients who will derive maximum

benefits from canakinumab (or other anti-inflammatory

agents) is crucial, and it is critically important to un-

derstand the role of sex in NLRP3 inflammasome–

mediated IL-1b and IL-1a-driven inflammation in

atherosclerosis. The present study lends support to

the impact of estrogens and testosterone on the

inflammasome in atherogenesis and yields important

information on biologic mechanisms underlying path-

ophysiological processes in atherosclerosis in both

sexes. The results of the present study may help

design future clinical trials, with the objective to

personalize cardiovascular care for men and women.



596

RE F E RENCE S

1. Maas AH, van der Schouw YT, Regitz-Zagrosek V, et al. Red alert for women’s heart: theurgent need for more research and knowledge oncardiovascular disease in women: proceedings ofthe workshop held in Brussels on gender differ-ences in cardiovascular disease, 29 September2010. Eur Heart J 2011;32:1362–8.

2. Shaw LJ, Bugiardini R, Merz CN. Women andischemic heart disease: evolving knowledge. J AmColl Cardiol 2009;54:1561–75.

3. Roger VL, Go AS, Lloyd-Jones DM, et al. Heartdisease and stroke statistics–2012 update: a reportfrom the American Heart Association. Circulation2012;125:e2–220.

4. Rosamond W, Flegal K, Friday G, et al.Heart disease and stroke statistics–2007 up-date: a report from the American HeartAssociation Statistics Committee and StrokeStatistics Subcommittee. Circulation 2007;115:e69–171.

5. Vaccarino V, Abramson JL, Veledar E,Weintraub WS. Sex differences in hospitalmortality after coronary artery bypass sur-gery: evidence for a higher mortality inyounger women. Circulation 2002;105:1176–81.

6. Argulian E, Patel AD, Abramson JL, et al.Gender differences in short-term cardiovascularoutcomes after percutaneous coronary in-terventions. Am J Cardiol 2006;98:48–53.
































597

7. Arbustini E, Dal Bello B, Morbini P, et al. Plaqueerosion is a major substrate for coronary throm-bosis in acute myocardial infarction. Heart 1999;82:269–72.

8. Campbell IC, Suever JD, Timmins LH, et al.Biomechanics and inflammation in atheroscleroticplaque erosion and plaque rupture: implicationsfor cardiovascular events in women. PLoS One2014;9:e111785.

9. Sattar N, Greer IA. Pregnancy complications andmaternal cardiovascular risk: opportunities forintervention and screening? BMJ 2002;325:157–60.

10. Veltman-Verhulst SM, van Rijn BB,Westerveld HE, et al. Polycystic ovary syndromeand early-onset preeclampsia: reproductive mani-festations of increased cardiovascular risk. Meno-pause 2010;17:990–6.

11. Ray JG, Vermeulen MJ, Schull MJ,Redelmeier DA. Cardiovascular health aftermaternal placental syndromes (CHAMPS):population-based retrospective cohort study.Lancet 2005;366:1797–803.

12. Fairweather D, Petri MA, Coronado MJ,Cooper LT. Autoimmune heart disease: role of sexhormones and autoantibodies in disease patho-genesis. Expert Rev Clin Immunol 2012;8:269–84.

13. Miller AM, McInnes IB. Cytokines as therapeu-tic targets to reduce cardiovascular risk in chronicinflammation. Curr Pharm Des 2011;17:1–8.

14. Matsuura E, Lopez LR, Shoenfeld Y, Ames PR.beta2-glycoprotein I and oxidative inflammation inearly atherogenesis: a progression from innate toadaptive immunity? Autoimmun Rev 2012;12:241–9.

15. Sollberger G, Strittmatter GE, Garstkiewicz M,Sand J, Beer HD. Caspase-1: the inflammasomeand beyond. Innate Immun 2014;20:115–25.

16. Jiang X, Wang F, Wang Y, et al. Inflamma-some-driven interleukin-1alpha and interleukin-1beta production in atherosclerotic plaques re-lates to hyperlipidemia and plaque complexity.J Am Coll Cardiol Basic Trans Science 2019;4:304–17.

17. Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis ASSO standard errorof the mean cited with adverse outcomes. N Engl JMed 2014;371:2488–98.

18. Beltrami-Moreira M, Vromman A, Sukhova GK,Folco EJ, Libby P. Redundancy of IL-1 isoformsignaling and its implications for arterial remod-eling. PLoS One 2016;11:e0152474.

19. Duewell P, Kono H, Rayner KJ, et al. NLRP3inflammasomes are required for atherogenesis andactivated by cholesterol crystals. Nature 2010;464:1357–61.

20. Menu P, Pellegrin M, Aubert JF, et al.Atherosclerosis in ApoE-deficient mice progressesindependently of the NLRP3 inflammasome. CellDeath Dis 2011;2:e137.

21. De Nardo D, Latz E. NLRP3 inflammasomes linkinflammation and metabolic disease. TrendsImmunol 2011;32:373–9.

22. Xu YJ, Sheng H, Bao QY, Wang YJ, Lu JQ, Ni X.NLRP3 inflammasome activation mediates

estrogen deficiency-induced depression- andanxiety-like behavior and hippocampal inflamma-tion in mice. Brain Behav Immun 2016;56:175–86.

23. Thakkar R, Wang R, Sareddy G, et al. NLRP3inflammasome activation in the brain after globalcerebral ischemia and regulation by 17beta-estra-diol. Oxid Med Cell Longev 2016:8309031.

24. Fairweather D. Sex differences in inflammationduring atherosclerosis. Clin Med Insights Cardiol2014;8:49–59.

25. Rossouw JE, Anderson GL, Prentice RL, et al.Risks and benefits of estrogen plus progestin inhealthy postmenopausal women: principal resultsfrom the Women’s Health Initiative randomizedcontrolled trial. JAMA 2002;288:321–33.

26. Hulley S, Grady D, Bush T, et al. Randomizedtrial of estrogen plus progestin for secondaryprevention of coronary heart disease in post-menopausal women. Heart and Estrogen/proges-tin Replacement Study (HERS) Research Group.JAMA 1998;280:605–13.

27. Wilson PW, Garrison RJ, Castelli WP. Post-menopausal estrogen use, cigarette smoking, andcardiovascular morbidity in women over 50. TheFramingham Study. N Engl J Med 1985;313:1038–43.

28. Ruh MF, Bi YH, D’Alonzo R, Bellone CJ. Effectof estrogens on IL-1 beta promoter activity.J Steroid Biochem 1998;66:203–10.

29. Ruh MF, Bi YH, Cox L, Berk D, Howlett AC,Bellone CJ. Effect of environmental estrogens onIL-1 beta promoter activity in a macrophage cellline. Endocrine 1998;9:207–11.

30. Wei Q, Guo PB, Mu K, et al. Estrogen sup-presses hepatocellular carcinoma cells through ERbeta-mediated upregulation of the NLRP3inflammasome. Lab Invest 2015;95:804–16.

31. Rettew JA, Huet YM, Marriott I. Estrogensaugment cell surface TLR4 expression on murinemacrophages and regulate sepsis susceptibilityin vivo. Endocrinology 2009;150:3877–84.

32. Bruck B, Brehme U, Gugel N, et al. Gender-specific differences in the effects of testosteroneand estrogen on the development of atheroscle-rosis in rabbits. Arterioscler Thromb Vasc Biol1997;17:2192–9.

33. Hanke H, Lenz C, Hess B, Spindler KD,Weidemann W. Effect of testosterone on plaquedevelopment and androgen receptor expression inthe arterial vessel wall. Circulation 2001;103:1382–5.

34. Alexandersen P, Haarbo J, Byrjalsen I,Lawaetz H, Christiansen C. Natural androgensinhibit male atherosclerosis: a study in castrated,cholesterol-fed rabbits. Circ Res 1999;84:813–9.

35. Gordon GB, Bush DE, Weisman HF. Reductionof atherosclerosis by administration of dehydro-epiandrosterone. A study in the hypercholester-olemic New Zealand white rabbit with aorticintimal injury. J Clin Invest 1988;82:712–20.

36. Nathan L, Shi W, Dinh H, et al. Testosteroneinhibits early atherogenesis by conversion toestradiol: critical role of aromatase. Proc Natl AcadSci U S A 2001;98:3589–93.

37. Oskui PM, French WJ, Herring MJ, Mayeda GS,Burstein S, Kloner RA. Testosterone and the cardio-vascular system: a comprehensive review of theclinical literature. JAmHeartAssoc2013;2:e000272.

38. Herring MJ, Oskui PM, Hale SL, Kloner RA.Testosterone and the cardiovascular system: acomprehensive review of the basic science litera-ture. J Am Heart Assoc 2013;2:e000271.

39. McCrohon JA, Jessup W, Handelsman DJ,Celermajer DS. Androgen exposure increases hu-man monocyte adhesion to vascular endotheliumand endothelial cell expression of vascular celladhesionmolecule-1. Circulation 1999;99:2317–22.

40. Ridker PM, Everett BM, Thuren T, et al. Anti-inflammatory therapy with canakinumab foratherosclerotic disease. N Engl J Med 2017;377:1119–31.

41. Ridker PM, MacFadyen JG, Everett BM, et al.Relationship of C-reactive protein reduction tocardiovascular event reduction following treat-ment with canakinumab: a secondary analysis fromthe CANTOS randomised controlled trial. Lancet2018;391:319–28.

42. Tumurkhuu G, Dagvadorj J, Porritt RA, et al.Chlamydia pneumoniae hijacks a host autor-egulatory IL-1 beta loop to drive foam cell for-mation and accelerate atherosclerosis. Cell Metab2018;28:432–48.

43. Chen S, Shimada K, Crother TR, Erbay E,Shah PK, Arditi M. Chlamydia and lipids engage acommon signaling pathway that promotesatherogenesis. J Am Coll Cardiol 2018;71:1553–70.

44. Tumurkhuu G, Shimada K, Dagvadorj J, et al.Ogg1-dependent DNA repair regulates NLRP3inflammasome and prevents atherosclerosis. CircRes 2016;119:e76–90.

45. Daugherty A, Tall AR, Daemen M, et al.Recommendation on design, execution, andreporting of animal atherosclerosis studies: a sci-entific statement from the American Heart Asso-ciation. Arterioscler Thromb Vasc Biol 2017;37:e131–57.

46. Tedgui A, Mallat Z. Cytokines in atheroscle-rosis: pathogenic and regulatory pathways. Phys-iol Rev 2006;86:515–81.

47. Kleemann R, Zadelaar S, Kooistra T. Cytokinesand atherosclerosis: a comprehensive review ofstudies in mice. Cardiovasc Res 2008;79:360–76.

48. Coronado MJ, Bruno KA, Blauwet LA, et al.Elevated sera sST2 is associated with heart failurein men <¼ 50 years old with myocarditis. J AmHeart Assoc 2019;8.

49. Robbins GR, Wen H, Ting JP. Inflammasomesand metabolic disorders: old genes in moderndiseases. Mol Cell 2014;54:297–308.

50. Kirii H, Niwa T, Yamada Y, et al. Lack ofinterleukin-1 beta decreases the severity ofatherosclerosis in ApoE-deficient mice. ArteriosclThrom Vas 2003;23:656–60.

51. Gardner SE, Humphry M, Bennett MR,Clarke MC. Senescent vascular smooth musclecells drive inflammation through an interleukin-1alpha-dependent senescence-associated secre-tory phenotype. Arterioscler Thromb Vasc Biol2015;35:1963–74.





































































































































































































598

52. Freigang S, Ampenberger F, Weiss A, et al.Fatty acid-induced mitochondrial uncouplingelicits inflammasome-independent IL-1alpha andsterile vascular inflammation in atherosclerosis.Nat Immunol 2013;14:1045–53.

53. Sheedy FJ, Moore KJ. IL-1 signaling inatherosclerosis: sibling rivalry. Nat Immunol 2013;14:1030–2.

54. Gross O, Yazdi AS, Thomas CJ, et al. Inflam-masome activators induce interleukin-1alphasecretion via distinct pathways with differentialrequirement for the protease function of caspase-1. Immunity 2012;36:388–400.

55. Dinarello CA. Immunological and inflammatoryfunctions of the interleukin-1 family. Annu RevImmunol 2009;27:519–50.

56. Paigen B, Holmes PA, Mitchell D, Albee D.Comparison of atherosclerotic lesions and HDL-lipid levels in male, female, and testosterone-treated female mice from strains C57BL/6, BALB/c, and C3H. Atherosclerosis 1987;64:215–21.

57. Smith JD, Trogan E, Ginsberg M, Grigaux C,Tian J, Miyata M. Decreased atherosclerosis inmice deficient in both macrophage colony-stimulating factor (op) and apolipoprotein E.Proc Natl Acad Sci U S A 1995;92:8264–8.

58. Tangirala RK, Rubin EM, Palinski W. Quantita-tion of atherosclerosis in murine models: correla-tion between lesions in the aortic origin and in theentire aorta, and differences in the extent of le-sions between sexes in LDL receptor-deficient andapolipoprotein E-deficient mice. J Lipid Res 1995;36:2320–8.

59. Marsh MM, Walker VR, Curtiss LK, Banka CL.Protection against atherosclerosis by estrogen isindependent of plasma cholesterol levels in LDLreceptor-deficient mice. J Lipid Res 1999;40:893–900.

60. Giefing-Kroll C, Berger P, Lepperdinger G,Grubeck-Loebenstein B. How sex and age affectimmune responses, susceptibility to infections, andresponse to vaccination. Aging Cell 2015;14:309–21.

61. Beagley KW, Gockel CM. Regulation of innateand adaptive immunity by the female sex hor-mones oestradiol and progesterone. FEMS Immu-nol Med Microbiol 2003;38:13–22.

62. Bhatia A, Sekhon HK, Kaur G. Sex hormonesand immune dimorphism. Scientific World Journal2014:159150.

63. Klein SL. The effects of hormones on sex dif-ferences in infection: from genes to behavior.Neurosci Biobehav Rev 2000;24:627–38.

64. Rupp MRG, Jorgensen TN. Androgen-inducedimmunosuppression. Front Immunol 2018;9:794.

65. Trigunaite A, Dimo J, Jorgensen TN. Suppres-sive effects of androgens on the immune system.Cell Immunol 2015;294:87–94.

66. Becerra-Diaz M, Strickland AB, Keselman A,Heller NM. Androgen and androgen receptor asenhancers ofM2macrophage polarization in allergiclung inflammation. J Immunol 2018;201:2923–33.

67. Liu SG, Wu XX, Hua T, et al. NLRP3 inflam-masome activation by estrogen promotes theprogression of human endometrial cancer. OncoTargets Ther 2019;12:6927–36.

68. Fan W, Gao X, Ding C, et al. Estrogen re-ceptors participate in carcinogenesis signalingpathways by directly regulating NOD-like re-ceptors. Biochem Biophys Res Commun 2019;511:468–75.

69. Wu XY, Cakmak S, Wortmann M, et al. Sex-and disease-specific inflammasome signatures incirculating blood leukocytes of patients withabdominal aortic aneurysm. Molecular Medicine2016;22:508–18.

70. Jaiswal S, Natarajan P, Ebert BL. Clonal he-matopoiesis and atherosclerosis. N Engl J Med2017;377:1401–2.

71. Sano S, Oshima K, Wang Y, Katanasaka Y,Sano M, Walsh K. CRISPR-mediated gene editingto assess the roles of Tet2 and Dnmt3a in clonalhematopoiesis and cardiovascular disease. Circ Res2018;123:335–41.

72. Sano S, Oshima K, Wang Y, et al. Tet2-mediated clonal hematopoiesis accelerates heart

failure through a mechanism involving the IL-1beta/NLRP3 inflammasome. J Am Coll Cardiol2018;71:875–86.

73. Fuster JJ, MacLauchlan S, Zuriaga MA, et al.Clonal hematopoiesis associated with TET2 defi-ciency accelerates atherosclerosis development inmice. Science 2017;355:842–7.

74. Jaiswal S, Natarajan P, Silver AJ, et al. Clonalhematopoiesis and risk of atherosclerotic cardio-vascular disease. N Engl J Med 2017;377:111–21.

75. Villablanca AC, Jayachandran M, Banka C.Atherosclerosis and sex hormones: current con-cepts. Clin Sci 2010;119:493–513.

76. Shlipak MG, Angeja BG, Go AS, et al. Hormonetherapy and in-hospital survival after myocardialinfarction in postmenopausal women. Circulation2001;104:2300–4.

77. Miller VM, Kararigas G, Seeland U, et al. Inte-grating topics of sex and gender into medicalcurricula-lessons from the international commu-nity. Biol Sex Differ 2016;7(Suppl):44.

78. Seeland U, Regitz-Zagrosek V. Sex and genderdifferences in cardiovascular drug therapy. HandbExp Pharmacol 2012;2014:211–36.

79. Potapov E, Schweiger M, Lehmkuhl E, et al.Gender differences during mechanical circulatorysupport. ASAIO J 2012;58:320–5.

80. Magnussen C, Bernhardt AM, Ojeda FM, et al.Gender differences and outcomes in leftventricular assist device support: the EuropeanRegistry for Patients with Mechanical CirculatorySupport. J Heart Lung Transplant 2018;37:61–70.

81. Clayton JA, Tannenbaum C. Reporting sex,gender, or both in clinical research? JAMA 2016;316:1863–4.

KEY WORDS atherosclerosis, IL-1b, NLRP3inflammasome, sex

APPENDIX For supplemental figures, pleasesee the online version of this paper.






























































































































EDITORIAL COMMENT

Taking Sex SeriouslyAn Oft-Overlooked Biological Variable*

Peter Libby, MD, Amélie Vromman, PHD

T he inflammasome, a supramolecular cyto-plasmic structure comprising multiple sub-units, senses danger in various guises and

activates its ultimate component, caspase-1, togenerate mature interleukin (IL)-1b and IL-18 fromtheir inactive precursors (Figure 1) (1). The inflamma-some participates in many inflammatory diseases,including those of the cardiovascular system. Dis-eases precipitated by crystalline structures such asmonosodium urate in gout and cholesterol monohy-drate in atherosclerosis, among other pathogen-associated molecular patterns (PAMPs) and danger-associated molecular patterns (DAMPs), trigger theinflammasome. Many lines of evidence converge todemonstrate the pivotal importance of the inflamma-some and its products in human diseases. Forexample, rare genetic disorders associated withconstitutive gain of function of the NLRP3

ISSN 2452-302X




From the Division of Cardiovascular Medicine, Department of Medicine,

Brigham and Women’s Hospital, Harvard Medical School, Boston, Mas-

sachusetts. Dr. Libby has received funding support from the National

Heart, Lung, and Blood Institute (R01HL080472 and 1R01HL134892), the

American Heart Association (18CSA34080399), and the RRM Charitable

Fund; is an unpaid consultant to, or involved in clinical trials for, Amgen,

AstraZeneca, Esperion Therapeutics, Ionis Pharmaceuticals, Kowa Phar-

maceuticals, Novartis, Pfizer, Sanofi-Regeneron, and XBiotech, Inc.; is a

member of the scientific advisory board for Amgen, Corvidia Therapeu-

tics, DalCor Pharmaceuticals, IFM Therapeutics, Kowa Pharmaceuticals,

Olatec Therapeutics, MedImmune, Novartis, and XBiotech, Inc.; serves

on the board of XBiotech, Inc.; and his laboratory has received research

funding in the last 2 years from Novartis. Dr. Vromman is supported by

the Harold M. English Fellowship Fund from Harvard Medical School.

Both authors participate in the Leducq Transatlantic Network on Clonal

Hematopoiesis.






inflammasome cause cryopyrin-associated periodicsyndrome or Muckle-Wells syndrome. NeutralizingIL-1b has proven transformative in the treatment ofthese diseases. CANTOS (Canakinumab Anti-inflammatory Thrombosis Outcomes Study) providedthe first proof of the success of targeting inflamma-tory pathways in cardiovascular disease by adminis-tration of a neutralizing antibody for theinflammasome product IL-1b in individuals withestablished coronary artery disease and persistentinflammation despite standard-of-care therapy,including high-intensity statin treatment (2). Down-stream of IL-1, IL-6 has proven causal in human athe-rothrombosis based on concordant Mendelianrandomization studies in humans. These convergentstrands of evidence all point to a pivotal role of theinflammasome in human disease and coronary heartdisease in particular.

We routinely use mice to test hypothesesregarding disease mechanisms. The ready geneticmanipulation in this species and the existence ofinbred strains facilitate our investigative quest tounravel disease pathogenesis. Exploration of the roleof the inflammasome using such techniques hasnonetheless proven confusing. A seminal study byDuewell et al. (3) assessed the effect of geneticallyinduced loss of function of the NLRP3 inflammasomein experimental atherosclerosis in mice; they re-ported an attenuation of atherogenesis in mice.Menu et al. (4) undertook similar experiments, albeitwith a different intensity of a cholesterol-enricheddiet, and showed no such moderation of atheromaformation. The Duewell et al. (3) experiments clearlyspecified the use of female mice. The null findingsreported by Menu et al. did not specify the sex of theanimals studied. In this issue of JACC: Basic toTranslational Science, Chen et al. (5), in a carefullydesigned head-to-head comparison, showed thatloss of function of the NLRP3 inflammasome atten-uates atherosclerosis in female but not in male mice






FIGURE 1 The NLRP3 Inflammasome, Depicted Here Artistically in Stained Glass,

Processes the Inactive Precursor Forms of the Proinflammatory Cytokines IL-1 Beta

and IL-18 to the Mature Active Mediators

Data presented by Chen et al. (5) suggest that in mice, inflammasome activation ac-

centuates atherosclerosis preferentially in females. These findings highlight the need to

consider sex as a biological variable in animal experiments and reinforce the necessity of

including women in clinical trials. These experimental findings should stimulate analyses

to seek sex differences in inflammatory pathways in humans. The stained glass image

shows a long-axis view of the NLRP3 inflammasome with an atherosclerotic artery in the

background (a design conceived by Dr. Peter Libby and biologist/stained glass artist Dr.

Joel Kowit, who created it). The image of the silhouettes of a man and a woman is from

the National Institutes of Health News in Health and was drawn by illustrator Alan

Defibaugh (https://newsinhealth.nih.gov/2016/05/sex-gender). IL ¼ interleukin.

Libby and Vromman J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Taking Sex Seriously J U N E 2 0 2 0 : 5 9 9 – 6 0 1

600

(5). Their careful observations help clarify theseemingly disparate findings of the 2 prior studiesdiscussed.

Furthermore, Chen et al. (5) found that castrationof male mice rendered them sensitive to inhibition ofatherosclerosis by NLRP3 inflammasome loss offunction. Ovariectomy of female mice renderedatherosclerotic lesion formation insensitive toinflammasome loss of function. These latter ablativeexperiments implicated sex hormones in the depen-dence of inflammatory processes mediated by theinflammasome. Nonetheless, these intriguing exper-iments do not establish the molecular mechanism bywhich gonadal products can modify the responses ofexperimental atherosclerosis to modulation by theinflammasome and its products.

We have long appreciated that autoimmune dis-eases can occur more frequently in women than men.Perhaps these mouse experiments lend some insightinto possible mechanisms that underlie the greatersusceptibility of women to certain inflammatory dis-eases such as lupus erythematosus, rheumatoidarthritis, and the like. In addition to sex hormones,sex chromosomes can influence these inflammatorydiseases. The X chromosome carries many genesinvolved in immune functions, including CD40 ligand(CD154) and IL-1 receptor-associated kinase-1. More-over, the X chromosome contains 10% of microRNA inthe genome. Many of these noncoding RNAs maymodulate cardiovascular diseases and immune re-sponses (6). These various observations highlight theneed to consider sex seriously as a biological variable.Too many experimental studies use animals of only 1sex or do not specify or assess response in both sexes.The story of the inflammasome in atherosclerosisrecounted here underscores the need to do so (5).Many studies have used male animals in a statedattempt to avoid potential confounding by cyclichormone changes in females. Is this justification foravoiding the study of female animals an extension ofpaternalistic thinking, or even of remnants of scien-tific misogyny persisting from times of yore? Ratherthan regarding the fluctuations as confounders, andthus excluding study of female mice or mixing thesexes willy-nilly, we should embrace the study ofdifferential responses between the sexes to deriveinsight into biologic mechanisms of capital impor-tance to half of our population. We must ask our-selves critically if excluding 1 sex or another in thedesign of experimental studies has a strong biologicaljustification or merely represents a convenientextension of habit or of traditional practices.

The recent heightened sensitivity of the impor-tance of sex as a biological variable both by fundingagencies and scientific journals has begun to reme-diate this deficiency in our enterprise. Data assem-bled in a recent important compilation inArteriosclerosis, Thrombosis, and Vascular Biologyshowed that <40% of publications in 2017 reportedboth sexes or provided justification of why only 1 sexwas studied (7). By the end of 2018, this number wasapproaching 100%. Most clinical trialists today striveto achieve greater inclusion of women and under-represented minorities in clinical trials. Although weseem to be improving in both laboratory and clinicalstudies in this regard, we must strive to do better toaddress both sexes experimentally. Funders shouldprovide sufficient budget to permit this effort. Thesponsors of clinical trials should provide the re-sources required to enroll representative numbers of

https://newsinhealth.nih.gov/2016/05/sex-gender

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Libby and VrommanJ U N E 2 0 2 0 : 5 9 9 – 6 0 1 Taking Sex Seriously

601

women and members of minority populations. Suchefforts can yield both expected and unexpectedbenefits. The inclusion of some 6,000 women inJUPITER (Justification for the Use of Statins in Pre-vention: an Intervention Trial Evaluating Rosuvasta-tin), a primary prevention trial with rosuvastatin,helped to dispel the myth that women do not benefitfrom statins (8). The fibrate arm of ACCORD (Action toControl Cardiovascular Risk in Type 2 Diabetes)showed directionally opposite point estimates ofevent reduction in men and in women (9). The recentPARAGON-HF (Prospective Comparison of ARNI withARB Global Outcomes in HF With Preserved EjectionFraction) study showed a possibly greater benefit of acombination of an angiotensin-receptor blockingagent with a neprilysin inhibitor in women than in

men with heart failure with preserved ejection frac-tion (10). This finding has important implications forfuture clinical trials. Thus, in both the laboratory andin the clinic, we have no excuse not to make everyeffort to include both males and females in our in-vestigations. We stand to learn more about underly-ing mechanisms of health and disease and about howto provide optimum care for each individual we treatin the clinic.

ADDRESS FOR CORRESPONDENCE: Dr. Peter Libby,Division of Cardiovascular Medicine, Department ofMedicine, Brigham and Women’s Hospital, HarvardMedical School, 77 Avenue Louis Pasteur, Boston,Massachusetts 02115. E-mail: [email protected].

RE F E RENCE S

1. Prochnicki T, Latz E. Inflammasomes on thecrossroads of innate immune recognition andmetabolic control. Cell Metab 2017;26:71–93.

2. Ridker PM, Everett BM, Thuren T, et al. Antiin-flammatory therapy with canakinumab foratherosclerotic disease. N Engl J Med 2017;377:1119–31.

3. Duewell P, Kono H, Rayner KJ, et al. NLRP3inflammasomes are required for atherogenesis andactivated by cholesterol crystals. Nature 2010;464:1357–61.

4. Menu P, Pellegrin M, Aubert JF, et al. Athero-sclerosis in ApoE-deficient mice progresses inde-pendently of the NLRP3 inflammasome. Cell DeathDis 2011;2:e137.

5. Chen S, Markman JL, Shimada K, et al. Sex-specific effects of the Nlrp3 inflammasome onatherogenesis in LDL receptor-deficient mice.J Am Coll Cardiol Basic Trans Science 2020;5:582–98.

6. Klein SL, Flanagan KL. Sex differences in im-mune responses. Nat Rev Immunol 2016;16:626–38.

7. Lu HS, Schmidt AM, Hegele RA, et al. Annualreport on sex in preclinical studies: arterioscle-rosis, thrombosis, and vascular biology publica-tions in 2018. Arterioscler Thromb Vasc Biol 2020;40:e1–9.

8. Ridker PM, Danielson E, Fonseca FA, et al.Rosuvastatin to prevent vascular events in men

and women with elevated C-reactive protein.N Engl J Med 2008;359:2195–207.

9. Ginsberg HN, Elam MB, Lovato LC, et al. TheACCORD Study Group. Effects of combination lipidtherapy in type 2 diabetes mellitus. N Engl J Med2010;362:1563–74.

10. Solomon SD, McMurray JJV, PARAGON-HFSteering Committee and Investigators. Angiotensin-neprilysin inhibition in heart failure with preservedejection fraction. N Engl J Med 2019;381:1609–20.

KEY WORDS atherosclerosis, gender,inflammation, IL-18, IL-1b, NLRP3 inflammasome,sex













































T H E C C B Y L I C E N S E ( h t t p : / / c r e a t i v e c o mm o n s . o r g / l i c e n s e s / b y / 4 . 0 / ) .


Highly Reactive Isolevuglandins PromoteAtrial Fibrillation Caused by Hypertension
Joseph K. Prinsen, DO, PHD,a,b Prince J. Kannankeril, MD, MSCI,c Tatiana N. Sidorova, PHD,a,b
Liudmila V. Yermalitskaya, MS,a,b Olivier Boutaud, PHD,a,b Irene Zagol-Ikapitte, PHD,a,b Joey V. Barnett, PHD,a,b

Matthew B. Murphy, PHARMD,a,b Tuerdi Subati, MD, PHD,a,b Joshua M. Stark, BA,a,b Isis L. Christopher, BS,a,b

Scott R. Jafarian-Kerman, MD, MSCI,a,b Mohamed A. Saleh, PHD,a,b Allison E. Norlander, PHD,a,b

Roxana Loperena, PHD,a,b James B. Atkinson, MD, PHD,d Agnes B. Fogo, MD,d James M. Luther, MD,a,b

Venkataraman Amarnath, PHD,a,b Sean S. Davies, PHD,a,b Annet Kirabo, PHD,a,b Meena S. Madhur, MD, PHD,a,b

David G. Harrison, MD,a,b Katherine T. Murray, MDa,b

ISSN 2452-302X

From the aDepartment of M

cology, Vanderbilt Universit

of Medicine, Nashville, Ten

School of Medicine, Nashvill

VISUAL ABSTRACT

e

y

n

e

Prinsen, J.K. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):602–15.

dicine, Vanderbilt University School of Medic

School of Medicine, Nashville, Tennessee; cDe

essee; and the dDepartment of Pathology, Mi

, Tennessee. This work was supported by N


ine, Nashville, Tennessee; bDepartment of Pharma-

partment of Pediatrics, Vanderbilt University School

crobiology, and Immunology, Vanderbilt University

ational Heart, Lung, and Blood Institute grants



http://creativecommons.org/licenses/by/4.0/

R E V I A T I O N S

D ACRONYM S

BA = 2-

xylbenzylamine

BA = 4-

xylbenzylamine

atrial fibrillation

II = angiotensin II

= atrial natriuretic

de

= B-type natriuretic

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Prinsen et al.J U N E 2 0 2 0 : 6 0 2 – 1 5 Isolevuglandins and Atrial Fibrillation

603

HIGHLIGHTS

� IsoLGs are highly reactive lipid dicarbonyl metabolites that constitute a major component of oxidative

stress-related injury, and they promote the formation of amyloid.

� In a hypertensive murine model, IsoLG adducts and PAOs developed in the atria, along with inducible AF.

� IsoLG and PAO accumulation and AF were prevented by the dicarbonyl scavenger 2-HOBA, but not by an

inactive analog 4-hydroxybenzylamine.

� Mechanically stretched atrial cells generated cytosolic IsoLG adducts and PAOs that were prevented by 2-HOBA.

� Natriuretic peptides generated cytotoxic oligomers, a process accelerated by IsoLGs, contributing to atrial PAO

formation.

� These findings identify a novel pathway during oxidative stress to increase AF susceptibility, and they support the

concept of preemptively scavenging reactive downstream mediators as a potential therapeutic approach to prevent AF.

AB B

AN

2-HO

hydro

4-HO

hydro

AF =

ang

ANP

pepti

BNP

peptide

blood pressure

= electrocardiogram
SUMMARY
BP =

ECG

G/R = green/red ratio

IsoLG = isolevuglandin

PAO = preamyloid oligomer

PBS = phosphate-buffered

saline

ROS = reactive oxygen species

HL

Bo

He

18

Ce

Am

is

23

su

rep

Co

affi

is

Em

Th

sti

the

Ma

Oxidative damage is implicated in atrialfibrillation (AF), but antioxidants are ineffective therapeutically. The authors

tested the hypothesis that highly reactive lipid dicarbonyl metabolites, or isolevuglandins (IsoLGs), are principal

drivers of AF during hypertension. In a hypertensive murine model and stretched atriomyocytes, the dicarbonyl

scavenger 2-hydroxybenzylamine (2-HOBA) prevented IsoLG adducts and preamyloid oligomers (PAOs), and AF

susceptibility, whereas the ineffective analog 4-hydroxybenzylamine (4-HOBA) had minimal effect. Natriuretic

peptides generated cytotoxic oligomers, a process accelerated by IsoLGs, contributing to atrial PAO formation.

These findings support the concept of pre-emptively scavenging reactive downstreamoxidative stressmediators as

a potential therapeutic approach toprevent AF. (J AmColl Cardiol Basic Trans Science 2020;5:602–15)©2020The

Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open access

article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

A trial fibrillation (AF) is epidemic in the UnitedStates and worldwide, and it often results indevastating outcomes such as stroke and

congestive heart failure (1). Nevertheless, currentlyavailable treatment designed to prevent or interruptthe AF substrate has met with only limited success,with the potential for serious adverse effects. Thus,there is a critical need for improved understandingof the underlying mechanisms causing AF and novelstrategies to treat it.

096844 and HL133127 to Dr. Murray and K01HL130497 to Dr. Kirabo, the N

utaud, the National Institute of General Medical Sciences grant T32 GM

alth, the American Heart Association, Southeast Affiliate grant 216

SFRN34230125 to Dr. Dan Roden (Dr. Murray is the Basic Project PI). Drs.

nter for Advancing Translational Sciences of the National Institute of

arnath and Murray have a pending patent application with Metabolic Tec

a patent holder for use of 2-HOBA, an isolevuglandin scavenger. Drs. Kira

2,615. Confocal microscopy and image analysis were performed through

pported by the National Institutes of Health [CA68485, DK20593, DK584

orted that they have no relationships relevant to the contents of this pape

llege of Medicine, University of Tennessee Health Science Center, Mem

liated with the Center for Drug Evaluation and Research, U.S. Food and Dr

currently affiliated with the Clinical Sciences Department, College of Me

irates.

e authors attest they are in compliance with human studies committees

tutions and Food and Drug Administration guidelines, including patient co

JACC: Basic to Translational Science author instructions page.

nuscript received January 2, 2020; revised manuscript received March 31

There is abundant evidence linking oxidativestress and reactive oxygen species (ROS) directly tothe pathogenesis and progression of AF (2). Inflam-matory cells generate ROS, and inflammation-mediated AF is the most common and costlycomplication of cardiac surgery, as well as themechanism of early recurrence following catheterablation (3–5). In addition, multiple risk factors forAF, including hypertension, obesity, and aging, aremechanistically linked to oxidative stress (6,7).

ational Institute of Aging grant 5R44AG005184 to Dr.

007569 to Dr. Prinsen at the National Institutes of

0035 to Dr. Murray, and National Center grant

Prinsen and Murray are supported by the National

Health under Award Number UL1 TR000445. Drs.

hnologies, Inc., and Vanderbilt University. Dr. Davies

bo and Harrison are coinventors on U.S. Patent # 14/

the Vanderbilt Cell Imaging Shared Resource (also

04, DK59637 and EY08126]). All other authors have

r to disclose. Dr. Stark is currently affiliated with the

phis, Tennessee. Dr. Jafarian-Kerman is currently

ug Administration, Silver Spring, Maryland. Dr. Saleh

dicine, University of Sharjah, Sharjah, United Arab

and animal welfare regulations of the authors’ in-

nsent where appropriate. For more information, visit

, 2020, accepted April 2, 2020.

http://creativecommons.org/licenses/by/4.0/


FIGURE 1 Mechanism of IsoLG Scavengers

The 1,4-dicarbonyl (red box) IsoLGs interact rapidly with lysines to form lactam adducts and crosslinking of proteins. The phenolic

amine pyridoxamine and its structural analog 2-HOBA (blue box) react with IsoLGs at a rate several orders of magnitude more rapidly

than they react with lysines, thus serving as scavengers to prevent adduct formation. 2-HOBA ¼ 2-hydroxylbenzylamine;

IsoLG ¼ isolevuglandin.

Prinsen et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Isolevuglandins and Atrial Fibrillation J U N E 2 0 2 0 : 6 0 2 – 1 5

604

Unfortunately, “upstream therapy” targeting ROSlevels directly with dietary antioxidants has beenineffective in clinical trials (8), in part because theyfail to actually reduce oxidative injury in humans.Nonspecific ROS scavenging may also interfere withphysiological ROS signaling.

Polyunsaturated fatty acid oxidation leads to theformation of highly reactive aldehydes. The mostreactive of these products are dicarbonyl compoundsknown as isolevuglandins (IsoLGs) (also calledg-ketoaldehydes or isoketals [9,10]) (Figure 1). Theyadduct proteins almost instantaneously, causingmisfolding and crosslinks (9). Tissue IsoLG adductsare elevated early in multiple diseases linked toinflammation and oxidative stress, including hyper-tension, obesity, atherosclerosis, and Alzheimer’sdisease (11–15). Moreover, IsoLGs induce multiple ef-fects that drive disease, including cytotoxicity, acti-vation of inflammation and cytokine secretion, andacceleration of amyloidosis. In Alzheimer’s, mis-folded protein amyloid b1-42 monomers coassembleinitially to form soluble preamyloid oligomers (PAOs),

now recognized to be the primary cytotoxic speciescorrelating with disease progression rather thandownstream amyloid fibril deposition (16,17). Impor-tantly, IsoLGs markedly accelerate the oligomeriza-tion of amyloid b1-42 (18,19), providing apathophysiological link between oxidative stress andproteotoxicity. As in the brain, amyloidosis developsin the human atrium with aging (20–22), and werecently identified PAOs in human atrial tissue (23).

In a cellular model simulating AF, we previouslyfound that rapid stimulation of atrial cells caused theformation of IsoLG adducts and protein oligomerswithin hours (24). We hypothesized that IsoLGs aremolecular drivers of the AF substrate, constituting anovel mechanism to increase arrhythmia suscepti-bility. We chose a model of hypertension to test thishypothesis for several reasons. First, we found thatthe presence of protein oligomers in the humanatrium was linked to hypertension (23). Second,considerable evidence implicates oxidative damageand inflammation in the development of hyperten-sion (11,25). Third, it was recently demonstrated that


605

IsoLG adducts are indeed formed during experi-mental hypertension, serving as neoantigens to pro-mote dendritic and T-cell activation (11). In thepresent studies, we report that IsoLGs and PAOsdevelop in the atrium during murine hypertensionand define a pathophysiological pathway linkingoxidative stress and AF susceptibility. The findingsidentify downstream mediators of ROS-related injuryas novel, alternative therapeutic targets for the pre-vention and treatment of AF.

METHODS

ANIMAL USE. Male C57Bl/6J mice were obtained fromJackson Laboratory (Bar Harbor, Maine) and studiedat 3 months of age. Hypertension was induced bycontinuous infusion of angiotensin II (ang II)(490 ng/kg/min) via osmotic minipumps (Alzet,Durect Corp., Cupertino, California) for 2 weeks. Bloodpressure (BP) was monitored using tail cuff measurementspreceded by acclimation. Oral 2-hydroxylbenzylamine(2-HOBA) (1 g/l), 4-hydroxylbenzylamine (4-HOBA)(1 g/l), or hydralazine þ hydrochlorothiazide (320 mg/land 60 mg/l, respectively) was delivered via drinkingwater (11).

ATRIAL HL-1 CELL CULTURE. Atrial HL-1 cells weregrown in Claycomb Medium (Sigma-Aldrich, Boston,Massachusetts) supplemented with 10% fetal bovineserum, 0.1 mmol/l norepinephrine, 2 mmol/l L-gluta-mine, and 0.1 mmol/l norepinephrine as describedpreviously (24,26). Near-confluent/confluent cells(grown on a BioFlex Culture Plate for 48 h; FlexcellInternational, Burlington, North Carolina) wereexposed to 10% cyclical stretch at a rate of 1 Hz for24 h using the Flexcell FX-5000 Tension System(Flexcell International) (27).

IsoLG ADDUCTS. Immunohistochemistry. Formalinfixed hearts were subjected to immunohistochemistryusing an anti–IsoLG-lysyl adduct single-chain anti-body (D11 ScFv) characterized previously (28). Imageswere captured using a high-throughput Leica SCN400slide scanner automated digital image system fromLeica Microsystems (Wetzlar, Germany). Whole slideswere imaged at 20� magnification to a resolution of0.5 mm/pixel. Tissue cores were mapped using AriolReview software (Leica Biosystems Richmond, Rich-mond, Illinois). Because rapid stimulation of atrialcells can produce IsoLGs and PAOs, atrial tissue wasanalyzed for these parameters only from animals notsubjected to electrophysiological studies.

QUANTITATION BY MASS SPECTROMETRY. Flash-frozen atria were thawed in 4 ml of phosphate-buffered saline (PBS) containing indomethacin

100 mmol/l (Sigma-Aldrich) to prevent formation ofIsoLGs via oxygenation by cyclooxygenase of arach-idonic acid released during the process, and pyri-doxamine 1 mmol/l (Sigma-Aldrich) as an IsoLGscavenger. Tissues were homogenized using a jawhomogenizer and tissue grind tubes, beforecentrifugation at 10,000 � g for 20 min at 4�C. Thesupernatant was collected for protein IsoLG ad-ducts analysis.

Cells subjected to stretch, and control cells simul-taneously cultured on BioFlex plates, but withoutstretch, were incubated with indomethacin and pyr-idoxamine, in 1 ml of PBS (pH 7.4) at 4�C for 30 minbefore harvest.

Protein concentrations in homogenized atria orcells were measured using a BCA Protein Assay kit(Pierce, Rockford, Illinois), and samples were sub-jected to complete enzymatic digestion to individualamino acids (15). A [13C6] internal standard wasadded, and the IsoLG-lysyl adducts were purified bysolid-phase extraction and high-performance liquidchromatography before being quantified by liquidchromatography-tandem mass spectrometry assayusing isotopic dilution as described previously (29).

QUANTITATION OF PAOs. Immunostaining was per-formed on optimal cutting temperature compound–embedded myocardial sections using a mousemonoclonal antibody specific for striated muscle(MF20, 1:10, Developmental Studies Hybridoma Bank,Iowa City, Iowa) to label myocardium, and a rabbitpolyclonal antibody (A11, 1:3,000, EMD Millipore,Darmstadt, Germany) recognizing a conformationalepitope common to all PAOs (30,31), with secondarygoat anti-mouse Alexa 568–conjugated and donkeyanti-rabbit Alexa 488–conjugated antibodies (Molec-ular Probes, Eugene, Oregon), respectively. Confocalimages were acquired from the tissue sections, and apreviously validated method was used to quantify therelative myocardial surface area (red) that containedPAOs (green), or green/red ratio (G/R), as a spatialrepresentation of PAO burden in an atrial sample (32).

QUANTITATION OF FIBROSIS. Atrial samples weresectioned (5 mm) and stained using a standard Mas-son’s trichrome procedure to visualize collagen-richtissue. Digitized images of the entire specimen wereacquired using a high-throughput Leica SCN400 slidescanner imaged at 20� magnification (resolution0.5 mm/pixel). Tissue cores were mapped using AriolReview software, and the number of blue pixels wasquantified as percentage of atrial myocardium.

ALKALINE CONGO RED STAINING. Tissue sectionswere stained in Congo red solution using standardmethods. Positive controls with known amyloid were



606

stained and examined concurrently, and demon-strated apple green birefringence under polarizedlight. Experimental samples were evaluated by apathologist (J.B.A., A.B.F.) blinded to experi-mental groups.

TRANSESOPHAGEAL ELECTROPHYSIOLOGICAL

STUDIES. AF was induced during a transesophagealelectrophysiological study by an operator blinded totreatment (33). Mice were anesthetized with iso-flurane, and a surface electrocardiogram (ECG) (leadI) recording was obtained using subcutaneous 27-ganeedles in each forelimb. The ECG channel wasamplified (0.1 mV/cm) and filtered between 0.05 and400 Hz. A 2-F octapolar electrode catheter (CIBercath, NuMED, Hopkinton, New York) was positionedin the esophagus with placement adjusted until reli-able atrial capture was obtained. Bipolar pacing wasperformed with a 1-ms pulse width at 3 mA. Baselineintervals were measured, and standard clinical elec-trophysiological pacing protocols were used todetermine the atrioventricular effective refractoryperiod and Wenckebach cycle length. AF inducibilitywas measured after burst atrial pacing (6 separate 15-strains delivered at cycle lengths of 50, 40, 30, 25, 20,and 15 ms, respectively). AF was defined as develop-ment of rapid atrial activity with an irregularlyirregular ventricular response lasting at least 1 s. Thestudy was terminated for an animal if AF lasting10 min occurred. Data were analyzed to quantitatetotal AF duration, representing the AF burden.

OLIGOMER GENERATION AND WESTERN BLOT

ANALYSIS. Synthetic a-atrial natriuretic peptide(ANP) (1-28) (SLRRSSCFGGRMDRIGAQSGLGCNSFRY-disulfide bond [C7-C23]) and B-type natriuretic pep-tide (BNP) (SPKMVQGSCFGRKMDRISSSSGLGC-KVLRRH-disulfide bond [C10 to C26]) peptides weregenerated by RS Synthesis (Louisville, Kentucky). Totest for oligomerization, peptide (10 mmol/l) was pre-pared in PBS buffer (pH 7.4) and incubated at roomtemperature for 24 h or up to 6 days. A separate samplewas incubated for 24 h with either 2 to 4 molar equiv-alent of synthetic IsoLGs or dimethyl sulfoxide(vehicle) as described (24). After incubation, peptideswere subjected to Western analysis. Briefly, equalamounts of peptide samples were resolved with aNuPage Bis-Tris 4-12% gel (Thermo Fisher Scientific,Waltham, Massachusetts) and transferred to a poly-vinylidene difluoride membrane at 30 V for 1 h on ice.Blotswere then blocked in 5% (w/v) nonfatmilk in Tris-buffered saline 0.1% Tween 20 buffer and incubated inanti–a-ANP or anti-BNP antibody (1:500, PhoenixPharmaceuticals, Burlingame, California) overnight.The antigens were detected by luminescence

method (enhanced chemiluminescent kit PierceECL Substrate, Thermo Fisher Scientific), usinghorseradish peroxidase–conjugated secondary (goatanti-rabbit) antibody (1:5,000, Jackson ImmunoR-esearch, West Grove, Pennsylvania).

IMMUNOHISTOCHEMISTRY FOR NATRIURETIC

PEPTIDES. Adjacent frozen sections of atrium wereimmunostained for A11 and either ANP or BNP. Fornatriuretic peptides, immunostaining was performedusing primary rabbit polyclonal anti–a-ANP (1-28;1:200) and anti-BNP (1:500) antibodies (PhoenixPharmaceuticals) as described previously forANP (23).

CYTOTOXICITY. BNP and ANP oligomers weregenerated by incubating the peptides at room tem-perature for 24 h, 3 days, and 7 days at a concentra-tion of 30 mmol/l in PBS. Atrial HL-1 cells wereplated at a density of 25,000 cells per 100 ml ClaycombMedium/well in a 96-well microplate (PerkinElmer, Waltham, Massachusetts) pre-coated withgelatin and fibronectin, and incubated overnight(37�C, 5% CO2). Cells were then treated with BNP andANP oligomers (0.45 mmol/l) for 24 h. At the end ofthe treatment, cytotoxicity of BNP and ANP oligo-mers on HL-1 cells were determined by measuringcellular ATP levels with an ATPlite assay (PerkinElmer) according to the manufacturer’s instructions.Luminescence was measured using a Lumicountmicroplate reader (Global Medical Instrumentation,Ramsey, Minnesota).

STATISTICAL ANALYSIS. Data are expressed as mean� SEM. For data with a skewed (non-normal) distri-bution, nonparametric Mann-Whitney U test wasused to compare the differences in IsoLG adducts,G/R values, AF inducibility, and fibrosis (Figures 2B to2E, 4A, and 4C, Supplemental Figures 1 and 2). Thetime and treatment effects on BP, as well as themodified effect of treatment by time, were analyzedusing 2-way analysis of variance for repeated mea-sures (Figure 4B). This is equivalent to a linear mixed-effects model with fixed effects on time, treatment,and their interaction and random intercept. The ef-fect of incubation times on protein oligomer cyto-toxicity was compared using 1-way analysis ofvariance with Tukey’s post hoc multiple pairwisecomparison test (Figure 5C). A p value of <0.05 wasconsidered statistically significant. Statistical analysiswas performed using GraphPad Prism softwareversion 7.02 (GraphPad Software, La Jolla, California).

STUDY APPROVAL. All animal procedures wereapproved by the Vanderbilt Institutional Animal Careand Use Committee. Mice were housed and cared forin accordance with the Guide for the Care and Use of


FIGURE 2 Hypertension Promotes the Formation of Atrial IsoLG Protein Adducts and PAOs, Which Is Inhibited by 2-HOBA

(A) During ang II–mediated hypertension (ang II), striking accumulation of IsoLG protein adducts is demonstrated in left (LA) and right (RA) atria using immunolabeling

with an anti–IsoLG-lysyl adduct antibody (D11 ScFv; n ¼ 2, 4 for sham and ang II–treated mice, respectively; scale bars ¼ 50 mm) compared with control mice (sham).

(B) Summary data are shown for quantitation of IsoLG adducts in LA and RA using liquid chromatography-tandem mass spectrometry assay (mean � SEM; n ¼ 5 each;

**p < 0.01 between indicated groups, ns is nonsignificant, nonparametric Mann-Whitney U test). (C) Representative mass spectrometry traces are shown for IsoLG

adduct quantitation in LA from sham, ang II, and ang IIþ2-HOBA–treated mice, along with the internal standard in red (Std). (D) Confocal images are shown for

myocardium (red) and PAOs (green) on the left, and PAOs localized to the myocardium on the right, from control and hypertensive mice, with PAO burden expressed

as G/R values (scale bars ¼ 20 mm). (E) Summary data are illustrated for oligomer burden in LA and RA (n ¼ 11, 16, 9, 5 per group for LA; n ¼ 5, 4, 9, 3 per group for RA;

*p < 0.05, **p < 0.01 between indicated groups, nonparametric Mann-Whitney U test) (Scale bars ¼ 50 mm). (F) 2-HOBA prevented development of IsoLG adducts

(upper panel) and PAOs (lower panel) during ang II–mediated hypertension (also see B and E), whereas the inactive analog 4-HOBA had minimal effect. ang

II ¼ angiotensin II; PAO ¼ preamyloid oligomer; other abbreviations as in Figure 1.


607

Laboratory Animals, U.S. Department of Health andHuman Services.

RESULTS

HYPERTENSION CAUSES FORMATION OF ATRIAL

IsoLGAdducts AND PAOs, WHICH IS PREVENTED BY

THE DICARBONYL SCAVENGER 2-HOBA. Given thatIsoLGs are formed in the vasculature during experi-mental hypertension (11), we hypothesized that thisalso occurs in the atrium. Immunohistochemistry wasperformed in the atria of mice rendered hypertensiveby minipump infusion of angiotensin II (ang II) (27)using a single-chain antibody (D11 ScFv) that recog-nizes IsoLG-lysyl adducts on any protein (28). Hy-pertension caused diffuse IsoLG protein adductaccumulation in both the left and right atria(Figure 2A), which was absent in the atria of normo-tensive sham animals. This finding was confirmed byquantifying IsoLG adducts using mass spectrometry,with a significant increase in adduct formation inboth atria of hypertensive animals (Figures 2B and 2C).

Small-molecule compounds, exemplified by 2-HOBA, have been identified that react with IsoLGs topre-emptively scavenge these and closely relateddicarbonyl mediators to prevent downstream protein

modification (34,35). When mice were cotreated with2-HOBA (starting 3 days before ang II infusion), theformation of IsoLG adducts during hypertension wasprevented (Figures 2B and 2F). A separate group ofhypertensive mice was treated with the related struc-tural analog 4-HOBA, which is a very poor scavenger ofIsoLGs (11,34). For these animals, IsoLG adduct levelswere not significantly different from those seen inmice treated with ang II alone (Figure 2B), indicatingthe specificity of the effects of 2-HOBA to scavengeIsoLGs.

We have previously shown that amyloid-relatedprotein oligomers develop in the atria of patientsundergoing cardiac surgery, where the oligomers arelinked to hypertension (23). To determine whetherPAOs are formed in murine atrium during hyperten-sion, we performed immunohistochemistry using aconformation-specific antibody (A11) recognizingPAOs derived from any protein irrespective of aminoacid sequence (30). Compared with normotensiveanimals, hypertension led to significant accumulationof PAOs in both the left and right atria (Figures 2D and2E). As for IsoLG adducts, this effect was abrogated by2-HOBA (Figures 2E and 2F), whereas the inactivestructural analog 4-HOBA failed to prevent PAO for-mation (Figure 2E).

FIGURE 3 IsoLG Adducts and PAOs Develop at an Early Point During Hypertension, When Histological Abnormalities Are Absent

For normotensive (sham), hypertensive (ang II), and 2-HOBA–treated hypertensive (ang IIþ2-HOBA) animals, columns from left to right

display representative atrial images after exposure to hematoxylin and eosin (H&E), Masson’s trichrome, and Congo red stains. There was no

evidence of myocardial structural abnormalities or amyloid, with minimal fibrosis that was similar between groups (see text). Scale

bars ¼ 50 mm. Abbreviations as in Figures 1 and 2.



608

IsoLG ADDUCTS AND PAOs DEVELOP EARLY DURING

HYPERTENSION. Histochemical staining was per-formed to determine whether additional myocardialabnormalities were present in this model (Figure 3).Hematoxylin and eosin staining showed no differencein atrial histology between hypertensive and normo-tensive control mice. In addition, Masson’s trichromestaining demonstrated minimal fibrosis in sham, angII–treated, and ang IIþ2-HOBA–treated animals(Supplemental Figure 2) (5.0 � 0.6%, 5.0 � 0.7%, and

4.6 � 0.5%, respectively; n ¼ 5 each), with no evidenceof amyloid formation by Congo red staining. Thus,IsoLGs and PAOs occurred early in the pathogenesis ofthis hypertensive model before the development ofsignificant atrial structural abnormalities.2-HOBA SUPPRESSES HYPERTENSION-MEDIATED

ATRIAL FIBRILLATION. AF susceptibility was inves-tigated in control and hypertensive mice usingtransesophageal electrophysiological studies thatemployed rapid atrial burst pacing (33). Compared


FIGURE 4 2-HOBA Prevented AF in Hypertensive Mice and Suppressed IsoLG Adduct and PAO Formation in Mechanically Stretched Atrial Cells

(A) Total AF burden was increased in hypertensive (ang II) mice compared with controls (sham; n ¼ 13, 22; **p < 0.01, nonparametric Mann-Whitney U test). During

hypertension, cotreatment with 2-HOBA reduced AF burden, whereas the inactive structural analog 4-HOBA had no effect (ang IIþ2-HOBA, ang IIþ4-HOBA; n ¼ 14, 7,

respectively; *p < 0.05, nonparametric Mann-Whitney U test). Blood pressure normalization with hydralazine/hydrochlorothiazide (H/H) and cessation of ang II also led

to a reduction AF (ang IIþH/H, and ang II recovery; n ¼ 7, 12, respectively; *p < 0.05, nonparametric Mann-Whitney test). (B) Summary data for systolic blood pressure

are illustrated for the groups studied (*p < 0.01 compared with sham, †p < 0.01 compared with ang II, 2-way analysis of variance for repeated measures). (C) Atrial

HL-1 cells were subjected to either no stretch or stretch (10% at 1 Hz) for 48 h and analyzed by liquid chromatography-tandem mass spectrometry assay. Stretch

caused robust development of IsoLG adducts, which was abrogated by 2-HOBA (n ¼ 6 each; **p < 0.01, nonparametric Mann-Whitney U test). (D) Immunostaining

demonstrates that atrial cells developed PAOs in response to stretch (lower left) compared with no stretch (upper right) or during stretch in the presence of 2-HOBA

(lower right). AF ¼ atrial fibrillation; other abbreviations as in Figures 1 and 2.


609

with control mice, the total amount or burden ofinducible AF was significantly increased in hyper-tensive mice (Figure 4A). The AF substrate wasreversible, with a 95% reduction in total AF burdenwithin 2 weeks after stopping ang II (Figure 4A)(associated with a 70% reduction in BP (SupplementalFigure 1) (n ¼ 12), providing further support thatIsoLGs were generated early in the development ofthe AF substrate. Cotreatment with 2-HOBA signifi-cantly reduced AF burden compared with ang II alone

(Figure 4A), whereas for mice receiving 4-HOBA, AFburden was comparable to that seen with animalsreceiving ang II alone. There were no effects of2-HOBA on any ECG or electrophysiological parame-ters (Table 1). Taken together with the results shownin Figure 2, these findings demonstrate that ang II–mediated hypertension promotes the formation ofatrial IsoLGs, PAOs, and AF susceptibility, withIsoLGs playing a critical role in the pathophysiolog-ical process.



FIGURE 5 ANP and BNP Form Cytotoxic Protein Oligomers in Hypertensive Atria

(A)Western blotting is shown following incubation of ANP peptide (10 mmol/l) at 22oC for 24 h or 6 days, compared with incubation with IsoLGs (synthetic 15-E2-IsoLG,

1 mmol/l) for 24 h, demonstrating time-dependent oligomerization that is markedly accelerated by IsoLGs. (B) Similar results are shown for BNP (10 mmol/l) following

0 to 3 days of incubation in the absence and presence of IsoLGs. (C and D) ANP and BNP (30 mmol/l) were allowed to oligomerize for 1, 3, and 7 days. Oligomers were

incubated with atrial HL-1 cells (0.45 mmol/l for 24 h), followed by quantitation of cellular ATP production expressed as % change from control untreated cells. Upon

exposure to oligomers, there was a reduction in ATP production indicative of cytotoxicity that declined significantly with increased oligomerization time for ANP

(mean � SEM; n ¼ 5 independent experiments; *p < 0.05, 1-way analysis of variance with Tukey’s multiple comparison test). (E) Immunofluorescent labeling with A11

(left) and ANP- or BNP-specific antibodies (middle) was performed in adjacent 5-mm atrial sections from a hypertensive mouse (scale bars ¼ 50 mm). Evidence of partial

colocalization of natriuretic peptides with PAOs (right) is indicated by lighter greenish yellow color. ANP ¼ atrial natriuretic peptide; ATP ¼ adenosine triphosphate;

BNP ¼ B-type natriuretic peptide; other abbreviations as in Figures 1 and 2.



610

ATRIAL STRETCH CAUSES IsoLG AND PAO FORMATION

THAT IS SUPPRESSED BY 2-HOBA. In a separate cohortof mice receiving ang II, BP was normalized by theconcomitant administration of hydralazine and hy-drochlorothiazide, and this was associated with a lowAF burden similar to that of sham-treated controlmice (Figures 4A and 4B). To investigate the role ofatrial myocyte stretch in the pathophysiological pro-cess, atrial HL-1 cells were cultured in the absenceand presence of 10% cyclical stretch. Exposure tostretch caused a substantial increase in IsoLG adducts(Figure 4C), as well as the generation of protein olig-omers (Figure 4D), and both effects were prevented inthe presence of 2-HOBA. These findings point to acausative role for atrial cell stretch in the patho-physiology of AF susceptibility during hypertension.

ISOLEVUGLANDINS ACCELERATE FORMATION OF

CYTOTOXIC NATRIURETIC PEPTIDE OLIGOMERS,

WHICH CONTRIBUTE TO HYPERTENSION-MEDIATED

PAOs IN THE ATRIA. The amyloid-forming proteinANP is a prominent component in aging-related (se-nile) atrial amyloidosis, and some studies support thepresence of BNP in these deposits as well (20,22,36).Given that ANP is a component of the PAOs that form

in both human atrium and rapidly stimulated atrialcells (23,24), we investigated the role of natriureticpeptides in hypertension-mediated PAOs usingseveral approaches. Purified ANP and BNP incubatedat room temperature demonstrated time-dependentoligomerization, indicated by the development ofadditional higher molecular weight bands on Westernblot analysis (Figures 5A and 5B). However, whenincubated in the presence of IsoLGs, PAO formationwas markedly accelerated. We then examinedwhether natriuretic peptide oligomers were detri-mental to atrial cells. Both ANP and BNP oligomersreduced ATP production in atrial HL-1 cells, indi-cating cytotoxicity (Figures 5C and 5D). This effect wasmost pronounced for oligomers formed during a 1-dayincubation, whereas cytotoxicity progressivelydeclined with longer incubation times, most promi-nently for ANP (Figure 5C). This time course is anal-ogous to that observed for amyloid b1-42 neuronalinjury: as monomers coalesce to oligomers and sub-sequently to less toxic fibrils, PAO formation andassociated cytotoxicity develops and then declines ina time- and concentration-dependent manner (17).Finally, adjacent sections of hypertensive mouse atriawere immunostained for PAOs and either ANP or BNP,

TABLE 1 Intergroup Comparison of Electrophysiological Parameters

Sham(n ¼ 9)

Ang II(n ¼ 14) p Value

Ang IIþ2-HOBA(n ¼ 10) p Value*

SCL, ms 127 � 5 119 � 3 0.25 117 � 4 0.76

PR, ms 39 � 1 38 � 1 0.58 38 � 1 0.55

QRS, ms 13 � 1 13 � 1 0.45 14 � 1 0.89

QT, ms 43 � 2 44 � 1 0.99 41 � 1 0.39

AVERP, ms 56 � 2 54 � 2 0.31 57 � 1 0.14

WCL, ms 77 � 2 76 � 2 0.23 77 � 1 0.16

Values are mean � SEM. *Comparison of angiotensin II (ang II) þ 2-hydroxylbenzylamine (2-HOBA) with ang II.

2-HOBA ¼ 2-hydroxylbenzylamine; AVERP ¼ atrioventricular effective refractory period;SCL ¼ sinus cycle length; WCL ¼ Wenckebach cycle length.


611

with results demonstrating evidence of partialcolocalization of natriuretic peptides with atrialoligomers (Figure 5E). Taken together, these findingssupport a role for cytotoxic ANP and BNP oligomers aspotential mediators of atrial pathophysiology dur-ing hypertension.

DISCUSSION

As the most common sustained cardiac arrhythmia,AF constitutes a significant public health problem forwhich optimal medical therapies are lacking. Eluci-dating early mechanisms that increase AF suscepti-bility are critical to develop effective preventativeand therapeutic strategies. In this study, we identi-fied a novel role for highly reactive IsoLGs in thepathophysiology of hypertension-mediated AF. Usinga murine model of hypertension, we found that atrialIsoLG adducts and cytotoxic protein oligomers weregenerated before histological abnormalities, associ-ated with AF susceptibility that was reversible whenBP declined. These detrimental effects were pre-vented by the dicarbonyl scavenger 2-HOBA, but notthe ineffective analog 4-HOBA, confirming the spec-ificity of this biochemical mechanism. Experimentsin vitro and in vivo revealed a critical role of atrialmyocyte stretch in the generation of IsoLG adductsduring the pathophysiological process. These findingssupport the concept of pre-emptively scavengingreactive downstream mediators of oxidative stress,rather than targeting ROS generation per se, as anovel therapeutic approach to prevent AF (Figure 6).

With inflammation and oxidative stress, peroxi-dation of fatty acids generates multiple reactive al-dehydes, including malondialdehyde (MDA), 4-oxo-2-nonenal, and IsoLGs (9,10,37,38). The toxicity of suchcompounds is markedly augmented by the presenceof 2 carbonyl groups (C¼O), and IsoLGs have a 1,4-dicarbonyl ring configuration that renders themextremely reactive (Figure 1) (9,34). These com-pounds react nearly instantaneously with proteinsand are the most reactive products of lipid peroxi-dation identified to date (9). Indeed, they modifyproteins so rapidly that they can only be detectedin vivo as adducts rather than their unreacted form,in contrast to other lipid oxidation products.

IsoLGs form covalent adducts with amines, notablythe epsilon amine of lysines in proteins, causingirreversible protein modifications. An intermediate inthis reaction is also highly reactive, generatingintramolecular crosslinks that cause dysfunction ofproteins, including structures relevant to car-diomyocyte homeostasis, such as ion channels(39,40), HDL (13,14,41), mitochondria (42), histones

(43), and proteasomes (44). IsoLGs can also adduct toDNA and phosphatidylethanolamines (45,46). TissueIsoLG adducts are elevated early in animal models ofcardiovascular risk factors, including hypertension,obesity, and hyperlipidemia (41), as well as athero-sclerosis (13,14). They are also increased in otherdiseases linked to oxidative injury/inflammation,such as chronic ethanol exposure (47), pulmonaryfibrosis (48), Alzheimer’s disease (15), and cancer(49). To date, IsoLG adducts identified in experi-mental models have emerged as critical mediators ofoxidative injury in the brain during Alzheimer’s dis-ease, and in the vasculature during hypertension andatherosclerosis (11,13,14,50).

Multiple risk factors for developing AF are associ-ated with increased atrial pressure that promotesatrial tension/enlargement, and our results support acritical role for atrial cell stretch in the pathophysio-logical process. Atrial myocyte stretch triggers ageneralized stress response, with activation of im-mediate early genes, dedifferentiation, activation ofhypertrophic signaling cascades, and increasedrelease/production of natriuretic peptides (51,52).Importantly, stretch of ventricular myocytes causesrapid production of superoxide (53). Similarly, in thepresent study, we found that atrial myocyte stretchcauses IsoLG adduct formation, indicative of atrialROS production. Prevention of AF susceptibility usinga dicarbonyl scavenger is consistent with the conceptthat stretch-mediated oxidative stress is an earlyevent in generating the AF substrate.

Diseases related to oxidative stress are increasinglylinked to proteotoxicity as a contributing mechanism(16,31,54), in particular for neurological and cardiacdysfunction (16,17,31). The generation of atrial PAOsin this model is not unexpected on the basis of severalconsiderations. First, the development of natriureticpeptide-related amyloidosis is almost universal in the

FIGURE 6 In Response to Hypertension, Oxidative Stress-Mediated IsoLGs Promote AF Susceptibility

Hypertension and atrial cell stretch, as well as the rapid activation of atrial cells, causes oxidative stress and formation of highly reactive

IsoLGs, that rapidly adduct and crosslink cellular proteins and other macromolecules. The generation of dysfunctional adducted proteins and

protein oligomers promote atrial myocyte dysfunction to increase AF susceptibility. Abbreviations as in Figures 1 and 4.



612

aging human atrium (20–22). Second, we showed thatIsoLGs markedly accelerate the oligomerization ofANP and BNP in vitro and in cells, yielding cytotoxicoligomers, as occurs with amyloid b1-42 (18,19).Finally, elevated concentrations of amyloidogenicproteins are a major factor that drives oligomer for-mation (55), and both local and systemic concentra-tions of natriuretic peptides are increased withstretch and rapid atrial contraction. Given thatoxidative stress-mediated IsoLG formation promotesproteotoxicity in both the heart and brain, this pro-vides a potential mechanism for the pathophysiolog-ical link between AF and dementia (56).

By targeting downstream mediators of ROS-relatedinjury, dicarbonyl scavengers represent a totally

novel therapeutic approach for diseases linked tooxidative stress. Contemporary antioxidants havebeen largely ineffective in such diseases, includingAF. However, therapeutically used doses of antioxi-dants such as vitamin E and fish oil are not effectiveto reduce in vivo measures of oxidative injury (e.g.,F2-isoprostanes, widely used sensitive markers ofoxidative stress) (57–59). Dicarbonyl scavengersrepresent an alternative strategy to leave ROS gen-eration intact, but to rapidly scavenge reactive lipidmediators as they form, rendering them inactive, sothat they cannot interact with their biological targets.For 2-HOBA, structure–activity relationship assaysdemonstrated that the close proximity of themethylamine to the hydroxyl group (Figure 1) is key to

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: The mechanism

whereby oxidative stress increases AF susceptibility is not

known. This paper identifies a novel molecular pathway by which

highly reactive lipid dicarbonyl metabolites constituting a major

component of oxidative stress-related injury are mechanistically

linked to AF susceptibility during hypertension, a disease also

linked to oxidative stress. Our findings also define a novel po-

tential mechanism whereby oxidative stress promotes amyloid

formation in the atria.

TRANSLATIONAL OUTLOOK: These findings identify a novel

pathway during oxidative stress to increase AF susceptibility, and

they support the concept of pre-emptively scavenging reactive

downstream mediators, rather than targeting generation of reac-

tive oxidative species per se, as a potential therapeutic approach

to prevent AF. The scavenger 2-HOBA has been well-tolerated in

initial Phase 1 clinical trials, and a Phase 2 trial will start within the

next few months to examine its efficacy to prevent AF.


613

scavenger potency (34,35). For the related analog 4-HOBA, this structural proximity is lost—hence, thiscompound is a very poor scavenger of dicarbonyls,enabling it to serve as a negative control. Impor-tantly, 2-HOBA and its analogs are not antioxidants inthat they do not react with O₂_̄ , OONO�, or H2O2 (11),and the reduction in IsoLG adduct levels has beenattributed directly to the dicarbonyl scavenging ef-fect, and not to inhibition of ROS production and/orlipid peroxidation. Although 2-HOBA reacts withIsoLG (a 1,4-dicarbonyl) much more rapidly than withMDA (a 1,3-dicarbonyl) or methylglyoxal (a 1,2-dicarbonyl), 2-HOBA is capable of scavenging theseother dicarbonyls in vivo (11,60,61). 2-HOBA does notinhibit COX1 or COX2, and thus the production ofphysiological prostaglandins is preserved (62). Todate, in vivo studies have demonstrated a beneficialeffect of 2-HOBA in animal models of Alzheimer’sdisease (50), hypertension (11), and atherosclerosis,with improvement in high-density lipoprotein func-tion (13,14,41). Interestingly, 2-HOBA has also beenshown to prolong the life span of Caenorhabditis ele-gans by w56% (63).STUDY LIMITATIONS. A limitation of the study is thatexperiments were performed in a single murinemodel. The specifications of these mice were selectedbased on a previously published study that demon-strated accumulation of IsoLG adducts in the heartand aorta in this model (11), as proof of concept. Inaddition, male mice were chosen because the BPresponse to ang II in female mice is considerablyreduced compared with males. Preliminary data in amouse model of obesity demonstrated a similarbeneficial effect of 2-HOBA to reduce AF susceptibil-ity (12), supporting the potential generalizability ofour findings to other conditions. Although cytotoxicprotein oligomers are generated during murine hy-pertension, our findings do not prove a causative rolefor PAOs in the pathogenesis of hypertension-mediated AF. In addition, our results suggest thatthe atrial oligomers formed in this model arecomposed of additional protein components besidesANP and BNP, given that immunostaining for PAOsand natriuretic peptides demonstrates partial over-lap. Clarifying the specific nature of injurious medi-ators and identification of other PAO-formingproteins is an important goal of future studies.Finally, although our data demonstrated a reductionin cytotoxicity with longer peptide incubation timesfor ANP supporting PAOs as the cytotoxic moiety, thiswas not observed for BNP. Nonetheless, these ex-periments were performed solely to assess oligomercytotoxicity, rather than the kinetics of PAO/amyloidformation for the natriuretic peptides.

CLINICAL IMPLICATIONS. Our results provide evi-dence for a novel pathophysiological pathway in thegenesis of the AF substrate. As highly reactive medi-ators of oxidative stress-related injury, IsoLGs arelogical candidates for targeted inhibition using smallmolecule scavengers. By scavenging IsoLGs preemp-tively, dicarbonyl scavengers may represent a para-digm shift in pharmacological strategy to preventinjurious oxidative protein modification that cancause AF. Of note, 2-HOBA has been well tolerated inPhase 1 trials, with a Phase 2 trial to commence in thenear future.

CONCLUSIONS

Our findings demonstrate that hypertension pro-motes concomitant IsoLG and PAO accumulationalong with arrhythmia susceptibility in the atrium,and they identify IsoLGs as a critical molecularcomponent of this pathophysiological process. Thesefindings provide a mechanistic link between hyper-tension, oxidative stress, proteotoxicity, and AFsusceptibility.

ADDRESS FOR CORRESPONDENCE: Dr. Katherine T.Murray, Division of Clinical Pharmacology, Room 559Preston Research Building, Vanderbilt UniversitySchool of Medicine, 2220 Pierce Avenue, Nashville,Tennessee 37232-6602. E-mail: [email protected].





614

RE F E RENCE S

1. Chugh SS, Havmoeller R, Narayanan K, et al.Worldwide epidemiology of atrial fibrillation: aGlobal Burden of Disease 2010 Study. Circulation2014;129:837–47.

2. Gutierrez A, Van Wagoner DR. Oxidant and in-flammatory mechanisms and targeted therapy inatrial fibrillation: an update. J Cardiovasc Phar-macol 2015;66:523–9.

3. Jacob KA, Nathoe HM, Dieleman JM, vanOsch D, Kluin J, van Dijk D. Inflammation in new-onset atrial fibrillation after cardiac surgery: asystematic review. Eur J Clin Invest 2014;44:402–28.

4. Richter B, Gwechenberger M, Socas A, et al.Markers of oxidative stress after ablation of atrialfibrillation are associated with inflammation,delivered radiofrequency energy and early recur-rence of atrial fibrillation. Clin Res Cardiol 2012;101:217–25.

5. Lim HS, Schultz C, Dang J, et al. Time course ofinflammation, myocardial injury, and pro-thrombotic response after radiofrequency catheterablation for atrial fibrillation. Circ Arrhythm Elec-trophysiol 2014;7:83–9.

6. Keaney JF Jr., Larson MG, Vasan RS, et al.Obesity and systemic oxidative stress: clinicalcorrelates of oxidative stress in the FraminghamStudy. Arterioscler Thromb Vasc Biol 2003;23:434–9.

7. Watanabe H, Tanabe N, Watanabe T, et al.Metabolic syndrome and risk of development ofatrial fibrillation: the Niigata preventive medicinestudy. Circulation 2008;117:1255–60.

8. Savelieva I, Kakouros N, Kourliouros A,Camm AJ. Upstream therapies for management ofatrial fibrillation: review of clinical evidence andimplications for European Society of Cardiologyguidelines. Part II: secondary prevention. Europace2011;13:610–25.

9. Brame CJ, Salomon RG, Morrow JD, Roberts LJ.Identification of extremely reactive g-ketoalde-hydes (isolevuglandins) as products of the iso-prostane pathway and characterization of theirlysyl protein adducts. J Biol Chem 1999;274:13139–46.

10. Brame CJ, Boutaud O, Davies SS, et al. Modi-fication of proteins by isoketal-containing oxidizedphospholipids. J Biol Chem 2004;279:13447–51.

11. Kirabo A, Fontana V, de Faria AP, et al. DCisoketal-modified proteins activate T cells andpromote hypertension. J Clin Invest 2014;124:4642–56.

12. Prinsen JK, Kannankeril PJ, Yermaliskaya LV,et al. Highly-reactive isolevuglandins promoteatrial fibrillation susceptibility in obesity (abstr).Circulation 2016;134:A20445.

13. Huang J, Yancey PG, Zhang LM, et al. Scav-enging dicarbonyls with 50-O-pentyl-pyridoxamineimproves insulin sensitivity and reduces athero-sclerosis through modulating inflammatory Ly6Chimonocytosis and macrophage polarization. Bio-Rxiv 2019(529339[Preprint]). https://doi.org/10.1101/529339.

14. Tao H, Huang J, Yancey PG, et al. Scavengingof reactive dicarbonyls with 2-hydroxybenzyl-amine reduces atherosclerosis in hypercholester-olemic Ldlr�/� mice. BioRxiv 2019(524884[Preprint]). https://doi.org/10.1101/524884.

15. Zagol-Ikapitte I, Masterson TS, Amarnath V,et al. Prostaglandin H2-derived adducts of pro-teins correlate with Alzheimer’s disease severity.J Neurochem 2005;94:1140–5.

16. Willis MS, Patterson C. Proteotoxicity andcardiac dysfunction–Alzheimer’s disease of theheart? N Engl J Med 2013;368:455–64.

17. Guerrero-Munoz MJ, Castillo-Carranza DL,Kayed R. Therapeutic approaches against commonstructural features of toxic oligomers shared bymultiple amyloidogenic proteins. Biochem Phar-macol 2014;88:468–78.

18. Boutaud O, Ou JJ, Chaurand P, Caprioli RM,Montine TJ, Oates JA. Prostaglandin H2 (PGH2)accelerates formation of amyloid b1-42 oligomers.J Neurochem 2002;82:1003–6.

19. Chen K, Kazachkov M, Yu PH. Effect of alde-hydes derived from oxidative deamination andoxidative stress on beta-amyloid aggregation;pathological implications to Alzheimer’s disease.J Neural Transm (Vienna ) 2007;114:835–9.

20. Rocken C, Peters B, Juenemann G, et al. Atrialamyloidosis: an arrhythmogenic substrate forpersistent atrial fibrillation. Circulation 2002;106:2091–7.

21. Steiner I, Hajkova P. Patterns of isolated atrialamyloid: a study of 100 hearts on autopsy. Car-diovasc Pathol 2006;15:287–90.

22. Leone O, Boriani G, Chiappini B, et al. Amyloiddeposition as a cause of atrial remodelling inpersistent valvular atrial fibrillation. Eur Heart J2004;25:1237–41.

23. Sidorova TN, Mace LC, Wells KS, et al. Hy-pertension is associated with preamyloid oligo-mers in human atrium: a missing link in atrialpathophysiology? J Am Heart Assoc 2014;3:e001384.

24. Sidorova TN, Yermalitskaya LV, Mace LC, et al.Reactive g-ketoaldehydes promote protein mis-folding and preamyloid oligomer formation inrapidly-activated atrial cells. J Mol Cell Cardiol2015;79:295–302.

25. Harrison DG, Guzik TJ, Lob HE, et al. Inflam-mation, immunity, and hypertension. Hyperten-sion 2011;57:132–40.

26. Yang Z, Shen W, Rottman JN, Wikswo JP,Murray KT. Rapid stimulation causes electricalremodeling in cultured atrial myocytes. J Mol CellCardiol 2005;38:299–308.

27. Wu J, Thabet SR, Kirabo A, et al. Inflammationand mechanical stretch promote aortic stiffeningin hypertension through activation of p38mitogen-activated protein kinase. Circ Res 2014;114:616–25.

28. Davies SS, Talati M, Wang X, et al. Localizationof isoketal adducts in vivo using a single-chainantibody. Free Radic Biol Med 2004;36:1163–74.

29. Boutaud O, Brame CJ, Chaurand P, et al.Characterization of the lysyl adducts of prosta-glandin H-synthases that are derived fromoxygenation of arachidonic acid. Biochemistry2001;40:6948–55.

30. Kayed R, Head E, Thompson JL, et al. Commonstructure of soluble amyloid oligomers impliescommon mechanism of pathogenesis. Science2003;300:486–9.

31. Glabe CG, Kayed R. Common structure andtoxic function of amyloid oligomers implies acommon mechanism of pathogenesis. Neurology2006;66 Suppl 1:S74–8.

32. Sidorova TN, Mace LC, Wells KS, et al. Quan-titative imaging of preamyloid oligomers, a novelstructural abnormality, in human atrial samples.J Histochem Cytochem 2014;62:479–87.

33. Faggioni M, Savio-Galimberti E,Venkataraman R, et al. Suppression of sponta-neous Ca elevations prevents atrial fibrillation incalsequestrin 2-null hearts. Circ Arrhythm Elec-trophysiol 2014;7:313–20.

34. Amarnath V, Amarnath K, Amarnath K,Davies S, Roberts LJ. Pyridoxamine: an extremelypotent scavenger of 1,4-dicarbonyls. Chem ResToxicol 2004;17:410–5.

35. Davies SS, Brantley EJ, Voziyan PA, et al. Pyri-doxamine analogues scavenge lipid-derivedg-ketoaldehydes and protect against H2O2-medi-ated cytotoxicity. Biochemistry 2006;45:15756–67.

36. Johansson B, Wernstedt C, Westermark P.Atrial natriuretic peptide deposited as atrial amy-loid fibrils. Biochem Biophys Res Commun 1987;148:1087–92.

37. Salomon RG, Miller DB, Zagorski MG,Coughlin DJ. Solvent induced fragmentation ofprostaglandin endoperoxides. New aldehydeproducts from PGH2 and novel intramolecular 1,2-hydride shift during endoperoxide fragmentationin aqueous solution. J Am Chem Soc 1984;106:6049–60.

38. Salomon RG, Batyreva E, Kaur K, et al. Iso-levuglandin-protein adducts in humans: productsof free radical-induced lipid oxidation through theisoprostane pathway. Biochim Biophys Acta 2000;1485:225–35.

39. Fukuda K, Davies SS, Nakajima T, et al.Oxidative mediated lipid peroxidation re-capitulates proarrhythmic effects on cardiac so-dium channels. Circ Res 2005;97:1262–9.

40. Nakajima T, Davies SS, Matafonova E, et al.Selective g-ketoaldehyde scavengers protectNav1.5 from oxidant-induced inactivation. J MolCell Cardiol 2010;48:352–9.

41. Huang J, Yancey PG, Tao H, et al. Reactivedicarbonyl scavenging effectively reduces MPO-mediated oxidation of HDL and preserves HDLatheroprotective functions. BioRxiv 2019(528026[Preprint]). https://doi.org/10.1101/528026.

42. Stavrovskaya IG, Baranov SV, Guo X,Davies SS, Roberts LJ, Kristal BS. Reactive g-ketoaldehydes formed via the isoprostanepathway disrupt mitochondrial respiration and

























































https://doi.org/10.1101/529339

https://doi.org/10.1101/529339

https://doi.org/10.1101/524884





















































































































https://doi.org/10.1101/528026






615

calcium homeostasis. Free Radic Biol Med 2010;49:567–79.

43. Carrier EJ, Zagol-Ikapitte I, Amarnath V,Boutaud O, Oates JA. Levuglandin forms adductswith histone H4 in a cyclooxygenase-2-dependentmanner, altering its interaction with DNA.Biochemistry 2014;53:2436–41.

44. Davies SS, Amarnath V, Montine KS, et al.Effects of reactive g-ketoaldehydes formed by theisoprostane pathway (isoketals) and cyclo-oxygenase pathway (levuglandins) on proteasomefunction. FASEB J 2002;16:715–7.

45. Carrier EJ, Amarnath V, Oates JA, Boutaud O.Characterization of covalent adducts of nucleo-sides and DNA formed by reaction with levu-glandin. Biochemistry 2009;48:10775–81.

46. Sullivan CB, Matafonova E, Roberts LJ,Amarnath V, Davies SS. Isoketals form cytotoxicphosphatidylethanolamine adducts in cells. J LipidRes 2010;51:999–1009.

47. Roychowdhury S, McMullen MR, Pritchard MT,Li W, Salomon RG, Nagy LE. Formation ofg-ketoaldehyde-protein adducts during ethanol-induced liver injury in mice. Free Radic Biol Med2009;47:1526–38.

48. Mont S, Davies SS, Roberts Second LJ, et al.Accumulation of isolevuglandin-modified proteinin normal and fibrotic lung. Sci Rep 2016;6:24919.

49. Yan HP, Roberts LJ, Davies SS, et al. Iso-levuglandins as a gauge of lipid peroxidation inhuman tumors. Free Radic Biol Med 2017;106:62–8.

50. Davies SS, Bodine C, Matafonova E, et al.Treatment with a g-ketoaldehyde scavenger

prevents working memory deficits in hApoE4 mice.J Alzheimers Dis 2011;27:49–59.

51. De Jong AM, Maass AH, Oberdorf-Maass SU,Van Veldhuisen DJ, Van Gilst WH, Van Gelder IC.Mechanisms of atrial structural changes caused bystretch occurring before and during early atrialfibrillation. Cardiovasc Res 2011;89:754–65.

52. De Jong AM, Maass AH, Oberdorf-Maass SU,De Boer RA, Van Gilst WH, Van Gelder IC. Cyclicalstretch induces structural changes in atrial myo-cytes. J Cell Mol Med 2013;17:743–53.

53. Prosser BL, Ward CW, Lederer WJ. X-ROSsignaling: rapid mechano-chemo transduction inheart. Science 2011;333:1440–5.

54. Klein WL, Krafft GA, Finch CE. Targeting smallAb oligomers: the solution to an Alzheimer’s dis-ease conundrum? Trends Neurosci 2001;24:219–24.

55. Millucci L, Paccagnini E, Ghezzi L, et al.Different factors affecting human ANP amyloidaggregation and their implications in congestiveheart failure. PLoS One 2011;6:e21870.

56. Singh-Manoux A, Fayosse A, Sabia S, et al.Atrial fibrillation as a risk factor for cognitivedecline and dementia. Eur Heart J 2017;38:2612–8.

57. Morrow JD, Hill KE, Burk RF, Nammour TM,Badr KF, Roberts LJ. A series of prostaglandin F2-like compounds are produced in vivo in humans bya non-cyclooxygenase, free radical-catalyzedmechanism. Proc Natl Acad Sci U S A 1990;87:9383–7.

58. Roberts LJ, Oates JA, Linton MF, et al. Therelationship between dose of vitamin E and

suppression of oxidative stress in humans. FreeRadic Biol Med 2007;43:1388–93.

59. Darghosian L, Free M, Li J, et al. Effect ofomega-three polyunsaturated fatty acids oninflammation, oxidative stress, and recurrenceof atrial fibrillation. Am J Cardiol 2015;115:196–201.

60. Amarnath V, Amarnath K. Scavenging 4-oxo-2-nonenal. Chem Res Toxicol 2015;28:1888–90.

61. Amarnath V, Amarnath K, Avance J, Stec DF,Voziyan P. 5’-O-Alkylpyridoxamines: Lipophilicanalogues of pyridoxamine are potent scavengersof 1,2-dicarbonyls. Chem Res Toxicol 2015;28:1469–75.

62. Zagol-Ikapitte I, Amarnath V, Bala M,Roberts LJ, Oates JA, Boutaud O. Characterizationof scavengers of g-ketoaldehydes that do notinhibit prostaglandin biosynthesis. Chem ResToxicol 2010;23:240–50.

63. Nguyen TT, Caito SW, Zackert WE, et al.Scavengers of reactive g-ketoaldehydesextend Caenorhabditis elegans lifespan andhealthspan through protein-level interactionswith SIR-2.1 and ETS-7. Aging (Albany NY)2016;8:1759–80.

KEY WORDS atrial fibrillation, atrialnatriuretic peptide, B-type natriureticpeptide, hypertension, isolevuglandins,oxidative stress, preamyloid oligomers

APPENDIX For supplemental figures, pleasesee the online version of this paper.


































































































EDITORIAL COMMENT

Removing the Stress FromHypertension-Induced Atrial Fibrillation*
Jessica A. Hennessey, MD, PHD,a Steven O. Marx, MDa,b
T he incidence of atrial fibrillation, the mostcommon arrhythmia, is increasing as ourpopulation ages. Traditionally, systemic and

cardiac disorders have been regarded as the causesof atrial fibrillation, leading to electrical and struc-tural remodeling of the atria. In only 10% to 20% ofcases, atrial fibrillation is primarily an electrical disor-der and is not associated with underlying systemicand cardiac disorders such as hypertension. Yet,most of the pharmacological and ablative therapiescurrently used for atrial fibrillation, which are limitedby the suboptimal efficacy, toxicities, and relativelyhigh rates of recurrence, are targeting electrical activ-ity within the heart rather than the upstream under-lying pathophysiological processes. Despiteremarkable progress, mechanistic insights into thesignaling pathways that promote the substrate forthe pathogenesis and maintenance of atrial fibrilla-tion are still required.

Oxidative stress, an imbalance between the gen-eration and neutralization of reactive oxygen species(ROS), is believed to be one mechanism throughwhich atrial fibrillation is initiated and sustained (1).In the atrial appendages of patients with atrial fibril-lation, compared with those patients in normal sinus

ISSN 2452-302X




From the aDivision of Cardiology, Department of Medicine, Columbia

University, Vagelos College of Physicians and Surgeons, New York, New

York; and the bDepartment of Pharmacology and Molecular Signaling,

Columbia University, Vagelos College of Physicians and Surgeons, New

York, New York. Dr. Hennessey is supported by National Institues of

Health grant T32 HL007854. Dr. Marx is supported by National Institutes

of Health grant R01 HL140934.






rhythm, increased inflammation and oxidative injurywere noted. The change in redox state leads to geneprogramming changes and activation of matrix met-alloproteinases leading to activation of atrial myofi-broblasts. Atrial myofibroblasts secrete extracellularmatrix and, through the production of cytokines andchemokines, trigger an inflammatory response. In-flammatory cells produce ROS. Additionally, underhemodynamic stress, noninflammatory cells,including cardiomyocytes can produce ROS, typicallythrough nicotinamide adenine dinucleotide phos-phate oxidases or mitochondrial dysfunction.Increased ROS in cardiomyocytes alters electrophys-iological properties of atrial ion channels, promotingarrhythmias.

Attenuating oxidative stress either by blocking theformation of ROS with antioxidants or blocking pro-tein adduction with scavenger molecules has beentested as therapeutic approaches for atrial fibrillation.In general, ROS scavengers have not yielded protec-tion against atrial fibrillation. Multiple antioxidantshave been studied, including N0acetyl-cysteine,ascorbic acid (vitamin C), and tocopherol (vitamin E),as well as apocynin, which directly inhibits nicotin-amide adenine dinucleotide phosphate oxidase. In ameta-analysis, antioxidant treatment reduced therisk of post-operative atrial fibrillation, althoughsubanalysis showed that only N0acetyl-cysteine andascorbic acid had a beneficial effect (2). The trialswere small, and it is unclear whether it was a directeffect of the drug itself, especially because the oralavailability of these drugs, specifically ascorbic acid,is low compared with that of intravenous adminis-tration. Perhaps suboptimal therapeutic targets havebeen studied and treatment of atrial fibrillation withmore specific antioxidative therapies is possible.

When the redox state is unbalanced, poly-unsaturated fatty acids can become oxidized toform highly reactive g-ketoaldehydes (KAs), the most






J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Hennessey and MarxJ U N E 2 0 2 0 : 6 1 6 – 8 Removing the Stress From Hypertension-Induced AF

617

reactive of which are the isolevuglandins (IsoLGs).These highly reactive aldehydes adduct proteinsessentially immediately, leading to protein misfold-ing and crosslinking and the formation of pre-amyloid oligomers in the atria. In the absence ofnormal protein homeostatic mechanisms to refold orrecycle these misfolded proteins, proteotoxicity oc-curs, leading to downward spiral of cellular and car-diovascular health. During aging, proteotoxicitynaturally occurs, and many proteins that are respon-sible for maintaining proteostasis are implicated incellular aging. Both hypertension and atrial fibrilla-tion are diseases more frequently observed in olderindividuals.

In humans with atrial fibrillation, pre-amyloidoligomers (precursors to amyloid deposition formedfrom protein adduction) have been found in the atria.Previously, the Murray laboratory demonstrated thatrapid pacing of cultured atrial cells in vitro caused thegeneration of superoxide and cytosolic pre-amyloidoligomers, which were inhibited by salicylamine, asmall molecule scavenger of g-KAs, but not by theantioxidant curcumin, which is incapable of scav-enging g-KAs (3). They also found that formation ofpre-amyloid oligomers in atria was independentlyassociated with hypertension in patients undergoingelective cardiac surgery without a history of atrialarrhythmias, congestive heart failure, cardiomyopa-thy, or amyloidosis (4). These results offer a mecha-nistic link between a specific type of oxidative stressand atrial cell injury.

In this issue of JACC: Basic to TranslationalScience, Prinsen et al. (5) proposed the hypothesisthat IsoLGs may be the final common pathway foratrial remodeling and atrial fibrillation in the settingof hypertension. To test this hypothesis, the Murraylaboratory used a well-established model of hyper-tension in which mice were rendered hypertensiveby minipump infusion of angiotensin II. Similar tohumans with hypertension, the mice developed adiffuse accumulation of IsoLGs and pre-amyloidoligomers, occurring before the development ofsignificant atrial structural abnormalities, whichwere absent in control animals and in mice treatedwith a scavenger of IsoLGs, 2-hydroxybenzylamine,starting 3 days prior to angiotensin II infusion. Inhypertensive mice, the total amount of inducibleatrial fibrillation was increased in the hypertensivemice, but decreased within 2 weeks of stoppingangiotensin II, demonstrating reversibility of theprocess. Remarkably, cotreatment with 2-hydroxybenzylamine, compared with angiotensin IIalone, significantly reduced the burden of atrial

fibrillation. Using atrial HL-1 cells, they demonstratethat stretch caused a substantial increase in IsoLGadducts and pre-amyloid oligomers, which wasattenuated by 2-hydroxybenzylamine. Finally, theMurray group demonstrates that the atrial oligomersmay contain atrial/brain natriuretic peptides, whichcan induce cytotoxicity by reducing adenosinetriphosphate production in atrial HL-1 cells. Takentogether, the study fills in many gaps from theoriginal in vitro studies, strongly supporting theconcept that an increase in IsoLGs may initiatedysfunction within the atria leading to atrial fibril-lation. Furthermore, the investigators propose thatscavenging reactive downstream mediators ofoxidative stress, namely IsoLGs, rather than target-ing the generation of ROS may be a better thera-peutic target to prevent atrial fibrillation. Given thedearth of new therapies for atrial fibrillation, thefindings offer some hope that more precise target-ing of this abnormally regulated pathway may yielda new therapeutic for at least some patients withatrial fibrillation.

Naturally, as in any study of this design, there arelimitations. Is the relatively short burst of inducibleatrial fibrillation in mice equivalent to atrial fibrilla-tion in humans? Probably not, but it is a sign of anunderlying propensity to develop an atrialarrhythmia. At minimum, the presence of IsoLGs andpre-amyloid oligomers correlate with hypertension inhumans. In this 2-week study, the IsoLG scavengerwas administered 3 days prior to initiating the in-duction of hypertension, which appears to be an un-likely treatment scenario in humans. The more likelytimeline would be that a patient would have hyper-tension for years prior to developing structural ab-normalities including atrial enlargement and fibrosis.Whether scavenging IsoLGs at a later stage of thismultifactorial disease process will be beneficialcannot be assessed based on the current data. Yet, thefindings certainly support additional studies in otheranimal models of atrial fibrillation.

What might be the downstream targets that cause achange in the susceptibility to develop atrial fibrilla-tion in this hypertension animal model? Do IsoLGsaffect the abundance, folding, and/or function of ionchannels that modulate the cardiac action potential,and are these ion channels conserved in humans? It islikely that angiotensin II causes a multitude ofchanges in the atria, some dependent on and othersindependent of increased oxidative stress. It isconceivable that IsoLGs and pre-amyloid oligomersare necessary, but not sufficient, for the induction ofatrial fibrillation. Sorting through these important

Hennessey and Marx J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Removing the Stress From Hypertension-Induced AF J U N E 2 0 2 0 : 6 1 6 – 8

618

details may take substantial time and effort, but thepotential of developing a new pharmacologicalapproach for the prevention and potentially treat-ment of atrial fibrillation should stimulate substantialinterest from the electrophysiology community.

ADDRESS FOR CORRESPONDENCE: Dr. Steven O.Marx, Vagelos College of Physicians and Surgeons,622 West 168th Street, PH-3 Center, New York, NewYork 10032. E-mail: [email protected].

RE F E RENCE S

1. Van Wagoner DR. Oxidative stress and inflam-mation in atrial fibrillation: role in pathogenesisand potential as a therapeutic target. J CardiovascPharmacol 2008;52:306–13.

2. Violi F, Pastori D, Pignatelli P, Loffredo L. An-tioxidants for prevention of atrial fibrillation: apotentially useful future therapeutic approach? Areview of the literature and meta-analysis. Euro-pace 2014;16:1107–16.

3. Sidorova TN, Yermalitskaya LV, Mace LC, et al.Reactive g-ketoaldehydes promote protein mis-folding and preamyloid oligomer formation inrapidly-activated atrial cells. J Mol Cell Cardiol2015;79:295–302.

4. Sidorova TN, Mace LC, Wells KS, et al. Hyper-tension is associated with preamyloid oligomers inhuman atrium: a missing link in atrial pathophysi-ology? J Am Heart Assoc 2014;3:e001384.

5. Prinsen JK, Kannankeril PJ, Sidorova TN,et al. Highly reactive isolevuglandins promoteatrial fibrillation caused by hypertension. J AmColl Cardiol Basic Trans Science 2020;5:602–15.

KEY WORDS atrial fibrillation, hypertension,isolevuglandins, oxidation






























Impact of Coronary Atherosclerosis onBioresorbable Vascular ScaffoldResorption and Vessel Wall Integration
Yanping Cheng, MD,a Marco Ferrone, MD,a Qing Wang, MD, PHD,b Laura E.L. Perkins, DVM, PHD,b
Jennifer McGregor, BS,a Björn Redfors, MD, PHD,a Zhipeng Zhou, MA,a Richard Rapoza, PHD,b

Gerard B. Conditt, RCIS,a Aloke Finn, MD,c Renu Virmani, MD,c Grzegorz L. Kaluza, MD, PHD,a Juan F. Granada, MDa

ISSN 2452-302X

From the aCRF Skirball Cent

Institute, Inc., Gaithersburg

employee of Abbott Laborato

VISUAL ABSTRACT

er

,

r

Cheng, Y. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):619–29.


for Innovation, Orangeburg, New York; bAbbott Vascular, Santa Clara, California; and the cCVPath

Maryland. This study was sponsored by Abbott Vascular, Santa Clara, California. Dr. Wang is an

ies. Dr. Perkins is an employee of and shareholder in Abbott Laboratories. Dr. Rapoza is an employee





AND ACRONYMS

BVS = bioresorbable vascular

scaffold

EES = everolimus-eluting stent

FHS = familial

hypercholesterolemic swine

IVUS = intravascular

ultrasonography

OCT = optical coherence

tomography

of Abbott L

disclose.

The author

stitutions a

the JACC: B

Manuscript

Cheng et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

BRS in Experimental Porcine Atherosclerosis J U N E 2 0 2 0 : 6 1 9 – 2 9

620

HIGHLIGHTS

� The bioresorption process of the Absorb BVS has been directly characterized only in a normal swine model and

indirectly (by imaging surrogates) in clinical studies.

� Using multimodality imaging and histology, this study indicates that in a diseased animal model, the resorption

and integration of BVS into the arterial wall is not affected by the presence of untreated hyperlipidemia and

atherosclerosis progression.

� Imaging and histology suggest that BVS degradation progresses similarly in the presence of atherosclerosis

compared with earlier data from nonatherosclerotic arteries. However, BVS is not immune to the development of

neoatherosclerosis.

SUMMARY

The integration of the Absorb bioresorbable vascular scaffold (BVS) into the arterial wall has never been tested in an

in vivo model of atherosclerosis. This study aimed to compare the long-term (up to 4 years) vascular healing responses of

BVS to an everolimus-eluting metallic stent in the familial hypercholesterolemic swine model of atherosclerosis. The

multimodality imaging and histology approaches indicate that the resorption and vascular integration profile of BVS is

not affected by the presence of atherosclerosis. BVS demonstrated comparable long-term vascular healing and

anti-restenotic efficacy to everolimus-eluting metallic stent but resulted in lower late lumen loss at 4 years.

(J Am Coll Cardiol Basic Trans Science 2020;5:619–29) © 2020 The Authors. Published by Elsevier on behalf of the

American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

T he clinical efficacy of metallic drug-elutingstents has been well established (1,2),although long-term clinical events related

to in-stent restenosis and late stent thrombosiscontinue to accrue over time (3). The permanent cag-ing of the treated artery by a metallic stent has beenproposed to be an important contributor to late de-vice failure (4). It has been hoped that bioresorbablescaffolds will overcome this limitation by allowingthe return of vasomotion and elasticity in the treatedsegment and, eventually, also by late lumen gain andstabilization of the atherosclerotic process at thetreated site. Randomized controlled trials comparingthe Absorb bioresorbable vascular scaffold (BVS)with metallic everolimus-eluting stents (EES) showedsimilar clinical efficacy up to 1 year (5). However,meta-analysis of pooled individual patient data fromthe ABSORB trials at 2 and 3 years showed increasedrates of adverse events for BVS compared with EES(6,7). The reasons for this unexpected adverse long-term result are not fully understood. Although

aboratories. All other authors have reported that they have no re

s attest they are in compliance with human studies committees

nd Food and Drug Administration guidelines, including patient co

asic to Translational Science author instructions page.

received December 20, 2019; revised manuscript received April

experimental studies in healthy swine show that theAbsorb BVS is completely resorbed by approximately3 years (8), the integration of the BVS has neverbeen tested with in vivo models of atherosclerosis.In this study, we aimed to evaluate the long-term (4years) biological effect of BVS on vascular adaptationand strut resorption compared with EES in the famil-ial hypercholesterolemic swine (FHS) model of spon-taneous atherosclerosis using endovascular imagingtechniques and histology.

METHODS

EXPERIMENTAL DESIGN. The study was approved bythe Institutional Animal Care and Use Committee andconducted in accordance with the Animal Welfare Actand the Guide for the Care and Use of LaboratoryAnimals (National Research Council, National In-stitutes of Health publication no. 85-23, revised 1996)at U.S. Department of Agriculture–licensed, Associa-tion for the Assessment and Accreditation of

lationships relevant to the contents of this paper to

and animal welfare regulations of the authors’ in-

nsent where appropriate. For more information, visit

2, 2020, accepted April 6, 2020.



FIGURE 1 Study Design

BVS ¼ bioresorbable vascular scaffold; EES ¼ everolimus-eluting stent; FHS ¼ familial hypercholesterolemic swine; f/u ¼ follow-up; OS ¼overstretch.

TABLE 1 Quantitative Coronary Angiography Analysis

Time Point/Parameter BVS EES p Value

Baseline 21 12

% Pre-implantation DS 12.05 � 8.89 9.05 � 7.06 0.325

Device-to-artery ratio 1.18 � 0.12 1.26 � 0.08 0.036

Post-implantation MLD, mm 2.79 � 0.44 3.70 � 0.28 <0.001

1-yr follow-up 21 12

RVD, mm 3.30 � 0.64 3.95 � 0.51 0.005

MLD, mm 2.22 � 0.32 3.04 � 0.61 0.006

%DS 36.9 � 9.6 31.2 � 17.6 0.319

Late lumen loss, mm 1.02 � 0.32 1.15 � 0.67 0.517

2-yr follow-up 8 5

RVD, mm 3.09 � 0.31 3.63 � 0.38 0.017

MLD, mm 1.98 � 0.34 2.65 � 0.42 0.010

%DS 24.4 � 11.9 26.9 � 9.8 0.701


3-yr follow-up 8 4

RVD, mm 2.77 � 0.36 3.30 � 0.61 0.085

MLD, mm 1.80 � 0.46 2.27 � 0.93 0.253

%DS 36.9 � 13.7 37.5 � 25.0 0.956


4-yr follow-up 5 3

RVD, mm 3.55 � 0.82 3.85 � 0.70 0.612

MLD, mm 1.85 � 0.64 1.83 � 0.22 0.954

%DS 39.2 � 16.0 53.4 � 2.6 0.118


Values are n or mean � SD.

DS ¼ diameter stenosis; MLD¼minimal lumen diameter; RVD ¼ reference vessel diameter; other abbreviationsas in Figure 1.

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Cheng et al.J U N E 2 0 2 0 : 6 1 9 – 2 9 BRS in Experimental Porcine Atherosclerosis

621

Laboratory Animal Care International–accredited an-imal research facility.

The study flowchart is shown in Figure 1. Interimimaging assessments, including coronary angiog-raphy, intravascular ultrasonography (IVUS), andoptical coherence tomography (OCT) were performedat 1-year follow-up for all the animals (n ¼ 11). At 2(n ¼ 4), 3 (n ¼ 4), and 4 (n ¼ 3) years, animals werekilled for histology evaluation after imaging analysis.

THE FHS MODEL. Eleven FHS (10 � 0.07 months ofage; weight: 70.8 � 7.1 kg) were used in thisstudy. Mean cholesterol levels at baseline was 709 �90 mg/dl (range 612 to 956 mg/dl). The genotypic andclinical characteristics and response to stent implan-tation of this model have been published (9). Animalswere maintained on a low-grade high-cholesterol diet(0.6% cholesterol) for the first 21 weeks to acceleratelesion development. At device implantation, themean cholesterol level was 654 � 65 mg/dl, and theanimals were switched to a standard porcine diet, yetcholesterol level remained markedly elevated at theend of the study at 361 � 64 mg/dl.CORONARY INJURY AND DEVICE IMPLANTATION

PROCEDURE. Under general anesthesia, arterial ac-cess was obtained, and activated clotting time of$250 s was achieved. Proximal coronary segmentswere balloon-injured, targeting at least approximately40% overstretch (day 0) (10). Twenty weeks afterinitial injury, either BVS (n ¼ 21; 3.0 � 18 mm or 3.5 �18mm) or EES (n¼ 12; 3.5� 18mmor 4.0� 18mm)wereimplanted in the previously injured segments, target-ing a device-to-artery ratio of 1.1:1, under angiographicguidance. All animals received oral aspirin (81 mg) andclopidogrel (75 mg) once daily beginning 1 day beforeinjury and implantation procedures and continued for

30 days post-injury and throughout the first yearpost-implantation.

QUANTITATIVE CORONARY ANGIOGRAPHY ANALYSIS.

Quantitative coronary angiography analysis was per-formed with QAngio XA Software, version 7.1.14.0(Medis Medical Imaging System, Leiden, the

TABLE 2 IVUS Parameter Changes Between 1 and 2, 1 and 3, and 1 and 4 Years

BVS EES p Value

1- to 2-yr follow-up interval 8 5

Lumen area

Absolute change, mm2 0.83 � 1.11 –0.34 � 1.93 0.186

% change 19 � 21 –2 � 20 0.105

Vessel area


% change 6 � 17 –2 � 15 0.423

Total plaque area

Absolute change, mm2–0.19 � 1.34 0.06 � 4.09 0.897

% change –1 � 27 –1 � 38 0.973


Lumen area

Absolute change, mm2 1.42 � 2.07 0.45 � 2.14 0.466

% change 31 � 54 1 � 33 0.333

Vessel area


% change 54 � 56 15 � 17 0.100

Total plaque area


% change 61 � 94 24 � 51 0.484


Lumen area


% change 3 � 15 –25 � 9 0.039

Vessel area


% change 26 � 42 20 � 10 0.810

Total plaque area


% change 34 � 76 70 � 31 0.470

Values are n or mean � SD.

IVUS ¼ intravascular ultrasonography; other abbreviations as in Figure 1.



622

Netherlands). A contrast-filled catheter was used forcalibration; the minimum lumen diameter (MLD) wasobtained from a single view with the lowest measure-ment, and the reference vessel diameter was auto-matically calculated. Percent diameter stenosis (%DS)was calculated from the MLD and the referencevessel diameter.

GRAY-SCALE IVUS ANALYSIS. IVUS pullback imageswere generated (Atlantis SR Pro 40MHz catheters andiLab system; Boston Scientific, Natick, Massachusetts)and analyzed with commercially available software(echoPlaque, Indec Systems Inc., Santa Clara, Califor-nia). Luminal, device, and vessel areasweremeasured,and the neointimal and total plaque areas were calcu-lated. To normalize the lumen changes to the varia-tions in the reference vessel size, patency ratio wascalculated as: (follow-up lumen area in the implantedsegment)/(follow-up reference vessel lumen area), andits changes were also evaluated at different timepoints (11).

OCT IMAGE ACQUISITION AND ANALYSIS. OCT im-ages were obtained with the C7-XR OCT imagingsystem (LightLab Imaging, Inc., St. Jude Medical, St.Paul, Minnesota). Qualitative analyses were per-formed at 1-mm intervals with commercial software(ILUMIEN OPTIS; St. Jude Medical). Cross-sectionlumen, device areas, and percent area stenosis weremeasured. Alterations of the BVS struts in their op-tical appearance at follow-up were categorized into 4subgroups that have been applied in the preclinicalstudy (12): preserved box, open box, dissolved brightbox, and dissolved black box.

HISTOLOGICAL ANALYSIS. An independent pathol-ogy laboratory (CVPath Institute Inc., Gaithersburg,Maryland) conducted the histological analysis. Sec-tions were collected and stained with hematoxylinand eosin and Movat’s pentachrome as previouslydescribed (8). Vessel injury (range 0 to 3), neointimalinflammation (range 0 to 4), adventitial inflammation(range 0 to 3), and fibrin (range 0 to 3) were semi-quantitatively scored for each section as previouslydescribed (8). All sections were also evaluated for thepresence of neoatherosclerosis, which is defined asthe presence of foam cells, cholesterol clefts, and/orcalcification in the neointima (13), and assigned ascore from 0 to 3 (14).

STATISTICAL ANALYSIS. Statistical analysis wasconducted with SAS, version 9.4 (SAS Institute, Cary,North Carolina). Values are expressed as the mean �SD. Groups (BVS and EES) were compared at eachtime point using Student’s t-test. Select IVUS and OCTparameters were also compared between animal co-horts killed at 1 year and subsequent time points (1 to2, 1 to 3, and 1 to 4 years), also by means of Student’st-test. p < 0.05 was considered significant.

RESULTS

QUANTITATIVE CORONARY ANGIOGRAPHY ANALYSIS.

No complications occurred at the time of balloon injuryand device implantation. Table 1 summarizes thequantitative coronary angiography analysis data. Thedegree of balloon injury achieved in the 2 groups wascomparable (balloon-to-artery ratios: BVS, 1.55 � 0.15vs. EES, 1.43 � 0.16; p ¼ 0.051). At the time of deviceimplantation, the mean pre-implantation %DS andimplant-to-artery ratios were also comparable be-tween BVS and EES. Post-implantation MLDwas largerwith EES than BVS due to the fact that larger (>3.5 mm)BVS were not available at the time of the study(Table 1). Angiographic late lumen loss and %DS werenot significantly different between BVS and EES at 1, 2,and 3-year follow-up time points. At 4 years, despitethe difference in device size used in this study favoring

FIGURE 2 Key Quantitative IVUS Data From Day 0 to 4 Years of Follow-Up

Lumen, scaffold/stent, vessel (EEL), neointimal, and total plaque area (PA) changes by IVUS from baseline (post-implantation) to 4 years. The vessel area and total plaque area

were increased in both groups at 3 and 4 years. There was no difference in the percent PA stenosis between the 2 devices at all time points. The patency ratio of BVS-treated

vessels remained unchanged from 2 to 4 years but decreased in the EES-treated vessels at 4 years. NIA ¼ neointimal area; other abbreviations as in Figure 1.


623

EES, BVS showed lower late lumen loss than EES (BVS:1.14 � 0.37 vs. EES: 2.09 � 0.12; p ¼ 0.0056), whereasthe difference in %DS between the 2 devices was notsignificant at this time point (Table 1).

GRAY-SCALE IVUS. Twenty weeks after injury, acomparable degree of plaque burden (% area stenosis)before device implantation was found in both groups(BVS: 25.5 � 10.3% vs. EES: 22.1 � 4.6%; p ¼ 0.207). Anoverall summary of IVUS data from post-implantationto 4 years is presented in Table 2 and Figure 2. Theimplanted scaffoldswere no longer discernible by IVUSby 3 years; thus, only lumen and vessel areas weremeasured. The average vessel area for all devices ofeach type that were available for analysis at each timepoint appeared to be higher at later time points than atbaseline (BVS: baseline [n ¼ 21], 13.31 � 3.12 mm2 vs. 1year [n ¼ 21], 12.42 � 3.27 mm2 vs. 2 years [n ¼ 8], 12.96� 2.92 mm2 vs. 3 years [n ¼ 8], 16.34 � 5.73 mm2 vs. 4years [n ¼ 4], 18.97 � 5.89 mm2; EES: baseline [n ¼ 12],16.99� 2.54mm2 vs. 1 year [n¼ 12], 17.95� 3.31mm2 vs.2 years [n¼5], 19.14� 5.67mm2 vs. 3 years [n¼4], 18.23� 1.13 mm2 vs 4 years [n ¼ 3], 21.63 � 2.53 mm2;p ¼ 0.281). Average total plaque areas were also higherin both groups at later time points than at baselinebecause of atherosclerosis progression (BVS: baseline,

5.82 � 2.41 mm2 vs. 1 year, 7.71 � 2.50 mm2 vs. 2 years,7.05 � 2.25 mm2 vs. 3 years, 10.89 � 6.80 mm2 vs. 4years, 12.47� 5.95mm2; EES: baseline, 4.39� 1.54mm2

vs. 1 year, 9.35� 2.27 mm2 vs. 2 years 10.44 � 4.87 mm2

vs. 3 years 10.32 � 4.14 mm2 vs. 4 years 14.63 �1.70 mm2). There was no difference in the percentplaque area stenosis between the 2 devices at any of thestudied time points (Figure 2).

As summarized in Table 2, compared to 1-yearvalues, the lumen area remained stable in BVS at 4years but significantly decreased in EES. The patencyratio of BVS-treated vessels appeared stable between2 and 4 years but dropped in the EES-treated vesselsat 4 years (Figure 2).

OCT ANALYSIS. In all animals, the scaffold strutswere no longer discernible along the length of theimplanted segments by OCT at 4 years (Figure 3). Thestrut count and its optical appearance changed overtime: the recognizable struts were decreased overtime, with 95% of preserved box appearance at 1 year,17% at 2 years, and 7% at 3 years; at 4 years, the strutswere not discernible in any implanted segments(Figure 4). The overall OCT findings are summarized inFigure 5, top panels. In addition, relative OCT param-eter differences from years 1 to 2, 1 to 3, and 1 to 4 for

FIGURE 3 Representative Angiographic and OCT Images

Representative angiography and OCT images of BVS and EES at 2, 3, and 4 years. The scaffold struts were no longer discernible by OCT at 4

years. OCT ¼ optical coherence tomography; other abbreviations as in Figure 1.



624

both devices are summarized in Figure 5, bottompanels. Compared to 1 year, the scaffold area washigher by 15% at 2 years, and the mean lumen area waslarger by 25% between 1 and 2 years and 37% between 1and 3 years. Consistent with IVUS findings, the lumenarea was significantly decreased by 34% in the EESgroup between 1 and 4 years,whereas no further lumenarea changes were observed in the BVS group.HISTOLOGICAL ANALYSIS. In all evaluated arteries,all lumens were widely patent, and struts werecompletely incorporated within neointimal growth(Figure 6). BVS struts at 2 years were readily visible asunstained rhombi sequestered within the neointima,whereas at 3 years, struts stained blue-green (Movat’spentachrome) and were faintly eosinophilic (hema-toxylin and eosin) (Figures 6 and 7). Evidence of BVSdismantling, defined as stacking or misaligned struts,was observed infrequently at 2 and 3 years. At 4 years,BVS struts were difficult to discern and were mainlyrecognized as discrete foci of fibrous tissue, althoughblue-green–tinted irregular to rhomboid-shaped

regions were rarely observed. In contrast to 2 years,birefringence under polarized light consistent withresidual polymer was not observed at 4 years(Figure 6). For both BVS and EES, the neointimalgrowth was moderate to severe in thickness andcomposed of primarily proteoglycan with scatteredsmooth muscle cells from 2 to 4 years (Figure 6). Injuryscores were moderate to marked; were comparablebetween BVS and EES at 2, 3, and 4 years; and werelikely related to the advancement of atherosclerosisand inflammation for both implants. Neointimalinflammation was moderate to severe in both BVS andEES at 2, 3 and 4 years (Figure 8). Fibrin deposition andred blood cell extravasation were absent to minimal inboth groups at each time point. Evidence of neo-atherosclerosis was observed in both BVS and EES andincluded focal to focally extensive foam cells, calcifi-cation, cholesterol clefts, and necrotic cores starting at2 years (Figure 7); the mean scores are provided inFigure 8. Similarly, the naive segments proximal anddistal to both BVS and EES implants demonstrated

FIGURE 4 Changes in Optical Appearance of Struts Over Time (“Box Analysis”)

Alterations of the scaffold struts in their optical appearance were categorized into 4 subgroups: preserved box, open box, dissolved black box,

and dissolved bright box. The recognizable struts were decreased over time, with 95% of preserved box appearance at 1 year, 17% at 2 years,

and 7% at 3 years; at 4 years, the struts were not discernible in any implanted segments. OCT ¼ optical coherence tomography.


625

marked atherosclerotic lesion progression at each timepoint, with foam cell accumulations, calcification, andcholesterol clefts (Figure 7).

DISCUSSION

In this study, we aimed to study the patterns of poly-mer resorption, vascular healing, and neoathero-sclerosis development after BVS implantation in aswine model of spontaneous untreated atheroscle-rosis. At 4 years, the main findings of this study are asfollows.

� The net effect on angiographic restenosis (lumenloss) was more favorable in BVS compared to EES.

� Imaging demonstrated that atherosclerotic plaqueprogression occurred in comparable proportions inrelation to both devices; however, the expansiveremodeling occurring in BVS resulted in a netlumen gain despite disease progression.

� The long-term healing profiles of both devices werecomparable throughout all time points.

� Multi-modality imaging and histology confirmedthat full strut resorption occurs between years 3

and 4; a timeline similar to that reported in normalhealthy arteries.

� BVS are not immune to neoatherosclerosisdevelopment.

One of the most intriguing biological effects of BVSis the induction of expansive vascular remodelingwithrecovery of in-segment pulsatility observed from 12 to48 months in healthy porcine coronary arteries (15). Inthe present study using atherosclerotic vessels, thescaffold and vessel area remained unchanged duringthe first year after scaffold implantation. Both IVUSand OCT confirmed evidence of expansive remodelingoccurring after the second year. In contrast, progres-sive lumen loss continued in the EES group due to thepermanent mechanical caging effect in the arterialwall. Overall, vessel and plaque/media area measuredby IVUS increased over time in both devices. However,because of the expansive remodeling effect, the neteffect of plaque progression on lumen loss over time isless pronounced in BVS compared to EES. Of note, along-term study of BVS in normal swine has featuredmuch less pronounced lumen loss and stenosis be-tween 12 and 42 months post-implantation than

FIGURE 5 Lumen, Scaffold/Stent, and Neointimal Area Changes by OCT at 1 to 4 Years of Follow-Up

The scaffold and luminal area enlargements were observed in the BVS group between 1 and 2 years and 1 and 3 years. The lumen area remained unchanged in the BVS

group between 1 and 4 years but decreased by 34% in the EES group. Abbreviations as in Figure 1.



626

reported here (8), suggesting that untreated athero-sclerosis amplified late lumen loss and stenosis pro-gression in our study. These data are consistent withpreclinical observations in normal swine (15) and withclinical observations described in the ABSORB II ran-domized trial, in which BVS showed frequent dynamicvessel remodeling with larger increase in mean lumenand vessel area compared to the EES at 3 years (16).With regard to other histological findings, the notabledisparity between EES and BVS in inflammation/mac-rophages at 4 years presumably results from thecontinued presence of metallic struts with EEScompared to complete resorption and tissue replace-ment for BVS. Otsuka et al. (8) reported mild to mod-erate inflammation in BVS through 42 months in thehealthy swine model, and the observed inflammationprogressively decreased, with largely restoredmorphological appearance of BVS-implanted arteriesafter 18months, which is the same period duringwhichthe most rapid mass loss occurred. However, althoughEES has demonstrated preclinical and clinical safetywith low inflammation to 4 years and beyond, thesustained inflammation observed in the treated arte-rial segments at years 3 and 4 is likely a function of theinherent highly inflammatory state of the arteries inthe unmitigated atherosclerotic model used in ourstudy.

This study has confirmed that complete scaffoldresorption occurs between years 3 and 4 in the pres-ence of atherosclerosis. At 4 years, the struts are nolonger visible on OCT, and histological evaluationconfirmed the full integration of the device describedas connective tissue replacing the pre-existing poly-meric struts. These findings indicate that the scaffoldbioresorption timeline defined in normal animals(8,17) is not significantly altered by atherosclerosis.Unlike in healthy animals, severe atherosclerosis waspresent in all vessels evaluated, and significant dis-ease progression was evident up to 4 years. Theseverity of the disease is not unexpected in thismodel and is likely attributed to the 4-year chronicexposure to supra-physiological levels of cholesterolin the absence of any cholesterol-lowering therapy.Indeed, the high cholesterol levels (>350 mg/dl) likelyexacerbated the progression of atherosclerosis in thisstudy. This was readily evidenced by the severity ofatherosclerosis observed in the proximal and distalhost (nonimplanted) regions.

The speculation that a more prolonged polymerresorption process may have taken place in patientswith complex and biologically active atherosclerosisoriginated from the clinical observations that uncov-ered intraluminal dismantling has been associatedwith the late biomechanical failure of the device (18).

FIGURE 6 Histologic Appearance of BVS and EES Over Time

Representative low (original magnification: �2) and high (original magnification: �20) power images of BVS- and EES-implanted vessels at 2,

3, and 4 years. Sections shown are stained by (left) MP (original magnification: �2), and (right) HE (original magnification: �20). Abbre-

viations as in Figure 1.


627

However, late discontinuities are part of the normallate phase of the scaffold dismantling process whenadequately covered with tissue. In the present study,there were no malapposed struts observed at baselinein this study; atypical alignment or strut stacking wasseen infrequently in BVS sections at 2 and 3 years andmay represent localized structural discontinuities. Allthe struts were completely embedded and incorpo-rated with neointimal growth; hence, no adverseevents occurred during the follow-up period. There-fore, in view of our study demonstrating a similarpolymer resorption timeline in the presence ofatherosclerosis, it is plausible that the intraluminaldismantling observed clinically is not caused bydelayed polymer resorption but, rather, by long-termadverse biomechanical consequences of problematicdeployment.

Finally, the responses to BVS and EES in this studywere comparable, with similar neointimal characterand inflammation and injury scores. Additionally, ev-idence of neoatherosclerosis (necrotic core, foam cells,cholesterol clefts and calcification) was observed overtime to an equivalent degree in both devices. Humanautopsy studies suggest that neoatherosclerosis indrug-eluting stent shows unstable characteristics by 2

years after implantation (19), and neoatherosclerosisdevelopment may relate to dysfunctional vessel heal-ing, persistent inflammation, platelet activation, andadverse immunologic responses (20). In our study,neointimal inflammation was moderate to severe inboth devices tested, and neoatherosclerosis wasevident by the second year after device implantation.These data are important because they show that BVS,also being a drug-eluting device, is not immune toneoatherosclerosis formation. The INVEST (INdepen-dent OCT Registry on VEry Late Bioresorbable ScaffoldThrombosis) registry (21), which represents the largestcohort and first international multicenter consortiumto investigate the mechanisms underlying very latescaffold thrombosis, has found that neoatherosclerosiswas observed as a mechanism underlying late scaffoldthrombosis in 18.4% of lesions at 26.9 � 11.3 monthsafter BVS implantation.

Several limitations were present in the current study.First, EES was implanted in larger vessels due to stent sizeavailability, but this was offset by normalizing the resultsto the stent/scaffold size, and efforts were made to keepthe arterial injury consistent, irrespective of anatomiclocation and device type by rigorous control of stent-to-artery ratio. Second, the sample size is modest in

FIGURE 7 Representative Histological Images of Neoatherosclerosis

Representative histological images of neoatherosclerosis in BVS- and EES-implanted sections and atherosclerosis in proximal/distal reference segments (NS) at 2 to 4

years. Atherosclerosis with necrotic core (blue arrow), foamy macrophages (red arrows), cholesterol clefts (yellow arrow), and calcification (black arrow) in the device

implanted and proximal/distal naïve segments. Sections shown are stained by MP (original magnification: �2).

FIGURE 8 Summar

Histological analysis s



628

comparison to previous studies in normal animals, thestudy was not longitudinal, and the imaging observationscould not be serial; still, valuable insights were rendered.Finally, our diseased model still differs from atheroscle-rosis seen in human coronary arteries, and the absence ofcholesterol-lowering therapy routinely expected in

y of Key Semiquantitative Histology Parameters

howed comparable vascular healing responses and evidence of neoatherosclerosi

patients treated for obstructive coronary disease presum-ably augments this difference. However, although theresulting lesions may be biologically different, we haveshown that the diseased vascular background used in thisstudymayunveildifferences invascularhealingotherwisenot shown by healthy animal models.

s between BVS and EES from 2 to 4 years. Abbreviations as in Figure 1.

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: The biological

effect of BVS on atherosclerotic plaque progression and its

effect on neointimal formation and composition have not been

well studied. This long-term (up to 4 years) study using mul-

timodality imaging and histology shows that the implantation

of BVS in an untreated animal model of atherosclerosis

resulted in expansive vascular remodeling and slower late

lumen loss compared to EES but a similar extent of

neoatherosclerosis.

TRANSLATIONAL OUTLOOK: Imaging and histology

demonstrate that BVS degradation follows the same progression

in the presence of atherosclerosis as it does in normal arteries;

however, BVS is not immune to the development of

neoatherosclerosis.


629

CONCLUSIONS

In the presence of untreated hyperlipidemia andatherosclerosis, by using multimodality imaging andhistology, BVS demonstrates comparable long-termvascular healing and anti-restenosis efficacycompared with EES, with lower late lumen loss at 4years attributable to favorable remodeling notattainable in the EES-caged segments. In addition,the integration process is complete at 4 years basedon OCT and histological findings, indicating that thescaffold bioresorption/integration timeline defined innormal animals is not significantly altered byatherosclerotic disease.

ADDRESS FOR CORRESPONDENCE: Dr. Juan F.Granada, CRF Skirball Center for Innovation, Cardiovas-cular Research Foundation, 8 Corporate Drive, Orange-burg, New York 10962. E-mail: [email protected].

RE F E RENCE S

1. Palmerini T, Benedetto U, Biondi-Zoccai G, et al.Long-term safety of drug-eluting and bare-metalstents: evidence from a comprehensive networkmeta-analysis. J Am Coll Cardiol 2015;65:2496–07.

2. Mahmoud AN, Shah NH, Elgendy IY, et al.Safety and efficacy of second-generation drug-eluting stents compared with bare-metal stents:an updated meta-analysis and regression of 9randomized clinical trials. Clin Cardiol 2018;41:151–8.

3. Niccoli G, Sgueglia GA, Montone RA, Roberto M,Banning AP, Crea F. Evolving management of pa-tients treated by drug-eluting stent: prevention oflate events. Cardiovasc Revasc Med 2014;15:100–8.

4. Onuma Y, Muramatsu T, Kharlamov A,Serruys PW. Freeing the vessel from metallic cage:what can we achieve with bioresorbable vascularscaffolds? Cardiovasc Interv Ther 2012;27:141–54.

5. Stone GW, Gao R, Kimura T, et al. 1-year out-comes with the Absorb bioresorbable scaffold inpatients with coronary artery disease: a patient-level, pooled meta-analysis. Lancet 2016;387:1277–89.

6. Ali ZA, Serruys PW, Kimura T, et al. 2-yearoutcomes with the absorb bioresorbable scaffoldfor treatment of coronary artery disease: a sys-tematic review and meta-analysis of seven rand-omised trials with an individual patient datasubstudy. Lancet 2017;390:760–72.

7. Ali ZA, Gao R, Kimura T, et al. Three-year out-comes with the absorb bioresorbable scaffold:individual-patient-data meta-analysis from theABSORB randomized trials. Circulation 2018;30:464–79.

8. Otsuka F, Pacheco E, Perkins LE, et al. Long-term safety of an everolimus-eluting

bioresorbable vascular scaffold and the cobalt-chromium Xience V stent in a porcine coronaryartery model. Circ Cardiovasc Interv 2014;7:330–42.

9. Tellez A, Seifert PS, Donskoy E, et al. Experi-mental evaluation of efficacy and healing responseof everolimus-eluting stents in the familial hy-percholesterolemic swine model: a comparativestudy of bioabsorbable versus durable polymerstent platforms. Coron Artery Dis 2014;25:198–207.

10. Pedersen SF, Thrysøe SA, Paaske WP, et al.CMR assessment of endothelial damage andangiogenesis in porcine coronary arteries usinggadofosveset. J Cardiovasc Magn Reson 2011;13:10.

11. Vahl TP, Gasior P, Gongora CA, et al. Four-yearpolymer biocompatibility and vascular healingprofile of a novel ultrahigh molecular weightamorphous PLLA bioresorbable vascular scaffold:an OCT study in healthy porcine coronary arteries.EuroIntervention 2016;12:1510–8.

12. Onuma Y, Serruys PW, Perkins LE, et al. Intra-coronary optical coherence tomography and his-tology at 1 month and 2, 3, and 4 years afterimplantation of everolimus-eluting bioresorbablevascular scaffolds in a porcine coronary arterymodel: an attempt to decipher the human opticalcoherence tomography images in the ABSORBtrial. Circulation 2010;122:2288–300.

13. Otsuka F, Byrne RA, Yahagi K, et al. Neo-atherosclerosis: overview of histopathologicfindings and implications for intravascular im-aging assessment. Eur Heart J 2015;36:2147–59.

14. Zhao HQ, Nikanorov A, Virmani R, Schwartz LB.Inhibition of experimental neointimal hyperplasiaand neoatherosclerosis by local, stent-mediated

delivery of everolimus. J Vasc Surg 2012;56:1680–8.

15. Lane JP, Perkins LEL, Sheehy AJ, et al. Lumengain and restoration of pulsatility after implanta-tion of a bioresorbable vascular scaffold in porcinecoronary arteries. J Am Coll Cardiol Intv 2014;7:688–95.

16. Serruys PW, Katagiri Y, Sotomi Y, et al. Arterialremodeling after bioresorbable scaffolds andmetallic stents. J Am Coll Cardiol 2017;70:60–74.

17. Campos CM, Ishibashi Y, Eggermont J, et al.Echogenicity as a surrogate for bioresorbableeverolimus-eluting scaffold degradation: analysisat 1-, 3-, 6-, 12- 18, 24-, 30-, 36- and 42-monthfollow-up in a porcine model. Int J CardiovascImaging 2015;31:471–82.

18. Stone GW, Granada JF. Very late thrombosisafter bioresorbable scaffolds. Cause for concern?J Am Coll Cardiol 2015;66:1915–7.

19. Park SJ, Kang SJ, Virmani R, Nakano M, Ueda Y.In-stent neoatherosclerosis: a final commonpathway of late stent failure. J Am Coll Cardiol2012;59:2051–7.

20. Borovac JA, D’Amario D, Vergallo R, et al.Neoatherosclerosis after drug-eluting stent im-plantation: a novel clinical and therapeutic chal-lenge. Eur Heart J Cardiovasc Pharmacother 2019;5:105–16.

21. Yamaji K, Ueki Y, Souteyrand G, et al. Mecha-nisms of very late bioresorbable scaffold throm-bosis: the INVEST registry. J Am Coll Cardiol 2017;70:2330–44.

KEY WORDS bioresorbable vascularscaffolds, familial hypercholesterolemicswine, metallic drug-eluting stents,neoatherosclerosis















































































































ª 2 0 2 0 P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E A M E R I C A N



EDITORIAL COMMENT

Scaffold Resorption Process Is Not theAchilles’ Heel of the Absorb BVSBut What Then?*

Laura S.M. Kerkmeijer, MD, Joanna J. Wykrzykowska, MD, PHD

B ioresorbable scaffolds were developed torestore vasomotion; allow for adaptive shearstress, luminal enlargement, and vessel wall

remodeling; and most importantly, improve long-term clinical outcomes after complete resorptioncompared with metallic drug-eluting stents. However,the AIDA (Amsterdam Investigator-Initiated AbsorbStrategy All-Comers Trial) trial and a meta-analysisof the ABSORB trials demonstrated that the Absorbbioresorbable vascular scaffold (BVS) (AbbottVascular, Santa Clara, California) increases the risk oftarget vessel myocardial infarction and device throm-bosis compared with the Xience everolimus-elutingstent (EES) (Abbott Vascular) during the time of scaf-fold resorption (1,2).

The cause of this device failure is not yet fullyunderstood. It is thought that the underlying mech-anism of very late scaffold thrombosis is mostlyscaffold discontinuity (3), which suggests an unfa-vorable resorption-related process. Although theAbsorb BVS is no longer commercially available,detailed knowledge on the in vivo interaction betweenthe vessel wall and scaffold is of the utmost impor-tance for further development of this technology.

ISSN 2452-302X




From the Amsterdam UMC, Heart Center; Department of Clinical and

Experimental Cardiology, Amsterdam Cardiovascular Sciences, Amster-

dam, the Netherlands. Dr. Kerkmeijer has reported that she has no re-

lationships relevant to the contents of this paper to disclose. Dr.

Wykrzykowska has received institutional research grant support from

Abbott.






In this issue of JACC: Basic to Translational Science,Cheng et al. (4) evaluated the vascular healingresponse of the Absorb BVS up to 4 years comparedwith the XIENCE EES in 11 familial hypercholesterol-emic swine using serial multimodality imaging andhistology. At 3 years, still 7% of struts appeared aspreserved boxes by optical coherence tomography(OCT), and at 4 years, all struts were no longerdiscernible. The 4-year late lumen loss was less pro-nounced in the Absorb BVS than in the Xience EES(1.14 � 0.37 mm vs. 2.09 � 0.12 mm; p ¼ 0.006). Thisdifference in luminal dimension was confirmed byintravascular ultrasound assessment of the absolutechange of lumen area within 4 years (0.21 � 0.95 mm2

vs. –2.37 � 1.25 mm2; p ¼ 0.026). For both devices, theaverage total plaque area increases over time (BVS atbaseline: 5.82 � 2.41 mm2, at 4 years: 12.47 �5.95 mm2; EES at baseline: 4.39 � 1.54 mm2, at 4years: 14.63 � 1.70 mm2). In addition, histologyshowed evidence of neoatherosclerosis in both de-vices starting at 2 years. The vessel response to theAbsorb BVS and Xience EES was similar, with com-parable injury scores, neointimal inflammation, andgrowth.

So far, the bioresorption process of the Absorb BVShas been directly investigated only in normal swineand has been indirectly investigated in clinicalstudies. Therefore, the present study is unique andprovides knowledge of the resorption process in hy-percholesterolemia plaques. Cheng et al. (4) confirmthat the resorption process of the Absorb BVS takesbetween 3 and 4 years. Although the study did notinclude serial observations, the results suggest thatpositive vessel wall remodeling and late lumenenlargement is possible in Absorb BVS–treated le-sions. The study is of great importance because itshows that the long-term vascular healing in the






J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Kerkmeijer and WykrzykowskaJ U N E 2 0 2 0 : 6 3 0 – 1 Scaffold Resorption Process Is Not the Achilles’ Heel of the Absorb BVS

631

Absorb BVS is comparable to the Xience EES, and bothdevices are not immune to the progress ofatherosclerosis.

Cheng et al. (4) characterized for the first time thevascular healing process of BVS in swine with familiarhypercholesterolemia. However, the lesions createdin the pigs do not represent the majority of lesionsseen in clinical practices. The diameter stenosis atbaseline was low (12.05 � 8.89% vs. 9.05 � 7.06%),and it is most likely that these lesions consisted ofhomogenous soft plaques. All struts were found to beapposed at baseline and were completely embeddedand incorporated with neointima during follow-up.Hence, no adverse events occurred in these swinemodels.

Therefore, it seems that when the struts are wellembedded and encapsulated early, scaffold disman-tling is a benign process. In cases in which struts arenot covered by neointima, discontinuity allows pro-trusion of part of the scaffold into the lumen, bringsthrombogenic proteoglycan into contact with blood,and could potentially cause scaffold thrombosis. Inhomogenous soft lesions, as in the swine models, fullembedment and incorporation of struts is easilyaccomplished. However, when lesions becomemore complicated, such as those seen in thereal-life clinical population, apposition, embedment,and early encapsulation of struts become anissue (2,3).

Accurate deployment and embedment of the strutsof the Absorb BVS is more difficult to obtain in sig-nificant heterogenic lesions because of the low radialstrength. Underdeployment and malapposition aremajor risk factors of scaffold thrombosis (5). Therewere hopes that the introduction of a specific im-plantation technique, so-called PSP (pre-dilatation,sizing, and post-dilatation), would diminish the riskof scaffold thrombosis. However, the COMPARE-

ABSORB (ABSORB Bioresorbable Scaffold vs. XienceMetallic Stent for Prevention of Restenosis FollowingPercutaneous Coronary Intervention in Patients atHigh Risk of Restenosis) trial showed us that evenwhen we apply routinely pre- and post-dilatationaccording to the PSP criteria, the concern for scaf-fold thrombosis remains (Pieter Smits, unpublisheddata, September 25, 2018).

Moreover, even though we manage to obtain well-apposed struts by using intravascular imaging, intra-luminal scaffold dismantling is still seen duringfollow-up (6). Encapsulation of thick struts with amature neointimal layer appears to be an issue aswell. It is also plausible that even good apposition atbaseline would not prevent the occurrence of ac-quired malapposition, as large plaque burden con-tinues to exert an inner force on the progressivelyweaker resorbing device. Therefore, one can hy-pothesize that thinner struts and better mechanicalproperties are the key factors for improvement of thisdevice, rather than changing the nature of the scaf-fold dismantling process. To allow for assessment ofnovel devices with more favorable mechanical prop-erties, we will certainly need preclinical models thatbetter mimic the complexity of human patients.However, the ultimate test of the device will stillneed to take place in the context of a well-designed,randomized, real-clinical-practice trial, with long-term follow-up performed and completed beforecommercialization will take place.

ADDRESS FOR CORRESPONDENCE: Dr. Joanna J.Wykrzykowska, Department of Clinical and Experi-mental Cardiology, Amsterdam Cardiovascular Sci-ences, Medical Center, University of AmsterdamMedical Center, Meibergdreef 9, 1105 AZ Amsterdam,the Netherlands. E-mail: [email protected].

RE F E RENCE S

1. Kerkmeijer LSM, Tijssen RYG, Hofma SH, et al.Comparison of an everolimus-eluting bio-resorbable scaffold with an everolimus-elutingmetallic stent in routine PCI: three-year clinicaloutcomes from the AIDA trial. EuroIntervention2019;15:603–6.

2. Ali ZA, Gao R, Kimura T, et al. Three-year out-comes with the Absorb bioresorbable scaffold:individual-patient-data meta-analysis from theABSORB randomized trials. Circulation 2018;137:464–79.

3. Yamaji K, Ueki Y, Souteyrand G, et al. Mecha-nisms of very late bioresorbable scaffold throm-bosis: the INVEST registry. J Am Coll Cardiol 2017;70:2330–44.

4. Cheng Y, Ferrone M, Wang Q, et al. Impact ofcoronary atherosclerosis on bioresorbable vascularscaffold resorption and vessel wall integration. J AmColl Cardiol Basic Trans Science 2020;5:619–29.

5. Sotomi Y, Suwannasom P, Serruys PW,Onuma Y. Possible mechanical causes of scaffoldthrombosis: insights from case reports with

intracoronary imaging. EuroIntervention 2017;12:1747–56.6. Onuma Y, Honda Y, Asano T, et al. Randomizedcomparison between everolimus-eluting bio-resorbable scaffold and metallic stent: multi-modality imaging through 3 years. J Am CollCardiol Intv 2020;13:116–27.

KEY WORDS bioresorbable vascular scaffold,neoatherosclerosis, scaffold thrombosis, vascularhealing




































STATE-OF-THE-ART REVIEW

Mechanisms of Cardiovascular Benefits ofSodium Glucose Co-Transporter 2(SGLT2) InhibitorsA State-of-the-Art Review

Gary D. Lopaschuk, PHD,a Subodh Verma, MD, PHDb

HIGHLIGHTS

� Treatment with SGLT2 inhibitors reduces the incidence of cardiovascular death and heart failure hospitalization in patients

with and without diabetes.

� This review discusses the potential mechanisms by which SGLT2 inhibitors exert their beneficial effects, including

beneficial effects on cardiac energy metabolism, reducing inflammation, improving kidney function, and increasing

erythropoiesis.

� Future studies are required to clarify how SGLT2 inhibitors exert their impressive cardiovascular effects, which will allow

for a more specific targeting of heart failure therapy.

SUMMARY

ISS

Fro

ge

ha

sh

Am

No

Re

Re

Th

sti

the

Ma

Recent clinical trials have shown that sodium glucose co-transport 2 (SGLT2) inhibitors have dramatic beneficial car-

diovascular outcomes. These include a reduced incidence of cardiovascular death and heart failure hospitalization in

people with and without diabetes, and those with and without prevalent heart failure. The actual mechanism(s)

responsible for these beneficial effects are not completely clear. Several potential theses have been proposed to explain

the cardioprotective effects of SGLT2 inhibition, which include diuresis/natriuresis, blood pressure reduction, erythro-

poiesis, improved cardiac energy metabolism, inflammation reduction, inhibition of the sympathetic nervous system,

prevention of adverse cardiac remodeling, prevention of ischemia/reperfusion injury, inhibition of the Naþ/Hþ-exchanger,

inhibition of SGLT1, reduction in hyperuricemia, increasing autophagy and lysosomal degradation, decreasing epicardial

fat mass, increasing erythropoietin levels, increasing circulating pro-vascular progenitor cells, decreasing oxidative stress,

and improving vascular function. The strengths and weaknesses of these proposed mechanisms are reviewed in an effort

to try to synthesize and prioritize the mechanisms as they relate to clinical event reduction.

(J Am Coll Cardiol Basic Trans Science 2020;5:632–44) © 2020 The Authors. Published by Elsevier on behalf of

the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

N 2452-302X https://doi.org/10.1016/j.jacbts.2020.02.004

m the aCardiovascular Research Centre, University of Alberta, Edmonton, Alberta, Canada; and the bDivision of Cardiac Sur-

ry, Li Ka Shing Knowledge Institute of St. Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada. Dr. Lopaschuk

s received speaking honoraria from AstraZeneca; has served as a consultant for Boehringer Ingelheim and Servier; and is a

areholder in Metabolic Modulators Research Ltd. Dr. Verma has received research grants and/or speaking honoraria from

gen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, EOCI Pharmacomm Ltd., Janssen, Merck,

vartis, Novo Nordisk, Sanofi, Sun Pharmaceuticals, and the Toronto Knowledge Translation Working Group; is a Tier 1 Canada

search Chair in cardiovascular surgery; and is the president of the Canadian Medical and Surgical Knowledge Translation

search Group, a federally incorporated not-for-profit physician organization.

e authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ in-

tutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit

JACC: Basic to Translational Science author instructions page.

nuscript received January 22, 2020; revised manuscript received February 5, 2020, accepted February 5, 2020.






J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Lopaschuk and VermaJ U N E 2 0 2 0 : 6 3 2 – 4 4 Mechanisms of Cardiovascular Benefits of SGLT2 Inhibitors

633

AB BR EV I A T I O N S

AND ACRONYM S

EPO = erythropoietin

LV = left ventricular

NLRP3 = nucleotide-binding

oligomerization domain,

leucine-rich repeat, and pyrin

domain-containing 3

ROS = reactive oxygen species

SGLT = sodium glucose co-

transporter

SNS = sympathetic nervous

system

T2DM = type 2 diabetes

mellitus

O ne of the most serious health concerns inthe world is heart failure, with over 50million people worldwide being afflicted

with this disease (1). Despite the advances made inthe treatment of heart failure, patients diagnosedwith heart failure still have a very poor prognosisand quality of life. It also remains the most commonreason for hospitalization in older individuals (1). Pa-tients with type 2 diabetes mellitus (T2DM) are particu-larly susceptible to developing heart failure, which isthe major cause of morbidity and mortality in these indi-viduals (2–4). It is therefore critical to develop new ther-apies and approaches to prevent and treat heart failure.

A number of sodium glucose co-transporter 2(SGLT2) inhibitors have been developed to treat hy-perglycemia in T2DM, which act by inhibiting glucosereabsorption in the proximal tubule of the kidney (5).A number of large clinical trials have been conductedto evaluate the safety and efficacy of SGLT2 inhibitorsin patients with diabetes (with established vasculardisease, multiple cardiovascular risk factors, or renalinsufficiency) and in those with established heartfailure and reduced ejection fraction (with andwithout type 2 diabetes) (6–12). The remarkable re-sults from the EMPA-REG OUTCOME (EmpagliflozinCardiovascular Outcomes Event Trial in Type 2 Dia-betes Mellitus Patients—Removing Excess Glucose)demonstrated that T2DM patients who were at highrisk of cardiovascular disease had an early reductionin major cardiovascular and renal outcomes (8). Thisincluded a marked reduction in cardiovascular deathand hospitalization for heart failure in patientstreated with empagliflozin. Subsequent large trialswith other SGLT2 inhibitors, such as canagliflozin(CANVAS [Canagliflozin Cardiovascular AssessmentStudy] [9] and CREDENCE [Canagliflozin and RenalOutcomes in Type 2 Diabetes and Nephropathy] trials[10]) and dapagliflozin (DECLARE-TIMI 58 [Dapagli-flozin and Cardiovascular Outcomes in Type 2 Dia-betes] trial (11), confirmed these observations in abroader population of primary and secondary pre-vention patients (see Zelniker et al. [12] and Vermaet al. [13] for discussions of trials).

Whereas the aforementioned trials provided robustevidence to suggest that SGLT2 inhibitors can preventincident heart failure, 2 important questions remainedunanswered. First, could these therapies also be usedin the treatment of prevalent heart failure, and sec-ond, is the benefit seen in people without type 2 dia-betes? Importantly, in this regard, the recentlycompleted DAPA-HF (Dapagliflozin and Prevention ofAdverse Outcomes in Heart Failure) trial, whichenrolled 4,744 patients with heart failure and reducedejection fraction demonstrated a marked reduction on

worsening heart failure or cardiovasculardeath on top of excellent heart failurestandard-of-care therapy (6). Furthermore,this benefit was similar in those with andwithout type 2 diabetes and was consistentacross the spectrum of A1c evaluated eithercategorically or continuously.

POTENTIAL MECHANISMS BY WHICH

SGLT2 INHIBITION IS

CARDIOPROTECTIVE

A substantial number of theories have beenproposed to explain the beneficial effects ofSGLT2 inhibitors (14–18). These includebeneficial effects of SGLT2 inhibition on the
following: 1) blood pressure lowering; 2) increasingdiuresis/natriuresis; 3) improving cardiac energymetabolism; 4) preventing inflammation; 5) weightloss; 6) improving glucose control; 7) inhibiting thesympathetic nervous system; 8) preventing adversecardiac remodeling; 9) preventing ischemia/reperfu-sion injury; 10) inhibiting the cardiac Naþ/Hþ
exchanger; 11) inhibiting SGLT1; 12) reducing hyper-uricemia; 13) increasing autophagy and lysosomaldegradation; 14) decreasing epicardial fat mass; 15)increasing erythropoietin (EPO) levels; 16) increasingcirculating provascular progenitor cells; 17) decreasingoxidative stress; and 18) improving vascular function(Figure 1). In the sections that follow, we provide asummary of these proposed mechanisms and a syn-thesis of what mechanism(s) are likely most importantin terms of the observed clinical results observed.

BLOOD PRESSURE LOWERING. Hypertension is aprevalent modifiable risk factor for the developmentof heart failure. Because SGLT2 inhibitors lower bloodpressure (19), some of the beneficial effects of SGLT2inhibitors in the setting of heart failure have beensuggested to be related to this blood pressureimproved cardiac energetics with SGLT2 inhibitionlowering effect. Although the exact mechanism(s) forthe antihypertensive effects of SGLT2 inhibition arenot fully understood, they are probably mediated bythe osmotic and diuresis effects of SGLT2 inhibitors asa result of an inhibition of sodium reabsorption in theproximal tubules of the kidney. SGLT2 inhibition canresult in a 30% to 60% increase in urinary sodiumexcretion (20). The antihypertensive effect of SGLT2inhibition is greater than that of the thiazide diureticswhen used in combination with ß-blockers or calciumantagonists (21,22). By lowering blood pressure,SGLT2 inhibitors may lower cardiac afterload, withresultant improvement in ventricular arterialcoupling and cardiac efficiency. This would be

Lopaschuk and Verma J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Mechanisms of Cardiovascular Benefits of SGLT2 Inhibitors J U N E 2 0 2 0 : 6 3 2 – 4 4

634

expected to benefit the failing heart. However, theblood pressure–lowering effects of SGLT2 inhibitionare modest and are unlikely to completely explain thebeneficial cardiovascular and kidney effects of thesedrugs. In addition, blood pressure lowering would beanticipated to have a greater effect on stroke ratescompared with other cardiovascular outcomes, whichwas not observed in the EMPA-REG OUTCOME trial(8). Finally, in the DAPA-HF trial, the reductions inblood pressure were quite modest and unlikely to berelated to the large reduction in failure events (6).

DIURESIS AND NATRIURESIS. SGLT2 inhibitors havebeen shown to promote natriuresis and glucosuria,and it has been suggested that the resultant osmoticdiuresis may improve heart failure outcomes. In fact,mediation analyses from the EMPA-REG OUTCOMEtrial suggested that hemoconcentration (presumed tobe secondary to volume contraction) accounted forabout 50% of the cardiovascular benefit observed (8).It is difficult to explain the benefits of SGLT2 in-hibitors purely based on diuresis, because otherdiuretic strategies per se have not been associatedwith an improved event reduction in heart failurestudies. It has been suggested that SGLT2 inhibitorsmay differ somewhat from classical diuretics. In astudy comparing dapagliflozin and hydrochlorothia-zide, for example, a reduction in plasma volume andincrease in erythrocyte mass was observed withdapagliflozin but not with hydrochlorothiazide (23).When compared with a loop diuretic (bumetanide),dapagliflozin was associated with more reduction ininterstitial versus intravascular volume (24). It hastherefore been speculated that SGLT2 inhibition mayafford a differential effect in regulating interstitialfluid (vs. intravascular volume), which may limit thereflex neurohumoral stimulation that occurs inresponse to intravascular volume contraction withtraditional diuretics.

IMPROVED CARDIAC ENERGY METABOLISM. Dra-matic changes in energy metabolism occur in thefailing heart. As heart failure progresses, a continualdecline in mitochondrial oxidative metabolism oc-curs, and the heart becomes more reliant on glycol-ysis as a source of energy (25). Mitochondrial glucoseoxidation decreases in the failing heart (26–28),leading to a decrease in energy production and a fuel-starved heart (29). The uncoupling between glycolysisand glucose oxidation in the failing heart also leads toincreased proton production that leads to a decreasein cardiac efficiency (cardiac work / O2 consumed)(25–28). This decrease in cardiac efficiency is notconfined to patients with heart failure and reducedejection fraction, but also occurs in patients with

heart failure and preserved ejection fraction with leftventricular (LV) hypertrophy who also have a reducedLV mechanical efficiency (30).

It has been proposed that the beneficial effects ofSGLT2 inhibitors in heart failure can occur byimproving cardiac energetics and improving cardiacefficiency. SGLT2 inhibitors increase circulating ke-tone levels, secondary to mobilizing adipose tissuefatty acids, which are then used by the liver forketogenesis (31–33). Circulating ketone levels can in-crease following SGLT2 inhibitor treatment even inthe absence of diabetes (33). These ketones have beenproposed to improve cardiac energetics and cardiacefficiency by being a “thrifty” fuel for the heart(34,35). However, we have shown that ketones are nota more efficient source of fuel for the heart, but theyare an additional source of fuel for the failing heart(36,37). The failing heart is “energy starved,” dueprimarily to a decrease in mitochondrial oxidativemetabolism (25,29). Ketone oxidation is increased inthe failing heart, which has been proposed to be anadaptive metabolic process (37–40). Increasingplasma ketone levels in the blood due to SGLT2 in-hibition does increase cardiac ketone oxidation ratesand therefore improves energy supply to the “starv-ing” failing heart (41). In diabetic cardiomyopathicmice, empagliflozin-induced increases in cardiac ke-tone oxidation provide an additional source of fuelfor the heart, which is associated with an improve-ment in cardiac performance (Figure 2) (41). In sup-port of this, Santos-Gallego et al. (42) have shown thatempagliflozin can decrease adverse remodeling andheart failure in a porcine model of heart failure, byimproving cardiac energetics. It has also been shownthat SGLT2 inhibitors can improve mitochondrialrespiratory function in diabetic rats (43), which alsomay contribute to improving energy production inthe heart. These changes in mitochondrial respirationhave been proposed to be partially mediated due tofavorable alterations in energy supply to the heart.Ketone infusions into patients with heart failure isalso associated with an improvement in contractileperformance (44). Of interest, this ketone-inducedimprovement in contractile performance is also notassociated with an increase in cardiac efficiency. Thismakes sense, as the oxidation of ketones, comparedwith the oxidation of glucose, are not a more efficientsource of energy. As a result, some of the beneficialeffects of SGLT2 inhibition in heart failure may occursecondary to increasing fuel supply to the failingheart, as opposed to supplying the heart with a moreefficient source of fuel. Additionally, the increase incardiac ketone oxidation in the failing heart is notassociated with a decrease in either glucose or fatty

FIGURE 1 Possible Mechanisms by Which SGLT2 Inhibitors Decrease the Severity of Heart Failure

Improved cardiac energetics with SGLT2 inhibition. NLRP3 ¼ nucleotide-binding oligomerization domain, leucine-rich repeat, and pyrin

domain-containing 3; SGLT2 ¼ sodium glucose co-transporter 2.


635

acid oxidation, resulting in an overall increase inadenosine triphosphate production (37,41). There-fore, whereas SGLT2 inhibitors may not increasecardiac efficiency in the failing heart, they may sup-ply the heart with an extra source of fuel, which maybe particularly beneficial for the energy-compromisedfailing heart.

REDUCTION IN INFLAMMATION. Inflammation is animportant contributor to heart failure severity, and

proinflammatory biomarkers are elevated in patientswith heart failure and correlate with the severity ofthe disease (45,46). This association between heartfailure and markers of inflammation is evident inpatients both with reduced and with preserved ejec-tion fraction (47). Inflammatory cytokines notonly cause endothelial dysfunction, they also canincrease extracellular matrix turnover and increasefibrosis. The SGLT2 inhibitors, empagliflozin (48),



636

canagliflozin (49), and dapagliflozin (50) have beenshown to attenuate or ameliorate the inflammatoryprofile in patients with diabetes. This decrease inanti-inflammatory properties of SGLT2 inhibitors hasthe potential to decrease molecular processes relatedto inflammation, such as extracellular matrix turn-over and fibrosis (51). In support of this, dapagliflozinhas been shown to have marked antifibrotic effects inthe post-infarct rat heart by suppressing collagensynthesis (51). Empagliflozin also significantly atten-uates cell-mediated extracellular matrix collagenremodeling (52).

How SGLT2 inhibitors modify the inflammatoryprocess is not exactly clear. Decreasing glucose levelswith SGLT2 inhibitors may decrease macrophage in-flammatory response, as macrophages preferentiallyutilize glucose from glycolysis as an energy source(53). Alternatively, SGLT2 inhibitors may directlytarget the inflammatory pathways independent ofglucose lowering per se. The nucleotide-bindingdomain-like receptor protein (specifically,nucleotide-binding oligomerization domain, leucine-rich repeat, and pyrin domain-containing 3 [NLRP3])inflammasome plays an important role in mediatinginflammation. The NLRP3 inflammasome also con-tributes to chronic inflammation in heart failure,thereby increasing heart failure severity (54). Recentevidence in the kidney (55), liver (56), macrophages(57), vasculature (58), and heart (59,60) suggests thatempagliflozin can inhibit the NLRP3 inflammasomeand that this can occur independent of glucoselowering per se. Whether this is a direct or indirecteffect of SGLT2 inhibitors on NLRP3 is not clear. Theketone ß-hydroxybutyrate is an effective blocker ofthe NLRP3 inflammasome-mediated inflammatoryprocess (61). Because SGLT2 inhibitors increasecirculating ß-hydroxybutyrate levels, it is possiblethat some of the beneficial effects of SLT2 inhibitioncould occur secondary to ketone inhibition of theNLRP3 inflammasome.WEIGHT LOSS. The excretion of glucose by the kid-ney with SGLT2 inhibitor treatment results in a loss ofcalories. As a result of this, there is a subsequentdecrease in body weight as fatty acids are mobilizedfrom adipose tissue stores. Clinical studies haveconsistently shown body weight reduction in patientstreated with SGLT2 inhibitors (62). Whereas thisweight loss may contribute to the beneficial effects ofSGLT2 inhibition, other mechanisms must also beinvolved, as weight loss strategies have been muchless effective in decreasing heart failure severitycompared with SGLT2 inhibition, therefore this isunlikely to be an important mechanism of the heartfailure benefit observed. For example, in the

DAPA-HF trial, whereas there was a modest numericreduction in weight observed, the magnitude ofwhich appeared to be greater in those with diabetes(63). Furthermore, the effects of SGLT2 inhibition onweight loss are moderate and diminish with time, duein part to counter-regulatory mechanisms (such asincreased energy intake) being activated to attempt tomaintain weight (64).

IMPROVING GLUCOSE CONTROL. Whereas SGLT2inhibitors are effective glucose-lowering agents, theefficacy on heart failure is unlikely related to im-provements in glucose lowering per se. Hyperglyce-mia itself has been shown to be a weak risk factor forcardiovascular disease (65). In addition, the rapidefficacy noted (within days of treatment initiation) isdifficult to reconcile with a glucose-lowering effect.Furthermore, the differences in glycemic control inthe cardiovascular outcome trials were small (in aneffort to fulfill the glycemic equipoise principle of thetrials), and post hoc analyses from trials suggestedthat baseline A1c or changes in A1c were not associatedwith any treatment modification with SGLT2 in-hibitors (65). Definitive proof of this concept emergedfrom the DAPA-HF trial wherein the efficacy ofdapagliflozin was entirely consistent in those withand without diabetes. Even in those without diabetes,efficacy was similar in those with pre-diabetes orimpaired glucose tolerance compared with in thosewho were truly euglycemic. When studied continu-ously, using fractional polynomial analyses, baselineA1c was unrelated to the efficacy of dapagliflozin toreduce heart failure and mortality in DAPA-HF. Inaddition, in experimental models of heart failure, thebenefit of SGLT2 inhibition has been observed inde-pendent of diabetes or hyperglycemia (60,66,67).

INHIBITING THE SNS. The observation that SGLT2inhibitors reduce blood pressure in the absence ofincreasing heart rate suggests, indirectly, that theseagents may be associated with a reduction in sym-pathetic nervous system (SNS) activity. In fact,accumulating data indicate that SGLT2 inhibition maylead to a reduction in sympathetic nerve activity,inhibit norepinephrine turnover in brown adiposetissue, and reduce tyrosine hydroxylase production(68–72). These sympathoinhibitory effects appear tobe observed in both animal models of diabetes as wellas those with obesity (without diabetes) (68–72). Ithas also been postulated that the effects of SGLT2inhibition to reduce SNS activity may be secondary toa reduction in renal stress with resultant inhibition ofrenal afferent sympathetic activation (73).

PREVENTING ADVERSE CARDIAC REMODELING.

Adverse cardiac remodeling is an important

FIGURE 2 SGLT2 Inhibition Increases Cardiac Energy Production

SGLT2 inhibitors can increase cardiac energy metabolism. Reproduced with permission from Verma et al. (41). ATP ¼ adenosine triphosphate;

SGLT2 ¼ sodium glucose co-transporter 2.


637

contributor to heart failure severity. This includes thedevelopment of cardiac hypertrophy, fibrosis,inflammation, and cardiomyocyte cell death. Severalexperimental and human studies have demonstratedbeneficial effects of SGLT2 inhibition on cardiacremodeling (23,67,74–77). In a randomized trial, peo-ple with type 2 diabetes and a history of coronary

artery disease were treated with empagliflozin versusplacebo for 6 months (76). The primary outcome—change in LV mass index (evaluated by cardiac mag-netic resonance imaging) was significantly lower inthose treated with empagliflozin versus in those whoreceived placebo. Although these data do not provideinsight about the exact mechanism of action, they do

FIGURE 3 Multiple Sites for the Beneficial Effects of SGLT2 Inhibition

Proposed renal mechanisms for increased erythropoietin (EPO) with sodium glucose co-transporter 2 (SGLT2) inhibitors. Reproduced with

permission from Mazer et al. (95).



638

suggest that even short-term exposure to SGLT2 in-hibitors can promote cardiac reverse remodeling(Figure 3). Inhibition of the mammalian target ofrapamycin pathway, a major pathway involved incardiac hypertrophy, by SGLT2 inhibition may also beinvolved (76). Prevention of adverse remodeling withSGLT2 inhibition is also associated with a decreasedfibrosis (52,78), which may in part be mediated by theanti-inflammatory actions of SGLT2 inhibition (seereduction in inflammation discussion). As a result,SGLT2 inhibition may reverse the cardiac remodelingseen in heart failure, thereby reducing LV wall stressand improving cardiac function.

PREVENTING ISCHEMIA/REPERFUSION INJURY.

Ischemia/reperfusion injury can promote car-diomyocyte cell death and heart failure. Recentexperimental evidence suggests that SGLT2

inhibition has a cardioprotective effect againstischemia/reperfusion injury in both diabetic andnondiabetic rats (79). This beneficial effect of SGLT2inhibition on ischemia/reperfusion injury is associ-ated with a decrease in calmodulin kinase II activity,resulting in improved sarcoplasmic reticulum Ca2þ

flux and increased contractility. However, whetherthis effect occurs in humans remains unclear.

INHIBITING THE CARDIAC NAD/HD EXCHANGER.

The Naþ/Hþ exchanger is increased in the failing heartand can lead to Naþ and Ca2þ overload in the heart(reviewed by Wakabayashi et al. [80]). Inhibition ofthe Naþ/Hþ exchanger has also been demonstrated toprotect the heart in several experimental models ofheart failure (see Wakabayashi et al. [80] for review).Inhibition of the Naþ/Hþ exchanger has been pro-posed to explain the beneficial effects of SGLT2


639

inhibitors in heart failure (81–85). The cardiac Naþ/Hþ

exchanger is inhibited by SGLT2 inhibitors, which canlower myocardial Naþ levels (81). However, it is notclear whether direct inhibition of the cardiac Naþ/Hþ

exchanger occurs at clinically relevant concentrationsof SGLT2 inhibitors. In addition, the development ofNaþ/Hþ exchanger inhibitors has not been shown tobe beneficial in the clinical setting of heart failure(85). Therefore, it is not clear whether some of thebeneficial effects of SGLT2 inhibition in the setting ofheart failure may occur secondary to direct inhibitionof the cardiac Naþ/Hþ exchanger.

INHIBITING SGLT1. Although the heart does not ex-press SGLT2, it does express some SGLT1. Inhibitionof SGLT1 in the heart has the potential to decreasemyocardial Naþ and glucose uptake and decreasehyperglycemia-induced generation of reactive oxy-gen species (ROS) (86). However, whereas some of theSGLT2 inhibitors will decrease SGLT1, such as cana-gliflozin, the concentration necessary to inhibit SGLT1is much higher than the plasma concentrations seenwith the clinical use of SGLT2 inhibitors. In addition,the use of dual inhibitors of SGLT2 and SGLT1 canexacerbate cardiac dysfunction post-myocardialinfarction in rats (87). As a result, it is unlikelythat the beneficial effects of SGLT2 inhibitorscan be explained by any secondary effect onSGLT1 inhibition.

REDUCING HYPERURICEMIA. SGLT2 inhibitorsdecrease plasma uric acid, which adversely affects theprognosis of heart failure (88). Small reductions inuric acid levels have been seen with SGLT2 inhibitortreatment (89). This may be attributed to theincreased glycosuria in the proximal tubules due toSGLT2 inhibition, which stimulates uric acid secretion(90). However, whether a reduction in hyperuricemiaby SGLT2 inhibition is a marker or plays a causal roleremains unknown.

INCREASING AUTOPHAGY AND LYSOSOMAL

DEGRADATION. Cardiac autophagy and lysosomaldegradation can be impaired in diabetes and heartfailure (77,91). By driving catabolic rates due to con-stant glycosuria, it has been proposed that SGLT2inhibition can promote autophagy and lysosomaldegradation, thereby improving mitochondrialmorphology and function (77). An SGLT2-mediatedinhibition of mammalian target of rapamycin mayalso stimulate autophagy and lysosomal degradation,leading to the enhanced degradation of dysfunctionalorganelles. It therefore cannot be ruled out that someof the benefit of SGLT2 inhibition in heart failuremay be secondary to their effects on stimu-lating autophagy.

DECREASING EPICARDIAL FAT MASS. High epicar-dial adipose tissue attenuation on computed tomog-raphy is associated with an increased risk ofcardiovascular events (92). Epicardial adipose tissuecan produce a number of bioactive molecules that cannegatively affect heart function and contribute tocoronary artery disease. In addition, SGLT2 inhibitorsreduce the accumulation and inflammation of peri-vascular adipose tissue, thus minimizing the secre-tion of leptin and its paracrine actions on the heart topromote fibrosis (84). In patients with diabetes whoalso have coronary artery disease, SGLT2 inhibitionreduces epicardial adipose tissue mass, as well as thelevels of bioactive molecules such as tumor necrosisfactor-a and plasminogen activator inhibitor-1 (93).This may contribute to a decreased adverse remod-eling of the failing heart.

INCREASING EPO LEVELS. The fact that SGLT2 in-hibitors raise the hematocrit (94), even in thosewithout diabetes (as seen in DAPA-HF), has led to thesuggestion that these agents may promote erythro-poiesis via enhanced EPO secretion by the kidney(95). Such an increase in EPO may serve to favorablyinfluence cardiomyocyte mitochondrial function,angiogenesis, cell proliferation, and inflammation, inaddition to directly enhancing myocardial tissue ox-ygen delivery (95). Mazer et al. (95) recently evalu-ated this in the EMPA-Heart CardioLink-6randomized clinical trial and demonstrated that EPOlevels increased significantly after 1 month of empa-gliflozin treatment in people with type 2 diabetes andcoronary artery disease accompanied by an increasein hematocrit, reduced ferritin and red blood cellhemoglobin concentration (Figure 3). Whether EPOlevels go up in people without type 2 diabetes re-mains unknown.

INCREASING CIRCULATING PROVASCULAR PROGENITOR

CELLS. Preliminary evidence in humans points to-ward an effect of SGLT2 inhibitors on the restorationof provascular progenitor cells in people with type 2diabetes. In one such completed study, Hess et al.(96) observed that empagliflozin treatment wasassociated with a reduction in the number proin-flammatory M1 cells while increasing the number ofM2 polarized, anti-inflammatory cells. By using theAldeflour assay (STEMCELL Technologies, Cam-bridge, Massachusetts), the investigators found thatSGLT2 inhibition reduced systemic granulocyteburden in individuals with T2DM, increased circu-lating ALDHhiSSCmid monocytes and induced a tran-sition from M1 to M2 polarization—all of which isconsistent with maturation of collateral vessels dur-ing arteriogenesis. The investigators concluded that

CENTRAL ILLUSTRATION Potential Direct Myocardial and Indirect � Systemic Effects of SGLT2i

Lopaschuk, G.D. et al. J Am Coll Cardiol Basic Trans Science. 2020;5(6):632–44.

CAMKII ¼ calmodulin-dependent protein kinase II; EPO ¼ erythropoietin; NHE ¼ sodium/hydrogen exchanger; NLRP3 ¼ nucleotide-binding

oligomerization domain, leucine-rich repeat, and pyrin domain-containing 3; SGLT2i ¼ sodium glucose co-transporter 1(2) inhibitor; SNS ¼sympathetic nervous system.



640

SGLT2 inhibition may uniquely serve as a strategy topromote the recovery of circulating provascular cellsin T2DM. Whether this occurs in people withoutT2DM is unknown.

DECREASING OXIDATIVE STRESS. Excessive cardiacmitochondrial ROS production is an importantcontributor to contractile dysfunction in human heartfailure and in animal models of heart failure (for re-view see Zhou and Tian [97]). During the develop-ment of heart failure, increased oxidative stress canresult in mitochondrial dysfunction. This increase inROS production may occur due to a stimulation of

mitochondrial respiration for adenosine triphosphateproduction and an increased electron transport chainactivity. In diabetic mice, increasing glycemic controlwith SGLT2 inhibition can decrease myocardial ROSproduction and cardiac fibrosis (97). In human coro-nary arterial endothelial cells, SGLT2 inhibition canalso decrease ROS generation (98). How SGLT2 inhi-bition decreases ROS production is not clear,although this may occur secondary to favorable ef-fects on the inflammatory process (see reduction ininflammation discussion), cardiac mitochondrialoxidative metabolism (see improved cardiac energy


641

metabolism discussion), or decreasing the potentialfor cardiac glucotoxicity (which can promote ROSproduction). Furthermore, there is a paucity of dataon SGLT2 inhibition on ROS production in theabsence of type 2 diabetes.

IMPROVING VASCULAR FUNCTION. Vascular smoothmuscle and endothelial dysfunction contributes tothe pathophysiology of heart failure (99,100). Its ex-istence in patients with heart failure increasesmorbidity and mortality. SGLT2 inhibition has beenshown to improve vascular function by attenuatingendothelial cell activation, inducing direct vaso-relaxation, reducing endothelial cell dysfunction andmolecular changes associated with early atherogen-esis decreasing arterial wall stiffness, and decreasingvascular resistance (101–104). A number of underlyingmechanisms for the beneficial actions of SGLT2 inhi-bition may occur, including beneficial effects ofSGLT2 inhibition on inhibiting the inflammatorypathways and improving mitochondrial function(101). Induction of vasodilation via activation of pro-tein kinase G and voltage-gated potassium channelsby SGLT2 inhibitors has also been proposed (105). Thedirect effects of SGLT2 inhibition on the vascular,combined with the natriuresis effects of SGLT2 inhi-bition, may contribute to the desirable hemodynamiceffects seen with SGLT2 inhibition.

FROM A LIST TO SYNTHESIS:

WHAT MECHANISMS MOST LIKELY

EXPLAIN THE BENEFITS OBSERVED?

A number of clinical trials have demonstrated thatSGLT2 inhibitors have impressive beneficial cardio-vascular effects in both patients with diabetes andpatients who do not have diabetes, but do have heartfailure. As a result, SGLT2 inhibitors are a new weaponthat can be added to the arsenal of weapons used totreat heart failure. However, it remains unclear exactlyhow SGLT2 inhibitors produce their impressive clinicalbenefits in patients with heart failure (CentralIllustration). It is clear that its classical actions oflowering blood glucose cannot fully explain thesebenefits. The early onset of the beneficial effects ofSGLT2 inhibitors in clinical trials suggests that thesebenefits are not occurring as a result of slowing theatherosclerotic process. Whereas the published re-ports are replete with numerous proposed mecha-nisms linking SGLT2 inhibition to cardiovascularprotection, how do we synthesize and prioritize these

to help explain the observed clinical benefits? Taking abedside-to-bench approach, the clinical data wouldsuggest that the mechanism(s) involved must accountfor the following key elements: 1) efficacy in thetreatment and prevention of heart failure; 2) efficacyon top of excellent background therapy, includingneprilysin inhibition; 3) rapid onset of the benefit; 4)efficacy independent of glycemic status; and 5) asso-ciation with renal protection. We opine that the renaleffects of SGLT2 inhibitors are an important mecha-nism of action and that some of the cardiovascularbenefits are secondary to this. According to this the-ory, SGLT2 inhibition result in an early hemodynamiceffect at the level of the proximal renal tubule. This inturn promotes sodium and water loss while also,through tubule-glomerular feedback, promotingafferent arteriolar constriction. The ensuring reduc-tion in intraglomerular pressure leads to renal pro-tection. Improving renal function and/or reducingrenal stress can indirectly improve cardiac functionthrough various pathways, including a reduction inafferent SNS activation, reduction in inflammation,and ROS generation. We propose that future studiesare required to examine these possibilities. We alsoargue that the renal hemodynamic effects areobserved independent of glycemia, because in theDAPA-HF trial an initial drop in estimated glomerularfiltration rate was observed in people both with andwithout diabetes. The increase in EPO production mayalso be secondary to an improvement in renal healthand may explain why the hematocrit is increased to asimilar extent in people both with and without dia-betes in DAPA-HF. Additional intriguing mechanismsfor the benefits of SGLT2 inhibition in heart failureinclude improving cardiac energy metabolism anddecreasing cardiac inflammation. Further studies areneeded to examine SGLT2 inhibition impacts on thesepathways in the setting of heart failure, as well as thepotential inter-relationship between these 2 pathways(because ketones can inhibit the inflammatorypathway). Further studies should help elucidateexactly how SGLT2 inhibitors exert these impressivecardiovascular effects.

ADDRESS FOR CORRESPONDENCE: Dr. Gary D.Lopaschuk, Cardiovascular Research Centre, 423Heritage Medical Research Centre, University ofAlberta, Edmonton, Alberta T6G 2S2, Canada. E-mail:[email protected].




642

RE F E RENCE S

1. Braunwald E. The war against heart failure: theLancet lecture. Lancet 2015;385:812–24.

2. Khan SS, Butler J, Gheorghiade M. Managementof comorbid diabetes mellitus and worsening heartfailure. JAMA 2014;311:2379–80.

3. Swoboda PP, McDiarmid AK, Erhayiem B, et al.Diabetes mellitus, microalbuminuria, and subclin-ical cardiac disease: identification and monitoringof individuals at risk of heart failure. J Am HeartAssoc 2017;6:e005539.

4. American Diabetes Association. Cardiovasculardisease and risk management: standards of medi-cal care in diabetes—2019. Diabetes Care 2019;42:S103–23.

5. Gallo LA, Wright EM, Vallon V. Probing SGLT2as a therapeutic target for diabetes: basic physi-ology and consequences. Diab Vasc Dis Res 2015;12:78–89.

6. McMurray JJV, Solomon SD, Inzucchi SE, et al.,for the DAPA-HF Trial Committees and In-vestigators. Dapagliflozin in patients with heartfailure and reduced ejection fraction. N Engl J Med2019;381. 1995–200.

7. Bhatt DL, Verma S, Braunwald E. The DAP-HFtrial: a momentous victory in the war againstheart failure. Cell Metab 2019;30:847–9.

8. Zinman B, Wanner C, Lachin JM, et al., for theEMPA-REG OUTCOME Investigators. Empagli-flozin, cardiovascular outcomes, and mortality intype 2 diabetes. N Engl J Med 2015;373:2117–28.

9. Neal B, Perkovic V, Mahaffey KW, et al., for theCANVAS Program Collaborative Group. Canagli-flozin and cardiovascular and renal events in type2 diabetes. N Engl J Med 2017;377:644–57.

10. Perkovic V, Jardine MJ, Neal B, et al., forCREDENCE Trial Investigators. Canagliflozin andrenal outcomes in type 2 diabetes and nephropa-thy. N Engl J Med 2019;380:2295–306.

11. Wiviott SD, Raz I, Bonaca MP, et al., for theDECLARE-TIMI 58 Investigators. Dapagliflozin andcardiovascular outcomes in type 2 diabetes. N EnglJ Med 2019;380. 380:347–357.

12. Zelniker TA, Wiviott SD, Raz I, et al. SGLT2inhibitors for primary and secondary prevention ofcardiovascular and renal outcomes in type 2 dia-betes: a systematic review and meta-analysis ofcardiovascular outcome trials. Lancet 2019;393:31–9.

13. Verma S, Juni P, Mazer CD. Pump, pipes, andfilter: do SGLT2 inhibitors cover it all? Lancet2019;393:3–5.

14. Verma S, McMurray JJV. SGLT2 inhibitors andmechanisms of cardiovascular benefit: a state-of-the-art review. Diabetologia 2018;61:2108–17.

15. Staels B. Cardiovascular protection by sodiumglucose cotransporter 2 inhibitors: potentialmechanisms. Am J Cardiol 2017;120:S28–36.

16. Butler J, Handelsman Y, Bakris G, Verma S. Useof sodium-glucose co-transport inhibitors in pa-tients with and without type 2 diabetes: implica-tion for incident and prevalent heart failure. Eur JHeart Fail 2020 Jan 11 [E-pub ahead of print].

17. Filippatos TD, Liontos A, Papakitsou I,Elisaf MS. SGLT2 inhibitors and cardioprotection: amatter of debate and multiple hypotheses. Post-grad Med 2019;131:82–8.

18. Lam CSP, Chandramouli C, Ahooja V, Verma S.SGLT-2 inhibitors in heart failure: current man-agement, unmet needs, and therapeutic pros-pects. J Am Heart Assoc 2019;8:e013389.

19. Mazidi M, Rezaie P, Gao HK, Kengne AP. Effectof sodium-glucose cotransport-2 inhibitors onblood pressure in people with type 2 diabetesmellitus: a systematic review and meta-analysis of43 randomized control trials with 22 528 patients.J Am Heart Assoc 2017;6:e004007.

20. Ferrannini E, Baldi S, Frascerra S, et al. Renalhandling of ketones in response to sodium-glucose cotransporter 2 inhibition in patientswith type 2 diabetes. Diabetes Care 2017;40:771–6.

21. Weber MA, Mansfield TA, Cain VA, et al. Bloodpressure and glycaemic effects of dapagliflozinversus placebo in patients with type 2 diabetes oncombination antihypertensive therapy: a rando-mised, double-blind, placebo-controlled, phase 3study. Lancet Diabetes Endocrinol 2016;4:211–20.

22. Kario K, Bohm M, Mahfoud F, et al. Twenty-four-hour ambulatory blood pressure reductionpatterns after renal denervation in the SPYRALHTN-OFF MED Trial. Circulation 2018;138:1602–4.

23. Lambers Heerspink HJ, de Zeeuw D, Wie L,Leslie B, List J. Dapagliflozin a glucose-regulatingdrug with diuretic properties in subjects with type2 diabetes. Diabetes Obes Metab 2013;15:853–62.

24. Hallow KM, Helmlinger G, Greasley PJ,McMurray JJV, Boulton DW. Why do SGLT2 in-hibitors reduce heart failure hospitalization? Adifferential volume regulation hypothesis. Dia-betes Obes Metab 2018;20:479–87.

25. Lopaschuk GD, Ussher JR, Folmes CD,Jaswal JS, Stanley WC. Myocardial fatty acidmetabolism in health and disease. Physiol Rev2010;90:207–58.

26. Zhang L, Jaswal J, Ussher JR, et al. Cardiacinsulin-resistance and decreased mitochondrialenergy production precede the development ofsystolic heart failure after pressure-overload hy-pertrophy. Circ Heart Fail 2013;6:1039–48.

27. Wang W, Zhang L, Battiprolu PK, et al. MalonylCoA decarboxylase inhibition improves cardiacfunction post-myocardial infarction. J Am CollCardiol Basic Trans Science 2019;4:385–400.

28. Mori J, Basu R, McLean BA, et al. Agonist-induced hypertrophy and diastolic dysfunction areassociated with selective reduction in glucoseoxidation: a metabolic contribution to heart failurewith normal ejection fraction. Circ Heart Fail 2012;5:493–503.

29. Neubauer S. The failing heart—an engine outof fuel. N Engl J Med 2007;356:1140.

30. AbouEzziddine OF, Kemp BJ, Borlaug BA, et al.Energetics in heart failure with preserved ejectionfraction. Circ Heart Fail 2019;12:e006240.

31. Ferrannini E, Muscelli E, Frascerra S, et al.Metabolic response to sodium-glucose cotrans-porter 2 inhibition in type 2 diabetic patients.J Clin Invest 2014;124:499–508.

32. Al Jobori H, Daniele G, Adams J, et al. De-terminants of the increase in ketone concentrationduring SGLT2 inhibition in NGT, IFG and T2DMpatients. Diabetes Obes Metab 2017;19:809–13.

33. Ferrannini E, Baldi S, Frascerra S, et al. Shift tofatty substrates utilization in response to sodium-glucose co-transporter-2 inhibition in nondiabeticsubjects and type 2 diabetic patients. Diabetes2016;65:1190–5.

34. Mudaliar S, Alloju S, Henry RR. Can a shift infuel energetics explain the beneficial cardiorenaloutcomes in the EMPA-REG OUTCOME study? Aunifying hypothesis. Diabetes Care 2016;39:1115–22.

35. Ferrannini F, Mark M, Mayoux E. CV Protectionin the EMPA-REG OUTCOME trial: a “thrifty sub-strate” hypothesis. Diabetes Care 2016;39:1108–14.

36. Lopaschuk GD, Verma S. Empagliflozin’s fuelhypothesis: not so soon. Cell Metab 2016;24:200–2.

37. Ho KL, Zhang L, Wagg C, et al. Increased ke-tone body oxidation provides additional energy forthe failing heart without improving cardiac effi-ciency. Cardiovas Res 2019;115:1606–16.

38. Bedi KC, Snyder NW, Brandimarto J, et al.Evidence for intramyocardial disruption of lipidmetabolism and increased myocardial ketone uti-lization in advanced human heart failure. Circula-tion 2016;133:706–16.

39. Aubert G, Martin OJ, Horton JL, et al. Thefailing heart relies on ketone bodies as a fuel.Circulation 2016;133:698–705.

40. Horton JL, Davidson MT, Kurishima C, et al.The failing heart utilizes 3-hydroxybutyrate as ametabolic stress defense. JCI Insight 2019;4:e124079.

41. Verma S, Rawat S, Ho KL, et al. Empagliflozinincreases cardiac energy production in diabetes:novel translational insights into the heart failurebenefits of SGLT2 inhibitors. J Am Coll CardiolBasic Trans Science 2018;3:575–87.

42. Santos-Gallego CG, Requena-Ibanez JA, SanAntonio R, et al. Empagliflozin amelioratesadverse left ventricular remodeling in nondiabeticheart failure by enhancing myocardial energetics.J Am Coll Cardiol 2019;73:1931–44.

43. Shao Q, Meng L, Lee S, et al. Empagliflozin, asodium glucose co-transporter-2 inhibitor, allevi-ates atrial remodeling and improves mitochondrialfunction in high-fat diet/streptozotocin-induceddiabetic rats. Cardiovasc Diabetol 2019;18:165.

44. Nielsen R, Moller N, Gormsen LC, et al. Car-diovascular effects of treatment with the ketonebody 3-hydroxybutyrate in chronic heart failurepatients. Circulation 2019;139:2129–41.

45. Dick SA, Epelman S. Chronic heart failure andinflammation: what do we really know? Circ Res2016;119:159–76.





























































































































































































643

46. Mehta JL, Pothineni NV. Inflammation in heartfailure: the holy grail? Hypertension 2016;68:27–9.

47. Briasoulis A, Androulakis E, Christophides T,Tousoulis D. The role of inflammation and celldeath in the pathogenesis, progression and treat-ment of heart failure. Heart Fail Rev 2016;21:169–76.

48. Iannantuoni F, de Marañon AM, Diaz-Morales N, et al. The SGLT2 inhibitor empagliflozinameliorates the inflammatory profile in type 2diabetic patients and promotes an antioxidantresponse in leukocytes. J Clin Med 2019;8:1814.

49. Heerspink HJL, Perco P, Mulder S, et al. Can-agliflozin reduces inflammation and fibrosis bio-markers: a potential mechanism of action forbeneficial effects of SGLT2 inhibitors in diabetickidney disease. Diabetologia 2019;62:1154–66.

50. Leng W, Wu M, Pan H, et al. The SGLT2 in-hibitor dapagliflozin attenuates the activity ofROS-NLRP3 inflammasome axis in steatohepatitiswith diabetes mellitus. Ann Transl Med 2019;7:429.

51. Lee T-M, Chang N-C, Lion S-Z. Dapagliflozin, aselective SGLT2 inhibitor, attenuated cardiacfibrosis by regulating the macrophage polarizationvia STAT3 signaling in infarcted rat hearts. FreeRadic Biol Med 2017;104:298–310.

52. Kang S, Verma S, Hassanabad AF, et al. Directeffects of empagliflozin on extracellular matrixremodelling in human cardiac myofibroblasts:novel translational clues to explain EMPA-REGOUTCOME results. Can J Cardiol 2019 Aug 29 [E-pub ahead of print].

53. Rotkvi�c PG, Berkovi�c MC, Bulj N, Rotkvi�c L.Minireview: are SGLT2 inhibitors heart savers indiabetes? Heart Fail Rev 2019 Aug 13 [E-pubahead of print].

54. Butts B, Gary RA, Dunbar SB, Butler J. Theimportance of NLRP3 inflammasome in heart fail-ure. J Card Fail 2015;21:586–93.

55. Benetti E, Mastrocola R, Vitarelli G, et al.Empagliflozin protects against diet-induced NLRP-3 inflammasome activation and lipid accumulation.J Pharmacol Exp Ther 2016;359:45–53.

56. Tahara A, Kurosaki E, Yokono M, et al. Effectsof SGLT2 selective inhibitor ipragliflozin on hy-perglycemia, hyperlipidemia, hepatic steatosis,oxidative stress, inflammation, and obesity in type2 diabetic mice. Eur J Pharmacol 2013;715:246–55.

57. Lee Y-H, Kim SR, Han DH, et al. Senescent Tcells predict the development of hyperglycemia inhumans. Diabetes 2019;68:156–62.

58. Leng W, Ouyang X, Lei X, et al. The SGLT-2inhibitor dapagliflozin has a therapeutic effect onatherosclerosis in diabetic ApoE�/� mice. Media-tors Inflamm 2016;2016:6305735.

59. Ye Y, Jia X, Bajaj M, Birnbaum Y. Dapagliflozinattenuates Naþ/Hþexchanger-1 in cardiofibro-blasts via AMPK activation. Cardiovasc DrugsTherap 2018;32:553–8.

60. Byrne NJ, Matsumura N, Maayah ZH, et al.Empagliflozin blunts worsening cardiac dysfunc-tion associated with reduced NLRP3 (nucleotide-binding domain-like receptor protein 3)

inflammasome activation in heart failure. CircHeart Fail 2020;13:e006277.

61. Youm YH, Nguyen KY, Trant RW, et al. Theketone metabolite b-hydroxybutyrate blocksNLRP3 inflammasome-mediated inflammatorydisease. Nat Med 2015;21:263–9.

62. Cai X, Yang W, Gao X, et al. The associationbetween the dosage of SGLT2 inhibitor and weightreduction in type 2 diabetes patients: a meta-analysis. Obesity (Silver Spring) 2018;26:70–80.

63. Pereira MJ, Eriksson JW. Emerging role ofSGLT-2 inhibitors for the treatment of obesity.Drugs 2019;79:219–30.

64. Lee PC, Ganguly S, Goh SY. Weight lossassociated with sodium-glucose cotransporter-2inhibition: a review of evidence and underlyingmechanisms. Obes Rev 2018;19:1630–41.

65. Sattar N. Revisiting the links between glycae-mia, diabetes and cardiovascular disease. Dia-betologia 2013;56:686–95.

66. Cannon CP, Perkovic V, Agarwal R, et al.Evaluating the effects of canagliflozin on cardio-vascular and renal events in patients with type 2diabetes and chronic kidney disease according tobaseline HbA1c, including those with HbA1c <7%:results from the CREDENCE trial. Circulation 2019;141:407–10.

67. Connelly KA, Zhang Y, Vistram A, et al.Empagliflozin improves diastolic function in anondiabetic rodent model of heart failure withpreserved ejection fraction. J Am Coll Cardiol BasicTrans Science 2019;4:27–37.

68. Chiba Y, Yamada T, Tsukita S, et al. Dapagli-flozin, a sodium-glucose co-transporter 2 inhibi-tor, acutely reduces energy expenditure in BAT vianeural signals in mice. PLoS ONE 2016;11:e0150756.

69. Yoshikawa T, Kishi T, Shinohara K, et al.Arterial pressure lability is improved by sodium-glucose cotransporter 2 inhibitor instreptozotocin-induced diabetic rats. HypertensRes 2017;40:646–51.

70. Matthews VB, Elliot RH, Rudnicka C, Hricova J,Herat L, Schlaich MP. Role of the sympatheticnervous system in regulation of the sodiumglucose cotransporter 2. J Hypertens 2017;35:2059–68.

71. Kimmerly DS, Shoemaker JK. Hypovolemia andneurovascular control during orthostatic stress.Am J Physiol 2002;282:H645–55.

72. Jordan J, Tank J, Heusser K, et al. The effect ofempagliflozin on muscle sympathetic nerve activ-ity in patients with type II diabetes mellitus. J AmSoc Hypertens 2017;11:604–12.

73. Sano M. A new class of drugs for heart failure:SGLT2 inhibitors reduce sympathetic overactivity.J Cardiol 2018;71:471–6.

74. Byrne NJ, Parajuli N, Levasseur JL, et al.Empagliflozin prevents worsening of cardiacfunction in an experimental model of pressureoverload-induced heart failure. J Am Coll CardiolBasic Trans Science 2017;2:347–54.

75. Shi L, Zhu D, Wang S, et al. Dapagliflozin at-tenuates cardiac remodeling in mice model ofcardiac pressure overload. Am J Hypertens 2019;32:452–9.

76. Verma S, Garg A, Yan AT, et al. Effect ofempagliflozin on left ventricular mass and dia-stolic function in individuals with diabetes: animportant clue to the EMPA-REG OUTCOME trial?Diabetes Care 2016;39:e212–3.

77. Esterline RL, Vaag A, Oscarsson J, Vora J.SGLT2 inhibitors: clinical benefits by restoration ofnormal diurnal metabolism? Eur J Endocrinol2018;178:R113–25.

78. Lee HC, Shiou YL, Jhuo SJ, et al. The sodium-glucose co-transporter 2 inhibitor empagliflozinattenuates cardiac fibrosis and improves ventric-ular hemodynamics in hypertensive heart failurerats. Cardiovasc Diabetol 2019;18:45.

79. Lim VG, Bell RM, Arjun S, Kolatsi-Joannou M,Long DA, Yellon DM. SGLT2 inhibitor, canagli-flozin, attenuates myocardial infarction in thediabetic and nondiabetic heart. J Am Coll CardiolBasic Trans Science 2019;4:15–26.

80. Wakabayashi S, Hisamitsu T, Nakamura TY.Regulation of the cardiac Na⁺/H⁺ exchanger inhealth and disease. J Mol Cell Cardiol 2013;61:68–76.

81. Baartscheer A, Schumacher CA, Wust CA, et al.Empagliflozin decreases myocardial cytoplasmicNaþ through inhibition of the cardiac Naþ/Hþ

exchanger in rats and rabbits. Diabetologia 2017;60:568–73.

82. Uthman L, Baartscheer A, Schumacher CA,et al. Direct cardiac actions of sodium glucosecotransporter 2 inhibitors target pathogenicmechanisms underlying heart failure in diabeticpatients. Front Physiol 2018;9:1575.

83. Packer M. Do sodium-glucose co-transporter-2 inhibitors prevent heart failure with a preservedejection fraction by counterbalancing the effectsof leptin? A novel hypothesis. Diabetes ObesMetab 2018;20:1361–6.

84. Iborra-Egea O, Santiago-Vacas E, Yurista SR,et al. Unraveling the molecular mechanism of ac-tion of empagliflozin in heart failure with reducedejection fraction with or without diabetes. J AmColl Cardiol Basic Trans Science 2019;4:831–40.

85. Mentzer RM, Bartels C, Bolli R, et al., for theEXPEDITION Study Investigators. Sodium-hydrogen exchange inhibition by cariporide toreduce the risk of ischemic cardiac events in pa-tients undergoing coronary artery bypass grafting:results of the EXPEDITION study. Ann Thorac Surg2008;85:1261–70.

86. Bell RM, Yellon DM. SGLT2 inhibitors: hy-potheses on the mechanism of cardiovascular pro-tection. Lancet Diabetes Endocrinol 2017;6:435–7.

87. Connelly KA, Zhang Y, Desjardins JF, Thai K,Gilbert RE. Dual inhibition of sodium-glucoselinked cotransporters 1 and 2 exacerbates cardiacdysfunction following experimental myocardialinfarction. Cardiovasc Diabetol 2018;17:99.

88. Hare JM, Johnson RJ. Uric acid predicts clinicaloutcomes in heart failure: insights regarding therole of xanthine oxidase and uric acid in diseasepathophysiology. Circulation 2003;107:1951–3.

89. Schernthaner G, Gross JL, Rosenstock J, et al.Canagliflozin compared with sitagliptin for pa-tients with type 2 diabetes who do not haveadequate glycemic control with metformin plus


















































































































































































































644

sulfonylurea: a 52-week randomized trial. Dia-betes Care 2013;36:2508–15.

90. Chino Y, Samukawa Y, Sakai S, et al. SGLT2inhibitor lowers serum uric acid through alterationof uric acid transport activity in renal tubule byincreased glycosuria. Biopharm Drug Dispos 2014;35:391–404.

91. Nussenzweig SC, Verma S, Finkel T. The role ofautophagy in vascular biology. Circ Res 2015;116:480–8.

92. Raggi P, Gadiyaram V, Zhang C, Chen Z,Lopaschuk G, Stillman AE. Statins reduce epicar-dial adipose tissue attenuation independent oflipid lowering: a potential pleiotropic effect. J AmHeart Assoc 2019;8:e013104.

93. Sato T, Aizawa Y, Yuasa S, et al. The effect ofdapagliflozin treatment on epicardial adipose tis-sue volume. Cardiovasc Diabetol 2018;17:6.

94. Sano M, Takei M, Shiraishi Y, Suzuki Y.Increased hematocrit during sodium-glucosecotransporter 2 inhibitor therapy indicates recov-ery of tubulointerstitial function in diabetic kid-neys. J Clin Med Res 2016;8:844–7.

95. Mazer CD, Connelly PW, Gilbert RE, et al.Effect of empagliflozin on erythropoietin levels,iron stores and red blood cell morphology in

patients with type 2 diabetes and coronary arterydisease. Circulation 2019 Nov 11 [E-pub ahead ofprint].

96. Hess DA, Terenzi DC, Trac JZ, et al. SGLT2Inhibition with empagliflozin increases circulatingprovascular progenitor cells in people with type 2diabetes mellitus. Cell Metab 2019;30:609–13.

97. Zhou B, Tian R. Mitochondrial dysfunction inpathophysiology of heart failure. J Clin Invest2018;128:3716–26.

98. Li C, Zhang J, Xue M, et al. SGLT2 inhibitionwith empagliflozin attenuates myocardial oxida-tive stress and fibrosis in diabetic mice heart.Cardiovasc Diabetol 2019;18:15.

99. Uthman L, Homayr A, Juni RP, et al. Empagli-flozin and dapagliflozin reduce ROS generation andrestore NO bioavailability in tumor necrosis factora-stimulated human coronary arterial endothelialcells. Cell Physiol Biochem 2019;53:865–86.

100. Endemann DH, Schiffrin EL. Endothelialdysfunction. J Am Soc Nephrol 2004;15:1983–92.

101. Patel AR, Kuvin JT, Pandian NG, et al. Heartfailure etiology affects peripheral vascular endo-thelial function after cardiac transplantation. J AmColl Cardiol 2001;37:195–200.

102. Gaspari T, Spizzo I, Liu H, et al. Dapagliflozinattenuates human vascular endothelial cell acti-vation and induces vasorelaxation: a potentialmechanism for inhibition of atherogenesis. Dia-betes Vasc Dis Res 2017;15:64–73.

103. Mancini SJ, Boyd D, Katwan OJ, et al. Cana-gliflozin inhibits interleukin-1b-stimulated cyto-kine and chemokine secretion in vascularendothelial cells by AMP-activated protein kinase-dependent and -independent mechanisms. Sci Rep2018;8:5276.

104. Juni RP, Kuster DWD, Goebel M, et al.Cardiac microvascular endothelial enhancementof cardiomyocyte function is impaired byinflammation and restored by empagliflozin.J Am Coll Cardiol Basic Trans Science 2019;4:575–91.

105. Li H, Shin SE, Seo MS, et al. The anti-diabeticdrug dapagliflozin induces vasodilation via acti-vation of PKG and Kv channels. Life Sci 2018;197:46–55.

KEY WORDS erythropoetin, inflammation,ketones, renal function, sympathetic nervoussystem













































































EDITOR’S PAGE

The National Institute of Allergy andInfectious Diseases Decision to Stopthe Adaptive COVID-19 TrialOn Solid Ethical and Scientific Grounds

Jessica Mozersky, PHD,a Douglas L. Mann, MD, Editor-in-Chief, JACC: Basic to Translational Science,b

James M. DuBois, DSC, PHDa

O n April 29, the National Institute of Healthdisclosed the preliminary findings of theACTT-1 (Adaptive COVID-19 Treatment

Trial) (NCT04280705), the first clinical trial in theUnited States to evaluate an experimental treatmentfor coronavirus disease-2019 (COVID-19). ACTT wasdesigned as an adaptive, randomized, double-blind,placebo-controlled trial to evaluate the safety and ef-ficacy of remdesivir (and eventually other drugs) inhospitalized adults diagnosed with COVID-19.Remdesivir is a nucleoside analog that interfereswith the action of viral ribonucleic acid–dependentribonucleic acid polymerase (1). Patients withCOVID-19 were randomized to receive remdesivir orplacebo as an inpatient and assessed for up to29 days as an outpatient. The primary outcome vari-able was time to recovery by day 29. An independentdata and safety monitoring board (DSMB) met onApril 27 to review data and shared their interim anal-ysis with the ACTT-1 study team. The DSMB notedthat when compared with placebo, remdesivir-treated patients had a shorter time to recovery, whichis an endpoint that is used frequently in influenza

ISSN 2452-302X

From the aBioethics Research Center, Department of Medicine, Wash-

ington University School of Medicine in St. Louis, St. Louis, Missouri; and

the bCenter for Cardiovascular Research, Department of Medicine,

Washington University School of Medicine in St. Louis, St. Louis, Mis-

souri.






trials. Preliminary results from ACTT show that pa-tients treated with remdesivir had a 31% faster timeto recovery than those who received placebo(p < 0.001). For the remdesivir treatment arm, themedian time to recovery was 11 days for patientstreated when compared with 15 days for those whoreceived placebo. There was a trend toward improvedsurvival in the remdesivir treatment arm (p ¼ 0.059).Based on these findings, the DSMB recommendedthat there was no need for a placebo-only group inthe next phase of the ACCT study, which planned totest baricitinib, a Janus kinase inhibitor, againstremdesivir. Following the DSMB meeting, the Na-tional Institute of Allergy and Infectious Diseases(NIAID) chose to offer patients in the placebo arm ofACCT-1 the opportunity to receive open label remde-sivir, rather than allowing these patients to remain inthe placebo arm, so that additional mortality datacould be collected.

Although the results of the ACCT-1 clinical trial fellshort of being a home run because of the short-termendpoint that was chosen, these results constitutethe first randomized scientific evidence that antiviralagents can be used to treat patients with COVID-19,and thus represent an important win for the healthcare community in the war against COVID-19. How-ever, the ACCT-1 trial has been sharply criticized bymany experienced clinical trialists because they feltthat the NIAID should have allowed the investigatorsto collect additional data to learn whether remdesivirsaved lives. Dr. Steven Joffe, who is a pediatriconcologist and bioethicist and the Interim Chair,Department of Medical Ethics and Health Policy at theUniversity of Pennsylvania, stated that he believed


https://clinicaltrials.gov/ct2/show/NCT04280705?term=NCT04280705&draw=2&rank=1





Mozersky et al. J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0

Editor’s Page J U N E 2 0 2 0 : 6 4 5 – 7

646

that the NIAID likely took the right steps in making itsdecision to give remdesivir to the placebo patients,but raised concerns about using time to improve-ment, not death, as the measure of success, in thefirst place. He stated, “I don’t find this endpoint verycompelling, and to me the real issue is the decision todesign the trial around the endpoint of time to re-covery defined in the way they defined recovery. . . .To me, the decisions that are this weighty ought to bebased on clinically important endpoints. (2)” Dr.Steven Nissen, a well-known experienced clinicaltrialist and cardiologist at the Cleveland Clinic, didnot believe that allowing placebo patients to takeremdesivir was the correct decision. He stated, “Ibelieve it is in society’s best interest to determinewhether remdesivir can reduce mortality, and withthe release of this information doing a placebo-controlled trial to determine if there is a mortalitybenefit will be very difficult.” He went on to say thathe viewed the NIAID’s decision as “a lostopportunity”(2).

We believe that the controversies surroundingthe NIAID’s decision to terminate the ACCT-1 trialin the midst of the COVID-19 epidemic raiseimportant questions about the goals and conduct ofclinical trials during a public health crisis. Inparticular, the ACCT-1 trial controversy illustratesthe tension between the competing goals of clinicalresearch: the production of ideal scientific knowl-edge that benefits society—in this case mortalityoutcome data—with the protection of individualparticipants in the face of a deadly pandemic; andthe need to make treatments quickly available tovery sick patients (3).

Based on the data that was available to the NIAIDand the ACCT-1 investigators, we believe there arevalid scientific reasons for designing and conductinga single arm versus placebo trial with remdesivir,without mortality as a primary outcome. First, amongthe lessons learned from treating patients with thehuman immunodeficiency virus is that effective cur-rent antiretroviral therapy has evolved from mono-therapy with azidothymidine to a combination of 2antiretroviral reverse transcriptase inhibitors to ahighly active triple antiretroviral therapy. There is nobiological precedence for first-generation antiviraltherapies reducing morality when given as a singleagent. Second, remdesivir was among the treatmentarms that were dropped in the NIAID-sponsoredPALM (Pamoja Tulinde Maisha [“Together SaveLives” in the Kiswahili language]) trial (4), becausethe remdesivir treatment arm had the highest mor-tality of the 4 different anti-Ebola treatment armsthat were being tested.

From an ethical standpoint, the ACCT-1 trial designand decision to stop the placebo arm early are justi-fiable, especially when considering prior researchconducted during pandemics. While examining theclinical trials conducted during the 2014/15 Ebolaoutbreak, the National Academies of Science, Engi-neering, and Medicine observed that the very idea ofrandomized clinical trials (RCTs) was highly contro-versial: “RCTs are the preferred research designbecause they allow researchers to directly comparethe outcomes of similar groups of people who differonly in the presence or absence of the investigationalagent. However, many stakeholders argued that RCTswould be unethical in the context of the Ebolaepidemic. The arguments against RCTs were varied,but most were primarily based on one centralassumption: that it was unethical and unacceptable todeprive patients of an agent that could potentiallyprevent or treat Ebola, given the high mortality rateand lack of known and available treatment options”(5). The National Academies of Science, Engineering,and Medicine committee supported RCTs at theoutset of the Ebola outbreak because it was unknownwhether any agents would be safe and effective; trueequipoise existed between the experimental treat-ment and placebo. Thus, the use of placebo in theACCT-1 trial was warranted based on established sci-entific and ethical grounds. However, at the pointwhen NIAID stopped the ACCT-1 trial, it would bedifficult to say that there was no effective agent tojustify the continued use of placebo in ACCT-1 or inthe adaptive clinical trials designs that will followACCT-1. The most recent version of the World HealthOrganization Declaration of Helsinki (2013) stateswhen a known effective treatment exists, new treat-ments must be tested against it except when “it isnecessary to determine the efficacy or safety of anintervention and the patients who receive any inter-vention less effective than the best proven one, pla-cebo, or no intervention will not be subject toadditional risks of serious or irreversible harm as aresult of not receiving the best proven intervention”(6). Could one really say that withholding remdesivirafter it was found to shorten time to recovery—theprimary trial outcome participants were told about inthe consent form—would not place patients who werehospitalized with COVID-19 at an additional risk of“serious or irreversible harm as a result of notreceiving the best proven intervention?” It is alsoworth noting that ACCT-1 was not designed norpowered for a mortality outcome. Moreover, during aperiod of pandemic, it is fair to assume that mosthospitalized patients want to try something that iseffective rather than continuing on placebo.

J A C C : B A S I C T O T R A N S L A T I O N A L S C I E N C E V O L . 5 , N O . 6 , 2 0 2 0 Mozersky et al.J U N E 2 0 2 0 : 6 4 5 – 7 Editor’s Page

647

Clinical research necessarily involves balancingcompeting priorities and values such as producinggeneralizable knowledge that benefits society whileminimizing harms and maximizing benefits to trialparticipants (3). The principles articulated in theDeclaration of Helsinki describe a common approachto balancing the goals of gaining new medicalknowledge and protecting patients who enroll inclinical trials: Even in ordinary times, we cannotwithhold a proven intervention when doing so risksharming patients. This alone could justify stopping atrial early. During times of pandemic, the urgent needfor an initial proven intervention just strengthens thecase for prioritizing patients’ needs over ideal scien-tific endpoints.

Nevertheless, broadly recognized principles ofpublic health ethics require that we infringe as littleas possible against competing values or prioritiessuch as knowledge regarding survival in clinical trialsas we pursue a goal such as providing acutely ill pa-tients with a proven intervention (7). That is to say,even if we prioritize the protection of study partici-pants and society’s need for an initial proven inter-vention, we should support efforts to determinewhether remdesivir reduces mortality to whateverextent is possible after dropping placebo controls.Conceivably this can be done with remdesivir in thecontext of nonrandomized clinical trials that usepropensity matching of treated patients and un-treated case control subjects. Admittedly, the resultswill not be as definitive as an RCT and may represent

a lost opportunity; however, this may be the type ofscientific compromise that is both reasonable andrequired in the context of the public health crisis thatwe are currently facing and will likely have to faceagain in the future. Many aspects of evidence-basedmedicine are based on less than perfect evidence,and we must never forget that the participantsenrolling in a clinical trial during a deadly pandemicare first and foremost patients who trust their physi-cians to do what is best for them. As always, wewelcome your thoughts on the ACCT-1 trial, eitherthrough social media (#JACC:BTS) or by e-mail([email protected]).

ADDENDUM

The partial results of the ACCT-1 trial were publishedonline (8) after this editorial was completed. InACCT-1, the Kaplan Meier estimates of mortality by 14days were, respectively, 7.1% and 11.9%, in theremdesivir and placebo groups (hazard ratio 0.70;95% CI 0.47-1.04). The Kaplan Meier estimates ofmortality for 28 days had not been analyzed at thetime this preliminary report was published becausepatients had not yet completed their 29th visit.

ADDRESS FOR CORRESPONDENCE: Dr. James M.DuBois, Bioethics Research Center, Department ofMedicine, Box 8005, Washington University School ofMedicine, St. Louis, Missouri 63108. E-mail:[email protected].

RE F E RENCE S

1. KyB,MannDL. COVID-19 clinical trials: a primer forthe cardiovascular and cardio-oncology communities.J Am Coll Cardiol Basic Trans Science 2020;5:501–17.2. Herper M. Inside the NIH’s controversial de-cision to stop its big remdesivir study. STAT[online news]. May 11, 2020. Available at:https://www.statnews.com/2020/05/11/inside-the-nihs-controversial-decision-to-stop-its-big-remdesivir-study/. Accessed May 17, 2020.

3. Emanuel EJ, Wendler D, Grady C. What makesclinical research ethical? JAMA 2000;283:2701–11.

4. Mulangu S, Dodd LE, Davey RT Jr., et al., for thePALM Consortium Study Team. A randomized,controlled trial of Ebola virus disease therapeutics.N Engl J Med 2019;381:2293–303.

5. Busta ER, Mancher M, Cuff PA, McAdam K,KeuschG, editors. Integrating Clinical Research intoEpidemic Response: The Ebola Experience. Wash-ington, DC: National Academies Press, 2017:7.

6. World Medical Association. World Medical As-sociation Declaration of Helsinki: ethical principles

for medical research involving human subjects.JAMA 2013;310:2191–4.

7. Childress JF, Faden RR, Gaare RD, et al. Publichealth ethics: mapping the terrain. J Law MedEthics 2002;30:170–8.

8. Beigel JH, Tomashek KM, Dodd LE, et al.Remdesivir for the treatment of Covid-19 - pre-liminary report. N Engl J Med 2020. Available at:https://www.nejm.org/doi/full/10.1056/NEJMoa2007764. Accessed May 26, 2020.






https://www.statnews.com/2020/05/11/inside-the-nihs-controversial-decision-to-stop-its-big-remdesivir-study/




















https://www.nejm.org/doi/full/10.1056/NEJMoa2007764

https://www.nejm.org/doi/full/10.1056/NEJMoa2007764

Documents

Full Issue PDF · San Francisco General Hospital, San Francisco, CA Roberta Gottlieb, MD, ... SOCIAL MEDIA COORDINATOR Tamika Edaire, BS, American College of Cardiology, Washington,