Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/273469001

Cannabinoidreceptor1isamajormediatorofrenalfibrosis

ArticleinKidneyInternational·March2015

DOI:10.1038/ki.2015.63·Source:PubMed

CITATIONS

3

READS

65

19authors,including:

StanislasGrassinDelyle

UniversitédeVersaillesSaint-Quentin

89PUBLICATIONS324CITATIONS

SEEPROFILE

NinoslavIvanovski

87PUBLICATIONS324CITATIONS

SEEPROFILE

AntoineDurrbach

FrenchInstituteofHealthandMedicalResea…

116PUBLICATIONS2,885CITATIONS

SEEPROFILE

HélèneFrançois

AssistancePublique–HôpitauxdeParis

93PUBLICATIONS773CITATIONS

SEEPROFILE

AllcontentfollowingthispagewasuploadedbyHélèneFrançoison04October2016.

Theuserhasrequestedenhancementofthedownloadedfile.Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,lettingyouaccessandreadthemimmediately.

Cannabinoid receptor 1 is a major mediator of renalfibrosisLola Lecru1, Christophe Desterke2, Stanislas Grassin-Delyle3, Christos Chatziantoniou4,Sophie Vandermeersch4, Aurore Devocelle1, Amelia Vernochet1, Ninoslav Ivanovski1, Catherine Ledent5,Sophie Ferlicot6, Meriem Dalia1, Myriam Saïd1, Séverine Beaudreuil1,6, Bernard Charpentier1,6,Aimé Vazquez1, Julien Giron-Michel1, Bruno Azzarone7, Antoine Durrbach1,6,8 and Hélène François1,6,8

1INSERM Unité 1014, Régulation de la survie et des Allogreffes, Institut André Lwoff, Hôpital Paul Brousse, Villejuif, France; 2INSERM Unité972, Transfert des gènes dans le foie: Applications Thérapeutiques, Institut André Lwoff, Hôpital Paul Brousse, Villejuif, France; 3Universitéde Versailles Saint-Quentin, Mass Spectrometry Facility, MasSpecLab, Versailles, France; 4UMRS 702, Remodelage et Réparation du TissuRénal, Hôpital Tenon, Paris, France; 5Institut de la Recherche Interdisciplinaire en Biologie Humaine Et Moléculaire, Bruxelles, Belgium;6AP-HP Hôpital Bicêtre, le Kremlin-Bicêtre, France; 7Immunology Department, Istituto Giannina Gaslini, Genova, Italy and 8InstitutFrancilien de Recherche en Néphrologie et Transplantation, le Kremlin-Bicêtre, France

Chronic kidney disease, secondary to renal fibrogenesis,is a burden on public health. There is a need to explore newtherapeutic pathways to reduce renal fibrogenesis. To studythis, we used unilateral ureteral obstruction (UUO) in miceas an experimental model of renal fibrosis and microarrayanalysis to compare gene expression in fibrotic and normalkidneys. The cannabinoid receptor 1 (CB1) was among themost upregulated genes in mice, and the main endogenousCB1 ligand (2-arachidonoylglycerol) was significantlyincreased in the fibrotic kidney. Interestingly, CB1 expressionwas highly increased in kidney biopsies of patients with IgAnephropathy, diabetes, and acute interstitial nephritis. Bothgenetic and pharmacological knockout of CB1 induced aprofound reduction in renal fibrosis during UUO. While CB2 isalso involved in renal fibrogenesis, it did not potentiate therole of CB1. CB1 expression was significantly increased inmyofibroblasts, the main effector cells in renal fibrogenesis,upon TGF-β1 stimulation. The decrease in renal fibrosisduring CB1 blockade could be explained by a direct action onmyofibroblasts. CB1 blockade reduced collagen expressionin vitro. Rimonabant, a selective CB1 endocannabinoidreceptor antagonist, modulated the macrophage infiltrateresponsible for renal fibrosis in UUO through a decrease inmonocyte chemoattractant protein-1 synthesis. Thus, CB1has a major role in the activation of myofibroblasts and maybe a new target for treating chronic kidney disease.Kidney International advance online publication, 11 March 2015;doi:10.1038/ki.2015.63

KEYWORDS: cannabinoid; chronic kidney disease; fibrosis; myofibroblast

Chronic kidney disease is a burden on public health systems,and it affects millions of individuals worldwide. Indepen-dently of its cause, chronic kidney disease is secondary to thereplacement of functional renal tissue by extracellular matrixproteins (that is, fibrosis) that ultimately impair kidneyfunction. Therefore, preventing renal fibrogenesis remainsa major goal for clinicians. There are few new therapeuticoptions available, underscoring the urgent need to explorenew potential therapeutic pathways to reduce renal fibro-genesis and chronic kidney disease.

To this end, we performed a microarray analysis thatcompared the gene expression profile of diseased and fibroticversus normal kidneys in mice using the unilateral ureteralobstruction (UUO) experimental model. This model wasthought best for this approach, as it is a rapid andreproducible model of renal fibrosis in mice, and it mimicsthe main steps of tubulointerstitial fibrosis in humans.1

Unexpectedly, among the well-known pathways involved inrenal fibrosis, expression of the Cnr1 gene (which encodescannabinoid receptor 1 (CB1)) was found to be among themost significantly upregulated genes in the UUO model.The cannabinoid pathway acts through G protein–coupledcannabinoid receptors 1 (CB1) and 2 (CB2) that are bestknown for their roles in regulating food intake, body weight,thymic functions, and behavior. Therefore, the CB1 pathwayis a major player in obesity and, more specifically, inmetabolic syndrome and its related complications, such asatherosclerosis,2 type 2 diabetes, and microalbuminuria.3

However, CB1 seems to be also directly involved in bothmetabolic and nonmetabolic causes of liver.4

In the kidney, CB1 expression is very low and is moreprominently expressed in the vasculature where it regulatesrenal hemodynamics.4 CB1 is involved in the development ofmicroalbuminuria during diabetes and before overt nephrop-athy develops.3,5 This decrease in albuminuria in diabeticmice, obtained by blocking CB1, may be explained by changes

http://www.kidney-international.org bas i c re sea r ch© 2015 International Society of Nephrology

Correspondence: Hélène François, Service de Néphrologie, Dialyse etTransplantation, Hôpital Bicêtre, 78 rue du Général Leclerc, 94275 Le KremlinBicêtre Cedex, France. E-mail: helene.francois@bct.aphp.fr

Received 3 July 2014; revised 13 January 2015; accepted 22 January2015

Kidney International 1

in renal hemodynamics as glomeruli were normal underelectron microscopy.5 Recently, another study found adifference in mesangial expansion after CB1 blockade, witha decrease in profibrotic genes during the early stages ofdiabetic nephropathy in mice.3,5 So far, the role of CB1 in thedevelopment of overt nephropathy or in renal fibrogenesishas not been established.

Therefore, the objective of our study was to demonstratethat the cannabinoid receptors, and particularly CB1,represent an entirely new therapeutic pathway during renalfibrogenesis and act independently of their metabolic andrenal hemodynamic effects.

RESULTSCB1 expression is increased during the UUO model of renalfibrosisWe first compared the gene expression profile of three fibrotickidneys with three undamaged contralateral kidneys in miceusing the UUO model. We found many overexpressed genesin obstructed kidneys that are classically involved in renalfibrosis, such as Tgfb, Mmp, and Timp. However, using theDAVID (Database for Annotation, Visualization and Inte-grated Discovery) bioinformatics resources, the Cnr1 gene,encoding for CB1, was found to be one of the most upregu-lated genes in fibrotic kidneys, as illustrated by heatmaps thatshowed hierarchical clustering of significant transcripts(Figure 1a). This was confirmed using real-time quantitativePCR (RT-qPCR) (Figure 1b).

Immunohistochemistry revealed that the CB1 receptordrastically increased during UUO and was expressed in thetubules, interstitium, and glomeruli. Indeed, no staining wasfound in the obstructed kidney of Cnr1−/− mice, used asnegative control (Figure 1c). We performed double immu-nostainings in kidney cortex, using specific markers for Tlymphocytes (CD3), macrophages (F4/80), and myofibro-blasts (α-smooth muscle actin (α-SMA)). Under confocalmicroscopy, neither CD3+ lymphocytes nor F4/80+ macro-phages expressed the CB1 receptor in obstructed kidneys.Nonetheless, we found coexpression of the CB1 receptor withthe α-SMA marker in interstitial cells, indicating that CB1 wasexpressed in myofibroblasts in the kidney during UUO(Figure 2).

Endocannabinoid synthesis is altered during UUOWe quantified anandamide and 2-arachidonoylglycerol(2-AG), the main endogenous CB1 ligands within the kidneyin mice during UUO. We found a significant increase of2-AG (Po0.01), whereas anandamide decreased in UUO(Figure 1f). However, 2-AG is 50-fold more abundant thananandamide in obstructed kidneys for a similar CB1 affinity(Supplementary Figure S1 online); therefore, we found aglobal increase in CB1 ligands during UUO. Using ourmicroarray analysis, we also found that the transcription ofgenes involved in endocannabinoid synthesis was significantlyaltered during UUO (Supplementary Figure S2a–c online).

CB1 expression is induced in human nephropathiesWe next studied CB1 expression in normal and diseasedkidneys in human renal biopsies. We found increased expres-sion of CB1 in acute interstitial nephritis (n= 5), IgA nephrop-athy (n= 5), and diabetic nephropathy (n= 6) compared with alow level of CB1 expression in normal kidneys, mostly inendothelial cells (n= 5). In injured kidneys, CB1 expressionwas found in tubules, interstitial cells, and podocytes indiabetes nephropathy, and in mesangial cells in IgA nephrop-athy (Figure 3a–c). In addition, CB1 expression correlates withkidney function (Po0.01) (Supplementary Figure S3 online).

Both CB1 genetic invalidation and CB1 pharmacologicalblockade reduced fibrosis in the UUO modelWe investigated the effect of CB1 deficiency on thedevelopment of renal fibrosis in the UUO model usinggenetic disruption of the CB1 gene (Cnr1) and specificpharmacological blockade of CB1. Histomorphometric ana-lytical software was used to quantify renal fibrosis within anentire kidney cross-section with Sirius red (Figure 4a–c).

After 8 days of UUO, as expected, we found a significantincrease in collagen in obstructed compared with nonob-structed kidneys. The obstructed kidneys of Cnr1−/− miceshowed a 33% reduction in collagen accumulation comparedwith wild-type mice (Figure 4b). Similarly, rimonabant-treated mice had a 65% reduction in fibrosis compared withvehicle-treated mice, using two different doses of rimonabant(Figure 4d). We also found a 25–60% reduction in fibrosis inAM6545 compared with vehicle-treated mice (Figure 4e).

Our results were confirmed by quantification of collagenIII by WB (Figure 4f) and of collagen I and III mRNA usingRT-qPCR (Figure 4g).

CB2 is not involved in the CB1 blockade–dependentantifibrotic effect during UUOBecause CB1 and CB2 may exert opposite physiological effectsduring liver fibrosis,6 we investigated CB2 expression duringUUO in CB1 blockade. We found a significant increasein Cnr2 (encoding the CB2 receptor) mRNA after UUOand during pharmacological and genetic CB1 blockade(Figure 5a). This result was confirmed in immunohistochem-istry (Supplementary Figure S4 online).

To investigate whether the CB2 pathway was involved inthe antifibrogenetic effect conferred by CB1 invalidationduring UUO, we used a specific CB2 antagonist (SR144528)and an agonist (JWH133). Neither of these treatments furtherreduced the development of fibrosis when compared with theCB1 pharmacological blockade. However, the CB2 antagonistalone aggravated the development of fibrosis. In addition, theCB2 agonist alone blunted the development of fibrosis duringUUO (Figure 5b).

Number of myofibroblasts is not altered by blockade of CB1during UUOWe next studied the expression of mesenchymal markers inthe kidney cortex during CB1 blockade. Mesenchymal cell

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

2 Kidney International

WT

#1

NO

0.0 0.9523212 1.56748310.

6854

911

0.84

2745

54

1.0

WT

#2

NO

WT

#3

NO

WT

#2

O

WT

#1

O

WT

#3

O

CB

1–0.31721264

0.21215981

0.74153227PlatMmp9Snai1Ccr2Col1a2Col3a1Serpinh1Itgb5Thbs2DcnItgb3LoxMmp14Mmp3Nfkb1Stat1Tgfbr1Mmp2Timp2Ccl12Edn1EngIl1bTgfb1Tgfb2Tgfb3Cxcr4Ccl3PdgfaTnfcnr1

Nonobstructed kidney

Interstitium Glomeruli

Cnr1–/– kidneyControl kidney

CB1

Cnr1

0.8

0.6

0.4

0.2

0.0Rel

ativ

e qu

antif

icat

ion

(AU

)

Con

cent

ratio

n (n

g/μg

tiss

ue)

Actin

Tubules

Obstructed kidney

Obs

truc

ted

kidn

ey

Cnr1–/– Obstructed kidney

20,000

15,000

10,000

5000

Anandamide 2-AG

* Nonobstructed kidney

Nonobstructed kidneys

*

Obstructed kidney

Obstructed kidneys

Brain

*0

Figure 1 |Cnr1 mRNA expression in the renal cortex is increased after 8 days of unilateral ureteral obstruction (UUO). (a) Heatmapsillustrating hierarchical clustering of significant transcripts expressed in UUO, for three wild-type (WT) mice nonobstructed (NO) andcontralateral-obstructed (O) kidneys. Colors represent the level of gene expression relative to the corresponding control level and mediancentered for each gene, with green, black, and red corresponding to lower, equal, and higher expression, respectively. (b) Expression of Cnr1mRNA evaluated by real-time quantitative PCR (RT-qPCR). *Po0.01 versus nonobstructed kidneys (n= 7, for each group). Data are mean± s.e.m.(n= 6–11). AU, arbitrary units. (c) Immunostaining for cannabinoid receptor 1 (CB1) in mouse kidneys. Bar scales = 100 μm. (d) Localizationof positive staining in the kidney cortex. (e) Western blot analysis of CB1 receptor expression. (f) Concentrations of anandamide and2-arachidonoylglycerol (2-AG) quantified in kidney and brain tissues by high-performance liquid chromatography (HPLC) analysis (232 ± 8 ng/μgof tissue in obstructed kidneys vs. 396 ± 16 ng/μg of tissue in nonobstructed kidneys for anandamide and 14,497 ± 2234 ng/μg of tissue inobstructed kidneys vs. 6543 ± 686 ng/μg of tissue in non-obstructed kidneys for 2-AG, n= 8 for each group, *Po0.05). Data are mean ± s.e.m.

Kidney International 3

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch

markers reflect myofibroblast number in the cortex, althoughspecific markers are lacking: indeed, these cells are highlyplastic by nature.7 Quantitative analysis showed that neither aspecific α-SMA nor S100A4 immunostaining was significantlydifferent during CB1 genetic disruption or CB1 pharmaco-logical blockade (Figure 6a–d). This suggests that the numberof myofibroblasts that express these markers was notmodified by the CB1 blockade.

Expression of TGF-β1 is not modified by a deficiency of CB1Transforming growth factor-β1 (TGF) is indisputably one ofthe key drivers of fibrosis. We quantified TGF-β1 and TGFBI(transforming growth factor-β-induced) mRNA expressionduring UUO. TGFBI is a target gene of TGF-β1 and correlateswith TGF-β1 activity in renal tissue.8 We found theirexpressions to be significantly greater during UUO. However,this increase was not modified by genetic or pharmacologicalinvalidation of CB1 (Figure 7a and b).

The role of the CB1 blockade in reducing renal fibrosisinvolves direct action on renal myofibroblastsWe next studied the role of CB1 in collagen synthesis using aprimary cell culture of myofibroblasts in vitro. Indeed, wehypothesized that CB1 acts downstream of myofibroblastdifferentiation. We found weak basal myofibroblast expres-sion of Cnr1 using RT-qPCR analysis and protein expressionin western blotting. CB1 expression was notably increasedafter TGF-β1 treatment (Figure 8a and b). In addition, wefound a significant amount of anandamide in culture super-natant that decreases after CB1 pharmacological blockade(Figure 8c). We found increased Col3a1 (encoding forCollagen III) expression after stimulation of myofibroblastsby TGF-β1 that was significantly blunted by rimonabant(Figure 8d). Similarly, Col3a1 expression remained lowand stable upon TGF-β1 stimulation in Cnr1− /− myofibro-blasts (Figure 8d).

We also analyzed the effect of CB1 antagonism on mono-cyte chemoattractant protein-1 (MCP-1) synthesis, a major

CB1

CB1 α-SMA Merge Magnification ×2

CB1 F4/80 Merge Magnification ×2

CB1 CD3 Merge Magnification ×2

Obs

truc

ted

kidn

eyO

bstr

ucte

d ki

dney

Obs

truc

ted

kidn

eyC

nr1–/

– ki

dney

Figure 2 |Cannabinoid receptor 1 (CB1) is expressed in α-smooth muscle actin (α-SMA)–positive cells, but not in infiltrating immunecells from obstructed kidneys in mice. Fluorescent coimmunostaining for CB1 receptor (green), and α-SMA, F4/80, or CD3 (red) in Cnr1− /−

and control obstructed kidneys. Bar scales = 200 μm. Arrows show myofibroblast cells.

4 Kidney International

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

Normal kidney

Normal kidney

Normal kidneyTu

bule

sIn

ters

titiu

mG

lom

erul

iV

esse

ls

IgA nephropathy

IgA nephropathy

******

IgA nephropathy

AIN

Acute interstitial nephritis

25

20

15

Pos

itive

are

a (%

)

10

5

0

Diabetic nephropathy

Normal kidney

+/–

+/–

++ ++ ++

++++

+

+

+ +0

Tubules

Glomeruli

Interstitium

IgA nephropathyAcute interstitial

nephritisDiabetic

nephropathy

Diabetic nephropathy

Diabetic nephropathyAIN

Figure 3 |Cannabinoid receptor-1 (CB1) expression is increased in multiple human nephropathies. (a) Immunostaining for the CB1receptor in control kidneys, IgA nephropathy, acute interstitial nephritis (AIN), or diabetic nephropathy. (b) Localization and intensity of positiveareas. (c) Quantification of positive areas by histomorphometry. *Po0.05, **Po0.05, ***Po0.05 versus a control kidney. Data are mean± s.e.m.(n= 6 for diabetic nephropathy, n= 5 for the other groups). Bar scales = 100 μm.

Kidney International 5

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch

WT

WT

Cnr1–/–

Cnr1–/–

Cnr1–/–10

8

6

Rel

ativ

e qu

antif

icat

ion

for

Col

3a1(

AU

)

Rel

ativ

e qu

antif

icat

ion

for

Col

1a2(

AU

)

4

2

00

50

100

150

Cnr1–/–

WT

WTWT

Cnr1–/–

Vehicle

###

Rimonabant 50 mg/kg

1.5

1.0

0.5

0.0

1.5

1.0

0.5

0.0 Col

lage

n III

pro

tein

expr

essi

on(n

orm

aliz

ed to

β-a

ctin

)

Col

lage

n III

pro

tein

expr

essi

on(n

orm

aliz

ed to

β-a

ctin

) ControlRimonabant 10 mg/kg

Rimonabant

RimonabantVehicleRimonabant

***

Nonobstructed Nonobstructed

Collagen III Collagen III

β-Actin β-Actin

Obstructed Obstructed

0 10 500 10

12

50

20

15

10

Fib

rosi

s ar

ea (

%)

5

0

WT

WT KO

Cnr1–/–

Nonobstructed

Rimonabant

Rimonabant 50 mg/kg per day

###***

##***

#**

*

**

*

VehicleRimonabant 10 mg/kg per day

Rimonabant 50 mg/kg per day

VehicleRimonabant 10 mg/kg per day

Control

Non

obst

ruct

ed

Obstructed

Nonobstructed Obstructed Nonobstructed

Fib

rosi

s ar

ea (

%)

Fib

rosi

s ar

ea (

%)

Obstructed

Obs

truc

ted

Non

obst

ruct

edO

bstr

ucte

d

#**

*

15

10

5

0

15

10

5

0

Figure 4 |Both cannabinoid receptor 1 (CB1) genetic disruption and pharmacological blockade reduce fibrosis during unilateralureteral obstruction (UUO). Representative sections, stained with Sirius red, of kidney (original magnification × 10) from (a) nonobstructed andobstructed kidneys from Cnr1− /− and wild-type (WT) mice after 8 days of UUO, and (c) of nonobstructed and obstructed kidneys fromrimonabant-treated and vehicle-treated mice after 8 days of UUO. KO, knockout. (b) Quantification of fibrosis in the renal cortex bymorphometry. Genetic disruption of the CB1 receptor reduced renal fibrosis during UUO: *Po0.05 versus nonobstructed kidneys from WT mice(12.90 ± 0.94% in WT mice vs. 8.50 ± 0.81% in Cnr1−/− mice), **Po0.05 versus nonobstructed kidneys from Cnr1− /− mice, and #P o0.01 versusobstructed kidneys from WT mice. (d) Pharmacological blockade by 10 or 50 mg/kg/day of rimonabant reduced renal fibrosis during UUO.*Po0.05 versus nonobstructed kidneys from vehicle-treated mice, **Po0.05 versus nonobstructed kidneys from rimonabant (10 mg/kg/day)-treated mice, ***Po0.05 versus nonobstructed kidneys from rimonabant (50 mg/kg/day)-treated mice, #Po0.01 and ##Po0.01 versusobstructed kidneys from vehicle-treated mice (9.62 ± 1.82% in vehicle-treated compared with 4.25 ± 0.61% and 4.28 ± 0.48% in rimonabant-treated mice at 10 or 50 mg/kg/day, respectively). (e) Pharmacological blockade by 10 or 20 mg/kg/day of AM6545 peripherally limited selectiveantagonism and reduced renal fibrosis. *Po0.05 versus nonobstructed kidneys from vehicle-treated mice, **Po0.05 versus nonobstructedkidneys from mice treated with AM6545 at 10 mg/kg/day, ***Po0.05 versus nonobstructed kidneys from mice treated with AM6545 at 20 mg/kg/day, #Po0.01 and ##Po0.01 versus obstructed kidneys from vehicle-treated mice (9.04 ± 0.75% in vehicle-treated versus 6.77 ± 0.38% and5.33 ± 0.72% in AM6545-treated mice, 10 or 20 mg/kg/day, respectively). (f) Western blot analysis of collagen III protein in kidney cortexes. Barcharts represent quantifications of the target protein expression normalized to β-actin. (g) Expression of Col1a2 and Col3a1 in obstructedkidneys evaluated by real-time quantitative PCR (RT-qPCR). *Po0.05 and #Po0.05 versus Cnr1− /− mice, **Po0.05 and ##Po0.05 versusrimonabant-treated mice. Data are mean± s.e.m. (n= 11 for Cnr1− /− mice and n= 6 for WT mice, n= 7 for all other groups). AU, arbitrary units.Bar scales = 100 μm.

6 Kidney International

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

inflammatory cytokine chemoattractant for macrophagesand produced by myofibroblasts, using enzyme-linkedimmunosorbant assay quantification. As expected, we foundincreased synthesis of MCP-1 after TGF-β1 stimulation thatwas significantly decreased after rimonabant treatment(Figure 8d).

We also examined the role of CB1 in tubules during UUO.We found no major difference in tubular dilation during CB1pharmacological blockade and a slight decrease in Crn1− /−

mice. Next, experiments carried out in HK-2 cells in vitrofound that CB1 may not have a major role in collagensynthesis or in epithelial to mesenchymal transition (Supple-mentary Figure S6 online).

Macrophage infiltration during UUO is modulated byrimonabant but not by CB1 blockadeInflammation is a key process during UUO and triggers thefirst steps of renal fibrogenesis.9 Because we found decreasedMCP-1 synthesis in rimonabant-treated myofibroblasts,

we performed semiquantitative analysis on cells that hadinfiltrated the kidney cortex during UUO by immuno-staining with T-lymphocyte (CD3) and macrophage (F4/80)markers (Figure 9a). UUO significantly increased the stainingof both CD3 and F4/80 in the kidney (data not shown).CB1 blockade had no effect on CD3-positive staining,whereas F4/80+ cell infiltration was significantly decreasedduring UUO.

We performed RT-qPCR analysis on selected M1- andM2-associated markers (Figure 9c–g) in mRNA from thekidney cortex. Both M1 (iNOS, MCP-1, and IL-23) and M2(Arg1 and Mrc1) markers were increased in obstructedkidneys. M1, but not M2, markers were reduced by rimona-bant treatment during UUO compared with vehicle-treated mice.

These results regarding macrophage infiltration duringrimonabant treatment were not found in Cnr1− /− mice orduring CB1 blockage using AM6545. As expected, kidneycortex MCP-1 mRNA was not modified during UUO inboth Cnr1− /− and AM6545-treated mice (SupplementaryFigure S5 online). Thus, rimonabant may act on macrophagesthrough noncanonical pathways.

DISCUSSIONOur study provides strong evidence, for the first time, for amajor role of CB1 in the development of renal fibrosis: wehave provided data from several human nephropathies, aswell as from the UUO model in mice.

The fact that the expression of Cnr1 was found among themost upregulated genes in our microarray analysis duringUUO was a first clue to its possible involvement in renalfibrogenesis. In addition, we found a very high increase in2-AG, the most abundant CB1 ligand, during UUO. More-over, CB1 expression is highly upregulated during UUO in thekidney tubules and podocytes, but also, more interestingly, ininterstitial myofibroblasts. We found no expression of CB1 inimmune-infiltrating cells in the kidney during UUO, inagreement with the literature on mice.10 The prominentexpression of CB1 in renal myofibroblasts strongly suggests amajor role for the cannabinoid system in the development ofrenal fibrosis. We also found, for the first time, an importantincrease in CB1 expression in human nephropathies that seemto correlate with kidney function, whereas CB1 expression innormal kidneys was very low, as previously reported.4 We havealso demonstrated, for the first time, that both geneticinvalidation of Cnr1 and CB1 pharmacological blockadedrastically decreased the development of fibrosis in mice duringUUO, proving the role of CB1 in the develop-ment of renal fibrosis. Moreover, genetic disruption and theCB1 pharmacological blockade were performed on twodifferent genetic backgrounds, and this makes the role of CB1in renal fibrogenesis even more relevant, as mice pheno-types are highly dependent on genetic background.11,12

We reproduced the reduction in renal fibrosis during UUOusing both rimonabant and a peripherally restricted CB1antagonist AM6545. This is an important result, as the use of

15

Obstructed kidneys

Nonobstructed kidneys

10

5

0

Rel

ativ

e qu

antif

icat

ion

(AU

)

Vehicle Rimonabant WT Cnr1–/–

§§§

***

#

##

15

10

5

0

Fib

rosi

s ar

ea (

%)

ObstructedNonobstructed

§§§§

**#

#####

§§

****

†††

†††

Control

Rimonabant

SR144528 + vehicle

SR144528 + rimonabant

JWH133 + vehicle

JWH133 + rimonabant

Figure 5 | Expression of Cnr2 mRNA in the renal cortex, andeffects of cannabinoid receptor 2 (CB2) agonism and antagonismassociated with CB1 blockade in renal fibrosis after 8 days ofunilateral ureteral obstruction (UUO). (a) *Po0.01 versusnonobstructed kidneys from vehicle-treated mice, **Po0.05 versusnonobstructed kidneys from rimonabant-treated mice, #Po0.01versus nonobstructed kidneys from wild-type (WT) mice, ##Po0.01versus nonobstructed kidneys from Cnr1− /− mice, §Po0.01 versusnonobstructed kidneys from WT mice, §§Po0.05 versusnonobstructed kidneys from vehicle-treated mice. AU, arbitrary units.(b) Quantification of fibrosis in the cortical area by morphometry.*Po0.01, **Po0.01, ***Po0.01, #Po0.01, ##Po0.01, ###Po0.01 versusnonobstructed kidneys from associated-treatment mice. §Po0.05,§§Po0.05, §§§Po0.05, †Po0.05, ††Po0.05 versus obstructed kidneysfrom vehicle-treated mice. †††Po0.01 obstructed kidneys fromSR144528 + vehicle-treated mice versus SR144528 + rimonabant-treated mice. Data are mean± s.e.m. (n=11 for Cnr1− /− mice andn= 11 for WT mice, n= 7 for all other groups).

Kidney International 7

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch

rimonabant in humans is limited because of its adverse effectson the central nervous system (it was withdrawn from themarket in 2008 (see Chorvat13) despite its positive results onmetabolic syndrome). Therefore, AM6545 could be usedwithout the drawback of having adverse effects on the centralnervous system. The proportional decrease in renal fibrosisduring both genetic and pharmacological blockade of CB1 isextremely important, and it is comparable to that usually foundduring the blockade of the renin–angiotensin system duringUUO.14 This also suggests that CB1 may be a major pathwayinvolved in fibrogenesis, as it is in other organs in metabolic15

and nonmetabolic diseases,14,15,16 although the complete UUOmodel may not completely be transferred into humanpathology.

Because CB1 is mostly implicated in the regulation of foodintake and glucose homeostasis,17 targeting CB1 with apharmacological blockade has been previously investigated instudies on diabetes and obesity.2,3 In addition to reducingbody weight and metabolic parameters, the CB1 blockadedecreases albuminuria during metabolic syndrome andduring the early stages of diabetic nephropathies in mice andrats,3,5 possibly through both hemodynamic effects and a

direct action on the podocyte itself.18 However, in diabeticmice, kidney lesions were absent5 or faint.18 Indeed, so far,none of the experimental models conducted on diabeticnephropathies in mice have been fully able to explore theovert renal fibrogenesis induced by diabetes in humandiabetic kidneys.11

Many reports suggest that CB2, which is mainly expressedin immune cells and less so in the central nervous system,may also be involved in the development of fibrosis. Contraryto CB1, CB2 disruption or pharmacological blockade hasbeen reported to promote liver, cardiac, and skin fibrosis,whereas CB2 agonists decrease fibrogenesis,6,19,20 Therefore,the blockade of renal fibrosis with CB1 invalidation duringUUO may have been caused by the unbalanced actions ofCB2; we also found increased expression of CB2. Interest-ingly, treatment with the CB2 agonist reduced fibrosis duringUUO, whereas the CB2 antagonist aggravated renal lesions. Inaddition, the role of CB2 agonists in decreasing renal fibrosishas not been published (as yet), although the CB2 pathwayhas a role in decreasing albuminuria during diabetes.21

We also found that the CB2 antagonist aggravates renalfibrosis during UUO. This result is in accordance with the

15

Pos

itive

are

a (%

)

10

5

0

20

Obstructed

ObstructedObstructed

S100A4

S10

0A4

WT

WT KO

Cnr1–/–

α-SMAα-

SM

AS

100A

4

Control

Vehicle

Rimonabant

Rimonabant

Vehicle

AM6545

α-S

MA

α-SMAα-SMAP

ositi

ve a

rea

(%)

S100A4

2.5

2.0

1.5

1.0

0.5

0

20

15

10

5

0

Figure 6 | Expression of myofibroblast markers in Cnr1− /− and rimonabant-treated mice with obstructed kidneys. Geneticdisruption and pharmacological blockade do not modify the mesenchymal markers in the renal cortex during unilateral ureteral obstruction(UUO). (a, b) Immunostaining for α-smooth muscle actin (α-SMA) and S100A4 in the renal cortex at 8 days after UUO in wild-type (WT), Cnr1− /− ,vehicle- and rimonabant-treated mice, and vehicle- and AM6545-treated mice. (c, d) Quantification of the positive area by histomorphometry.Data are mean± s.e.m. (n=11 for Cnr1− /− mice and n= 6 for WT mice, n=7 for all other groups). KO, knockout. Bar scales = 100 μm.

8 Kidney International

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

work by Barutta et al.,22 who found direct participation ofpodocytes expressing CB2 during diabetic nephropathy inCB2 knockout mice. Moreover, renal fibrosis was reduced ina similar proportion during rimonabant treatment alone andduring the association of rimonabant with a specific CB2antagonist. Therefore, although the CB2 pathway alone seemsto be involved in renal fibrosis, CB2 signaling is not necessaryfor mediating the antifibrotic effect of the CB1 pharmacolo-gical blockade. Similarly, adding a CB2 agonist duringrimonabant treatment did not further decrease the develop-ment of renal fibrosis during UUO, and this rules out anyadditive or potentialization effect in the kidney, at least inour model.

As our report provides solid evidence for the majorimplications of CB1 in renal fibrosis, we also attempted todecipher the exact mechanisms involved in reducing renalfibrosis. First, we explored the TGF-β1 pathway. TGF-β1promotes myofibroblast activation or differentiation7,23 and isa major profibrotic cytokine in renal fibrosis, particularly inthe UUO model.24 We did not find any difference in TGF-β1expression or in TGF-β1 target gene expression when CB1was invalidated by either a pharmacological blockade or

genetic disruption. This suggests that the activity of CB1 inour model took place downstream of TGF-β1, that is, onTGF-β1 target cells such as myofibroblasts.

Although its precise origin is still highly controversial,21,24

the myofibroblast remains the main effector cell during renalfibrogenesis. During UUO in mice, we were able todemonstrate that cells expressing interstitial α-SMA, that is,myofibroblasts, also strongly expressed CB1. However, wewere unable to demonstrate a decrease in cells that expressedα-SMA or S100A4 in the CB1 blockade during UUO.Myofibroblasts are highly plastic cells, and not all myofibro-blasts express these markers, although they may still beinvolved in collagen synthesis.8 Nevertheless, it is more likelythat the CB1 blockade does not alter myofibroblastdifferentiation in vivo. We next used a primary cell cultureof kidney myofibroblasts and found that rimonabant bluntedthe increased Col3 expression induced by TGF-β1. Similarly,we found no increase in collagen Col3 synthesis in Cnr1− /−

myofibroblasts upon TGF-β1 stimulation. Thus, CB1 isnecessary for extracellular matrix protein synthesis in renalfibrosis, although its absence does not seem to altermyofibroblast differentiation in vivo. This result is expectedgiven that undifferentiated myofibroblasts express veryweakly, if not at all, the CB1 receptor. Although CB1 isexpressed in tubules in UUO, it does not seem to have amajor role in epithelial to mesenchymal transition in vitro.

Inflammation is a key mediator of renal fibrogenesis in theUUO model, and renal fibrosis can be blunted whenmacrophage and T-cell recruitment is reduced.25,26 It is stillcontroversial whether there is significant CB1 expression inimmune mouse cells, as some reports show weak27 or noexpression.10 However, we found no significant expression ofCB1 in CD3 lymphocytes or F4/80 macrophages duringUUO. Nonetheless, there was a decrease in MCP-1 secretionby activated myofibroblasts in vitro during the CB1 blockadeby rimonabant, and also a decreased number of M1-typemacrophages during UUO, similar to that reported in theliterature during atherosclerosis.2 This could also account forthe decreased fibrosis by rimonabant, as the role of M1-typemacrophages promote renal fibrosis.28 However, this is mostlikely a noncanonical effect of rimonabant as macrophageinfiltration and MCP-1 transcription are not altered duringAM6545 treatment or in Cnr1− /− mice in kidney cortex inUUO. Therefore, the antifibrotic effect of rimonabant couldalso act through other receptors. Moreover, during CB1blockade, endocannabinoids, which we found increasedduring UUO, could bind to receptors other than CB2, suchas TRPV1. This hypothesis is in accordance with the work byWang and Wang29 who found that TRPV1− /− mice haveexaggerated renal damage in deoxycorticosterone acetate–salthypertension. Overall, our work demonstrates that the mainmechanism by which CB1 promoted renal fibrogenesis was itsdirect action on renal myofibroblast function. Thus, CB1 maybe a major player in renal fibrosis, regardless of the initialrenal injury: thus, it has promising therapeutic potential, aswell as a good cardiovascular profile.

4000

3000

2000

1000

0Rel

ativ

e qu

antif

icat

ion

(AU

)

TGF-β1 TGFBI

#

##

***

2500

2000

1500

1000

500

0Rel

ativ

e qu

antif

icat

ion

(AU

)

TGF-β1 TGFBI

###

Vehicle nonobstructedRimonabant nonobstructedVehicle obstructedRimonabant obstructed

WT nonobstructedCnr1–/– nonobstructed

Cnr1–/– obstructedWT obstructed

***

Figure 7 | Expression of transforming growth factor-β1 (TGF-β1)and TGFBI mRNA in the renal cortex evaluated by real-timequantitative PCR (RT-qPCR). Data are mean± s.e.m. (a) *Po0.05versus nonobstructed kidneys from wild-type (WT) mice, **Po0.05versus nonobstructed kidneys from Cnr1− /− mice for TGF-β1,#Po0.05 versus nonobstructed kidneys from WT mice, ##Po0.05versus nonobstructed kidneys from Cnr1− /− mice for TGFBI. AU,arbitrary units. (b) *Po0.05 versus nonobstructed kidneys fromvehicle-treated mice, **Po0.05 versus nonobstructed kidneys fromrimonabant-treated mice for TGF-β1, #Po0.01 versus nonobstructedkidneys from vehicle-treated mice, ##Po0.01 versus nonobstructedkidneys from rimonabant-treated mice for TGFBI. Data aremean± s.e.m. (n= 11 for Cnr1− /− mice and n=6 for WT mice,n= 7 for all other groups).

Kidney International 9

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch

0.25

0.20

0.15

0.10

Rel

ativ

e qu

antif

icat

ion

(AU

)

0.05

0.00

0.00

0.02

0.04

70

55CB1

Actin

0

+/+ –/–Myofibroblasts Cnr1

TGF-β1 (ng/ml) 05

AnandamideAnandamide

*

Con

cent

ratio

n (n

g/m

l)

Con

cent

ratio

n (n

g/m

l)

0.06

0.00

0Col3a1 Col3a1

MCP-1

###

*

2.0

1.5

1.0

0.5

0TGF-β15 ng/ml

– – – + + +

Nor

mal

ized

con

cent

ratio

ns (

AU

)

0.00

0.01

0.02

0.03

0.04

**

*

1

2

3

4

Rel

ativ

e qu

antif

icat

ion

(AU

)

Rel

ativ

e qu

antif

icat

ion

(AU

)

0.02

0.04

0.06

Cnr1

***

Basal myofibroblasts

Myofibroblasts + TGF-β1 5 ng/ml 24 h

Myofibroblasts + TGF-β1 5 ng/ml 48 h

Basal myfibroblasts Cnr1–/– + DMSO

Myofibroblasts Cnr1–/– + TGF-β1 5ng/ml 1h + DMSO

Myofibroblasts Cnr1–/– + TGF-β1 5ng/ml 1h + rimonabant 1 μM

Basal myofibroblasts Cnr1–/– + DMSO

Myofibroblasts Cnr1–/– + TGF-β1 5 ng/ml 1 h + DMSO

Myofibroblasts Cnr1–/– + TGF-β1 5 ng/ml 1 h + rimonabant 1 μM

ControlRimonabant 1 μM

AM6545 1 μM

Basal myofibroblasts + DMSO

Myofibroblasts + TGF-β1 5 ng/ml 1 h + DMSO

Myofibroblasts + TGF-β1 5 ng/ml 1 h + rimonabant 1 μM

Basal myofibroblasts + DMSO

Myofibroblasts + TGF-β1 5 ng/ml 1 h + DMSO

Myofibroblasts + TGF-β1 5 ng/ml 1 h + rimonabant 1 μM

Figure 8 | The role of cannabinoid receptor 1 (CB1) blockade in reducing renal fibrosis involves a direct action on renal myofibroblasts.Data are mean ± s.e.m. (a) Relative quantification of Cnr1 expression in the primary culture of mouse kidney myofibroblasts. *Po0.05, **Po0.05versus basal myofibroblasts. AU, arbitrary units. (b) Transforming growth factor-β1 (TGF-β1) increases CB1 receptor expression (60 kDa) in renalmyofibroblasts after 48 h of treatment. Protein expression was evaluated by western blot analysis. As expected, CB1 receptor expression was notdetected in myofibroblasts isolated from Cnr1− /− obstructed kidneys. (c) Quantification of anandamide in primary myofibroblast supernatant,isolated from C57BL/6 or Cnr1− /− mice. *Po0.05 versus myofibroblasts treated with TGF-β1 (5 ng/ml, 1 h). 2-arachidonoylglycerol (2-AG) wasnot detectable (o2 ng/ml). DMSO, dimethyl sulfoxide. (d) Relative quantification of Col3a1 expression in the primary culture of mouse kidneymyofibroblasts. *Po0.05 versus basal fibroblasts +DMSO, **Po0.05 versus fibroblasts + TGF-β1 5 ng/ml 1 h+DMSO. Col3a1 expression was notsignificantly modified by TGF-β1 treatment with or without rimonabant on primary myofibroblasts isolated from Cnr1− /− mice. (e) CB1pharmacological blockade reduces monocyte chemoattractant protein-1 (MCP-1) synthesis in renal myofibroblasts after 48, 72, and 96 h ofTGFβ1 treatment. MCP-1 was quantified by enzyme-linked immunosorbant assay (ELISA). Experiments were repeated three times. Results arenormalized over the mean for each experiment. *Po0.05 versus control myofibroblasts not treated with TGF-β1, #Po0.05 versus myofibroblaststreated with rimonabant, ##Po0.05 versus myofibroblasts treated with AM6545.

10 Kidney International

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

MATERIALS AND METHODSMicroarray analysesThe hybridization procedure was performed according to the Agilent60-mer oligo microarray processing protocol using an Agilent GeneExpression Hybridization Kit (Agilent Technologies, Santa Clara, CA).Agilent Feature Extraction Software was used to read out and process themicroarray image files. Genes were considered to be differentiallyexpressed if the Significance Microarray Analysis algorithm passed a falsediscovery rate of o0.01 and a fold change of 42. Functional analyses ofthe data were performed using the DAVID (The Database for Annotation,Visualization and Integrated Discovery) (v6.7) bioinformatics resource.Microarray data were deposited in GEO DataSets under accession numberGSE38117. See Supplementary Materials and Methods online.

DrugsRimonabant was a gift from Sanofi-Aventis R&D (Montpellier, France)and AM6545 was purchased from Sigma-Aldrich (Saint Louis, MO).

JWH133 (Tocris Bioscience, Bristol, UK) and SR144528 (CaymanChemicals, Ann Arbor, MI) were prepared in phosphate-bufferedsaline containing 10% dimethyl sulfoxide and 0.1% Tween-80.JWH133 and SR144528 were administered at doses of 1mg/kg/daysubcutaneously and 3mg/kg/day intraperitonealy, respectively. All Ki

values are given in Supplementary Figure S1 online.

AnimalsMice, genetically invalidated for CB1 (Cnr1−/−) and wild-typemice (Cnr1+/+) (CD1 background) were obtained from Dr C Ledent(Université Libre de Bruxelles, Brussels, Belgium) and gene-rated as previously described30 (also see Supplementary Materialsand Methods online). All mice were housed under stan-dardized conditions with a 12-h dark/light cycle, and unlimitedaccess to food and water. All animal procedures wereapproved by the animal ethics advisory committee of Paris SudUniversity.

20

15

10

5

0

CD3

MCP-1IL-23iNOS

Vehicle

Vehicle

20 4

3

2

1

0

400

300

200

100

00

2

4

6

8

0

2000

4000

6000

Nonobstructed Obstructed

#

#

******

Rel

ativ

e qu

antif

icat

ion

(AU

)

Rel

ativ

e qu

antif

icat

ion

(AU

)

15

10

5

0

Rel

ativ

e qu

antif

icat

ion

(AU

)R

elat

ive

quan

tific

atio

n (A

U)

Rel

ativ

e qu

antif

icat

ion

(AU

)

Control

CD

3

Rimonabant 10 mg/kg day

Rimonabant

Rimonabant VehicleRimonabant

VehicleRimonabant

VehicleRimonabant

#

*VehicleRimonabant

Arg1 Mrc1

F4/80

F4/

80

*

Obstructed

Pos

itive

are

a (%

)

Nonobstructed Nonobstructed

Nonobstructed

Obstructed Obstructed

Nonobstructed ObstructedObstructed

#*

#*

Figure 9 |Characterization of kidney cortex infiltration during unilateral ureteral obstruction (UUO). Rimonabant-treated mice showedreduced infiltration of macrophages during UUO. Conversely, CD3 lymphocyte infiltration was not modified. (a) Immunostaining for CD3 andF4/80 in the renal cortex at 8 days after UUO. (b) Quantification of the positive area by histomorphometry. *Po0.05 versus obstructed kidneysfrom vehicle-treated mice. Relative quantification for (c) inducible nitric oxide synthase (iNOS): *Po0.05 versus nonobstructed kidneys fromvehicle-treated mice, #Po0.05 versus nonobstructed kidneys from rimonabant-treated mice. Relative quantification for (d) interleukin-23 (IL-23):*Po0.01 versus nonobstructed kidneys from vehicle-treated mice, #Po0.01 versus nonobstructed kidneys from rimonabant-treated mice,**Po0.05 obstructed kidneys from vehicle-treated mice versus rimonabant-treated mice. Relative quantification for (e) monocytechemoattractant protein-1 (MCP-1): *Po0.01 versus nonobstructed kidneys from vehicle-treated mice, #Po0.01 versus nonobstructed kidneysfrom rimonabant-treated mice, **Po0.05 obstructed kidneys from vehicle-treated mice versus rimonabant-treated mice. Relative quantificationfor (f) Arg1: *Po0.01 versus nonobstructed kidneys from vehicle-treated mice, #Po0.01 versus nonobstructed kidneys from rimonabant-treatedmice. Relative quantification for (g) Mrc1: *Po0.01 versus nonobstructed kidneys from vehicle-treated mice, #Po0.01 versus nonobstructedkidneys from rimonabant-treated mice. Data are means ± s.e.m. (n=7). AU, arbitrary units. Bar scales = 100 μm.

Kidney International 11

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch

Endocannabinoid quantificationEndocannabinoids were quantified in kidneys, brain, and myofibro-blast culture supernatants with liquid chromatography coupled toFourier transform mass spectrometry (see Supplementary Materialsand Methods online).

Unilateral ureteral obstructionAfter giving general anesthesia (isoflurane 2%), the left ureter wasexposed via a mid-abdominal incision. UUO was performed withdouble silk sutures by complete ligation of the left ureter. Thecontralateral kidney served as an intraindividual control. Animalswere killed on day 8 after treatment.

Histopathological analysis of renal fibrosisKidneys were fixed in 10% formalin and embedded in paraffin.Tissue sections (4 μm) were stained with Picrosirius red. Renalcortex fibrosis was quantified using computer-based morphogenicanalysis software (TRIBVN CaloPix, Chatillon, France) in a blindedmanner. The total kidney cortex of an entire kidney sagittal cross-section was selected for quantification from each mouse, represent-ing ∼ 7 mm2 surface area. The positive area was expressed as apercentage of the total cortical kidney section.

ImmunohistochemistryParaffin-embedded kidney tissues were stained with an anti-α-SMA(Abcam, Cambridge), anti-CD3 (DakoCytomation, Glostrup, Denmark),

anti-S100A4 (Abcam, Cambridge, UK), anti-F4/80 (Abcam), andanti-CB1 and CB2 (Abcam). Staining was quantified by a morpho-metry software.

Fluorescent immunostaining was revealed with anti-rabbit (AlexaFluor 488, Jackson Immunoresearch, West Grove, PA) or anti-ratantibodies (DyLight 549, Jackson Immunoresearch). The slides wereexamined by confocal laser microscopy (Leica TCS SP5 AOBSTandem Confocal System, Wetzler, Germany). See SupplementaryMaterials and Methods online.

Human kidney specimensParaffin sections of renal biopsies from patients were retrospectivelyanalyzed. Informed written consent was given by the patients for theuse of part of the biopsy for scientific purposes. All procedures andthe use of tissues were performed in accordance with the Declarationof Helsinki principles.

Real-time quantitative PCR (RT-qPCR)RNA was extracted from tissues or cells using a QIAGEN RNeasyMini kit, according to the manufacturer’s instructions (QIAGEN,Venlo, Netherlands). The cDNA was synthesized using a Revert AidH minus First-Strand DNA Synthesis kit (Fermentas, Villebon-sur-Yvette, France) and amplified by PCR with a LightCycler 480 (RocheDiagnostics, Meylan, France). See Supplementary Materials andMethods online and Table 1 for primers sequences.

Cell culturePrimary cultures of mouse kidney myofibroblasts were isolatedfrom C57/BL6 mice, as previously described.31 See SupplementaryMaterials and Methods online. HK-2 cells were purchased fromAmerican Type Culture Collection (ATCC, Manassas, VA). SeeSupplementary Materials and Methods online.

Enzyme-linked immunosorbant assayThe quantitative determination of mouse MCP-1 concentrationsin primary myofibroblast supernatants was performed using aQuantikine enzyme-linked immunosorbant assay CCL2/MCP-1 kit(R&D Systems, Minneapolis, MN) according to the manufacturer’sinstructions.

Western blottingCells were lysed in NP-40 buffer containing protease inhibitors(Roche, Mannheim, Germany). Proteins, separated in a 12% sodiumdodecyl sulfate–polyacrylamide gel electrophoresis, were transferred to apolyvinylidene fluoride membrane and then incubated with primaryantibodies. See Supplementary Materials and Methods online.

Statistical analysesWe compared differences using the Mann–Whitney U-test or thepaired Student’s t-test, as appropriate. All analyses were performedusing GraphPad Prism 5.0 software (La Jolla, CA). Data wereconsidered statistically significant when the P-value was o0.05. Alldata are expressed as mean± s.e.m.

DISCLOSUREAll the authors declared no competing interests.

ACKNOWLEDGMENTSWe thank Aurélie Perier, Séverine Lecourt, and Djenaba Ba for theirtechnical support and scientific advice, and Xavier Biolchini from the

Table 1 |Primers used for RT-qPCR

mRNA Strand Sequence

Arg1 Sense 5′-CCTGAAGGAACTGAAAGGAAAG-3′Antisense 5′-TTGGCAGATATGCAGGGAGT-3′

Cnr1 Sense 5′-GGGCAAATTTCCTTGTAGCA-3′Antisense 5′-GGCTCAACGTGACTGAGAAA-3′

Cnr2 Sense 5′-CAGCTCTTGGGACCTACGTG-3′Antisense 5′-TGTCCCAGAAGACTGGGTGT-3′

GusB Sense 5′-CTCTGGTGGCCTTACCTGAT-3′Antisense 5′-CTCAGTTGTTGTCACCTTCACC 3′

IL-23 Sense 5′-TCCCTACTAGGACTCAGCCAAC-3′Antisense 5′-TGGGCATCTGTTGGGTCT-3′

iNOS Sense 5′-GGGCTGTCACGGAGATCA-3′Antisense 5′-CCATGATGGTCACATTCTGC-3′

MCP-1 Sense 5′-GTTGGCTCAGCCAGATGCA-3′Antisense 5′-AGCCTACTCATTGGGATCATCTTG-3′

MRC1 Sense 5′-CCACAGCATTGAGGAGTTTG-3′Antisense 5′-ACAGCTCATCATTTGGCTCA-3′

TGF-β1 Sense 5′-TGGAGCAACATGTGGAACTC-3′Antisense 5′-GTCAGCAGCCGGTTACCA-3′

TGFBI Sense 5′-GAGCTGCTTATCCCAGATTCA-3′Antisense 5′-GGCAGTGGAGACGTCAGATT-3′

Col3a1 Sense 5′-TCCCCTGGAATCTGTGAATC-3′Antisense 5′-TGAGTCGAATTGGGGAGAAT-3′

Col1a2 Sense 5′-GCAGGTTCACCTACTCTGTCCT-3′Antisense 5′-CTTGCCCCATTCATTTGTCT- 3′

CNR1 Sense 5′-ATCCTAGATGGCCTTGCAGA-3′Antisense 5′-GTTGTCTCCCGCAGTCATCT-3′

COL4A3 Sense 5′-CACGGGTTCCAAAGGTGTAA-3′Antisense 5′-AGTCCGTAAGGCCCGGTAT-3′

VIMENTIN Sense 5′-TGGCCGACGCCATCAACACC-3′Antisense 5′-CACCTCGACGCGGGCTTTGT-3′

Abbreviations: iNOS, inducible nitric oxide synthase; IL-23, interleukin-23; MCP-1,monocyte chemoattractant protein-1; RT-qPCR, real-time quantitative PCR; TGF-β1,transforming growth factor-β1; TGFBI, transforming growth factor-β-induced.

12 Kidney International

bas i c resea rch L Lecru et al.: CB1 receptor and renal fibrosis

animal facility. This work was partially supported by Sanofi-Aventis,NRB Vaincre le Cancer, and Institut Francilien de Recherche enNéphrologie et Transplantation. Rimonabant was a gift fromSanofi-Aventis.

SUPPLEMENTARY MATERIALFigure S1. Comparative table of the different Ki of CB1 and CB2ligands, pharmacological agonists and antagonists.Figure S2. Microarray analysis of the expression of genes involved inthe regulation of anandamide and 2-AG synthesis by unsupervisedmultivariate methods.Figure S3. Linear regression analysis of serum creatinine and CB1expression quantified by immunohistochemistry (IgA nephropathy,AIN or diabetic nephropathy, normal kidneys n=18, po0.01, Pearson).Figure S4. Immunostaining of CB2 receptors in mouse kidneyshowing positive staining mostly in the interstitium (myofibroblastsand immune cells), compared to a low level in glomeruli (mainlypodocytes) and tubules in obstructed kidneys.Figure S5. Infiltration of T lymphocytes and macrophages inobstructed kidneys is not modified neither by genetic disruption norby AM6545 treatment.Figure S6. (A) Quantification of tubular dilation during UUO.Figure S7. Recapitulative scheme illustrating the hypothetic role ofCB1 and CB2 receptors during renal fibrosis induced by UUO.Supplementary material is linked to the online version of the paper athttp://www.nature.com/ki

REFERENCES1. Bascands JL, Schanstra JP. Obstructive nephropathy: insights from

genetically engineered animals. Kidney Int 2005; 68: 925–937.2. Dol-Gleizes F, Paumelle R, Visentin V et al. Rimonabant, a selective

cannabinoid CB1 receptor antagonist, inhibits atherosclerosis in LDLreceptor-deficient mice. Arterioscler Thromb Vasc Biol 2008; 29: 12–18.

3. Janiak P, Poirier B, Bidouard JP et al. Blockade of cannabinoid CB1receptors improves renal function, metabolic profile, and increasedsurvival of obese Zucker rats. Kidney Int 2007; 72: 1345–1357.

4. Deutsch DG, Goligorsky MS, Schmid PC et al. Production and physiologicalactions of anandamide in the vasculature of the rat kidney. J Clin Invest1997; 100: 1538–1546.

5. Barutta F, Corbelli A, Mastrocola R et al. Cannabinoid receptor 1 blockadeameliorates albuminuria in experimental diabetic nephropathy. Diabetes2010; 59: 1046–1054.

6. Julien B, Grenard P, Teixeira-Clerc F et al. Antifibrogenic role of thecannabinoid receptor CB2 in the liver. Gastroenterology 2005; 128:742–755.

7. Boor P, Floege J. The renal (myo-)fibroblast: a heterogeneous groupof cells. Nephrol Dial Transplant 2012; 27: 3027–3036.

8. Mesnard L, Keller AC, Michel M-L et al. Invariant natural killer T cells andTGF-beta attenuate anti-GBM glomerulonephritis. J Am Soc Nephrol 2009;20: 1282–1292.

9. Bascands J-L, Schanstra JP. Obstructive nephropathy: insights fromgenetically engineered animals. Kidney Int 2005; 68: 925–937.

10. Han KH, Lim S, Ryu J et al. CB1 and CB2 cannabinoid receptorsdifferentially regulate the production of reactive oxygen species bymacrophages. Cardiovasc Res 2009; 84: 378–386.

11. Brosius FC 3rd, Alpers CE, Bottinger EP et al. Mouse models of diabeticnephropathy. J Am Soc Nephrol 2009; 20: 2503–2512.

12. Le TH, Fogo AB, Salzler HR et al. Modifier locus on mouse chromosome 3for renal vascular pathology in AT1A receptor-deficiency. Hypertension2004; 43: 445–451.

13. Chorvat RJ. Peripherally restricted CB1 receptor blockers. Bioorg MedChem Lett 2013; 23: 4751–4760.

14. Marquart S, Zerr P, Akhmetshina A et al. Inactivation of the cannabinoidreceptor CB1 prevents leukocyte infiltration and experimental fibrosis.Arthritis Rheum 2010; 62: 3467–3476.

15. Teixeira-Clerc F, Julien B, Grenard P et al. CB1 cannabinoid receptorantagonism: a new strategy for the treatment of liver fibrosis. Nat Med2006; 12: 671–676.

16. Slavic S, Lauer D, Sommerfeld M et al. Cannabinoid receptor 1 inhibitionimproves cardiac function and remodelling after myocardial infarctionand in experimental metabolic syndrome. J Mol Med Berl Ger 2013; 91:811–823.

17. André A, Gonthier M-P. The endocannabinoid system: its roles in energybalance and potential as a target for obesity treatment. Int J Biochem CellBiol 2010; 42: 1788–1801.

18. Nam DH, Lee MH, Kim JE et al. Blockade of cannabinoid receptor 1improves insulin resistance, lipid metabolism, and diabetic nephropathyin db/db mice. Endocrinology 2012; 153: 1387–1396.

19. Defer N, Wan J, Souktani R et al. The cannabinoid receptor type 2promotes cardiac myocyte and fibroblast survival and protects againstischemia/reperfusion-induced cardiomyopathy. FASEB J Off Publ Fed AmSoc Exp Biol 2009; 23: 2120–2130.

20. Servettaz A, Kavian N, Nicco C et al. Targeting the cannabinoid pathwaylimits the development of fibrosis and autoimmunity in a mouse model ofsystemic sclerosis. Am J Pathol 2010; 177: 187–196.

21. Barutta F, Piscitelli F, Pinach S et al. Protective role of cannabinoidreceptor type 2 in a mouse model of diabetic nephropathy. Diabetes 2011;60: 2386–2396.

22. Barutta F, Grimaldi S, Franco I et al. Deficiency of cannabinoid receptor oftype 2 worsens renal functional and structural abnormalities instreptozotocin-induced diabetic mice. Kidney Int [Internet] 2014; 86:979–990.

23. Wynn TA. Common and unique mechanisms regulate fibrosis in variousfibroproliferative diseases. J Clin Invest 2007; 117: 524–529.

24. Sato M, Muragaki Y, Saika S et al. Targeted disruption of TGF-beta1/Smad3 signaling protects against renal tubulointerstitial fibrosisinduced by unilateral ureteral obstruction. J Clin Invest. 2003; 112:1486–1494.

25. Eis V, Luckow B, Vielhauer V et al. Chemokine receptor CCR1 but not CCR5mediates leukocyte recruitment and subsequent renal fibrosis afterunilateral ureteral obstruction. J Am Soc Nephrol 2004; 15: 337–347.

26. Henderson NC, Mackinnon AC, Farnworth SL et al. Galectin-3 expressionand secretion links macrophages to the promotion of renal fibrosis.Am J Pathol 2008; 172: 288–298.

27. Schatz AR, Lee M, Condie RB et al. Cannabinoid receptors CB1 andCB2: a characterization of expression and adenylate cyclase modu-lation within the immune system. Toxicol Appl Pharmacol 1997; 142:278–287.

28. Wang Y, Harris DCH. Macrophages in renal disease. J Am Soc Nephrol 2011;22: 21–27.

29. Wang Y, Wang DH. Aggravated renal inflammatory responses in TRPV1gene knockout mice subjected to DOCA-salt hypertension. Am J PhysiolRenal Physiol 2009; 297: F1550–F1559.

30. Ledent C, Valverde O, Cossu G et al. Unresponsiveness to cannabinoidsand reduced addictive effects of opiates in CB1 receptor knockout mice.Science 1999; 283: 401–404.

31. Grimwood L, Masterson R. Propagation and culture of renal fibroblasts.Methods Mol Biol 2009; 466: 25–37.

Kidney International 13

L Lecru et al.: CB1 receptor and renal fibrosis bas i c resea rch