A New Animal Model for Pulmonary HypertensionBased on the Overexpression of a Single Gene,Angiopoietin-1Danny Chu, MD, Christopher C. Sullivan, MS, Lingling Du, MD, Augustine J. Cho, BA,Masakuni Kido, MD, Paul L. Wolf, MD, Matthew D. Weitzman, PhD,Stuart W. Jamieson, MB, FRCS, and Patricia A. Thistlethwaite, MD, PhDDivision of Cardiothoracic Surgery, University of California, San Diego, and Department of Pathology, Veterans AdministrationMedical Center and Laboratory of Genetics, The Salk Institute for Biological Studies, La Jolla, California

Background. Angiopoietin-1 gene expression in humanpulmonary hypertensive lungs is directly proportional toincreasing pulmonary vascular resistance. We hypothe-sized that targeted overexpresssion of angiopoietin-1 inthe lung would cause persistent pulmonary hypertensionin an animal model.

Methods. We injected 2 � 1010 genomic particles ofadeno-associated virus-angiopoietin-1 (AAV-Ang-1) intothe right ventricular outflow tract of 30 Fischer rats whileusing adeno-associated virus-lacZ (AAV-lacZ) injectedrats and carrier-injected rats as our control groups. Allanimals underwent survival surgery and were sacrificedat serial timepoints postgene delivery. At each timepoint,pulmonary artery pressures were measured and pulmo-nary angiography using the Microfil polymer perfusiontechnique was performed. The lungs were harvested forpathologic analysis, mRNA analysis, Western blot as-says, and in situ RNA hybridization to localize geneexpression.

Results. Pulmonary artery pressures of AAV-Ang-1

injected rats were significantly increased compared withthe control groups (p < 0.01) at all timepoints. Pathologicanalysis of AAV-Ang-1 lung specimens demonstratedincreased smooth muscle cell proliferation within themedial layer of arterioles with obliteration of smallvessels similar to that seen in human pulmonary hyper-tension. Angiograms of AAV-Ang-1 injected lungsshowed blunting of small peripheral arterioles consistentwith advanced pulmonary hypertension. In situ RNAhybridization localized angiopoietin-1 expression to thevascular wall of small-caliber pulmonary vessels. Proteinand mRNA assays confirmed persistent angiopoietin-1expression in the lung for up to 60 days postgenedelivery.

Conclusions. Overexpression of angiopoietin-1 usingan adeno-associated virus vector causes pulmonary hy-pertension in rats. These data provide a novel physio-logic animal model for pulmonary hypertension.

(Ann Thorac Surg 2004;77:449–57)© 2004 by The Society of Thoracic Surgeons

Pulmonary hypertension is a pathologic processcaused by diffuse smooth muscle cell hyperplasia

and hypertrophy of the distal pulmonary vasculaturewhich eventually results in obliteration of small pulmo-nary arterioles [1]. Pulmonary vessel smooth musclehyperplasia and hypertrophy occur in all forms of humanpulmonary hypertension [2]. Despite the similar patho-logic process seen in both primary and secondary pul-monary hypertension, the exact mechanism of this dis-ease remains a mystery.

In embryonic development, the establishment andremodeling of blood vessels are controlled by paracrinesignals between endothelial cells and smooth musclecells [3]. Angiopoietin-1 (Ang-1), a molecule secreted bysmooth muscle cells, plays an essential role in the forma-tion of arteries and arterioles in utero and has beenimplicated in pathologic angiogenesis [4]. As a muscle-

secreted ligand, Ang-1 signals vascular endothelial cellsthrough the endothelial-specific receptor, TIE2, to recruitand stimulate the proliferation of smooth muscle cellsaround nascent endothelial tubes to create mature arte-rial structures [5]. Ang-1 has also been shown to actsynergistically with vascular endothelial growth factor(VEGF) to potentiate vascular network maturation in vivo[6]. Knockout mice deficient in Ang-1 die in utero fromdefects in embryonic vascular development includingsevere vascular malformation of the lungs [7]. Endothe-lial cells of these animals were arranged into tubularstructures and were present in normal numbers but thevessels in these genetically engineered mice had nosmooth muscle cell encasement and lacked branch net-works. These results suggested that Ang-1 plays anessential role in arterial vascular formation andremodeling.

Because Ang-1 plays a role in the muscularization ofarteries in utero we hypothesized that its aberrant ex-pression in the adult lung could be a key molecular stepin the genesis of pulmonary hypertension. In this study,

Accepted for publication April 28, 2003.

Address reprint requests to Dr Thistlethwaite, Division of CardiothoracicSurgery, University of California, San Diego, 200 West Arbor Dr, SanDiego, CA 92103-8892; e-mail: pthistlethwaite@ucsd.edu.

© 2004 by The Society of Thoracic Surgeons 0003-4975/04/$30.00Published by Elsevier Inc doi:10.1016/S0003-4975(03)01350-X

GEN

ERA

LT

HO

RA

CIC

we sought to determine if constitutive Ang-1 expressionin the adult lung causes pulmonary hypertension and tocreate a reproducible animal model for this disease. Wehave previously shown that Ang-1 is overexpressed inlung tissues of humans with different forms of pulmonaryhypertension [8] and that the level of Ang-1 expression isdirectly proportional to the severity of the disease [9].Our earlier results suggested that Ang-1 correlated withthe pulmonary hypertension phenotype but did notprove it played a causal role in the disease process. Inorder to establish whether up-regulation of the Ang-1gene product causes pulmonary hypertension we createdan animal model whereby Ang-1 was constitutively ex-pressed in the adult rodent lung. Our results demon-strate that Ang-1 induces pulmonary hypertensive vas-cular pathology and clinical development of pulmonaryhypertension leading to right ventricular failure anddeath. This is the first reproducible model of this diseasebased on tissue-specific alteration of the expression of asingle gene.

Material and Methods

Viral Plasmid and Vector ProductionpXX2 and pXX6 were used for adeno-associated virus(AAV) vector production. pXX2 contains the AAV rep andcap genes whereas pXX6 contains the adenoviral geneproducts which are required to facilitate AAV replication[10]. pAAV-Ang-1 and pAAV-lacZ plasmids were createdby inserting either the Ang-1 or lacZ genes into thepAAV-shuttle vector generated via standard subcloningtechniques. This shuttle vector has two inverted terminalrepeats (ITRs) required for AAV production, cytomega-lovirus (CMV) immediate early promoter/enhancer, mul-tiple cloning site, and a simian virus 40 (SV40) polyade-nylation tail signal. The lacZ gene, a common reportergene, was used as an insert in our control vector (AAV-lacZ) to show that any lung pathology generated byAAV-Ang-1 was not due to nonspecific effects of thevector or gene transfer technique.

Large scale recombinant serotype 2 AAV containingthe Ang-1 gene (AAV-Ang-1) and AAV containing thelacZ gene (AAV-lacZ) were produced and purified bycotransfecting 293T cells. Triple cotransfections withpXX2, pXX6, and pAAV-Ang-1/pAAV-lacZ plasmidswere accomplished with Polyfect Transfection Reagent(Qiagen, Valencia, CA). Cells were collected and resus-pended in media and underwent two freeze/thaw cyclesto lyse their membranes. Benzonase (Sigma-Aldrich, St.Louis, MO) was added to the mixture to eliminate un-packaged DNA. Cellular debris was separated from thecleared cell lysate containing AAV via centrifugation.

AAV vector high-grade purification was modified fromthe single-step gravity-flow column technique describedby Auricchio and associates [11]. Type I heparin columns(Sigma-Aldrich, St. Louis, MO) were equilibrated withphosphate-buffered saline (PBS)/1 mmol/L MgCl2 usingan Econo-Pump (Bio-Rad, Hercules, CA). Crude AAVvector was added directly to the column with a flow rate

of 0.2 ml/min and subsequently washed with PBS/1mmol/L MgCl2/0.1 mol/L NaCl. The virus was elutedwith PBS/1 mmol/L MgCl2/0.4 mol/L NaCl and furtherdialyzed against PBS/1 mmol/L MgCl2 gradient. Theconcentration of the vectors was determined by real-timepolymerase chain reaction (PCR) (7700 Sequence Detec-tor, Applied Biosystems, Foster City, CA) with SYBRGreen detection kit (Applied Biosystems). All viral prep-arations had at least 1 � 1011 genomic particles/ml. Thepurified vectors were then retested for transgene expres-sion by infecting 293 cells and nucleic acid sequenced toconfirm transgene integrity.

Gene Transfer Protocol Into Rat Lungs In VivoTwelve week-old pathogen-free Fischer rats (Harlan, SanDiego, CA) housed in micro-isolator boxes (2–3 animalsper box) were anesthetized with intraperitoneal injectionof ketamine (50 mg/kg) and xylazine (10 mg/kg). Theanimals were intubated and ventilated using a Harvardrodent ventilator Model 683 (Harvard Apparatus, Hollis-ton, MA). The heart was exposed via a left antero-lateralthoracotomy at the fourth intercostal space. 2 � 1010

genomic particles of AAV-Ang-1 in 200 �l of PBS/1mmol/L MgCl2 were injected directly into the rightventricular outflow tract (just beneath the pulmonicvalve) of the hearts of 30 rats in our test group whereasthe same amount of AAV-lacZ was injected in 30 rats inour first control group. Thirty sham animals forming oursecond control group were injected with 200 �l of thecarrier solution, PBS/1 mmol/L MgCl2 alone. Rats fromeach group were sacrificed at 1-month and 2-monthspostgene delivery timepoints and organs/blood collectedfor tissue and molecular analysis. Four animals in eachgroup were followed for one year or until death (ifearlier) in order to observe the natural history of theirpulmonary disease. All animals received care in accor-dance with the “Guide for the Care and Use of Labora-tory Animals” published by the National Institutes ofHealth (N.I.H. publication 85 to 23, revised 1985). Thisstudy was approved by the University of California, SanDiego Animal Subjects Program.

Hemodynamic Measurements and PulmonaryAngiographyAt two serial timepoints after the gene delivery operationpulmonary artery systolic and diastolic pressures as wellas heart rate and systemic blood pressures were mea-sured just before the sacrifice of the animals. To accom-plish this a sternotomy was performed after anestheticinduction and intubation in order to expose the heart. A22-gauge angiocatheter was inserted directly into theright ventricular outflow tract and advanced into thepulmonary artery. The pulmonary artery pressures weremeasured using a standardized pressure transducer(SpaceLabs, Inc., Issaquah, WA) whereas systemic bloodpressures were measured by direct aortic transduction.Afterwards the animals were sacrificed by exsanguina-tion through the abdominal aorta. For 5 animals in eachgroup at each timepoint the pulmonary vasculature wasflushed with saline and perfused with Microfil (Flow-

450 CHU ET AL Ann Thorac SurgANGIOPOIETIN-1 IN PULMONARY HYPERTENSION 2004;77:449–57

GEN

ERA

LT

HO

RA

CIC

Tech, Inc., Carver, MA), a liquid silicon polymer, at 0.25ml per minute for 4 min using a syringe infusion pump(Harvard Apparatus, Holliston, MA). After 12 h of 4°storage the lungs were sequentially bathed in increasingconcentrations of ethanol, placed in methyl-salicylate,and photographed with a digital camera (DSC-S30, SonyCorporation, Japan) through a dissecting microscope(WILD M651, Leica, Switzerland).

Protein and mRNA AssaysRat lung protein extracts and Western blot analyses wereperformed according to standard procedures using 100mg of whole tissue extracts. Both the polyclonal goatAng-1 and actin antibodies were obtained from SantaCruz Biotechnology. Total RNA was isolated using RNA-zol B reagent (Tel-Test, Friendswood, TX). For all PCRanalyses, 2 �g of total RNA was digested with DNase Iand reverse-transcribed according to the manufacturer’sinstructions (Superscript II kit, Invitrogen, Carlsbad, CA).PCR amplifications were performed under standard con-ditions and each PCR product was electrophoresed on1% SeaKem LE (BioWhittaker Molecular Applications,Walkersville, MD) agarose gels stained with ethidiumbromide. Each gene was analyzed using specific primerpairs: viral-specific Ang-1: 5�- GGTACCCGAATGA-CAGTTTTCC-3� 5�-CTCCATTTCTAAGATTTTGT-GCTC-3� rodent Ang-1: 5�-GAGAAGCAACTTCTCCAA-CAG-3 � 5 �- CTCGTTCCCGAGCCAATATTC-3 �glyceraldehyde-3-phosphate dehydrogenase (GAPDH): 5�-CATCATCTCTGCCCCCTCTG-3� and 5�-CCTGCTTCAC-CACCTTCTTG-3�. The viral-specific Ang-1 primer se-quence overlapped with the multiple cloning site sequencein pAAV-shuttle vector and therefore only recognized theAng-1 gene sequence delivered by AAV.

Immunohistochemistry and In Situ HybridizationAt each timepoint, The lungs from 5 rodents per groupwere fixed in 4% paraformaldehyde, embedded in paraf-fin, and sectioned at 5 �m thickness. After hematoxylinand eosin staining the specimens were analyzed by apulmonary pathologist in blinded fashion. In situ hybrid-izations were performed using RNA probes subcloned inour lab specific to the SV40 Poly-A domain of the pAAV-shuttle vector expression cassette. The SV40 Poly-A do-main was inserted into pBluescript II SK(�) (Stratagene,LA Jolla, CA) and served as a template for the synthesisof digoxigenin-labeled antisense RNA probes using T3polymerase and the DIG RNA labeling mix (Roche,Indianapolis, IN). Sense probes were made using T7polymerase and served as controls. Hybridization wascarried out at 50°C with 50% formamide/0.3 mol/L so-dium chloride, 0.03 mol/L sodium citrate (2xSSC). Thedigoxigenin (DIG)-label was detected by an anti-DIG Fab(Roche) coupled to alkaline phosphatase using nitro bluetetrazolium/5-bromo-4-chloro-3-indolyl-phosphate(Roche).

Statistical AnalysisValues for the pulmonary artery systolic/diastolic pres-sure were expressed as the mean � standard error of the

mean. Continuous variables were compared with a Stu-dent’s t test. A p value of less than 0.05 was consideredstatistically significant.

Results

Constitutive Ang-1 Expression in Rodent LungsOur study compared the physiologic and pathologicresult of injection of an adeno-associated virus contain-ing Ang-1 gene into the pulmonary circulation as com-pared with the injection of a control virus (AAV-lacZ) orcarrier solution alone. We chose the right ventricularoutflow tract for viral delivery because it allowed easyaccess to the pulmonary arterial circulation without re-sulting in excessive bleeding from direct needle injectioninto the thin-walled pulmonary artery.

Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis confirmed that vector-specific Ang-1 tran-scripts were present in the lungs of animals at 1 and 2months after injection with AAV-Ang-1 (Fig 1A). The 561base pair PCR product representing wild-type endoge-nous Ang-1 was present in small amounts in all rat lungsamples tested. Because our AAV-Ang-1 injected rodentsexpressed not only wild-type but also virally deliveredAng-1 this group of animals had greater overall steady-state levels of Ang-1 mRNA in their lung tissue comparedwith control animals. We found that levels of GAPDHmRNA, our control for RNA amount in each lane, wasconsistently equal among the samples studied. Westernblot protein analysis using an antibody against the car-boxy-terminus of murine Ang-1 demonstrated increasedoverall steady-state levels of Ang-1 protein expression inlungs of animals injected with AAV-Ang-1 comparedwith control animals injected with AAV-lacZ (Fig 1B).Equal amounts of the internal control, actin filamentprotein, confirmed that all samples studied by Westernblot analysis had equivalent amounts of protein presentin each lane.

To determine whether virally produced Ang-1 waspresent in other organs we assayed a variety of tissues forthe presence of virus-specific Ang-1 transcripts. By PCRanalysis vector-specific Ang-1 mRNA was found only inthe lung and not in the brain, intestine, kidney, heart, orliver. Serum Ang-1 protein was not detected in theAAV-Ang-1 injected rats nor in the two control groups byWestern analysis (data not shown).

High Levels of Ang-1 in the Lung Induces PulmonaryArteriolar Muscular Hyperplasia and HypertrophyAnimals constitutively expressing high levels of Ang-1protein in the lung from injection of our AAV-Ang-1vector showed diffuse pulmonary pathologic changesconsistent with advanced pulmonary hypertension. Byhistologic analysis these lungs demonstrated severe mus-cular hyperplasia and hypertrophy in the medial layer ofsmall pulmonary arteries and arterioles measuring � 500�m in diameter (Fig 2A). This small vessel medial hyper-trophy/hyperplasia was present in more than 80% of thearterioles examined at 1 and 2 months post AAV-Ang-1

451Ann Thorac Surg CHU ET AL2004;77:449–57 ANGIOPOIETIN-1 IN PULMONARY HYPERTENSION

GEN

ERA

LT

HO

RA

CIC

gene injection and occurred throughout all lobes of bothlungs. By micrometer analysis on review of ten fields/lung slide at 40� magnification we found that smallpulmonary vessels on average increased their vessel walldiameter by 400% (range 210%–630%) and increased thenumber of myocytes per vessel wall area by 300% (range250%–430%) 1 month after gene transfer. One fourth ofall pulmonary arterioles examined in animals constitu-tively expressing the virally delivered Ang-1 gene wereoccluded from severe medial hyperplasia/hypertrophy ateach time point. There was no lymphocytic infiltration in

or around the pulmonary vessels with this pathology ashas been shown for adenovirally transduced genes. Incontrast to the human form of this disease few ( � 1%)stenosed vessels in the AAV-Ang-1 treated animals man-ifest plexiform pathology at 1 and 2 month intervals aftergene injection. Rats injected in the right ventricularoutflow tract with either AAV-lacZ or an equal volume ofPBS/1 mmol/L MgCl2 demonstrated normal lung histol-ogy (less than 2% of pulmonary arterioles � 500 �mbeing muscularized) without evidence of small vesselmuscle cell hypertrophy or hyperplasia.

Fig 1. Constitutive angiopoietin-1transgene mRNA expression bypolymerase chain reaction methods(A) and protein expression by West-ern blot assay (B) in lungs of rodentsinjected with adeno-associated virus-angiopoietin-1 (AAV-Ang-1) com-pared with rodents injected with ad-eno-associated virus-lacZ (AAV-lacZ). Lung gene expressionmeasured at 1 and 2 months afterviral gene transfer. (Ang-1 � angio-poietin-1, bp � base pair, GAPDH� glyceraldehyde-3-phosphate dehy-drogenase, kD � kilodalton.)

Fig 2. (A) Photomicrograph of rodent lungsdemonstrating severe hypertrophy/hyperpla-sia of medial layer of small arterioles mea-suring 40 �m compared with sham and con-trol animals. (Bar � 25 �m.) (B)Photomicrograph of rodent lungs showingsevere muscle hypertrophy of pulmonary ar-terioles measuring 300 �m compared withsham and control animals. (Bar � 150 �m.)(Hematoxylin & eosin stain, magnification�40.) (Ang-1 � animals injected with ad-eno-associated virus-angiopoietin-1 [AAV-Ang-1]; lacZ � animals injected with adeno-associated virus-lacZ [AAV-lacZ]; Saline �animals injected with carrier saline.)

452 CHU ET AL Ann Thorac SurgANGIOPOIETIN-1 IN PULMONARY HYPERTENSION 2004;77:449–57

GEN

ERA

LT

HO

RA

CIC

Animals With High Levels of Ang-1 in the LungDevelop Pulmonary HypertensionBy pulmonary artery transduction each of the 26 animalsconstitutively expressing Ang-1 at 1 and 2 months post-gene delivery demonstrated a significant increase inpulmonary artery systolic and diastolic pressures (Fig 3)that was reproducible and averaged over four measure-ments taken during a 10 min period before sacrifice of theanimal. This represented a selective elevation in pulmo-nary artery pressure as systemic arterial blood pressureswere unchanged compared with base line measurementstaken before gene transfer (data not shown). Systolic,diastolic, and mean pulmonary artery pressures for ani-

mals expressing the virally delivered Ang-1 gene productwere significantly elevated (p � 0.01) compared withpressures measured in control animals (n � 13 at eachtimepoint) injected with either AAV-lacZ or PBS/1mmol/L MgCl2 at the same timepoint.

We next confirmed that animals expressing high levelsof Ang-1 in the lung had pulmonary angiograms demon-strating severe small vessel pruning similar to that seenin human pulmonary hypertension (Fig 4). A continuousin vivo Microfil polymer perfusion technique that dis-tends vessels evenly and results in reproducible assess-ments was used to assess the pulmonary vascular tree in5 animals in each group. Control animals demonstratedvessel branching and vascular blush throughout the lungparenchyma including the periphery of the lung. Incontrast animal lungs with measured pulmonary hyper-tension following AAV-Ang-1 injection showed normalto slightly enlarged hilar vessels with minimal angio-graphic blush in the lung periphery (Fig 4). The lack ofvessels visualized in the periphery of the lung with thistechnique represents the fact that most pulmonary arter-ies � 500 �m in diameter (or greater than four branchdivisions from the main pulmonary artery) were nearlyoccluded. A similar angiographic pattern is seen in thehuman form of this disease.

Localization of Ang-1 Expression in the Lungs ofAAV-Injected RatsTo investigate the cell specificity of virally deliveredgenes in the lung we performed RNA in situ hybridiza-tion using a probe specific to the 3� untranslated region ofthe Ang-1 or lacZ genes in our viral constructs. Lungspecimens from animals injected either with AAV-Ang-1or AAV-lacZ showed staining for viral-specific tran-scripts produced in the vessel wall of small pulmonaryarteries and arterioles compared with control animals(Fig 5). This suggests that paracrine expression of theAng-1 protein in the pulmonary vessel wall correlateswith the location of pathology in the lung. Localization ofthe virally transduced Ang-1 and lacZ transcripts per-sisted in small pulmonary vessel walls up to 60 days aftergene delivery.

The Natural History of Pulmonary Hypertension inAnimals Overexpressing Ang-1 in the LungFour animals from each test group were followed for oneyear or until near-death (if earlier). AAV-Ang-1 injectedanimals lived a mean of 5.2 months (range 2.7–6.3months) whereas AAV-lacZ and PBS-carrier injectedcontrol animals were healthy and sacrificed at 12 months.On autopsy animals injected with AAV-Ang-1 demon-strated end-stage pulmonary hypertension with rightventricular hypertrophy, diffuse pulmonary arteriolarmuscular hyperplasia, and widespread small vessel ste-nosis and occlusion. Despite subtotal obliteration ofsmall pulmonary arterioles plexiform pathology was onlyminimally present. Control animals injected with eitherAAV-lacZ or PBS-carrier demonstrated normal heart andlung pathology at autopsy.

Fig 3. Pulmonary artery pressures of rodents at (Top) 30 and (Bot-tom) 60 days postgene delivery demonstrating persistent signifi-cantly elevated pressures of adeno-associated virus-angiopoietin-1(AAV-Ang-1) injected animals compared with control animals.(Ang-1 � animals injected with AAV-Ang-1; lacZ � animals in-jected with adeno-associated virus-lacZ (AAV-lacZ); mm Hg �millimeters of mercury; PAD � pulmonary artery diastolic pressure;PAS � pulmonary artery systolic pressure.)

453Ann Thorac Surg CHU ET AL2004;77:449–57 ANGIOPOIETIN-1 IN PULMONARY HYPERTENSION

GEN

ERA

LT

HO

RA

CIC

Comment

Pulmonary hypertension is a debilitating and lethal dis-ease characterized by small vessel vasculopathy in thelung. To date, the cause of this organ-specific pathologyis unknown. In the early phases of pulmonary hyperten-sion distal pulmonary vessels exhibit exaggerated vaso-constriction with normal histology [12]. As timeprogresses the medial layer of the pulmonary arteriolesdevelops muscular hyperplasia and hypertrophy. Thissmall vessel wall thickening narrows and eventuallyobstructs blood flow [13]. The clinical manifestation ofthis vasculopathy is markedly increased pulmonary ar-tery pressure compared with systemic arterial pressure.

Hypoxia, chemical damage, increased pulmonary circu-lation, and thromboembolism have all been implicated totrigger pulmonary hypertension [14–17]. However, themolecular mechanism responsible for the common lungpathology seen with each of these inciting factors re-mains unknown. Although treatments such as prostacy-clin analogs, nitric oxide, or endothelin receptor antago-nists alleviate the symptoms of the disease, none of thesedrugs has been shown to halt the progressive muscular-ization of small pulmonary vessels which is the patho-logic hallmark of this disease [18, 19].

The aim of our study was to understand the molecularbasis by which normal pulmonary vessels develop

Fig 4. Whole lung gross Microfilpulmonary artery angiograms illus-trate severe small vessel blunting inanimals injected with adeno-associ-ated virus-angiopoietin-1 (AAV-Ang-1) compared with control ani-mals. (Left) Ang-1. (Right) lacZ.(Bar � 5 mm.) (Ang-1 � animalinjected with AAV-Ang-1; lacZ �animal injected with adeno-associ-ated virus-lacZ [AAV-lacZ].)

Fig 5. In situ RNA hybridization photomicrograph shows viral-specific angiopoietin-1 transcript being expressed in the walls of pulmonaryarterioles. (Left) Ang-1. (Right) Saline. (Bar � 25 �m, nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate stain, magnification 40�.)

454 CHU ET AL Ann Thorac SurgANGIOPOIETIN-1 IN PULMONARY HYPERTENSION 2004;77:449–57

GEN

ERA

LT

HO

RA

CIC

smooth muscle hyperplasia and hypertrophy whicheventually causes clinical manifestations of pulmonaryhypertension. In this study we show that high levels ofAng-1 protein induced by viral injection of transgene intothe pulmonary circulation leads to the development ofclinical and pathologic pulmonary hypertension in ro-dents. This pathology persisted after transgene expres-sion had ceased. The phenotype observed with genetransfer was not due to the viral vector, method ofsurgical delivery, nor carrier solution used. Our findingssuggest that in the pulmonary vasculature, supranormallevels of Ang-1 facilitate the development of profusemuscular hyperplasia of small pulmonary blood vesselswhich ultimately leads to the development of pulmonaryhypertension. From a mechanistic point of view webelieve that derangement in expression of a single gene,Ang-1, may be the molecular trigger for this organ-specific vasculopathy.

Using sustained gene transfer into the lung we havecreated a novel physiologic animal model to study themechanisms responsible for pulmonary hypertension.Although several animal models have been used in thepast to study pulmonary hypertension none of themaccurately mimics the human form of the disease. In onemodel, the subcutaneous injection of a pyrrolizidinealkaloid, monocrotaline, in rodents results in global in-flammation of the lung with thickening of arterioles andvenules and the eventual sporadic development of lungadenocarcinomas [20, 21]. The inflammatory componentinduced by this chemical damage in the lung is not seenin our animal model or in the human form of pulmonaryhypertension. Other pulmonary hypertensive animalmodels rely on rodents with both systemic and pulmo-nary hypertension or on placing rodents into hypoxic orhigh altitude conditions mimicking hypoxia [22, 23]. Al-though low partial pressure of oxygen will induce thevasoreactive component of this disease in humans, themajority of patients with pulmonary hypertension de-velop this disease under normoxic conditions. Supranor-mal expression of Ang-1 in our rodent lung modelinduced elevated pulmonary artery pressures and in-creased muscularization of distal pulmonary vessels sim-ilar to the human disease without chemically inducedlung inflammation or pulmonary hypoxia. We found thatthe viral vector, AAV, itself introduced into the pulmo-nary circulation did not stimulate perivascular lympho-cytic infiltration or inflammatory reaction in lung tissueas has been seen with gene delivery systems to the lung[24].

Clues as to why Ang-1 plays a role in pulmonaryhypertension comes from earlier studies of its function inangiogenesis and vasculogenesis. Ang-1 is a vascularsmooth muscle cell growth factor in the developing lungand its expression is shut-off once development is com-plete [7]. Constitutive Ang-1 expression in the adult lungleads to an exaggerated and organ-specific vascularsmooth muscle cell proliferative response. Ang-1 isknown to bind the endothelial-restricted promoter, TIE2,resulting in tyrosine phosphorylation and activation of

this receptor into an active tyrosine kinase [6]. TIE2intracellular signaling involves tyrosine phosphorylationof several intracellular proteins [25] although the finaldownstream effect of this signal transduction in the lungis unknown. We suspect that Ang-1 signaling throughthe endothelial-specific TIE2 receptor results in the re-lease of a paracrine muscle-specific growth factor byvascular endothelial cells. In the normal adult lung whereAng-1 expression is minimal vascular smooth musclecells are quiescent, whereas in the adult pulmonaryhypertensive lung where Ang-1 expression is de-regulated vascular smooth muscle cells are proliferative.Recent evidence has suggested that haploid mutations inthe bone morphogenetic receptor type 2 (BMPR2) areseen in patients with the rare form of inherited primarypulmonary hypertension [26, 27]. The molecular mecha-nism by which Ang-1 causes pulmonary hypertensionmay be based on co-signaling with this membrane boundreceptor.

Several limitations of this study deserve comment.First our animal experiments are based on the transduc-tion of a specific gene product (Ang-1) for only 60 dayspostgene delivery. This is an inherent limitation of genetransfer experiments and does not necessarily reflect thenatural time course of the human disease. Second al-though overexpression of Ang-1 in the lung resulted insmooth muscle cell hyperplasia and hypertrophy, itrarely resulted in the development of plexiform lesions inthe lung. We recognize that the end-stage pathology ofhuman pulmonary hypertension includes a complex ar-ray of stenosed, occluded, and plexogenic vessels. End-stage disease in our model system was limited to vesselstenosis and occlusion by smooth muscle cell prolifera-tion and hypertrophy. Despite these limitations we showthat constitutive expression of Ang-1 in the rodent lunginduces pulmonary hypertension and that this vascu-lopathy is similar to that seen in the human form of thisdisease. Our results suggest that Ang-1 may be a molec-ular signal for vascular muscle cell hyperplasia in thelung. Future inhibition of the expression or effect ofAng-1 in the adult pulmonary vasculature may be auseful strategy to prevent and treat pulmonary hyperten-sion in humans.

Cotransfecting plasmids, pXX2 and pXX6, were gifts from Dr R.Jude Samulski, University of North Carolina, Chapel Hill. Finan-cial support for this project was provided by the Charles B.Wang Foundation and National Institutes of Health Grant 1R01HL70852–01 to Dr Thistlethwaite.

References

1. Gaine SP, Rubin LJ. Primary pulmonary hypertension. Lan-cet 1998;352:719–25.

2. Rabinovitch M. Pulmonary hypertension: updating a myste-rious disease. Cardiovasc Res 1997;34:268–72.

3. Folkman J, D’Amore PA. Blood vessel formation: what is itsmolecular basis? Cell 1996;87:1153–5.

4. Takahama M, Tsutsumi M, Tsujiuchi T, et al. Enhancedexpression of Tie2, its ligand angiopoietin-1, vascular endo-

455Ann Thorac Surg CHU ET AL2004;77:449–57 ANGIOPOIETIN-1 IN PULMONARY HYPERTENSION

GEN

ERA

LT

HO

RA

CIC

thelial growth factor, and CD31 in human non-small celllung carcinomas. Clin Cancer Res 1999;5:2506–10.

5. Hanahan D. Signaling vascular morphogenesis and mainte-nance. Science 1997;277:48–50.

6. Asahara T, Chen D, Takahashi T, et al. Tie2 receptor ligands,angiopoietin-1, and angiopoietin-2, modulate VEGF-induced postnatal neovascularization. Circ Res 1998;83:233–40.

7. Suri C, Jones PF, Patan S, et al. Requisite role of angiopoi-etin-1, a ligand for the TIE2 receptor, during embryonicangiogenesis. Cell 1996;87:1171–80.

8. Du L, Sullivan CC, Chu D, et al. Signaling molecules innon-familial pulmonary hypertension. N Engl J Med 2003;348:500–9.

9. Thistlethwaite PA, Lee SH, Du LL, et al. Human angiopoietingene expression is a marker for severity of pulmonaryhypertension in patients undergoing pulmonary thrombo-endarterectomy. J Thorac Cardiovasc Surg 2001;122:65–73.

10. Xiao X, Li J, Samulski RJ. Production of high-titer recombi-nant adeno-associated virus vectors in the absence of helperadenovirus. J Virol 1998;72:2224–32.

11. Auricchio A, Hildinger M, O’Connor E, Gao GP, Wilson JM.Isolation of highly infectious and pure adeno-associatedvirus type 2 vectors with a single-step gravity-flow column.Hum Gene Ther 2001;12:71–6.

12. Giaid A, Yanagisawa M, Langleben D, et al. Expression ofendothelin-1 in the lungs of patients with pulmonary hyper-tension. N Engl J Med 1993;328:1732–9.

13. Rubin LJ. Pathology and pathophysiology of primary pulmo-nary hypertension. Am J Cardiol 1995;75:51A–4.

14. O’Blenes SB, Fischer S, McIntyre B, Keshavjee S, Rabino-vitch M. Hemodynamic unloading leads to regression ofpulmonary vascular disease in rats. J Thorac CardiovascSurg 2001;121:279–89.

15. Maruyama J, Yokochi A, Maruyama K, Nosaka S. Acetylcho-line-induced endothelium-derived contracting factor in hy-poxic pulmonary hypertensive rats. J Appl Physiol 1999;86:1687–95.

16. Jamieson SW, Kapelanski DP. Pulmonary endarterectomy.Curr Probl Surg 2000;37:165–252.

17. Rich S, Rubin L, Walker AM, Schneeweiss S, Abenhaim L.

Anorexigens and pulmonary hypertension in the UnitedStates: results from the surveillance of North Americanpulmonary hypertension. Chest 2000;117:870–4.

18. Hoeper MM, Schwarze M, Ehlerding S, et al. Long-termtreatment of primary pulmonary hypertension with aerol-ized iloprost, a prostacyclin analogue. N Engl J Med 2000;342:1866–70.

19. Bradley SP, Auger WR, Moser KM, Fedullo PF, ChannickRN, Bloor CM. Right ventricular pathology in chronic pul-monary hypertension. Am J Cardiol 1996;78:584–7.

20. Cowan KN, Heilbut A, Humpl T, Lam C, Ito S, RabinovitchM. Complete reversal of fatal pulmonary hypertension inrats by a serine elastase inhibitor. Nat Med 2000;6:698–702.

21. Ghodsi F, Will JA. Changes in pulmonary structure andfunction induced by monocrotaline intoxication. Am JPhysiol 1981;240:H149–55.

22. Zamora MR, Stelzner TJ, Webb S, Panos RJ, Ruff LJ, Demp-sey EC. Overexpression of endothelin-1 and enhancedgrowth of pulmonary artery smooth muscle cells from fawn-hooded rats. Am J Physiol 1996;270:L101–9.

23. Rabinovitch M, Gamble W, Nadas AS, Miettinen OS, Reid L.Rat pulmonary circulation after chronic hypoxia: hemody-namic and structural features. Am J Physiol 1979;236:H818–27.

24. Thorne PS, McCray PB, Howe TS, O’Neill MA. Early-onsetinflammatory response in vivo to adenoviral vectors in thepresence or absence of lipopolysaccharide-induced inflam-mation. Am J Respir Cell Mol Biol 1999;20:1155–64.

25. Jones N, Master Z, Jones J, et al. Identification of Tek/Tie2binding partners. Binding to a multifunctional docking sitemediates cell survival and migration. J Biol Chem 1999;274:30896–905.

26. Lane KB, Machado RD, Pauciuo MW, et al. Heterozygousgermline mutations in BMPR2, ecoding a TGF-beta receptor,cause familial primary pulmonary hypertension. The Inter-national PPH Consortium. Nat Genet 2000;26:81–4.

27. Machado RD, Pauciulo MW, Thomson JR, et al. BMPR2haploinsufficiency as the inherited molecular mechanism forprimary pulmonary hypertension. Am J Hum Genet 2001;68:92–102.

INVITED COMMENTARY

Pulmonary hypertension is a debilitating disease associ-ated with disparate factors, ranging from pulmonarydisorders, thromboembolism, and familial genetic aber-rations to the use of appetite suppressants. The exactmechanisms of the disease remain unknown, and mightvary for different forms. A more general hypothesis isevolving, however, and the present paper by Dr Chu andcoworkers from Dr Thistlethwaite’s group provides sig-nificant support for it.

The vascular endothelial growth factor (VEGF) familyand angiopoietin family of growth factors are known toact in concert on the blood vasculature. The VEGFsinitiate vessel growth, whereas the angiopoietins seem topromote vessel maturation and maintenance. Angiopoi-etin-1 acts through the receptor tyrosine kinase Tie2.Previous work from the same group has linked thissignaling pathway to the bone morphogenic proteinreceptor (BMPR) family. In addition to affecting bonegrowth, these receptors are involved in regulation ofsmooth muscle proliferation. The family includes a pri-mary gene for inherited pulmonary hypertension,

named BMPR2. Since angiopoietin-1 downregulates theBMP pathway, overexpression may lead to the samedysregulation of vascular smooth muscle cell inhibitionas seen with BMPR2 mutations in the familial forms ofthe disease. The result is increased proliferation ofsmooth muscle cells and obstruction of small pulmonaryvessels.

The present paper employs gene transfer methodsallowing local angiopoietin-1 overexpression in the pul-monary vasculature of adult rats. This model probablycomes closer to the clinical situation than the traditionalmethod that uses intraperitoneal injection of mono-crotaline to induce pulmonary hypertension. As pre-dicted by their hypothesis, the authors showed that therats developed clinical and pathological changes con-sistent with advanced pulmonary hypertension, whichwas unrelated to the gene transfer technique itself. Thepaper very nicely demonstrates hypertrophy/hyperplasiaof the media of pulmonary vessels following angiopoi-etin-1 overexpression. However, a number of publica-tions on different forms of pulmonary hypertension

456 CHU ET AL Ann Thorac SurgANGIOPOIETIN-1 IN PULMONARY HYPERTENSION 2004;77:449–57

© 2004 by The Society of Thoracic Surgeons 0003-4975/04/$30.00Published by Elsevier Inc doi:10.1016/j.athoracsur.2003.06.009

GEN

ERA

LT

HO

RA

CIC