Intravitreal AAV2.COMP-Ang1 Prevents … · 2015-11-17 · Christina L. O’Neil,2 Reinhold J....

Judd M Cahoon1 Ruju R Rai1 Lara S Carroll1 Hironori Uehara1 Xiaohui Zhang1

Christina L OrsquoNeil2 Reinhold J Medina2 Subtrata K Das1 Santosh K Muddana1

Paul R Olson1 Spencer Nielson1 Kortnie Walker1 Maggie M Flood1

Wyatt B Messenger1 Bonnie J Archer1 Peter Barabas1 David Krizaj1

Christopher C Gibson3 Dean Y Li4 Gou Y Koh5 Guangping Gao6 Alan W Stitt2

and Balamurali K Ambati1

Intravitreal AAV2COMP-Ang1 PreventsNeurovascular Degeneration in aMurineModel of Diabetic RetinopathyDiabetes 2015644247ndash4259 | DOI 102337db14-1030

Diabetic retinopathy (DR) is the leading cause of blind-ness in the working-age population in the US The vision-threatening processes of neuroglial and vasculardysfunction in DR occur in concert driven by hypergly-cemia and propelled by a pathway of inflammationischemia vasodegeneration and breakdown of the bloodretinal barrier Currently no therapies exist for normaliz-ing the vasculature in DR Here we show that a singleintravitreal dose of adeno-associated virus serotype 2encoding a more stable soluble and potent form ofangiopoietin 1 (AAV2COMP-Ang1) can ameliorate thestructural and functional hallmarks of DR in Ins2Akitamice with sustained effects observed through sixmonths In early DR AAV2COMP-Ang1 restored leuko-cyte-endothelial interaction retinal oxygenation vasculardensity vascular marker expression vessel permeabilityretinal thickness inner retinal cellularity and retinalneurophysiological response to levels comparable withnondiabetic controls In late DR AAV2COMP-Ang1 en-hanced the therapeutic benefit of intravitreally deliveredendothelial colony-forming cells by promoting their in-tegration into the vasculature and thereby stemmingfurther visual decline AAV2COMP-Ang1 single-dosegene therapy can prevent neurovascular pathologysupport vascular regeneration and stabilize vision in DR

Diabetes affects 258 million people in the US and itsprevalence is expected to triple in the next 20 years (1)Diabetic retinopathy (DR) is the leading cause of blind-ness in the working-age population (2) The leading causeof vision loss in DR is diabetic macular edema (DME)a condition in which fluid accumulates underneath thecentral macula due to a breakdown of the blood retinalbarrier (BRB)

Current treatments for DME include laser photocoag-ulation intravitreal agents that block vascular endothelialgrowth factor (VEGF) andor intravitreal corticosteroidsSuch treatments address the downstream consequencesbut not the vascular endothelial cell loss and ischemiaunderlying DME (3) Moreover these therapies improvevision in only a minority of patients (4) Merely 23ndash33of patients treated with ranibizumab (5) and 34 ofpatients treated with aflibercept (6) achieve significantvisual gains Because it creates small burns that can in-terfere with peripheral vision and overall visual perfor-mance traditional treatment with laser photocoagulationis primarily used to retard rather than reverse retinal non-perfusion (7) Intravitreal steroids have served as an alter-native for patients who have contraindications or areresistant to anti-VEGF agents but are inferior to VEGF

1Department of Ophthalmology and Visual Sciences John A Moran Eye CenterUniversity of Utah Salt Lake City UT2Centre for Vision and Vascular Science Queenrsquos University Belfast BelfastIreland3Department of Biomedical Engineering University of Utah Salt Lake City UT4Program in Molecular Medicine Department of Medicine University of Utah SaltLake City UT5Korean Advanced Institute for Science and Technology Daejeon SouthKorea6Department of Molecular Genetics and Microbiology University of MassachusettsWorcester MA

Corresponding author Balamurali K Ambati balaambatiutahedu

Received 3 July 2014 and accepted 23 August 2015

This article contains Supplementary Data online at httpdiabetesdiabetesjournalsorglookupsuppldoi102337db14-1030-DC1

RRR and HU are condashsecond authors

copy 2015 by the American Diabetes Association Readers may use this article aslong as the work is properly cited the use is educational and not for profit andthe work is not altered

Diabetes Volume 64 December 2015 4247

COMPLIC

ATIO

NS

inhibitors in recovering visual acuity and are associatedwith side effects like cataract and intraocular hypertension(8) Given the suboptimal outcomes we developed a differ-ent approach to DME focusing on the reversal of retinalvascular damage and restoration of normal perfusion

The underlying pathogenesis of DR is largely due tohyperglycemia (9) Hyperglycemia triggers an inflammatoryresponse leading to leukocyte adhesion microvascular occlu-sion and consequent hypoxia (1011) Further hyperglyce-mia instigates pericyte loss compromising endothelialstability and BRB integrity Eventual capillary degenerationleads to retinal nonperfusion exacerbating retinal hypoxia(12) Consequent pathological VEGF-induced angiogenesis isuncoordinated and results in immature leaky vessels withinadequate perfusion creating a vicious cycle of hypoxia-driven VEGF secretion and DME (13) Retinal ganglion cell(RGC) loss neuronal dysfunction and changes in vision arealso seen in patients with DR concurrently with vascularpathology (14) Hence as a therapeutic goal vascular stabi-lization could promote normal perfusion of metabolicallydemanding retinal neurons and thereby avert the sight-threatening sequelae of ischemia and hyperpermeability

One therapeutic target is angiopoietin 1 (Ang1) a vasculargrowth factor that has an abnormally low concentration inthe vitreous of patients with DR (15) Ang1 via binding tothe Tie2 endothelial receptor fosters vessel quiescence andmaturation and suppresses vascular leakage by preventingVEGF-induced degradation of vascular endothelial (VE)-cadherin a transmembrane protein in the adherens junctionbetween endothelial cells that promotes vascular integrityand decreases vascular permeability (1617) Ang1 also pro-motes the survival of damaged vascular endothelial cellsthrough the phosphatidylinositol 3-kinaseAkt cascade (18)Thus restoration of Ang1 signaling could serve as apossible solution for preventing endothelial loss retinalischemia and abnormal VEGF expression in DR (19)

Pharmaceutical development of Ang1 as a viable ther-apy has been hindered by its insolubility and aggregationTen years ago a novel stable soluble and more potentversion of Ang1 cartilage oligomeric matrix protein Ang1(COMP-Ang1) was bioengineered to overcome the limi-tations of native Ang1 (20) Although the benefits ofCOMP-Ang1 have been demonstrated for vasculopathicdisorders in animal models of the cardiac (21) nervous(22) and renal (23) systems this study is the first totest its effectiveness for the prevention and treatmentof DR

We hypothesized that if introduced early in diabetesconstitutive expression of COMP-Ang1 via adeno-associatedvirus serotype 2ndashmediated gene therapy (AAV2COMP-Ang1) either as a monotherapy or in combination withhuman-derived endothelial colony-forming cells (ECFCs)could protect retinal neurovascular structure and functionin a type 1 diabetic Ins2Akita mouse model of DR Wefound that COMP-Ang1 prevents the endothelial losscapillary dropout BRB instability leukocyte dysfunctionneuroretinal attenuation and visual decline associated

with DR In addition COMP-Ang1 enhances the thera-peutic efficacy of ECFCs in mitigating DR through vascu-lar and visual rehabilitation

RESEARCH DESIGN AND METHODS

MiceThis research protocol was approved by the InstitutionalAnimal Care and Use Committee of the University of Utahand conforms to the standards in the Associaton forResearch in Vision and Ophthalmology Statement for theUse of Animals in Ophthalmic and Vision Research Thediabetic C57BL6-Ins2AkitaJ (Ins2Akita) mouse and itsbackground wild-type (WT) strain C57BL6J were used(24) Mice heterozygous for the Ins2 mutation experienceprogressive retinal abnormalities 12 weeks after the onsetof hyperglycemia including apoptosis (ie endothelialand RGC loss) and functional deficits (increased vascularpermeability and decreased neuronal function) (25)

Mice were randomly assigned to one of three experi-mental groups AAV2COMP-Ang1 AAV2AcGFP (Aequoreacoerulescens green fluorescent protein) or PBS At 2 monthsof age (Supplementary Fig 3) each mouse was anesthetizedby injection of 125 tribromoethanol (Sigma-AldrichSt Louis MO) at a dose of 0025 mLg body wt ip Eachmouse was treated with either 2 mL AAV2 solution (203 109

particles) or PBS injected into the vitreous cavity of botheyes with a 33-gauge microsyringe (Hamilton CompanyReno NV) An additional cohort of mice was treated asdescribed above at 6 months of age Two weeks later thiscohort received a second intravitreal injection with 1 mLof 1 3 105 ECFCs (Supplementary Fig 3)

Virus Vector ConstructionThe plasmids pAAVCOMP-Ang1 and pAAVAcGFP werecreated by incorporating the COMP-Ang1 cDNA fromthe pCMV-dhfr2-COMP-Ang1 (donated generously by theKoh Laboratory KAIST Daejeon) into pAAV-MCS (Agi-lent Technologies Santa Clara CA) while pAAVAcGFPwas created by the same technique with AcGFP cDNAfrom the pIRES2-AcGFP1 plasmid (Clontech LaboratoriesMountain View CA) (Supplementary Fig 3)

For in vivo assays requiring imaging with the SpectralisHRA+OCT (Heidelberg Engineering Heidelberg Germany)mice were anesthetized by an inhalation of 3 isofluraneO2

mixture in a closed canister at a flow rate of 10 Lpm Pupilswere dilated with a 1 tropicamide

Retinas were harvested and processed for in situhybridization (ISH) according to standard protocols usingprocedures to avoid RNAse contamination

RT-PCR for COMP-Ang1 mRNA ExpressionPrimer sequences for COMP-Ang1 (University of Utah GeneCore Salt Lake City UT) were as follows COMP-Ang1 F59-GCTCTGTTTTCCTGCTGTCC-39 and COMP-Ang1 R 59-GTGATGGAATGTGACGCTTG-39 Primer sequences for the internalcontrol GAPDH were as follows GAPDH F 59-AACTTTGGGATTGTGGAAGGG-3 and GAPDH R 59-ACCAGTGGATGCAGGGATGAT-39G

4248 AAV2COMP-Ang1 for Diabetic Retinopathy Diabetes Volume 64 December 2015

Immunoprecipitation and Western BlotRetinas were harvested and protein lysate sampleswere immunoprecipitated with anti-FLAG M2 affinity gel(Sigma-Aldrich) Eluted proteins samples were run on12 SDS-PAGE Overall protein levels were comparedwith anti-GAPDH (13000 Abcam Cambridge MA) andantindashb-actin (13000 Abcam) Samples were tested forantindashVEGF-A (1200 Santa Cruz Biotechnology SantaCruz CA) antindashVE-cadherin (11000 Abcam) and antindashphospho-Src (PY419 1 mgmL RampD Systems Minneap-olis MN)

Immunofluorescence of Retinal Flat Mounts forVascular MarkersEnucleated globes were fixed in 4 paraformaldehydefollowed by retinal dissection Specimens were stainedwith 1200 a-smooth muscle actin antibody or neuron-glial antigen 2 antibody conjugated with cyanine dye(Cy3) (Sigma-Aldrich) and 5 mgmL Alexa Fluor 647ndashconjugated isolectin GS-IB4 (Invitrogen) in blockingbuffer overnight at 4degC After washing the retina wasflat mounted on a glass slide Full retinal field immuno-fluorescence (IF) images were captured at low magnifica-tion followed by increasing magnification of each quadrantwith scanning laser confocal microscopy (OlympusAmerica)

Trypsin Digest for Retinal Vascular ArchitectureEnucleated globes were fixed in 2 formalin The retinawas detached around the subretinal space The optic nervewas cut under the disc The specimen was digested at 37degCin 25 trypsin02 molL TRIS at pH 80 for 30ndash60 minSpecimens were transferred to distilled water and to 05triton X-100 surfactantdistilled water and left at roomtemperature for another hour Lastly the specimen wasmoved to 01 triton X-100distilled water for mountingand dried in a 37degC incubator Samples were stained withperiodic acid Schiff staining and imaged with light micros-copy (Olympus Center Valley PA) Acellular capillariesand pericytes were counted in each of a total of fivehigh-powered fields per retinal quadrant

Transendothelial Electrical Resistance for BRBIntegrityIn vitro measurements of transendothelial electrical re-sistance (TER) were performed with an electrical cellsubstrate impedance sensing (ECIS) system (AppliedBiophysics Troy NY) Human retinal microvascularendothelial cells (HrMVECs) (Cell Systems Kirkland WA)were seeded (50000 cellswell) onto fibronectin-coatedgold microelectrodes in ECIS culture wells (8W10E+Applied Biophysics) and incubated overnight at 37degC incomplete medium (EBM-2 plus EGM2-MV supplementsLonza Group Basel Switzerland) until cell resistancereached a plateau Cells were serum starved for 1 h untilresistance was stabilized (1200 V) Each well receivedone of three experimental treatments COMP-Ang1 protein(Enzo Life Sciences Farmingdale NY) VEGF protein (RampDMinneapolis MN) or PBS Monitoring was continued for

21 h The data from triplicate wells were averaged and pre-sented as normalized resistance versus time

Miles Assay for Retinal VasopermeabilityMice were administered Evans blue (EB) (Sigma-Aldrich)at a dosage of 20 mgkg through tail vein injection After4 h the vasculature was perfused with PBS The retinaswere next harvested and placed in formamide at 70degC for18 h Samples were centrifuged for 2 h at 40000g ina 02-mm filter EB concentration was detected spectro-photometrically by subtracting absorbance at 620 nmfrom 740 nm

Acridine Orange Fluorography for LeukocyteTransendothelial MigrationAcridine orange (Acros Organics Geel Belgium) (010PBS) was filtered with a 022-mm filter The solution(005 mLmin for a total of 1 min) was injected into thetail vein Imaging used Spectralis HRA+OCT (HeidelbergEngineering) with a 488-nm argon blue laser with a stan-dard 500-nm long-pass filter Images were acquired fromboth eyes with a 55-degree lens using the movie mode onthe Spectralis HRA+OCT

Flow Assay for Leukocyte-Endothelial InteractionHrMVECs were cultured in parallel-plate fibronectin-coated flow chambers (m-Slide VI 04 ibidi USA Madi-son WI) until 80 confluent and exposed to tumornecrosis factor-a (TNF-a) (10 ngmL) (RampD Systems)or vehicle control for 3 h Human leukocytes were iso-lated as previously described in accordance with Institu-tional Review Board guidelines (26) and diluted inwarmed ultrasaline (Lonza Group) to 1 3 106 cellsmLLeukocytes were pumped through the parallel plate flowchambers using a syringe pump (Harvard ApparatusHolliston MA) at 1 dynescm2 (typical venous shearstress) Differential interference contrast images wereacquired at a rate of 1s for 1 min and the number ofleukocytes adhered to or rolling on the monolayer wasquantified as leukocytesframessecond Three indepen-dent flow wells were averaged to attain the reportedvalues

Immunoperoxidase Staining for Retinal HypoxiaMice received an intraperitoneal injection of the biore-ductive hypoxia marker pimonidazole (HypoxyprobeBurlington MA) at 60 mgkg (3) Three hours later ret-inas were harvested and stained with a hypoxyprobe-1monoclonal antibody conjugated to fluorescein isothiocya-nate to detect reduced pimonidazole adducts (Hypoxyprobe)forming in pO2 10 mmHg

Optical Coherence Tomography for Retinal ThicknessMice were imaged bilaterally with optical coherencetomography (OCT) (Spectralis HRA+OCT HeidelbergEngineering) Retinal cross-sectional thickness was mea-sured 250 mm relative to the optic nerve head using theen face image as a guide Measurements were recorded foreach retinal quadrant and averaged to attain the reportedvalues

diabetesdiabetesjournalsorg Cahoon and Associates 4249

Microsphere Fluorescence for RetinalVasopermeabilityEach mouse received a dosage of 100 mL20 g tail veininjections with 100 nm microspheres (Magsphere Pasa-dena CA) conjugated to either the near-infrared fluoro-phore ZW800 (Flare Foundation Boston MA) (4) or GFP(Magsphere) Bilateral imaging by Spectralis HRA+OCT(Heidelberg Engineering) was performed with fluoresceinangiography (FA) and indocyanine green modalities

Immunofluorescence for Retinal Thickness andVE-CadherinEnucleated globes were fixed in 4 paraformaldehydeThe globes were cut in 10-mm sections and stainedwith antindashVE-cadherin antibody (1200 Abcam) and DAPI(Sigma-Aldrich) Sections were captured with scanninglaser confocal microscopy (Olympus America)

Immunofluorescence of Retinal Flat Mounts for RGCDensityFour-month-old mice received intravitreal injections ofeither AAV2COMP-Ang1 or PBS as described above Sixmonths later retinas were fixed and dissected as de-scribed above for flat mounts Specimens were labeledwith 1200 pan-Brn3 antibody (Santa Cruz Biotechnol-ogy) followed by Alexa Fluor 546ndashconjugated secondaryantibody (Invitrogen) and counterstained with DAPI(Sigma-Aldrich) Eight fields were imaged for each retinausing the confocal340 oil objective these comprised fourevenly spaced fields (one per quadrant) adjacent to theoptic nerve and four fields (similarly spaced) near the flatmount periphery Images were counted blind by two sep-arate investigators Counts were averaged for each retinaand compared across control and experimental groups

Electroretinography for Retinal FunctionMice were dark adapted anesthetized with ketaminexylazine (90 mg10 mgkg body wt) and placed on acontrolled warming plate (TC-1000 CWE InstrumentsArdmore PA) (5) Electroretinograms (ERGs) were takenbetween a gold corneal electrode and a stainless steelscalp electrode with a 03- to 500-Hz band-pass filter(UTAS E-3000 LKC Technologies Gaithersburg MD)The photoflash unit was calibrated to deliver 25 cd sm2

at 0 dB flash intensity and scotopic measurements wererecorded with flash intensities increasing from 00025 to250 cd sm2 The b-wave amplitudes were determined inscotopic conditions and the mean values at each stimu-lus intensity were compared with an unpaired two-tailedt test

Optokinetic Tracking for Visual AcuityOptomotor reflex-based spatial frequency threshold testswere conducted in a visuomotor behavior measuringsystem (OptoMotry CerebralMechanics Lethbridge ABCanada) Tracking was defined as a reproducible smoothpursuit with a velocity and direction concordant with thestimulus Trials of each direction and spatial frequencywere repeated until the presence or absence of thetracking response could be established unequivocally

Rotation speed (12degs) and contrast (100) were keptconstant

ECFCsFresh human cord blood donated under full ethicalapproval by healthy volunteers at the Northern IrelandBlood Transfusion Service (Belfast UK) underwent densitygradient fractionation for the isolation of mononuclear cellsand was selected for ECFCs via resuspension in completemedium (EBM-2 plus EGM-2 MV supplements LonzaGroup) supplemented with 10 FBS and seeding onto24-well culture plates precoated with rat tail collagen type1 (BD Biosciences Bedford UK) at a density of 1 3 107

cellsmL Cells were labeled (Qtracker 655 InvitrogenLife Technologies Carlsbad CA) per the manufacturerrsquosinstructions

Immunofluorescence of Retinal Flat Mounts for ECFCEngraftmentFrom the mice that had received intravitreal ECFCsharvested retinas were fixed and dissected as describedabove for flat mounts (Supplementary Fig 3B) Specimenswere stained with 5 mgmL Alexa Fluor 647ndashconjugatedisolectin GS-IB4 (Invitrogen) and mounted on a glass slideas described above ECFC integration was counted in fourhigh-powered fields in each retinal quadrant

Scratch Assay for ECFC MigrationECFCs were plated on rat-tail collagen-coated sixwellplates prelabeled with traced lines (27) When cellswere 90 confluent a uniform straight scratch wasmade in the monolayer using a 200-mL pipette tip Afterinjury cells were washed medium was changed and ref-erence photographs were taken within each regionmarked by the lines using a phase contrast microscope(Eclipse E400 Nikon Tokyo Japan) Wells were incu-bated with the experimental treatment (doses ofCOMP-Ang1 protein from 0ndash1000 ngmL) and imageswere captured at hourly intervals Endothelial cell migra-tion was quantified by calculating the proportion of de-nuded area

Matrigel Assay for ECFC TubulogenesisECFCs were labeled with a fluorescent dye (PKH CellLinker kit for General Cell Membrane Labeling Sigma-Aldrich) as previously described (28) Next they weresuspended in growth factor reduced basement membranematrix (Matrigel Becton Dickinson Biosciences FranklinLakes NJ) and 50-mL aliquots were spotted onto a24-well plate Spots were covered with DMEM contain-ing 5 porcine serum and treated with either control orincreasing doses of COMP-Ang1 After 24 h wells wereassessed for the presence of tubules Images were ac-quired by using a laser confocal microscope (Nikon)

Statistical AnalysisAll numerical data were analyzed in Excel (MicrosoftRedmond WA) and presented as the mean 6 SD Studenttwo-tailed t test with a level 005 was used to comparedifferences between two samples ANOVA test with


P 005 followed by Tukey post hoc analysis was usedto compare differences between four groups

RESULTS

Intravitreal AAV2 Gene Therapy Is Safe for the MouseRetinaThe long-term safety of intraocular AAV-mediated genetherapy has been validated by numerous animal andhuman studies in which vector inoculation was welltolerated and unaccompanied by any structural orfunctional defects (29) We recently reported that theintraocular injection of exogenous AAV2 constructs didnot impact retinal thickness influence ERG responsesor increase the risk of retinochoroidal apoptosis (30)Concordant with prior results the current study foundno significant differences (P = 03 at 230 dB 06 at0 dB and 06 at 20 dB) in scotopic and photopicb-wave amplitudes on ERG between WT mice treatedwith intravitreal AAV2COMP-Ang1 AAV2AcGFP orPBS (Supplementary Fig 1A) Furthermore optokinetictracking (OKT) response of no-injection WT controlmice did not change appreciably compared with PBS-treated mice (P = 05) over the course of 5 weeks (Sup-plementary Fig 1B)

Intravitreal AAV2COMP-Ang1 Expresses COMP-Ang1in the Mouse RetinaAAV2 localization and transfection in the retina wereconfirmed by in vivo and ex vivo detection of thefluorescent gene product of the sham viral control byconfocal microscopy further COMP-Ang1 mRNA andprotein expression was verified by ex vivo immunoassay

AcGFP fluorescence was initially observed at 1 weekpostinjection and persisted through to the 4 monthspostinjection end point (Supplementary Fig 2A) Asanticipated based on previous experience with intra-vitreal AAV2 (31) AcGFP signal was visualized in allretinal quadrants at the level of the ganglion cellndashinner plexiform layer (GC-IPL) (Supplementary Fig 2Band C)

COMP-Ang1 production in the mouse retina wasdemonstrated via semiquantitative RT-PCR (Supplemen-tary Fig 2D) ISH (Supplementary Fig 2FndashH) and immu-noblotting (Supplementary Fig 2E) at the 4 monthspostinjection end point In parallel with AcGFP fluores-cence ISH revealed increased amounts of COMP-Ang1mRNA in the inner retina predominantly in the GC-IPL(Supplementary Fig 2FndashH)

COMP-Ang1 Prevents Breakdown of VascularStructureThe principal morphologic features of DR are pericyteendothelial cell and capillary dropout (32) Concordantlyretinal vessel density was discernably reduced in IF andtrypsin digest images of Ins2Akita control (PBS andAAV2AcGFP) versus WT retinas (Fig 1AndashC) Endothelialcell (P 001) (Fig 1D) and pericyte (P 001) (Fig1E) coverage was significantly decreased in the diabetic

controls compared with WT while capillary acellularitywas increased (P 001) (Fig 1F)

IF and trypsin digest images of AAV2COMP-Ang1exhibit an improvement in vessel density compared withdiabetic controls (Fig 1AndashC) accompanied by a significantdrop in acellular capillary density (P 001 vs AAV2AcGFPP 001 vs PBS and P 001 vs WT) (Fig 1F)AAV2COMP-Ang1 significantly rescued endothelial cellloss relative to controls (WT 232 AAV2COMP-Ang1205 PBS 155 AAV2AcGFP 153 P 001)(Fig 1D) but not pericyte coverage (WT 65 AAV2COMP-Ang1 38 PBS 43 AAV2AcGFP 39 P = 09)(Fig 1E)

Previous reports have demonstrated that the rescueof pericytes the major source of endogenous Ang1 inthe capillary unit (33) is not necessary for Ang1 rescueof function as long as Ang1 is available in adequatequantities (34) Our data suggest that in lieu of peri-cytes AAV2COMP-Ang1 can provide sufficient Ang1endothelial trophic signaling to prevent capillary drop-out (Fig 1F)

COMP-Ang1 Promotes BRB IntegrityIschemia from capillary dropout in DR stimulates VEGFproduction and consequent vascular hyperpermeability(3) Accordingly the TER of HrMVECs decreased aftertreatment with VEGF compared with PBS in vitro (P 001) (Fig 2A) while EB extravasation was elevated inPBS- and AAV2AcGFP-treated diabetic controls (38-and 31-fold respectively) (Fig 3A) compared with WTmice (P 001) Microsphere leakage was similarly ele-vated in diabetic controls (Fig 3B)

COMP-Ang1 significantly increased the TER (P 001)(Fig 2A) of HrMVECs and decreased EB extravasation(P 001) (Fig 3A) and microsphere (Fig 3B) leakagein diabetic mice

These results indicate that COMP-Ang1 restores thebarrier function of the retinal vasculature in DR Toexplore the molecular underpinnings of this finding weinvestigated the influence of COMP-Ang1 on VEGF-Athe proto-oncogene nonreceptor tyrosine kinase Src andthe intercellular junction adhesion molecule VE-cadherinVEGF-A is known to induce vessel leakage in DR throughSrc-mediated downregulation of VE-cadherin while Ang1is known to upregulate VE-cadherin (1635)

COMP-Ang1 decreased Src phosphorylation (Fig 2B)and increased VE-cadherin expression in HrMVECs(Fig 2C) Both Ins2Akita and WT retinas treated withAAV2COMP-Ang1 had reduced levels of VEGF-A (Fig 2Dand Supplementary Fig 4) and increased levels of VE-cadherin (Fig 2E and Supplementary Fig 4)

Ang1 acts on the phosphatidylinositol 3-kinaseAktcascade to prevent the apoptosis of damaged vascularendothelial cells (18) Considering the reduction in capillaryacellularity and endothelial cell loss with COMP-Ang1 weexplored Akt phosphorylation as a possible mechanismfor COMP-Ang1ndashmediated survival of endothelial cells


COMP-Ang1 increased Akt phosphorylation at the serine473 residue in both ECFCs and human umbilical veinendothelial cells (Fig 2F)

COMP-Ang1 Reduces Leukocyte-EndothelialAdhesion and LeukostasisDR is characterized by a chronic subclinical inflammatoryresponse that is thought to play a critical role in itspathogenesis (36) The less deformable and more activatedleukocytes in DR are conjectured to contribute to retinalnonperfusion and capillary dropout through increased

attachment to endothelial cells and entrapment withinthe capillaries (37)

Leukocyte adhesion to the vascular wall is mediatedin part by TNF-a (38) Correspondingly the endothelialmonolayer of HrMVECs exposed to TNF-a experienced anabnormally high rate of leukocyte adherence Treatmentwith COMP-Ang1 protein decreased the number of adher-ent leukocytes per minute by 80 (P 001) (Fig 2G)

On Acridine orange leukocyte fluorography (39) leu-kocyte rolling was significantly elevated in Ins2Akita con-trol (98 cellsmin) versus WT retinas (3 cellsmin P 001)

Figure 1mdashAAV2COMP-Ang1 mitigates diabetic retinal capillary dropout A Representative retinal flat mounts prepared from 6-month-oldmice and stained for isolectin (endothelial cell marker [green]) and a-smooth muscle actin (smooth muscle marker [red]) B Magnified viewof retina stained with isolectin and neuron-glial antigen 2 (pericyte marker) Ins2Akita mice experienced pericyte and endothelial dropoutthe latter was prevented with a single intravitreal dose of AAV2COMP-Ang1 C Trypsin digest featuring retinas representative of eachgroup Black arrowheads denote acellular capillaries Quantification using ImageJ of endothelial coverage (D) and pericyte coverage (E) FAcellular capillaries were manually counted and averaged over an area 1 mm2 Eight eyes were used in each analysis Data are mean6 SDP lt 001 ANOVA Post hoc comparisons with a Tukey test to compare means of each group Scale bars = 600 mm (A) 100 mm (B) and200 mm (C)


(quantitative image Fig 3C and representative image Fig 3Dwith white arrows pointing to leukocyte aggregations at thebifurcation) AAV2COMP-Ang1 was able to reduce this rateto below the disease-free baseline (28 cellsmin P 001)(Fig 3C and Supplementary Videos 1ndash4)

These results indicate that the improvement in vascularparameters by COMP-Ang1 may have an anti-inflammatorycomponent

COMP-Ang1 Reduces HypoxiaLeukostasis has been proposed as a mechanism of capillarynonperfusion and retinal hypoxia (40) Since hypoxia isa potent inducer of VEGF-A we further assessed the re-lationship between COMP-Ang1 and hypoxia Pimonida-zole staining was increased in the diabetic control micerelative to WT mice whereas it was reduced nearly tobaseline levels in AAV2COMP-Ang1 mice (Fig 3E)

Collectively our outcomes demonstrate a pathway forCOMP-Ang1ndashmediated retinal vascular functional stabi-lization The inhibition of leukocyte adhesion and

stimulation of Akt phophorylation lead to the preserva-tion of perfusion and normalization of tissue oxygenationwhereas the inhibition of VEGF-A and Src phos-phorylation lead to the preservation of VE-cadherin andnormalization of permeability

COMP-Ang1 Prevents Retinal Neuronal DysfunctionDR causes neural degeneration of the inner retina (14)Consistent with this the retinas on OCT and IF imagesfrom Ins2Akita control mice (185 mm) were qualitativelyand quantitatively thinner than in WT mice (210 mm P001) (Fig 4A-D) In parallel GC-IPL cell density was alsodecreased in Ins2Akita control (68 nuclei300 mm length)versus WT (45 nuclei300 mm length) retinas (P 0001)(Fig 4BndashD) a 34 loss of cells

AAV2COMP-Ang1 preserved retinal thickness (WT210 mm AAV2COMP-Ang1 205 mm AAV2AcGFP 181 mmand PBS 185 mm P 001) (Fig 4A and D) and preventedloss of cells in the GC-IPL (65 nuclei300 mm length P = 003)in Ins2Akita mice (Fig 4BndashD)

Figure 2mdashCOMP-Ang1 increases endothelial integrity A Representative graph of ECIS of HrMVECs with COMP-Ang1 (100 ngmL) VEGF(50 ngmL) or control (PBS) added to the media COMP-Ang1 increased resistance of HrMVECs (n = 3) Increases in endothelial resistancewere correlated with deceased Src phosphorylation (B) and increased VE-cadherin (C) in HrMVECs as demonstrated by Western blot (n =3) Western blot from Ins2Akita mouse retinas demonstrating decreased VEGF-A (D) and increased VE-cadherin (E) in mice treated withAAV2COMP-Ang1 COMP-Ang1 increased Akt phosphorylation at the serine 473 residue in both ECFCs and human umbilical veinendothelial cells (HUVECs) G COMP-Ang1 reduced TNF-andashinduced leukocyte rolling in cultured HrMVECs (n = 6 per group) P lt001 ANOVA hrs hours


Further characterization of GC-IPL cells with the RGC-specific marker Brn3 revealed no difference in RGC countswithin the central retinas of PBS versus COMP-Ang1ndashtreated Ins2Akita mice (P = 07) (Fig 4E) However pe-ripheral retinas showed a 17 loss of ganglion cells (Fig4F) Although this difference was not statistically signifi-cant (P = 007) the trend shows an effect size similar tothat reported for RGC loss in the peripheral retina ofdiabetic versus WT mice (41) suggesting that a largersample size would provide sufficient power to confirman effect

In sum these data suggest that AAV2COMP-Ang1 isbeneficial in preventing diabetes-induced GC-IPL atrophyand peripheral RGC cell loss but may also target orrecruit non-RGC cell types within the inner retina forneuroprotection

COMP-Ang1 Prevents Visual DysfunctionPatients with DR manifest with visual deficits early in thedisease and animals exhibit changes in visual acuity andcontrast sensitivity through impaired visual trackingbehavior and delayed retinal electrical responses (42) In

Figure 3mdashCOMP-Ang1 enhances vascular barrier function and reduces retinal hypoxia in the diabetic retina A EB extravasation from theretina of Ins2Akita mouse was increased compared with control treatment with AAV2COMP-Ang1 returned vascular hyperpermeabilty tocontrol levels Eyes from eight mice were used in each analysis data are mean6 SD Plt 001 compared with WT P = 002 compared withAAV2AcGFP B FA did not reveal any leakage in diabetic mice however with GFP or the near-infared fluorophore ZW800 conjugated toaminated latex microspheres (GFP-ms or ZW800ndashms 100 nm in diameter) in vivo leakage was captured using the FA or indocyanine green(ICG) imaging modality on Spectralis respectively Note that background GFP fluorescence of the AAV2AcGFP-treated diabetic mice maskedsignal from the GFP microspheres C Diabetes induced leukocyte rolling in the retinal vasculature as captured by acridine orange leukocytefluorography D Representative image of acridine orange leukocyte fluorography with white arrowheads pointing to adherent and rollingleukocytes COMP-Ang1 prevents leukostasis and inflammation in this model of diabetic retinopathy (See also Supplementary Video 1) ERepresentative retinas (four mice per group) from mice treated with hypoxyprobe (pimonidazole) COMP-Ang1 reduced hypoxia in diabeticmouse retinas Scale bars 600 mm (E) P lt 001 ANOVA Post hoc comparisons with a Tukey test to compare means of each group


line with this ERG and OKT responses were within normallimits for WT mice but abnormally depressed in diabetic con-trol mice (at 240 dB WT 127 mvolts PBS = 57 mvolts andAAV2AcGFP 73 mvolts P 001 at 4 months postinjectionWT = 0388 cyclesdegree PBS = 0184 cyclesdegree andAAV2AcGFP = 0174 cyclesdegree P 001) (Fig 5AndashC)

AAV2COMP-Ang1 treatment diminished the dampen-ing in scotopic b-wave amplitudes (185 mvolts P 001)(Fig 5A and B) caused by anomalous photoreceptor-bipolarcommunication in DR (43) Likewise Ins2Akita mice treated

with AAV2COMP-Ang1 were able to avert the deteriorationof OKT (0312 cyclesdegree P 001) (Fig 5C)

Together spatial resolution and ERG data show thatAAV2COMP-Ang1 can preserve retinal neurophysiologi-cal function

COMP-Ang1 Enhances ECFC Treatment EffectThe recellularization and resultant refunctionalization ofacellular capillaries could theoretically halt the nonperfu-sion at the root of DR pathophysiology but in diabetes

Figure 4mdashAAV2COMP-Ang1 prevents diabetes-induced GC-IPL degeneration A Representative figures from OCT measuring retinalthickness The red line generated by OCT software indicates the retinal surface and Bruch membrane Scale bars = 200 mm B Cross-sections of 6-month-old retinas from WT or Ins2Akita mice treated with PBS AAV2AcGFP or AAV2COMP-Ang1 stained with DAPI CView of the GC-IPL frommice stained for VE-cadherin (red) or nuclei (DAPI blue) demonstrating increased VE-cadherin and nuclear stainingScale bars = 30 mm (right) D Quantification of retinal thickness from OCT showing that AAV2COMP-Ang1 prevented diabetes-induced retinal thinning as measured in vivo (P lt 001 vs both WT and AAV2AcGFP) AAV2COMP-Ang1 prevented diabetes-inducedinner retinal layer loss (P = 003 ANOVA with post hoc Tukey test) as measured by nuclei counted in the GC-IPL in retinal cross-sectionsE Representative retinal flat mounts stained with BRN3 (red [a marker for retinal ganglion cells]) isolectin (green [marker for vessels]) and DAPI(blue) Qualitatively fewer peripheral RGCs are observed in PBS-treated Akita retina F Ins2Akita mice showed no difference in central RGCswith either treatment but there was a trend toward reduced peripheral RGCs in PBS-treated mice vs COMP-Ang1ndashtreated mice (17 fewerganglion cells in PBS group [P = 007]) At least six eyes from each group were tested Data are mean 6 SD P lt 001 ANOVA Post hoccomparisons with a Tukey test to compare means of each group c central p peripheral


the endogenous reparative cells responsible for this taskhave a decreased ability to associate with existing vascularnetworks (44) The exogenous delivery of human-derivedECFCs as evidenced by their utility in oxygen-inducedretinopathy (45) may be able to compensate for this de-ficiency but have not yet been explored in the context ofDR We know that ECFCs express high levels of the Ang1receptor Tie2 (46) and that Ang1 promotes the differen-tiation of stem cells into vasculogenic cells for vessel en-graftment and reformation Therefore we tested theregenerative potential of dual therapy with COMP-Ang1and ECFCs

COMP-Ang1 demonstrated a dose-dependent increase in6-h migration (25-fold increase over control at 10 ngmLP 001) (Fig 6A) and 24-h tubulogenesis (43-fold overcontrol P 001) (Fig 6B) of ECFCs in vitro as assessedby scratch migration assay and matrigel tube formation as-say respectively In vivo aged 26-week-old Ins2Akita micetreated with AAV2COMP-Ang1 had increased 72-h ECFCvessel integration on confocal microscopy (Fig 6C and Sup-plementary Videos 5 and 6) and 2-month visual responsewith OKT (AAV2COMP-Ang1 = 0307 cyclesdegree AAV2AcGFP = 0251 cyclesdegree and PBS = 0263 cyclesdegreeP 001) (Fig 6D)

Our results show that COMP-Ang1 boosts the capacityof ECFCs to rebuild vessels and counteract vision loss inDR

DISCUSSION

This study established the salutary effects of COMP-Ang1in ameliorating pivotal pathogenic events in the trajectorytoward DME through neurovascular normalization Struc-tural and functional indices of neurovascular restorationto a state more consistent with a homeostatic disease-free phenotype by COMP-Ang1 included endothelial and

capillary density vessel permeability VEGF-A phospho-Src VE-cadherin and phospho-Akt levels leukocyte-endothelial interaction and retinal hypoxia neuroretinalthickness GC-IPL cellularity and peripheral RGC densityand most importantly vision Furthermore COMP-Ang1augmented the effectiveness of ECFCs in regeneratingvessels and stabilizing vision in advanced DR Based onthese results our working model for further study centerson the premise of COMP-Ang1 directly or indirectlysuppressing inflammation and modulating the actions ofVEGF Src VE-cadherin and Akt at the molecular levelthereby influencing downstream processes of perfusionand BRB reinforcement vital for neurovascular stability

Our study corroborates and extends work that showedthat AAV-mediated Ang1 gene therapy could suppressvascular leakage in stroke (47) and intravitreal recombi-nant COMP-Ang1 protein could suppress vascular leakagein choroidal neovascularization (48) Our results advancethe field by 1) demonstrating that sustainable delivery ofCOMP-Ang1 to the retina can be achieved with a singleintravitreal injection 2) testing the therapeutic effects ofCOMP-Ang1 in an animal model of DR and 3) introduc-ing the potential for COMP-Ang1 to be used in conjunc-tion with ECFCs for the treatment of advanced DR

In addition our data raise a number of intriguingquestions about the mechanism of action for COMP-Ang1in DR Intriguingly we found that both diabetic andnondiabetic mice experienced decreases in VEGF pro-duction due to COMP-Ang1 exposure inviting inquiryabout the nature of this relationship and how it isintertwined with oxygen supply and macrophage secre-tion (48) Equally fascinating and consistent with evi-dence of Ang1 neuroprotection in the central nervoussystem (49) was our discovery that COMP-Ang1 seemsto preserve peripheral RGC density Ins2Akita mice

Figure 5mdashAAV2COMP-Ang1 prevents diabetes-induced neural dysfunction A Representative example of ERG response from all groupsof mice Electrical retinal response was elicited and the amplitude of b-wave during scotopic conditions at 2362 log (Cd sm2) (240 dB)2262 log (Cd sm2) (230 dB) and 2162 log (Cd sm2) (220 dB) intensity was recorded B Decreased amplitudes were recorded inIns2Akita mice treated with PBS or AAV2AcGFP compared with WT mice and AAV2COMP-Ang1 prevented the decrease in amplitudeP lt 001 ANOVA P lt 001 compared with AAv2AcGFP Assessing visual acuity was accomplished by testing optomotor trackingresponse of Ins2Akita mice treated with AAV2COMP-Ang1 or control compared with WT C Ins2Akita mice exhibited decreased opto-kinetic tracking response (measured as cyclesdegree) AAV2COMP-Ang1 (COMP) prevented the decrease in visual response at least sixmice from each group were tested Data are mean6 SD Plt 001 ANOVA Post hoc comparisons with a Tukey test to compare means ofeach group


reportedly lose RGCs in the peripheral but not centralretina within 3 months of developing diabetes (41) Wespeculate this geographic predilection for both RGC apo-ptosis and COMP-Ang1 rescue may be tied to the conceptof increased vulnerability to hypoxia with distance fromthe central artery along with our finding that COMP-Ang1 reduces hypoxia Moreover since RGCs accountedfor only 50ndash55 of the GC-IPL cell density conserved byCOMP-Ang1 in our study an exploration into other neu-roretinal cell types (eg cholinergic amacrine cells [50]) astargets for COMP-Ang1 rescue is merited

We have shown here that COMP-Ang1 is a safe andeffective replacement for endogenous Ang1 which canadequately compensate for deficient Ang1 secretion bypericytes Our results thus far indicate that COMP-Ang1suppresses the pathognomonic features of nonprolifer-ative DR and in contrast to existing therapies decreasesthe nonperfusion and ischemia critical to the genesis ofproliferative DR This latter finding along with the long-term duration of action for a single intravitreal injectionof AAV2COMP-Ang1 relative to anti-VEGF agents holdsimmense promise for fulfilling an unmet need in the

Figure 6mdashAAV2COMP-Ang1 enhances ECFC engraftment into the diabetic retina and prevents further visual decline A ECFCs wereplated on collagen-coated wells and assayed for migration potential under increasing doses of COMP-Ang1 B Additionally three-dimensional tube formation was tested in matrigel COMP-Ang1 increased migration and tube formation in a dose-dependent mannerwith maximal effects exerted at 500 ngmL Qdot-655ndashlabeled ECFCs were injected intravitreally into aged diabetic mice (6 months[arrow in D]) after the mice had been treated with COMP-Ang1 or control Three days later retinas were harvested and stained for bloodvessels (isolectin 546) and flat mounted for confocal analysis COMP-Ang1 increased ECFC integration into the diabetic retinal vascu-lature D Mice treated with COMP-Ang1 or control plus ECFCs were analyzed for visual tracking ability COMP-Ang1 plus ECFCsprevented further declines in spatial frequency threshold P lt 0001 in vitro experiments were performed in triplicate on three differentECFC clones (total of nine experiments per condition) In vivo experiments were performed on five mice per group (10 eyes) Scale bars(C ) 600 mm (top) 150 mm (middle) and 90 mm (bottom) P lt 001 ANOVA Post hoc comparisons with a Tukey test to compare meansof each group au arbitrary units


management of DR if COMP-Ang1 can be successfullytranslated from the bench to the bedside Future researchwill focus on investigating the mechanism of action bywhich COMP-Ang1 safeguards the retina and the appli-cation of COMP-Ang1 to human models of DR

Acknowledgments The authors thank Srinivas P Sangly (IndianaClinical and Translational Sciences Institute) Jayakrishna Ambati (University ofKentucky) and Valeria Tarallo and Derick Holt (University of Utah) for insightfuland constructive discussions The authors also acknowledge Andrew WeyrichGuy Zimmerman and Robert Campbell from the University of Utah for isolatingleukocytesFunding This work was partly supported by National Eye Institute NationalInstitutes of Health grants 5R01EY017950 5R01EY017182 and P30EY14800the University of Utah T-32 Neuroscience Training grant 5T32DC008553-05 theUniversity of Utah Metabolism T-32 Training grant 5T32-DK-091317 James AHaley Veteransrsquo Hospital (the VA SPIRE 5 I21 RX001597-02) the Diabetes Re-search Center at Washington University in St Louis (grant 5 P30 DK-020579)and the University of Utah Diabetes and Metabolism Center seed grant This workwas also supported in part by an unrestricted grant from Research to PreventBlindness Inc New York NY to the Department of Ophthalmology and VisualSciences University of UtahDuality of Interest JMC HU and BKA have filed a provisional patentapplication relating to the content of this article No other potential conflicts ofinterest relevant to this article were reportedAuthor Contributions JMC RRR LSC HU XZ CLO RJMPRO SN MMF and WBM performed animal studies JMC RRR HUKW and AWS performed cell culture studies JMC RRR PRO WBMPB and DK performed visual functional studies JMC KW BJA CCGand DYL performed transendothelial electrical resistance studies HU GYKand GG developed the plasmids and viral vectors SKD and SKM wereresponsible for in situ hybridization JMC RRR LSC BJA and BKAprepared and wrote the manuscript BKA is the guarantor of this work and assuch had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis

References1 Yau JWY Rogers SL Kawasaki R et al Meta-Analysis for Eye Disease(META-EYE) Study Group Global prevalence and major risk factors of diabeticretinopathy Diabetes Care 201235556ndash5642 Wild S Roglic G Green A Sicree R King H Global prevalence of diabetesestimates for the year 2000 and projections for 2030 Diabetes Care 2004271047ndash10533 Frank RN Diabetic retinopathy N Engl J Med 200435048ndash584 Agarwal A Sarwar S Sepah YJ Nguyen QD What have we learnt about themanagement of diabetic macular edema in the antivascular endothelial growthfactor and corticosteroid era Curr Opin Ophthalmol 201526177ndash1835 Mitchell P Bandello F Schmidt-Erfurth U et al RESTORE study group TheRESTORE study ranibizumab monotherapy or combined with laser versus lasermonotherapy for diabetic macular edema Ophthalmology 2011118615ndash6256 Heier JS Boyer D Nguyen QD et al CLEAR-IT 2 Investigators The 1-yearresults of CLEAR-IT 2 a phase 2 study of vascular endothelial growth factor trap-eye dosed as-needed after 12-week fixed dosing Ophthalmology 20111181098ndash11067 Campochiaro PA Wykoff CC Shapiro H Rubio RG Ehrlich JS Neutralizationof vascular endothelial growth factor slows progression of retinal nonperfusion inpatients with diabetic macular edema Ophthalmology 20141211783ndash17898 Elman MJ Aiello LP Beck RW et al Diabetic Retinopathy Clinical ResearchNetwork Randomized trial evaluating ranibizumab plus prompt or deferred laseror triamcinolone plus prompt laser for diabetic macular edema Ophthalmology20101171064ndash1077e35

9 Dornan T Mann JI Turner R Factors protective against retinopathy in in-sulin-dependent diabetics free of retinopathy for 30 years Br Med J (Clin Res Ed)19822851073ndash107710 Joussen AM Murata T Tsujikawa A Kirchhof B Bursell SE Adamis APLeukocyte-mediated endothelial cell injury and death in the diabetic retina Am JPathol 2001158147ndash15211 Joussen AM Poulaki V Le ML et al A central role for inflammation in thepathogenesis of diabetic retinopathy FASEB J 2004181450ndash145212 Mizutani M Kern TS Lorenzi M Accelerated death of retinal microvascularcells in human and experimental diabetic retinopathy J Clin Invest 199697

2883ndash289013 Hammes H-P Lin J Wagner P et al Angiopoietin-2 causes pericytedropout in the normal retina evidence for involvement in diabetic retinopathy

Diabetes 2004531104ndash111014 van Dijk HW Kok PHB Garvin M et al Selective loss of inner retinal layerthickness in type 1 diabetic patients with minimal diabetic retinopathy Invest

Ophthalmol Vis Sci 2009503404ndash340915 Patel JI Hykin PG Gregor ZJ Boulton M Cree IA Angiopoietin concen-trations in diabetic retinopathy Br J Ophthalmol 200589480ndash48316 Gavard J Patel V Gutkind JS Angiopoietin-1 prevents VEGF-induced en-dothelial permeability by sequestering Src through mDia Dev Cell 20081425ndash3617 Thurston G Rudge JS Ioffe E et al Angiopoietin-1 protects the adultvasculature against plasma leakage Nat Med 20006460ndash46318 Augustin HG Koh GY Thurston G Alitalo K Control of vascular morpho-genesis and homeostasis through the angiopoietin-Tie system Nat Rev Mol Cell

Biol 200910165ndash17719 Joussen AM Poulaki V Tsujikawa A et al Suppression of diabetic reti-nopathy with angiopoietin-1 Am J Pathol 20021601683ndash169320 Cho C-H Kammerer RA Lee HJ et al COMP-Ang1 a designed angio-poietin-1 variant with nonleaky angiogenic activity Proc Natl Acad Sci U S A20041015547ndash555221 Syrjaumllauml SO Nykaumlnen AI Tuuminen R et al Donor heart treatment withCOMP-Ang1 limits ischemia-reperfusion injury and rejection of cardiac allograftsAm J Transplant 2015152075ndash208422 Moon HE Byun K Park HW et al COMP-Ang1 Potentiates EPC Treatment ofIschemic Brain Injury by Enhancing Angiogenesis Through Activating AKT-mTORPathway and Promoting Vascular Migration Through Activating Tie2-FAK Path-

way Exp Neurobiol 20152455ndash7023 Kim DH Jung YJ Lee AS et al COMP-angiopoietin-1 decreases lipo-polysaccharide-induced acute kidney injury Kidney Int 2009761180ndash119124 Barber AJ Antonetti DA Kern TS et al The Ins2Akita mouse as a model ofearly retinal complications in diabetes Invest Ophthalmol Vis Sci 2005462210ndash221825 Han Z Guo J Conley SM Naash MI Retinal angiogenesis in the Ins2(Akita)

mouse model of diabetic retinopathy Invest Ophthalmol Vis Sci 201354574ndash58426 Zhu W London NR Gibson CC et al Interleukin receptor activates

a MYD88-ARNO-ARF6 cascade to disrupt vascular stability Nature 2012492252ndash25527 Medina RJ OrsquoNeill CL Devine AB Gardiner TA Stitt AW The pleiotropic

effects of simvastatin on retinal microvascular endothelium has important im-plications for ischaemic retinopathies PLoS One 20083e258428 Medina RJ OrsquoNeill CL OrsquoDoherty TM et al Myeloid angiogenic cells act as

alternative M2 macrophages and modulate angiogenesis through interleukin-8Mol Med 2011171045ndash105529 Simonelli F Maguire AM Testa F et al Gene therapy for Leberrsquos congenital

amaurosis is safe and effective through 15 years after vector administration MolTher 201018643ndash65030 Zhang X Das SK Passi SF et al AAV2 delivery of Flt23k intraceptors in-hibits murine choroidal neovascularization Mol Ther 201523226ndash234


31 Yin L Greenberg K Hunter JJ et al Intravitreal injection of AAV2 transducesmacaque inner retina Invest Ophthalmol Vis Sci 2011522775ndash278332 Cogan DG Toussaint D Kuwabara T Retinal vascular patterns IV Diabeticretinopathy Arch Ophthalmol 196166366ndash37833 Davis S Aldrich TH Jones PF et al Isolation of angiopoietin-1 a ligandfor the TIE2 receptor by secretion-trap expression cloning Cell 1996871161ndash116934 Uemura A Ogawa M Hirashima M et al Recombinant angiopoietin-1 re-stores higher-order architecture of growing blood vessels in mice in the absenceof mural cells J Clin Invest 20021101619ndash162835 Thurston G Complementary actions of VEGF and angiopoietin-1 on bloodvessel growth and leakage J Anat 2002200575ndash58036 Huang H Gandhi JK Zhong X et al TNFalpha is required for late BRBbreakdown in diabetic retinopathy and its inhibition prevents leukostasis andprotects vessels and neurons from apoptosis Invest Ophthalmol Vis Sci 2011521336ndash134437 Chibber R Ben-Mahmud BM Chibber S Kohner EM Leukocytes in diabeticretinopathy Curr Diabetes Rev 200733ndash1438 Vinores SA Xiao W-H Shen J Campochiaro PA TNF-alpha is critical forischemia-induced leukostasis but not retinal neovascularization nor VEGF-induced leakage J Neuroimmunol 200718273ndash7939 Cahoon JM Olson PR Nielson S et al Acridine orange leukocyte fluo-rography in mice Exp Eye Res 201412015ndash1940 Miyamoto K Khosrof S Bursell SE et al Vascular endothelial growth factor(VEGF)-induced retinal vascular permeability is mediated by intercellular adhesionmolecule-1 (ICAM-1) Am J Pathol 20001561733ndash173941 Gastinger MJ Kunselman AR Conboy EE Bronson SK Barber AJDendrite remodeling and other abnormalities in the retinal ganglion

cells of Ins2 Akita diabetic mice Invest Ophthalmol Vis Sci 2008492635ndash264242 Aung MH Kim MK Olson DE Thule PM Pardue MT Early visual deficits instreptozotocin-induced diabetic long evans rats Invest Ophthalmol Vis Sci 2013541370ndash137743 Hombrebueno JR Chen M Penalva RG Xu H Loss of synaptic connectivityparticularly in second order neurons is a key feature of diabetic retinal neuropathyin the Ins2Akita mouse PLoS One 20149e9797044 Caballero S Hazra S Bhatwadekar A et al Circulating mononuclear pro-genitor cells differential roles for subpopulations in repair of retinal vascularinjury 2013543000ndash300945 Medina RJ OrsquoNeill CL Humphreys MW Gardiner TA Stitt AW Outgrowthendothelial cells characterization and their potential for reversing ischemic ret-inopathy Invest Ophthalmol Vis Sci 2010515906ndash591346 Medina RJ OrsquoNeill CL Sweeney M et al Molecular analysis of endothelialprogenitor cell (EPC) subtypes reveals two distinct cell populations with differentidentities BMC Med Genomics 201031847 Shen F Walker EJ Jiang L et al Coexpression of angiopoietin-1 with VEGFincreases the structural integrity of the blood-brain barrier and reduces atrophyvolume J Cereb Blood Flow Metab 2011312343ndash235148 Lee J Kim KE Choi D-K et al Angiopoietin-1 guides directional angio-genesis through integrin avb5 signaling for recovery of ischemic retinopathy SciTransl Med 20135203ra127ndash749 Shin HY Lee YJ Kim HJ et al Protective role of COMP-Ang1 in ischemic ratbrain J Neurosci Res 2010881052ndash106350 Gastinger MJ Singh RSJ Barber AJ Loss of cholinergic and dopaminergicamacrine cells in streptozotocin-diabetic rat and Ins2Akita-diabetic mouse reti-nas Invest Ophthalmol Vis Sci 2006473143ndash3150


inhibitors in recovering visual acuity and are associatedwith side effects like cataract and intraocular hypertension(8) Given the suboptimal outcomes we developed a differ-ent approach to DME focusing on the reversal of retinalvascular damage and restoration of normal perfusion

The underlying pathogenesis of DR is largely due tohyperglycemia (9) Hyperglycemia triggers an inflammatoryresponse leading to leukocyte adhesion microvascular occlu-sion and consequent hypoxia (1011) Further hyperglyce-mia instigates pericyte loss compromising endothelialstability and BRB integrity Eventual capillary degenerationleads to retinal nonperfusion exacerbating retinal hypoxia(12) Consequent pathological VEGF-induced angiogenesis isuncoordinated and results in immature leaky vessels withinadequate perfusion creating a vicious cycle of hypoxia-driven VEGF secretion and DME (13) Retinal ganglion cell(RGC) loss neuronal dysfunction and changes in vision arealso seen in patients with DR concurrently with vascularpathology (14) Hence as a therapeutic goal vascular stabi-lization could promote normal perfusion of metabolicallydemanding retinal neurons and thereby avert the sight-threatening sequelae of ischemia and hyperpermeability

One therapeutic target is angiopoietin 1 (Ang1) a vasculargrowth factor that has an abnormally low concentration inthe vitreous of patients with DR (15) Ang1 via binding tothe Tie2 endothelial receptor fosters vessel quiescence andmaturation and suppresses vascular leakage by preventingVEGF-induced degradation of vascular endothelial (VE)-cadherin a transmembrane protein in the adherens junctionbetween endothelial cells that promotes vascular integrityand decreases vascular permeability (1617) Ang1 also pro-motes the survival of damaged vascular endothelial cellsthrough the phosphatidylinositol 3-kinaseAkt cascade (18)Thus restoration of Ang1 signaling could serve as apossible solution for preventing endothelial loss retinalischemia and abnormal VEGF expression in DR (19)

Pharmaceutical development of Ang1 as a viable ther-apy has been hindered by its insolubility and aggregationTen years ago a novel stable soluble and more potentversion of Ang1 cartilage oligomeric matrix protein Ang1(COMP-Ang1) was bioengineered to overcome the limi-tations of native Ang1 (20) Although the benefits ofCOMP-Ang1 have been demonstrated for vasculopathicdisorders in animal models of the cardiac (21) nervous(22) and renal (23) systems this study is the first totest its effectiveness for the prevention and treatmentof DR

We hypothesized that if introduced early in diabetesconstitutive expression of COMP-Ang1 via adeno-associatedvirus serotype 2ndashmediated gene therapy (AAV2COMP-Ang1) either as a monotherapy or in combination withhuman-derived endothelial colony-forming cells (ECFCs)could protect retinal neurovascular structure and functionin a type 1 diabetic Ins2Akita mouse model of DR Wefound that COMP-Ang1 prevents the endothelial losscapillary dropout BRB instability leukocyte dysfunctionneuroretinal attenuation and visual decline associated

with DR In addition COMP-Ang1 enhances the thera-peutic efficacy of ECFCs in mitigating DR through vascu-lar and visual rehabilitation

RESEARCH DESIGN AND METHODS

MiceThis research protocol was approved by the InstitutionalAnimal Care and Use Committee of the University of Utahand conforms to the standards in the Associaton forResearch in Vision and Ophthalmology Statement for theUse of Animals in Ophthalmic and Vision Research Thediabetic C57BL6-Ins2AkitaJ (Ins2Akita) mouse and itsbackground wild-type (WT) strain C57BL6J were used(24) Mice heterozygous for the Ins2 mutation experienceprogressive retinal abnormalities 12 weeks after the onsetof hyperglycemia including apoptosis (ie endothelialand RGC loss) and functional deficits (increased vascularpermeability and decreased neuronal function) (25)

Mice were randomly assigned to one of three experi-mental groups AAV2COMP-Ang1 AAV2AcGFP (Aequoreacoerulescens green fluorescent protein) or PBS At 2 monthsof age (Supplementary Fig 3) each mouse was anesthetizedby injection of 125 tribromoethanol (Sigma-AldrichSt Louis MO) at a dose of 0025 mLg body wt ip Eachmouse was treated with either 2 mL AAV2 solution (203 109

particles) or PBS injected into the vitreous cavity of botheyes with a 33-gauge microsyringe (Hamilton CompanyReno NV) An additional cohort of mice was treated asdescribed above at 6 months of age Two weeks later thiscohort received a second intravitreal injection with 1 mLof 1 3 105 ECFCs (Supplementary Fig 3)

Virus Vector ConstructionThe plasmids pAAVCOMP-Ang1 and pAAVAcGFP werecreated by incorporating the COMP-Ang1 cDNA fromthe pCMV-dhfr2-COMP-Ang1 (donated generously by theKoh Laboratory KAIST Daejeon) into pAAV-MCS (Agi-lent Technologies Santa Clara CA) while pAAVAcGFPwas created by the same technique with AcGFP cDNAfrom the pIRES2-AcGFP1 plasmid (Clontech LaboratoriesMountain View CA) (Supplementary Fig 3)

For in vivo assays requiring imaging with the SpectralisHRA+OCT (Heidelberg Engineering Heidelberg Germany)mice were anesthetized by an inhalation of 3 isofluraneO2

mixture in a closed canister at a flow rate of 10 Lpm Pupilswere dilated with a 1 tropicamide

Retinas were harvested and processed for in situhybridization (ISH) according to standard protocols usingprocedures to avoid RNAse contamination

RT-PCR for COMP-Ang1 mRNA ExpressionPrimer sequences for COMP-Ang1 (University of Utah GeneCore Salt Lake City UT) were as follows COMP-Ang1 F59-GCTCTGTTTTCCTGCTGTCC-39 and COMP-Ang1 R 59-GTGATGGAATGTGACGCTTG-39 Primer sequences for the internalcontrol GAPDH were as follows GAPDH F 59-AACTTTGGGATTGTGGAAGGG-3 and GAPDH R 59-ACCAGTGGATGCAGGGATGAT-39G




























RESULTS














































DISCUSSION























































RESULTS














































DISCUSSION




























































DISCUSSION





























Intravitreal AAV2.COMP-Ang1 Prevents … · 2015-11-17 · Christina L. O’Neil,2 Reinhold J....

Documents

Transcript of Intravitreal AAV2.COMP-Ang1 Prevents … · 2015-11-17 · Christina L. O’Neil,2 Reinhold J....