Gene Transfer of Prolyl Hydroxylase Domain 2 Inhibits Hypoxia-inducible Angiogenesis in a Model of Choroidal Neovascularization

Cellular responses to hypoxia are mediated by the hypoxia-inducible factors (HIF). In normoxia, HIF-α proteins are regulated by a family of dioxygenases, through prolyl and asparagyl hydroxylation, culminating in proteasomal degradation and transcriptional inactivation. In hypoxia, the dioxygenases become inactive and allow formation of HIF transcription factor, responsible for upregulation of hypoxia genes. In ocular neoangiogenic diseases, such as neovascular age-related macular degeneration (nAMD), hypoxia seems pivotal. Here, we investigate the effects of HIF regulatory proteins on the hypoxia pathway in retinal pigment epithelium (RPE) cells, critically involved in nAMD pathogenesis. Our data indicates that, in ARPE-19 cells, prolyl hydroxylase domain (PHD)2 is the most potent negative-regulator of the HIF pathway. The negative effects of PHD2 on the hypoxia pathway were associated with decreased HIF-1α protein levels, and concomitant decrease in angiogenic factors. ARPE-19 cells stably expressing PHD2 impaired angiogenesis in vitro by wound healing, tubulogenesis, and sprouting assays, as well as in vivo by iris-induced angiogenesis. Gene transfer of PHD2 in vivo resulted in mitigation of HIF-mediated angiogenesis in a mouse model of nAMD. These results may have implications for the clinical treatment of nAMD patients, particularly regarding the use of gene therapy to negatively regulate neoangiogenesis.


Protein expression analysis
ARPE-19 cells were aliquoted into 6-well plates and subsequently transfected with 500 ng of plasmid DNA (pDNA) encoding FLAG-tagged HIF-1a, HIF-2a, PHD1, PHD2, PHD3, FIH-1, VHL, or IPAS; plasmid concentrations were kept normalized to 1000 ng with empty pCMX vector. Cells were allowed to recover for 24 h and, as depicted in figure legends, kept at normoxia or exposed to hypoxia or treated with CoCl 2 for 16 h. Whole-cell extracts and immunoblots are described below.

Reporter gene assay
ARPE-19 cells were pre-aliquoted into 24-well plates and transfected with a pDNA mix containing 280 ng HRE-Luc and 20 ng rLuc, as a reporter system. FLAG-tagged PHD1, PHD2, PHD3, VHL, FIH-1, IPAS, or empty CMX (as a negative control) were added independently at 25 ng of pDNA to the reporter system mix. DNA amounts were equilibrated to a total of 500 ng using pCMX. The cells were allowed to recover for 16 h post-transfection, and exposed to normoxia, hypoxia, or CoCl 2 for 24 h. Cells were lysed with 200 µL passive lysis buffer (Promega, Madison, WI, USA), and 40 µL of extract was used for dual-luciferase reporter (DLR) assay (Promega), according to the manufacturer's instructions. Luminometry was carried out using a Tecan Infinity F200 plate reader with dual injectors (Grödig, Austria). Data was normalized to normoxic pCMX DLR ratio.

Reoxygenation assay
ARPE-19 cells were seeded as for protein expression analysis and transfected with 1 μg of PHD1, PHD2, PHD3, VHL, FIH-1, IPAS expression vectors, or empty CMX. All cells recovered for 36 h before exposure to 8 h of hypoxia. After incubation in hypoxia, cells were transferred to normoxia and whole-cell extracts were prepared after 0, 5, 10, 15, 30, and 60 min of reoxygenation. Experiments were analyzed by immunoblotting.

RPE medium conditioning
RPE-puro and RPE-PHD2 cells were grown to 80 % confluence in 10-cm diameter dishes, changed to reduced medium, and exposed to 24, 48, or 72 h of hypoxia, together with normoxia controls. Whole-cell extracts were prepared and the conditioned medium was clarified by centrifugation at 4000 rpm for 4 min at room temperature (RT), aliquoted, zap-frozen in liquid nitrogen, and stored at -80° C until analysis.

VEGF-capturing assay
Twenty µL slurry Dynabeads protein G (ThermoFisher Scientific Corp.) were blocked with 1 % BSA (Sigma-Aldrich Corp.) in Tris-buffered saline (TBS; BioRad Laboratories, Hercules, CA, USA) for 30 min at RT under rotation, followed by immunization with 125 µg Bevacizumab, a clinically used VEGF-neutralizing monoclonal antibody (Avastin; Roche, Welwyn Garden City, UK), for an additional 30 min. Immunization solution was removed and replaced with 1.5 mL RPE-conditioned medium for 1 h under rotation, followed by extensive washes with TBS. Immunocomplexes were eluted from Dynabeads using 20 µL denaturing Laemmli buffer (dLB; BioRad Laboratories) at 95° C for 5 min and analyzed by immunoblotting.

ELISA assay
Fifty µL RPE-conditioned media were incubated overnight (ON) on a VEGF human ELISA kit (Abcam, Cambridge, UK) and analyzed according to the manufacturer's instructions on a Tecan Infinity F200 plate reader equipped with a 450 nm filter. Sample quantification was performed using a regression standard curve, as described by the manufacturer.

Wound healing assay
HUVE cells were grown to confluence on 24-well plates and incubated for 1 h in reduced medium, prior to scratching with a tip to induce wounding. After two phosphate-buffered saline (PBS; ThermoFisher Scientific Corp.) washes, wounded cell cultures were changed into RPE-conditioned media, collected from the 24 h hypoxia-exposed RPE-puro or RPE-PHD2 cells. Images from 1, 3, 6, and 12 h after wounded HUVE cell cultures were exposed to RPE-conditioned media were acquired using a PrimoVert contrast phase microscope (Zeiss, Gottingen, Germany) coupled to a Visicam TC 10 (VWR, Lutterworth, UK). To allow reproducibility, only central field images were acquired and an average of three measurements (top, center, and bottom) of the scratched area using ImageJ freeware were used to assess wound healing.

Tubulogenesis assay
HUVE cells were seeded (4x10 4 cells/well) onto 24-well plates pre-coated with growth factor reduced matrigel (BD Biosciences, San Jose, CA, USA), as described by the manufacturer. Cells were allowed to attach for 1 h in reduced medium before exposure to RPE-conditioned media from RPE-puro or RPE-PHD2 cells exposed to 24 h of hypoxia. To grant reproducibility, central field images were acquired as described for wound healing assay. Tube-like cells and nuclei were counted using ImageJ, and data presented as a tubes/nuclei ratio.

Sprouting assay
RE or CE cells were aliquoted into non-adherent round-bottom 96-well plates (1x10 3 cells/well) in medium supplemented with 0.4% methylcellulose (Sigma-Aldrich Corp.) and allowed to form spheroids. Cultures containing spheroids (100 µL) were transferred onto flat-bottom adherent 96-well plates containing 25 µL matrigel, and incubated at 37° C for 1 h. Excess media was carefully removed and replaced with RPE-conditioned media from RPE-puro or RPE-PHD2 cells exposed to 24 h of hypoxia. Alternatively, 3D cultures were obtained by mixing cell suspensions of RE or CE cells with either RPE-puro or RPE-PHD2, at a ratio of 2:1 (endothelium:epithelium) prior to formation of spheroids. After transferred onto matrigel, 3D cultures were incubated in reduced medium. As before, reproducibility was granted by acquiring central field images of 36 h sprouting spheroids and 3D cultures. Sprouts were counted using ImageJ, and data presented as number of sprouts per spheroid.

Animals
Nine 12.5-day-old BalbC and 36 8-week-old C57Bl6J mice (Charles River, Cologne, Germany) were used in accordance with the ARVO statement for the Use of Animals in Ophthalmologic and Vision Research, and the study protocols were approved by Stockholm's Committee for Ethical Animal Research. All mice were acclimatized for approximately 1 week on arrival and housed in social groups of 4-6 animals with litter, sizzle-nest, and housing enrichment. Mice were kept in an IVC system (Allentown Inc., Allentown, NJ, USA) with food and water ad libitum, a light/dark cycle of 12 h, and room conditions averaging 23 °C temperature and 55 % humidity. All animals were monitored daily. At the end of experimental procedures, mice were euthanized by cervical dislocation by trained personal, as approved by the ethical committee.

Iris angiogenesis
BalbC mice were anesthetized using 4 % isoflurane (Baxter, Kista, Sweden) and injected intravitreally with 1 µL reduced medium (vehicle) or 1 µL of conditioned media from RPE-puro or RPE-PHD2 exposed to 24 h of hypoxia. All media was sterilized through a 0.22 µm filter prior to injection. Intravitreal injections were repeated every fourth day. On day 15 after the first injection, mice were deeply anesthetized with ketamine (90 mg/Kg; Pfizer, Cambridge, UK) and xylazine (15 mg/Kg; Bayer, Leverkusen, Germany). One mL of Dextran-FITC (Sigma-Aldrich Corp.; 25 mg/mL in PBS) was injected into the left ventricle of each animal and allowed to circulate for 10 minutes. The animals were euthanized and the eyes were enucleated and cleared from extraneous tissues. After fixing for 6 h at RT in 4 % PBS-buffered formaldehyde (PFA; Solveco, Rosersberg, Sweden), irises were microdissected as whole-mounts for fluorescence microscopy (refer to Immunofluorescence for details). Comparative fluorescence intensity was maintained using fixed camera exposures for all samples. Data were analyzed as fluorescence densitometry (intensity of signal per area) and compared to vehicle.

CNV induction and gene transfer
Two CNV lesions were induced nasally and temporally in C57Bl6J mice as previously described 3 . Four days post-laser induction, mice were anesthetized and subretinally injected with 1 µg of plasmid DNA encoding FLAG-tagged PHD2 or empty CMX in 1 µL endotoxin-free Tris-EDTA buffer (Qiagen, Hilden, Germany). Gene transfer to the RPE layer of the retina was achieved by electroporation using a NEPA21 system with a CUY650P5 electrode (NepaGene, Chiba, Japan). Electroporation was conducted by 3 repeats of a poring pulse (3 pulses of 60 V for 30 ms, with 50 ms interval over 10 % decay rate and dual polarity) followed by a transfer pulse (5 pulses of 20 V for 50 ms, with 50 ms interval over 40 % decay rate and dual polarity). On days 7 and 14 after laser (equivalent of 3 and 10 days of DNA expression), mice were euthanized and eyes microdissected into posterior eye segments, comprehending RPE-choroid-sclera complexes.

Immunoblotting assays
As previously described, whole-cell extracts were prepared by lysis with RIPA buffer (Sigma-Aldrich Corp.) 1 while whole-tissue extracts were obtained using CelLytic-MT (Sigma-Aldrich Corp.) 3 . Total protein extracts were prepared with dLB, separated by SDS-PAGE, and transferred onto nitrocellulose membranes (all reagents and equipment from BioRad Laboratories). Blots were blocked using 5 % non-fat milk (nfm) in TBS, and incubated with antibodies using 1 % nfm/TBS containing 0.05 % Tween-20 (Sigma-Aldrich Corp.) (TBS-T). For primary antibodies, membranes were incubated with dilutions at 1:500 of: anti-HIF-1a (Bio-Techne Corp.), anti-HIF-2a (Bio-Techne Corp.), anti-FLAG (Sigma-Aldrich Corp.), anti-VEGF (Abcam), anti-Actin (Sigma-Aldrich Corp.). Secondary antibodies were used at a 1:2000 dilution of anti-rabbit IgGhorseradish peroxidase conjugate (Dako, Carpinteria, CA, USA). After both incubations with antibodies, extensive washes with TBS-T were performed. Visualization of the proteins of interest was achieved using enhanced chemiluminescence (Thermo Scientific) and exposed to hyperfilm ECL autoradiography films (VWR). Due to sample limitations, membranes were cropped into regions of interest, prior to western blot analyses, and data is presented as a full representation of the cropped and blotted area. Paralleled exposures across each figure panel are presented for best comparison. When described, ImageJ (NIH freeware) was used to analyze densitometry.

Protein arrays
Protein profiler arrays (Bio-Techne Corp.) for angiogenic factors and cytokines were preformed according to the manufacturer's instructions. For human soluble factors, 1 mL of 24 h hypoxia-exposed RPE-conditioned media was used. One-hundred µL of whole-tissue extract was used for mouse experiments. Densitometric analysis was determined using ImageJ against two internal positive controls, of the same exposure, across three independent experiments or biological replicates, respectively.

Statistical analysis
Immunoblots display representative results of at least three independent experiments from the same exposure. Quantitative in vitro experiments were performed three times with technical duplicates (n=6). Iris angiogenesis was determined in 3 mice per group, and data was quantified from all eyes (n=6). CNV area analysis was quantified from 3 mice as for iris angiogenesis (n=6). Total RNA (and total protein) from CNV mice was collected from 3 mice as a pool of 2 posterior eye segments from 2 independent animals (n=3). Results are presented as mean ± standard error. Student's t-test was used for paired statistical evaluation while one-way ANOVA with Bonferroni corrected post-hoc tests was used for multiple comparisons, and P<0.05 was considered statistically significant.