Anti-Syndecan 2 Antibody Treatment Reduces Edema Formation and Inflammation of Murine Laser-Induced CNV

Abstract

Purpose: Alteration of visual acuity in wet age-related macular degeneration (AMD) is mostly driven by vascular endothelial growth factor A (VEGF-A)–induced edema from leaky newly forming blood vessels below the retina layers. To date, all therapies aimed at alleviation of this process have relied on inhibition of VEGF-A activity. Although effective in preventing vascular leak and edema, this approach also leads to the loss of normal vasculature and multiple related side effects.

Methods: We have developed an alternative strategy that uses anti–syndecan-2 polyclonal antibody (anti-Sdc2 pAb) to block VEGF-A–induced permeability without interfering with other VEGF-A activities. The effect of anti-Sdc2 pAb therapy was assessed in vitro using a transendothelial electrical resistance (TEER) assay, as well as staining of the endothelial cell junction, and in vivo in the laser-induced choroidal neovascularization (CNV) model.

Results: Anti-Sdc2 pAb blocked VEGF-A–induced permeability in vitro, and both local intravitreal injections and systemic intravenous treatments with anti-Sdc2 pAb were as effective as intravitreal anti-VEGF therapy in reducing edema, size of retinal lesions, and local inflammation in this model. Post-injury neovascularization was not affected by treatment with anti-Sdc2 pAb.

Conclusions: These findings indicate that anti-Sdc2 pAb therapy can be an effective alternative to anti–VEGF-A approaches for suppression of edema and to prevent retinal lesions in wet neovascular AMD (nAMD).

Translational Relevance: Intravitreal anti-Sdc2 treatment may avoid side effects observed with the long-term anti-VEGF therapy, and systemic treatment with an anti-Sdc2 pAb antibody can address the issues associated with repeated intravitreal injections.

Introduction

Age-related macular degeneration (AMD) is a major cause of vision loss in patients over the age of 50 years. It is expected that 288 million individuals worldwide will be affected by AMD in 2040.^1,2 Neovascular AMD (nAMD) is the advanced stage of the disease characterized by fluid leakage from neo-vessels forming from choroidal vasculature. The result of this process is subretinal edema, inflammation and tissue disruption with devastating consequences for visual acuity. The standard approved therapy for nAMD consists of repeated intravitreal injections of anti–vascular endothelial growth factor A (VEGF-A) monoclonal antibodies (bevacizumab and ranibizumab) or a soluble VEGF trap chimera of VEGF receptors 1 and 2 (aflibercept) which leads to regression of edema and satisfactory restoration of visual acuity in most patients.³ Nevertheless, treatment of nAMD still presents a number of challenges, including the required frequency of injections, the need for intravitreal administration, and the side effects of this therapy in the long term.⁴ In particular, complications from repeated intravitreal injections include subconjunctival hemorrhage, bacterial infections within the eye, intraocular inflammation, and increases in intraocular pressure, among others.^5,6 Consequently, a number of clinical trials are in progress aiming to evaluate better therapeutic options.⁷ 

Syndecan-2 (Sdc2), a heparan sulfate proteoglycan transmembrane protein that carries heparan sulfate chains,^8,9 is highly expressed in the vascular endothelium. Compared to other proteoglycans, its heparan sulfate chains have enhanced VEGF-A binding capacity. As a result, Sdc2 functions as a preferential VEGF-A binding proteoglycan.¹⁰ In the presence of VEGF-A, Sdc2 forms a trimolecular complex with the growth factor and its signaling receptor, vascular endothelial growth factor receptor-2 (VEGFR2). VEGF-A binding to VEGFR2 initiates a cascade of events that result in VEGFR2 dimerization and a conformational change that activates its kinase domain, leading to cross-phosphorylation of key tyrosines (Y951, Y1175, and Y1241, human nomenclature). When phosphorylated, each of the phosphotyrosines initiates a specific signaling pathway that result in activation of endothelial permeability, proliferation, and migration.¹¹ Critically, both Y1175 and Y1214 (but not Y951) signaling cascades are required for endothelial viability as long-term withdrawal of VEGF-A signaling input leads to the loss of capillary vasculature.¹² At the same time, disruption of the phospho-Y951–activated signaling specifically disrupts initiation of vascular permeability while not affecting endothelial viability. Currently, available anti–VEGF-A agents act by either neutralizing VEGF-A itself or blocking its binding to VEGFRs that leads to the elimination of all VEGF-A signaling, affecting both its permeability and endothelial viability effector pathways. 

An unusual aspect of Sdc2 biology is its direct association with the density-enhanced protein (DEP) tyrosine phosphatase 1 (DEP1). A previous study from our laboratory demonstrated that, in the absence of Sdc2 or when its binding to DEP1 is prevented, the phosphatase binds to VEGFR2 specifically dephosphorylating Y951, thereby inhibiting VEGF-A–induced permeability while maintaining the remainder of VEGF-A signaling cascades intact.¹³ This mechanism of action opens a possibility of selective regulation of VEGF-A signaling. Here, we investigated the ability of an antibody that blocks DEP1 binding to Sdc2 to inhibit vasogenic edema and prevent retinal damage in a mouse model of laser-induced choroidal neovascularization (CNV). 

Materials and Methods

Animals

All animals were used in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The study protocols were approved by Stockholm's committee for ethical animal research (Dnr 7053-2020). A total of 28 C56BL/6J mice (Charles River Laboratories, Wilmington, MA) were kept with a 12-hour light/dark cycle and ad libitum food and water access. Mice were anesthetized by inhaled 4% isoflurane/room air (VDG9623C; Baxter, Deerfield, IL) in room air. Both pupils were dilated by topical administration of 0.5% tropicamide (Alcon, Geneva, Switzerland). CNV lesions were induced with the laser-coupled MICRON IV (Phoenix-Micron, Bend, OR) set for 50-µm spot, 180-mW intensity, 100-ms duration, and four shots per eye in both eyes as approved in protocols. Successful induction of Bruch’s membrane rupture was estimated by formation of a characteristic vapor bubble. All visual hemorrhagic lesions were excluded from the study. On day 7 post-laser, mice were anesthetized for in vivo imaging. All animals were euthanized on day 8, and the eyes were enucleated and fixed in 4% buffered formaldehyde overnight at 4°C. 

Development of Anti-Mouse Sdc2 Polyclonal Antibody

The anti-Sdc2 polyclonal antibody (anti-Sdc2 pAb, referred to as pAb3) was generated by immunization of rabbits with a peptide corresponding to the DEP1-binding motif of mouse Sdc2 (spanning amino acids 124–141) and purified via affinity chromatography (GenScript, Piscataway NJ). We have previously characterized the specificity and biological activity of this antibody in vitro and in vivo.¹³ 

Pharmacological Treatments

Animals were randomly distributed into three groups, and both eyes received intravitreal injections at days 3 and 6 after laser photocoagulation of 1 µL of vehicle (phosphate-buffered saline [PBS], control) or 1 µg (1µL) of recombinant mouse VEGFR1-Fc chimera protein (anti-VEGF, 471-F1-100; Biotechne, Minneapolis, MN), or 1 µg (1 µL) of anti-mouse Sdc2 polyclonal antibody (anti-Sdc2 pAb). Two additional groups received systemic administration via tail vein injection of 100 µL PBS (control) or 100 µg (100 µL) of anti-Sdc2 pAb on days 3 and 6 after laser induction. 

In Vivo Imaging

Seven days after laser induction, mice were anesthetized with isoflurane as described, and their pupils were dilated by tropicamide eyedrops. Spectral domain optical coherence tomography (SD-OCT) was performed with the assistance of a live fundus image (Phoenix-Micron). B-scans were acquired categorically midline to the lesion center. Lesion volume was calculated as previously described¹⁴ using InSight software (Phoenix-Micron). Fluorescence fundus angiography was performed after the acquisition of OCT B-scans by subcutaneous injection of 30-mg/kg fluorescein sterile solution (Alcon Nordic, Copenhagen, Denmark) and imaged with fixed camera exposure settings 5 minutes after injection. Area of leakage was measured using ImageJ (National Institutes of Health, Bethesda, MD). Exclusion criteria for CNV lesions followed the recommended outliers.¹⁵ 

Electroretinography

Assessment of retinal function by electroretinography was performed in healthy mice that received anti-Sdc2 intravitreal injections but not laser treatment. Mice were dark adapted overnight prior to recordings, and their manipulation was performed under dim red light. The area studied was acquired by positioning the optic nerve at the center in the field of view under red light (to maintain dark adaptation), representing a total retinal spot of 1.5-mm diameter with the Phoenix-Micron–coupled electroretinography module. Mixed cone and rod responses were recorded averaging 10 sweeps at flash strengths of log –0.5, 0.0, and 1 cd·s/m² of intensity. The amplitude of the a-wave was measured from the baseline to the first negative trough, and the b‐wave amplitude was measured from the trough of the a‐wave to the peak of the b‐wave, in compliance with the International Society for Clinical Electrophysiology guidelines. 

Clearance of Anti-Sdc2 pAb In Vivo

For measurements of half-life in the vitreous humor and whole retina, 1 µg of anti-Sdc2 was injected intravitreally, and then mice were euthanized at the indicated time points. Vitreous was collected, transferred into 100 µL of saline solution, mechanically homogenized, and kept at 4°C until measurement. 

The whole retina/choroid structure was isolated, placed in 100 µL radioimmunoprecipitation assay (RIPA) buffer, and homogenized with a QIAGEN TissueLyser (QIAGEN, Hilden, Germany). Laemmli buffer (2× final concentration) was added to this suspension, and the sample was boiled for 10 minutes. For measurement of plasma half-life, 100 µg of anti-Sdc2 was injected systemically, blood was collected in ethylenediaminetetraacetic acid (EDTA) tubes at indicated time points, and spun at 1000g for 10 minutes to obtain plasma. The levels of anti-sdc2 in vitreous and plasma were assessed using the Easy-Titer Rabbit IgG Kit (Thermo Fisher Scientific, Waltham, MA). Anti-Sdc2 in whole-retina was assessed by western blot analysis and using an anti-rabbit IgG antibody conjugated to HRP (PI-1000-1; Vector Laboratories, Newark, CA). 

Evaluation of Complete Blood Count

Two doses of anti-Sdc2 pAb were administered systemically as described earlier in the Pharmacological Treatments section. Venous blood was collected directly in EDTA-coated tubes 2 days after the second anti-Sdc2 injection and immediately analyzed with an automated hematology analyzer (HEMAVET; Drew Scientific, Plantation, FL). 

Retinal Pigment Epithelium/Choroid Complex Immunostaining

The choroid tissue was isolated as previously described.¹⁶ Briefly, eyes were fixed in 1 mL of 1% paraformaldehyde at room temperature overnight. The cornea, lens, and the neuroretina were carefully removed from the eyecup, and the retinal pigment epithelium (RPE)/choroid layers were isolated, supported by the sclera. Four unbuckling cuts from the edge to the center between the laser impacts were performed. The choroid tissue was then washed in PBS for 30 minutes and afterward permeabilized in 0.5% Triton X-100 in PBS for 30 minutes under agitation at room temperature. The blocking step was done in PBS containing 10% bovine serum albumin (BSA) for 1 hour under agitation at room temperature. The choroid tissue was then incubated overnight at 4°C with the following primary antibodies: CD31 (AF3628, dilution 1:200; R&D Systems, Minneapolis, MN); F4/80 (MF48000, dilution 1:100; Thermo Fisher Scientific); transcription factor ERG (ab92513, dilution 1:200; Abcam, Cambridge, UK). After three washes for 20 minutes in PBS, the choroid tissue was incubated for 1 hour with the appropriate secondary antibodies for 90 minutes. The choroid tissue was then washed three times for 20 minutes under agitation in PBS followed by mounting in medium for microscope analysis. Images of the four laser impacts were taken with confocal microscopy (TCS SP8 confocal laser scanning microscope; Leica, Wetzlar, Germany) using a 10× objective lens (zoom factor, 3.00; Z-stack step size, 0.30 µm). Secondary antibodies were Donkey anti-Goat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (A-11055, used in CD31 staining; Thermo Fisher Scientific); Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 568 (A-11011, used in staining of transcription factor ERG; Thermo Fisher Scientific); and Goat anti-Rat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 647 (A-21247, used in F4/80 staining; Thermo Fisher Scientific). All secondary antibodies were used at a 1:250 dilution. 

Analysis of Neovessel Morphology

To characterize the morphology of neovessels in laser-injured areas we first performed bleaching of the retina/choroid tissue to remove pigmentation of the RPE layer as previously reported.¹⁷ Briefly, fixed tissue was incubated with a solution of 1% H₂O₂ in PBS at 55°C for 6 to 8 hours until discoloring occurred, then the tissue was washed four times with PBS, permeabilized with 0.5% Triton X-100 for 30 minutes, and blocked with 10% BSA for 1 hour. Immunostaining was then performed as already described above using primary antibodies against transcription factor ERG (ab92513, dilution 1:200; Abcam) and VE-cadherin (AF1002, 2.5 µg/mL; R&D Systems) 

In Vitro Permeability and Proliferation of Mouse Retinal Endothelial Cells

Permeability and proliferation assays were performed using the xCELLigence RTCA eSight System (Agilent Technologies, Santa Clara, CA). For permeability assays, mouse primary retinal microvascular endothelial cells (mRECs; C57-6065; Cell Biologics, Chicago, IL) were seeded onto an E-Plate 16 (5469830001; Agilent Technologies) at a density of 30,000 mRECs per well. Cells were kept in 200 µL/well of full media (M1168, with growth factors; Cell Biologics) for 48 hours to allow formation of a mature endothelial monolayer. At that point, 100 µL of media were replaced with 100 µL of fresh media containing 5% fetal bovine serum (FBS) without growth factors added. After 16 hours, either 1 µL of PBS (vehicle) or VEGF-A (293-VE-010; R&D Systems) at a final concentration of 100 ng/mL was added to each well. VEGF-A was allowed to induce disruption of the endothelial monolayer for 12 hours, then 3 µL/well of Sdc2 pAb was added at the indicated final concentration. Permeability curves were recorded for at least 24 hours, with cell index measurements taken every 15 minutes. For proliferation assays, 5000 cells/well were seeded in E16 plates in media (M1168; Cell Biologics) containing 5% FBS, VEGF-A (100 ng/mL), and either vehicle or Sdc2 pAb (40 µg/mL). Proliferation curves showed ∼48 hours of cell index monitoring (measured at 15-minute intervals). 

Staining of mRECs

mRECs were seeded onto glass slides and kept at confluence in full media for 48 hours. Confluent cells were then starved overnight in EBM (5% FBS) and stimulated with VEGF-A (100 ng/mL). After 12 hours of VEGF-A stimulation, either vehicle or anti-Sdc2 pAb (40 µg/mL) was added. After additional 24 hours of incubation, mRECs were fixed and stained for markers of endothelial junctions: VE-cadherin (AF1002, dilution 1:200; R&D Systems) and zonula occludens-1 (ZO-1; 339100, dilution 1:200; Life Technologies, Carlsbad, CA). Donkey anti-Goat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (A-11055, 1:400; Thermo Fisher Scientific) and Donkey anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 647 (A-31571, 1:400; Thermo Fisher Scientific) were used as secondary antibodies. 

Statistical Analysis

Statistical significance was determined by one-way analysis of variance (ANOVA) using Prism 9 (GraphPad, Boston, MA), followed by multiple comparisons post-test with specific sample sizes as disclosed in figure legends. The results are expressed as mean ± standard error of the mean (SEM). Differences with P < 0.05 were considered significant. 

Results

Intravitreal Delivery of Anti-Sdc2 pAb Does not Alter Retinal Function

To gain insight into the feasibility of the anti-Sdc2 pAb treatment, we first assessed whether intravitreal administration of the antibody impairs the normal retinal function in mice that did not receive laser treatment. Electroretinography (ERG) was used as a reliable method to detect drug-induced retinal injury.^18,19 ERG was carried out 8 days after intravitreal injection of either 1 µL of vehicle (PBS) or 1 µL of 1-µg/µL anti-Sdc2 pAb. Neither a- nor b-wave recordings showed any differences between anti-Sdc2 pAb and vehicle-treated eyes (Figs. 1A–C). This was confirmed by quantitative analysis that showed no statistical differences for wave amplitudes between eyes injected with either vehicle or anti-Sdc2 pAb across all light-stimulus intensities (Figs. 1D–I). This is consistent with the lack of retinal toxic events and indicates safety of the intraocular administration of anti-Sdc2 pAb. 

Figure 1.

 

Electroretinographic recording shows no abnormalities after anti-Sdc2 pAb administration in healthy mice that were not treated with laser. Either vehicle (PBS) or anti-Sdc2 pAb was administered to healthy mice via intravitreal injection, and ERG recording was performed on day 8. (A–C) Mixed cone and rod responses recorded at light intensities of −0.5, 0, and 1 log cd·s/m² (a representative sweep for each group is shown). (D–I) Quantification of a- and b-wave amplitudes at different flash strengths indicated no differences in a- or b-waves between PBS- and Sdc2 pAb–treated eyes. Graphs are presented as scatterplots with mean ± SEM (n = 10 vehicle, n = 9 Sdc2 pAb–treated eyes). Data were analyzed by unpaired two-tailed Student's t-test.

Anti-Sdc2 pAb Blocks Blood Vessel Leakage and Reduces Retinal Lesion Size After Intravitreal or Systemic Administration

We next set out to compare the effectiveness of anti-Sdc2 pAb in preventing vascular leakage and lesion size when administered either locally (intravitreally) or systemically, and we compared it to the standard-of-care intravitreal injection of VEGFR1-Fc chimera protein (anti-VEGF), a mouse equivalent of the clinically used aflibercept. Intravitreal and intravenous injections of PBS were used as controls. The presence and extent of the vascular leak around areas of laser damage (7 days after initial injury) was evaluated by fluorescein angiography (FA), which readily detected hyperfluorescent areas indicative of abnormal blood vessel permeability (Figs. 2A–E). Quantitative analysis of the area of the lesions, a measure of the extent of vascular leak, showed no significant differences between animals receiving either systemic (intravenous) or intravitreal vehicle (Fig. 2F). In contrast, both systemic (P = 0.007) and intravitreal administration of anti-Sdc2 pAb (P = 0.001) significantly reduced the leakage area compared to vehicle controls, with no discernible differences between the two anti-Sdc2 Ab therapy routes (Fig. 2F). Importantly, the effectiveness of both intravitreal and intravenous Sdc2 pAb therapy was comparable to that of the intravitreal anti-VEGF (Fig. 2F). 

Figure 2.

 

Anti-Sdc2 pAb administration effectively reduces CNV leakage and lesion size. FA was performed on day 7 after laser photocoagulation. (A–E) Illustrative FA images of eyes intravitreally injected with vehicle, anti-Sdc2 pAb, or anti-VEGF or systemically treated (vehicle, anti-Sdc2 pAb). (F) Quantification of leakage area by FA of eyes intravitreally injected with vehicle (n = 34 lesions), anti-Sdc2 pAb (n = 30 lesions), or anti–VEGF-A (n = 18 lesions) or systemically treated with vehicle (n = 23 lesions) or anti-Sdc2 pAb (n = 20 lesions). (G–K) Illustrative en face OCT images of eyes intravitreally injected with vehicle, anti-Sdc2 pAb, or anti-VEGF or systemically treated with vehicle or Sdc2 pAb. (L) Quantification of lesion volume by SD-OCT of eyes intravitreally injected with vehicle (n = 33 lesions), anti-Sdc2 pAb (n = 29 lesions), or anti-VEGF (n = 41 lesions) or systemically treated by vehicle (n = 30 lesions) or anti-Sdc2 pAb (n = 31 lesions). Data are presented as mean ± SEM and analyzed by one-way ANOVA followed by Tukey multiple-comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001.

To further corroborate these results, we analyzed volumes of the vascular leak lesions using spectral-domain optical coherence tomography (SD-OCT) B-scans (Figs. 2G–K and Supplementary Fig. S1). Quantitative analysis indicated a significant 45% reduction in the CNV lesion volume following both systemic and intravitreal administration of the anti-Sdc2 pAb compared to vehicle controls (systemic, P < 0.001; intravitreal, P = 0.001), with no statistically significant difference between the two routes of administration (Fig. 2L). Furthermore, there were no significant differences in the extent of lesion volume reduction between either intravitreal and intravenous anti-Sdc2 pAb versus anti-VEGF-treated animals (Fig. 2L). 

To investigate the effect of anti-Sdc2 pAb therapy at the tissue level, we performed immunostaining of the RPE and choroidal layer with anti-CD31 (platelet endothelial cell adhesion molecule 1, PECAM-1) antibody to evaluate the extent of neovascularization of laser-induced lesions. Laser injury induced a significant angiogenic response that was readily evident in vehicle-treated mice tissues (Fig. 3A). As expected, peri-lesional angiogenesis was significantly reduced (69.5%) following intravitreal anti-VEGF administration (Figs. 3B, 3F), whereas intravitreal administration of the anti-Sdc2 pAb had no effect on the extent of angiogenic response (Figs. 3C, 3F). 

Figure 3.

 

Analysis of RPE complexes. (A–E) Immunocytochemistry analysis of mice retinas 8 days after indicated treatments. (F) CD31 staining and associated quantification of CD31 staining (n = 12 lesions). (G–L) ERG staining (endothelial marker) (G–K) and associated quantification (L) (n = 12–16 lesions). (M–R) F4/80 staining (macrophage marker) (M–Q) and associated quantification (R) (n = 12 lesions). Data are presented as column (mean) and mean ± SEM and analyzed by one-way ANOVA followed Šidák multiple-comparison test. *P < 0.05, **P < 0.01, ***P < 0.001.

Interestingly, unlike the intravitreal administration, systemic anti-Sdc2 pAb therapy resulted in a significant reduction (44.8%) of CD31-positive areas (Figs. 3D–F). Because CD31 antigen is expressed by both endothelial cells and macrophages,^20,21 the size of CD31-positive areas reflects not only the extent of neovascularization but also the presence of blood-derived macrophages. To distinguish these possibilities and characterize a potential effect of Sdc2 pAb therapy on macrophage recruitment to laser injury lesion sites, choroid tissue was also stained with an antibody against transcription factor ERG, a highly specific endothelial nuclear marker²² that accurately reflects the extent of angiogenesis, and an F4/80 antibody, a marker of blood-derived macrophages.²³ In agreement with CD31 staining results, anti-VEGF trap treatment resulted in a significant reduction in staining of transcription factor ERG (Figs. 3G, 3H, 3L), indicating reduced angiogenesis, whereas neither intravitreal (Figs. 3G, 3I, 3L) nor systemic (Figs. 3J–L) anti-Sdc2 pAb therapy had any anti-angiogenic effect. 

As expected, intravitreal anti-VEGF treatment led to a significant reduction in the number of F4/80-positive macrophages in the retina (Figs. 3M, 3N, 3R). At the same time, both intravitreal (Figs. 3M, 3O, 3R) and systemic (Figs. 3P–R) anti-Sdc2 pAb administration resulted in a significant reduction (33% and 59%, respectively) in the recruitment of F4/80-positive macrophages that was comparable to that of anti-VEGF therapy. 

To better characterize the effect of anti-Sdc2 pAb on neovessel morphology, we used bleached choroid (to remove dark interference of RPE layer) and then performed immunostaining for VE-cadherin (a specific endothelial marker of adherens junctions) and the transcription factor ERG. We observed that control animals displayed patchy membrane expression of VE-cadherin, indicating the presence of immature and leaky blood vessels (Supplementary Fig. S2A). Conversely, animals treated with anti-Sdc2 pAb showed the expected cell-surface staining of VE-cadherin with uniform demarcation of contact points between adjacent endothelial cells. As expected, staining of transcription factor ERG did not show differences in the number of endothelial cells in the lesions between the two animal groups, confirming the absence of anti-angiogenic effects of anti-Sdc2 pAb (Supplementary Fig. S2A). 

We then assessed tissue clearance of anti-Sdc2 antibody in vivo. First, we looked at vitreous humor and found that anti-Sdc2 pAb became undetectable by ELISA at 48 hours after intravitreous injection (Supplementary Fig. S2b). These results are in line with published literature estimating a vitreous half-life of 1.5 to 4 hours in mice.²⁴ To confirm the kinetics of the anti-Sdc2 pAb retinal clearance, we performed western blots of whole retina lysates following intravitreal injection. Here, we found that the anti-Sdc2 pAb clearance kinetics and half-lives in the retina followed a profile similar to that of the vitreous humor, although a low signal was still detectable 48 hours after injection (Supplementary Fig. S2C). Finally, quantification of anti-Sdc2 pAb in plasma over time showed a detectable signal at day 8 after injection (Supplementary Fig. S2D). We also checked whether systemic injections of anti-Sdc2 pAb could induce alterations of the complete blood count (CBC) as a sign of acute toxicity. We did not detect statistically significant differences in any component of white blood cells, red blood cells, or platelets (Supplementary Fig. S3). 

Treatment With Anti-Sdc2 pAb Restores Barrier Integrity in VEGF-Treated mRECs

The in vivo results described above indicate that anti-Sdc2 pAb therapy reduced the extent of vascular leakage following laser-induced injury without specifically affecting angiogenesis. To confirm these findings in vitro, we set out to characterize the effect of anti-Sdc2 pAb on the barrier function of mRECs using an in vitro transendothelial electrical resistance (TEER) permeability assay. To mimic the leaky endothelium observed in wet nAMD and after choroidal laser injury described above, mRECs were plated at confluence and were allowed to form a tight monolayer as measured by TEER (Fig. 4A). The mREC barrier disruption was induced by the addition of VEGF-A (100 ng/mL) for 12 hours and monitored by TEER. As expected, there was a significant decrease in TEER (Fig. 4A, green vs. magenta line). When the disruption was fully established, either anti-Sdc2 pAb (three different concentrations) or vehicle was added. The addition of Sdc2 pAb resulted in a dose-dependent restoration of the mREC barrier integrity as indicated by the recovery of normalized TEER over time. (Figs. 4A, 4B). The highest tested concentration (40 µg/mL) led to a complete restoration of the endothelial barrier integrity as early as 15 hours after antibody treatment (Figs. 4A, 4B). 

Figure 4.

 

Sdc2 pAb restores endothelial barrier function in mRECs treated with VEGF-A. (A) Confluent mRECs were stimulated for 12 hours with VEGF-A or vehicle, which led to disruption of the endothelial barrier, as measured by a drop in normalized cell index (green vs. magenta line). Administration of anti-Sdc2 pAb at 12 hours after VEGF-A led to concentration-dependent recovery of endothelial barrier integrity (red, cyan, and blue lines) compared to treatment with vehicle alone (green line). (B) Quantification of the differences in cell index 24 hours after initiation of anti-Sdc2 pAb treatment. (C–H) Staining with junction markers (VE-cadherin and ZO-1) of mRECs stimulated with VEGF-A for 12 hours and then treated with either vehicle or Sdc2 pAb for 24 hours. (I, J) Assessment of VEGFA-induced proliferation in the presence of Sdc2 pAb versus vehicle. *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with Šidák multiple-comparison test (A, B, I, J).

To further validate the effect of Sdc2 pAb on endothelial permeability, we stained mRECs with anti–VE-cadherin and ZO-1 antibodies that, respectively, mark adherence and tight junctions. To this end, confluent mRECs were first incubated in VEGF-A for 12 hours, followed by the addition of either vehicle or anti-Sdc2 pAb for 24 hours. Control (untreated) cells did not receive VEGF-A or anti-Sdc2 pAb. In agreement with prior publications,²⁵ VEGF-A stimulation led to the loss of VE-cadherin from endothelial cell–cell junctions (Figs. 4C, 4D), fragmentation of VE-cadherin staining, and appearance of gaps between cells in the monolayer (Fig. 4D, white arrows). However, the addition of anti-Sdc2 pAb completely restored the morphology of cell–cell contacts, which appeared undistinguishable from those of control cells (Figs. 4C, 4E). 

Analysis of tight junctions using ZO-1 staining produced similar results. mRECs treatment with VEGF-A led to a marked reduction and disorganization of ZO-1 staining (Figs. 4F, 4G), a finding compatible with the loss of tight junction integrity. This effect was completely rescued by the addition of anti-Sdc2 pAb (Figs. 4F, 4H) 

Finally, to confirm that anti-Sdc2 pAb does not impair VEGF-A–driven endothelial cell proliferation, mRECs were cultured in the presence of VEGF-A with or without the addition of the anti-Sdc2 pAb. Time-course quantitative assessment indicated no differences in VEGF-A-induced proliferation in the presence or absence of anti-Sdc2 pAb (Figs. 4I, 4J). 

Discussion

The results of this study show that both ocular and systemic administration of anti-Sdc2 pAb leads to reduced edema formation in subretinal layers following laser injury in the mouse nAMD model. The extent of edema reduction with anti-Sdc2 pAb therapy was similar to that achieved with an intravitreal administration of a VEGF receptor trap. Importantly, prevention of vessel leakage observed with both intravitreal and systemic anti-Sdc2 pAb led to a strong reduction in retinal lesion volumes that was equal to the anti-VEGF therapy. 

Anti-VEGF therapy in nAMD achieves its salutary therapeutic effect by blocking the formation of new vessels in the retina or subretinal layers, as well as inhibiting increased permeability associated with local inflammation and formation of the new vasculature. In contrast, anti-Sdc2 therapy has no effect on the extent of angiogenesis or the survival of the existing vasculature, but it restores the integrity of leaky new blood vessels, thereby preventing subretinal edema formation. These findings become particularly compelling in light of the accumulating experience with long-term anti-VEGF treatments. In particular, local side effects of this therapy include RPE tears, underlying fast progressing geographic atrophy, and anatomical changes of the choriocapillaris, among others.^26–30 These side effects are likely a direct consequence of the complete suppression of VEGF signaling by current anti-VEGF therapies that not only inhibit VEGF-induced permeability but also block pro-survival effects of VEGF signaling that are necessary for maintenance of the capillary bed.¹² Regulation of nAMD-related edema by modulation of a specific VEGF signaling pathway that induces edema formation but leaving other aspects of VEGF signaling intact, as previously described in other settings,³¹ could minimize some of the current side effects observed with sustained anti-VEGF regimens. 

Our findings highlight the unusual role played by Sdc2 in endothelial biology and VEGF-A signaling and demonstrate that therapeutic targeting of this signaling pathway can be used to selectively target and inhibit pathological edema and inflammation without affecting endothelial cell survival. Previous studies in an acute stroke model in mice demonstrated the ability of anti-Sdc2 pAb to block peri-infarct edema formation in the brain, resulting in a significant reduction in stroke size.¹³ This neuroprotective effect of anti-Sdc2 pAb has been associated with inhibition of VEGF-A–induced vasogenic edema and preservation of blood–brain barrier integrity, a key determinant of the infarct size in the ischemic stroke.^32–35 

The anti-Sdc2 antibody used in Corti et al.¹³ and in our present study targets the DEP1 binding site on Sdc2, thereby preventing the two transmembrane proteins from interacting with each other. Free from its Sdc2 chaperone, DEP1 can bind to and dephosphorylate VEGFR2 on its Y951 site, thereby inhibiting vascular leak. The data in our present study support the following interpretation of the role of Sdc2 role in VEGF-A biology: Treatment of mREC in vitro or in vivo in mice following choroidal laser injury results in selective inhibition of VEGF-A–induced endothelial permeability, but endothelial cell proliferation and survival are not affected. Interestingly, although a deletion of DEP1 led to an increase in endothelial cells proliferation,³⁶ anti-Sdc2 pAb did not affect VEGF-A–induced endothelial cell proliferation. The difference can be explained by the fact that Sdc2 controls DEP1 at the surface level without affecting its total expression.¹³ The effect of DEP1 deletion would be very different from simply moving DEP1 from the plasma membrane into the cytoplasm, given that DEP1 can phosphorylate a number of intracellular proteins involved in migration and proliferation, such as the Src family kinases (SFKs), extracellular signal-regulated kinases (ERKs), and VE-cadherin, among others, and not just VEGFR2.³⁷ 

Importantly, anti-Sdc2 pAb therapy was effective in restoring the integrity of the endothelial barrier after the initiation of injury (or VEGF-A)-induced vascular leak as shown both in vitro and in vivo. It is also interesting to note that both systemic and intravitreal anti-Sdc2 pAb therapies were equally effective in suppressing laser-induced edema and were as potent as intravitreal anti-VEGF therapy. Additionally, the high effectiveness of systemic anti-Sdc2 pAb may also stem from the fact the antibody does not have to cross the retinal–brain barrier as it targets Sdc2-DEP1 interaction at the cell membrane of endothelial cells. 

Another interesting aspect of anti-Sdc2 therapy is its anti-inflammatory effect. Both intravitreal and systemic anti-Sdc2 pAb therapy resulted in a significant reduction in macrophage accumulation at sites of laser-induced lesions in the mouse model of CNV. This likely reflects the fact that the protein-rich edema fluid itself is pro-inflammatory.³⁸ A reduced presence of pro-inflammatory mediators at injury sites would be expected then to result in decreased macrophage accumulation. Indeed, there is a good correlation between the extent of edema reduction and the extent of macrophage accumulation following anti-Sdc2 pAb treatment. However, it is also possible that Sdc2 blockade may affect macrophage recruitment and/or accumulation at injury sites. Regardless, these results highlight the link between edema and inflammation that deserves further study. 

While anti–VEGF-A antibody can be effective when given intravenously in patient with nAMD,³⁹ this mode of therapy has not been pursued clinically, given the numerous well-established side effects of systemic anti-VEGF therapies.^40–43 Conversely, multiple lines of evidence suggest that anti-Sdc2 pAb therapy is likely to be safe when given systemically. First, both global and endothelial Sdc2 knockout mice develop normally, are healthy, and do not exhibit cardiovascular or systemic abnormalities.¹⁰ In particular, they are not hypertensive and exhibit no evidence of renal dysfunction. The only reported phenotype in these mice is a mild reduction in developmental retinal angiogenesis that resolves over time. In agreement with this, although deletions of either VEGF-A or VEGFR2 genes lead to embryonic lethality, knock-in mice that express VEGFR2 with the single mutation at the Y951 site recapitulate the molecular effect of Sdc2 pAb (selective inhibition of Y951 phosphorylation).^31,44 Second, anti-Sdc2 pAb does not interfere with non-permeability effects of VEGF-A signaling, including having no effects on endothelial survival, proliferation, and migration.¹³ Third, anti-Sdc2 pAb therapy shows high specificity and does not block other well-characterized permeability mediators such as histamine,¹⁰ thrombin, or tumor necrosis factor α (TNF-α) or interleukin-1β (IL-1β) (unpublished data). 

Intravitreal injection of anti-Sdc2 pAb did not induce evident signs of retinal toxicity. This was evaluated by ERG, a technique that has been successfully used to assess retinal toxicity in multiple preclinical studies.^45–47 Similarly, no change in CBCs was observed between controls and treated animals when anti-Sdc2 pAb was given systemically. Although it is difficult from this study to infer potential chronic toxicity of an anti-Sdc2 pAb therapy, we have observed that mice treated repeatedly with anti-Sdc2 pAb or following either a global or endothelial-specific Sdc2 deletion remain healthy in the long term. Another limitation of this study is the lack of ERG measurements following laser injury as a further readout to confirm the efficacy of anti-Sdc2 pAb and highlight potential differences with anti-VEGF treatment. Indeed, both full-field^48,49 and focal⁵⁰ retinography have been previously used to assess treatment efficacy in preventing retinal damage in the CNV model. 

In summary, our data demonstrate that both ocular and systemic therapy with anti-Sdc2 pAb is as effective as intravitreal anti-VEGF strategies at inhibiting edema formation, inflammation, and retinal lesions in a mouse nAMD model. A selective inhibition of edema formation utilizing anti-Sdc2 pAb may offer a new approach to therapy of these challenging patients and offer a better tolerability and side-effects profile than currently available therapeutic options. 

Acknowledgments

The authors thank Anders Kvanta for support and discussions during the preparation of this work. 

Supported in part by a grant from the National Institutes of Health (HL149343 to MS), the Open Philanthropy Foundation (MS), and VST-Bio, Inc. (MS). 

Disclosure: F. Corti, VST Bio (O); F. Locri, None; F. Plastino, None; P. Perrotta, None; K. Zsebo, VST Bio (O); E. Ristori, None; X. Yin, None; E. Song, None; H. André, None; M. Simons, VST Bio SAB (O)