Anti-Scg3 Gene Therapy to Treat Choroidal Neovascularization in Mice

Neovascular age-related macular degeneration (nAMD) with choroidal neovascularization (CNV) is a leading cause of blindness in the elderly in developed countries. The disease is currently treated with anti-angiogenic biologics, including aflibercept, against vascular endothelial growth factor (VEGF) but with limited efficacy, treatment resistance and requirement for frequent intravitreal injections. Although anti-VEGF gene therapy may provide sustained therapy that obviates multiple injections, the efficacy and side effects related to VEGF pathway targeting remain, and alternative strategies to block angiogenesis independently of VEGF are needed. We recently reported that secretogranin III (Scg3) induces only pathological angiogenesis through VEGF-independent pathways, and Scg3-neutralizing antibodies selectively inhibit pathological but not physiological angiogenesis in mouse proliferative retinopathy models. Anti-Scg3 antibodies synergize dose-dependently with VEGF inhibitors in a CNV model. Here, we report that an adeno-associated virus-8 (AAV8) vector expressing anti-Scg3 Fab ameliorated CNV with an efficacy similar to that of AAV-aflibercept in a mouse model. This study is the first to test an anti-angiogenic gene therapy protocol that selectively targets pathological angiogenesis via a VEGF-independent mechanism. The findings support further safety/efficacy studies of anti-Scg3 gene therapy as monotherapy or combined with anti-VEGF to treat nAMD.


Introduction
Neovascular age-related macular degeneration (nAMD), also known as exudative or wet AMD, is a leading cause of blindness in the elderly populations of developed countries, with the incidence currently estimated to be >1.5 million cases in the US [1]. Wet AMD manifests as choroidal neovascularization (CNV) in the macula and was traditionally treated with thermal laser or photodynamic therapy but with poor efficacy and potential side effects [2]. The advent and approval of anti-angiogenic biological drugs against vascular endothelial growth factor (VEGF), such as ranibizumab and aflibercept, delivered by intravitreal injections represent a major breakthrough for nAMD therapy but with limited efficacy. Clinical trials reported visual acuity improvement (>15 letters) only in 14.6-40.3% of patients treated with anti-VEGF vs. 3.8-5.6% in sham controls [3]. The requirement for frequent intravitreal injections is a major drawback. The half-lives of intravitreally injected ranibizumab and aflibercept are 9 and 11 days, respectively, so monthly injections are required to maintain therapeutic efficacy [4,5]. Frequent intravitreal injections increase the risk of eye complications, including ocular pain, cataract, retinal and vitreous hemorrhage, retinal detachment, endophthalmitis and elevated intraocular Neutralizing activity of anti-Scg3 mAb to block Scg3-induced proliferation of HUVECs. n = 3 wells/group. (C) Functional validation of AAV-mediated transgene expression. HEK293 cells were transduced by indicated AAVs. Conditioned media were analyzed for binding activity to immobilized Scg3, VEGF or BSA (negative control) using ELISA. n = 3 wells/group. (D) AAV-mediated FLAG-tagged transgene gene expression in mouse retinas. Indicated AAVs were intravitreally injected into mice, and retinas were isolated one month after AAV injection and analyzed using Western blot with anti-FLAG mAb. ± SEM, one-way ANOVA test.

Characterization of AAV Plasmids and Viral Vectors
Human embryonic kidney 293 cells (HEK293) were seeded on 6-well plates at 2 × 10 5 cells/well in Dulbecco's modified Eagle's minimum essential medium (DMEM) (Gibco/Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco), 1x GlutaMAX (Thermo Fisher) and 1% penicillin/streptomycin (Gibco) and incubated at 37 °C overnight in a humidified atmosphere with 5% CO2. The cells were transduced with AAV-anti-Scg3Fab, AAV-aflibercept or AAV-mCherry at 5 × 10 6 vg/mL. The medium was replaced with a serum-free 293SFM II medium (Thermo Neutralizing activity of anti-Scg3 mAb to block Scg3-induced proliferation of HUVECs. n = 3 wells/group. (C) Functional validation of AAV-mediated transgene expression. HEK293 cells were transduced by indicated AAVs. Conditioned media were analyzed for binding activity to immobilized Scg3, VEGF or BSA (negative control) using ELISA. n = 3 wells/group. (D) AAV-mediated FLAG-tagged transgene gene expression in mouse retinas. Indicated AAVs were intravitreally injected into mice, and retinas were isolated one month after AAV injection and analyzed using Western blot with anti-FLAG mAb. ± SEM, one-way ANOVA test.

Laser-Induced Choroidal Neovascularization
Mice were subjected to laser photocoagulation to induce CNV at 1 or 4 months after the AAV injection by using the following procedures: For pupil dilation, anesthetized mice received a topical eye drop of 1% tropicamide (Akorn, Lake Forest, IL, USA) and 2.5% phenylephrine (Paragon BiTeck, Portland, ME, USA). A green laser beam (532 nm, 240 mW, 150 ms, 50 µm spot) was applied to the retina around the optic disk (4 spots/retina) using a Micron IV retinal imaging system (Phoenix Research Labs, Pleasanton, CA, USA). Gaseous bubbles formed at laser spots indicated the rupture of Bruch's membrane. Lesions with retinal hemorrhage on Day 0 and linear or fused lesions on Day 7 were excluded.

Fluorescein Angiography
Fluorescein angiography was conducted on Day 7 after laser photocoagulation. All fluorescein angiography images were taken 6 min after an injection of fluorescein sodium (0.1 mL/mouse, 2.5%, Akorn) in anesthetized mice with standardized instrument settings using a Spectralis Tracking OCTA system (Heidelberg Engineering, Franklin, MA, USA). Fluorescein angiography images were analyzed using ImageJ software (NIH). The area and intensity of the laser spots were normalized to cognate the entire viewing field of the eye. After fluorescein angiography, the retinal pigment epithelium (RPE)-choroid-sclera eyecups (RPE eyecups) were isolated from the euthanized mice, fixed, stained with Alexa Fluor 488-isolectin B4 (AF488-IB4, 10 µg/mL, Thermo Fisher), flat-mounted, and analyzed using a Keyence BZ-X810 structured illumination microscope (SIM) and Keyence software.

Immunohistochemistry
The mice with CNV were euthanized with CO 2 inhalation after fluorescein angiography, and they were immediately perfused intracardially with PBS, followed by 4% paraformaldehyde (PFA) and eye enucleation. The anterior segments, including the cornea and lens, were removed to yield RPE eyecups that were embedded in the optimal cutting temperature (OCT) compound (Tissue-Tek; Miles Scientific, Napierville, IL, USA) and cryosectioned with a 10 µm thickness. The retinal sections were immunostained with anti-FLAG mouse M2 mAb (Sigma, #F1804, dilution 1:200), followed by Alexa Fluor 594conjugated anti-mouse IgG F(ab') 2 (Cell Signaling, Danvers, MA, USA; #8889S; dilution 1:1000) and Hoechst staining, and analyzed using SIM microscopy.

Western Blot
Western blots were performed as previously described [20]. Briefly, the total protein was isolated from the retinas and homogenized in a RIPA buffer (Sigma) supplemented with a protease inhibitor cocktail (Sigma, Cat. #P8340). The total protein was quantified using a BCA protein assay Kit (Thermo Fisher), separated by SDS-PAGE (20 µg/lane), and transferred onto nitrocellulose membranes (Millipore). The membranes were probed with anti-FLAG M2 mAb (1:1000) and a horseradish peroxidase (HRP)-conjugated secondary antibody (Ab) for chemiluminescence signal detection, followed by stripping and reprobing with anti-β-actin mAb.

Statistical Analysis
Data are expressed as mean ± SEM. A statistical analysis was performed using a one-way ANOVA test. p < 0.05 was considered significant.

Construction and In Vitro Characterization of AAVs
We chose anti-Scg3 mAb over the related humanized antibody (hAb) [20,23] for this project to minimize potential mouse anti-human IgG Ab that may attenuate the efficacy of anti-Scg3 gene therapy. We constructed a mouse anti-Scg3 mAb Fab fragment in an AAV2 vector with a CAG promoter [28]. The heavy and light chains contained the same human/murine IgG heavy chain signal peptide, separated by a furin GT2A cleavage site ( Figure 1A) [24,25]. A FLAG tag was attached to the C-terminus of the light chain. AAV-aflibercept was constructed in a similar fashion. Anti-Scg3 mAb Fab was verified for its neutralizing activity to block Scg3-induced endothelial proliferation of HUVECs in a culture before AAV construction ( Figure 1B). After packaging into AAV8, purified AAVs were used to transduce HEK293 cells. Conditioned media were collected and analyzed for binding activity to immobilized VEGF or Scg3 using ELISA. The results show that the HEK293 cells transduced with AAV-anti-Scg3Fab and AAV-aflibercept secreted functionally active anti-Scg3 Fab and aflibercept with corresponding binding activity to Scg3 and VEGF, respectively ( Figure 1C). The results suggest that the heavy and light chains of anti-Scg3 Fab are appropriately processed by the furin protease cleavage and assembled into a functionally active Fab fragment. By contrast, a similar construct of anti-Scg3 Fab without the second signal peptide produced non-functional Fab without Scg3-binding activity.

Transgene Expression In Vivo
To characterize gene expression in mouse retinas, we intravitreally injected AAV-anti-Scg3Fab, AAV-aflibercept and AAV-mCherry into mice and analyzed gene expression one month post-injection. Western blot detected the expressions of the FLAG-tagged anti-Scg3 Fab light chain and aflibercept at approximately 27 kDa and 67 kDa, respectively, under reduced and denaturing conditions ( Figure 1D). The predicted molecular weight was 26.7 kDa for anti-Scg3 light chain-FLAG and 51.4 kDa for aflibercept-FLAG without glycosylation. The size of the anti-Scg3 light chain also confirmed the cleavage by the furin protease. According to FDA drug information, the aflibercept dimer produced from CHO cells is 97 kDa (monomer 48.5 kDa) but migrates as 115 kDa, with the additional 15% of the total molecular weight attributed to glycosylation [29]. Therefore, aflibercept expressed in the mouse retina by AAV-aflibercept was glycosylated with a~30% increase in the molecular weight.
To independently validate the transgene expression pattern in the mouse retinas, we performed immunohistochemistry one month after the AAV injection using anti-FLAG mAb. The assay detected the expressions of FLAG-tagged anti-Scg3 Fab and aflibercept in the retinal ganglion cell (RGC) layer, inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), photoreceptor inner segments (PISs) and RPE. A reduced expression was also detected in the outer nuclear layer (ONL) and photoreceptor outer segments (POSs) (Figure 2). These expression patterns suggest that the transgenes are expressed throughout the entire retina. No or a minimal FLAG signal was detected for AAV-mCherry or retinal sections without the primary Ab, supporting the signal specificity.
AAV-mCherry or retinal sections without the primary Ab, supporting the signal specificity.

Anti-Angiogenic Gene Therapy to Inhibit CNV
To compare efficacies, we intravitreally injected AAV-anti-Scg3Fab, AAV-aflibercept and AAV-mCherry into mice, followed by CNV induction one month after the AAV injection. We quantified the CNV leakage and related therapeutic efficacy in anesthetized mice using fluorescein angiography 7 days after CNV induction. The results indicate that AAV-

Anti-Angiogenic Gene Therapy to Inhibit CNV
To compare efficacies, we intravitreally injected AAV-anti-Scg3Fab, AAV-aflibercept and AAV-mCherry into mice, followed by CNV induction one month after the AAV injection. We quantified the CNV leakage and related therapeutic efficacy in anesthetized mice using fluorescein angiography 7 days after CNV induction. The results indicate that AAV-anti-Scg3Fab significantly ameliorated CNV leakage in terms of the leakage area and intensity ( Figure 3A-C). AAV-aflibercept as a positive control reduced the CNV leakage area and intensity by the same degree as the vehicle control vector AAV-mCherry.
anti-Scg3Fab significantly ameliorated CNV leakage in terms of the leakage area and intensity ( Figure 3A-C). AAV-aflibercept as a positive control reduced the CNV leakage area and intensity by the same degree as the vehicle control vector AAV-mCherry. To quantify CNV lesions using histopathology, we euthanized the mice after fluorescein angiography, and we isolated and stained the RPE-choroid-sclera eyecups with AF488-IB4 to label CNV vessels. An SIM microscopy analysis confirmed that AAV-anti-Scg3Fab and AAV-aflibercept significantly reduced CNV lesion size and 3D volume with similar efficacy (Figure 3D-F).

Long-Term Efficacy in Alleviating CNV
To investigate whether gene therapy in this model provides long-term therapeutic benefits, we induced CNV in mice 4 months after AAV transduction and analyzed therapeutic efficacy via fluorescein angiography and immunostaining with AF488-IB4 7 days after the CNV induction. The results show that both AAV-anti-Scg3Fab and AAV-aflibercept significantly and similarly reduced CNV lesion size and 3D volume ( Figure 4D-F). However, unlike the 1-month protocol, the CNV leakage size and intensity were not significantly reduced by either treatment at 4 months ( Figure 4A-C). Compared to transgene expression at one month after the AAV injection, immunohistochemistry revealed that the transgene expression at 4 months was markedly reduced (Figure 2 vs. Figure 5). To quantify CNV lesions using histopathology, we euthanized the mice after fluorescein angiography, and we isolated and stained the RPE-choroid-sclera eyecups with AF488-IB4 to label CNV vessels. An SIM microscopy analysis confirmed that AAV-anti-Scg3Fab and AAV-aflibercept significantly reduced CNV lesion size and 3D volume with similar efficacy (Figure 3D-F).

Long-Term Efficacy in Alleviating CNV
To investigate whether gene therapy in this model provides long-term therapeutic benefits, we induced CNV in mice 4 months after AAV transduction and analyzed therapeutic efficacy via fluorescein angiography and immunostaining with AF488-IB4 7 days after the CNV induction. The results show that both AAV-anti-Scg3Fab and AAV-aflibercept significantly and similarly reduced CNV lesion size and 3D volume ( Figure 4D-F). However, unlike the 1-month protocol, the CNV leakage size and intensity were not significantly reduced by either treatment at 4 months ( Figure 4A-C). Compared to transgene expression at one month after the AAV injection, immunohistochemistry revealed that the transgene expression at 4 months was markedly reduced (Figure 2 vs. Figure 5).

Discussion
We provide the first evidence that an Scg3 antagonist with a signaling pathway distinct from that of VEGF reduced CNV lesion size and leakage in a gene therapy protocol that is quantitatively equivalent to AAV-aflibercept administered according to the same regimen and gene dosing. Both genes were delivered by AAV8, a serotype appropriate for ocular indications [25]. The vectors delivered equivalent levels of expression at one month after intravitreal injections in mice with 5.0 × 10 8 vg/eye and generated secreted gene products that displayed the expected selective binding to immobilized VEGF and Scg3 ligands. Immunohistochemistry revealed common distributions of the respective AAV-aflibercept and AAV-anti-Scg3Fab gene products in all retinal layers (Figure 2). AAV-anti-Scg3Fab and AAV-aflibercept reduced the CNV 3D volume and maximal lesion area by the same degrees, indicating equivalent efficacy at both short and extended time intervals after AAV transduction. However, protection against CNV leakage was significantly effective only in the 1-month group and not at 4 months after transduction. We attribute the diminished efficacy to the relatively low vector dose, resulting in a gradual reduction in transgene expression over time (Figure 2 vs. Figure 5). The reduced transgene expression over time may be due to epigenetic regulations after AAV integration. Although the sustained expression of AAV in ocular tissues is well established [25,[30][31][32], Liu et al. reported a strict dose-dependent expression and therapeutic efficacy of AAV8 anti-VEGF Fab at 1 month in a mouse CNV model with a markedly reduced efficacy at doses below 10 9 vg/eye [25]. This is also consistent with the high dose of 2 × 10 12 vg/eye used to express AAV-aflibercept in African green monkeys, equivalent to~2 × 10 9 vg/eye in mice assuming a factor of~660× for primate versus mouse vitreous volume [30][31][32][33]. Such a low-dose instability may account for the variable results seen in some gene therapy trials. In a Phase I trial of intravitreal AAV2-sFLT01, gene expressions in the aqueous humor of patients injected with <2 × 10 10 vg/eye were below the limits of detection [34]. An insufficient transgene expression of 1 × 10 11 vg/eye AAV2 via subretinal delivery in patients is suspected for the low efficacy and failure of sFLT01 clinical trials [35,36]. The results suggest a narrow window or threshold of the AAV vector dose to achieve sustained therapeutic gene expression [25]. In future studies, we will characterize the doseresponse curves of AAV-anti-Scg3Fab and AAV-aflibercept in parallel to determine the dose requirement for persistent anti-Scg3 gene therapy. Additionally, we will investigate whether AAV2 and AAV2.7m8 further improve therapeutic duration [34,37].
Aflibercept contains the Ig domains of VEGFR1 and VEGFR2 fused to the Fc region of human IgG and functions as a soluble VEGF decoy receptor that binds and neutralizes VEGF-A, VEGF-B and placental growth factor (PIGF) [38]. Aflibercept is currently a firstline therapy for nAMD, macular edema following retinal vein occlusion, DR and diabetic macular edema [38]. The related and competing first-line VEGF-neutralizing reagents that are also FDA-approved for the same or similar ocular indications include ranibizumab; bevacizumab (used off-label for nAMD); brolucizumab; and more recently faricimab, a bispecific anti-VEGF and anti-Ang2 Ab [39]. To achieve stable efficacy for nAMD, all the approved drugs of this category require frequent intraocular injections, a requirement that adversely affects patient quality of life and treatment compliance [9,10]. Multiple studies have confirmed the requirement for rigid adherence to injection regimens and the loss or reversal of therapy caused by non-compliance [7,8]. Consequently, intense efforts are underway to circumvent monthly injections by devising sustained-release technologies and/or gene therapy that promises persistent therapy via a single intraocular injection [37]. After the failure of early efforts to translate endogenous angiogenesis inhibitors, such as angiostatin/endostatin and pigment epithelial-derived factor, into gene therapy applications, the field focused on adapting the current FDA-approved repertoire of VEGF pathway blockers [40]. After promising preclinical results [37], ADVM-022, an optimized AAV2 vector encoding aflibercept, is currently in a Phase 2 trial to treat nAMD [19,41]. In a second ongoing Phase I/IIa trial, RGX-314, an AAV8 vector expressing a mAb fragment similar to ranibizumab, delivered by a subretinal injection is being evaluated in patients with nAMD [36]. The safety/efficacy results at 1.5 years reported an improved or stabilized best-corrected visual acuity (BCVA), a reduced central retinal thickness (CRT) and markedly decreased requirements for supplemental anti-VEGF injections [42].
The progression of anti-angiogenesis protocols to gene therapy clinical trials promises to relieve the constraints of frequent intravitreal injections for patients with nAMD but fails to address the confounding issues of suboptimal therapy and other adverse side effects associated with current anti-VEGF therapy. Recent studies reported that VEGF inhibitors are effective in alleviating CNV in young but not aged animal models [43,44], implying that the promised efficacy of anti-VEGF in young animals may not be translated quantitatively to aged patients with nAMD. The alternating use and/or combination of these reagents with inhibitors of VEGF-dependent accessory factors, such as PDGF, Ang2 and semaphorin 6A [45], are unlikely to significantly improve efficacy as next-generation pharmacology or gene therapy for CNV indications [11,12]. Central to this dilemma is that the independence of VEGF signaling is a likely prerequisite for synergistic combinations with current anti-VEGF protocols, and, except for anti-Scg3, no inhibitors of pathological angiogenesis that work independently of VEGF signaling have been described.
Scg3 was discovered by our group using a novel comparative ligandomics technology to screen for disease-restricted ligands in mouse models of DR and CNV [20,22]. Scg3 is a disease-restricted angiogenic factor that selectively binds to diseased but not healthy vessels. Scg3-neutralizing Abs alleviated CNV, DR and retinopathy of prematurity (ROP) in mouse models with an efficacy equivalent to that of aflibercept [20][21][22][23][46][47][48]. The inhibition of angiogenesis by anti-Scg3 Abs in all cases was independent of VEGF, consistent with separate angiogenic signaling pathways and synergy between anti-Scg3 hAb and aflibercept [23]. In Scg3, we appear to have uncovered a hitherto invisible but long-soughtafter disease-restricted proangiogenic pathway that operates in parallel but independent of VEGF in pathological states [20]. Such a property that restricts Scg3 actions to pathological angiogenesis is consistent with our findings of safety and wide therapeutic windows of anti-Scg3 vs. anti-VEGF [46,47]. Because, like anti-VEGF biologics, intraocular anti-Scg3 hAb is expected to have a short therapeutic duration, and clinical applications will also require repeated injections. Therefore, translation from protein pharmacology to gene therapy is an important next step to relieve these constraints.
Safety concerns have overshadowed the progression of anti-VEGF gene therapy since its inception. In addition to indiscriminate binding to diseased and healthy vessels, VEGF possesses neurotrophic and neuroprotective properties that promote neuronal growth and survival [49][50][51]. We and others have shown that intravitreal aflibercept induces abnormalities in electroretinography (ERG) and retinal structure in animal models [47,48,52]. Similar adverse effects on retinal function and structure were also reported in patients treated with VEGF inhibitors in some clinical studies [53][54][55][56]. Clinical studies also revealed that intravitreal anti-VEGF agents can interfere with the function of the central nervous system [57][58][59]. Long-term anti-VEGF therapy for nAMD may increase the risk of geographic atrophy [60]. A recent study reported trends of b-wave suppression by AAVaflibercept in adult non-human primates 19 months after gene therapy [37]. Aged neurons in patients with AMD may be more susceptible to low-ambience VEGF ligands caused by anti-VEGF therapy, thereby contributing to a limited improvement in visual acuity [3]. Indeed, the possible neurotoxicity of anti-VEGF was implied in clinical trials of nAMD and diabetic macular edema (DME), in which high-dose ranibizumab inversely reduced long-term visual acuity despite improving vascular symptoms [61,62]. It is unclear whether neurotoxicity is caused by the direct effects of VEGF blockade on neurons or indirectly through the suppression of healthy vessels. It seems possible that patients with nAMD are more sensitive to manipulations of VEGF homeostasis due to age and comorbidities. Safety concerns related to chronic VEGF suppression prompted the testing of multiple strategies to regulate anti-VEGF gene therapy [63][64][65]. Because our studies indicate that Scg3 regulates only pathological angiogenesis, with no effect on healthy vessels or neurons [24,26], such safety issues may not apply, and long-term constitutive gene therapy with anti-Scg3 hAb is expected to be safe. Our findings warrant further investigation to compare the efficacy and safety of optimized AAV-anti-Scg3 and AAV-anti-VEGF for monotherapy or combination therapy in large animal models. Informed Consent Statement: Not applicable.

Data Availability Statement:
All data related to this study are reported herein.