Calcitriol and Non-Calcemic Vitamin D Analogue, 22-Oxacalcitriol, Attenuate Developmental and Pathological Ocular Angiogenesis Ex Vivo and In Vivo

Aberrant ocular blood vessel growth can underpin vision loss in leading causes of blindness, including neovascular age-related macular degeneration, retinopathy of prematurity and proliferative diabetic retinopathy. Current pharmacological interventions require repeated invasive administrations, lack efficacy in some patients and are associated with poor patient compliance and tachyphylaxis. Small molecule vitamin D has de novo pro-differentiative, anti-proliferative, immunomodulatory, pro-apoptotic and anti-angiogenic properties. Here, our aim was to validate the anti-angiogenic activity of the biologically active form of vitamin D, calcitriol, and a selected vitamin D analogue, 22-oxacalcitriol, across a range of ocular angiogenesis models. First, we validated the anti-angiogenic activity of calcitriol, showing calcitriol to significantly inhibit choroidal sprouting in an ex vivo mouse choroidal fragment sprouting assay. Viability studies in human RPE cell line, ARPE-19, suggested non-calcemic vitamin D analogues have the least off-target anti-proliferative activity compared to calcitriol and additional analogues. Thereafter, the ocular anti-angiogenic activity of non-calcemic vitamin D analogue, 22-oxacalcitriol, was demonstrated in the ex vivo mouse choroidal fragment sprouting assay. In zebrafish larvae, 22-oxacalcitriol was anti-angiogenic, inducing a dose-dependent reduction in choriocapillary angiogenesis. Inhibition of mouse retinal vasculature development was not induced by systemically delivered calcitriol. However, both calcitriol and 22-oxacalcitriol administered intraperitoneally significantly attenuate choroidal neovascularisation lesion volume in the laser-induced CNV mouse model. 22-oxacalcitriol presented with a more favourable safety profile than calcitriol. In summary, calcitriol and 22-oxacalcitriol attenuate ex vivo and in vivo choroidal vasculature angiogenesis. Vitamin D has potential as a preventative or interventional treatment for ophthalmic neovascular indications.


Introduction
Growth of pathological ocular blood vessels can underpin vision loss in leading causes of blindness including neovascular age-related macular degeneration (nAMD) and proliferative diabetic retinopathy (PDR). Worldwide 8.7% of blindness results from AMD. nAMD accounts for 10% of AMD cases but ˃80% of cases with poor visual acuity [1][2][3][4]. nAMD and rapid vision loss is driven by pathological choroidal vasculature angiogenesis. This pathological choroidal vasculature can be deficient in tight junctions, leak plasma or blood, cause scarring, project through the Bruch's membrane, cause retinal pigmented epithelium (RPE) detachment and disrupt normal perfusion of the retina [5][6][7]. Worldwide 382 million people suffer from diabetes, ~35% of whom develop DR, making this the leading cause of blindness in the working age population [8]. Severe vision loss is a consequence of macular oedema and the sprouting of poorly formed retinal vasculature into the vitreous [8]. Retinal neovascularisation triggered by insufficient perfusion can result in haemorrhaging and retinal detachment [9].
After development, a tightly regulated balance of pro-and anti-angiogenic factors maintain the mature quiescent vasculature. Pathological insults such as hypoxia can disrupt this equilibrium and promote neovascularisation [10,11]. VEGF is a pivotal regulator of nAMD, clinically evident from the success of anti-VEGF targeting therapies. Ranibizumab (LUCENTIS®), bevacizumab (Avastin®) and aflibercept (Eylea®) are utilised in the treatment of ocular neovascularisation [12]. These anti-VEGF therapies cause vessel regression and improve visual function [4]. Despite representing the standard of care, several treatment limitations exist. Firstly, with molecular weights between 50-149 kDa, current interventions require administration via intravitreal injection [13]. This places a burden both on patients and clinicians, resulting in inadequate dosing, exemplified by the CATT/IVAN trial, where an average of 4-5 treatments were administered compared to the recommended 7-8 [7]. Secondly, repeat administrations are required: aflibercept has the greatest intravitreal half-life yet injections are required every 2 months [4,14]. Thirdly, anti-VEGF therapies are associated with a severe economic burden, with aflibercept costing approximately €1,000 per injection [15]. Finally, anti-VEGF therapy can lack efficacy in "non-responsive" populations and tachyphylaxis can occur. A non-responsive population of 45% is reported for bevacizumab [16]. Tachyphylaxis is postulated to be a consequence of compensatory VEGF upregulation or generation of neutralising antibodies [14]. These limitations highlight the need to identify and develop safe, efficacious, cost effective anti-angiogenics with a less invasive route of administration.
Vitamin D is a fat-soluble vitamin and steroid hormone with pleiotropic health implications.
Recognised as having pro-differentiative, anti-proliferative, immunomodulatory, pro-apoptotic and anti-angiogenic properties; vitamin D is under examination for malignant, cardiovascular, cognitive, metabolic, infectious and autoimmune indications [17][18][19]. Interestingly, the vitamin D receptor (VDR) is expressed in the cornea, lens, ciliary body, RPE, ganglion cell layer and photoreceptors, supporting ocular functions [20]. In 2017, we reported calcitriol (vitamin D) and diverse VDR agonists including vitamin D2 analogues, vitamin D3 analogues and a pro-hormone to attenuate in vivo ocular vasculature development in zebrafish [21]. Further interrogation of the anti-angiogenic activity was needed in mammalian models to assess the therapeutic potential of vitamin D.
Here, we examined the anti-angiogenic activity of calcitriol or 22-

Mouse choroidal sprouting angiogenesis assay
The mouse choroidal sprouting procedure was adapted from Shao et al [22]. C57BL/6J mice aged between 6-12 weeks were euthanised by CO2 asphyxiation, eyes immediately enucleated and placed in ice cold Endothelial Cell Growth Medium (PromoCell). Eyes were cut along the pars planar, cornea removed, lens removed, and four incisions made facilitating flattening of the eyecup. The neural retina was removed from the RPE/choroid complex and six-eight 1 x 1 mm RPE-choroid explants cut from each quadrant. Explants were transferred to 30 µl thawed Matrigel® (BD Biosciences) in a 24 well plate, uniformly orientated, incubated for 20 min at 37°C (5% CO2, 95% O2) and 500 μl medium applied. Following 1 day of culturing, medium was exchanged, and vehicle control, calcitriol and 22oxacalcitriol treatments applied with a final well volume of 500 μl. Treatments were replenished on day 3-4. Culturing was ended on day 7 and explants were imaged live, calcein stained or fixed in 4% PFA overnight.

Image acquisition and sprouting area quantification
Brightfield images were acquired using Olympus SZX16 or Zeiss Axiovert 200 M microscopes with Cell^F or Zeiss Axiovision image analysis software. Sprouting area and explant area were manually quantified using ImageJ freehand tool and explant area subtracted from overall area. Statistical differences between vehicle-and drug-treated samples were determined by one-way ANOVA with Dunnett's post-hoc test. Statistical analyses were performed with PRISM 5 software and significance accepted where P≤0.05.
Ultra-thin ocular cross sections, 1 µm, were acquired using a diamond knife and Leica EM UC6 microtome. Sections were toluidine blue stained, cover-slipped (DPX mounting medium) and representative images acquired (Nikon Eclipse E80i Microscope, Canon camera). Mouse retina morphology between vehicle control and calcitriol treated samples was compared to identify deviations in retina cell organisation, retinal thickness and pyknotic nuclei presence.

Mouse model of retinal vasculature development
Experimental design followed Yagasaki et al [23]. Dams along with their pups were raised in standard light (12 h light and 12 h dark cycle) and standard air conditions for the duration of the study.

Calcitriol dosing
Calcitriol was prepared as previously reported [24]. Calcitriol (Selleckchem) was dissolved to 1 mg/ml in ethanol and working dilutions to 1 µg/ml in PBS prepared. Calcitriol multiple injection study: Pups received a 3.75 ng calcitriol or vehicle control s.c. treatment on P1, P3, P5 and P7. Animal welfare was monitored daily until P4 or P8. On P4 and P8 mouse pups were euthanised by cervical dislocation, eyes immediately enucleated and fixed with 4% PFA overnight at 4°C.

Retina flatmount
Fixed eyes were positioned on a made-for-purpose indented dental wax strip, in PBS under a dissecting microscope. The eye was gripped at the optic nerve with a Dumont no. 5 forceps and excess exterior muscle removed using a springbow microdissection scissors. The optic nerve was removed, the eye was pierced along the pars planar, the anterior eye removed using a springbow dissection scissors and lens removed. The remaining eyecup was transferred to a glass slide and 4 incisions made 2 mm from the site of the optic nerve with a no. 11 scalpel blade dividing the retina into 4 quadrants. Using a no. 11 blade the periphery of the quadrants was cut, straightening the outer edge and preventing curling of the retina. Retinal flat mounts were stored in perm/block buffer composed of PBS with 0.5% Triton-X100, 1% goat serum, and 0.1 mM CaCl2 (Sigma-Aldrich).

Isolectin staining
Isolectin staining was performed as previously reported [25]. Retina flat mounts underwent permeabilization with perm/block buffer overnight at 4°C. Perm/block buffer was replaced with 20 µg/ml GS isolectin B4 (ThermoFisher Scientific) in perm/block buffer and incubated overnight at 4°C.
Flat-mounts underwent 8 perm/block buffer washes over 4 hours at 37°C with 30 min interval changes. Flat mounts were stained with Alexa-streptavidin-564 (Thermo Fisher Scientific) diluted in perm/block buffer 1:500 overnight at 4°C. Flat-mounts underwent 8 perm/block buffer washes over 4 hours at 37°C with 30 min interval changes. Flat mounts were stored in PBS with 0.1 mM CaCl2.
Flat-mounts were transferred to a glass slide, the retina cover-slipped with Aqua-Poly mount (Polyscience Inc) and stored protected from light at 4°C.

Flat-mount image acquisition and retinal vasculature area quantification
Fluorescent images were acquired using a Zeiss AxioVert 200M fluorescent microscope, Andor IQ2 software with Andor montaging or Olympus SZX16 fluorescence microscope with Cell^F software.
Retinal superficial vasculature development was expressed as vasculature area compared to total flat mount area. Area measurements were performed using ImageJ software freehand tool.

Mouse Model of Laser-Induced Choroidal Neovascularisation
Wild-type female C57BL/6J mice, 6-8 weeks of age, were purchased from the Jackson Laboratory (Bar Harbor, ME), and housed under standard conditions [26]. Intraperitoneal injections (i.p.) of 60 mg/kg ketamine hydrochloride and 2.5 mg/kg xylazine mixture were used for anaesthesia, and isoflurane overdose for euthanasia. Body weights were determined daily. All analyses were performed by a masked investigator.

Calcitriol dosing
Calcitriol (1000 ng/mL in almond oil; Professional Compounding Centers of America, Houston, TX) was purchased from Indiana School of Medicine Laboratory Animal Resource Center's drug distribution center. Each mouse received once-daily i.p. injections of 5 µg/kg calcitriol or almond oil vehicle for 14 days (5 days on/2 days off). The dose was determined based on published observations [27].

22-oxacalcitriol dosing
Pure crystalline solid 22-oxacalcitriol (Cayman Chemical, Ann Arbor, MI) was dissolved in ethanol as previously reported to yield a 2 µg/µL stock solution [28,29]. This stock solution was diluted in PBS to working solution, 2 µg/mL 22-oxacalcitriol in 0.1% ethanol-PBS on day of the injection. Each mouse in the treatment group received once-daily i.p. injections of 15 µg/kg 22-oxacalcitriol or ethanol-PBS vehicle every day for 14 days.

Laser-induced choroidal neovascularisation
The L-CNV mouse model was performed as previously described [30]. Briefly, both eyes of 6-8 week old C57BL/6J mice were dilated using tropicamide, and subjected to laser treatment using 50 µm spot size, 50 ms duration and 250 mV pulses of an ophthalmic argon green laser wavelength 532 nm, coupled to a slit lamp. Three laser burns per eye were created around the optic nerve at 12, 3 and 9 o' clock positions. Optical coherence tomography (OCT) was performed in L-CNV mice as described previously [30], on days 7 and 14 post laser, using a Micron III intraocular imaging system (Phoenix Research Labs, Pleasanton, CA, USA). Briefly, eyes of anesthetised mice were dilated with 1% tropicamide solution (Alcon, Fort Worth, TX, USA) and lubricated with Gonak hypromellose ophthalmic solution (Akorn, Lake Forest, IL, USA). Horizontal and vertical OCT images were taken per lesion and L-CNV lesion volumes were obtained using the quantification method previously established [30,31]. To assess vascular leakage, fluorescein angiography was performed on day 14 post L-CNV by i.p. injection of 50 μL of 25% fluorescein sodium (Fisher Scientific, Pittsburgh, PA, USA). Fundus images were taken using the Micron III system and Streampix software. The posterior eyecups were then washed three times with PBS and mounted in fluorescent mounting medium (VectaShield; Vector Laboratories, Inc.) and cover-slipped. Confocal imaging and analysis of L-CNV lesion volume were performed as previously described [31]. Treatments were compared by unpaired T-test (two tailed) with Welch's correction, while mouse body weights were compared by two-way repeated-measures ANOVA with Holm-Sidak post hoc tests using GraphPad Prism v. 6.

Mouse eye RNA extraction, cDNA synthesis and QRT-PCR
Mouse eyes were harvested on day 14 post L-CNV induction, lens and cornea removed and eye stored in RNAlater. Total ocular RNA was extracted using mirVana™ miRNA Isolation Kit

Calcitriol attenuates mouse ex vivo choroid-RPE fragment sprouting angiogenesis.
Previously, we demonstrated calcitriol and seven other VDR agonists to inhibit ocular vasculature development in zebrafish larvae [21]. To identify the most active anti-angiogenic VDR agonist in mammalian models, the tubule formation assay, a late stage in vitro angiogenesis model, was performed. Human dermal-derived microvascular endothelial cells, HMEC-1 cells, were seeded in matrix and cultured with 10 µM calcitriol, 22-oxacalcitriol, tacalcitol or vehicle control and tubule formation quantified after 16 h. The Angiogenesis Analyzer for ImageJ was utilised for automatic unbiased measurement of tubule formation properties. Surprisingly, VDR agonist-treated HMEC-1 cells exhibited no significant difference in tubule formation compared to vehicle controls (Supp Figure   1A-B). Tubule formation properties are influenced by cell type (primary or immortalised), derivation (human or non-human) and tissue origin [32]. With ocular selective anti-angiogenic activity previously identified in zebrafish larvae, tubule formation was also investigated in human retinal-derived microvascular endothelial cells (HREC). HREC cells were seeded in a matrix and cultured with 10 µM calcitriol for 16 h. Again, no significant tubule formation difference was identified between vehicle control and calcitriol treated HREC cells (Supp Figure 1C-D).
To investigate the anti-angiogenic activity of calcitriol in a more physiologically relevant model, the ex vivo mouse choroidal sprouting angiogenesis assay was employed. This system is multicellular in nature and accounts for micro-environmental cues which support angiogenesis [22]. Calcitriol treatments between 5-10 µM significantly (p<0.001) reduced choroidal sprouting area by up to 93% compared to vehicle control. No significant difference in sprouting was identified with 1 µM calcitriol treatments ( Figure 1B-D). Calcein staining confirmed explant and sprout viability after 1-10 µM treatments ( Figure 1D).

Calcitriol attenuates RPE cell viability, while non-calcemic vitamin D3 analogues show a greater RPE cell safety profile.
Pro-apoptotic and anti-proliferative properties of calcitriol are known [33]. Such actions on endothelial cells could underpin the anti-angiogenic mechanism of calcitriol. However, induction of apoptosis is undesirable in neighbouring cells such as the RPE. Thus, we sought to identify VDR agonists with negligible effects on RPE cell viability. VDR agonist-induced changes in ARPE-19 cell number were determined by the surrogate measure of metabolic activity, quantified using the MTT assay. Calcitriol was tolerated over 24 h in ARPE-19 cells, with no significant change in cell viability with concentrations ≤20 µM (Figure 2A). However, treatments with ≥10 µM calcitriol for 48 h significantly reduced ARPE-19 cell viability in a concentration-dependent manner, 10 µM (p≤0.05), 15 µM (p≤0.01) and 20 µM (p≤0.001) ( Figure 2B). Cell viability in response to a range of VDR agonist treatments was subsequently investigated in ARPE-19 cells over 96 h. Vitamin D2 analogue, doxercalciferol, reduced ARPE-19 cell viability (~42%) only with 10 µM treatment (p≤0.01) ( Figure 2C). Vitamin D2 analogue, paricalcitol, had no significant effect on APRE-19 cell viability with treatments ≤10 µM ( Figure 2D). Vitamin D3 analogues, tacalcitol and calcipotriol induced a significant reduction (~42 and 29%, respectively) in ARPE-19 cell viability with concentrations ≥5 µM (Figure 2E-F). Non-calcemic vitamin D3 analogues were better tolerated. No significant change in ARPE-19 cell viability was identified with 22oxacalcitriol or EB 1089 treatments ≤10 µM (Figure 2G-H). Therefore, non-calcemic vitamin D3 Prior to in vivo assessment of the anti-angiogenic activity of calcitriol, ocular safety was evaluated in adult C57BL/6J mice. Adult mice received a single 50 ng subcutaneous calcitriol or vehicle control treatment, and retinal histology was investigated 7 days later. Vehicle controls and calcitriol treated mice present with animal welfare scores comparable to uninjected mice, with mouse weight recorded daily ( Figure 5B). On day 7, mice were euthanised, eyes enucleated, fixed and sectioned. Ultra-thin toluidine blue stained cross sections were investigated for the presence of pyknotic nuclei and deviations in the highly-organised cell lamination of the eye. Calcitriol treatments appeared well tolerated in mice, with no observable difference between vehicle and calcitriol treated retinal structures ( Figure 5A).
Retinal vasculature development after birth in mice provides a unique opportunity to study developmental angiogenesis. Normal mouse retinal vasculature growth is well documented, and drugs can inhibit this growth [34,35]. Calcitriol attenuates zebrafish ocular developmental angiogenesis, therefore, we sought to investigate if this response translated to the mouse. To validate the model,

22-oxacalcitriol inhibits mouse laser-induced choroidal neovascularisation without adverse effects
Previously calcitriol was reported to reduce retinal neovascularization in the mouse oxygen induced retinopathy (OIR) model of retinopathy of prematurity [27]. To extend upon this finding, we evaluated the effect of calcitriol in L-CNV. To visualize L-CNV and consequential vascular leakage, in vivo optical coherence tomography (OCT) imaging and fluorescein angiography were performed.
Choroidal neovascularisation was inhibited by 5 µg/kg/day calcitriol administered intraperitoneally ( Figure 6A-D) and fluorescein angiography revealed vascular leakage of CNV lesions was reduced in calcitriol treated mice ( Figure 6B). However, calcitriol treatment had adverse effects on body weight ( Figure 6D), perhaps due to hypercalcemia-induced toxicity [36]. Therefore, we sought to evaluate the effect of 22-oxacalcitriol, the non-calcemic bioactive analogue of calcitriol [37]. Notably

DISCUSSION
The ocular vasculature systems support retinal development and visual function. Pathological ocular neovascularisation is a hallmark in numerous diseases causing vision loss, including nAMD. Antiangiogenics have revolutionised nAMD treatment, yet their long-term safety profiles remain controversial and they require frequent intravitreal injection due to their molecular weights greater than 50 kDa [38,39]. Here, we investigated the anti-angiogenic activity and safety of agonists targeting the VDR in zebrafish larvae, human cell cultures and murine models. We report four significant  [22]. The choriocapillaris is the vasculature located immediately posterior to the RPE and highly sensitive to VEGF-A signalling during development and in pathologies such as AMD [1]. In initial studies, calcitriol attenuated ex vivo choroidal sprouting angiogenesis. This anti-angiogenic response is consistent with previous studies wherein calcitriol inhibits angiogenesis in a chick chorioallantoic membrane (CAM) model [40], an OIR model [27] and a transgenic mouse model of retinoblastoma [40]. Interestingly, here calcitriol did not inhibit in vitro tubule formation in human dermal-or retinal-derived endothelial cells. This is contrary to previous findings which demonstrated calcitriol to inhibit mouse retinal endothelial cell capillary network formation on a basement membrane [27]. Interestingly, Bao et al reported calcitriol to exert no effect on HUVEC tubule formation, yet attenuated HUVEC tubule formation when stimulated with prostate cancer cell conditioned medium [41]. These differences suggest that the anti-angiogenic activity of calcitriol is context-or environmental-dependent. As the choroidal explant assay employed here is multi-cellular in nature, it is plausible that anti-angiogenic responses are not directly induced at the endothelial cells, but instead through regulation from neighbouring cells.
Calcitriol treatment induces oedema and impairs visual function in zebrafish larvae [42]. Therefore, the safety profile of calcitriol was assessed here in a mammalian system. In mice, calcitriol treatment was well-tolerated in developing pups with no ocular morphological welfare concerns or weight loss at selected dose. We hypothesised that in vivo calcitriol would stall development of the mouse retinal vasculature, correlating with previous observations in zebrafish larvae [42]. However, mouse retinal vascular plexus development at P4 or P8 was not attenuated by 0.00375 μg calcitriol subcutaneously administered as a single dose or repeatedly on alternating days. The lack of efficacy could be a consequence of suboptimal dosing or short drug half-life (t½). Indeed, the calcitriol t½ is only a few hours, ~100 times quicker than 25(OH)D3 [43]. Alternatively, lack of anti-angiogenic activity could result from poor distribution to the eye, suboptimal vehicle selection or the early developmental status of the retinal vasculature. Calcitriol in vivo was previously reported to inhibit retinal neovascularization in a pathological OIR model of retinopathy of prematurity at a concentration of 5 μg/kg via intraperitoneal administration daily from P12 to P17 [27]. Subsequently, we hypothesised that calcitriol may exert an anti-angiogenic effect in vivo on the choroidal vasculature, consistent with our data in the ex vivo choroidal explants. Significantly, 5 µg/kg calcitriol administered intraperitoneally daily attenuated L-CNV in adult mice. However, the short calcitriol t½ suggests that frequent administration is required for chronic conditions [43]. Of further concern, the calcitrioltreated arm in the L-CNV study presented with reduced body weight, a response also reported by Albert et al in OIR studies [27]. Weight loss in calcitriol treated animals is likely a consequence of hypercalcemia.
Vitamin D regulates calcium mobilisation, therefore high-dose vitamin D treatment can induce hypercalcemia. VDR agonists including 22-oxacalcitriol were developed to reduce calcemic responses.
22-oxacalcitriol is a calcitriol analogue with an oxygen substituted for carbon at position 22, and is approved for the treatment of psoriasis due to its therapeutic activity and reduced calcaemic responses [44]. Interestingly, 22-oxacalcitriol inhibits CAM angiogenesis in a dose-dependent manner [45]. Here calcitriol decreased ARPE-19 cell viability, a result not observed with 22-oxacalcitriol.
Differing responses induced by vitamin D and analogues are the result of altered protein binding, metabolism, receptor affinity, dimerization and co-regulator recruitment [44]. 22-oxacalcitriol is reported to have reduced vitamin D-binding protein affinity, up to 500 times lower calcitriol [46]. 22-oxacalcitriol attenuates zebrafish hyaloid vasculature development, a lens associated system which metamorphs into a retina-associated vasculature system [42,47]. Here Vitamin D traditionally mediates its effects though the VDR, a nuclear receptor expressed diversely throughout the body which regulates the transcription of hundreds of genes [44]. The anti-angiogenic effects of vitamin D appear VDR-dependent. First, knockout of the VDR affects tumour vasculature integrity, resulting in vessel enlargement and reduced pericyte coverage [48]. Second, increased expression of pro-angiogenic factors including VEGF has been identified in tumours from VDR KO mice [48]. Third, calcitriol attenuates oxygen-induced neovascularisation in mice, a response reduced in VDR KO mice [49]. We previously reported calcitriol treatments to regulate VEGF expression in the developing eye [42]. However, here in the L-CNV model, ocular VEGFA expression was not altered by         T-test (two tailed) with Welch's correction, Mean ± SEM, n=7 eyes. [23].