Anti-sphingosine-1-phosphate monoclonal antibodies inhibit angiogenesis and sub-retinal fibrosis in a murine model of laser-induced choroidal neovascularization

Abstract

The efficacy of novel monoclonal antibodies that neutralize the pro-angiogenic mediator, sphingosine-1-phosphate (S1P), were tested using in vitro and in vivo angiogenesis models, including choroidal neovascularization (CNV) induced by laser disruption of Bruch's membrane. S1P receptor levels in human brain choroid plexus endothelial cells (CPEC), human lung microvascular endothelial cells, human retinal vascular endothelial cells, and circulating endothelial progenitor cells were examined by semi-quantitative PCR. The ability of murine or humanized anti-S1P monoclonal antibodies (mAbs) to inhibit S1P-mediated microvessel tube formation by CPEC on Matrigel was evaluated and capillary density in subcutaneous growth factor-loaded Matrigel plugs was determined following anti-S1P treatment. S1P promoted in vitro capillary tube formation in CPEC consistent with the presence of cognate S1P(1-5) receptor expression by these cells and the S1P antibody induced a dose-dependent reduction in microvessel tube formation. In a murine model of laser-induced rupture of Bruch's membrane, S1P was detected in posterior cups of mice receiving laser injury, but not in uninjured controls. Intravitreous injection of anti-S1P mAbs dramatically inhibited CNV formation and sub-retinal collagen deposition in all treatment groups (p<0.05 compared to controls), thereby identifying S1P as a previously unrecognized mediator of angiogenesis and subretinal fibrosis in this model. These findings suggest that neutralizing S1P with anti-S1P mAbs may be a novel method of treating patients with exudative age-related macular degeneration by reducing angiogenesis and sub-retinal fibrosis, which are responsible for visual acuity loss in this disease.

Figures

Fig. 1

CPEC express receptors for S1P. Quantitative real-time PCR of the cognate S1P receptors expressed by diverse human vascular endothelial cells. RT-PCR data are plotted showing relative S1PR1–5 expression in HREC, CPEC and LMEC. Human HREC and CPEC express significantly higher levels of S1P2 and S1P3 mRNA by comparison to S1P1, with S1P3 being by far the most abundant of all the S1P receptors expressed by CPEC. Both LMEC and CD34+ EPC express S1P1 most abundantly. All samples were normalized to b-actin and studies were performed in triplicate. The data represent mean values ± SEM that are relative to S1P1 receptor expression.

Fig. 2

LT1009 inhibits tube formation in vitro. CPEC, either unstimulated or pre-incubated with S1P with or without LT1009, were plated on Matrigel-coated wells and allowed to incubate for 12 h. S1P alone (1 nM or 10 nM, A and B, respectively) stimulated tube formation, compared to unstimulated cells (F). Co-incubation of 10 nM S1P with 25 μg/mL LT1009 (C) greatly reduced microvessel tube formation by these cells, while 50 μg/mL LT1009 (D) completely abrogated tube formation. (E) A positive control (CPEC in the presence 50 ng/mL VEGF and 50 ng/mL FGF-2). All phase contrast micrographs are typical of at least three wells per condition, and were taken at 10× original magnification.

Fig. 3

LT1009 shows efficacy in the in vivo Matrigel neovascularization assay by reducing microvascular vessel formation induced by the pro-angiogenic factors VEGF and FGF-2. Matrigel plugs loaded with 50 ng/mL VEGF and 50 ng/mL FGF-2 were implanted in female C57BL/6 mice. Mice received drug treatments consisting of equal volumes of 0.33–81 mg/ kg LT1009 or saline beginning 1 day prior to the implantation of Matrigel plugs; each treatment was then administered intraperitoneally twice daily for 14 days. At the end of the experiment, the plugs were collected, sectioned and stained for CD31. (A) Representative image of sections from controls. (B) Representative image of sections from mice harboring VEGF/FGF-2 loaded plugs; black arrows indicate representative vessels. (C) Representative image of sections from mice harboring VEGF/FGF-2 loaded plugs and treated with LT1009; note the reduced number of vessels (black arrows). (D) Quantification of the reduction of microvascular density (MVD) in the plugs (n ≥ 6 plugs per treatment). Statistical analysis was performed using 1-way ANOVA (***p < 0.0001) followed by Bonferroni’s post test (*p < 0.01).

Fig. 4

Immunohistochemical localization of S1P in the murine retina. (A) Composite image of red (B) and green (C) channels of a posterior cup from an uninjured (negative control) eye reacted with rhodamine-conjugated agglutinin, S1P antibody and FITC-conjugated secondary antibody. (D) Composite image of red (E) and green (F) channels of a posterior cup from an eye that underwent laser rupture of Bruch’s membrane, and reacted as described above for the eye in panels A, B, C. Note the lack of S1P immunoreactivity in the untreated control eye and widespread S1P immunoreactivity in the laser-treated eye in association with the CNV lesion. Scale bar = 50 μm.

Fig. 5

Time-dependent efficacy of the anti-S1P mAb LT1002 in reducing CNV lesions in vivo. CNV lesion volume was measured on days 7, 14 and 28 after laser rupture of Bruch’s membrane (3 burns/eye, 1 eye per animal). The areas for each burn were converted to a volume and the volumes averaged to produce a single CNV lesion volume for each animal. Representative fluorescent micrographs of flat-mounted posterior eye cups from injured eyes stained for with R. communis agglutinin are shown. (A), (C) and (E) depict eyes that received non-specific antibody 7, 14 and 28 days post injury, respectively. (B), (D) and (F) depict eyes treated with intravitreal injection of 0.5 μg per eye of LT1002 7, 14, and 28 days post injury, respectively. (G) CNV lesion size at all time points was quantified and is shown in graph form. * Denotes p < 0.05 compared to non-specific antibody at days 7, 14 and 28. Scale bar in (A) = 50 μm and is applicable to all of the micrographs. n = 3 animals per treatment condition at each time point.

Fig. 6

A comparison of LT1002 and LT1009 efficacy on reducing experimental CNV lesion volume in mouse eyes injured by laser rupture of Bruch’s membrane. All animals received intravitreal injection of either non-specific antibody (NSA), injection of 0.5 μg per eye of LT1002 or LT1009 at the time of injury and weekly thereafter for the duration of the experiment. At this time point (28 days post injury) weekly injections of LT1009 were three-fold more effective than weekly injections of LT1002. n = 15 for non-specific antibody, n = 16 for LT1002, and n = 14 for LT1009. * Denotes p < 0.05 compared to non-specific antibody. ** Denotes p < 0.05 compared to LT1002.

Fig. 7

Use of LT1002 resulted in a reduction of scarring in the CNV wound bed. Cross sections stained with Masson’s Trichrome. Lesions are shown within the white box. Histological images show collagen deposition (blue staining) within and posterior to the vascular lesion after laser injury in eyes treated with 0.5 μg per eye of non-specific antibody on day 7 (A), and compared to LT1002-treated (0.5 μg) eyes on days 7 (B), 14 (C) and 28 (D). (E) A naive (untreated control) eye for comparison. OS, outer segments; ONL, outer nuclear layer; INL, inner nuclear layer; ILM, internal limiting membrane; sc, sclera; ch, choroid; he, hemorrhage; *Representative locations where collagen deposition was measured. (F) A summary plot of the collagen deposition measurements in all eyes examined (n = 3 eyes per treatment condition at each time point). * Denotes p < 0.05 versus treatment with nonspecific antibody.