The Role of LOX and LOXL2 in the Pathogenesis of an Experimental Model of Choroidal Neovascularization

Abstract

Purpose: We investigated whether lysyl oxidase (LOX) and lysyl oxidase-like 2 (LOXL2) play a role in an experimental model of choroidal neovascularization (CNV). The therapeutic potential of antibodies against LOX (M64) and LOXL2 (AB0023) was evaluated in a murine laser-induced CNV model.

Methods: Expression of LOX and LOXL2 in the posterior eye cups (including retina, retinal pigment epithelium, choroid, and sclera) was studied by qRT-PCR and immunohistochemistry. In the murine model of CNV, both antibodies were administered intraperitoneally every other day until the day killed. On different time points after laser, treatment outcome was studied by immunohistochemical analysis of inflammation, angiogenesis and fibrosis, and by transcript analysis of different cytokines.

Results: Levels of LOX and LOXL2 in the posterior eye cups were increased after CNV-induction at different time points after laser. At day 35, their protein expression patterns appeared to correlate with retinal glial cells and endothelial cells, respectively. Both antibodies significantly inhibited fibrosis, whereas AB0023 also significantly reduced angiogenesis and inflammation. Transcript levels of α-1 type I collagen (COL1A1) in the posterior eye cups were significantly decreased in lasered mice treated with either M64 or AB0023. Vascular endothelial growth factor expression was also reduced only after AB0023 treatment, whereas activated fibroblast marker α-smooth muscle actin (αSMA) levels were not significantly changed.

Conclusions: This study suggests that LOX and LOXL2 may play an important role in the pathogenesis of AMD. Targeting LOXL2 could have a broader efficacy than targeting LOX, by reducing angiogenesis and inflammation, as well as fibrosis.

The most common blinding disease of the posterior segment of the eye, caused by abnormalities in the choroidal and retinal vasculature, is AMD. This degenerative disease is not only characterized by choroidal neovascularization (CNV), but also by macular edema, retinal, vitreous hemorrhages and fibrovascular scarring.¹ Indeed, the primary pathophysiological process that causes blindness in AMD is the retinal response to injury, that is, chronic wound healing leading to subretinal fibrosis.¹ This process of fibrosis can lead to destruction of the surrounding tissues, such as photoreceptors and retinal pigment epithelium, which can be associated with permanent vision loss. Importantly, it is described that 10% to 15% of the patients with AMD will lose central vision as a direct effect of choroidal neovascularization and fibrosis.² At present, anti-VEGF agents such as bevacizumab and ranibizumab are the gold standard in the treatment of CNV. However, due to its essential role in blood vessel formation and maintenance and in neuronal survival, VEGF inhibiting has caused a number of severe local and systemic adverse events.^3–5 Moreover, inhibiting angiogenic molecules does not address the underlying pathophysiological processes of inflammation and fibrosis. Therefore, a need remains for alternative strategies to treat AMD. 

Lysyl oxidase (LOX) and lysyl oxidase-like 2 (LOXL2) are the two most studied members of the lysyl oxidase (LOX(L)) family. Lysyl oxidases are amino oxidase enzymes that can, via generation of aldehydes on lysine residues, catalyze the crosslinking of fibrils of collagen and elastin in the extra-cellular matrix (ECM). The expression of these LOX(L) family members is highly controlled during normal development,⁶ but they have also been described as critical contributors to the development of a variety of fibrosis-related diseases, such as liver and lung fibrosis, as well in many cancers.^7,8 It is indeed known that LOXL2 is overexpressed in the environment of solid tumors and that LOXL2-mediated crosslinking of collagen may stimulate tumor growth.^8,9 Moreover, targeting LOXL2 with an inhibitory monoclonal antibody (AB0023)¹⁰ has been shown to be efficacious in rodent models of cancer, as well as in different fibrosis models.⁸ Inhibition of LOXL2 has resulted in a marked reduction in activated fibroblasts and endothelial cells, decreased production of growth factors and cytokines, and decreased TGF-β pathway signaling.⁸ Lysyl oxidase-like 2 has also been shown to be expressed in disease-associated endothelial cells and to participate in angiogenic processes.^8,11,12 

Although a majority of blinding ocular diseases are associated with a disruption of the tissue architecture in the eye, caused by vascular leakage and fibrosis,² little information is available about the involvement of LOX and LOXL2 in eye diseases.^13,14 Very recently, our research group showed that targeting LOXL2 with an inhibitory monoclonal antibody reduced pathological angiogenesis, inflammation, and fibrosis in a rabbit model of glaucoma surgery.¹⁵ The role of LOX(L) in the process of wound healing in AMD, however, is still unknown. Therefore, the first aim of this study was to characterize and localize the expression pattern of LOX and LOXL2 in pathologic wound healing after CNV-related AMD. Next, the therapeutic antiangiogenic, anti-inflammatory, and antifibrotic potential of anti-LOX (M64) and anti-LOXL2 (AB0023) antibodies was investigated in a murine model of CNV. 

Materials and Methods

All animal procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The Institutional Animal Care and Research Advisory Committee of the KU Leuven approved all experimental animal procedures. 

CNV Laser Model

We anesthetized C57BL/6J mice (8–10 weeks old, Charles River Laboratories, Lyon, France) with 10 times-diluted (60 mg/kg final dose) sodium pentobarbital (Nembutal, 60 mg/mL; Ceva Sante Animale, Belgium) and the pupils were dilated with tropicamide (Tropicol, Thea Pharma, Schaffhaussen, Switzerland). Three laser burns were placed with a 532-nm green laser at 9, 12, and 3 o'clock positions around the optic disk using a slit lamp delivery system with a handheld cover slide as a contact lens and using lubricating eye drops (GenTeal Gel; Novartis Pharma nv-sa, Vilvoorde, Belgium). Each spot was placed with a spot size of 50 μm, laser duration of 100 ms, and a power of 400 mW. Bubble production, indicating rupture of the Bruch's membrane, was necessary for inclusion of the spot. In a first experiment, 10 mice per time point were used to evaluate LOX and LOXL2 expression on days 2, 4, 7, 14, 28, and 35 after laser photocoagulation by qRT-PCR and immunohistochemistry. One eye was lasered and the contralateral nonlasered eye served as control. In a second experiment, treatment with the anti-LOX antibody (M64; n = 30 mice) or anti-LOXL2 antibody (AB0023; n = 30 mice) was initiated on day 0 just after laser photocoagulation both eyes and was given intraperitoneally (IP) every other day at 0.75 mg/dose (200 μL) until the day killed. Origin and characterization of both antibodies is described in detail in literature.^8,10 Control lasered animals (n = 30) were treated with the vehicle (PBS + 0.01% Tween 20) given IP according to the same treatment scheme. Six nonlasered mice were used as the naïve controls and were not treated. Mice were checked for pain-distress every other day and body weight was measured at day 0 before laser and at day killed. 

Quantitative Real-Time RT-PCR

Cytokine levels in posterior eye cups (including retina, RPE, choroid, and sclera) were analyzed by quantitative RT-PCR (qRT-PCR). The tissues were dissected immediately after enucleation of the eyes, freshly frozen in liquid nitrogen and stored at −80°C until analysis. We isolated RNA from these samples using the RNeasy Mini kit (Qiagen, Valencia, CA, USA) at different time points after laser photocoagulation. Expression was normalized to that of the housekeeping gene 60S ribosomal protein L19 (RPL19). The following forward (for) and reverse (rev) primers and probes (pro) labeled with a fluorescent dye (FAM) and quencher (BHQ-1) were used: Murine LOX: for 5′-CAAGAGGGAAGCAGAGCCTTC-3′; rev 5′-GCACCTTCTGAATGTAAGAGTCTC-3′; pro 5′-ACCAAGGAGCACGCACCACAACGA-3′. Murine LOXL2: for 5′-GCTATGTAGAGGCCAAGTCCTG-3′; rev 5′-CAGTGACACCCCAGCCATTG-3′; pro 5′-TCCTCCTACGGTCCAGGCGAAGGC-3′. Murine VEGF: for 5′-TGCACCCACGACAGAAGG-A-3′; rev 5′-GGCAGTAGCTTCGCTGGTAGAC-3′; pro 5′-CAGAAGTCCCATGAAGTGATCAAGTTCATGGA-3′. Murine α-1 type I collagen (COL1A1): for 5′-CGGCTGTGTGCGATGACG-3′; rev 5′-ACGTATTCTTCCGGGCAGAAAG-3′; pro 5′-CAGCACTCGCCCTCCCGTCTTTGG-3′. Murine α-smooth muscle actin (αSMA): for 5′-TCTGCCTCTAGCACACAACTG-3′; rev 5′-AAACCACGAGTAACAAATCAAAGC-3′; pro 5′-TGTGGATCAGCGCCTCCAGTTCCT-3′. Murine RPL19: for 5′-AGAAGGTGACCTGGATGAGAA-3′; rev 5′-TGATACATATGGCGGTCAATCT-3′; pro 5′-CTTCTCAGGAGATACCGGGAATCCAAG-3′. 

Immunohistochemistry and Immunofluorescent Stainings

Mice were killed via cervical dislocation on day 5 (n = 10/group to evaluate inflammation); day 14 (n = 10/group to evaluate angiogenesis); or day 35 (n = 10/group to evaluate inflammation, neovascularization and fibrosis) after laser treatment. Eyes were enucleated and the tissues were fixed overnight in 1% paraformaldehyde (PFA; Merck, Darmstadt, Germany), dehydrated, and embedded in paraffin. Serial sections (7 μm) were cut and subjected to different immunohistochemical stainings. 

To correlate the LOX expression pattern with retinal glial cells in the diseased retina, immunofluorescent staining was performed using a monoclonal rabbit anti-mouse glial fibrillary acidic protein (GFAP, 1/100; CP040B; BioCare Medical, Concord, PA, USA) antibody and the polyclonal rabbit anti-mouse LOX (5 μg/mL) antibody. The immunofluorescent stainings were performed on fresh frozen 5-μm sections post-fixed with 4% PFA, treated with hydrogen peroxide (Peroxidazed; BioCare Medical) for 15 minutes, blocked with a blocking reagent (Background Sniper; BioCare Medical) for 20 minutes, incubated with primary antibody (Da Vinci Green Diluent; BioCare Medical) for 60 minutes, incubated with anti-rabbit AlexaFluor 488 (Invitrogen, Life Technologies, Grand Island, NY, USA) for 60 minutes, and mounted in medium with DAPI (VectaMount; Vector Laboratories, Burlingame, CA, USA). Also a double immunofluorescent staining was performed at 35 days after laser, to correlate the LOX expression pattern with retinal glial cells, using a polyclonal goat anti-mouse glial fibrillary acidic protein (GFAP, 1/100; ab53554; Abcam, Cambridge, UK) antibody and the polyclonal rabbit anti-mouse LOX (5 μg/mL,) antibody. Slides were incubated with anti-rabbit AlexaFluor 488 and anti-goat AlexaFluor 568 (Invitrogen, Life Technologies) for 60 minutes, incubated with DAPI for 10 minutes, and mounted in mounting media (ProLong Gold; Invitrogen, Life Technologies). 

To identify localization of LOXL2 in the retina and to correlate its expression with blood vessels, polyclonal rabbit anti-mouse LOX(L2) and rat anti-mouse CD31 immunohistochemistry was performed on formalin-fixed paraffin-embedded (FFPE) sections using BioCare Medical (Concord, CA, USA) reagents. Slides were deparaffinized and antigen retrieval was performed with a commercial instrument (Decloaking Chamber; Biocare Medical), using buffer solution and reagents (Universal Decloaker and Hot Rinse; Biocare Medical). The slides were then treated with Peroxidazed for 15 minutes and blocked with Background Sniper for 20 minutes. The slides were incubated with primary antibodies (in-house rabbit polyclonal antibodies recognizing LOX (5 μg/mL) and LOXL2 (12.5 μg/mL) and rat anti-mouse CD31 (1/100; Abcam, Cambridge, UK) in Da Vinci Green Diluent for 30 minutes. The Mach 2 polymer kit was used for antigen detection by adding anti-rabbit secondary antibody (conjugated to horseradish peroxidase) for 30 minutes. DAB (3, 3′ diaminobenzidine) chromogen was added to the slides for 5 minutes, followed by a single rinse in distilled water. The slides were then counterstained with CAT hematoxylin, manually dehydrated with graded alcohol and mounted with Entellan mounting media. 

To investigate the effect of LOX- and LOXL2-inhibitors on the different processes of wound healing after laser, different (immuno)histological stainings for inflammation (days 5 and 35) and fibrosis (day 35) and fluorescent perfusion of the blood vessels (days 14 and 35) were executed. To analyze the effect on inflammatory cell infiltration in the CNV lesions, the retina was removed from the dissected posterior segments. These posterior eye cups, which included RPE, choroid, and the sclera, were stored in PBS. To stain all leukocytes, a rat anti-mouse CD45 antibody (1/100; Pharmingen, Erembodegem, Belgium) was used overnight, diluted in Tris-buffered saline (TBS)-Triton 0.3%. The following day, the posterior eye cups were incubated for 2 hours with rabbit anti-rat biotin labeled antibody (1/300; DakoCytomation A/S, Copenhagen, Denmark), diluted in TBS-Triton 0.3%. Antibody binding was visualized by fluorescent staining using streptavidin-AlexaFluor 568 (1/200; Molecular Probes, Life Technologies, Eugene, OR, USA) in TBS-Triton 0.3% for 2 hours. Choroids were placed on mounting media (Invitrogen, Life Technologies) with DAPI (Molecular Probes, Life Technologies). Angiogenesis was investigated using retrobulbar perfusion with 200 μL of fluorescein isothiocyanate (FITC)-conjugated dextran (50 mg/mL, Mr 2 × 10⁶ Da; Sigma-Aldrich, Diegem, Belgium) for 2 minutes. The next day, the RPE-choroid-sclera complexes were dissected and flatmounted on a slide containing a drop of mounting medium (Vectashield; Vector Laboratories) to prevent the fluorescent dyes from bleaching. 

Deposition of collagen was analyzed by Sirius red staining. Serial sections were cut at 7 μm thickness in five series on five glass slides, resulting in a set of five almost identical slides. First, hematoxylin and eosin (H&E) staining was performed to localize the laser spots. The consecutive slides were used for Sirius red staining. Slides were deparaffinized, washed, and placed in Sirius red solution (direct red 80 and 1.3% picric acid solution; Sigma Aldrich) for 60 minutes. Sections were then placed in 0.01 N HCl (Prolabo, Leuven, Belgium) for 2 minutes, dehydrated and mounted with DPX mounting medium (Prosan, Merelbeke, Belgium). 

Chromogenic In Situ Hybridization

To determine LOXL2 mRNA expression in the diseased retina at postoperative day 35, red chromogenic in situ hybridization (CISH) was performed using the standard high definition red assay protocol (RNAscope 2.0; Advanced Cell Diagnostics, Hayward, CA, USA), utilizing the RNA target Mus musculus LOXL2 probe (VS probe–Mm-LOXL2; Advanced Cell Diagnostics) as well as the appropriate positive and negative control probes (VS Probe–Mm-Ppib and negative control VS Probe-DapB, respectively; Advanced Cell Diagnostics). The protocol was followed on serial FFPE sections (5.0 μm), according to the manufacturer's recommendations (Advanced Cell Diagnostics). The only tissue-specific protocol step included a 10-minute incubation time in the boiling Pretreatment 2 reagent (Advanced Cell Diagnostics). 

Imaging and Analysis

Images were obtained using a microscope (Leica Microsystems, Wetzlar, Germany) with a digital camera (Axiocam MrC5; Carl Zeiss, Oberkochen, Germany) at a magnification of ×20 and a resolution of 2584 × 1936 pixels. Morphometric analyses were performed using commercial software (Axiovision; Carl Zeiss). To correct for spot size variability, the density of leukocytes and blood vessels was quantified by calculating respectively the CD45-positive (days 5 and 35) and the FITC-dextran-positive area (days 14 and 35) as a proportion to the total CNV lesion area in the samples. The total neovascular area (the surface of the lesions) on FITC-perfused choroids was measured as an indication of the neovascular membranes and the aggressiveness of the CNV spot. Fibrosis was evaluated using Sirius red staining on day 35 and was determined by measuring the percentage of the area of mature collagen fibers compared with the total lesion area. Polarized light was used to distinguish mature (appearing yellow and orange) from immature (green) collagen fibers. For each lesion, the middle on the H&E section was first defined and analysis was performed on five serial sections in the middle of each spot. The area of these five sections was averaged to have 1 value per laser spot, leading to three values per eye. 

Statistical Analysis

All histological data were analyzed using the Student's t-test for independent samples. Data at individual time points were analyzed using mixed model analysis for repeated measures (using GraphPad Prism 5.02). Kaplan-Meier survival analysis was performed for bleb failure using the log-rank test. Values of P ≤ 0.05 were considered to be statistically significant. Data are represented as mean ± SEM. 

Results

Ocular Expression of LOX and LOXL2

Expression of LOX and LOXL2 at the mRNA level in the mouse eye after CNV induction was evaluated by qRT-PCR at multiple time points and by immunohistochemical stainings at day 35. Quantitative RT-PCR analyses showed that the levels of LOX in the posterior eye cups (including retina, RPE, choroid, and sclera) were significantly upregulated in the lasered eyes as compared to the control eyes after lasering at day 4 (2.94-fold; P = 0.001); day 7 (1.86-fold; P = 0.005); and day 28 (1.42-fold; P = 0.03; n = 10; Fig. 1A). Analysis showed that LOX protein was significantly upregulated in the lasered eyes compared with the control eyes over the period of 35 days after laser (overall P = 0.02). The immunohistochemical staining for LOX at day 35 confirmed these results and showed that LOX expression was increased in the region of laser injury, in the disrupted inner and outer plexiform and nuclear layers, the choroid, and the sclera (Figs. 1B–E). Of note, in nonlasered eyes, baseline expression of LOX was seen in the inner and outer plexiform layers of the retina (Fig. 1C). Moreover, the LOX immunofluorescent expression pattern appeared to colocalize to a subpopulation of glial cells, likely Müller cells, identified by the GFAP glial cell marker. Expression of LOX and GFAP was in the inner (IPL) and outer plexiform layers (OPL) of the retina and in the disrupted choroid and sclera of the region of the laser-induced injury. Both LOX and GFAP-expressing cells were seen extending into the outer nuclear layer (ONL) and exterior the ONL and in the laser-induced spot (Figs. 1F–I, Supplementary Fig. S1A). Double immunofluorescence for LOX and GFAP confirmed coexpression of these proteins in the lesion (Supplementary Fig. S1B). Interestingly, a subset of cells in the fibrotic area of the laser-spot was positively stained for LOX, but not for GFAP (Supplementary Fig. S1B). 

Figure 1

 

Expression of LOX in the region of the laser-induced injury. (A) Expression of LOX at mRNA level in posterior eye cups (including retina, RPE, choroid, and sclera) was determined by quantitative RT-PCR. Levels of LOX were upregulated compared with nonlasered eyes at days 4, 7, and 28 after CNV induction in mice (10 mice/time point). The level of mRNA was normalized to that of the housekeeping gene 60S ribosomal protein L19 (RPL19). *P < 0.05. (B–E) Images show representative pictures of H&E and LOX immunohistochemistry (IHC) of the noninjured (B, C) and injured retina at day 35 after lasering (D, E), respectively. In nonlasered eyes, baseline expression of LOX was seen in the inner and outer plexiform layers of the retina (C). In the lasered eyes, LOX was also upregulated in these layers and in the disrupted choroid and sclera of the region of the laser-induced injury (D). (F–H) The LOX immunofluorescent expression pattern (G–I) appeared to colocalize to a subpopulation of glial cells identified by the GFAP glial cell marker (F–H). Expression of LOX and GFAP in the IPL and OPL of the retina within (indicated by filled arrows) and outside the lesion (indicated by empty arrows). Also LOX-positive cells were seen in the disrupted choroid and sclera (indicated by asterisk) of the region of the laser-induced injury. Both LOX and GFAP-expressing cells were seen extending into the ONL and exterior the ONL and in the laser-induced spot (indicated by arrowheads). CH, choroid; RGC, retinal ganglion cell layer; R&C, rod and cones; SCL, sclera.

Quantitative RT-PCR analysis also showed that the expression of LOXL2 was significantly increased compared with control samples at day 4 after laser (1.68-fold; P = 0.03; n = 10; Fig. 2A). The overall expression over 35 days of LOXL2 protein was not significantly different compared with control eyes (overall, P = 0.44). Chromogenic in situ hybridization confirmed expression of LOXL2 mRNA at day 35 in a subset of cells in the inner and outer nuclear and ganglion cell layers, including in the area of laser-induced injury (Supplementary Fig. S2). Moreover, immunostaining for LOXL2 and CD31 in the nondiseased area of the retina demonstrated that LOXL2 was expressed in retinal endothelial cells, strongly suggesting that LOXL2 might be involved in the retinal vascular biology (Figs. 2B–E). 

Figure 2

 

Expression of LOXL2 in region of the laser-induced injury. (A) Analysis of qRT-PCR showed that LOXL2 at mRNA level was upregulated in posterior eye cups (including retina, RPE, choroid, and sclera) at day 4 after lasering (10 mice/time point). The mRNA level was normalized to that of the housekeeping gene 60S ribosomal protein L19 (RPL19; *P < 0.05). (B–E) Representative pictures of CD31 and LOXL2 IHC at day 35 post injury showed clear expression of LOXL2 in the retinal endothelial cells.

Thus, LOX and LOXL2 were upregulated at different time points in the posterior eye cups after CNV-induction. As such, these two proteins are implicated in the pathogenesis of this disease. Importantly, since LOX and LOXL2 mRNA expression showed a peak at day 4 after CNV induction, they could play an important role in the early stages of postoperative wound healing in the pathogenesis of AMD. 

Effect of M64 and AB0023 in a Murine Model of CNV

To determine the therapeutic potential of LOX and LOXL2 inhibition in a murine model of CNV, mice were treated with repeated injections of anti-LOX antibody (M64; 0.75 mg IP every other day), anti-LOXL2 antibody (AB0023; 0.75 mg IP every other day) or PBS-Tween (vehicle). All animals were clinically examined every other day and neovascularization, inflammation, and collagen deposition were analyzed at study termination. Daily clinical examination revealed that none of the animals showed severe pain-distress. One mouse in the anti-LOX group died unexpectedly at 16 days post-lasering. No treatment-related differences in pre- and post-treatment body weights at day 35 were detected (data not shown). 

Analysis of CD45 stained (days 5 and 35) and FITC perfused (days 14 and 35) choroids showed a significant reduction in inflammation and neovascularization and in the eyes of the group treated with the anti-LOXL2 antibody (AB0023; Fig. 3). Inflammation was reduced by 34% ± 6% compared with control (day 5; n = 10; P = 0.01; Figs. 3A, 3B), while blood vessel density was reduced by 47% ± 11%, compared with vehicle-treated mice (day 14; n = 10; P = 0.003; Figs. 3C, 3D). In the anti-LOX treated mice, neovascularization and inflammation were comparable with the control (PBS-T) group (n = 10; P = 0.10 and P = 0.43, respectively; Figs. 3A, 3C). The same inhibitory effect of the antibodies was seen at day 35 after laser (Figs. 3A, 3C). Besides the blood vessel density (neovascular area/total spot area), the total neovascular area of each spot was also measured at day 14, as an indication of the neovascular membranes and the aggressiveness of the CNV spot. The values of three spots per eye were averaged and analysis showed that anti-LOXL2 was able to significantly reduce the total neovascular area with 18% compared with anti-LOX and PBS-treated mice (P = 0.02; Fig. 3E). Both antibodies also significantly reduced fibrosis at day 35 after laser. Sirius red staining revealed that collagen deposition was significantly decreased by 42% ± 7% in the anti-LOX treated group (n = 9; P = 0.00005, Figs. 4A, 4B) and by 27% ± 4% in the anti-LOXL2 group (n = 10; P = 0.006; Figs. 4A, 4B) compared with the vehicle-treated group. Of note, the nonlasered group of mice showed a normal blood vessel density and sporadically some glial cells in the choroid. No mature collagen fibers were detectable in naive choroid (data not shown). 

Figure 3

 

Inflammation and neovascularization in lasered mice eyes after treatment. (A–C) Analysis of CD45-stained and FITC-perfused posterior eye segments (including RPE, choroid, and sclera) showed significant reduction in inflammation (days 5 and 35) and neovascularization (days 14 and 35) in the group treated with anti-LOXL2 antibody compared with nontreated eyes. No significant differences relative to control eyes were seen after anti-LOX antibody (M64) treatment. (B–D) The images show representative pictures of CD45 stained or FITC-perfused laser spots of eyes treated with the anti-LOX or anti-LOXL2 antibody or with vehicle (PBS-T) at day 5 and day 14 after laser, respectively (Scale bar: 50 μm). Edges of the laser spots are marked by a dotted line. (E) Total neovascular area after anti-LOX, anti-LOXL2 and PBS-Tween treatment. At day 14, anti-LOXL2 treatment was able to significantly reduce the total neovascular area (CNV aggressiveness) with 18% compared with anti-LOX and PBS-treated mice (P = 0.02).

Figure 4

 

(A) Collagen deposition in lasered mice eyes after treatment. Collagen deposition at 35 days post injury was significantly reduced in the anti-LOX and anti-LOXL2 antibody-treated groups (by 42% and 27%, respectively) compared with the vehicle-treated group. *P < 0.05. (B) The images show representative pictures of Sirius red (left) and H&E (right) of eyes treated with the anti-LOX or anti-LOXL2 antibody or with vehicle (PBS-T). Scale bar: 50 μm. Edges of the laser spots are marked by a dotted line.

To add some mechanistic support to our (immuno)histological stainings and fluorescent perfusion of the blood vessels, the expression of different growth factors and cytokines in the posterior eye cups (including retina, RPE, choroid, and sclera) was also examined by quantitative RT-PCR at day 35 after laser. Transcript levels of VEGF, as an angiogenic marker, were first investigated. Analysis revealed a significant reduction of 18% of VEGF mRNA in lasered eyes from mice treated with AB0023 (anti-LOXL2) compared with eyes from vehicle or M64 (anti-LOX) treated mice (P = 0.003; Fig. 5A). Alpha-1 type I collagen (COL1A1) was selected as a marker for fibrosis and transcript analysis showed a 17% reduction of COL1A1 mRNA in mice treated with AB0023 (P = 0.01), whereas a reduction of 25% was observed in M64-treated eyes (P = 0.01), compared with vehicle-treated eyes (Fig. 5B). However, the differences were not significantly different, transcript analysis of activated fibroblast marker αSMA revealed a 24% reduction (down to levels observed in naïve eyes) in lasered eyes from mice treated with either M64 or AB0023 compared with eyes from vehicle-treated mice (Fig. 5C). Of note, the reduction of levels of αSMA might be of great importance considering that only a subset of fibroblasts exhibits an activated phenotype in disease and that other cells types that are not directly implicated in disease also express αSMA (e.g., pericytes). 

Figure 5

 

Transcript analyses of different cytokines in lasered mice after treatment. Transcript analysis of activated fibroblast marker (A) VEGF, (B) COL1A1, and (C) αSMA levels post injury revealed reduced expression at day 35 after laser in the posterior eye cups (including retina, RPE, choroid, and sclera) of lasered eyes from mice treated with either anti-LOX (M64) or anti-LOXL2 (AB0023) antibody, compared with eyes from vehicle-treated mice.

Thus, anti-LOX reduced collagen deposition at 35 days after laser, whereas anti-LOXL2 significantly reduced not only collagen deposition, but also blood vessel density at day 14 and inflammation at day 5 after laser. 

Discussion

Age-related macular degeneration is amongst the most frequent causes of irreversible vision loss due to subretinal inflammation and fibrosis, and its prevalence is increasing.¹ This disease is indeed becoming an important socio-economic problem, as the proportion of the aging population is continuously increasing. Anti-VEGF therapy seems to be effective to a certain level, but is mostly anti-angiogenesis driven,¹⁶ whereas inflammation and fibrosis are also important players in the pathogenesis of AMD.² Developing targeted therapies, that attack the disease process on multiple fronts, might help to fight the expected exponential rise in number of legally blind patients. Since the LOX/L family plays an important role in the fibrotic process by cross-linking collagen and elastin,^13,17 we hypothesized that these enzymes might be good targets for anti-fibrotic strategies in AMD. To investigate the role of LOX and LOXL2 in the wound healing process and to explore the therapeutic potential of blocking antibodies, the laser-induced CNV mouse model for AMD was used.¹⁸ This laser model is the most widespread used, albeit acute, model for CNV induction.^19–21 First, a time-course study of fibrosis was performed by staining collagen fibers, since this method is well described in literature in the CNV model for exploring fibrosis.^22,23 Sirius red was selected since this staining has the advantage of distinguishing mature collagen fibers from immature fibers under polarized light. Analysis on different time points after lasering showed a peak of fibrosis on day 35 (Supplementary Fig. S3). Therefore, this time point was selected to investigate the antifibrotic properties of both antibodies in this study. 

Levels of LOX and LOXL2 in posterior eye cups (including retina, RPE, choroid, and sclera) were first determined by qRT-PCR and analyses showed a peak upregulation at day 4 after laser for both proteins and levels trailed off from there. Unfortunately, no significant differences were seen in LOX(L2) expression in treated eyes 35 days after laser. Since the half-life of both proteins in the ECM has not been elucidated, transcript levels may not necessarily be indicative for protein levels. Indeed, immunohistochemical analysis of LOX demonstrated that the protein was expressed in different layers of the retina. Moreover, immunofluorescent analysis of LOX and GFAP clearly showed that the expression pattern appeared to correlate with known phenotypes of Müller-type retinal glial cells, and that these proteins were coexpressed in a subset of cells in the lesion. Indeed, it has already been described that different members of the LOX-family are expressed by Müller cells.^24,25 Interestingly, LOX and GFAP were also coexpressed in what looks to be a fibrotic scleral region of the laser-induced injury. From the literature, it is known that Müller retinal glial cells play an important role in CNV-formation and that these cells might migrate outside of their normal domain into the lesion as a response to the injured retina.^26–28 It is indeed described that Müller cells can be activated (and induce GFAP expression) at the site of injury starting to extend their distal processes into the lesion and invade in the fibrovascular complex of the laser spot.²⁸ Moreover, Müller cells can transdifferentiate into myofibroblasts in the injured retina and migrate onto the surfaces of the detached retina and proliferate there.²⁷ This can explain the presence of GFAP-negative and fibroblast-like LOX-positive cells at day 35 after laser in the fibrotic area of the laser-induced injury, although we cannot exclude that these cells are fibroblasts of different origin. Expression of LOXL2 in lesions was difficult to detect by immunohistochemistry and we were not able to assess LOXL2 induction at sites of injury, but CISH analysis provided evidence that LOXL2 was being expressed and could be involved in disease progression in these areas. 

Recent studies showed that polymorphisms of LOXL1 are associated with a significantly increased risk of AMD,²⁹ suggesting a potential role for the LOX family in the pathogenesis of this disease. Moreover, a previous study showed that 100% inhibition of LOXL1 (in LOXL1-deficient mice) led to more aggressive CNV growth. A systemic defect in elastic fiber deposition affected the integrity of the Bruch membrane by accumulation of soluble elastin peptides in these transgenic mice.³⁰ In our study, immunohistochemical analyses revealed that administration of LOX and LOXL2 antibodies reduced CNV severity. Both anti-LOX and anti-LOXL2 treatment significantly decreased collagen deposition at 35 days after laser as shown by Sirius red stainings, which was associated with a reduction in COL1A1 mRNA levels in posterior eye cups. These data suggest that LOX and LOXL2 participate in the process of retinal fibrosis. In addition to being secreted by disease-associated Müller-type glial cells, LOX may also be expressed in differentiated myofibroblasts in the laser-spot, since it has already been described that myofibroblasts express LOX.³¹ 

Importantly, besides collagen deposition, anti-LOXL2 therapy also significantly reduced inflammation at day 5, blood vessel density and total neovascular area at day 14. These results indicate that AB0023 treatment might not only able to reduce the different processes after lasering, but also reduces the aggressiveness of the laser spots. This may be relevant for clinical practice, since this suggests that anti-LOXL2 treatment might also be able to reduce/prevent the formation of aggressive CNV formation. Barry-Hamilton et al.⁸ also observed LOXL2 induction in disease-associated fibrosis and neovascularization, and efficacy in models of fibrosis and cancer after AB0023 treatment. Zaffryar-Eilot et al.¹¹ reported as well an antiangiogenic effect of AB0023 treatment in neovascularization and tumor vascular development. Moreover, our findings are also consistent with our previously published results on the effect of both antibodies in a rabbit model of glaucoma filtration surgery.¹⁵ Although the effect of LOXL2 inhibition was not directly compared to bevacizumab (gold standard in clinical practice for CNV) in this study, comparison to previous results of our lab showed that the antibody against LOXL2 induced a similar range of inhibition in neovascularization (of 47%) compared with bevacizumab (inhibition of 34%) in the CNV-model.³² Notably, in comparison to anti-VEGF treatment, LOXL2 inhibition also affected inflammation and fibrosis in lasered mice. Importantly, in this study, we showed that LOXL2 was expressed in retinal endothelial cells, suggesting that LOXL2 could be involved in retinal vascular biology and that its expression could explain the strong antiangiogenic effect of the LOXL2 inhibitor. Our observations of an antiangiogenic effect with AB0023 treatment are in line with a recent study that showed that LOXL2 was expressed in angiogenic endothelial cells as a hypoxia-target and accumulated in the endothelial ECM.¹² It is also known that induction of angiogenesis regulators, such as VEGF, occurs with similar kinetics as LOXL2 in mouse postischemic revascularization,³³ suggesting that LOXL2 plays an important role in the endothelial cell angiogenic response. Moreover, Zaffryar-Eilot et al.¹¹ demonstrated that administration of LOXL2-inhibitor to endothelial cells partially reduced VEGF-induced phosphorylation of ERK 1/2, known to be a major mediator of VEGF induced proliferation. These data suggest that the antiangiogenic effect of AB0023 may result from inhibition of VEGF-induced signaling in endothelial cells. Indeed, we showed that administration of the anti-LOXL2 antibody was associated with reduced transcription levels of VEGF in choroid and retina. Zaffryar-Eilot et al.³⁴ hypothesize that the effect of LOXL2-inhibition on VEGF signaling can be mediated by putative LOXL2-receptors on endothelial cells; however, further research is necessary. Here, it might well be possible that the observed anti-inflammatory effect is a consequence of the reduced angiogenesis after LOXL2 antibody treatment (i.e., less opportunity for inflammatory cells to reach the disease site). 

In conclusion, this study showed that LOX and LOXL2 play a role in the pathogenesis of CNV. Targeting LOXL2 with an inhibitory monoclonal antibody (AB0023) had a broader efficacy than targeting LOX, reducing angiogenesis and inflammation, in addition to fibrosis. We hypothesize that the role of LOXL2 in the disease progression of CNV is likely an early event and the efficacy characterized at study termination is a consequence of inhibition of the LOXL2 induced shortly after laser. 

Acknowledgments

The authors thank Sofie Beckers, Magda Bressink, and Ann Verbeek for their technical support. 

Supported by Fund for Research in Ophthalmology and by Arresto Biosciences (acquired by Gilead Sciences). 

Disclosure: T. Van Bergen, P; R. Spangler, Gilead Sciences (E); D. Marshall, Gilead Sciences (E), P; K. Hollanders, None; S. Van de Veire, None; E. Vandewalle, None; L. Moons, None; J. Herman, None; V. Smith, Gilead Sciences (E), P; I. Stalmans, Gilead Sciences (F), P