Angiopoietin-1 Is Regulated by miR-204 and Contributes to Corneal Neovascularization in KLEIP-Deficient Mice

Abstract

Purpose.: Corneal neovascularization can cause loss of vision. The introduction of anti-VEGF therapy has been a major improvement in therapeutic options. Recently, we established Kelch-like Ect2-interacting protein (KLEIP/KLHL20) knockout mice as a model of spontaneous corneal neovascular dystrophy. The aim of the present study was to characterize corneal neovascularization in progressive corneal dystrophy in KLEIP^−/− mice, to evaluate the efficacy of anti-VEGF therapy, and to identify novel molecular regulators in this experimental model.

Methods.: Corneal neovascularization was assessed by immunohistochemistry. Vascular endothelial growth factor signaling was inhibited by injection of a blocking antibody. Microarrays were used to measure expression of mRNA and microRNA (miRNA) in dystrophic corneae. Results were validated by immunohistochemistry and Western blotting.

Results.: Blood vessels and lymphatics grew from the limbus toward the dystrophic epithelium in corneae of KLEIP^−/− mice. Blocking VEGF signaling did not reduce phenotype progression. Correspondingly, microarray analysis revealed no upregulation of canonical vascular growth factors in late dystrophy. During phenotype progression, angiopoietin-1 expression increased while miR-204 expression decreased. Bioinformatic analysis identified a binding site for miR-204 in the angiopoietin-1 gene. Validation by in vitro experiments confirmed regulation of angiopoietin-1 by miR-204.

Conclusions.: Vascular endothelial growth factor does not act as a major player in corneal neovascularization in KLEIP^−/− mice. However, the proangiogenic factor angiopoietin-1 was strongly upregulated in late-stage phenotype, correlating with loss of miR-204 expression. Correspondingly, we identified miR-204 as a novel regulator of angiopoietin-1 in vitro. These findings may explain the incomplete efficacy of anti-VEGF therapy in the clinic and may provide new candidates for pharmaceutical intervention.

Introduction

Corneal translucence is a highly complex result of multiple processes and is essential for unimpaired vision. ¹ Lack of blood vessels is an important factor in maintaining corneal translucence. ^2,3 The pathological growth of blood vessels into the cornea is called corneal neovascularization (corneal NV). Several corneal pathologies associated with corneal NV can cause blindness in humans. ^4,5 In the United States, extended wear of soft contact lenses, mechanical injury, infection, and other causes lead to corneal NV in up to 1.4 million patients. ^4,6 It is estimated that in up to 12% of these patients, corneal NV may be correlated to loss of visual acuity. ⁴ Globally, severe sight-threatening diseases such as infection with herpes simplex virus 1 (HSV-1), infection with Chlamydia trachomatis , and onchocerciasis are causes of corneal blindness. ^4–6 Corneal neovascularization is a common feature of these infectious diseases. ^7–9  

In addition, corneal NV is a major risk factor in corneal allograft transplantation, the most common allograft transplantation in patients. ^3,10 Presence of corneal neovessels doubles the risk of graft failure. ¹¹  

Therefore, understanding the mechanisms of corneal NV is critical for understanding the pathophysiological basis for loss of vision caused by a wide range of pathologies. 

Physiologically, a balance of proangiogenic and antiangiogenic factors maintains corneal avascularity. ^2,3,12 In various pathologies, this balance is disturbed and blood vessels sprout into the cornea. Several proangiogenic factors have been identified in corneal pathologies, most importantly VEGF, fibroblast growth factor (FGF), platelet-derived growth factor (PDGF), and placental growth factor (PLGF). ^2,3,6,13,14  

Conventional pharmacological treatment of corneal NV consists of topical steroids and nonsteroidal anti-inflammatory drugs. Recently, anti-VEGF agents have been introduced for studying corneal NV in experimental settings and for treating corneal NV in clinical trials. ^3,15 Still, the efficacy of anti-VEGF monotherapy is limited in clinical settings. ³ Therefore, it is necessary to identify causes for the moderate efficacy of anti-VEGF agents in corneal NV and to identify and characterize additional proangiogenic transmitters in corneal NV. 

Apart from corneal NV, corneal dystrophies are a cause of corneal blindness. ^16,17 Corneal dystrophies are genetic disorders causing epithelial and stromal aberrations. ^17,18 Genetic models suitable for studying corneal dystrophy in model organisms are sparse. Available genetic modifications that cause corneal dystrophy comprise knockout of Notch1, ¹⁹ proteoglycans, ²⁰ keratins, ²¹ and transcription regulators. ²² Recently, we identified Kelch-like Ect2-interacting protein (KLEIP) as essential for corneal structure and translucence. ²³ KLEIP^−/− mice spontaneously develop corneal dystrophy, which can be accelerated by superficial epithelial abrasion. 

Dystrophy in KLEIP^−/− mice is accompanied by corneal NV. ²³ In contrast to other models, corneal NV in KLEIP^−/− mice develops spontaneously. ²⁴ Thus, KLEIP^−/− mice can serve as a genetic model of corneal dystrophy and corneal NV. 

However, the molecular mechanisms underlying corneal epithelial dystrophy and stromal NV in KLEIP^−/− mice are not known. Most mouse models of corneal NV are dependent on VEGF or FGF2. ^3,25–27 It is not known whether these growth factors are responsible for all forms of corneal NV, including KLEIP^−/− corneal neovascular dystrophy. 

In the present study, we aimed at understanding mechanisms responsible for corneal NV in dystrophic corneae of KLEIP^−/− mice. First, we developed a morphological staging system to classify dystrophy development in KLEIP^−/− mice. Then, we used immunohistological staining to get a specific understanding of corneal NV in KLEIP^−/− corneae. Because anti-VEGF treatment is currently in the spotlight of antiangiogenic therapy in corneal NV, we investigated the effects of inhibition of VEGF in the KLEIP^−/− neovascular dystrophy model. Finally, using RNA microarrays and immunohistology, we molecularly characterized corneal NV and epithelial dystrophy in KLEIP^−/− mice. We identified a novel interaction of miR-204 and angiopoietin-1 as a possible mechanism for VEGF-independent neovascularization in corneal disease. 

Materials and Methods

KLEIP^−/− Mice

For all experiments, we used KLEIP^−/− mice as described before. ²³ Mice were kept under specific pathogen-free conditions according to the animal facility regulations of the Medical Faculty Mannheim, Heidelberg University. All animal experiments were approved by the Regierungspräsidium Karlsruhe (protocol numbers 35-9185.81/G-82/11 and I-07/03) and were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Intravitreal Injection of VEGF-Blocking Antibody

Fourteen mice were treated with anti-VEGF-A intravitreal injections. Anti-VEGF-A antibody (1 μg; 2G11-2A05, BLD-512808; BioLegend, San Diego, CA, USA) was administered to 3-week-old mice without corneal phenotype (n = 7 KLEIP^−/− eyes and n = 7 KLEIP^+/+ eyes). Parallel intravitreal injection of the equivalent amount of isotype control (TK2758, BLD-400516; BioLegend) served as control (n = 7 KLEIP^−/− eyes and n = 7 KLEIP^+/+ eyes). To achieve maximum comparability, animals received anti-VEGF-A treatment on one eye and control treatment on the other eye. The rate of dystrophy development was assessed 3 and 6 weeks after initial treatment. 

Immunohistology

For whole-mount immunostaining, corneae were excided along the corneal limbus, fixed with 4% paraformaldehyde, and blocked for 1 hour using bovine serum albumine (BSA)–based blocking solution. They were incubated with rat anti-mouse CD31 antibody (1:250; MEC 13.3; BD Pharmingen, Heidelberg, Germany) and rabbit anti-mouse LYVE-1 antibody (1:100; cat. no. 103-PA50AG; Reliatech, Wolfenbüttel, Germany) in normal goat serum–based buffer at −20°C overnight. Corneae were washed, then incubated with goat anti-rat IgG Alexa 488–conjugated antibody (1:500; Molecular Probes, Darmstadt, Germany) and goat anti-rabbit IgG Alexa 546–conjugated antibody (1:500; Molecular Probes) for 1 hour and 4′,6-diamidino-2-phenylindole (DAPI) (1:5000) for 3 minutes. At least 25 whole-mounted corneae were analyzed for each staining (25 for LYVE-1 and 31 for CD31). For staining of corneal sections, eyes were enucleated, embedded in optimal cutting temperature compound (OCT; Sakura Finetek, Alphen aan den Rijn, The Netherlands), and quick-frozen in liquid nitrogen, and 8-μm-thick sections were cut. Sections were washed, fixed in −20°C methanol for 10 minutes, then blocked for 1 hour using BSA-based blocking solution. Sections were incubated with the following primary antibodies for 12 hours: rat anti-mouse CD31 antibody (1:250; as above), rabbit anti-mouse LYVE-1 antibody (1:100; as above), rabbit anti-mouse angiopoietin-1 antibody (1:1000; ab8451; Abcam, Cambridge, UK), rabbit anti-mouse Tie2 antibody (1:100; sc-324; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Then they were stained with secondary antibodies and DAPI as described above. Both corneal whole mounts and sections were viewed under a fluorescence microscope (Carl Zeiss, Oberkochen, Germany) using Zeiss AxioVision software. Corneal sections from at least 16 different tissue samples were used for each staining. 

Image Analysis and Statistics

ImageJ (National Institutes of Health, Bethesda, MD, USA) was used for measurement of corneal vascularized area (CD31: n = 4–14 corneae; LYVE-1: n = 3–12 corneae), ²⁸ measurement of distance between blood vessel and Bowman's membrane (in n = 3 corneae), and angiopoietin-1 (Angpt1) mean fluorescence intensity (MFI; measured in n = 16 stage 3 corneal sections and n = 10 stage 0 control sections). R programming language (r-project.org [in the public domain]) and MATLAB (Mathworks, Inc., Natick, MA, USA) were used for statistical analyses. 

RNA Sample Acquisition

Corneae were quick-frozen in liquid nitrogen. In total, 12 corneae were analyzed: three early-stage dystrophic corneae (KLEIP^−/− stage 1 and 2; dystrophy was induced by mechanical abrasion as described before ²³ ), three late-stage dystrophic corneae (KLEIP^−/− stage 3), and six control corneae (KLEIP^−/− stage 0). Total RNA was extracted using RNeasy Mini Kit (Qiagen, Hilden, Germany) and TissueLyser II system (Qiagen) according to the manufacturer's protocols. Complementary DNA target was synthesized, fragmented, and biotin-labeled using the Whole Transcript Target Labeling and Control Reagents (Affymetrix, Santa Clara, CA, USA), according to the Affymetrix defined protocol. Labeled and fragmented cDNA was hybridized to Affymetrix MoGene-2_0-st microarrays using Affymetrix defined protocol. Microarrays were washed using an Affymetrix fluidics station 450 and were scanned with Affymetrix GeneChip Scanner 3000 controlled by Affymetrix GeneChip Command Console (AGCC). 

RNA Microarray

Gene expression profiling was performed using arrays of MoGene-2_0-st-type (Affymetrix). This microarray chip detects both mRNA and microRNA (miRNA). The raw fluorescence intensity values were normalized applying quantile normalization. Differential gene expression was analyzed with one-way ANOVA. All statistical analyses were performed using a commercial software package, SAS JMP10 Genomics, version 5, from SAS (SAS Institute, Cary, NC, USA). A false-positive rate of α = 0.05 with false discovery rate (FDR) correction was taken as the level of significance. 

Angiopoietin-1 Expression After Mir-204-5p Transfection of HUVEC

Human umbilical vein endothelial cells (HUVEC; passage number P5) were grown until 60% confluence in six-well culture plates and transfected with 100 nM control GFP-22 siRNA (cat. no. 1022064; Qiagen) or 100 nM Syn-hsa-miR-204-5p mimic (cat. no. MSY0000265; Qiagen) using Transfection Reagent and Medium (Santa Cruz Biotechnology) according to the respective manufacturers' instructions. Additionally, nontransfected controls were used. Cells were incubated for 24 hours, medium was changed to regular endothelial cell culture medium, and cells were harvested 48 hours after medium change. For protein extraction, 75 μL Triton-based lysis buffer was used per well. Western blot membranes were stained with anti-angiopoietin-1 antibody (1:500; ab8451; Abcam) and with Ponceau S for loading control. MiR-204 knockdown and Western blotting was repeated independently at least six times. 

Results

Phenotypical Staging of Dystrophy Progression in KLEIP^−/− Mice

KLEIP^−/− mice develop both corneal opacification and corneal stromal neovascularization. ²³ To assess corneal dystrophy semiquantitatively, we developed a staging system for corneal dystrophic phenotype and classified corneal phenotypes as “no dystrophy,” “early dystrophy,” or “late dystrophy” according to the following criteria (Fig. 1A). A completely translucent physiological cornea was defined as stage 0, or no dystrophy. Corneal opacity detectable by 4-fold magnification of a dissecting microscope was defined as stage 1. Corneal opacity seen without magnification that still allowed some translucence was defined as stage 2. Stage 1 and stage 2 were considered stages of early dystrophy. Complete opacity of the cornea without any translucence was defined as stage 3, or late dystrophy. Mice underwent early dystrophy stages before reaching late dystrophy. Late dystrophy was most prominent in the corneal center (Fig. 1A). 

Figure 1

 

Development of corneal neovascularization (corneal NV) in corneal dystrophy in KLEIP^−/− mice. (A) Spontaneous dystrophy development in KLEIP^−/− progresses from stage 0 (no dystrophy) to stage 3 (epithelial metaplasia). (B, C) Blood vessels (green, CD31) and lymphatics (red, LYVE-1) sprout from the corneal limbus toward the dystrophic corneal center (scale bars [B, C]: 500 μm). (D) Corneal NV was quantified by measuring corneal area covered by blood vessels (green, CD31) and lymphatics (red, LYVE-1). Error bars represent standard deviation; *P < 0.05.

Distribution of Blood Vessels and Lymphatics Within the Corneal Stroma

To assess the pattern of blood vessel growth in dystrophic corneae, we performed whole-mount immunostaining of blood vessels and lymphatics in corneae of all dystrophy stages (n ≥ 3 corneae for each stage; Fig. 1B). In stage 0, blood vessels and lymphatics were found only at the corneal limbus and did not sprout into the translucent parts of the cornea. Starting in early dystrophy (stage 1), blood vessels and lymphatics sprouted from the corneal limbus toward the corneal center. In late dystrophy, all vessels reached the corneal center (Figs. 1B, 1C). Corneal area covered by vessels was significantly greater in early-stage dystrophy compared to no dystrophy (stage 0 versus stage 1; CD31: P = 0.0364, LYVE-1: P = 0.0380; Fig. 1D). Also, the difference between early stage and late stage was significant (stage 2 versus stage 3; CD31: P = 0.0083, LYVE: P = 0.0405; Fig. 1D). Stages 1 and 2, both considered early dystrophy, did not differ significantly in terms of vessel coverage (CD31: P = 0.1976, LYVE-1: P = 0.8923). The difference in corneal vascularized area between no dystrophy and late dystrophy was found to be highly significant (CD31: P < 0.0001, LYVE-1: P = 0.0008). 

To assess apical–basal vessel distribution, we measured the distance of vessel profiles from the epithelial–stromal border (i.e., Bowman's membrane; Fig. 2). With increasing distance from the epithelium, vessel density decreased exponentially (n = 256 vessel profiles in n = 3 corneae). Most vessels were located directly under the epithelium; barely any vessels were present in the basal stroma (Fig. 2). The directional growth of neovessels toward dystrophic areas suggests a gradient of angiogenic factor(s) originating from the corneal epithelium. 

Figure 2

 

Blood vessel density within the corneal stroma decreases exponentially with increasing distance from the epithelium, suggesting a gradient of soluble angiogenic factors. (A) Blood vessels in stage 3 dystrophic corneae (KLEIP^−/−) are restricted to the apical half of the stroma (green, CD31; blue, DAPI; scale bar: 50 μm). (B) Histogram plot of n = 256 vessel profiles in n = 3 KLEIP^−/− stage 3 mice. On the vertical axis, the position within the corneal stroma is plotted, corresponding to (A).

Injection of VEGF-Blocking Antibody Does Not Attenuate Neovascular Dystrophy

Intravitreal injection of VEGF-neutralizing antibody is a common and effective strategy both in clinical and in experimental settings. ^3,29 We assessed whether intravitreal injection of VEGF-blocking antibody inhibits development of corneal neovascular dystrophy. As bevacizumab does not effectively neutralize murine VEGF, we used an antibody with proven efficacy against murine VEGF. ^30,31 Vascular endothelial growth factor–neutralizing antibody was injected into seven eyes of KLEIP^+/+ mice and into seven eyes of KLEIP^−/− mice without any corneal dystrophy (i.e., stage 0). As an additional control, we injected unspecific IgG2α into the contralateral eye (Table 1). Within 6 weeks, none of the KLEIP^+/+ mice developed corneal neovascular dystrophy. In contrast, two of seven KLEIP^−/− mice of the IgG2α group and five of seven KLEIP^−/− mice of the anti-VEGF group developed stage 3 dystrophy and neovascularization (Table 1). Interestingly, IgG2α-treated mice showed even less corneal neovascular dystrophy than the anti-VEGF treatment group. Yet the difference was not statistically significant, which does not support the conclusion that anti-VEGF treatment is actually harmful but disproves our initial hypothesis. Thus, intravitreal injection of anti-VEGF failed to inhibit neovascularization in KLEIP^−/− corneae. 

Table 1

 

Inhibition of VEGF Did Not Inhibit Development of Neovascularization and Dystrophy

	Sample Size, Eyes	Neovascular Dystrophy, i.e., Stage 3	No Neovascular Dystrophy, i.e., Stage 0	χ2
KLEIP+/+ IgG2α	7	0	7
KLEIP+/+ anti-VEGF	7	0	7
KLEIP−/− IgG2α	7	2	5	NS
KLEIP−/− anti-VEGF	7	5	2

 Fourteen KLEIP^−/− eyes and 14 KLEIP^+/+ eyes were treated with 1 μL either IgG2α or anti-VEGF (intravitreal injection). No KLEIP^+/+ eyes developed dystrophy within 6 weeks. No significant difference was present in dystrophy development after 6 weeks between KLEIP^−/− IgG2α and KLEIP^−/− anti-VEGF groups.

Microarray Analysis Identifies Dystrophy- and Opacity-Related Genes

To identify molecular players involved in corneal dystrophy and angiogenesis, we performed two RNA microarray experiments. Global gene expression was measured in late dystrophic KLEIP^−/− corneae (n = 3) versus nondystrophic KLEIP^−/− control corneae (n = 3) and in early dystrophic KLEIP^−/− corneae (n = 3) versus nondystrophic KLEIP^−/− control corneae (n = 3). For both microarray experiments, within-group correlation was high and between-group correlation was low (Supplementary Fig. S1). 

Microarray results were used to assess expression levels for genes that are known to play a role in corneal dystrophies. One mechanism that leads to loss of corneal translucence is transformation of stromal keratocytes into activated fibroblasts. Quiescent stromal keratocytes express crystalline proteins such as Aldh1a1 (aldehyde dehydrogenase) and Tkt (transketolase). ² Activated fibroblasts do not express these proteins and thus contribute to corneal opacification. Correspondingly, we observed an increasing downregulation of these genes during phenotype progression (Table 2). 

Table 2

 

Expression of Highly Regulated Genes in KLEIP^−/− Corneae That Are Correlated to Corneal or Epithelial Pathologies

Gene Symbol	Implied in:	Ref.	Fold Change in Early Dystrophy	Fold Change in Late Dystrophy
Alox12b	Autosomal recessive congenital ichthyosis, human	32	1.13	72.35*
Saa3	Gelatinous drop-like corneal dystrophy, human	33	2.69	63.46†
Dmkn	Skin wound healing	34	1.18	61.72*
Sprr1b	Squamous metaplasia in dry eye disease, human	35	1.12	52.28*
Defb4	Resistance to Pseudomonas aeruginosa keratitis	36	1.43	50.26†
Tkt	Corneal crystallins: genes for maintenance of corneal transparency, human	2	0.71‡	0.40*
Aldh1a1		2	0.41*	0.35†

  Alox12b, Saa3, Dmkn, Sprr1b, Defb4: genes that were extremely upregulated in late-stage corneal dystrophy have previously been implied in corneal or epithelial pathologies in humans. These genes are related to specific human or murine corneal dystrophies or epithelial proliferative processes. Tkt and Aldh1a1 are crystalline proteins that are important for maintenance of corneal transparency. Alox12b, arachidonate 12-lipoxygenase, 12R type; Saa3, serum amyloid A 3; Dmkn, dermokine; Sprr1b, small proline-rich protein 1B; Defb4, defensin beta 4; Aldh1a1, aldehyde dehydrogenase; Tkt, transketolase.

 * P < 0.001.

 † P < 0.01.

 ‡ P < 0.05.

Among the most significantly upregulated genes in late-stage dystrophy, we identified several that are related to corneal dystrophy in particular and epithelial metaplasia in general: Alox12b (arachidonate 12-lipoxygenase, 12R type), Saa3 (serum amyloid A3), Dmkn (dermokine), Sprr1b (small proline-rich protein 1B), and Defb4 (defensin beta 4; Table 2). Expression of these genes was highly elevated by more than 50× in late-stage versus control corneae. These genes are related to specific corneal dystrophies or epithelial proliferative processes (Table 2). ^32–36 This high magnitude of upregulation suggests an involvement of these genes in dystrophy development in KLEIP^−/− mice. 

Angiopoietin-1, but Not VEGF, FGF, PDGF, or PLGF, Contributes to Pronounced Neovascularization in the Transition Between Early and Late Dystrophy

To identify molecular mechanisms responsible for corneal NV, microarray data were analyzed for regulation of proangiogenic and antiangiogenic transmitters as well as vascular markers. Microarray analysis showed a significant upregulation of CD31 and LYVE-1 in early- and late-stage dystrophic corneae (Fig. 3). This corresponds to immunohistological staining showing no vessels in nondystrophic KLEIP^−/− corneae and many vessels in late dystrophic KLEIP^−/− corneae (Figs. 1B, 1C). However, canonical vascular growth factors such as VEGF-A, VEGF-B, VEGF-C, and FGF2 were not significantly upregulated in late-stage corneae (Fig. 3). Platelet-derived growth factor-B and PLGF were only minimally upregulated in late-stage dystrophy (PDGF-B: late versus control fold change = 1.32, P = 0.048; PLGF: late versus control fold change = 1.17, P = 0.0019; Fig. 3) and upregulated even less in early dystrophy (Fig. 3). Inflammatory proangiogenic factors TNF-α, IL-1β, IL-6, and erythropoietin were not significantly upregulated in early- or late-stage dystrophy (data not shown). 

Figure 3

 

Tie2 and angiopoietin-1 are upregulated in KLEIP^−/− late-stage dystrophic corneae while other canonical vascular growth factors are not. Gene expression of proangiogenic genes and vascular markers expressed in fold change (KLEIP^−/− early dystrophy versus KLEIP^−/− control and KLEIP^−/− late dystrophy versus KLEIP^−/− control, total n = 12 corneae). Growth factors from the VEGF, PLGF, PDGF, or FGF family are not regulated more than 1.5-fold in late dystrophy. Only angiopoietin-1 (Angpt1) and its receptor Tie2 are significantly upregulated more than 4-fold in late dystrophy. *P < 0.05, **P < 0.01, ***P < 0.001; dashed line: level of gene expression in control corneae (KLEIP^−/− without dystrophy).

As described above, corneal NV is already detectable in early stages of corneal dystrophy. Initial vessel growth was associated with increased VEGF-A expression (Fig. 3). However, much more vessel growth took place in the transition of stage 2 to stage 3 phenotype, where we found strongly increased expression of angiopoietin-1 (late-stage versus control fold change = 4.25; P = 0.0001) and its receptor Tie2 (late-stage versus control fold change = 4.38; P = 0.0001) while VEGF-A expression returned to baseline level (Fig. 3). 

Based on these data, we concluded that canonical vascular growth factors, particularly VEGF-A, possibly contributed to initial vessel sprouting but did not contribute to the considerable amount of angiogenesis taking place between early- and late-stage phenotype. These findings strongly support the results of anti-VEGF injection experiments, which did not show a reduction of corneal NV upon neutralization of VEGF (Table 1). 

We subsequently stained corneal sections for angiopoietin-1 and Tie2. CD31-positive vessels were positive for Tie2 (Fig. 4A). Additionally, we detected perivascular CD31-negative Tie2-positive cells (Fig. 4A). Angiopoietin-1 was mainly detected in the cytoplasm of corneal epithelial cells but not in stromal cells (Fig. 4B). Corneal epithelial angiopoietin-1 fluorescence was significantly increased in stage 3 dystrophic corneae from KLEIP^−/− mice (n = 16 KLEIP^−/− stage 3 corneal sections and n = 10 KLEIP^−/− stage 0 control sections; relative mean fluorescence intensity = 4.14 ± 1.98; P = 0.00001; Fig. 4C). 

Figure 4

 

Tie2 and angiopoietin-1 are strongly expressed in late-stage dystrophy in KLEIP^−/− mice. (A) Corneal blood vessels expressed Tie2 (asterisk). The arrow indicates nonendothelial perivascular Tie2-expressing cells; scale bar (lower right): 10 μm. (B) Angiopoietin-1 (Angpt1) was not detectable in KLEIP^−/− control corneae but was highly expressed in the epithelium of KLEIP^−/− late-stage dystrophic corneae; scale bar (lower right): 100 μm. (C) Quantification of angiopoietin-1 (Angpt1) immunofluorescence in tissue sections (n = 16 KLEIP^−/− stage 3 corneal sections and n = 10 KLEIP^−/− stage 0 control sections; MFI, mean fluorescence intensity). This result is in quantitative agreement with the microarray data. Error bars represent standard deviation; ***P < 0.001.

MiR-204 and MiR-184 Are Highly Downregulated in Dystrophic Corneae

Among the most highly regulated genes in both microarray experiments, we identified two miRNAs, miR-204 (early dystrophy versus control: fold change = 0.32, P = 0.0002; late dystrophy versus control: fold change = 0.05, P = 0.00003; Fig. 5A) and miR-184 (early dystrophy versus control: fold change = 0.20, P = 0.0027; late dystrophy versus control: fold change = 0.16, P = 0.0007). Both miR-204 and miR-184 are implicated in physiological eye development and ocular pathologies. ^37–42 Twenty-fold downregulation of miR-204 in KLEIP^−/− late-stage dystrophic corneae corresponds to an almost complete suppression of the transcript. 

Figure 5

 

Angiopoietin-1 is a target of miR-204. (A) Expression level of miR-204 in corneae. (B) Expression levels of angiopoietin-1 (Angpt1) and previously validated negative targets of miR-204 are elevated or unchanged in dystrophic corneae where miR-204 is almost completely absent. (C) An octamer in the 3′ untranslated region of angiopoietin-1 is complementary to an octamer in miR-204-5p. Sequence from http://www.targetscan.org/ version 6.2 (Whitehead Institute for Biomedical Research, Cambridge, MA, USA). (D) Western blot of HUVEC lysates. Upper image: Detection of angiopoietin-1. Lower image: Corresponding loading control (Ponceau S). (E) Quantification of six independent Western blot experiments. Error bars represent standard deviation; *P < 0.05, **P < 0.01, ***P < 0.001; dashed line (A, B): level of gene expression in control corneae (KLEIP^−/− without dystrophy).

Evidence for a relevant effect of ablation of miR-204 in dystrophic corneae was provided by an analysis of 19 validated target genes that are either downregulated or upregulated by the presence of miR-204. ⁴³ Angiopoietin-1 and previously validated negative targets of miR-204 are shown in Figure 5B (Cxcl12, Tgfbr2, Jun, Igf2r, Pdgfb, Snai1, Snai2, Smad4, Cgn, Smad3). Of these 10 negative target genes, only 1 gene (Smad3) was regulated contrary to expectation (Fig. 5B). Likewise, of eight validated positive target genes, only one gene (Pcdh18) was regulated contrary to expectation in early dystrophy, and none was regulated contrary to expectation in late dystrophy (genes: Lrat, Rpe65, Pcdh18, Cldn16, Ttr, Cldn19, Slc4a4, Cldn10; data not shown). 

MiR-204 Is a Novel Negative Regulator of Angiopoietin-1

Finally, we identified a binding site for miR-204 in the angiopoietin-1 gene using the miRWalk-database (Fig. 5C). ⁴⁴ Six of 10 miRNA prediction tools included in miRWalk predicted angiopoietin-1 as a target of miR-204. This is consistent with the observation that during phenotype progression, miR-204 expression showed a significant decline (Fig. 5A) while angiopoietin-1 expression showed a significant increase (Fig. 5B). To verify these data in an independent experiment, we transfected endothelial cells with miR-204-5p mimic and measured angiopoietin-1 protein levels by Western blotting. In miR-204–transfected cells, we found significantly less angiopoietin-1 protein than in both controls. Angiopoietin-1 level in miR-204–transfected cells was 70.6 ± 27.3% of control while angiopoietin-1 level in GFP-22-siRNA–transfected cells was 99.0 ± 25.5% of control (miR-204 versus control: P = 0.027; miR-204 versus GFP: P = 0.046; Figs. 5D, 5E). 

Collectively, these data demonstrate that miR-204 regulates expression of angiopoietin-1 and contributes to KLEIP^−/− corneal neovascular dystrophy. 

Discussion

Corneal neovascularization is a relevant cause of morbidity worldwide. In the present study we used KLEIP^−/− mice to identify molecular mechanisms involved in corneal NV. We show that corneal NV in KLEIP^−/− mice is not chiefly regulated by VEGF but rather by angiopoietin-1. Furthermore, we provide evidence for a previously unknown regulation of angiopoietin-1 by miR-204. These findings open up a perspective for non-VEGF-targeting therapies for corneal vascular pathologies. 

Vessels in the Corneal Stroma Grow Toward the Dystrophic Epithelium

In KLEIP^−/− corneal neovascular dystrophy, stromal neovessels grow directionally toward dystrophic epithelium both in a planar dimension (toward the center) and in an apical–basal dimension (toward the epithelium). This observation suggests that the dystrophic corneal epithelium produces proangiogenic factors stimulating the growth of blood vessels and lymphatics within the stroma. KLEIP^−/− mice spontaneously develop corneal dystrophy and neovascularization and thus will serve as a model both for corneal dystrophies and corneal NV. ²³ To the best of our knowledge, the KLEIP^−/− corneal NV model is the first spontaneous in vivo corneal NV model that is not chiefly dependent on VEGF, FGF, PDGF, or PLGF. 

Angiopoietin-1, Not VEGF, Is Involved in Corneal Angiogenesis in KLEIP^−/− Mice

Most studies of corneal NV highlight the importance of VEGF isoforms and demonstrate partial efficacy of VEGF-targeting drugs. ^3,6 In the present study, we showed that most of the angiogenesis occurs between early and late phenotype stages in the KLEIP^−/− model and that during this period, corneal NV is not dependent on VEGF. Evidence was provided both by microarray analysis of VEGF expression (which was not increased in late-stage dystrophy) and by injection of VEGF-neutralizing antibody (which failed to attenuate corneal NV). Furthermore, microarray experiments showed that neither FGF, PLGF, nor PDGF cause corneal NV in KLEIP^−/− corneal dystrophy. 

However, angiopoietin-1 was significantly upregulated in late-stage dystrophic corneae. Immunohistology confirmed these findings. We could also histologically show that the source of angiopoietin-1 is the dystrophic corneal epithelium. This corresponds to above-mentioned data showing that blood vessels grow toward these abnormal epithelial cells, resulting in an inhomogeneous distribution of blood vessels within the corneal stroma. 

These findings suggest that dystrophic corneal epithelial cells secrete angiopoietin-1, thereby attracting stromal blood vessels. This is consistent with a study by Kanayama et al., ⁴⁵ who described the expression of angiopoietin-1 in cultured rabbit corneal epithelial cells and the subsequent attraction of human endothelial cells. ⁴⁵ Also, Morisada et al. ⁴⁶ reported corneal angiogenesis and lymphangiogenesis after local application of synthetic angiopoietin-1 to mouse corneae. 

The identification of angiopoietin-1 as the pivotal proangiogenic factor in KLEIP^−/− corneal NV is consistent with its biological role in angiogenesis. In blood vessels of adult mice, angiopoietin-1 is not needed for vessel quiescence but modulates vascular function in response to injury. ⁴⁷ This notion of the role of angiopoietin-1 in the vasculature is compatible with its function in KLEIP^−/− corneal dystrophy. 

Loss of miR-204 Promotes Angiopoietin-1 Expression

KLEIP^−/− late-stage dystrophic corneae display epithelial metaplasia, stromal neovascularization, and corneal haze. ²³ Correspondingly, crystalline proteins and other genes known to play a role in corneal dystrophies or epithelial proliferative processes showed pronounced regulation in microarray experiments. Most interestingly, two miRNAs (miR-204, miR-184) were strongly downregulated in KLEIP^−/− dystrophic corneae. These miRNAs are known to regulate eye development in mammals. ³⁹ Also, both are implied in secondary cataract in a mouse model. ⁴⁰ As several relevant target genes of miR-204 showed significant regulation, we conclude that miR-204 contributes to corneal phenotype development in KLEIP^−/− mice. Even more strikingly, bioinformatic analysis predicted regulation of angiopoietin-1 by miR-204. This novel regulation was validated by in vitro experiments with human cells. These data show that loss of miR-204 leads to increased expression of angiopoietin-1. Collectively, we provide strong evidence for a novel miR-204–angiopoietin-1 pathway in KLEIP^−/− corneal NV. 

Clinical Perspective

Corneal neovascularization can lead to loss of vision. Until a few years ago, the only available pharmaceuticals to treat corneal NV were corticosteroids and nonsteroidal anti-inflammatory drugs. ³ Introduction of VEGF-targeting agents such as bevacizumab, ranibizumab, and pegaptanib has been a certain improvement in treatment of corneal NV. ^3,15 Still, the problem of corneal NV has not been ultimately solved, which suggests that other non-VEGF–related mechanisms are relevant in corneal NV. Yet most studies focus on the role of VEGF in induction of corneal NV. Vascular endothelial growth factor–independent angiogenesis as seen in the KLEIP^−/− corneal NV model could potentially explain the incomplete efficacy of VEGF-targeting drugs in corneal NV patients. This will therefore serve as a model to evaluate VEGF-independent targeted therapy options for corneal NV. The previously unknown regulation of angiopoietin-1 by miR-204 especially suggests these molecules as potential targets for pharmaceutical intervention. 

Supplementary Materials

pdf S1 

Acknowledgments

We are grateful to Maria Muciek and Marlene Hausner for excellent technical assistance. 

Supported by grants from Deutsche Forschungsgemeinschaft (SFB/TR23: Integrated Research Training Group [JNK, JK]; KR1887/5-1 [JK]) and Studienstiftung des deutschen Volkes (JNK). 

Disclosure: J.N. Kather, None; J. Friedrich, None; N. Woik, None; C. Sticht, None; N. Gretz, None; H.-P. Hammes, None; J. Kroll, None