Studies on Axenfeld–Rieger syndrome patients and mice reveal Foxc1's role in corneal neovascularization

Normal vision critically depends on the transparency of the cornea overlying the lens. This tissue is made up of precisely oriented layers of fibroblasts perfectly interwoven with the collagen and other ECM components that they produce. Importantly, in the healthy eye, there are no blood vessels or lymphatic vessels in the stroma, making it what is known as an “angiogenically privileged” site. However, in a number of human ocular disorders, and in response to injuries like chemical burns, blood vessels do grow into the cornea from around the junction between the iris and the sclera (the limbus). This pathological angiogenesis can severely disrupt clarity of vision and necessitate corneal transplants. Consequently, there is intense interest in understanding the processes that normally restrict blood vessel growth into the cornea. In PNAS, Seo et al. (1) provide compelling evidence from human and mouse studies that implicate one member of the large Forkhead box (Fox) transcription factor family [Forkhead box transcription factor c1 (Foxc1)] in the regulation of corneal avascularity. Specifically, they suggest that Foxc1 regulates the normal delicate balance between factors in the cornea that can promote angiogenesis (e.g., VEGF and VEGF receptor) and those that inhibit growth of blood vessels and lymphatics (e.g., soluble VEGF receptor).

Corneal fibroblasts in the stroma and endothelial cells in the limbus originate from the neural crest (2), a transient embryonic tissue along the lateral margins of the neural fold. The study by Seo et al. (1) sheds light on the interplay between neural crest cells (NCs) and endothelial cells (ECs) in the limbus, and its implications for corneal vascularity associated with the rare autosomal dominant disorder Axenfeld–Rieger syndrome (ARS) in humans. Some patients with ARS develop defects in their teeth, heart, and abdominal region, as well as in their eyes. More specifically, the eye phenotype affects the anterior segment, which comprises the cornea, iris, ciliary body, and lens. FOXC1, PITX2, and PAX6 genes have been implicated in the etiology of ARS (3). Seo et al. (1) identify that ARS patients with FOXC1 gene deletions or duplications showed pathological corneal angiogenesis consistent with previous findings on the role of FOXC1 gene dosage in mammalian ocular and CNS development (4). In mice, Foxc1 is expressed in the NCs that make up the cornea, as well as in the blood vessels. Conditional deletion of Foxc1 specifically in the neural crest lineage induces neovascularization in the cornea with no change in the lens vasculature, which was also observed postnatally. Interestingly, corneas of NC conditional Foxc1^+/− mice are more susceptible to neovascularization after alkali burn, which makes the gene dosage issue clinically relevant.

The study by Seo et al. (1) proposes a molecular explanation for the anterior segment abnormalities observed in ARS. They suggest that the increased expression of matrix metalloproteases (MMPs; MMP3, MMP9, and MMP19) in the cornea of the NC-deleted Foxc1^−/− mice is likely a contributing factor to the vascularization in this tissue. MMPs are enzymes that are responsible for degrading a wide range of ECM components, thereby facilitating ECs’ migration, coalescence, and tube formation. Specifically, MMP9 expressed by inflammatory cells in close apposition to the vasculature is known to trigger the angiogenic switch in tumors by mobilizing VEGF from tumor matrix and making it available to bind to VEGF receptor 2 on ECs and to transmit signal (5). Similarly, in the cornea, MMP9-triggered degradation of the matrix and fragmentation of the collagen fibers in stroma are hypothesized to release the matrix-sequestered VEGF in the local milieu, thereby altering the balance between pro- and antiangiogenic factors (Fig. 1A). Recent evidence in zebrafish that foxc1 is essential to maintain vascular basement membrane integrity (6) supports this interpretation. In contrast, Seo et al. (1) do not observe changes in Vegf mRNA expression in NC-Foxc1 mutant mice but observe an increase in both Vegf receptor 1 (Vegfr1) and soluble Vegf receptor 1 (sVegfr1) levels. The soluble form of VEGFR1 (also called sFLT1), comprising only the extracellular domain of VEGFR1, is thought to act as a “growth factor sink” because it binds to VEGF and other family members without transducing signal. This dysregulation of the balance between pro- (VEGFR1) and antiangiogenic (sVEGFR1, a nonfunctional receptor) molecules could significantly alter the finely tuned VEGF pathway and trigger neovascularization in the cornea. This finding is corroborated in past elegant studies by Ambati et al. (7), which used a host of manipulations of sVEGFR1 to demonstrate convincingly that this mechanism of corneal avascularity was conserved across evolution, ranging from manatees (the only known species to have vascularized corneas), to marine mammals (dugongs, dolphins, and whales), to elephants, and finally to humans. Thus, the loss of Foxc1 function resulting in changes in MMP expression and VEGF-receptor ratio levels is likely to alter the balance between pro- and antiangiogenic factors, thereby increasing neovascularization in the limbus (Fig. 1A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

(A, B, and D) Three hypotheses are depicted under physiological and pathological conditions. (A) In hypothesis 1, under physiological conditions, Foxc1 in NCs negatively regulates MMP levels, and concomitantly maintains the levels of svegfr1 and vegfr1 in the cornea, leading to low VEGF bioavailability and thereby maintaining an avascular cornea. In pathological situations, this regulation is lost, resulting in high VEGF bioavailability leading to corneal neovascularization. (B) In hypothesis 2, under physiological conditions, Foxc1 functions in NCs to alter differentiation toward EC or smooth muscle cell lineage. Because NC-Foxc1^−/− mice show reduced smooth muscle cells [figure S3D in the study by Seo et al. (1)], we hypothesize that loss of Foxc1 drives cells to the EC lineage as indicated by the bold arrow in B. (C) Corneas of WT control mice (Left) and Ra^Op/+ heterozygous adult mice (Right). (D) In hypothesis 3, under physiological conditions, Foxc1 and SoxF transcription factors function synergistically to act as negative regulators of corneal angiogenesis. Under pathological conditions, such as loss of Foxc1 or SoxF gene function, this regulation is disturbed, leading to corneal neovascularization (red lines).

In addition to the model that Seo et al. (1) propose, there are other possible ways in which reducing the dosage of Foxc1 could lead to corneal vascularity. In the chick, NCs are some of the first cells to differentiate into smooth muscle cells that surround the arteries. Similarly, in the mouse, cell-fate mapping studies confirmed that the majority of corneal stromal and ECs are NC-derived (2). The chick and mouse models suggest that NCs have abilities to differentiate into ECs and smooth muscle cell lineages. In the PNAS study, Seo et al. (1) use a Wnt1-Cre line to delete Foxc1 conditionally in cells derived from the neural crest. Assuming that the deletion of Foxc1 gene occurred in a multipotent neural crest population (8) that resides in the cornea, one possibility is that the absence of Foxc1 promotes EC fate over the smooth muscle cell differentiation pathway, leading to overabundance of blood vessels in the cornea (Fig. 1B). This hypothesis is supported by data in figure S3D in the study by Seo et al. (1), where iris smooth muscle cell differentiation is impaired in NC-Foxc1^−/− mutant mice. This concept is also reinforced by previous reports suggesting that forkhead genes (foxc1 and foxc2) can act as molecular switches to induce either para-axial or intermediate mesoderm cell fate in vertebrate organogenesis (9).

To date, transcription factors that affect the development of artery, vein, and lymphatic vessels in mammals include gridlock (Hey2), SoxF (SoxF-7, SoxF-17, and SoxF-18), Foxc1/c2, Coup-TFII, and Prox1 (10). All these factors act as proangiogenic/lymphangiogenic factors. The study by Seo et al. (1) suggests that Foxc1 acts as an inhibitory factor for ECs, because mutation in the Foxc1 gene led to neovascularization of the cornea. Interestingly, a phenocopy of the Foxc1 mutation is also observed in the loss of Sox-F function (Fig. 1C). The ragged Opossum (Ra^Op) mouse, carrying a SOX18-dominant negative mutant allele, displays aberrant vascularization of the cornea in adult heterozygous animals (Fig. 1C, Ra^Op/+). The Ra^Op mutant mice carry a natural mutation in the Sox18 [SRY (sex determining region Y) box 18] gene (11, 12) that triggers the production of a truncated nonfunctional protein that suppresses F-group SOX (Sox7, Sox17, and Sox18) proteins in blood vessels. During lymphangiogenesis, Sox18 is expressed in venous cells to induce Prox-1 expression (11); at the same embryonic stages, Foxc1 and Foxc2 are required to maintain artery vs. vein specification (13). This observation enables us to suggest that a subset of ECs coexpress SoxF and Foxc transcription factors in vivo and that cooperation between these factors is part of a mechanism responsible for maintaining corneal avascularity. Whether genetic and direct physical interactions of Sox18 and Foxc1 occur is not known and would be worth assessing. We hypothesize that Foxc or SoxF acts as a negative regulator of angiogenesis to restrict the vasculature in the limbus (Fig. 1D). Because both SOX18 and FOXC1 are mutated in the human syndromes hypotrichosis-lymphoedema-telangiectasia (14) and ARS, respectively, a more global analysis of the interplay of Sox-F/Foxc would enable the identification of a regulatory network of genes affected by the association of these two factors during development and disease. Previous work supports this concept in that Foxc transcription factor interacts with E26 transformation specific factors synergistically to control ∼27% of endothelial-specific promoters (15). Seo et al.'s discovery (1) thus propels Foxc1 as a clinical target. This finding may have an impact on other fields of biology, such as hemangioma (vascular tumors) research, because recent reports have shown that proliferating infantile hemangioma is composed primarily of primitive mesodermal cells with a neural crest stem cell phenotype (16).

The study by Seo et al. (1) establishes a molecular link between the blood and lymphatic vascular endothelium and NC cells in the corneal limbus tissue. The nature of this interplay is still debatable; however, this discovery further advances our understanding of the etiology of ARS and positions Foxc1 as a potential molecular target for corneal neovascularization.