Wnt5a/β-catenin-mediated epithelial-mesenchymal transition: a key driver of subretinal fibrosis in neovascular age-related macular degeneration

Age-related macular degeneration (AMD) is characterized by the progressive degeneration of the neuronal retina, retinal pigment epithelium (RPE) and choriocapillaris complex, leading to severe and permanent central visual impairment among aging population in developed countries [1]. Neovascular AMD (nAMD), accounting for up to 90% of AMD-associated vision loss and blindness [2], is primarily characterized by the presence of choroidal or retinal neovascularization (CNV or RNV) in the macular region, ultimately resulting in the formation of fibrotic scar [2]. The prevalence of fibrosis at baseline, 12, 24, and 60 months in patients with nAMD was reported as follow: 13%, 32%, 36%, and 56%, respectively, according to a recent epidemiological report [3]. Despite the efficacy of anti-vascular endothelial growth factor (anti-VEGF) therapy as the cornerstone in nAMD management, capable of stabilizing or even enhancing the visual function [4], approximately 1/3 of the patients with subretinal fibrosis still experience poor prognosis, and even deterioration of the visual acuity [5]. Therefore, it is imperative and urgent to explore the molecular mechanisms underlying the pathogenesis of subretinal fibrosis to develop effective therapies for subretinal fibrosis secondary to nAMD.

The RPE, a highly specialized monolayer of polarized, pigmented epithelial cells, is located between the outer segment of photoreceptors and the choriocapillaris endothelial cells, playing an essential role in maintaining visual function [6]. The primary functions of RPE encompass light energy absorption focused by the lens onto the retina [7]; transportation of nutrients and ions between photoreceptors and the choriocapillaris [8]; secretion of various factors crucial for maintaining structural integrity in both retina and choriocapillaris [9]; facilitation of phagocytosis in photoreceptor outer segments; and preservation of the blood-retinal barrier [10]. Dysfunction of the RPE, particularly its transition from an epithelial to a mesenchymal phenotype, is implicated in the pathogenesis of subretinal fibrosis, a hallmark of the end-stage complications of nAMD. Recently, several signaling pathways regulating the epithelial–mesenchymal transition (EMT) of RPE cells have been identified, among which the canonical Wnt/β-catenin pathway has been extensively studied and well-characterized [11]. Jung Woo Han et al. demonstrated a crucial role of Wnt/β-catenin signaling in RPE proliferation and EMT using a laser-induced mouse model, while also observing a significant upregulation of the non-canonical Wnt ligand Wnt5a expression in laser-treated RPE compared to control RPE [12]. With accumulating evidence indicating a substantial role for the non-canonical Wnt ligand Wnt5a in the processes of EMT and fibrosis across diverse tissues [13‐16], the precise molecular mechanisms underlying Wnt5a-triggered signaling and its interplay with canonical Wnt/β-catenin in EMT, as well as its contribution to the development of subretinal fibrosis in the context of nAMD remain poorly elucidated.

The Wnt signaling pathway plays a crucial role in both development and various diseases. Overall, two primary Wnt signaling pathways can be distinguished: the canonical Wnt/β-catenin pathway, which functions through β-catenin as a transcriptional cofactor, and the non-canonical, β-catenin-independent Wnt pathway, encompassing the Wnt/Ca²⁺ pathway and the planar cell polarity (PCP) pathway [17]. Wnt5a, a highly conserved non-canonical Wnt ligand, functions as a secreted signaling molecule that plays pivotal roles in regulating PCP, convergent extension, and epithelial–mesenchymal interaction during embryonic morphogenesis [18]. Although Wnt5a primarily signals through the non-canonical Wnt signaling pathways by binding to various members of the Frizzled (FZD)- and receptor tyrosine kinase-like orphan receptor (ROR)-family receptors, thereby initiating intracellular signaling cascades, under certain circumstances, it also possessed the capacity to activate the canonical Wnt signaling pathway [19]. This activation led to stabilization of β-catenin and modulation of gene transcription regarding cell proliferation, survival, differentiation, and migration [20]. In the field of cancer biology, the role of Wnt5a exhibits intricate and context-dependent characteristics. Emerging evidence suggests that the dysregulated activation or inhibition of Wnt5a signaling remains an important event in cancer progression, exerting both oncogenic and tumor suppressive effects depending on the availability of key receptors, thereby highlighting the paradoxical role of Wnt5a across different cancers [18]. The involvement of Wnt5a in regulation of cancer cell invasion, metastasis, metabolism and inflammation renders it a subject of intense research in oncology [21]. Furthermore, Wnt5a has been widely recognized as an essential factor in tissue repair and regeneration. For example, in some instances, Wnt5a acting as a particularly attractive growth factor stimulates tissue regeneration (such as colonic crypt [22], osteochondral regeneration [23] and cartilage interface integration [23]) and wound healing [24]. While in other cases, the upregulation of Wnt5a has been implicated in aggravating a variety of tissue fibrosis and scar formation, such as cardiac fibrosis under pressure overload [25], myocardial fibrosis following myocardial infarction [26], atrial fibrosis [25], keloid scarring caused by aberrant genetic activation [13], idiopathic pulmonary fibrosis [27], renal fibrosis [28], and liver fibrosis [29]. However, the specific involvement of Wnt5a in the EMT process of RPE cells and subretinal fibrosis formation in nAMD remains elusive.

Therefore, the objective of this study to elucidate the molecular mechanisms of Wnt5a, a non-canonical Wnt ligand, and its crosstalk with the canonical Wnt signaling pathway involved in EMT of RPE cells, as well as its contribution to subretinal fibrosis progression in nAMD. Additionally, we aim to identify potential therapeutic targets for preventing or halting fibrotic progression in nAMD patients, thereby preserving visual function and improving the quality of life of individuals affected by this devastating disease.

Subretinal fibrosis, as the most prevalent natural sequelae of nAMD, causally impacts the photoreceptors, RPE, and choriocapillaris, thereby leading to irreversible central vision loss [38]. Despite extensive research on the pathogenesis of nAMD, the molecular mechanisms underlying subretinal fibrosis remain insufficiently characterized. Although the involvement of canonical Wnt/β-catenin activation in subretinal fibrosis has been reported [12, 39], the role of the non-canonical ligand Wnt5a in EMT and subretinal fibrosis remains obscure.

In the present investigation, we observed a significant upregulation of Wnt5a expression in both in vivo and in vitro models simulating subretinal fibrosis or EMT (Fig. 3A and Fig. 4B). Above molecular and phenotypic changes were evidently inhibited by inactivating β-catenin using FH535 (Figs. 2, 3, 5 and 6, and Additional file 1: Fig. S3), thereby corroborating the indispensable role of the canonical Wnt/β-catenin signaling in facilitating EMT [12], as well as its transcriptional effect on the non-canonical ligand Wnt5a. The action of the small molecule FH535 is achieved by interfering with the nuclear translocation of β-catenin in the canonical Wnt signaling pathway, thereby impeding its entry into the cell nucleus and subsequent binding with transcription factors, ultimately resulting in the reduction of downstream target gene expression [40], such as Wnt5a [41]. Furthermore, the inhibition of Wnt5a through Box5 or shRNA could notably reverse the activation of β-catenin, EMT, and subretinal fibrosis both in vivo and in vitro (Figs. 3, 7 and 8). The causative role of Wnt5a in promoting EMT and its positive regulation on β-catenin could be further reinforced in a Wnt5a agonist assay, demonstrating that Wnt5a alone is sufficient to activate β-catenin and initiate EMT (Fig. 6). Based on the aforementioned data, activation of Wnt5a and β-catenin establishes a mutually reinforcing interaction loop involving "Wnt5a/β-catenin" significantly promoting the EMT process and subsequent subretinal fibrosis (Fig. 8I). However, the activation sequence of these two pathogenic factors in the development of EMT or fibrosis remains unclear.

The expression of Wnt5a in myofibroblasts was reported to be induced by the profibrotic factor TGFβ, playing a crucial role in the regulation of fibrotic matrix proteins induced by TGFβ in the context of liver fibrosis. This effect was reversed after silencing Wnt5a [42]. It was also reported that TGFβ-mediated regulation of Wnt5a was confirmed in primary cells, with Smad binding sites identified within the Wnt5a promoter [43, 44]. Thus, it is possible to propose that TGFβ1 mediates the upregulation of Wnt5a transcription, promoting the activation of β-catenin, which ultimately contributes to subsequent EMT and subretinal fibrosis. The precise mechanism underlying TGFβ1-induced upregulation of Wnt5a requires further exploration in the context of subretinal fibrosis secondary to nAMD.

The intricate Wnt signaling system plays a pivotal role in regulating the processes implicated in the pathogenesis of EMT as well as various fibrotic diseases. Although the canonical and non-canonical Wnt signaling pathway are two independent cellular signaling pathways, they can interact and cross-regulate each other in cell development, tissue repair, and disease processes [45]. Wnt5a is recognized to stimulate non-canonical signaling independent of β-catenin [46]. Nonetheless, this non-canonical Wnt ligand can trigger the activation of the canonical Wnt signaling pathway [47, 48]. For example, Wnt5a overexpression in certain cases resulted in an enhancement of TCF/ lymphoid enhancer-binding factor (LEF) transcriptional activity and direct activation or cooperation with the Wnt/β-catenin pathway [49]. Additionally, one study demonstrated that myometrial cells, acting as niche components, regulate the self-renewal activity of endometrial mesenchymal stem-like cells (eMSCs) via Wnt5a-dependent activation of the Wnt/β-catenin signaling pathway [49]. Meanwhile, in specific circumstances, the activation of the canonical Wnt/β-catenin signaling pathway can induce the binding of the transcription factor TCF/LEF to the promoter region of Wnt5a gene, thereby enhancing its transcription [50]. One of the limitations of this study is that we have not yet determined the direct transcriptional impact of β-catenin on Wnt5a, which may necessitate the use of a ChIP-qPCR technique in our future work.

After proposing and validating the concept of the "Wnt5a/β-catenin" interaction loop, we also discovered that FH535, Box5, and Wnt5a shRNA displayed comparable inhibitory effects on subretinal fibrosis and EMT (Figs. 3 and 5–8). In other words, either the blockage of Wnt5a or β-catenin could disrupt this interaction loop (Fig. 8). Notably, although no apparent changes were observed in the expression of Wnt3a and Wnt1 both in vivo and in vitro (Figs. 3A and 4A), it is important to acknowledge that the canonical Wnt/β-catenin signaling pathway is not solely regulated by these ligands alone. Therefore, we cannot disregard the potential involvement of additional Wnt ligands in β-catenin activation within the scope of this study. One research validated our findings by demonstrating elevated level of Wnt5a and unaltered expression of wnt3a in laser-treated RPE compared to control RPE, as determined through RT-qPCR analysis. However, this study also identified significant increases in other Wnt ligands, such as Wnt1, Wnt2b, Wnt3, Wnt7a, Wnt7b, and Wnt10b, which might also contribute to the activation of β-catenin [12]. Furthermore, we cannot exclude the potential involvement of the Wnt5a-mediated non-canonical Wnt signaling activation in the pathogenesis of EMT [51, 52] and subretinal fibrosis, which also merits further study.

The underlying mechanisms of subretinal fibrosis secondary to nAMD are complicated and remain incompletely elucidated. Extensive literature corroborated the involvement of additional factors, such as pericyte-myofibroblast transition (PMT) [53], EndMT [54], activated microglia [55], macrophages [56] and Müller glia [57] in the laser-induced CNV mouse model, which contribute to the accumulation of differentiated myofibroblasts and deposition of ECM, ultimately leading to fibrosis formation in nAMD. Therefore, the contribution of Wnt5a or Wnt/β-catenin in the MT of other cell types during subretinal fibrosis remains to be elucidated. Notably, the critical role of EMT undergone by RPE cells during subretinal fibrosis has been extensively recognized and corroborated [58, 59]. The intricate molecular networks governing EMT exhibit interconnectedness and reciprocal interactions. This process involves the activation of not only the Wnt signaling but also the canonical TGF-β-Smad signaling, as well as the non-canonical TGF-β signaling pathways, including phosphatidylinositol-3-kinase/Akt (PI3K/Akt), Rho/Rho kinase (ROCK), mitogen-activated protein kinases (MAPKs), Jagged/Notch, Hedgehog (Hh), and other signaling pathways [60]. It is proposed that patients with nAMD who exhibit an inadequate response to anti-VEGF therapy could potentially benefit from a combined therapeutic approach that targets both VEGF and the EMT process of RPE cells within the fibrovascular scar, thereby achieving a more favorable prognosis.

Subretinal fibrosis often occurs concurrently with neovascularization in nAMD, as exemplified by the co-localization of collagen I and IB4 in the CNV lesions visualized by confocal microscope in Fig. 2C and D. The early intervention of pathological neovascularization potentially mitigates the infiltration of immune cells and inflammatory response, thereby facilitating the prevention of subsequent subretinal fibrosis [61]. Despite its seemingly dual-functionality in physiological angiogenesis, Wnt5a signaling is essential for pathological angiogenesis [62]. Experimental studies on various vascular eye diseases, including nAMD, diabetic retinopathy (DR), retinopathy of prematurity (ROP), and corneal neovascularization, indicate that an aberrantly heightened Wnt signaling pathway is among the potential causative factors for pathological ocular neovascularization [17]. The activation of the Wnt pathway has been reported in laser-induced CNV models, indicating its pathogenic role in CNV. Inhibition of Wnt signaling through the use of an anti-LRP6 antibody demonstrates therapeutic potential for treating CNV [63]. Therefore, it is difficult to assert that the attenuation of subretinal fibrosis through inhibition of Wnt5a/β-catenin signaling pathway partially stemmed from its inhibitory impact on CNV, which merits further investigation. In order to mitigate the confounding effect of CNV and subretinal fibrosis, we administered the inhibitor injection at a later time point (Day 14) after laser induction in mice. This decision was based on the report that neovascularization in this mouse model reaches its peak stage at day 7 and progresses gradually [64]. Remarkably, our findings demonstrate that inhibition of Wnt5a/β-catenin pathway remained highly effective in combating fibrosis and EMT during the late stage (Day 21) of this mouse model (Fig. 3).

These findings and discussion should be taken into account when developing specific drugs for the treatment of subretinal fibrosis, as only targeting the mesenchymal transition and activation of multiple retinal cells is insufficient. Instead, a combination approach with inhibition of pathological neovascularization should be employed to effectively inhibit both the neovascularization and subretinal fibrosis. In developed countries, a significant proportion of vision impairment arises from abnormalities in the retinal or choroidal vasculature. These conditions, encompassing ailments such as nAMD [2], DR [65], ROP [66], and neovascular glaucoma [67], share the hallmark features of macular edema, retinal and vitreous hemorrhage, or the formation of fibrovascular scars. A shared underlying mechanism in these diverse conditions lies within the retina's response to damage, eventually setting off an ongoing cycle of chronic wound healing that may culminate in fibrotic tissue formation and, in the worst cases, irreversible vision impairment [68]. Therefore, the mechanistic insights gleaned from the nAMD study and the combined inhibition of subretinal fibrosis and neovascularization mentioned above has important clinical significance in developing a treatment for the prevention of the subretinal fibrosis and neovascularization in other ocular diseases.

In summary, our research findings unequivocally demonstrated the pivotal role of the positive interaction loop of Wnt5a/β-catenin in inducing EMT of RPE cells, facilitating cell migration, and contributing to subretinal fibrosis in nAMD, as illustrated in the schematic diagram presented in Fig. 8I. Furthermore, exploring the involvement of Wnt5a in the pathogenesis of AMD-related conditions beyond EMT, such as EndMT, PMT, and glial cell activation, holds promising potential for further exploration. Moreover, considering the pivotal role of the Wnt signaling pathway in pathological neovascularization across diverse diseases, it is plausible to speculate that Wnt5a/β-catenin might serve as a key role in driving angiogenesis of RNV and CNV secondary to nAMD. Our findings shed light on one of the intricate molecular pathways underlying subretinal fibrosis and offer novel insights into potential therapeutic strategies for the management of the fibrovascular diseases by targeting Wnt5a/β-catenin-mediated EMT. In future, when developing novel therapies for treating nAMD, these findings should be taken into careful consideration, aiming to employ minimal drug combinations to address the pathological neovascularization, subretinal fibrosis, inflammation, and the various cellular processes involved in EMT or activation within the retina. This approach seeks to optimize the therapeutic strategy for the management of nAMD and other fibrovascular diseases.