Introduction

A well-organized vascular system is crucial for the proper development of most tissue and organ morphogenesis^[1,2]. Consequently, angiogenic defects have been associated with several congenital human diseases^[3,4]. Familial exudative vitreoretinopathy (FEVR) is a heritable vitreoretinopathy characterized by defective retinal angiogenesis and various complications, such as extensive neovascularization, exudation, retinal folds and detachments, and vision loss⁵. Based on an ocular screening of over sixty thousand Chinese newborns, the estimated prevalence of FEVR is around 0.46%⁶. Due to the significant genetic heterogeneity, FEVR displays all Mendelian forms of inheritance: autosomal dominant (AD), autosomal recessive (AR), or X-linked recessive (XR)^[7-9]. To date, mutations in 17 genes and 1 locus have been identified to cause FEVR, including Norrin (NDP)¹⁰, frizzled 4 (FZD4)¹¹, low-density lipoprotein receptor related protein 5 (LRP5)^[12,13], low-density lipoprotein receptor-related protein 6 (LRP6)¹⁴, tetraspanin-12 (TSPAN12)^[15,16], α-catenin (CTNNA1)¹⁷, β-catenin (CTNNB1)^[18-20], p120-catenin (CTNND1)²¹, zinc finger protein 408 (ZNF408)²², kinesin family member 11 (KIF11)²³, atonal homolog 7 (ATOH7)²⁴, exudative vitreoretinopathy 3 (EVR3)²⁵, integrin-linked kinase (ILK)²⁶, jagged canonical Notch ligand 1 (JAG1)²⁷, discs large MAGUK scaffold protein 1 (DLG1)²⁸, transforming growth factor beta receptor 2 (TGFBR2)²⁹, sorting nexin 31 (SNX31)³⁰, and ER membrane protein complex subunit 1 (EMC1)³¹. Nevertheless, these mutations can explain only approximately 50% of FEVR cases^[32,33].

Calcyphosine like (CAPSL) is a protein-coding gene whose function remains elusive. In 2019, Julia Schreml pointed out that CAPSL is a promising candidate gene for multiple symmetric lipomatosis (MSL), a rare adipose tissue disorder of largely unknown etiology³⁴. Nevertheless, the molecular mechanisms by which CAPSL regulates cell processes and signaling cascades have yet to be elucidated.

MYC gene family is one of the most widely investigated cancer-causing genes, being implicated in the formation, maintenance, and progression of several different cancer types, such as breast cancer, lymphoma, and prostatic cancer^[35-38]. c-MYC is a master regulator in MYC signaling, which participates in cell metabolism, proliferation, differentiation, and migration³⁹. The important role of c-Myc in vascularization and angiogenesis during tumor development has already been reported in mouse models^[40,41]. Vascular endothelial cell (EC)-specific knockout of c-Myc in mice leads to impaired vascular expansion, thinned and poorly branched vascular, and reduced endothelial proliferation⁴². Conversely, EC-specific c-Myc over-expression caused sustained vascular outgrowth, increased endothelial cell proliferation, and vessel density⁴². These studies indicate that c-MYC is a regulating factor in coordinating endothelial cell behavior and vessel development.

In the current study, from a large cohort of FEVR patients, we identified a missense mutation and a stop-gain mutation in the CAPSL gene from four FEVR patients by whole exome sequencing (WES) analysis. Furthermore, the endothelial cell (EC)-specific Capsl knockout mouse model exhibited FEVR-like retinal vascular defects, emphasizing the indispensable role of Capsl for the retinal vascular architecture. Knockdown of CAPSL led to compromised proliferation of stalk ECs and retarded migration of tip ECs by regulating the MYC signaling axis. Collectively, this study not only identifies CAPSL as a candidate gene for FEVR, but also presents that a compromised MYC signaling axis might constitute a previously unrecognized factor contributing to FEVR.

Results

WES analysis revealed two CAPSL variants in FEVR patients

To evaluate the causative variants that account for FEVR, we applied WES on genomic DNA samples from FEVR families. Sanger sequencing was further applied to validate the variants and genotype-phenotype co-segregation analysis. Thus, two novel candidate variants in the CAPSL gene (NM_144647) were identified in patients with FEVR, and the variants were predicted to be pathogenic by Mutation taster, Polyphen-2, and PROVEAN (Table S1). In family 3036, a heterozygous c.88C>T (p.R30X) variant was identified in an infant (age 0-5 years old) (Figure 1A). His father was a heterozygous carrier and manifested vascular defects in the retina, including peripheral neovascularization area and leakage by fundus fluorescein angiography, whereas the wild-type mother showed normal vision (Figures 1B and D). The affected amino acid is highly conserved among different species (Figure 1C). The proband (II:1) in the 3104 family has been diagnosed with FEVR and identified with a heterozygous variant c.247C>T (p.L83F) (Figure 1A). His father has also been identified as a carrier of this heterozygous variant, manifested with FEVR symptoms, according to the medical records.

Figure 1.

To determine the pathogenesis of CAPSL variants for FEVR, we first investigated the effect of variants on the expression of CAPSL. Wild-type and mutant coding sequences of CAPSL were introduced to the GFP-tagged pcDNA3.1 vector, and an IRES2 box was inserted to prevent the fusion expression of CAPSL and GFP. Immunoblot analysis of cells transfected with plasmids revealed that the CAPSL-R30X protein was undetectable, and the L83F variant resulted in a considerably reduced protein level (Fig.1E and F). Conversely, neither variant influenced the mRNA levels of CAPSL (Fig.1G). Thus, these variants may cause FEVR by impairing CAPSL expression.

Depletion of Capsl in vascular endothelial cells caused FEVR-liked phenotypes

We, generated an inducible endothelial (EC)-specific Capsl-knockout mouse by breeding mice carrying a loxp-flanked allele of Capsl with tamoxifen-inducible Pdgfb-iCreER⁴³ transgenic animals to determine whether depletion of Capsl in mouse endothelial cells could cause FEVR-like phenotypes (Fig.S1A). Capsl^loxp/loxp, Pdgfb-CreER mice (hereafter named Capsl^iECKO/iECKO) and their control littermates (Capsl^loxp/loxp and Pdgfb-CreER, hereafter named Ctrl) were induced by consecutive intraperitoneal injection of 50 µg tamoxifen per pup at postnatal day 1 (P1) to P3 before sacrifice (Fig.S1B and C). Capsl^iECKO/iECKO mice showed indistinguishable overall appearance compared to their Ctrl littermates. The efficiency of EC-specific deletion of Capsl was confirmed by real-time quantitative PCR (RT-qPCR) and western blot analysis of P35 mouse lung tissues (Fig.S1D). The results showed a pronounced reduction in CAPSL expression, at both mRNA and protein levels, in Capsl^iECKO/iECKO mice compared to Ctrl littermates.

To assess the role of CAPSL in the growth and patterning of retinal vasculature, we employed flat-mount analysis to visualize the retinal vessels at different postnatal ages. As expected, the radial vascular growth, as well as vessel density and vascular branching, are dramatically reduced in P5 Capsl^iECKO/iECKO retina relative to Ctrl mice (Fig.2A and D-F). The number of retinal vascular tip cells, which leads EC migration toward high VEGFA by extending filopodia at the angiogenic front⁴⁴, was considerably decreased and sparsely distributed in Capsl^iECKO/iECKO retinae compared to that of control littermates (Fig.2B-C and G-H).

Figure 2.

The onset of mouse retinal vascular development occurs postnatally at P0, coinciding with the regression of the hyaloid vasculature^[45,46]. Reproducing the hyaloid phenotypes observed in the FEVR-associated mouse model, the hyaloid vasculature in Capsl^iECKO/iECKO at P10 displayed delayed regression in comparison to the Ctrl littermates (Fig.3A). Following the expansion of the superficial vessel plexus at around P8, retinal vessels extend perpendicularly into the deep retina, forming secondary and deep third capillary layers⁴⁷. To investigate whether the depletion of Capsl hinders deep capillary vascular development, we conducted immunostaining on P10 retinal sections. The results revealed a delayed development of deep retinal vessels in Capsl^iECKO/iECKO mouse retinae compared to that of Ctrl (Fig.3B). In addition, immunostaining of P14 flat-mount retina confirmed an incomplete architecture of the deep vascular layer in Capsl^iECKO/iECKO mouse (Fig.3C).

Figure 3.

The development of retinal vasculature is a complex process encompassing not only the coordinated behavior of endothelial cells but also the intricate interactions among different cell types and cytokines within the retina⁴⁸. Astrocytes form a mesh as a template prior to vascular growth and guide angiogenic expansion by secreting gradient distributed VEGFA^[49-51]. Abnormal activation of astrocytes and subsequent inflammatory responses have been observed in multiple FEVR mouse models^{[15,17,47,52]}. To explore whether the gradient and expression level of VEGFA were affected by the depletion of Capsl, we conducted immunostaining of VEGFA on the retina. However, the results showed no alteration of VEGFA (Fig.S2A). Meanwhile, the activation state and the pattern of the vascular template for EC outgrowth formed by astrocytes remained normal, indicated by GFAP (Fig.S2B). Pericyte recruitment is necessary for vessel stability and maturation⁵³. The immunostaining of Desmin or NG2 staining exhibited no difference between Ctrl and Capsl^iECKO/iECKO mice (Fig.S2C). To assess the integrity of the retinal vascular barrier, we performed the intraperitoneal injection of biocytin-TMR (869 Da), a small-molecular-weight fluorescent tracer⁵⁴. The results showed no notable leakage in the Capsl^iECKO/iECKO retina, suggesting the intact blood barrier despite the depletion of ECs-Capsl (Fig.S2D). The extracellular matrix (ECM) coordinates the behavior of ECs by regulating cell-cell communication processes⁵⁵. The deficiency and abnormal accumulation of ECM can both disrupt the adhesion, proliferation, and migration of the endothelial cells, leading to defects in blood vessel architecture^[56-58]. We thus analyzed the extracellular matrix accumulation using fibronectin 1, and the result showed normal deposition of ECM (Fig.S2E). In conclusion, endothelial cell-specific knockout of Capsl may mainly lead to defects in retinal vascular ECs rather than other vascular cells.

CAPSL controls vascular proliferation and migration

Considering the compromised vascular progression and decreased vessel density in Capsl^iECKO/iECKO mouse retinae, we first performed the EdU assay on retina flat-mounts co-stained with ETS transcription factor ERG (endothelial cell nuclear marker) to assess the proliferation of ECs. The results showed a strongly reduced proliferation of Capsl-defective ECs compared to that of Ctrl (Fig.4A). As previously reported, the regressed retinal vessels exhibited sleeve-like positivity for basement matrix component Collagen IV and negativity for Isolectin IB4⁵⁹. To compare the vascular regression between Capsl^iECKO/iECKO and Ctrl retinae, we conducted co-staining of the Collagen IV and IB4. The results revealed that depletion of Capsl leads to an increased vascular regression, evidenced by a significant increase in collagen IV basal membrane sleeves lacking IB4 both in the capillary plexus and angiogenic front (Fig.4B).

Figure 4.

Additionally, considering the pivotal role of tip cells in the direction of vascular development, we then asked whether depletion of Capsl impairs the polarity of tip cells by analyzing the morphology of cell nuclei at the angiogenic front. This analysis was based on the observation that the nuclei of actively migrating tip cells are elliptical and exhibit increased sphericity (decreased ellipticity) in static tip cells^[60,61]. In Ctrl mice retinae, the nuclei of tip ECs were predominantly elliptical, pointing towards the avascular area (Fig.4C). Conversely, in the retinae of Capsl^iECKO/iECKO mice, the nuclei of tip ECs appeared more spherical and did not orient towards the avascular area (Fig.4C). These findings suggest that CAPSL is critical for EC proliferation, vascular regression, and cell polarity.

Endothelial CAPSL is required for cell polarity and filopodia/lamellipodia formation

To investigate the functional mechanism and signaling cascade by which CAPSL regulates EC behaviors, we performed in vitro experiments on primary human retinal microvascular endothelial cells (HRECs). Stable CAPSL knockdown HRECs were generated using the lentivirus-mediated shRNA system. RT-qPCR analysis showed that the mRNA level of CAPSL was significantly decreased (Fig.S1E). In line with the overall reduced vascular coverage in Capsl^iECKO/iECKO mice retina, CAPSL-knockdown HRECs (shCAPSL-ECs) exhibited considerably impaired tube formation in vitro (Fig.5A). EdU labeling also revealed decreased cell proliferation of shCAPSL-ECs compared to that of shCtrl-ECs (Fig.5B). We further conducted flow cytometry to determine whether cell cycle arrest occurred upon depletion of CAPSL in HRECs. The redistribution of cell cycle phases indicated that the cell cycle was arrested during the G0/G1 phase in the shCAPSL-ECs (Fig.5C-D). To further investigate the role of CAPSL in EC migration, we performed the scratch-wound assay in confluent shCAPSL-ECs and shCtrl-ECs, as previously reported^[61,62]. Nine hours after scratching, ECs were co-stained with GM130 and phalloidin to visualize the Golgi apparatus and the cytoskeleton. The results indicated that the nuclei of shCAPSL-ECs were more spherical than those of shCtrl-ECs (Fig.5E-F). Given that the axial polarity can reflect the direction and migration of ECs^[63,64], we further measured the cell polarity (nucleus-to-Golgi polarity axis) of ECs at the wound edge^[61,62]. As a result, the Golgi apparatus of shCtrl-ECs was predominantly positioned around the wound edge, while that of shCAPSL-ECs showed more random positioning⁶⁵ (Fig.5H).

Figure 5.

At the angiogenic front, the dynamic behaviors of tip cells largely rely on the rearrangement and organization of the actin cytoskeleton⁶⁶. Tip cells extend actin-driven filopodia and lamellipodia to probe chemotactic guidance cues, providing direction and migration for the developing vascular network^[67-69]. Notably, shCAPSL-ECs exhibited fewer filopodia, smaller lamellipodia, and significantly impaired wound closure compared to shCtrl-ECs at the wound scratch edge (Fig.5I). Also, as expected, shCAPSL-ECs exhibited significantly impaired wound closure at the same time point after scratching relative to shCtrl-ECs (Fig.5J). Small Rho GTPases, including CDC42 and Rac, are essential for lamellipodial extension and cell migration^[70,71]. Western blot analysis uncovered markedly attenuated expression of CDC42, RHOA, and Rac1 (Fig.5K). Moreover, both MYL9 and phosphorylated MYL9 at Ser19, which are crucial regulators of contractile force, were also drastically reduced in shCAPSL-ECs compared to those of shCtrl-ECs (Fig.5K). These results indicated that CAPSL regulates EC polarity, filipodia/lamellipodia formation, and migration by modulating the actomyosin cytoskeleton.

CAPSL regulates endothelial cell proliferation through the MYC signaling axis

Given the fact that FEVR is predominantly characterized by the compromised Norrin/β-catenin signaling pathway as its main causative factor^{[10,15,17,18]}, we first asked whether loss of CAPSL function causes FEVR through Norrin/β-catenin signaling. TopFlash reporter gene system was used to determine the activity of the Norrin/β-catenin signaling pathway. Knocking down CAPSL expression in human embryonic kidney 293 (HEK293) SuperTopFlash (STF) cells led to similar luciferase activity, compared to cells transfected with shCtrl (Fig.S3A). Furthermore, transfection of shRNA-resistant CAPSL plasmid (WT, R30X, or L83F) in CAPSL-knockdown 293STF cells resulted in no significant difference in luciferase activity (Fig.S3B). These results indicated that CAPSL was not a major player in Norrin/β-catenin signaling to regulate retinal vascular development.

We next performed unbiased transcriptomic and proteomic analyses on the control and CAPSL-knockdown HRECs to explore the underlying mechanism by which CAPSL regulates EC proliferation and migration. RNA sequencing (RNA-seq) identified 1961 up-regulated genes and 2205 down-regulated genes in shCAPSL-ECs compared to shCtrl-ECs (logFC>0.58, Pvalue<0.05), 1342 up-regulated genes and 1218 down-regulated genes in ECs overexpressing CAPSL (LentiOE-ECs) compared to Ctrl-ECs (Fig.6A). Gene set enrichment analysis (GSEA) on transcriptomic data showed multiple down-regulated signaling axes upon the depletion of CAPSL, including E2F targets (NES=-3.2944438, P value=0), G2/M checkpoints (NES=-3.2851038, Pvalue=0), MYC targets (NES=-2.8795638, P value=0), and mitotic spindle (NES=-2.646523, Pvalue=0) (Fig.6B-C). In line with this, overexpression of CAPSL in HRECs also caused a significant increase in E2F targets (NES=2.7765026, P value=0), G2/M checkpoints (NES=2.6528199, Pvalue=0), and MYC targets (NES=2.6528199, Pvalue=0) (Fig.6D-E). Meanwhile, proteomic profiling showed similar expression patterns with that of transcriptomic profiling upon knockdown or overexpression of CAPSL (Fig.6F-I). Furthermore, we applied correlation analysis on the transcriptomic and proteomic data, which divided the genes into 9 quadrants (Fig. 6J). GSEA analysis was performed on genes in the third and seventh quadrants, which were regulated at the transcriptional level (Fig. 6K). Interestingly, the down-regulated signaling pathways are highly consistent with those observed in GSEA analysis of both transcriptomic and proteomic analyses (Fig.6K). Given that the c-MYC was an established driver of cell growth and migration⁷², and that the genes in the MYC signaling were down-regulated upon CAPSL-depletion (Fig.6L), we speculated that CAPSL might play a pivotal role in MYC signaling in HRECs.

Figure 6.

We next incorporated an analysis of transcriptome profiling of human umbilical vein endothelial cells (HUVECs) infected with lentiviruses encoding control-or c-MYC-targeting gRNAs (GSE161815)⁷³. Intriguingly, GSEA analysis revealed that ablation of cMYC in HUVECs led to compromised E2F targets, MYC targets, and G2/M checkpoint signaling activity (Fig.7A-B), largely consistent with those observed in the absence of CAPSL. This also suggests that cMYC might play a role in the upstream regulation of E2F targets and G2/M checkpoint signaling. Venn diagram analysis of down-regulated genes (LogFC<-0.585, P<0.05) in CAPSL-depleted HRECs and c-MYC-depleted HUVECs revealed a large proportion of shared genes, indicative of similar transcriptional regulatory patterns between CAPSL and cMYC (Fig.7C). Additionally, these genes are predominently enriched in the MYC targets signaling (Fig.7D-F).

Figure 7.

We, therefore, asked whether CAPSL regulates c-MYC expression. Notably, the depletion of CAPSL significantly decreased the c-MYC protein level, rather than the mRNA level in HRECs (Fig.7G). Subsequently, western blot analysis revealed that expression levels of c-MYC targets, such as MCM family, E2F1, CyclinD family, and PCNA, were decreased in the absence of CAPSL in HRECs (Fig.7H-I). Taken together, these findings suggest c-MYC as a potential functional regulator downstream of CAPSL.

Discussion

Over the last decades, major progress has been made in understanding the molecular mechanisms underlying FEVR, and 17 FEVR-causing genes have been identified^[10-31]. However, the pathogenic genes and mechanisms underlying approximately 50% of clinical cases remain elusive^[20,21]. In this study, we report a new FEVR candidate gene CAPSL and uncover the pivotal roles of CAPSL in retina vascular development. The EC-specific Capsl-knockout mouse model exhibited FEVR phenotypes, including delayed vascular progression, retarded hyaloid vessel regression, and decreased vessel density. Deficiency of CAPSL in ECs resulted in compromised cell proliferation and defective EC migration, providing insights into the regulatory roles of CAPSL in retina vascularization (Fig.2-5).

During retinal development, angiogenesis comprises various processes such as endothelial cell (EC) specification, adhesion, proliferation, migration, and pruning⁴⁷. The regulation of angiogenesis entails the control of EC behaviors and the interactions between ECs, extracellular matrix (ECM), and other types of retinal cells⁶⁵. In this study, we observed impaired radial expansion and vertical invasion, as well as the increased pruning of the retinal vascular ECs upon inactivating endothelial Capsl (Fig.2-3). However, neither the expression level nor the gradience of VEGFA was disturbed in Capsl^iECKO/iECKO mice, which could be attributed to the intact blood barrier in the Capsl^iECKO/iECKO retina (Fig.S2). In addition, the loss of Capsl resulted in the presence of more spherical tip cell nuclei, which were directed more randomly to the avascular area (Fig.4). These results provide evidence that CAPSL acts as a novel regulator for collective EC behavior.

In vitro functional studies demonstrated that depletion of CAPSL impaired tube formation, EC proliferation ability, and EC polarity (Fig.5). Moreover, the formation of filopodia and lamellipodia is also disturbed by the depletion of CAPSL (Fig.5). Using the unbiased transcriptomic and proteomic sequencing, bioinformatic analysis, and western blot assay, we demonstrated that the defects in CAPSL affect EC function by down-regulating the MYC signaling cascade (Fig.6-7). c-MYC has been shown to be a critical mediator of anabolic metabolism, cell growth, and migration^[74,75]. Aberrant upregulation of the MYC signaling pathway is frequently observed in many types of cancers^[76,77]. Interestingly, the phenotypes of retinal vessels in Capsl^iECKO/iECKO mice resemble the vascular defects in c-Myc^iECKO/iECKO mice, exhibiting FEVR-like phenotypes, such as impaired vascular progression, decreased branch points, and reduced EC proliferation^[39,42].

The Norrin/β-catenin signaling pathway is a subset of Wnt signaling pathway that is specifically activated during vascular development of the central nervous system^[20,78]. Although several signaling pathways have been implicated in the pathogenesis of FEVR, variants in crucial components of Norrin/β-catenin signaling account for most of the FEVR cases^[15,17,18]. It is worth mentioning that c-MYC is widely recognized as a downstream target of Wnt signaling, which is transcriptionally regulated during Wnt activation^[79-81]. However, to our knowledge, there is a lack of literature reporting on the correlation between the MYC signaling pathway and FEVR. Interestingly, here we proved that the absence of CAPSL leads to a decrease in c-MYC protein level, which down-regulates MYC signaling without affecting Norrin/β-catenin signaling (Fig.7 and S3). Consequently, we propose c-MYC as a potential therapeutic target for FEVR, a hypothesis that warrants substantiation through subsequent investigations.

Additionally, the formation of filopodia and lamellipodia, which is reported to be intimately related to the actomyosin dynamics and the activity of small Rho GTPases such as CDC42^[82,83], is significantly reduced in CAPSL-defective HRECs. In our study, we confirmed that the knockdown of CAPSL led to a significant reduction of CDC42, which might be the causative mechanism behind the disturbed formation of filopodia and lamellipodia (Fig.5).

Taken together, our study demonstrated that CAPSL is potentially a novel candidate gene associated with FEVR. We also demonstrated that variants in the CAPSL gene, independently of canonical Norrin/β-catenin signaling, cause FEVR through inactivating MYC signaling, expanding FEVR-involved signaling pathway and providing a potential therapeutic target for the intervention of FEVR.

Methods

DNA sequencing of patients and controls

The study was approved by the Institutional Research Committees of Sichuan Provincial People’s Hospital and Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. Informed consent was obtained from all participants and from guardians of minors involved in genetic testing. The patients were clinically diagnosed with exudative vitreoretinopathy. Genomic DNA was extracted from the peripheral blood of FEVR families and control individuals using a blood DNA extraction kit (Qiagen, Germantown, Maryland, USA) according to standard protocol. Whole-exome sequencing (WES) was performed on DNA samples to identify candidate pathogenic genes for FEVR as previously described¹⁷. Family ID numbers 3036 and 3104 are not known to anyone outside the research group.

Mouse model and genotyping

All mice used in the study were on a C57BL/6J genetic background, and Capsl^loxp/+ mice were bred with Pdgfb-iCreER⁴³ transgenic mice to generate Capsl^loxp/loxp, Pdgfb-iCreER mice. Genomic DNA samples extracted from mouse tails were genotyped using PCR to detect loxp-flanked Capsl and Pdgfb-iCreER. Primer sets used for loxp were: 5’-GGCAGGTAAGATGGTGTC-3’ and 5’-TCTGTTTGTGGATCAATGTG-3’; and primer sets used for Pdgfb-iCreER were: 5’-GCCGCCGGGATCACTCTCG-3’ and 5’-CCAGCCGCCGTCGCAACTC-3’. All mice were maintained under specific pathogen-free conditions with standard rodent water and diet, both males and female mice were used for the analysis. Experiments involving animals were conducted in accordance with institutional guidelines and following the protocols approved by the Institutional Animal Care and Use Committee of Sichuan Provincial People’s Hospital.

Preparation of flat-mount retinae and retina sections

Enucleated eyes were fixed for 15 min in 4% paraformaldehyde (PFA) at room temperature, followed by immersion in PBS for the same duration. The retinae were then dissected and partially cut into four quadrants as previously described⁸⁴. The dissected retinal cups were post-fixed overnight and then stained as described below. As for retina sections, after fixation in 4% PFA for 2h, the enucleated eyes were dehydrated until they settled to the bottom of the tube with 30% sucrose and then embedded in Tissue Freezing Medium (Servicebio). Specimens were sectioned, cleaned, and stained as previously described⁸⁵.

Hyloaid vessel preparation

Neonatal eyes were fixed in 4% PFA for 2h before the cornea, lens, and iris were removed. The eye cup was immersed in 5% gelatin at 4 ℃ overnight. The solidified gelatin was extricated and heat melted on a glass slide, washed with warm water, air-dried, and imaged with DAPI staining.

Immunohistochemistry

All immunostainings of retinae were performed using littermates with similar body size and treated under the same conditions. The flat-mount retina and retina sections were permeabilized and blocked in PBS containing 5% fetal bovine serum and 0.2% Triton X-100 at room temperature for 2h. Next, they were rinsed in PBS and incubated in directly conjugated Isolectin-B4 Alexa 488 or Alexa 594 (1:200; Thermo Fisher, USA) and primary antibodies in a blocking buffer at 4℃ overnight. After three washes, the retinae were processed for multiple labeling or flat-mounted onto microscope glass slides with Fluoromount (Sigma-Aldrich, USA).

For in vivo analysis of cell proliferation, each pup was injected intraperitoneally with 200ug EdU 3h before the mice were euthanasia. Retinae were dissected and permeabilized as previously described. EDU-positive labeling was stained and detected by means of a Click-iT EDU Alexa Fluor-647 Imaging Kit ( C10640, Thermo Fisher Scientific, USA) according to the manufacturer’s instruction. For paracellular BRB integrity analysis, each pup was injected intraperitoneally with 2% fluorescent tracer, 5-(and-6)-tetramethylrhodamine biocytin, Biocytin-TMR (CAT#T12921, Thermo Fisher Scientific, USA) for 24 hours prior to sacrifice, followed by flat-mount preparation, blocking, and staining of Isolectin B4.

Images of stained flat-mounted retinae, retina sections, and HRECs were acquired using Zeiss LSM 800 confocal microscope (Thornwood, NY, USA) and processed with Zeiss processing software, Angio Tool, and Adobe Photoshop. The detailed information about antibodies used for immunofluorescence is listed in Table S2.

Cell lines and primary cell culture

Human retinal microvascular endothelial cells (HRECs) were purchased from Cell System and cultured in EBM™-2 media (CC-3156, Lonza, Switzerland) containing 5% fetal bovine serum, 0.4% hydrocortisone, 4% Hfgf-b, 0.1% VEGF, 0.1% R3-IGF, 0.1% ascorbic acid, 0.1% hEGF 0.1% GA-1000, and 0.1% heparin at 37°C in a 5% CO₂ incubator. HRECs at passages 3-7 were used for experiments.

HEK-293T were obtained from the American Type Culture Collection (ATCC) and cultured in DMEM (SH30023.01, Hycolon, USA) with 10% fetal bovine serum and cells were maintained at 37°C in a 5% CO₂ incubator.

Gene knockdown and overexpression strategies

HRECs were seeded on 6 cm dishes at the day before transfected with lentivirus carrying shRNA for human CAPSL (5’-AAGACCTTCGTGAAGTATA-3’, Genechem, Shanghai, China) or negative control shRNA (5’-TTCTCCGAACGTGTCACGT-3’) according to the protocol of the manufacture. Cells were incubated for at least 72h before used for further experiments.

For overexpression of exogenous protein in HEK-293T cells, cells were transfected with Lipofectamine-3000 (Invitrogen, USA) according to the manufacturer’s instructions.

Matrigel EC tube formation assay

HRECs transfected with the corresponding shRNA after 72h were counted and planted onto the matrigel as previously described⁸⁶. After 6h of incubation at 37℃ in a 5% CO₂ incubator, the images were captured under an anatomical lens (Carl Zeiss, Germany).

Immunofluorescence of HREC

After cell counting, the same number of HRECs transfected with shCtrl and shCAPSL were seeded on slides in 24-well plates. Cells were fixed with 4% PFA at room temperature for 15min. After being rinsed in PBS three times, cells were blocked in the blocking buffer at room temperature for 2h. The cells were labeled with primary antibodies overnight at 4℃. Following three washes in PBS, Alexa-Fluor-594 or Alexa-Fluor-647 conjugated antibodies were administered along with DAPI, and cells were incubated at room temperature for 1h. After three washes in PBS, stained cells were mounted on microscope glass slides with Fluoromount (Sigma-Aldrich, USA).

For in vitro analysis of cell proliferation, 20 µM EdU (Thermo Fisher Scientific, USA) was added into 96-well plated 3h before cells were harvested. EdU-positive cells were stained and detected with the Click-iT EDU Alexa Fluor-647 Imaging Kit (C10640, Thermo Fisher Scientific, USA) according to the manufacturer’s instruction.

Wound healing assay

72h after shRNA transfection, HRECs were seeded in 6-well plates with complete medium for 24h until grown to confluence. Plates were then coated with 10 µg/mL fibronectin for half an hour in 37 ℃ before cell culture. 200 µL pipette tips were used to generate wounds, and each well was washed twice with Utrasalin A (Lonza, Switzerland) to remove detached cells. Cells were then starved in the EBM-2 medium at 37°C in a 5% CO2 incubator. Images of the wounds were captured with an optical microscope at 0h, 12h, and 16h after the wounds were made, respectively. The wound closure areas were measured using ImageJ software.

For analyses of Golgi apparatus polarization, cells were maintained on 24-well glass slides and fixed with 4% PFA for 15 min at 9 h after cell migration was initiated.

RNA extraction and RT-qPCR

Total RNA was extracted from cultured HRECs or fresh lungs from mouse using TRIzol reagent (Life Technologies, USA) according to the manufacturer’s instructions. RNA concentration and quality were measured with NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA), and 1 µg total RNA was reverse transcribed using EasyScript One-Step RT-PCR SuperMix (TransGen Biotech, Beijing, China) following the standard protocol. RT-qPCR was performed using a 7500 Real-Time PCR System (Applied Biosystems, USA) with TransStart Tip Green qPCR Supermix (TransGen Biotech). At least three experiments were performed, and PCR reactions were performed in triplicate. GAPDH was used as a reference gene. The primer sets are listed in Table S3.

Protein extraction and western blotting

HRECs or lung tissue from mice were lysed in standard PIPA lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, pH 7.4) supplemented with Protease and Phosphatase Inhibitor Cocktail (Roche, USA). After sonication, centrifugation, and protein quantification, the supernatant was diluted in 2x SDS loading buffer. Equal amounts of protein (20 µg) were loaded and resolved on 10-15% SDS-polyacrylamide, and Western blotting was performed. After gels were transfected onto a Polubinylidene Difluoride (PVDF) membrane (GE Healthcare, USA), the membranes were incubated in a blocking buffer containing 8% skimmed milk (9999, Cell Signaling Technology) in TBST for 1h at room temperature on a rocking platform, followed by overnight incubation at 4℃ with the primary antibodies diluted in blocking buffer. Membranes were then washed in TBST three times and incubated in the blocking buffer containing secondary HRP-conjugated antibodies for 1h at room temperature. Protein blots were visualized with e-BLOT Touch Imager (e-BLOT, China) and blots were relatively quantified with ImageJ. GAPDH was used as an internal reference. Detailed information about antibodies used for western blot is listed in Table S4.

Luciferase Assays

Dual-luciferase assays were performed as previously described¹⁷. Briefly, 293STF cells were plated on 24-well plates at a confluency of ∼30%. The following day, the cells were transfected with corresponding plasmids with lip3000. 48 hours post-transfection, cells were harvested, and the lysates were used to measure Firefly and Renilla luciferase activity according to the manufacturer’s instructions (TransGen).

Quantification of retinal parameters

All quantifications were done using Zeiss processing software on high-resolution confocal images. Vascular progression was measured in a straight line from the optic nerve to the angiogenic front of the retinal plexus for each retina quadrant (n=6, each group). Vessel density and branch points were calculated with Angio Tool software from flat-mounted retinae (n=6, each group). Endothelial tip cell numbers were measured by counting endothelial sprouts at the angiogenic front of the entire vascular plexus of the same length (n=6, each group). Endothelial tip cell filopodia density was calculated with high-resolution images (60× objective) of 6 randomly selected angiogenic front fields from 6 retinae per genotype (n=6, each group). Endothelial cell proliferation was calculated by measuring the number of both EDU-positive and ERG-positive cells within retinae of the same size and same position at the edge of the vascular plexus (n=6, each group). Retinal pericyte coverage was quantified by comparing the NG2⁺ or Desmin⁺ pericytes with IB4⁺ vessels (n=6, each group). The nuclear ellipticity of the cells at the scratch edge was calculated by the distance of the nuclear long axis (Height) divided by the maximum vertical distance to the nuclear long axis (Width).

Transcriptomic profiling

Total RNA was extracted from shCtrl-ECs, shCAPSL-ECs, and LentiOE-ECs and quantified using NanoDrop2000 Spectrophotometer (Thermo Fisher Scientific, USA). RNA integrity was determined by Bioanalyzer RNA 6000 Nano assay kit (Agilent, China). RNA library construction was performed with Illumina TruseqTM RNA sample prep Kit and RNA sequencing was performed with Truseq SBS Kit (300 cycles) in BIOZERON (BIOZERON, China). RNA-Seq reads were performed using the TopHat software tool to acquire the alignment file, which was used to quantify mRNA expression and determine the differentially expressed genes. All gene expression values were changed to log2 values for further analysis, and a P value of less than or equal to 0.05 was considered to indicate significance. The gene set enrichment analysis (GSEA) was performed with the omicshare platform (https://www.omicshare.com).

Proteomic profiling

Total protein was extracted from shCtrl-ECs, shCAPSL-ECs, and LentiOE-ECs and quantified using a BCA protein quantification kit (Beyotime, China). Samples were sent to PTM Biolab (PTM Biolab, China) to perform proteomic profiling using 4D-Labelfree. All protein expression values were changed to log2 values for further analysis, with a corrected P value< 0.05. The gene set enrichment analysis (GSEA) was performed with the omicshare platform (https://www.omicshare.com).

Image acquisition and statistical analysis

Immunofluorescence images were obtained using a laser scanning microscope (LSM 900, Zeiss). Statistical analyses were performed with GraphPad Prism 8.0 software. Data were analyzed using the Student’s t-test and plotted as mean ± SD. At least three independent experiments were performed. Image J was used to compare results for the tube formation assay experiments and bands of immunoblot. Results were considered significant when *P<0.05, **P<0.01, ***P<0.001, or ****P< 0.0001. “ns” stands for “no significance.”

Study approval

This study was approved by the institutional review board of Sichuan Provincial People’s Hospital (approval number: 2014-009) and Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, and informed consent was obtained from all participants or legal guardians for minors. All animal studies were performed according to established ethical guidelines approved by the animal care committee of Sichuan Provincial People’s Hospital.

Competing interests

All authors declare that there are no competing interests.

Author contributions

XZ, ZY, and PZ conceived the research. WL, SL, MY, and JM performed the animal mouse model research and cell biology, transcriptome, and proteome data analysis. LL, PF, and QX performed animal breeding and genotyping. PZ recruited the participants and performed the WES study. LH performed the Sanger sequencing analysis. WL, SL, MY, and XZ wrote the manuscript, with input from other authors.

Funding

This study was supported by the National Natural Science Foundation of China (82371083, 82121003, 82101153, 82101160), the Department of Science and Technology of Sichuan Province (2023ZYD0172, 2023NSFSC1661, 2023NSFSC0036, and 2022ZYD0059), the CAMS Innovation Fund for Medical Sciences (2019-12M-5-032), Sichuan Intellectual Property Office (China) (No. 2022-ZS-0070), the Fundamental Research Funds for the Central Universities of Ministry of Education of China (ZYGX2022J023), and Sichuan Provincial People’s Hospital Postdoctoral fund (2022BH019). The funders had no role in the study design, data collection and analysis, or preparation of the manuscript.

Data Availability

All data produced in the present study are available upon reasonable request to the authors.