TMEM30A deficiency in endothelial cells impairs cell proliferation and angiogenesis

Phosphatidylserine (PS) is asymmetrically distributed in the exoplasmic and cytoplasmic leaflet of the plasma membrane (Halleck et al., 1999; Paulusma and Oude Elferink, 2005; Tang et al., 1996). Such an asymmetrical distribution of aminophospholipids is maintained by P4-ATPases called PS flippases, which are required for many biological processes, such as protein trafficking, apoptosis, neurodevelopment and sperm capacitation (Chen et al., 1999; Op den Kamp, 1979; Saito et al., 2004; Segawa et al., 2014; Seigneuret and Devaux, 1984). There are 14 P4-ATPases in the mammalian genome, and these flippases work together to maintain the phospholipid asymmetry between the outer and inner leaflets of the plasma membrane (Takatsu et al., 2017).

Mutations in several P4-ATPases have been reported to be associated with human diseases, indicating the importance of P4-ATPases. Mutations in ATP8B1 lead to progressive familial intrahepatic cholestasis and hearing loss (Bull et al., 1998; Stapelbroek et al., 2009). Mutations in ATP8B2 cause a severe neurological disorder that is characterized by cerebellar ataxia, mental retardation and disequilibrium syndrome (Cacciagli et al., 2010; Maríin-Hernández et al., 2016). Animal model experiments have provided additional evidence that P4-ATPases are critical for normal physiological function. Mice deficient in Atp11c have recently been reported to exhibit B-cell lymphopenia, anemia and intrahepatic cholestasis (Arashiki et al., 2016; Siggs et al., 2011a,b; Yabas et al., 2014; Yabas et al., 2011). Mutations in Atp8a2 result in axonal degeneration and retinal degeneration.

Proper folding and trafficking of most P4-ATPases except ATP9A and ATP9B is dependent on a β-subunit from the transmembrane 30 protein [TMEM30; also known as cell division control protein 50 (CDC50)] protein family, which includes TMEM30A, TMEM30B and TMEM30C (also known as CDC50A, CDC50B and CDC50C, respectively) (Bryde et al., 2010; Paulusma et al., 2008; Saito et al., 2004; Takatsu et al., 2011; Zhang et al., 2017). These genes encode N-glycosylated proteins with two transmembrane domains and play critical roles in the folding and function of P4-ATPases (Katoh and Katoh, 2004; van der Velden et al., 2010). Our previous studies have demonstrated the essential roles of the TMEM30A gene in retinal development, bile salt transportation and the survival of hematopoietic cells (Li et al., 2018; Liu et al., 2017; Zhang et al., 2017).

Angiogenesis is a critical process in vascular development and abnormal angiogenesis is implicated in many diseases, such as age-related macular degeneration (AMD), familial exudative vitreoretinopathy (FEVR) and cancer (Carmeliet and Jain, 2000; Folkman, 1971; Huang et al., 2016; Yang et al., 2006; Zhang et al., 2016). Angiogenesis is a complex process regulated by multiple signaling pathways. Understanding the network controlling angiogenesis is of great importance for unveiling various disease conditions (Chung and Ferrara, 2011; Potente et al., 2011). Given the essential roles of TMEM30A in organism development and no previous studies of its roles in vascular development, we set out to study the roles of TMEM30A in angiogenesis.

In the present study, we show that TMEM30A plays essential roles in the angiogenesis process. We found that knockdown of TMEM30A in primary human retinal endothelial cells (HRECs) led to impaired tube formation in vitro. Moreover, we examined retina from mice with an endothelial cell (EC)-specific deletion of mouse Tmem30a (Tmem30a^i△EC) and these showed a retarded vascular extension toward the peripheral zone and deep layer, a hyperpruned vascular network with decreased tip ECs at the vascular front, and blunted-end, aneurysm-like tip ECs with fewer filopodia. By using an induced global Tmem30a knockout mouse model (Tmem30a-iKO), we also proved that deletion of Tmem30a impaired the vessel barrier integrity. We further demonstrated that TMEM30A controls EC proliferation during sprouting angiogenesis by regulating VEGF-induced signaling.

To verify whether P4-ATPases were expressed in the vascular system, we carried out reverse transcriptional polymerase chain reaction (RT-PCR) experiments to study the expression of the 14 P4-ATPases in human HRECs obtained from Cell Systems. The results showed that several P4-ATPase-encoding genes, including ATP8B1, ATP8B2, ATP9A, ATP9B, ATP10D, ATP11A, ATP11B and ATP11C, were expressed in HRECs (Fig. 1A). The expression of multiple P4-ATPases in HRECs indicates that they could have redundant functions, making it challenging to dissect the role of each gene in angiogenesis. Proper folding of most P4-ATPases requires their β-subunit TMEM30A (Montigny et al., 2016), which is expressed in HRECs, human umbilical vein ECs (HUVECs) and human microvascular ECs (HMVECs) (Fig. 1B,C). To assess the roles of TMEM30A in HRECs, we used a lentivirus carrying TMEM30A shRNA to knockdown (KD) the expression of TMEM30A in HRECs (shTMEM30A HRECs). Analyses of TMEM30A RNA expression by RT-PCR and protein expression by western blotting showed that the expression of TMEM30A in these cells was reduced to ∼10% of that in control cells (Fig. 2A,B). Tube formation was impaired in TMEM30A-knockdown HRECs (Fig. 2C).

As Tmem30a global knockout and vascular EC-specific knockout mice, when induced by Tie2-Cre (Tie2 is an endothelial cell-specific promoter), are embryonic lethal (Kisanuki et al., 2001), loss of Tmem30a might impair vasculogenesis and angiogenesis. To assess the role of Tmem30a in ECs, we crossed a Tmem30a conditional knockout allele with Pdgfb-CreER (Claxton et al., 2008) to generate an inducible vascular EC-specific knockout line. In Pdgfb-CreER mice, the expression of an improved Cre gene fused to ERT2 is controlled by the mouse Pdgfb promoter. Mice homozygous for the Tmem30a floxed allele (Tmem30a^loxp/loxp) were mated to Tmem30a^loxp/+ Pdgfb-CreER mice to generate Tmem30a^loxp/loxp Pdgfb-CreER (named Tmem30a^iΔEC) and Tmem30a^loxp/loxp mice (WT), as shown in Fig. 3A. PCR methods were used to identify the genotypes of the offspring (Fig. 3B). Tmem30a^iΔEC and WT mice were induced by intraperitoneal injection of tamoxifen daily starting at post-natal day (P)2 for 3 consecutive days. The specificity and efficacy of Cre-mediated recombination were assessed by means of the tdTomato reporter. To do this, we crossed ROSA26-tdTomato reporter mice with Tmem30a^loxp/loxp Pdgfb-CreER mice to generate Tmem30a^loxp/loxp Pdgfb-CreER Rosa-tdTomato mice. In the presence of the Cre enzyme, the stop codon before the tdTomato expression cassette was removed, and red fluorescent protein (RFP) is expressed in Cre-positive cells. As shown in Fig. 3C, after induction by tamoxifen, the expression of RFP was restricted to vascular ECs in Tmem30a^iΔEC mice, here examined in the retina, and was absent in WT mice, indicating the specificity of this Cre line.

The fact that knockdown of TMEM30A in HRECs led to impaired tube formation in vitro (see above) promoted us to investigate the roles of Tmem30a in angiogenesis in mouse retinas. Mouse retinas are avascular at birth, and a single, superficial layer of blood vessels grows progressively from the center toward the periphery from P1 until P7. Then, the superficial blood vessels grow into the deep retinal layers. Retina whole-mount staining revealed that growth of the superficial vasculature toward the periphery zone in Tmem30a^iΔEC mouse retinas was delayed at P7 and exhibited a hyperpruned vascular network compared with those of WT mice (Fig. 4A–D). These results indicate that endothelial Tmem30a promoted angiogenesis sprouting and vascular network formation in the retina during the early postnatal period. Strikingly, at the frontier of the vascular network, the number of tip ECs was reduced dramatically and these exhibited a blunt-end, aneurysm-like structure with fewer and malformed filopodia compared with those of WT mice (Fig. 4E–G). To further characterize the retinal vascular phenotypes of the Tmem30a^iΔEC mice, we performed an immunofluorescence assay on frozen sections of P9 retinas. In the WT mouse retinas, the vascular branches were established in the outer plexiform layer (OPL) and inner plexiform layer (IPL) at P9, whereas vascular branches were found only in the ganglion cell layer of the Tmem30a^iΔEC mouse retinas, indicating a delayed vascular development toward the deep layers in the retina (Fig. S1).

Our studies using the induced EC-specific deletion of Tmem30a mouse model (Tmem30a^iΔEC mice) demonstrated that loss of Tmem30a in ECs impaired the retinal vascular angiogenesis process. However, other cells, such as pericytes, are also indispensable during maturation of the blood vasculature. To study the global roles of Tmem30a in angiogenesis process, we generated induced global Tmem30a knockout mice (Tmem30a-iKO) by crossing Tmem30a^loxP/loxP mice with Tmem30a^loxP/loxP CAG-Cre mice (Fig. 3A,B). The Tmem30a-iKO mice and their littermates without CAG-Cre were injected intraperitoneally with tamoxifen at P2 to P4 for 3 consecutive days. Retinal whole mounts were analyzed by staining with isolectin B4 (IB4) at P7. Angiogenesis in Tmem30a-iKO mice was also impaired, similar to that in Tmem30a^iΔEC mice (Fig. 4A–G). Staining of red blood cells using Ter119 antibody revealed blood vessel leakage in the Tmem30a-iKO retinas (Fig. 4H). Taken together, our results demonstrate that Tmem30a mainly contributes to vascular sprouting in ECs, while in cells other than ECs, Tmem30a plays an important role in vessel barrier integrity.

The above results indicate that Tmem30a plays critical roles in the angiogenesis process, while its role in maintaining the vascular network still needs to be elucidated. To this end, we started to administer tamoxifen injections to mice when the retinal vascular was mature, specifically, from P21 for 3 consecutive days. Then, the retinal whole mounts were analyzed by IB4 staining at P30. Our results revealed that the retinal vascular density of Tmem30a-iKO mice did not change compared to that of the WT mice (Fig. 5A,B).

To obtain mechanical insights into how TMEM30A regulates angiogenesis, we performed a series of in vitro and in vivo analyses in HRECs and Tmem30a-iKO mice. As the tip ECs numbers and vascular density and branches in Tmem30a^iΔEC mice were reduced dramatically, we first considered whether apoptosis had occurred during angiogenesis. However, terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) assays on the whole-mount retinas from Tmem30a^iΔEC mice at P7 did not show any signs of apoptosis (Fig. S2). In addition, as shown in Fig. 5, the vascular density did not decrease when Tmem30a was removed from the mature retinal vascular, verifying that apoptosis was not ongoing in the ECs. However, the question that then arises is what causes the retarded vascular angiogenesis and reduced tip ECs upon Tmem30a deletion? Because VEGF is a key growth factor in angiogenesis, we stained retinal whole mounts from Tmem30a-iKO mice with anti-VEGFa antibody. There was no difference in the distribution of VEGFa that surrounded tip ECs between WT and Tmem30a-iKO retinas, indicating that VEGFa secretion was not disturbed when Tmem30a was removed (Fig. 6A). Furthermore, we labeled the proliferating ECs with 5-ethynyl-2'-deoxyuridine (EDU) to assess cell proliferation. The number of EDU-positive ECs in Tmem30a-iKO mouse retinas were reduced dramatically (Fig. 6B). To unveil the reasons behind this finding, we first performed RNA sequencing (RNA-seq) analysis on total RNAs extracted from the WT and HRECs shTMEM30A HRECs. Statistical analyses of the RNA-seq results revealed that mRNA expression of the genes involved in the cell cycle (CCNA1, CCNB1, CCND1, CCDN3, CDC6, DBF4 ESPL1, PCNA1 and STAG1) was significantly downregulated in the shTMEM30A HRECs (Fig. 7A). These RNA-seq results were verified by real-time quantitative PCR (RT-qPCR) (Fig. 7B). Next, we also assessed the expression of those genes in the lung of Tmem30a-iKO mice by RT-qPCR analysis. Similar to the results in HRECs, the expression of these cell-cycle-related genes was also downregulated in the lungs of Tmem30a-iKO mice at P7 (Fig. 7C). Therefore, based on the RNA expression results, we propose that the main reason for the impaired angiogenesis after deletion of Tmem30a is compromised proliferation of ECs.

These findings prompted us to ask another question: how does Tmem30a regulate EC proliferation? VEGF is a key player governing sprouting angiogenesis. VEGF ligands bind to VEGFR2 receptors to stimulate endothelial stalk cells to differentiate to endothelial tip cells, which is a critical initial step in angiogenesis. Based on the fact that the number of retinal endothelial tip cells was greatly reduced in Tmem30a deletion mice, we inferred that the VEGF-induced signaling was impaired. We performed western blot analyses on the phosphorylation status of VEGFR2, an indicator of VEGF signaling pathway activity. In both the shTMEM30A HRECs and the lungs of Tmem30a-iKO mice (at P7), the Tyr1175 phosphorylation level of VEGFR2 was reduced compared to that of the WT controls (Fig. 8A,B). We further examined the phosphorylation of other proteins involved in VEGF induced signaling. Phosphorylation of PLCγ1, SRC, ERK1 and ERK2 (ERK1/2, also known as MAPK3 and MAPK1, respectively), p38 MAPK family proteins (α, β and γ forms; also known as MAPK14, MAPK11 and MAPK12, respectively) and MEK1 and MEK2 (MEK1/2, also known as MAP2K1 and MAP2K2, respectively) were also downregulated in shTMEM30A HRECs and in the lung tissue of Tmem30a-iKO mice (Fig. 8A,B). The phosphorylation of these proteins play critical roles in the angiogenesis process, such as cell proliferation and cell migration (Olsson et al., 2006). However, the phosphorylation level of AKT1 (hereafter AKT), which is responsible for ECs survival, seemed to be unchanged when Tmem30a was removed (Fig. S2) (Koch et al., 2011). Consistent with this observation, no obvious apoptosis was observed in Tmem30a^iΔEC mice (Fig. S3A). An ER stress assay using anti-CHOP antibody (CHOP is also known as DDIT3) also did not reveal any difference between WT and Tmem30a-iKO samples (Fig. S3B). Taken together, we demonstrated that deletion of TMEM30A led to an extensive disturbance on the activity of VEGF-induced signaling and further impaired the angiogenesis process.

In this study, we uncovered the important roles of TMEM30A, the β-subunit of P4-ATPases, in the sprouting angiogenesis process. P4-ATPases are members of a family of proteins that transport aminophospholipids from the exoplasmic to the cytoplasmic leaflet of cell membranes by utilizing ATP, thereby maintaining the aminophospholipid asymmetry in the cellular membrane (Halleck et al., 1999; Holthuis and Levine, 2005). Several P4-ATPases have been reported to have important functions in physiological and pathological conditions (Bull et al., 1998; Cacciagli et al., 2010; Darland-Ransom et al., 2008; Wang et al., 2004; Liu et al., 2017; Zhang et al., 2017). By using RT-PCR, we demonstrated that eight of 14 P4-ATPases are expressed in HRECs (Fig. 1A), indicating that they might have redundant roles, which would make it difficult to study their functions. To this end, we focused on the β-subunit of these P4-ATPases, namely TMEM30A. TMEM30A is expressed in three primary EC lines, as revealed by RT-PCR and western blotting methods (Fig. 1B,C). Defective tube formation of shTMEM30A-treated HRECs in vitro prompted us to assess what would happen when Tmem30a was deleted from ECs in a genetically modified mouse model. As constant deletion of Tmem30a in ECs when induced by Tie2-Cre is prenatally lethal (Table S2), we generated a tamoxifen-inducible endothelial cell-specific deletion model (Tmem30a^i△EC) using PDGFb-Cre. Deletion of Tmem30a in ECs resulted in retarded vascular sprouting, a greatly reduced number of tip ECs, and blunt-end, aneurysm-like structures with fewer and malformed filopodia in the retinal vessels (Fig. 4). EDU labeling also revealed that there was reduced EC proliferation in this model system (Fig. 6).

To uncover the mechanisms underlying these phenotypes, we carried out a series of RNA-seq, RT-qPCR and western blotting experiments. Our RNA-seq and RT-qPCR analysis showed that mRNA expression of those genes involved in the cell cycle (CCNA1, CCNB1, CCND1, CCDN3, CDC6, DBF4, ESPL1, PCNA1 and STAG1) was reduced in TMEM30A-knockdown HRECs and Tmem30a-iKO lung tissues (Fig. 7). Further western blot analyses of proteins involved in VEGF-induced signaling revealed significantly reduced Tyr1175 phosphorylation level of VEGFR2 in both the shTMEM30A HRECs and the lungs of Tmem30a-iKO mice (at P7), indicating compromised formation of the VEGFA–VEGFR2 complex in the absence of TMEM30A. The phosphorylation levels of other proteins involved in VEGF signaling, including PLCγ1, SRC, p38 MAPK family proteins and ERK1/2, were also decreased (Fig. 8). Our data demonstrate that TMEM30A governs the angiogenesis sprouting process by regulating the VEGF-induced signaling. In addition, an apoptosis assay (TUNEL) revealed no obvious apoptosis (Fig. S3A). Consistent with this, the phosphorylation level of AKT, which involved in VEGF-mediated EC survival, was not reduced (Fig. S1). One possible explanation is that the phosphorylation of AKT is maintained via feedback-mediated upregulation of another growth factor whose signaling is not affected by Tmem30a. Indeed, lack of upregulation of the stress-induced CHOP staining confirmed no apoptosis due to cellular stress (Fig. S2B). Therefore, we propose the following mechanism for the role of Tmem30a in angiogenesis: loss of Tmem30a leads to reduced phosphorylation on Tyr1175 of VEGFR2, which in turn leads to reduced PLCγ1 phosphorylation, which, via PKC, causes reduced phosphorylation of p38 MAPK and ERK1/2 proteins (Takahashi et al., 2001). This ultimately leads to decreased cell proliferation and retarded angiogenesis. Further work is warranted to dissect the relationship bewteen PS distribution and VEGF signaling.

Angiogenesis is a fundamental process in neovascularization, which plays important roles in multiple pathological conditions, such as AMD and cancer (Carmeliet and Jain, 2000). Interventions into the angiogenesis process by chemical or antibody-based drugs have been used to treat certain diseases, such as AMD and cancer (Bizzaro et al., 2018; Calugaru and Calugaru, 2018; Feng et al., 2018; Gorsi et al., 2018; Hutton-Smith et al., 2018; Martini et al., 2018). Our findings in this research have unveiled indispensable roles for TMEM30A, the β-subunit of P4-ATPases, in angiogenesis, which could provide new targets for the treatment of these diseases.

In summary, our study has delineated the essential roles of TMEM30A for sprouting angiogenesis and blood vessel barrier integrity. We further demonstrated that deletion of Tmem30a inhibited angiogenesis by disturbing the VEGF-induced signaling. Our results also show that TMEM30A is a potential target for treatment of pathological angiogenesis.

HRECs (obtained from Cell Systems, Seattle, WA) were cultured in EGM™-2 media (Lonza, Rochester, NY) at 37°C in a 5% CO₂ incubator. HRECs at passages 3–7 were transfected with a lentivirus carrying shRNA targeting TMEM30A (5′-TACAATTACCCTGTACATT-3′, Genechem, Shanghai, China) or negative control shRNA (5′-TTCTCCGAACGTGTCACGT-3′) according to the manufacturer's protocol.

Total RNA of HRECs or retinal tissues was extracted. The cDNA was synthesized by a Superscript cDNA Synthesis Kit (Invitrogen, Waltham, MA, USA) and then used as template for PCR with different primers. The primers used for RT-PCR are listed in Table S1.

After extraction of total RNA from HRECs using RNeasy Mini kits (QIAGEN), the RNA purity was checked using a K5500 spectrophotometer (Kaiao, Beijing, China). RNA integrity and concentration were assessed using an RNA Nano 6000 Assay Kit of a Bioanalyzer 2100 system (Agilent Technologies, CA). For library preparation, a total of 2 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEB Next, Ultra RNA Library Prep Kit for Illumina^® (Cat# E7530L, NEB, USA), following the manufacturer's recommendations, and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature conditions in NEBNext First Strand Synthesis Reaction Buffer (5×). First-strand cDNA was synthesized using random hexamer primers and RNase H. Second-strand cDNA synthesis was subsequently performed using dNTPs, DNA polymerase I and RNase H. The library fragments were purified with QiaQuick PCR kits and elution with EB buffer, and then terminal repair, A-tailing and addition of an adapter were implemented. The RNA concentration of the library was measured using a Qubit RNA Assay Kit in Qubit 3.0 to preliminarily quantify and then dilute to 1 ng/μl. Insert size was assessed using the Agilent Bioanalyzer 2100 system (Agilent Technologies), and the qualified insert size was accurately quantified using a StepOnePlus™ Real-Time PCR System (library valid concentration>10 nM). The clustering of the index-coded samples was performed on a cBot cluster generation system using a HiSeq PE Cluster Kit v4-cBot-HS (Illumina, CA, USA) according to the manufacturer's instructions. After cluster generation, the libraries were sequenced on an Illumina HiSeq 2500 platform, and 150 bp paired-end reads were generated. All gene expression values from RNA-seq were changed to a log2 value and analyzed further. Then, those gene sets that had less than a 0.05 nominal P value were called.

Total RNA from HRECs or lung tissues was extracted with RNeasy Mini kits (QIAGEN), and 1 μg total RNA was reverse transcribed with EasyScript One-Step RT-PCR SuperMix (TransGen Biotech, China) following the manufacturer's instructions. The cDNAs were amplified using TransStart Tip Green qPCR SuperMix (TransGen Biotech, China) in a 7500 Fast Real-Time PCR System (Applied Biosystems, CA). The primers used for RT-qPCR are listed in Table S1.

Cells or tissues from mice were lysed in SDS lysis buffer (2% SDS and 62.5 mM Tris-HCl pH 6.8, containing protease inhibitor cocktail tablets ordered from Roche) and sonicated three times for 5 s. Equal amounts of protein (20 μg) were loaded onto a 10% polyacrylamide gel and analyzed by immunoblotting. The antibodies used for western blotting were against VEGFR2 [Cat# 2479S, Cell Signaling Technology (CST), Danvers, MA,, 1:2000 dilution], phosphorylated (p)-VEGFR2 (Cat# 2478S, CST, 1:2000 dilution), ERK1/2 (Cat# 16443-1-AP, Proteintech, Wuhan, China, 1:3000 dilution), p-ERK1/2 (Cat# 4370S, CST, 1:2000 dilution), MEK1 (Cat# 2352S, CST, 1:2000 dilution), p-MEK1/2 (Cat# 2338S, CST, 1:2000 dilution), p-PLCγ (Cat# 8713T, CST, 1:2000 dilution), SRC (Cat# 2123T, CST, 1:2000 dilution), p-SRC (Cat# 6943T, CST, 1:2000 dilution), p38 MAPK family proteins (Cat# 8690T, CST, 1:2000 dilution), p-p38 MAPK family proteins (Cat# 4511T, CST, 1:2000 dilution), AKT (Cat# 10176-2-AP, Proteintech, 1:2000 dilution), p-AKT (Cat# 4060S, CST, 1:2000 dilution) and GAPDH (Cat# 60004-1-Ig, Proteintech, 1:3000 dilution). HRP-conjugated goat anti-rabbit-IgG secondary antibody (Cat# 7074, CST, 1:5000 dilution) and goat anti-mouse-IgG secondary antibody (Cat# 7076, CST, 1:5000 dilution) were used for immunoblotting.

The in vitro Matrigel tube formation assays were carried out according to a previously reported method (Shao et al., 2009), with a few modifications. Specifically, 10 μl of liquid Matrigel (Corning, Cat# 354234) was carefully added to each inner well of a μ-Slide Angiogenesis plate (Ibidi, Germany, Cat# 81506) and then incubated at 37°C for half an hour. After the Matrigel was solidified, HRECs were trypsinized, and the cell number was determined using an automated cell counter (Inno-Alliance Biotech). The HRECs were then diluted to 1×10⁶ cells/ml, and 50 μl was added to each well of the μ-Slide angiogenesis plate. After 3 or 6 h, the images of controls or shTMEM30As were captured under an anatomical lens (Carl Zeiss, Germany).

All animal study protocols were approved by the Animal Care and Use Committee of Sichuan Provincial People's Hospital. All experimental procedures and methods were carried out in accordance with the approved study protocols and relevant regulations. Mice were raised in cyclic lighting conditions with a 12-h-light and 12-h-dark cycle.

Tmem30a^loxp/+ mice were mated to Pdgfb-CreER mice to generate Tmem30a^loxp/+Pdgfb-iCre mice (Claxton et al., 2008). Tmem30a^loxp/+Pdgfb-CreER mice were crossed with Tmem30a^loxp/loxp homozygous mice to generate an inducible EC deletion of Tmem30a (Tmem30a^loxp/loxp Pdgfb-iCre). Tmem30a^loxp/+mice were crossed with broadly expressed inducible CRE CAG-Cre-ER mice (Hayashi and McMahon, 2002) [stock Tg(CAG-cre/Esr1*) 5Amc/J, https://www.jax.org/strain/004453] to generate inducible adult knockout mice. To monitor the efficiency of Cre-mediated deletion of the floxed exon, a tdTomato reporter was used [strain name, B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J, also named Ai14D, http://jaxmice.jax.org/strain/007914.html]. The reporter contains a loxP-flanked STOP cassette that prevents transcription of the downstream CAG promoter-driven red fluorescent protein variant tdTomato. In the presence of Cre recombinase, reporter mice will have the STOP cassette removed in the Cre-expressing tissue(s) and will express tdTomato. Because this CAG promoter-driven reporter construct was inserted into the Gt(ROSA)26Sor locus, tdTomato expression is determined by the tissue(s) that expressed Cre recombinase.

Tamoxifen salt (100 mg; Sigma, St Louis, MO) was dissolved in 10 ml of ethanol as a stock solution. On the day of injection, a 1 mg/ml working solution was prepared by mixing the 10 mg/ml stock solution with corn oil (Sigma) at a ratio of 1:9 and mixed well. Mice were intraperitoneally injected with a daily dosage of 25 mg/kg body weight for 3 consecutive days.

Retinal dissection was carried out as described previously, and whole-mounted retinas were preserved in 0.4% paraformaldehyde (PFA). Enucleated eyes were fixed with 4% PFA and embedded in Tissue-Tek optimal cutting-temperature compound (OCT, Sakura Finetek). Before immunostaining, whole-mounted retinas and cryosections (12 μm, Leica CM1950) were rinsed in PBS three times (5 min/rinse) and blocked in PBS containing 5% fetal bovine serum and 0.2% Triton X-100 for 30 min at room temperature, followed by incubation with primary antibodies at 4°C overnight. Primary antibodies were diluted in blocking buffer as follows: isolectin GS-IB4 (1:200; Thermo Fischer, Waltham, MA; CAT# I21411), rat anti-mouse Ter-119 (1:20 dilution; 553670; BD Biosciences, San Jose, CA), and goat anti-mouse Vegf164 (1:100 dilution; AF-493; R&D Systems, Minneapolis, MN), and C/EBP homologous protein (CHOP; 1:500 dilution, 15204-1-AP; Proteintech, Chicago, IL). Then, the sections were washed with PBS three times and labeled for 1–4 h with Alexa Fluor™-488- or Alexa Fluor™-594-labeled goat anti-rat- or anti-rabbit-IgG or donkey anti-goat-IgG secondary antibody (1:500; Thermo Fischer). To detect EC proliferation in the retinas, 200 μg EDU (Thermo Fischer) per pup was injected intraperitoneally 3 h before animals were euthanized. EDU-positive cells were visible by subsequent staining with a Click-iT EDU Alexa Fluor-488 Imaging Kit (C10337; Thermo Fischer).

Retinal dissection was carried out as described previously, and whole-mounted retinas were fixed in 4% PFA and preserved in 0.4% PFA. Whole-mounted retinas were rinsed in PBS three times (5 min/rinse). Retinas were incubated in -ermeabilisation solution (0.1% Triton X-100 in 0.1% sodium citrate, freshly prepared) for 8–10 min. Then the samples were rinsed twice with PBS. After the slides were dried, 50 μl of TUNEL reaction mixture (Roche, In Situ Cell Death Detection Kit, POD Cat# 11684817910) was added to the specimen, and left to react at 37°C for 1 h in a dark humid chamber. For the positive controls, recombinant DNase I (3 U/ml in 50 mM Tris-HCl pH 7.5, 10 mM MgCl₂ and 1 mg/ml BSA) was added and incubated for 10 min at 25°C to induce DNA strand breaks, prior to labeling procedures. Slides were rinsed three times with PBS. Images were captured via fluorescence microscopy after slides were sealed with anti-fluorescence quenching liquid.