Enhancing Gpx1 palmitoylation to inhibit angiogenesis by targeting PPT1

Abstract

The significance of protein S-palmitoylation in angiogenesis has been largely overlooked, leaving various aspects unexplored. Recent identification of Gpx1 as a palmitoylated protein has generated interest in exploring its potential involvement in novel pathological mechanisms related to angiogenesis. In this study, we demonstrate that Gpx1 undergoes palmitoylation at cysteine-76 and -113, with PPT1 playing a crucial role in modulating the depalmitoylation of Gpx1. Furthermore, we find that PPT1-regulated depalmitoylation negatively impacts Gpx1 protein stability. Interestingly, inhibiting Gpx1 palmitoylation, either through expression of a non-palmitoylated Gpx1 mutant or by expressing PPT1, significantly enhances neovascular angiogenesis. Conversely, in PPT1-deficient mice, angiogenesis is notably attenuated compared to wild-type mice in an Oxygen-Induced Retinopathy (OIR) model, which mimics pathological angiogenesis. Physiologically, under hypoxic conditions, Gpx1 palmitoylation levels are drastically reduced, suggesting that increasing Gpx1 palmitoylation may have beneficial effects. Indeed, enhancing Gpx1 palmitoylation by inhibiting PPT1 with DC661 effectively suppresses retinal angiogenesis in the OIR disease model. Overall, our findings highlight the pivotal role of protein palmitoylation in angiogenesis and propose a novel mechanism whereby the PPT1-Gpx1 axis modulates angiogenesis, thereby providing a potential therapeutic strategy for targeting PPT1 to combat angiogenesis.

1. Introduction

Pathologic angiogenesis occurs when new blood vessels are formed under abnormal physiological conditions e.g., infection and inflammation, oxidative stress, and oxygen-induced retinopathy (OIR) [[1], [2], [3], [4]]. These abnormal blood vessel formations can lead to a variety of health problems, including cancer, retinopathy, and chronic inflammation [5]. Angiogenesis depends on the expression of pro-angiogenic factors, with VEGF widely recognized as a key regulator [6] and hypoxia being one of the key mechanisms involved [3]. Under hypoxic conditions, HIF-1α translocates to the nucleus and binds to HIF-1β, and HIF-1α/HIF-1β dimer promotes angiogenesis by inducing VEGF expression [7]. Although blockers of the VEGF pathway have been extensively studied, their expanded application often result in the consequence of promoting drug-induced resistance, which makes it particularly important to find new targets or pathways to restrict the expression of VEGF [7].

Protein S-palmitoylation (hereafter referred to as palmitoylation) is a reversible lipid modification that characterized by the formation of a thioester linkage between palmitic acid and a cysteine residue [8,9]. Due to the hydrophobic nature of palmitate itself, protein palmitoylation often plays a role in regulating protein stability, membrane localization, and signaling transduction [9,10]. Notably, the reversibility of protein palmitoylation is mediated by the action of palmitoylation, catalyzed by ZDHHC palmitoyltransferases (ZDHHC 1–9, and 11–25, with no ZDHHC10), and depalmitoylation, catalyzed by protein thioesterases (APT1/2, PPT1/2, and Abhd17) [[11], [12], [13]]. Although the roles of palmitoylation in neuroscience and tumorigenesis have been extensively studied [11,14], its function and regulatory mechanism in angiogenesis remain obscure. To address this, we used the OIR disease model to mimic the pathological processes of angiogenesis induced by hypoxia [15], the lysates of mice retina were collected and subjected to palm-proteomics analysis. Remarkably, the MS data identified Gpx1 as a potent palmitoylated protein candidate.

Gpx1, a member of the glutathione peroxidase family, is an important antioxidant enzyme involved in the downregulation of intracellular organic hydroperoxides and hydrogen peroxide (H₂O₂) [16,17]. Dysregulation of GPX1 has been identified as a biomarker for various pathological conditions e.g., malignant tumors [18], oxidative stress [[19], [20], [21]], and cardiovascular disease [[22], [23], [24]]. Interestingly, the deletion of Gpx1 leads to increased pathological angiogenesis in the OIR disease model compared to WT control mice [25], suggesting that Gpx1 plays a critical role in angiogenesis. However, the detailed molecular mechanism underlying this role remains unclear.

Herein, we aimed to investigate the palmitoylation of Gpx1 and explore its molecular function and regulatory mechanism in pathological angiogenesis. Interestingly, we confirmed that Gpx1 undergoes palmitoylation, and we discovered that the dynamic palmitoylation of Gpx1 (referred to as palm-Gpx1) is regulated by PPT1. We observed that reducing palm-Gpx1 levels leads to increased angiogenesis, while increasing palm-Gpx1 levels by DC661, an inhibitor of PPT1, significantly suppresses VEGF expression and angiogenesis. These findings highlight the crucial roles of the PPT1-Gpx1 axis in mediating angiogenesis, and suggest that enhancing palm-Gpx1 by inhibiting PPT1 could have beneficial effects in clinical settings.

2. Results

2.1. Gpx1 undergoes palmitoylation predominantly at Cysteine-76 and Cysteine-113

To confirm the palmitoylation of Gpx1, we conducted Acyl-Biotin-Exchange (ABE) analysis on endogenously expressed Gpx1 in HEK293T cells, Human Umbilical Vein Endothelial Cells (HUVECs), and wildtype (WT) mouse retina. The results consistently demonstrated that Gpx1 undergoes palmitoylation both in vitro and in vivo (Fig. 1A–C). Additionally, we investigated the dynamic regulation of Gpx1 palmitoylation by treating cultured HUVECs with PalmB (an inhibitor of protein depalmitoylation) or 2BP (an inhibitor of protein palmitoylation). Intriguingly, PalmB treatment significantly increased the level of palm-Gpx1 (Fig. 1D and E), whereas 2-BP treatment dramatically decreased the level of palm-Gpx1 (Fig. 1F and G), providing further evidence that Gpx1 palmitoylation is a reversible process.

Fig. 1.

Gpx1 undergoes palmitoylation predominantly at Cysteine-76 and Cysteine-113.

(A–C) Gpx1, expressed endogenously in mouse retina (A), HEK293T cells (B), or HUVECs (C), was analyzed for protein palmitoylation using the ABE assay. (HA refers to hydroxylamine, HA + for treatment with HA, HA-for treatment without HA). (D–G) Cultured HUVECs were treated with either PalmB (D) or 2-BP (E), and lysates were evaluated for the level of Gpx1 palmitoylation using the ABE assay. The palmitoylation levels were quantified (F–G). (H) Cysteine conservation was analyzed by aligning the protein sequences of Gpx1 from different species. (I) Purified Gpx1 was probed using Mass-Spectrometry, and a mass-shift of 238 Da linked to cysteine was observed as a hallmark of palmitoylation. (J–K) Gpx1 constructs with point mutations (cysteine to serine) were expressed in HUVECs and subjected to evaluation of palmitoylation levels using the ABE assay (J), and the levels were quantified (K). (L) The palmitoylation modification sites of Gpx1 were analyzed by mPEG labeling assay in HUVECs. (M) HUVECs were metabolically labeled with 17-ODYA and the lysates were reacted with biotin-azide and enriched with streptavidin-agarose beads, hydroxylamine (HA) was used to hydrolyze thioester linkage and remove 17-ODYA labeling. Blots shown are representative, and data represents the mean ± S.E.M. from three biological replicates. Statistical significance is denoted as ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001, determined by student t-test or one-way ANOVA followed by Tukey's post hoc test.

To determine the specific cysteine residues involved in Gpx1 palmitoylation, we conducted a sequence alignment of Gpx1 protein sequences from various species, with a focus on cysteine residues. The alignment results revealed that human Gpx1 contains a total of 5 cysteines, three of which are conserved among different species. This conservation suggests that these cysteines (76, 133, and 153) are potential candidates for palmitoylation (Fig. 1H).

To pinpoint the exact sites of palmitoylation on Gpx1, we ectopically expressed Flag-tagged Gpx1 in HEK293T cells and purified the protein using Flag-conjugated agarose beads (Fig. S1A). The purified Gpx1 was then subjected to Mass Spectrometry (MS) analysis. The MS data demonstrated that Gpx1 is indeed modified with palmitate at cysteine-76 (C76, with a mass alteration of 238Da) and cysteine-113 (C113, also with a mass alteration of 238Da) (Fig. 1I and Figs. S1B–C).

To validate these findings, we introduced point mutations to replace the cysteines with serines (CS mutation). Interestingly, a single mutation either at C76 (C76S) or C113 (C113S) did not noticeably affect the level of palm-Gpx1. However, when both cysteines were mutated (Gpx1-2CS), there was a significant reduction in Gpx1 palmitoylation (Fig. 1J and K). In parallel, mPEG labeling assay confirmed that Gpx1 displayed two distinct migrating bands in +HA lane (caused by the substitution of 5 kD mPEG for palmitate). H-Ras was assessed as a positive control [26] (Fig. 1L). Also, click chemistry conjugated with17-ODYA metabolic labeling confirmed that Gpx1 was palmitoylated (Fig. 1M).

This supports the conclusion that C76 and C113 are the major sites of palmitoylation in Gpx1.

2.2. Palmitoylation regulates the protein stability of Gpx1

Repeatedly, it was noted that the expression of Gpx1-2CS is markedly lower than that of WT Gpx1, suggesting impaired expression of Gpx1-2CS. To determine if the mRNA transcription of Gpx1-2CS is altered, we extracted total mRNA from cells expressing either WT or mutant Gpx1. qRT-PCR results showed no significant change in the mRNA transcription of Gpx1-2CS compared to WT Gpx1 (Fig. S2A), indicating that the protein stability of Gpx1-2CS maybe modulated instead. To investigate this, HUVECs expressing either Flag-Gpx1 or Flag-Gpx1-2CS were incubated with cycloheximide (CHX) to blocking de novo protein synthesis for varied time periods. Indeed, the results showed that the degradation of Flag-Gpx1-2CS was significantly accelerated compared to WT Flag-Gpx1 (Fig. 2A and B). Furthermore, to understand the potential pathways involved in Gpx1 degradation, we introduced a panel of inhibitors (Chloroquine (CHQ) blocks lysosome degradation, 3-Methyladenine (3 MA) blocks autophagy, MG132 blocks proteasome). Intriguingly, the results showed that WT Gpx1 is primarily degraded via the autophagy and lysosome pathway (Fig. 2C and D), while the inhibition of these pathways was not effective for Gpx1-2CS (Fig. 2E and F), suggesting the involvement of more sophisticated degradation pathways.

Fig. 2.

Palmitoylation regulates the protein stability of Gpx1.

(A–B) HUVECs expressing Flag-Gpx1 or Flag-Gpx1-2CS were treated with cycloheximide (CHX) for various time intervals, and the protein levels of Gpx1 were assessed (A) and quantified (B). The blots presented are representative, and the data represent the mean ± S.E.M. from three biological replicates. (C–D) HUVECs expressing Flag-Gpx1 were treated with CHX and various combinations of inhibitors targeting protein degradation pathways for different time periods. Subsequently, the protein levels of Gpx1 were assessed (C) and quantified (D). (E–F) HUVECs expressing Flag-Gpx1-2CS were treated with CHX and various combinations of inhibitors targeting protein degradation pathways for different time periods. Subsequently, the protein levels of Gpx1 were assessed (E) and quantified (F). Blots shown are representative, and data represents the mean ± S.E.M. from three biological replicates. Statistical significance is denoted as ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, determined by student t-test.

Overall, these findings support the role of palmitoylation in positively regulating the protein stability of Gpx1.

2.3. The depletion of Gpx1 promotes angiogenesis in HUVECs

To assess a potential role of Gpx1 in angiogenesis, Gpx1 was deleted in HUVECs (Fig. 3A and Figs. S2B–C). Initially, we observed that the rate of cell proliferation was not significantly altered in Gpx1-KO compared to WT HUVECs under normal conditions (Fig. 3B). However, considering that Gpx1 is primarily involved in maintaining hydroperoxide homeostasis, we examined whether treatment with H₂O₂ would yield different results. Interestingly, our findings revealed that cell proliferation was significantly enhanced in Gpx1-KO HUVECs compared to WT cells upon H₂O₂ treatment (Fig. 3C). Moreover, additional experiments demonstrated that the depletion of Gpx1 markedly promoted vessel tube formation (Fig. 3D) but did not affect cell migration (Fig. 3F) in HUVECs under normal conditions. However, under H₂O₂ exposure, both angiogenesis (Fig. 3E) and cell migration (Fig. 3G) were significantly enhanced in Gpx1-KO cells compared to WT cells.

Fig. 3.

The depletion of Gpx1 promotes angiogenesis in HUVECs.

(A) The loss of Gpx1 expression was confirmed in WT and Gpx1-KO HUVECs through Western blot analysis. (B–C) The rate of cell proliferation in WT and Gpx1-KO HUVECs, treated with (C) or without (B) H₂O₂ (3h), was assessed using CCK8 assays. The data presented are mean ± S.E.M. from four biological replicates. (D–E) WT and Gpx1-KO HUVECs, treated with (E) or without (D) H₂O₂ (3h), were subjected to the tube formation assay, and the relative number of tubule nodes was quantified. (F–G) WT and Gpx1-KO HUVECs, treated with (G) or without (F) H₂O₂ (3h), were subjected to the wound healing assay, and the percentage of relative wound healing was quantified.

Blots shown are representative, and data represents the mean ± S.E.M. from three biological replicates. Statistical significance is denoted as ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, determined by student t-test.

Combined, these results suggest that Gpx1 plays a vital role in regulating cell proliferation, migration and angiogenesis, particularly under hypoxic conditions.

2.4. The regulation of angiogenesis by Gpx1 relies on its palmitoylation

To investigate the role of palmitoylation in Gpx1, we aimed to rescue Gpx1-KO HUVECs by introducing either Gpx1 or Gpx1-2CS. Firstly, the results indicated that cell proliferation remain unchanged under normoxia (room air) (Fig. S3A) or hypoxia (after H₂O₂ treatment) conditions (Fig. S3B). Secondly, while rescuing Gpx1-KO cells with WT Gpx1 significantly inhibited cell migration (Fig. 4C and D) and angiogenesis (Fig. 4A and B), the rescue with Gpx1-2CS completely abolished these effects and prominently enhanced cell migration (Fig. 4C and D) and angiogenesis (Fig. 4A and B). Lastly, similar outcomes were observed in Gpx1-KO HUVECs rescued with either Gpx1 or Gpx1-2CS under hypoxia conditions. Specifically, the re-expression of Gpx1 in Gpx1-KO cells significantly suppressed cell migration (Fig. 4G and H) and angiogenesis (Fig. 4E and F), while the expression of Gpx1-2CS in Gpx1-KO cells markedly enhanced cell migration (Fig. 4G and H) and angiogenesis (Fig. 4E and F).

Fig. 4.

The regulation of angiogenesis by Gpx1 relies on its palmitoylation.

(A–B) Tube formation assay was performed on Gpx1-KO HUVECs rescued with either Flag-Gpx1 or Flag-Gpx1-2CS, and the relative number of tubule nodes was quantified. (C–D) Wound healing assay was conducted on Gpx1-KO HUVECs rescued with either Flag-Gpx1 or Flag-Gpx1-2CS, and the percentage of relative wound healing was quantified. (E–F) Tube formation assay was performed on Gpx1-KO HUVECs expressing either Flag-Gpx1 or Flag-Gpx1-2CS after treatment with H₂O₂ (3h), and the relative number of tubule nodes was quantified. (G–H) Wound healing assay was conducted on Gpx1-KO HUVECs expressing either Flag-Gpx1 or Flag-Gpx1-2CS after treatment with H₂O₂ (3h), and the percentage of relative wound healing was quantified. Representative images are shown, and the data represent the mean ± S.E.M. from three biological replicates. Statistical significance is denoted as ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001, determined by one-way ANOVA followed by Tukey's post hoc test.

Jointly, these findings demonstrate that Gpx1 regulates angiogenesis in a manner that depends on palmitoylation.

2.5. PPT1 downregulates Gpx1 palmitoylation and modulates angiogenesis

To understand the dynamics of Gpx1 palmitoylation, we explored the enzymes responsible for the depalmitoylation of Gpx1. Gpx1-Flag was coexpressed with various thioesterases (APT1/2, PPT1/2, and Abhd17a), and the results demonstrated that the expression of PPT1 significantly reduces the levels of palm-Gpx1 in HUVECs (Fig. 5A and B). Considering that palmitoylation influences Gpx1 stability, we also examined whether coexpression of PPT1 could downregulate its protein stability. Indeed, the findings revealed that the coexpression of PPT1 with Gpx1 significantly diminishes the protein stability of Gpx1 compared to Gpx1 expression alone (Fig. 5C and D).

Fig. 5.

PPT1 downregulates Gpx1 palmitoylation and modulates angiogenesis.

(A–B) Thioesterases (APT1/2, PPT1/2, or ABHD17a) were co-expressed with Flag-Gpx1 in HEK293T cells, and protein palmitoylation was evaluated using the ABE assay (A), and quantified (B). (C–F) The Oxygen-Induced Retinopathy (OIR) disease model was established using WT and PPT1-KO mice, and mouse retinas were collected for the evaluation of protein palmitoylation by ABE (C–D) or flat-mounting to examine the relative area of neovascularization (E–F) at post-natal day 17 (P17). n = 7 for WT and 13 for PPT1-KO. (G–H) The rate of cell proliferation in HUVECs expressing either an empty vector or PPT1-Flag constructs, treated with (H) or without (G) H₂O₂ (3h), was examined using CCK8 assays. (I–L) HUVECs expressing an empty vector or PPT1-Flag, treated with (K–L) or without (I–J) H₂O₂ (3h), were subjected to the tube formation assay, and the relative number of tubule nodes was quantified. (M–P) HUVECs expressing an empty vector or PPT1-Flag, treated with (O–P) or without (M–N) H₂O₂ (3h), were subjected to the wound healing assay, and the percentage of relative wound healing was quantified. Blots and images shown are representative, and data represents the mean ± S.E.M. from three biological replicates or otherwise indicated. Statistical significance is denoted as ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, determined by student t-test or one-way ANOVA followed by Tukey's post hoc test.

Furthermore, to evaluate the role of Gpx1 palmitoylation in vivo, we utilized PPT1-deficient mice (PPT1-KO) [27] to establish the OIR disease model (Fig. S3C). Briefly, the mice were exposed to 75 % oxygen from P7–P12 to suppress angiogenesis, followed by a return to normoxia to induce angiogenesis due to hypoxia. The timepoint at P17 represents the peak of angiogenesis. Notably, the results from flat-mounting demonstrated a significant decrease in the relative area of neovascularization in PPT1-KO mice compared to WT control mice (Fig. 5E and F). Additionally, enhanced levels of palm-Gpx1 were observed in the retinas of PPT1-KO mice compared to control mice (Fig. 5G and H). Consistently, in vitro experiments using HUVECs revealed that ectopic expression of PPT1 significantly enhances cell migration (Fig. 5M − N) and angiogenesis (Fig. 5I and J), but does not affect cell proliferation (Figs. S3D–E), compared to control conditions. Furthermore, similar results were obtained upon treatment with H₂O₂ (Fig. 5K and L and Fig. 5O and P).

Collectively, these findings support the notion that elevated levels of palm-Gpx1, regulated by PPT1, restrain angiogenesis.

2.6. Increase in Gpx1 palmitoylation using DC661 attenuates angiogenesis

Therefore, we aimed to test the hypothesis that inhibiting PPT1 with DC661 [28], a PPT1 inhibitor, may have beneficial effects in suppressing pathological angiogenesis. To investigate this, we incubated HUVECs with different concentrations of DC661 and found that DC661 at doses higher than 1.5 μM significantly suppressed cell migration and angiogenesis (Fig. 6A and B and Fig. 6C and D). Notably, DC661 treatment also markedly impaired HUVECs proliferation (Fig. 6E).

Fig. 6.

Increase in Gpx1 palmitoylation using DC661 attenuates angiogenesis.

(A–B) HUVECs were incubated with different dosages of DC661 and subjected to the tube formation assay (A), with quantification of the relative number of tubule nodes (B). (C–D) HUVECs were incubated with similar dosages of DC661 and subjected to the wound healing assay (C), with quantification of the percentage of relative wound healing (D). (E) HUVECs were incubated with different dosages of DC661 and subjected to the CCK-8 assay. (F–G) HUVECs were incubated with the same dosage of DC661 for varied time periods and subjected to the ABE assay (F), with quantification (G). (H–K) The WT mouse was used to establish the OIR disease model, and mouse retinas were collected for the evaluation of protein palmitoylation by ABE (H–I) or flat-mounting to examine the relative area of neovascularization (J–K) at P17. n = 6 for WT mouse. Blots and images shown are representative, and data represents the mean ± S.E.M. from three biological replicates or otherwise indicated. Statistical significance is denoted as ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001, determined by student t-test or one-way ANOVA followed by Tukey's post hoc test.

Moreover, we examined the correlation between the level of palm-Gpx1 and DC661 treatment, and observed that DC661 significantly increased the level of palm-Gpx1 in a time-dependent manner in HUVECs (Fig. 6F and G). Importantly, in an in vivo setting using an OIR mouse model, we found that injection of DC661 into the vitreous cavity dramatically increased the level of palm-Gpx1 at P17 compared to the control group treated with DMSO (Fig. 6H and I). Consequently, this treatment significantly reduced the area of neovascularization at P17 in the DC661-treated group compared to the DMSO control group (Fig. 6J and K).

In summary, these findings confirm that increasing Gpx1 palmitoylation by targeting PPT1 confers beneficial effects for attenuating pathological angiogenesis both in vitro and in vivo.

2.7. Gpx1 palmitoylation modulates angiogenesis in human retinal endothelial cell

To verify the findings from HUVECs to human retinal endothelial cell (HRECs), HRECs were introduced. Consistently, the results show that endogenously expressed Gpx1 is palmitoylated and that PPT1 significantly downregulates the level of Gpx1 palmitoylation in HRECs (Figs. S4A–B). Functionally, the downregulation of Gpx1 palmitoylation noticeably enhances cell proliferation in HRECs, both with and without H₂O₂ treatment (Figs. S4C–D). Additional experiments demonstrated that a reduced level of Gpx1 palmitoylation markedly promotes vessel tube formation but does not affect cell migration in HRECs under normal conditions (Fig. S4E-F and Fig. S4I-J). However, both angiogenesis and cell migration are significantly enhanced under H₂O₂ exposure (Fig. S4G-H and Fig. S4K-L).

Conversely, to test whether inhibiting PPT1 with DC661 could increase the level of Gpx1 palmitoylation, cultured HRECs were incubated with DC661. The results showed that DC661 notably upregulates the level of Gpx1 palmitoylation (Figs. S5A–B). Functionally, this increased level of Gpx1 palmitoylation significantly suppresses cell proliferation in HRECs upon DC661 treatment (Fig. S5C). Furthermore, additional experiments demonstrated that the augmented level of Gpx1 palmitoylation markedly inhibits vessel tube formation (Figs. S5D–E) and cell migration in HRECs (Figs. S5F–G).

Together, these findings extend the evidence from HUVECs to HRECs, reinforcing the conclusion that the PPT1-Gpx1 palmitoylation axis plays a crucial role in regulating angiogenesis.

2.8. Gpx1 palmitoylation facilitates the downregulation of ROS

To elucidate the mechanism underlying palm-Gpx1 regulated angiogenesis, we aimed to examine the level of reactive oxygen species (ROS), which is a hallmark of oxidative stress and a direct substrate of Gpx1, upon the exposure to H₂O_2. Notably, the experiments showed that the level of ROS is significantly elevated after 1 h of H₂O₂ treatment but recovers to the normal level after 3 h in WT HUVECs (Fig. 7A and B). In contrast, such treatment results in a significantly higher level of ROS in Gpx1-KO cells at both timepoints (1h or 3h) (Fig. 7A and B), implying that Gpx1 is a critical regulator of cellular ROS.

Fig. 7.

Gpx1 palmitoylation facilitates the downregulation of ROS.

(A–B) Cultured WT and Gpx1-KO HUVECs were treated with H₂O₂ for varying time periods, stained with CellROX® Deep Red for Fluorescence Imaging (A), and quantified (B). (C–D) Gpx1-KO HUVECs rescued with Gpx1 or Gpx1-2CS were treated with H₂O₂ for varying time periods, stained with CellROX® Deep Red for Fluorescence Imaging (C), and quantified (D). (E–F) Cultured WT HUVECs were treated with H₂O₂ for varying time periods, and collected for the evaluation of protein palmitoylation by ABE assay (E), with quantification (F). (G) The Gpx1 activity was measured in Gpx1-KO HUVECs rescued with either WT Gpx1 or Gpx1-2CS. Blots and images shown are representative, and data represents the mean ± S.E.M. from three biological replicates or otherwise indicated. Statistical significance is denoted as ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001, determined by student t-test.

To further assess the role of palmitoylation, Gpx1-KO cells were rescued with either WT Gpx1 or Gpx1-2CS. The experiments showed that the re-expression of WT Gpx1 markedly downregulates the cellular ROS level, while the rescue with Gpx1-2CS fails to reduce intracellular oxidative stress at either 1h or 3h after H₂O₂ treatment (Fig. 7C and D). These findings indicate that the downregulation of ROS by Gpx1 is dependent on its palmitoylation. Moreover, the level of palm-Gpx1 is significantly elevated upon exposure to H₂O₂ (Fig. 7E and F), suggesting a dynamic regulation of Gpx1 palmitoylation in response to ROS and the involvement of palmitoylation in modulating Gpx1 activity.

Indeed, the Gpx1 activity was evaluated by measuring GSH concentration, and the results demonstrated that Gpx1-KO cells rescued with WT Gpx1 exhibit significantly higher enzymatic activity compared to Gpx1-KO cells rescued with Gpx1-2CS (Fig. 7G).

Combined, these findings suggest that Gpx1 palmitoylation plays a crucial role in downregulating ROS levels by regulating its enzymatic activity.

2.9. Gpx1 palmitoylation regulates angiogenesis by modulating the expression levels of HIF-1α and VEGF

To further dissect the molecular events underlying angiogenesis related to Gpx1 palmitoylation, the expression levels of HIF-1α and VEGF were measured. The results showed that both HIF-1α and VEGF expression levels are significantly upregulated in Gpx1-KO compared to WT HUVECs when exposed to H₂O₂ (3h). However, in normoxia, the VEGF level is markedly elevated while the HIF-1α level is downregulated (Fig. 8A and B).

Fig. 8.

Gpx1 palmitoylation regulates angiogenesis by modulating the expression levels of HIF-1α and VEGF.

(A–B) WT and Gpx1-KO HUVECs were treated with or without H₂O₂ for 3 h and subjected to the evaluation of the expression levels of HIF-1α (A) and VEGF (B). (C–D) Gpx1-KO HUVECs rescued with Gpx1 or Gpx1-2CS were treated with or without H₂O₂ for 3 h, and the expression levels of HIF-1α (C) and VEGF (D) were evaluated. (E–F) WT HUVECs expressing an empty vector or PPT1-Flag were treated with or without H₂O₂ for 3 h, and the expression levels of HIF-1α (E) and VEGF (F) were evaluated. (G–J) WT mice were used to establish the OIR disease model, and mouse retinas were collected at different time points (P12–P25) for the evaluation of protein palmitoylation by ABE assay (G–H), and the expression levels of HIF-1α (I) and VEGF (J). (K–L) WT mice were used to establish the OIR disease model and treated with either DMSO or DC661 (injection into the vitreous cavity). Subsequently, mouse retinas were collected and subjected to the evaluation of the expression levels of HIF-1α (K) and VEGF (L). Blots shown are representative, and data represents the mean ± S.E.M. from three biological replicates or otherwise indicated. Statistical significance is denoted as ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001, determined by student t-test or one-way ANOVA followed by Tukey's post hoc test.

Moreover, when palm-Gpx1 is inhibited, such as in Gpx1-KO cells rescued with Gpx1 or Gpx1-2CS (Fig. 8C and D), or in HUVECs ectopically expressing PPT1 (Fig. 8E and F), the expression levels of HIF-1α and VEGF are markedly increased upon H₂O₂ treatment. Similar patterns persist, except that HIF-1α is downregulated in Gpx1-KO cells rescued with Gpx1 in normoxia.

Physiologically, the level of palm-Gpx1 remains relatively constant from P12–P21 but decreases at P25 in normoxia. However, this pattern is altered in hypoxia (OIR mouse model), where the level of palm-Gpx1 gradually decreases from P12–P25 (Fig. 8G and H). Consistently, while the expression of HIF-1α and VEGF remains relatively constant in normoxia, their levels are dramatically upregulated in hypoxia conditions (Fig. 8I and J).

Remarkably, the treatment of the PPT1 inhibitor, DC661, significantly lowers the expression levels of both HIF-1α and VEGF in pathological angiogenesis induced by hypoxia (Fig. 8K and L). In total, these data suggest that Gpx1 palmitoylation is dysregulated in pathological angiogenesis, and the onset of the latter can be largely explained by the altered expression levels of HIF-1α and VEGF.

2.10. Alternative modulators for HIF-1α and VEGF in vitro and in vivo

We observed significant changes in the expression of HIF-1α and VEGF between control and H₂O₂-treated conditions. Since HIF-1α activation or dimerization is typically induced by hypoxia or oxidative stress, this suggests the presence of alternative modulators for HIF-1α and VEGF under these conditions.

To further explore this mechanism, we conducted an unbiased mRNA-seq analysis of both WT and Gpx1-KO cells. Our results revealed significant alterations in the Cytokine-cytokine receptor interaction, TNF, and NF-κB signaling pathways between the WT and Gpx1-KO genotypes (Fig. S). These pathways were also notably affected by H₂O₂ treatment in both genotypes (Fig. S). These findings suggest that the TNFα–NF–κB-mediated inflammatory axis may play a crucial role in regulating angiogenesis. Previous studies indicate that ROS interacts with NF-κB signaling pathways in a context-dependent manner, as ROS can both activate and inhibit NF-κB pathways due to its ability to act in multiple ways and locations simultaneously [29].

To validate these findings at the protein level, we performed mass spectrometry analysis using in vivo models. WT and PPT1-KO mice were used to establish the OIR disease model. On day 17 (P17), the mouse retinas were dissected and prepared for mass spectrometry analysis. KEGG analysis revealed that Glycosaminoglycan binding proteins, Complement and coagulation cascades, and PI3K-AKT signaling pathways were upregulated in the WT-OIR group compared to the WT control. Interestingly, these pathways were significantly downregulated in the OIR disease model of PPT1-KO compared to the WT control group. These data suggest that PI3K-AKT-mediated immune response and inflammatory pathways may play a crucial role in regulating the initiation and development of OIR in vivo.

For further confirmation, we examined multiple marker proteins involved in the AKT–NF–κB-STAT3-VEGF signaling pathway and assessed the level of HIF-1α. Results showed that phosphorylated AKT (p-AKT) was significantly reduced in the OIR disease model of PPT1-KO compared to the WT control. Similarly, the NF-κB/p–NF–κB/STAT3 signaling cascade was significantly upregulated in WT-OIR compared to the WT control, whereas this cascade was significantly downregulated in the OIR disease model of PPT1-KO compared to the WT control. Consequently, the elevated expression levels of HIF-1α and VEGF in the WT-OIR group were significantly reduced in the PPT1-KO-OIR group. These findings align with in vitro data from HUVECs, indicating that VEGF expression is likely modulated by AKT–NF–κB-STAT3 inflammatory pathways.

Lastly, to examine superoxide levels in vivo, mouse retinas (P17) were frozen, dissected, and stained with dihydroethidium (DHE) to visualize ROS in both WT and PPT1-KO mice. The results showed that while the basal level of ROS in the mouse retina is similar between WT and PPT1-KO mice under normoxia, ROS levels in WT-OIR are significantly reduced, indicative of a state of relative hypoxia. Notably, ROS levels in PPT1-KO-OIR remain not notably changed compared to PPT1-KO.

3. Discussion

Although the role of protein palmitoylation in angiogenesis has remained elusive, the identification of Gpx1 as a candidate palmitoylated protein presents a unique opportunity to elucidate this mechanism. In this study, we demonstrated that Gpx1 is dynamically palmitoylated in endothelia cells, with its depalmitoylation being mediated by PPT1. Notably, an increased level of Gpx1 palmitoylation enhances its protein stability, and vice versa. Interestingly, Gpx1 palmitoylation negatively regulates angiogenesis. Physiologically, upon the induction of ROS, the level of Gpx1 palmitoylation is elevated. This, in turn, downregulates ROS and modulates HIF-1α expression, as well as the AKT–NF–κB-STAT3 inflammatory pathways, thereby influencing VEGF expression to restrain angiogenesis. Importantly, the targeting of PPT1 by DC661 significantly elevates Gpx1 palmitoylation, alleviates HIF-1α and VEGF expression, and attenuates angiogenesis in vivo (Fig. 9A).

Fig. 9.

Proposed mechanism of angiogenesis regulated by Gpx1 palmitoylation and targeted intervention.

(A) Gpx1 is dynamically palmitoylated in endothelial cells, with depalmitoylation mediated by PPT1. Notably, an increased level of Gpx1 palmitoylation enhances its protein stability, and vice versa. Interestingly, Gpx1 palmitoylation negatively regulates angiogenesis. Physiologically, upon the induction of ROS, the level of Gpx1 palmitoylation is elevated, which then downregulates ROS and modulates the expression levels of HIF-1α and VEGF to restrain angiogenesis. Consistently, targeting PPT1 with DC661 significantly elevates Gpx1 palmitoylation, alleviates HIF-1α and VEGF, and attenuates angiogenesis in vivo.

Previously, it was reported that Gpx1 activity may be enhanced by its phosphorylation [30], while the oxidation of selenium in Gpx1 may inactivate the enzyme, considering that Gpx1 is a selenocysteine-containing protein [31,32]. In this study, we showed that Gpx1 is modified by palmitoylation, primarily at cysteine-76 and cysteine-113 (Fig. 1). Whether this novel modification also modulates the enzymatic activity of Gpx1 remains unknown. Indirectly, it was shown that the levels of ROS (Fig. 7, Fig. 1H) and palm-Gpx1 (Fig. 7E and F) are concomitantly upregulated upon exposure to H₂O₂ in WT HUVECs, consequently, this results in the downregulation of ROS levels (Fig. 7, Fig. 3H). Meanwhile, the re-expression of Gpx1-2CS in Gpx1-KO HUVECs cannot efficiently alleviate the burden of ROS (Fig. 7C and D). These findings suggest that palmitoylation may be involved in regulating Gpx1 activity, which warrants further investigation.

As Gpx1 is a major antioxidant enzyme involved in metabolizing intracellular ROS to maintain cellular redox balance [33], its expression level is tightly regulated [34,35]. Hence, dysregulated Gpx1 expression has been associated with various pathological conditions e.g., oxygen tension [36,37], oxidative stress [[19], [20], [21]], and tumorigenesis [18,35]. The regulation of Gpx1 expression occurs at different levels, including transcriptional [20,38,39], translational [[40], [41], [42]], and post-translational [31,32]. In this study, we discovered that palmitoylation positively modulates the protein stability of Gpx1. This finding unveils a new means of post-translational regulation that controls Gpx1 levels and offers a novel intervention strategy for manipulating its protein expression.

Functionally, previous studies have demonstrated that Gpx1 is abnormally elevated in most types of cancer but its roles as tumor suppressor or promoter vary depending on the specific cancer type [34,35]. In this study, however, we have shown that an enhanced level of palm-Gpx1 attenuates angiogenesis, while inhibiting palm-Gpx1 or deleting Gpx1 promotes angiogenesis. This observation aligns with previous finding that increased pathological angiogenesis occurs in Gpx1-KO mice [25]. Interestingly, the potential role of Gpx1 palmitoylation in tumorigenesis and other related diseases remains to be determined.

Mechanistically, to elucidate whether VEGF regulation is exclusively dependent on ROS–HIF–1α signaling, we employed both mRNA-seq and mass spectrometry techniques. The integrated data analysis suggests that VEGF expression is also likely influenced by the AKT–NF–κB-STAT3 inflammatory pathways (Figs. S6–S7). This finding is consistent with previous studies that have demonstrated a close interaction between ROS and NF-κB signaling pathways. The relationship is complex, as ROS can both activate and inhibit NF-κB pathways due to its ability to act in multiple ways and locations simultaneously [29].

Strategically, since PPT1 is responsible for depalmitoylating palm-Gpx1 (Fig. 5A and B), and increasing Gpx1 palmitoylation attenuates angiogenesis (Fig. 5E and F), we thought to target PPT1 using DC661 to enhance the levels of palm-Gpx1. Intriguingly, we found that inhibiting PPT1 with DC661 suppresses angiogenesis both in vitro (cultured HUVECs, Fig. 6A–D) and in vivo (OIR mouse model, Fig. 6J and K). These findings suggest that intervening in Gpx1 palmitoylation, rather than directly targeting Gpx1 itself[[43], [44], [45]], could be a promising alternative therapeutic approach for combating pathological angiogenesis in varied types of diseases.

Taken together, our findings not only uncover the important roles of protein palmitoylation in angiogenesis but also suggest a mechanism wherein the PPT1-Gpx1 palmitoylation axis modulates angiogenesis and therefore identify PPT1 as a valuable target for suppressing pathological angiogenesis.

4. Materials and methods

4.1. Cell culture, transfection, and treatments

HEK-293T (CRL-11268) and HUVECs (CRL-1730) were obtained from ATCC. HRECs was obtained from BeNa Culture Collection (BNCC377527). HEK-293T and HUVECs were cultured in DMEM high glucose medium (Gibco) supplemented with 10 % fetal bovine serum (Gibco), 100 μg/mL streptomycin, and 100 U/mL penicillin at 37 °C in a 5 % CO₂ incubator. HRECs were cultured in Endothelial Cell Medium (ECM) medium (ScienCell), supplemented with 10 % fetal bovine serum (ScienCell), 100 μg/mL streptomycin, and 100 U/mL penicillin at 37 °C in a 5 % CO₂ incubator. For transfection experiments, Flag-Gpx1-C76S, Flag-Gpx1-C113S, Flag-Gpx1-C76/113S, and PPT1-Flag plasmids were transfected into HUVECs cells. Transfections were performed using Lipofectamine 3000 reagent (Invitrogen) and the reduced serum medium Opti-MEM (Life Technologies) following the manufacturer’s instructions. The cells were transfected for 24 h. The following reagents were used for cell treatments: DMSO (Sigma-Aldrich, Cat #D8418), 2-BP (Sigma-Aldrich, Cat #238422), PalmB (Millipore, Cat #178501), and DC661 (2 mg; MCE; Cat #HY-111621/CS-0088759).

4.2. Mass spectrometric analysis of palmitoylation in Gpx1

30 μg purified Flag-Gpx1 were digested using FASP [46]. Briefly, disulfide bonds were broken and blocked using 2 mM Tris (2-carboxyethyl) phosphine (TCEP) and 10 mM iodoacetamide (IAA), then proteins were transferred to 10 K filter, and cleaned sequentially using 8 M urea and 50 mM Tris-Hcl pH 6.8 at 13000 g, 20 °C. Proteins were digested with trypsin (V5113, Promega) at 1:50 (mass/mass) in 50 mM Tris-Hcl pH 7.5 at 37 °C for 16 h. Peptides were collected at 13000g, 20 °C, lyophilized and stored at −80 °C until use. Raw files were acquired with data dependent acquisition mode using Orbitrap Fusion Lumos (San Jose, Thermo Fisher). Peptide mixture were separated on EasyNano LC1200 system (San Jose, Thermo Fisher) using both C18 (3 μm, 75 μm × 15 cm, homemade) and C4 column (2.6 μm, 75 μm × 15 cm, Thermo Fisher) at a flowrate of 300 nl/min. For peptide separation with C18 columns, a 60-min linear gradient was set as follows: 5 % B (0.1 % FA in 80%ACN/H2O)/95 % A (0.1 % FA in H2O) to 10 % B in 5 min, 10 % B to 28 % B in 38 min, 28 % B to 38 % B in 8 min, 38 % B to 100 % B in 2 min and stayed 7 min for 100 % B. For peptide separation with C4 column, a 60-min linear gradient was set as follows: 5 % B (0.1 % FA in 80%ACN/H2O)/95 % A (0.1 % FA in H2O) to 10 %B in 5 min, 10 % B to 28 % B in 14 min, 28 % B to 38 % B in 18 min, 38 % B to 100 % B in 16 min and stayed 7 min for 100 % B. For the data acquisition a top 20 scan mode with MS1 scan range m/z 350–1550 was used and other parameters were set as below: MS1 and MS2 resolution was set to 120K and 30K; AGC for MS1 and MS2 was 4e5 and 1e5; isolation window was 1.6 Th, dynamic exclusion time was 15 s. Each precursor ion was fragmented with both HCD and EThcD. Collision energy of HCD was set to 32, and for EThcD collision energy was 25 and ETD reaction time was set automatically according to m/z and ion charge state of each precursor. Raw files were searched against target protein sequence using Byonic v2.16.11 (Protein Metrics). Searching parameter was set as follows: target-decoy searching algorithm; enzyme of trypsin (semi) with maximum number of 3 missed cleavages; precursor and fragment ion mass tolerance was set to 30 ppm and 50 ppm; variable modification was set to oxidation of M, deamidation of N, Q, carbamidomethylation of C, acetylation of protein N-term, palmitoylation of C, Y, S, T, W. For better identification of MS2 spectrums, a wildcard search function with ±500 Da mass range was applied to all identified peptides. An automatic score cut was used to remove low score peptides. A manual check was applied to further filter high confident palmitoylated cysteine sites. Modified peptides only with continuous b and y product ions can be considered as a high confident modified site.

4.3. Cycloheximide treatment

Cells were treated with Cycloheximide (CHX; 100 μM; Sigma, Cat #C7698) for varying periods (e.g., 0–8 h), after which the treated cells were harvested and processed for Western blot analysis.

4.4. Deletion of Gpx1 in HUVECs cells

We utilized the online tool http://crispor.tefor.net to design single-guide RNA (sgRNA) targeting the third exon of Gpx1 (Gene ID: ENST00000419783.3). To facilitate cloning, we introduced a BbsI restriction site. The sgRNA sequences were synthesized by Shanghai Bioligo Biotechnology. The Px458 vector was digested using BbsI (New England Biolabs #R3539), annealed with the gRNA, and then transferred to DH5α competent cells (Beijing Protein Biotechnology) for transformation. Ampicillin-resistant positive clones were screened and sequenced by Wuhan Genecreate Bioengineering. The recombinant plasmid was extracted using the Tiangen kit (Tiangen Biotechnology) for maximal yield. HUVECs cells were prepared and transfected with two pX458 vectors using Lipofectamine 3000 Transfection Reagent (Invitrogen). After 48 h, single fluorescent cells were sorted into a 96-well plate using FACS (BD Biosciences). Approximately two weeks later, single-cell colonies were obtained. DNA was extracted from these cells and subjected to PCR amplification and sequencing to identify positive clones. PCR conditions included 35 cycles: 94 °C for 2 min, followed by 30 s at 94 °C, 30 s at 55 °C, and 30 s at 72 °C, with a final extension at 72 °C for 2 min.

4.5. Protein extraction and acyl-biotin exchange (ABE)

To extract proteins from the testes, RIPA buffer (P0013B; Beyotime) supplemented with a protease inhibitor cocktail (P1006; Beyotime) was used to lyse the testes. The lysate was subjected to continuous rotation at 4 °C for 1 h to facilitate cracking. Subsequently, the lysate was centrifuged at 12,000 g for 10 min, and the supernatant was transferred to a new tube. Protein concentrations were determined using the BCA Protein Quantification Kit (E112-02; Vazyme Biotech Co., Ltd). For protein alkylation, the protein lysates were treated with 50 mM N-ethyl maleimide (NEM) for 90 min at room temperature to block free thiols. Excess NEM was removed through chloroform-methanol extraction and subsequently washed twice with methanol. To cleave the Cys-palmitoyl thioester linkages and biotinylate the proteins, the sample was incubated at room temperature for 120 min in the presence of 2 M hydroxylamine (H3NO) and 4 mM biotin-HPDP. After chloroform-methanol precipitation and two washes with methanol, the biotinylated proteins were affinity-bound to streptavidin agarose and subjected to overnight rotation at 4 °C. The bound proteins were eluted using a washing buffer and precipitated using methanol/chloroform precipitation. Finally, the proteins were dissolved in 50 μl of 8 M urea for Western blot analysis.

4.6. Western blot analysis and antibodies

Samples were separated on standard SDS-PAGE gels and subsequently transferred to Immobilon-P PVDF membranes (pore size 0.2 μm; EMD Millipore). The membranes were then blocked in 5 % (wt/vol) skimmed milk in TBS containing 0.1 % (vol/vol) Tween-20 (TBST) for 120 min. After blocking, the membranes were washed with TBST and incubated with primary antibodies overnight at 4 °C. Following TBST washing, the membranes were incubated with an appropriate horseradish peroxidase (HRP)-labeled secondary antibody for 1 h, and the signals were detected using an ECL kit (Tanon). The following primary antibodies were used: Gpx1 (Abcam, Cat # ab22604; 1:1000 for immunoblot), HRP-conjugated Mouse anti DDDDK-tag mab (ABclonal Cat # AE024; 1:1000 for immunoblot), β-Actin Rabbit mAb (ABclonal Cat # AC026; 1:10000 for immunoblot). For immunoblot, the following secondary antibody was used: HRP goat anti-rabbit IgG (H + L) (ABclonal, Cat # AS014; 1:5000 for immunoblot). β-Actin was utilized as an internal reference for equal sample loading.

4.7. Tube formation assay

To perform the tube formation assay, 50 μl of Matrigel was added to each well of a 96-well plate and incubated at 37 °C for 40 min in a 5 % CO₂ incubator. Subsequently, 2.0 × 10⁴ HUVECs were seeded into each well on the Matrigel. Tube formation was assessed by microscopy after 4 h. Three random images were captured per well, and Image-Pro Plus software was utilized to measure the length of the tubes. Each experiment was repeated three times.

4.8. CCK8 assay

For the cell proliferation assay, cells (3 × 10³/well) were seeded in gelatin-coated 96-well plates. Cell proliferation was assessed using a Cell Counting Kit-8 (CCK-8; Invigentech, USA). In the CCK-8 assay, 10 μL of CCK-8 solution was added to each well, and the 96-well plate was incubated at 37 °C for 2 h. The absorbance at 450 nm of each well was then measured using a Biotek reader (Infinite M200 Pro, Tecan) to determine the number of viable cells. All treatments were replicated at least three times.

4.9. Wound healing assay

The ibidi Culture-Insert 2 Well was placed on the cell culture surface, and two cell culture vessels were prepared. A suspension containing 1 × 10⁴ cells in 100 μL was applied to each well of the Culture-Insert 2 Well. Cells were incubated at 37 °C and 5 % CO₂ for at least 12 h until a confluent cell layer was formed. The Culture-Insert 2 well was gently removed using sterile forceps. After removal, the cell layer was washed with PBS to eliminate cell debris and non-adherent cells. Subsequently, 2 mL of medium containing 0.35 % serum was added to fill the μ-Dish. The culture dish was placed under a microscope, and the wound area was captured using a 10x objective lens at 0 h and 8 h, respectively.

4.10. Detection of ROS using CellROX® Deep Red

Cells were plated on coverslips and transfected with the indicated plasmids as described. They were then incubated with 800 μM H₂O₂ for 1 h and 3 h, or without H₂O₂. After that, the medium was replaced with medium containing CellROX® Deep Red reagent, and the cells were incubated at 37 °C in the dark for 0.5 h. Following incubation, the cells were washed three times with PBS and fixed with 4 % paraformaldehyde for 20 min. They were then immersed and washed three times with PBS. The slides were mounted with Dapi-Fluoromount-G (Electron Microscopy Sciences, Cat #17984–24). Images were captured using a bright-light Nikon (TI2-FP) microscope with 20 × objectives, using the same settings for all samples.

4.11. RNA extraction, cDNA synthesis, and RT-qPCR

Total RNA from cells and retinal tissue was isolated using an RNA Easy Fast Tissue/Cell Kit (DP451, Tiangen) according to the manufacturer's instructions. 1 mg of total RNA was transcribed into cDNA using PrimeScript™ RT Master Mix (RR036Q, Takara). Relative mRNA expression analyses were performed using TB Green™ Premix Ex Taq II (RR820A, Takara) on a Cobas™ z 480 (Roche) instrument. Real-time PCR was carried out in a total reaction volume of 20 μL, comprising 2 μL (100 ng) of cDNA, 10 μL of PCR master mix, 0.6 μL of each primer (10 μM), and 6.8 μL of RNase-free water. The sample expression results were averaged from triplicate amplification. The following specific primers were used:β-actin (From human): F 5′ to −3′, R 5′- CATGTACGTTGCTATCCAGGC-3′; VEGFA (From human): F 5′-AGGGCAGAATCATCACGAAGT-3′, R 5′-AGGGTCTCGATTGGATGGCA-3′; HIF-1α (From human): F 5′-ATCCATGTGACCATGAGGAAATG-3′, F 5′-ATCCATGTGACCATGAGGAAATG-3′; GPX1 (From human): F 5′-GCGGGGCAAGGTACTACTTA-3′, R 5′-CTCTTCGTTCTTGGCGTTCT-3′.β-Actin (From mouse): F 5′-GGCTGTATTCCCCTCCATCG-3′, R 5′-CCAGTTGGTAACAATGCCATGT-3′; VEGF (From mouse): F 5′-GCACATAGAGAGAATGAGCTTCC-3′, R 5′-CTCCGCTCTGAACAAGGCT-3′;

HIF-1α (From mouse): F 5′-GGGGAGGACGATGAACATCAA-3′, R 5′-GGGTGGTTTCTTGTACCCACA-3′. NF-κB (From human): F 5′-AACAGAGAGGATTTCGTTTCCG-3′ R 5′-TTTGACCTGAGGGTAAGACTTCT-3′; IL-6 (From human): F 5′- ACTCACCTCTTCAGAACGAATTG -3′, R 5′- CCATCTTTGGAAGGTTCAGGTTG -3′. The specific primers were synthesized by Wuhan Genecreat Co. Ltd (Wuhan, China), and all reactions were performed in triplicate with β-actin as the control.

4.12. Assay of Gpx1 activity

Gpx1-KO cells rescued with Gpx1 or Gpx1-2CS were subjected to treatment with or without H₂O₂ for 1 and 3 h. Following treatment, the cells were suspended in PBS and lysed on ice. The lysed cells were then centrifuged at 10,000g for 10 min at 4 °C, and the resulting supernatant was collected. The concentration of GSH was measured using a reduced glutathione (GSH) assay kit (Servicebio, Cat# G4305-48T) as per the manufacturer's instructions.

4.13. Mouse animal management

C57BL/6J (B6J) were purchased from Beijing Vital River Laboratory Animal Technology. PPT1-KO mice is a gift from Dr. Anil B. Mukherjee. The male and female mouse pups were designated as P0 when they were born. The pups and suckling mice were provided with free access to food and water and maintained on a 12/12 h-day/night cycle in a normal air environment for 7 days (P0–P7). On day P7, healthy pups and suckling mice were placed together in a glass oxygen chamber (Aipulab; Zhejiang; China), with the oxygen flow controlled at 0.50–0.75 L/min and the oxygen concentration maintained at 75 ± 2 %. The animals were kept in the glass oxygen chamber for five consecutive days. At P12, the pups were returned to a normal air environment and continued to be co-bred with the suckling mice. All animal procedures were performed according to guidelines approved by the committee on animal care at Xinxiang Medical University.

4.14. Animal treatment

Male and female OIR-C57BL/6J pups were randomly assigned to the control group (0.1 % DMSO) and the DC661 group (n = 3–5 in each group). DC661 was freshly prepared in a solution of 0.1 % DMSO diluted with corn oil. Pups in the control group were administered an equal amount of 0.1 % DMSO. One microliter of DC661 was administered via vitreoretinal injection at P14 and P16.

4.15. Flat-mounting of the retina

Mice were euthanized by carbon dioxide inhalation at 17 days after infection. The eyes were fixed with 4 % PFA at room temperature for 0.5–1 h, a hole was made at the pupil, and then fixed for 10 min. The eyes were then washed three times with PBS, 5 min each time. Retinas were dissected and stained overnight by immersion in Isolectin B4-594 (Alexa Fluor Cat #594-I21413, diluted 1:100) in PBS containing 1 mM CaCl2. The following day, the retinas were washed twice with PBS, flat-mounted on microscope slides, and embedded in fluorescence mounting medium (EMS 17984–25). Immunofluorescence images of the entire retina were captured using LEICA DMi8 STED (Ernst Leitz, Germany). In each whole mount, the number of pixels in the neovascular tufts was measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

4.16. Nuclear and cytoplasmic separation

Cytoplasmic and nuclear fractions were separated using a Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, Shanghai, China, #P0028) according to the manufacturer’s instructions. Protein concentrations were determined using a BCA protein assay kit and then subjected to WB analysis.

4.17. Frozen sections & DHE staining

Fresh retinal tissue was flash-frozen in liquid nitrogen, embedded in OCT, and stored at −80 °C for 1–2 h. Serial frozen coronal sections (10 μm) were cut using a cryostat (Leica CM1950). The sections were stained with DHE (Beyotime, Shanghai, China, #S0063) in the dark at 37 °C for 25 min, then washed three times with PBS. Slides were mounted with Dapi-Fluoromount-G (Electron Microscopy Sciences, Cat #17984–24). Images were captured using a bright-field Nikon TI2-FP microscope with a 20 × objective, applying the same settings for all samples. The intensity of DHE fluorescence was quantified using ImageJ software, showing means for DHE fluorescence from GCL to ONL. Product instructions were strictly followed to ensure the accuracy and reliability of the staining results.

4.18. Statistical analysis

Basic descriptive data are presented as means±standard errors of means (s.e.m). Statistical analyses were carried out using GraphPad Prism software. Statistical significance for two groups was determined by unpaired two-tailed Student’s t-test. For determining differences between more than two groups, one-way ANOVA followed by Tukey’s post hoc test was used. Statistical significance mentioned with ∗ represents significant differences. Statistical significance was defined as: ns, not significant, ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. The n value is reported as the number of cells or mice per group from at least 3 replicate experiments.

Funding

This study received funding from the National Natural Science Foundation of China (Grant No. 32371309) awarded to EYK. Introduction Plan of High-Level Foreign Experts (G2022026027L) to GEM, Major Program of Medical Science Challenging Plan of Henan province (SBGJ202102190) to GEM. The genetic modification of cell lines and mice was specifically supported by the 111 program (Grant No. D20036).

Data and materials availability

All materials are available upon reasonable request from the corresponding author. All data needed to evaluate the conclusions in the paper are present in the main paper or the supplementary materials.

CRediT authorship contribution statement

Yidan Ma: Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft. Xinxin Yuan: Formal analysis, Investigation, Methodology, Software, Visualization. Aodong Wei: Investigation, Software, Validation. Xiaopeng Li: Methodology, Visualization. Azim Patar: Data curation, Methodology, Supervision, Validation. Shaobo Su: Conceptualization, Methodology, Resources, Visualization. Songtao Wang: Methodology, Resources. Gaoen Ma: Conceptualization, Methodology, Project administration, Resources, Visualization. Jiangli Zhu: Conceptualization, Data curation, Formal analysis, Methodology, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Eryan Kong: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

YDM, JLZ, and EYK is listed as inventor on pending patent covering the targeting of PPT-Gpx1 axis as intervention strategy for treating pathological angiogenesis. All other authors declare that they have no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article.