Lactate-binding protein DNMT3A in HRMECs promotes angiogenesis by upregulating VEGFA through HIF-1α lactylation

Abstract

Background

Ocular neovascularization is a leading cause of blindness. Hypoxia is associated with retinal angiogenesis. Hypoxia results in lactate accumulation, which typically precedes protein lactylation, and this plays a crucial role in ocular neovascularization. However, the underlying mechanism remains unclear. Here, we investigate the role of the DNA methyltransferase, DNMT3A, in regulating lactylation following hypoxia in ocular neovascularization.

Results

DNMT3A controls endothelial cell angiogenesis via regulating lactate-derived HIF-1α lactylation. During oxygen-induced retinopathy progression, we detected increased levels of DNMT3A, HIF-1α lactylation, and vascular endothelial growth factor (VEGF) in endothelial cells. Exogenous lactate administration significantly enhances vascularization, while inhibiting lactate uptake in the presence or absence of DNMT3A prevents endothelial cell angiogenesis and attenuates HIF-1α lactylation. We demonstrate that modulating DNMT3A expression alters HIF-1α lactylation levels, influencing angiogenesis in vitro and in vivo.

Conclusions

DNMT3A facilitates lactate transport into the nucleus, where it promotes VEGFA upregulation through HIF-1α lactylation, thereby stimulating endothelial cell angiogenesis. Targeting the lactate-DNMT3A/HIF-1α lactylation/VEGFA pathway could provide new therapeutic strategies for treating ocular neovascularization disorders.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13059-025-03845-7.

Background

Retinal neovascularization (NV) diseases, such as retinopathy of prematurity (ROP), proliferative diabetic retinopathy (PDR), and wet age-related macular degeneration (wAMD), are among the most vision-threatening vascular conditions worldwide [1–3]. These diseases share common pathological features, including blood-retinal barrier damage, vascular leakage, inflammation, and ocular pathologic NV, potentially leading to retinal degeneration, detachment, and other severe complications [4–6]. An oxygen-induced retinopathy (OIR) animal model can partly simulate certain pathological processes of ROP and PDR [7], providing a crucial basis for preventing and treating retinal NV in humans. The OIR mouse model is extensively utilized in retinopathy-related studies, including ROP and PDR, as well as vascular pathogenesis of retinopathy (quantification of OIR in mice: a model of vessel loss or regrowth and pathological angiogenesis).

Hypoxia, which is influenced by microenvironmental changes, is a major factor promoting retinal NV in ischemic retinopathy [8]. Intracellular glycolytic metabolism increases under hypoxic conditions, leading to lactate accumulation. Although once considered merely a byproduct of energy metabolism, lactate has emerged as a potentially pivotal player in reshaping energy metabolism [9–11]. Beyond its traditional metabolic role, lactate functions as a novel signaling molecule, regulating macrophage polarization, T-cell activation, and tumor growth [12–15]. Lactate plays crucial regulatory roles in diverse cellular physiological and pathological processes, such as immune responses, neuronal energy metabolism, and neurodevelopment, facilitated by its transport carrier, the monocarboxylic acid transporter (MCT) [16–18]. MCTs constitute a family of integral membrane proteins that efficiently transport monocarboxylic acids, including lactate, across cell membranes. MCT1, which is widely distributed, is involved in endothelial cell (EC) lactate uptake [18, 19]. Lactate can serve as an epigenetic regulator, influencing the expression of specific genes through histone and nonhistone modifications [20, 21].

Lactylation, a posttranslational modification (PTM) induced by lactate, plays crucial roles in regulating macrophage polarization and transcription, accelerating ocular melanoma tumorigenesis, and facilitating cellular reprogramming [22–25]. Lactylation also influences protein function and participates in multiple physiological and pathological processes. For instance, lactate induces normal development through epigenetic modification of histone H3 acetylation on key eye-field transcription factors (Rax and Six3) [26]. The upregulation of YY1 lactylation in microglia is critical in NV pathogenesis [20]. We previously reported that lactylation of non-histone Ikzf1 (164 K) and YY1 (183 K) promotes autoimmune uveitis [15, 21]. Although lactate/lactylation signaling is known to play a significant role in regulating angiogenesis, its potential significance as a trigger in NV pathogenesis remains enigmatic and warrants further investigation. Insufficient tissue perfusion during hypoxia prompts organs to stimulate capillary growth through efficient metabolic and oxygen-sensitive mechanisms [27]. Hypoxia-inducible factor (HIF) might be the best means to explain the regulatory mechanisms of oxygen-sensitizing molecules [28]. Once activated, HIF promotes angiogenic factor production, initiating vascular growth [29]. Moreover, HIF plays a role in maintaining retinal vascular growth and dynamic homeostasis by regulating angiogenic factors such as vascular endothelial growth factor (VEGF) [30, 31]. NV involves various cellular processes and signaling pathways and is regulated by angiogenic factors such as VEGF and HIF [32]. Current studies have shown that VEGFA is recognized as a key mediator in ocular neovascularization and is a major target for current therapeutic interventions.

DNA methyltransferase 3 A (DNMT3A) plays a crucial role in genome-wide ab initio methylation and in establishing DNA methylation patterns during development [33]. In acute myeloid leukemia, HIF-1α forms a positive regulatory loop with DNMT3A, resulting in DNA hypermethylation [34]. Nonhistone protein methylation promotes acylation at lysine sites of relevant target genes, regulating downstream gene transcription. For example, methylation of p53 at lysine 372 (K372) enhances acetylation at lysine 382 (K382), activating the transcriptional regulatory activity of p53 [35]. DNMT3A is activated through its effects on histone deacetylase activity, leading to transcription inhibition [36].

Lactic acid accumulation, which typically occurs before lactylation through hypoxia in ocular NV, is involved in retinal angiogenesis; however, the underlying mechanism remains unclear. Therefore, this study aimed to examine whether DNMT3A overexpression or knockdown regulates HIF-1α lactylation and VEGFA expression following hypoxia. Notably, increased lactate levels significantly enhanced DNMT3A expression, exacerbating retinal angiogenesis by inducing VEGFA levels through HIF-1α lactylation. In vitro, lactate promoted hypoxia-induced activation of DNMT3A/HIF-1α lactylation/VEGFA in ECs. These findings could enhance our understanding of the role of lactate-DNMT3A in retinal angiogenesis, providing the foundation for innovative therapies targeting ocular neovascular diseases.

Results

Lactate promotes retinal angiogenesis in vivo

In the OIR mouse model, the hyperoxic conditions (P7-P12) inhibited angiogenesis in mouse pups, whereas hypoxic conditions promoted angiogenesis, peaking on day P17. We measured retinal lactate levels at seven time points in the OIR model to evaluate the relationship between lactate and vascular pathogenesis of retinopathy. Lactate levels correlated positively with angiogenesis severity (Fig. 1a), aligning with the results of a previous study [20]. Subsequently, to assess whether lactate affects OIR progression, we administered a lactic acid compound (L-lactate). We observed significantly enhanced pathological angiogenesis following L-lactate injection (Fig. 1b), indicating the pivotal role of lactate in retinal NV.

Fig. 1.

a Lactate production in the OIR retina at seven time points (P4, P7, P12, P14, P17, P22, and P24; n = 4 mice per group). b Representative images of CD31 by single-label immunofluorescence in the OIR and L-lactate-treated groups. The vascular fluorescence intensity of each retinal leaflet was calculated using ImageJ. The relative vascular fluorescence intensity in the experimental and control groups was calculated for statistical analysis. The red dotted line indicates neovascularization. Scale bar: top panel, 500 μm; below panel, 50 μm; *p < 0.05 and **p < 0.01

Lactate promotes retinal angiogenesis in vitro

Owing to the importance of hypoxia in lactate production, we explored the role of lactate in ECs by exposing human retinal microvascular endothelial cells (HRMECs) to hypoxia. The findings of several studies have indicated the role of monocarboxylate transporter protein 1 (MCT1) in brain ECs, which has been established to be associated with facilitating the transport of lactate, pyruvate, and ketone bodies across cell membranes to maintain lactate homeostasis [37]. To confirm the important role of lactate, we treated HRMECs with an MCT1 inhibitor (AZD3965) under hypoxic conditions and also investigated the effects of lactate on the angiogenic properties of HRMECs under hypoxia/lactate/AZD3965 conditions. Compared with the control group cells, HRMECs in the hypoxia and lactate groups were found to be characterized by enhanced spheroid sprouting, migration, proliferation, and tube formation (Fig. 2e, g, i and Additional file 1: Figs. S1d–g, S2a). However, these abilities were mitigated in the AZD3965 group (Fig. 2f, h, j and Additional file 1: Fig. S2b). VEGFA serves as a key angiogenic factor [38], and compared with HRMECs in the normoxic groups, we observed elevated levels of MMP9, VEGFA, and overall lactylation in hypoxic HRMECs (Additional file 1: Fig. S1a, b). To confirm the role of lactate in ECs, we treated HRMECs with L-lactate, the results of which revealed an upregulated expression of VEGFA and an overall increase in lactylation (Fig. 2a, b), whereas HRMECs treated with the MCT1 inhibitor AZD3965 were found to be characterized by reduction in the expression of VEGFA and overall levels of lactylation (Fig. 2c, d). These findings accordingly provided evidence to indicate that lactate regulates the angiogenic functions of ECs.

Fig. 2.

a HRMECs were treated with L-lactate (20 mM). The levels of classic angiogenic factors in HRMECs were measured via western blotting (WB) in the control and L-lactate groups (n = 3 per group). N, no significance; **p < 0.01. b Lactylation levels in HRMECs in the control and L-lactate (20 mM) groups were measured via WB. c HRMECs were treated with AZD3965 (8 nM) under hypoxic conditions. The levels of classic angiogenic factors in HRMECS were measured via WB in the control and AZD3965 groups (n = 3 per group). N, no significance; **p < 0.01. d Lactylation levels in HRMECs in the control and AZD3965 (8 nM) groups were measured via WB. e The angiogenic capacity of HRMECs was evaluated via a sprouting assay. Representative images of sprouting in the control and L-lactate (20 mM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. **p < 0.01. f Representative images of spheroid sprouting in the control and AZD3965 (8 nM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. **p < 0.01. g Migration was evaluated using a Transwell assay. Representative images in the control and L-lactate (20 mM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. **p < 0.01. h Representative images of cell migration in the control and AZD3965 (8 nM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. *p < 0.05. i Proliferation was evaluated via EdU staining. Representative images of proliferation in the control and L-lactate (20 mM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 100 µm. *p < 0.05. j Representative images of proliferation in the control and AZD3965 (8 nM) groups (n = 3 independent experiments, 3 images for each group); Scale bar: 100 µm. ***p < 0.001

Identification of specific lactate-binding proteins

We validated the critical role of lactate in angiogenesis in vitro and in vivo. To identify key lactate-binding proteins in HRMECs, we used the HuProt microarray, incubating it with biotin-labeled lactate (biotin-lactate) or biotin. We then employed Cy5-coupled streptavidin (Cy5-SA) to identify proteins with lactate-binding capacity (Fig. 3a). Figure 3b shows the molecular structures of lactate and biotin-lactate, whereas Fig. 3c displays a partially enlarged view of the biotin and biotin-lactate proteome microarrays. After filtering out nonspecific signals, we identified 222 candidate lactate-binding proteins from the HuProt microarray data. To standardize the data, we evaluated the mean IMean ratio (set to ≥ 1.414) of the two repeated spots. We then selected 222 proteins in descending order of the IMean ratio (Fig. 3d) and conducted Gene Ontology (GO) analysis (Fig. 3e). To further elucidate the functional roles of lactate-binding proteins, we performed Kyoto Encyclopedia of Genes and Genomes conducted a (KEGG) pathway enrichment analysis on these 222 proteins (Fig. 3e, f). Molecular function and KEGG analyses showed significant enrichment in actin binding and the focal adhesion pathway, indicating enhanced cell-extracellular matrix attachment activity and lactate endocytosis.

Fig. 3.

a Schematic of the 20 K Chip. b Molecular structures of L-lactate and biotin-labeled L-lactate. c Chip showcase. d Histogram from the 20 K Chip. e Gene Ontology enrichment analysis of 20 K Chip data. f KEGG pathway analysis of 20 K Chip data

Among these specific lactate-binding proteins, DNMT3A immediately caught our attention owing to its high IMean ratio (3.56) and its role as a multifunctional DNA methyltransferase with regulatory effects on angiogenesis [39]. We used another independent biophysical method to test binding of L-lactate to DNMT3A: we performed microscale thermophoresis (MST) using labeled RED-NHS-loaded DNMT3A. We measured a dissociation constant of 3.58 ± 0.27 μM for the interaction between L-lactate and DNMT3A by MST (Additional file 1: Fig. S1c), consistent with the previous results of L-lactate binding to DNMT3A. HIF-1α/VEGF signaling plays a significant role in angiogenesis in cerebral insult and cardiac remodeling [40]. We observed an upregulation of DNMT3A and HIF-1α levels in HRMECs subjected to hypoxic conditions and upregulated levels of DNMT3A without significant changes in HIF-1α levels in L-lactate-treated HRMECs (Fig. 4a, b). In contrast, there was a downregulation in the expression of DNMT3A without significant changes in HIF-1α levels in AZD3965-treated HRMECs under hypoxic conditions (Fig. 4c). Furthermore, hypoxia and lactate treatment increased HIF-1α lactylation in HRMECs, whereas AZD3965 treatment decreased HIF-1α lactylation, as validated via immunoblotting (IB) after immunoprecipitation (IP) assays (Fig. 4d-f). These results indicated that L-lactate-specific binding protein DNMT3A has a positive effect on HIF-1α lactylation.

Fig. 4.

a HRMECs were subjected to hypoxia for 24 h. DNMT3A and HIF-1α levels in HRMECs were measured via WB in the control and hypoxic groups (n = 3 per group). *p < 0.05 and ***p < 0.001. b HRMECs were treated with L-lactate (20 mM). DNMT3A and HIF-1α levels in HRMECs were measured via WB in the control and L-lactate groups (n = 3 per group). N, no significance; *p < 0.05. c HRMECs treated with AZD3965 (8 nM) under hypoxic conditions. DNMT3A and HIF-1α levels in HRMECs were measured via WB in the control and L-lactate groups (n = 3 per group). N, no significance; *p < 0.05. d Hypoxia increases HIF-1α lactylation. HIF-1α lactylation in control or hypoxic HRMECs was detected using the Pan anti-Kla antibody (n = 3 per group). *p < 0.05. e L-lactate increases HIF-1α lactylation. HIF-1α lactylation in control or L-lactate-treated HRMECs was detected using the Pan anti-Kla antibody (n = 3 per group). *p < 0.05. f AZD3965 decreases HIF-1α lactylation. HIF-1α lactylation in control or AZD3965-treated HRMECs was detected using the Pan anti-Kla antibody (n = 3 per group). ***p < 0.001

DNMT3A may promote angiogenesis through HIF-1α lactylation after hypoxia in HRMECs

To explore the role of DNMT3A in angiogenesis, we overexpressed and knocked down DNMT3A in HRMECs. Significantly increased DNMT3A, VEGFA, Pan-Kla, and HIF-1α-Kla levels were observed in the overexpression group, whereas DNMT3A, VEGFA, Pan-Kla, and HIF-1α-Kla levels were decreased in the knockdown group (Fig. 5a–e). To further confirm the role of DNMT3A, we assessed the spheroid sprouting, migration, proliferation, and tube formation capacities of HRMECs. These abilities were enhanced in DNMT3A-overexpressing HRMECs under hypoxic conditions and lactate treatment; however, these abilities were reduced in the DNMT3A-knockdown group compared with those in the control group (Fig. 5f, h, j and Additional file 1: Fig. S2c-f, h). Furthermore, no difference in the abilities of HRMECs to overexpress or knock down DNMT3A was observed following AZD3965 treatment (Fig. 5g, i, k and Additional file 1: Fig. S2g), suggesting that DNMT3A requires lactate to influence endothelial functions. To confirm the target of HIF-1α in response to lactate treatment, we performed ChIP-qPCR with HIF-1α on the VEGFA promoter, which revealed that lactate promotes HIF-1α binding to the VEGFA promoter (Additional file 1: Fig. S2i). Moreover, no difference was observed in the methylation levels of the HIF-1α promoter in HRMECs following lactate treatment, ruling out the involvement of DNMT3A in promoting the downstream cascade reaction through HIF-1α methylation (Additional file 1: Fig. S3a, b). These findings suggested the potential involvement of the lactate/DNMT3A/HIF-1α-Kla/VEGFA axis in EC-mediated angiogenesis.

Fig. 5.

a HIF-1α, DNMT3A, and VEGFA levels in DNMT3A-overexpressing HRMECs (n = 3 per group). N, no significance; *p < 0.05 and **p < 0.01. b HIF-1α, DNMT3A, and VEGFA levels in DNMT3A-knockdown HRMECs (n = 3 per group). N, no significance; **p < 0.01, ***p < 0.001. c Lactylation levels in HRMECs in the overexpressing (oe)-control and oe-DNMT3A groups were measured via WB. d Lactylation levels in HRMECs in the silencing (sh)-control and sh-DNMT3A groups were measured using WB. e Oe-DNMT3A increased HIF-1α lactylation, whereas sh-DNMT3A decreased HIF-1α lactylation (n = 3 per group). *p < 0.05. f Representative images of sprouting in the control and DNMT3A overexpression/knockdown groups after L-lactate treatment (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. *p < 0.05, **p < 0.01. g Representative images of sprouting in the control and DNMT3A overexpression/knockdown groups after AZD3965 treatment (n = 3 independent experiments, 3 images for each group); N, no significance; Scale bar: 50 µm. h Representative images of cell migration in the control and DNMT3A overexpression/knockdown groups after L-lactate treatment (n = 3 independent experiments, 3 images for each group); Scale bar: 50 µm. **p < 0.01, ***p < 0.001. i Representative images of migration in the control and DNMT3A overexpression/knockdown groups after AZD3965 treatment (n = 3 independent experiments, 3 images for each group); N, no significance; Scale bar: 50 µm. j Representative images of proliferation in the control and DNMT3A overexpression/knockdown groups after L-lactate treatment (n = 3 independent experiments, 3 images for each group); Scale bar: 100 µm. *p < 0.05. k Representative images of proliferation in the control and DNMT3A overexpression/knockdown groups after AZD3965 treatment (n = 3 independent experiments, 3 images for each group); N, no significance; Scale bar: 100 µm

Lactate/DNMT3A/HIF-1α-Kla/VEGFA signaling pathway may promote angiogenesis in vivo

We used the OIR model to explore the role of the lactate/DNMT3A/HIF1α-Kla/VEGFA signaling pathway in ECs during the vascular pathogenesis of retinopathy. Following modeling, significantly increased levels of VEGFA, Pan-Kla, DNMT3A, and HIF-1α were observed on P17, coinciding with peak NV in OIR, within retinal neovascular ECs (Fig. 6a, b, d and Additional file 1: Fig. S4a). We also observed significantly upregulated HIF-1α lactylation levels in the retina of OIR mice compared with that in the age-matched control group (Fig. 6c). Given the critical proangiogenic role of the lactate/DNMT3A/HIF-1α-Kla/VEGFA axis in ECs, we evaluated the therapeutic potential of targeting DNMT3A with L-lactate in vivo. We observed upregulated HIF-1α lactylation and increased levels of VEGFA and DNMT3A in the retinal ECs of L-lactate-treated mice (Fig. 6e–h). To further confirm the role of DNMT3A in lactate transport, we determined the intracellular distribution of GFP-DNMT3A at three time points via live-cell imaging. Lactate treatment promoted the nuclear influx of GFP-DNMT3A, with increased time-dependent influx at selected time points (Fig. 7a and Additional files 2, 4 and 5: movies S1-S2). Subsequent to L-lactate treatment, we examined the lactate content in HRMECs in the control and DNMT3A-knockdown groups, which revealed that compared with the control group, lactate contents were lower in the DNMT3A-knockdown group (Fig. 7b). Furthermore, we found that the addition of L-lactate after DNMT3A knockdown in HRMECs did not increase intracellular lactate content (Fig. 7c), consistent with previous hypotheses that DNMT3A protein facilitates lactate transport.

Fig. 6.

a Representative images of CD31 co-stained with VEGFA in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ****p < 0.0001. b Representative images of CD31 co-stained with Pan-Kla in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ****p < 0.0001. c Representative images of HIF-1α co-stained with Pan-Kla in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ***p < 0.001. d Representative images of CD31 co-stained with DNMT3A in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. **p < 0.01. e Representative images of CD31 co-stained with VEGFA in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ***p < 0.001. f Representative images of CD31 co-stained with Pan-Kla in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ****p < 0.0001. g Representative images of HIF-1α co-stained with Pan-Kla in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. ***p < 0.001. h Representative images of CD31 co-stained with DNMT3A in retina samples. The relative fluorescence intensity was calculated and measured by Image J (n = 4 images per group); Scale bar, left, 500 μm, right, 50 μm. **p < 0.01

Fig. 7.

a Representative images of GFP-DNMT3A distribution at three time points in HRMECs (Left panel); statistical plot (right panel) (n = 3 per group); Scale bar: 5 µm. *p < 0.05, **p < 0.01. b Quantification of lactate in the control and DNMT3A-knockdown groups after L-lactate treatment. (n = 4 per group). *p < 0.05. c Quantification of lactate in the control and L-lactate groups in DNMT3A-knockdown HRMECs (n = 4 per group). N, no significance

Discussion

Lactate/lactylation plays vital roles in proliferative vasculopathy, including ROP, PDR, and wAMD [20, 41]. In vivo injections of lactate compounds significantly ameliorate OIR severity, and adoptive transfer of L-lactate into untreated hosts can induce OIR [20]. Our study demonstrated that the lactate-specific binding protein DNMT3A, combined with lactate-derived HIF-1α lactylation, plays an important role in regulating angiogenesis. We observed upregulated DNMT3A, lactate, VEGFA, and HIF-1α-Kla levels in ECs of OIR mice, with L-lactate injection promoting OIR progression (Fig. 8). 20 K Chip sequencing revealed enrichment in the focal adhesion pathway, associated with strong cell-extracellular matrix communication. Knockdown/overexpression of DNMT3A regulated retinal NV through the lactate/HIF-1α-Kla/VEGFA signaling axis, suggesting that DNMT3A plays a crucial role as an important lactate transport-associated protein.

Fig. 8.

Schematic illustrating the molecular mechanism by which lactate-DNMT3A in HRMECs contributes to angiogenesis. Under hypoxic conditions, lactate-DNMT3A is upregulated in HRMECs in response to hypoxia. HIF-1ɑ lactylation promotes VEGFA levels and contributes to NV

The VEGF family of growth factors regulates pathological angiogenesis and increases vascular permeability in significant eye diseases, including wAMD and PDR [42, 43]. VEGFA serves as a critical protein in wAMD and PDR onset and progression, acting as a major regulator of blood vessel formation and functions [44, 45]. It regulates several EC processes, including sprouting, migration, and proliferation [46–48]. HIF-1α typically regulates VEGFA, and the HIF-1α/VEGFA/ERK pathway contributes to treating VEGF-nanofiber membranes against chronic cerebral hypoperfusion-induced neuronal injury [49]. HIF-1α-induced VEGFA expression offers cardioprotective effects and aids in infarcted heart repair [49, 50]. However, previous studies have not investigated classical proangiogenic factors in relation to EC lactylation. Furthermore, the potential responsiveness of EC functions to lactylation-induced factors remains unexplored.

DNA methylation, facilitated by epigenetic alterations, is established and maintained by a group of DNA methyltransferases (DNMT1, DNMT3A, DNMT3B, and DNMT3L) [51]. DNMT3A critically regulates the epigenetic silencing of the EC marker genes [52–54]. It negatively regulates HEY2 and EFNB2 expression in human umbilical artery ECs in a promoter-metal-independent manner, thereby inhibiting angiogenesis [53]. DNMT3A localizes to the promoter region of ubiquitin B (UBB) and epigenetically inhibits UBB transcription, enhancing VEGFA-dependent angiogenesis [54]. DNMT3A methylation can both inhibit and promote angiogenesis. Based on our experiments and the 20 K Chip analysis results, we propose that DNMT3A may function as a lactate transport-associated protein or lactase. A recent study demonstrated that the sequence of lactyl-CoA (Ydif) in Escherichia coli is highly conserved, particularly in the EXGXXG and GXGGF motifs [55]. However, we did not find similar motifs in HRMEC cells. We explored the regulatory effects of DNMT3A as the lactate transport-associated protein and its effect on HRMEC angiogenesis.

Lactylation, a posttranslational modification (PTM), promotes and activates profibrotic activity in macrophages [56, 57]. Glycolysis-derived lactate serves as a substrate for histone and non-histone lactylation, directly activating downstream gene transcription [20, 22]. Apart from its role as a glycolysis end product, lactate functions as a signaling molecule, revealing novel functions through lactylation [58]. These findings provide novel insights into retinal angiogenesis.

VEGFA effectively is an effective cytokine that regulates angiogenesis through HIF-1α. Hypoxia induces HIF stabilization and increases VEGFA levels in retinal pigment epithelium (RPE) cells [59]. Under hypoxic conditions, HIF translocates to the nucleus, binds to hypoxia response elements, and induces VEGFA expression [60]. Our study revealed that HIF-1α hyperlactylation promotes VEGFA expression, enhancing angiogenesis. These findings explain the mechanisms by which PTMs regulate VEGFA expression in retinal NV.

The progression of angiogenesis has been established to be closely associated with metabolic pathway alterations, particularly a dysregulation of glycolysis [20, 61]. Lactylation has emerged as a novel regulatory mechanism in biological processes [22]. Our findings in the present study indicate that HIF-1α lactylation represents a previously unrecognized mechanism, whereby DNMT3A promotes the transport of lactate into the nucleus, thereby having deleterious effects on angiogenesis. This accordingly presents new avenues for studying PTMs in disease progression and may contribute to the development of therapeutic approaches for the treatment of pathological NV.

Nevertheless, our study also has some limitations. First, although our results showed that DNMT3A may function as a potential lactate transporter protein, promoting HIF-1α lactylation in vitro, we were unable to fully elucidate the mechanisms whereby lactate upregulates DNMT3A. We also could not rule out the possibility of indirect effects of DNMT3A or or identify the specific HIF-1α lactylation sites that promote retinal angiogenesis. Further studies, including WT/DNMT3A conditional knockdowns, are warranted to validate our in vivo findings. In addition, the regulation of other angiogenesis-related genes through HIF-1α lactylation requires further exploration. Second, although lactylation appears to significantly regulate angiogenesis and differential protein lactylation was observed in HRMECs, the enzymes involved in lactylation (writers, readers, and erasers) and their interactions with other modifications remain largely unknown. Further research is warranted to elucidate these underlying mechanisms.

Conclusions

In this study, we employed the 20 K Chip to map lactate-specific binding proteins and identified the role of DNMT3A in facilitating lactate entry into the nucleus. Our findings also revealed that hypoxia increases HIF-1α lactylation in HRMECs, and HIF-1α hyperlactylation enhances VEGFA expression, contributing to NV. These findings highlight the potential role of lactylation in regulating angiogenesis progression. We further investigated the mechanism by which DNMT3A knockdown or overexpression attenuates or enhances its inhibitory/promotional effects on angiogenesis. Our study also provides preliminary insights into lactate binding proteins, their effect on angiogenesis-related gene expression, and PTM regulation. These findings lay the groundwork for future scientific investigations and clinical applications.

Methods

Animal handling

Male and female mouse pups were randomly assigned to the control (PBS) and experimental (L-lactate) groups. At postnatal day 14 (P14), 1 µL of L-lactate (1 g/kg; 10 mM, L7022; Sigma-Aldrich) [62, 63], was administered via intravitreal injection into one eye, whereas an equal volume of PBS was injected into the other eye. OIR was induced as previously described [20]. All experimental pups were housed under a 12-h light/dark cycle at 25 °C.

HRMECs cultures

HRMECs (Cell Systems, USA) were cultured in EGM-2 BulletKit medium (CC-3156CC-4176; LONZA) at 37 °C and 5% CO₂. All cells were identified and confirmed to be mycoplasma-free using the Applied Biosystems PCR Mycoplasma Test Kit (4,460,623, Applied Biosystems). For in vitro experiments, HRMECs were cultured for 24 h as previously described (20 mM L-lactate and 8 nM AZD3965 [5 mg, HY-12750; Med Chem Express] were added separately) [63–65].

Quantification of lactate concentration

Retinal tissue lactate content was determined using an LA Assay Kit (BC2230; Solarbio) following the manufacturer's instructions. Samples from different treatment groups were subjected to a reduction reaction with reducing agent solutions. The absorbance of the resulting purple substance was measured at 570 nm (Thermo Varioskan LUX, USA) to determine the lactate content.

Immunofluorescence

Eyes at day P17 were fixed with 4% paraformaldehyde (2 h). Retinal samples were separated from the eyeballs and blocked with 0.3% Triton-X-100 (P0096; Beyotime) and normal goat serum (at 37 °C, 1 h). The samples were incubated overnight at 4 °C with primary antibodies. The retinas were washed four times with PBS and incubated with the respective secondary antibodies for 1 h at 37 °C. Images were captured using a confocal microscope (Leica, Germany). Primary antibodies included CD31 (ab9498, 1:800; Abcam), DNMT3A (rabbit, 20,954–1-AP, 1:1000; Proteintech), Pan-Kla (PTM-1401, 1:250), VEGFA (rabbit, 19,003–1-AP, 1:200; Proteintech), and HIF-1α (36,169, 1:1000; Cell Signaling Technology).

Western blotting

Proteins from retinal tissues and cell lysates were extracted using the RIPA lysis buffer (HY-K1001; Med Chem Express). Samples (20 μg) were separated via 4–20% polyacrylamide electrophoresis gel and transferred onto PVDF membranes (Millipore, MA, USA). The following primary antibodies were used: Pan-Kla (PTM-1401, 1:1000); DNMT3A (rabbit, 20,954–1-AP, 1:1000; Proteintech), HIF-1α (#36,169, 1:1000; Cell Signaling Technology), HIF-1α (mouse, 66,730–1-lg, 1:1000; Proteintech), VEGFA (rabbit, 19,003–1-AP, 1:1000; Proteintech), MPP2 (sc-13594, 1:400; Santa Cruz Biotechnology), MPP9 (sc-393859, 1:400; Santa Cruz Biotechnology), and β-actin (20,536–1-AP, 1:5000; Proteintech). Veri-Blot for IP detection (ab131366, 1:1000; Abcam) was used for IB after IP.

20 K Chip analysis

Protein-binding microarray chips were provided by CDI Laboratories (Johns Hopkins Medical Institutions, USA). Microarray proteome analysis was performed by the Wayen Biotechnology Company (Shanghai, China). The proteome microarrays were blocked with blocking buffer (5% bovine serum albumin in 1 × Tris-buffered saline with Tween 20) for 1.5 h at 25 °C and then washed with 1 × TBST for 5 min. The blocked proteome microarrays were incubated with biotin-L-lactate (10 µmol/L) for 1 h and then washed thrice with TBST (5 min each). Cy5-streptavidin (1:1000) was applied and incubated, followed by three 5-min TBST washes. The microarrays were spun dry for 2 min and scanned using an Axon GenePix 4000 B scanner. Images were extracted using the GenePix Pro 6.0 software (Axon Instruments, USA). Protein spots with a Z-score < 2.8 in biotin-treated microarrays were identified as candidate-positive proteins. Enrichment analysis (KEGG and GO) was conducted using the ClusterProfiler package in R Studio.

Microscale thermophoresis (MST)

The equilibrium dissociation constant (K_D) value of the binding of L-lactate to DNMT3A was measured using NanoTemper Monolith X (Nanotemper Technologies, Germany). DNMT3A (P0647, Fine Test) carrying a polyhistidine-tag (His-tag) was labeled with the RED-NHS dye for 30 min at room temperature in the dark according to the manufacturer’s instructions (MO-L011, Nanotemper Technologies). L-lactate solution was serially diluted in the reaction buffer (50 mM HEPES buffer, pH = 7.4, containing 0.05% Tween-20 for L-lactate). Then, labeled DNMT3A protein was added to the serial dilution of the compound in a 1:1 volume ratio. After incubation for 30 min at room temperature, the samples were examined with NanoTemper Monolith X in Monolith Capillaries (MO-K022, Nanotemper Technologies). K_D values were determined using the Monolith Affinity Analysis K_D fit from triplicate experiments.

Lentivirus infection

HRMECs were seeded in 6-well plates (3 × 10⁵ cells/well) with ECM containing 10% FBS for 24 h before lentiviral infection. Lentiviral vectors, including the empty negative control (pcSLenti-EF1-EGFP-P2A-Puro-CMV-MCS-3 × FLAG-WPRE), DNMT3A-overexpression vector (pcSLenti-EF1-EGFP-P2A-Puro-CMV-DNMT3A × FLAG-WPRE), and DNMT3A-underexpression vector (pSLenti-U6-shRNA (DNMT3A)-CMV-EGFP-F2A-Puro-WPRE) were obtained from OBiO Technology (Shanghai, China). The lentiviruses were added to cells at an MOI of 30 with polybrene (1 μg/μL) following the manufacturer's instructions. The medium was replaced after 24 h of incubation, and cells were cultured for an additional 48 h before use.

Coimmunoprecipitation

Target proteins were immunoprecipitated using an IP kit (ab206996; Abcam). Protein samples were incubated with IP antibodies overnight at 4 °C following the manufacturer's instructions. Protein A/G beads (30 µL) were prewashed twice with 1 × wash buffer and then incubated with the lysates for 6 h. After three washes, bound proteins were processed via IB using the Veri-Blot for IP Detection reagent. In subsequent western blot experiments, a 10-µL input was used for experimental group protein samples, whereas a 5-µL input was used for HIF-1α-IP samples.

Sprouting assay

The sprouting assay was performed as reported previously [66]. Cells (80 000 per group) were transferred to a 15-mL conical tube. EGM-2 BulletKit medium with FBS was added to 4 mL total volume, followed by 1 mL methyl cellulose stock solution, and incubated at 37 °C and 5% CO₂ for 24 h. Spheroids were suspended in 2 mL collagen medium (4 mL collagen original solution, 0.5 mL 10 × medium 199, 200 μL 0.2 N NaOH) supplemented with 20% FBS. Subsequently, spheroid–collagen was seeded in 24-well plates (1 mL per well) and incubated at 37 °C and 5% CO₂ for an additional 24 h. Images were obtained under a microscope (Leica, Germany), and the total sprout length was quantified by measuring the cumulative sprouting length of all sprouts per spheroid using ImageJ (1.8.0; National Institute of Mental Health, USA).

Migration assay

HRMECs (1.0 × 10⁴ cells/mL) were seeded in 8-μm pore size filters of Transwell chambers (14,347; BIOFIL) and incubated at 37 °C and 5% CO₂ for 24 h. Subsequently, HRMECs were fixed with 4% paraformaldehyde for 30 min and stained with 1% crystal violet for 15 min. Migrated cells were counted under a microscope (Leica, Germany). Each experiment was performed in triplicate.

Proliferation assay

HRMECs were seeded in 48-well plates (3 × 10⁴ cells/well) with EGM-2 containing 10% FBS and randomly assigned to four treatment groups: normoxia, hypoxia, normoxia + lactate, and hypoxia + lactate. Each group received the respective treatment. Cells were treated for 24 h before being incubated in 500 μL EGM containing 100 μM EdU for 30 min. Cells were then fixed, permeabilized, and stained following the manufacturer's instructions. Images were captured under a fluorescence microscope (Leica, Germany) and analyzed using the ImageJ software. EdU-positive cells were counted in nine random areas per group.

Tube formation assay

The tube-like structure formation capacities of HRMEC cells in different groups were evaluated using Matrigel (3432–010-01; R&D Systems), which was applied to the wells of 96-well plates, followed by incubation at 37 °C for 50 min. Treated HRMECs (2 × 10⁴ cells/well) were seeded on the Matrigel and then incubated at 37 °C for 6 h. Tubes formed in three randomly selected fields were photographed under a microscope (Leica, Germany), the lengths of which were measured using the ImageJ software (1.8.0; National Institute of Mental Health, USA).

Methylation-specific PCR (MSP)

MSP was used to determine the methylation status of the HIF-1α promoter. Genomic DNA was extracted from HRMECs using the TIANamp Genomic DNA Kit (DP304; TIANGEN) following the manufacturer's instructions. The methylation‐specific primers used for MSP amplification were HIF-1α-MF (5′-AGATTAAAGGAAGGGTTTGTTGTTAC-3′) and HIF-1α-MR (5′-CTCGACCTCAATACTAAACACGAT-3′). For the unmethylation‐specific reaction, HIF-1α-UF (5′-GATTAAAGGAAGGGTTTGTTGTTAT-3′) and HIF-1α-UR (5′-CCTCAACCTCAATACTAAACACAAT-3′) primers were employed. Purified DNA samples were treated with bisulfite (RQM011; Bio. Ruqi). PCR amplification was performed using 50-μL reactions using TaKaRa EpiTaq™ HS (R110A; TaKaRa). PCR conditions comprised 40 cycles of predenaturation (98 °C, 10 s), annealing (55 °C, 30 s), and extension (72 °C, 30 s). The products were analyzed using gel electrophoresis.

ChIP assay

ChIP assays were performed as described previously [20]. Briefly, after fixation, 1 × 10⁷ HRMECs were resuspended in ChIP buffer containing a 1 × Protease Inhibitor Cocktail, then cross-linked with 1% formaldehyde, with immunoprecipitation being performed overnight using a HIF-1α specific antibody (#36,169; Cell Signaling Technology) and IgG (#3900; Cell Signaling Technology) as a control, following the instructions provided with a SimpleChIP Plus Enzymatic Chromatin IP Kit (9004S; Cell Signaling Technology). The PCR primers used for the VEGFA site in the human HIF-1α promoter have been described previously [67]. To validate HIF-1α binding to the VEGFA promoter, we used the forward primer 5ʹ-AGACTCCACAGTGCATACGTG-3ʹ and reverse primer 5ʹ-AGTGTGTCCCTCTGACAATG-3ʹ for RT-quantitative PCR analysis [20].

Statistical analysis

Data are presented as the mean ± SD and were analyzed using SPSS 24.0. Comparisons between the two groups were performed using the Student's t-test or Mann–Whitney U test based on data normality. Multiple group comparisons were conducted using one-way analysis of variance or the Kruskal–Wallis test (*p < 0.05, **p < 0.01, ***p < 0.001).

Authors’ contributions

Author information XTW and SPH conceived this study and designed the experiments. XTW, JXH, and WF performed the experiments, analyzed the data, and wrote the manuscript. XRL and WQL conducted in vitro experiments and contributed to data analysis; RNL, HY, and WXY helped to conduct the Co-IP and IB experiments. XTW, NL, QZ, and JYL analyzed the 20K-Chip data. SPH and NL conceptualized the study, supervised the experiments, and revised the manuscript. All authors read and approved the final manuscript.

Authors’ information

XTW and SPH conceived this study and designed the experiments. XTW, JXH, and WF performed the experiments, analyzed the data, and wrote the manuscript. XRL and WQL conducted in vitro experiments and contributed to data analysis; RNL, HY, and WXY helped to conduct the Co-IP and IB experiments. XTW, NL, QZ, and JYL analyzed the 20 K-Chip data. SPH and NL conceptualized the study, supervised the experiments, and revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by National Natural Science Foundation Project of China (82371045, 82401240), Youth Beijing Scholar (No.076), Beijing Municipal Public Welfare Development and Reform Pilot Project for Medical Research Institutes (PWD&RPP-MRI, JYY2023-6), China Postdoctoral Science Foundation (2023M740450), and the Chongqing Natural Science Foundation (CSTB2022NSCQ-MSX0144).

Data availability

The raw data of protein-binding microarray chip are available in the Protein Microarray Database with accession number PMDA195 [68]. Microscopy images are available on Figshare (10.6084/m9.figshare.30043045.v1) [69].

Declarations

Ethics approval and consent to participate

Experimental mice (male and female C57BL/6 J) were obtained from Chongqing Medical University (Chongqing Experimental Animal Center, Chongqing, China). All animal experiments were approved by the Ethics Committee of First Affiliated Hospital of Chongqing Medical University (approval number: 2019–101). All procedures followed the guidelines provided in the Animals in Ophthalmic and Vision Research statement.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.