Conformational Dynamics of PKM2 Regulate Hypoxia-Induced Pathological Retinal Angiogenesis

Abstract

Purpose

Because anti-vascular endothelial growth factor therapies have limited effect in some cases of retinal neovascularization, there might be other potential pathways involved. Pyruvate kinase M2 (PKM2), a major isoform of pyruvate kinase in tumor cell and human retina, has been broadly studied in oncology and believed to play key roles in tumor growth and invasion. However, its role in retinal angiogenesis remains unclear. In this study, we aimed to explore the contribution of PKM2 conformational dynamics on pathological neovascularization of the retina.

Methods

We used the oxygen-induced retinopathy (OIR) mouse model and the hypoxia-exposed human retinal microvascular endothelial cell (HRMEC) model to evaluate PKM2 conformational dynamics. DASA-58, a small-molecule activator that increase formation of tetrameric PKM2, was used to evaluate the effects of PKM2 tetramerization.

Results

Hypoxia induced phosphorylation-dependent monomerization of PKM2. The monomer translocated to the nucleus, where it interacted with hypoxia-inducible factor-1α (HIF-1α) to promote angiogenic and glycolytic gene expression. In vitro, treatment with DASA-58 induced the formation of tetrameric PKM2 and prevented its nuclear translocation, which further led to suppressed HIF-1α signaling, reduced glycolysis, and inhibited retinal neovascularization. Pyruvate dehydrogenase kinase 1 (PDHK1) was identified as an upstream modulator of PKM2 phosphorylation. In vivo, DASA-58 treatment led to the reduction in neovascularization, vascular leakage, and preserved retinal thickness, and improved visual function in OIR mice.

Conclusions

PKM2 conformational dynamics represents a novel regulatory mechanism of hypoxia-induced retinal neovascularization. Transformation of PKM2 tetramers induced by DASA-58 may provide a sound therapeutic approach for neovascular diseases including retinopathy of prematurity (ROP).

Retinopathy of prematurity (ROP) is a proliferative disorder affecting the retinal vasculature and poses a serious threat to visual health in premature infants.¹ In recent decades, the incidence of ROP has increased with the rising survival rate of preterm infants.² The treatment of ROP mainly includes laser therapy and intraocular injection of anti-vascular endothelial growth factor (VEGF) agents, all of which carry a risk of complications. Given the limitations of VEGF-targeted therapies, there is a growing need to explore alternative molecular pathways that regulate pathological retinal neovascularization in ROP.

The pyruvate kinase (PK) is the rate-limiting enzyme in the glycolysis and converts phosphoenolpyruvate to pyruvate. Mammalian cells contain two PK genes, one encoding liver-type PK (PKL) and red blood cell-type PK (PKR), and the other encoding pyruvate kinase M1 (PKM1) and pyruvate kinase M2 (PKM2).³ The four isoforms of PK are differentially expressed in various human tissues.⁴ PKM2 is the major isoform of PK in tumors and retinas.⁵ Whereas PKM1 primarily functions as a pyruvate kinase, PKM2 exhibits a more complex function as it exists in various forms, including monomer, dimer, and tetramer.⁴ In tumors and the retina, pyruvate is converted to lactate abundantly despite the presence of oxygen, a process called “aerobic glycolysis” or the “Warburg effect,” which is thought to be closely related to PKM2.⁵ The main function of tetrameric PKM2 is to convert phosphoenolpyruvate to pyruvate, whereas monomeric PKM2 promotes cellular glycolysis and biosynthesis, favoring cell proliferation.^6,7 It is potentially involved in multiple biological processes, including hypoxia response, glucose metabolism, neovascularization, proliferation, and inflammation through hypoxia-inducible factor-1α (HIF-1α),⁸ which is important in promoting neovascularization in ROP.⁹ Previous studies have shown that monomeric PKM2 acts as a transcription factor, stabilizing factor, and co-activator of HIF-1α.^10,11

In this study, we investigated the role of PKM2 in neovascularization using human retinal microvascular endothelial cells (HRMECs) and a mouse model of oxygen-induced retinopathy (OIR). DASA-58, a small-molecule activator that stabilizes PKM2 tetramers, provides a valuable tool to explore the structural-function relationship of PKM2 in this context.¹² This small-molecule increases PKM2 activity by promoting the tetrameric state, and has been used in numerous studies on tumor proliferation and immunity.^13–15 We further examined the potential role of pyruvate dehydrogenase kinase 1 (PDHK1) as an upstream regulator of PKM2 phosphorylation.

Methods

Reagents

DASA-58 (Selleck) was used in both in vitro and in vivo experiments. The compound was dissolved and stored according to the manufacturer's instructions. For cell culture, HRMECs were treated with 50 µM DASA-58 prior to and during exposure to hypoxia (1% O₂, 5% CO₂, and 94% N₂) for 24 hours. For in vivo studies, 0.5 µL of 500 µM DASA-58 was intravitreally injected into the right eye of OIR mice at postnatal day (P)12.¹⁶

Cell Culture

The HRMECs (Cell Biologics) were incubated in endothelial cell medium (ScienCell) containing 10% fetal bovine serum (FBS; Gibco), 100 U/mL penicillin (Gibco), and 100 µg/mL streptomycin (Gibco). The cell cultures were maintained at 37°C in a humidified atmosphere comprising 95% air and 5% CO₂, and passaged by trypsinization (0.05% trypsin-ethylenediaminetetraacetic acid; Gibco) every 2 or 3 days. Cells under passage 7 were used for the study.

Whole Protein Extraction

Whole protein was extracted from HRMECs and retinal tissues using RIPA lysis buffer (Beyotime). HRMECs were washed with cold PBS and lysed on ice for 30 minutes. Retinal tissues were homogenized, sonicated for 10 seconds, and lysed on ice for 30 minutes. Lysates were centrifuged at 12,000 × g for 15 minutes at 4°C, and supernatants were collected. Protein concentrations were determined by BCA assay (Beyotime). Samples were mixed with loading buffer (4:1) and boiled at 100°C for 10 minutes before analysis.

Nuclear and Cytoplasmic Protein Extraction

Nuclear and cytoplasmic proteins were isolated from HRMECs and retinal tissues using a commercial extraction kit (Beyotime) following the manufacturer's instructions. Protein concentrations were determined by BCA assay (Beyotime), and samples were mixed with loading buffer (4:1) and boiled at 100°C for 10 minutes before analysis.

Western Blotting

Samples were electrophoretically resolved on 12% to 20% gradient tris-glycine sodium dodecyl sulfate–polyacrylamide gels and transferred to polyvinylidene difluoride membranes using a Trans-Blot Turbo Transfer System (Bio-Rad). The membranes were then blocked with 5% skim milk in tris-buffered saline containing Tween for 1 hour and incubated with the following antibodies: anti-proliferating cell nuclear antigen (PCNA; 1:10000, Proteintech), anti-HIF-1α (1:1000, Cell Signaling Technology), anti-PKM2 (1:1000, Cell Signaling Technology), anti-phosphorylated-PKM2 (p-PKM2; 1:1000, Cell Signaling Technology), anti-pyruvate kinase M1 (PKM1; 1:1000, Cell Signaling Technology), anti-VEGFA (1:1000, Abcam), anti-pyruvate dehydrogenase kinase-1 (PDHK1; 1:1000, Cell Signaling Technology), anti-glucose transporter-1 (GLUT1; 1:2000, Proteintech), anti-lactate dehydrogenase-A (LDHA; 1:2000, Proteintech), anti-tubulin (1:5000, Proteintech), anti-histone H3 (1:5000, Proteintech), and anti-β-actin (1:10,000, Cell Signaling Technology). After incubation with HRP-conjugated secondary antibodies (Yesen), signals were detected using chemiluminescence (Thermo Fisher). β-Actin and Tubulin were used as loading controls for cytoplasmic proteins, and Histone H3 for nuclear proteins. Densitometric analysis of the bands was performed using ImageJ software.

Immunofluorescence Staining

HRMECs cultured on glass-bottom chamber slide were fixed with 4% paraformaldehyde for 15 minutes. The cells were washed with PBS, permeabilized in 0.5% Triton X-100 in PBS for 10 minutes, and blocked in PBS containing 5% BSA for 1 hour. The cells were then incubated with the following antibodies: anti-HIF-1α (1:100, Cell Signaling Technology), anti-PKM2 (1:200, Cell Signaling Technology), anti-PKM1 (1:200, Cell Signaling Technology), anti-GLUT1 (1:200, Proteintech), anti-LDHA (1:200, Proteintech), anti-PCNA (1:500, Proteintech), anti-PDHK (1:200, Cell Signaling Technology), and p-PKM2 (Tyr105) antibody (1:200, Cell Signaling Technology) overnight at 4°C and subsequently washed with PBS, followed by Alexa Fluor-conjugated secondary antibody (1:1000, Thermo Fisher Scientific) and Alexa Fluor 488-phalloidin (1:1000, Thermo Fisher Scientific) for 1 hour at room temperature in the dark. Nuclei were counterstained with DAPI. The cells were imaged using a confocal microscope (Leica Microsystems).

Metabolic Assays

HRMECs were passaged and cultured for 24 hours before treatments. Cells were then maintained under normoxia, hypoxia (1% O₂), or hypoxia with DASA-58 (50 µM, 1 hour pretreatment) for an additional 24 hours. Lactate concentration in the culture medium was measured using a lactate assay kit (Jiancheng Bioengineering Institute). Cell viability was assessed using the Cell Counting Kit-8 (CCK-8, Dojindo, Japan), which measures the metabolic activity of viable cells through the reduction of WST-8 by cellular dehydrogenases. Intracellular ATP content was determined from cell lysates using an enhanced ATP detection kit (Beyotime) according to the manufacturers’ protocols.

Endothelial Cell Functional Assays

HRMECs were passaged and cultured for 24 hours before treatments. Cells were then maintained under normoxia, hypoxia (1% O₂), or hypoxia with DASA-58 (50 µM, 1 hour pretreatment) for an additional 24 hours before following functional assays.

Wound healing assay: Confluent monolayers were scratched with a 200 µL pipette tip, and wound closure was photographed after 10 hours. Scratch areas were quantified with ImageJ software (National Institutes of Health).

Transwell migration assay: 1 × 10⁴ HRMECs were seeded in the upper chamber of 8 µm-pore Transwell inserts (Corning) with serum-free medium, whereas medium containing 10% FBS was added to the lower chamber. After 8 hours, migrated cells were fixed, crystal violet-stained, and counted in 5 random fields per insert.

Tube formation assay: HRMECs (2 × 10⁵/mL) were seeded on Matrigel-coated (Corning) coverslips in 24-well plates and incubated for 10 hours. Tube length, meshes, and branch points were quantified using ImageJ software.

Endothelial Cell Proliferation Assay

HRMECs were passaged and cultured for 24 hours before treatments. Cells were then maintained under normoxia, hypoxia (1% O₂), or hypoxia with DASA-58 (50 µM, 1 hour pretreatment) for an additional 24 hours. Cell proliferation was assessed using a 5-ethynyl-2′-deoxyuridine (EdU) Cell Proliferation Assay Kit (Beyotime) following the manufacturer's instructions. The proliferation index was calculated as the ratio of EdU⁺ to DAPI⁺ nuclei in five random fields.

Crosslinking Proteins

HRMECs were incubated in normoxic, hypoxic (1% O₂), or hypoxic with DASA-58 (50 µM, 1 hour pretreatment) conditions for 24 hours. The cells were washed with cold PBS (pH 8.0) to remove amine-containing culture media and proteins from the cells and crosslinked with 5 mM disuccinimidyl suberate (Thermo Fisher) for 30 minutes at room temperature, followed by quenching with 20 mM Tris·HCl (Beyotime) for 15 minutes. The cell lysates were subjected to Western blotting. The retinal samples were first digested to form a cell suspension and collected by centrifugation before crosslinking.

Coimmunoprecipitation

Coimmunoprecipitation (CoIP) was performed using a CoIP commercial kit (Thermo Fisher Scientific). PKM2 antibody was coupled to resin. HRMECs exposed to hypoxic conditions were lysed and incubated with the immobilized PKM2 antibody overnight at 4°C. After washing and elution, interacting proteins were collected for analysis.

OIR Model

All animal experiments were performed in accordance with the ARRIVE guidelines and were approved by the ethics committee of the eye and ENT Hospital of Fudan University. Pregnant C57BL6/J mice were purchased from JIESIJIE laboratory animal company. On P7, male pups were randomly assigned to 2 groups: controls maintained under normoxia and OIR mice exposed to 75% ± 3% oxygen for 5 days. At P12, OIR mice received an intravitreal injection of 0.5 µL of 500 µM DASA-58, delivering a total dose of 0.25 nmol per eye. This was calculated to achieve an estimated theoretical initial intraocular concentration of approximately 50 µM, assuming instantaneous and uniform distribution within the approximately 5 µL vitreous body.

Retinal Flat Mount

The eyes of P17 mice were fixed using 4% paraformaldehyde for 30 minutes, after which the retina was carefully dissected from the eyecup. The stripped retinas were treated with 0.1% Triton X-100 and 5% bovine serum albumin in PBS for 1 hour at room temperature, and incubated with Griffonia simplicifolia Lectin I isolectin B4 (IB4, Vector Laboratories) overnight at 4°C. After being washed five times, the retinas were cut into four petals and mounted on glass slides for imaging with a confocal microscope (SP8, Leica Microsystems). Supplementary Figure S2 demonstrates the process used to quantify the percentages of the neovascular and avascular areas.

Electroretinogram

Retinal function was evaluated by electroretinography (ERG) in three groups of mice: normoxia control, OIR, and OIR treated with DASA-58. Recordings were performed using a scotopic ERG system (Espion Electrophysiology System, Diagnosys LLC). The mice were dark-adapted for 24 hours prior to the test and subsequently anesthetized. Pupils were dilated with 0.5% tropicamide combined with 0.5% phenylephrine (Santen Pharmaceutical Co.), followed by topical application of oxybuprocaine (Santen Pharmaceutical Co.) for corneal anesthesia and carbomer gel (Bausch & Lomb) for corneal hydration. Under dim red light, platinum ring electrodes were placed on the corneal surface, with a reference electrode inserted subcutaneously in the frontal region and a ground electrode placed subdermally at the tail. After 10 minutes of dark adaptation under anesthesia, ERG responses were recorded. Dark-adapted ERG single-flash stimuli were delivered at 0.1, 1, and 10 cd·s/m² at 1 hertz (Hz), with each intensity repeated five times. The a-wave amplitude was defined as the voltage from baseline to the first negative trough, and the b-wave amplitude as the voltage from baseline to the subsequent positive peak.

Frozen Sectioning and Hematoxylin and Eosin Staining

Eyeballs were enucleated and fixed in 4% PFA at room temperature for 2 hours. After dehydration in 20% and 30% sucrose at 4°C (12 hours each), tissues were embedded in optimum cutting temperature (OCT) and frozen at −80°C. Sagittal sections (10 µm) were cut using a cryostat. Sections were stained with hematoxylin for 5 minutes, rinsed, differentiated, and counterstained with eosin for 1 minute. After dehydration in graded ethanol and clearing in xylene, sections were mounted and observed under a light microscope.

Measurement of Retinal Vascular Leakage

Retinal vascular permeability was evaluated by Evans Blue assay in three groups of mice (normoxia control, OIR, and OIR + DASA-58). Mice were deeply anesthetized, and 0.5 mL of Evans Blue (1.0 mg/mL; Sigma-Aldrich) was administered by retro-orbital injection. After 1 minute of systemic circulation, the animals were euthanized and the eyes were immediately enucleated. The globes were fixed in 4% paraformaldehyde for 1 hour, and retinas were carefully dissected and flat-mounted on glass slides with coverslips. Retinal vasculature and dye leakage were examined with a confocal microscope (SP8, Leica Microsystems).

Statistical Analyses

Data are presented as mean ± SD. Group differences were assessed using 1-way ANOVA with least significant difference post hoc test. Any P < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 20 (International Business Machines Corporation).

Results

Activation of the Canonical Function of PKM2 Modulates Hypoxia-Induced Angiogenic Signaling via HIF-1α

Transcriptomic analysis revealed that hypoxia promoted HRMEC proliferation and activated the HIF-1 signaling pathway, neovascularization-associated genes, and glycolysis-related pathways. These effects were significantly suppressed by DASA-58 treatment (Figs. 1A, 1B), suggesting that PKM2 may exert non-canonical (non-pyruvate kinase) functions under hypoxic conditions. Western blotting and immunofluorescence confirmed this treatment suppressed hypoxia-induced expression of HIF-1α and VEGFA (see Figs. 1B, 1C), whereas PKM1 expression remained unaffected by DASA-58 treatment.

Figure 1.

Activation of the canonical function of PKM2 modulates hypoxia-induced angiogenic signaling via HIF-1α. (A) Transcriptomic analysis of HRMECs showing hypoxia-induced activation of HIF-1α, angiogenic, and glycolytic pathways, which was reversed by DASA-58. (B, C) Hypoxia increased HIF-1α, PKM2, and VEGFA, whereas DASA-58 suppressed these changes; PKM1 was unaffected. (D, E) Western blot and immunofluorescence showing that GLUT1 and LDHA expression were upregulated by hypoxia and reduced by DASA-58. (F, G) Hypoxia increased extracellular and intracellular lactate levels, which was reversed by DASA-58. (H) Hypoxia decreased ATP production, which was restored by DASA-58. (I) Hypoxia decreased HRMEC viability, whereas DASA-58 treatment improved cell survival.

Given that glucose transporter 1 (GLUT1) and lactate dehydrogenase A (LDHA) are key glycolytic enzymes downstream of HIF-1α, their expression levels were further examined. Western blot and immunofluorescence showed that hypoxia significantly upregulated GLUT1 and LDHA in HRMECs compared with controls, whereas DASA-58 treatment reduced their expression toward baseline levels (Figs. 1D, 1E).

Consistent with these molecular changes, hypoxia markedly increased both extracellular and intracellular lactate levels in HRMECs, whereas DASA-58 treatment reduced lactate accumulation toward control levels (Figs. 1F, 1G). Moreover, hypoxia decreased intracellular ATP content and cell viability, whereas DASA-58 significantly restored both (Figs. 1H, 1I), which suggested that PKM2 activation could improve cellular bioenergetics, possibly by enhancing aerobic respiration.

Together, these findings demonstrated that PKM2 activation could attenuate hypoxia-induced HIF-1α activation, suppress downstream glycolytic reprogramming and restore energy metabolism, thereby limiting angiogenic signaling in HRMECs.

PKM2 Activation Attenuates Hypoxia-Induced Angiogenic Behaviors

To further investigate the functional relevance of PKM2 activation under hypoxia, we assessed its role in HRMECs using in vitro angiogenesis-related assays. Hypoxic conditions significantly enhanced HRMECs’ wound closure (Fig. 2A), migratory ability (Fig. 2B), and tube-formation capacity (Fig. 2C). In addition, EdU-based proliferation assays and measurements of proliferating cell nuclear antigen (PCNA) expression were conducted (Fig. 2D–F). The results showed that hypoxia significantly increased HRMECs’ capability of proliferation. Notably, DASA-58 pretreatment significantly suppressed all hypoxia-induced angiogenic behaviors.

Figure 2.

PKM2 activation attenuated hypoxia-induced angiogenic behaviors. (A–C) Hypoxia increased HRMECs’ wound healing (A), migration (B), and tube formation (C). These effects of hypoxia were significantly reduced by PKM2 activation (DASA-58 pretreatment). (D–F) EdU staining and PCNA expression indicate that hypoxia increased HRMECs proliferation. These effects were significantly reduced by PKM2 activation. Data are presented as the mean ± SEM and P values were obtained using 1-way ANOVA. **P < 0.01; ***P < 0.001; ****P < 0.0001. Quantifications were based on six independent experiments.

PKM2 Phosphorylation Promotes Monomer Formation, Nuclear Translocation, and Interaction With HIF-1α, a Process Regulated by PDHK1

To investigate the molecular mechanism by which PKM2 contributes to hypoxia-induced retinal neovascularization, we examined its subcellular localization and phosphorylation status. Protein crosslinking assays revealed that hypoxia induced PKM2 monomerization, whereas treatment with DASA-58 restored its tetrameric conformation (Fig. 3A). Notably, phosphorylated PKM2 (p-PKM2) was detected only in monomeric form (Fig. 3B). Subcellular fractionation showed that hypoxia promoted nuclear translocation of p-PKM2 (Fig. 3C). Crosslinking combined with nuclear-cytoplasmic fractionation revealed that nuclear PKM2 predominantly existed in monomeric form, whereas cytoplasmic PKM2 was present as both monomers and tetramers (Fig. 3D). Similarly, p-PKM2 was detected exclusively in monomeric form in both cytoplasm and nucleus (Fig. 3E). Co-immunoprecipitation confirmed that PKM2, but not PKM1, interacted with HIF-1α under hypoxia (Fig. 3F).

Figure 3.

PKM2 phosphorylation promoted monomer formation, nuclear translocation, and interaction with HIF-1α, a process regulated by PDHK1. (A) Crosslinking assays showing hypoxia-induced PKM2 monomerization, which was reversed by DASA-58. (B) Phosphorylated PKM2 (p-PKM2) was detected only in monomeric form. (C) Cytoplasmic–nuclear fractionation showing nuclear accumulation of p-PKM2 under hypoxia, which was attenuated by DASA-58. (D, E) Crosslinking combined with subcellular fractionation confirming nuclear PKM2 exists predominantly as monomers, while cytoplasmic PKM2 includes both monomers and tetramers; p-PKM2 was exclusively monomeric in both compartments. (F) Co-immunoprecipitation showing that PKM2, but not PKM1, interacted with HIF-1α under hypoxia. (G) Western blot and immunofluorescence showing that PDHK1 expression paralleled p-PKM2 changes under hypoxia and was suppressed by DASA-58. (H, I) PDHK1 inhibition by TM-1 decreased PKM2 phosphorylation in a dose-dependent manner, confirmed by Western blot and immunofluorescence. (J) Co-immunoprecipitation verifying direct interaction between PDHK1 and PKM2.

To identify upstream regulators of PKM2 phosphorylation, we examined the role of pyruvate dehydrogenase kinase 1 (PDHK1). PDHK1 expression followed a similar pattern to p-PKM2 (Fig. 3G), and its inhibition through TM-1 suppressed PKM2 phosphorylation in a dose-dependent manner (Fig. 3H). Co-immunoprecipitation further verified a direct interaction between PDHK1 and PKM2 (Fig. 3I). These results identified PDHK1 as an upstream modulator of PKM2 phosphorylation that promotes PKM2 monomerization and nuclear translocation under hypoxic conditions.

PKM2 Tetramerization and Phosphorylation Dynamics in the OIR Model

To validate our in vitro findings, we examined PKM2 conformational changes and phosphorylation status in the OIR mouse model. Western blot analysis showed that the retinas of OIR mice had elevated expression of HIF-1α and VEGFA compared with controls, whereas DASA-58 treatment suppressed these changes (Fig. 4A). Consistent with the in vitro results, crosslinking assays revealed that OIR markedly increased PKM2 monomer formation, whereas DASA-58 restored tetramerization (Fig. 4B). The p-PKM2 was detected only in monomeric form, and its abundance was elevated in OIR but reduced after DASA-58 treatment (Fig. 4C). Subcellular fractionation confirmed that nuclear PKM2 in OIR retinas was predominantly monomeric, with increased nuclear p-PKM2 under hypoxia. Importantly, DASA-58 treatment reduced nuclear PKM2 and p-PKM2 levels while promoting cytoplasmic tetramerization (Fig. 4D). These findings were consistent with our in vitro observations, demonstrating that in the OIR model, hypoxia drives PKM2 phosphorylation, monomerization, and nuclear translocation to activate HIF-1α signaling, whereas DASA-58-induced PKM2 tetramer stabilization counteracts these effects.

Figure 4.

PKM2 tetramerization and phosphorylation dynamics in the OIR model. (A) Western blot showing increased expression of HIF-1α, PKM2, and VEGFA in OIR retinas, which was reduced by DASA-58; PKM1 was unchanged. (B) Crosslinking assays demonstrating that OIR induced PKM2 monomerization, whereas DASA-58 restored tetramer formation. (C) The p-PKM2 was detected only in monomeric form and increased in OIR retinas, whereas DASA-58 reduced its abundance. (D) Cytoplasmic–nuclear fractionation showing nuclear accumulation of PKM2 and p-PKM2 in OIR retinas, which was attenuated by DASA-58, whereas cytoplasmic PKM2 tetramerization was promoted. Data are presented as the mean ± SEM and P values were obtained using 1-way ANOVA. **P < 0.01; ***P < 0.001; ****p < 0.0001. Quantifications were based on six independent experiments.

PKM2 Tetramer Stabilization Suppresses Pathological Neovascularization and Improves Retinal Function in the OIR Model

We next evaluated whether the molecular changes induced by DASA-58 translated to improved retinal structure and function in vivo. Intravitreal injection of DASA-58 had no adverse effect on normal retinal vascular development, as evidenced by unchanged vascular morphology and density in neonatal mice under normoxic conditions (Figs. 5A, 5B), which supported its safety profile. In the OIR model, DASA-58 treatment led to a marked reduction in pathological neovascularization, as demonstrated by a significant decrease in neovascular tuft area and avascular zones (Fig. 5C). Additionally, the number of vascular filopodia, key indicators of angiogenic activity, was significantly reduced in the retinas of DASA-58–treated mice (Fig. 5D).

Figure 5.

PKM2 tetramer stabilization suppressed pathological neovascularization and improved retinal function in the OIR model. (A, B) Intravitreal injection of DASA-58 did not alter retinal vascular morphology or density under normoxia. (C) DASA-58 reduced neovascular tuft formation and avascular areas. (D) Quantification showing decreased vascular filopodia in DASA-58-treated OIR retinas. (E) Electroretinography (ERG) demonstrated impaired a- and b-wave amplitudes in OIR mice, which were partially restored by DASA-58. (F) Evans Blue assays showing increased vascular leakage in OIR mice, attenuated by DASA-58. (G) Hematoxylin and eosin (H&E) staining revealed increased retinal thickness and preretinal neovascularization in OIR mice, both reduced by DASA-58 treatment. Data are presented as the mean ± SEM and P values were obtained using 1-way ANOVA. **P < 0.01; ***P < 0.001; ****P < 0.0001. Quantifications were based on ≥10 independent experiments.

We next evaluated the effect of PKM2 modulation on retinal function in the OIR model using ERG. Compared with control mice, the amplitudes of both a-wave and b-wave were markedly reduced in OIR mice, indicating impaired retinal function. Treatment with DASA-58 partially restored these responses (Fig. 5E), which suggested that inhibition of PKM2 alleviated visual dysfunction in OIR. Evans Blue assays showed that vascular leakage was significantly increased in OIR mice compared with control mice, indicating impaired vascular function. Treatment with DASA-58 markedly reduced the leakage in OIR mice (Fig. 5F). Histological analyses revealed prominent neovascularization on the retinal surface in the OIR group. Meanwhile, the average retinal thickness of OIR mice was increased, which may be the combined effect of factors, including neovascular membranes, vascular leakage, and hypoxia. Following DASA-58 treatment, neovascularization on the retinal surface was reduced, and OIR-induced retinal thickening was also attenuated, which indicated a protective effect of DASA-58 on retinal morphology (Fig. 5G).

These in vivo findings extended our cellular observations by demonstrating that PKM2 tetramer stabilization could not only modulate hypoxia-induced molecular and metabolic changes but also confer structural and functional protection to the retina.

Discussion

Current therapies for ROP, such as laser ablation and anti-VEGF agents, effectively suppress pathological neovascularization but carry risks of disrupting physiological retinal development^17–21 and causing systemic side effects,^21–24 which underscores the need for alternative strategies targeting upstream regulators. Here, we identified the conformational dynamics of PKM2 as a critical upstream mechanism driving hypoxia-induced angiogenesis in ROP.

In this study, we proposed and preliminarily validated a previously unknown mechanism of neovascularization in ROP. Transcriptomic and protein analyses showed that hypoxia increased HIF-1α, VEGFA, and glycolysis-related effectors including GLUT1 and LDHA. These changes were accompanied by elevated lactate accumulation and enhanced cell proliferation. Importantly, pharmacological stabilization of PKM2 tetramers by DASA-58 reversed these alterations, suppressed glycolysis, and improved cellular energy status. These findings suggest that conformational change of PKM2 contributes to angiogenesis not only through its metabolic role but also by amplifying HIF-1α signaling under hypoxia.

Functionally, hypoxia markedly enhanced endothelial migration, tube formation, and proliferation, which validated the link between metabolic reprogramming and angiogenic behaviors. DASA-58 pretreatment significantly suppressed all these responses, supporting the concept that PKM2 monomerization drives angiogenic phenotypes in endothelial cells. This dual effect on both metabolism and angiogenesis highlights PKM2 as a regulatory hub that integrates metabolic and signaling cues to promote pathological vascular growth.

The mechanistic experiments revealed that hypoxia induced phosphorylation-dependent PKM2 monomerization, nuclear accumulation, and interaction with HIF-1α. Phosphorylated PKM2 was detected exclusively in monomeric form, indicating that phosphorylation is tightly coupled to its conformational state. Nuclear-cytoplasmic fractionation confirmed that monomeric PKM2 accumulated in the nucleus under hypoxia, whereas DASA-58 reduced nuclear PKM2 and promoted cytoplasmic tetramerization. Moreover, we identified PDHK1 as an upstream regulator via following facts: (1) PDHK1 directly interacted with PKM2, (2) its expression paralleled that of p-PKM2, and (3) PDHK1 inhibition suppressed PKM2 phosphorylation and monomerization.

HIF-1α regulates multiple proangiogenic factors (especially VEGF) which contributes to angiogenesis²⁵ as an important signaling factor in the development of ROP.⁹ HIF-1 consists of two subunits, HIF-1β, which is continuously and stably expressed in cells, and HIF-1α, which is regulated by the partial oxygen pressure. HIF-1α is disassembled by ubiquitination upon binding to von Hippel-Lindau tumor-suppressor protein (VHL) in normoxic conditions. Under hypoxic conditions, HIF-1α fails to bind to VHL, resulting in elevated HIF-1α activity and binding to hypoxia response elements, thereby increasing the activation of many proangiogenic genes.^25,26 Retinal hypoxia in ROP leads to abnormal neovascularization by strongly upregulating the expression of these proangiogenic genes via HIF-1α.²⁷ Based on our results, when PKM2 tetramerization is promoted by DASA-58, PKM2 monomerization is reduced, which directly decreases the expression of HIF-1α. Furthermore, the activation of the expression downstream proangiogenic genes by PKM2 as an HIF-1α coactivator is reduced by DASA-58.

Our study unveils a novel regulatory axis wherein PDHK1, known as a downstream effector of HIF-1α that redirects glycolytic flux by inhibiting pyruvate dehydrogenase complex,²⁸ also acts upstream to directly control PKM2 conformational dynamics. We demonstrate that under hypoxia, PDHK1 interacts with and phosphorylates PKM2, promoting its monomerization and nuclear translocation. This is supported by our findings that PDHK1 inhibition reduced PKM2 phosphorylation and that the two proteins physically interacted.

Most importantly, we propose that this interaction initiates a positive feedback loop that amplifies hypoxic signaling: hypoxia-induced HIF-1α upregulates PDHK1 expression, which phosphorylates PKM2 to facilitate its nuclear translocation. Nuclear PKM2 then serves as a coactivator for HIF-1α, further enhancing the transcription of HIF-1α target genes including PDHK1 itself. This self-reinforcing cycle (HIF-1α → PDHK1 → p-PKM2 → nuclear PKM2 → enhanced HIF-1α activity) creates a powerful amplifier for pathological angiogenesis.

The observed decrease in total PKM2 expression following DASA-58 treatment further supports the existence of this regulatory circuit. As HIF-1α is an established transcriptional activator of PKM2,^10,29 disruption of the feedback loop through PKM2 tetramerization would naturally reduce PKM2 transcription, ultimately diminishing the pool of PKM2 available to drive angiogenic signaling.

An in vivo OIR model corroborated our in vitro findings. The retinas of OIR mice showed increased PKM2 monomers, elevated nuclear p-PKM2, and enhanced HIF-1α/VEGFA expression, all of which were reversed by DASA-58. Functionally, DASA-58 reduces pathological changes (neovascular tufts, avascular areas, and vascular leakage), restores ERG responses, and alleviates OIR-induced retinal edema and thickening. These results confirm that PKM2 conformational regulation not only drives molecular and cellular angiogenic responses but also drives pathological neovascularization and retinal dysfunction in vivo.

The goal of treating ROP is to reduce VEGF production by the immature retina and suppress neovascularization.³⁰ Because VEGF is also essential for the development of normal blood vessels,³¹ we evaluated the toxicity of DASA-58 on the retinal vascular network. We found no obvious signs of retinal toxicity following intravitreal injection of DASA-58 at P12. However, this experiment was a short-term observation and further data are needed to support its safety profile.

Although our study provides mechanistic and functional evidence supporting PKM2 tetramer formation as a therapeutic strategy, further work is required to evaluate the long-term safety and efficacy of DASA-58 in vivo, particularly given the essential role of VEGF in normal vascular development. Nonetheless, targeting PKM2 conformational dynamics may represent a promising approach to modulate hypoxia-driven angiogenesis in ROP and potentially other ischemic retinopathies.

Conclusions

Our study identifies PKM2 conformational dynamics as a key regulator of hypoxia-induced retinal neovascularization. Under hypoxia, HIF-1α-driven PDHK1 activity phosphorylated PKM2, promoting its monomerization, nuclear translocation, and interaction with HIF-1α, thereby establishing a positive feedback loop that amplifies glycolytic reprogramming and angiogenic signaling. Pharmacological stabilization of PKM2 tetramers with DASA-58 disrupted this loop, suppressed glycolysis and endothelial proliferation in vitro, and preserved retinal vascular morphology and function in vivo. These findings highlight PKM2 as a novel therapeutic target in ROP and suggest that PKM2 tetramer stabilization may provide a promising strategy to modulate pathological angiogenesis in retinopathies.