Endothelial SP1 lactylation promotes bronchopulmonary dysplasia via regulation of Cdkn1a expression Authors: Ziming Zhang, Qiqi Li, Xinmin Ju, Lingchen Li, Tian Xie, Lu Zu, Qianyun Cao, Mengting Hu, Yun Ling, Lingling Yan, Tieshuai Liu, Jiajing Ge, Xiaolu Ma, Zheng Chen, Jiajun Zhu, Jia He, Wei Shi, Lizhong Du Published: 2026 ABSTRACT AbstractBronchopulmonary dysplasia (BPD) is a major complication of prematurity, and endothelial dysfunction contributes to its pathogenesis. Although lactate metabolism is dysregulated in BPD, whether lactate-derived lysine lactylation contributes to endothelial dysfunction remains unknown. This study investigated the role of lactylation in endothelial injury and the involvement of SP1-Cdkn1a axis in BPD. We found that lactate levels and global lactylation levels were increased in both in vitro and in vivo models of BPD, accompanied by impaired endothelial proliferation, migration and angiogenic capacity. Integrated bioinformatic analysis identified Cdkn1a as a glycolysis-associated hub gene in BPD, and its upregulation in pulmonary vascular endothelial cells (PVECs) was confirmed experimentally. Subsequent Cdkn1a knockdown alleviated hyperoxia-induced endothelial dysfunction. In addition, inhibition of glycolysis reduced lactylation level in PVECs, suppressed Cdkn1a expression, and partially improved pulmonary microvascular injury. To explore the upstream mechanism, lactylome profiling identified SP1, a known transcription factor of Cdkn1a, as a lactylated candidate. Increased SP1 lactylation and enhanced binding at the Cdkn1a promoter were verified in PVECs from BPD mice. K640R mutation attenuated SP1-dependent Cdkn1a promoter activation. Overall, this study reveals the involvement of endothelial lactylation in pathogenesis of BPD and implicates the SP1-Cdkn1a axis as a potential therapeutic target.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56295-x.Keywords: Bronchopulmonary dysplasia, Lactylation, Endothelial dysfunction, Hyperoxia, Cdkn1a, SP1Subject terms: Neonatology, Preterm birth Background Bronchopulmonary dysplasia (BPD) is the most common complication of preterm infants and is associated with lifelong consequences1. The mortality rates of infants with moderate to severe BPD reach 40–50%2–4. Current therapeutic options remain limited5. Pathologically, BPD is characterized by disrupted alveolar development, alveolar simplification, and impaired pulmonary microvascular growth6,7. Endothelial dysfunction plays a central role in its pathogenesis8,9. Impaired angiogenesis disrupts alveolar development10,11, whereas maintaining endothelial survival improves lung structure12,13. Although numerous pharmacological and non-pharmacological approaches have been suggested, current therapeutic options remain limited14. Therefore, investigating the drivers of endothelial injury is crucial for developing effective early prevention strategies in BPD. Recent evidence suggests that lactate metabolism is dysregulated in BPD15. Elevated lactate levels have been detected in urine or cord blood from infants who later developed BPD16,17. Animal studies also demonstrated enhanced lactate accumulation in lung tissue of BPD mice18. Moreover, lactate is now increasingly recognized as the biochemical source of lysine lactylation (Kla), a recently identified post-translational modification19. Emerging evidence indicates that lactylation could regulate cell behavior under pathological conditions, including proliferation, migration, and angiogenesis. For example, YY1 lactylation promotes microglial proliferation and migration in autoimmune uveitis20. c-Myc lactylation enhances tubular epithelial cell proliferation in renal ischemia-reperfusion injury21. IDH2 K272 lactylation facilitates the proliferative, migratory, and angiogenic capacity of cardiac microvascular endothelial cells in diabetic myocardial infarction22. However, whether lactylation contributes to endothelial dysfunction in BPD remains unknown. Among downstream effectors potentially involved in endothelial dysfunction, cyclin-dependent kinase inhibitor 1 A (Cdkn1a) is a plausible candidate23. In addition to its canonical role in cell-cycle regulation, Cdkn1a has been implicated in endothelial proliferation, migration, and angiogenesis. For instance, suppression of Cdkn1a contributed to the pro-proliferative, pro-migratory, and pro-angiogenic effects of miR-93 in endothelial cells24. Endothelial Cdkn1a deletion in a developmental vascular regression model was associated with endothelial overproliferation25, while increased Cdkn1a expression in senescent endothelial progenitor cells was linked to impaired proliferative and angiogenic capacity26. Moreover, Cdkn1a is regulated by several transcription factors, including specificity protein 1 (SP1)27,28. Given that Cdkn1a has been upregulated in the lungs of BPD mice29,30, it may contribute to endothelial dysfunction in BPD. However, whether SP1-Cdkn1a contributes to endothelial dysfunction in BPD, and whether this process is regulated by lactylation, remains unclear. In this study, we investigated whether lactylation promotes endothelial dysfunction in BPD through SP1-dependent regulation of Cdkn1a. Materials and methods Establishment of BPD mouse model and 2-DG administration All animal experiments were performed in accordance with relevant guidelines and regulations. All animal experiments were approved by the Institutional Animal Care and Use Committee of Zhejiang University (approval no. ZJU20230034) and complied with the ARRIVE guidelines (https://arriveguidelines.org). A hyperoxia-induced neonatal mouse model was used to study BPD pathogenesis31. Briefly, newborn C57BL/6 pups were randomly assigned to room air or 85% O2 from postnatal day 0 to 14. On postnatal day 14, right ventricular systolic pressure (RVSP) was measured under isoflurane anesthesia using an open-chest technique described in previous studies32. Pressure waveforms were recorded using a physiological data acquisition system (BIOPAC Systems, Inc., USA). In the 2-DG treatment group, mice received intraperitoneal injections of 2-deoxy-D-glucose (2-DG, 50 mg/kg, Shanghai Yuanye Bio-Technology Co., Ltd, China) once daily from postnatal day 7 to 14, with the dose based on previous studies33. Isolation of PVECs and HUVECs Umbilical cords from preterm infants (< 30 weeks gestation) were collected from Guiyang Maternal and Child Health Care Hospital with approval from the Research Ethics Commission of Maternal and Child Health Hospital of Guiyang City (approval no.2023-11). All procedures involving human participants were performed in accordance with relevant guidelines and regulations and with the Declaration of Helsinki. Human umbilical vein endothelial cells (HUVECs) were isolated from umbilical cords using previously described methods34. Pulmonary vascular endothelial cells (PVECs) were isolated from mouse lungs according to established protocols35. EdU assay Cell proliferation of PVECs was assessed using an EdU assay kit (RiboBio, China) according to the manufacturer’s instructions. Briefly, cells were cultured under the indicated conditions and incubated with EdU solution for 2 h. Cells were then fixed, permeabilized, and subjected to the Click reaction, followed by Hoechst 33,342 staining. Images were captured using a fluorescence microscope (Axio Scope.A1, Zeiss, Germany), and the percentage of EdU-positive cells was quantified using Fiji/ImageJ software (ImageJ version 1.51d, NIH, USA). Tube formation assay Tube formation assays of HUVECs and PVECs were performed using Matrigel Basement Membrane Matrix (Corning, USA). Matrigel was added to 96-well plates and allowed to polymerize at 37 °C. Cells (1.5 × 10⁴ per well) were then seeded onto the Matrigel-coated plates for tube formation assays. After 6 h, tube-like structures were imaged using a phase-contrast microscope (Axio Scope.A1, Zeiss, Germany), and angiogenesis was quantified by total branching points and segment lengths using Fiji/ImageJ software (ImageJ version 1.51d, NIH, USA) with the Angiogenesis Analyzer plugin. Transwell assay Cell migration of HUVECs and PVECs was assessed using a Transwell assay performed in Transwell chambers (24-well plates, 8.0 μm pore size; Corning, USA). Briefly, a total of 1 × 10⁵ cells were seeded into the upper chambers in endothelial growth medium without FBS (ScienCell, USA), while endothelial growth medium supplemented with 5% FBS (ScienCell, USA) was added to the lower chambers. After 24 h, cells remaining on the upper surface of the membrane were gently removed using a cotton swab. Migrated cells on the lower surface were fixed, stained, and washed. Images were captured using an optical microscope (Axio Scope.A1, Zeiss, Germany), and migrated cells were quantified using Fiji/ImageJ software (ImageJ version 1.51d, NIH, USA). Wound healing assay To evaluate cell migration, HUVECs were plated in six-well plates and grown to confluence. A sterile 200µL pipette tip was used to create scratches in the monolayer. Cells were then gently washed with PBS and maintained in endothelial growth medium without FBS (ScienCell, USA). Wound closure was imaged at 0 h and 24 h using an optical microscope (Axio Scope.A1, Zeiss, Germany) and quantified using Fiji/ImageJ software (ImageJ version 1.51d, NIH, USA) by calculating the percentage of wound area closure. Public datasets Public gene expression datasets were retrieved from the NCBI Gene Expression Omnibus (GEO) database. Two bulk transcriptomic datasets, GSE29632 and GSE25286, were included in the present study36,37. Because GSE29632 includes both wild-type and Nrf2-knockout samples, only the 21 wild-type lung samples were retained for analysis. GSE25286 contains 10 lung samples from wild-type and hyperoxia-induced BPD mice, and all samples were included. In total, 31 samples were retained for subsequent analyses. Batch effects between bulk datasets were corrected using the ComBat function implemented in the “sva” R package, thereby improving inter-dataset comparability. For single-cell transcriptomic analysis, GSE266988 was used38. All mutant samples were excluded, and only the room air control sample (GSM8257789) and the hyperoxia control sample (GSM8257790) were retained for downstream analysis. WGCNA and differential expression analyses After batch correction, normalized bulk transcriptomic expression matrices were used to construct a weighted gene co-expression network using the WGCNA R package39. A soft-thresholding power was selected based on the scale-free topology criterion, and genes were clustered into different co-expression modules. Module–trait relationship analysis was then performed to identify the module most strongly associated with BPD. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was subsequently performed for genes within the module most strongly associated with BPD to identify enriched biological pathways. KEGG pathway annotations were obtained from the KEGG database40–42 and used with permission from Kanehisa Laboratories. In parallel, differential expression analysis between the BPD group and the control group was performed using the limma R package. Genes meeting the criteria of | logFC | > 1.5 and adjusted P value < 0.01 were defined as differentially expressed genes (DEGs). Candidate gene selection To identify candidate genes associated with BPD, DEGs were intersected with genes from the BPD-associated WGCNA module. The overlapping genes were then subjected to least absolute shrinkage and selection operator (LASSO) regression using the glmnet R package to identify core candidate genes. Genes retained at the optimal λ, as determined by minimum cross-validated error, were selected for subsequent analyses. Protein–protein interaction (PPI) analysis of candidate genes was performed using the STRING database and visualized in Cytoscape. Immunoprecipitation assay Immunoprecipitation was performed using an IP/Co-IP Kit with Protein A/G magnetic beads (MedChemExpress, USA) according to the manufacturer’s instructions. Cells were lysed in IP lysis buffer supplemented with protease inhibitors (MedChemExpress, USA), and the lysates were incubated with anti-SP1 antibody (Proteintech, China) or control IgG (Proteintech, China) together with Protein A/G magnetic beads. After extensive washing, bound proteins were eluted and subjected to western blot analysis using anti-SP1 (Proteintech, China; 1:3000) and pan-Kla antibodies (PTM Bio, China; 1:1000). Protein bands were visualized with BeyoECL Star chemiluminescent substrate (Beyotime, China) and imaged using a G: BOX imaging system (Syngene, UK). Band intensities were quantified using Fiji/ImageJ software (ImageJ version 1.51d, NIH, USA). CUT&Tag-qPCR CUT&Tag-qPCR was performed using a CUT&Tag assay kit (Vazyme, China) according to the manufacturer’s instructions. A total of 1.2 × 10⁵ PVECs were incubated with 1 µg of anti-SP1 antibody (Proteintech, China) or control IgG (Proteintech, China), followed by secondary antibody and pA/G-Tnp Pro. After tagmentation, DNA Spike-in provided in the kit was added as an internal normalization control for CUT&Tag-qPCR analysis, and CUT&Tag enrichment was quantified by qPCR using ChamQ Universal SYBR qPCR Master Mix (Vazyme, China) on a CFX Opus 96 Real-Time PCR System (Bio-Rad, USA). The sequences of primers used for Cdkn1a are listed in Supplementary Table S3. Luciferase reporter assay HEK-293T cells were co-transfected with a mouse Cdkn1a promoter luciferase reporter, the indicated mouse Sp1 expression constructs (wild-type or mutant), and a Renilla luciferase plasmid. All plasmids were provided by Tsingke Biotech Co., Ltd. (China). Luciferase activity was measured 48 h later using the Dual-Luciferase Reporter Assay System (Promega, USA) on a Synergy HTX Multimode Reader (BioTek, USA) and firefly luciferase activity was normalized to Renilla luciferase activity. Statistical analysis Statistical analyses were performed using R software (version 4.5.0) and GraphPad Prism 9 (GraphPad Software, USA). Comparisons between two groups were performed using Student’s t-tests, and comparisons among multiple groups were conducted using two-way ANOVA followed by Tukey’s post hoc test. A P value < 0.05 was considered statistically significant. Results Increased lactylation and endothelial dysfunction are observed in an in vitro BPD model To investigate whether hyperoxic stress impairs endothelial function in BPD, we established an in vitro model using human umbilical vein endothelial cells (HUVECs) isolated from preterm infants and exposed them to hyperoxia (Fig. 1A). The identity of HUVECs was confirmed by the characteristic cobblestone morphology (Fig. 1B). The results showed that hyperoxic exposure significantly reduced the proliferative capacity of HUVECs (Fig. 1C). In addition, migratory and angiogenic capacities were also markedly impaired under hyperoxic conditions (Fig. 1D-E). Given the emerging role of lactylation as a molecular interface between metabolic reprogramming and cellular dysfunction43, we next examined whether hyperoxia altered lactate metabolism and protein lactylation in HUVECs. As a result, intracellular lactate levels were significantly increased in HUVECs under hyperoxic conditions compared with normoxia (Fig. 1F). Consistently, protein lysine lactylation (Kla) levels were markedly increased in hyperoxia-exposed HUVECs (Fig. 1G and H). Taken together, these findings indicate increased lactate-dependent lactylation along with endothelial dysfunction in an in vitro BPD model. Increased lactylation and endothelial dysfunction are observed in an in vitro BPD model. (A) Schematic timeline for establishment of the hyperoxia-induced in vitro BPD model. (B) Representative image showing the typical cobblestone-like morphology of primary HUVECs; scale bar: 200 μm. (C) CCK-8 assay showing the proliferative capacity of HUVECs in control and hyperoxia groups. (D) Representative images and quantification of HUVECs migration in control and hyperoxia groups, scale bar: 50 μm. (E) Representative images and quantification of tube formation by HUVECs in control and hyperoxia groups, scale bar: 100 μm. (F) Intracellular lactate concentrations in HUVECs in control and hyperoxia groups. (G) Representative immunoblot and quantification of total Kla protein levels in HUVECs in control and hyperoxia groups. (H) Representative immunofluorescence images and quantification of Kla fluorescence intensity in HUVECs in control and hyperoxia groups, scale bar: 50 μm. HUVECs, human umbilical vein endothelial cells; CCK-8, Cell Counting Kit-8; Kla, lysine lactylation. Data are presented as mean ± SEM and were analyzed using a two-tailed unpaired Student’s t-test. *** p < 0.001, ** p < 0.01, * p < 0.05. Increased lactylation along with endothelial dysfunction is observed in BPD mouse model To validate the in vivo relevance of these findings, we next established a neonatal murine BPD model via hyperoxic exposure (Fig. 2A). Hemodynamic assessment showed that BPD mice exhibited significantly increased right ventricular systolic pressure and right ventricular hypertrophy compared with control mice (Fig. 2B-C). In addition, the lung-to-body weight ratio was significantly elevated in BPD mice (Fig. 2D). Histological examination further demonstrated impaired alveolar development in these mice, characterized by reduced radial alveolar count and increased mean linear intercept (Fig. 2E). Increased lactylation along with endothelial dysfunction is observed in BPD mouse model. (A) Experimental timeline for BPD mice model. (B) Representative images and quantification of RVSP of BPD and control mice. (C) RVHI of BPD and control mice. (D) Lung to body weight ratio of BPD and control mice. (E) Representative lung histology images and quantification of RAC and MLI in BPD and control mice, scale bar: 50 μm.(F) CCK-8 assay showing proliferation of PVECs isolated from BPD and control mice. (G) Representative images and quantification of EdU assay showing proliferation of PVECs isolated from BPD and control mice, scale bar: 50 μm. (H) Representative images and quantification of migration of PVECs isolated from BPD and control mice, scale bar: 50 μm. (I) Lactate concentration in PVECs isolated from BPD and control mice. (J) Representative ECAR trace and quantification in PVECs isolated from control and BPD mice. (K) Protein level of Kla (lactylation) in PVECs isolated from BPD and control mice. RVSP, right ventricular systolic pressure; RVHI, right ventricular hypertrophy index; RAC, radial alveolar count; MLI, mean linear intercept; PVECs, pulmonary vascular endothelial cells; ECAR, Extracellular Acidification Rate. Data are presented as mean ± SEM and analyzed by two-tailed unpaired Student’s t test. *** p < 0.001, ** p < 0.01, * p < 0.05. We then isolated pulmonary vascular endothelial cells (PVECs) from control and BPD mice to further explore the functional alterations. As a result, PVECs isolated from BPD mice exhibited significantly reduced proliferative capacity and impaired migratory ability as well (Fig. 2F-H). Intracellular lactate concentrations were significantly increased, and Seahorse analysis revealed enhanced glycolytic activity in PVECs derived from BPD mice (Fig. 2I-J). In parallel, protein lysine lactylation (Kla) levels were markedly increased in PVECs isolated from BPD mice (Fig. 2K). Together, these findings indicate that endothelial dysfunction in the BPD mouse model is accompanied by increased lactate accumulation and elevated lactylation. These observations prompted us to investigate the downstream transcriptional mediators that might link these metabolic alterations to endothelial dysfunction in BPD. Integrated transcriptomic analyses identify Cdkn1a as a glycolysis-related hub gene in BPD To identify co-expression modules associated with BPD and prioritize candidate genes, we first performed weighted gene co-expression network analysis (WGCNA)39 using bulk RNA-seq data from control and BPD mouse lungs. After sample clustering, genes were assigned to multiple co-expression modules (Fig. 3A-B). Among these modules, the turquoise module showed the strongest association with BPD status (Fig. 3C). Genes within this module also showed a strong positive correlation between module membership and gene significance (Fig. 3D). KEGG analysis further showed that genes in the turquoise module were significantly enriched in glycolysis/gluconeogenesis-related pathways (Fig. 3E). To identify key genes within the glycolysis-enriched turquoise module that were also significantly dysregulated in BPD, we first performed differential expression analysis between control and BPD mouse lungs to obtain differentially expressed genes (DEGs) (Fig. 3F). By taking the intersection of these DEGs with the genes in the turquoise module, we obtained 13 overlapping candidate genes (Fig. 3G). Among these candidates, Cdkn1a and Zmat3 were prioritized using LASSO regression (Fig. 3H–I). Protein-protein interaction analysis subsequently showed that Cdkn1a exhibited higher connectivity within the candidate network compared with Zmat3 (Fig. 3J). Combined with previous evidence linking Cdkn1a to cell dysfunction44, it was selected for further investigation. Consistently, single-cell RNA-seq analysis lung tissues revealed increased Cdkn1a expression specifically in endothelial cells from BPD mice (Fig. 3K–M). Collectively, these results identify Cdkn1a as a hub gene within a glycolysis-enriched transcriptional program in BPD and support its potential role in endothelial dysfunction. Integrated transcriptomic analyses identify Cdkn1a as a glycolysis-related hub gene in BPD. (A) Hierarchical clustering dendrogram of bulk RNA-seq samples from lung tissues of control and BPD mice. (B) Cluster dendrogram of genes used for WGCNA. (C) Module-trait relationships showing the correlations between WGCNA modules and BPD status. (D) Relationship between module membership and gene significance in the turquoise module. (E) KEGG enrichment analysis of genes in the turquoise module, showing significant enrichment of glycolysis/gluconeogenesis-related pathways. (F) Volcano plot showing DEGs between control and BPD mice lung from bulk RNA-seq data. (G) Overlap of DEGs and genes in the WGCNA turquoise module, identifying overlapping candidate genes. (H) LASSO regression analysis of the overlapping candidate genes for key gene selection. (I) Cross-validation plot for LASSO model selection. (J) Protein-protein interaction network of the top candidate genes from the overlapping gene set, identifying Cdkn1a as a key candidate gene associated with glycolysis-related transcriptional changes in BPD. (K) UMAP plot showing cell clusters identified from scRNA-seq data of lung tissues from control and BPD mice. (L) Feature plots showing Cdkn1a expression in endothelial cells from control and BPD mice. (M) Quantification of Cdkn1a expression in endothelial cells from control and BPD mice in the single-cell RNA-seq dataset. * p < 0.05. WGCNA, weighted gene co-expression network analysis; DEGs, differentially expressed genes; KEGG, Kyoto Encyclopedia of Genes and Genomes; LASSO, least absolute shrinkage and selection operator; UMAP, uniform manifold approximation and projection. Permission has been obtained from Kanehisa laboratories for using KEGG pathway database40–42. Cdkn1a is upregulated and contributes to endothelial dysfunction under hyperoxia Based on its identification as an upregulated candidate gene, we next examined Cdkn1a expression in the BPD mouse model. Cdkn1a expression levels were significantly upregulated in whole-lung tissues and PVECs isolated from BPD mice (Fig. 4A-C). To determine whether this upregulation was involved in endothelial injury, Cdkn1a was silenced in HUVECs under hyperoxic conditions, with knockdown efficiency confirmed (Fig. 4D). As a result, knocking down Cdkn1a significantly improved endothelial proliferation, migration, and angiogenic capacity under hyperoxic conditions, as evidenced by increased cell viability, enhanced migratory activity, and improved tube formation (Fig. 4E-H). Collectively, these findings indicate that Cdkn1a contributes to BPD-associated endothelial dysfunction. Cdkn1a is upregulated and contributes to endothelial dysfunction under hyperoxia. (A) Relative mRNA levels of Cdkn1a in lung tissues isolated from control and BPD mice. (B) Relative mRNA levels of Cdkn1a in PVECs isolated from control and BPD mice. (C) Protein level of Cdkn1a in PVECs isolated from control and BPD mice. (D) Knockdown efficiency of Cdkn1a in HUVECs. (E) CCK-8 assay showing proliferation of HUVECs following knockdown of Cdkn1a under hyperoxia. (F) Representative images and quantification of migrated HUVECs following knockdown of Cdkn1a under hyperoxia, scale bar: 50 μm. (G) Representative images and quantification of wound healing assay showing migration area of HUVECs after knockdown of Cdkn1a under hyperoxia, scale bar: 100 μm. (H) Representative images and quantification of tube formation assay showing total segment length and total branching length of HUVECs following knockdown of Cdkn1a under hyperoxia, scale bar: 100 μm. PVECs, pulmonary vascular endothelial cells; HUVECs, human umbilical vein endothelial cells. Data are presented as mean ± SEM and analyzed by two-tailed unpaired Student’s t test. *** p < 0.001, ** p < 0.01, * p < 0.05. Inhibition of glycolysis downregulates Cdkn1a expression and alleviates microvascular impairment in BPD Given that Cdkn1a was prioritized from a glycolysis-enriched transcriptional program and involved in endothelial dysfunction, we next explored whether inhibition of glycolysis could attenuate lactylation and thereby regulate Cdkn1a expression in vivo. To address this, BPD mice were treated with 2-deoxy-D-glucose (2-DG), a pharmacological inhibitor of glycolysis. Treatment with 2-DG did not significantly affect body weight compared with untreated BPD mice (Fig. 5A), but effectively reduced intracellular lactate levels in PVECs isolated from the 2-DG-treated BPD group (Fig. 5B). Consistent with the lactate decrease, total protein lysine lactylation (Kla) and Cdkn1a protein levels in PVECs were markedly reduced in the 2-DG-treated BPD group compared with untreated BPD group (Fig. 5C–D). Inhibition of glycolysis downregulates Cdkn1a expression and alleviates microvascular impairment in BPD. (A) Body weight comparison between BPD and 2-DG-treated BPD mice at postnatal day 14. (B) Lactate concentration in PVECs isolated from BPD and 2-DG-treated BPD mice. (C) Protein levels of pan-Kla in PVECs isolated from BPD and 2-DG-treated BPD mice. (D) Protein level of Cdkn1a in PVECs isolated from BPD and 2-DG-treated BPD mice. (E) Representative images and quantification of RVSP of BPD and 2-DG-treated BPD mice. (F) RVHI of BPD and 2-DG-treated BPD mice. (G) Representative images of lung histology and quantification of MLI and RAC of BPD and 2-DG-treated BPD mice, scale bar: 50 μm. (H) Representative images and quantification of EdU assay showing proliferation of PVECs isolated from BPD and 2-DG-treated BPD mice, scale bar: 50 μm. (I) Representative images and quantification of tube formation assay showing total branching length of PVECs isolated from BPD and 2-DG-treated BPD mice, scale bar: 100 μm. RVSP, right ventricular systolic pressure; MLI, mean linear intercept; RAC, radial alveolar count; PVECs, pulmonary vascular endothelial cells. Data are presented as mean ± SEM and analyzed by two-tailed unpaired Student’s t test. ** p < 0.01, * p < 0.05. In addition, 2-DG-treated BPD mice exhibited significantly decreased right ventricular systolic pressure (RVSP) and attenuated right ventricular hypertrophy, as measured by right ventricular hypertrophy index (RVHI) (Fig. 5E-F). Histological analysis further showed partial restoration of alveolar architecture (Fig. 5G). In parallel, PVECs isolated from 2-DG-treated BPD mice exhibited enhanced proliferative activity and improved angiogenic capacity compared with PVECs from untreated BPD mice (Fig. 5H-I). Together, these results suggest that inhibition of glycolysis decreased lactylation levels in PVECs, suppressed Cdkn1a expression, and partially ameliorated pulmonary microvascular injury in BPD mice. SP1 lactylation is increased in PVECs from BPD mice and promotes Cdkn1a transcription To investigate upstream regulators linking lactylation to Cdkn1a expression, pooled PVECs isolated from BPD mice were subjected to LC-MS/MS-based lactylome profiling (Fig. 6A). Among the identified lactylated proteins, SP1 was prioritized because its known role in regulating Cdkn1a transcription2728,. Immunoprecipitation of SP1 followed by pan-Kla immunoblotting confirmed increased SP1 lactylation in PVECs from BPD mice (Fig. 6B). To determine whether SP1 participates in Cdkn1a transcriptional regulation, we then assessed its binding at the Cdkn1a promoter. Motif analysis identified multiple putative SP1-binding elements within the promoter region, including E1, E2, E3 and E4 regions (Fig. 6C). CUT&Tag-qPCR demonstrated that SP1 enrichment was increased at the E1/E2 and E3 regions in PVECs from BPD mice, whereas no significant difference was detected at E4 (Fig. 6D). Moreover, LC-MS/MS identified three lactylation sites on SP1, namely K640, K686, and K694 (Fig. 6E). To assess the functional relevance of these sites, K640R, K686R, and K694R mutants were generated by site-directed mutagenesis. Among these mutants, K640 showed the greatest reduction in Cdkn1a mRNA expression (Fig. 6F). A representative MS/MS spectrum of the K640-lactylated peptide is shown in Fig. 6G. Furthermore, luciferase reporter assays showed that hyperoxia-induced activation of the Cdkn1a promoter was attenuated by mutation of SP1 K640 (Fig. 6H). Together, these findings reveal that increased SP1 lactylation, particularly at K640, promotes Cdkn1a transcription in PVECs isolated from BPD mice (Fig. 7). SP1 lactylation is increased in PVECs from BPD mice and promotes Cdkn1a transcription. (A) Schematic workflow of lactylated peptide enrichment and LC-MS/MS analysis in pooled PVECs isolated from BPD mice. (B) Input and IP-Western blots showing SP1 lactylation in PVECs isolated from control and BPD mice. (C) Predicted SP1-binding motifs and amplicon regions within the Cdkn1a promoter. (D) CUT&Tag-qPCR analysis showing SP1 enrichment at the Cdkn1a promoter in PVECs isolated from control and BPD mice. (E) Summary of lactylation sites identified on SP1 by mass spectrometry. (F) Relative Cdkn1a mRNA levels in HEK-293T cells transfected with SP1 wild-type, K640R, K686R, or K694R constructs under normoxic and hyperoxic conditions. (G) Representative MS/MS spectrum of the K640-lactylated SP1 peptide. (H) Luciferase reporter assay showing Cdkn1a promoter activity in HEK-293T cells transfected with SP1 wild-type or SP1 K640R expression plasmid under normoxic or hyperoxic conditions. Data are presented as mean ± SEM and analyzed by two-tailed unpaired Student’s t test or two-way ANOVA. *** p < 0.001, ** p < 0.01, * p < 0.05. Hyperoxia-induced SP1 lactylation promotes Cdkn1a expression, causing endothelial dysfunction and contributing to bronchopulmonary dysplasia. Discussion In this study, we found that lactylation levels were increased in vitro and in vivo models of BPD and accompanied by endothelial dysfunction. Integrated transcriptomic analyses subsequently identified Cdkn1a as a candidate hub gene, and further demonstrated it was upregulated in PVECs isolated from BPD mice. Inhibition of glycolysis reduced lactylation level, suppressed Cdkn1a expression, and partially alleviated pulmonary microvascular impairment in vivo. Moreover, increased SP1 lactylation in PVECs from BPD mice was accompanied by enhanced SP1 binding to the Cdkn1a promoter, while mutation of K640 reduced Cdkn1a transcription. These findings suggest that SP1 lactylation promotes Cdkn1a transcription, thereby contributing to endothelial dysfunction and the development of BPD. Post-translational modifications are increasingly recognized as important regulators of endothelial injury in hyperoxia-induced BPD models45. Previous studies have mainly focused on modifications such as acetylation and phosphorylation. For example, early hyperoxia exposure induces persistent hyperacetylation of H2aZac and H3K9ac at the NOS3 and STAT3 loci in PVECs46. In neonatal mice exposed to hyperoxia, Smad2 and Smad3 are hyperphosphorylated in PVECs, driving endothelial-to-mesenchymal transition47. Furthermore, intrapulmonary arteries from BPD rats exhibit decreased eNOS phosphorylation at the Ser1177 activation site, which is associated with impaired endothelium-dependent relaxation to acetylcholine48. These observations support the concept that PTM-dependent signaling is a key component of endothelial dysfunction in BPD. Our findings extend this framework by identifying lactylation as another PTM associated with BPD-related endothelial injury. SP1 is a GC-box-binding transcription factor involved in cell-cycle regulation49. It binds GC-rich promoter motifs through three conserved Cys2His2 zinc fingers that form its DNA-binding domain50. In our study, SP1 K640 lactylation was increased in PVECs from BPD mice and mutation of K640 reduced Cdkn1a transcription, supporting the idea that lactylation at this site may influence SP1-dependent transcription under hyperoxic conditions. Similar PTM-dependent regulation of SP1 has also been described. For instance, in endothelial cells exposed to shear stress, serine-phosphorylated SP1 shows enhanced occupancy at the MT1-MMP promoter, thereby repressing MT1-MMP transcription, a pathway linked to endothelial migration and angiogenesis51. Under hyperglycemic conditions, O-GlcNAcylation enhances SP1 binding to the VEGF-A promoter, thereby upregulating VEGF-A expression and contributing to proangiogenic signaling relevant to diabetic retinopathy52. During aging and oxidative stress, K16-SUMOylated SP1 displays reduced binding to the Prdx6 promoter, resulting in Prdx6 downregulation, ROS accumulation, and impaired cell survival53. Collectively, these findings suggest that post-translational modifications can reshape SP1-dependent promoter binding and transcriptional programs in response to metabolic or stress signals. Our study still has several limitations. First, 2-DG is a broad glycolytic inhibitor. Although it reduced lactylation and partially improved BPD-related phenotypes, its effects may not be exclusively mediated by endothelial lactylation. Second, considering recent studies have reported sex-dependent responses to hyperoxic exposure5455,, sex-specific differences remain to be further explored. Third, given the emerging heterogeneity of endothelial cells, our study did not address how lactylation affects distinct endothelial subpopulations in BPD. Further studies are needed to clarify how lactylation regulates subtype-specific endothelial changes during BPD progression. Despite these limitations, our study provides novel insights into the role of endothelial lactylation and its associated genes in the pathogenesis of BPD. In conclusion, our study suggests that endothelial lactylation contributes to impaired pulmonary vascular development in BPD, at least in part through SP1 K640 lactylation-associated regulation of Cdkn1a. These findings identify the SP1-Cdkn1a axis as a potential therapeutic target in BPD. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Acknowledgements We would like to express our heartfelt gratitude to the editors and the reviewers for their time, effort, and thoughtful contributions to our manuscript. Abbreviations Bronchopulmonary dysplasia Human umbilical vein endothelial cells Pulmonary vascular endothelial cells Specificity protein 1 Cyclin-dependent kinase inhibitor 1 A lysine lactylation Author contributions Lizhong Du, Wei Shi and Jia He designed this study. Tieshuai Liu and Lingchen Li contributed to the bioinformatics analysis. Ziming Zhang, Qiqi Li, Xinmin Ju, Lu Zu, Qianyun Cao established the animal models. Qiqi Li, Xinmin Ju, Mengting Hu, Lu Zu, Lingling Yan and Yun Ling performed molecular experiments. Ziming Zhang and Qiqi Li analysed the data. Ziming Zhang drafted the manuscript. Tian Xie, Jiajing Ge, Xiaolu Ma, Zheng Chen and Jiajun Zhu contributed to clinical sample processing. Lizhong Du, Jia He, Wei Shi and Ziming Zhang revised the manuscript. All authors approved the final manuscript. Funding This study was supported by grants from the Zhejiang Provincial Natural Science Foundation (No. LMS26H040003), National Natural Science Foundation of China (No. 82201890, No. 82241017, No. 32200748) and Wu Jieping Medical Foundation (No. 320.6750.2025-9-25). Data availability Publicly available datasets were utilized in this study. The datasets analyzed during the current study are available in the NCBI Gene Expression Omnibus (GEO) under accession numbers GSE29632, GSE25286, and GSE266988. Other data generated or analyzed during this study are available from the corresponding authors upon reasonable request. Declarations Competing interests The authors declare no competing interests. Ethics approval and consent to participate The Research Ethics Commission of Maternal and Child Health Hospital of Guiyang City approved the study. Informed consent was obtained from all subjects’ legal guardians. All procedures involving human participants were performed in accordance with the Declaration of Helsinki. All the animal experimental protocols were approved by Institutional Animal Care and Use Committee of Zhejiang University. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Ziming Zhang, Qiqi Li and Xinmin Ju. Contributor Information Jia He, Email: hejia1005@zju.edu.cn. Wei Shi, Email: 0021084@zju.edu.cn. Lizhong Du, Email: dulizhong@zju.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Data Availability Statement Publicly available datasets were utilized in this study. The datasets analyzed during the current study are available in the NCBI Gene Expression Omnibus (GEO) under accession numbers GSE29632, GSE25286, and GSE266988. Other data generated or analyzed during this study are available from the corresponding authors upon reasonable request.