Long non-coding RNA LINC00607 epigenetically regulates endothelial TSPAN18 to promote hypoxia-induced thromboinflammation

TL;DR

This study identifies a long non-coding RNA that promotes blood clotting and inflammation in low oxygen conditions by regulating a specific protein in blood vessel cells.

Contribution

The study reveals a novel epigenetic regulatory axis involving LINC00607, BRG1, and TSPAN18 in hypoxia-induced thromboinflammation.

Findings

LINC00607 is a hypoxia-inducible RNA that enhances TSPAN18 expression through chromatin remodeling.

The LINC00607–TSPAN18 axis promotes endothelial activation and monocyte adhesion under hypoxia.

BRG1-dependent chromatin remodeling is essential for TSPAN18 activation by LINC00607.

Abstract

Hypoxia promotes endothelial dysfunction and thrombosis through transcriptional and epigenetic mechanisms that remain incompletely understood. Long non-coding RNAs have emerged as important regulators of endothelial gene expression, yet their contribution to hypoxia-driven thromboinflammatory signaling is poorly defined. Here, we identified the endothelial-enriched long non-coding RNA LINC00607 as a hypoxia-inducible regulator linking chromatin remodeling to pro-thromboinflammatory endothelial activation. Integrative transcriptomic and chromatin analyses revealed that hypoxia upregulates LINC00607 and its downstream effector TSPAN18, a member of the tetraspanin superfamily of transmembrane proteins, accompanied by increased enhancer acetylation and chromatin accessibility under hypoxia. Loss- and gain-of-function experiments demonstrated that LINC00607 is required and sufficient for TSPAN18 induction. Mechanistically, LINC00607 associates with the chromatin remodeler BRG1 and facilitates BRG1-dependent activation of the TSPAN18 enhancer, while pharmacological inhibition of BRG1 attenuated this response. Functionally, the activation of LINC00607–TSPAN18 axis enhanced store-operated calcium entry and endothelial–monocyte adhesion under hypoxic conditions, key processes underlying endothelial activation and thromboinflammation. Although the ETS transcription factor ERG contributed to enhancer maintenance, LINC00607-mediated regulation of TSPAN18, persisted independently of ERG under hypoxia. Collectively, these findings define a hypoxia-responsive LINC00607–BRG1–TSPAN18 regulatory axis that integrates epigenetic remodeling with calcium-dependent endothelial activation, providing a new insight into the molecular basis of hypoxia-induced thromboinflammatory responses.

Figures

Full text

Venous thromboembolism (VTE), defined by deep vein thrombosis and pulmonary embolism, is a common and serious vascular condition with significant global impact (1, 2). VTE arises from a complex interplay of risk factors, including acute and subacute triggers, underlying demographic and behavioral factors, physiological and genetic predispositions, and acquired clinical conditions (3). Beyond these well-established factors, emerging evidence suggests that hypoxia acts as an independent risk factor by promoting thromboinflammation and enhancing thrombus formation (4, 5, 6). Hypoxic conditions such as high-altitude exposure, during severe COVID-19, or in obstructive sleep apnea have been increasingly associated with elevated VTE risk through mechanisms involving inflammation and enhanced coagulation (7, 8, 9). These observations highlight the need for continued investigation into the contribution of hypoxia to thrombotic risk, guided by growing but still evolving evidence.

The endothelium, a key component of Virchow’s triad, has long been underrecognized in its crucial role in preventing pathological venous thrombosis (10). In the presence of cardiovascular risk factors, endothelial cells (ECs) lose their regulatory function, and growing mounting evidence highlights their central role in thrombosis and their potential as therapeutic targets (11, 12). Endothelial activation promotes leukocyte adhesion and the release of procoagulant factors, contributing to the initiation of thrombotic events (13). Hypoxic conditions can disturb the balance between procoagulant and anticoagulant factors, modulating thrombotic regulation (4, 14). Acute hypoxia rapidly activates ECs, whereas prolonged hypoxia induces transcriptional reprogramming through the activation of specific transcription factors (15). Despite these insights, studies exploring hypoxia-induced endothelial gene expression and their regulatory mechanisms remain limited, particularly in the context of hypoxia-driven thrombosis. Recent studies have examined hypoxia-induced endothelial gene expression, highlighting the roles of hypoxia-regulated genes and epigenetic modulators in angiogenesis, pulmonary arterial hypertension, endothelial-to-mesenchymal transition (EndMT), and HIF1α/HIF2α-mediated transcriptional regulation (16, 17, 18, 19, 20).

Long non-coding RNAs (lncRNAs), classically defined as transcripts over 200 base pairs lacking canonical protein-coding function, regulate gene expression and protein interactions and have been implicated in multiple cardiovascular pathologies, including myocardial infarction, heart failure, and arrhythmias (21). The lncRNAs regulate gene expression by shaping chromatin architecture and epigenetic states, interacting with DNA and chromatin-associated proteins, and controlling transcription and post-transcriptional processes (22, 23). Abnormal lncRNA expression is associated with a spectrum of disease conditions owing to its ability to modulate genome, transcriptome, proteome, and epigenome dynamics (24). In recent years, several lncRNAs such as MEG3, LEENE, LINC00607, SENCR, and MIR181A1HG have been reported to regulate endothelial function and contribute to the development of cardiovascular pathologies (25, 26, 27, 28). MALAT1 and ANRIL are linked to atherosclerosis and regulate endothelial cell functions, while several lncRNAs, including Sirt1-AS, ANRIL, and NEAT1, have been associated with DVT, and our previous research has demonstrated the regulatory roles of LINC00659 and UXT-AS1 in high-altitude-induced thrombosis (29, 30, 31, 32, 33, 34, 35). Endothelial cells show altered expression of several lncRNAs such as MEG3, TUG1, MIR503HG, LINC00323, H19, MIR210HG, TTTY15, MALAT1, LINC00607, and GATA**6-AS under hypoxic conditions (27, 36, 37, 38, 39, 40). While lncRNAs are increasingly linked to vascular disease progression, their involvement in hypoxia-driven prothrombotic phenotype is not well understood. Studying their role in regulating endothelial homeostasis under hypoxia may provide valuable insights into cardiovascular disease mechanisms.

In this study, we investigated the impact of hypoxia on endothelial gene expression, with a particular focus on the role of LINC00607 in the epigenetic regulation of the prothrombotic gene Tetraspanin18 (TSPAN18). Our findings provide new insight into how hypoxia-driven transcriptional and chromatin remodeling events modulate store-operated calcium entry (SOCE) and contribute to endothelial activation and prothromboinflammatory responses. More broadly, this work underscores the functional significance of the non-coding transcriptome in regulating core endothelial signaling mechanisms under hypoxic stress.

Results

Transcriptomic landscape of hypoxia-exposed endothelial cells reveals activation of thromboinflammatory and vascular hallmarks

To understand the transcriptional regulation of endothelial cells under hypoxia, we analyzed the RNA-seq dataset GSE70330, comprising human umbilical vein endothelial cells (HUVECs) exposed to 1% O_2_ for 24 h compared to normoxic controls. Differential expression analysis identified a robust set of hypoxia-responsive protein-coding genes (DEGs) and long non-coding RNAs (DElncRNAs). The volcano plots (Fig. 1, A and B) show a marked shift in the endothelial transcriptome, with 676 protein-coding genes significantly upregulated (red) and 1234 downregulated (blue). Among lncRNAs, 693 were upregulated, and 210 were downregulated under hypoxic stress (|log_2_FC| ≥ 1, adjusted p < 0.05). Among the DElncRNAs, LINC00607 and HIF1A-AS3 emerged as prominent hypoxia-responsive lncRNAs, suggesting a potential regulatory role in endothelial alteration to low oxygen. Gene ontology (GO) enrichment analysis of upregulated DEGs (Fig. 1C) revealed significant enrichment of biological processes related to extracellular matrix organization, cell-substrate adhesion, angiogenesis, leukocyte chemotaxis, and response to hypoxia, indicating endothelial remodeling and activation of inflammatory signaling. In contrast, downregulated DEGs (Fig. 1D) were predominantly associated with cell cycle-associated processes, such as chromosomal segregation and DNA replication, suggesting proliferative arrest under hypoxia. KEGG pathway analysis (Fig. 1, E and F) further demonstrated that upregulated genes were enriched in PI3K-Akt, NF-κB, and TNF signaling pathways, along with focal adhesion and cytoskeletal regulation, consistent with endothelial activation and pro-inflammatory signaling. Conversely, downregulated genes were associated with DNA replication, homologous recombination, and mismatch repair, reflecting global transcriptional repression of DNA synthesis and repair mechanisms.Figure 1**Transcriptomic and co-expression network analysis reveals hypoxia-induced thromboinflammatory signatures in endothelial cells.**A, Volcano plot depicting differentially expressed protein-coding genes (DEGs) in HUVECs exposed to hypoxia (1% O_2_, 24 h) compared to normoxia (GSE70330 dataset). Significantly upregulated and downregulated genes are shown in red and blue, respectively (|log_2_FC| ≥ 1, adjusted p < 0.05). B, Volcano plot of differentially expressed long non-coding RNAs (DElncRNAs) under hypoxia. The top 10 most significantly dysregulated lncRNAs are highlighted, including LINC00607, HIF1A-AS3 (C) Gene Ontology (Biological Process) enrichment of upregulated DEGs demonstrates activation of pathways associated with extracellular organization, response to hypoxia, leukocyte chemotaxis, angiogenesis, and vascular development. D, GO-BP enrichment of downregulated DEGs reveals suppression of cell cycle–related processes, including chromosomal segregation, mitotic division, and DNA replication. E, KEGG pathway analysis of upregulated DEGs highlights enrichment in PI3K-Akt, NF-κB, and TNF signaling pathways, cytoskeletal regulation, and focal adhesion, consistent with endothelial activation. F, KEGG pathway analysis of downregulated DEGs shows significant enrichment of DNA replication, mismatch repair, and homologous recombination pathways, indicating transcriptional repression of proliferative programs. G, disease ontology-based gene set enrichment analysis (DO-GSEA) reveals associations with vascular, thrombotic, and inflammatory disorders, including cardiovascular and connective tissue diseases. H, Co-expression network constructed between the top 10 significantly dysregulated lncRNAs and 50 DEGs selected from hallmark hypoxia, angiogenesis, coagulation, and inflammatory response gene sets.

Disease Ontology-based gene set enrichment analysis (DO-GSEA; Fig. 1G) revealed significant associations of the hypoxia-induced transcriptome with vascular, cardiovascular, and connective tissue diseases, underscoring the development of a pro-thrombotic endothelial phenotype under hypoxia. To gain further insight into the potential regulatory architecture underlying these changes, we focused on hallmark gene sets related to hypoxia, angiogenesis, coagulation, and inflammatory response from the Molecular Signature Database (MSigDB) (https://www.gsea-msigdb.org/gsea/msigdb/human/collections.jsp) (Table S1). Additionally, a PubMed literature survey (2015–2025) using the keywords “thromboinflammation” and “endothelial” was conducted to identify genes recently implicated in endothelial thromboinflammatory signaling. Integrating these datasets, we identified 114 DEGs that represent the functional backbone of hypoxia-mediated vascular and thromboinflammatory pathways.

Given that co-expressed genes often exhibit related biological functions, we constructed a correlation-based co-expression network to infer the potential roles of the top 10 most significantly dysregulated lncRNAs (highlighted in Fig. 1B). Pearson correlation coefficients were calculated between these DElncRNAs and the top 50 DEGs from the hallmark and literature-derived thromboinflammatory gene sets, considering correlations above 0.85 as statistically significant. A co-expression network highlighting the co-expressed genes grouped into different modules was constructed based on correlation values to identify co-expressed lncRNA-mRNA pairs using igraph package in R. Among these, LINC00607 showed a strong positive correlation with TSPAN18 within a defined (cyan) module enriched for vascular and coagulation-associated genes. Notably, both LINC00607 and TSPAN18 exhibited endothelial cell-enriched expression in GTEx (https://gtexportal.org/home/) and protein-atlas single-cell datasets (https://www.proteinatlas.org/search/tspan18), suggesting a cell type-specific regulatory relationship.

Together, these integrative transcriptomic and network analyses highlight LINC00607 as a hypoxia-responsive lncRNA potentially involved in endothelial thromboinflammatory signaling. These findings prompted us to further investigate the regulatory relationship between LINC00607 and TSPAN18, and to functionally characterize their roles under hypoxic conditions.

LINC00607 knockout reverses hypoxia-induced endothelial transcriptional programs and attenuates TSPAN18 activation

Given the strong transcriptional association of LINC00607 with TSPAN18 and other hypoxia-responsive genes identified (Fig. 1), we next examined whether LINC00607 functionally contributes to the hypoxia-induced endothelial transcriptome. To address this, we analysed publicly available LINC00607 knockout (KO) RNA-seq data in HUVECs and compared it with the hypoxia-exposed transcriptome (1% O_2_, 24 h). A comparative analysis of DEGs revealed a substantial reciprocal relationship between the two datasets (Fig. 2A). Specifically, 199 genes were upregulated under hypoxia but downregulated upon LINC00607 depletion, while 200 genes displayed the opposite trend being downregulated under hypoxia but upregulated in LINC00607 KO. This reciprocal pattern indicates that LINC00607 positively regulates a subset of hypoxia-responsive genes, supporting its functional involvement in hypoxia-mediated endothelial activation. Gene Ontology (GO, Biological Process) enrichment analysis of genes that were upregulated under hypoxia but downregulated upon LINC00607 knockout revealed significant enrichment in processes associated with extracellular matrix organization, endothelial differentiation, and endothelium development (Fig. 2B). In contrast, genes that were upregulated upon LINC00607 KO but downregulated under hypoxia were enriched for cell cycle, DNA replication, and RNA metabolic processes (Fig. 2C), indicating that the loss of LINC00607 may partially reverse the adaptive endothelial transcriptional program induced by hypoxia. Consistent with these findings, KEGG pathway analysis (Fig. 2, D and E) demonstrated that the PI3K-Akt and Rap1 signaling pathways, which were activated under hypoxia, were suppressed upon LINC00607 depletion, whereas DNA repair and cell cycle regulatory pathways became enriched, reflecting a shift from endothelial activation toward a proliferative transcriptomic state. To further understand shared biological pathways, we performed hallmark gene set enrichment analyses across the hypoxia and LINC00607 KO datasets (Fig. 2F). The overlapping signatures included hypoxia response, angiogenesis, coagulation, and inflammatory response pathways, key processes that characterize the thromboinflammatory endothelial phenotype observed under hypoxic stress. These overlaps were corroborated by the Gene Ontology Molecular Function (MF) and Cellular Component (CC) categories shown in Figure S1, which further demonstrated enrichment of extracellular matrix components, cytoskeletal structures, and ATP-dependent chromatin regulatory activities.Figure 2Integrative transcriptomic and chromatin accessibility analysis reveals that LINC00607 regulates TSPAN18 expression under hypoxia.A, Venn diagram showing overlap between differentially expressed genes (DEGs) in hypoxia-exposed HUVECs and LINC00607 knockout (KO) transcriptomes. A total of 199 genes were upregulated under hypoxia but downregulated in LINC00607 KO, while 200 genes were upregulated in LINC00607 KO and downregulated under hypoxia, suggesting reciprocal regulation. B and C, Gene Ontology (Biological Process) enrichment of the reciprocal gene sets demonstrates that genes upregulated under hypoxia but downregulated in LINC00607 KO are associated with extracellular matrix organization, angiogenesis, and endothelial differentiation. Conversely, genes upregulated in LINC00607 KO but downregulated under hypoxia are enriched for cell cycle, DNA replication, and RNA processing pathways, indicating loss of adaptive transcriptional programs upon LINC00607 depletion. D and E, KEGG pathway analysis of the reciprocal gene sets reveals activation of PI3K–Akt, Rap1, and NF-κB signaling pathways under hypoxia, whereas LINC00607 KO suppresses these while inducing DNA repair and cell cycle pathways. F, Heatmap showing overlapping hallmark gene sets enriched in both *LINC00607*-regulated and hypoxia-responsive DEGs. The overlapping genes correspond to hallmark pathways associated with hypoxia response, angiogenesis, coagulation, and inflammatory signaling, defining a thromboinflammatory endothelial signature. G–I, Volcano plots showing expression trends of LINC00607 and TSPAN18 across datasets. Both genes are upregulated under hypoxia, significantly reduced upon LINC00607 KO, and downregulated under combined Acriflavine (HIF-1α inhibitor) and hypoxia treatment, confirming HIF-dependent regulation of the *LINC00607-TSPAN**18* axis. J and K, GRO-seq strand-specific IGV tracks of LINC00607 and TSPAN18 under normoxia and hypoxia demonstrate increased nascent transcription under hypoxic conditions, consistent with transcriptional activation. L and M, ATAC-seq tracks at the LINC00607 and TSPAN18 loci show enhanced chromatin accessibility under hypoxia compared to normoxia, suggesting open chromatin states at regulatory elements. N, ATAC-seq tracks of TSPAN18 in LINC00607 KO cells show reduced accessibility compared to control, supporting that LINC00607 facilitates chromatin opening at the TSPAN18 locus.

To validate these transcriptomic associations, we examined independent RNA-seq datasets of hypoxia-treated HUVECs (1% O_2_ for 24 h and 48 h) and observed consistent upregulation of both LINC00607 and TSPAN18 (Fig. S1). Conversely, in an independent LINC00607 knockdown dataset, TSPAN18 expression was markedly reduced, reinforcing a positive regulatory relationship between the two transcripts. We further evaluated the role of HIF signaling in this axis by analyzing RNA-seq data from Acriflavine-treated HUVECs, where HIF-1α activity is pharmacologically inhibited. Both LINC00607 and TSPAN18 were significantly downregulated under combined Acriflavine + hypoxia treatment (Fig. 2I), confirming that the LINC00607-TSPAN18 regulatory axis is HIF-dependent. Genome-wide nascent transcription analysis using GRO-seq strand-specific IGV tracks revealed increased transcriptional activity at both LINC00607 and TSPAN18 loci under hypoxia compared to normoxia (Fig. 2, J and K), consistent with direct transcriptional activation. Corresponding ATAC-seq profiles demonstrated enhanced chromatin accessibility at both loci in hypoxia (Fig. 2, L and M), supporting the notion that hypoxia induces open chromatin states at regulatory regions of these genes. Strikingly, LINC00607 KO resulted in reduced ATAC-seq signal intensity at the TSPAN18 locus (Fig. 2N), indicating that LINC00607 facilitates chromatin accessibility and transcriptional activation of TSPAN18 under hypoxic conditions. To evaluate whether other hypoxia-responsive lncRNAs identified in our co-expression network (Fig. 1H) exhibit similar regulatory effects, we analyzed TSPAN18 expression following HIF1A-AS3 knockdown. Unlike LINC00607, depletion of HIF1A-AS3 did not alter TSPAN18 expression, suggesting that the LINC00607-TSPAN18 relationship is specific rather than a general feature of hypoxia-induced lncRNAs. Collectively, these findings demonstrate that LINC00607 is an essential component of the hypoxia-induced endothelial transcriptional network, functioning as a positive regulator of pro-coagulatory and angiogenic gene expression. Loss of LINC00607 disrupts these adaptive programs, reverses hypoxia-induced endothelial activation, and reduces chromatin accessibility at the TSPAN18 locus, implicating the LINC00607-TSPAN18 axis as a central HIF-dependent regulatory mechanism in hypoxia-driven thromboinflammation.

LINC00607 interacts with BRG1 chromatin remodeling complex to regulate TSPAN18 transcription under hypoxia

Previous studies have shown that LINC00607 exerts chromatin-associated regulatory functions via the BRG1 (SMARCA4) chromatin remodeling complex, modulating endothelial and gene programs (27, 41). To determine whether a similar mechanism operates in the context of hypoxia-driven thromboinflammation, we investigated the transcriptional and chromatin-level interplay between LINC00607, BRG1, and TSPAN18 in endothelial cells. RNA Polymerase II (Pol II) ChIP-seq profiles revealed increased Pol II occupancy across the LINC00607 locus under hypoxia compared to normoxia (Fig. 3, A and B), consistent with the transcriptional activation observed in RNA-seq and GRO-seq datasets. A similar pattern was observed at the TSPAN18 locus (Fig. 3, C and D), confirming increased Pol II engagement and active transcription in response to hypoxic stimulation.Figure 3Hypoxia induces transcriptional activation of the *LINC00607-TSPAN**18* axis through BRG1-mediated chromatin remodeling.**A and B, RNA Polymerase II ChIP-seq tracks and occupancy profiles across the LINC00607 locus under hypoxia and normoxia demonstrate increased Pol II enrichment under hypoxia, indicating transcriptional activation. C and D, similar Pol II accumulation is observed at the TSPAN18 locus under hypoxia, confirming enhanced transcriptional engagement consistent with RNA-seq and GRO-seq data. E, iCLIP signal tracks reveal BRG1 interaction sites across LINC00607 under UV-crosslinking conditions, suggesting that LINC00607 directly associates with BRG1 at the RNA level. F, RNA immunoprecipitation (RIP) assay followed by qPCR validation shows significant enrichment of LINC00607 transcripts in BRG1 immunoprecipitates compared to IgG control (∗∗p < 0.001), confirming physical interaction between LINC00607 and BRG1. G, CUT&RUN BRG1 profiles demonstrate robust BRG1 occupancy at enhancer regions within the TSPAN18 locus. H, CUT&RUN BRG1 profiles demonstrate loss of BRG1 occupancy at enhancer regions within the TSPAN18 locus under LINC00607 knock-down condition, demonstrating LINC00607 dependent occupancy at TSPAN18 locus. I, targeted BRG1 degradation using PROTAC treatment leads to complete loss of BRG1 occupancy at the TSPAN18 locus, followed by partial recovery at later time points, confirming BRG1’s essential role in maintaining chromatin engagement at this site. J, ATAC-seq profiles reveal a marked reduction in chromatin accessibility at the TSPAN18 locus upon BRG1 depletion via PROTAC treatment, highlighting BRG1’s requirement for open chromatin conformation and transcriptional competence at TSPAN18.

To investigate whether LINC00607 physically associates with chromatin modifiers, we analyzed BRG1 iCLIP-seq data, which revealed strong UV-dependent crosslinking signals along the LINC00607 transcript (Fig. 3E), supporting a potential RNA–protein association between LINC00607 and BRG1. This finding was independently validated by RNA immunoprecipitation (RIP), using BRG1-specific antibodies, which showed significant enrichment of LINC00607 transcripts in BRG1 immunoprecipitates compared to IgG control (∗p < 0.001; Fig. 3F), confirming that LINC00607 directly binds BRG1 in endothelial cells. While these data are consistent with an association between LINC00607 and BRG1, they do not exclude the possibility of an indirect interaction mediated by additional chromatin-associated factors. We next examined whether BRG1 occupancy regulates the transcriptional activity of TSPAN18. CUT&RUN profiling of BRG1 revealed robust BRG1 binding at enhancer regions within the TSPAN18 locus (Fig. 3G), co-localizing with sites of open chromatin and elevated Pol II density. Notably, targeted degradation of BRG1 using PROTAC treatment resulted in complete loss of BRG1 occupancy at the TSPAN18 enhancer (Fig. 3H) and a concurrent reduction in chromatin accessibility, as shown by ATAC-seq analysis (Fig. 3I). Partial recovery of BRG1 binding at later time points confirmed the specificity and dynamic reversibility of BRG1-dependent chromatin remodeling at this locus. Collectively, these findings establish that BRG1 recruitment is essential for maintaining open chromatin and transcriptional competence at the TSPAN18 enhancer, and that LINC00607 likely facilitates this recruitment under hypoxic conditions.

To complement these findings, we assessed the subcellular localization of LINC00607 (Fig. S2). Fractionation analysis revealed that LINC00607 is predominantly chromatin-associated, with a measurable presence in the cytoplasmic fraction, in contrast to prior reports describing it as exclusively nuclear. Its chromatin enrichment profile was comparable to that of MALAT1, used as a nuclear/chromatin control, suggesting that LINC00607 may shuttle between nuclear and cytoplasmic compartments while exerting chromatin-regulatory functions. Analysis of BRG1 iCLIP crosslink data further identified LINC00607 among the top 50 BRG1-bound lncRNAs (Fig. S2C), reinforcing its relevance as a BRG1-associated transcript. To validate the broader chromatin engagement pattern, we examined BRG1 overexpression (OE) CUT&RUN profiles in a PANC1 cell line dataset. Despite the distinct cellular context, BRG1 OE led to pronounced enrichment of BRG1 peaks at the TSPAN18 enhancer (Fig. S2, D and E), supporting the notion that BRG1 binding at this locus represents a conserved enhancer feature across cell types. Collectively, these findings establish that LINC00607 interacts with BRG1 through direct RNA-protein interaction and facilitates BRG1-mediated chromatin remodeling at the TSPAN18 enhancer, thereby promoting transcriptional activation under hypoxia. These observations align with recent reports implicating LINC00607–BRG1 complexes in oxygen-sensitive chromatin regulation and expand their role to endothelial thromboinflammatory signaling. LINC00607 positively regulates TSPAN18 expression in a hypoxia- and BRG1-dependent manner.

To functionally validate the transcriptomic, chromatin, and interaction-based observations from Figure 1, Figure 2, Figure 3, we next examined whether LINC00607 directly regulates TSPAN18 expression in endothelial cells under hypoxic conditions. Consistent with our RNA-seq and GRO-seq results, quantitative PCR analysis in EA.hy926 cells showed that LINC00607 expression increased significantly under hypoxia (1% O_2_, 24 h) and following DMOG treatment, a pharmacological stabilizer of HIF-1α (Fig. 4, A and B). Importantly, TSPAN18 expression mirrored this induction, showing robust upregulation under both hypoxia and DMOG (Fig. 4, C and D), confirming that both transcripts are HIF-responsive and integral to the hypoxia-regulated transcriptional network identified earlier. To determine whether LINC00607 directly influences TSPAN18 expression, we performed loss-of-function experiments using two independent locked nucleic acid (LNA) antisense oligonucleotides (LNA1 and LNA2) targeting LINC00607 under hypoxic conditions. Both LNAs achieved efficient LINC00607 knockdown (Fig. 4, E and G), accompanied by a marked reduction in TSPAN18 transcript levels relative to LNA controls (Fig. 4, F and H). These results indicate that endogenous LINC00607 is required for optimal TSPAN18 expression under hypoxia, consistent with our knockout transcriptome and chromatin-accessibility findings. Conversely, LINC00607 overexpression (LINC00607_OE) in endothelial cells led to a significant increase in TSPAN18 mRNA abundance compared to pcDNA controls (Fig. 4, I and J), further supporting a positive and directional regulatory relationship. Together, the loss- and gain-of-function results strongly validate the co-expression and chromatin-based predictions, establishing LINC00607 as an upstream regulator of TSPAN18. Given the robust interaction between LINC00607 and BRG1 shown in Figure 3, we next evaluated whether BRG1 activity is required for TSPAN18 induction. Treatment of hypoxic cells with PFI3, a selective inhibitor of the BRG1 bromodomain (Fig. 4K), significantly attenuated TSPAN18 expression, confirming BRG1-dependent activation of TSPAN18. This is consistent with the BRG1 CUT&RUN and PROTAC-degradation data showing BRG1 occupancy and chromatin modulation at the TSPAN18 enhancer. Finally, efficient siRNA-mediated knockdown of TSPAN18 under hypoxia (Fig. 4L) was validated to support downstream functional assays investigating the role of the LINC00607-TSPAN18 axis in endothelial thromboinflammation. Collectively, these results demonstrate that LINC00607 regulates TSPAN18 expression through a hypoxia- and BRG1-dependent mechanism, integrating transcriptional, chromatin, and functional layers of regulation. These findings position the LINC**00607-BRG1-TSPAN**18 axis as a key regulatory pathway driving endothelial activation under hypoxic stress, providing a mechanistic foundation for subsequent functional interrogation of its role in thromboinflammatory endothelial phenotypes. Both LINC00607 and TSPAN18 were significantly upregulated upon DMOG treatment and downregulated with Acriflavine dataset (Fig. 2I), confirming HIF1α-dependent transcriptional control. However, analysis of public HIF1α ChIP-seq data in endothelial cells revealed no direct HIF1α binding peaks across the LINC00607 locus (Fig. S1), including its enhancer region. These findings suggest that HIF1α regulates LINC00607 indirectly, likely through secondary transcriptional mediators or epigenetic modifiers activated under hypoxic conditions.Figure 4**qPCR validation of hypoxia-induced regulation of LINC00607 and TSPAN18 in endothelial cells.**A–D, quantitative PCR analysis of LINC00607 and TSPAN18 expression in EA.hy926 endothelial cells exposed to normoxia or hypoxia (1% O2, 24 h), and under normoxic conditions with DMOG treatment (0.5 mM, 24 h). Both LINC00607 and TSPAN18 were significantly upregulated under hypoxia and DMOG treatment, confirming HIF-dependent induction. E–H, knockdown of LINC00607 using two independent locked nucleic acid (LNA) oligonucleotides (LNA1 and LNA2) under hypoxic conditions resulted in a marked reduction of both LINC00607 and TSPAN18 transcript levels compared to LNA control, indicating that LINC00607 positively regulates TSPAN18 expression. I and J, Overexpression of LINC00607 (LINC00607_OE) significantly enhanced TSPAN18 transcript abundance compared to the pcDNA control, further supporting a direct regulatory relationship. K, treatment with PFI3 (20 μM, 24 h), a selective inhibitor of BRG1 bromodomain activity, attenuated TSPAN18 expression under hypoxia, confirming that TSPAN18 upregulation is BRG1-dependent. L, efficient siRNA-mediated knockdown of TSPAN18 under hypoxic conditions was validated by qPCR and used in subsequent functional assays. Data represent mean ± SEM from at least three independent experiments. Statistical significance was determined using Student’s t test: p < 0.05 (∗), p < 0.01 (∗∗), p < 0.001 (∗∗∗), p < 0.0001 (∗∗∗∗).

LINC00607 and BRG1 regulate the membrane-associated protein TSPAN18 under hypoxia

To determine whether the transcriptional regulation of TSPAN18 by LINC00607 and BRG1 extends to the protein level, we assessed TSPAN18 levels by Western blotting and immunocytochemistry under conditions of hypoxia, LINC00607 modulation, and BRG1 inhibition. Consistent with transcriptomic findings, TSPAN18 protein abundance was significantly increased in endothelial cells exposed to hypoxia (1% O2, 24 h) or DMOG treatment (0.5 mM, 24 h) (Fig. 5, A and B), confirming HIF-dependent induction. Quantitative analysis of normalized band intensities showed a robust upregulation of TSPAN18 under both conditions (p = 0.00028). Silencing LINC00607 using two independent LNAs under hypoxia resulted in a marked decrease in TSPAN18 protein levels compared with LNA control (Fig. 5, C and D; p = 2.3 × 10-5), indicating that LINC00607 is required to maintain TSPAN18 expression. Conversely, LINC00607 overexpression (LINC00607_OE) under normoxia significantly increased TSPAN18 protein abundance relative to pcDNA-transfected cells (Fig. 5, E and F; p < 0.01), demonstrating that elevated LINC00607 is sufficient to increase TSPAN18 expression. Inhibition of BRG1 activity with the selective bromodomain inhibitor PFI3 (20 μM, 24 h) significantly attenuated hypoxia-induced TSPAN18 upregulation (Fig. 5, G and H; p < 0.05), consistent with BRG1’s requirement for chromatin activation at this locus. Immunocytochemical analysis (Figs. S4 and S5) further confirmed these findings. TSPAN18, a membrane-localizing tetraspanin protein, displayed strong surface fluorescence under hypoxia and DMOG, whereas LINC00607 knockdown or BRG1 inhibition markedly reduced membrane signal intensity. In contrast, LINC00607 overexpression enhanced TSPAN18 membrane localization, consistent with the Western blot quantifications. Together, these findings establish that TSPAN18 induction and membrane localization under hypoxia require coordinated regulation by LINC00607 and BRG1, defining a LINC00607-BRG1-TSPAN18 signaling axis as a key mechanism operating at both transcriptional and protein levels. This regulatory pathway sets the molecular basis for subsequent calcium signaling and prothromboinflammatory activation of endothelial cells under hypoxic conditions.Figure 5**Western blot validation of hypoxia-induced regulation of TSPAN18 and its dependence on LINC00607 and BRG1.A and B, Western blot analysis of TSPAN18 protein levels in EA.hy926 cells under normoxia, hypoxia (1% O_2_, 24 h), and DMOG treatment (0.5 mM, 24 h). Quantification of normalized band intensity (β-actin as loading control) shows significant upregulation of TSPAN18 under hypoxia and DMOG treatment (ANOVA, p = 0.00028). C and D, Western blot and quantification showing that knockdown of LINC00607 using two independent LNA oligonucleotides (LNA1 and LNA2) under hypoxic conditions markedly reduced TSPAN18 protein levels compared to LNA control (ANOVA, p = 2.3 × 10^–5^), confirming *LINC00607*-mediated regulation. E and F, overexpression of LINC00607 (LINC00607_OE) led to a significant increase in TSPAN18 protein abundance relative to pcDNA control (∗∗p < 0.01), demonstrating that LINC00607 is sufficient to drive TSPAN18 upregulation. G and H, pharmacological inhibition of BRG1 using PFI3 (20 μM, 24 h) under hypoxia significantly attenuated TSPAN18 expression (p < 0.05), establishing BRG1 dependence of TSPAN18 regulation. All protein levels were normalized to β-actin, and data are represented as mean ± SEM from three independent biological replicates. Statistical significance was determined using one-way ANOVA followed by Tukey’s post hoc test: p < 0.05 (∗), p < 0.01 (∗∗), p < 0.001 (∗∗∗), p < 0.0001 (∗∗∗∗).

ERG influence the enhancer activity of the LINC00607- TSPAN18 axis under hypoxia

Previous studies have shown that LINC00607 interacts with BRG1 to modulate ERG binding at chromatin (27), while ERG itself has been identified as a master regulator of endothelial enhancers (42). Based on these findings, we sought to delineate the contribution of ERG to the regulation of the LINC00607-TSPAN18 axis under hypoxia.

To begin, we examined enhancer-associated histone marks H3K27ac and H3K4me1 across the TSPAN18 and LINC00607 loci under normoxic and hypoxic conditions. Both loci exhibited a clear enrichment of these marks under hypoxia, consistent with enhancer activation (Fig. 6, A–D). Given ERG’s established role in enhancer regulation, we analyzed ERG ChIP-seq datasets and identified distinct ERG occupancy peaks within the TSPAN18 locus under normoxia, indicating direct ERG binding (Fig. 6E). Consistent with this observation, H3K27ac ChIP-seq following ERG knockdown (siERG) revealed reduced acetylation at the TSPAN18 enhancer, confirming ERG-dependent enhancer activation (Fig. 6F).Figure 6**ERG modulates enhancer activity and transcriptional dynamics of the LINC00607–TSPAN18 axis under hypoxia.**A–D, genome browser (IGV) tracks showing enhancer-associated histone marks H3K27ac and H3K4me1 across the TSPAN18 and LINC00607 loci under normoxia and hypoxia, demonstrating increased enrichment of both marks under hypoxic conditions, consistent with enhancer activation. E, ERG ChIP-seq tracks displaying distinct ERG binding peaks within the TSPAN18 locus under normoxia, indicating direct ERG occupancy at its regulatory region. F, H3K27ac ChIP-seq tracks under ERG knockdown (siERG) showing decreased acetylation at the TSPAN18 locus, confirming ERG-dependent enhancer activity. G–I, quantitative PCR validation under hypoxia showing: (G) efficient ERG knockdown (p < 0.001), (G) paradoxical upregulation of TSPAN18 expression upon ERG depletion (p < 0.05), and (H) increased LINC00607 levels under the same condition (p < 0.05). These findings indicate that while ERG contributes to enhancer acetylation at TSPAN18, its loss under hypoxia leads to compensatory upregulation of LINC00607, which may sustain TSPAN18 expression through an alternative regulatory mechanism involving BRG1.

To validate these findings at the transcriptional level, we performed qPCR under hypoxia and confirmed efficient ERG knockdown (p < 0.001). Unexpectedly, however, TSPAN18 expression increased rather than decreased following ERG depletion, while LINC00607 levels were also upregulated (Fig. 6, G–I). These results suggest that although ERG facilitates enhancer acetylation at TSPAN18, loss of ERG triggers a compensatory mechanism leading to upregulation of LINC00607.This rise in LINC00607 may be sufficient to maintain TSPAN18 expression through an ERG-independent mechanism, likely involving LINC**00607-BRG1chromatin remodeling axis. To investigate this further, we examined the LINC00607 enhancer status following ERG knockdown. H3K27ac ChIP-seq analysis revealed a marked increase in acetylation at the LINC00607 locus under siERG conditions (Fig. 7A), corroborating the qPCR results and suggesting that ERG loss enhance LINC00607 activation. To assess whether this upregulation of LINC00607 mediates the compensatory activation of TSPAN18, we performed co-transfection experiments with siERG and LINC**00607-targeting LNA2 under hypoxia. As expected, co-transfection resulted in efficient ERG silencing, a modest reduction in LINC00607 (likely due to counteracting activation upon ERG loss), and no significant change in TSPAN18 expression (Fig. 7, B–D). These findings collectively indicate that LINC00607 is required for sustaining TSPAN18 expression and that its induction under ERG knockdown represents a compensatory feedback mechanism.Figure 7**ERG and FLI1 cooperatively modulate enhancer activity of LINC00607 and TSPAN18 under hypoxia.**A, H3K27ac ChIP-seq tracks showing increased acetylation at the LINC00607 locus under ERG knockdown (siERG), consistent with qPCR results in Figure 6, indicating ERG loss enhances LINC00607 enhancer activation. (B–D) qPCR analysis following co-transfection of siERG and LNA2 (targeting LINC00607) under hypoxia. B, ERG transcript levels were efficiently reduced. C, LINC00607 expression showed a minor decrease due to ERG-mediated activation offsetting knockdown efficiency, and (D) TSPAN18 expression exhibited a non-significant increase, suggesting that ERG loss alone does not suppress TSPAN18 when LINC00607 remains partially active. E, ChIP-seq tracks for ERG and FLI1 showing overlapping binding sites within the LINC00607 locus, indicating potential cooperative binding of ETS transcription factors in endothelial regulation. F and G, H3K27ac ChIP-seq profiles under combined ERG and FLI1 knockdown reveal a marked reduction in acetylation at both LINC00607 and TSPAN18 loci, confirming that both transcription factors are required to maintain enhancer activity. These findings suggest that FLI1 may compensate for ERG loss at the LINC00607 enhancer, whereas dual depletion abrogates enhancer activation across the LINC00607–TSPAN18 regulatory axis.

Given that ERG and FLI1 have been shown to cooperatively maintain endothelial gene programs and enhancer identity (43, 44), we next examined whether FLI1 might compensate for ERG loss. ChIP-seq tracks revealed overlapping ERG and FLI1 binding peaks within the LINC00607 locus (Fig. 7E), suggesting shared regulatory occupancy. Further, H3K27ac ChIP-seq under dual ERG and FLI1 knockdown showed a loss of enhancer acetylation at both the LINC00607 and TSPAN18 loci (Fig. 7, F and G), implying that FLI1 preserves enhancer activity of LINC00607 in the absence of ERG. These results suggest a model in which ERG and FLI1 cooperatively maintain enhancer activation, while LINC**00607-BRG1 regulation of TSPAN18 operates independently of ERG signaling.

Finally, to determine whether enhancer activation is accompanied by higher-order chromatin reorganization, we analyzed Hi-C data at the TSPAN18 locus. Virtual 4 C plots using the TSPAN18 transcription start site as a viewpoint showed local contact enrichment under both normoxia and hypoxia, consistent with stable promoter–enhancer interactions (Fig. S3A). However, insulation score analysis across chr11:44.4 to 45.2 Mb revealed reduced insulation under hypoxia (Fig. S3B), indicating weaker domain boundaries and increased chromatin accessibility, in line with the observed enhancer activation.

Although our data indicate that ERG modulates enhancer activity at both the LINC00607 and TSPAN18 loci, the precise nature of its regulatory influence on endothelial enhancers under hypoxia remains undefined. The observed increase in LINC00607 expression following ERG knockdown could suggest that ERG may act as a transcriptional repressor of LINC00607 under certain conditions, and its loss leads to de-repression and enhancer activation. Alternatively, this response may arise from compensatory activation by related ETS factors such as FLI1, which can bind overlapping genomic regions and maintain endothelial enhancer function. These findings underscore the complexity and plasticity of ETS factor networks in endothelial chromatin regulation. Future studies integrating chromatin occupancy, enhancer dynamics, and non-coding RNA interactions will be necessary to elucidate the precise role of ERG and its cooperation with FLI1 in maintaining endothelial enhancer integrity under hypoxic stress.

LINC00607–TSPAN18 axis regulates store-operated calcium entry in endothelial cells

Given that TSPAN18 has been linked to store-operated calcium entry (SOCE) and Ca^2+^-dependent endothelial activation (45, 46), we next examined whether the hypoxia-induced LINC**00607-TSPAN**18 axis functionally influences Ca^2+^ dynamics. Intracellular calcium levels were measured in in EA.hy926 endothelial cells using live-cell fluorescence kinetics under various experimental conditions. Exposure to hypoxia (1% O_2_, 24 h) or treatment with DMOG (0.5 mM, 24 h) significantly increased Ca^2+^ influx compared with normoxic controls (Fig. 8A). The enhanced Ca^2+^ entry following store depletion indicates that hypoxic signaling activates SOCE, consistent with the upregulated expression of TSPAN18 observed in our transcriptomic data and subsequent validation. To determine whether LINC00607 mediates this calcium influx, we silenced LINC00607 using two independent locked-nucleic-acid oligonucleotides (LNA1 and LNA2) under hypoxia. Both knockdowns markedly reduced SOCE amplitude relative to LNA controls (Fig. 8B), indicating that LINC00607 mediates full activation of hypoxia-induced Ca^2+^ influx. Because LINC00607 recruits BRG1 to regulatory enhancers (Figure 3, Figure 4, Figure 5), we tested whether BRG1 contributes to Ca^2+^ regulation. Pharmacological inhibition of BRG1 with PFI3 (20 μM, 24 h) under hypoxia significantly attenuated Ca^2+^ entry (Fig. 8C), confirming that BRG1-dependent chromatin remodeling is necessary for transcriptional control of genes mediating SOCE. Conversely, LINC00607 overexpression (LINC00607OE) under normoxic conditions increased Ca^2+^ influx compared with pcDNA controls (Fig. 8D), establishing LINC00607 as a positive regulator of endothelial calcium signaling even in the absence of hypoxia. These findings corroborate the transcriptional and protein-level evidence that LINC00607 promotes TSPAN18 expression and endothelial activation. Consistently, knockdown of TSPAN18 (siTSPAN18) significantly reduced SOCE amplitude under hypoxia (Fig. 8E), phenocopying the effect of LINC00607 silencing. This confirms that TSPAN18 acts as a downstream effector of LINC**00607-mediated calcium entry, linking the lncRNA-chromatin regulatory module to a key physiological endothelial signaling pathway. Interestingly, ERG knockdown (siERG) also decreased Ca^2+^ influx under hypoxia (Fig. 8E), even though TSPAN18 transcript levels were paradoxically increased (Fig. 7B). This suggests that ERG influences calcium homeostasis through additional downstream targets or chromatin contexts, and that TSPAN18 upregulation alone is insufficient to sustain full SOCE in the absence of ERG. Together, these data highlight the context-dependent role of ERG and reinforce the centrality of the LINC**00607-BRG1-TSPAN**18 axis in orchestrating hypoxia-induced endothelial calcium signaling.Figure 8LINC00607-TSPAN**18 axis modulates store-operated calcium entry (SOCE) in endothelial cells.**A, hypoxia (1% O_2, 24 h) and DMOG treatment (0.5 mM, 24 h under normoxia) enhanced Ca^2+^ influx compared to normoxia, indicating activation of SOCE under hypoxic signaling. B, knockdown of LINC00607 with two independent LNAs (LNA1 and LNA2) under hypoxia resulted in reduced Ca2+ influx relative to LNA control, confirming LINC006**07-dependent modulation of SOCE. C, inhibition of BRG1 with PFI3 (20 μM, 24 h) under hypoxia decreased Ca^2+^ entry compared to hypoxia alone, suggesting BRG1-dependent transcriptional control. D, overexpression of LINC00607 (LINC00607_OE) increased Ca^2+^ influx compared with pcDNA control, reinforcing its role as a positive regulator of SOCE. E, TSPAN18 knockdown (siTSPAN18) diminished SOCE amplitude, validating TSPAN18 as an effector of calcium entry under hypoxia. ERG knockdown (siERG) also reduced Ca^2+^ influx despite increased TSPAN18 expression, likely reflecting context-dependent ERG regulation of endothelial calcium homeostasis. All experiments were repeated independently at least three times, and representative kinetics are shown.

LINC00607-TSPAN18 axis drives hypoxia-induced endothelial–monocyte adhesion

To functionally validate the contribution of the LINC00607-TSPAN18 regulatory axis to endothelial activation, we assessed monocyte adhesion to endothelial monolayers under hypoxic and pharmacological inhibition conditions. Endothelial-monocyte adhesion represents an early and critical step in vascular inflammation and thrombosis (47), making it a robust functional readout of endothelial activation. Endothelial cells exposed to hypoxia (1% O_2_, 24 h) exhibited a marked increase in adherent monocytes compared with normoxic controls (Fig. 9, A and B). Similarly, treatment with DMOG (0.5 mM, 24 h under normoxia), which stabilizes HIF-1α, reproduced this effect, confirming that hypoxia-responsive signaling enhances endothelial adhesiveness. These findings establish a functional correlate to the transcriptional and chromatin activation of LINC00607 and TSPAN18 observed under hypoxia. Pharmacological inhibition of BRG1 with PFI3 (20 μM, 24 h) significantly reduced monocyte adhesion relative to hypoxia alone (Fig. 9, C and D). This indicates that BRG1-dependent chromatin remodeling is required to maintain the pro-adhesive phenotype. This observation aligns with our above findings suggesting that BRG1 is recruited by LINC00607 to activate TSPAN18 transcription. Knockdown of TSPAN18 (siTSPAN18) resulted in a pronounced reduction in monocyte adhesion (Fig. 9, E and F), establishing TSPAN18 as a key effector that promotes endothelial-monocyte interactions. In contrast, ERG knockdown led to a mild but non-significant increase in adhesion, further emphasizing that ERG acts as a contextual modulator rather than a direct determinant of the adhesive phenotype. Silencing LINC00607 using two independent LNA oligonucleotides (LNA1 and LNA2) under hypoxia markedly decreased monocyte adhesion compared to LNA control (Fig. 9, G and H). Conversely, LINC00607 overexpression (LINC00607OE) under normoxic conditions significantly enhanced monocyte adhesion relative to pcDNA control (Fig. 9, I and J). Together, these gain- and loss-of-function experiments demonstrate that LINC00607 acts upstream of TSPAN18 and BRG1 to promote hypoxia-induced endothelial activation.Figure 9**LINC00607–TSPAN18 axis promotes endothelial–monocyte adhesion under hypoxia.**A and B, representative fluorescence microscopy images and quantification of monocyte adhesion to EA.hy926 endothelial cells under normoxia, hypoxia (1% O_2, 24 h), and DMOG treatment (0.5 mM, 24 h, normoxia). Both hypoxia and DMOG significantly increased monocyte adhesion compared to normoxia (ANOVA, p = 6.2 × 10^-9^), indicating hypoxia-induced endothelial activation. C and D, pharmacological inhibition of BRG1 using PFI3 (20 μM, 24 h) during hypoxia reduced monocyte adhesion relative to hypoxia alone (∗∗∗p < 0.0001), suggesting BRG1-dependent regulation of the adhesive phenotype. E and F, knockdown of ERG (siERG) resulted in a non-significant increase in monocyte adhesion, whereas TSPAN18 knockdown (siTSPAN18) markedly reduced adhesion (ANOVA, p = 1.1 × 10^-5^), establishing TSPAN18 as a key effector mediating endothelial–monocyte interactions. G and H, knockdown of LINC00607 with two independent LNAs (LNA1 and LNA2) under hypoxia significantly decreased monocyte adhesion compared to LNA control (ANOVA, p = 3.1 × 10^-7^), consistent with the loss of TSPAN18 expression and downstream endothelial activation. I and J, overexpression of LINC00607 (LINC00607_OE) enhanced monocyte adhesion compared with pcDNA control (∗∗∗p < 0.001), confirming its role as an upstream regulator promoting pro-adhesive endothelial property. Statistical significance was determined using one-way ANOVA followed by Tukey’s post hoc test: p < 0.05 (∗), p < 0.01 (∗∗), p < 0.001 (∗∗∗), p < 0.0001 (∗∗∗∗).

Discussion

Hypoxia is a major pathophysiological driver in the onset and progression of cardiovascular diseases, exerting profound effects on vascular homeostasis through transcriptional and epigenetic reprogramming. Emerging evidence indicates that hypoxia influences endothelial gene expression through multiple epigenetic mechanisms, including DNA methylation, histone and non-histone protein modifications, and non-coding RNA-mediated regulation (48, 49). Among these, long non-coding RNAs have gained prominence as key regulators of chromatin dynamics and transcriptional output, fine-tuning endothelial responses that govern angiogenesis, inflammation, and thrombosis (50, 51, 52).

In this study, we identified a hypoxia-induced endothelial regulatory axis involving the long non-coding RNA LINC00607 and the prothrombotic membrane protein Tetraspanin18 (TSPAN18). Integrative transcriptomic and chromatin accessibility analyses revealed that LINC00607 epigenetically regulates TSPAN18 under hypoxic conditions, establishing a mechanistic link between non-coding RNA driven chromatin remodeling and Ca^2+^-dependent thromboinflammatory signaling in endothelial cells. Co-expression network analysis and knockout transcriptomics further demonstrated that LINC00607 and TSPAN18 are co-regulated within a hypoxia-responsive gene module, with overlapping differentially expressed genes highlighting the central role of LINC00607 in hypoxia-induced transcriptional reprogramming. Notably, loss of LINC00607 reduced both chromatin accessibility and expression of TSPAN18, whereas LINC00607 overexpression led to a marked increase in TSPAN18 transcript and protein levels, confirming a direct and positive regulatory relationship. Together, these findings define a LINC00607–TSPAN18 epigenetic circuit that mediates endothelial activation in response to hypoxia through a coordinated transcriptional and chromatin-level control.

LINC00607 has been reported to play roles in several cancer types (53, 54, 55) and shows endothelial-enriched expression, as reflected in GTEx datasets (27, 56). It has been characterized as a super-enhancer derived lncRNA involved in vascular gene regulation through chromatin interactions and modulation of chromatin modifier activity, particularly in angiogenic pathways (27, 56, 57). Moreover, a single-nucleotide polymorphism (rs78529201) within the LINC00607 locus has been associated with an increased risk of peripheral arterial disease, further supporting its relevance to vascular homeostasis and cardiovascular pathology (58). Previous studies have identified LINC00607 as a hypoxia-responsive lncRNA regulated by HIF1α (27). Consistent with these studies, both LINC00607 and TSPAN18 were upregulated under hypoxia and following DMOG-mediated HIF1α stabilization. Importantly, knockdown of LINC00607 under hypoxia significantly reduced TSPAN18 expression, indicating that TSPAN18 upregulation is dependent on LINC00607 rather than a direct HIF1α effect. Analysis of HIF1α ChIP-seq datasets show no evidence of direct HIF1α binding within the LINC00607 locus or its enhancer region, suggesting that HIF1α likely regulates LINC00607 indirectly, possibly through secondary transcriptional mediators activated under hypoxic stress.

BRG1, the catalytic subunit of the SWI/SNF chromatin-remodeling complex, plays a central role in regulating enhancer accessibility through interactions with transcription factors and non-coding RNAs (59). Several lncRNAs, including MALAT1, Evf2, and Mhrt, have been shown to directly bind BRG1 and modulate its recruitment to specific genomic loci, thereby altering enhancer activity and gene transcription (60, 61, 62). Broader reviews also highlight SWI/SNF-lncRNA interactions as a common mechanism for directing chromatin remodeling toward cell type-specific regulatory elements (59, 63). In line with these observations, recent endothelial studies identified LINC00607 as a trans-acting lncRNA capable of guiding BRG1 to super-enhancers, thereby influencing ERG-dependent gene programs (27, 41). Recent structural study further demonstrated that the AT-hook domain of BRG1 binds enhancer RNAs (eRNAs) and is required for proper BRG1 enhancer engagement, supporting a model in which RNA molecules act as guides to position SWI/SNF complexes at lineage-defining enhancers (64).

Our findings further support a role for BRG1 in regulating the LINC00607-TSPAN18 axis under hypoxia. Analysis of BRG1 CUT&RUN and PROTAC-mediated degradation datasets showed a clear loss of BRG1 occupancy and chromatin accessibility at the TSPAN18 enhancer, consistent with BRG1-dependent activation of this locus. Combined with our RIP-qPCR validation demonstrating possible LINC**00607-BRG1 interaction in endothelial cells, these results suggest that LINC00607 may facilitate BRG1 recruitment to the TSPAN18 enhancer during hypoxic stimulation. This aligns with the broader pattern in which lncRNAs serve as scaffold or guide molecules to position SWI/SNF complexes at cell-type-specific enhancers. However, we emphasize that our study does not claim to fully resolve the molecular mechanism of how BRG1 is targeted to TSPAN18. While the available data strongly indicate a BRG1-dependent mechanism downstream of LINC00607, definitive mapping of BRG1’s locus-specific recruitment or its interaction dynamics with ETS factors will require future investigation. Nonetheless, the convergence of BRG1 datasets, hypoxia-induced enhancer activation, and functional loss-of-function experiments provides a coherent outline in which BRG1 acts as a key mediator linking LINC00607 to TSPAN18 transcriptional activation under hypoxic stress.

Endothelial enhancer networks are dynamically regulated by transcription factors of the ETS family, including ERG and FLI1, which act cooperatively to maintain vascular homeostasis and transcriptional identity (42, 43, 44). Consistent with these studies, our findings confirm that ERG modulates enhancer activity at both the LINC00607 and TSPAN18 loci under hypoxia, with distinct effects on histone acetylation and transcriptional output. The hypoxia-induced enrichment of H3K27ac and H3K4me1 marks suggests that both loci acquire an active enhancer configuration, supporting transcriptional activation of endothelial genes involved in stress response. The presence of ERG peaks within the TSPAN18 locus and decreased acetylation following ERG knockdown confirm ERG-dependent enhancer activity. However, the paradoxical upregulation of LINC00607 and TSPAN18 following ERG depletion indicates complex regulatory feedback. This upregulation may reflect either de-repression of LINC00607 upon ERG loss or compensatory activation by FLI1, which shares overlapping genomic binding sites. The observation that dual ERG and FLI1 depletion reduces enhancer acetylation at both loci further supports their cooperative and redundant roles in maintaining endothelial enhancer integrity. Mechanistically, our data suggest that LINC**00607-BRG1-mediated enhancer regulation functions as an active mechanism under hypoxia, driving TSPAN18 transcription independently of ERG. While ERG maintains basal enhancer acetylation under normoxia, hypoxia induces LINC00607 upregulation, which recruits BRG1 to remodel chromatin and activate TSPAN18 expression. Interestingly, ERG silencing under hypoxia further amplifies LINC00607 and TSPAN18 expression, implying that ERG may normally exert a restraining influence on this pathway. Thus, loss of ERG amplifies the LINC**00607-BRG1-driven activation loop, highlighting a context-dependent regulatory balance between ETS factors and non-coding RNA-chromatin interactions during hypoxic endothelial activation.

Tetraspanins are membrane-organizing proteins that coordinate cell adhesion, signaling, and trafficking by forming specialized surface microdomains (64). Several tetraspanins, including CD9, TSPAN1, and CD82, have been reported to respond to hypoxic stress in various cell types (65, 66, 67). In the present study, we identify TSPAN18 as a hypoxia-responsive, endothelial-enriched tetraspanin that is epigenetically upregulated under hypoxic conditions. Functionally, TSPAN18 serves as a key mediator of store-operated calcium entry (SOCE) by stabilizing STIM1-Orai1 interactions, thereby enhancing calcium influx and facilitating von Willebrand factor (vWF) release (45, 46). Consistent with this role, our results show increased SOCE activity under hypoxia and DMOG treatment, which was significantly reduced upon LINC00607 or TSPAN18 knockdown, this establish the LINC00607-TSPAN18 axis as a positive regulator of hypoxia-induced calcium influx. In endothelial biology, calcium influx acts as a central signal that promotes endothelial activation and leukocyte adhesion, critical events in vascular inflammation and thrombosis (68, 69, 70). Previous studies have demonstrated that hypoxia enhances monocyte-endothelial adhesion (71, 72, 73) and that these interactions initiate the expression of adhesion molecules, cytokines, and procoagulant factors that contribute to atherogenesis and thrombosis (74, 75). In our study, both hypoxia and DMOG treatment increased monocyte adhesion, whereas inhibition of BRG1 or silencing of LINC00607 or TSPAN18 significantly reduced this effect, establishing their role in promoting endothelial-monocyte interactions. Overexpression of LINC00607, conversely, enhanced adhesion, further reinforcing its function as an upstream epigenetic regulator of this proinflammatory phenotype. Together, these findings link LINC**00607-driven TSPAN18 activation to enhanced Ca^2+^ signaling and increased endothelial adhesiveness under hypoxia. This identifies the LINC00607–TSPAN18–SOCE axis as a novel epigenetic circuit that couples hypoxic signaling to endothelial activation and thromboinflammation, providing mechanistic insight into how non-coding RNAs regulate vascular dysfunction under low-oxygen stress.

In summary, our study identifies a hypoxia-responsive regulatory cascade in endothelial cells in which LINC00607 functions as an upstream modulator that links epigenetic remodeling to vascular activation. Under hypoxic conditions, LINC00607 expression is induced in a HIF-dependent manner and facilitates recruitment of the chromatin remodeler BRG1 to enhancer regions, including those associated with TSPAN18. This interaction promotes enhancer acetylation and transcriptional activation of TSPAN18, leading to increased SOCE and enhanced endothelial-monocyte adhesion, two hallmarks of a pro-thromboinflammatory state. Although ERG contributes to maintaining enhancer activity, its depletion does not disrupt the LINC00607-TSPAN18 regulatory link, indicating that LINC**00607-BRG1 constitutes the core hypoxia-driven regulatory axis. Together, these findings suggest that LINC00607 integrates hypoxic signaling with chromatin remodeling to fine-tune endothelial responses relevant to vascular inflammation and thrombosis.

While our study identifies TSPAN18 as a robust and functionally validated downstream target of LINC00607 under hypoxia, we acknowledge that additional BRG1-dependent endothelial genes may also be regulated by LINC00607. Mapping the full scope of these targets will require further investigation. Complementary in vivo will be important in future to confirm physiological relevance. Additionally, although we observe compensatory regulation involving ERG and FLI1, the broader ETS network and its integration with lncRNA-chromatin interactions under hypoxia requires deeper exploration. Despite these limitations, the combined transcriptomic, epigenetic, and functional datasets presented here provide a strong foundation for defining the LINC00607-TSPAN18 axis as a key component of hypoxia-induced endothelial activation.

Experimental procedures

Cell culture

EA.hy926 endothelial cells were a kind gift from Dr Jagavelu Kumaravelu (CSIR–Central Drug Research Institute, Lucknow, India). Cells were cultured in 75 cm^2^ flasks in Dulbecco’s Modified Eagle Medium (DMEM; Gibco; 11,965,092) supplemented with 10% fetal bovine serum (FBS; Gibco; A5256701) and 1% penicillin–streptomycin (MCE; HY-K1006). Cultures were maintained at 37 °C in a humidified incubator (Forma Steri-Cycle CO_2_ Incubator, Thermo Fisher Scientific) with 5% CO_2_. For passaging, cells were detached using TrypLE reagent (Thermo Fisher Scientific). For hypoxia treatment, cells were exposed to 1% O_2_ for 24 h in a hypoxic cell culture incubator (Forma Steri-Cycle i160 CO_2_ Incubator, Thermo Fisher Scientific).

THP-1 human monocytic cells (NCCS, Pune, India) were maintained in RPMI-1640 medium (Gibco; 11875093) supplemented with 10% FBS and 1% penicillin–streptomycin under identical culture conditions. All other culture conditions were similar.

Experimental design and treatment groups

To investigate the regulatory mechanisms involving LINC00607, and TSPAN18 in endothelial cells, EA.hy926 cells were subjected to a series of defined experimental conditions. Cells maintained under standard culture conditions (21% O_2_) served as the normoxic control, while hypoxic signaling was induced by exposing cells to 1% O_2_ for 24 h. To mimic hypoxia-mediated HIF1α stabilization under normoxia, cells were treated with 0.5 mM dimethyloxalylglycine (DMOG; MCE; HY-15893) for 24 h. The functional role of BRG1 during hypoxic responses was evaluated by treating cells with 20 μM PFI-3 (MCE; HY-12409), a selective bromodomain inhibitor of BRG1/BRM. To assess the contribution of LINC00607, cells were transfected under normoxic conditions with a pcDNA3.1(+)-LINC00607 expression construct, while the empty vector served as a negative control. Loss-of-function studies were performed under hypoxia using two distinct LNA gapmers (Qiagen) targeting LINC00607 (LNA1 and LNA2) or a non-targeting LNA control (LNA CTRL) at a final concentration of 15 nM. In parallel, siRNA-mediated silencing of ERG (siERG) and TSPAN18 (siTSPAN18) or a non-targeting siRNA control (siCTRL) was carried out under hypoxic conditions at a final concentration of 10 nM. Additionally, co-transfection experiments with siERG and LNA2 were performed under hypoxic conditions to investigate potential cooperative regulatory effects, with siCTRL and LNA CTRL serving as the corresponding controls. Cells from each treatment group were subsequently processed for downstream analyses.

RNA isolation and quantitative real-time PCR

Total RNA was isolated from cells using the GeneAll Hybrid-R RNA extraction kit according to the manufacturer’s instructions. One microgram of total RNA was reverse transcribed into cDNA using the PrimeScript first Strand cDNA Synthesis Kit (Takara Bio; 6110A). Quantitative real-time PCR (qRT-PCR) was performed using the Luna Universal qPCR Master Mix (New England Biolabs; M3003 L) on a QuantStudio six Flex Real-Time PCR System (Applied Biosystems) in a 10 μl reaction volume. Ct values for each target gene were normalized to 18S rRNA as an internal control, and relative expression levels were calculated using the ΔΔCt method. Primer sequences used for qRT-PCR are listed in Table S1.

RNA immunoprecipitation (RIP)

RNA immunoprecipitation (RIP) was performed as previously described by Martindale et al. (2020) (76), with minor modifications. Briefly, EA.hy926 endothelial cells were cultured to confluency in two 100-mm dishes. Protein A magnetic beads (MCE; HY-K0203) were pre-coated overnight at 4 °C with 5 μg of either anti-BRG1 antibody (Cell Signaling Technology; 49360T) or normal rabbit IgG (Cell Signaling Technology; 2729S) on a rotator. Cells were washed twice with ice-cold PBS, scraped into PBS, and pelleted by centrifugation at 1000g. Pellets were lysed in polysome extraction buffer supplemented with RNase inhibitor (Biohelix; RI001–0125) and protease inhibitor cocktail (MCE; HY-K0010) and incubated on ice for 15 min. Lysates were clarified by centrifugation at 10,000g for 15 min at 4 °C, and the resulting RNP-containing supernatants were pre-cleared with 10 μg IgG on a rotator at 4 °C. Total protein concentration was quantified using a BCA protein assay kit (G-Biosciences; 786–570). For each RIP reaction, 1000 μg of protein lysate was used with either anti-BRG1 antibody or IgG as a negative control. Immunoprecipitation was performed following the protocol described by Martindale et al. (2020). Fifty microlitre of pre-coated Protein A magnetic beads-antibody complex was aliquoted into microcentrifuge tubes. A master mix was prepared for each reaction consisting of 300 μl NT2 buffer (50 mM Tris HCl, pH 7.5, 150 mM NaCl, 1 mM MgCl2, 0.05% IGEPAL CA-630), 10 μl of 0.1 M DTT (added to the buffer and not directly to the beads to avoid antibody reduction), 10 μl RNase inhibitor, and 33 μl of 0.5 M EDTA. To this mixture, 1000 μg of RNP lysate was added, and the total reaction volume was adjusted to 1 ml with NT2 buffer. The reactions were incubated for 1.5 h at 4 °C with gentle rotation to allow formation of RNA-protein-antibody complexes. Following incubation, immune complexes were captured using a magnetic stand, and the supernatant was carefully removed. Following immunoprecipitation, the bead-protein-RNA complexes were washed four times with 1 ml ice-cold NT2 buffer. For each wash, beads were gently resuspended and collected using a magnetic stand, and the supernatant was carefully discarded. After the final wash, beads were gently resuspended in 100 μl NT2 buffer containing 10 μl DNase I (Promega; M6101) and incubated at 37 °C for 10 min to eliminate contaminating DNA. At this step, care was taken not to resuspend the beads harshly; instead, the solution was gently dispensed and incubated with minimal agitation. Shaking was strictly avoided to prevent disruption of the RNA-protein complexes. Following DNase I treatment, 1 ml NT2 buffer was added, beads were captured using a magnetic stand, and the supernatant was discarded. Protein digestion was then performed by adding 100 μl of a freshly prepared master mix containing 100 μl NT2 buffer, 2.5 μl Proteinase K (Genei; 20 mg/ml), and 1 μl 10% SDS (Merck). Samples were incubated at 55 °C for 20 min with gentle mixing to release RNA from the complexes. Beads were subsequently immobilized on a magnetic stand, and the supernatant (∼100 μl) was transferred to a fresh tube. To ensure maximal RNA recovery, beads were washed once with 200 μl NT2 buffer, and the supernatant was collected and combined with the initial eluate. Beads were then discarded. For RNA extraction, 700 μl TRIzol reagent was added to the combined supernatants, followed by vigorous vortexing for 1 min at room temperature. Immunoprecipitation was then carried out following the protocol of Martindale et al. (2020). RNA was extracted from the immunoprecipitates using the GeneAll Hybrid-R RNA isolation kit, and cDNA was synthesized using the Takara First Strand cDNA Synthesis Kit. Relative transcript abundance of LINC00607 was quantified by RT-qPCR using the CFX Opus 96 Dx Real-Time PCR System (Bio-Rad).

Western blot

Whole-cell lysates were prepared in ice-cold RIPA buffer (Sigma-Aldrich; R0278) supplemented with protease and phosphatase inhibitor cocktails (MCE; HY-K0010; HY-K0021). Lysates were clarified by centrifugation (14,000g, 15 min, 4 °C), and protein concentration was determined using the BCA Protein Assay Kit (G-Biosciences; 786–570). Equal amounts of protein (20–40 μg) were mixed with 2 × Laemmli buffer (Biorad; 1610737) containing 50 mM β-merceptoethanol, denatured at 95 °C for 5 min, resolved on 12% SDS–PAGE gels, and transferred onto PVDF membranes (Merck; IPVH00010)) by wet transfer (100 V, 90 min, 4 °C). Membranes were blocked for 1 h at room temperature with 5% (w/v) skimmed milk prepared in PBST (PBS containing 0.1% Tween-20) and incubated overnight at 4 °C with primary antibodies: anti-TSPAN18 (1:1000; Abclonal) or anti-β-actin (1:1500; Affinity Biosciences). After three washes (5 min each) with 0.1% PBST, membranes were incubated with HRP-conjugated anti-rabbit secondary antibody (1:3000; Genei) for 1 h at room temperature. The specificity of the anti-TSPAN18 antibody was validated by siRNA-mediated knockdown of TSPAN18, which resulted in a marked reduction of the corresponding protein band compared to siRNA control. Blots were washed again three times in 0.1% PBST and developed using enhanced chemiluminescence (ECL; MCE; HY-K1005) detection reagent. Protein bands were visualized using a CCD-based imaging system (Biorad), and band intensities were quantified with ImageJ (NIH). TSPAN18 expression levels were normalized to β-actin as a loading control. All experiments were performed in at least three independent biological replicates.

Calcium assay

The cytosolic Ca^2+^ influx was quantified with Fluo-4 AM (Invitrogen; F14201) according to the manufacturer’s instructions. The Ea.hy926 endothelial cells were treated as per the groups defined above. The cells were washed twice with Ca^2+^/Mg^2+^ free Hank's Balanced Salt Solution (HBSS). The cells were then loaded with 2 μM Fluo-4 AM with 0.02% Pluronic F-127 (Sigma; P2443) for 45 min and washed 3 times with Ca^2+^/Mg^2+^ free HBSS (137 mM NaCl, 5.4 mM KCl, 0.34 mM Na_2_HPO_4_, 0.44 mM KH_2_PO_4_, and 5.5 mM D-glucose). The cells were then incubated in Ca^2+^/Mg^2+^ free HBSS for 10 min to allow complete de-esterification of the dye. To ensure the SOCE-meditated Ca^2+^ entry, the cells were then treated with 1 μM Thapsigargin (Tg; Merck; 586005) and incubated at 37 °C for 15 min. After depleting intracellular Ca^2+^, the Ca^2+^ influx was measured by adding 2 mM extracellular Ca^2+^. The kinetics were recorded using an Agilent BioTek Synergy HTX plate reader at excitation 485/20 and emission 528/20.

Immunocytochemistry

The expression and subcellular localization of TSPAN18 were examined by immunocytochemical staining. EA.hy926 endothelial cells were seeded on sterile coverslips placed in 12-well plates and treated according to the experimental groups described above. Following treatment, cells were washed three times with sterile PBST (PBS containing 0.1% Tween-20) and fixed with 4% paraformaldehyde (Sigma-Aldrich; 158127) for 10 min at room temperature (RT). After two washes with ice-cold PBS, cells were permeabilized with 0.1% PBST for 5 min and washed three times with PBS. Non-specific binding was blocked by incubating cells with 3% bovine serum albumin (BSA; Himedia, Cat. No. MB083) for 30 min at RT. Cells were then incubated overnight at 4 °C with rabbit anti-TSPAN18 (Invitrogen, PA5-48957) primary antibody (1:200 dilution). After two washes with PBS (5 min each), cells were incubated with Alexa Fluor 488–conjugated anti-rabbit secondary antibody (Invitrogen, Cat. No. A11008) for 1.5 h at RT in the dark. Following three final PBS washes, coverslips were mounted on glass slides using a DAPI-containing mounting medium (Invitrogen, Cat. No. 00-4959-52). Fluorescence images were captured using a fluorescence microscope (Cilika). Corrected total cell fluorescence (CTCF) was quantified using ImageJ software (NIH) to compare expression levels across different treatment groups.

Adhesion assay

The Ea.hy926 endothelial cells were treated as per the groups defined above and grown up to 90% confluence for endothelial monocyte adhesion assay and then activated by 10 μM histamine (MCE; HY-B1204) for 5 min. THP-1 cells were labeled with 0.2 ug/ml calcein-AM (Cayman; 14948) for 45 min at 37 °C and then added onto endothelial cell monolayer in an equal volume. After 90 min co-cultured cells were washed with 1 × PBS containing 1% bovine serum albumin (BSA), and then imaging was done using ZOE Fluorescent Cell Imager (Biorad). The relative adhesion quantification was done by measuring the relative fluorescence using ImageJ software.

Transfection (knockdown)

Ea.hy926 endothelial cells were transfected for post-transcriptional gene silencing using Xfect RNA Transfection Reagent (Takara; 631450) following the manufacturer’s instructions. The LNA GapmeRs (Qiagen) were used against LINC00607 and a non-targeting control (NTC) GapmeR was used at a final conc. of 15 nM. The oligonucleotide sequences for LINC00607 LNA gapmers were: LNA1 5′-ATAGGTCACGCATTCT-3′; LNA2 5′-AGGATTGGATAGGTCA-3′ and the NTC (LNA CTRL), 5′-AACACGTCTATACGC-3′. For targeting the post-transcriptional silencing of protein coding genes (siTSPAN18 and siERG) Flexitube siRNAs (Qiagen) and Allstars negative control (siCTRL) were used at final conc. of 10 nM. The oligonucleotide sequences for were siTSPAN18 5′-CACGGTGATCCTCAACACCTT-3′; and siERG 5′-CAGATCCTACGCTATGGAGTA-3′.

Plasmid overexpression

The cDNA sequence of LINC00607 was synthesised with flanking KpnI (5′ end) and XhoI (3′ end) restriction sites commercially from Azenta Life Sciences (USA) and subcloned into mammalian expression vector pcDNA3.1(+). The integrity and cloning efficiency were then verified by sanger sequencing and followed by single and double restriction digestion and run on agarose gel to confirm the insert size. To assess the regulatory effect of LINC00607 on TSPAN18 expression, EA.hy926 endothelial cells were transiently transfected with the pcDNA3.1(+) vector containing the full-length LINC00607 insert. For each well, 250 ng of the LINC00607 plasmid DNA was diluted in OptiMEM (Mix A) and incubated for 10 min at room temperature. Separately, 750 ng of polyethylenimine (PEI) (3:1 PEI:DNA ratio) was also diluted in Opti-MEM (Invitrogen; 31985070) (Mix B) and incubated for 10 min. Mix A and Mix B were then combined and incubated at room temperature for 20 min to allow complex formation. Empty pcDNA3.1(+) was used as a control. EA.hy926 cells were seeded to reach 70 to 80% confluency on the day of transfection. Cells were washed with PBS, the medium was replaced with fresh complete growth medium, and the transfection mix was added dropwise to the cells. Plates were incubated at 37 °C in a CO_2_ incubator for 24 h. Following incubation, total RNA was extracted from both transfected and control cells, and RT-PCR was performed to confirm the overexpression of LINC00607.

Subcellular fractionation

To determine the subcellular localization of LINC00607 and evaluate its potential chromatin association, subcellular fractionation was performed in EA.hy926 endothelial cells exposed to hypoxia (1% O_2_) for 24 h. Upon reaching confluency in T-75 flasks, cells were washed twice with ice-cold PBS (MCE), detached using TrypLE reagent (Thermo), and pelleted by centrifugation. The pellet was resuspended in cell lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.15% Igepal CA-630) supplemented with RNase and protease inhibitors. The lysate was layered onto a cold sucrose cushion (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 24% sucrose, 1 mM EDTA) and centrifuged at 3500g for 10 min to separate the cytoplasmic (supernatant) and nuclear (pellet) fractions. The cytoplasmic fraction was clarified by centrifugation at 14,000g for 1 min. The nuclear pellet was washed with PBS-EDTA, resuspended in glycerol buffer (20 mM Tris-HCl pH 7.4, 75 mM NaCl, 0.5 mM EDTA, 50% glycerol), and mixed with an equal volume of urea-based nuclear lysis buffer (10 mM Tris-HCl pH 7.4, 1 M urea, 0.3 M NaCl, 7.5 mM MgCl_2_, 0.2 mM EDTA, 1% Igepal CA-630). After vortexing and incubation on ice, samples were centrifuged at 13,000g for 2 min to separate the nucleoplasmic (supernatant) and chromatin-bound (pellet) fractions. RNA was isolated from all three fractions using the GeneAll RNA Isolation Kit. The chromatin pellet was homogenized in TRIzol with a 21-gauge needle prior to RNA extraction. Complementary DNA (cDNA) synthesis was performed from each fraction, The relative transcript level quantification of LINC00607 was done CFX Opus 96 Dx RT-PCR System. The nuclear retained lncRNA MALAT1 was used as positive control.

Transcriptomic data analysis

Raw RNA-seq reads were subjected to quality control and adapter trimming using fastp (77). High-quality reads were then aligned to the human reference genome (GRCh38.p14) using HISAT2 (78), and the resulting alignment files were sorted and processed with featureCounts (79) to obtain gene-level read counts. Differential gene expression analysis was performed using the DESeq2 R package (80). Genes with an absolute log_2_ fold change (|log_2_FC|) > 1 and an adjusted p-value < 0.05 were considered significantly differentially expressed. The same thresholds were applied to identify differentially expressed long non-coding RNAs (DElncRNAs).

Volcano plots highlighting the top 10 most significant differentially expressed genes and lncRNAs were created with ggplot2 (https://ggplot2.tidyverse.org).

Gene Ontology and pathway enrichment analysis

Gene Ontology (GO) enrichment analysis for Biological Process and Cellular Component categories was performed using the ClusterProfiler (81) and enrichplot (https://bioconductor.org/packages/enrichplot) R packages. KEGG pathway enrichment analysis was also conducted with ClusterProfiler to identify significantly enriched molecular pathways associated with differentially expressed genes. Additionally, Disease Ontology Gene Set Enrichment Analysis (GSEA) of hypoxia-associated differentially expressed genes was carried out using the gseDO function from the DOSE (82) package to explore disease-related functional associations.

lncRNA–mRNA Co-expression analysis

Pearson correlation analysis was performed in R to assess the co-expression relationships between the top 10 most significant DElncRNAs (as identified in the volcano plot, Fig. 1B and the top 50 differentially expressed genes (DEGs) among the 114 hypoxia-responsive genes associated with hallmark genes (Table S2). Correlations with a coefficient greater than 0.85 were considered significant. The analysis was carried out using the dplyr (https://github.com/tidyverse/dplyr), stringr (83), and tidyverse (84) packages.

The resulting correlation matrix was visualized as a heatmap using the gplots package, and a co-expression network was constructed from this matrix to illustrate the co-expressed modules, which were plotted using the igraph (85) package.

iCLIP analysis

Publicly available BRG1 iCLIP datasets were downloaded from the NCBI Sequence Read Archive (SRA) using the SRA Toolkit (https://trace.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?view=software). Adapter trimming was performed using Cutadapt with the standard 3′ iCLIP adapter sequence (AGATCGGAAGAGCGGTTCAGCAGGAATGCCGAG). Quality-controlled reads were then aligned to the human reference genome (GRCh38/hg38) using Bowtie2 (86). The resulted SAM files from Bowtie2 were converted to binary BAM format, sorted, and indexed with SAMtools (87). To enable visualization of BRG1 binding profiles, normalized BigWig coverage tracks were generated from the aligned BAM files using deepTools (88). Crosslink-induced truncation events were identified at single-nucleotide resolution using PureCLIP (89), and annotated gene models from GENCODE v48 were used to restrict the analysis to lncRNA biotypes (transcript_type "lncRNA"). Crosslink positions were intersected with lncRNA annotations using BEDTools (90), and the number of crosslink events per lncRNA was quantified. The top 50 BRG1-bound lncRNAs were ranked based on crosslink density and visualized in R using the ggplot2 package.

GRO-seq analysis

Raw sequencing data (FASTQ format) were demultiplexed and subjected to initial quality assessment using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Adapter trimming and quality filtering were performed with fastp (77) using a minimum read length of 15 bp and base quality threshold of Q20.

Trimmed reads were aligned to the human reference genome (GRCh38/hg38) using Bowtie2 (86) with the very-sensitive preset. Alignments with a mapping quality score below 30 were discarded using samtools (87), and PCR duplicates were removed with Picard MarkDuplicates (https://broadinstitute.github.io/picard/). Resulting filtered BAM files were indexed with samtools (87) and used for downstream analyses. Because GRO-seq reads represent nascent RNA transcription and are inherently strand-specific, strand-separated coverage profiles were generated from the filtered BAM files using deepTools bamCoverage (88). Coverage files were normalized to counts per million mapped reads (CPM) with a bin size of 10 bp. Forward (plus) and reverse (minus) strand signals were generated separately using the --filterRNAstrand forward and --filterRNAstrand reverse options, respectively. This approach ensured that nascent transcription from each strand was captured independently, preserving strand polarity of the transcriptional landscape.

Strand-specific BigWig files were visualized using the Integrative Genomics Viewer (IGV genome browser). Plus- and minus-strand tracks were overlaid for each replicate to assess reproducibility and visualize strand-specific nascent transcription dynamics under normoxic and hypoxic conditions.

ChIP-seq analysis

Raw sequencing reads were quality-filtered and adapter-trimmed using fastp (77). High-quality reads were aligned to the human reference genome (GRCh38) with Bowtie2 (86). PCR duplicates and low-quality alignments were removed using SAMtools (87). Genome-normalized coverage tracks (BigWig files) were generated from filtered BAM files with deepTools bamCoverage (88) using --normalizeUsing RPGC and a bin size of 10 bp. Coverage tracks were visualized using IGV and used for downstream metagene analysis. Peak calling was performed using MACS2 (91) with input DNA as a control wherever applicable. Narrow peaks were called for transcription factors, RNA polymerase II, and promoter-associated histone marks such as H3K4me3 (--nomodel --extsize 200), while broad peaks were called for enhancer-associated histone marks such as H3K27ac and H3K4me1 (--broad --broad-cutoff 0.1). During polII ChIP-Seq analysis for promoter-proximal occupancy and transcriptional dynamics analysis, normalized read density was computed within defined windows around transcription start sites (TSS) and gene bodies using deepTools computeMatrix (88). Average Pol II occupancy profiles ±2 kb around TSS were visualized using plotProfile, and the pausing index was calculated as the ratio of Pol II density at the promoter (TSS − 50 bp to + 300 bp) to that across the gene body. Representative metagene plots were generated for key loci, such as TSPAN18, illustrating differential Pol II loading and elongation under hypoxic versus normoxic conditions.

ATAC-seq analysis

Raw ATAC-seq reads in FASTQ format were processed using a standardized bioinformatics workflow. Adapter sequences and low-quality bases were trimmed with fastp (77) using default settings and a minimum read length threshold of 30 bp. Quality-controlled reads were then aligned to the human reference genome (GRCh38/hg38) using Bowtie2 (86) with the --very-sensitive preset and a maximum fragment size of 2000 bp. Resulting SAM files were converted to BAM format, sorted, and indexed with SAMtools (87).

Duplicate reads and mitochondrial alignments were removed to minimize technical artifacts. Only properly paired, uniquely mapped reads with a mapping quality score ≥30 were retained. Genome-wide chromatin accessibility profiles were generated by converting filtered BAM files into coverage tracks using deepTools bamCoverage (88). Reads were extended to the estimated fragment length, and per-base coverage was normalized as reads per genomic content (RPGC) with an effective genome size of 2,913,022,398 bp (human genome, GRCh38). Signal was calculated in 10-bp bins, and blacklisted genomic regions (ENCODE hg38 blacklist) were excluded to avoid artefactual peaks. Final normalized signal tracks were exported in BigWig format for visualization in IGV Genome Browser. The resulting BigWig files were used for comparative visualization of chromatin accessibility across experimental conditions and integration with transcriptomic and epigenomic datasets.

CUT&RUN analysis

Raw paired end CUT&RUN sequencing data were downloaded from the NCBI Sequence Read Archive (SRA). Adapter trimming and quality filtering were performed with fastp (77). Cleaned reads were aligned to the human reference genome (GRCh38/hg38) using Bowtie2 (86) with the --very-sensitive option. Post-alignment processing, including sorting, filtering for properly paired reads with mapping quality ≥30, mitochondrial read removal, and duplicate removal, was carried out with samtools (87). Genome-wide signal tracks were generated using deepTools (88) bamCoverage with RPGC normalization (effective genome size: 2.91 × 10^9^ bp) to produce high-resolution BigWig files for visualization.

Publicly available datasets

Various sequencing datasets used in this study were obtained from NCBI-GEO (92): RNA-seq datasets - HUVECs exposed to normoxia/hypoxia GSE70330 (36), GSE76743 (37), HUVECs with LINC00607 Knock-out GSE199877 (27), HUVECs with LINC00607 Knock-down GSE197954 (56), Acriflavine treated HUVECs exposed to hypoxia GSE186297 (93). ATAC-seq datasets – HUVECs exposed to normoxia/hypoxia GSE145774 (94), HUVECs with LINC00607 Knock-out GSE199876 (27), HUVECs with PROTACs targeting BRG1 GSE262057 (41). Chip-seq data – HUVECs exposed to hypoxia/normoxia H3K27Ac GSE38555 (95), HUVECs with ERG transcription factor knockdown H3K27Ac GSE124891 (42), ERG transcription factor in HUVECs GSE128382 (96), HIF1α transcription factor in HUVECs GSE89836, Pol II binding in HUVEC exposed to hypoxia/normoxia GSE50144 (95), H3K4me1 histone marks in HUVECs exposed to hypoxia/normoxia GSE39089 (97), ERG and FLI1 transcription factors in HUVECs GSE109695 (43), HUVECs with ERG and FLI1 transcription factors knockdown H3K27Ac GSE109695 (43). Cut & Run Seq datasets – BRG1 in HUVECs with BRG1 KO GSE201824 (27), BRG1 in HUVECs with siLINC00607 GSE262061 (41), BRG1 in HUVECs with PROTACs targeting BRG1 GSE262059 (41), BRG1, H3K4me1, H3K4me3, H3K27Ac in PANC1 cells under BRG1 overexpression GSE266350 (98). iCLIP dataset – HUVEC BRG1 iCLIP GSE262068 (41). Hi-C data (TCC) HUVECs exposed to hypoxia/normoxia GSE94872 (99). GRO-Seq data – HUVEC exposed to hypoxia/normoxia GSE136813 (100).

Statistical analysis

All statistical analyses were performed in R (version 4.3.1) using the tidyverse, ggpubr, and ggplot2 packages. Data are presented as mean ± standard error of the mean (SEM), unless otherwise stated. Comparisons between two groups were performed using unpaired two-tailed Student’s t test. For comparisons involving more than two groups, one-way ANOVA followed by pairwise two-tailed Student’s t-tests was used. Statistical significance was defined as p < 0.05 (∗), p < 0.01 (∗∗), p < 0.001 (∗∗∗), and p < 0.0001 (∗∗∗∗); non-significant differences were denoted as “ns.” All bar graphs display mean values with SEM error bars, and significance annotations were added using the stat_compare_means () function in ggpubr. Calcium influx kinetic data were plotted and analyzed using GraphPad Prism (ver. 9). ImageJ was used for densitometric quantification of Western blot bands. Unless otherwise specified, all experiments were performed in at least biological triplicate and repeated independently to ensure reproducibility.

Data availability

All data generated or analyzed during this study are included in this published article and its supporting information files. Raw data is available upon request.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.