Role of lncRNAs in stress-associated gene regulation following chromatin silencing: Mechanistic insights from an in vitro cellular model of glucocorticoid receptor gene overexpression
Stress significantly impacts brain function and is a major contributor to major depressive disorder. This study explores the role of long noncoding RNAs (lncRNAs) in stress-induced chromatin remodeling and gene regulation, particularly in the context of glucocorticoid receptor (GR) activation, a hallmark of stress-related disorders. Using a cellular model, GR activation regulated 79 lncRNAs: 44 upregulated and 35 downregulated. Chromosome-wise analyses of lncRNA regulation revealed complex patterns, with specific chromosomes, chr1 and chr15, showing a higher prevalence of downregulated genes, while chr11 and chr12 exhibited upregulated genes, indicating the differential roles of lncRNAs in gene silencing and activation. The differentially regulated lncRNAs were further classified into intergenic, overlap_antisense and overlap_sense, revealing their diverse roles in regulating gene expression. An immunoprecipitated ribonucleoprotein assay followed by sequencing confirmed the interaction of lncRNAs with a histone methyltransferase primarily targeting histone H3 lysine 27 (H3K27), with 87 lncRNAs enriched in the pull-down group, and 51 enriched in the EZH2 pull-down, a part of Polycomb Repressive Complex 2 (PRC2), linking these lncRNAs to PRC2-mediated chromatin repression. An integrative analysis revealed an inverse correlation between specific lncRNAs and the transcriptional activity of proximal genes, supporting the role of lncRNAs in gene silencing. Gene ontology of suppressed genes highlighted enrichment in functions related to neuronal death and synaptic transmission. Altogether, our study demonstrates that lncRNAs, particularly those interacting with PRC2, may serve as potential mediators of stress-induced neuronal changes by influencing chromatin accessibility, and potentially highlight their roles in stress-related disorders.
Introduction
As a critical environmental factor, stress is frequently associated with severe adverse effects on various brain functions (1). Depending on the type and magnitude of environmental influences, stress can play a key role in developing debilitating neuropsychiatric conditions, including major depressive disorder (MDD) (2–5). Recent studies suggest that persistent gene dysregulation triggered by stress can contribute to the etiopathology of MDD, which often results from complex regulatory mechanisms and their dynamic alterations (6, 7). In this context, epigenetic modifications, particularly those involving architectural changes in chromatin, have recently emerged as a key area of investigation (8–10). However, the exact molecular nature and associated mechanisms underlying these chromatin changes remain poorly understood.
lncRNAs, a family of noncoding RNAs, are increasingly recognized as critical regulators of chromatin accessibility in the brain (11–15). More specifically, nuclear lncRNAs interact with chromatin-modifying complexes to alter chromatin accessibility by switching euchromatin to a heterochromatin state (15–18). In this context, a few studies have recently highlighted the role of polycomb repressive complex 2 (PRC2) as a modifier of chromatin structure in the brains of rodents with behavioral deficits (19–22). Reports from our lab and others show that this interaction facilitates the deposition of repressive histone marks on chromatin, such as H3K27me3 (23–25) and triggers a heterochromatic state (26). Following the changes in chromatin accessibility, lncRNAs can alter the expression of genes proximal to the repressed chromatin domains (15, 17, 27). This mechanism represents a novel pathway through which stress could induce molecular and phenotypic changes in the brain (28, 29).
Stress-induced hyperactivity of the hypothalamic-pituitary-adrenal (HPA) axis is a hallmark of MDD (2, 5, 30, 31). Under normal circumstances, stressors trigger the release of corticotropin-releasing factor from the hypothalamus, which in turn leads to the secretion of adrenocorticotropic hormone. This, in turn, initiates the synthesis and release of glucocorticoids (GCs), which bind to glucocorticoid receptors (GRs) in limbic and prelimbic brain regions, where they play a crucial role in modulating the stress response (32). However, under stress, this negative feedback system becomes impaired, resulting in maladaptive physiological and morphological changes, including elevated cortisol levels, altered dendritic spine morphometry, and synaptic reorganization (33). Thus, GR functions as a master regulator of the HPA axis, mediating cellular responses to stress through GC signaling (6, 34). Although GR-induced changes in gene expression have been studied (35), the potential role of lncRNAs in mediating GR-mediated chromatin remodeling and transcriptional repression has yet to be explored. Given the growing recognition of lncRNAs in neuropsychiatric disorders (29, 36–38), elucidating the interplay between GR activation, lncRNA expression, and chromatin regulation is a critical step toward understanding the transcriptional irregularities in the brain under stressful conditions.
This study aimed to explore the mechanistic relationship between GR activation and lncRNA-mediated chromatin modulation that could lead to the selective repression of genes in the proximity of silenced chromatin. Using a cellular model of stress following GR overexpression, we performed next-generation RNA sequencing to comprehensively profile the lncRNA transcriptome and identify stress-induced changes in lncRNA expression. To further investigate the functional role of these lncRNAs, we conducted RNA immunoprecipitation (RIP) assays to detect lncRNAs associated with EZH2, a key component of the RNA-binding protein, PRC2, along with the H3K27me3 repressive chromatin mark (26, 39, 40). By coupling RIP with next-generation sequencing, we pinpointed specific lncRNAs that were enriched in PRC2- and H3K27me3-associated complexes following GR activation (41). Additionally, mRNA transcriptome profiling was performed on the same GR-overexpressed cells to assess the downstream effects of lncRNA-mediated chromatin changes on gene expression. The integration of lncRNA and mRNA data revealed a strong negative correlation between the expression of specific lncRNAs and the transcriptional activity of proximal genes. Altogether, our study provides the first mechanistic evidence that GR-induced lncRNA expression drives stress-mediated gene regulation through chromatin remodeling and suggests that similar mechanisms may contribute to stress-mediated transcriptional dysregulation in the brain of subjects with neuropsychiatric conditions such as MDD.
Results
lncRNA expression profile in SHSY5Y cellular model following GR activation
We first sought to determine whether GR overactivation, mimicking a hyperactive HPA axis, leads to changes in lncRNA expression in the SH-SY5Y cellular model. The transcriptome profiling of lncRNAs using next-generation sequencing revealed significant alterations in lncRNA expression in the GR-overexpressed cells. Out of 12,075 differentially expressed lncRNAs, 6259 were upregulated and 5816 lncRNAs were downregulated under GR activation. However, with a statistical threshold of p < 0.05, 79 lncRNAs were found to be significantly differentially expressed: 44 upregulated and 35 downregulated. A heatmap (Figure 1A) of the top 30 differentially expressed lncRNAs showed apparent clustering between GR-overexpressing and control cells, indicating distinct transcriptional changes. A volcano plot (Figure 1B) illustrates the distribution of 12,075 differentially expressed lncRNAs. The red and blue colored dots show significantly upregulated and downregulated lncRNAs, respectively, following adjustment with p < 0.05. The differential expression changes based on mean normalized counts are presented with an MA plot (Figure 1C), where the differences were transformed into a log2 fold-change (log2FC) scale showing higher expression in the GR-transfected group.
Citation: Genomic Psychiatry 2025; 10.61373/gp025h.0107
Genomic mapping and biotyping of stress-responsive lncRNAs following GR activation
To better understand the genomic distribution of lncRNAs expressed under GR overexpression, significantly upregulated and downregulated lncRNAs were mapped across 23 autosomes and the two sex chromosomes (X and Y) in a control versus GR overexpression comparison. As illustrated in the circular chromosomal plots (Figure 1D), an even proportion of upregulated lncRNAs was distributed across the autosomes, except for the two sex chromosomes, where the distribution was uneven. On the other hand, downregulated lncRNAs were evenly distributed across all autosomes and sex chromosomes. Under GR overexpression, notable variability in the intrachromosomal distribution was observed for downregulated lncRNAs compared to the upregulated class.
Understanding the various lncRNA biotypes is crucial for interpreting their distinct functional and regulatory roles. When characterized by genomic location, orientation, and relationship to protein-coding genes, lncRNAs exhibit a wide variety of biotypes. A stacked bar diagram shown in Figure 1E illustrates all the differentially expressed lncRNAs categorized by their biotype. These include bidirectional, axon-sense overlapping, intergenic, intron-sense overlapping, intronic antisense, and natural antisense. Each one of the classes showed a mixed or balanced distribution of up- and down-regulated sets of lncRNAs. Additionally, a Manhattan plot was generated to visualize the log2FC of differentially expressed lncRNAs across the various biotype classifications (Figure 1F), demonstrating the relationship between biotype and expression levels of lncRNAs.
Coexpression network analysis of lncRNAs responsive to GR activation
Coexpression network analyses were conducted to explore potential interactions among significantly differentially regulated lncRNAs in the GR overexpressed group (threshold: R > 0.7 and p < 0.01) (Figure 1G). These network analyses also demonstrated a scale-free topology, featuring a small number of lncRNAs serving as highly connected hubs. As displayed with dark magenta color, six nodes acting as hub lncRNAs in the differential network map were LINC02674, ENSG00000242375, ENSG00000233651, ENSG00000275478, ENSG0000026189, and ENSG00000260035. These findings suggest that GR-induced changes in the expression of lncRNAs are in higher concordance in their expression cohesiveness and display a close association in terms of their network topology.
Genome-wide mapping of lncRNAs on repressed heterochromatin as marked with H3K27me3 modification
GRactivation-induced lncRNA expression prompted us to investigate further their role in gene regulation via chromatin modification. To assess this, RIP was performed targeting H3K27me3, a repressive histone mark. Differential analysis identified 3982 enriched and 5324 depleted lncRNAs in GR-overexpressed cells, with 87 and 89 showing significant enrichment and depletion, respectively (p < 0.05). A volcano plot (Figure 2A) highlighted top H3K27me3-associated lncRNAs, including BMS1P4, PLPPR5-AS1, PLS3-AS1, PMS2P4, PNISR-AS1, PNN-AS1, ENSG00000226041, ENSG00000289449, FMR1-AS1, and ENSG00000228392 (p < 0.005). Notably, several GR-induced lncRNAs enriched in H3K27me3 complexes overlapped with upregulated lncRNAs from the transcriptome profiling (Supplementary Table S1). Their distribution by biotype is shown in a stacked bar diagram (Figure 2B), while a chord diagram (Figure 2C) maps their chromosomal origins, illustrating the genomic spread of H3K27me3-associated lncRNAs.
Citation: Genomic Psychiatry 2025; 10.61373/gp025h.0107
EZH2-mediated recruitment of GR-induced lncRNAs on repressed chromatin domain
To examine the recruitment of GR-induced lncRNAs to PRC2, we performed RIP targeting EZH2, a core component of PRC2. Sequencing of EZH2-RIP samples revealed 4752 enriched and 4639 depleted lncRNAs in GR-overexpressed cells, with 51 and 41 reaching significance, respectively (p < 0.05). A volcano plot (Figure 2D) highlighted top EZH2-associated lncRNAs, including ENSG00000289750, ENSG00000290674, ENSG00000286673, PSMB8-AS1, ENSG00000228318, GBP1P1, ENSG00000286553, ENSG00000289396, and LINC01050 (p < 0.005). A stacked bar diagram (Figure 2E) depicts the biotype distribution of differentially regulated lncRNAs, while a chord diagram (Figure 2F) maps their chromosomal origins. Notably, several EZH2-bound lncRNAs overlapped with H3K27me3-associated lncRNAs, supporting their role in establishing repressive chromatin states (Supplementary Table S2).
GR-induced differential mRNA sequencing reveals gene repression in the silenced chromatin domain potentially mediated by lncRNAs
To assess how lncRNA-mediated heterochromatization influences gene expression, we profiled mRNA transcripts under GR overexpression and identified 3237 differentially expressed genes , including 1879 upregulated and 1358 downregulated genes (p < 0.01). Differential expression patterns were visualized by volcano and MA plots (Figure 3A–B) and a heatmap of the top 30 upregulated and downregulated genes (Figure 3C). Pearson correlation further revealed a significant inverse relationship between global gene and lncRNA expression (R = –0.21, p < 0.005; Figure 3D).
Citation: Genomic Psychiatry 2025; 10.61373/gp025h.0107
Next, to identify genes affected by the PRC2-based chromatin silencing, we analyzed the lncRNA expression data obtained from two RIP experiments (EZH2 and H3K27me3) and separately compared them with gene expression data. Expression of lncRNA enriched for H3K27me and EZH2 showed a significantly strong negative correlation (r=–0.071 and –0.037, p < 0.0001) with nearby gene expression at repressed chromatin loci (Figure 3E–F).
Integrated lncRNA–mRNA analysis identified three GR-induced lncRNAs (ENSG00000225963.8, ENSG00000228412.9, and ENSG00000254211.6) that were consistently upregulated in GR-overexpressed cells and enriched in both H3K27me3- and EZH2-associated pull-down complexes.
Functional impact of lncRNA-mediated chromatin silencing on gene repression during GR activation
Gene Ontology (GO) enrichment analysis of downregulated genes revealed a strong impact on synaptic biology, including presynaptic and postsynaptic processes, vesicle mobilization, and neurotransmitter receptor regulation (transport and internalization) (Figure 4A). These GO categories are critical for maintaining synaptic plasticity and signal transmission, processes that are consistently impaired in stress-related neuropathology and a hallmark of MDD. Keyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis identified two core pathways affected by lncRNA-mediated repression in the GR-overexpression model—calcium signaling (p < 0.01) and glycosylphosphatidylinositol biosynthesis (p < 0.05) (Figure 4B–C). Both pathways are integral to neuronal excitability, receptor anchoring, and intracellular communication, suggesting that GR-driven repression of lncRNAs may destabilize fundamental neuronal signaling mechanisms. Although KEGG analysis yielded a conservative outcome, Reactome-based curation uncovered 33 significantly altered pathways (Supplementary Table S3). Among these, nerve growth factor (NGF)-independent tropomyosin receptor kinase A (TRKA) activation, multiple NTRK2 signaling branches (via RAC1, CDK5, PI3K, RAS, and FRS2/3), estrogen biosynthesis, FGFR1/FGFR2-mediated phospholipase C cascades, SLC transporter disorders, and negative regulation of the PI3K-AKT network stood out as central regulators of neuronal survival, dendritic remodeling, synaptic vesicle dynamics, and stress hormone responsiveness—all processes disrupted in MDD (42–44). Importantly, TrkA/TrkB signaling cascades and PI3K-AKT pathways are recognized as convergence points for stress-induced plasticity deficits and antidepressant action (43), underscoring the relevance of these findings. Finally, screening the downregulated genes against the DisGeNET database revealed strong enrichment for psychiatric phenotypes, with suicide, mood disorders, and panic disorder (Supplementary Table S4). This functional convergence highlights that repression of synaptic and neurotrophic signaling networks by GR-induced heterochromatin remodeling may represent a molecular mechanism linking stress to the pathophysiology of MDD.
Citation: Genomic Psychiatry 2025; 10.61373/gp025h.0107
Citation: Genomic Psychiatry 2025; 10.61373/gp025h.0107
Discussion
Studies in the recent past have significantly advanced our understanding of chromatin organization, the underlying mechanisms of three-dimensional (3D) genome structure, and its role in transcriptional regulation (8, 45). However, a large swath of mechanisms underlying this 3D chromatin organization remains unexplored, particularly in the context of neuropsychiatric disorders. Our present study examined how GR overactivation reprograms lncRNA landscapes and modulates chromatin architecture, establishing a foundation for transcriptional repression that could underlie stress-related neuropathology. More specifically, we investigated how GR overactivation in a cellular model that mimics HPA axis hyperactivity affects the expression landscape of lncRNAs. Additionally, we examined how lncRNA-mediated chromatin modulation may selectively repress genes near silenced chromatin. Our findings reveal that GR overexpression remodels the lncRNA transcriptome and subsequently mediates widespread gene repression through chromatin silencing mechanisms.
The altered lncRNA profiling in response to GR activation demonstrated significant upregulation of 44 and downregulation of 35 lncRNAs. These altered lncRNAs were distributed across various chromosomes and exhibited diverse biotypes, including intergenic, antisense, and sense-overlapping. The notable diversity in lncRNA biotypes further indicates that GR-induced lncRNAs may have multiple regulatory roles, such as modulating chromatin structure, influencing transcription, and interacting with other RNA species. Mapping differentially expressed lncRNAs across all chromosomes revealed a broad genomic footprint, with notable variability in intrachromosomal distribution. These findings resonate with reports showing that stress-induced epigenetic modifications, such as DNA methylation and histone modifications, are not restricted to specific loci but can affect chromatin globally (9). Moreover, the observation that sex chromosomes were relatively spared in upregulated lncRNA responses mirrors previous studies suggesting that autosomal genes, rather than sex-linked genes, are more dynamically regulated by GC signaling (46). Notably, our study found several upregulated lncRNAs to be associated with repressive chromatin marks, specifically H3K27me3, and the PRC2 component EZH2. This association suggests a mechanism by which GR-induced lncRNAs could contribute to gene silencing through chromatin remodeling (47).
Until recently, emerging evidence has highlighted the crucial role of lncRNAs in nuclear organization, including chromatin structuring and the formation of nuclear bodies (28, 48, 49). However, it is still a relatively new area of research, and the precise mechanisms by which lncRNAs influence 3D genome architecture are under active investigation across different biological contexts (50). Recent data suggest that due to their unique ability to interact with RNA-binding proteins, lncRNAs form an extensive network of ribonucleoprotein (RNP) complexes with numerous chromatin regulators (as RNA-binding proteins) (48). This complex molecular interaction then guides their associated enzymatic activities to appropriate locations in the genome (28). Moreover, lncRNAs can serve as modular scaffolds to specify higher-order chromatin organization in RNP complexes, causing localized changes in chromatin conformation that result in either heterochromatic or euchromatic states (51). Furthermore, the recruitment of lncRNAs to repressive chromatin regions aligns with previous studies demonstrating the role of lncRNAs in guiding chromatin-modifying complexes to specific genomic loci, thereby influencing gene expression patterns (19). One such complex is the PRC2, which silences chromatin by catalyzing H3K27me2/3 through its core subunits EZH1/2, SUZ12, and EED, with EZH2 serving as the primary catalytic component (19, 23, 52). Our RIP sequencing (RIP-seq) analyses of H3K27me3 and EZH2 revealed that a subset of GR-induced lncRNAs associate with facultative heterochromatin domains (53). This aligns with previous studies, which demonstrate that lncRNAs such as HOTAIR and MEG3 direct PRC2 to target loci (54, 55). It is believed that the chromatin-binding ability of the PRC2 complex could be primarily attributed to its interaction with select lncRNAs (56, 57). It has been suggested that upon its transcription, specific lncRNA can act in cis to recruit PRC2 via the SUZ12 subunit and direct H3K27 trimethylation of the locus with respect to a particular chromatin domain (52). For instance, Kcnq1ot1, an antisense noncoding RNA, plays a pivotal role in epigenetic regulation by interacting with chromatin and histone methyltransferases, such as G9a and the PRC2 complex (58). This lncRNA recruits the chromatin remodeling complexes to specific regions, resulting in the formation of extended heterochromatin regions marked by repressive histone modifications such as H3K9me3 and H3K27me3 (59). This process involves localizing these regions to the nucleolar compartment, creating a transcriptionally repressive environment (60, 61).
In neurons, the above-described mechanisms are increasingly appreciated. For example, PRC2-mediated repression governs neurodevelopmental timing and is dysregulated in psychiatric conditions (21, 22, 62, 63). From our findings, the enrichment of GR-upregulated lncRNAs in H3K27me3 and EZH2 complexes suggests that these molecules may act as mediators of stress-induced gene repression (59). Their convergence with transcriptome-wide repression further points to a potential lncRNA–chromatin interface as a downstream effector of HPA axis hyperactivity. Our mRNA sequencing (mRNA-seq) analysis further supports this notion, revealing that GR overexpression results in the downregulation of a significant proportion of genes associated with synaptic vesicle transport, neurotransmitter receptor activity, and calcium signaling pathways (42, 44, 64, 65, 66). These gene categories are foundational to synaptic efficacy, network oscillations, and neurocognitive function—all known to be impaired in depression and chronic stress states (43, 67–70), further suggesting the role of those stress-responsive lncRNAs in mediating the transcriptional repression via altered chromatin accessibility.
Our pathway enrichment analyses (KEGG and Reactome) converged on neurotrophic signaling cascades including TRK (NTRK2), PI3K-Akt, and FGFR1/2 pathways. As the literature suggests, these signaling axes are tightly linked to dendritic spine maintenance, neurogenesis, and affective resilience, and are suppressed in the brains of MDD subjects (71–74). Interestingly, our data mechanistically connect this suppression to GR-induced lncRNA activity, suggesting that chronic GC exposure hijacks epigenetic silencing programs to produce lasting reductions in synaptic gene expression (33). It is worth noting that these molecular changes mirror the synaptic deficits observed in MDD and other stress-related disorders (75). Furthermore, the inverse correlation between the expression of lncRNAs and nearby gene expression underscores the potential of lncRNAs to act in cis, repressing adjacent genes through chromatin-based mechanisms (59). This mode of action has been earlier implicated in the regulation of genes critical for neuronal function and has been observed in various models of stress and depression (17, 21, 22, 76, 77).
Intriguing changes were noted in the downregulated gene expression, which potentially highlights the translational relevance of our findings in various stress-responsive psychiatric conditions. By intersecting differentially repressed genes with a curated disease-gene association dataset (DisGeNET), we found robust enrichment for psychiatric phenotypes, most prominently, suicide, major depression, and panic disorder. This is in agreement with evidence that GR hypersensitivity is a core endophenotype in stress-related disorders, and that synaptic gene downregulation constitutes a convergent molecular signature across depressive and suicidality transcriptomes (78–80). Moreover, our data suggest that a specific set of three chromatin-bound lncRNAs (ENSG00000225963.8, ENSG00000228412.9, and ENSG00000254211.6) is negatively correlated with the expression of adjacent genes. These GR-induced lncRNAs engage EZH2, the catalytic subunit of PRC2, promoting H3K27me3 deposition and leading to transcriptional silencing of genes involved in synaptic and neuronal signaling. This GR-lncRNA-PRC2 axis constitutes a novel cellular mechanism linking stress-induced epigenetic repression to HPA axis activation and heightened vulnerability to neuropsychiatric disorders (Figure 5).
Altogether, for the first time, our study provides valuable insights into the role of GR-induced lncRNAs in chromatin regulation. Our study identified a novel axis of stress pathology wherein GR overactivation induces a distinct lncRNA program that engages polycomb-mediated chromatin silencing, leading to widespread repression of synaptic and neuronal signaling genes. Uncovering the involvement of H3K27me3 and PRC2 in this process provides a mechanistic framework for understanding the molecular basis of stress responses and their dysregulation in neuropsychiatric disorders. Our findings additionally offer mechanistic insight into how HPA axis activation produces long-lasting transcriptional impairments and position lncRNAs as crucial epigenetic effectors linking stress to psychiatric vulnerability. In doing so, we advance the current neurobiological models of stress-related disorders by integrating lncRNA-based chromatin control into the framework of GC-induced neuropathology. These findings could potentially open new avenues for therapeutic interventions targeting lncRNAs and chromatin dynamics to mitigate the effects of stress on the brain. Future studies using genome-wide chromatin interaction assays, such as Hi-C or Chromatin isolation by RNA purification-sequencing, could provide a comprehensive understanding of lncRNA-mediated chromatin remodeling (16). Additionally, functional validation of candidate lncRNAs through loss-of-function and gain-of-function studies will confirm their roles in stress-induced gene regulation.
Methods
A comprehensive methodology has been provided in the Supplemental section.
Cell culture and transfection of NR3C1 (GR) cDNA expression clone
Cellular transfection was performed using a previously published method with slight modifications (81). GR (NR3C1) was overexpressed by transfecting a cDNA clone (Sino Biologicals, USA) into SH-SY5Y cells using Lipofectamine 3000 (Invitrogen, USA). We chose GR overexpression rather than ligand stimulation to achieve a sustained and uniform activation of GR signaling, thereby avoiding variability arising from ligand pharmacokinetics or receptor desensitization. Harvested lysates were processed with TRIzol (Waltham, USA) for RNA isolation (INPUT) or with RIP lysis buffer for RIP.
RNA-immunoprecipitation assay
RIP was performed following a previously published method (39), with slight modifications. Specifically, prewashed magnetic protein A/G beads (Millipore Sigma, USA) were conjugated either with EZH2 (Cell Signaling Technology, USA, catalog no. 5246T) or H3K27me3 (Active Motif, USA, Catalog no. 39155) antibodies for separate pull-down experiments, incubated overnight with cell lysates, and the resulting RIP complexes were used to extract RNA with TRIzol (Waltham, USA). Total RNA from both INPUT and RIP samples was isolated using TRIzol with minor modifications to enrich small RNAs, including glycogen supplementation during overnight precipitation at −20°C (81, 82). RNA quality was assessed by Nanodrop (260/280 nm ≥ 1.8), agarose gel electrophoresis, and the Agilent Bioanalyzer 2100 (RNA Integrity Number [RIN] ≥ 8), and only high-quality samples were used for sequencing.
Transcriptome-wide lncRNA profiling following RNA sequencing
Strand-specific RNA sequencing (RNA-seq) libraries were prepared after rRNA depletion, followed by sequencing on the Illumina HiSeq X Ten platform. Reads were processed, aligned to GRCh37, assembled with StringTie, quantified with Kallisto, and differentially expressed lncRNAs were identified using Sleuth. Mapping of lncRNAs on Repressed Chromatin Domains (EZH2 and H3K27me3 RIP-seq).
RIP-seq was performed in GR-transfected SH-SY5Y cells using EZH2 and H3K27me3 antibodies, followed by library preparation and sequencing on Illumina HiSeq4000. Reads were trimmed, aligned with STAR, and differential enrichment was assessed using DESeq with FDR correction (p < 0.05).
Transcriptome-wide mRNA profiling of GR-transfected cellular RNAs
Poly(A)-enriched or rRNA-depleted RNA was used to construct stranded mRNA-seq libraries and sequenced on Illumina HiSeq4000. Reads were quality-checked, trimmed, aligned with Hisat2, quantified using StringTie/Ballgown, and differentially expressed genes were identified.
Expression correlation analysis
Correlation between lncRNA and mRNA expression was examined using Pearson and Spearman methods (83), with significance set at |R| ≥ 0.3 and p < 0.05.
Functional enrichment analysis
GO and KEGG pathway analyses were performed using DAVID and Enrichr to characterize genes near H3K27me3-silenced domains, applying Fisher's test with FDR correction (p < 0.05) (84, 85).
Data availability
All the data are provided in the manuscript and in the supplemental tables.
Acknowledgments
We sincerely acknowledge Kevin Prall for his assistance with cell culture studies.
Author contributions
YD conceptualized, designed, and supervised the study. AKV performed experiments. AKV and BR analyzed the data and prepared figures. BR, AKV, and YD co-wrote the manuscript. YD edited and finalized the manuscript. YD received funding. The manuscript has been read and approved by all authors. All authors take full responsibility for all data, figures, and text and approve the content and submission of the study. No related work is under consideration elsewhere. All authors state that all unprocessed data are available, and all figures provide accurate presentations of the original data.
Corresponding Author: Contact Professor Yogesh Dwivedi for any aspect of the work. The corresponding author takes full responsibility for the submission process.
Funding sources
This work was supported by funding from the National Institute of Mental Health (R01MH130539, R01MH124248, R01MH118884, R01MH128994, R01MH107183, and R56MH138596) to YD.
Author disclosures
The authors declare no conflict of interests.
Supporting Online Material
Transcriptomic profiling and analysis of lncRNAs in GR-overexpressing SH-SY5Y cells. (A) Heatmap of the top 30 differentially expressed lncRNAs showing hierarchical clustering between GR-overexpressing and control cells. (B) Volcano plot of 12,075 lncRNAs displaying log₂ fold-change versus –log₁₀(p-value) significance. Significantly upregulated lncRNAs (adjusted p < 0.05) are shown in red, and downregulated lncRNAs in blue. (C) MA plot representing log₂ fold-change versus mean normalized counts of differentially expressed lncRNAs under GR-overexpression. (D) Circular chromosomal plots showing the genomic distribution of significantly upregulated (track 1, blue bars) and downregulated (track 2, red bars) lncRNAs across all chromosomes. The innermost track presents both upregulated (blue) and downregulated (red) lncRNAs as scatter dots. (E) Stacked bar plot showing the distribution of differentially expressed lncRNAs by biotype, including bidirectional, exon-sense overlapping, intergenic, intron-sense overlapping, intronic antisense, and natural antisense categories. The Y-axis is representative of the number of counts of each biotype. (F) Manhattan plot of log₂ fold-change values of differentially expressed lncRNAs grouped by biotype and chromosomal origin. Some of the significantly regulated lncRNA gene symbols are labeled on the plot. (G) Coexpression network of significantly differentially expressed lncRNAs (|log₂FC| > 1, p < 0.05) constructed using Pearson correlation (R > 0.7, p < 0.01). Nodes represent lncRNAs, and edges indicate significant coexpression relationships. Hub lncRNAs are highlighted in dark magenta.
Identification and chromatin association of GR-induced lncRNAs with H3K27me3 and EZH2 following RNA-induced immunoprecipitation sequencing (RIP-seq) analysis. (A) Volcano plot of RIP-enriched lncRNAs from H3K27me3 immunoprecipitation in GR-overexpressing cells, showing log₂ fold-change versus –log₁₀(p-value) significance levels. Top enriched lncRNAs are highlighted with their gene symbols. (B) Stacked bar plot showing the distribution of H3K27me3-associated differentially expressed (up and down) lncRNAs categorized by biotype, including bidirectional, exon-sense overlapping, intergenic, intron-sense overlapping, intronic antisense, and natural antisense types. The y-axis is representative of the percentage counts of each biotype. (C) Chord diagram mapping top H3K27me3-associated lncRNAs to their chromosomal origins. (D) Volcano plot of RIP-enriched lncRNAs from EZH2 immunoprecipitation in GR-overexpressing cells, with log₂ fold-change versus –log₁₀(p-value) significance. Top enriched lncRNAs are highlighted with their gene symbols. (E) Stacked bar plot showing the distribution of EZH2-associated differentially expressed lncRNAs categorized by biotype, including bidirectional, exon-sense overlapping, intergenic, intron-sense overlapping, intronic antisense, and natural antisense types. The Y-axis is representative of the percentage counts of each biotype. (F) Chord diagram mapping top EZH2-associated lncRNAs to their chromosomal origins.
Coding gene expression changes and their correlation with GR-induced lncRNAs and repressive chromatin marks. (A) Volcano plot of differentially expressed coding genes in GR-overexpressing SH-SY5Y cells, showing log₂ fold-change versus –log₁₀(p-value). (B) MA plot displaying log₂ fold-change versus mean normalized counts of coding genes. (C) Heatmap of the top 30 upregulated and downregulated coding genes, illustrating expression differences between GR-overexpressing and control groups. (D) Scatter plot showing correlation between genome-wide coding gene expression and differential lncRNA expression (Pearson r = –0.21, p < 0.005). (E, F) Scatter plots displaying correlations between expression of lncRNAs enriched in H3K27me3 (E) or EZH2 (F) RIP-seq datasets and expression of genes (Pearson r = –0.071 and –0.037, respectively; p < 0.0001) located in the vicinity of repressed chromatin.
Functional analysis of genes inversely linked with GR-induced, repressed chromatin-associated lncRNAs. (A) Gene Ontology enrichment analysis of downregulated coding genes associated with upregulated lncRNAs enriched in H3K27me3 and EZH2 complexes, highlighting synaptic and neurotransmission-related processes. (B) KEGG pathway analysis showing significant enrichment of the calcium signaling pathway (p < 0.01). (C) KEGG pathway analysis showing enrichment of the glycosylphosphatidylinositol-anchor biosynthesis pathway (p < 0.05).
GR-induced lncRNAs mediate chromatin silencing via polycomb engagement during stress response. This figure illustrates the mechanistic model uncovered in our study, highlighting how glucocorticoid receptor (GR) activation induces a distinct set of long noncoding RNAs (lncRNAs)—ENSG00000225963.8, ENSG00000228412.9, and ENSG00000254211.6—that are chromatin-bound and exhibit negative correlation with expression of neighboring genes. These lncRNAs engage EZH2 as part of polycomb repressive complex 2 (PRC2), promoting H3K27me3 deposition and leading to transcriptional silencing of synaptic and neuronal signaling genes. The model underscores a novel regulatory axis through which GR overactivation mediates epigenetic repression, offering insights into the long-lasting transcriptional effects of hypothalamic-pituitary-adrenal (HPA) axis activation. This pathway provides a mechanistic link between stress exposure, chromatin remodeling, and increased vulnerability to neuropsychiatric disorders, positioning lncRNAs as key epigenetic mediators of glucocorticoid-induced neuropathology.
Contributor Notes
Publisher's note: Genomic Press maintains a position of impartiality and neutrality regarding territorial assertions represented in published materials and affiliations of institutional nature. As such, we will use the affiliations provided by the authors, without editing them. Such use simply reflects what the authors submitted to us and it does not indicate that Genomic Press supports any type of territorial assertions.
