A shift in chromatin binding of phosphorylated p38 precedes transcriptional changes upon oxidative stress

Abstract

P38 mitogen‐activated protein kinases are key in the regulation of the cellular response to stressors. P38 is known to regulate transcription, mRNA processing, stability, and translation. The transcriptional changes mediated by phosphorylated p38 (P‐p38) in response to extracellular stimuli have been thoroughly analyzed in many tissues and organisms. However, the genomic localization of chromatin‐associated P‐p38 remains poorly understood. Here, we analyze the chromatin binding of activated P‐p38 and its role in the response to reactive oxygen species (ROS) in Drosophila S2 cells. We found that P‐p38 is already bound to chromatin in basal conditions. After ROS exposure, chromatin‐associated P‐p38 relocates towards genes involved in the recovery process. Our findings highlight the role of P‐p38 dynamic chromatin binding in orchestrating gene expression responses to oxidative stress.

Chromatin‐associated phosphorylated p38 (P‐p38) is linked with active genes in Drosophila S2 cells under control conditions. In response to oxidative stress, P‐p38 relocates, favoring previously inactive genes. These genes are then activated during the recovery phase, highlighting the role of P‐p38 in gene regulation during stress responses.

Abbreviations

ChIP‐seq, chromatin immunoprecipitation sequencing

CTL, control

FDR, false discovery rate

GO, gene ontology

H ₂ O ₂ , hydrogen peroxide

H3K4me3, histone H3 lysine 4 trimethylation

MAPKs, mitogen‐activated protein kinases

P‐p38, phosphorylated p38

qPCR, quantitative PCR

SAPKs, stress‐activated protein kinases

STR, stress

TES, transcription end site

TF, transcription factor

TSS, transcription start site

Many intracellular signaling pathways that are activated by environmental stimuli rely on the post‐translational modifications of proteins, such as phosphorylation, driven by the balance between kinases and phosphatases. Mitogen‐activated protein kinases (MAPKs) are highly conserved serine/threonine protein kinases involved in multiple signal transduction pathways [1]. MAPKs undergo sequential phosphorylation and activation by upstream kinases. They function both in the cytoplasm and in the nucleus, where they may interact with chromatin through direct binding to DNA or chromatin‐associated substrates [2]. Exposure of cells to stress results in the rapid activation of the MAPKs JNK and p38, also known as stress‐activated protein kinases (SAPKs) [1, 3, 4]. P38 is activated by several physical and chemical stressors and is involved in a broad range of cellular processes, such as proliferation, differentiation, regeneration, migration, and apoptosis [3, 4, 5, 6, 7, 8, 9, 10, 11].

In response to extracellular stimuli, specific upstream kinases phosphorylate p38 to activate its kinase domain and, subsequently, phosphorylated p38 (P‐p38) will phosphorylate its substrates either in the cytoplasm or in the nucleus [1, 4, 5, 6, 12]. The nuclear translocation of p38 depends on p38 phosphorylation and on the action of importins [13, 14, 15, 16, 17]. In the nucleus, P‐p38 substrates include different transcription factors (TFs), chromatin‐modifying enzymes and remodelers, and core elements of the transcriptional machinery [18, 19, 20, 21, 22, 23]. Moreover, p38 can also localize at chromatin [24, 25], indicating a direct regulation of gene expression and function at the chromatin level. The transcriptional response mediated by p38 has been analyzed in several organisms. P38 signaling is critical to elicit the early gene expression program required for mouse embryonic fibroblasts adaptation to stress [26]. In yeast, p38 homolog Hog1 plays a key role in global gene regulation under saline stress [27]. Additionally, p38 mediates changes in gene expression in cellular processes unrelated to stress. It regulates TEF‐1 and C/EBPbeta transcriptional activity in proliferating cardiomyocytes [18], promotes muscle‐specific gene expression [28], and mediates tumor necrosis factor alpha signaling [29].

The production of reactive oxygen species (ROS), as byproducts of metabolism, is a source of oxidative stress that can be harmful for the cell. However, at physiological levels, ROS are also considered beneficial as they can act as signaling molecules [30, 31, 32]. ROS can be produced during aerobic respiration in the mitochondria as well as by several enzymes located in different cellular compartments, including peroxisomes, the endoplasmic reticulum, and the plasma membrane [32]. The most abundant type of ROS inside the cell is hydrogen peroxide (H₂O₂), which is produced by different stimuli [32]. In Drosophila, physiological levels of ROS have been shown to control key cellular and developmental processes, such as neuronal plasticity, stem cell proliferation and maintenance, and differentiation of immune cells upon infection [32, 33, 34]. A role for P‐p38 by reducing ROS levels has also been reported during embryonic wound repair in flies [35]. Moreover, ROS‐induced JNK and p38 signaling are required to activate the cytokines Unpaired (Upd) during regeneration of wing imaginal discs [10] and to mediate the regenerative response in the adult midgut [36]. In this study, we explore the chromatin localization of P‐p38 in response to oxidative stress in Drosophila S2 cells. We find that P‐p38 is already bound to chromatin in control conditions, but it changes its localization in the genome after exposure to H₂O₂.

A Catalan version of the article's abstract is available at: https://doi.org/10.5281/zenodo.13388878

Materials and methods

Induction of oxidative stress in Drosophila S2 cells

Drosophila S2 cells (ATCC CRL‐1963) were grown in Schneider's medium (Sigma‐Aldrich, St. Louis, MO, USA) supplemented with 10% FBS (Gibco, Waltham, MA, USA), 100 mg·mL⁻¹ of streptomycin, and 100 mg·mL⁻¹ of penicillin at 25 °C. For the induction of oxidative stress, cells were grown to 5 × 10⁶ cells·mL⁻¹ and incubated with three different concentrations of H₂O₂ (Sigma‐Aldrich, 516813) diluted in Schneider's medium: 5, 10, or 20 mm. The same volume of Schneider's medium was added to the control cells. Cells were incubated at 25 °C for the indicated time. To calculate the cell survival upon H₂O₂ treatment, we stained the cells using Trypan Blue and counted the number of living and dead cells using a Neubauer chamber.

Antibodies

The antibodies used in these experiments were: anti‐P‐p38 (Cell Signaling Technology, Danvers, MA, USA, 9211), anti‐histone H3 trimethyl Lys4 (H3K4me3; Abcam, Cambridge, UK, ab8580), anti‐histone H3 (Abcam, ab1791), and anti‐α‐tubulin (Invitrogen, Waltham, MA, USA, A11126). The commercial secondary antibodies used were coupled to horseradish peroxidase (Jackson ImmunoResearch, West Grove, PA, USA) or Alexa fluorophores (Invitrogen).

Immunostaining

For the immunostaining experiments, cells were attached to slides previously treated with concanavalin A (0.5 mg·mL⁻¹; Sigma, C5275), washed with PBS for 10 min with agitation and fixed in 4% paraformaldehyde for 15 min. After fixation, cells were washed with agitation for 15 min in PBS, and in PBST (PBS‐0.3% Triton X‐100) containing 0.2% BSA for 10 min twice. The anti‐P‐p38 (1 : 100) primary antibody was diluted in PBST‐0.2% BSA and added to the cells, which were incubated overnight with slow agitation at 4 °C. The next day, cells were washed three times with PBST‐0.2% BSA and incubated with the secondary antibody (1 : 200) diluted in PBST‐0.2% BSA for 1 h at room temperature with agitation. Cells were then washed twice for 10 min in PBST and incubated with Phalloidin (1 : 75; Invitrogen, A12379) for 30 min. Cells were then washed four times with PBS and mounted in SlowFade Diamond Antifade Mountant with DAPI (Invitrogen, S36964).

Images were recorded on a Zeiss LSM 880 microscope. Nuclear and cytoplasmic P‐p38 signal intensity was measured for 10 cells per condition. Two‐way ANOVA followed by Sidak's multiple comparison test were used to address differences in cytoplasmic and nuclear P‐p38 signal among different conditions. One‐way ANOVA followed by Sidak's multiple comparison test were used to address differences in the nuclear to cytoplasmic P‐p38 signal among different conditions.

Chromatin immunoprecipitation (ChIP)

Cells were collected and fixed in 1% formaldehyde for 10 min at room temperature by gentle mixing. Glycine was added to a final concentration of 125 mm to stop crosslinking. After 5 min, cells were spun down for 2 min at 1500  g and washed with 5 mL of PBS. Cells were then resuspended in 10 mL of wash buffer A (10 mm HEPES pH 7.9, 10 mm EDTA, 0.5 mm EGTA, and 0.25% Triton X‐100) and incubated for 10 min at 4 °C on a rotating wheel. Cells were spun down again and resuspended in 10 mL of wash buffer B (10 mm HEPES, pH 7.9, 100 mm NaCl, 1 mm EDTA, 0.5 mm EGTA, and 0.01% Triton X‐100), incubated for 10 min on a rotating wheel at 4 °C and spun down again. Cells were lysed in 5 mL of TE (10 mm Tris–HCl, pH 8, and 1 mm EDTA) and 1% SDS. Chromatin was washed three times with 5 mL of TE and resuspended in TE containing 1 mm phenylmethanesulfonyl fluoride (PMSF) and 0.1% SDS. Chromatin was sonicated in a Bioruptor sonicator (Diagenode, Seraing, Belgium) to obtain fragments of 200–500 bp. Lysates were adjusted with 1% Triton X‐100, 0.1% sodium deoxycholate (DOC), and 140 mm NaCl, before being incubated for 10 min on a rotating wheel at 4 °C. Chromatin was then recovered by centrifugation.

For each experiment, 40 μL of chromatin were used for the input sample, while two aliquots of 400 μL were used for the immunoprecipitation (IP) of P‐p38 and H3K4me3. IPs were carried out using RIPA buffer (140 mm NaCl, 10 mm Tris–HCl, pH 8, 1 mm EDTA, 1% Triton X‐100, 0.01% SDS, and 0.1% DOC). The preclearing of chromatin samples was performed on a rotating wheel for 1 h at 4 °C with 30 μL of 50% (v/v) protein A‐Sepharose CL4B beads (GE Healthcare, Chicago, IL, USA, 17‐0780‐01) previously blocked with RIPA‐1% BSA. The antibody was added, and the incubation was performed overnight at 4 °C on a rotating wheel. IPs were performed by adding 40 μL of 50% (v/v) protein A‐Sepharose CL4B beads previously blocked with RIPA‐1% BSA and incubating the samples on a rotating wheel for 3 h at 4 °C. Beads were washed five times for 5 min in 1 mL of RIPA, once for 5 min in 250 mm LiCl buffer (250 mm LiCl, 10 mm Tris–HCl, pH 8, 1 mm EDTA, 0.5% NP‐40, and 0.5% DOC) and twice for 5 min in TE. Then, the beads were resuspended in 40 μL of TE, and DNase‐free RNase A was added at 0.25 μg·mL⁻¹ to the IPs and input samples and incubated for 30 min at 37 °C. Samples were adjusted with 1% SDS, 0.1 m NaHCO₃, and 0.2 mg·mL⁻¹ of Proteinase K and incubated overnight at 65 °C for decrosslinking. DNA was purified with phenol‐chloroform extraction. The antibodies used were anti‐P‐p38 and anti‐H3K4me3. Library preparation and sequencing using the HiSeq 2500 system were undertaken at the CRG Genomics Unit (Barcelona, Spain). Two biological replicates and their corresponding inputs were sequenced per condition.

ChIP‐seq data processing and analysis

Data were processed using the chip‐nf pipeline (https://github.com/guigolab/chip‐nf; revision: 47c471b6f4 [v0.2.3]; nextflow v19.07.0). Reads were continuously mapped to the fly genome (dm6), with up to two mismatches using the GEM mapper [37]. Only alignments for reads mapping to 10 or fewer loci were reported. Duplicated reads were removed using Picard (http://broadinstitute.github.io/picard/). The fragment length was set to 200 bp. We ran MACS2 [38] to identify the regions significantly enriched on ChIP‐Seq reads from each sample in comparison to the normalized input control. We retained peaks that had at least 50% overlap in each replicate using the bedtools intersectbed tool [39]. Peak scores were rescaled to conform to the format supported by the UCSC genome browser (score must be < 1000). Peaks spanning < 50 bp or located in non‐canonical chromosomes were discarded. Peaks showing ≥ 50% overlap in control and stress conditions were considered the same peak. Coordinates of identified P‐p38 and H3K4me3 peaks are shown in Table S1 and Table S2, respectively.

P‐p38 peaks were classified based on their genomic location into: promoter, if the peak center was positioned within −500 bp and +100 bp from the TSS of a gene; gene body, if the peak overlapped at least 1 bp with a gene body; and intergenic, if none of the previous applied. H3K4me3 peaks were classified into: promoter, if the peak overlapped at least 1 bp with the region located ±500 bp from a TSS; and gene body, if the peak overlapped at least 1 bp with a gene body. Intergenic H3K4me3 peaks were not considered. Peak classification was mutually exclusive in the following rank: promoter > gene body > intergenic.

Average plots and heatmaps were calculated for identified promoter, gene body, and intergenic P‐p38 peaks in Fig. S2, and for promoter and gene body H3K4me3 peaks in Fig. S3. Bigwig files from input and experimental replicates for control and stress conditions were used. Metagene plots for genes associated with P‐p38 and H3K4me3 peaks shown in Fig. 2C,F, respectively, were generated using the bigwig average of the two experimental replicates per condition.

Fig. 2.

P‐p38 chromatin localization changes upon oxidative stress. (A) Number of P‐p38 chromatin peaks identified in untreated control cells (Control), upon H₂O₂ exposure (Stress), and in both conditions. (B) Genomic distribution of P‐p38 chromatin peaks for each condition. (C) Metagene plot showing the distribution of P‐p38 signal along the genes associated with P‐p38 peaks in control (left plot) and in stress (right plot). (D) Number of H3K4me3 chromatin peaks identified in untreated control cells (Control), upon H₂O₂ exposure (Stress), and in both conditions. (E) Genomic distribution of H3K4me3 chromatin peaks for each condition. (F) Metagene plot showing the distribution of H3K4me3 signal along the genes associated with H3K4me3 peaks in control (left plot) and in stress (right plot). (G) Screenshots of the UCSC Genome Browser showing P‐p38 and H3K4me3 ChIP‐seq tracks in Control and Stress conditions. Screenshot on top shows a P‐p38 control peak present at the gene body of eEF1gamma. Screenshot at the middle shows a P‐p38 stress peak in an intergenic region of chr 2R. Screenshot at the bottom shows a P‐p38 peak present in control and stress conditions at the promoter of the lncRNA:Hsromega. ChIP‐seq tracks shown are normalized to input. (H) Proportion of genes associated with P‐p38 that contain H3K4me3 marks at their promoter or gene body. Proportions are calculated separately for genes associated with P‐p38 in the promoter, gene body, or intergenic regions. (I) Expression of genes associated with P‐p38 in the control, upon stress, or in both conditions. Relative gene expression in the control and after 30 min of 10 mm H₂O₂ exposure is shown and is represented as the log₁₀ of TPMs plus a pseudocount of 0.1. CTL, control; STR, stress. Chi‐squared test with Bonferroni correction for multiple comparisons was used to assess significance in B, E, and H. Two‐way ANOVA and Dunnett's test for multiple comparisons were used to assess significance in I. ***P < 0.001; n.s., non‐significant.

RNA extraction and quantitative PCR (qPCR)

For expression analysis, total RNA was isolated from 10⁷ cells using Trizol (Ambion, Waltham, MA, USA) and the RNA Clean and Concentrator kit (Zymo Research, Irvine, CA, USA). A total of 1 μg of RNA was used as template for cDNA synthesis using Moloney Murine Leukemia Virus reverse transcriptase (M‐MLV) (Invitrogen).

Reactions containing FastStart Universal SYBR Green Master (Rox) (Roche, Basel, Switzerland), the appropriate cDNA, and primers were run in a 7500 Real‐Time PCR System (Applied Biosystems, Foster City, CA, USA). The levels of sply were used to normalize the samples, and relative RNA expression was calculated using the ddCt method. Three technical replicates were used for each reaction, and three separate biological replicates were collected for each experiment. Primer sequences used are shown in Table S3.

RNA‐seq library preparation, data processing, and analysis

Sequencing libraries were prepared using the TruSeq Stranded mRNA Library Prep kit (Illumina, San Diego, CA, USA), following the manufacturer's instructions. Paired‐end sequencing was performed in a HiSeq 2500 sequencer. Library preparation and sequencing were undertaken at the CRG Genomics Unit (Barcelona, Spain). Three biological replicates were sequenced per condition.

Data were processed using grape‐nf (available at https://github.com/guigolab/grape‐nf; revision: 5fb9c88236 [v0.2.1]; nextflow version 19.07.0). RNA‐seq reads were aligned to the fly genome (dm6) using the star 2.4.0j software [40], with up to 4 mismatches per paired alignment using the FlyBase genome annotation r6.36 [41]. Only alignments for reads mapping to 10 or fewer loci were reported. Gene and transcript TPMs were quantified using RSEM [42]. Differential gene expression analysis (DEA) was performed using DESeq2 [43]. Low count genes (less than 10 counts across all samples) were filtered out before DEA. No outlier sample was indicated by Cook's distance method available in DESeq2. Shrinkage of log fold change values was performed using ashr method [44]. A fold change ≥ |1.5| and an adjusted P‐value < 0.05 were used to consider a gene to be differentially expressed. Kruskal–Wallis test followed by Dunn's test for multiple comparisons were used to analyze differences in gene expression between groups. A minimum P value < 0.05 was set for significance. The list of differentially expressed genes is shown in Table S4.

Hierarchical clustering of genes associated with P‐p38

For gene clustering, we used available RNA‐seq data from S2 cells in control conditions (Pre‐treatment), after 3 h of sodium arsenite exposure (Post‐treatment), and after 3 h of recovery in control media (Recovery) [45]. Three biological replicates were used for each condition. We discarded genes that were not expressed in any sample by removing all genes whose sum of counts was < 10. Complete hierarchical clustering of genes associated with P‐p38 was based on the Euclidean distance calculated using z‐scores for each gene. Three main dendrogram branches were used for the assignment of genes into Clusters 1, 2, and 3.

Functional annotation of genes and motif enrichment analysis

We used the clusterprofiler tool version 3.10.1 [46] from Bioconductor to identify the enriched Gene Ontology (GO) terms in our study. We searched for enriched biological process terms setting a < 0.01 P‐value cutoff and using False Discovery Rate (FDR) as adjustment method.

We used the ame tool version 5.5.3 [47] to search for significantly enriched motifs in P‐p38 peaks. We selected the input motifs from the JASPAR 2022 Insects database [48]. We selected a < 10 E‐score to consider significant hits and used the default options for the analysis. Only motifs from TFs expressed at least 1 TPM in at least one condition (Pre‐treatment, Post‐treatment, or Recovery) were considered.

Statistical analysis

To analyze differences across different treatments in Figs 1F,G, 2I and 3A', and S1, we used a two‐way ANOVA followed by Dunnett's test for multiple comparisons. For evaluating differences in proportions among the different groups depicted in Figs 2B,E,H and 3C, we used a contingency table followed by a Chi‐squared test with Bonferroni correction for multiple comparisons.

Fig. 3.

Stress recovery genes are more associated with P‐p38. (A, A') Expression of genes associated with P‐p38 peaks. Fold change of TPMs from RNA‐seq experiments using S2 cells in control media (Pre‐treatment), after 3 h of sodium arsenite exposure (Post‐treatment), and after 3 h of recovery in control media (Recovery) are shown from Singh and colleagues [45] (A). Normalized ddCt from the qPCR of S2 cells incubated with control media (CTL), after 30 min of 10 mm H₂O₂ incubation (STR), and after 1 h of recovery in control media (REC) (A'). Error bars in A' represent the standard deviation (SD) of 3 biological replicates. (B) Heatmap of genes containing P‐p38 only in the control, only upon stress, in control and stress conditions, and of genes not associated with P‐p38. RNA‐seq data from Singh et al. [45]. Three biological replicates are represented per condition. Normalized Z‐score is shown for each gene. Complete hierarchical clustering was used to generate gene and sample dendrograms. (C) Proportion of genes assigned to each cluster separated by their association with chromatin P‐p38. (D) Selected Gene Ontology (GO) biological process terms enriched in each cluster and condition. Major biological processes are shown for each group of GO terms. Gene Ratio is represented as bubble sizes, while bubble color represents the false discovery rate (FDR). (E) Selected TF motifs enriched in P‐p38 chromatin peaks for each cluster. Only TFs expressed Pre‐treatment, Post‐treatment or in Recovery samples were considered. Two‐way ANOVA and Dunnett's test for multiple comparisons were used to assess significance in A'. Chi‐squared test with Bonferroni correction for multiple comparisons was used to assess significance in C. ***P < 0.001; *P < 0.05; n.s., non‐significant.

All statistical tests were two‐tailed, with significance determined at P‐values less than 0.05. All statistical tests were conducted using graphpad prism 9 (GraphPad Software, Boston, MA, USA) or r.

Results and Discussion

P‐p38 accumulates in the nucleus upon oxidative stress

To explore the response of P‐p38 to oxidative stress, we used Drosophila S2 cells, a model system previously employed to investigate p38 activation in stress conditions [49, 50, 51]. In agreement with prior research [49, 51], we observed low levels of P‐p38 in S2 cells under control conditions, predominantly located in the nucleus (Fig. 1A). To investigate the effect of oxidative stress in the nuclear localization of P‐p38, we first determined the survival curve at various hydrogen peroxide (H₂O₂) concentrations and incubation times (Fig. S1). A low dose of 5 mm H₂O₂ did not induce changes in the P‐p38 signal after a short exposure of 30 min (Fig. 1B,F), but it significantly increased the nuclear signal of P‐p38 after 1 h of incubation (Fig. 1C,F). Conversely, a higher dose of 10 mm H₂O₂ significantly increased the nuclear signal of P‐p38 after 30 min of treatment (Fig. 1D,F), but these levels did not show further increase after a longer exposure time (Fig. 1E,F).

Fig. 1.

Phosphorylated p38 accumulates in the nucleus upon exposure to H₂O₂. (A–E) Immunostaining of S2 cells after incubation with control media (A), after incubation with 5 mm H₂O₂ for 30 min (B) or 1 h (C), and after incubation with 10 mm H₂O₂ for 30 min (D) or 1 h (E). Top row (A–E) shows Phalloidin labeling the cytoplasm in yellow, DAPI labeling the nuclei in blue, and P‐p38 signal in magenta. Middle row (A'–E') shows DAPI. Bottom row (A”–E”) shows P‐p38. Scale bar = 10 μm. (F) Quantification of P‐p38 immunostaining. Mean signal intensity of P‐p38 in the cytoplasm and in the nucleus is shown. Error bars represent the standard error (SD) of N = 10 cells. (G) Nuclear to cytoplasmic P‐p38 signal ratio. Median and interquartile range of N = 10 cells are represented. Two‐way ANOVA and Dunnett's test for multiple comparisons were used to assess significance in F and G. ***P < 0.001; *P < 0.05; n.s., non‐significant.

In response to oxidative stress, we observed a dose‐ and time‐dependent nuclear accumulation of P‐p38, which aligns with earlier studies describing a rapid nuclear translocation of active p38 in reaction to various stimuli [6]. We also found that subjecting S2 cells to a higher dose of 10 mm H₂O₂ for a short duration of 30 min maximized the P‐p38 nuclear to cytoplasmic signal ratio (Fig. 1G). Consequently, subsequent experiments involved a 30‐min incubation of S2 cells with 10 mm H₂O₂.

Oxidative stress induces changes in the chromatin localization of P‐p38

Next, we analyzed the chromatin‐binding profile of nuclear P‐p38 by chromatin immunoprecipitation sequencing (ChIP‐seq) in untreated cells (Control) and in 10 mm H₂O₂‐treated cells (Stress). We found that the number of P‐p38 chromatin peaks was comparable in both conditions: 1552 peaks in the control and 1616 peaks in the stressed cells, with only 109 peaks shared between them (3.5% of the total identified P‐p38 peaks) (Fig. 2A; Fig. S2). We classified peaks according to their position relative to the transcription start site (TSS) of their closest annotated gene and defined 3 types of peaks: promoter peaks (located −500 bp to +100 bp from the TSS), gene body peaks (overlapping exons or introns), and intergenic peaks. In control conditions, we found 7.3% of P‐p38 peaks associated with promoter regions, 89.6% localized in the gene body, and only 3.1% in intergenic regions (Fig. 2B). However, exposure to H₂O₂ induced a redistribution of chromatin P‐p38 across the genome, leading to an increase of intergenic peaks (33.3%), and a concurrent reduction of promoter and gene body peaks to 3.9% and 62.8%, respectively (Fig. 2B). Peaks present in both conditions showed intermediate percentages between control and stress peaks (Fig. 2B). Metagene plots reveal that the P‐p38 signal is found along the entire gene body region, from the TSS to the transcription end site (TES) (Fig. 2C). Our results lead to two main conclusions: (a) P‐p38 primarily associates with gene body regions, and (b) oxidative stress induces a redistribution of P‐p38 to intergenic regions.

The major presence of P‐p38 in gene body regions suggests that it may operate not only in the initiation, but also in the elongation of transcription. In yeast and mammals, p38 is present along the coding sequence and interacts with different elongation factors and the RNA polymerase machinery [21, 24, 52]. Indeed, in yeast, the phosphorylation of the transcription elongation factor Spt4 by p38 homolog Hog1 facilitates transcriptional elongation after osmotic shock [21]. A similar situation may occur in S2 cells, although further experiments are required to corroborate this hypothesis, as there have been no reports yet on the interaction of p38 and Spt4 in Drosophila. After stress, however, there is an increase in P‐p38 bound to intergenic regions, suggesting a potential role for these regions as enhancers.

As a first approach to relate chromatin‐associated P‐p38 to gene expression, we examined the tri‐methylation of lysine 4 on histone H3 (H3K4me3) by ChIP‐seq in control and stressed cells. H3K4me3 is a chromatin modification typically restricted to narrow regions around the TSS of expressed genes, although it can also be found covering extensively the coding regions of some genes [53]. The majority of H3K4me3 peaks identified were present in both conditions (5427 peaks; 84.4% of total), while a small fraction was present only in control (671 peaks; 10.4%) or stress (333 peaks; 5.2%) (Fig. 2D; Fig. S3). As expected, most H3K4me3 peaks were primarily located close to the TSS (Fig. 2E,F). The presence of H3K4me3 marks in almost identical positions in control and stressed cells suggests that very few changes in gene expression occurred after 30 min of H₂O₂ exposure (Fig. 2D–F).

Next, we analyzed the correlation between P‐p38 and H3K4me3. We observed that genes associated with P‐p38 in the promoter or gene body after stress were less marked by H3K4me3 than genes containing P‐p38 in the same genomic regions in control or in both conditions (Fig. 2G,H). This suggests that, upon stress, P‐p38 preferentially binds to lowly expressed genes. We then assigned P‐p38 intergenic peaks to the nearest gene [54]. We found that these genes also displayed lower levels of H3K4me3 in their promoter and gene body regions (Fig. 2G,H), indicating less expression. To further link P‐p38 binding to gene expression, we performed RNA‐seq. In accordance with the levels of H3K4me3, we observed that the expression of genes associated with P‐p38 under stress was significantly lower than that of genes bound by P‐p38 in the control or in both conditions (Fig. 2I). Probably due to the short exposure time to H₂O₂ (30 min), we found very few differential‐expressed genes (202 genes; 2.6% of all expressed genes). Altogether, our results suggest that the shift observed in P‐p38 binding towards intergenic regions upon stress may occur prior to changes in gene expression.

P‐p38 targets recovery genes in response to oxidative stress

To get more insight into the expression profiles of genes associated with P‐p38 in response to stress, we used available RNA‐seq data obtained from Drosophila S2 cells after acute oxidative stress [45]. These data include three different conditions: cells incubated in standard media (Pre‐treatment), cells subjected to acute oxidative stress through exposure to sodium arsenite for 3 h (Post‐treatment), and cells incubated in control media for 3 h after the sodium arsenite exposure (Recovery). First, we selected several genes associated with chromatin P‐p38 and plotted their expression following sodium arsenite treatment from the available RNA‐seq (Fig. 3A). Next, we analyzed the expression of the same genes after incubation with H₂O₂ using qPCR (Fig. 3A '). To mimic the sodium arsenite recovery stage, we treated S2 cells with control media for 1 h after H₂O₂ exposure (REC). We noted consistent expression patterns for most tested genes: hsp26, hsp68, gadd45, and rho highly increased their expression after oxidative stress, particularly during the recovery stage, whereas the expression of dom, pnr, and Psc remained relatively similar to control in both treatments (Fig. 3A,A'). Given the concordance in transcriptional changes following both oxidative treatments, we subsequently used the RNA‐seq data obtained after sodium arsenite incubation [45] to further explore the putative function of P‐p38 chromatin‐associated genes.

We removed non‐expressed genes and defined 4 different gene sets: (a) genes associated with P‐p38 chromatin peaks in the control (708 genes), (b) genes associated with P‐p38 peaks upon stress (355 genes), (c) genes associated with P‐p38 peaks present in both conditions (67 genes), and (d) genes not associated with P‐p38 (15 070 genes). Then, each set was subjected to complete hierarchical clustering, revealing 3 separated gene clusters based on their expression: genes upregulated pre‐treatment (Cluster 1), genes upregulated post‐treatment (Cluster 2), most of which remain activated in recovery, and genes mainly upregulated in the recovery phase (Cluster 3) (Fig. 3B; Fig. S4). The 3 clusters were observed in the four gene sets, but the proportion of genes belonging to each cluster clearly differed. The percentage of Cluster 1 genes was higher among genes not associated with P‐p38 compared to any other gene set (Fig. 3C), while genes associated with P‐p38 belonged preferentially to Clusters 2 (genes bound by P‐p38 only in the control) and 3 (genes bound by P‐p38 only under stress or in both conditions) (Fig. 3C). These differences point to distinct trends of gene expression based on P‐p38 binding: (a) genes not bound by P‐p38 tended to be downregulated upon stress, (b) genes associated with P‐p38 only in the control were preferentially upregulated post‐treatment, and (c) genes associated with P‐p38 upon stress or in both conditions were mainly upregulated in the recovery stage.

We next performed a Gene Ontology (GO) term enrichment analysis of genes belonging to each cluster. In the set of genes associated with P‐p38 only in control conditions, Cluster 1 was enriched in biological processes that usually occur on hemocytes, such as the formation of cell protrusions, proliferation, migration, differentiation, and immunity‐related processes (Fig. 3D). These findings were consistent with S2 cells originating from a macrophage‐like lineage [55]. On the other hand, Cluster 2 genes were specifically enriched in translation‐related processes, including ribosome assembly and translational elongation (Fig. 3D). Genes from Clusters 1 and 2 that were associated with P‐p38 only in the control were enriched in the MAPK cascade, which includes p38 itself. While genes from Cluster 1 that were not bound by P‐p38 also showed enrichment in the regulation of MAPK signaling, only genes from Cluster 2 showed enrichment in stress‐activated MAPKs (Fig. 3D). Probably due to the smaller number of genes, no relevant enrichment was found for genes associated with P‐p38 upon stress belonging to Cluster 2. Finally, Cluster 3 genes were enriched in multiple processes occurring during the recovery process upon stress exposure, including gene silencing, protein folding, stress response, and immunity‐related terms (Fig. 3D). Interestingly, Cluster 3 genes lacking P‐p38 were not enriched in any of these GO terms, suggesting that P‐p38 may bind preferentially to genes involved in the recovery process.

Although MAPKs, including p38, lack a DNA binding domain, previous studies have reported their potential interaction with chromatin through the binding to chromatin‐associated substrates, such as TFs, RNA polymerase subunits, histone modifying complexes, and ATP‐dependent chromatin modifiers [2]. In mammals, TFs are involved in the anchoring of p38 to the chromatin [52]. Because similar mechanisms may exist in Drosophila, we searched for TF motifs within the P‐p38 chromatin peaks associated with genes in each cluster (Fig. 3E). P‐p38 peaks from Cluster 1 genes contained a small number of motifs mostly found in peaks from other clusters. In contrast, P‐p38 peaks from genes of Clusters 2 and 3 showed enrichment in specific motifs, with clear differences between control and stress conditions. TFs enriched exclusively in stress‐related P‐p38 peaks included pannier (pnr), a gene of the GATA family, whose members are known for their role in the C. elegans response to oxidative stress [56], and Ets at 21C (Ets21C), which coordinates a regeneration‐specific gene regulatory network activated by apoptosis in Drosophila imaginal discs [57]. The burst of ROS resulting from injuries in these discs is known to be required for the activation of the JNK and p38 signaling pathways, necessary for regenerative growth [58].

Several studies have investigated the relationship between p38 and chromatin [22, 25, 52] but, to our knowledge, this is the first report on chromatin‐associated P‐p38. In yeast and mammalian cells, the early stages of the response to various stimuli, including oxidative stress, have been shown to be highly dependent on p38 [26, 27]. Hog1, the yeast homolog of p38, controls stress responses by regulating the expression of genes required for a rapid stress response and for the adaptation to future exposure [59, 60]. Similarly, in Drosophila S2 cells, we found that P‐p38 in chromatin was primarily associated with genes that were upregulated after stress exposure and during recovery. Furthermore, a fraction of the chromatin P‐p38 is associated with genes involved in major S2 cell functions, such as the formation of cell protrusions, migration, or differentiation. This indicates that, similar to mammalian cells [18, 20, 29], Drosophila P‐p38 also plays a role in cellular processes unrelated to stress. Finally, as previously reported for p38 [18, 25, 26, 27, 28, 29] we found that P‐p38 binds to both up and downregulated genes upon exposure to stress, which suggests that the phosphorylated form of p38 may also be involved in the activation and repression of transcription.

Author contributions

CC‐R, PC‐C, and MC conceived and designed the experiments. CC‐R, PC‐C, and PL‐G performed the experiments. CCK pre‐processed the sequencing data. CC‐R, PC‐C, and PL‐G analyzed the data. CC‐R, PC‐C, and MC wrote the manuscript. CC‐R, PC‐C, PL‐G, FS, and MC revised and edited the manuscript. FS and MC acquired the funding.

Peer review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1002/1873‐3468.15006.

Supporting information

Fig. S1. Survival curve of S2 cells in control conditions and after exposure to 5 mM, 10 mM, and 20 mM H₂O₂.

Fig. S2. Analyses of P‐p38 ChIP‐seq peaks.

Fig. S3. Analyses of H3K4me3 ChIP‐seq peaks.

Fig. S4. Expression analyses of genes bound by P‐p38.

Table S1. List of P‐p38 peaks and coordinates.

Table S2. List of H3K4me3 peaks and coordinates.

Table S3. List of oligonucleotides used in this work.

Table S4. List of differentially expressed genes.

Data accessibility

Raw and processed ChIP‐seq and RNA‐seq data from this study have been submitted to GEO under the accession number GSE220761. Processed BigWig files used to generate the plots in Figs S2 and S3 are available in Zenodo repository under the DOI number 10.5281/zenodo.13292560.