Epigenetic landscape of intestinal cell line HT29 cocultured with Lacticaseibacillus

Abstract

Aim: To investigate the changes in epigenetic landscape of HT29 cells upon coculture with the Lacticaseibacillus.

Materials & methods: Histone and m6A mRNA modifications were examined by biochemical and NGS-based methods including western blotting, colorimetric assays, ChIP-Seq and direct mRNA sequencing. LC-MS was performed to identify Lacticaseibacillus secretome.

Results: In cocultured HT29 cells global enrichment of H3K9ac and H3K4me3 and depletion of H3K9me3 mark was observed; mean genic positional signals showed depletion of H3K9ac and H3K4me3 at the TSS but enrichment in the upstream region; m6A methylation was altered in mRNAs corresponding to specific gene pathways; Lacticaseibacillus HU protein interacts with histone H3.

Conclusion:  Lacticaseibacillus can epigenetically alter specific genetic pathways in human intestinal cells.

Plain Language Summary

Lactocaseibacillus, considered as a good bacterium, is present in human gut and helps in maintaining good health of an individual. In this study, we have examined how this bacterium influences the regulation of gene expression in the intestinal cells. We observed that L. rhamnosus alters the packaging of DNA into chromatin by altering histone modifications and methylation of adenine residues in the mRNA molecules. This was found to be correlated with interaction of Lactocaseibacillus histone-like protein, HU, with histone H3 in the intestinal cell nucleus.

Plain language summary

Article highlights.

• Global and local distribution of activating and repressive histone modifications in HT29 cells are altered upon coculture with L. rhamnosus.

• mRNAs in control and cocultured HT29 cells are differentially methylated for m6A.

• Histone modifications and m6A mRNA methylation acts in conjunction to differentially regulate gene expression in cocultured HT29 cells.

• Epigenetic effector proteins like histone-like protein, HU, are components of L. rhamnosus secretome.

• L. rhamnosus histone-like protein, HU, has the capability of interacting with histone H3 inside HT29 cell nucleus.

1. Background

Microbes form an important part of our environment, coexisting with humans and other organisms in a commensal or pathogenic relationship. Most human-microbe studies have focused on the interaction of pathogenic bacterium with human cells, but it is becoming quite evident, especially from microbiome studies, that constant interaction of commensal bacteria with human cells has the ability to alter the cellular functions including immune response, nutrient availability, etc. [1]. However, the mechanisms underlying this crosstalk of human cells with commensals remain largely unknown.

Modulation of host cellular functions by epigenetic modifications including DNA and RNA modification and histone modifications is well established in response to environment. Recent studies have also shown the role of mRNA modification particularly the cross talk of m6A mRNA methylation and chromatin modifications in regulation of cellular functions [2]. Previous research from our group has identified secretory mycobacterial factors including Rv2966c (a DNA methyltransferase) and Rv1988 (a histone methyltransferase) modulate the host epigenome [3,4]. Work from other laboratories has also shown that epigenetic mechanisms play an important role in the alteration of the host cellular functions [5,6]. Commensal bacteria like Lacticaseibacillus and Bifidobacterium, present in our gut, have been shown to play role in drug metabolism, maintenance of gut mucosal barrier, protection against pathogens, immunomodulation, xenobiotic, nutrient metabolism, etc. [7]. Bacteria of the genus Lacticaseibacillus, have been shown to regulate expression of specific genes in epithelial cells and vaginal keratinocytes [8]. These changes have been correlated with gene-specific alteration in DNA and histone modifications [9]. A few groups have also studied global DNA methylation in human intestinal cell lines upon incubation with these bacteria [10–12].

Bacteria of the genus Lacticaseibacillus unlike mycobacteria do not enter the host cell and remain in the extracellular environment or adherent to cell membrane during their interaction with the host cell. Therefore, if any of the bacterial proteins were to have the capability of interacting with the host epigenome, they would not only have to be secretory in nature and be present in the extracellular milieu of the host cell but also have the capability to enter the host cell and its nucleus. In the present study, we set out to identify epigenetic changes in the host cell upon coculturing with the commensal bacteria Lacticaseibacillus rhamnosus, and putative secretory epigenetic modifiers produced by the bacteria.

2. Materials & methods

2.1. Bacterial strains & culture

Bacterial strain used in the study was L. rhamnosus GG. Cultures were grown in MRS (De Man, Rogosa and Sharpe) broth as complete medium at 37°C under anaerobic conditions, following approved IBSC guidelines (CSIR-IMTECH/IBSC/2021/June13). RPMI was used as a minimal medium for secretome studies.

2.2. Mammalian cell culture & transfection

HT29 cells were cultured in DMEM (Gibco 11995-065) with 10% FBS (Gibco 16000-044) and antibiotics (Gibco 15240-062) at 37°C in 5% CO₂. HEK293T cells were cultured in DMEM and transfected with pCDNA-SFB3.1 using PEI in Opti-MEM (Gibco-11058021).

2.3. HT29 coculture with L. rhamnosus

L. rhamnosus bacterial cells (1 × 10⁹) grown till mid log phase were resuspended in antibiotic free DMEM and further cocultured with HT29 cells (grown till a confluency of 70–80%) for 3 h. Before the coculture, HT29 cells were washed with PBS to remove the traces of antibiotics in the culture medium.

2.4. Secretome analysis

For the secretome analysis of L. rhamnosus alone, the bacteria were cultured in RPMI (used as a minimal media). For both, bacteria grown in minimal media as well as upon coculture with HT29 cells, the culture media (supernatant) was collected after harvesting the bacterial cells at low speed (5000 rpm). The supernatant was then again centrifuged at 10,000 rpm to remove cell debris and the clear supernatant obtained was then passed through a 3KDa cutoff Amicon filter and concentrated to 500 μl. The concentrate was then washed with PBS to remove excess salts and electrophoresed on SDS PAGE. The CBB stained gel slice was cut and analyzed by LC-MS at Taplin facility, Harvard University.

2.5. Pull down assay

To examine possible interactome of Lacticaseibacillus HU protein in human cells, recombinant His tagged HU was used as bait protein, incubated with HT29 cell lysate and interacting partners were determined by Ni-NTA bead-based pull-down followed by mass spectrometric analysis. Similar strategy was applied to identify L. rhamnosus proteins interacting with host histone H3. To further validate the interaction of Lacticaseibacillus HU with histone H3, HEK293T cells expressing SFB-HU were lyzed in NETN buffer at 4°C for 30 min. Lysate was centrifuged at 10,000 rpm at 4°C to remove cell debris. The clear supernatant was incubated with streptavidin beads for 4 h at 4°C with gentle mixing. Proteins bound to the beads were washed with NETN buffer, electrophoresed on SDS-PAGE followed by western blotting with FLAG. To check for interaction with histone H3, the blot was probed with H3 antibody.

2.6. Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was carried out following Jägle et al. protocol with slight modifications [24]. Briefly, control or cocultured HT29 cells were cross linked with 1% formaldehyde solution at RT for 10 min followed by quenching with 125 mM glycine for 10 min at RT with gentle shaking. Cells were washed three-times with ice cold PBS and harvested by scraping. Cells were centrifuged at 1200 rpm for 10 min at 4°C. The pellet was resuspended in 1 ml sonication buffer (50 mM HEPES pH 8.0, 140 mM NaCl, 1 mM EDTA, 1% Triton-X100, 0.1% sodium-deoxycholate, 0.1% SDS, 1X protease inhibitor cocktail) and kept on ice for 30 min followed by sonication using Diagenode bioruptor (30 s ON/OFF cycle, medium intensity for 15 min followed by 30 s ON/OFF cycle at high intensity for 2.5 min). Sonicated cells were centrifuged at 12,000 rpm for 10 min at 4°C to remove cell debris. Supernatant was pre-cleared using 20 μl of magnetic A/G beads (Biorad) for 2 h at 4°C. The pre-cleared supernatant was then divided into input (10%), sample for ChIP using IgG antibody (45%) and sample for ChIP using histone modification antibody (45%). Pulldown was carried out using the desired antibody at 4°C overnight on a hula mixer and the input fraction was stored at -20°C. 50 μl of magnetic A/G beads were added to the ChIP samples and were allowed to bind for 4 h at 4°C. Beads were washed 4X for 5 min with sonication buffer, followed by wash buffer A (50 mM HEPES pH 8.0, 500 mM NaCl, 1 mM EDTA, 1% Triton-X100, 0.1% sodium-deoxycholate, 0.1% SDS, 1X protease inhibitor cocktail), wash buffer B (20 mM Tris-HCl pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5% sodium-deoxycholate, 0.5% NP-40, 1X protease inhibitor cocktail) and finally TE buffer (10 mM Tris, 1 mM EDTA). All the washes were carried out at 4°C on the hula mixer. 150 μl elution buffer (50 mM Tris pH8.0, 1 mM EDTA, 1% SDS, 50 mM NaHCO₃) was added and kept at 80°C for 15 min at 1400 rpm on a shaking mixer. Elute was collected in a fresh tube and the elution step repeated. Both the eluates together and the input fraction were reverse cross-linked by adding 20 μg RNase A and 300 mM NaCl and incubation at 67°C for 5 h followed by proteinase K digestion at 45°C for 1 hour. The DNA in the samples was purified by phenol-chloroform method and stored in nuclease free water at -80°C.

2.7. RNA isolation, colorimetric estimation of m6A levels & qRT-PCR

RNA isolation from control and cocultured HT29 cells was done by Qiagen RNeasy kit (Qiagen 74106). RNA was eluted in 40 μl of RNase free water and stored at -80°C. mRNA was isolated from total RNA using NEXTFLEX Poly (A) beads (PerkinElmer – Nova-512980) kit, strictly following the manufacturer protocol. 300 ng of purified total RNA or mRNA was used for colorimetric estimation using m6A RNA methylation Quantification kit (colorimetric) (Abcam ab185912). For qRT-PCR analysis, 2 μg of total RNA was converted to cDNA using random hexamers with SSRTIII (Invitrogen 18080051) and qRT-PCR was performed using DyNAmo ColorFlash SYBR Green qPCR Kit (Invitrogen F416L) and fold change was calculated using ΔΔCt method.

2.8. Next generation sequencing

RNA-Seq and ChIP-Seq library preparation and sequencing was performed on Illumina platform at CDFD or CCMB, Hyderabad. For m6A mRNA methylation sequencing, libraries were prepared and examined on GridIon Nanopore platform (using Direct RNA sequencing SQK-RNA-002 kit) following manufacturer's instructions.

2.9. RNA-Seq Data analysis

FastQC and MultiQC was performed on raw files to determine the quality of the reads. Adapters were trimmed using Trimmgalore (https://github.com/FelixKrueger/TrimGalore) and reads were aligned using STAR [25] using standard parameters. Transcript counting was done using featureCounts and differential gene expression analysis was performed using DESeq2 [26]. Heatmap and Volcano plot were generated using R, cluster mapping was done using STRING and visualized using Cytoscape [27]. Pathway enrichment was carried out using ShinyGO v0.77 [28]. Venn diagrams were generated using Venny2.1.0.

2.10. ChIP-Seq data analysis

FastQC and MultiQC was performed on raw files to determine the quality of the reads. Adapters were trimmed using fastP and alignment to Hg38 reference genome was carried out using bowtie2 [29] with default parameters. Peaks were called using MACS2 with broad peak parameter. Pile up values from the MACS2 data was used for downstream analysis using ChIP-Seeker [30]. Motif analysis was carried out using MEME-ChIP suite on galaxy server [31] and transcription factor binding site prediction was done using Factorbook [32]. Enrichment across transposable elements was determined using T3E (available at Github repository https://github.com/michelleapaz/T3E). Deeptools coverage feature was used to generate bigwig coverage files for individual alignment (bam) files. The bigwig files for all the replicates of control and cocultured samples were merged using IGV merge tracks feature. The merged bigwig file for control was subtracted from merged cocultured bigwig file using IGV merge tracks feature, bigwig track file depicting the increment or decrement of histone modification enrichment across the genomic regions was generated. The regions with positive enrichment were highlighted as green and regions with lower enrichment were highlighted as red and visualized on IGV.

2.11. m6A RNA methylation sequencing analysis

Raw fast5 files were basecalled using Guppy and the fastq files generated were used for detection of modified adenines by ELIGOS2 [33]. Bedtools makewindow feature was used to make 1Mb non-overlapping windows in human genome sequence and bedtools coverage feature was used to determine the number of m6A residues in those 1Mb non-overlapping windows [34], followed by Circos plot generation using command line version of circos version 0.69–9 [35]. Bedtools getfasta feature was used to determine the fasta sequence and used for motif enrichment analysis carried out using BaMMotif version 1.4.0 (https://bammmotif.soedinglab.org/). Dotplot was generated using ggplot in R.

2.12. FITC labelling

4 mg/ml solution FITC (Sigma) in bicarbonate buffer (pH 9.0) was prepared and incubated with 200 μg of purified Lacticaseibacillus HU or BSA (NEB) for 4 h at 4°C with gentle mixing. The labelled proteins were dialyzed against PBS using 3KDa cutoff membrane overnight at 4°C. HT29 cells seeded in 4 chambered slides were incubated with labelled protein for 15–60 min. Cells were washed gently with PBS and fixed with 4% paraformaldehyde for 10 min at RT and imaged using confocal microscopy. Nucleus was stained with DAPI. All the steps were performed under dark conditions.

3. Results

3.1. Coculture of L. rhamnosus with HT29 cells modifies histone modification profile

3.1.1. At global level

To determine if coculture with L. rhamnosus affects chromatin organization in HT29 cells, we examined the overall levels of a few histone H3 modifications including H3K9me3, H3K9ac and H3K4me3 by western blot analysis and quantified by densitometry analysis. Enrichment of the activating histone marks H3K9ac and H3K4me3 and a decrease in repressive H3K9me3 histone marks in HT29 cells upon coculture with L. rhamnosus (hence forth referred as cocultured HT29 cells) was observed (Figure 1A & B).

Figure 1.

Global histone modification levels and mean genic differential enrichment profile in HT29 cells. The level of histone modifications H3K4me3, H3K9ac and H3K9me3 was examined by western blot analysis (representative blots shown in (A) and the levels were calculated by densitometry scanning of the western blot for control and cocultured HT29 cells (B). Expression levels of histone H3 was used as loading control. Densitometry analysis was performed for at least 3 biological replicates. Densitometry analysis was performed using ImageJ. Student's t-test was used for statistical analysis. Error bars represent standard deviation (SD). * indicates p-value < 0.05. Distribution profiling of enrichment peaks from ChIP-Seq data for H3K4me3 (brown and red), H3K9ac (dark blue, blue) and H3K9me3 (dark green and green) across different genomic and non-genomic regions was analyzed by ChIP-Seeker (an R package) for mean genic positions. The enrichment of these histone modifications was plotted for (C) Transcription Start Site (TSS), (D) promoter and upstream region and (E) Transcription Termination site (TTS) for control (brown, dark blue, dark green) and cocultured HT29 cells (red, blue, green). (F) The enrichment of these histone modifications for 5′UTR, exons, introns and 3′ UTR within the gene proper.

3.1.2. ChIP Seq analysis

As we observed changes in H3K9me3, H3K9ac and H3K4me3 histone modifications at the global level, genome-wide distribution profiling was undertaken by ChIP-Seq experiment (raw data has been uploaded to NCBI SRA, project ID PRJNA1096287). Similar number of genes in control and cocultured HT29 cells were enriched for the three modifications (Supplementary Table S1). Around 3–4% of the genes were differentially modified for the active histone marks H3K4me3 and H3K9ac. For the inactive-specific histone mark H3K9me3 approximately 10% genes were differentially modified (Supplementary Figure S1).

An interesting dichotomy in the enrichment of the active histone modifications was observed when the mean genic positional signals for the histone modifications were calculated and plotted for all the human genes in control and cocultured HT29 cells. The average enrichment of active chromatin-specific histone modifications H3K9ac and H3K4me3 at the Transcription Start Sites (TSS) was lower in cocultured HT29 cells as compared with control (Figure 1C). However, the mean genic positional signals in the upstream region (TSS to -10 Kb) especially between -2 Kb and -8 Kb for TSS, H3K9ac and H3K4me3 were significantly enriched in the cocultured HT29 cells. The average enrichment for H3K9ac and H3K4me3 enrichment at around -10 Kb upstream of TSS mirrored the profile of TSS and was lower in cocultured HT29 cells in comparison to control (Figure 1D).

Analysis of mean genic positional signals for H3K9me3 indicated that it was poorly represented at TSS in both the conditions. However, in the promoter region (0 to -3 Kb upstream of TSS) and up to +3 Kb downstream of the TSS, the H3K9me3 profile in cocultured HT29 cells seemed to be a mirror image of the control HT29 cells. In the near promoter region (0 to -1.5 Kb) and gene body region closer to the TSS (0 to 1.5 Kb), H3K9me3 was higher in cocultured HT29 cells but in the far promoter (-1.5 to -3 Kb) and middle region of the gene body the enrichment was lower than the control (Figure 1C). The same was evident in heat map analysis for all these three histone modifications (Supplementary Figure S2). All the three histone modifications were poorly enriched near Transcription Termination Site (TTS, Figure 1E) in both the conditions.

Histone modification peak distribution was also examined specifically for exons, introns and UTRs (Figure 1F). We observed a mirror image of peak distribution for H3K9me3 across 5′UTR, 3′UTR and exonic regions in cocultured HT29 cells as compared with control. No appreciable differences were observed for intronic region. For 5′UTR, 3′UTR and intronic regions, H3K9ac and H3K4me3 peak distribution was found to be similar in both the control and cocultured HT29 cells. The enrichment toward the mid exonic region was lower in cocultured HT29 cells for both H3K9ac and H3K4me3.

The human genome has several overlapping genes which could imply that promoter region for one gene could overlap with the CDS of other gene or CDS of one gene could overlap with intergenic region for some other gene. We plotted Upset plot for the peak distribution data to determine the overlap of peaks in different genomic regions. For H3K9me3, majority of peaks in both the conditions belong to intergenic and distal intergenic regions. On the other hand, for H3K9ac and H3K4me3, majority of peaks belong to genic and intronic region (Supplementary Figure S3).

The differential histone modifications across major class of transposable elements (TEs) including LINE, SINE and LTRs was also examined in cocultured HT29 cells. A few LTRs showed approximately twofold enrichment for H3K4me3 and H3K9ac in cocultured samples. H3K9me3 enrichment was observed in the other DNA transposon elements in cocultured samples (Supplementary Figure S4 & Supplementary Table S2). This could imply that apart from directly regulating the gene expression by differential binding at their gene body and promoters, differential regulation of TEs could also be one of the mechanisms to regulate the gene expression in cocultured HT29 cells.

To examine the possibility that the histone modifications might be getting enriched in the same gene but at different genomic regions leading to differential regulation, the differential distribution of various histone marks for a few genes upregulated (ATF2, CYP1B1, FZD5, ADRB2) and downregulated (MYC, MAP2K3, FADD, TNFRSF10A) in cocultured HT29 cells (see section on HT29 cell transcriptome and Supplementary Table S3) was plotted in the Integrative Genome Viewer (IGV). The representative plots showing the region of enrichment (green peaks) and depletion (red peaks) of the particular histone modification at the specific genetic loci in cocultured HT29 cells are shown in Figure 2. As expected, for all upregulated genes, H3K4me3 and/or H3K9ac marks at the promoter showed enrichment in cocultured HT29 cells. For ADRB2, H3K9ac was enriched whereas H3K4me3 modification showed depletion. For downregulated genes like MYC and FADD, both H3K9ac and H3K4me3 were depleted from the promoter and 5′ region of the gene around the TSS. Genes like CASP2, TNFRSF10A, which show only minor changes in gene expression, no significant correlation was observed for levels of H3K9ac and H3K4me3.

Figure 2.

Gene-specific IGV snapshots for differential enrichment profiles for the indicated histone modifications in the cocultured HT29 cells as compared with control. IGV snapshots for genes showing: Enrichment of activating histone marks (H3K4me3 & H3K9ac) at their promoters (left panel); Depletion of activating histone marks (H3K4me3 & H3K9ac) at their promoters (middle panel); and with multiple changes in all histone marks as indicated (right panel) in cocultured HT29 cells. Green peaks indicate enrichment and red colored peaks denotes depletion of indicated histone modifications at the particular genomic region in cocultured HT29 cells.

Pathways analysis indicated G-α signaling and O-linked glycosylation related genes were more enriched for H3K9me3 in cocultured HT29 cells. H3K9ac enrichment was observed for Ubiquitination and proteasome degradation pathway genes in cocultured HT29 cells whereas MAPK signaling pathway genes were enriched for H3K4me3 modification in control HT29 cells (Supplementary Figure S5).

Motif enrichment analysis followed by transcription factor binding site prediction revealed that the consensus DNA sequence to which JUN binds as transcription factor was enriched for H3K9me3 in control HT29 cells. Upon coculture with L. rhamnosus, this consensus DNA sequence was enriched with H3K9ac and H3K4me3 histone modifications (Supplementary Figure S6). This could imply that JUN regulated genes might get activated in HT29 cells upon coculture. AHRR, one of the genes regulated by JUN showed overexpression upon coculture.

3.2. Coculture with L. rhamnosus affects m6A levels in HT29 cell transcriptome

3.2.1. In total RNA

Along with histone modifications, recent studies have also reflected the role of RNA epitranscriptomics in regulating gene expression [13]. To examine whether m6A methylation profile of the RNA in HT29 cells cocultured with L. rhamnosus shows any changes as compared with control cells, colorimetric analysis of m6A in total RNA was performed. The analysis revealed a significant increase (37%) in m6A content in the total RNA of cocultured HT29 cells (Figure 3A).

Figure 3.

Differential m6A mRNA methylation. (A) Bar plots showing quantitative colorimetric analysis of global levels of m6A modification in total RNA and mRNA in control and cocultured HT29 cells for 3 or more biological replicates. Student's t-test was used for statistical analysis. * indicates p-value < 0.05. Error bars represent standard deviation (SD). (B) Venn diagram showing count of genes whose mRNAs show m6A methylation in control and cocultured HT29 cells. (C) Circos plot showing mapping of differential m6A mRNA methylation in 1Mb non-overlapping windows of human genome categorized by chromosome numbers. Green and red indicate genomic regions with higher and lower m6A mRNA methylation respectively in cocultured HT29 cells as compared with control. (D) Motif enrichment analysis performed for sequences enriched with m6A mRNA methylation in control and cocultured HT29 cells. The conserved m6A DRACH motif is shown in the middle.

3.2.2. In mRNA

Since mRNA and its modifications have been shown to directly regulate gene expression [14], the levels of m6A modification in mRNA was examined. As can be seen in Figure 3A, we observed a small but statistically significant decrease of around 10% in m6A modification in the mRNA isolated from cocultured HT29 cells. Since mRNA constitute only 10–15% of total RNA population, this decrease in m6A in mRNA could have been masked by a considerable increase of m6A in rRNA, the most abundant RNA species.

3.2.3. m6A-Seq analysis

Next, we sought to determine the sites at which differential m6A modification occurs in the mRNA population by direct RNA sequencing using Nanopore technology. Excluding mitochondrial genes and genes derived from the contigs in Hg38 construct, 7313 m6A sites in control and 7186 m6A sites in cocultured HT29 cell mRNA were identified (Supplementary Table S4). The enrichment of m6A sites mapping to the same mRNA species was calculated using bedtools. The 7313 m6A sites in control HT29 cell were present in mRNAs coded by 2959 genes whereas 7186 m6A sites in cocultured HT29 cell mapped to 2859 genes. For 627 and 527 unique genes, m6A mRNA methylation was present exclusively either in control or cocultured HT29 cells respectively (Figure 3B). 2332 genes showed m6A mRNA methylation in both the samples. Using bedtools coverage feature, mRNA species which are hypo- or hypermethylated for m6A modification in cocultured HT29 cells were identified (Supplementary Table S5). The top 25 hypo- and hypermethylated genes are shown in Supplementary Table S6. Furthermore, m6A coverage for mRNAs mapping to different genes in 1 Mb non-overlapping windows was calculated across the various human chromosomes as described in a flow chart in Supplementary Figure S7. As shown by Circos plot in Figure 3C, a number of mRNA species mapping across various chromosomes showed significant differential m6A methylation.

Gene enrichment analysis showed that the genes belonging to ribosome, proteasome and spliceosome pathways had m6A enriched mRNAs in both the samples (Supplementary Figure S8). Although spliceosome pathway is enriched in both the cases, few key genes including Prp16, ACINUS, CCDC12, U2A′, SPF45 were enriched in cocultured whereas Prp43, PRL1, AD002, SKIP, PPIL1, magoh, CBP8020 were enriched in control HT29 (Supplementary Figure S9).

In human cells, m6A in RNA is mostly enriched at adenines present in the “DRACH” motif. Motif search across 7bp upstream and downstream sequence of the target adenine residue in the Hidden Markov Model based BaMMmotif search engine showed that m6A enriched motifs in control HT29 cell mRNA was indeed DRACH. However, when m6A enriched mRNA from cocultured HT29 cell was analysed the most enriched motif was not similar to the DRACH motif (Figure 3D). Moreover, only 60% of the m6A sites were same in both the conditions (Supplementary Table S4 & Supplementary Figure S10).

3.3. Correlation of histone modifications, mRNA m6A methylation & gene expression in HT29 cell cocultured with L. rhamnosus

m6A modifications are known to regulate multiple functions in a cell including modulation of mRNA expression and downstream protein expression profile [13]. As epigenetic modifications are known to regulate gene expression, we examined by RNA–Seq based transcriptomic analysis, whether coculture with L. rhamnosus lead to changes in the transcriptional profile of HT29 cells. Using an FDR cut off of <0.05, 1785 differentially expressed genes (DEGs) were identified (Figure 4A). Figure 4B shows the heat map for the top 25 up and down regulated genes. Gene enrichment analysis showed that colorectal cancer, apoptosis and p53 signaling pathway associated genes were differentially expressed (Supplementary Figure S11). String analysis showed clustering for genes associated with chromatin, apoptosis, PPARA, angiogenesis (Figure 4C).

Figure 4.

Differential RNA expression and its correlation with epigenetic modifications. (A) Volcano plot depicting differential gene expression (RNA-Seq data) in cocultured HT29 cells as compared with control. Red dots indicate overexpressed and blue indicate downregulated genes. Dataset with 1785 genes (genes with Ensembl IDs) showing non-zero Log2FC value was used for plot. Names of a few genes showing higher level of differential expression are shown (B) Heat map for the top 25 up and downregulated genes in cocultured HT29 cells. (C) 422 out of 1785 genes that showed differential expression of Log2FC >1 or Log2FC <-1 were selected and used for cluster analysis using STRING, clusters were visualized using Cytoscape. (D) Dot plot depicting the correlation between m6A methylation status of a gene with its expression profile in HT29 cells cocultured with L. rhamnosus. Only 1736 out of 1785 genes in the differential gene expression data sets that were annotated, were taken for the analysis. (E) Validation of differential expression by quantitative RT-PCR. For a few genes, as indicated below the graph, which showed differential expression in the RNA-seq data or showed differential m6A mRNA were selected and their expression level quantified by real-time RT-PCR. Student's t-test was used for statistical analysis. * indicates p-value < 0.05. Error bars represent standard deviation (SD). The data is for 3 or more biological replicates. (F) Correlation of gene expression with histone modifications and m6A mRNA methylation. IGV snapshot showing regions of enrichment (green) or depletion (red) for the indicated histone modifications and m6A mRNA methylation sites (vertical blue lines) for the genes shown in (E).

To examine whether any correlation existed between the gene sets either showing differential m6A mRNA methylation or differential gene expression, genes having both differential m6A mRNA methylation and differential expression were identified. Of the 1736 annotated genes among the 1785 DEGs (Supplementary Table S7), 60 genes showed m6A methylation exclusively in control, whereas mRNAs for 61 genes were exclusively methylated in cocultured HT29 cells. mRNA for 206 differentially expressed genes were methylated in both control and cocultured HT29 cells (Supplementary Figure S12). Scatter plot for Log2FC of differential mRNA expression (X-axis) and differential mRNA m6A modification (Y-axis) indicated that a small subset of genes including CYP1A1 were upregulated as well showed hyper m6A methylation (Figure 4D).

To investigate the correlation between the differential histone modifications and gene expression. The DEGs were classified into four categories based on extent of magnitude of gene expression change (Log2FC <-1, Log2FC -1 to 0, Log2FC 0 to 1 and Log2FC >1) and the histone modification enrichment profile for these genes were plotted for both control and cocultured HT29 cells within 3 Kb upstream and downstream of TSS (Supplementary Figure S13) and across the gene body (Supplementary Figure S14). Downregulated genes (Log2FC <-1 and Log2FC -1 to 0) showed enrichment for H3K9me3 upstream of the TSS in the promoter region in cocultured HT29 cells. Additionally, for these genes, H3K9ac and H3K4me3 modification was relatively less enriched at the TSS in the cocultured as compared with control HT29 cells (Supplementary Figure S13). For upregulated genes (Log2FC 0 to 1 and Log2FC >1), no significant change in H3K9ac and H3K4me3 distribution was observed, however, H3K9me3 modification enrichment was substantially reduced near the TSS (Supplementary Figure S13).

We next tested the correlation between gene expression, histone modifications and m6A mRNA methylation. Based on RNA-Seq & qRT-PCR (Figure 4D & E) a few genes were selected for this analysis. ChIP-Seq data for the various histone modifications and m6A mRNA methylation NGS data was plotted together on IGV for these genes (Figure 4F). CYP1A1 and FOS showed significant increase in their expression levels (Figure 4E) and m6A methylation of their mRNAs exclusively in cocultured HT29 cells (Figure 4F). While CYP1A1 showed significant enrichment for only one of the active marks, H3K9ac, around its TSS and the gene body, FOS showed significant enrichment of both the active chromatin marks H3K4me3 and H3K9ac in its gene body. Both JUND and JMJD8 mRNAs show (i) differential m6A methylation, with cocultured HT29 cells showing a greater number of m6A sites and (ii) decline in the level of active chromatin-specific histone marks, H3K9ac and H3K4me3, in cocultured HT29 cells (Figure 4F). But only JUND showed significant overexpression in cocultured cells (Figure 4E). BAX and BID are both overexpressed and associated with active chromatin marks near their TSS in cocultured cells but do not show any differential m6A mRNA methylation. In fact, BID does not show any m6A methylation in both the conditions.

3.4. Identification of epigenetic effector proteins in L. rhamnosus secretome

L. rhamnosus is a free-living extracellular microbe and interacts with the host cells either by association with the host cell membrane or through secretory bacterial factors [15]. To examine the possibility of L. rhamnosus secretory factor(s) being responsible for the observed epigenetic modifications in the host HT29 cells, we decided to investigate the presence of putative Lacticaseibacillus epigenetic modifiers in its secretome by mass spectrometric based analysis. For the secretome analysis, culture filtrate of (i) L. rhamnosus cultured in minimal media; (ii) and after coculture with HT29 cells, was concentrated and examined by mass spectrometric analysis. Data mining for proteins in the culture filtrate of L. rhamnosus cultured in minimal media (RPMI) revealed 537 proteins (Supplementary Table S8). 58 L. rhamnosus proteins were detected in the culture media of cocultured cells (Supplementary Table S8). Low number of L. rhamnosus proteins were detected probably because of the abundance of human proteins in the culture media of the cocultured cells. The proteins present in the secretome were classified into categories depending upon their function and annotation and the major categories are shown in Figure 5A & B.

Figure 5.

Lacticaseibacillus HU is a secretory protein with capability to enter HT29 cells and interact with histone H3. (A & B) Pie charts showing the major components of L. rhamnosus secretome in minimal media (A) and cocultured with HT29 cells (B) identified by mass spectrometric analysis for 2 biological duplicates. Functional categorization of proteins in the secretome was done manually using protein description obtained from Uniprot database. Only DNA and RNA associated secretome proteins are color labelled. (C) Localization of FITC conjugated Lacticaseibacillus HU protein in HT29 cells. Representative confocal image showing localization of FITC conjugated L. rhamnosus HU protein (green) both in the cytoplasm and nuclei of HT29 cells. Nuclei are stained with DAPI (blue). Scale bar: 5 μm. (D) Cluster analysis for human proteins interacting with L. rhamnosus HU identified by mass spectrometric analysis (for 2 biological replicates). Clusters are visualized using Cytoscape. (E) Representative western blot depicting interaction between L. rhamnosus HU and human histone H3 identified using pulldown assay. First two lanes show Input samples for control (pCDNA-SFB3.1 transfected in HEK293T cells) and SFB-Hup (pCDNA-SFB3.1-HU transfected in HEK293T cells) respectively. PD lanes show pulldown using streptavidin beads for control and SFB-HU (SFB-Hup). Anti-FLAG panel shows the presence of SFB or SFB-HU in the respective input and pulldown lanes.

From the list of secreted proteins, we have highlighted proteins that had the potential to act as epigenetic modifier in the host including proteins capable of binding or epigenetically modifying DNA, RNA or histone proteins in Table 1. It was interesting to find RNA as well as DNA methyltransferases, Histone like protein (HU) and an N-acetyltransferase in the list of secretory proteins.

Table 1.

List of probable epigenetic modifier present in L. rhamnosus secretome.

Gene name	Gene description	Identified by
DNA binding and methylation
C2JWW8	DNA-binding protein HU	Mass spec
A0A179YI80	DNA methyltransferase	Mass spec
C2JZM5	Site-specific DNA-methyltransferase	Mass spec
RNA binding and methylation
A0A249DDR8	RNA methyltransferase	Mass spec
A0A249DF01	Ribosomal RNA small subunit methyltransferase H	Mass spec
C2JW46	23S rRNA (Uracil-5-)-methyltransferase RumA	Mass spec
C2JY45	tRNA (guanine-N (7)-)-methyltransferase	Mass spec
Protein modification
A0A171J9Q7	Serine hydroxymethyltransferase	Mass spec
A0A171J9Q7	Serine hydroxymethyltransferase	Mass spec
A0A249N3M0	Tyrosine-protein kinase CpsD	Mass spec
C2JXU0	Non-specific serine/threonine protein kinase	Mass spec
C2JZF8	Homoserine kinase	Mass spec
C2K1I2	Nucleoside diphosphate kinase	Mass spec
F3MS19	Uridylate kinase (Uridine monophosphate kinase)	Mass spec
A0A2A5L9F2	Histidine phosphatase family protein	Mass spec
C2K1Q5	Ser/Thr phosphatase family protein	Mass spec
K8QF76	dUTP diphosphatase (EC 3.6.1.23)	Mass spec
A0A0J6UM07	GNAT family acetyltransferase	Mass spec
A0A0D6U4V8	N-acetyltransferase (RibT protein)	Mass spec
C2JVL0	S-adenosylmethionine synthase (AdoMet synthase)	Mass spec

3.5. L. rhamnosus HU can enter the nucleus of HT29 cells & interact with chromatin

The mechanism underlying epigenetic changes in the host HT29 cells upon coculture with L. rhamnosus could be due to host-specific epigenetic effectors. However, as has been seen in pathogenic bacteria [3,4], the possibility exists of L. rhamnosus epigenetic effectors proteins being part of this mechanism as well. As seen in Table 1, several epigenetic effector molecules produced by L. rhamnosus were present in the culture medium during bacterial coculture with HT29 cells. For a bacterial protein, produced by an extracellular bacterial species like L. rhamnosus that does not enter the host cell, to act as epigenetic modifier, it should be able to enter the host cell and localize in nucleus; and interact with components of the host epigenetic effector proteins. HU protein was found in the L. rhamnosus secretome and since it regulates bacterial gene expression, it was sought to determine whether HU could interact with chromatin in the host cell and modulate the host epigenetic machinery.

FITC labelled L. rhamnosus HU protein was incubated with HT29 cells and its localization was visualized by confocal imaging. As can be seen in Figure 5C, HU was found to be localized to both cytoplasm and nuclei in HT29 cells. Control FITC-BSA remained localized to HT29 cell membrane.

Next, we sought to identify host cell proteins that could interact with L. rhamnosus HU protein. We prepared HT29 cell lysate and incubated with purified HU (bait protein) and the pull-down sample was analyzed by LC-MS. We found histones, VIRMA/KIAA1429 (m6A RNA methyltransferase complex protein), HMG, RBM and several ribosome and mRNA surveillance pathway associated proteins to be interacting with HU (Figure 5D & Supplementary Table S9). To confirm the interaction of histone H3 with L. rhamnosus HU protein, histone H3 was used as a bait and incubated with L. rhamnosus lysate followed by pull down assay using Ni-NTA beads followed by mass spectrometry analysis. As can be seen in Supplementary Table S9, along with proteins DnaK, RpoA and RpoC (RNA polymerase subunits), DNA binding protein HU was found to be interacting with histone H3. To further validate the interaction of Lacticaseibacillus HU with histones H3, SFB-HU construct was transfected into HEK293T cells, streptavidin beads were used for pulldown of the SFB-HU and Western blotting was performed using histone H3 antibodies. As can be seen in Figure 5E, HU was found to interact with histone H3.

4. Discussion

A few recent studies have revealed the influence of microbiota, through epigenetic mechanisms, on cellular physiology of human cells, be it in the gut, skin, the respiratory tract or the reproductive organs. In this study, we demonstrate epigenetic reprogramming in HT29 colorectal cancer cells upon interaction with a probiotic bacterium, L. rhamnosus. Previous studies have shown the role of bacterial proteins in regulating the host epigenome [3,4]. Considering the potential of L rhamnosus's secreted HU protein to not only enter the intestinal cells but also interact with the host chromatin, our study provides another evidence of modulation of cellular function of host cells by bacterial produced factors through epigenetic mechanisms.

Previous studies have noted reduction in the global H3 and H4 acetylation in the human colon cell line, Caco-2, when treated with L. rhamnosus and L. fermentum in the presence of inflammatory stimulus of E. coli [16]. A few other studies have also examined the DNA methylation or histone acetylation profile of specific genes in human cells, upon treatment with bacteria of various Lacticaseibacillus species [8]. One of the primary goals of this study was to examine the range of epigenetic changes in human cells of intestinal origin cocultured with probiotic L. rhamnosus. The present study provides a comprehensive map of active histone modifications, namely H3K9ac, H3K4me3 and inactive-chromatin associated H3K9me3 along with m6A mRNA methylation in HT29 intestinal cell line cocultured with L. rhamnosus. HT29 cell cocultured with Lacticaseibacillus showed changes at multiple loci across its genome for both active and inactive chromatin-specific histone modifications. Interestingly, analysis of mean genic positional signals across genic and upstream regions for active chromatin-associated histone modifications, H3K9ac and H3K4me3, showed a dichotomy in their enrichment at the TSS and upstream regions. TSS showed lower enrichment whereas upstream region showed higher enrichment in cocultured HT29 cells. On the other hand, inactive chromatin mark, H3K9me3, was poorly enriched at the TSS in both conditions but showed enrichment profiles in control and cocultured HT29 cells that were inverse of the other in the gene body and upstream region. At the global level, m6A mRNA methylation in the cocultured HT29 cells was lower than control. At specific mRNA level, differential methylation was observed for multiple genes in the cocultured HT29 cells. Changes in epigenetic modifications including histone modifications and m6A mRNA methylation are correlated with gene expression [17,18]. Of the 1785 differentially expressed genes, only a very small number of differentially expressed genes showed differential m6A mRNA methylation. This fits well with the knowledge that m6A is not only correlated with gene expression but has multitude of other functions including nuclear transport, interaction with auxiliary factors, pri-miRNA processing etc [19]. Moreover, it also indicates that m6A mRNA methylation is not the only epigenetic factor correlate of gene expression.

These small number of differentially expressed genes that show differential m6A mRNA methylation also showed changes in histone modifications and majorly belong to the apoptosis, spliceosome, tight and gap junction, RNA degradation, mRNA surveillance, cell cycle and colorectal cancer pathways. It has been previously shown that coculture with Lacticaseibacillus inhibits the growth of the colonic carcinoma cell line HT29. FOS and JUND are the AP1 associated transcription factors involved in apoptosis. In this study, FOS and JUND were found to have increased expression, their mRNA was hypermethylated with m6A modification and showed enrichment for H3K9ac and H3K4me3 histone modification across their gene body. Since AP-1 associated proteins regulate apoptosis, it is possible that the commensal bacteria like Lacticaseibacillus induces apoptosis in colorectal cancer cells by epigenetically modulating the AP-1 pathway. MYC or c-MYC belongs to a class of protein that is responsible for tumorigenesis in cancer cells and plays a central role in maintaining their self-renewal. MYC was found to be downregulated with m6A mRNA methylation upon coculture. Studies have shown involvement of CYP1A1 in mediating apoptosis and it can also bring imbalance in cellular homeostasis [20,21]. CYP1A1 mRNA was observed to be hypermethylated with m6A modification, its promoter and upstream region showed higher enrichment for H3K9ac modification and was overexpressed in the cocultured HT29 cells. While the exact underlying mechanism needs to be delineated, our results suggests that the coculturing of HT29 colonic carcinoma cell line with L. rhamnosus leads to epigenetic activation of the canonical apoptotic pathway genes. Further work on examining the influence of Lacticaseibacillus on differentiated intestinal enterocytes would help us understand the influence of gut microbiota on the epigenome of intestinal cells.

Changes in epigenetic modifications and mRNA methylation are concomitant with the presence of bacterial epigenetic modifiers including histone like protein HU, RNA methyltransferase RsmD in the culture medium. Our finding that HU, the histone like protein of L. rhamnosus is a secretory protein corroborates similar findings for other Lacticaseibacillus spp. and adds to the growing list of studies including those from M tuberculosis and Vibrio parahaemolyticus, which have shown function of HU protein in microbe-host interaction [22,23]. Further work would be required to understand the reason for the use of a nucleoid associated protein in inter-species interactions.

5. Conclusion

Using ChIP-seq and direct RNA sequencing based analysis, we show genome-wide changes in the profile of specific histone modifications and differential m6A mRNA epitranscriptome of HT29 intestinal cells cocultured with L. rhamnosus. Concomitantly, we also observed secretion of multiple epigenetic modifiers including histone-like protein HU from L. rhamnosus in these cocultures. The Lacticaseibacillus HU protein was found to have the capability of entering the intestinal cells and localize to both the nucleus and cytoplasm and interacting with histone H3.

Funding Statement

A Sinha is the recipient of Junior and Senior Research Fellowship of the University Grants Commission (UGC) in pursuit of a PhD degree at CDFD. He was registered with Regional Centre for Biotechnology, DBT for his PhD degree. S Khosla is a JC Bose Fellow of the Department of Science and Technology, India and his laboratory is supported by grants from CSIR and DBT, Government of India.

Supplemental material

Supplemental data for this article can be accessed at https://doi.org/10.1080/17501911.2024.2377949

Author contributions

S Khosla conceived the study. S Khosla and A Sinha designed the experiments. A Sinha performed the experiments. S Khosla and A Sinha analyzed the data and wrote the manuscript.

Financial disclosure

A Sinha is the recipient of Junior and Senior Research Fellowship of the University Grants Commission (UGC) in pursuit of a PhD degree at CDFD. He was registered with Regional Centre for Biotechnology, DBT for his PhD degree. S Khosla is a JC Bose Fellow of the Department of Science and Technology, India and his laboratory is supported by grants from CSIR and DBT, Government of India. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.

Data availability statement

Complete ChIP-seq, RNA-seq and m6A methylation sequencing files have been deposited in NCBI's Sequence Read Archive (SRA, https://submit.ncbi.nlm.nih.gov/subs/sra/) and are accessible under the BioProject accession number PRJNA1096287. The mass spectrometric data was provided by Taplin facility, Harvard University.