ATAC-seq and RNA-seq libraries were prepared from three human bone marrow-derived MSCs (BM-MSCs) treated with or without 10 ng/ml PDGF-BB for 48 hours. Changes in chromatin accessibility in response to PDGF-BB were analyzed using assay for transposase accessible chromatin with sequencing (ATAC-seq) and integrated with RNA-seq data. An average of 30 million high quality RNA-seq reads were mapped to the human reference genome, assembly hg38. Open chromatin regions from each sample were combined into a merged region set, resulting in a total of 248,171 regions. Our findings showed that these regions were located within 3 million base pairs around the transcription start site (TSS) of genes.
To visualize differences between the PDGF-BB treated and non-treated MSC samples, we performed principal component analysis (PCA) using normalized counts of the open chromatin regions (Figure S1A) and of the RNA-seq reads (Figure S1B). PCA results of both ATAC-seq and RNA-seq presented the same trend for each MSC sample. In both plots, PDGF-BB-treated and control cells from the same donor clustered together, suggesting that MSC cell origin was strongly correlated with donor-specific chromatin and gene expression patterns.
PDGF-BB treatment changes chromatin accessibility in BM-MSC
Differential accessibility analysis was performed using edgeR (14, 15). We identified a total of 37,548 differential accessibility regions (DAR) at FDR-adjusted p value < 0.05 following PDGF-BB treatment of MSCs (Fig. 1A). Among these regions, 19,973 regions display increased accessibility, while 17,575 regions display decreased accessibility. Examples of increased and decreased chromatin accessibility are shown in Figs. 1B and 1C.
Approximately 17% of all chromatin-accessible regions were in promoter regions of genes (defined as the region between 5kb upstream and 1kb downstream of the transcription start site), ~ 40% of all open regions were in gene body regions, and the remaining 44% were in intergenic regions. Among all the regions with significant changes in chromatin accessibility following PDGF-BB treatment, there was a reduction in the percentage of open regions associated with promoters. Conversely, the percentage of open chromatin regions increased in gene body and intergenic regions (Fig. 1D). This finding suggests that changes in chromatin accessibility occur in both promoter and non-promoter regions, with a higher frequency observed in the non-promoter region.
RNA-seq analysis indicates that PDGF-BB treatment of bone marrow-derived MSCs promote expression of genes involved in cell cycle and cell proliferation
Differential expression analysis detected 2,478 significantly upregulated and 2,482 significantly downregulated genes in the PDGF-BB treated MSCs compared to control (Fig. 2A, Table S1-S2) at FDR-adjusted p value < 0.05. Functional analysis of differential expression genes using hallmark gene sets further revealed that up-regulated genes were enriched in pathways that included cell cycle, cell division, DNA repair, mTOR signaling, IL2-STAT5 signaling, and KRAS signaling. Genes exhibiting down-regulation were found to be enriched in pathways including NFKB signaling, hypoxia, early estrogen response, epithelial mesenchymal transition, P53 signaling, and angiogenesis. (Fig. 2B). Significant enrichment of both up-regulated and down-regulated genes was noted in relation to exosome-related genes (Fig. 2C), indicating the important roles of PDGF-BB signaling in cell to cell communication.
Although numerous genes within the hallmark angiogenesis gene set show down-regulation, several angiogenic growth factors, cytokines, and chemokines exhibit up-regulation. This includes vascular endothelial growth factor C (VEGF-C), hepatocyte growth factor (HGF), chemokines like KC (keratinocyte chemoattractant), FGF-1, IL-8, MMP1, and insulin-like growth factor-binding proteins (IGFBP)-1.
Motif enrichment analysis reveals that changes in chromatin accessibility after PDGF-BB treatment are associated with the bindings of AP-1 family transcription factors, TEAD, CEBPA, and RUNX2.
To gain a better understanding of what regulatory pathways are altered in MSCs after PDGF-BB treatment, we investigated the enrichment of motifs related to transcription factors (TFs) within the differentially accessible chromatin regions. Using all genomic regions as the background, we performed enrichment analyses for all accessible regions using HOMER (16). Our findings revealed a comparable set of enriched motifs, consistent with those illustrated in Ho et al. (17).
The TF motif enrichment in the more accessible regions in the PDGF-BB treated samples (Fig. 3A) indicated that most of the enriched transcription factors belong to the AP-1 family. Many of these transcription factors are known to regulate the expressions of the genes which are associated with cell proliferation, differentiation, cell apoptosis, and response to various stimuli such as cytokines, growth factors, and stress signals. Notably, transcription factors ATF3, FOS, FOSL1, FOSL2, JUN, and JUNB were significantly enriched, with p values < 10-1000. Most of these transcription factors exhibited high expression levels, and some also demonstrated differential expressions. When combined with motif enrichment analysis of differentially accessible regions in PDGF-BB treated samples, this observation underscores a significant regulatory impact exerted by these transcription factors.
The TF motif enrichment in the less accessible regions in PDGF-BB treated samples (Fig. 3B) showed that the AP-1 family members JUN and ATF4 were enriched. It suggests that the transcriptional program in PDGF-BB treated samples may be orchestrated primarily through the AP-1 family of transcription factors, which possibly mediate this regulation by changing the chromatin accessibility landscape. In addition, transcription factors critical for regulating osteogenesis and adipogenesis such as TEAD, CEBPA, and RUNX2 (18, 19) were enriched in the closed regions.
Integrative transcriptome and chromatin accessibility analysis discloses pathways and processes involved in PDGF-BB treatment effects
To investigate the effects of changes in chromatin accessibility on gene expression after PDGF-BB treatment of MSCs, we integrated the RNA-seq and ATAC-seq of the same samples. We observed that genes possessing increased accessibility in their promoter or enhancer regions exhibited more significant expression changes when comparing PDGF-BB treated samples with controls, and conversely, genes with less accessible promoter or enhancer regions showed smaller expression changes (Fig. 4A). Overlaps of the significant differentially expressed genes and the genes with differentially accessible promoters or enhancers (Fig. 4B) are statistically significant. The change in chromatin accessibility correlated with gene expression (Fig. 4C), which demonstrates the consistency of chromatin accessibility and the gene expression. As a result, we identified 760 genes that were consistently up or down regulated both at the chromatin accessibility and transcriptional levels, which represented a shortlist of PDGF-BB-related candidate genes.
To elucidate the biological pathways dysregulated with treatment of PDGF-BB, we examined the enrichment of hallmark gene sets of the consistently PDGF-BB up or down-regulated candidate genes using hypergeometric test (Figure. 4D). Genes with both up-regulated expression and increased accessibility in PDGF-BB treated samples are enriched in PI3K-AKT signaling, IL2-STAT5 signaling, KRAS signaling, and E2F targets. In contrast, genes with both down-regulated expression and decreased accessibility in PDGF-BB treated samples are enriched for estrogen response, epithelial-mesenchymal transition, and coagulation. This may indicate that PDGF-BB stimulates PI3K-AKT signaling, IL2-STAT5 signaling, KRAS signaling, and E2F targets to regulate gene expression.
In a previous study of PDGF effects on MSCs, it was reported that distinct pathways with opposing actions were activated by PDGF. The PI3K/Akt signaling was identified as the main contributor to MSC proliferation in response to PDGFRα activation. Additionally, activation of Erk by PDGFR signaling strongly inhibited the adipocytic differentiation of MSCs by blocking PPAR and CEBP expression(20). The induced Akt and Erk pathways regulate opposing fate decisions of proliferation and differentiation (21). In our study, the PI3K/Akt signaling was not enriched based on gene expression changes alone (Fig. 2B). However, when considering chromatin accessibility changes, it became apparent that this pathway is enriched in more accessible, up-regulated genes. This indicates that changes in chromatin accessibility were a consequence of the regulation in the PI3K/Akt signaling pathway.
PDGF-BB treatment changes gene expression and chromatin accessibility patterns of key genes in MSCs and in osteogenic, adipogenic, and chondrogenic differentiated cells
To further elucidate differential potential of MSCs after PDGF-BB treatment, we investigate the gene expression and chromatin accessibility changes of key marker genes of MSCs, osteogenic, adipogenic, and chondrogenic cells. There are mixed results, while some marker genes are up-regulated, some are not changed or down-regulated. This indicates the diverse effects of PDGF-BB treatment on MSCs for the selected marker genes. The majority of examined adipogenic marker genes demonstrate elevated gene expression and increased accessibility. In contrast, most examined osteogenic marker genes show decreased expression under our experimental conditions.
We examined the transcriptional and chromatin accessibility changes associated with the differentiation of PDGF-treated MSCs towards adipocytes compared to osteoblasts (Fig. 5A). Genes selectively associated with osteogenesis, adipogenesis, and those exhibiting characteristics of both osteogenic and adipogenic differentiation were derived from Rauch's research (22). In their study, it was found that undifferentiated MSCs express osteoblast-selective genes at higher levels than adipocyte-selective genes, as demonstrated by bulk RNA-seq and single-cell RNA- seq. After differentiation, the expression of osteoblast genes was higher compared to the expression of adipogenic genes. Overall, these data demonstrate that osteogenesis involves the induction of numerous genes already active in MSCs, while adipogenesis involves the downregulation of several MSC genes along with the significant upregulation of a substantial group of genes not active in MSCs. The study's analysis, presented in Fig. 5A, compares the ratio of changes in chromatin accessibility and transcriptional expression among selective gene groups. It reveals that the percentage change of osteogenic selective genes is more significant than that of adipogenic selective genes. This trend is observed not only in osteogenic selective genes but also in both osteogenic and adipogenic selective genes, as well as all genes collectively. These findings suggest that osteogenic selective genes experience more comprehensive changes overall in response to PDGF-BB treatment compared to adipogenic selective genes, encompassing alterations in all genes examined.
The transcription factor (TF) – chromatin regulatory network following PDGF-BB treatment reveals key factors involved transcriptional and chromatin accessibility changes
The regulatory network was inferred using PECA (23) from both RNA-seq and ATAC-seq data (Fig. 6A). Among the major TFs in the network, their target genes are enriched for extracellular matrix, ossification, blood vessel development, the WNT signaling pathway, the BMP signaling pathway, ECM-receptor interaction, the PI3K-Akt signaling pathway, focal adhesion and paracrine signaling- important processes or functions within MSCs (Fig. 6B).
Several transcription factors (TFs), including CEBPB, CEBPD, ETS1, FOXC1, FOXO3, JUN, MAFF, MAFG, MAFK, TCF7, TEAD1, and TEAD3, have been identified. Their predicted target genes are significantly associated with the extracellular matrix. This finding underscores the intricate processes and functions that are involved in the regulation of the extracellular matrix, highlighting the diverse roles these TFs play in their modulation and maintenance.
Similar to earlier findings from motif enrichment analysis, TEAD and CEBP transcription factors are predominantly featured within the control-specific network. The downregulation of these factors has a consequential impact on their target genes, subsequently influencing numerous biological processes, including ECM-receptor interactions and the PI3K-Akt signaling pathway.
The target genes of TCF7 are implicated in a wide array of biological functions, including protein digestion and absorption, ECM-receptor interactions, response to BMP, cellular response to growth factor stimulus, PI3K-Akt signaling pathway, and the transforming growth factor beta (TGF-β) receptor signaling pathway. This highlights the multifaceted role of TCF7 in regulating critical pathways that contribute to cellular processes and responses.