Integrated gene and miRNA expression analysis of prostate cancer associated fibroblasts supports a prominent role for interleukin-6 in fibroblast activation

Abstract

Tumor microenvironment coevolves with and simultaneously sustains cancer progression. In prostate carcinoma (PCa), cancer associated fibroblasts (CAF) have been shown to fuel tumor development and metastasis by mutually interacting with tumor cells. Molecular mechanisms leading to activation of CAFs from tissue-resident fibroblasts, circulating bone marrow-derived fibroblast progenitors or mesenchymal stem cells are largely unknown. Through integrated gene and microRNA expression profiling, we showed that PCa-derived CAF transcriptome strictly resembles that of normal fibroblasts stimulated in vitro with interleukin-6 (IL6), thus proving evidence, for the first time, that the cytokine is able per se to induce most of the transcriptional changes characteristic of patient-derived CAFs. Comparison with publicly available datasets, however, suggested that prostate CAFs may be alternatively characterized by IL6 and TGFβ-related signatures, indicating that either signal, depending on the context, may concur to fibroblast activation. Our analyses also highlighted novel pathways potentially relevant for induction of a reactive stroma. In addition, we revealed a role for muscle-specific miR-133b as a soluble factor secreted by activated fibroblasts to support paracrine activation of non-activated fibroblasts or promote tumor progression.Overall, we provided insights into the molecular mechanisms driving fibroblast activation in PCa, thus contributing to identify novel hits for the development of therapeutic strategies targeting the crucial interplay between tumor cells and their microenvironment.

Keywords: cancer associated fibroblasts; gene expression; interleukin-6; microRNA; prostate cancer.

Conflict of interest statement

CONFLICTS OF INTEREST

The authors of this manuscript have no conflict of interest to declare.

Figures

Figure 1. Unsupervised analysis of prostate fibroblast gene expression profiles

Gene expression profiles of CAFs, adjacent normal fibroblasts (HPF) and either TGFβ- or IL6-activated fibroblasts were subjected to hierarchical clustering using Euclidean distance and average linkage. Resulting dendrogram was reported and patient ID was color-coded according to the legend.

Figure 2. Gene set enrichment analysis

A. Graphical representation of positively (red) or negatively (blue) enriched gene sets in CAFs (n = 3) compared with HPFs (n = 3). Node size is proportional to the number of genes included in the gene set, whereas link thickness is proportional to the number of common genes. B. Heatmap summarizing significantly enriched gene sets in at least two of the evaluated comparisons (CAF vs HPF, IL6-HPF vs HPF, TGFβ-HPF vs HPF, n = 3 for each group).

Figure 3. Selected gene set enrichment plots

Example of gene sets having divergent enrichment between CAF/IL6-HPFs and TGFβ-HPFs (vs HPF). The analysis was performed using GSEA that calculate an Enrichment Score (ES) by walking down the list of genes ranked according with the t-statistics value of the reported comparison, increasing a running-sum statistic when a gene is in the gene set and decreasing it when it is not (green line). A positive ES indicates gene set enrichment at the top of the ranked list (red part of the horizontal bar); a negative ES indicates gene set enrichment at the bottom of the ranked list (blue part of the horizontal bar). The middle portion of the plot shows where the members of the gene set appear in the ranked list of genes. The bottom portion of the plot shows the value of the ranking metric as you move down the list of ranked genes.

Figure 4. IL6-induced phenoconversion of normal prostate fibroblasts into activated fibroblasts

A. Representative immunofluorescence images of primary normal prostate fibroblasts (HPFs) treated with IL6 or TGFβ, the latter used as positive control for myofibroblast-like differentiation (Magnification: 40×). α-SMA and Collagen1A1 were stained in green and red respectively. IL6-activated fibroblasts showed increased staining for both markers, suggesting an activated phenotype, though maintaining a characteristic spindle-like shape with extended cellular processes compared to TGFβ-HPFs. B. Representative immunoblotting for activation and senescence markers. Fibroblast activation of both IL6- and TGFβ-stimulated HPFs was confirmed by increased α-SMA and FAP protein amounts. IL6-HPFs also showed enhanced expression of senescence-related p21 and p16 proteins. C. Representative images of 3meH3K9 (magnification: 40×) and γH2AX (magnification: 60×) staining confirmed senescent-like phenotype of IL6-activated fibroblasts.

Figure 5. Correlation between activated fibroblast signatures and prostate stroma expression profiles

Three distinct centroids were derived from differentially expressed genes between CAF vs HPF, IL6-HPF vs HPF or TGFβ-HPF vs HPF (n = 3 for each group), and correlated with expression profiles from microdissected tumor/normal stroma (Planche dataset, upper panel) or tumor/normal fibroblasts (Zhao dataset, lower panel). Results were summarized as a heatmap, positive correlations are in red and negative correlations in blue.

Figure 6. miRNA expression analysis

A–D. Significantly up- (A) and down-regulated (B) miRNAs in IL6-HPFs (n = 3) or up- (C) and down-regulated (D) in TGFβ-HPFs (n = 3) compared with HPFs (n = 3) were used as miRNA sets tested for enrichment in the comparison between CAF and HPF miRNA expression profiles. Coherent or divergent expression of miRNAs within miRNA sets with respect to CAF/HPF trend is reported. MiRNAs coherently modulated in all types of activated fibroblasts are highlighted in red. In green are miRNAs belonging to miR-17~92 cluster, in bold muscle-specific miRNAs, both groups being coherently modulated in IL6-activated fibroblasts and patient-derived CAFs. E. qRT-PCR validation of miRNAs found to be up-regulated in patient-derived CAFs, IL6-HPFs and TGFβ-HPFs. Date were reported as -ΔΔCt with respect to serum-starved HPFs of three independent experiments. **P < 0.01.

Figure 7. Integration of gene and miRNA expression data

A. Gene expression data were correlated with expression levels of miR-590-5p, miR-210 or miR-143 respectively. Genes were then ranked according with the correlation value and subjected to gene set enrichment analysis. Enriched gene sets are summarized as a heatmap. Red indicates positive enrichment and blue negative enrichment; darker colors indicate higher statistical significance according to the legend. B. The same analysis in (A) was repeated for miR-133b and here represented as a graph. Node size is proportional to the number of genes included in the gene set, whereas link thickness is proportional to the number of common genes. Positive enrichment is in red and negative enrichment in blue.

Figure 8. miR-133b induces fibroblast activation and acts as soluble factor for paracrine stimulation of fibroblast and tumor cells

A. miR-133b expression levels were evaluated in conditioned medium (CM) from CAFs, IL6- or TGFβ-activated fibroblasts by qRT-PCR. miR-133b release in CM increased upon stimulation of HPFs with IL6 and, to a lesser extent, with TGFβ. Enhanced secretion of miR-133b was also found in CAFs. Data were analyzed using spike-in non-human synthetic miRNA as normalization control and were reported as −ΔΔCt with respect to CM from serum-starved HPFs of three independent experiments. B. miR-133b expression levels were evaluated in CM and extracellular exosomes from CAFs and HPFs by qRT-PCR. CAFs showed enhanced release of both total and exosome-associated miR-133b compared to HPFs. However, starting from equal RNA amounts, Cts were higher in exosomes (hence miR-133b expression lower) with respect to total media, suggesting that miR-133b may be secreted in forms other than exosomes. C. qRT-PCR measurement of endogenous miR-133b levels in HPFs transfected with miR-133b mimics (vs cells transfected with miR-Neg), of miR-133b in the media of HPFs transfected with miR-133b mimics (vs media form cells transfected with miR-Neg), and of endogenous miR-133b in PC3 cells stimulated with CM from miR-133b-transfected cells (vs cells stimulated with CM from miR-Neg-transfected HPFs). Data were reported as −ΔΔCt respect to specific control of three independent experiments. D. Representative bright field and immunofluorescence (α-SMA and FAP staining) images of HPFs transfected with miR-133b mimics or miR-Neg (Magnification: 10×). Ectopic expression of miR-133b induced morphological changes and increase of activation markers reminiscent to those observed in IL6-activated fibroblasts. E. qRT-PCR assessment of ACTA2, FAP, S100A4 and COL4A2 to confirm fibroblast activation upon miR-133b overexpression in HPFs. qRT-PCR data were reported as −ΔΔCt respect to miR-Neg-transfected HPFs of three independent experiments. F. pri-miR-133b, miR-133b, pri-miR-210 and miR-210 expression levels were assessed in PC3 cells upon stimulation with CM from HPF-IL6, HPF-TGFβ or from CAFs by qRT-PCR. Results showed direct uptake of miR-133b in PC3 cells from fibroblast media, as shown by paradoxical down-regulation of endogenous pri-miR-133. In contrast, fibroblast stimulation increased miR-210 expression in PC3 cells by enhancing transcription of pri-miR-210, suggesting a different mechanism. Data were reported as −ΔΔCt with respect to PC3 cells stimulated with CM from HPFs. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 9. PC3 cells directly internalize miR-133b from fibroblast media

A. Merged bright field and fluorescence images of PC3 cells grown in a medium supplied with Cy3-labeled spike-in miRNA (Cy3™ Dye-Labeled Pre-miR Negative Control #1, Life Technologies) or miR-Neg as a control (Magnification: 20×, insert 40×). B. qRT-PCR showing ath-miR-159a expression levels in PC3 cells were grown in a medium supplied with ath-miR-159a spike-in miRNA or miR-Neg as a control. Data are presented as ΔCt compared to RNU48. Notably, signal for ath-miR-159a in miR-Neg sample was undetermined and Ct set to 40. C. Expression levels of endogenous miR-133b in CAFs transfected with siCTR or siDROSHA-1. Data are presented as relative quantity vs HPFs. D. Expression levels of miR-133b in the media from CAFs transfected with siCTR or siDROSHA-1. Data are presented as relative quantity vs CM from HPFs. E. Expression levels of miR-133b in PC3 cells stimulated with the media from CAFs transfected with siCTR or siDROSHA-1. Data are presented as relative quantity vs PC3 cells stimulated with CM from HPFs. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.