Limits...
Confrontation of fibroblasts with cancer cells in vitro: gene network analysis of transcriptome changes and differential capacity to inhibit tumor growth.

Alexeyenko A, Alkasalias T, Pavlova T, Szekely L, Kashuba V, Rundqvist H, Wiklund P, Egevad L, Csermely P, Korcsmaros T, Guven H, Klein G - J. Exp. Clin. Cancer Res. (2015)

Bottom Line: The combination of our methods pointed to proteins, such as members of the Rho pathway, pro-inflammatory signature and the YAP1/TAZ cascade, that warrant further investigation via tools of experimental perturbation.We also demonstrated functional congruence between the in vitro and ex vivo models.The microarray data are made available via the Gene Expression Omnibus as GSE57199.

View Article: PubMed Central - PubMed

Affiliation: Department of Microbiology, Tumor and Cell Biology (MTC), Karolinska Institutet, Stockholm, Sweden. andrej.alekseenko@scilifelab.se.

ABSTRACT

Background: There is growing evidence that emerging malignancies in solid tissues might be kept under control by physical intercellular contacts with normal fibroblasts.

Methods: Here we characterize transcriptional landscapes of fibroblasts that confronted cancer cells. We studied four pairs of in vitro and ex vivo fibroblast lines which, within each pair, differed in their capacity to inhibit cancer cells. The natural process was modeled in vitro by confronting the fibroblasts with PC-3 cancer cells. Fibroblast transcriptomes were recorded by Affymetrix microarrays and then investigated using network analysis.

Results: The network enrichment analysis allowed us to separate confrontation- and inhibition-specific components of the fibroblast transcriptional response. Confrontation-specific differences were stronger and were characterized by changes in a number of pathways, including Rho, the YAP/TAZ cascade, NF-kB, and TGF-beta signaling, as well as the transcription factor RELA. Inhibition-specific differences were more subtle and characterized by involvement of Rho signaling at the pathway level and by potential individual regulators such as IL6, MAPK8, MAP2K4, PRKCA, JUN, STAT3, and STAT5A.

Conclusions: We investigated the interaction between cancer cells and fibroblasts in order to shed light on the potential mechanisms and explain the differential inhibitory capacity of the latter, which enabled both a holistic view on the process and details at the gene/protein level. The combination of our methods pointed to proteins, such as members of the Rho pathway, pro-inflammatory signature and the YAP1/TAZ cascade, that warrant further investigation via tools of experimental perturbation. We also demonstrated functional congruence between the in vitro and ex vivo models. The microarray data are made available via the Gene Expression Omnibus as GSE57199.

No MeSH data available.


Related in: MedlinePlus

Pathway scores from network enrichment analysis for DEG lists compiled of genes most differing by mRNA expression level between low and high inhibitory cell lines in vitro (Wh1 vs. Cr9) and ex vivo (PrNFB1 vs. PrTFB2). The network analysis (see “Network enrichment analysis” in Methods) was performed on 300 genes with highest fold change between each pair of cell lines. Horizontal and vertical grey lines denote levels of significance in NEA as false discovery rate = 0.01. Red line displays the linear fit. The text labels are given for pathways with NEA FDR < 0.001 in the both conditions
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4472614&req=5

Fig5: Pathway scores from network enrichment analysis for DEG lists compiled of genes most differing by mRNA expression level between low and high inhibitory cell lines in vitro (Wh1 vs. Cr9) and ex vivo (PrNFB1 vs. PrTFB2). The network analysis (see “Network enrichment analysis” in Methods) was performed on 300 genes with highest fold change between each pair of cell lines. Horizontal and vertical grey lines denote levels of significance in NEA as false discovery rate = 0.01. Red line displays the linear fit. The text labels are given for pathways with NEA FDR < 0.001 in the both conditions

Mentions: Similarly to the principal component analysis presented above, these individual gene values were thus only weakly informative on the biological process. We therefore decided to investigate the DEGs using NEA. First, we had to demonstrate that the lists of DEGs between low and high inhibitory cells collected functionally relevant genes despite the absence of replicates. Assuming that the fold change values reported true differential expression, we expected to find functional interrelations between at least a fraction of the members of DEG lists. This could be proven with NEA by estimating internal connectivity. Indeed, we found that the respective DEGs were significantly interconnected with network links (NEA FDR <10−8 in all cases). As a negative control, NEA calculated enrichment for randomly generated gene sets of the same size and matching network topological properties. Expectedly, no enrichment was found for such sets. For final evidence that the expression differences between the low and high inhibitory cells in the in vivo and ex vivo pairs were consistent, we compared network enrichment of DEGs against relevant pathways. Each such pathway could potentially summarize a group of related DEGs, so that individual pathway-linked genes could be regarded as replicates in a statistical analysis with pathways as factor levels. Unlike the raw gene expression values, the pathway scores (Fig. 5) were indeed correlated: DEGs from in vitro and ex vivo fibroblasts were highly associated with the same pathways (Spearman rank R = 0.686, p0 < 10−18).Fig. 5


Confrontation of fibroblasts with cancer cells in vitro: gene network analysis of transcriptome changes and differential capacity to inhibit tumor growth.

Alexeyenko A, Alkasalias T, Pavlova T, Szekely L, Kashuba V, Rundqvist H, Wiklund P, Egevad L, Csermely P, Korcsmaros T, Guven H, Klein G - J. Exp. Clin. Cancer Res. (2015)

Pathway scores from network enrichment analysis for DEG lists compiled of genes most differing by mRNA expression level between low and high inhibitory cell lines in vitro (Wh1 vs. Cr9) and ex vivo (PrNFB1 vs. PrTFB2). The network analysis (see “Network enrichment analysis” in Methods) was performed on 300 genes with highest fold change between each pair of cell lines. Horizontal and vertical grey lines denote levels of significance in NEA as false discovery rate = 0.01. Red line displays the linear fit. The text labels are given for pathways with NEA FDR < 0.001 in the both conditions
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4472614&req=5

Fig5: Pathway scores from network enrichment analysis for DEG lists compiled of genes most differing by mRNA expression level between low and high inhibitory cell lines in vitro (Wh1 vs. Cr9) and ex vivo (PrNFB1 vs. PrTFB2). The network analysis (see “Network enrichment analysis” in Methods) was performed on 300 genes with highest fold change between each pair of cell lines. Horizontal and vertical grey lines denote levels of significance in NEA as false discovery rate = 0.01. Red line displays the linear fit. The text labels are given for pathways with NEA FDR < 0.001 in the both conditions
Mentions: Similarly to the principal component analysis presented above, these individual gene values were thus only weakly informative on the biological process. We therefore decided to investigate the DEGs using NEA. First, we had to demonstrate that the lists of DEGs between low and high inhibitory cells collected functionally relevant genes despite the absence of replicates. Assuming that the fold change values reported true differential expression, we expected to find functional interrelations between at least a fraction of the members of DEG lists. This could be proven with NEA by estimating internal connectivity. Indeed, we found that the respective DEGs were significantly interconnected with network links (NEA FDR <10−8 in all cases). As a negative control, NEA calculated enrichment for randomly generated gene sets of the same size and matching network topological properties. Expectedly, no enrichment was found for such sets. For final evidence that the expression differences between the low and high inhibitory cells in the in vivo and ex vivo pairs were consistent, we compared network enrichment of DEGs against relevant pathways. Each such pathway could potentially summarize a group of related DEGs, so that individual pathway-linked genes could be regarded as replicates in a statistical analysis with pathways as factor levels. Unlike the raw gene expression values, the pathway scores (Fig. 5) were indeed correlated: DEGs from in vitro and ex vivo fibroblasts were highly associated with the same pathways (Spearman rank R = 0.686, p0 < 10−18).Fig. 5

Bottom Line: The combination of our methods pointed to proteins, such as members of the Rho pathway, pro-inflammatory signature and the YAP1/TAZ cascade, that warrant further investigation via tools of experimental perturbation.We also demonstrated functional congruence between the in vitro and ex vivo models.The microarray data are made available via the Gene Expression Omnibus as GSE57199.

View Article: PubMed Central - PubMed

Affiliation: Department of Microbiology, Tumor and Cell Biology (MTC), Karolinska Institutet, Stockholm, Sweden. andrej.alekseenko@scilifelab.se.

ABSTRACT

Background: There is growing evidence that emerging malignancies in solid tissues might be kept under control by physical intercellular contacts with normal fibroblasts.

Methods: Here we characterize transcriptional landscapes of fibroblasts that confronted cancer cells. We studied four pairs of in vitro and ex vivo fibroblast lines which, within each pair, differed in their capacity to inhibit cancer cells. The natural process was modeled in vitro by confronting the fibroblasts with PC-3 cancer cells. Fibroblast transcriptomes were recorded by Affymetrix microarrays and then investigated using network analysis.

Results: The network enrichment analysis allowed us to separate confrontation- and inhibition-specific components of the fibroblast transcriptional response. Confrontation-specific differences were stronger and were characterized by changes in a number of pathways, including Rho, the YAP/TAZ cascade, NF-kB, and TGF-beta signaling, as well as the transcription factor RELA. Inhibition-specific differences were more subtle and characterized by involvement of Rho signaling at the pathway level and by potential individual regulators such as IL6, MAPK8, MAP2K4, PRKCA, JUN, STAT3, and STAT5A.

Conclusions: We investigated the interaction between cancer cells and fibroblasts in order to shed light on the potential mechanisms and explain the differential inhibitory capacity of the latter, which enabled both a holistic view on the process and details at the gene/protein level. The combination of our methods pointed to proteins, such as members of the Rho pathway, pro-inflammatory signature and the YAP1/TAZ cascade, that warrant further investigation via tools of experimental perturbation. We also demonstrated functional congruence between the in vitro and ex vivo models. The microarray data are made available via the Gene Expression Omnibus as GSE57199.

No MeSH data available.


Related in: MedlinePlus