by ecancer reporter Clare Sansom

All solid tumours grow on a matrix of stromal cells: host cells that are non-malignant but distinguishable from normal cells of the same type by inflammatory and other markers.

Interactions between tumour and stromal cells in the tumour micro-environment play an important part in cancer growth and progression.

Many of the genetic changes that are present in epithelial tumours are also present in pre-cancerous lesions; cellular senescence is known to be one of the mechanisms that prevents this progression.

Although the increased senescence of stromal cells has been implicated in the increase in some cancers with age, the stroma surrounding epithelial tumours is characterised by the proliferation of non-senescent fibroblasts expressing a different set of markers, termed cancer-associated fibroblasts (CAF).

A DNA-binding protein known as CSL, which can function as either a repressor or an activator of transcription (the latter following Notch activation) has been implicated in the induction of both the senescent and the CAF phenotype in fibroblasts.

A group of researchers led by G. Paolo Dotto of the University of Lausanne, Epalinges, Switzerland, has now explored the function of this protein using both in vitro and in vivo methods and developed a model for the transformation of normal fibroblasts into CAFs.

They first observed that mice in which the CSL gene had been deleted in connective tissue precursor cells were born with signs of ageing skin and rapidly developed skin tumours.

Fibroblasts in these mice also proliferated rapidly.

A comparison of cultured dermal fibroblasts from mice with and without CSL deletions showed that those from the mutant mice developed senescence-related properties, but only after two passages.

Down-regulation of CSL in primary human dermal fibroblasts (HDFs) and CSL silencing in other fibroblasts using small hairpin RNA or small interfering RNA induced similar phenotypes.

Activation of the Notch receptor in these cells converted CSL from a repressor to an activator of transcription and induced genes associated with both the senescent and the CAF phenotypes.

The researchers found that CAFs derived from squamous cell carcinomas (SCCs) of human skin had CSL levels that were decreased compared to those in HDFs.

Similarly, fluorescently-labelled skin SCC cells expanded more slowly in a mouse model when mixed with CAFs with increased CSL levels than when mixed with unaltered CSL.

They then compared gene expression profiles of CAFs from skin, head and neck and lung cancers to those of normal fibroblasts taken from the same sites, and to HDFs in which CSL had been silenced.

They found that 56% of the genes with expression patterns that were significantly altered by CSL silencing had expression patterns that were altered in a similar way in CAFs.

Functional analysis of this gene set showed that genes that were down-regulated in both cases were typically involved in development or migration, whereas genes involved in matrix remodelling and smooth muscle proliferation were up-regulated.

Fibroblast growth factor receptor 2 (FGFR2) was among the genes that were up-regulated in CAFs and down-regulated in normal fibroblasts with CSL silencing.

Other genes in this category were involved in apoptosis; some of these genes are implicated in senescence, and some are targets of the tumour suppressor p53.

From these results, Dotte and his co-workers developed the hypothesis that CSL functions as a repressor of genes that promote senescence and of those that drive the CAF phenotype separately.

Bioinformatics analysis supported this by identifying CSL binding sites in the promoter and enhancer regions of a number of genes in both categories.

They also suggested that CSL was likely to interact with p53, and pulldown assays showed direct binding between these proteins.

The researchers then analysed gene expression levels in fibroblasts associated with pre-malignant skin lesions and with SCC, finding that CSL was down-regulated in both cell types but p53 only in SCC stroma.

Paracrine signalling by FGF2 but not by other growth factors was found to induce down-regulation of p53 in HDFs, and this could be restored and cellular senescence induced with specific small-molecule inhibitors of FGFR signalling.

Finally, the researchers injected human skin SCCs in which CSL, p53 or both genes had been knocked down in the stromal cells into mouse ears and monitored their proliferation.

Both the tumour and the stromal cells expanded much faster when both genes were knocked down.

Taken together, these results suggested a model of stromal fibroblasts in which active CSL and p53 together suppress the expression of genes involved in both senescence and transformation to a CAF phenotype.

Loss of CSL expression alone is sufficient to induce the senescence genes, whereas loss of both genes is necessary for CAF-related gene induction.

As CAF activation is induced by FGFR signalling, inhibitors of this kinase might be useful in preventing this process and thus co-evolution of the tumour and stroma.

Reference

Procopio, M-G., Laszlo, C., Al Labban, D. and 15 others (2015). Combined CSL and p53 downregulation promotes cancer-associated fibroblast activation. Nature Cell Biology, published online ahead of print 24 August 2015.