In this study, we report - for the first time - that IRF5 is expressed in different thyroid carcinoma histotypes and in multiple thyroid cancer cell lines but is not detectable in normal human thyrocytes (Figure
1A). In primary normal and tumor thyroid cancer cells IFNα does not modulate IRF5 levels (Figure
1B) while in immortalized cell lines, IRF5 seems oddly responsive to IFNα, since exposure to this cytokine reduces IRF5 in two thyroid cancer cells and consistently increases its expression in one cell line (Figure
1C). We found both endogenous and ectopic IRF5 in the cytoplasm of thyroid cancer cells and IFNα was unable to relocate the protein in the nucleus (Figure
2A, B and Additional file
2). Not surprisingly, IRF5 did not induce the p21 promoter (Figure
2C, D) [
17]. On the contrary, IRF5 stimulated thyroid cancer proliferation (Figure
3B, C), protected malignant thyroid cells from the cytotoxic effects of different antiblastic compounds (Figure
4), and significantly increased their colony-forming ability (Figure
5B). Indeed, silencing of endogenous IRF5 by lentiviral shRNA reduces the clonogenic potential of most thyroid cancer cells (Figure
5A). The induction of B-Raf expression detected in thyroid cells transduced with IRF5 could partially explain its tumor-promoting effects (Figure
5C).
These findings pose the pivotal question of what role is fulfilled by IRF5 in thyroid cancer. The initial observation that the IRF5 protein is not expressed in normal thyrocytes but is detected in neoplastic thyroid cells is somewhat surprising as IRF5 has been usually associated with tumor-suppressor rather than tumor-promoting activities [
8,
21,
22]. Our finding that thyroid cancer cells localize IRF5 to the cell cytoplasm implies that the protein is transcriptionally inactive. Indeed, IRF5 fails to induce a previously characterized target such as p21 and does not arrest cell-cycle progression. On the contrary, IRF5 lentiviral over-expression significantly increases the proliferation rate of malignant thyroid cells. Taken together these findings suggest two alternative scenarios: a) IRF5 might be inactivated due to point mutations or increased tyrosine phosphorylation as previously reported [
22]; b) IRF5 might be either inducing the expression of tumor-promoting genes or inhibiting the promoters for tumor-suppressors. Our supplemental data seem to exclude the first scenario as direct sequencing of IRF5 revealed no point mutations and we failed to observe any variations in IRF5 tyrosine phosphorylation levels (Additional file
3). The second hypothesis seems much more likely as suggested by: i) our findings showing increased B-Raf expression in thyroid cancer cells lentivirally transduced with IRF5; ii) the protective effect of IRF5 on thyroid cells exposed to different cytotoxic drugs; iii) the reduced colony-forming potential of malignant thyrocytes displaying reduced IRF5 levels.
In summary, the present study indicates that IRF5 favors the thyroid tumoral phenotype. However, the exact mechanisms underscoring this unexpected biological function remain partially unresolved.