Low tumour PPM1H indicates poor prognosis in colorectal cancer via activation of cancer-associated fibroblasts

Abstract

Background: Vimentin (VIM) is considered a prognostic marker in colorectal cancer (CRC). Our aim is to identify genes that fulfil a "X-low implies VIM-high" Boolean relationship and to evaluate their prognostic value and potential mechanism.

Methods: Potential biomarkers related to VIM expression were searched using a bioinformatics approach across gene-expression arrays. Based on subgroup analysis of 2 CRC cohorts, the selected gene was tested for its association with patient's survival outcomes. The regulatory link between the selected gene and VIM was further examined with in vitro models.

Results: PPM1H was identified as the top candidate in our search. Patients with PPM1H-low tumours have a lower 5-year disease-free survival rate than patients with PPM1H-high tumours in 2 independent cohorts. In multivariate Cox analysis, patients with PPM1H-low tumours were independently associated with relapse in both the discovery cohort (hazard ratio [HR], 1.362; 95% confidence interval [CI], 1.015-1.826; P = 0.039) and the validation cohort (HR for DFS, 4.052; 95% CI, 2.634-6.234; P < 0.001). PPM1H knockdown in CRC cells and growth in the corresponding conditional medium increased VIM expression and colon fibroblast proliferation, indicating a transformation of cancer-association fibroblasts (CAFs). Conversely, educated CAFs also facilitated the growth of CRC cells with low PPM1H expression.

Conclusions: Lack of tumour PPM1H expression identifies a patient subgroup with a high relapse risk, and CRC cells with low expression of PPM1H activate CAFs and inversely get promoted by CAFs.

Fig. 1

Outline diagram of this study. CRC colorectal cancer, IBD inflammatory bowel disease, DFS disease-free survival, DSS disease-specific survival

Fig. 2

Relationship between the patient subgroups classified by mRNA expression of PPM1H and VIM and disease-free survival in the discovery data set. (a) Classification of the subgroups with high or low PPM1H mRNA expression; (b) survival analysis between patients with PPM1H-high tumours or PPM1H-low tumours; (c) three patient subgroups defined by PPM1H and VIM mRNA expression; (d) survival analysis among the subgroups defined by the PPM1H/VIM system. The location of the cut lines was defined according to StepMiner analysis

Fig. 3

Relationship between the patient subgroups classified by PPM1H protein expression and survival outcomes in the validation data set. (a) Represented protein expression of cytoplasmic PPM1H; (b) DFS analysis between patient subgroups with high or low PPM1H protein expression; (c) DSS analysis between patient subgroups with high or low PPM1H protein expression. The red bar in the figure represents 100 μM. DFS, disease-free survival; DSS, disease-specific survival

Fig. 4

Knockdown of PPM1H in CRC cells increased VIM expression in fibroblasts and its proliferation, and activated fibroblasts facilitated CRC cell proliferation and invasion. (a) VIM mRNA or protein expression of CCD-18Co with respect to the different CMs from the indicated CRC cells after siPPM1H or siControl interference; (b) proliferation of CCD-18Co with respect to the different CMs from the indicated CRC cells after siPPM1H or siControl interference; (c) proliferation of SW480 and CaCO2 cultured with the different CMs from activated CCD-18Co and CCD-18Co-control; (d) cell invasion of SW480 and CaCO2 cultured with different CMs from activated CCD-18Co and CCD-18Co-control. CMs, conditional media. Each assay was repeated thrice