Noncanonical TGFβ Pathway Relieves the Blockade of IL1β/TGFβ-Mediated Crosstalk between Tumor and Stroma: TGFBR1 and TAK1 Inhibition in Colorectal Cancer

PURPOSE

The aim of the study is blocking the recruitment of a protective stroma by altering the crosstalk between normal stromal cells and tumor cells for stripping tumors of the protection conferred by the microenvironment.

EXPERIMENTAL DESIGN

A transcriptomic analysis of cocultured normal colonic fibroblasts and colorectal tumor cells was performed. We focused on the study of molecules that mediate the communication between both compartments and that entail fibroblasts' activation and the alteration of the sensitivity to chemotherapy. We identified targets for the blocking of the tumor-stroma interaction. Finally, we tested, in vivo, the blockade of the tumor-stroma interaction in orthotopic models derived from patients and in models of acquired resistance to oxaliplatin.

RESULTS

IL1β/TGFβ1 are the triggers for fibroblasts' recruitment and conversion into carcinoma-associated fibroblasts (CAF) in colorectal cancer. CAFs then secrete proinflammatory factors that alter sensitivity in tumor cells, activating JAK/STAT and PI3KCA/AKT pathways. Blocking such crosstalk with a neutralizing IL1β antibody and a TGFBR1 inhibitor is relieved by the TAK1-mediated activation of the noncanonical TGFβ pathway, which induces a change in the cytokine/chemokine repertoire that maintains a sustained activation of AKT in tumor cells. TAK1 plus TGFBR1 inhibition blocks IL1β/TGFβ1-mediated fibroblast activation, decreasing the secretion of proinflammatory cytokines. In turn, tumor cells became more sensitive to chemotherapy. In vivo, the combination of a TAK1 inhibitor plus TGFBR1 inhibitor reduced the metastatic capacity of tumor cells and the recruitment of fibroblasts.

CONCLUSIONS

Our findings provide a translational rationale for the inhibition of TAK1 and TGFBR1 to remove the chemoprotection conferred by CAFs.

A monotonic and prognostic genomic signature from fibroblasts for colorectal cancer initiation, progression, and metastasis

The differential gene expression patterns between normal colonic fibroblasts (NCF), carcinoma-associated fibroblasts from primary tumors (CAF-PT), and CAFs from hepatic metastasis (CAF-LM) are hypothesized to be useful for predicting relapse in primary tumors. A transcriptomic profile of NCF (n = 9), CAF-PT (n = 14), and CAF-LM (n = 11) was derived. Prediction Analysis of Microarrays (PAM) was used to obtain molecular details for each fibroblast class, and differentially expressed transcripts were used to classify patients according to recurrence status. A number of transcripts (n = 277) were common to all three types of fibroblasts and whose expression level was sequentially deregulated according to the transition: NCF→CAF-PT→CAF-LM. Importantly, the gene signature was able to accurately classify patients with primary tumors according to their prognosis. This capacity was exploited to obtain a refined 19-gene classifier that predicted recurrence with high accuracy in two independent datasets of patients with colorectal cancer and correlates with fibroblast migratory potential. The prognostic power of this genomic signature is strong evidence of the link between the tumor-stroma microenvironment and cancer progression. Furthermore, the 19-gene classifier was able to identify low-risk patients very accurately, which is of particular importance for stage II patients, who would benefit from the omission of chemotherapy, especially T4N0 patients, who are clinically classified as being at high risk.

IMPLICATIONS

A defined stromal gene expression signature predicts relapse in patients with colorectal cancer.