The TGF-β signalling pathway has the unusual ability to suppress the initial stages of tumorigenesis, but to promote later stages such as metastasis. An unanswered question, though, is whether the response to TGF-β is dependent on cell type or the specific genetic alterations that induce tumour formation, or whether TGF-β can actually switch roles within a single cell lineage. Lalage Wakefield and colleagues have investigated this, and suggest the latter hypothesis is true.

The group worked with three cell lines that were derived by Fred Miller and colleagues from MCF10A breast epithelial cells (M-I), which are immortal but otherwise normal, and so are non-tumorigenic. Transfection with the HRAS oncogene led to the MCF10 AT1k.cl2 (M-II) line, cells from which normally form benign hyperplastic lesions when injected into mice. However, a few of these lesions progress to carcinomas, and these were used to generate the MCF10Ca1h (M-III) line, cells from which form well-differentiated carcinomas, and the MCF10Ca1a.cl1 (M-IV) line, cells from which form poorly differentiated, metastatic carcinomas. All of these cell lines express TGF-β receptors and their proliferation can be inhibited by TGF-β.

So, what effect would inhibiting the pathway have on the four cell lines? The authors transfected them with a dominant-negative TGF-β receptor II, and showed that this could largely block TGF-β signalling for 18 hours. This, in turn, resulted in de-repression of the TGF-β target gene MYC and resistance to growth inhibition by TGF-β. Next, these transfected cells were injected into mice, to investigate the effect on tumorigenicity. Although M-I cells were still unable to form tumours — indicating that loss of TGF-β is not sufficient to induce tumour formation — M-II transfected cells formed rapidly growing lesions, to an extent that was statistically significant compared with the M-II control cells. The M-III cell line also changed from one that formed slow-growing tumours on injection into mice, to one that formed fast-growing tumours with a shorter latency on inhibition of the TGF-β signalling pathway. These were of a higher histological grade and were less well differentiated.

By contrast, the tumorigenicity of the M-IV cell line was not affected by inhibition of the TGF-β signalling pathway; injected cells formed tumours at a similar rate and of a similar histology, regardless of the status of the TGF-β pathway. However, metastatic ability was significantly affected. Injection of control M-IV cells into the tail vein of nude mice resulted in lung metastases within 8 weeks in all mice, but only 40% of mice that were injected with transfected cells formed metastases. This indicates that in advanced tumours, the pro-metastatic ability of TGF-β wins out.

So, it seems that TGF-β signalling really can switch from being tumour suppressive to pro-metastatic within a single cell lineage and that this switch can be initiated by a single oncogenic event. This could have implications for therapy that targets TGF-β — whether the signalling pathway needs to be restored or inhibited would depend on the disease stage.