Proceed With Caution

For many years, cancer researchers have believed that metastasis relies on the transition of tumor cells from an epithelial to a mesenchymal phenotype. Even after tumor analysis revealed that the cells of secondary cancers exhibit epithelial characteristics, this was ascribed to a reversal of the transition – from mesenchymal back to epithelial phenotype. Why has this belief persisted so strongly despite uncertainty and debate – and why have the recent papers by Fischer et al. (see "Tracking the Transition") and Zheng et al. (see "The PDAC Key") had such an impact on the research landscape?

EMT is an embryonic process required for proper development. It has been observed in tissue culture upon expression of various transcription factors, and following treatment with different cytokines. In vitro, EMT is associated with increased cell migration and invasion. In many cases, the increased invasion observed in vitro translates into increased metastasis in mouse tumor models. But clinical evidence supporting EMT in human tumors has been somewhat limited, due to the difficulty in distinguishing mesenchymally transformed cancer cells from reactive fibroblasts within a tumor. This has led to some debate regarding the importance of EMT in tumor dissemination in the clinical setting. That’s where the two new studies may shed light.

Pros and cons

EMT is reversible; it’s currently believed that epithelial cells transition into a mesenchymal state, then revert to the epithelial state upon reaching the distal site. The plasticity and transient nature of EMT has made it difficult to follow these cells from the time they transition to a mesenchymal state, through invasion into the blood, and to the point of colonization at distal sites. The two studies reported in Nature are particularly interesting because they both addressed this problem, albeit using very different approaches. Fischer et al. used green fluorescent protein (GFP) expression as a proxy for mesenchymal transition and traced lineage-switched epithelial tumor cells from inception to metastatic colonization in two different mouse mammary tumor models. Zheng et al. knocked down the EMT-inducing transcription factors Snail and Twist in the pancreatic epithelium of mice so that they could monitor the consequences of EMT in the metastatic dissemination of pancreatic tumor cells. These approaches allowed definitive, real-time monitoring of the tumor cells and concluded that EMT is dispensable for metastatic colonization, but plays a role in drug resistance.

That doesn’t mean that these studies are without limitations (1). First, EMT relies on a complex signaling network that involves multiple transcription factors and signaling proteins, in some instances with redundant functions. Whether lineage-tracing studies with single genes can accurately mimic this complex process is unclear. Second, cancer progression involves a continually evolving genomic and epigenetic landscape, so it’s unlikely that mouse tumor models driven by only a few oncogenic events fully recapitulate this process. The studies certainly show that EMT is dispensable for metastasis, but readers must recognize the limitations of the mouse models.

Interestingly, in the mouse model generated by Fischer et al., epithelial tumor cells that switch to a mesenchymal state are permanently marked with GFP expression, and illustrate that a small subset of such cells do indeed spontaneously transition into a mesenchymal state (although it isn’t required to drive overt metastasis). The prevalence of green cells following drug treatment suggests that cells with a history of EMT, regardless of their current state, are more resistant to drugs. The mechanism by which EMT increases cell survival under adverse conditions is not yet known – but perhaps our new understanding of EMT will provide a springboard for further research.

Old theories have been challenged – now what?

The new studies tell a very different story compared with the prevailing narrative. Why? No one can say for certain, but the EMT models used for in vitro research represent powerful induction of the transition by EMT-inducing transcription factors and cytokines. Spontaneous EMT in clinical specimens might be much more subtle, and could account for – or at least contribute to – the discrepancy between these two studies and those that have previously been reported. Another consideration is that EMT relies on the activation of complex and sometimes redundant signaling modules, an aspect not reflected by the mouse models used in the Nature studies. Although those models do show that EMT is dispensable for metastasis, the findings need to be evaluated within the context of the complexity of tumor progression, which involves an ever-evolving genomic and epigenetic landscape.

EMT is an attractive concept to define the process of metastasis: it involves loss of cell-cell interaction and gain of cell motility. But there are other cellular mechanisms of tumor dissemination, like collective epithelial cell migration or tumor microemboli, that may drive the spread of cancer. And metastasis isn’t the end of the story – EMT is also emerging as an important contributor to drug resistance, a phenomenon supported by the findings from both Nature papers. In my own recent work, my colleagues and I demonstrated drug-induced shifts in the epithelial and mesenchymal tumor populations of breast cancer patients. So although the new findings raise questions about EMT’s role in metastasis, they also show that the transition does occur in tumors – and not without a purpose, as cells that switch lineage are more resistant to drugs. It’s now critical to gain further insight into the molecular nature of this process, so that we can use that information to research better treatments and more accurate prognoses.

As the field moves toward a more complete understanding of the tropism exhibited by tumor cells shed into blood and the role of EMT in drug resistance, I have one word of caution for researchers and clinicians alike. It’s important to carefully evaluate what we learn about metastasis from cell culture and mouse models against both human clinical samples derived from repeat biopsies or tumor cells circulating in the blood and freshly established tumor cell cultures. By keeping an open mind to both new information and the limitations of pioneering studies, we can ensure that we’re able to focus on the “big picture” of how cancer metastasis happens and what we can do to combat it.

Shyamala Maheswaran is Associate Professor of Surgery at Harvard Medical School and Assistant Molecular Biologist at the Center for Cancer Research, Massachusetts General Hospital, Boston, USA.

Proceed With Caution

Tracking the Transition

The PDAC Key

First published in The Pathologist, a sister publication of The Translational Scientist.