Studying cancers within their specific organ microenvironments yields important insights
In 1982, Mina Bissell, PhD, wrote an article for the Journal of Theoretical Biology in which she argued that a cell's microenvironment drove its gene expression and that the extracellular matrix (ECM) “talked” to the chromatin and the chromatin “talked back” (J Theor Biology 1982;99:31–68). That, she noted, meant that the unit of function in higher organisms was larger than the cell. A few years later, she posited that the unit of function is the organ itself—and, in 2001, that tumors are emerging evolutionary organs in fast forward. Although many researchers were skeptical of these unorthodox ideas, she persevered, ultimately shifting our understanding of the role of the microenvironment in cancer biology. Bissell, a distinguished scientist at Lawrence Berkeley National Laboratory, spoke with Cancer Discovery's Suzanne Rose about her work.
Why did it take so long to recognize the importance of the tumor microenvironment?
Part of it is a desire to simplify. If one gene is identified, it can be targeted. In a way, that kind of thinking is fine. By targeting single genes, we've helped lots of cancer patients live longer. Yet we now know that the context of that gene is crucial. Why do people with BRCA1 or BRCA2 mutations only get breast and ovarian cancer? Why do people with the APC mutation get colon cancer when they have the mutation in every cell in their body?
The sequence of the genome by itself doesn't answer everything. By the time you have cancer cells, the genome is a mess; the cells have genomic instability. The tumor is so heterogeneous. When we sequence a tumor, we get “the average.” If we sequenced individual cells, they most likely would each be different.
We need to learn about the normal organ before we can truly understand how cancer of that organ arises and how we should treat it. Tumor cells hijack the normal pathways that operate in a given organ. We need to open our minds to the fact that tumors are organs, and they constantly evolve as they encounter different microenvironments.
How are we learning from organ models?
We have made organ-specific models to understand breast ccancer, and various groups are modeling other organs. A given oncogenic pathway may work in a general way within the larger context of an organ, such as the breast or liver, but the details will be different because breast cancer and liver cancer are not the same cancer, and the devil is in the details! Their form and function are different. Once we destroy the form, we destroy the function.
We can take normal breast cells or cells from a breast tumor and put them in a 3-dimensional (3D) ECM gel we developed in 1992 with Ole Petersen. Every nonmalignant cells will form breast-like acini because all signaling pathways will integrate in our 3D assays. But every tumor cell will develop into a multicellular tumor-like mass. They do not integrate correctly. If we inhibit even one of the oncogenic signaling pathways, we “revert” these malignant cells to a “normal” phenotype despite their malignant genome. Thus phenotype is dominant over genotype even in tumors. This is well demonstrated by our own organs: The cells in the breast and liver have the same genome and yet they look different and do very different things!
Genomic instability and mutations are, of course, important. However, we have demonstrated that unless the “dynamic reciprocity” between the microenvironment and the organ or tissue architecture is lost, the cells will not become overtly cancerous or metastasize. Thus, we need to treat the “tumor organ” in patients, meaning both the tumor cells and their microenvironment.
We wrote a review in 2011 called “Why Don't We Get More Cancer?” (Nat Med 2011;17:320–9) in which we outlined these concepts. We argued that the normal microenvironment prevents premalignant cells from becoming completely malignant. But when the microenvironment becomes abnormal, it not only allows these cells to become malignant, but also may initiate tumors.
What's a recent project that highlights your approach?
We used the wisdom of thinking in 3D and succeeded in making a robust screen. In a paper that came out this past fall in the Journal of Clinical Investigation, we looked at resistance to EGFR-tyrosine kinase inhibitors (J Clinic Invest 2012;122:3211–20). Some patients with EGFR-mutated tumors who take lapatinib (Tykerb; GlaxoSmithKline) do not respond at all and some live 1 month longer. We reasoned that maybe we do not know all of the players in the EGFR pathway. We hypothesized that there are other molecules downstream of EGFR that we haven't yet discovered despite the fact that we know the sequence of the genome.
So we used our robust assay to revert the malignant phenotype with EGFR inhibitors. We made a cDNA library of the tumor cells, put it back in with the same cells, and looked for colonies that did not revert, meaning that those cells were now resistant to EGFR inhibitors because the cDNA had disrupted crucial signals. This occurred in 1% to 2% of the colonies.
When we pulled out the colonies that had become resistant to EGFR inhibitors and still exhibited a malignant phenotype, we discovered 5 new genes. One of them was FAM83A. We characterized FAM83A and showed that it is downstream of EGFR and PI3 kinase, and if you overexpress it in breast cancer cells, they become resistant to lapatinib.
In the same issue of JCI, using a completely different screen, Cipriano and colleagues reported another family member, FAM83B (J Clin Invest 2012;122:3197–210), and the 2 reports make a compelling argument about the importance of this 6-member family as a new oncogene family.
We found these genes because we used our heads in a way that was different from what other people were doing. We didn't say, “Let's find something better to target EGFR.” You could do everything against EGFR in those patients who have FAM83 mutations or overexpression and it won't work.
People are learning a lot about the brain, but we are not learning enough about cancer. That's because we treat cancer as if it's one thing. It's not. Cancer is organ specific.
For more news on cancer research, visit Cancer Discovery online at http://CDnews.aacrjournals.org.