Research in my laboratory has been focused on the mechanisms by which carcinoma cells – largely those arising in the mammary gland – generate aggressive tumors and metastatic outgrowths. The process of metastasis formation subsumes multiple distinct successive steps, which are termed in aggregate the invasion-metastasis cascade. Included among these are the steps of local invasiveness of carcinoma cells in the primary tumor, intravasation, transport through the circulation, lodging in distant microvessel, extravasation, and colonization, the last step referring to the ability of extravasated micrometastatic colonies to spawn macroscopic, life-threatening metastatic growths.
We have been focusing on the cell-biological program termed the epithelial-mesenchymal transition (EMT), which imparts to epithelial cells, specifically neoplastic ones, many of the attributes associated with cells of high-grade malignancy. In the context of the primary tumor, we have previously demonstrated that experimental activation of the previously latent EMT program in otherwise-non-metastatic primary tumor cells imparts to them the ability to disseminate and form hundreds of micrometastases in the lungs in contrast to comparable cells in which the EMT program is not activated and micrometastases are not readily detected. Hence, carcinoma cells that have sustained multiple heritable genetic and epigenetic changes during primary tumor formation do not require additional genetic changes in order to metastasize, since activation of the EMT program does not involve genetic alterations.
Of note, carcinoma cells that have activated their EMT program are placed in a state from which they can readily enter into a stem-cell state and thereby become cancer stem cells (CSCs). As we previously showed, such CSCs, which exhibit a mixture of epithelial and mesenchymal phenotypes, arise in >100-fold larger numbers once the EMT program has been activated in populations of non-CSCs, i.e., populations of breast cancer cells that are more epithelial and lack stemness.
We have been interested in the proclivity with which primary carcinoma cells are able to activate the EMT program. In previous work, we showed that more epithelial, non-CSCs may spontaneously enter into the CSC state via activation of their previously latent EMT programs. This indicated that there is great plasticity in these cells, in that there can be bidirectional interconversion between the CSCs and non-CSCs in experimentally transformed cells; we presume that similar dynamics operate in spontaneously arising tumors as well.
We examined in much greater detail the interconversion between non-CSCs and CSCs, specifically in the context of breast cancer. As we found, the non-CSCs of luminal carcinomas breast tumors, which carry a generally good prognosis, rarely spawn CSCs, either spontaneously or in response to applied TGF-b. Conversely, the non-CSCs of basal carcinomas, which often carry a worse prognosis, readily spawn CSCs, either spontaneously or in response to TGF-b. Associated with the formation of CSCs, as implied above, is entrance into a more mesenchymal, aggressive state.
Spontaneous activation of the gene encoding the ZEB1 master regulator of the EMT program is essential for entrance of basal non-CSCs into the CSC state. We examined the promoter of the ZEB1 gene and discovered that in luminal carcinoma cells, it is shut down by repressive chromatin marks that do not change in response to TGF-b treatment. Conversely, this promoter in basal carcinomas of the breast, the promoter is in a “bivalent” configuration with both coexisting repressive and inductive histone marks; this bivalent state allows it to rapidly spring into action in response to spontaneously arising signals or, more specifically TGF-b. Taken together, this might indicate that the future behavior of a primary breast cancer can be predicted by examining the configuration of its ZEB1 gene promoter, which may be the key determinant to its proclivity to activate its previously latent EMT/CSC program.
In another area of work, we have examined the processes by which recently extravasated carcinoma cells are able to gain an initial foothold in the parenchyma of lung tissue. As we discovered, post-extravasation proliferation of such cells depends on their ability to form mature adhesion plaques between their own cell-surface integrins and components of the extracellular matrix (ECM) of the host tissue. The formation of these plaques depends in turn upon the extension of filopodia-like protrusions, which are covered with integrins. These protrusions are exhibited in abundance by metastatic carcinoma cells but not by their more benign, non-metastatic counterparts. Of special interest is that the formation of these filopodia-like protrusions is activated by the EMT program, indicating that this cell-biological program extends its reach as far as the post-extravasation behavior of disseminated carcinoma cells.
The final step of invasion-metastasis cascade, termed colonization, which appears to depend on the adaptation of cells originating in one tissue to the foreign microenvironment of another tissue, is not addressed by this research. It would seem to depend on epigenetic changes that enable already-disseminated cells to utilize the contextual signals released by the microenvironment of their newfound homes to provide the types of trophic and mitogenic signals that were available in the tissue-of-origin of the cancer cells. We do not intend to examine the nature of these adaptive programs.
In a final project related to the latter steps of the invasion-metastasis cascade, we have examined the mechanisms that enable carcinoma cells that have lodged in the microvessels of the lung to extravasate. As we have found, the vast majority of these cells (>90%) are lost from the lungs with several hours of their initial arrival. We have examined the ancillary cell types that enable intraluminal survival and subsequent extravasation of the carcinoma cells. In one tumor model, involving the 4T1 mammary carcinoma cells, tumor-bearing mice develop splenomegaly as a consequence of the GM-CSF that the carcinoma cells release into the circulation; the latter then induces the formation and mobilization of myeloid cells – neutrophils and monocytes – into the circulation, which proceed to accumulate in large numbers in the spleen.
These splenocytes, specifically neutrophils, subsequently associate with trapped carcinoma cells in the lung microvasculature. Here they exercise two functions: they protect the carcinoma cells from elimination by NK cells, thereby extending their intraluminal dwell time in the lung microvessels, and they facilitate the extravasation of the carcinoma cells, in part through the secretion of matrix metalloproteinases that create gaps in the walls of the microvessels that facilitate extravasation. Hence, these later steps of the invasion-metastasis cascade, like earlier steps, depend on heterotypic interactions of carcinoma cells with the inflammatory cells of the innate immune system.