MIT Department of Biology: Robert A. Weinberg

Overview
Research in our laboratory is increasingly focusing on three major areas: First, what molecular and biochemical mechanisms are responsible for triggering cell senescence? Second, how does the stroma of a tumor, such as a carcinoma, influence the biology of the tumor as a whole? Third, how do cancer cells within a primary tumor acquire the ability to invade and metastasize?

Research Summary
Molecular and biochemical mechanisms of cell senescence
Senescence is a cell phenotype that has been defined in the context of in vitro culture of cells. Thus, after a certain number of cell divisions, many types of cultured cells will halt proliferation and enter into a non-growing state that is often termed replicative senescence. The precise molecular mechanisms that are responsible for triggering entrance into this state are complex and confounding.

Much evidence implicates suboptimal conditions of culture as a key element that is responsible for triggering entrance into senescence. Thus, if cells are cultured at a level of ambient oxygen that more closely approximates that within living tissues, in vitro proliferation is extended. Moreover, certain types of epithelial cells can be coaxed to proliferate longer if they are provided with stromal support, i.e., support by mesenchymal cells such as fibroblasts and the extracellular matrix that the latter construct.

One key source of senescence is clearly the reactive oxygen species (ROS) that are generated as a consequence of normal metabolism and may conspire with ambient oxygen to induce oxidized DNA bases. The most commonly observed of these bases is 8-oxo-dG. Thus, the accumulation of these oxidized bases in the DNA may overwhelm the ability of the DNA repair apparatus to restore normal DNA structure, resulting in turn in the accumulation of these bases in the DNA. This in turn might trigger a halt to cell proliferation that is induced by proteins such as p53 and p16^INK4A.

In fact, this oxidation may alter the guanosine before it is incorporated into the DNA. This oxidation is normally countered by an enzyme, 8-oxo-dGTPase, normally termed MTH1, which hydrolyzes this oxidized nucleoside triphosphate in order to avoid its inadvertent incorporation into the DNA. We have found that knockdown of MTH1 expression induces rapid senescence and the accumulation of significant damage in the genomic DNA of cells. This provides us with a powerful tool to measure the influence of DNA oxidation on the entrance into cell senescence.

It remains unclear precisely how a halt to cell proliferation is imposed, once a cell has sensed extensive physiologic stress and incurred damage. A prominent effector of this halt is p16^INK4A, which blocks the advance of cells through the G1 phase of the cell cycle. We have recently been exploring the possibility that its mechanism of activation, which has been elusive until now, depends on the activation of stress-activate protein kinases of the p38 family and are currently exploring how inhibition of these enzymes will affect the expression of the important p16^INK4A cyclin-dependent kinase inhibitor.

Tumor stroma and the growth of carcinomas
Carcinomas arise in epithelial tissues that are composed of both epithelial and stromal cells. The latter encompass a variety of mesenchymal cell types, including fibroblasts, myofibroblasts, adipocytes, macrophages, mast cells, endothelial cells, pericytes, and lymphocytes. Normal epithelial cells depend on various types of cell physiologic support in order to sustain their survival and proliferation. While the process of tumor progression yields cells that acquire increasing independence from stromal support, the great majority of carcinomas are formed from neoplastic cells that continue to be dependent on nearby stroma.

We have been interested in the possibility that the stroma of a tumor changes as tumor progression advances. Thus, we have undertaken a series of experiments in which weakly growing carcinoma cells are mixed with the stromal cells extracted from a number of human breast carcinomas. The growth the resulting mixed tumors has then been followed and has revealed that the stromal cells from the majority of breast cancers are more competent to drive tumor progression than are the stromal cells from the normal human breast. This change in biological make-up is reflected by the change in the types of fibroblasts that are present in the tumor-associated stroma: myofibroblasts increasingly replace fibroblasts. Tumors arising through the admixture of myofibroblasts show a greatly increased vascularization, indicating that angiogenesis is a key factor in limiting tumor growth, and that the latter can be accelerated by providing myofibroblasts with potent angiogenesis-inducing powers.

The mechanism(s) used by myofibroblasts to accelerate angiogenesis are quite interesting. These cells release a cytokine, call variously stroma-derived factor-1 (SDF-1) or CXCL12. The latter encourages the recruitment of endothelial precursor cells (EPCs) from the circulation into the tumor-associated stroma, whereupon the EPCs are induced to differentiate into the endothelial cells that proceed to construct the capillaries forming the tumor-associated neovasculature. Because angiogenesis is a rate-limiting step in tumor formation, the presence of these myofibroblasts in the tumor-associated stroma greatly accelerates the growth of the tumor as a whole.

A topic of great interest is the origins of the tumor-associated stromal cells. While many may originate through proliferation of stromal cells in the normal, adjacent host stroma, others may originate in the circulation and thus in the bone marrow of the host. Indeed, we have found that primary tumors are able to encourage the mobilization and recruitment of stromal fibroblasts from the bone marrow, indicating that such tumors extend their reach to distant corners of the body in order to expedite their own agenda of proliferation. Our current research is focused on identifying the nature of the signals released by primary tumors in order to stimulate this stromal recruitment and on the identities of the bone marrow cell populations that serve as precursors to tumor-associated stromal cells.

Mechanisms of tumor cell invasiveness and metastasis
Carcinomas constitute ~80% of the tumors encountered in the oncology clinic, and metastases are responsible for ~90% of all cancer-associated deaths. These figures have focused our attentions on the mechanisms that enable carcinoma cells to invade and metastasize. In fact, the ability to invade and metastasize is a complex, multistep process that involves a number of distinct changes in cell phenotype. Thus, the invasion-metastasis cascade has been proposed to be constituted of the following discrete steps: local invasiveness, intravasation (invasion into blood and lymphatic vessels), transport through the circulation, extravasation (escape from blood vessels into the surrounding tissue parenchyma), formation of a micrometastasis, and finally, colonization (growth of a micrometastasis into a macroscopic metastasis).

The biological complexity of this cascade rivals that of the initial steps of tumorigenesis, raising the question of whether a number of distinct mutations must occur within tumor cells in order to enable them to execute this series of complex biological processes. At the same time, it provokes the question of how cancer cells are able to acquire these multiple abilities in a relatively short period of time.

We have been working over the past several years with a series of transcription factors that are normally active during early embryogenesis and during wound healing. These transcription factors all are capable of inducing epithelial cells (the progenitors of carcinomas) to undergo the epithelial-mesenchymal transition (EMT), a transdifferentiation process that allows epithelial cells to acquire many of the attributes of motile, invasive stromal cells such as fibroblasts. Each of these transcription factors is capable of acting pleiotropically to induce the multiple cellular changes that are associated with the EMT. These include the acquisition of fibroblastic morphology, the downregulation of E-cadherin and cytokeratins, the induction of N-cadherin and vimentin, (often) the secretion of matrix metalloproteinases (MMPs), and the acquisition of invasive behavior.

To date, we have studied in some detail the actions of four of these transcription factors, which play key roles in specific steps of embryogenesis involving EMTs, such as gastrulation and the emigration of cells from the neural crest. The Twist, FOXC2, Goosecoid, and Slug transcription factors all seem capable of programming much, if not all, of the EMT program when expressed ectopically in epithelial cells. We have arrived at these transcription factors through a variety of experimental routes. Twist and FOXC2 were uncovered through an expression array screen of genes that were expressed in highly metastatic mouse breast cancer cells but not in their non-invasive, non-metastatic counterparts. Goosecoid was identified because of its known role in specifying the Spemann organizer, which helps to program gastrulation. And Slug was identified because of its known role in enabling neural crest cells to become motile and invasive.

The role of several of these transcription factors in facilitating metastasis has been demonstrated by inhibiting their expression in otherwise metastatic cells. Thus, shutdown of Twist and Slug expression in metastatic mouse breast cancer cells and human melanoma cells respectively results in suppression of their metastatic behavior. Unproven by these experiments is whether ectopic expression of one or another of these transcription factors in normally non-invasive cancer cells enables the latter to acquire all of the capabilities needed to execute the entire invasion-metastasis cascade.

Significantly, the expression of several of these transcription factors is associated with specific subtypes of human malignancies. For example, Twist is expressed preferentially in invasive lobular carcinomas of the breast, which are known to be highly invasive and carry a poor clinical prognosis, while rarely being expressed in invasive ductal carcinomas, which have a far more favorable clinical course. Similarly, FOXC2 is expressed in almost half of the basaloid subclass of human breast cancers, which carry a poor prognosis, while being rarely expressed in the epithelioid tumors that generally have a good prognosis for the breast cancer patient.

Observations like these suggest that various types of human cancers opportunistically upregulate specific early embryonic transcription factors in order to gain many of the attributes of the cells associated with high-grade malignancy. By activating these normally latent transcription factors, cancer cells gain access to their pleiotropic activities and thus are able to acquire the multiple traits associated with the EMT and invasiveness in a single step.

Research does not yet reveal how expression of these various transcription factors is induced. Nonetheless there is clear evidence that carcinoma cells in the epithelial compartment of a tumor receive various signals from the nearby reactive stroma that causes some of the transcription factors to be expressed in the carcinoma cells, enabling them to activate the EMT and acquire invasive and metastatic powers. Our current research is focused on uncovering the signals that are involved in inducing expression of these transcription factors and the mechanisms that enable them to communicate with one another.

Selected Publications
Orimo, A., Gupta, P.B., Sgroi, D.C., Arenzana-Seisdedos, F., Delaunay, T., Rizwan, N., Carey, V.J., Richardson, A.L., and Weinberg, R.A. Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and angiogenesis through elevated SDF-1/CXCL12 secretion. Cell, 121: 335-348. (2005)

Ben-Porath, I. and Weinberg, R.A. When cells get stressed:an integrative view of cellular senescence. Journal of Clinical Investigation, 113(1):8-13. (2004)

Kuperwasser, C., Chavarria, T., Wu, M., Magrane, G., Gray, J.W., Carey, L., Richardson, A., and Weinberg, R.A. Recreation of functional normal and malignant human breast tissues in mice. PNAS, 101: 4966-4971. (2004)

Yang, J., Mani, S.A., Liu Donaher, J., Richardson, A., Ramaswamy, S., Gitelman, I., and Weinberg, R.A. Twist, a master regulator of morphogenesis plays an essential role in tumor metastasis. Cell, 117:927-939. (2004)

Rangarajan, A., Hong, S.J. Gifford, A., and Weinberg R. A. Species-and cell type-specific requirements for cellular transformation. Cancer Cell, 6:171-183. (2004)

Dessain, S.K., Adekar, S.P., Stevens, J.B., Carpenter, K.A., Skorski, M.L., Goldsby, R.A., and Weinberg, R.A. High efficiency creation of human monoclonal antibody-producing hybridomas. The Journal of Immunological Methods, 291:109-122. (2004)

Ben-Porath, I., and Weinberg, R.A. The signals and pathways activating cellular senescence. International Journal of Biochemistry and Cell Biology, 961-976. (2004)

Weinberg, R.A. Inadvertent Cancer Research. Cancer Biology & Therapy, 3:100-101. (2004)

Search PubMed for Weinberg lab publications.