Robert Benezra, PhD

Molecular Mechanisms of Differentiation

Our attention is focused on the regulation of the basic helix-loop-helix (bHLH) family of transcription factors. These proteins have been shown to control commitment and differentiation in multiple cell lineages in organisms ranging from Drosophila to humans.

We have shown that the activity of bHLH proteins can be controlled in different cell lineages by a common mechanism involving the dominant negative HLH proteins of the Id family (Id1, 2, 3, and 4). The Id proteins contain an HLH dimerization motif but lack a basic DNA binding domain, and as a result heterodimers between Id and bHLH proteins cannot bind DNA. By sequestering a family of ubiquitously expressed bHLH proteins (called E proteins), which are themselves obligate partners of tissue-specific bHLH transcription factors (such as MyoD and NeuroD), the Id proteins inhibit the activity of the bHLH family until the Id proteins themselves are downregulated transcriptionally during the differentiation process. This dominant negative mode of inhibition of DNA binding activity is widely used in the cell, as it is also employed by members of the leucine zipper and homeodomain protein families.

Overexpression studies by numerous laboratories have demonstrated that Id is sufficient to inhibit the activity of bHLH proteins, resulting in alterations in cell fate. The next question we addressed was whether it is necessary. Targeted disruption of 2 of the 4 known Id genes (namely Id1 and Id3), which have nearly identical expression patterns during murine development, has demonstrated the following:

Laboratory Research Summary

My research program has focused on 2 areas of cell biology: the molecular mechanisms of mammalian cell differentiation and mitotic checkpoint control. Though seemingly unrelated, both areas as described below continue to provide fertile ground for the analysis of molecular mechanisms and direct application to cancer biology. In the most general terms, we are exploring the role of the Id proteins in controlling the growth and differentiation of tumor cells and the vasculature that supports their growth.

In addition, we are examining the role of the mitotic checkpoint gene Mad2 in maintaining genome stability by ensuring proper chromosome segregation during mitosis.

The Id Genes are Required to Inhibit the Expression and Activity of Neural Determination Genes during Mammalian Development

Loss of Id1,3 function leads to premature neural differentiation which correlates with ectopic expression of a subset of neural bHLH proteins (NeuroD1, MATH1, MATH2, and MATH3). This is the first demonstration that the Id proteins are required to prevent premature differentiation and that the expression and the activity of the HLH family are under Id control.

The same mechanism for Id activity described above may be at work in both processes: i.e., Id may control the activity of neural HLH proteins involved in commitment (such as MATH1 and MATH3), which themselves regulate the transcription of downstream HLH proteins that initiate differentiation (such as Neuro D1 and MATH2). Thus, the regulation of a single protein-protein interaction (namely Id/E) may play a role in both the commitment and differentiation of neural precursors in mammalian cells.

The Id Genes are Required for Normal Angiogenesis

The Id1,3 knockout mice display a defect in angiogenesis in the neuroectoderm of the brain characterized by a decrease in expression of endothelial cell-specific markers (including the VEGF receptor). The brain specificity of the defect can be explained by the expression of Id1, 2, and 3 in the vasculature of all regions of the embryo except the CNS, and the absence of Id2 expression within the CNS. Our data establish for the first time that HLH proteins control the fate of endothelial cells in mammalian development. Future studies will be directed at determining the consequences of the loss of Id expression on bHLH protein expression and activity in this cell type.

The defect in angiogenesis observed in the Id1,3 knockout mice is similar to that observed in the av integrin knockout mice. This led us to test and confirm the hypothesis that stimulation of av in endothelial cells can activate the Id1 enhancer. We have determined that this activation is independent of growth factor in endothelial cells in culture, but that fibroblasts retain their growth factor dependence whether or not the integrins are engaged.

These results establish for the first time a direct, growth factor-independent link between integrin signaling and the expression of a delayed early response gene (Id1), and will allow us to map the response elements in the Id1 promoter/enhancer that mediate this effect. We hope ultimately to identify the factors involved in this pathway.

The Id Genes are Required to Establish a Normal Tumor Vasculature

The defect in angiogenesis observed in the Id1,3 double knockout mice prompted us to determine if reduced Id dosage would affect the growth of tumor xenografts in adult animals.

Mice lacking 3/4 alleles (Id1-/- Id3+/-) were inoculated with 3 different tumor types and were shown to be resistant to growth and/or metastasis in all cases, independent of any immune-mediated response. Histological analysis has established unequivocally that the knockout mice fail to establish a normal vasculature in and around the supplanted tumor cells. A thickening of the extracellular matrix was associated with the defective vasculature, most probably due to a loss of av integrin and matrix metalloproteinase-2 expression. How Id controls the expression of these genes is currently under investigation.

We hope now to extend these studies to the analysis of tumors that arise in animals genetically predisposed to the development of various cancers. We have initiated crosses between the Id knockout mice and mice with alterations in p53, PTEN, and her2/neu expression. In addition, in collaboration with others, we will be examining the effects of reduced Id dosage on the Min (multiple intestinal neoplasia) mice and mice with other APC mutations that lead to colon cancers.

In this way, we hope to define the classes of cancers that will or will not be susceptible to reduced Id dosage. Conditional disruption of Id in the mouse will establish whether loss of Id in the tumors or the vasculature is responsible for any observed effect and whether Id is truly dispensable in the adult. In addition, we can use our conditional Id knockout mice to determine if loss of Id after the initiation of a tumor will lead to tumor regression, an important consideration in the design of anti-Id treatment strategies (see below).

Our studies have established that angiogenesis in tumors and in the brain are distinguishable processes. Loss of Id in the embryo results in the alteration of markers (VEGF-R2, smooth muscle actin) that are unaffected in the vasculature of tumors (where av integrin and MMP2 are affected). Further subdivisions of the angiogenic process may now be possible as the pathways upstream and downstream of Id expression are identified.

Importantly, therefore, our analysis of the Id knockout mice has lead to the identification of 3 novel signaling pathways: Id as an antagonist of bHLH protein expression; av integrin signaling to the Id promoter/enhancer; and Id as a regulator of av integrin expression. The details of the latter 2 pathways may provide clues for the design of drugs that can interfere with angiogenesis.

Molecular Analysis of the Primary Id Target: The E Proteins

While the E proteins have been shown to interact with tissue-specific bHLH proteins, such as MyoD, to form active transcription factors, in B cells only, it appears that E protein homodimers are the active species. We have shown that at least one means by which E protein homodimers are restricted to B cells is the formation of a B-cell-specific intermolecular disulfide bond, which stabilizes the homodimer and prevents its association with Id. This observation, which runs counter to the widely held misconception that all nuclear proteins are fully reduced, has been confirmed by an independent published report.

We have gone on to show that this disulfide bond can be reduced by a member of the protein disulfide isomerase (PDI) family. Whether PDI modulates the formation of this bond during B-cell development or is inhibited from acting in non-B cells is yet unknown.

Our analysis of E proteins has recently allowed us to shed light on a long-standing question in the bHLH field: since E protein homodimers bind muscle specific enhancers in vitro with high affinity, how is the binding or activation of these homodimers on muscle enhancers inhibited in B cells? We have shown that a highly conserved 11 amino acid motif N-terminal to the bHLH domain of all E proteins can act to repress the primary activation domain (called AD1) present in this protein family only when the homodimers are bound inappropriately to a muscle-specific (MCK) enhancer.

In addition, we showed that this activation domain must also be repressed when E protein/MyoD heterodimers are bound to the MCK enhancer in order to allow for a second activation domain (AD2) to participate with MyoD in activating transcription. We are currently trying to determine if this repression domain communicates with the MyoD basic region to generate the muscle-specific activation code that was postulated by Lassar, Davis, and Weintraub nearly a decade ago.

Complete Loss of Mad2 Function in the Mouse Leads to Early Embryonic Lethality due to Aberrant Chromosome Segregation

In collaboration with the Sorger Laboratory at MIT, we have shown that Mad2-/- mouse embryos die at the blastocyst stage. These blastocysts have a defective mitotic checkpoint, as evidenced by a failure to arrest in nocodazole, and in addition, show aberrant chromosome segregation events under normal growth conditions. This work was extremely challenging technically and required the combined expertise of both laboratories. It has confirmed for the first time the suggestion from Mad2 antibody microinjection studies that Mad2 is an essential gene in mammals required during each mitosis to ensure the accurate segregation of sister chromatids.

We are currently exploring phenotypes present in Mad2+/- adult mice. We have observed hyperproliferation of lymphoid cells in the spleen and extramedullary hematopoiesis in the liver and are currently trying to determine, in collaboration with the Cardon-Cardo Laboratory, if this is due to chromosome missegregation events that ultimately lead to the loss of growth control.

In addition, we will be challenging these mice with low doses of mitotic spindle inhibitors in order to determine if certain tissue types are predisposed to chromosome missegregation events and transformation relative to wild type controls.

Preliminary results indicate that partial loss of mitotic checkpoint function in human cancer cells may be the event that predisposes them to aneuploidy. It is likely that complete loss of mitotic checkpoint function leads to cell death in higher eucaryotic cells. Careful quantitative analysis of mitotic checkpoint function in human cancers will be necessary to establish a link with alterations in genome stability.

Finally, since human Mad2+/+ cells die at a much higher rate after nocodazole exposure than Mad2+/- cells (which survive after endoreduplication), partial loss of mitotic checkpoint function may paradoxically lead to resistance to mitotic spindle inhibitors and polyploidy in human cancers.

Mad2 Can Block the Mitotic Exit Machinery by Direct Interaction with the Proteolytic Degradation Machinery

Specifically, we have shown:

• Components of the spindle assembly checkpoint are localized to the kinetochore prior to spindle attachment but are lost once the microtubule attachment is complete.
• Mad2 associates with a ubiquitin ligase (called the anaphase promoting complex or APC) prior to completion of the spindle assembly, and this association is lost once the spindles become attached. (Activation of the APC is necessary to initiate the metaphase to anaphase transition and mitotic exit).
• Purified Mad2 can inhibit the activity of the APC in the context of a Xenopus extract.

These observations established for the first time a direct biochemical link between the kinetochore and the mitotic exit machinery and have been supported by genetic analysis in both S. pombe and S. cerevisiae.