HUMAN/YEAST COMPLEMENTATION APPROACHES
TO STUDY OF ONCOGENIC SIGNALLING

The yeast S. cerevisiae (baker's yeast) reproduces by budding. Shape of the new bud and placement of the bud on the cell surface are governed by a complex series of interactions between intracellular signalling proteins responsible for regulating organization of the cytoskeleton. Many of these yeast signalling proteins have been identified, and have been found to have higher eukaryotic relatives similarly responsible for controlling cell shape and division properties in mammalian cells. Significantly, a number of the mammalian genes encoding these proteins are oncogenes and anti-oncogenes, implying the essential importance of regulated cell shape control as a decision point between normal and cancerous cell division.

Our laboratory has taken multiple approaches to explore the connections between yeast budding control and the regulation of human cell growth. In one basic approach, we expressed a library of human proteins in yeast, and identified human genes capable of converting the yeast budding pattern from a vegetative to a "pseudohyphal" form; this process is sensitive to nitrogen limitation (a nutritional stressor) and involves basic changes in cell division polarity and cell elongation control. From this screen, we isolated HEF1 (human enhancer of filamentation 1), a gene that appears to be a key intermediate in mammalian growth control signalling related to cell adhesion and cell cycle progression through G2. This screen also yielded a number of HEI (human enhancer of invasion) genes that affect both cell cycle and specific signal transduction cascades. In complementary work, we have continued development of reagents for the Interaction Trap two-hybrid system, a yeast based screen to isolate partner proteins for genes of interest; in the past year, this has included development of a novel Dual Bait system, and delineation of the mechanism of false positive activity.

The HEF1 protein is subject to extensive post-translational modification and cleavage, which may play a key regulatory role in controlling its action as a putative coordinator of cell adhesion and cell cycle progression. In continuing work on the characterization and functional analysis of HEF1, we have focused on the post-translational processing of HEF1. We have described the formation of several derivatives, namely, p115 HEF1, p105 HEF1, p55 HEF1, p65 HEF1 and p28 HEF1, which are observed endogenously and following transfection of a full length HEF1 cDNA into cells. Treatment of cell lysates with lambda phosphatase demonstrates that p115 represents hyper-phosphorylated p105 HEF1; these species localize to focal adhesions. p55 HEF1 constitutes the amino terminal half of HEF1 and includes the SH3 domain that interacts with a polyproline motif in focal adhesion kinase (FAK). The endogenous p55 HEF1 protein product accumulates during mitosis in synchronized cell populations and is found localized predominantly to the mitotic spindle. We have demonstrated that p55 HEF1 arises from cleavage of full length HEF1 at a candidate caspase cleavage site (DLVD).

p65 HEF1 comprises the remaining carboxy-terminal residues of HEF1; under some circumstances, it is additionally processed to form p28 HEF1 consisting of the extreme carboxy-terminus of HEF1. A divergent helix-loop-helix motif is located in the sequence of p28 HEF1, and potentially mediates HEF1 homodimerization and heterodimerization with other cellular proteins. These two HEF1 species, p65 HEF1 and p28 HEF1, are not detectable during mitosis, except following treatment with a proteosome inhibitor, suggesting that these proteins may be degraded during mitosis. We noted that HeLa cells transiently transfected with full length HEF1 cDNA undergo considerable cell death, leading us to examine HEF1 processing in apoptosis. When MCF7 breast carcinoma cells expressing only endogenous HEF1 are treated with the apoptosis inducing agent TNFa, p55 HEF1, p65 HEF1 and p28 HEF1 are all produced. In addition, following TNFa treatment, the time course of HEF1 cleavage is similar to the time course of FAK cleavage, a known caspase substrate and HEF1 interacting partner. Cleavage of both HEF1 and FAK following TNFa treatment is blocked by the addition of caspase inhibitory peptides. Additionally, we have prepared MCF7 cells expressing HEF1 under the control of a tetracycline repressor. Induction of HEF1 in these cells results in cleavage of poly-ADP ribose polymerase (PARP), a marker of apoptosis. We are now attempting to relate the time course of apoptosis in TNFa-induced cell death to the induction of HEF1 in these cells.

As mentioned above, we have shown by yeast two-hybrid studies, co-immunoprecipitation, and co-localization by immunofluourescence that the SH3 domain of HEF1 (encompassed by p55 HEF1) interacts with FAK and is subsequently phosphorylated. FAK is known to be a key regulator of cell adhesion and motility. The association of HEF1 with FAK suggests that it may also be involved in this pathway, and has led us to undertake study of HEF1 phosphorylation in MCF7 cells in response to adhesion. Initially, we are characterizing the phosphorylation status of HEF1 when MCF7 cells are plated on plastic, poly-lysine, fibronectin, or poly-HEMA (a matrix to which cells are unable to attach causing them to grow in suspension). Phosphorylation of HEF1 varies in response to cell adhesion, with the hyperphosphorylated p115 HEF1 species lost in cells held in suspension, although abundant in adherent or newly attached cells. Suspended MCF7 cells form large multicellular aggregates and remain viable for extended periods of time. During this time (up to 8 days) the cells maintain a large and constant amount of unphosphorylated p105 HEF1 that does not appear to be significantly cleaved into the smaller HEF1 protein products, as might be expected. It is reported that such multicellular aggregates are stalled in the G1 phase of the cell cycle, possibly explaining why HEF1 is not further processed and why the cells are not undergoing loss of anchorage-induced apoptosis. We are performing further studies on these multicellular aggregates of MCF7 cells in an attempt to define the mechanism by which HEF1 is prevented from being phosphorylated and cleaved in these cells.

The multi-domain structure of HEF1, in common with the other family members p130Cas and Efs, has implicated these proteins in a central role in the coordination of growth signal processing. It now appears that the domains of HEF1 may be differentially processed; certain forms appear to be selectively stabilized and others degraded, depending on the cell status. Current experiments address the balance of HEF1 stabilization and degradation with respect to cell cycle, differentiation, and death.

HsDim1p was isolated in a two-hybrid screen as a protein interacting with the amino-terminal substrate-binding domain of HEF1. HsDim1p is highly conserved evolutionarily (79% identity from S. pombe to H. sapiens), and may be essential for the entry of cells into mitosis and for chromosome segregation. A recent study by Berry and Gould (J. Biol. Chem. 137:1337, 1997) showed that deletion of Dim1 from the S. pombe genome produced a lethal G2 arrest phenotype that can be rescued by HsDim1p. Given the lethal phenotype involves a defect in mitotic spindle formation, and p55 HEF1 localizes to the mitotic spindle, it was of interest to pursue HsDim1p function. As one approach, we are performing structure-function studies of HsDim1p, guided by molecular modeling.

Threading methods for sequence alignment, and additional programs for secondary structure prediction, suggested that Dim1p might be structurally similar to thioredoxin. Using this relationship as a guide, we have introduced mutations and truncations into Dim1p based on predictions of important residues in thioredoxins (e.g., active site cysteines, predicted regulatory regions). In collaboration with K. Gould at Vanderbilt University, we are also assessing the effect of these altered forms of HsDim1p on S. pombe cell cycle. To understand the structural basis of Dim1p function and its interaction with HEF1, we are determining the solution structure of HsDim1p by multi-dimensional NMR. 15N labeled, as well as 15N, 13C double-labeled HsDim1p have been purified from E. coli. A series of 3D NMR spectra have been collected and assignments of the resonances of the main chain that aligned with thioredoxin were made. Side chain resonance assignments and NOE assignment is under way. Using Chemical shift index, the secondary structure of thioredoxin aligned with part of HsDim1p has been determined. The secondary structure determination has confirmed our prediction that Hsdim1p and thioredoxin fold similarity.

To gain additional insight into Dim1p activity, we used HsDim1p as a bait in a two-hybrid screening and isolated specific partner proteins that were previously defined as regulators of splicing, as well as a novel protein, Dimit. Dimit encodes a small protein (265 amino acids) with sequences that suggest it may also have a role in splicing. It has a unique domain structure comprising: an N-terminal WW domain that binds to proline rich peptides, an RD domain that usually binds to RNA, a region that contains a number of RS di-peptides that is characteristic of a family of pre-mRNA splicing factors, and a potential C-terminal C2 domain that may indicate calcium binding activity. Dimit is predominantly expressed in the nucleus. Of note, while it is distributed throughout the nucleus, a significant proportion of detectable Dimit is present in nuclear speckles that colocalize with the SC35 antigen, further supporting the idea of association with splicing complexes. Our current hypothesis is that Dimit, like HEF1, may be a substrate for Dim1p activity. Whether a Dim1-Dimit interaction is responsible for Dim1's function in the cell cycle is not clear, but it is intriguing that a number of recent studies suggest that cell cycle regulation of pre-mRNA splicing is part of the array of signals that control mitotic entry (K. Lundgren et al., Mol. Biol. Cell 7:1083, 1996; and, references therein). Functional studies are in progress.

We have used a human/yeast genetic complementation screen to identify a number of human genes that can stimulate pseudohyphal/invasive growth in S. cerevisiae, screening either for induction of filamentation (HEF genes) or invasion (HEI genes). Based on current models of endogenous S. cerevisiae control of this process, isolated genes might be expected to include regulators of mitogen-activated protein kinase (MAPK) signaling cascades, Ras functions, protein kinase C (PKC) signaling, morphoregulatory GTPases, and cell cycle control genes. We have isolated known and novel genes falling into a number of these categories, including: HEI 7 (PRK2), which encodes a PKC involved in actin cytoskeletal regulation; HEI 10 (novel), whose product contains a zinc finger motif and resembles the cell cycle control protein, MAT1; HEI 15 (HEIC), which encodes an a helical protein that may be involved in regulating cytoskeletal changes, and two-hybrid predicts to interact with specific MAPK cascades; and, HEF2 and HEF3, whose protein product sequence homologies predict involvement in cell cycle regulation and RNA splicing/metabolism, respectively.

We have begun to further characterize these genes in relationship to the well-conserved signaling control pathways between humans and yeast. We are particularly interested in studying how these genes may be involved in specific pathways related to activation of MAPK cascades, cell polarity/cytoskeletal GTPases, changes in cell wall integrity, and regulation of cell cycle controls. We are using functional assays in yeast to determine whether; specific HEI or HEF genes activate the STE11/7/12 filamentation-specific cascade (e.g., activate FLO11 transcription); hyperactivate Ras (e.g. reduce glycogen accumulation, induce sensitivity to heat shock); or, increase the length of the G2 phase of the cell cycle (e.g., by FACS analysis). To date, we have been able to sort the human genes into a number of discrete functional groups by such approaches. Focusing on HEIC and HEI10 in particular, we are currently engaged in characterizing the function of these proteins in mammalian cells.

We have had a long-term interest in the development of reagents for performing two-hybrid screens in yeast to detect protein-protein interactions. Past work has included: establishment of the Interaction Trap variant of two-hybrid screening system; calibration of the affinity of interactions detected by this type of system; and, development of a Web database characterizing common false positive proteins ( http://www.fccc.edu:80/research/labs/golemis/InteractionTrapInWork.html). We also have investigated the effects of overexpressing a series of putative false positives in yeast, in the presence or absence of the DNA-binding-domain-fused bait used to identify them. Our findings provide an explanation for the selection of some proteins as false positives, and have additional implications for potential indirect effects of protein overexpression on transcription in other biological systems.

In the past year, we have developed a new set of reagents that extend the capabilities of the two-hybrid systems to discriminate specific protein-protein interactions. Existing reagents assayed the interaction of an activation-domain-fused Protein 1 (prey) with a DNA-binding-domain-fused Protein 2 (bait) by their ability to activate transcription of two DNA-binding-domain responsive reporters. In the new reagent system, termed Dual Bait, a single strain of yeast contains two separately responsive reporter systems. In a model system assaying the interaction of the related GTPases, Ras and Krev-1, with their preferred partners Raf and Krit1, and their shared partner RalGDS, the dual bait system clearly differentiates higher affinity versus lower affinity interactions either in grids of yeast streaked to selective plates, or in library screening imitation. This system allows for the isolation of specifically interacting protein pairs against a vast excess of non-interacting pairs. These properties support the idea that these reagents will be useful in library screening and genome-scale applications. In collaboration with Invitrogen, the reagents described here were successfully tested in performing two independent, simultaneous screens in a single yeast. The dual bait reagents can be similarly used for mutational analysis. For example, in collaboration with M. Reeder and J. Chernoff, the reagents have been recently used to successfully identify mutations in Pak1 kinase that selectively reduce interaction affinity for either of two partners, the Cdc42 or Rac GTPases. Further, these reagents have been constructed to potentially supplement any of the currently existing two-hybrid variants, including the recently described membrane-based SOS system. Together, these developments have the potential to greatly expand two-hybrid system contributions to studies of biological interactions.

GOLEMIS, E.A., SEREBRIISKII, I., Recent developments in two-hybrid technology. In 3^rd edition, Molecular Cloning. edited by Sambrook, et. al. Cold Spring Harbor Press (in press).

KHAZAK, V., ESTOJAK, J., CHO, H., MAJORS, G., SONODA, G., TESTA, J.R., GOLEMIS, E.A. Analysis of the interaction of the novel RNA polymerase II subunit hsRPB4 with its partner hsRPB7 and with pol II. Mol. Cell Biol. 18:1935-1945, 1998.

LAW, S.F., ZHANG, Y.-Z., KLEIN-SZANTO, A., GOLEMIS, E.A. Cell-cycle regulated processing of HEF1 to multiple protein forms differentially targeted to multiple subcellular compartments. Mol. Cell. Biol. 18:3540-3551, 1998.

PETERMANN, R., MOSSIER, B.M., ARYEE, D.N.T., KHAZAK, V., GOLEMIS, E., KOVAR, H. Oncogenic EWS-Fli1 interacts with hsRPB7, a subunit of human RNA polymerase II. Oncogene 17:603-610, 1998.

Papers in press at time of previous report:

ASTIER, A., MANIE, S.N., LAW, S., CANTY, T., HAGHEYEGHI, N., DRUKER, B.J., SALGIA, R., GOLEMIS, E.A., FREEDMAN, A.S. Association of the Cas-like molecule HEF1 with CrkL following integrin and antigen receptor signaling in human B cells. Possible relevance to neoplastic lymphohematopoietic cells. Leuk. Lymphoma 28:65-72, 1997.

TOBY, G., LAW, S.F., GOLEMIS, E.A. Vectors to target protein domains to different cellular compartments. Biotechniques 24:637-640, 1998.

^§   Fox Chase researcher

^a   K. Gould: Vanderbilt University, Nashville, TN 37212

^b   D. Stillman: University of Utah, Salt Lake City, UT 84112

^c   Invitrogen Corporation: Carlsbad, CA 92008

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998