ROLE OF PROTEIN PHOSPHORYLATION IN
ADHESION-DEPENDENT SIGNALING PATHWAYS

Our laboratory focuses on two major topics: 1) the regulation and role of protein tyrosine phosphatases (PTPs) in adhesive signal transduction; and 2) the mechanism by which small guanosine triphosphatases (GTPases) of the Rho family exert their effects on cell growth and morphology. We are studying two PTPs, PTP1B and PTEN. Both of these proteins are cytosolic proteins that negatively regulate mitogenic signals. Our goal is to examine in greater detail the regulation of these PTPs and to determine the mechanisms by which these enzymes suppress growth signals. In addition, we are exploring the mechanisms by which the Ras-related GTPases, Cdc42 and Rac1, induce changes in cell growth and morphology. Both of these GTPases bind to and activate a class of protein kinases known as p21-activated protein kinases (Paks). Thus, Paks are prime candidates as GTPase effectors. We are seeking to determine the regulatory mechanisms and physiological functions of these kinases in mammalian cells and in fission yeast.

When overexpressed in Src-transformed mammalian fibroblasts, PTP1B induces phenotypic reversion, as reflected in restoration of anchorage dependence, serum dependence, and normal morphology. Because Src is a tyrosine kinase, one simple explanation for this phenomenon is that PTP1B dephosphorylates Src, thereby inactivating it. However, PTP1B also induces the reversion of both Crk- and Ras-transformed cells. As neither Crk nor Ras are tyrosine kinases, these results suggest that PTP1B targets one or more key signaling molecules that are commonly activated by these three oncoproteins.

To find such signaling molecules, we used a combination of two-hybrid trapping and in vitro binding techniques. We established that PTP1B complexes with a variety of Src-homology 3 (SH3)-containing proteins, including the adaptor protein Crk and the cytoskeletal docking protein p130Cas. PTP1B contains two proline-rich motifs, the first of which serves as a ligand for the SH3 domains of these proteins. The importance of the interaction of PTP1B with such proteins is highlighted by the fact that, while overexpression of wild-type (WT) PTP1B causes phenotypic reversion of Src, Crk, and Ras-transformed cells, overexpression of a proline-to-alanine (PA) mutant PTP1B, which bears a mutation within one of its proline-rich motifs, has no effect on transformation by these oncoproteins. As such proline-rich mutants are catalytically competent, these results strongly suggest that interactions with SH3-containing proteins are vital to PTP1B's function.

Since Crk and p130Cas themselves form a complex in response to cell adhesion, these results also suggest that PTP1B may have a role in cytoskeletal assembly and/or integrin-mediated signaling pathways in normal (i.e., non-transformed) cells. To address this possibility, we overexpressed various forms of PTP1B in rat 3Y1 fibroblasts and measured signaling in response to growth factors and to cell adhesion. Neither WT nor PA-PTP1B had any effect on signals generated by epidermal growth factor. In contrast, WT, but not PA-PTP1B, severely impeded adhesion signaling, as assessed by impaired activation of mitogen-activated protein kinase, cell spreading, motility, and focal adhesion complex assembly. As these same phenomena are observed in cells expressing various mutant forms of p130Cas, these data are consistent with our previous binding studies which indicated that p130Cas may represent a key physiological target for PTP1B.

To find potential substrates for PTEN, we performed a three-hybrid analysis in yeast. In this system, a catalytically inactive, substrate-trap form of PTEN is introduced as `bait,' a mammalian library is introduced as `fish,' and a tyrosine kinase is included to phosphorylate the `fish.' Tyrosine phosphorylated fish are expected to stably bind to the substrate-trap `bait,' thus activating transcriptional reporters. From a screen of about 200,000 clones, we isolated 52 interactors. Fifty of these were not dependent on the tyrosine kinase (i.e, they were not three-hybrid dependent). These fifty clones all encoded the same protein, a novel 20 kilodalton protein similar to the regulatory subunit of the serine/threonine protein phosphatase, Calcineurin. As the regulatory subunit of Calcineurin binds and confers calcium sensitivity to this enzyme, it is possible that PTEN is also regulated in a similar manner by calcium.

The remaining two PTEN interactors were three-hybrid dependent and encoded the same protein. This protein does not contain any recognizable motifs, but is closely related to a protein in the C. elegans database. Mutation of a tyrosine residue within the C-terminus of this protein abolishes its interaction with PTEN, showing that the interaction indeed requires tyrosine phosphorylation. Thus, this protein may represent a bona-fide substrate for PTEN.

Using a variety of mutant forms of human Pak1, we have shown that Pak1 function is required for activation of stress-activated protein kinases by Cdc42 and Rac1, as well as for transformation by activated forms of these GTPases. We have further shown that activated forms of Pak1 induce rapid formation of membrane ruffles and focal complexes. Thus, Paks are potentially bifunctional proteins, affecting both gene transcription (through a kinase cascade) and actin dynamics (via an unknown mechanism). We are currently attempting to identify key targets for Pak, both in mammalian fibroblasts and in fission yeast (see below).

One of the more interesting features of Pak's ability to remodel filamentous actin in mammalian fibroblasts is that this activity is partly independent of its kinase function. That is, kinase-dead forms of Pak have cytoskeletal activity, as does the amino-terminal (non-catalytic) domain of this protein. Mutations in certain conserved proline residues, which we have previously shown to serve as ligands for SH3 ligands in vivo, reduce the cytoskeletal activity of Pak, suggesting that SH3-containing effector proteins bind to these regions and mediate the effects of Pak on the actin cytoskeleton. Thus, the primary affect of the protein kinase Pak on cytoskeletal architecture is not through its kinase activity (though this may have some role), but rather because of its ability to recruit additional effectors to its amino terminus. As the relevant proline-rich domains in the N terminus are highly conserved, these data suggest that the mechanisms used by Pak to affect cell shape and polarity are likewise conserved. We are, therefore, currently using interaction-traps and other biochemical means to isolate proteins that associate with this the amino terminus of Pak.

Analysis of gene function in fission yeast is useful because signal transduction systems in this single-celled haploid organism closely resemble those found in higher eukaryotes. We, therefore, isolated two Paks from fission yeast, termed pak1 and pak2. We found that the pak1 gene encodes a protein kinase that is essential for viability. By overexpressing WT or mutant, catalytically inactive forms of Pak1, we established that this kinase not only regulates the mating pathway (like its homolog Ste20 in budding yeast), but is also involved in the control of cell polarity. The related pak2 gene is not essential for viability, but appears to have partially overlapping function with pak1. We are currently using a genetic screen to identify downstream targets of Pak1 and Pak2 that regulate cell morphogenesis.

ADAM, L., VADLAMUDI, R., KONDAPAKA, S.B., CHERNOFF, J., MENDELSOHN, J., KUMAR, R. Heregulin regulates cytoskeletal reorganization and cell migration through the p21-activated kinase-1 via phosphatidylinositol-3 kinase. J. Biol. Chem. 273:28238-28246, 1998.

GRAVES, J.D., DRAVES, K.E., GOTOH, Y., AMBROSE, D., CHERNOFF, J., CLARK, E.A., KREBS, E.G. Caspase-mediated cleavage and activation of the mammalian Ste20-like kinase Mst1 during apoptosis. EMBO J. 17:2224-2234, 1998.

SELLS, M.A., BARRATT, J.T., CAVISTON, J., OTTILIE, S., LEBERER, E., CHERNOFF, J. Characterization of Pak2, a pleckstrin-homology-domain-containing p21-activated protein kinase from fission yeast. J. Biol. Chem. 273:18490-18498, 1998.

WARDENBURG, J.B., PAPPU, R., BU, J.-Y., CHERNOFF, J., STRAUS, D., CHAN, A.C. Regulation of the T cell cytoskeleton and Jnk activation by the co-localization of Vav and Nck through phosphorylation of SLP-76. Immunity (in press).

ZHANG, B., CHERNOFF, J., ZHENG, Y. Interaction of Rac1 with GTPases-activating proteins and putative effectors. J. Biol. Chem. 273:8776-8782, 1998.

LIU, F., SELLS, M.A., CHERNOFF, J. Protein tyrosine phosphatase 1B negatively regulates integrin signaling. Curr. Biol. 8:173-176, 1998.

^a   F. Liu*: Memorial Sloan-Kettering Cancer Center, New York, NY 10021

^b   J. Barratt*: Temple University Medical School, Philadelphia, PA 19140

*   Formerly with J. Chernoff's laboratory

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

Fox Chase Cancer Center

Scientific Report 1998