Activated PAK4 Regulates Cell Adhesion and Anchorage-Independent Growth

doi:10.1128/MCB.21.10.3523-3533.2001

Mol Cell Biol. 2001 May; 21(10): 3523–3533.

doi: 10.1128/MCB.21.10.3523-3533.2001

PMCID: PMC100274

Activated PAK4 Regulates Cell Adhesion and Anchorage-Independent Growth

Jian Qu,¹ Marta S. Cammarano,¹ Qing Shi,² Kenneth C. Ha,³ Primal de Lanerolle,³ and Audrey Minden¹^*

Department of Biological Sciences, Columbia University, New York, New York 10027¹; Duke University Medical Center, Department of Cell Biology, Durham, North Carolina 27710²; and Department of Physiology and Biophysics, University of Illinois at Chicago, Chicago, Illinois 60612³

^*Corresponding author. Mailing address: Columbia University, Biological Sciences MC 2460, Sherman Fairchild Center, Room 813, 1212 Amsterdam Ave., New York, NY 10027. Phone: (212) 854-5632. Fax: (212) 854-7655. E-mail: agm24/at/columbia.edu.

Received October 25, 2000; Revisions requested December 28, 2000; Accepted February 14, 2001.

This article has been cited by other articles in PMC.

Abstract

The serine/threonine kinase PAK4 is an effector molecule for the Rho GTPase Cdc42. PAK4 differs from other members of the PAK family in both sequence and function. Previously we have shown that an important function of this kinase is to mediate the induction of filopodia in response to activated Cdc42. Since previous characterization of PAK4 was carried out only with the wild-type kinase, we have generated a constitutively active mutant of the kinase to determine whether it has other functions. Expression of activated PAK4 in fibroblasts led to a transient induction of filopodia, which is consistent with its role as an effector for Cdc42. In addition, use of the activated mutant revealed a number of other important functions of this kinase that were not revealed by studying the wild-type kinase. For example, activated PAK4 led to the dissolution of stress fibers and loss of focal adhesions. Consequently, cells expressing activated PAK4 had a defect in cell spreading onto fibronectin-coated surfaces. Most importantly, fibroblasts expressing activated PAK4 had a morphology that was characteristic of oncogenic transformation. These cells were anchorage independent and formed colonies in soft agar, similar to what has been observed previously in cells expressing activated Cdc42. Consistent with this, dominant-negative PAK4 mutants inhibited focus formation by oncogenic Dbl, an exchange factor for Rho family GTPases. These results provide the first demonstration that a PAK family member can transform cells and indicate that PAK4 may play an essential role in oncogenic transformation by the GTPases. We propose that the morphological changes and changes in cell adhesion induced by PAK4 may play a direct role in oncogenic transformation by Rho family GTPases and their exchange factors.

Members of the Rho family of small GTPases, including Cdc42, Rac, and Rho, were first identified as proteins that have key roles in regulating the organization of the actin cytoskeleton. They were shown to induce the production of filopodia, lamellipodia, and stress fibers, respectively (17, 30, 37, 38). Subsequently they were found to have other functions, including the regulation of cell proliferation and activation of the JNK and p38 mitogen-activated protein kinase (MAP kinase) pathways (3, 5, 8, 11, 28, 51). When improperly regulated, the Rho GTPases also play key roles in oncogenic transformation and tumor invasiveness (26, 27, 33, 36, 52) which may be directly related to changes in morphology and activation of specific signal transduction pathways.

The characterization of molecular targets for the Rho proteins is important for understanding their functions. The PAK family of serine/threonine kinases interacts directly with GTP-loaded Rac and Cdc42 through a GTPase binding domain (GBD) (1, 2, 4–6, 23, 24, 42). The first members of the family to be identified include the closely related human PAK1 and -2, mouse PAK3, and the rat homologues PAK α, β, and γ (4, 5, 23, 24). One possible function of the PAKs is the regulation of the organization of the actin cytoskeleton. PAK1 has been reported to induce filopodia and membrane ruffles (43) and to localize to polymerized actin (10, 43). These cytoskeletal changes, however, occur independently of PAK1's ability to bind Rho GTPases and are partly independent of PAK1's kinase activity (43). Others have found that PAK α and PAK2 do not induce filopodia or lamellipodia but instead have a role in the dissolution of stress fibers, down-regulation of focal adhesions, and cell retraction (22, 50). Thus, the exact roles for PAKs 1, 2, and 3 in cytoskeletal regulation remain to be fully clarified.

In addition to cytoskeletal organization, the Rho proteins also play key roles in cell proliferation and oncogenic transformation. Constitutively active mutants of the exchange factors for the GTP binding proteins, members of the Dbl family, are potent oncogenes (7). Cdc42, Rac, and Rho were all shown to be necessary for oncogenic transformation by oncogenic Dbl, and each GTPase appears to contribute to different aspects of transformation, including anchorage-independent growth, superoxide production, and loss of contact inhibition, respectively (20). Although PAK1 may have a role in transformation and cell survival induced by oncogenic Ras (44–46) in some cells, a direct role for the PAKs in transformation by the Rho family GTPases has not been demonstrated. In fact, Cdc42C40 and RacC40, effector mutants which cannot bind to PAKs 1, 2, or 3 (1, 13, 18), maintain the ability to induce cytoskeletal changes as well as some of the hallmarks of oncogenic transformation (13, 18). These results indicate that PAKs 1, 2, and 3 may not be essential for the induction of oncogenic transformation in response to these GTPases.

The newest member of the PAK family is PAK4 (1). PAK4 is a new type of PAK protein and it differs significantly from the other PAKs in sequence. PAK4 lacks several key features characteristic of PAKs 1, 2, and 3, including four proline-rich motifs, an autoinhibitory domain, and a putative G βγ binding site (9, 16). PAK4 does contain a modified GBD, however, and it interacts with GTP-loaded Cdc42 and has a weaker interaction with GTP-loaded Rac. Unlike the other PAKs, PAK4 even interacts with the effector loop mutant Cdc42C40 (1). Importantly, PAK4 was found to be a link between Cdc42 and filopodium formation. Expression of Cdc42V12 together with PAK4 leads to a prolonged induction of filopodia (1), and this depends strictly on PAK4's kinase activity and on its binding to Cdc42. PAK4 does not mediate all of Cdc42's functions, however. For example, it is a relatively weak activator of the signaling pathways that lead to activation of the JNK and p38 MAP kinase pathways (1). It is therefore thought to be primarily a mediator of the cytoskeletal changes induced by Cdc42 rather than a regulator of the JNK or p38 pathways.

In order to gain a better understanding of the functions of PAK4, we have generated and analyzed a constitutively active PAK4 mutant, PAK4(S445N). Expression of PAK4(S445N) in fibroblasts resulted in a transient induction of filopodia, a reduction of stress fibers, decreased adhesion to the extracellular matrix, and cell rounding. Interestingly, cells expressing PAK4(S445N) had a transformed morphology. These cells formed colonies in soft agar, similar to what has been described for cells expressing activated Cdc42 (19, 20, 34). Furthermore, dominant-negative PAK4 partially inhibited focus formation in response to oncogenic Dbl in fibroblasts. Taken together, our results suggest that PAK4 is an essential mediator of anchorage-independent growth induced by Cdc42 and by Dbl family exchange factors.

MATERIALS AND METHODS

Plasmids.

Expression plasmids encoding influenza hemagglutinin (HA)-tagged PAK4wt, HA-tagged PAK4(S474E), and HA-tagged PAK4(K350M) are described in reference 1. HA-PAK4R (1) containing the PAK4 regulatory domain was generated by inserting the HindIII/PstI site from 3HA-PAK4-pBS into the HindIII/BglIII site of the SRα vector, together with a pair of BglII/PstI adapter primers. PAK4RΔGBD (1) is the same as PAK4R but lacks the first 113 amino acids, which include the GBD. To generate activated PAK4, serine 445 on PAK4 was changed to an asparagine using site-directed mutagenesis (Stratagene QuickChange kit), and asparagine 445 was introduced into both SRα-PAK4wt and -PAK4(S474E) to generate PAK4(445N) and PAK4(474E/445N), respectively. PAK4(445N) and PAK4(474E/445N) had equivalent levels of kinase activity and induced identical types of morphological changes in transient transfections (including a rounded morphology and loss of stress fibers and focal adhesions). However, PAK4(474E/445N) was expressed at higher levels (its expression level was comparable to that of wild-type PAK4) and was therefore used in all subsequent experiments. This mutant is subsequently referred to as PAK4(S445N). To construct Myc-tagged PAK4(S445N), an EcoRI fragment containing the entire coding sequence of PAK4(S445N) (without the HA tag) was removed from SRα and inserted into the EcoRI site of the pCAN-Myc2 vector. To subclone PAK4 into the pIRES2-EGFP vector (Clontech), HA-PAK4wt was removed pBluescript KS II(+) as an XhoI/StuI fragment and inserted into the XhoI/SmaI cut of the pIRES2-EGFP vector. To construct the pIRES2-EGFP vector containing PAK4(S445N), an EcoRI/BamHI fragment of pIRES2-EGFP-PAK4 was replaced by the EcoRI/StuI fragments of SRα-PAK4(S445N). To construct pLPC-PAK4-WT and pLPC-PAK4 mutants, HindIII/XhoI fragments from pCAN-Myc2-PAK4 were inserted into the HindIII/XhoI sites of the pLPC vector. The pLPC vector is a retroviral expression vector with a puromycin resistance marker (a gift from R. Prywes). JNK, Rac2L61, Dbl, and Cdc42V12 have been described previously (28).

Antibodies and phalloidin.

The anti-HA antibody (MMS-101P) was obtained from Covance, and anti-Myc (sc-40) and anti-Rho antibodies (sc-418) were obtained from Santa Cruz Biotechnology. Mouse monoclonal anti-PAK4 antibody was generated against full-length human PAK4 protein (purified using the Gibco BRL Bac to Bac baculovirus system) in conjunction with PharMingen. Fluorescein isothiocyanate (FITC)-conjugated phalloidin (F-432) was obtained from Molecular Probes, antivinculin antibody (hVin-1) was obtained from Sigma, and rhodamine-conjugated goat anti-mouse immunoglobulin G (IgG) antibody (31663) was obtained from Pierce. Antipaxillin antibody (tetramethyl rhodamine isothiocyanate conjugated) was obtained from Transduction Laboratories. Alkaline phosphatase-conjugated goat anti-mouse IgG (59296) and peroxidase-conjugated goat anti-mouse IgG (55550) for Western blots were obtained from ICN. Antibodies to the regulatory 20-kDa myosin light chain (MLC₂₀) were produced by immunizing rabbits with MLC₂₀ purified from chicken gizzard smooth muscle. The antibodies were affinity purified by applying the immune serum to an MLC₂₀-Sepharose 4B column. Western blot analyses showed that these antibodies react specifically with purified smooth muscle MLC₂₀ and MLC₂₀ found in cell extracts prepared from smooth muscle and nonmuscle cells. Antibody against phosphorylated MLC was a gift from F. Matsumura.

Cell culture and transfection.

293T and Rat1 cells were grown at 37°C in 5% CO₂ and cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum. NIH 3T3 cells and v-Ki-Ras-transformed NIH 3T3 cells (31) were cultured in DMEM containing 10% bovine calf serum. Rat1 and NIH 3T3 cells stably transfected with either pLPC, PAK4, or PAK4 mutants were grown as described above in the presence of 1.5 or 2.0 μg of puromycin per ml, respectively. All media were supplemented with 100 U of penicillin/ml, 100 μg of streptomycin/ml, and 1 mM glutamine. Transient transfection assays were carried out using the Lipofectamine (Gibco-BRL) method according to the manufacturer's protocol. Stable cell lines were generated by retroviral infection. Briefly, 293T cells were transfected with empty pLPC vector, pLPC-Myc-PAK4, pLPC-Myc-PAK4(S474E), or pLPC-Myc-PAK4(S445N), together with helper plasmid ψ by the calcium phosphate precipitation method. Supernatants containing the released viruses were collected from 293T cells 2 days after transfection and filtered through a 0.45-μm-pore-size filter. The virus was then used to infect either Rat1 cells or NIH 3T3 cells. Cells were selected with puromycin (1.5 μg/ml for Rat1 cells and 2.0 μg/ml for NIH 3T3 cells) and colonies were picked approximately 2 weeks after selection. Expression of PAK4 was determined by Western blotting and immunofluorescence microscopy using a monoclonal antibody against the Myc tag.

Protein kinase assays.

To assay the kinase activity of PAK4, NIH 3T3 cells were transfected with either empty SRα expression vector or expression vectors containing either wild-type PAK4, PAK4(S474E), or PAK4(S445N) fused with the indicated epitope tags. Cells were harvested in M2 buffer (29) 48 h after transfection. PAK4 was immunopurified from approximately 100 μg of cell extracts using antibody generated against the epitope tag. PAK4 protein kinase activity was then assessed as described previously (1). JNK, ERK, and p38 activities were measured as described previously (28). MLC kinase (MLCK) assays were carried out as described previously (41).

Western blots.

Western blots were carried out as described previously (1).

Immunofluorescence microscopy.

Cells stably expressing wild-type or mutant PAK4 were grown on 22-mm glass coverslips in complete medium and 24 h later were fixed in 4% paraformaldehyde for 10 min at room temperature. For transient transfection experiments, cells were fixed 48 h after transfection. Fixed cells were permeabilized with phosphate-buffered saline (PBS) containing 0.1% Triton X-100 for 20 min. Cells were then incubated with the primary monoclonal anti-HA antibodies or anti-Myc antibodies for 60 min. The coverslips were washed with PBS containing 0.1% Triton X-100 and incubated for 30 min with the secondary rhodamine-conjugated anti-mouse antibody (Molecular Probes). To visualize F-actin, cells were washed again and were incubated with FITC-conjugated phalloidin. For vinculin staining, the antivinculin monoclonal antibody (hVIN-1; Sigma) was used as the primary antibody, and the rhodamine-conjugated anti-mouse antibody was used as the secondary antibody. For paxillin staining, cells were stained with paxillin-tetramethyl rhodamine isothiocyanate for 60 min after fixation and permeabilization. Fluorescence photomicroscopy was carried out with appropriate filters for fluorescence detection.

Soft agar assays.

Approximately 10⁴ cells were suspended in 2 ml of 0.3% Bacto agar in DMEM containing 10% fetal bovine serum (Rat1 cells) or 10% bovine calf serum (NIH 3T3 cells), antibiotics, and glutamine and were overlaid on 2 ml of 0.6% Bacto agar in the same medium in 35-mm-diameter dishes. Each cell line was tested in duplicate wells. Colonies were visualized under an inverted light microscope after 2 to 3 weeks.

Focus formation assays.

Focus formation assays with NIH 3T3 cells were carried out as described previously (28).

Cell spreading assays.

Equal numbers of cells (5 × 10³) were plated onto either 96-well plates or 35-mm-diameter plates in the spreading medium (DMEM, 0.5% bovine serum albumin, 2 mM l-glutamine, penicillin-streptomycin, 20 mM HEPES) and incubated at 37°C. At 10- or 15-min intervals the medium from each well was removed and cells were fixed with fixative solution (4% paraformaldehyde in phosphate buffer) for 30 min at 4°C followed by incubation with dye solution (0.5% [wt/vol] toluidine blue [Sigma] in fixative solution) at 4°C for 1 h. Cells were rinsed with PBS and then visualized under a 10× objective, and the fraction of spread cells was determined at each time point.

RESULTS

PAK4(S445N) has increased kinase activity compared to wild-type PAK4 but does not lead to an increased activation of the JNK pathway.

Two PAK4 mutants were made in an attempt to generate an activated PAK4. The first mutation was PAK4(S474E) in which one of the predicted autophosphorylation sites was converted to a glutamate (1). The second mutant was PAK4(S445N) in which a serine-to-asparagine mutation was introduced at amino acid 445 in the catalytic loop within kinase subdomain VIb. Mutation of this residue is predicted to stabilize the catalytic loop (47). Expression vectors containing epitope-tagged PAK4(S445N), wild-type PAK4, or PAK4(S474E) were transfected into NIH 3T3 cells. After transient expression, PAK4 was immunopurified from cell lysates using antibodies directed against the epitope tags and used to phosphorylate histone H4 (HH4) in an in vitro kinase assay. PAK4(S474E) exhibited a small increase in its autophosphorylation activity compared with wild-type PAK4. Although serine 474 is a predicted autophosphorylation site and is necessary for PAK4 activity (1), our results suggest that additional sites within PAK4 may be weakly phosphorylated by the PAK4(S474E) mutant. In spite of this, PAK4(S474E) was actually weaker than wild-type PAK4 in HH4 phosphorylation (Fig. (Fig.1A).1A). Strikingly, however, PAK4(S445N) showed an approximately 30-fold increase in both autophosphorylation and HH4 phosphorylation compared with wild-type PAK4 (Fig. (Fig.1A).1A). Both Cdc42 and Rac are strong activators of the JNK and p38 MAP kinase pathways. Although PAK4 can activate the JNK pathway, its activation ability is weak compared with that by activated Rac or Cdc42 and its primary function is most likely cytoskeletal regulation rather than activation of the JNK pathway (1). To determine whether PAK4(S445N) was a stronger activator of the JNK pathway than wild-type PAK4, NIH 3T3 cells were transfected with Myc-tagged PAK4 expression vectors or empty SRα vector together with an expression vector for HA-tagged JNK. After transient expression, HA-JNK was immunopurified using antibodies against the HA tag. Immune complex kinase assays were carried out using glutathione S-transferase–c-Jun as substrate. As shown in Fig. Fig.1B,1B, PAK4(S445N) did not activate the JNK pathway any more than wild-type PAK4 did. Like wild-type PAK4, PAK4(S445N) was not an efficient activator of the p38 or ERK pathways (data not shown). These results suggest that despite its high activity, PAK4(S445N) retains the substrate specificity of wild-type PAK4.

FIG. 1

PAK4(S445N) has increased kinase activity compared with wild-type PAK4 but is not enhanced in its activation of the JNK pathway. (A) NIH 3T3 cells were transiently transfected with either empty vector (Con) or vectors containing HA-tagged wild-type PAK4 (more ...)

Transient expression of PAK4(S445N) causes cell rounding in NIH 3T3 cells.

To determine whether PAK4(S445N) has an effect on cell morphology, NIH 3T3 cells were transiently transfected with either empty vector, wild-type PAK4, or PAK4(S445N) in a bicistronic vector containing an internal ribosome entry sequence and enhanced green fluorescent protein (EGFP). After 24 h, PAK4(S445N)-expressing cells had a distinctly rounded appearance and adhered poorly to the surface of the dish (Fig. (Fig.2).2). Trypan blue exclusion experiments confirmed the viability of the PAK4(S445N)-expressing cells (data not shown). Wild-type PAK4-expressing cells were more spread out and looked similar to cells containing the empty vector. Similar results were found in several other cell types that were transfected with PAK4(S445N), including 293T cells and PAE cells (data not shown).

FIG. 2

Transient expression of PAK4(S445N) results in cell rounding. NIH 3T3 cells were transfected with 1 μg of either EGFP expression vector (top), PAK4-IRES2-EGFP (middle), or PAK4(S445N)-IRES2-EGFP (bottom). Twenty-four hours after transfection, (more ...)

Fibroblasts stably expressing PAK4(S445N) show distinctive morphological and cytoskeletal changes.

To further characterize the morphological changes induced by constitutively active PAK4, stable cell lines were generated in which Rat1 cells and NIH 3T3 cells overexpressed either empty vector (pLPC), Myc-tagged wild-type PAK4, PAK4(S474E), or PAK4(S445N). Expression of PAK4 in the stable cell lines was analyzed by Western blotting (Fig. (Fig.3A)3A) and immunofluorescence microscopy. Immunofluorescence analysis indicated that each cell line was homogenous for expression of PAK4 (data not shown). The morphologies of the pLPC- and PAK4(S445N)-expressing Rat1 cells are shown in Fig. Fig.3B3B and C. Figure Figure3B3B shows the morphologies of the cells 10 min and 1 h after plating onto fibronectin-coated surfaces. At 10 min, PAK4(S445N)-expressing cells were rounded but had numerous long filopodia. These filopodia were quite motile as assessed by video microscopy, and a representative cell that was photographed at 20-s intervals approximately 10 min after plating is shown in panels i, ii, and iii. At 1 h, some of the cells had a more flattened morphology, but filopodia could still be visualized at the edges of the cells. See panel iv for an example of a cell that had begun to spread onto the surface. Filopodia could be visualized up to 3 h after plating. The filopodia were somewhat transient, however, because by 24 h filopodia could not be visualized in most of the Rat1 cells (see below). In contrast to the PAK4(S445N) cells, very few filopodia could be detected on the control pLPC cells either 10 min or 1 h after plating (panels v and vi). These results indicate that like wild-type PAK4 (1), PAK4(S445N) can induce filopodia. Unlike wild-type PAK4, however, PAK4(S445N) can induce filopodia even without cotransfection of activated Cdc42.

FIG. 3

Morphological changes in fibroblasts stably expressing PAK4(S445N). (A) Representative Western blot analysis of Myc-tagged PAK4 expression in the different stable cell lines. Cell lysates (25 μg) from Rat1 cells or NIH 3T3 cells stably expressing (more ...)

The cell lines were further analyzed after the cells had been allowed to settle onto the dish for 24 h. By this time the Rat1/pLPC cells had a spread morphology similar to that of the parental cells. In contrast, PAK4(S445N)-expressing cells consisted of many rounded cells, together with cells that had a more elongated morphology (Fig. (Fig.3C,3C, top panels). Filopodia were no longer visible in most of the cells by this time. To observe the actin cytoskeleton and focal adhesions, the stable cell lines were fixed 24 h after plating onto fibronectin-coated coverslips and stained with either FITC-conjugated phalloidin or antivinculin antibody. The middle and bottom panels of Fig. Fig.3C3C show representative photos of Rat1/pLPC cells and clusters of the more elongated Rat1/PAK4(S445N) cells. (Because of their compact morphologies, the more rounded cells could not be easily discerned by immunofluorescence. They can be visualized as bright spots of fluorescence in the middle panels.) Strikingly, we found that the cells containing PAK4(S445N) had a dramatic reduction in stress fibers (middle panels). When stained with antibodies against vinculin, the PAK4(S445N)-expressing cells showed a nearly complete loss of vinculin, suggesting a loss of focal adhesions (bottom panels). Similar results were seen when cells were stained with antibodies against paxillin, another component of focal adhesions (data not shown). For all experiments described above, cells expressing wild-type PAK4 and PAK4(S474E) appeared similar to cells containing the empty pLPC vector, although there was a slightly increased number of rounded cells in the PAK4(S474E) cells (data not shown).

The morphological changes induced by PAK4(S445N) were not due to clonal variations, because similar results were observed in NIH 3T3 cells (Fig. (Fig.3D).3D). Strikingly, however, in NIH 3T3 cells the filopodia were less transient and could be observed in many of the cells even 24 h after plating. Transient transfection of PAK4(S445N) led to morphological changes that were similar to those seen in the stable cell lines. In fact, in NIH 3T3 cells at least 90% of cells transfected with PAK4(S445N) showed a decrease in focal adhesions and dissolution of stress fibers (data not shown).

PAK4 does not inhibit MLC phosphorylation.

The dissolution of stress fibers is reminiscent of previous results seen with activated PAK1. Activated PAK1 was shown to cause the dissolution of stress fibers by phosphorylating MLCK and inhibiting MLCK activity, thereby decreasing MLC phosphorylation on serine 19 (41). PAK1 thus inhibits stress fiber formation, since MLC phosphorylation is required for the actin-myosin interactions that are present in actin stress fibers. In contrast to PAK1, however, activated PAK4 did not phosphorylate MLCK (Fig.(Fig.4A),4A), although in parallel experiments MLCK was phosphorylated by activated PAK1 (data not shown). To examine MLC phosphorylation in PAK4(S445N)-expressing cells, equal volumes of lysates from pLPC- and PAK4(S445N)-expressing NIH 3T3 cells were blotted and probed with anti-phospho-MLC antibody and anti-MLC antibody (Fig. (Fig.4B).4B). Although a lower amount of MLC was detected in the lysates from the PAK4(S445N)-expressing cells, there was not a corresponding decrease in MLC phosphorylation (Fig. (Fig.4B).4B). Thus, PAK4 appears to differ from PAK1 in its substrate specificity and in the mechanism by which it induces cytoskeletal reorganization.

FIG. 4

PAK4 does not phosphorylate MLCK or regulate MLC phosphorylation. (A) Wild-type PAK4, PAK4(S474E), or PAK4(S445N) were immunopurified from NIH 3T3 stable cell lines using anti-Myc antibody. Empty pLPC vector cells were used as a control. Immunoprecipitates (more ...)

Fibroblasts expressing PAK4(S445N) are deficient in their ability to spread on extracellular matrix-coated surfaces.

Because of the low level of vinculin staining in the PAK4(S445N)-expressing cells, we were interested in determining whether PAK4-expressing cells had a defect in cell spreading. To analyze cell spreading, the Rat1 stable cell lines were plated onto either fibronectin- or collagen-coated wells. At 10- to 15-min intervals cells were fixed and stained as described in Materials and Methods. The number of spread cells was counted at each time point. The results indicate that Rat1 cells containing PAK4(S445N) spread significantly more slowly on both surfaces than did Rat1 cells containing empty vector (Fig. (Fig.5).5).

FIG. 5

Inhibition of spreading in cells expressing PAK4(S445N). Equal numbers of stable Rat1 cells expressing empty vector (pLPC) (diamonds) or PAK4(S445N) (squares) were plated onto dishes that were coated with fibronectin or collagen. At the indicated time (more ...)

PAK4-overexpressing cells exhibit anchorage-independent growth in soft agar.

The rounded morphologies of PAK4(S445N)-overexpressing cells and the loss of stress fibers and focal adhesions resembled fibroblasts that have been transformed with oncogenes such as the Ras oncogene. One of the hallmarks of oncogenic transformation is the loss of anchorage-dependent growth as demonstrated by the ability to form colonies on soft agar. Because the PAK4(S445N)-expressing cells had a decrease in cell adhesion without any noticeable change in cell viability, we were interested in determining whether PAK4(S445N) cells exhibited a lack of anchorage-dependent growth. We analyzed two independent Rat1 cell lines and two NIH 3T3 cell lines overexpressing PAK4(S445N), wild-type PAK4, or empty pLPC vector. For comparison, we analyzed NIH 3T3 cells that were stably transfected with oncogenic v-Ki-Ras (31). Equal numbers of each cell type were plated in soft agar as described in Materials and Methods. We found that all of the Rat1 and NIH 3T3 cell lines containing PAK4(S445N) produced colonies on soft agar similar to Ras-transformed NIH 3T3 cells (Fig. (Fig.6A6A and B). In each case [both PAK4(S445N)- and v-Ki-Ras-transformed cells] approximately 18% of the total plated cells produced foci. Neither Rat1 cells nor NIH 3T3 cells expressing wild-type PAK4 or pLPC formed any colonies in soft agar (Fig. (Fig.6A6A and B). Likewise, PAK4(S474E)-expressing cells formed very few colonies in soft agar (data not shown). These differences reflect the anchorage-independent growth characteristic of PAK4(S445N)-expressing cells rather than an increase in their growth rates, as PAK4(S445N)-expressing cells did not grow at an enhanced growth rate compared with the pLPC-containing cells under normal adherent culture conditions (data not shown).

FIG. 6

PAK4(S445N) induces anchorage-independent growth, and PAK4 is required for oncogenic Dbl-induced focus formation. (A and B) PAK4(S445N) induces anchorage-independent growth. Soft agar growth assays were carried out as described in Materials and Methods. (more ...)

To determine whether PAK4 is necessary for transformation, NIH 3T3 cells were transfected with oncogenic Dbl together with either empty vector, dominant-negative Cdc42 (Cdc42N17), or one of three different dominant-negative PAK4 mutants. The first mutant was PAK4(K350M), which is a full-length PAK4 without kinase activity (1). The second was PAK4R (1), which contains only the PAK4 regulatory domain and no kinase domain. The third was PAK4RΔGBD (1), which lacks both the kinase domain and the GBD so that it can no longer bind to Cdc42. Foci were scored 2 weeks after transfection. Expression of all three dominant-negative PAK4 mutants resulted in a substantial reduction in Dbl-induced foci, indicating that PAK4 plays an important role in transformation by Dbl (Fig. (Fig.6C).6C). Dominant-negative Cdc42 completely blocked focus formation by oncogenic Dbl. When used at the same concentrations, none of the dominant-negative mutants effectively blocked focus formation by oncogenic Ras (Fig. (Fig.66C).

DISCUSSION

PAK4 was originally identified as a molecular target for the Rho GTPase Cdc42, and it plays a key role in Cdc42's ability to induce filopodia (1). Here we have generated a constitutively active mutant of PAK4 [PAK4(S445N)] in order to better characterize the function of this serine/threonine kinase. Fibroblasts expressing PAK4(S445N) transiently induced filopodia when they were plated onto fibronectin, which is consistent with PAK4's function as a target for Cdc42. Activated PAK4 also had a number of other functions which were not revealed previously by studying wild-type PAK4. Most importantly, expression of PAK4(S445N) caused a transformed morphology in fibroblasts. Fibroblasts stably expressing PAK4(S445N) lost their anchorage-dependent growth requirement and formed colonies in soft agar. Similar results were shown previously for cells expressing activated Cdc42 (19, 20, 34). Consistent with this, dominant-negative PAK4 mutants inhibited focus formation by oncogenic Dbl, an exchange factor for the Rho GTPases. We propose that PAK4(S445N)'s ability to transform cells is due at least in part to its ability to induce morphological changes, including cell rounding, loss of stress fibers and focal adhesions, and decreased spreading.

PAK4(S445N) is the first full-length activated PAK4 to be generated. PAK4 as well as the PAK4 Drosophila melanogaster homologue mushroom body tiny (mbt) (25) and an uncharacterized Caenorhabditis elegans homologue, C45B11.1, all encode for proteins with a serine at residue 445. Most serine/threonine protein kinases, including other members of the PAK family, however, have an asparagine residue at the corresponding site. The asparagine at this position serves to stabilize the catalytic loop by hydrogen bonding to a conserved aspartate (corresponding to D440 in PAK4) within the loop (47). The serine-to-asparagine conversion in PAK4 is therefore thought to function by strengthening the catalytic loop. Although PAK4(S445N) has strong kinase activity, its substrate specificity appears to be unaltered. Like wild-type PAK4, for example, PAK4(S445N) does not phosphorylate MLCK. Furthermore, PAK4(S445N) does not activate the ERK or p38 pathways and does not activate the JNK pathway any further than wild-type PAK4 does. This is consistent with our previous prediction that PAK4 is primarily a mediator of the cytoskeletal changes induced by Cdc42 rather than a key player in the Cdc42-to-JNK pathway (1).

The morphological changes resulting from PAK4(S445N) expression do not appear to be nonspecific effects due to the high activity of this kinase. Rather, they seem to be an amplification of wild-type PAK4's activity. Overexpression of even wild-type PAK4 in some cells, such as HeLa cells, results in a slightly rounded shape and lowered adhesion (N. Gnesutta and A. Minden, unpublished results). Likewise, another activated PAK4 mutant, PAK4Δ, in which the PAK4 regulatory domain is deleted (1), induced some of the morphological changes induced by PAK4(S445N), such as a decrease in stress fibers and focal adhesions, but not filopodium formation (data not shown). Furthermore, some of the functions of PAK4(S445N) are consistent with PAK4's role as an effector for Cdc42. For example, the transient induction of filopodia and the ability to confer anchorage-independent growth are functions which are shared by PAK4 and activated Cdc42 (17, 19, 20, 30, 34). Importantly, the activities of PAK4(S445N) appear to be specific to PAK4. For example, unlike activated PAK4, constitutively active PAK1(T423E) does not induce anchorage-independent growth (data not shown) (45). PAK4(S445N) also differs from activated PAK1 in substrate specificity and the mechanism by which it induces cytoskeletal changes. In particular, we have found that unlike PAK1, the dissolution of stress fibers by PAK4 does not appear to be mediated by MLCK phosphorylation and the subsequent regulation of MLC phosphorylation.

While many of the functions of PAK4(S445N) are characteristic of Cdc42 functions, some of its functions, such as cell rounding and the dissolution of stress fibers, may be distinct from Cdc42 functions. The identification of new PAK4 activators other than Cdc42 will be important in order to better understand whether it also serves as a target for other types of signaling and cytoskeletal regulatory proteins. The mechanisms by which PAK4 induces its various morphological effects are not yet clearly understood. For example, as discussed above, unlike for PAK1, the dissolution of stress fibers by PAK4 is not due to a decrease in MLC phosphorylation. It is interesting to note, however, that PAK4 does lead to a consistent decrease in MLC expression which could potentially contribute to the dissolution of stress fibers. Another possibility is that PAK4 could inhibit Rho activity, which normally functions to stimulate stress fiber formation. We have in fact found that Rho activity is decreased (but not abolished) in PAK4(S445N)-expressing cells (data not shown) and it will be interesting to determine whether this is related to PAK4's role in stress fiber dissolution. It should be noted, however, that activated Rho triggers stress fiber formation by a mechanism that involves multiple steps which include increased phosphorylation of MLC (15) and activation of LIMK1 (21, 32), yet we do not see a decrease in MLC phosphorylation (Fig. (Fig.4)4) or an inhibition of LIMK1 activity (data not shown) in PAK4(S445N)-expressing cells. Ultimately it will be important to identify new substrates for PAK4 in order to fully understand the mechanism by which it induces specific morphological changes.

The ability of PAK4(S445N) to induce a transformed phenotype in fibroblasts is especially intriguing because the regulation of cell proliferation, progression through the cell cycle, and oncogenic transformation are important functions of the Rho proteins (14, 18, 34–36). Most of the Dbl family exchange factors are potent oncogenes, and all three GTPases have important contributions to oncogenic transformation by Dbl (20). A number of target proteins for the Rho GTPases have been identified which may be necessary for oncogenic transformation, such as the Rho effector protein ROCK (39, 40, 48) and the Cdc42 target protein γ-COP (49). Neither of these, however, is sufficient to transform cells on its own. Likewise, activated PAK1 does not induce transformation on its own (44, 45), even though it has been shown to be necessary for transformation by oncogenic Ras in some cells (44, 45). Furthermore, effector loop mutants for Cdc42 and Rac which do not bind to PAKs 1, 2, or 3 can still induce many of the hallmarks of oncogenic transformation (13, 18), suggesting that these PAK family members are not required for transformation by the Rho family GTPases.

In contrast to other PAKs, we have found that activated PAK4 confers anchorage-independent growth on fibroblasts and leads to focus formation in soft agar assays. This work is the first study to show a direct role for a PAK protein in transformation. Importantly, constitutively active and cycling mutants of Cdc42 also induce anchorage-independent growth in fibroblasts (19, 20, 34). The induction of anchorage-independent growth has in fact been shown to be a major contribution of Cdc42 to transformation by oncogenic Dbl (20). Not only does activated PAK4 trigger anchorage-independent growth, but dominant-negative PAK4 also inhibits transformation by oncogenic Dbl, a potent activator of all three Rho family GTPases. This inhibition is not due merely to sequestering Cdc42, because a mutant that lacks the GBD also inhibits transformation. Dominant-negative PAK4 was less efficient at inhibiting Dbl-induced foci than was dominant-negative Cdc42 (Cdc42N17). However, it should be noted that the PAK4 mutants and Cdc42N17 cannot be directly compared with each other because we do not know whether they have the same capacities to act as dominant-negative mutants, i.e., to inhibit their endogenous counterparts. Furthermore, it should be noted that Cdc42N17 can bind directly to Dbl and therefore may be a particularly effective inhibitor of Dbl activity.

The mechanism by which PAK4 can regulate oncogenic transformation is not yet known. Our results suggest that activated PAK4 can induce anchorage-independent growth in the absence of any strong activation of several known signal transduction pathways, such as the JNK, p38, and ERK pathways, that lead to changes in gene expression patterns. Rather, we propose that changes in cell morphology and adhesion play a major role in PAK4's ability to transform cells. We do not know, however, whether all of the different cytoskeletal changes induced by PAK4(S445N) contribute to the oncogenic process. Further investigation will be necessary in order to determine which of the morphological changes induced by PAK4 are directly related to its role in transformation. Furthermore, as discussed above, although PAK4 was originally identified as a target for Cdc42, we cannot rule out the possibility that it could regulate transformation by a mechanism that is either partly or entirely independent of Cdc42. For example, recently we have found that PAK4 can protect cells against apoptosis and inhibit caspase activation in response to a variety of different stimuli, including serum withdrawal, UV irradiation, and tumor necrosis factor alpha stimulation (12). Since inhibition of apoptosis is an important part of oncogenic transformation, such a protective role is also likely to contribute to PAK4's role in oncogenic transformation.

We do not rule out the possibility that PAK4 may also mediate oncogenic transformation or morphological changes in response to other signaling enzymes besides Cdc42. We have found, however, that dominant-negative PAK4 is an inefficient inhibitor of Ras transformation. This is initially surprising because previous work using dominant-negative Cdc42N17 has indicated that Ras requires Cdc42 to produce foci (34). It should be noted, however, that the requirement for Cdc42 by Ras and Dbl cannot be directly compared, because Cdc42N17 is a much more effective inhibitor of Dbl foci than of Ras foci (Fig. (Fig.6).6). In future work it will be interesting to determine whether other oncogenes require signals through PAK4 to induce transformation and whether PAK4 can mediate responses that are independent of Cdc42.

ACKNOWLEDGMENTS

We thank M. Sheetz and A. Beg for critically reading the manuscript, R. Prywes and C. Prives for helpful discussions, and B. Dubin-Thaler for assistance with time lapse microscopy.

This work was supported by grant R01 HL 59618 to P.D.L. and grant R01 CA76342 and an American Scientist Development Grant Award from the American Heart Association to A.M.

REFERENCES

1. Abo A, Qu J, Cammarano M S, Dan C, Fritsch A, Baud V, Belisle B, Minden A. PAK4, a novel effector for Cdc42Hs, is implicated in the reorganization of the actin cytoskeleton and in the formation of filopodia. EMBO J. 1998;17:6527–6540. [PubMed]

2. Bagrodia S, Cerione R A. PAK to the future. Trends Cell Biol. 1999;9:350–355. [PubMed]

3. Bagrodia S, Derijard B, Davis R J, Cerione R A. Cdc42 and PAK-mediated signaling leads to Jun kinase and p38 mitogen-activated protein kinase activation. J Biol Chem. 1995;270:27995–27998. [PubMed]

4. Bagrodia S, Taylor S J, Creasy C L, Chernoff J, Cerione R A. Identification of a mouse p21 Cdc42/Rac activated kinase. J Biol Chem. 1995;270:22731–22737. [PubMed]

5. Brown J L, Stowers L, Baer M, Trejo J, Coughlin S, Chant J. Human Ste20 homologue hPAK1 links GTPases to the JNK MAP kinase pathway. Curr Biol. 1996;6:598–605. [PubMed]

6. Burbelo P D, Drechsel D, Hall A. A conserved binding motif defines numerous candidate target proteins for both Cdc42 and Rac GTPases. J Biol Chem. 1995;270:29071–29074. [PubMed]

7. Cerione R A, Zheng Y. The Dbl family of oncogenes. Curr Opin Cell Biol. 1996;8:216–222. [PubMed]

8. Coso O A, Chiariello M, Yu J C, Teramoto H, Crespo P, Xu N, Miki T, Gutkind J S. The small GTP-binding proteins Racl and Cdc42 regulate the activity of the JNK/SAPK signaling pathway. Cell. 1995;81:1137–1146. [PubMed]

9. Daniels R H, Bokoch G M. p21-activated protein kinase: a crucial component of morphological signaling? Trends Biochem Sci. 1999;24:350–355. [PubMed]

10. Dharmawardhane S, Sanders L C, Martin S S, Daniels R H, Bokoch G M. Localization of p21-activated kinase 1 (PAK1) to pinocytic vesicles and cortical actin structures in stimulated cells. J Cell Biol. 1997;138:1265–1278. [PMC free article] [PubMed]

11. Dutartre H, Davoust J, Gorvel J P, Chavrier P. Cytokinesis arrest and redistribution of actin-cytoskeleton regulatory components in cells expressing the Rho GTPase CDC42Hs. J Cell Sci. 1996;109:367–377. [PubMed]

12. Gnesutta, N., J. Qu, and A. Minden. The serine/threonine kinase PAK4 prevents caspase activation and protects cells from apoptosis. J. Biol. Chem., in press. [PubMed]

13. Joneson T, McDonough M, Bar-Sagi D, Van Aelst L. RAC regulation of actin polymerization and proliferation by a pathway distinct from Jun kinase. Science. 1996;274:1374–1376. [PubMed]

14. Khosravi-Far R, Solski P A, Clark G J, Kinch M S, Der C J. Activation of Racl, RhoA, and mitogen-activated protein kinases is required for Ras transformation. Mol Cell Biol. 1995;15:6443–6453. [PMC free article] [PubMed]

15. Kimura K, Ito M, Amano M, Chihara K, Fukata Y, Nakafuku M, Yamamori B, Feng J, Nakano T, Okawa K, Iwamatsu A, Kaibuchi K. Regulation of myosin phosphatase by Rho and Rho-associated kinase (Rho-kinase) Science. 1996;273:245–248. [PubMed]

16. Knaus U G, Bokoch G M. The p21Rac/Cdc42-activated kinases (PAKs) Int J Biochem Cell Biol. 1998;30:857–862. [PubMed]

17. Kozma R, Ahmed S, Best A, Lim L. The Ras-related protein Cdc42Hs and bradykinin promote formation of peripheral actin microspikes and filopodia in Swiss 3T3 fibroblasts. Mol Cell Biol. 1995;15:1942–1952. [PMC free article] [PubMed]

18. Lamarche N, Tapon N, Stowers L, Burbelo P D, Aspenstrom P, Bridges T, Chant J, Hall A. Rac and Cdc42 induce actin polymerization and G1 cell cycle progression independently of p65PAK and the JNK/SAPK MAP kinase cascade. Cell. 1996;87:519–529. [PubMed]

19. Lin R, Bagrodia S, Cerione R, Manor D. A novel Cdc42Hs mutant induces cellular transformation. Curr Biol. 1997;7:794–797. [PubMed]

20. Lin R, Cerione R A, Manor D. Specific contributions of the small GTPases Rho, Rac, and Cdc42 to Dbl transformation. J Biol Chem. 1999;274:23633–23641. [PubMed]

21. Maekawa M, Ishizaki T, Narumiya S. Signaling from Rho to the actin cytoskeleton through protein kinases ROCK and LIM-kinase. Science. 1999;285:895. [PubMed]

22. Manser E, Huang H Y, Loo T H, Chen X Q, Dong J M, Leung T, Lim L. Expression of constitutively active alpha-PAK reveals effects of the kinase on actin and focal complexes. Mol Cell Biol. 1997;17:1129–1143. [PMC free article] [PubMed]

23. Manser E, Leung T, Salihuddin H, Zhao Z S, Lim L. A brain serine/threonine protein kinase activated by Cdc42 and Racl. Nature. 1994;367:40–46. [PubMed]

24. Martin G A, Bollag G, McCormick F, Abo A. A novel serine kinase activated by racl/CDC42Hs-dependent autophosphorylation is related to PAK65 and STE20. EMBO J. 1995;14:1970–1978. [PubMed]

25. Melzig J, Rein K H, Schafer U, Pfister H, Jackle H, Heisenberg M, Raabe T. A protein related to p21-activated kinase (PAK) that is involved in neurogenesis in the Drosophila adult central nervous system. Curr Biol. 1998;8:1223–1226. [PubMed]

26. Michiels F, Collard J G. Rho-like GTPases: their role in cell adhesion and invasion. Biochem Soc Symp. 1999;65:125–146. [PubMed]

27. Michiels F, Habets G G, Stam J C, van der Kammen R A, Collard J G. A role for Rac in Tiam 1-induced membrane ruffling and invasion. Nature. 1995;375:338–340. [PubMed]

28. Minden A, Lin A, Claret F X, Abo A, Karin M. Selective activation of the JNK signaling cascade and c-Jun transcriptional activity by the small GTPases Rac and Cdc42Hs. Cell. 1995;81:1147–1157. [PubMed]

29. Minden A, Lin A, Smeal T, Derijard B, Cobb M, Davis R, Karin M. c-Jun N-terminal phosphorylation correlates with activation of the JNK subgroup but not the ERK subgroup of mitogen-activated protein kinases. Mol Cell Biol. 1994;14:6683–6688. [PMC free article] [PubMed]

30. Nobes C D, Hall A. Rho, Rac, and Cdc42 GTPases regulate the assembly of multimolecular focal complexes associated with actin stress fibers, lamellipodia, and filopodia. Cell. 1995;81:53–62. [PubMed]

31. Noda M, Selinger Z, Scolnick E M, Bassin R H. Flat revertants isolated from Kirsten sarcoma virus-transformed cells are resistant to the action of specific oncogenes. Proc Natl Acad Sci USA. 1983;80:5602–5606. [PubMed]

32. Ohashi K, Nagata K, Mackawa M, Ishizaki T, Narumiya S, Mizuno K. Rho-associated kinase ROCK activates LIM-kinase 1 by phosphorylation at threonine 508 within the activation loop. J Biol Chem. 2000;275:3577–3582. [PubMed]

33. Olson M F, Ashworth A, Hall A. An essential role for Rho, Rac, and Cdc42 GTPases in cell cycle progression through Gl. Science. 1995;269:1270–1272. [PubMed]

34. Qiu R-G, Abo A, McCormick F, Symons M. Cdc42 regulates anchorage-independent growth and is necessary for Ras transformation. Mol Cell Biol. 1997;17:3449–3458. [PMC free article] [PubMed]

35. Qiu R G, Chen J, Kirn D, McCormick F, Symons M. An essential role for Rac in Ras transformation. Nature. 1995;374:457–459. [PubMed]

36. Qiu R G, Chen J, McCormick F, Symons M. A role for Rho in Ras transformation. Proc Natl Acad Sci USA. 1995;92:11781–11785. [PubMed]

37. Ridley A J, Hall A. The small GTP-binding protein Rho regulates the assembly of focal adhesions and actin stress fibers in response to growth factors. Cell. 1992;70:389–399. [PubMed]

38. Ridley A J, Paterson H F, Johnston C L, Diekmann D, Hall A. The small GTP-binding protein Rac regulates growth factor-induced membrane ruffling. Cell. 1992;70:401–410. [PubMed]

39. Sahai E, Alberts A S, Treisman R. RhoA effector mutants reveal distinct effector pathways for cytoskeletal reorganization, SRF activation and transformation. EMBO J. 1998;17:1350–1361. [PubMed]

40. Sahai E, Ishizaki T, Narumiya S, Treisman R. Transformation mediated by RhoA requires activity of ROCK kinases. Curr Biol. 1999;9:136–145. [PubMed]

41. Sanders L C, Matsumura F, Bokoch G M, de Lanerolle P. Inhibition of myosin light chain kinase by p21-activated kinase. Science. 1999;283:2083–2085. [PubMed]

42. Sells M A, Chernoff J. Emerging from the Pak: the p21-activated protein kinase family. Trends Cell Biol. 1997;7:162–167. [PubMed]

43. Sells M A, Knaus U G, Bagrodia S, Ambrose D M, Bokoch G M, Chernoff J. Human p21-activated kinase (Pak1) regulates actin organization in mammalian cells. Curr Biol. 1997;7:202–210. [PubMed]

44. Tang Y, Chen Z, Ambrose D, Liu J, Gibbs J B, Chernoff J, Field J. Kinase-deficient Pak1 mutants inhibit Ras transformation of Rat-1 fibroblasts. Mol Cell Biol. 1997;17:4454–4464. [PMC free article] [PubMed]

45. Tang Y, Marwaha S, Rutkowski L, Tennekoon G I, Phillips P C, Field J. A role for Pak protein kinases in Schwann cell transformation. Proc Natl Acad Sci USA. 1998;95:5139–5144. [PubMed]

46. Tang Y, Zhou H, Chen A, Pittman R N, Field J. The Akt proto-oncogene links Ras to Pak and cell survival signals. J Biol Chem. 2000;275:9106–9109. [PubMed]

47. Taylor S S, Knighton D R, Zheng J, Ten Eyck L F, Sowadski J M. Structural framework for the protein kinase family. Annu Rev Cell Biol. 1992;8:429–462. [PubMed]

48. Tran Quang C, Gautreau A, Arpin M, Treisman R. Ezrin function is required for ROCK-mediated fibroblast transformation by the Net and Dbl oncogenes. EMBO J. 2000;19:4565–4576. [PubMed]

49. Wu W J, Erickson J W, Lin R, Cerione R A. The gamma-subunit of the coatomer complex binds Cdc42 to mediate transformation. Nature. 2000;405:800–804. [PubMed]

50. Zeng Q, Lagunoff D, Masaracchia R, Goeckeler Z, Cote G, Wysolmerski R. Endothelial cell retraction is induced by PAK2 monophosphorylation of myosin II. J Cell Sci. 2000;113:471–482. [PubMed]

51. Zhang S, Han J, Sells M A, Chernoff J, Knaus U G, Ulevitch R J, Bokoch G M. Rho family GTPases regulate p38 mitogen-activated protein kinase through the downstream mediator Pak1. J Biol Chem. 1995;270:23934–23936. [PubMed]

52. Zohn I M, Campbell S L, Khosravi-Far R, Rossman K L, Der C J. Rho family proteins and Ras transformation: the RHOad less traveled gets congested. Oncogene. 1998;17:1415–1438. [PubMed]

Articles from Molecular and Cellular Biology are provided here courtesy of
American Society for Microbiology (ASM)