Activation of Rac-1, Rac-2, and Cdc42 by hemopoietic growth factors or cross-linking of the B-lymphocyte receptor for antigen

Small guanosine triphosphatases (GTPases) of the Ras and Rho families cycle between an inactive, guanosine diphosphate (GDP)–bound form, and a GTP-bound form with affinity for a series of effector proteins that control signal transduction cascades regulating a multiplicity of cellular changes. Rac-1, Rac-2, and Rac-3 form a closely related subfamily of the Rho GTPases. Cdc42 is more distantly related and, like Rac-1, is expressed ubiquitously.1-3 In contrast, the expression of Rac-2 is restricted to cells of the lymphohemopoietic system.1 Most clues to the function of the Rac family have come from experiments involving either overexpression of constitutively active or dominant-negative mutants of Rac-1 or microinjection of constitutively active Rac-1.4-6They have implicated Rac-1 in actin-mediated processes including membrane-ruffling and chemotaxis, and in signaling paths involving the p21 activated kinase (PAK) family of serine/threonine kinases, p38 mitogen–activated protein kinase (p38 MAPK), and the c-jun-terminal kinase (JNK).4-6 Similar approaches have implicated Rac in a variety of functions in the immune system including phagocytosis, chemotaxis, production of oxidants by neutrophils, cell-mediated cytotoxicity, and the differentiation of lymphocytes of the T helper-1 subclass.7-14 

An obvious caveat to experiments that involve supraphysiologic levels of constitutively active or dominant-negative mutant proteins is that they may not accurately identify physiologic functions. The interpretation of experiments using dominant-negative mutants of Rac is particularly problematic because there are exchange factors (eg, mSos-1, Ras GRF1 and 2) with 2 catalytic domains that are able to bind to and activate, respectively, members of the Ras and Rac families.4,15-18 Thus, dominant-negative Rac mutants will bind to and sequestrate activators of Rac that also activate members of the Ras pathway, and vice versa. Analysis of hemopoietic cells from gene-targeted mice lacking Rac-2 has provided clear evidence for specific functions of Rac-2 that are not complemented by other Rac family members. Thus, the absence of Rac-2 resulted in defects in chemotaxis, adhesion, and superoxide production in neutrophils,19 and growth, survival, chemotaxis, adhesion, and degranulation in mast cells.20 There were also surprises not predicted by previous experimentation with the methodologies discussed above, including indications that Rac-2 was upstream of Erk MAPK,19 and Akt.20 There were also observations revealing an unexpected interregulation of Rac-2 and Rac-1 and Cdc42. Mast cells from mice lacking Rac-2 expressed much higher levels of Rac-1 (which were not diminished when Rac-2 was reintroduced),20 whereas hemopoietic stem cells lacking Rac-2 exhibited increased activation of Cdc42.21 These observations emphasize the limitations to our current picture of signaling events involving Rac.

There is only one report of a direct biochemical examination of the ability of hemopoietic growth factors to activate endogenous Rac. Surprisingly, this study concluded that, whereas treatment of neutrophils with fMetLeuPhe (fMLP) induced strong activation of Rac, treatment with granulocyte-macrophage colony-stimulating factor (GM-CSF), or tumor necrosis factor-α (TNF-α) did not.22 

Biochemical evidence for the activation of Rac by platelet-derived growth factor (PDGF), fMLP, leukotrienes, and cross-linking of the T-cell receptor has been previously reported.23-26 The general mechanisms of activation are thought to involve the recruitment to the plasma membrane of guanine-nucleotide exchange factors (GEFs), which trigger the exchange of Rac-bound GDP for GTP. With one notable exception,27,28 the GEFs that are active on the Rho family contain pleckstrin homology (PH) domains that bind to or have the potential to bind to membrane-bound lipid products of the action of phosphatidyl-inositol-3 kinase (PI-3K).29,30 Moreover, studies with pharmacologic inhibitors of PI-3K (Ly294002 and wortmannin) have demonstrated that PI-3K is necessary for activation of Rac by many stimuli.23-25,31 However, activation of Rac-2 by treatment of neutrophils with phorbol myristate acetate (PMA) was not blocked by an inhibitor of PI-3K activity, pointing to the existence of Rac GEFs that do not require PI-3K activity for their action on Rac.25 

Overexpression of dominant-active or dominant-inhibitory mutants of Rac have implicated it in the activation of both families of “stress activated kinases” (p38 MAPK and JNK) in fibroblasts,32-35 and of JNK in hemopoietic cells.36,37 Hemopoietic growth factors, whether acting through receptor tyrosine kinases (eg, Steel locus factor [SLF] or colony-stimulating factor 1 [CSF-1]), or receptors of the cytokine receptor superfamily (eg, interleukin-3 [IL-3] or GM-CSF) activate both p38 MAPK and JNK.36,38-40 Activation of p38 MAPK results from dual phosphorylation on threonine and tyrosine residues of a TGY motif in its activation loop by the upstream MAPK kinases (MKKs), MKK3 and MKK6.41-43 

Here we have directly investigated activation of endogenous Rac-1, Rac-2, and Cdc42 using antibodies we have demonstrated to be operationally specific for these molecules. We have tested as stimuli a variety of hemopoietic growth factors active on 2 classes of receptors and ligation of the B-cell receptor for antigen (BCR). Further, we have examined whether activation of one Rac isoform, Rac-1, correlates with activation of p38 MAPK. We observed that endogenous Rac-1 was activated by 5 hemopoietic growth factors acting through 2 different classes of receptor and by cross-linking of the BCR. However, activation of p38 MAPK was not always correlated with the degree of activation of Rac-1. Thus, whereas activation of Rac-1 via the IL-3 receptor was largely PI-3K dependent (70% inhibition by Ly294002), activation of p38 MAPK remained unaffected by inhibition of PI-3K activity. Moreover, the activation of Rac-1 induced by overexpression of the constitutively active mutants of 3 isoforms of Ras, H-Ras, N-Ras, and M-Ras, was completely insensitive to Ly294002. These results suggest that IL-3 can activate Rac-1 and p38 MAPK via mechanisms that do not depend on PI-3K and may involve Ras.

The vectors encoding G12V H-Ras, Q61K N-Ras, V12L Rac-1, and constitutively active PI-3K (p110*) were gifts from Dr A. Ambrosini (DiBiT, H. San Raffaele, Milan, Italy), Dr Rob Kay (The Terry Fox Laboratories, Vancouver, BC, Canada), Dr Frank McCormick (University of California, San Francisco Cancer Research Institute), and Dr Lewis T. Williams (Chiron, Emeryville, CA), respectively. M-Ras was cloned as described previously.44 The bacterial expression vector, pGEX, encoding a glutathione-S-transferase (GST)–PAK fusion protein (amino acids 59-145 of human PAK-1β; GST-PAK) and the mammalian expression vector, pRK5, encoding myc tagged Rac-1 were kind gifts from Dr Alan Hall (MRC Laboratory for Molecular Cell Biology, London, England). The pEGFP-C1 vector was purchased from Clontech (Palo Alto, CA). The Rac-1 mouse monoclonal antibody was purchased from Upstate Biotechnology (Lake Placid, NY). Rabbit polyclonal antibodies specific for Rac-225 and Cdc42 (Figure1A-B and data not shown) were from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies specific for phosphorylated molecules, including phospho-Erk (mouse monoclonal), phospho-serine 473 of Akt (rabbit polyclonal), and phospho-p38 MAPK (rabbit polyclonal) were from New England Biolabs (Beverly, MA). Polyclonal rabbit antibodies recognizing unphosphorylated Erk and p38 MAPK were from Santa Cruz Biotechnology and Akt was from New England Biolabs. The F(ab′)₂ fragments of goat antimouse IgM were purchased from Jackson Immunoresearch Laboratories (West Grove, PA). Recombinant murine cytokines: CSF-1, SLF, and IL-5 were obtained from R & D Systems (Windsor, ON, Canada). Dr Ian Clark-Lewis (The Biomedical Research Centre, Vancouver, BC, Canada) generously provided synthetic murine IL-3 and GM-CSF. Ly294002 was obtained from Calbiochem-Novabiochem (La Jolla, CA) and stock solutions were made at 50 mM in 100% ethanol unless indicated otherwise.

Cells were grown at 37°C in humidified incubators gassed with 5% CO₂ using RPMI 1640 medium (Stem Cell Technologies, Vancouver, BC, Canada), supplemented with 10% (vol/vol) fetal calf serum (FCS; Cansera, Rexdale, ON, Canada), 50 μM β-mercaptoethanol, 0.2 mM l-glutamine, and 1 mM sodium pyruvate. The IL-3–dependent cell lines, Baf/3, WEHI 274.3, R6/X, and MC/9, were cultured in 4% WEHI 3B-conditioned medium as a source of IL-3. Primary bone marrow–derived mast cells (BMMCs) were generated from 4- to 8-week-old (C57BL/6 x DBA/2) F1 hybrid mice (BDF1) or Balb/c mice by culturing bone marrow cells in 4% WEHI 3B-conditioned medium for 3 to 4 weeks to generate mature BMMCs as described previously.45 Primary B-cell blasts were generated from the spleens of 4- to 8-week-old BDF1 hybrid mice. A single-cell suspension of splenocytes was obtained and red cells lysed using red cell removal buffer (0.017 M NH₄Cl, 0.14 M Tris [tris(hydroxymethyl)aminomethane], pH 7.2). After depletion of adherent cells by incubation for 2 hours on tissue culture–treated plastic dishes in 10% FCS and RPMI, splenocytes were resuspended at 2 × 10⁶ cells/mL with 15 μg/mL lipopolysaccharide (LPS) and cultured for 72 hours to generate B-cell blasts.

Prior to stimulation, factor-dependent cell lines or BMMCs were cultured overnight in one tenth the concentration of growth factor in which they were normally grown. Cells were then washed 2 times in serum-free media (SFM) containing 20 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid; pH 7.2). MC/9 or WEHI 274.3 were resuspended at 10⁷ cells/mL and BMMCs at 1.5 × 10⁷ cells/mL in SFM for 1.5 hours at 37°C. Cells were stimulated in 1 mL SFM by addition of saturating concentrations of recombinant or synthetic cytokines: IL-5 (50 ng/mL), SLF (50 ng/mL), CSF-1 (200 ng/mL), IL-3 (5 μg/mL), and GM-CSF (10 μg/mL). After incubation with LPS for 72 hours, cultures of splenocytes were washed 2 times in SFM, resuspended in SFM at 1.5 × 10⁷ cells/mL, and incubated for 3 hours at 37°C in SFM. B lymphocytes (1.5 × 10⁷ cells in 1 mL) were stimulated by addition of 40 μg F(ab′)₂ fragments of goat antimouse IgM. Ly294002 or solvent was added to the cells during the final 15 minutes of serum starvation (prior to stimulation) where indicated.

Baf/3, MC/9, or R6/X cells (2.0 × 10⁷) were incubated in 500 μL SFM with 10 μg/mL diethylaminoethyl (DEAE) dextran for 20 minutes at 37°C. The suspension was transferred to a 0.4-cm electroporation cuvette and electroporated at 300 V and 975 μF using a Gene Pulser II from Bio-Rad (Hercules, CA) yielding a time constant of 20 to 25 milliseconds. After electroporation, cells were incubated at 37°C in RPMI, 10% FCS, and 4% WEHI 3B-conditioned medium for 10 to 16 hours at 10⁶ cells/mL. Cells were then washed twice in SFM and incubated at 1.0 × 10⁷ cells/mL in SFM for 2 hours. Where indicated, cells were treated with Ly294002 for 30 minutes during the last half-hour of serum starvation.

Recombinant GST-PAK was prepared by growing Escherichia coli (DH5α) to an optical density (600 nm) of 0.6 at 37°C and inducing fusion protein production for 2 hours at 25°C with 0.1 mM isopropyl β-d-thiogalactoside (IPTG). Cells were pelleted and sonicated in phosphate-buffered saline (PBS) containing 1% Triton X-100 and protease inhibitors (protease inhibitor cocktail tablets; Boehringer Mannheim, Mannheim, Germany). The soluble fraction of bacterial sonicate was applied to glutathione (GT) Sepharose 4B beads (Amersham Pharmacia Biotech, Baie d'Urfe, QC, Canada) for 30 minutes at 4°C, which were washed twice in PBS containing 0.1% Tween-20 (PBST) and 1 mM dithiothreitol (DTT) and twice in PBST without DTT. To assess Rac/Cdc42 activation, equal numbers of cells were lysed in a buffer containing 50 mM Tris, pH 7.5, 10% glycerol, 1% Nonidet P-40, 30 mM MgCl₂, 150 mM NaCl, 1 mM sodium molybdate, 200 mM sodium orthovanadate, 1 mM sodium fluoride, 50 mM β-glycerol phosphate, 1 μg/mL microcystin, 10 μg/mL aprotinin, 10 μg/mL soy bean trypsin inhibitor, 2 μg/mL leupeptin, 0.7 μg/mL pepstatin, and 40 μg/mL phenylmethylsulfonyl fluoride (PMSF). Total protein in whole cell lysates (WCLs) was quantified using a BCA protein assay kit (Pierce, Rockford, IL). Aliquots of WCLs were added to 20 μL GT beads with bound GST-PAK and were rotated for 30 minutes at 4°C. The supernatant was removed from the beads and retained for immunoblotting. Beads were then washed 3 times with 1 mL of the same buffer used to lyse the cells. Sodium dodecyl sulfate (SDS) sample buffer was applied to aliquots of whole cell lysates or bead pellets and heated to 80°C for 10 minutes. Known aliquots of SDS eluates of beads or WCLs were subject to SDS-polyacrylamide gel electrophoresis (SDS-PAGE; 12% acrylamide) and immunoblotting. Blots were probed with a Rac-1–specific mouse monoclonal antibody or rabbit polyclonal antibodies specific for Cdc42 or Rac-2 to quantify levels of the respective activated GTPases. The ratio of GTP-bound GTPase in stimulated cells versus unstimulated controls was estimated from measurements of arbitrary units of optical density of bands corresponding to the GTPase in immunoblots of the respective samples. The ratio of the optical density of the band corresponding to the GTPase in immunoblots of eluates from the GST-PAK–coated beads, and in an immunoblot of a known aliquot of the WCLs, provided an estimate of the absolute amount of GTP-bound GTPase as a fraction of the total cellular GTPase. The amount of total Rac/Cdc42 in the sample of WCLs, which was subjected to SDS-PAGE and immunoblotting, corresponded to 1% to 4% of that applied to the GST-PAK–coated beads as indicated. The degree of activation of GTPases was also expressed as the ratio of the optical density of the immunoblotted band of activated GTPase precipitated from stimulated cells to the optical density of the immunoblotted band of activated GTPase precipitated from unstimulated cells. Goat antibodies specific for rabbit or mouse immunoglobulins and conjugated to horseradish peroxidase (Dako, Glostrup, Denmark) were used as secondary antibodies. Immunoblots were developed using enhanced chemiluminescence (Amersham Pharmacia Biotech).

Blots of WCLs (50-100 μg total protein) were probed with phospho-specific antibodies to assess levels of activated p38 MAPK, Akt, or Erk as described above. To confirm equivalency of loading, membranes were stripped of bound antibodies by exposure to a solution containing 62.5 mM Tris, 0.2% SDS, and 100 mM β-mercaptoethanol for 30 minutes at 55°C, followed by washing with PBST and reprobing with antibodies specific for the unphosphorylated forms of p38 MAPK, Erk, or Akt.

The activation of Rac-1, Rac-2, and Cdc42 was measured by specifically coprecipitating GTP-bound forms of these GTPases using a recombinant fusion protein (GST-PAK) of GST and amino acid residues 59-145 of PAK-1β, including the Cdc42/Rac-interactive binding domain.46,47 Because the GST-PAK coprecipitates a variety of activated Rho family GTPases, it was critical to use antibodies specific for the desired GTPase. To assess the specificity of the antibodies we have used to identify Rac-1 and Rac-2, we expressed myc-tagged constitutively active mutants of these proteins in Baf/3 cells. We coprecipitated the exogenous GTPases with GST-PAK and split the precipitates into aliquots that were subjected to SDS-PAGE and immunoblotting with either a monoclonal antibody specific for Rac-1 or polyclonal antibodies specific for Rac-2. As shown in Figure 1A, the Rac-1–specific monoclonal antibody specifically recognized constitutively active Rac-1 while detecting absolutely none of the constitutively active Rac-2 present. The converse was observed with immunoblotting with polyclonal antibodies specific for Rac-2 (Figure1A), in agreement with previous assessments of the specificity of these antibodies.25 Moreover, as shown in Figure 1B, the Rac-1–specific monoclonal antibody was able to specifically detect overexpressed myc-tagged Rac-1, but not myc-tagged Rac-2 while recognizing endogenous Rac-1. The converse was observed when fractions of the same coprecipitation were immunoblotted with the Rac-2–specific polyclonal antibodies. These results indicated that the antibodies used to detect Rac-1 and Rac-2 were indeed operationally specific for their respective targets and had the necessary specificity and sensitivity for use in assessing activation of specific endogenous isoforms that occur naturally as a mixture in murine hemopoietic cells. The polyclonal antibodies specific for Cdc42 have been used previously23,24,25 and we confirmed that they recognized recombinant Cdc42, but did not cross-react with Rac-1 or p21 Ras (data not shown).

The mouse mast cell-like cell line, MC/9, was chosen as a convenient model for study of the activation of endogenous Rac-1 because it responds to multiple hemopoietic growth factors. These include IL-3, GM-CSF, and IL-5, all of which act through receptors of the hemopoietin receptor family, as well as SLF, which acts through a receptor-tyrosine kinase.

In contrast to published data indicating that GM-CSF failed to activate Rac in human neutrophils,22 we observed rapid activation of Rac-1 by GM-CSF (Figure 1C). This activation was maximal at 1 minute and was maintained for at least 10 minutes. Stimulation of MC/9 cells with IL-3 also induced activation of Rac-1, although the kinetics were slower with maximal activation occurring after 5 minutes and being maintained for at least 10 minutes (Figure 1C). Similar results were observed with IL-5 (Figure 1C). The levels of Rac-1 GTP reached after stimulation by these 3 growth factors was estimated to range from 0.8% to 1.4% (0.8% for GM-CSF, 1.2% for IL-3, and 1.4% for IL-5) as a percentage of the total Rac-1 present in WCL.

We also used cultures of BMMCs to confirm that IL-3 activated endogenous Rac-1 in primary cells. IL-3 treatment of BMMCs resulted in greater increases in the levels of activated Rac-1 over those seen in resting cells (4.3-fold), than those seen in MC/9 (2.4-fold; Figure2A compared with Figure 1C). The estimated percentage of total cellular Rac-1 activated in IL-3–treated BMMCs was also greater than that seen in IL-3–treated MC/9 cells (3.6% in BMMCs compared with 1.2% in MC/9).

Because Rac-2 is specifically expressed in hemopoietic cells, it was interesting to examine activation of this Rac isoform.1,48Endogenous Rac-2, as detected by antibodies specific for Rac-2 (Figure1),25 was clearly activated by IL-3 in BMMCs (Figure 2B), although limitations of the Rac-2 antibodies prevented accurate quantitation of the percentage of the total cellular Rac-2 that was activated. IL-3 stimulation of BMMCs also induced increased levels of GTP-bound Cdc42 (activation of approximately 1% of total cellular Cdc42; Figure 2C). We observed that CSF-1 and SLF, 2 other hemopoietic growth factors, acting through a different class of receptor (receptor tyrosine kinases), also activated Rac-1. SLF stimulation of MC/9 cells induced maximal increases in levels of Rac-1 GTP by 1 minute, but levels had dropped considerably by 5 minutes (Figure3A). SLF also induced activation of endogenous Rac-1 in primary BMMCs (Figure 3B). Stimulation of the myelomonocytic cell line, WEHI 274.3, with CSF-1, led to rapid activation of Rac-1, with levels reaching a maximum at 1 minute, and dropping significantly by 5 minutes (Figure 3C). The percentages of Rac-1 GTP (as a fraction of the total Rac-1 pool present) in cells activated by SLF (4% in BMMCs and 2.4% in MC/9) or by CSF-1 (5%) were again relatively small.

To investigate whether Rac-1 was activated by ligation of the endogenous receptor for antigen on B lymphocytes (BCRs), we used primary cultures of B lymphoblasts that had been generated by treatment of resting splenocytes with LPS. We cross-linked the BCR without engaging the Fc receptors by using the F(ab′)₂ fragment of antibodies to IgM. This resulted in rapid activation of Rac-1, with maximal stimulation occurring at 1 minute (5.3-fold over background; Figure 4). This level of activated Rac-1 was maintained for 10 minutes. Once again, the absolute amount of Rac-1 activated by BCR-mediated stimulation represented only a small fraction of the total cellular Rac-1 (about 1%).

All known Rho family GEFs, which contain the Dbl homology (DH) catalytic domain, also contain a PH domain that is likely to bind phosphoinositides generated by PI-3K.29,49 Thus, if activation of Rac-1 by hemopoietic growth factors or ligation of the BCR involves these GEFs, it is likely to depend on PI-3K. To test this hypothesis, we investigated the effect of a pharmacologic inhibitor of PI-3K, Ly294002. The activation of Rac-1 induced by cross-linking the BCR was reduced by approximately 75% with concentrations of Ly294002 (3.1 μM) that reduced Akt phosphorylation to background levels (Figure 5B). This suggested that the remaining 25% of Rac-1 activation did not require increases in levels of PI-3K activity. With higher concentrations of Ly294002, levels of Akt phosphorylation were reduced to well below the background level in unstimulated cells. We observed that when background PI-3K activity was so reduced, Rac-1 activation was also reduced to near background levels (93% inhibition at 25 μM Ly294002; Figure 5B). This suggests that low background levels of overall PI-3K activity may be permissive for Rac-1 activation.

Similar results were obtained with IL-3 stimulation of BMMCs, where activation of Rac-1 was only partially inhibited (by 70%) by Ly294002 even at high concentrations (50 μM; Figure 5A). Immunoblotting of cell lysates with an antibody specific for phosphorylated Akt confirmed that the IL-3–induced increase in PI-3K activity had been completely inhibited. These results suggested the possible existence of a pool of Rac-1 that was activated by IL-3 or ligation of the BCRs through mechanism(s) that did not require stimulus-mediated increases in PI-3K activity. However, because levels of Akt phosphorylation were never reduced below background to undetectable levels, we cannot exclude the possibility that a low level of PI-3K activity was permissive for IL-3–induced Rac-1 activation.

There were also differences in the sensitivity to inhibition by Ly294002 of activation of the p38 MAPK and Erk induced by IL-3 or by cross-linking of the BCR. Thus, Ly294002 had no inhibitory effect on the IL-3–induced activation of p38 MAPK or Erk (Figure 5A). The absence of any effect of Ly294002 on activation of these MAP family kinases was surprising given evidence implicating the Rac pathway in their activation,33-35 and our finding that Ly294002 inhibited IL-3–induced activation of Rac by 70%. These results suggested that, in the context of signaling through endogenous IL-3 receptors, part of the activation of Rac and all of the activation of p38 MAPK and Erk, was independent of IL-3–induced increases in PI-3K activity. However, in the case of ligation of the BCR, the presence of Ly294002 resulted in partial inhibition of the activation of both p38 MAPK and Erk 1/2 (Figure 5B). Thus, activation of Erk and p38 MAPK downstream of the BCR involved both PI-3K–dependent and PI-3K–independent pathways.

Our observation that IL-3–induced activation of Rac-1 was not completely inhibited by concentrations of Ly294002 that completely blocked increases in PI-3K–mediated activation of Akt suggested the existence of alternative mechanisms of activating Rac-1 that did not require increased PI-3K activity. There is evidence that Ras proteins are upstream activators of Rac,4,18 and hemopoietic growth factors are known to activate Ras family members.50,51 To test the hypothesis that members of the Ras family could activate Rac-1 in hemopoietic cells, and to investigate whether this occurred through a mechanism involving increased PI-3K activity, we coexpressed wild-type Rac-1 and a constitutively active mutant of 1 of 2 p21 Ras isoforms, H-Ras and N-Ras, in the IL-3–dependent cell line, Baf/3. A plasmid encoding GFP, pEGFP-C1, was coexpressed in all samples to monitor cell viability and electroporation efficiency. Overexpression of either a constitutively active mutant of p21 H-Ras (G12V H-Ras) or N-Ras (Q61K N-Ras) induced activation of coexpressed Rac-1 (Figure6A). To investigate whether Ras-induced activation of Rac-1 was dependent on Ras-mediated increases in PI-3K activity, we examined the effects of treating cells with Ly294002. To determine the concentration of Ly294002 needed to give maximal inhibition of Akt phosphorylation in Baf/3 cells exhibiting high levels of PI-3K activity, we treated Baf/3 cells transfected with either constitutively active H-Ras (G12 H-Ras) or PI-3K (p110*) with titrated doses of Ly294002 (Figure 6C). We observed that 25 μM Ly294002 resulted in the reduction of Akt phosphorylation to background levels in cells overexpressing either the G12 H-Ras or the constitutively active PI-3K. Moreover, there was no further decrease in Akt phosphorylation in cells treated with 50 μM Ly294002. These experiments showed that, despite the fact that Ly294002 (25 μM) reduced the phosphorylation of Akt induced by expression of G12 H-Ras or Q61 N-Ras to background levels, it had no effect on the activation of exogenous Rac-1 (Figure 6A). Thus, expression of constitutively active p21 H or N-Ras activated coexpressed Rac-1 through a mechanism that did not depend on overall increases in PI-3K activity. Because we have shown that the method used to assay activation of “p21 Ras” by hemopoietic growth factors in published studies does not discriminate between isoforms of p21 Ras or the “nonclassical” member of the Ras family, M-Ras,44 it was important to determine whether M-Ras was upstream of Rac-1 as well. This was particularly relevant, because we have shown (A. Schallhorn and J.W.S., unpublished observations, September 2000) that hemopoietic growth factors such as IL-3 were strong activators of M-Ras but not of p21 H-Ras. Expression of a constitutively active mutant of M-Ras (Q71L M-Ras) in Baf/3 cells resulted in activation of coexpressed Rac-1 (Figure 6B). Moreover, treatment of cells expressing constitutively active M-Ras with Ly294002 (25 μM) had no effect on activation of coexpressed Rac-1, although the phosphorylation of Akt induced by expression of Q71L M-Ras was reduced to background levels (Figure 6B). Therefore, we have concluded that both p21 Ras isoforms (H-Ras and N-Ras) and M-Ras are capable of activating exogenous Rac-1 through mechanisms that do not depend on overall increases in PI-3K activity.

To determine whether increased PI-3K activity alone was sufficient for activation of Rac in cells of hemopoietic origin, we performed parallel experiments in which groups of cells were cotransfected with Rac-1 and a constitutively active mutant of the p110 catalytic subunit of PI-3K. These experiments demonstrated that overexpression of activated PI-3K in hemopoietic cells was sufficient for activation of coexpressed Rac-1 (Figure 6A).

To assess whether activation of Ras, Rac-1, or PI-3K was sufficient for activation of p38 MAPK in hemopoietic cells, constitutively active mutants were transiently overexpressed in 2 IL-3–dependent cell lines, Baf/3 and R6/X. In both cell lines, overexpression of G12V H-Ras, Q61K N-Ras, or V12L Rac-1 was sufficient for the induction of increased p38 MAPK phosphorylation (Figure7A-B and data not shown). However, overexpression of constitutively active PI-3K did not result in increased phosphorylation of p38 MAPK in either cell type (Figure 7A-B). Microscopic observation of levels of GFP expression in transfected cells ruled out the possibility that overexpression of the p110* construct was toxic (data not shown). Moreover, immunoblotting for phospho-Akt demonstrated not only that overexpression of p110* resulted in increased PI-3K activity, but that the levels induced were considerably higher than those observed in cells that were overexpressing G12V H-Ras and in which activation of p38 MAPK was induced. To investigate the role of PI-3K in the induction of p38 MAPK activation downstream of Ras, cells expressing Q61K N-Ras were treated with Ly294002. Treatment with Ly294002 did not inhibit activation of p38 MAPK induced by overexpression of activated Q61K N-Ras, instead inducing a reproducible increase in phosphorylation of p38 MAPK (data not shown).

We show here using both primary cells and cell lines that endogenous Rac-1 is activated by stimulation of receptors of the hemopoietin family (the receptors for IL-3, GM-CSF, or IL-5), of the tyrosine kinase family (receptors for SLF and CSF-1), or of the receptor for antigen on BCRs. IL-3 induced activation of both Rac-2 and the related GTPase, Cdc42 (Figure 2). Although stimulation of endogenous receptors induced significant increases in levels of GTP-bound Rac-1 or Cdc42 relative to those in unstimulated cells (2.5- to 12-fold factors of induction; Figures 1-4), the absolute amount of activated Rac-1 or Cdc42 remained small (ranging from approximately 1% to 5% depending on the stimuli). These results are in broad agreement with reported levels of GTP-bound Rac-2 in neutrophils stimulated with fMLP (2.5% of total cellular Rac-2), or in porcine aortic endothelial cells stimulated with PDGF (4% of total cellular Rac-1).23,24 These results should be taken as minimal estimates of Rac activation, as in vivo levels of Rac-1 or 2 GTP will rapidly decrease due to the high intrinsic GTPase activity of Rac GTPases. Moreover, some activated Rac may be associated with a detergent insoluble fraction, for example, the cytoskeleton, meaning it will be unavailable for detection by this assay.

Our conclusion that IL-3 induced activation of Rac-1 and 2 was based on the use of antibodies, which we demonstrated were operationally specific for each Rac isoform (Figure 1A-B). The notion that IL-3 induces activation of Rac-1 is consistent with observations that bone marrow cells from mice that are homozygous null for Rac-2 as a result of gene targeting still retain the capacity to proliferate, survive, and differentiate to BMMCs in response to IL-3.20 The defects that are observed in these Rac-2–deficient cells reflect the differences in the specific functions of Rac-1 and Rac-2.20 Antibodies specific for Rac-3 were not available, so there is also the caveat that some of the Rac-GTP we detected may be Rac-3. A detailed analysis of this question will require the development of specific antibodies capable of differentiating between Rac-1, Rac-2, and Rac-3.

One likely component of the PI-3K–dependent mechanism mediating the activation of Rac-1 downstream of the BCR is the Rac GEF, Vav. Vav contains a PH domain, has enhanced GEF activity in the presence of lipid products of PI-3K,30 and is critical for activation of B cells via the BCR.52 Moreover, Vav becomes tyrosine phosphorylated (an event associated with enhanced GEF activity) in response to IL-3, SLF, or GM-CSF,53 and thus may be responsible for the PI-3K–dependent activation of Rac-1 induced by IL-3. In that the Tec family kinase, Btk, is required for a portion of the activation of Erk induced by ligation of the BCR54 and is activated by products of PI-3K activity,55 it is likely to be involved in the portion of Erk activation that we observed was dependent on PI-3K (Figure 5B). Our observations that overexpression of constitutively active PI-3K resulted in activation of Rac-1, but not of p38 MAPK (Figures 6A and 7), are consistent with the report that overexpression of constitutively active PI-3K was sufficient to induce membrane ruffling, but not activation of transcription factors known to require phosphorylation by p38 MAPK.56 Nevertheless, these observations with overexpressed proteins do not establish that physiologic increases in PI-3K activity are sufficient for activation of Rac-1. Indeed we were unable to observe activation of Rac-1 by IL-4 (B.G. and J.W.S., unpublished observations, April 2001), even though we have previously shown that IL-4 induces increases in PI-3K activity.57 Because IL-4 also fails to activate the Ras/MAPK pathways in lymphohemopoietic cells,38,39,51,58this is compatible with the notion that activation of Rac-1 requires both signals downstream of PI-3K and others downstream of Ras.

It was somewhat unexpected to find that a component of the activation of Rac-1 downstream of the IL-3 receptor and the BCR was not dependent on receptor-induced increases in PI-3K activity (Figure 5A). In this connection, however, it was of interest that expression of constitutively active mutants of H-Ras, N-Ras, or M-Ras resulted in activation of coexpressed Rac-1 through mechanism(s), which did not depend on increases in PI-3K activity (Figure 6). There is indirect evidence consistent with our observation. Thus, dominant active mutants of H-Ras induced membrane ruffling that probably reflected activation of Rac and was not inhibited by Ly294002 or wortmannin.59,60 In Rat-1 fibroblasts, both PI-3K–dependent and –independent pathways led from Ras to the Rac effector PAK.61 Certainly, our results indicate that in hemopoietic cells, activation of Rac-1 downstream of Ras occurs without simultaneous increases in overall PI-3K activity. It should be noted that our results do not exclude that small levels of PI-3K activity corresponding to that in resting, unstimulated cells might still have a permissive role in Rac-1 activation. Indeed this is consistent with our data in B lymphocytes, where high doses of Ly294002 that in these cells (but not mast cells) reduced levels of PI-3K activity to below background levels, blocked Rac-1 activation almost completely. The molecular mechanism for PI-3K–independent activation of Rac-1 is unclear. Only, one candidate, Rac GEF, lacks a PH domain. This protein smgGDS27,28 is structurally unrelated to the other Ras and Rho family GEFs and lacks DH or cdc25 catalytic domains. Nevertheless, smgGDS has GEF activity on a broad range of small GTPases, including Rac, Rap, Ral, and some members of the Ras family such as Ki-Ras 4B,62-64 and M-Ras (C. Korherr, M. Quadroni, and J.W.S., unpublished observations, July 1999). However, there is no evidence that smgGDS is downstream of Ras or is involved in any responses to extracellular signals. Our results raise the possibility that the activation of Rac-1 induced by IL-3, which occurs in the absence of increased PI-3K activity, may be mediated by activation of p21 Ras isoforms, M-Ras, or both. We and others reported that IL-3 and SLF activate Ras isoforms that bound to the monoclonal antibody Y13-259 and were hitherto thought to be p21-Ras alone,50,51 but are now known to include M-Ras.44 However, we have recently shown that IL-3 activates M-Ras and to a lesser extent p21 H-Ras (A. Schallhorn and J.W.S., unpublished observation, September 2001), making both candidate components of IL-3–mediated mechanisms for activation of Rac-1.

The question of whether some or all of the Rac activation induced by hemopoietic growth factors is downstream of Ras is difficult to address experimentally. Transient overexpression of a dominant-negative mutant of p21-Ras, S17N H-Ras, resulted in decreased activation of Rac-1 by IL-3 (B.G., unpublished observations, January 2001). However, dominant-negative Ras mutants such as S17N H-Ras will sequestrate Ras GEF, such as mSos-1 and Ras GRF-1 and -2, that are also direct activators of Rac.15-17 Therefore, we cannot exclude that S17N H-Ras might be exerting its inhibitory effect on activation of Rac-1, not through inhibiting activation of Ras, but by directly sequestrating a Rac GEF or targeting it for destruction.

Our observations on the lack of correlation between activation of Rac-1 and of p38 MAPK in hemopoietic cells contrast with evidence from experiments with other types of cells that suggest that activation of Rac was necessary and sufficient for activation of JNK and p38 MAPK.32-35 Our observations that the activation of Rac-1 resulting from stimulation of the receptors for hemopoietic growth factors or the BCR could be inhibited without corresponding proportional decreases in activation of p38 MAPK (Figure 5) suggested that activation of p38 MAPK was not entirely dependent on activation of Rac-1. Thus, whereas inhibition of PI-3K activity below background levels almost completely abolished the activation of Rac-1 induced by ligation of the BCR, the effect on p38 MAPK activation was much less (Figure 5B). Even greater differences in the effect of Ly294002 on activation of Rac-1 and p38 MAPK occurred in response to IL-3, where Ly294002 reduced activation of Rac-1 by approximately 70%, while having no effect on activation of p38 MAPK (Figure 5A). Similar results were reported in mast cells stimulated with SLF, where PI-3K inhibitors failed to block p38 MAPK activation,65 although not the activation of JNK.65,66 The latter suggests differences in the mechanisms of activation of JNK and p38 MAPK by SLF. This notion is also supported by our observation that overexpression of activated p21 Ras was sufficient to activate p38 MAPK in Baf/3 cells (Figure 7), whereas our published data and that of others show that overexpression of activated Ras was not sufficient to activate JNK.36,37 

Despite the lack of correlation between increases in activity of Rac-1 and p38 MAPK seen when inhibitors of PI-3K were used to reduce levels of activation of endogenous Rac-1, overexpression of constitutively active V12L Rac-1 was sufficient for activation of p38 MAPK (Figure 7). However, in that we and others have shown that the levels of activated Rac-1 induced by physiologic stimuli are relatively low (< 5% of cellular Rac), results of experiments involving overexpression of constitutively active Rac-1 (where cellular levels of activated Rac-1 exceed those observed with physiologic activation of endogenous Rac-1) may not be biologically relevant. Consistent with this notion, overexpression of the Rac GEF, Vav, which probably resulted in levels of activated endogenous Rac-1 that are closer to physiologic levels, failed to induce activation of p38 MAPK in RBL cells.67The notion that physiologically relevant levels of activated Rac-1 are alone, insufficient to activate p38 MAPK, is supported by our observation that overexpression of constitutively active PI-3K in 2 IL-3–dependent cell lines was insufficient for activation of p38 MAPK (Figure 7), although it activated coexpressed, exogenous Rac-1 (Figure6A). These results are consistent with the report that overexpression of constitutively active PI-3K was sufficient to induce membrane ruffling, but not activation of transcription factors known to require phosphorylation by p38 MAPK, including AP-1 and Elk-1.56 

The differential effect of inhibition of PI-3K activity on activation of Rac-1 and p38 MAPK by IL-3 might be explained in several ways. First, p38 MAPK activation might not depend on activation of Rac-1 at all. In that we have shown that IL-3 activates Cdc42 (Figure 2C) and the activation of Cdc42 induced by PDGF occurs independent of PI-3K catalytic activity,68 it is possible that Cdc42 is responsible for the PI-3K–independent activation of p38 MAPK. However, we failed to observe consistent activation of p38 MAPK in cells of hemopoietic origin (Baf/3 or R6X) in which constitutively activated Cdc42 had been overexpressed (data not shown). Alternatively, IL-3 might activate 2 pools of Rac-1 in hemopoietic cells, the minor one of which is activated via a PI-3K–independent mechanism and is responsible for activation of p38 MAPK. Finally, it is possible that the 30% of IL-3–induced Rac-1 activation that was independent of increases in PI-3K activity may synergize with another signal to activate p38 MAPK. Our observations that overexpression of activated Ras (H-Ras, N-Ras, or M-Ras) induced activation of Rac-1 and p38 MAPK (Figures 6 and 7), the latter through mechanisms that could only involve physiologic levels of endogenous Rac-1, raise the possibility that this second pathway may involve Ras. Activation of p38 MAPK induced by granulocyte-colony-stimulating factor (G-CSF) was dependent on activation of Ras.40 Certainly our data show that increased PI-3K activity alone, even at levels greater than those associated with p38 MAPK activation in cells overexpressing constitutively active p21 Ras (Figure 7), was insufficient for activation of p38 MAPK. Finally, activation of p38 MAPK by IL-3 may involve Rac-independent mechanism(s), which as discussed above, appear to be involved in activation of p38 MAPK downstream of cross-linking of the BCR as well.

The activation of Rac by hemopoietic growth factors is likely to be involved in many of their actions in the regulation of growth, survival, differentiation, and function, all of which are influenced by the absence of Rac-2.19,20 One critical function of the Rac pathway in growth may involve stathmin, first recognized as a protein overexpressed in leukemic blast cells and now known to be a key regulator of microtubule assembly.69 The Rac and Cdc42 pathways activate p65 PAK (PAK-1), which phosphorylates stathmin on serine 16, inactivating it so that it releases tubulin.70We identified stathmin as one of the proteins undergoing serine/threonine phosphorylation after stimulation with IL-3 and showed that one of the sites phosphorylated was serine 16 (M. Quadroni and J.W.S., unpublished observation, June 1997). Our demonstration that IL-3 induces activation of Rac-1, Rac-2, and Cdc42 accounts for the IL-3–induced activation of PAK-171 and phosphorylation of stathmin on serine 16, an essential prelude to cell division. Overexpression of activated Rac promotes survival of hemopoietic cells72 and hemopoietic cells from Rac-2–deficient mice exhibited decreased expression of the antiapoptotic protein, Bcl-X_L, and increased expression of the proapoptotic molecule, BAD.20 Hemopoietic growth factors such as GM-CSF, SLF, and CSF-1 are capable of inducing cellular migration and chemotaxis, which in the case of CSF-1 has been shown to be Rac dependent.10,73 Finally, our demonstration that IL-3 and CSF-1 induce activation of Rac provides a molecular basis for observations that these cytokines induce membrane ruffling.73,74 Our findings that hemopoietic growth factors activate Rac-1, Rac-2, and Cdc42 confirms the central role of these GTPases in growth factor–mediated signaling. Further, it raises new questions about the existence of novel mechanisms of activating these GTPases, which do not depend on increases in PI-3K activity and potentially involve novel effectors of the Ras family.

We thank the Canadian Institutes of Health Research (CIHR) for supporting this work.