Ras Regulates the Association of Serum Response Factor and CCAAT/Enhancer-binding Protein β*

The serum response element (SRE) is a promoter element essential for transcriptional activation of immediate early genes, such as c-fos and early growth response-1, by mitogenic signals. Several transcription factors bind the SRE, including the serum response factor (SRF), the ternary complex factor, and the CCAAT/enhancer-binding protein β (C/EBPβ). The C/EBPβ mRNA encodes three translation products of 38, 35, and 20 kDa. p35-C/EBPβ activates transcription of the SRE in an SRF-dependent fashion, whereas p20-C/EBPβ, which initiates at an internal in-frame methionine, lacks a transactivation domain and inhibits transcription. We show that SRF and C/EBPβ interact in vivo through the DNA binding domain of SRF and the C terminus of C/EBPβ common to p35/38 and p20. Therefore, like the ternary complex factor, C/EBPβ may be recruited to the SRE not only by binding to the DNA, which is not a high affinity site, but also by protein-protein interactions with SRF. Strikingly, in both the mammalian two-hybrid assay and in vivocoimmunoprecipitations, the association of SRF and p35-C/EBPβ but not p20-C/EBPβ is dramatically stimulated by activated Ras. Furthermore, mutation of the threonine within a mitogen-activated protein kinase consensus motif in the C terminus of C/EBPβ eliminates the response to Ras. These results suggest a new mechanism by which mitogenic signals may influence transcription activity of the SRE by selectively promoting protein-protein interactions between SRF and the transactivator p35-C/EBPβ.

The c-fos proto-oncogene is a prototypic member of the set of "immediate early" genes whose transcription is rapidly activated in the absence of protein synthesis by mitogenic signals (1). Within the promoter of the c-fos gene is a 26-bp 1 regulatory sequence, the serum response element (SRE), that mediates the rapid and transient induction of the c-fos gene by serum and growth factors (2). The SRE binds a ubiquitous transcription factor, the serum response factor (SRF), which is essential but not sufficient for serum and growth factor activation of SRE (3)(4)(5). SRF is an obligatory member of a multiprotein complex that occupies the SRE and serves as a convergence point for multiple signal transduction pathways. Thus, growth factors that activate the Ras-Raf-extracellular signal-regulated kinase cascade and phorbol esters target a family of transcription factors containing an N-terminal Ets domain that form a ternary complex with SRF and SRE DNA (6). c-fos promoter mutants that cannot bind the ternary complex factor (TCF) remain responsive to serum induction through a second TCFindependent pathway that still requires SRF and can involve certain Rho family GTPases (5,7,8). The role of accessory factors in the TCF-independent pathway has yet to be completely defined.
We have observed that maximal TCF-independent serum induction of the SRE requires SRF and another transcription factor, C/EBP␤ (9,10). C/EBP␤, which binds to the SRE overlapping and immediately 3Ј to SRF, is a member of the basic leucine zipper family of transcription factors (11,12). The C/EBP␤ gene contains 3 in-frame methionines that give rise to three translation products of 38, 35, and 20 kDa (13). (Values are for the murine or rat genes; the human and avian genes are slightly larger). The N-terminal half of p38 and p35 contains a strong transactivation domain, whereas p20-C/EBP␤ lacks this transactivation domain and acts as an inhibitor of transcription. Many fibroblast cell types, such as NIH 3T3 cells, express both the long (35 kDa) and short (20 kDa) forms of C/EBP␤ at approximately equal levels (14). Both p35 and p20 homodimers as well as p35-p20 heterodimers bind to the SRE with equivalent affinity. Moreover, in transient transfection assays p20-C/ EBP␤ completely blocks serum induction of the c-fos SRE, whereas p35-C/EBP␤ potentiates activity approximately 20 -30-fold (10). Interestingly, transactivation by p35-C/EBP␤ is dependent upon SRF binding to the c-fos SRE. Moreover, we have shown that C/EBP␤ is able to bind SRF through proteinprotein interactions in vitro (10). Taken together, these results suggest the requirement for SRF binding in p35-C/EBP␤ transactivation of the SRE reflects some physical interaction between these two transcription factors that is essential for transactivation.
In this study we confirm that SRF and C/EBP␤ associate in vivo using both a mammalian two-hybrid assay for demonstrating protein-protein interactions and coimmunoprecipitation studies. The domains of each protein that are sufficient for their interaction in vivo are comparable to the domains previously identified in vitro (10) and include the DNA binding domain of SRF (amino acids 133-265) and the C terminus of C/EBP␤ (amino acids 152-297) common to p35 and p20. Strikingly, the association of SRF and p35-C/EBP␤ but not p20 C/EBP␤ is dramatically stimulated by activated Ras. Furthermore, mutation of the threonine within a MAP kinase consensus motif PGTP in the C terminus of C/EBP␤ eliminates the response to Ras. These results indicate that C/EBP␤-SRF protein-protein interactions may be one target for TCF-independent signaling pathways to the SRE.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-HeLa cells (kindly provided by Dr. R. Chalkley, Vanderbilt University) were maintained in Dulbecco's modified Eagle's medium and nutrient mixture F-12 (Life Technologies, Inc.) containing 10% donor calf serum with iron (Life Technologies, Inc.), 0.24% sodium bicarbonate, 25 units of penicillin G sodium/ml, and 25 mg of streptomycin sulfate/ml. NIH 3T3 fibroblasts, from the American Type Culture Collection, were grown in Dulbecco's modified Eagle's medium with 10% calf serum (Colorado Serum Company), 0.22% sodium bicarbonate, and 4 mM L-glutamine, penicillin, and streptomycin. COS-7 cells (kindly provided by Dr. S. Hann, Vanderbilt University) were maintained in the same medium as described for NIH 3T3 fibroblasts except that 10% fetal bovine serum (Hyclone) instead of calf serum was added.
HeLa and COS-7 cell transfections were performed using the calcium phosphate technique (15). Cells were plated at a density of 5 ϫ 10 5 cells/dish 1 day before transfection. The cells were exposed to the CaPO 4 -DNA precipitate for 8 h. The medium was removed and replaced with complete medium for an additional 36 h before harvesting. NIH 3T3 cell transfections were performed using a CalPhos Maximizer (CLONTECH) as described by the manufacturer. Cells at 60 -70% confluence were exposed to the CaPO 4 -DNA precipitate for 8 h before the medium was removed and replaced with complete medium for 24 h. Cells were then serum-deprived for 36 -40 h in Dulbecco's modified Eagle's medium supplemented as above except containing 0.5% calf serum before harvesting. Cell extracts were prepared, and chloramphenicol acetyl transferase (CAT) assays were performed on extracts containing equivalent cell protein as described previously (16).
Two-hybrid Assay-The pGAL4, pVP16, and pG5CAT plasmids used are from the Mammalian MATCHMAKER Two-Hybrid Assay Kit (CLONTECH). pGAL4-SRF and pGAL4-SRF(133-265) were constructed by polymerase chain reaction amplification of a 1563-bp fragment (encoding amino acids 1-508 of SRF) or a 420-bp fragment (encoding amino acids 133-265 of SRF), respectively, from pGEM3.5 (gift of R. Treisman) as described previously (10). The amplified fragments with EcoRI ends were inserted into the EcoRI site of pGAL4. pVP16-LIP was constructed by digestion of pRSETA-LIP (14) with BamHI, treatment with DNA polymerase I at 4°C to generate blunt ends, and subsequent digestion with HindIII to release a 582-bp fragment. This fragment was inserted into a pVP16 plasmid, digested with MluI, treated with DNA polymerase I at 4°C, and digested with HindIII. CMV-LAP was a gift of U. Schibler (University of Geneva, Geneva, Switzerland), and CMV-RasV.12 was a gift of E. Ruley (Vanderbilt University).
Coimmunoprecipitation-COS-7 cells were harvested in phosphatebuffered saline containing 0.1 mM sodium vanadate and collected by low speed centrifugation. Cells were resuspended in antibody lysis buffer (10 mM Tris, pH 7.5, 1 mM EDTA, 50 mM NaCl, 0.25% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 1 g of aprotinin/ml, 0.1 mM sodium vanadate, 10 mM sodium molybdate, and 10 mM ␤-glycerol phosphate) and lysed by sonication with a microtip on setting 2 and 20% duty cycle for 10 s. After clarification by centrifugation at 12,000 ϫ g for 10 min, the cell extract was incubated with T7 tag antibody-agarose beads (Novagen) for 2 h. The beads were collected by low speed centrifugation and washed 3 times with antibody lysis buffer. All steps were done at 4°C. pcDNA3.1/His-SRF was constructed by cloning the full-length 1563-bp fragment obtained from polymerase chain reaction amplification of pGEM3.5, as described above, into EcoRI digested pcDNA3.1/ HisC (Invitrogen).
Immunoblots-The samples from the coimmunoprecipation described above were analyzed by electrophoresis on an SDS-12% polyacrylamide gel for 3 h at 160 V. The gel was equilibrated in transfer buffer (33 mM Tris base, 192 mM glycine, 20% methanol) for 15 min before the transfer of proteins to Immobilon-P membrane (Millipore Corp.). After transfer, an immunoblot was performed as described previously (10), except a 1:2000 dilution of anti-C/EBP␤ C-terminal peptide antibody (Santa Cruz Biotechnology) and a 1:5000 dilution of goat anti-rabbit antibody (Roche Molecular Biochemicals) were used. The secondary antibody was detected using SuperSignal Chemiluminescent Substrate (Pierce).

C/EBP␤ and SRF Associate in Vivo-
To extend the validity of the C/EBP␤-SRF protein interactions previously demonstrated by GST pulldown assays in vitro (10), we have also developed a mammalian "two-hybrid" system to monitor the possible association of SRF and p35-C/EBP␤ (also called LAP) or p20-C/EBP␤ (also called LIP) in vivo. Initially, we prepared a construct in which the DNA binding domain of Gal4 (amino acids 1-80) is fused to SRF. When pGal4-SRF is transfected into HeLa cells along with pG5CAT (CLONTECH), a CAT reporter gene-driven by five copies of the Gal4 DNA binding site linked to the E1B promoter, at best only a 1.5-fold increase transcription is observed with the average of four experiments being 1.1-fold (Fig. 1). Thus, although SRF contains a C-terminal transactivation domain, it is not particularly potent in this assay. Other investigators have also observed very low activity with Gal4-SRF (17). However, a much larger increase in transactivation is observed when pCMV-LAP (encoding p35-C/ EBP␤) is cotransfected with pGal4-SRF and the pG5CAT reporter. At the highest level of p35-C/EBP␤, an average 7.5-fold increase in CAT activity was observed with up to a 15-fold increase in some experiments. This increase is not observed when p35-C/EBP␤ alone is expressed with pG5CAT as expected, because Gal4 sites, not C/EBP␤ sites, are upstream of the promoter driving the CAT gene. The results presented in Fig. 1 indicate that SRF, when affixed to the DNA through a Gal4 DNA binding domain, can recruit p35-C/EBP␤ to the promoter whereby the transactivation domain of p35-C/EBP␤, either alone or in concert with the transactivation domain of SRF, results in a strong increase in CAT activity. In the strictest sense, this approach does not constitute a two-hybrid assay, because only one hybrid protein, Gal4-SRF, was employed. Because p35-C/EBP␤ intrinsically contains a strong activation FIG. 1. SRF and p35-C/EBP␤ associate in vivo. HeLa cells were transfected with 6 g of pG5CAT either alone or with 2 g of pGal4-SRF and/or the indicated amount of pCMV-LAP (encoding p35-C/EBP␤). Total DNA in each transfection was adjusted to 16 g with pUC19. Cells were harvested 48 h after transfection, and CAT activity was measured as described in Ref. 16. Data are the average of four determinations in duplicate; average standard error was 33%. domain, we did not need to append another. However, because the approach follows the concept of the two-hybrid assay and because we do in one instance (Fig. 4) use two-hybrid proteins Gal4-SRF and VP16-p20-C/EBP␤, we will for simplicity sake refer to this experimental approach as the two-hybrid assay.
Next we extended the two-hybrid system to examine possible mitogenic regulation of SRF-p35-C/EBP␤ interactions in NIH 3T3 cells. In these experiments we used a Ras protein with an activating mutation at codon 12 (Ras.V12) to simulate a variety of growth-promoting signals. As shown in Fig. 2A, dramatic increases (190 -240-fold) in pG5CAT reporter gene expression are observed when pGal4-SRF and pCMV-LAP (encoding p35-C/EBP␤) are cotransfected with an expression construct for activated Ras, compared with a 10 -13-fold increase in the absence of Ras. This result suggests that the interaction of SRF and p35-C/EBP␤ could be regulated in vivo by a Ras-dependent signaling pathway. It is also possible that activated Ras promotes the "unmasking" of a dormant transactivation domain in p35-C/EBP␤ as has been observed in some cell types. Because p35-C/EBP␤, when overexpressed, activates the c-fos SRE nearly equivalently in quiescent as well as stimulated cells (10), it does not appear that the transactivation domain of p35-C/EBP␤ necessarily requires unmasking by a signal-dependent pathway in NIH 3T3 cells. However, to address this possibility directly we repeated the experiment in Fig. 2A with a Gal4-LAP construct. pGal4-LAP activated pG5CAT but no stimulation by activated Ras was observed (data not shown). In Fig. 2B we demonstrate that the DNA binding domain of SRF (amino acids 133-265) is sufficient for both the interaction with p35-C/EBP␤ in vivo and stimulation of this interaction by activated Ras, although the magnitude of the effect is somewhat smaller, particularly in the case of activation by Ras (note the difference in y axis scales in Fig. 2, A and B). Nonetheless, given that the DNA binding domain of SRF is instrumental in TCF-independent serum activation of the SRE (5, 7), this result is highly significant.
The Association of SRF and C/EBP␤ Is Stimulated by Activated Ras in Vivo-Although the two-hybrid system is a standard and well accepted approach for analyzing protein-protein interactions in vivo, we cannot exclude the possibility that the stimulation by activated Ras observed in Fig. 2 is because of other DNA binding factors and/or general transcription factors that have been modified to make them more responsive to the effects of SRF/p35-C/EBP␤ on the promoter. We therefore sought direct evidence, via coimmunoprecipitation, that SRF and p35-C/EBP␤ associate in vivo and that their interaction is stimulated by activated Ras. COS cells were transfected with an expression vector for a 6ϫ histidine-tagged version of SRF carrying the 10 (T7 tag) epitope sequence in the presence or absence of pCMV-LAP (encoding p35-C/EBP␤). Because endogenous COS cell C/EBP␤ proteins are larger in size (42 and 45 kDa, similar to the human C/EBP␤ gene, NFIL6), the exogenous p35 rat C/EBP␤ is easily distinguished (compare Fig. 3,  lanes 8 and 9). Because p35-C/EBP␤ is overexpressed, to avoid overexposure of the p35-C/EBP␤ signal, the film exposure in lane 8 is shorter than in lane 9 (10 versus 60 s). This difference is reflected in the correspondingly lower signal of endogenous p42/45 and p20-C/EBP␤ in lane 8 compared with lane 9. Cell lysates were incubated with T7 tag antibody-agarose beads, and the immunoprecipitated proteins were subjected to immunoblotting with C/EBP␤ antibody. As shown in Fig. 3, p35-C/ EBP␤ coprecipitated with tagged SRF on the antibody beads but only when activated Ras was included in the cotransfections (lane 5). We are aware of the caveats of using overexpressed proteins in this type of assay, because the high level of expression may promote their fortuitous association. However, in this case nonspecific interactions are rendered less likely by the fact that the SRF-p35-C/EBP␤ interaction is only observed in the presence of activated Ras. Fig.  2 we determined that the DNA binding domain of SRF is sufficient for association with C/EBP␤ in vivo, in agreement with the GST pulldown experiments in vitro (data not shown). We also wanted to determine the domain of C/EBP␤ sufficient for interaction with SRF in vivo. Because p20-C/EBP␤ binds to SRF in GST pulldown assays (10), we proceeded to determine whether it would also interact with SRF in vivo. Because p20-C/EBP␤ lacks a transactivation domain, we used a chimeric VP16-p20-C/EBP␤ protein in the two-hybrid assay shown in Fig. 4. Gal4-SRF was able to recruit VP16-p20-C/EBP␤ to the

FIG. 2. Ras stimulates the interaction of SRF and p35-C/EBP␤ in vivo.
A, NIH 3T3 cells were transfected with 5 g of pG5CAT alone, with 2 g of pGAL4-SRF, or with 2 g of pGAL4-SRF and 1 or 2 g of pCMV-LAP in the absence or presence of 2 g of an activated Ras expression vector. Total DNA in each transfection was adjusted to 11 g with pUC19. Cells were serum-deprived for 40 h prior to harvesting. CAT activity was measured as described in Ref. 16. B, same as A except that 2 g of pGAL4-SRF(133-265) were used. Data are the average of four-six determinations in duplicate; average standard error in A was 33% and in B was 30%.  1, 2, 4, and 5) and 10 g of pCMV-LAP (lanes 2, 3, 5, 6, and 8) in the presence or absence of 2 g of an activated Ras expression construct as indicated. Cells were harvested 40 h posttransfection, and whole cell lysates were analyzed by immunoblotting with C/EBP␤ antibody (lanes 8 and 9) or incubated with T7 tag antibody (T7tag Ab)-agarose beads followed by immunoblotting of precipitated proteins (lanes 1-7).
pG5CAT promoter resulting in a 9-fold activation in transcription (Fig. 4A). We consistently required more VP16-p20-C/ EBP␤ expression construct to obtain the same degree of activation as observed with p35-C/EBP␤. However, it is difficult to determine if this is reflective of a weaker interaction between Gal4-SRF and p20-C/EBP␤ because we do not know the relative potencies of transactivation domains, although VP16 is generally considered to be quite strong. VP16-p20-C/EBP␤ consistently gave a small (3-fold) increase in transcription in the absence of Gal4-SRF. Because there are no DNA binding sites for p20-C/EBP␤ in the Gal4-dependent promoter, we are unsure of the mechanism for this increase. Because no increase is observed with VP16 alone, it may involve some protein-protein interaction between p20-C/EBP␤ and a component of the initiation machinery, which although weak, leads to a detectable increase in transcription because of the extremely strong nature of the VP16 transactivation domain. However, the key finding is that activated Ras completely failed to stimulate the association of Gal4-SRF and VP16-LIP in the two-hybrid assay (Fig. 4B). In fact, a 3-4-fold inhibition was observed so that the level of transactivation by VP16-p20-C/EBP␤ was no different than that observed upon expressing a high level of VP16 alone. These data suggest that selectively stimulating the physical association of p35-C/EBP␤, but not p20-C/EBP␤, with SRF may be a mechanism for Ras-dependent signaling pathway(s) to increase transcriptional activity of the c-fos SRE.
Ras-stimulated Association with SRF Requires Threonine 235 in C/EBP␤-Mitogen-activated protein kinases play pivotal roles in mediating signal transduction upon activation of GTPases such as Ras. The mitogen-activated protein kinases are proline-directed kinases, recognizing the consensus motif PX(S/T)P. Such a consensus motif, PGTP, is found in the C terminus of C/EBP␤ (common to LAP and LIP) and is conserved among C/EBP␤ genes from all species. Moreover, Nakajima et al. (23) found that phosphorylation of threonine 235 (numbering for human gene) is induced by cotransfecting Ras.V12 in NIH 3T3 cells. Nakajima et al. (23) kindly provided us with a site-directed mutant of human C/EBP␤ (NFIL6) in which they had substituted alanine for threonine 235. We used an expression construct for this mutant NFIL6 protein to determine if Ras-activated association with SRF in the two-hybrid assay might require phosphorylation of Thr 235 . Substitution of Ala for Thr 235 resulted in only a small (but reproducible) reduction in NFIL6 interaction with Gal4-SRF in the unstimulated state (Fig. 5). However, stimulation by activated Ras was completely abolished with the mutant NFIL6; in fact, activated Ras was slightly inhibitory. These results suggest that Rasactivated association of SRF with C/EBP␤ requires phosphorylation of Thr 235 by a Ras-dependent mitogen-activated protein kinase.

DISCUSSION
In this report we have presented evidence that C/EBP␤ and SRF associate in vivo and this protein-protein interaction is regulated by Ras-dependent signaling pathway(s). These transcription factors interact in vivo through the DNA binding domain of SRF, which is known to be required for TCF-independent signaling to the SRE (5, 7) and the C terminus of C/EBP␤ common to p20 and p35-C/EBP␤. Both the two-hybrid assay in mammalian cells and in vivo coimmunoprecipitations were used to show that activated Ras dramatically stimulates the association of p35-C/EBP␤ and SRF. These data support the involvement of a physical association of C/EBP␤ and SRF in transcriptional activation of the SRE, which is stimulated by Ras-dependent signaling pathways. However, future development of a mutant C/EBP␤ unable to bind SRF will be necessary to confirm the role of C/EBP␤-SRF protein-protein interactions in SRE function.
Interestingly, activated Ras completely failed to stimulate the association of Gal4-SRF and VP16-p20-C/EBP␤ in the two-FIG. 5. Ras-activated association with SRF requires threonine 235 in C/EBP␤. NIH 3T3 cells were transfected with 5 g of pG5CAT alone, 2 g of pGAL4-SRF, or 2 g of pGAL4-SRF and 1 or 2 g of pCMV-NFIL6 or pCMV-NFIL6(235T-A) as indicated in the absence or presence of 2 g of an activated Ras expression vector. Total DNA in each transfection was adjusted to 11 g with pUC19. Cells were serumdeprived for 40 h prior to harvesting. CAT activity was measured as described in Ref. 16. Data are the average of four determinations in duplicate; average standard error was 24%.
FIG. 4. The association of SRF and p35-C/EBP␤, but not SRF and p20-C/ EBP␤, is stimulated by Ras in vivo. A, NIH 3T3 cells were transfected with 5 g of pG5CAT alone or with 2 g of pGAL4-SRF, 1 or 2 g of pCMV-LAP, 5 g of VP16-p20-C/EBP␤, or 10 g of VP-16 as indicated. Total DNA in each transfection was adjusted to 19 g with pUC19. Cells were serum-deprived for 40 h prior to harvesting. CAT activity was measured as described in Ref. 16. B, same as A except that 2 g of an activated Ras expression vector were included. Data are the average of 6 -16 determinations. Average standard error was 52% for A and 40% for B. hybrid assay; in fact, a 3-4-fold inhibition was observed. Unfortunately, we have been unable to confirm the lack of Rasstimulated association of SRF and p20-C/EBP␤ in the coimmunoprecipitation studies because of an unacceptably high level of nonspecific binding of p20-C/EBP␤ (or VP16-p20-C/EBP␤) to the antibody-agarose beads. We presume this is because of the highly charged nature of p20-C/EBP␤ and are currently working on developing satisfactory precautions to prevent this. Nonetheless, our observations with the two-hybrid assay suggest a new model for SRE function in which protein-protein interactions with the DNA binding domain of SRF are a critical target for TCF-independent mitogenic regulation because they are key to determining which isoform of C/EBP␤ stably occupies the SRE. Our data indicate that p35 and p20-C/EBP␤ are recruited to the SRE not only by binding to the DNA, which is not a high affinity site, but also by protein-protein interactions with SRF. If SRF and p20 homodimers occupy the SRE, no transactivation occurs. When SRF and p35 homodimers or a p35/20 heterodimer occupy the SRE, transactivation results. Mitogenic signals that activate Ras dramatically enhance the interaction of SRF selectively with p35-C/EBP␤, ultimately resulting in efficient transactivation.
This mechanism is compatible with the fact that both p35 and p20 are located in the nucleus in similar amounts in serum-deprived or -stimulated NIH 3T3 cells. Moreover, the distribution of homodimers and heterodimers comprising C/EBP␤ does not appear to change in gel shift assays performed with nuclear extracts from serum-deprived or -stimulated cells (data not shown), suggesting that direct regulation of p20-C/EBP␤ or p35-C/EBP␤ DNA binding activity is unlikely. In vivo footprinting experiments have not found detectable changes in protein occupancy of the SRE upon serum stimulation (18). However, the exchange of p35-C/EBP␤ for p20-C/EBP␤ would be compatible with this observation in that both proteins share the same DNA binding domain and thus would be expected to exhibit extremely similar, if not identical, footprints on SRE DNA.
The Ras-dependent signaling pathway(s), which selectively promote the interaction of p35-C/EBP␤ with SRF, have yet to be defined. Although Raf was the first effector protein located downstream of Ras to be identified, it is now clear that there are multiple Ras effector proteins (for review, see Ref. 19). Moreover, Ras can activate another family of small GTPases, the Rho family GTPases (for review, see Ref. 20). Recent stud-ies on MAP kinase signaling pathways have uncovered the presence of at least three distinct MAP kinase pathways that can be activated by Ras and Rho family GTPases. The terminal kinases in these three pathways are the extracellular signalregulated kinases, Jun N-terminal kinases/stress-activated protein kinases, and p38 kinases (for review, see Refs. 21 and 22). The effect of Ras on p35-C/EBP␤-SRF protein-protein interactions was not observed with the Thr 235 -Ala C/EBP␤ mutant, suggesting that phosphorylation of threonine 235 strongly enhances the interaction of p35-C/EBP␤ with SRF. Threonine 235 is contained in a MAP kinase consensus sequence conserved among C/EBP␤ genes from all species, and this site is known to be an in vitro target for partially purified MAP kinase (23). It will be interesting to investigate which of the terminal MAP kinase(s) may phosphorylate C/EBP␤ at Thr 235 in response to mitogenic signals in vivo and identify the upstream kinases that link this phosphorylation event to Ras activation.