Mek1 Phosphorylation Site Mutants Activate Raf-1 in NIH 3T3 Cells

MAP (mitogen-activated protein) kinases are acti- vated by a family of dual specificity kinases called Meks (MAP kinase/Erk kinase). Mek1 can be activated by Raf by phosphorylation on serine 218 and serine 222. Muta- tion of these sites to acidic residues leads to constitutively active Mek1 in some cases. When fibroblast lines were infected with high titer retroviral stocks carrying these Mek1 genes, the resultant transformation and morphological changes correlated with the kinase activity of the respective Mek1 enzymes. Although [Asp 218 ]-and [Asp 218 ,Asp 222 ]Mek immunoprecipitated from clonal cell lines could phosphorylate kinase-inactive Erk1 equally well in vitro , the endogenous MAP kinase activity was 5–7-fold greater in [Asp 218 ]Mek1-infected clonal lines, and did not correlate with the degree of transformation. Analysis of the Erk1 pathway revealed Raf-1 activation, which correlated qualitatively with the MAP kinase activity seen in the [Asp 218 ]- and [Asp 218 ,Asp 222 ]Mek1-infected clonal cell lines. Expression of dominant negative Ras did not affect the ele- vated Raf-1 activity observed in these cells, however. These data suggest that Mek1 phosphorylation site mu- tants activate Raf-1 and MAP kinase by a Ras-independ-ent pathway and that the mechanism by which transfor- mation occurs may utilize pathways that are MAP kinase-independent. Cell growth and differentiation are controlled by a variety of extracellular signals. Many of these signals, including insulin, epidermal growth factor, nerve growth factor, and thrombin, activate a

that the activation of the MAP kinase pathway may be responsible for accelerated growth in NIH 3T3 and Rat 1a cells (5)(6)(7).
Other studies indicate that the activation of MAP kinases is concurrent with the differentiation of PC12 cells induced by nerve growth factor (8,9). These correlative observations, however, do not exclude the possibility that the activation of MAP kinases may be merely a secondary effect of the signaling processes.
Activation of MAP kinases requires the phosphorylation of both threonine and tyrosine residues in a conserved "TEY" region of the catalytic domain (10,11). A family of dual specificity kinases called Meks (MAP kinase/Erk kinase) are responsible for the phosphorylation and activation of MAP kinases (10,12). Mek1 is activated by phosphorylation on serine 218 and 222 by Raf (13,14). Mutation of these two serine sites on Mek1 to acidic residues, particularly Asp 218 , Asp 218 /Asp 222 , and Glu 218 /Glu 222 , produces constitutively active forms of Mek1 that can activate MAP kinases both in vitro (13,15) and when they are transiently expressed in COS cells (16,17).
Stable expression of the constitutively active Mek1 mutants causes neuronal differentiation of PC12 cells (16) and oncogenic transformation of fibroblast cell lines (16,18,19). The transformed fibroblast lines exhibited increased AP-1 transcriptional activity and induced rapid tumor formation when they were injected into nude mice.
We had previously generated constitutively active Mek1 mutants that have various specific activities (15,17). To correlate Mek activity with transformation, a retroviral packaging system (20) was used to obtain high titer retroviruses carrying the constitutively active Mek1 mutants. Infection of NIH 3T3 and Swiss 3T3 cells with these retroviral constructs demonstrated that the transformation potential of the Mek1 mutants is closely related to their kinase activity, although growth in soft agar did not correlate with MAP kinase activity. Further analysis of [Asp 218 ]-and [Asp 218 ,Asp 222 ]Mek1-infected clonal lines revealed that Raf1 activity is elevated.

MATERIALS AND METHODS
Production of Retroviruses-The HindIII-XbaI fragments containing Mek1 mutants from pG⅐MEK⅐Cglu (17) were subcloned into the Hin-dIII-AvrII site of LC7⌬SX (a gift from Benjamin Neel, Beth Israel Hospital, Boston, MA). The LC7⌬SX constructs were transfected into BOSC23 cells to produce retroviruses following the method of Pear et al. (20). Briefly, BOSC23 cells were grown to subconfluence in selection medium (Dulbecco's modified Eagle's medium (DMEM), 10% dialyzed fetal bovine serum (FBS) (JRH Biosciences), 250 g/ml xanthine, 2 mM glutamine, 1 ϫ HAT supplement (Sigma), 25 g/ml mycophenolic acid, 2 g/ml aminopterin, 6 g/ml thymidine, 100 units/ml penicillin, 100 g/ml streptomycin) and plated on 6-cm dishes at a density of 2 ϫ 10 6 cells/dish the day prior to transfection. Transfection was carried out by mixing equal volumes of HEPES-buffered saline (pH 7.05) and 250 mM calcium chloride solution containing 10 g of DNA. This mixture was then added to the BOSC23 cells, and they were incubated for 10 h at 37°C in DMEM, 10% FBS, 100 units/ml penicillin, 100 g/ml streptomycin, 25  Supernatants containing retroviruses were collected 48 and 72 h posttransfection in 2.5-ml volumes. Cell lines were then infected as described below.
Infection and Characterization of Cell Lines-NIH 3T3 and mouse embryo fibroblast (MEF) cells were grown in DMEM, 10% calf serum (CS), 100 units/ml penicillin, 100 g/ml streptomycin. Swiss 3T3 cells were grown in DMEM, 10% FBS, 100 units/ml penicillin, 100 g/ml streptomycin. Prior to infection, 1.5 ϫ 10 5 cells were seeded in a 6-cm plate and incubated for 24 h. The cells were incubated at 37°C for 5 h with a mixture of 0.5 ml of retrovirus, 0.5 ml of medium, 8 g/ml hexadimethrine bromide (Sigma) and then grown for 48 h before they were subcultured and characterized. MEF cells were seeded at 5 ϫ 10 3 cells on 3-cm plates 24 h prior to infection. The cells were incubated at 37°C for 5 h with 2 ml of retrovirus in the presence of 4 g/ml hexadimethrine bromide. MEFs were subcultured and characterized in the same manner as NIH 3T3 and Swiss 3T3 cells.
Growth in soft agar was measured following the method of Cowley et al. (16). In brief, 6-cm plates were coated with 5 ml soft agar (20% 2 ϫ DMEM, 10% serum, 50% DMEM, 20% 2.5% agar, 100 units/ml penicillin, 100 g/ml streptomycin, melted and combined at 45°C). Duplicates of 1 ϫ 10 5 cells were suspended in 0.5 ml of medium, mixed with 1 ml of soft agar and added to each plate. Cells were fed weekly with 2 ml of a mixture of 33% medium and 67% soft agar. After 14 days, the number of colonies (with more than 20 cells) from 10 randomly selected areas of each plate was counted to calculate the number of colonies per plate and the efficiency of colony formation.
To measure saturation density, 1 ϫ 10 5 cells were seeded in a 6-cm plate and grown in 4-ml medium for 10 days. The number of cells in triplicate plates was counted on a Coulter Counter ® ZM (Coulter Electronics Ltd.) Fluorescence-aided Cell Sorting (FACS) Analyses-Three days after retroviral infections, 5 ϫ 10 5 cells were trypsinized and pelleted by centrifugation at 300 ϫ g for 2 min. The cells were washed twice with 0.5 ml of ice-cold staining medium (Hank's balanced salt solution, 3% FBS, 0.05% sodium azide) and suspended in 50 l of an anti-CD7 monoclonal antibody conjugated to phycoerythrin (Sera Lab, VTR-144-5) (1:10 diluted in staining medium). The cells were incubated on ice for 30 min, washed three times with 0.5 ml of ice-cold staining medium, and suspended in 100 l of fixing medium (Hank's balanced salt solution, 1% formaldehyde, 2% glucose, 0.05% sodium azide). The percentage of cells stained positive for CD7 was measured on a FACScan (Becton Dickinson).
Isolation and Characterization of Clonal Lines-Clonal lines were established by picking colonies that had grown for 2-4 weeks in soft agar. The colonies were first placed in 96-well plates containing DMEM, 10% CS, 100 units/ml penicillin, 100 g/ml streptomycin and gradually expanded to 10-cm plates. Growth in soft agar was done as described above, except that 5 ϫ 10 4 cells/6-cm dish were plated.
Antibodies and Western Analyses-The anti-EE monoclonal antibody, which reacts with the sequence used as an epitope tag, was a gift of G. Walter (21). The anti-Mek1 monoclonal antibody, 3D9, was a gift of C. M. Crews. The C-16 anti-Erk1 antibody, the C-12, and the C-20 anti-Raf1 antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-pan Ras Ab-3 was from Oncogene Science. Western blots were performed as described previously (17) using the C-16 anti-Erk1 antibody, the C-20 anti-Raf1 antibody, and the anti-pan Ras Ab-3 antibody.
Expression of Dominant-negative Ras-Cells were transfected with either pBabe puro vector alone, or with pBabe puro containing dominant-negative RasN17. Transfectants were selected in 2 g/ml puromycin. Clonal lines were picked, expanded, and further analyzed.

Stable Expression of Mek1 Phosphorylation Site Mutants-
Epitope-tagged Mek1 phosphorylation site mutants (17) were placed under the control of an SV40 early promoter in a retroviral vector that also carries a human CD7 gene under the control of an LTR promoter. The retroviral constructs were transfected into BOSC23 cells (20) to produce high titer viruses. NIH 3T3, Swiss 3T3 and MEF cells were infected with the retroviruses. The percentage of infection was measured by FACS analysis using a monoclonal antibody against human CD7. For all Mek1 constructs, more than 90% of NIH 3T3 cells were infected, whereas Swiss 3T3 and MEF cells were infected 40 -50% and 30 -50%, respectively (Table I).
The expression of Mek1 in NIH 3T3 cells was examined by Western blotting with a monoclonal antibody against Mek1. The level of epitope-tagged Mek1 was 3-5-fold higher than that of endogenous Mek1 (Fig. 1A). To determine the specific activity of the Mek1 mutants, a monoclonal antibody against the epitope tag was used for immunoprecipitation, and the kinase activity in the immunocomplex was assayed using kinase-inactive Erk1 as substrate (Fig. 1B). Consistent with our data from COS-7 cells (15) ]infected cells was more than three times higher than that of control cells; and that of [Glu 218 ,Glu 222 ]-infected cells was 1.8 times higher (Fig. 3).
An important characteristic of transformed cells is anchorage-independent growth. NIH 3T3, Swiss 3T3, and MEF cells expressing the Mek1 mutants were tested for efficiency of colony formation in soft agar (Table I) Fig. 1B).
Isolation and Characterization of Clones Expressing Mek1 Mutants-It is possible that individual infected cells in the general population express Mek1 at significantly different levels, which would result in differential growth in soft agar. To address this concern, we isolated NIH 3T3 colonies that had grown in soft agar and examined their levels of Mek1 expression by Western blotting (Fig. 4). Without exception, the relative levels of epitope-tagged Mek1 to endogenous Mek1 in these clones were, for most parts, similar. Thus the ability of these clones to grow in soft agar was not due to an inordinately high level of Mek1.
The clonal cell lines showed growth rates similar to those of the general populations from which they had been isolated (data not shown). When these clonal lines were tested for efficiency of colony formation in soft agar, however, the results (Table II) differed from those of the general populations (Table  I). There was a significant variation in efficiency of colony    (Table II). To see if the MAP kinase-specific phosphatase (MKP-1) (22) may be involved in this apparent difference, we analyzed the levels of MKP-1 RNA. Our data show that there was no significant increase in message levels of MKP-1 in [Asp 218 ,Asp 222 ]Mek1-infected clonal lines as compared to [Asp 218 ]Mek1-infected clonal lines. Furthermore, MKP-1 levels in all of the Mek-transformed lines were lower than in NIH 3T3 cells (Fig. 6). Therefore, the difference in MAP kinase activity cannot be attributed to an increase in MKP-1 in these clonal lines, though other phosphatases (23-25) and/or inhibitors may be involved.
Raf-1 Kinase Activity-We analyzed signaling molecules upstream of Mek-1 in order to further investigate the activation of Erk-1 in the clonal lines, especially DS2 and DS4. Fig. 5A shows that Raf-1 activity is elevated in the clonal lines, and this activity qualitatively mirrors the Erk-1 kinase activity (Fig. 5C), in that the highest Raf-1 activity was observed in DS2 and DS4. Expression of dominant-negative RasN17 in these lines did not affect Raf-1 activity, suggesting that Raf-1 activation results from a Ras-independent pathway (Fig. 7B). Furthermore, these lines can tolerate high levels of RasN17 (Fig. 7A), suggesting that growth of these Mek-1-transformed cell lines can be compensated by Ras-independent pathways.
Dominant-negative Ras Expression-As mentioned above, RasN17 was stably expressed in the Mek1-transformed lines using a puromycin-resistant vector. Both pooled transfectants and clonal lines were characterized. The pooled puromycinresistant NIH 3T3 cells expressed very low levels of RasN17 (2-fold greater than endogenous Ras), while pooled puromycinresistant v-Src 3T3 did not express detectable levels of RasN17 (Fig. 8A) Mek-transformed 3T3 stably expressed high levels of dominant-negative Ras, more than 20-fold greater levels than endogenous Ras. No reversion of transformation was observed in these pooled Mek1-transformed lines (data not shown). Ras activity has been shown to be required for v-Src transformation (26,27), but repeated attempts to stably express RasN17 in these pooled cells failed. Clonal cell lines were therefore established to determine if a small percentage of the RasN17-transfected v-Src-transformed cells unable to be detected in a pooled    (Fig. 7A). Unlike the pooled cells, these lines did exhibit some slowed growth and a modest decrease in refractility, as well as partial inhibition of colony formation in soft agar (data not shown). However, the cells were still distinctly morphologically transformed, and were much more refractile than native NIH-3T3 (Fig. 9). Significantly, the single v-Src clone identified as expressing detecatable levels of RasN17 was severely morphologically reverted, exhibiting slowed growth and complete flattening of the cells (Fig. 9). DISCUSSION In this report we show that activated Mek1 mutants lead to cellular transformation, but this process does not correlate with Erk activity (Table II, Fig. 5C). We also observe an in-crease in Raf1 activation in the [Asp 218 ]Mek1-infected clonal lines that mirrors Erk activation (Fig. 5, A and C). Expression of dominant-negative Ras does not inhibit Raf1 activation (Fig.  7B), suggesting that this increase in activity occurs through a Ras-independent pathway. Although the [Asp 218 ]Mek1-infected clonal lines expressing RasN17 show a slight reversion morphologically (Fig. 9), this does not correlate with Raf1 activity, which remains elevated. Furthermore, when we expressed RasN17 in v-Src-and [Asp 218 ,Asp 222 ]Mek1-transformed cells, we observed severe reversion of v-Src-expressing cells, and only slight reversion of [Asp 218 ,Asp 222 ]Mek1expressing cells (Fig. 9). These data suggest that maintenance of transformation by Mek1 phosphorylation site mutants occurs through a Ras-independent pathway, and that the degree of transformation is independent of Raf1 and Erk1 activity.
We have expressed Mek1 phosphorylation site mutants in NIH 3T3, Swiss 3T3, and MEF cells using a retroviral vector. Because of the high infection rate, we were able to directly analyze and characterize the general populations of NIH 3T3 cells without selection of transformed foci. The Mek1 phosphorylation site mutants [Glu 218 ,Glu 222 ]Mek1, [Asp 218 ]Mek1, and [Asp 218 ,Asp 222 ]Mek1 had been previously shown to be 7-, 50-, and 80-fold, respectively, more active than wild type Mek1 as measured by their capacity to phosphorylate Erk1 (16). In soft agar growth assays, cells infected with these mutants had colony forming efficiencies of 0.2%, 5%, and 22%, respectively. The [Asp 218 ]-and [Asp 218 ,Asp 222 ]-infected cells were morphologically distinct from control cells and reached a saturation density 3-fold greater than that of control cells. The [Glu 218 ,Glu 222 ]-infected cells were morphologically similar to control cells and reached a saturation density 1.8 fold greater than that of control cells. Overall, the transformation potentials of the Mek1 phosphorylation site mutants appeared to correlate with their kinase activities.
There is, however, evidence that secondary events that promote transformation occur in these cells. We find for example significant variation in soft agar growth between the clones that were analyzed (Table II). There is also a striking increase in colony-forming efficiency of [Glu 218 ,Glu 222 ]Mek1 clones after they are picked from the initial population of infected cells (Table I and Table II). Since the level of mutant Mek1 expres-  (Fig. 4), other factors appear to contribute to the observed cloning efficiency and transformation (vide infra). The data reported here are relevant to earlier studies by others (16,19), in which transformed cells were analyzed after an initial selection by focus formation and an unspecified number of divisions. Our data suggest that the history of the cells analyzed and the contribution of the initial selection should be considered in the interpretation of Mek1 mutant transformation.
Although Swiss 3T3 cells expressing Mek1 mutants showed properties similar to those of NIH 3T3 cells expressing Mek1 mutants, our soft agar growth assay indicates that MEF failed to be transformed by any of the constitutively active Mek1 mutants. This, however, is not surprising since primary embryo cells are known to be resistant to oncogenic transformation (28,29).
Because MAP kinases are direct downstream targets of Mek1, the activation of MAP kinases would be expected in cells expressing constitutively active Mek1 mutants. Among the Mek1 mutants that caused transformation, we found moderate activation of MAP kinases only in [Asp 218 ]Mek1-infected NIH 3T3 cells. In the analysis of the clonal lines, we found that the [Asp 218 ]Mek1 clones exhibited the highest level of MAP kinase activation yet yielded the lowest percentage of soft agar colonies (Table II). It is not clear, however, why MAP kinases were more active in [Asp 218 ]-infected cells than in [Asp 218 ,Asp 222 ]infected cells despite the fact that both Mek1 mutants have similar specific activities when immunoprecipitated and assayed on GST-Erk1(K63M) (Fig. 1B). MKP-1 message levels are not elevated in the [Asp 218 ]-or [Asp 218 ,Asp 222 ]Mek1-infect-ed clonal lines compared to NIH 3T3 cells (Fig. 6). Similar results were recently reported by others (30) using Mek1 N3-S222D. Whether other regulators of MAP kinase activity are involved in these cell lines remains to be determined.
Our data show that Raf-1 kinase activity is elevated in the [Asp 218 ]Mek1-transformed cells. Overexpression of dominantnegative Ras in these cells does not decrease Raf-1 activity (Fig.  7B), and therefore this activation occurs through a Ras-independent pathway. Mek1-activated GST-Erk1 could not stimulate Raf-1 kinase activity in vitro (data not shown). How Raf-1 is in fact activated in [Asp 218 ]Mek1-transformed cells has yet to be determined. One can postulate that since [Asp 218 ]Mek1 still has an unphosphorylated Ser 222 , it remains in a complex with Raf-1 until the mutant Mek1 is fully phosphorylated and activated; [Asp 218 ,Asp 222 ]Mek1 mimics a fully phosphorylated Mek1, and therefore does not remain complexed to Raf-1 for an extended period of time. It is possible that by maintaining a tighter association with Raf-1, [Asp 218 ]Mek1 may place a factor or factors in close proximity to Raf-1, activating the kinase. We are currently trying to elucidate this phenomenon.
It is noteworthy that the Mek-transformed lines are able to stably express high levels of dominant-negative RasN17. Most published work involving RasN17 expression has been performed in transient or inducible systems because it is likely that Ras function is necessary for cell growth. This implies that the cells transformed by activated Mek1 at least partially bypass a requirement of Ras for cell growth. This is not surprising, as Mek1 has been shown to function downstream of Ras in growth factor signaling (31). However, the modest change in phenotype of the clonal RasN17-expressing Mek1-transformed cell lines and the lack of correlation of Erk activity with transformation suggests that the steady-state system for maintaining cell growth and transformation is more complex than simple phosphorylation of Erk by activated Mek1.
Since MAP kinase activity does not correlate with the efficiency of growth in soft agar, component(s) other than MAP kinases may be responsible for transducing the mitogenic signal downstream from Mek1 in these transformed cells. No Mek1 substrate other than MAP kinases has been identified to date. Transformation by oncogenic Ras has been shown to involve Rac, a pathway independent of the MAP kinase cascade (32,33). Furthermore, expression of MEKK1 in mammalian cells led to the constitutive activation of Mek1 and Mek2 but not of Erk2 (34). This suggests that these various Mek1 phosphorylation mutants may form complexes with distinct proteins and/or may localize to different parts of the cell where they would not normally be present. The elucidation of other pathway(s) involved in transformation by Mek1, together with the identification of new Mek1 substrates, will shed further light on our understanding of cell proliferation and differentiation.