Biochemical characterization of rolledSem, an activated form of Drosophila mitogen-activated protein kinase.

The rolled (rl) gene of Drosophila encodes a homologue of vertebrate mitogen-activated protein kinases. Genetic analyses have shown that the gain-of-function mutation rolledSevenmaker (rlSem) is sufficient to activate developmental pathways controlled by distinct receptor tyrosine kinases, such as Sevenless, Torso, and the Drosophila epidermal growth factor receptor homologue. Here we show that mutant RlSem protein, immunoprecipitated from transiently transfected COS cells, exhibits a moderate increase in kinase activity compared with wild-type Rl protein. Time course studies revealed that RlSem is more active than Rl following short term as well as prolonged treatment with epidermal growth factor. Interestingly, a more pronounced difference in kinase activity is observed when the proteins are immunoprecipitated from extracts of Drosophila rl and rlSem larvae. In fact, the kinase activity of RlSem from larvae extracts is comparable to the kinase activity of larvae expressing either an activated Sevenless receptor or an activated Raf kinase. We also demonstrate that Dsor1, which has been placed upstream of rl genetically, is able to phosphorylate and activate Rl in vitro.

The mitogen-activated protein kinase (MAPK) 1 cascade is a major signaling system by which cells translate extracellular signals into intracellular responses. The mammalian extracellular signal-regulated kinases (ERKs) are the best studied members of the MAP kinase family (for reviews, see Refs. 1 and 2). They are activated by phosphorylation on threonine and tyrosine residues by dual specificity MAPK kinases (MAPKKs). On activation, MAP kinases translocate to the nucleus (3) and activate transcription factors through phosphorylation of serine or threonine residues in the motif P/LXT/SP. One of the best characterized target proteins of MAPK is the ternary complex factor ELK-1, which is phosphorylated by MAP kinases in vitro on sites essential for trans-activation by the serum response element on the c-fos promotor in vivo (4,5).
Many steps of the MAP kinase cascade are conserved in different species, and homologous components have been identified in mammals, yeast, Drosophila melanogaster, and Caenorhabditis elegans. In Drosophila, for example, the specifica-tion of the R7 photoreceptor cell fate in each ommatidium of the developing eye is dependent on the components of the MAP kinase cascade. In the R7 precursor cells the Sevenless (Sev) receptor tyrosine kinase is activated by its ligand, the Brideof-sevenless (Boss) protein. The signal is then transduced via Drk (an SH3-SH2-SH3 adaptor protein, homologous to Grb2), Sos, Ras1, and the protein kinases Raf, Dsor1 (MAPKK), and Rl (MAPK) to the nucleus (reviewed in Refs. 6 and 7). A gainof-function mutation in the rl gene, rl Sem , was identified in a genetic screen for dominant mutations that result in the specification of R7 cells in the absence of the inducing signal, the Boss ligand. This dominant mutation is caused by a single amino acid substitution (D334N) in the catalytic domain of the kinase (8). Interestingly, this mutation not only activates the Sev pathway in the developing eye but also induces the formation of additional wing veins and causes female sterility mimicking the activation of the Drosophila EGF receptor homologue (DER) and Torso receptor tyrosine kinases, respectively. Therefore, Rl function is not only necessary but also sufficient to activate multiple receptor tyrosine kinase pathways in Drosophila. The homologue of the MAP kinase activating kinase (MAPKK) encoded by the Dsor1 gene is also required for multiple developmental pathways (9). Using a temperature-sensitive mutation in Dsor1, Hsu and Perrimon (10) showed that Dsor1 operates like rl in the pathways controlled by Torso, DER, and Sev.
Two nuclear target genes have been shown genetically to act downstream of rl in the Sev pathway, pointed P2 (pntP2) and yan, encoding two members of the ETS family of transcription factors (11,12). Furthermore, it has been shown that wild-type Rl directly phosphorylates Yan and PntP2 in vitro (11).
The ordering of the individual components in the Sev pathway has been determined largely by constructing double mutants of different members of the cascade to examine epistatic relationships between the different gene products. Although this type of genetic analysis allows ordering of individual components by functional criteria, it does not imply direct physical interactions, nor does it reveal the biochemical mechanisms by which the signals are transduced. To investigate the biochemical properties of the first identified gain-of-function mutation in a MAP kinase gene, rl Sem , we examined the kinase activity of the corresponding mutant protein, immunoprecipitated from transfected COS cell lysates and Drosophila larvae extracts. We demonstrate that Rl Sem , expressed in mammalian COS cells or purified from bacteria, is 2-3-fold more active than wild-type Rl. However, when isolated from Drosophila larvae, Rl Sem exhibited a 5-9-fold higher kinase activity than Rl. In addition we present evidence that Dsor1 is a direct activator of Rl in vitro.  1 The abbreviations used are: MAPK, mitogen-activated protein kinase; ERK, extracellular signal-regulated kinase; DmERKA, Drosophila melanogaster ERKA; MAPKK, mitogen-activated protein kinase kinase; GST, glutathione S-transferase; MBP, myelin basic protein; EGF, epidermal growth factor. subcloned as BamHI-EcoRI fragments into the pGEX2T vector and expressed exactly as described for the GST fusion proteins Rl, PntP2, PntP2 T151A and Yan-A (8). Mammalian GST-MAPKK fusion protein was provided by Sally Cowley (Chester Beatty Laboratories, London, United Kingdom) and the pGEX1-Yan-A construct was provided by Gerry Rubin (University of California, Berkeley, CA) (13).

Construction of Recombinant Expression Plasmids and
Wild-type and mutant forms of rl were cloned as Asp 718 -EcoRI fragments into the mammalian, SV40-based cell culture vector pMT21 (Genetics Institute). To construct pMT21-rl-myc and its mutant derivatives, a polymerase chain reaction fragment was generated using a reverse primer that deleted the natural termination codon and added an additional sequence encoding the amino acids of the Myc epitope (GGEQKLISEEDL) followed by a termination codon and a XhoI recognition site.
Site-directed mutagenesis was performed to generate the constitutively activated (Dsor1 act ) and kinase-defective (Dsor1 kd ) mutants of Dsor1 (wild-type Dsor1 cDNA was provided by Y. Nishida, Aichi Cancer Center Research Institute, Nagoya, Japan). To generate Dsor1 act , the oligonucleotide 5Ј-GGTGCCCACAAATTCGTTGGCCATCTCGTCGA-TCAGTTG-3Ј (antisense) was used to introduce AG to CT nucleotide changes at positions 1755 and 1756 of the genomic Dsor1 DNA sequence (GenBank sequence D 13782), altering Ser 234 to a Glu codon and to introduce AGG to CTT nucleotide changes at positions 1767-1769, altering Ser 238 to a Glu codon.
The Asp 224 codon was mutated to an Ala codon to generate the Dsor1 kd mutant by introducing a single T to G nucleotide change at position 1726 of the genomic Dsor1 DNA sequence, using the following oligonucleotide: 5Ј-GGAGACGCCGAAAGCACAGATCTTGAT-3Ј (antisense). In each case the presence of the mutated sequence was confirmed by sequencing.
The entire coding regions of wild-type and mutant Dsor1 cDNA fragments were amplified by polymerase chain reaction, using a primer containing a 5Ј-BamHI recognition sequence, and subcloned as BamHI fragments into the T7 promotor-containing pET3a expression vector (14). For protein expression, these plasmids were transformed into the Escherichia coli strain Bl21 (DE3) PLysS, and the recombinant proteins were expressed and purified as described (15,16).
Transient Transfections of COS-1 Cells-COS-1 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum. Transfections were carried out as described previously (17). In brief, transient transfections were performed by the DEAEdextran method, and cells were starved in serum-free DMEM 16 h before harvesting. Where indicated, cells were stimulated with EGF (10 ng/ml) for 10 min immediately prior to cell harvesting. Cells were washed twice in ice-cold Tris-buffered saline, scraped off the plate, and lysed for 10 min on ice in 20 mM Tris (pH 8), 40 mM Na 4 P 2 O 7 , 50 mM NaF, 5 mM MgCl 2 , 100 M Na 3 VO 4 , 10 mM EGTA, 1% Triton X-100, 0.5% sodium deoxycholate, 20 g/ml aprotinin, and 3 mM phenylmethylsulfonyl fluoride (S buffer). Cells were centrifuged for 10 min at 4°C, and supernatants were stored at Ϫ70°C.
Determination of MAPK Activation-40 g of protein from each COS-1 cell lysate was subjected to SDS-polyacrylamide gel electrophoresis and Western blotting to detect the slow-migrating, phosphorylated form of MAPK. In detail, after electrophoresis proteins were transferred to nitrocellulose, blots were blocked in 5% nonfat dried milk in TBST (10 mM Tris (pH 7.5), 150 mM NaCl, and 0.05% Tween 20) for 1 h and then probed with 9E10 monoclonal antibody (BAbCO), diluted 1:3000 in blocking buffer, for 1 h. After washing in TBST and incubating with goat anti-mouse serum coupled with horseradish peroxidase (diluted 1:15000 in TBST), for 1 h, the blots were washed extensively and developed using the ECL detection system (Amersham Corp.).
Determination of MAPK Activity-COS-1 cells were transfected with wild-type and mutant Rl constructs, serum deprived, EGF stimulated, and lysed as described above. The lysates, containing equal amounts of MAPK protein, were incubated for 1.5 h at 4°C with rotation in the presence of monoclonal antibody 9E10 prebound to protein G Sepharose beads (Sigma). The beads were washed two times with S buffer (without phenylmethylsulfonyl fluoride), once with 30 mM Tris (pH 8), and once with kinase buffer (30 mM Tris (pH 8), 20 mM MgCl 2 , and 2 mM MnCl 2 ). Each immunoprecipitate was incubated for 30 min at 30°C in kinase buffer containing 10 M unlabeled ATP, 7.5 g of MBP and 5 Ci of [␥-32 P]ATP, and the reactions were terminated by adding SDS-polyacrylamide gel electrophoresis sample buffer. Incorporation of 32 P into MBP was quantified by running the samples on 15% polyacrylamide gels, transferring to nitrocellulose, autoradiography, scanning into a PhosphorImager (Molecular Dynamics), and analyzing the intensity of each band using ImageQuant Software.
To analyze MAPK activity in Drosophila larvae extracts, 20 third instar larvae for each genotype were homogenized in S buffer and centrifuged for 10 min at 4°C. The extracts, split into two, were incubated for 1.5 h at 4°C with rotation in the presence of Rl-specific polyclonal DmERKA antibody (18), precoated to protein A-Sepharose beads (Sigma). Kinase assays on the immunoprecipitates were performed as described above. Where indicated, heat shocks were applied for 1 h at 37°C, followed by 2 h of recovering at 25°C prior to homogenization.
Kinase assays with bacterially expressed recombinant Rl, Yan-A, and mammalian MAPKK proteins were carried out as described previously (11).
Determination of MAPKK Activity-Different combinations of wildtype, activated, and kinase-inactive versions of recombinant Dsor1 and Rl proteins were subjected to kinase reactions, which were carried out essentially as described previously (11), using MBP as a substrate.

Generation of Activated and Kinase-negative Mutant Versions of Rl and Dsor1
-To study the biochemical properties of Rl and Dsor1, eukaryotic and bacterial expression vectors encoding wild-type, activated, and inactive versions of these proteins were generated. A kinase-inactive version of Rl Sem was generated by changing Lys 67 in the putative ATP-binding site to Met (Rl kd (kinase defective)), and a kinase-inactive Dsor1 was generated by changing Asp 224 of the conserved protein kinase motif Asp-Phe-Gly (19) to Ala (Dsor1 kd ).
To generate an active version of Dsor1, Ser 234 and Ser 238 , analogous to those sites that are phosphorylated to activate mammalian MAPKK, were changed to Glu (Dsor1 act ), thus introducing a negative charge at these sites in an attempt to mimic phosphorylation (20,21). The activated version of Rl, Rl Sem , was identified genetically as a dominant gain-of-function mutation and was shown to result in a single amino acid substitution (Asp 334 3 Asn) in kinase domain XI (8). To easily distinguish and purify recombinant proteins, a sequence encoding an epitope from the c-Myc protein, recognized by the 9E10 monoclonal antibody (22), was fused to the 3Ј-end of wild-type and mutant Rl coding sequences. Fig. 1 shows a schematic diagram of the proteins analyzed in this study.
Phosphorylation and Activation of Rl in Mammalian Cells-To investigate the biochemical properties of Rl and Rl Sem , we took advantage of the well established growth factor induction system of mammalian tissue culture and investigated phosphorylation and activation of these proteins.
We transiently transfected COS-1 cells with expression vectors encoding wild-type and mutant Rl, harvested from either quiescent or EGF-stimulated cells, and subjected the corresponding cell lysates to Western blot analysis using the 9E10 anti-Myc antibody. Wild-type, kinase-inactive (Rl kd ), and activated (Rl Sem ) versions of Rl were phosphorylated in lysates of EGF-stimulated cells, as indicated by a band of reduced electrophoretic mobility ( Fig. 2A). This suggests that Rl is phosphorylated in this heterologous COS cell system in an EGF-de-  To measure the activity of wild-type and mutant Rl kinases, COS-1 cells were transfected with the corresponding constructs and stimulated with EGF. Rl proteins were immunoprecipitated from lysates of these cells, and their ability to phosphorylate myelin basic protein (MBP, a standard MAPK substrate) was determined. Immunoprecipitates from cells transfected with the empty vector or expressing the kinase-inactive mutant (Rl kd ) exhibited no MBP kinase activity. In contrast, EGFstimulated Rl as well as Rl Sem showed kinase activity, measured by incorporation of radioactive phosphate into MBP (Fig.  2B). The activated Rl Sem was 2-3-fold more active than wildtype Rl on EGF stimulation, a difference that was consistently observed in duplicate experiments and was not due to differences in the expression level of the two forms of Rl. In the absence of EGF stimulation, Rl Sem exhibits a slightly higher intrinsic kinase activity than wild-type Rl, visible only after prolonged exposure times (see below).
To elucidate potential differences in the mode of action of Rl and Rl Sem , we analyzed their activity following EGF treatment over a more extensive time course (Fig. 3). The activation of Rl is triphasic, with a first phase peaking at 5 min, a second phase peaking at about 30 min, and a third broad phase with a peak at 2 h followed by a slow decline to basal levels at 9 h after the initial addition of EGF. For mammalian MAPK, a similar activity was reported with a first peak at 5 min and a second peak about 3 h after EGF treatment (23,24). In contrast, the activation of Rl Sem seems to be only monophasic. The activation reaches its peak at about 5 min and is 2-3-fold higher than Rl, as we have shown before. Then the activity continuously declines to near basal levels after 9 h. Notably, throughout this time course, the activity of Rl Sem is consistently higher than that of Rl.
Phosphorylation of Nuclear Target Proteins by Rl and Rl Sem -In the experiments described above, we have examined the kinase activity of Rl toward the artificial substrate MBP. It is possible, however, that the rl Sem mutation affects the substrate specificity of MAP kinase. Therefore, we wanted to test whether Rl and Rl Sem behave differently toward their natural substrates. We have demonstrated previously that pntP2 and yan, two genes encoding nuclear factors that, like ELK-1, belong to the ETS domain family of transcription factors, act downstream of rl in the Sev pathway. We showed that wildtype Rl phosphorylates Yan and PntP2 in vitro (11). To extend these in vitro studies, we generated and expressed recombinant wild-type and mutant Rl proteins fused to GST and performed kinase assays using bacterially expressed PntP2 and Yan-A as substrates.
Rl and Rl Sem were able to phosphorylate Yan (Fig. 4) and PntP2 (data not shown) in the presence of activated mammalian MAPKK (a kind gift from Sally Cowley and Chris Marshall, Chester Beatty Laboratories), whereas kinase-inactive Rl kd showed no activity. Rl Sem exhibited 2-3-fold higher activity toward PntP2 and Yan than Rl. This difference is similar to that observed in assays using MBP as a substrate (Fig. 2B). It is noteworthy here that Rl Sem exhibited weak intrinsic kinase activity toward Yan that was observed in the absence of activated MAPKK (Fig. 4).
Effects of Dsor1 on Rl Activity in Vitro-So far we have shown that mammalian MAPKK is able to phosphorylate Drosophila Rl kinase in COS cells. The Drosophila homologue of MAPKK, Dsor1, has been shown genetically to act in several different receptor tyrosine kinase-mediated pathways in Drosophila. By analogy to the function of MAPKK as a direct activator of MAPK in vertebrates, Dsor1 is believed to act upstream of rl. There is, however, neither genetic nor biochemical evidence demonstrating that Dsor1 directly activates Rl.

FIG. 2. Rl and Rl Sem are phosphorylated and activated in COS-1 cells.
A, Rl, Rl Sem and Rl kd phosphorylation was revealed as a shift in electrophoretic mobility in EGF-induced COS-1 cells, using antibody 9E10 as described under "Materials and Methods." As a control, COS-1 cells were also transfected with the empty vector. The positions of the active (MAPK-P) and inactive MAPK are indicated. B, MAPK was immunoprecipitated from COS cell lysates with the antibody 9E10, and the MBP kinase activity of the immunoprecipitates was determined as described under "Materials and Methods." The results were quantified with a PhosphorImager and expressed relative to the kinase activity of Rl (1). Similar results were obtained in four separate experiments.

FIG. 3. Time course of MBP kinase activation in COS cells in response to
EGF. Cells were transfected with Rl (q) and Rl Sem (E), respectively, serum deprived, and stimulated with EGF for 5, 10, 20, and 30 min and 1, 1.5, 2, 2.5, 3, 3.5, 4, 6, and 9 h. The first (1), second (2) and third (3) phases of Rl activity are indicated. Results were quantified with a PhosphorImager and expressed as -fold stimulation. The data presented are representative of five experiments, which gave similar results. Therefore, we wanted to test whether Dsor1 is able to phosphorylate and activate Rl in vitro. We expressed bacterial fusion proteins of a putatively activated and kinase-defective version of Dsor1 (see above), analogous to the mutations made in mammalian MAPKK (21). We carried out MBP kinase assays with bacterially expressed Dsor1 and Rl mutant versions in different combinations. Only the activated version of Dsor1, (Dsor1 act ) was able to phosphorylate and activate Rl and Rl Sem in vitro (Fig. 5). Wild-type Dsor1 and kinase-inactive Dsor1 (Dsor1 kd ) did not exhibit any kinase activity. Experiments were also performed using PntP2 and Yan as substrates for the kinase assays, and similar results were obtained (data not shown). These results demonstrate that Dsor1 directly activates Rl in vitro.
Activity of Rl-and Rl Sem Kinase in Drosophila Larvae Extracts-Surprised by the marginal differences in kinase activity of Rl and Rl Sem in vitro, considering the relatively strong effects of this mutation in vivo, we decided to investigate biochemically the activity of Rl-and Rl Sem kinase isolated from Drosophila larvae. Protein extracts were prepared from larvae of different genotypes and immunoprecipitated with a polyclonal Rl-specific antibody, and their MBP kinase activity was determined. We used wild-type (ϩ/ϩ) larvae containing two copies of the rl gene, rl Sem /ϩ larvae containing one copy of rl Sem and a wild-type copy of rl and, also rl Sem /rl 10a larvae containing only one copy of rl Sem , since the rl 10a allele represents a deficiency for the entire rl locus. MAP kinase activity was detected in all extracts (Fig. 6). Rl Sem /ϩ extracts showed 5-6-fold higher activity and rl Sem /rl 10a showed 8 -9-fold higher stimulation than wild-type extracts. Wild-type and mutant kinases showed the same kinetics. The kinase assays were linear with respect to incubation time, the amount of immunoprecipitated protein, and the amount of added substrate protein (data not shown).
As a control, we also heat shocked larvae of the genotype rl D334N /ϩ, transformants carrying the Sem mutation (8), immunoprecipitated Rl, and determined MBP kinase activity. These extracts exhibited 4 -5-fold higher kinase activity than wild-type extracts, exactly matching the activity of rl Sem /ϩ extracts.
These experiments demonstrate that the difference between wild-type Rl and Rl Sem kinase activity in larvae is significantly more pronounced than in heterologous systems. The higher activity in rl Sem /rl 10a (only one copy of rl Sem ) than in rl Sem /ϩ (one copy of rl and one copy of rl Sem ) is consistent with our genetic data. We have shown previously that the phenotype of rl Sem is enhanced in the absence of one copy of the wild-type rl gene (8).
To analyze the activity of endogenous Rl in larvae in which the MAP kinase pathway was activated by upstream components, we prepared lysates from larvae expressing activated forms of raf (raf torY9 ; Ref. 25) and Sev (Sev S11 ; 26). The larvae were heat shocked for 1 h at 37°C to induce ubiquitous expression of the corresponding transgenes; then Rl was immunoprecipitated, and MBP kinase activity was determined. On heat shock induction, Rl, immunoprecipitated from raf torY9 and Sev S11 extracts, was 9-and 4-fold more active, respectively, than Rl immunoprecipitated from wild-type extracts (Fig. 6). Consistent with our genetic analysis of rl Sem , it seems that the increase in kinase activity, generated by the Sem point mutation, is comparable with that generated by constitutive activation of upstream signaling molecules. DISCUSSION The genetic characterization of the rl Sem gain-of-function mutation demonstrated that Rl plays an essential role in various developmental pathways in Drosophila. It did not permit insight, however, into the mechanisms by which the rl Sem mutation affects the biochemical properties of MAP kinase in vivo.
Here we have shown that Rl Sem , isolated from COS-1 cells or bacteria, possesses only weakly enhanced intrinsic kinase activity toward the generic substrate MBP and its natural substrates PntP2 and Yan. Both, wild-type and mutant Rl proteins are phosphorylated, and their kinase activity is stimulated by endogenous mammalian or recombinant Drosophila MAPKK, although phosphorylated Rl Sem kinase is 2-3-fold more active

FIG. 4. Phosphorylation of Yan-A by wild-type and mutant Rl proteins in vitro.
Bacterially expressed recombinant proteins were subjected to kinase assays using Yan-A as a target. In the presence of activated mammalian MAPKK (act.MAPKK), Rl Sem exhibited higher kinase activity than wild-type Rl. The kinase-defective version, Rl kd , showed no activity. No phosphorylation of Yan-A was observed in the presence of activated MAPKK, only. GST protein, as a control, also showed no kinase activity. Note that Rl Sem exhibited weak kinase activity without addition of activated MAPKK.

FIG. 5. Activation of Rl by Dsor1.
Bacterially expressed recombinant wildtype and mutant forms of Rl and Dsor1 were mixed in different combinations and subjected to kinase assays using MBP as a substrate. Each protein alone showed no kinase activity (data not shown). 32 P incorporation into MBP was quantified with a PhosphorImager and expressed as -fold activation with respect to the kinase activity of the GST protein. Experiments were performed twice with identical results. WT, wild-type. than the wild-type Rl protein. Similar results have been reported for mammalian ERK2, containing the same mutation in the homologous position. D319N ERK2 exhibits 2-fold higher kinase activity on EGF stimulation in comparison with wildtype ERK2 (27). In Xenopus, injection of Rl Sem mRNA and also of ERK Sem mRNA (p44 MAPK carrying the corresponding D to N exchange) into embryos was sufficient to induce the expression of brachyury (Xbra), an immediate early mesoderm response gene (28,29). Therefore, it appears that this point mutation in the kinase subdomain XI has a conserved effect on different members of the MAP kinase family.
There are a number of possibilities why Rl Sem is more active on stimulation than the wild-type protein. The D334N mutation in Rl Sem could affect the subcellular localization of the protein and thus its accessibility to activating or inactivating signals. However, in immunofluorescence studies we did not observe a difference in the subcellular localization of wild-type and mutant proteins in Drosophila imaginal discs 2 or when expressed in COS-1 cells (data not shown). Alternatively, the enhanced kinase activity of Rl Sem could be due to an increased affinity toward its physiological substrates. However, the difference in activity of Rl Sem kinase compared with wild-type Rl was similar for both its natural substrates, PntP2 and Yan, and for the artificial substrate MBP. This implies that Rl Sem does not possess a noticeably altered substrate specificity. Another explanation for the increased activity is that Rl Sem is inactivated at a reduced rate, resulting in more persistent activity. It has been shown that mammalian ERK2, carrying the Sem mutation (D319N ERK2), is more resistant to the action of the MAP kinase-specific phosphatase CL100 than wild-type MAPK (27). The results of our time course experiments also promote this hypothesis. We showed that the activation of Rl is triphasic in response to EGF stimulation in COS cells. A rapid activation occurring at 5 min, a short decline at 20 min, and a second activation at 30 min with a minimum at 1 h is followed by a broader wave with a peak at about 2 h after stimulation. A similar, but only biphasic, activation was already reported for mammalian MAPK in CCL39 cells, a hamster fibroblast cell line, showing a fast activation at 5 min and a broader phase at about 3 h after EGF stimulation (23,24). In contrast, the activation of Rl Sem seems to be only monophasic. It peaks at about 5 min and then continuously declines to near basal levels. Throughout this time course the activity of Rl Sem is consistently higher than that of Rl.
It is interesting to note that the duration of MAP kinase activation has been postulated as the determinative factor in the decision between differentiation and proliferation in PC12 pheochromocytoma cells. Stimulation of PC12 cells with EGF results in a transient activation of MAP kinase and induces proliferation, whereas NGF stimulation, also mediated by the Ras-Raf pathway, results in a persistent activation of MAP kinase and induces neurite outgrowth (30; for review, see Ref. 31). In mutant rl Sem flies, the phenotypes observed are related to changes in differentiation, such as the excess recruitment of R7 cells, differentiation of additional wing veins, and suppression of the differentiation of the segmented trunk region of the embryo. Rl Sem does not affect the proliferation of cells, although rl is clearly required for cell proliferation (32). The more persistent activity of rl Sem may preferentially trigger differentiation as opposed to proliferation.
In Drosophila larval extracts the kinase activity of Rl Sem in comparison with wild-type Rl is considerably higher than observed in EGF-stimulated COS-1 cells or when isolated from bacteria. In fact, the level of activity was comparable to that observed in protein extracts of larvae expressing an activated Sev or Raf kinase. The activity was highest in rl Sem /rl 10a larval extracts (8 -9-fold), in which the wild-type rl allele is missing, corroborating genetic data showing that the phenotype of rl Sem is enhanced in the absence of the wild-type copy of the rl gene (8). The question arises why in Drosophila extracts the difference in kinase activity of Rl and Rl Sem is higher than in heterologous systems. It is possible that the resistance of Rl Sem to inactivation by phosphatases is more pronounced in larval extracts than in COS cells. In addition, genetic experiments suggest that the activity of Rl Sem is dependent on components acting upstream in the pathway. Sos and Ras1 act as dominant 2 D. Brunner, personal communication.
FIG. 6. Kinase activity of Drosophila larvae extracts. Larvae were heat shocked (hs) where indicated, and extracts were immunoprecipitated with DmERKA antibody. A, MAPK activity of the immunoprecipitates was determined and measured as the amount of 32 P incorporated into MBP. Activity was quantified with a PhosphorImager and expressed relative to the kinase activity of wild-type (wt) larvae (1). ϩ, parental wild-type chromosomes. These data are representative of three separate experiments. B, to confirm that equal amounts of Rl had been immunoprecipitated from each lysate, Western blot analyses were performed with DmERKA polyclonal antibody. Immunoprecipitates of rl Sem /rl 10a extracts showed about one-half of the amount of protein, since rl 10a represents a deficiency. Thick band above MAPK, immunoglobulin heavy chain.
suppressors of the rl Sem phenotype. 3 In this respect the behavior of rl Sem differs from other gain-of-function mutations in the Sev pathway. For instance, the recruitment of additional R7 cells in flies carrying a gain-of-function mutation in sev is not suppressed by mutations in boss, which codes for the Sev ligand. The dependence of Rl Sem on upstream components suggests that some sort of feedback loop is important in maintaining Rl Sem activation. Here we have shown biochemically that Rl Sem can respond to activating signals with increased activity, which is consistent with the hypothesis of a positive feedback loop. Furthermore, this feedback loop may explain how, even in the absence of an inducing signal, the basal activity of the signaling cascade is sufficient to activate Rl Sem above a threshold required for eliciting certain cellular responses, such as R7 differentiation in the eye, vein formation in the wing, and suppression of segmentation in the embryo. Therefore, the slight increase in kinase activity may be sufficient to trigger this feedback loop and the activation of various differentiation pathways.
A task for the future is to prove the existence of such a feedback loop in vivo. The genetic characterization of second site mutations that suppress or enhance the dominant phenotype of the rl Sem mutation should lead to the identification of further components acting in concert with rl Sem in the signal transduction pathway.