Identification of phospholipase C-gamma1 as a mitogen-activated protein kinase substrate.

The discovery of sequence motifs that mediate protein-protein interactions, coupled with the availability of protein amino acid sequence data, allows for the identification of putative protein binding pairs. The present studies were based on our identification of an amino acid sequence in phosphatidylinositol-specific phospholipase C-gamma1 (PLC-gamma1) that fits the consensus sequence for a mitogen-activated protein kinase (MAPK) binding site, termed the D-domain. Extracellular signal-regulated kinase 2 (ERK2), an MAPK, and phospho-ERK2 were bound by an immobilized peptide sequence containing the identified PLC-gamma1 D-domain. Furthermore, a peptide containing the PLC-gamma1 D-domain was able to competitively inhibit the in vitro phosphorylation of recombinant PLC-gamma1 by recombinant phospho-ERK2, whereas a control peptide derived from a distant region of PLC-gamma1 was ineffective. Similarly, the peptide containing the PLC-gamma1 D-domain, but not the control peptide, competitively inhibited the in vitro phosphorylation of Elk-1 and c-Jun catalyzed by recombinant phospho-ERK2 and phospho-c-Jun N-terminal kinase 3 (phospho-JNK3), another type of MAPK, respectively. Incubation of anti-PLC-gamma1 immunocomplexes isolated from rat brain with recombinant phospho-ERK2 opposed the increase in PLC-gamma1-catalyzed hydrolysis of phosphatidylinositol 4,5-P(2) (PtdIns(4,5)P(2)), which was produced by a tyrosine kinase associated with the immunocomplexes, whereas in vitro phosphorylation of recombinant PLC-gamma1 by recombinant phospho-ERK2 did not alter PLC-gamma1-catalyzed PtdIns(4,5)P(2) hydrolysis. These studies have uncovered a previously unidentified mechanism for the integration of PLC-gamma1- and ERK2-dependent signaling.

The discovery of sequence motifs that mediate protein-protein interactions, coupled with the availability of protein amino acid sequence data, allows for the identification of putative protein binding pairs. The present studies were based on our identification of an amino acid sequence in phosphatidylinositol-specific phospholipase C-␥1 (PLC-␥1) that fits the consensus sequence for a mitogen-activated protein kinase (MAPK) binding site, termed the D-domain. Extracellular signal-regulated kinase 2 (ERK2), an MAPK, and phospho-ERK2 were bound by an immobilized peptide sequence containing the identified PLC-␥1 D-domain. Furthermore, a peptide containing the PLC-␥1 D-domain was able to competitively inhibit the in vitro phosphorylation of recombinant PLC-␥1 by recombinant phospho-ERK2, whereas a control peptide derived from a distant region of PLC-␥1 was ineffective. Similarly, the peptide containing the PLC-␥1 D-domain, but not the control peptide, competitively inhibited the in vitro phosphorylation of Elk-1 and c-Jun catalyzed by recombinant phospho-ERK2 and phospho-c-Jun N-terminal kinase 3 (phospho-JNK3), another type of MAPK, respectively. Incubation of anti-PLC-␥1 immunocomplexes isolated from rat brain with recombinant phospho-ERK2 opposed the increase in PLC-␥1-catalyzed hydrolysis of phosphatidylinositol 4,5-P 2 (PtdIns(4,5)P 2 ), which was produced by a tyrosine kinase associated with the immunocomplexes, whereas in vitro phosphorylation of recombinant PLC-␥1 by recombinant phospho-ERK2 did not alter PLC-␥1-catalyzed PtdIns(4,5)P 2 hydrolysis. These studies have uncovered a previously unidentified mechanism for the integration of PLC-␥1-and ERK2-dependent signaling.
Mitogen-activated protein kinases (MAPKs) 1 are proline-directed, serine/threonine kinases having a minimal consensus substrate sequence of (S/T)P, where S, T, and P represent the amino acids serine, threonine, and proline, respectively (1). The presence of a proline residue in the Ϫ2 position is favorable and yields the optimal consensus substrate sequence of PX(S/T)P, where X is any amino acid (1). Three families of MAPK have been identified: extracellular signal-regulated protein kinase (ERK), c-Jun N-terminal kinase (JNK, also called stress-activated protein kinase), and p38 stress-activated protein kinase (p38; also called RK/CSBP). Multiple members of each of these MAPK families have been cloned. MAPKs exist in an inactive, unphosphorylated form and an active, phosphorylated form. MAPKs are converted to the active, phosphorylated form by dual-specificity kinases termed MAPK kinases or MAPK/ERK kinases (MEKs), which phosphorylate threonine and tyrosine (Y) residues in the enzymes (2)(3)(4).
It has been estimated that ϳ90% of all proteins contain an (S/T)P sequence, yet, not all of these proteins are substrates for MAPKs (5). This indicates that MAPK-dependent phosphorylation of a substrate involves the interaction (docking) of the kinase with a site on the substrate that is distinct from the phosphoacceptor site. Kornfeld and colleagues (5-6) have identified two ERK binding motifs: 1) the FXFP motif and 2) the D-domain motif. In addition to conferring specificity, docking domains may increase the efficiency of substrate phosphorylation (7)(8)(9)(10)(11). The location of the MAPK docking site can be either N-or C-terminal to the phosphorylation site (10,12). The sites on ERKs that are involved in substrate binding have also been identified: the common docking site, which binds the D-domain of the substrate (13), and a distinct hydrophobic pocket for FXFP binding formed between the MAP kinase insert, the p ϩ 1 site, and an ␣F helix (14).
Cross-talk between the MAPK signaling cascade and PLC-␥1 has been demonstrated in many studies. For example, Morrison et al. (23) showed that the MEK, Raf, co-immunoprecipitates with, as well as phosphorylates, PLC-␥1. Rong et al. (24) reported that activation of the Raf/MEK/MAPK pathway in PC12 cells by nerve growth factor requires PLC-␥1 enzyme activity. We demonstrate that ERK2 and phospho-ERK2 interact with PLC-␥1 both in vitro and within rat brain and that this interaction has functional significance.

EXPERIMENTAL PROCEDURES
Animals-Female Sprague-Dawley rats (150 -200 days old) were maintained as described in Weeber et al. (25). The rats were given unlimited access to standard rat chow and tap water. All procedures employed for the housing, handling, and sacrificing of rats were approved by the University of New Mexico Health Sciences Center Institutional Animal Care and Use Committee.
Peptide Synthesis-All peptides and peptide columns were commercially synthesized by BIOSOURCE International (Hopkinton, MA). Both of the peptides used in the kinase inhibition assays demonstrated the same solubility in water. Neither peptide altered the pH of the reaction buffers in these assays.
Phosphorylation of c-Jun and Elk-1-MAPK-dependent phosphorylation of c-Jun and Elk-1 was preformed employing a procedure based on a method described by Ho et al. (28). Phosphorylation of Elk-1 was performed as follows. His-tagged phospho-ERK2 (12.5 ng) was incubated (20 min, 30°C, 25-l reaction volume) with 1 g of recombinant fusion protein of GST-tagged Elk-1, residues 301-428, in phospho-ERK2 kinase reaction buffer (see above), 13.5 mM MgCl 2 , and 90 M [␥-32 P]ATP (15 Ci) in the absence or presence of the PLC-␥1 D-domain peptide (in a 1:2 dilution series starting at 100 M and ending at 12.5 M) or 100 M PLC-␥1 control peptide. The phospho-ERK2 was preincubated with the appropriate peptide for 10 min at 30°C prior to adding the substrate to initiate the kinase reaction. The peptide sequences for the PLC-␥1 D-domain and Control peptides were the same as used in the PLC-␥ in vitro phosphorylation studies. Reactions were terminated and processed as described above for in vitro phosphorylation of PLC-␥1. Equal loading of PVDF membranes was assessed using Brilliant Blue R-250 staining as described above.
Kinase reactions (25 l) for phospho-JNK3 (1 ng) phosphorylation of GST-c-Jun (300 ng) were performed in JNK3 kinase reaction buffer (see above), 13.5 mM MgCl 2 , 90 M [␥-32 P]ATP (15 Ci), and the indicated concentration of peptide. Again, phospho-JNK3 kinase was preincubated with the appropriate peptide for 10 min at 30°C prior to adding the substrate. Reactions were incubated for 10 min at 30°C and were analyzed as described above.
Tissue Preparation and Subcellular Fractionation-Preparation of Triton X-100 extracts of rat whole brain postnuclear (S1) fraction and hippocampal formation postnuclear membrane (P2) fraction were performed as described in Buckley and Caldwell (29) except buffers also contained 20 mM ␤-glycerophosphate, 20 mM sodium pyrophosphate, and 10 mM sodium fluoride.
Affinity Capture of Anti-PLC-␥1 Immune Complexes-Anti-PLC-␥1 immunocomplexes were isolated as follows. For determinations of anti-ERK2 immunoreactivities, whole rat brain S1 fraction (100 g of protein) was incubated with 6 g of rabbit anti-PLC-␥1 antibody overnight with mixing at 4°C. The immunocomplexes were recovered with protein A-Sepharose beads and washed twice with 1 ml of extraction buffer supplemented with 20 mM ␤-glycerophosphate, 20 mM sodium pyrophosphate, and 10 mM sodium fluoride, as described in Weeber et al. (25). For GST pull-down assays, whole rat brain S1 fraction (100 g of protein) was incubated with glutathione-agarose precoupled to a GST fusion protein of the PLC-␥1 SH2-SH2-SH3 domains or glutathione-agarose control beads for 4 h at 4°C. The beads were collected by centrifugation, and the supernatant was decanted and incubated with 6 g of rabbit anti-PLC antibodies or non-immune rabbit IgG. The immunocomplexes and controls were processed as described above. For in vitro PLC activity measurements, anti-PLC-␥1 immunocomplexes were immobilized on 96-well plates, as described elsewhere (29).
Immunoblotting-Anti-ERK2 immunoreactivity associated with anti-PLC-␥1 immunocomplexes or bound by immobilized D-domain peptides was measured as follows. Fifty microliters of 2ϫ SDS-PAGE sample buffer was added to the sample, and the mixture was boiled for 5 min. Eluted proteins were separated using 7.5% (w/v) SDS-PAGE gels prior to transfer to PVDF membranes. Membranes were blocked in 5% (w/v) fat-free milk in Tris-buffered saline and blotted with mouse anti-ERK2 antibody (1:1000) followed by anti-mouse horseradish peroxidase-conjugated antibody (1:3000). The immunoreactive proteins were detected using enhanced chemiluminescence.
In Vitro Treatment of Anti-PLC-␥1 Immunocomplexes with Phospho-ERK2-Anti-PLC-␥1 immunocomplexes were captured from 20 g of rat hippocampal P2 preparation, then incubated (20 min, 35°C) in the presence of one of the following buffers: A) Assay Dilution Buffer I plus 20 M protein kinase C (PKC) inhibitor peptide, 2 M protein kinase A (PKA) inhibitor peptide, and 20 M Compound R24571; B) Buffer A containing 0.4 unit of recombinant phospho-ERK2; C) Buffer A supplemented with 13.5 mM MgCl 2 and 90 M ATP; or D) Buffer C with 0.4 unit of recombinant phospho-ERK2. The wells were washed three times (5 min each) with 1.25ϫ PLC assay buffer (43.75 mM sodium phosphate, pH 6.8, 87.5 mM KCl, 1.0 mM EGTA, 1.0 mM CaCl 2 ) prior to quantification of PLC activity (see below). The in vitro treatment of recombinant PLC-␥1 with recombinant phospho-ERK2 was performed as described for the anti-PLC-␥1 immunocomplexes captured from rat hippocampal P2 preparation, except 100 ng of recombinant PLC-␥1 was first captured onto the microtiter plate prior to the in vitro treatment, or not, with recombinant phospho-ERK2. In addition, incubation of recombinant PLC-␥1 with Buffer C was omitted due to the failure of detecting [␥-32 P]phosphate incorporation in the absence of kinase (Fig.  3A). Recombinant PLC-␥1 lipase activity was then quantified as described below.
Immunocomplex PLC Activity Measurement-PLC activity associated with anti-PLC-␥1 immunocomplexes was quantified as described in Buckley and Caldwell (29). Activity was calculated as nanomoles of Ins(1,4,5)P 3 product formed/min/mg of protein present in the sample from which the enzyme was affinity captured.

Identification of a Putative MAPK Docking Motif in PLC-␥1-
We searched the primary structure of PLC-␥1 (Fig. 1A) for one or more amino acid motifs that fit the consensus sequences of ERK docking sites, as identified by Kornfeld and colleagues (5-6): 1) the FXFP motif, having the consensus sequence of FXFP and 2) the D-domain, having two possible consensus sequences: (K/ R)X(X/K/R)(K/R)X (1-4) (L/I)X(L/I) or (K/R)(K/R)(K/R)X (1-5) (L/I)X(L/ I). We did not find an FXFP sequence but did identify overlapping sequences that conform to the two consensus D-domain sequences: 945 RRKKIAL(EL). In addition, we searched for potential ((S/ T)P) and optimal (PX(S/T)P) MAPK phosphorylation sites in the primary sequence of PLC-␥1 (Fig. 1A). Seven (S/T)P sequences were identified, but none fit the optimal sequence for MAPK phosphorylation. We also searched the primary structure of ERK2 (Fig. 1B) for possible PLC-␥1-interacting domains. We identified a sequence ( 224 PIFP) that fits the minimum consensus sequence (PXXP) of an SH3 domain binding motif (30). In addition, we identified a sequence ( 112 KLLKTQHLSNDHI) that demonstrates homology to a PLC-␥1 binding site (KLLMIIH-DRREFA) on the cytoplasmic domain of integrin ␤ 1A (31).
We tested the hypothesis that ERK2 found associated with anti-PLC-␥1 immunoprecipitates was bound, either directly or indirectly (i.e. mediated by a second, intermediary protein), to the SH2 and/or SH3 domains of PLC-␥1. We performed preliminary experiments in which we incubated rat whole brain S1 fraction (100 g) with a fusion protein comprising the N-terminal SH2, C-terminal SH2, and SH3 domains of rat PLC-␥1 pre-coupled to agarose (50 g), or with each of these three domains individually coupled to agarose (50 g). These studies demonstrated specific binding of ERK2 to the combined SH2-SH2-SH3 agarose, as well as the C-terminal SH2 domain, but not with the N-terminal SH2 domain or the SH3 domain (data not shown). ERK2 within rat brain S1 extracts failed to bind to C-terminal SH2 beads following pre-clearing of the extracts with the C-terminal SH2 beads (50 g), even though a significant amount of ERK2 was still present within the sample (data not shown). These studies demonstrated that rat brain ERK2 indirectly binds to the PLC-␥1 C-terminal SH2 domain in vitro. However, pre-clearing of rat whole brain S1 preparations (100 g) with an excess (50 g) of agarose-coupled GST-SH2-SH2-SH3 domains of PLC-␥1 prior to isolation of anti-PLC-␥1 immunocomplexes did not dissociate the ERK2⅐PLC-␥1 complex (Fig. 2B). This approach is analogous to a procedure described by Snyder and colleagues (32), who reported that in the presence of 10 g of GST-PLC-␥1 SH3 domain co-immunoprecipitation of PLC-␥1 and phosphoinositide 3-kinase enhancer is blocked. Thus, these studies demonstrate that, although ERK2 can bind indirectly to the PLC-␥1 C-terminal SH2 domain in vitro, the PLC-␥1 SH2 and SH3 domains do not significantly contribute to the binding of ERK2 within anti-PLC-␥1 immunocomplexes isolated from rat brain.
Phospho-ERK2 Directly Interacts with and Phosphorylates PLC-␥1 in Vitro -The identification of a putative ERK2 docking site in PLC-␥1 and co-immunoprecipitation of PLC-␥1 and ERK2 indicated that the two proteins may directly bind to each other. To test this hypothesis and to determine whether PLC-␥1 is a substrate for phospho-ERK2-catalyzed phosphorylation, we incubated recombinant PLC-␥1 and PLC-␥2 with recombinant phospho-ERK2 in the presence of [␥-32 P]ATP. We also sought to determine whether other MAPKs, specifically p38␥ and JNK3, were able to phosphorylate PLC-␥1. Fig. 3A shows that PLC-␥1 serves as an in vitro substrate for phospho-ERK2, as well as phospho-p38␥ and phospho-JNK3, whereas negligible phosphorylation of PLC-␥2 was detected in the presence of these MAPKs (Fig. 3B). Control reactions incubated in the absence of recombinant MAPK demonstrated that the phosphorylation of PLC-␥1 was catalyzed by the recombinant MAPK.
Recombinant ERK2 and Phospho-ERK2 Bind to Immobilized PLC-␥1 D-domain Peptide in Vitro -To determine whether the identified PLC-␥1 D-domain sequence is capable of binding ERK2, the sequence of the PLC-␥1 D-domain (Fig. 4A) and a control peptide sequence outside of the PLC-␥1 D-domain (Fig.  4A) were coupled to thiol-agarose and incubated with recombinant ERK2. Similarly, we incubated the immobilized peptides with recombinant phospho-ERK2. Both recombinant ERK2 (Fig. 4B) and recombinant phospho-ERK2 (Fig. 4C) were capable of binding to the immobilized D-domain peptide. In contrast, neither ERK2 nor phospho-ERK2 was bound by the PLC-␥1 control peptide, confirming that the binding was specific for the D-domain.
The PLC-␥1 D-domain Inhibits the MAPK-catalyzed Phosphorylation of MAPK Substrates-To substantiate the claim that the peptide sequence that we identified in PLC-␥1 functions as a putative D-domain, we assessed the ability of a peptide containing the PLC-␥1 D-domain (Fig. 5A) to competitively inhibit the phosphorylation of a known phospho-ERK2 substrate, Elk-1, and a known JNK substrate, c-Jun. Several investigators (e.g. Refs. 8 -10) have demonstrated that D-domain peptides competitively inhibit the binding of MAPKs to their substrates in vitro. The PLC-␥1 D-domain peptide inhibited both c-Jun phosphorylation by active JNK-3 (Fig. 5B) and Elk-1 phosphorylation by active ERK2 (Fig. 5C) in a concentration-dependent manner, whereas the PLC-␥1 control peptide did not inhibit either Elk-1 or c-Jun phosphorylation by the appropriate kinase. The Elk-1 substrate contains multiple MAPK phosphorylation sites, accounting for the appearance of three phosphorylated bands in the autoradiogram. The PLC-␥1 D-domain peptide inhibited the phosphorylation of the upper two bands, whereas it did not alter labeling of the lower, least phosphorylated band, suggesting that it may result from phos-

FIG. 2. Co-immunoprecipitation of PLC-␥1 and ERK2
is not affected by preclearing rat brain S1 preparations of proteins that bind to PLC-␥1 SH2 and SH3 domains. A, rat brain S1 fraction was incubated overnight with PLC-␥1 rabbit polyclonal antibodies, or normal rabbit IgG. Immunocomplexes were recovered using Protein A-coated Sepharose beads and washed three times with buffer. Bound proteins were eluted with SDS-PAGE loading buffer, fractionated on SDS-polyacrylamide gels, and transferred to PVDF membrane. Anti-ERK-2 immunoreactivity was determined as described under "Experimental Procedures." B, rat brain S1 fraction was mixed with either glutathione-agarose or agarose-bound fusion protein consisting of the PLC-␥1 N-SH2, C-SH2, and SH3 domains fused with glutathione S-transferase. The beads were collected by centrifugation, the supernatant was removed, and PLC-␥1 was immunoprecipitated as in A. For both A and B the results are representative of three independent experiments. phorylation of phosphoacceptor sites independent of interaction of phospho-ERK2 with the D-domain of Elk-1. Similar results have been reported by Bardwell et al. (10).
Treatment of Anti-PLC-␥1 Immunocomplexes with Phospho-ERK2 Reverses the Tyrosine Kinase-dependent Stimulation of Immunocomplex PLC Activity-Phospho-ERK2 phosphorylation of recombinant PLC-␥1 did not alter lipase activity (Fig.  7A). In contrast, phospho-ERK2 exerted significant effects on anti-PLC␥1 immunocomplex PLC activity (Fig. 7B). Anti-PLC-␥1 immunocomplexes were isolated from rat hippocampal formation P2 preparations, then treated with or without ATP and with or without phospho-ERK2. Subsequently, PLC activity was measured. PLC-␥1 enzyme activity was significantly increased by treatment with ATP alone. We have shown previously that this effect of ATP is produced by a tyrosine kinase that is associated with the anti-PLC-␥1 immunocomplex (29). In the absence of ATP, phospho-ERK2 did not significantly alter PLC-␥1 enzyme activity. In the presence of ATP, phospho-ERK2 significantly reduced, but did not completely reverse, the increase in PLC-␥1 enzyme activity that was produced by a tyrosine kinase associated with the immunocomplexes. DISCUSSION We have made several important and novel observations in these studies. First, we provide evidence that PLC-␥1 associates with ERK2 in rat brain extracts and that dual phosphorylation of ERK2 in its activation loop is not critical for the association as demonstrated by minimal immunoreactivity of a phospho-specific antibody for ERK1/2 in a Western blot of PLC-␥1 immunoprecipitates. This association, therefore, was probably not mediated via binding to a phosphoacceptor site found within the PLC-␥1 sequence due to conformational constraints imposed on ERK2 by the non-phosphorylated "activation lip" of ERK2. However, the interaction of PLC-␥1 with phospho-ERK2 was direct, because recombinant PLC-␥1 was a substrate for recombinant phospho-ERK2 in vitro. This, in turn, is a second novel finding; i.e. PLC-␥1 is a MAPK substrate. The finding that PLC-␥2 was minimally phosphorylated by MAPKs under the same conditions demonstrated specificity of the kinase reactions. Although our results demonstrate that PLC-␥2 is a poor phospho-MAPK substrate, it is possible that MAPK-dependent phosphorylation of PLC-␥2 may occur under other conditions: e.g. after PLC-␥2 has been phosphorylated by another protein kinase. Third, using both peptide-binding and MAPK substrate phosphorylation assays, we identified a peptide sequence within the primary sequence of PLC-␥1 that conforms to the consensus sequence for an ERK-docking site, a D-domain. A peptide version of the PLC-␥1 D-domain inhibited the in vitro phosphorylation of PLC-␥1 by phospho-ERK2. This result demonstrates that the interaction between PLC-␥1 and phospho-ERK2 is dependent on the common docking domain of phospho-ERK2 and strongly implicates the D-domain of PLC-␥1 as the ERK2-binding site on PLC-␥1. Furthermore, it indicates that interactions between other sites on phospho-ERK2 (e.g. putative PLC-␥1 interaction motifs shown in Fig.  1B) and PLC-␥1 do not play a significant role in PLC-␥1phospho-ERK2 binding. Finally, phospho-ERK2 treatment of anti-PLC-␥1 immunocomplexes under conditions allowing for substrate phosphorylation revealed that phospho-ERK2 opposes tyrosine kinase-dependent stimulation of PLC-␥1 enzyme activity, whereas phosphorylation of recombinant PLC-␥1 by phospho-ERK2 under similar conditions did not affect the lipase activity of PLC-␥1.
Identification of the PLC-␥1 D-domain sequence was based on consensus sequences defined by Kornfeld and colleagues (5)(6). Other investigators (e.g. Refs. 3, 12-13, 33) have identified similar, but different, MAPK docking sequence motifs. In general, MAPK binding motifs consist of peptide sequences of ϳ20 amino acids, or less. Rather than specifying consensus sequences, Sharrocks and colleagues (34) have developed models of MAPK docking domains that identify regions of sequence similarity: specifically, basic, LXL, and hydrophobic motifs. These regions play differing roles in dictating specificity for interactions with members of the ERK, JNK, and p38 stressactivated protein kinase families. The p38 kinases bind proteins that have a basic region and hydrophobic region, whereas the intervening LXL motif is dispensable; ERKs require the presence of all motifs, although the hydrophobic motif is less important than the other two regions (34). The PLC-␥1 Ddomain that we identified consists of a basic region and an (L/I)XL motif; in addition, an adjacent hydrophobic region ( 953 LVV) is readily identifiable. This indicates that, in addition to binding ERKs, the PLC-␥1 D-domain may bind p38 and JNK family members. In support of this proposal, we have found that PLC-␥1 is an in vitro substrate for p38␥ and JNK3.
PLC-␥1⅐ERK2 protein complexes were immunoprecipitated from rat brain lysates, indicating that the observed interactions between recombinant proteins are not simply in vitro artifacts. However, we were able to detect only minimal antiphospho-ERK2 immunoreactivity associated with anti-PLC-␥1 immunocomplexes. It is worth noting that, similar to our data, Husi et al. (35) reported that the N-methyl-D-aspartic acid receptor is isolated in association with ERK2, but not phospho-ERK2, and Loeb et al. (36) reported that the TrkA receptor is found in association with ERK1, but not phosph-ERK1, or either form of ERK2. Interestingly, PLC-␥1 has been shown to co-immunoprecipitate with an unidentified protein of molecular mass of ϳ42,000 Da, and the amount of phosphorylated protein (molecular mass ϳ 42,000 Da) associated with anti-PLC-␥1 immunoprecipitates is significantly increased following receptor activation (37,38). Our results indicate that this protein may be ERK2.
Results of the experiments in which we assessed the effects of phospho-ERK2 pretreatment on in vitro PLC activity indicate that PLC-␥1 is regulated, either directly or indirectly, by phospho-ERK2. Phospho-ERK2 treatment opposed stimulation of PLC-␥1 enzyme activity by a tyrosine kinase that co-immunoprecipitated with PLC-␥1. Two possible mechanisms may underlie this effect. Phospho-ERK2-catalyzed phosphorylation PLC-␥1 was affinity captured as recombinant PLC-␥1 (A), or from rat hippocampal formation postnuclear particulate (P2) fraction (B). Captured recombinant PLC-␥1 was treated with one of the following: buffer, buffer containing recombinant phospho-ERK2 (ϩpERK2); or buffer containing recombinant phospho-ERK2 and ATP (ϩpERK2 ϩ ATP). Captured PLC-␥1 from rat hippocampal formation postnuclear particulate (P2) fraction was treated with buffer; buffer containing ATP (ϩATP); buffer containing recombinant phospho-ERK2 (ϩpERK2); or buffer containing recombinant phospho-ERK2 and ATP (ϩpERK2 ϩ ATP). Subsequently, PLC activity associated with the immunocomplex was determined. PLC activity is calculated as nanomoles of Ins(1,4,5)P 3 product formed/min/mg of protein, then expressed within each experiment as a percentage of the mean of the buffer value. Each point is the average Ϯ S.E. (n ϭ 3 preparations) after background subtraction. In B, one-way analysis of variance revealed a significant overall effect of treatment F (3, 11) ϭ 12.61, p Ͻ 0.002. Newman-Kuels multiple comparison test revealed p Ͻ 0.01 (**) for buffer versus ϩATP and p Ͻ 0.05 for ϩATP versus ϩpERK2ϩATP.
of PLC-␥1 may oppose tyrosine kinase-dependent phosphorylation of the enzyme, similar to the mechanism proposed by Rhee and colleagues (39) for the regulation of PLC-␥1 activity by PKA and PKC. Tyrosine phosphorylation, and consequent activation, of PLC-␥1 may be reduced either as the result of reduced efficiency of PLC-␥1 phosphorylation by one or more tyrosine kinases or increased efficiency of PLC-␥1 dephosphorylation by one or more tyrosine phosphatases. Alternatively, phospho-ERK2 may indirectly decrease PLC-␥1 enzyme activity as the result of inhibiting the catalytic activity of the tyrosine kinase that is responsible for the genistein-inhibited activation of the isozyme. Studies aimed at testing these possibilities are being conducted in our laboratory. Finally, it should be noted that, because PLC-␥1 is commonly positioned early within a signal transduction pathway, relatively small changes in PLC-␥1 enzyme activity become amplified, often producing 2-to 4-fold changes in physiologic responses (e.g. functioning of ion channels, changes in gene expression) (40).
The physiologic significance of associations between PLC-␥1 and MAPKs is currently uncertain. Several possibilities exist. First, MAPKs may regulate the catalytic activity of PLC-␥1. Our results support a model (see Fig. 8) in which phospho-ERK2 acts to oppose tyrosine kinase-dependent activation of PLC-␥1, indicating that MAPKs, at least phospho-ERK2, may act to down-regulate PLC-␥1 signaling. This is in contrast to the reported stimulatory effect of phospho-ERK1/2 on PLC-␤1 (41). A second possible significance of MAPK interactions with PLC-␥1 is that PLC-␥1 may target MAPKs to signaling complexes containing components (e.g. PKC, Raf, and MEK) of pathways that produce MAPK activation, thus increasing the efficiency of transducing signals from PLC-␥1 to MAPKs. Once activated, the phospho-MAPK may, or may not, dissociate from PLC-␥1 and regulate downstream effectors (e.g. transcription factors). In support of this proposal are reports of PLC-␥1 association with PKC-␣ (42) and Raf (23), and the identified roles of PLC-␥1 in the control of cellular processes that are dependent on gene transcription (43)(44)(45). Third, association of a MAPK with a PLC isozyme may allow for the MAPK to gain access to, and phosphorylate, a protein (e.g. one bound to the SH2 and/or SH3 domains of PLC-␥1) that does not itself directly bind to the MAPK. This is analogous to the demonstration that JNK can phosphorylate proteins lacking JNK-binding sites but that are bound to c-Jun (46).
PLC-␥1 has been reported to directly interact with various classes of cellular proteins, including receptor and non-receptor tyrosine kinases (26,(47)(48), ion channel-forming receptors (49), phospholipase D2 (50), the synaptic vesicle protein, synaptojanin (51), the p21Ras-specific guanine nucleotide exchange factor SOS1 (21), and the serine/threonine kinase Raf1 (23). Our studies demonstrate that ERK2 and phospho-ERK2 can be added to this list. Thus, in addition to regulating cellular functions via controlling Ins(1,4,5)P 3 and 1,2-DAG production, PLC-␥1 appears to act as a scaffolding protein, which may account for the observations that some PLC-␥1-mediated cellular responses are independent of catalytic activity (19 -22). Additional studies are needed to determine whether MAPK-dependent phosphorylation of PLC-␥1 regulates its functioning as a scaffolding protein.