INTRODUCTION

The v-fms oncogene of feline sarcoma virus encodes a modified receptor tyrosine kinase that differs from the cellular receptor for the colony-stimulating factor 1 (CSF-1, also termed M-CSF)¹ only in seven amino acid positions and in the C-terminal sequence (1, 2, 3, 4). Both proteins are known to share overall structural similarities with the platelet-derived growth factor (PDGF) receptor (5, 6). The c- and v-Fms molecules contain a large extracellular domain that binds CSF-1 and a cytoplasmic tyrosine kinase domain, split by an insertion of approximately 70 amino acids termed the kinase insert (KI) region (Fig. 1). Furthermore, a segment of 35 amino acids, termed the juxtamembrane (JX) domain, separates the membrane spanning domain from the first tyrosine kinase domain (Fig. 1). Through two of the amino acid substitutions in the extracellular domain and the replacement of the C terminus, the v-Fms molecule is thought to have gained biochemical properties that are observed with the c-Fms polypeptide only transiently upon binding to CSF-1 (3, 7). Activation of the tyrosine kinase leads to autophosphorylation of the cytoplasmic domain of the Fms molecule at multiple sites. The newly formed phosphotyrosine residues constitute binding sites for Src homology 2 (SH2) domain-containing cytoplasmic proteins that may participate in the control of mitogenic pathways, cell metabolism, or cell morphology. The entire cytoplasmic domain of v-Fms (408 amino acids) contains 18 tyrosine residues that are conserved among human, mouse, feline, and chicken c-Fms. All c-Fms proteins, however, contain an additional tyrosine residue at the C-terminal end (3, 8, 9).² Three tyrosine phosphorylation sites in the KI domain, Tyr⁶⁹⁶, Tyr⁷⁰⁵, and Tyr⁷²⁰ (corresponding to Tyr⁶⁹⁷, Tyr⁷⁰⁶, and Tyr⁷²¹ in the mouse c-Fms), and Tyr⁸⁰⁷ in the second kinase (K2) domain, have been mapped previously (10, 11, 12, 13, 14, 15, 16, 17). Tyrosine phosphorylation sites of Fms-related tyrosine kinase receptors and the corresponding proteins binding to such sites have been studied extensively (18). For the PDGF  receptor, nine tyrosine phosphorylation sites have been mapped, two of which are located in the JX domain (19, 20). These two tyrosine phosphorylation sites, Tyr⁵⁷⁹ and Tyr⁵⁸¹, provide the binding sites for polypeptides belonging to the Src family (19, 20) and Shc (21). Although it has been suggested that the corresponding tyrosine, Tyr⁵⁵⁸, in the Fms JX domain provides a similar function (22), neither phosphorylation of this residue nor the potential regulatory role for downstream cascades have been studied.

In this paper, we show that Tyr⁵⁴³ in contrast to Tyr⁵⁵⁸ is indeed a major autophosphorylation site of the v-Fms polypeptide. We demonstrate that cellular proteins including the growth factor receptor bound protein 2 (Grb2) and p85 and p110 subunits of phosphatidylinositol (PI) 3 kinase bind specifically to recombinant proteins containing the KI domain but not to those containing the JX domain. A fusion protein containing this latter domain binds a yet undefined 55-kDa cellular protein in a Tyr⁵⁴³ phosphorylation-dependent manner.

EXPERIMENTAL PROCEDURES

Mouse NIH 3T3 cells, wt-v-Fms cells (NIH 3T3 cells expressing the wild type v-fms gene), or Y696F-, Y705F-, or Y807F-v-Fms cells (expressing mutant v-Fms proteins in which a single tyrosine in position 696, 705, or 807, respectively, was replaced by phenylalanine) were grown in Dulbecco modified Eagle's medium supplemented with 10% fetal calf serum. FDCP-1Mac11 cells (23) were maintained in Dulbecco modified Eagle's medium supplemented with 10% fetal calf serum and WEHI3BD-conditioned medium as a source of interleukin-3 at a concentration that stimulated optimal cell growth.

An anti-v-Fms antiserum was used as described previously (24). Monoclonal antibodies against phosphotyrosine (4G10), Shc, and the p85 subunit of PI 3 kinase were purchased from Upstate Biotechnology Incorporated (Lake Placid, NY). Polyclonal cross-reactive antibody that bound to the SH2 domains of both the p85 and p55 subunits of PI 3 kinase, and monoclonal antibodies against Grb2 or Nck were from Transduction Laboratory (Lexington, KY). A polyclonal antibody against the p110 subunit of PI 3 kinase was from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-cst antibody was kindly provided by S. A. Courtneidge (SUGEN, Redwood City, CA), monoclonal antibody against p120Ras GTPase-activating protein (GAP) was provided by S. J. Parsons (University of Virginia Health Science Center, Charlottesville, VA) (25). Antibody against protein tyrosine phosphatase, SHP-1, was provided by A. Ullrich (Munich, FRG).

Generation of the Y705F and Y807F mutations and transfection of eukaryotic cells were performed as described previously (16). The Y696F mutation was made with an oligonucleotide-directed in vitro mutagenesis system (Amersham Buchler, Braunschweig, FRG) employing the synthetic oligonucleotide 5-GATGTTCTTGAAGCCAGCGCC-3.

Several glutathione S-transferase (GST) v-Fms fusion proteins were generated in the pGex system (Pharmacia Biotech Inc.), which contained various v-Fms sequences at their C terminus. GST-JX-Fms contained residues 535 to 571 with and without Y537F or Y543F mutations; GST-KI-Fms contained residues 617 to 759 or, as specified, residues 674 to 724 with and without Y720F mutation; GST-K2-Fms contained residues 757 to 944; and GST-CT-Fms contained residues 904-944. For numbering of v-Fms residues see Ref. 3.

Strain TKX-1 (Stratagene, La Jolla, CA) was used for the isolation of phosphorylated GST Fms fusion proteins. Production of phosphorylated GST fusion proteins was performed as recommended by the manufacturer. The corresponding nonphosphorylated molecular species were isolated from Escherichia coli DH5 or BL21(DE3)-pT-Trx (a kind gift of S. Ishii, RIKEN, Ibaraki, Japan) (26). Purified GST fusion proteins (50 µg) were bound for 1 h at 4 °C to glutathione (GT)-agarose beads (40 µl of slurry; Pharmacia) suspended in binding buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl₂, 1 mM EGTA, 1% Trasylol, 100 µg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 200 µM sodium orthovanadate, 10 mM sodium pyrophosphate, and 10 mM sodium fluoride). Precharged beads were incubated overnight at 4 °C with lysates from in vivo Tran[³⁵S]-labeled (Met/Cys; ICN, Meckenheim, FRG) cells or unlabeled NIH 3T3 or FDCP-1Mac11 cells in a total volume of 2 ml of binding buffer. Beads were washed five times with binding buffer, and pellets were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE). For identification of proteins by immunoblotting, proteins were transferred onto Immobilon-P sheets (Millipore, Bedford, MA) by a semi-dry blotting technique. Bound antibody was visualized by incubation of blots in 3 ml of 20 mM Tris-HCl, pH 7.6, containing 137 mM NaCl and 2 µCi of ¹²⁵I-labeled anti-species-specific immunoglobulin G (IgG, ICN). Bound radioactivity was quantified with a model BAS1000 bio-imaging analyzer (Fuji Photo Film Co., Kanagawa, Japan).

Kinase assays, phosphoamino acid analyses, and tryptic peptide mapping were performed as described previously (24, 27). Fms-specific immune complexes were incubated for 20 min at room temperature with 3 µCi of [gamma-³²P]ATP (Amersham Buchler) in the presence of 10 mM MnCl₂ and analyzed by SDS-PAGE. For phosphoamino acid analyses, material was eluted from the gels and hydrolyzed for 2 h in 6 M HCl at 110 °C. Samples were analyzed by two-dimensional electrophoresis (24). For tryptic peptide mapping, material was digested with trypsin (300 µg/ml) for 20 h. Products were analyzed by two-dimensional electrophoresis and chromatography, as described (27).

RESULTS

The cytoplasmic domain of v-Fms contains 18 tyrosine residues in positions identical with those in the feline c-Fms molecule. The latter contains, however, an additional tyrosine residue, Tyr⁹⁷⁷, at the C-terminal end. This tyrosine was suggested to regulate tyrosine kinase activity (28). To compare the autophosphorylation sites of v-Fms and feline c-Fms, we isolated the two proteins from cell lysates by immunoprecipitation labeled them by in vitro autophosphorylation and performed phosphoamino acid analyses and tryptic peptide mapping (Fig. 2). Both gp150^c-fms and gp140^v-fms (Fig. 2A) were phosphorylated exclusively on tyrosine residues (Fig. 2B) and yielded closely related tryptic phosphopeptide maps with six identical major and four minor phosphopeptides. We observed, however, a single minor phosphopeptide (marked with an arrowhead in Fig. 2C, 1) that was unique for the c-Fms molecule. It is likely, therefore, that the peptide underlying this spot contains Tyr⁹⁷⁷.

A compilation of published data on Fms autophosphorylation sites involves Fms molecules from several animal species (10, 11, 12, 13, 14, 15, 16, 17) and suggests that by analogy, Tyr⁶⁹⁶, Tyr⁷⁰⁵, Tyr⁷²⁰, and Tyr⁸⁰⁷ of the v-Fms molecule (numbering according to Ref. 3) should also be phosphorylated. Experimentally, however, this has been proven only for Tyr⁷⁰⁵ (16) and Tyr⁸⁰⁷ (16, 17). We first wanted to assess the role of Tyr⁶⁹⁶. For this purpose, we expressed mutant v-Fms proteins in mouse NIH 3T3 cells in which Tyr⁶⁹⁶, Tyr⁷⁰⁵, or Tyr⁸⁰⁷ were individually replaced by phenylalanines. Mutant and wild type v-Fms molecules were isolated and phosphorylated as above and analyzed by tryptic peptide mapping (Fig. 3). In agreement with previous studies, spots e or c were absent from samples lacking Tyr⁷⁰⁵ or Tyr⁸⁰⁷, respectively. Upon mutation of Tyr⁶⁹⁶, spots b and b disappeared. These data show that Tyr⁶⁹⁶ of v-Fms is indeed a major autophosphorylation site. Three major phosphopeptides designated a, d, and f and four minor phosphopeptides, designated g, h, i, and j remain to be assigned to phosphorylation of specific sites.

To define additional tyrosine phosphorylation sites, we incubated recombinant GST fusion proteins containing various C-terminal v-Fms sequences in autophosphorylation assays together with wild type v-Fms-containing immune complexes (Fig. 4). Three of the fusion proteins, GST-JX-Fms, GST-KI-Fms, and GST-K2-Fms, became phosphorylated on tyrosine, whereas GST-CT-Fms (containing Tyr⁹²¹) and GST were no substrates for the v-Fms kinase. The level of phosphorylation of GST-K2-Fms was about 5-fold lower than observed with GST-JX-Fms and GST-KI-Fms (data not shown). We have shown previously that the second kinase domain of v-Fms contains Tyr⁸⁰⁷ as a major phosphorylation site, yielding spot c upon phosphopeptide mapping (17). GST-KI-Fms (containing Tyr⁶⁶², Tyr⁶⁶⁵, Tyr⁶⁹⁶, Tyr⁷⁰⁵, and Tyr⁷²⁰) yielded four phosphopeptides, migrating in positions b, b, e, and f in Fig. 5C. The deletion of Tyr⁶⁶² and Tyr⁶⁶⁵ did not alter this pattern of phosphopeptides (data not shown), suggesting that these residues are not phosphorylated. A replacement of Tyr⁷²⁰ by phenylalanine (GST-Y720F-KI-Fms) resulted in the disappearance of spot f (data not shown). Together, these data indicate that material underlying spots b, b, and e contain phosphotyrosines Tyr⁶⁹⁶ and Tyr⁷⁰⁵ and that spot f represents a peptide containing Tyr⁷²⁰. Importantly, tryptic peptide maps from GST-JX-Fms revealed one major and one minor phosphopeptide that co-migrated with peptides a and g derived from autophosphorylated full-length v-Fms (Fig. 5, A and B). This suggests that the juxtamembrane domain of the Fms tyrosine kinase also contains autophosphorylation sites.

Which tyrosine residue of the JX domain is phosphorylated? The GST-JX-Fms contains 5 tyrosine residues at amino acid positions 537, 543, 553, 558, and 563 and could potentially yield three Fms-specific fragments upon exhaustive trypsin digestion. The first fragment contains Tyr⁵³⁷, the second contains Tyr⁵⁴³, and the third contains tyrosine residues Tyr⁵⁵³, Tyr⁵⁵⁸, and Tyr⁵⁶³. We generated mutant GST-JX-Fms fusion proteins in which Tyr⁵³⁷ or Tyr⁵⁴³ were individually replaced by phenylalanine residues and used them as substrates for the v-Fms tyrosine kinase. Whereas GST-Y537F-JX-Fms was phosphorylated to the same extent as the GST-wt-JX-Fms (Fig. 6, A and B, lanes 2), the Y543F point mutation completely abolished phosphorylation (Fig. 6, A and B, lanes 3), suggesting that Tyr⁵⁴³ is indeed one of the main phosphorylation targets of the v-Fms tyrosine kinase. To further support that phosphorylation of Tyr⁵⁴³ correlates with the presence of spot a, we compared the phosphorylation pattern full-length wt-v-Fms and Y543F-v-Fms as derived from transfected NIH 3T3 cells. Fig. 6 (C and D) shows that the mutant protein exhibited unchanged tyrosine kinase activity but specifically lacked spot a in phosphopeptide analyses.

In Vivo Phosphorylated GST-KI-Fms Fusion Protein Associates with Grb2 and the p110 and p85 Subunits of PI 3 Kinase, Whereas GST-JX-Fms Protein Associates with p55

To examine whether tyrosine phosphorylation of the JX domain generates novel binding sites for cellular proteins, we purified GST, GST-JX-Fms, and GST-KI-Fms from E. coli strain TKX-1, which expresses the active tyrosine kinase Elk. Fig. 7B shows that GST-Fms proteins from this strain indeed contained phosphotyrosine (lanes 4 and 6), as demonstrated by phosphotyrosine immunoblotting. Control preparations of the same proteins from E. coli strain DH5 lacked phosphotyrosine (Fig. 7B, lanes 3 and 5). Regarding an interaction of c-Fms with cellular proteins, several SH2 domain-containing proteins including PI 3 kinase (14), Grb2 (15), members of Src-family (29), and p120RasGAP (17) were shown to bind to the activated receptor molecule. In hematopoietic cells, p150, a protein with an as yet unknown function, was shown to associate with c-Fms (30), and furthermore, SHP-1 (previously termed PTP-1C; Refs. 31 and 32) appears to be one of the major substrates of the c-Fms tyrosine kinase in these cells (33). For these reasons, we employed cell lysates from the premyeloid cell line FDCP-1Mac11 (23) and NIH 3T3 cells to study binding of cellular proteins to the above phosphorylated GST-Fms fusion proteins. In agreement with previous work (14), the p85 subunit of PI 3 kinase bound to phosphorylated GST-KI-Fms (GST-pY-KI-Fms, Fig. 7C). In addition, the adaptor protein Grb2 was detected in the GST-pY-KI-Fms fraction (15). No other proteins such as Shc, SHP-1, Nck, p120RasGAP, or members of the Src-family were detected in this fraction. Furthermore, none of these proteins bound to the GST-pY-JX-Fms (Fig. 7C).

Binding of Grb2 and PI 3 kinase to in vivo phosphorylated GST-KI-Fms.

To learn whether additional cellular proteins associated with the individual GST-pY-Fms proteins, we next employed metabolically labeled cell lysates from the same cells. As expected, p85 and p110, the two subunits of PI 3 kinase, were found to associate with GST-pY-KI-Fms (Fig. 8, A, open circles, and B). Neither of the two proteins were found in the eluate from GST-pY-JX-Fms, again underscoring the specificity of binding to the KI domain. Instead, two proteins with molecular masses of 55 and 80 kDa, respectively, were detected in the GST-pY-JX-Fms-bound fraction (Fig. 8A, closed circles). The 55-kDa species did not represent the p55 subunit of PI 3 kinase (34), because it was not detected with a cross-reactive antibody that bound to the SH2 domains of both the p85 and p55 subunits (Fig. 8B). Furthermore, p55 was clearly distinct from Shc, because it was not reactive with a Shc-specific monoclonal antibody (Fig. 7C). It should be noted that similar negative results were obtained with cell lysates from v-Fms-transformed NIH 3T3 cells and from CSF-1-stimulated c-Fms overexpressing FDCP-1Mac11 cells (data not shown).

To further examine whether this binding correlated with tyrosine phosphorylation of the Fms-specific segment, we employed the two GST-JX-Fms fusion proteins in which either Tyr⁵³⁷ or Tyr⁵⁴³ were replaced by phenylalanine residues. In agreement with the in vitro phosphorylation data (Fig. 6), the wild type and the GST-Y537F-JX-Fms fusion proteins contained phosphotyrosine when isolated from E. coli TKX-1 (Fig. 8C, lanes 1 and 2). In contrast, the level of tyrosine-phosphorylation was reduced by more than 80% in GST-Y543F-JX-Fms (Fig. 8C, lane 3), suggesting that Tyr⁵⁴³ is also a major phosphorylation site for Elk tyrosine kinase. Wild type GST-JX-Fms and the two mutants were incubated with radiolabeled FDCP-1Mac11 cell lysates. As shown in lane 3 of Fig. 8D, the point mutation of Tyr⁵⁴³ abolished the binding of p55, but not of p80. In contrast, GST-Y537F-pY-Fms continued to bind p55. Together, this is strong evidence that Tyr⁵⁴³ indeed constitutes a phosphorylation-dependent binding site for p55. Additional evidence for the specificity of interaction between p55 and the Fms tyrosine kinase would be provided, if p55 by itself was also phosphorylated on tyrosine. To address this issue, we analyzed the eluate from GST-JX-Fms-agarose beads by phosphotyrosine immunoblotting. Fig. 8E shows that p55 from v-Fms-transformed NIH 3T3 or CSF-1-stimulated FDC-P1Mac11 cells is indeed phosphorylated on tyrosine. Furthermore, we performed co-immunoprecipitation studies using wild type v-Fms and Y543F-v-Fms expressing cells. As shown in Fig. 8F, phosphotyrosine containing the 55-kDa protein was co-immunoprecipitated with the wild type v-Fms (lane 1) but not with the Y543F-v-Fms mutant polypeptide (lane 2).

DISCUSSION

Generally, tyrosine phosphorylation sites in growth factor receptors serve two purposes: (i) to control the state of activity of the kinase and (ii) to create binding sites for downstream signal transduction molecules, which in many cases also are substrates for the kinase.

In this paper, we show for the first time that Tyr⁵⁴³ in the JX domain of the v-Fms tyrosine kinase represents a major autophosphorylation site, and we demonstrate that phosphorylation at this site is a prerequisite for binding of a cellular protein both in premyeloid cells and in fibroblasts. Furthermore, we show that Tyr⁶⁹⁶, Tyr⁷⁰⁵, and Tyr⁷²⁰ of the KI domain and Tyr⁸⁰⁷ of the second kinase domain are phosphorylated. No phosphorylation was observed with mutant proteins that contained the C-terminal segment of v-Fms including Tyr⁹²¹. The above conclusions are supported by phosphopeptide analyses involving mutant v-Fms proteins derived from (i) mammalian cells, (ii) various bacterial strains as tyrosine-phosphorylated GST fusion proteins, or (iii) recombinant proteins labeled by in vitro tyrosine kinase reaction.

Regarding the mapping of tyrosine phosphorylation sites, the PDGF  receptor has been characterized in greater detail. Nine tyrosine residues including Tyr⁵⁷⁹ and Tyr⁵⁸¹ in the JX domain, Tyr⁷¹⁶, Tyr⁷⁴⁰, Tyr⁷⁵¹, and Tyr⁷⁷¹ of the KI domain, Tyr⁸⁵⁷ in the second kinase domain, and Tyr¹⁰⁰⁹ and Tyr¹⁰²¹ in the C-terminal end domain have been mapped as autophosphorylation sites (18). Three phosphorylation sites of the v-Fms molecule involving Tyr⁶⁹⁶, Tyr⁷²⁰, and Tyr⁸⁰⁷ reside in regions that exhibit significant sequence homology with the corresponding Tyr⁷¹⁶, Tyr⁷⁴⁰, and Tyr⁸⁵⁷ containing regions of the PDGF  receptor, thus indicating that these three residues might serve similar functions in the two receptor molecules.

The KI domain of the PDGF  receptor associates with Grb2, p85 of PI 3 kinase, Nck, and p120RasGAP (35, 36, 37, 38, 39, 40), and Tyr⁷¹⁶ and Tyr⁷⁴⁰ were shown to specifically bind Grb2 and PI 3 kinase (36, 37). These residues are located at positions corresponding to Tyr⁶⁹⁶ and Tyr⁷²⁰ in the KI domain of the v-Fms tyrosine kinase. We show that in agreement with this similarity, both molecules bound tightly to the GST-pY-KI-Fms protein in vitro.

Interestingly, binding of p85 was paralleled by stoichiometric binding of p110, the catalytic subunit of PI 3 kinase. This finding is in line with the notion that binding of p110 to the KI domain is mediated through the p85 subunit. The efficiency of p85 binding, however, varied drastically depending on the source of p85. Whereas p85 from FDCP-1Mac11 cells bound nearly quantitatively, less than 5 or 20% of p85 from nontransformed or v-Fms-transformed NIH 3T3 cells, respectively, associated with the GST fusion protein. The significance of this finding is unclear but may depend on cell culture conditions. It has been shown previously that growth factor-induced tyrosine phosphorylation of p85 may alter the affinity of its SH2 domains for phosphotyrosine residues (41).

Our finding that the JX domain of v-Fms fails to bind members of the Src tyrosine kinase family is in contrast to results obtained with the PDGF  receptor that can bind Src, Fyn, or Yes (19, 20) and contradicts co-immunoprecipitation studies performed with human c-Fms (22). Neither the GST-JX-Fms fusion protein used in this study nor a cross-reactive peptide-specific antibody recognizing the C-terminal regions of Src, Fyn, and Yes provided any evidence for a direct interaction between these cytoplasmic tyrosine kinases and v-Fms. It is possible, however, that under in vivo conditions, a growth factor-mediated activation of the Fms tyrosine kinase may also cause an activation of additional tyrosine kinases such as Tyk2 (42). It is possible, therefore, that Tyk2 in turn creates novel association sites on the Fms molecule for members of the Src family in a manner similar to that observed for the EGF receptor (43). The major finding of this study is the detection of binding of a 55-kDa cellular protein to pY543. The number of proteins with molecular masses of 45-70 kDa, which were identified as SH2 containing phosphotyrosine binding proteins, include Shc (p55), Src (p60), SHP-1 or SHP-2 (68 kDa), the p55 subunit of PI 3 kinase, and NCK (p48). None of these proteins, however, was detected in the bound fraction of GST-pY-JX-Fms. Our finding that the v-Fms-associated p55 contains phosphotyrosine raises the question as to whether p55 serves as a substrate for the v-Fms or yet another tyrosine kinase. Furthermore, it opens the possibility that p55 is regulated through tyrosine phosphorylation. Characterization of p55 should shed light on the biological function of this molecule and on the JX domain of the Fms tyrosine kinase.