p150 TSP , a Conserved Nuclear Phosphoprotein That Contains Multiple Tetratricopeptide Repeats and Binds Specifically to SH2 Domains*

Src homology 2 (SH2) domains are structural modules that function in the assembly of multicomponent signaling complexes by binding to specific phosphopeptides. The tetratricopeptide repeat (TPR) is a distinct structural motif that has been suggested to mediate protein- protein interactions. Among SH2-binding phosphoproteins purified from the mouse B cell lymphoma A20, a 150-kDa species was identified and the corresponding complementary DNA (cDNA) was molecularly cloned. This protein encoded by this cDNA, which we have termed p150 TSP (for TPR-containing, SH2-binding phos-phoprotein), is located predominantly in the nucleus and is highly conserved in evolution. The gene encoding p150 TSP ( Tsp ) was mapped to chromosome 7 of the mouse with gene order: centromere- Tyr - Wnt11 - Tsp - Zp2. The amino-terminal two-thirds of p150 TSP consist almost entirely of tandemly arranged TPR units, which mediate specific, homotypic protein interactions in transfected cells. The carboxyl-terminal third of p150 TSP , which is serine- and glutamic acid-rich, is essential for SH2 bind- ing; this interaction is dependent on serine/threonine phosphorylation but independent of tyrosine phospho- rylation. The sequence and binding properties of p150 TSP

Src homology 2 (SH2) 1 domains are conserved structural modules of about 100 amino acid residues that have been identified in tyrosine kinases of the Src family and in more than 60 other proteins (1). SH2 domains bind subsets of phosphotyrosine-containing peptides with high affinity (K D Ϸ 10 -000 nM) (2)(3)(4)(5); these interactions mediate assembly of diverse multicomponent signaling complexes. In addition to phospho-tyrosine-dependent interactions, phosphoserine/threonine-dependent binding to SH2 domains has also been reported (6 -9). In Bcr-Abl chimeras that are implicated in the pathogenesis of chronic myelogenous leukemia, the Bcr segment contains serine/threonine-and glutamic acid-rich regions that bind SH2 domains in a phosphorylation-dependent manner but independent of phosphotyrosine (6). Phosphotyrosine-independent binding of Raf-1 to the SH2 domains of Fyn and Src has also been described (8). More recently, we described SH2 binding by the cyclin-dependent kinase homologue p130 PITSLRE (9). This interaction is mediated by a serine-and glutamic acid-rich region of p130 PITSLRE and is likely to involve the same site in the SH2 domain that binds phosphotyrosine-containing peptides.
The tetratricopeptide repeat (TPR) is a 34-amino acid motif found in proteins that function in diverse processes, including cell cycle control, transcriptional repression, protein transport, and protein dephosphorylation (10). TPRs contain eight consensus residues whose size, hydrophobicity, and spacing are conserved. TPRs are predicted to form a pair of amphipathic, ␣-helical domains (A and B) that have been proposed to mediate TPR-TPR interactions (11,12). While there is as yet no evidence that TPR motifs interact directly, they have been shown to participate in interactions between TPR-containing proteins. For example, the TPR-containing proteins CDC23 and CDC27 form part of a complex that promotes anaphase (13,14); a mutation in the TPR region of CDC27 impairs its ability to interact with CDC23 (14). There is also evidence that TPRs mediate interactions with non-TPR-containing proteins: the transcriptional repression protein SSN6 (Cyc8), for example, interacts with specific DNA-binding proteins by means of its TPR region (15).
In work described here, SH2-binding phosphoproteins from the B-lymphoid cell line A20 were isolated by affinity chromatography. Internal peptide sequences from one of these proteins were used to molecularly clone a complementary DNA that encodes a hitherto unidentified protein of 150 kDa. This protein, which we have termed p150 TSP (for TPR-containing, SH2-binding phosphoprotein), contains 1173 amino acid residues and is located predominantly in the nucleus. The aminoterminal portion of p150 TSP contains a tandem array of 15 TPRs; the TPR-containing region mediates p150 TSP self-association in transfected cells. Specific binding of p150 TSP to SH2 domains is mediated by a serine-and glutamic acid-rich region near the carboxyl terminus. This interaction requires serine/ threonine phosphorylation but is independent of tyrosine phosphorylation. The sequence and binding properties of p150 TSP suggest that it may mediate interactions between TPR-containing and SH2-containing proteins.

MATERIALS AND METHODS
Protein Isolation and Peptide Sequencing-SH2-binding proteins were isolated from the B-lymphoid cell line A20 by affinity chromatography, fractionated by electrophoresis, and transferred to PVDF membranes as described previously (9). Generation, isolation, and sequenc-ing of tryptic peptides were performed by Drs. David F. Reim  Isolation of Complementary DNA Clones Encoding p150 TSP -Based on the sequence of peptide 1 (VQADVPPEILNNVGALHFR), a unique, 57-mer oligonucleotide probe (5Ј GTG CAG GCT GAT GTG CCC CCT GAG ATC CTG AAC AAT GTG GGC GCC CTG CAC TTC CGG 3Ј) was synthesized. The probe was labeled with 32 P using T4 polynucleotide kinase to a specific activity of 5 ϫ 10 8 cpm/g and hybridized to 5 ϫ 10 5 recombinant bacteriophage from a mouse spleen cDNA library in the vector Lambda Zap II (Stratagene). Hybridization was carried out overnight at 45°C in 6 ϫ SSCPE, 20% formamide, 5 ϫ Denhardt's solution, 10% dextran sulfate (Pharmacia Biotech Inc.), 0.1% SDS and 100 g/ml salmon sperm DNA at an oligonucleotide concentration of 1 g/liter.
Filters were washed twice for 6 min in 2 ϫ SSC, 0.1% SDS at room temperature and twice for 6 min in 2 ϫ SSC, 0.1% SDS at 45°C. Positive bacteriophage were isolated by two additional rounds of plaque purification, and cDNAs were recovered as recombinant pBluescript plasmids, using an excision protocol supplied by the manufacturer.
Recombinant plasmid DNA was carried through a second round of hybridization screening. A Conserved TPR-containing Protein That Binds SH2 Domains 6954 peptides 2 and 3, respectively, were synthesized. These were labeled with 32 P and hybridized sequentially to plasmid DNA that had been digested with SalI and NotI, fractionated by agarose gel electrophoresis, and transferred to nitrocellulose. Hybridization was carried out overnight in 6 ϫ SSCPE, 20% formamide, 5 ϫ Denhardt's solution, 10% dextran sulfate (Pharmacia), 0.1% SDS, and 100 g/ml salmon sperm DNA at an oligonucleotide concentration of 33 g/liter; hybridization was performed at 42°C for SD945 and at 44°C for SD944. Membranes were washed in 2 ϫ SSC, 0.1% SDS twice for 6 min at room temperature and then once for 8 min at 42°C (for SD945) or once for 10 min at 44°C (for SD944). Between hybridizations,the membrane was stripped of probe by heating for 30 min at 68°C in 0.1 ϫ SSC, 0.1% SDS.
Mapping of the Mouse Tsp Gene-For genetic mapping we analyzed the progeny of the cross (NFS/N ϫ Mus spretus) ϫ M. spretus or C58/J (16), which have been typed for more than 650 markers, including the chromosome 7 markers Tyr (tyrosinase), Wnt11 (wingless-related gene 11), and Zp2 (zona pellucida 2) (16,17). Parental mouse DNAs were screened for restriction fragment length polymorphisms of Tsp using a full-length Tsp cDNA probe (clone . The Tsp probe detected HindIII fragments of 8.6, 6.3, 3.0, 2.5, and 2.4 kb in NFS/N and C58/J, and HindIII fragments of 9.2, 7.0, 2.9, 2.5, and 2.4 kb in M. spretus. Inheritance of these fragments was compared with inheritance of 650 markers previously typed in these crosses and mapped to positions on all 19 autosomes and the X chromosome. Data were stored and analyzed using the program LOCUS developed by C. E. Buckler (NIAID, National Institutes of Health, Bethesda, MD). Recombinational distances were determined according to Green (18), and markers were ordered by minimizing the number of recombinants.
Localization of p150 TSP by Immunofluorescence-NIH3T3 cells were seeded on glass coverslips and grown for 24 -48 h to about 50% confluence. Cells were washed three times with phosphate-buffered saline. The cells were fixed and permeabilized with cold (Ϫ20°C) methanol for 5 min, then rehydrated with phosphate-buffered saline. Between subsequent steps, the coverslips were washed three to five times for 1 min each with KB buffer (150 mM NaCl, 10 mM Tris-Cl (pH 7.6), 0.1% bovine serum albumin), supplemented with 0.1% Nonidet P-40 (KB ϩ Nonidet P-40). The cells were incubated with primary antibodies for 1 h, washed with KB ϩ Nonidet P-40, then incubated with secondary reagents for 1 h. The cells were again washed with KB ϩ Nonidet P-40, then stained for 1 min with 4,6-diamidino-2-phenylindole (1 g/ml) in KB buffer. Finally, cells were washed twice with KB buffer and mounted in a glycerol solution (70% glycerol, 3% n-propyl gallate).
For immunofluorescence, affinity-purified Ab1543 and Ab1544 or total IgG from the corresponding preimmune sera were used at 0.3 g/ml in KB ϩ Nonidet P-40. Mouse anti-p150 TSP antibody Ab635 and mouse preimmune serum were used at 1:500 dilutions in KB ϩ Nonidet P-40. For double immunofluorescence experiments, microtubules were stained with the mouse anti-tubulin antibody TU27B (19) at 1:75 dilution or with rabbit anti-tubulin antibody Ra53 (provided by Dr. D. Murphy, Johns Hopkins University School of Medicine) at 1:50 dilution.
Secondary reagents used in these experiments included fluorescein isothiocyanate-conjugated horse anti-mouse IgG (Vector Laboratories, Burlington, CA) and fluorescein isothiocyanate-conjugated swine antirabbit IgG (Accurate Chemicals, Westbury, NY). Biotinylated goat antirabbit IgG or biotinylated horse anti-mouse IgG (Vector Laboratories) were also used, in conjunction with Texas Red-streptavidin (Life Technologies, Inc.). For competition experiments, 150 l of diluted, primary antibodies were preincubated for 30 min with 15 g of a purified GST fusion protein containing residues 1059 -1173 of p150 TSP .
Generation of Epitope-tagged p150 TSP -The nonapeptide influenza hemagglutinin (HA) epitope YPYDVPDYA, which is recognized by the mouse monoclonal antibody 12CA5, was fused to the carboxyl terminus of p150 TSP as follows. The 4.2-kb SalI-NotI insert from clone 19-4 was cloned into the vector pET-21 (Novagen) to yield pET-21-p150 TSP . Oligonucleotides 5Ј CAT GTGGCC CGG GCA CGC AAG 3Ј (sense) and 5Ј TATTTTTTTGCGGCCGCTGTCGACTCA AGC GTA GTC TGG GAC GTC GTA TGG GTA GTC GCT ATC ATC TGA CCC ATG 3Ј (antisense) were used as polymerase chain reaction primers to amplify a 1.1-kb fragment from pET-21-p150 TSP . The resulting cassette was exchanged for the SrfI-NotI fragment of pET21-p150 TSP . The entire SalI-NotI insert, encoding p150 TSP fused at its carboxyl terminus to the HA epitope, was then subcloned into pCIS2 to yield pCIS-p150 TSP -HAC.
Phosphorylation Dependence of SH2 Binding by p150 TSP -p150 TSP was expressed in 293 cells and labeled metabolically with 32 P. Cells (2 ϫ 10 7 ) were lysed in 10 ml of C buffer (100 mM NaCl, 25 mM Tris-Cl (pH 7.6), 25 mM NaF, 1 mM EDTA, 2 mM Na 3 VO 4 , 1 mM Na 2 MoO 4 , 1 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40, 10 g/ml leupeptin, 10 g/ml aprotinin, and 5 g/ml pepstatin), and lysates were clarified by centrifugation; p150 TSP was immunoprecipitated in reactions containing 1 ml (2 ϫ 10 6 cell eq) of lysate and 10 g each of Ab1543 and Ab1544, affixed to protein A-Sepharose beads. Immunoprecipitations were carried out for 3 h at 4°C; beads were collected by centrifugation and washed four times for 6 min each with 1 ml of C buffer. Immunoprecipitates were treated with calf intestine alkaline phosphatase (Sigma) as described (9). Samples were split into three aliquots, fractionated by electrophoresis, and transferred to PVDF membranes. One membrane was immunoblotted with a mixture of antibodies Ab1604 and Ab1544, and the remaining membranes were assayed for binding to biotinylated GST or GST-BlkSH2 fusion protein (9).

Isolation of p150 TSP by SH2 Affinity Chromatography and Molecular
Cloning of TSP cDNA-We have shown previously that SH2 domains from the tyrosine kinases Blk, Fyn(T), and Lyn bind distinct sets of phosphoproteins from the B-lymphoid cell line A20 (21) and have described the use of preparative scale SH2 affinity chromatography to identify specific SH2 ligands by protein microsequence analysis and molecular cloning (9). Among the SH2 ligands identified in lysates from A20 cells, we detected a phosphoprotein of apparent molecular mass 150 kDa. Amino acid sequences of six internal tryptic peptides from this protein were obtained (Fig. 1A). Based on one of these peptides, P1 (VQADVPPEILNNVGALHFR), we designed a 57mer oligonucleotide probe of unique sequence. This was used to screen 5 ϫ 10 5 recombinant bacteriophage from a mouse spleen cDNA library. Twenty positive clones were carried through a second round of screening by hybridization to two degenerate oligonucleotides, corresponding to residues 8 through 13 of peptide P2 (QAQYHV) and residues 1 through 7 of peptide P3 (DKGNFYE). Four clones hybridized to both degenerate oligonucleotides, including clones 19 -3 (3591 base pairs) and 19 -4 (4167 base pairs). The sequences of these and two overlapping clones, 13-1 and 17-3, define an open reading frame of 1173 codons, specifying a protein that we have termed p150 TSP (Fig.  1A). The p150 TSP protein sequence can be divided into two regions: the amino-terminal 815 residues contain 15 occurrences of the 34-amino acid TPR motif; the carboxyl-terminal 358 residues are rich in serine, glutamic acid, and aspartic acid and contain multiple potential casein kinase 2 phosphorylation sites (Fig. 1B). While the Tsp coding sequence predicts a protein with a molecular mass of 129 kDa, transcription and translation of Tsp cDNA in vitro yielded a predominant product whose apparent molecular mass was 150 kDa, in agreement with the size of the protein originally purified from A20 cells (data not shown) and with the size of endogenously expressed p150 TSP as detected by immunoblotting (see below).
Interrogation of nucleotide and protein sequence data bases using the TBLASTN algorithm (22) revealed 98.6% amino acid sequence identity between p150 TSP and a hypothetical, 1173codon open reading frame in the human genome (GenBank TM accession number D63875). In addition, p150 TSP was found to share 31% amino acid sequence identity with a hypothetical, 1245-amino acid protein encoded at locus B0464.2 of Caenorhabditis elegans (23). The homology between p150 TSP and the putative B0464.2 product extends from near the amino terminus (residue 14 of p150 TSP ) through the TPR-rich region and includes most of the carboxyl-terminal domain (to residue 1111 of p150 TSP ) (Fig. 2). Thus, B0464.2 is likely to encode a C. elegans homologue of p150 TSP . Remarkably, interrogation of the dbEST data base of expressed sequence tags (24) revealed homology between p150 TSP and the conceptual translation product of an expressed sequence tag from the higher plant Arabidopsis thaliana (T46289; 47% identity over 204 residues). The similarity between p150 TSP and the arabidopsis expressed sequence tag includes a TPR unit but extends beyond it (Fig. 2). This suggested that a progenitor of p150 TSP first appeared before the animal and plant kingdoms diverged. Consistent with this suggestion, a TBLASTN search also detected a hypothetical, 1045-amino acid open reading frame in Saccharomyces cerevisiae (AOE1045) that exhibits significant (smallest sum probability P(N) ϭ 1.4 ϫ 10 Ϫ55 , N ϭ 12) homology with p150 TSP (Fig. 2). The existence of p150 TSP homologues in nematodes, plants, and yeast indicates an extraordinary degree of evolutionary conservation.
Mapping of the Tsp Gene to Mouse Chromosome 7-To map the Tsp gene in the mouse, DNA samples from the progeny of a multilocus cross were examined for inheritance of a restriction enzyme length variant of Tsp as described under "Materials and Methods." The observed pattern of inheritance was compared with that of 650 markers, including the chromosome 7 markers Tyr (tyrosinase), Wnt11, and Zp2 (zona pellucida glycoprotein 2). Tsp was mapped to a position on chromosome 7 proximal to Zp2. The data indicate the following gene order and distances: Expression of Tsp RNA in Mouse Tissues-A probe specific for Tsp RNA was radiolabeled and hybridized to polyadenylated RNA from various mouse tissues. The Tsp probe detected a single RNA species of about 5.0 kb in every tissue examined (Fig. 3). This transcript is somewhat longer than the longest Tsp cDNA clone obtained (4.2 kb) suggesting that 5Ј-or 3Јuntranslated sequences are incompletely represented in the cDNA. This interpretation is consistent with the observation that the cDNA sequence lacks a polyadenosine tract.
Expression of p150 TSP and Binding to an SH2 Domain-To obtain additional evidence that the Tsp open reading frame encodes a physiologic gene product, we used antibodies directed against proteins encoded by TSP to detect immunoreactive species in cell lysates. Rabbit antibody Ab1544, which was raised against a GST fusion containing residues 1059 -1173 of p150 TSP , detected a 150-kDa protein in total lysates of the B-lymphoid cell lines A20 and WEHI231 (Fig. 4,  lanes 3 and 4). This antibody also detected a comigrating species among proteins from A20 cells that were specifically retained by a GST-BlkSH2 affinity matrix (Fig. 4, lane 6); this species was not detected in eluates from an affinity matrix containing GST alone (Fig. 4, lane 5). Expression of Tsp cDNA by transfection into 293 cells yielded a 150-kDa protein that was immunoreactive with Ab1544 and which comigrated with the endogenous, 150-kDa species from A20 and WEHI231 cells (Fig. 4, lane 1); longer exposure revealed a similar immunoreactive species in the extract of 293 cells transfected with vector alone (data not shown), which likely represents endogenous p150 TSP . In the A20 cell lysate, and to a lesser extent in the WEHI231 lysate, an additional species of about 120 kDa was also observed (Fig. 4, lanes 3 and 4); this may represent a proteolytic product of p150 TSP , as its yield was variable. The discrepancy between the predicted and observed mobilities of p150 TSP may reflect anomalous electrophoretic mobility caused by the acidic region. Phospho-rylation also contributes to this difference, as dephosphorylation of p150 TSP in vitro results in a 5-kDa diminution in apparent molecular mass (see Fig. 8).
p150 TSP Is Localized to the Cell Nucleus-The intracellular distribution of endogenous p150 TSP in NIH3T3 cells was examined by immunofluorescence microscopy (Fig. 5). In interphase cells stained with the rabbit anti-p150 TSP antibody Ab1544, speckled nuclear fluorescence was observed (Fig. 5A); fluorescence was reduced to background by an excess of the corresponding specific antigen (Fig. 5B). Similar nuclear staining was seen in cells probed with a mouse anti-p150 TSP antibody directed against residues 1-821 of p150 TSP (Fig. 5C), but not with the corresponding preimmune serum (Fig. 5D). We conclude that p150 TSP accumulates predominantly or exclusively in the cell nucleus.
Specificity of SH2 Binding by p150 TSP -The experiment of Fig. 4 demonstrated that endogenously expressed p150 TSP is retained by an SH2 affinity matrix; we proceeded to examine the specificity of this interaction. p150 TSP was expressed by transfection in 293 cells and labeled metabolically with 32 P. Cell lysates were adsorbed to wild-type or mutant GST-BlkSH2 fusion proteins, or with GST alone; retained proteins were fractionated by SDS-PAGE and visualized by autoradiography. Wild-type GST-BlkSH2 beads retained a 150-kDa, 32 P-labeled protein (Fig. 6, lane 1); this protein comigrated with the predominant species immunoprecipitated by anti-p150 TSP antibodies Ab1544 and Ab1543 (Fig. 6, lanes 7 and 8). (Several smaller 32 P-labeled species were also retained by the SH2 matrix; these likely represent polypeptides derived from p150 TSP , as they were immunoprecipitated by Ab1543 and weakly by Ab1544.) Beads coated with GST alone did not retain p150 TSP (Fig. 6, lane 6). Binding of p150 TSP to GST-BlkSH2 was abolished by a phosphotyrosine-containing peptide (EPQ-(pY)EEIOIYL) with high affinity for the SH2 domain of Src (4, 25) (Fig. 6, lanes 3 and 4), but was unaffected by an irrelevant, unphosphorylated peptide, SD12 (Fig. 6, lane 2). Binding was greatly reduced by a serine-to-cysteine substitution in the conserved FL(I/V)RESE region (S147C in Blk) (Fig. 6, lane 5). Taken together, these observations indicate that binding of p150 TSP to the SH2 domain involves a site that overlaps or coincides with the site that binds phosphotyrosine-containing peptides.
The Acidic Region of p150 TSP Mediates Phosphotyrosine-independent Binding to SH2 Domains-To define the region of p150 TSP responsible for SH2 binding, we tested a series of p150 TSP deletion mutants for retention by an SH2 affinity matrix. Proteins were expressed by transfection in 293 cells and labeled metabolically with [ 35 S]methionine and [ 35 S]cysteine. Expression and intracellular accumulation of each p150 TSP fragment was verified (data not shown). Cell lysates were adsorbed to a GST-BlkSH2 affinity matrix; retained proteins were fractionated by SDS-PAGE and visualized by autoradiography. Wild-type p150 TSP (Fig. 7, lane 2) and fragments of p150 TSP spanning residues 497-1173 (Fig. 7, lane 5) or residues 887-1173 (Fig. 7, lane 6) were retained by the SH2 affinity matrix; fragments spanning residues 1-821 (Fig. 7, lane 3) or residues 497-821 (Fig. 7, lane 4) were not retained. In this way, the SH2 binding site(s) of p150 TSP was localized to the interval between residues 887 and 1173. Because this region is devoid of tyrosine residues, it seemed likely that the binding of p150 TSP to SH2 is independent of phosphotyrosine; it remained formally possible, however, that binding was not direct but rather mediated by a third protein. To determine whether p150 TSP bound the SH2 domain directly, and whether this interaction was dependent on phosphorylation of p150 TSP , we used a filter immobilization assay.
SH2 Binding by p150 TSP Is Direct and Dependent on Phosphorylation-Wild-type p150 TSP was expressed by transfection in 293 cells and labeled metabolically with 32 P. Cells were lysed, and p150 TSP was immunoprecipitated with Ab1543 and Ab1544. Immunoprecipitates were treated with calf intestinal alkaline phosphatase (Fig. 8, lanes 1, 3, 5 , 7, 9, and 11) or left untreated (Fig. 8, lanes 2, 4, 6, 8, 10, and 12). Then each sample was split three ways and fractionated by SDS-PAGE. Protein was transferred to PVDF membranes and assayed in parallel for binding to biotinylated GST-BlkSH2 fusion protein (Fig. 8,  lanes 1-4), biotinylated GST (Fig. 8, lanes 5-8), or an anti-p150 TSP antibody mixture (Fig. 8, lanes 9 -12). The GST-BlkSH2 protein was observed to bind directly to p150 TSP (Fig.  8, lane 2). Treatment of the immunoprecipitates with alkaline phosphatase, however, substantially reduced binding of the GST-BlkSH2 fusion protein to p150 TSP (Fig. 8, lane 1). Phosphatase treatment reduced the amount of p150 TSP -associated phosphate by about 5-fold (Fig. 8, compare lanes 4, 8, and 12 to lanes 3, 7, and 11) but did not significantly affect the recovery of p150 TSP (Fig. 8, compare lanes 9 and 10). Binding of the GST-BlkSH2 protein to p150 TSP was dependent on the SH2 moiety, as little or no binding was observed with biotinylated GST alone (Fig. 8, lanes 5 and 6). Thus, the binding of BlkSH2 to p150 TSP is direct and phosphorylation-dependent. Because p150 TSP truncation mutants lacking tyrosine retain their ability to bind SH2 (Fig. 7), we conclude that SH2 binding by p150 TSP requires phosphorylation at serine or threonine residues.
Self-association of p150 TSP in Transfected Cells-Based on secondary structure predictions, TPR motifs have been proposed to mediate homotypic interactions (11,12) and have been shown to participate in the formation of complexes between TPR-containing proteins (14). The presence of an extensive TPR-containing region suggested that p150 TSP might undergo self-association. To test this, p150 TSP was tagged at its carboxyl terminus with a 9-amino acid influenza HA epitope (p150 TSP -HAC) and coexpressed in 293 cells with fragments of p150 TSP spanning residues 1-821, 497-1173, 497-821, or 887-1173. Protein was labeled metabolically with 35 S and immunoprecipitated with the anti-HA antibody 12CA5 in the presence (Fig. 9,  lanes 2, 4, 6, and 8) or absence (Fig. 9, lanes 1, 3, 5, and 7) of a specific HA competitor peptide. Immunoprecipitations were carried out in parallel with anti-p150 TSP antibody Ab1544 (Fig.  9, lanes 9 -12). Precipitated proteins were fractionated by SDS-PAGE and detected by autoradiography. A 35 S-labeled protein corresponding to p150 TSP -HAC was precipitated from each of ; beads were subsequently washed, and bound protein was fractionated. Protein was transferred to a PVDF membrane, and protein was detected by immunoblotting with antibody Ab1544. Bound primary antibody was detected using a horseradish peroxidase-conjugated anti-rabbit antibody and an enhanced chemiluminescence assay. The apparent sizes (in kilodaltons) and positions of molecular mass standards are indicated at left. the transfected cell lysates by the 12CA5 antibody (Fig. 9, lanes  1, 3, 5, and 7, closed arrow); precipitation of this protein was greatly reduced in the presence of an HA peptide competitor (Fig. 9, lanes 2, 4, 6, and 8). p150 TSP fragments comprising residues 1-821, 497-1173, and 497-821 were observed to coprecipitate with p150 TSP -HAC (Fig. 9, lanes 1, 3, and 5). The fragment spanning residues 887-1173, however, was not precipitated (Fig. 9, lane 7), despite the fact that all four fragments could be immunoprecipitated from lysates of transfected cells by Ab1544 (Fig. 9, lanes 9 -12). Precipitation of fragments 1-821, 497-1173, and 497-821, like that of p150 TSP -HAC, was greatly reduced in the presence of the HA peptide (Fig. 9, lanes  2, 4, and 6). Fragments 1-821 and 497-821 were also present in immunoprecipitates of p150 TSP -HAC performed with FIG. 5. p150 TSP is localized predominantly to the cell nucleus. NIH3T3 cells were grown to subconfluence, fixed with methanol, and stained (red fluorescence) with the affinity-purified, rabbit anti-p150 TSP antibody Ab1544 in the absence (A) or presence (B) of 15 g of a purified, GST-p150 TSP fusion protein containing amino acid residues 1059 -173 of p150 TSP . Cells were similarly stained with a 1:500 dilution of mouse anti-p150 TSP antiserum Ab635 (C) or a 1:500 dilution of the corresponding preimmune serum (D). Binding of biotinylated secondary antibodies was detected with Texas Red-streptavidin. Microtubules were stained with the mouse anti-tubulin antibody TU27B or with rabbit anti-tubulin serum Ra53 (green fluorescence). DNA was visualized by staining with 4,6-diamidino-2-phenylindole (blue fluorescence).

FIG. 6. Specific binding of Tsp products to the BlkSH2 domain in vitro.
The 293 cell line was transfected with a plasmid encoding p150 TSP and labeled metabolically with 32 P. Lysate was adsorbed to beads coated with the following proteins: GST-BlkSH2 (lanes 1-4), GST-BlkSH2 S147C (lane 5), or GST alone (lane 6). Binding was carried out in the absence of competitor (lanes 1, 5, and 6 Ab1544 (Fig. 9, lanes 9 and 10) (Because Ab1544 recognizes a carboxyl-terminal p150 TSP epitope, the presence of fragments 497-1173 and 887-1173 in Fig. 9, lanes 11 and 12, is uninformative.) Thus, fragments of p150 TSP derived from the TPRcontaining region are able to associate, directly or indirectly, with p150 TSP in transfected cells. DISCUSSION We have used SH2 affinity chromatography to isolate SH2binding proteins from the B-lymphoid cell line A20. By partial peptide sequence determination and molecular cloning, one of these SH2 ligands was identified as a hitherto undescribed, ubiquitously expressed protein of 1173 amino acid residues, which we have termed p150 TSP . p150 TSP has a predicted molecular mass of 129 kDa but migrates as a protein of 150 kDa in SDS-polyacrylamide gels; anomalous mobility may be conferred by the acidic, carboxyl-terminal portion of the protein.
Comparison of the individual TPR motifs of p150 TSP provides the consensus (I/L/V)xxx(I/L/V)xL(A/G)xx(Y/F)xxxx(D/E)xxx-Axxx(F/Y)xxAL(R/K)xxxxx. This is in close agreement with the canonical TPR motif, xxxWxxLGxxYxxxxxxxxAxxxFxxAxxxxPxx (11,12). The p150 TSP TPR consensus differs from the canonical sequence in that tryptophan is not well conserved at position 4; nonetheless, in 10 out of the 15 TPR motifs in p150 TSP , hydrophobic residues are found at that position. Another difference from the canonical TPR motif is the poor conservation of proline at position 32. This difference, however, is not unique to p150 TSP ; for example, in the human serine/threonine phosphatase PP5 only one of four TPRs contains proline at that position (26). Structural, genetic, and biochemical observations have suggested that TPRs mediate formation of specific protein complexes (10). Consistent with these data, we have shown that p150 TSP undergoes self-association and that this interaction is mediated by the amino-terminal, TPR-containing region. Whether this association is mediated by direct interactions between TPR motifs has yet to be demonstrated.
The carboxyl-terminal acidic region of p150 TSP mediates binding to SH2 domains. While binding of GST-BlkSH2 to filter-immobilized p150 TSP was observed at a fusion protein concentration of 100 nM, estimation of the affinity of SH2 binding by p150 TSP is complicated by several factors, including the possible existence of multiple binding sites in the acidic region of p150 TSP , multimerization of p150 TSP through interactions between TPR-containing regions, and the ability of GST-SH2 fusion proteins to dimerize. Two lines of evidence indicate that SH2 binding by p150 TSP is dependent on phosphorylation but independent of phosphotyrosine. First, a 287amino acid fragment of p150 TSP which lacks tyrosine residues retains the ability to bind SH2. Second, SH2 binding was greatly reduced when p150 TSP was dephosphorylated by treatment with an alkaline phosphatase. Despite the lack of a requirement for phosphotyrosine, p150 TSP appears to interact with the same site on the SH2 domain that binds phosphoty- FIG. 8. Binding of p150 TSP to SH2 is direct and requires phosphorylation. The 293 cell line was transfected with an expression construct encoding p150 TSP and labeled metabolically with [ 32 P]orthophosphate. Cells were lysed and p150 TSP was immunoprecipitated with a mixture of antibodies Ab1544 and Ab1543. Immunoprecipitates were treated with calf intestinal alkaline phosphatase (lanes 1, 3, 5, 7, 9, and 11) or left untreated (lanes 2, 4, 6, 8, 10, and 12). Treated and untreated samples were each split into three portions which were fractionated by electrophoresis through a 7.5% SDS-polyacrylamide gel; protein was transferred to a PVDF membrane. Pairs of treated and untreated samples were assayed for binding to a biotinylated GST-BlkSH2 fusion protein (lanes 1-4) or biotinylated GST (lanes 5-8); another pair of samples was assayed for p150 TSP by immunoblotting with a mixture of antibodies Ab1544 and Ab1543 (lanes 9 -12). Membrane-bound biotinylated proteins or antibodies were detected by enhanced chemiluminescence (ECL, lanes 1, 2, 5, 6, 9, and 10). 32 P-Labeled proteins were detected by autoradiography after quenching of chemiluminescence ( 32 P, lanes 3, 4, 7, 8, 11, and 12). The apparent sizes (in kilodaltons) and positions of prestained molecular mass standards are indicated at right.
FIG. 9. Self-association of p150 TSP . p150 TSP was tagged at its carboxyl terminus with a nonapeptide influenza HA epitope. The HAtagged p150 TSP derivative was coexpressed with each of the following p150 TSP fragments in 293 cells by transient transfection: p150 (1-821) (lanes 1, 2, and 9); p150 (497-821) (lanes 3, 4, and 10); p150 (497-1173) (lanes 5, 6, and 11); and p150 (887-1173) (lanes 7, 8, and 12). Transfected cells were labeled metabolically with [ 35 S]methionine/cysteine and protein was immunoprecipitated from cell lysates with the anti-HA monoclonal antibody 12CA5 (lanes 1-8) in the absence (lanes 1, 3, 5, and 7) or presence (lanes 2, 4, 6, and 8) of an HA competitor peptide. Alternatively, protein was immunoprecipitated with the anti-p150 TSP antibody Ab1544 (lanes 9 -12). Immunoprecipitated protein was fractionated by electrophoresis through a 10% SDS-polyacrylamide gel. 35  rosine-containing peptides. Binding was abolished by excess free phosphotyrosine and by the phosphotyrosine analogue phenylphosphate (21); furthermore, a phosphotyrosine-containing peptide that binds Src-type SH2 domains with high affinity was able to compete specifically with p150 TSP for SH2 binding. Consistent with the results of specific competition experiments, binding of p150 TSP was greatly reduced by mutation of a single residue in the Blk FLI/VRES motif, Ser 147 , which is predicted on the basis of structural data to interact with phosphotyrosine (27,28). While it is possible that impairment of p150 TSP binding by free phosphotyrosine or the phosphotyrosine-containing peptide reflects an allosteric interaction between separate binding sites, the observation that the Ser 147 mutation also impairs binding makes this interpretation less likely.
We recently showed that another protein, p130 PITSLRE , also binds SH2 domains in a phosphorylation-dependent, phosphotyrosine-independent fashion (9). SH2 binding by both p150 TSP and p130 PITSLRE is mediated by an acidic region that contains multiple casein kinase II phosphorylation sites; in the case of p130 PITSLRE , phosphorylation of bacterially expressed protein by casein kinase II was sufficient to confer SH2 binding ability. While the structural basis of SH2 binding by p150 TSP and p130 PITSLRE remains to be determined, we note that several potential casein kinase II sites in the acidic regions of these proteins exhibit the amino acid sequence SEEE. Three-dimensional structures of Src and Lck SH2 domains in complex with the high-affinity peptide EPQ(pY)EEIOIYL have been determined (25,29). In these complexes, the SH2 domain makes critical contacts with glutamic acid residues at Tyr(P) ϩ1 and Tyr(P) ϩ2 . It is plausible that the SEEE sites in p150 TSP and p130 PITSLRE , when phosphorylated, mimic the high-affinity SH2-binding site (pY)EEI.
The biological significance of phosphotyrosine-independent SH2 interactions has yet to be established, and physiologic ligands of p150 TSP and p130 PITSLRE have not yet been identified. We have been unable to co-immunoprecipitate Blk and p150 TSP , and p150 TSP does not appear to be a substrate for sIgG-activated tyrosine kinases. Nonetheless, the ability of p150 TSP and p130 PITSLRE to bind SH2 domains in a phosphorylation-dependent, phosphotyrosine-independent fashion suggests that the number of proteins that interact with the classical phosphopeptide binding sites of SH2 domains may be substantially larger than appreciated.
Proteins homologous to p150 TSP can be found in other species. A search of nucleic acid and protein sequence data bases identified a putative C. elegans coding sequence specifying a protein 31% identical with p150 TSP . In its overall structure, including the arrangement of the TPR motifs and the sequence of the acidic region, the hypothetical C. elegans homologue resembles p150 TSP . In general, the homology between the nematode and mouse TPR motifs extends beyond the consensus residues; an exception is the seventh repeat, which is apparently not conserved in the nematode protein. The gene that encodes p150 TSP in the mouse was mapped to chromosome 7 between Wnt11 and Zp2; the putative C. elegans coding sequence is located on chromosome 3 at locus B0464.2 (23). No mutations in the mouse or in C. elegans have yet been mapped to those loci. A search of the dbEST data base identified a partial cDNA from the flowering plant A. thaliana which, when translated, specifies a 68-amino acid sequence with 47% identity to p150 TSP . Strikingly, an anonymous, 1045-amino acid open reading frame in the genome of S. cerevisiae (30) also exhibits significant homology to p150 TSP . The TPR-containing region of the hypothetical yeast protein is most similar to that of mouse p150 TSP in regions corresponding to the second, tenth, thirteenth, and fourteenth repeats of the mouse protein. Homology between the yeast and mouse proteins is not restricted to TPR consensus residues or to the TPR-containing region, suggesting that the yeast protein is a homologue of mouse p150 TSP and indicating an extraordinary degree of evolutionary conservation. While the function of p150 TSP in higher eukaryotes is unknown, we have found that homologous disruption of the yeast homologue is associated with mitotic chromosomal instability and temperature-sensitive defect in cell growth. 2 In recent years, it has become apparent that assembly of a diverse group of multicomponent protein complexes is mediated by a relatively small number of conserved structural modules, such as SH2 and SH3 domains, that bind specific target sites with high specificity (31). Some proteins contain multiple ligand-binding modules and apparently function as linking molecules. GRB-2, for example, which contains two SH3 domains and a single SH2 domain, functions as a bridge between transmembrane signaling complexes and SOS, a guanine nucleotide exchange factor for p21 ras (32)(33)(34)(35)(36). The presence of TPR motifs and an SH2-binding region within p150 TSP suggests that this protein may be able to mediate interactions between TPR-containing and SH2-containing proteins.