Oncogene (1998) 17, 1893 ± 1901  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc The germinal center kinase (GCK)-related protein kinases HPK1 and KHS are candidates for highly selective signal transducers of Crk family adapter proteins

Wolf Oehrl1, Christian Kardinal1, Sandra Ruf1, Knut Adermann2, John Gro€en3, Gen-Sheng Feng4, John Blenis5, Tse-Hua Tan6 and Stephan M Feller1

1Laboratory of Molecular Oncology, Institute for Medical Radiation and Cell Research (MSZ), Bavarian Julius-Maximilians University, 97078 WuÈrzburg, Germany; 2Lower Saxony Institute for Peptide Research (IPF), 30625 Hannover, Germany; 3Department of Molecular Pathology, Childrens Hospital , Los Angeles, California 90027, USA; 4Department of Biochemistry and Molecular Biology, Walther Oncology Center, Indiana University School of Medicine, Indianapolis, Indiana 46202, USA; 5Department of Cell Biology, Harvard Medical School, Boston, Massachusetts 02115, USA; 6Department of Microbiology and Immunology, Baylor College of Medicine, Houston, Texas 77030, USA

Adapter proteins function by mediating the rapid and 1992; ten Hoeve et al., 1993; reviewed in Feller et al., speci®c assembly of multi-protein complexes during the 1994a; Birge et al., 1996). SH2 domains bind to speci®c signal transduction which guards proliferation, di€er- sequence motifs on protein surfaces which usually entiation and many functions of higher eukaryotic cells. contain a phosphorylated tyrosyl residue followed by To understand their functional roles in di€erent cells it is one or several nearby residues that determine the important to identify the selectively interacting proteins binding speci®city (Birge and Hanafusa 1993). SH3 in these cells. Two novel candidates for signalling domains often recognize speci®c sequences with helical partners of Crk family adapter proteins, the hemato- structure that are rich in prolines (Williamson et al., poietic progenitor kinase 1 (HPK1) and the kinase 1994, Cohen et al., 1995). The ®rst SH3 domains of homologous to SPS1/STE20 (KHS), were found to bind Crk and Crk-like (CRKL) recognize a unique with great selectivity to the ®rst SH3 domains of c-Crk consensus motif P-x-L-P-x-K which is not compatible and CRKL. While KHS bound exclusively to Crk family with binding to most other SH3 domains (Wu et al., proteins, HPK1 also interacted with both SH3 domains 1995; Knudsen et al., 1995; Sparks et al., 1996). Much of Grb2 and weakly with Nck, but not with more than 25 of the function of these adapter proteins can be learned other SH3 domains tested. The interaction of HPK1 from the identi®cation and analysis of their direct SH2 with c-Crk and CRKL was studied in more detail. and SH3 binding partners in di€erent cell types. HPK1-binding to the ®rst SH3 domain of CRKL is Several cellular proteins which bind with great direct and occurs via proline-rich motifs in the C- selectivity to the SH3(1) domains of Crk and CRKL terminal, non-catalytic portion of HPK1. In vitro have been identi®ed, including the guanine nucleotide complexes were highly stable and in vivo complexes of exchange factor (GEF) C3G (Knudsen et al., 1994; c-Crk and CRKL with HPK1 were detectable by co- Tanaka et al., 1994) which activates the small GTPases immunoprecipitation with transiently transfected cells Rap1A and Rap1B (Gotoh et al., 1995; van den but also with endogenous proteins. Furthermore, c-Crk II Berghe et al., 1997) and the less well characterized and, to a lesser extent, CRKL were substrates for proteins DOCK180 (Hasegawa et al., 1996) and Eps15 HPK1. These results make it likely that HPK1 and (Schumacher et al., 1995). More promiscuous Crk and KHS participate in the signal transduction of Crk family CRKL SH3 binding proteins which also interact with adapter proteins in certain cell types. other small adapter proteins, namely Nck and Grb2, are the Abl family kinases c-Abl, Arg and the Keywords: Crk; CRKL; SH3; HPK1; KHS; kinase; oncogenic Bcr-Abl (Feller et al., 1994b; Ren et al., adapter 1994; Sattler et al., 1996; Wang et al., 1996), as well as the Ras-speci®c GEF-proteins SoS1 and SoS2 (Feller et al., 1995a; Okada and Pessin, 1996). It should be pointed out that these proteins contain multiple Introduction proline-rich motifs and that the SH3-mediated binding to di€erent adapter proteins does not always occur via Adapter proteins are crucial for the signal transduction a single site. of cells since they mediate the rapid and selective Crk family proteins appear to mediate basal, as well formation of multi-protein signalling complexes which as cell type speci®c signalling events. Recent work guard many aspects of cell growth, cell di€erentiation, provides intriguing evidence for an involvement of cell-cell communication and other cellular responses CRKL, C3G and Rap1 in the induction of T-cell (Cohen et al., 1995; Pawson, 1995). The small adapter anergy (Boussiotis et al., 1997), a desensitization proteins of the Crk family are largely composed of Src process of the T-cell antigen receptor which occurs homology (SH2 and SH3) domains (Reichman et al., after its stimulation without concomitant stimulation of a co-receptor (reviewed in Schwartz, 1996, 1997). c- Crk and CRKL are also candidates for signal Correspondence: SM Feller transducers in a variety of systems, including integrin Received 17 December 1997; revised 6 May 1998; accepted 6 May signalling in ®broblasts (Vuori et al., 1996) and 1998 lymphocytes (Minegishi et al., 1996), B-cell receptor Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1894 activation (Ingham et al., 1996; Smit et al., 1996), stress CRKL SH3(1) binding motifs are obvious in the kinase activation (Tanaka et al., 1997) and possibly related germinal centre kinase GCK (Katz et al., also PC12 di€erentiation (Ribon and Saltiel 1996; 1994). Interestingly, the also homologous human Torres and Bogenmann 1996) and apoptosis (Evans et kinase YSK1 (Osada et al., 1997) contains a similar al., 1997). Moreover, various growth factors and motif (PAEPVK), while the virtually identical human cytokines may signal through Crk family proteins kinase SOK1 (Pombo et al., 1996) di€ers in the (Barber et al., 1997; Husson et al., 1997; and references corresponding sequence in a position expected to be therein), although little hard evidence is available at crucial for potential SH3 binding (PADAVK). this point. Motifs with an arginine instead of a lysine (P-x-x-P- Despite the recent progress in the analysis of this x-R) often bind well to the Grb2SH3(N) domain but protein family with respect to their signalling partners, also, with lower anity, to Crk and CRKLSH3(1) their function in T-cell anergy and their pathological (Knudsen et al., 1995; Wu et al., 1996). One such motif role in Bcr-Abl-positive human CMLs (Sattler et al., 1996; Senechal et al., 1996; Tari et al., 1997), much can still be learnt about the roles of Crk family proteins in Table 1 Examples for known and putative CrkSH3(1) binding normal cellular settings. The continuing detection of sequences. o not yet tested yet unidenti®ed or uncharacterized Crk and CRKL binding proteins in di€erent cell types (Feller et al., 1995a,b; Matsuda et al., 1996; Schumacher et al., 1995; and this manuscript) demonstrate that additional functions and signalling partners for Crk and CRKL remain to be discovered. The current study reports the identi®cation of highly selective and stable interactions of adapter proteins with the hematopoietic progenitor kinase HPK1; (Hu et al., 1996; Kiefer et al., 1996) and KHS (kinase homologous to SPS1/STE20; Tung and Blenis 1997) which belong in the germinal centre kinase (GCK)- subfamily of the rapidly growing family of Ste20/ PAK65-related kinases. Both, HPK1 and KHS were previously reported to activate stress kinases of the SAPK/JNK family in transfection experiments (Hu et al., 1996; Kiefer et al., 1996; Tung and Blenis 1997). Furthermore, HPK1 has been recently implicated to mediate this activation of JNKs (Hu et al., 1996) through TAK1 (TGF-b activated kinase 1; Wang et al., 1997), a kinase which is also a candidate for the mediation of TGF-b and BMP e€ects (Yamaguchi et al., 1995). The biological functions and details of the signalling mechanisms of KHS and HPK1 are currently unknown. The interaction with c-Crk and CRKL depends largely on their ®rst SH3 domains, which recognize proline-rich motifs in the C-terminal half of the kinases. Several motifs in this region conform to the consensus P-x-L-P-x-K which was previously de®ned for selective, high anity Crk and CRKL binding (Knudsen et al., 1994, 1995; Sparks et al., 1996).

Results

Crk and CRKLSH3(1) binding proteins share a common binding consensus As listed in Table 1, the consensus motif P-x-L-P-x-K which is found in nearly all Crk and CRKL SH3(1) binding proteins and in the binding sequences detected by phage display (Knudsen et al., 1994, 1995; Sparks et al., 1996) is present in several copies in the C- terminal half of the PAK65/STE20-related kinases HPK1 (Hu et al., 1996; Kiefer et al., 1996) and in KHS (Tung and Blenis, 1997). GLK (GCK-like kinase), a very recently cloned kinase (Diener et al., 1997), which has not been analysed in this study, also contains such a motif, while no potential Crk or Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1895 (M1) is also found in HPK1. The presence of multiple various cells, including those expressing HPK1 highly potential c-Crk/CRKL SH3(1) binding motifs (MOLT4, Raji) or moderately (HL60) were compared prompted an investigation into whether HPK1 and to cells expressing the kinase KHS at high concentra- KHS can speci®cally bind and thus possibly mediate tions (HEK293, COS7; Figure 4a). Even upon long signalling events of Crk family adapter proteins. exposure (Figure 2a) no signal beyond background was visible in HEK293 and COS7 lysates, indicative of the speci®city of the HPK1 antiserum. Subsequently, total Crk and CRKL are expressed ubiquitously RIPA lysates from MOLT4 and Raji cells were Initially, MOLT4 T-cells and Raji B-cells which are precipitated with an anity puri®ed GST-fusion known to highly express HPK1 (Hu et al., 1996) were protein containing both SH3 domains of CRKL compared to other cell lines of diverse origins, primary (CRKLDSH2), or with GST alone. The precipitates ®broblasts and whole mouse embryo extracts by were washed with bu€er containing a moderate Western blot for expression of c-Crk and CRKL. concentration of detergent and then analysed by The CRKL protein is expressed in these cells in a Western blot with anti-HPK1. HPK1 bound efficiently concentration similar to that of many other cell types to GST-CRKLDSH2 but not GST alone (Figure 2b). (Figure 1). c-Crk II expression is easily detectable in To test the binding stability and to determine if the Raji cells and less prominent but detectable in MOLT4 SH3(1) domains of c-Crk and CRKL are sucient for cells. The shorter CRKL-immunoreactive band de- binding, GST-SH3 fusion proteins containing the ®rst tected in many cells after RIPA-extraction is likely a SH3 domains of c-Crk or CRKL, respectively, were proteolytic fragment of CRKL because it was greatly incubated with proteins from total MOLT4 cell lysate reduced by direct SDS-lysis of several cell lines tested or a cytosolic MOLT4 protein fraction (S100) and then (not shown). washed with high stringency bu€er (RIPA). Western blotting with HPK1-antiserum con®rmed that HPK1 HPK1 binds stably to the ®rst SH3 domains of c-Crk, CRKL and Grb2 To ensure speci®city, the HPK1 antiserum was tested for cross-reactivity with the later cloned, homologous kinase KHS (Tung and Blenis 1997). Total lysates from

Figure 1 The adapter proteins c-Crk and CRKL are widely expressed in hematopoietic and tissue-derived cells. 100 mgof total cell RIPA-protein extracts were separated by 11% SDS ± PAGE and probed with anti-Crk or anti-CRKL. The following Figure 2 HPK1 binds to the SH3(1) domain of Crk and CRKL cells and lysates were used: P19 (mouse embryo carcinoma), CEF under high stringency conditions. (a)50mg of total RIPA cell (chicken embryo ®broblast); mur. embryo (whole E16 mouse lysates of di€erent cell lines (cell line origin mostly described in embryo), 32Dcl2 (mouse granuloytic precursor cells), BaF3 the legend of Figure 1; FDCP Mix A1: mouse multipotent (mouse pro-B-cells), Raji (hu, Burkitt's lymphoma), Jurkat (hu. hematopoietic precursor cell; Ford et al., 1992) were separated by acute T-cell leukaemia), Namalwa (hu, Burkitt's lymphoma), 8% SDS ± PAGE and blotted with polyclonal anti-HPK1 (Hu et MOLT-4 (hu. T-cell lymphoblastic leukaemia), HuT78 (hu. T-cell al., 1996). 1 mg of HL60 total cell RIPA lysate was used for the lymphoma), NB4 (hu. acute promyelocytic leukaemia), HL60 (hu. immunoprecipitation with anti-HPK1 or a preimmune serum granulocyte precursor from acute promelocytic leukaemia), U937 included in the same blot. (b)10mg of GST or an equimolar (hu. monocytic leukaemia), K562 (hu. CML cells from blast amount of GST-CRKLDSH2 immobilized on GSH-Sepharose crisis), HEL (human erythroleukaemia), HeLa (hu. cervical were incubated with 1 mg of MOLT4 total cell RIPA lysate, carcinoma), SH-SY5Y (hu. neuroblastoma), HepG2 (hu. liver washed as and analysed by Western blot as indicated. (c) 250 mg carcinoma), HEK293 (hu. embryonic kidney), COS-7 (African of MOLT4 cystolic protein (S100) or Triton-X 100-extracted total green monkey kidney cells, ®broblast-like), 3Y1 (rat ®broblasts), cell proteins were incubated with 10 mg of GST or equimolar SR3Y1 (p60 v-Src transformed 3Y1), 3T3 (mouse NIH3T3 amounts of GST-SH3 fusion proteins, washed with RIPA and ®broblasts), 3T3 Bcr-Abl (p210 Bcr-Abl transformed NIH3T3), blotted as before. 50 mg of total MOLT4 protein were used as PC12 (rat phaeochromocytoma) positive control Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1896 can stably bind to the SH3(1) domains of Crk and KHS binds with great selectivity to the SH3(1) domains CRKL (Figure 2c). of c-Crk and CRKL The SH3 domain binding preference of HPK1 was then analysed with a large panel of GST-fusion The SH3 binding properties and subcellular localiza- proteins containing single or multiple SH3 domains. tion of the HPK1-related kinase KHS were studied in Ecient binding of HPK1 was found with all comparison. KHS is endogenously expressed in simian constructs containing the ®rst SH3 domains of c- COS7 and human HEK293 cells (Figure 4a), which do Crk, CRKL or Grb2, respectively (Figure 3). Less not contain detectable amounts of HPK1 (Figure 2a). ecient binding was observed with a fusion protein It binds RIPA-bu€er stable and with great selectivity containing all domains of Nck and even weaker to the ®rst SH3 domains of c-Crk and CRKL (Figure binding was seen with the C-terminal SH3 domain 4b). Only a very weak interaction beyond background of Grb2 and the HckSH3 domain. Longer exposure of the blot suggested that the binding to Nck occurs primarily through the NckSH3(2) domain, which is also known to bind the SoS and Abl proteins (data not shown). Subcellular fractionation of human T-cells (MOLT4, HuT78) into P1 (nuclei with attached cytoskeletal structures and membranes), P10 (mito- chondria and granules), P100 (small vesicles, poly- somes) and S100 (cytosol) followed by anti-HPK1 Western blot revealed that HPK1 is largely found in the cytosol (S100) but is also detectable in the P1 fraction and, to a lesser degree, in the P100 fraction (WO, unpublished data). This subcellular distribution is similar to that of its potential binding partners c- Crk, CRKL and Grb2.

Figure 4 KHS di€ers from HPK1 in its expression pattern, its adapter protein binding selectivity and its subcellular localization. (a) 150 mg of total RIPA cell lysate of the indicated cell lines were separated by 8% SDS ± PAGE and blotted with anti-KHS. (b) 3 mg of HEK293 total RIPA lysate were incubated with 10 mgof GST or equimolar amounts of the fusion proteins indicated which were immobilized on GSH-sepharose overnight. Samples were washed with RIPA bu€er, precipitated proteins separated by 8% SDS ± PAGE and blotted with KHS-antiserum. 100 mgof HEK293 total RIPA lysate were loaded as positive control. (c) HEK293 cells were fractionated as described in Materials and methods and analysed for the subcellular distribution of the KHS protein (left panel) and its potential to bind to the CrkSH3(1) domain (right panels). 100 mg of protein extract from each Figure 3 HPK1 binds eciently and selectively to the ®rst SH3 fraction or total RIPA lysate were separated by 8% SDS ± PAGE domains of c-Crk, CRKL and Grb2. 250 mg of MOLT4 RIPA- and blotted with anti-KHS for the analysis of the subcellular extracted total cell proteins were incubated with 10 mg of GST or distribution of KHS. 1 mg of protein extract as indicated and equimolar amounts of GST-SH3 constructs, washed with RIPA 20 mg of GST-CrkSH3(1) immobilized on GSH-beads were used and blotted as before. 50 mg of total MOLT4 RIPA-extracted for the precipitation analysis. Samples were washed with RIPA protein were used as positive control and analysed for the presence of KHS as before Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1897 (GST-tag alone) was found with the SH3 domains of 4c, right panels), possibly indicating conformational adapter proteins like Grb2, the Grb2-related GRAP di€erences or di€erences in protein modi®cation (Feng et al., 1996) or Nck, as well as other signalling between these protein pools. proteins after extended blot exposure (data not shown). KHS is contained in substantial amounts in the HPK1 forms complexes with c-Crk and CRKL in cells particulate (P) fractions after subcellular fractionation of HEK293 (Figure 4c, left panel). Importantly, this To determine if stable complexes of Crk family adapters `particulate' KHS-pool is eciently precipitated with and HPK1 are detectable in cells, FLAG-tagged HPK1 GST-c-CrkSH3(1) while the KHS-protein from cyto- (kinase inactive mutant M46; Hu et al., 1996) was solic S100 fraction is very poorly precipitated (Figure transiently co-transfected with CRKL or c-Crk II, respectively, into COS7 cells. Precipitation with anti- FLAG mAb but not a control mAb followed by Western blotting with anti-CRKL or anti-c-Crk demonstrated the existence of stable HPK1(M46)-CRKL and -c-Crk- complexes (Figure 5a). Lysates from COS7 cells co- transfected with CRKL or c-Crk II and pCI-neo control vector did show precipitation of CRKL with anti-FLAG mAb as determined by Western blot. In addition, a Crk SH2 mutant but not an SH3(1) mutant protein could stably bind to HPK1(M46) (data not shown). Complexes of endogenously expressed c-Crk with HPK1 were detectable in lysates of MOLT4 cells by co-immunoprecipitation with an anti-Crk monoclonal antibody (Figure 5b). However, this interaction was much weaker than the binding to the kinase inactive mutant protein FLAG-HPK1(M46) transfected into COS cells. Since c-Crk II is a substrate of HPK1 the inactive HPK1 may bind more stably to Crk than the catalytically active enzyme. Endogenously expressed CRKL could not be analysed since signi®cant amounts of HPK1 bound nonspeci®cally to several rabbit preimmune sera tested (not shown) and a monoclonal CRKL antibody was not available.

Direct binding of HPK1 to CRKL via proline-rich motifs The HPK1 interaction with Crk and CRKL was analysed in further detail. To con®rm a direct binding of HPK1 to CRKL, the kinase domain and the non- catalytic portion of HPK1 were expressed separately as GST-fusion proteins, anity puri®ed and equal amounts immobilized on a membrane following SDS ± PAGE. After blocking and renaturation, the membrane was probed with 35S-labelled GST- CRKLDSH2. As expected, the C-terminal region (CD) but not the kinase domain (KD) of HPK1 was detectable with this probe (Figure 6a). While the GST-HPK1-KD-protein appeared largely at the expected size in Coomassie-stained gels, much of the GST-HPK1-CD protein (HPK1 amino acids 292 ± 833) was truncated in six di€erent bacterial strains tested, although protein corresponding to the expected Figure 5 c-Crk and CRKL form complexes with HPK1 in cells. size of the full length fusion protein was also present. (a)10mg of pCIneo-FLAG-HPK1(M46) (Hu et al., 1996) and As expectable from its size, the truncated fragments 8 mg of pSG5-CRKL (ten Hoeve et al., 1994) or 20 mgof pMEXneo-c-Crk II (Fajardo et al., 1993), respectively, were indicated in the ®gure apparently still contained the transfected into COS7 cells grown to 70% con¯uency per 10 cm essential proline-rich motifs (clustered between amino é culture dish. Cells were harvested after 36 ± 48 h, lysed in bu€er acids 300 and 480) and were therefore eciently containing 1% Triton-X100 and precipitated with 4 mg of anti- recognized by the 35S-labelled GST-CRKLDSH2 FLAG mAb or control mAb. Precipitated protein was washed three times with lysis bu€er, separated by 11% SDS ± PAGE and probe. analysed for the presence of c-Crk or CRKL as indicated by Of the four motifs (M1 to M4) from human HPK1 Western blot. (c) MOLT4 cells were lysed with 1% Triton-X100 which are listed in Table 1, hu. HPK1-M1 is a good containing bu€er and 2 mg of extracted protein was precipitated candidate for binding to the Grb2SH3(N) domain, with 1 mg of anti-Crk mAb or control mAb. The precipitates were while hu. HPK1-M2 (sequence fully conserved between washed three times with lysis bu€er, separated by 8% SDS ± PAGE, blotted and probed with anti-HPK1. 30 mg of total mouse and human) is expected to be the primary site MOLT4 lysate were also loaded on the gel for Crk/CRKLSH3(1) binding. In contrast, the hu. Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1898 HPK1-M3 sequence is only partially conserved when Phosphorylation of Crk by HPK1 compared to the corresponding region in the murine protein but may still serve as a docking site for the It was then analysed whether HPK1 can phosphorylate Crk/CRKL SH3(1) domains in both species. c-Crk II or CRKL in vitro, using GST-fusion proteins Finally, the hu. HPK1-M4 sequence is not of c-Crk II and CRKL as substrates. No phosphoryla- conserved in the mouse protein and is therefore not tion of the GST-tag alone was detectable with likely essential for the interaction with Crk/CRKL. endogenous HPK1 precipitated from MOLT4 cells Synthetic peptides corresponding to the M2 and M3 (Figure 7), but GST-c-Crk II was eciently phos- regions of human HPK1 were compared to similar phorylated. A much weaker phosphorylation of GST- peptides from known Crk/CRKLSH3(1) binding CRKL was observed. proteins and a high anity CrkSH3(1) binding peptide derived from an extensive mutagenesis study CRKL activates the stress kinase JNK1 but has little (Posern et al., 1998) in a competition assay. In this e€ect on JNK1 activation by HPK1 assay, a GST-fusion protein containing the high anity CrkSH3(1) binding peptide (GST-HACBP) is It has been previously reported that HPK1 can activate immobilized on GSH-Sepharose beads and is then the JNK/SAPK stress response kinases in COS cells used to precipitate CRKL protein from lysates of (Hu et al., 1996; Kiefer et al., 1996). Therefore a K562 cells. This interaction can be completed through possible e€ect of CRKL on JNK1-activation by HPK1 the addition of soluble peptide which binds to the was investigated in transiently transfected COS7 cells. CRKLSH3(1) domain. In this assay, the M2 and the Expression of CRKL alone resulted in a slight, but M3 peptides can both signi®cantly reduce CRKL reproducible activation of JNK1 (Figure 8), reminis- binding to GST-HACBP (Figure 6b). The M2 peptide cent of data recently reported for v-Crk (Tanaka et al., is more ecient than M3, and its eciency at least as 1997). Expression of HPK1 alone was, as reported good as the best binding (CB-2) peptide of DOCK180, previously, sucient to strongly activate JNK1, but a well established Crk/CRKLSH3(1) binding protein even massive overexpression of CRKL did not seem to of yet unde®ned function (Hasegawa et al., 1996). modulate this e€ect (Figure 8).

Figure 6 HPK1 interacts with the CRKL SH3 domains via proline-rich motifs in the non-catalytic region. (a) Far Western blot with 35S-labelled GST-CRKLDSH2 (10 mg/ml) on 300 ng of membrane bound, anity-puri®ed GST-fusion proteins of the HPK1 kinase domain (KD, amino acids 1 ± 291 or the noncatalytic region (CD, aa 292 ± 833). (b) A peptide corresponding to amino acids 393 ± 403 in human HPK1 (Table 1; HPK1-M2) can block binding of CRKL protein from K562 cells to a high anity CRKLSH3(1)-interacting GST-fusion peptide (GST-HACBP) in a solution phase competition assay. 10 mg of GST-HACBP were immobilized on GSH-sepharose for 2 h in the presence of 500 mg inactivated, soluble E. coli lysate (see Posern et al., 1998 for details of preparation), which was added to saturate non-speci®c protein binding. The fusion peptide was then incubated with 900 mgof K562 total RIPA lysate in the presence or absence of peptides as indicated overnight. The ®nal total volume of the sample was 300 ml and the ®nal peptide concentration 100 mM. Samples were washed thrice with RIPA and precipitated CRKL protein was detected by anti-CRKL Western blot after SDS ± PAGE Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1899

Figure 7 c-Crk II is an in vitro substrate for HPK1. 300 mg of MOLT4 total cell Triton-X 100 (TX) lysate was precipitated with polyclonal anti-HPK1 or normal rabbit serum (NRS), washed twice with 1% TX lysis bu€er and once with kinase bu€er. 10 mgof GST, 20 mg of GST-c-Crk II, GST-CRKL or no exogenous substrate was added as indicated and the kinase reaction was carried out for 30 min at 258C. Samples were separated by SDS ± PAGE as indicated and gels exposed to X-ray ®lm overnight. Autophosphorylated HPK1, and the migration of the added substrate proteins are indicated by closed arrowheads. The open arrowheads indicate a prominent additional band of approximately 60 kDa which is detected only in anti-HPK1 precipitates, possibly an HPK1-fragment or a co-precipitating protein of unknown identity

Discussion important interactions of Crk and CRKL with receptor proteins is still weak. In addition, the The data currently available indicate that HPK1 is interaction of HPK1 with Grb2, which has not been mainly expressed in hematopoietic cells where the the focus of the current study, indicates that Grb2 may adapter proteins Crk and CRKL have been suggested couple HPK1 to a variety of signalling proteins to be involved in various signal transduction systems, including various receptors or their signalling part- including B-cell receptor signalling, T-cell receptor- ners. A report which analysed the Grb2 ± HPK1 induced anergy and leukocyte-integrin signalling. The interaction in more detail was published while this selective and stable interactions of HPK1 with c-Crk/ manuscript was being prepared (Ana® et al., 1997). CRKL and Grb2 reported here suggest that these Cell lines of the myeloid-monocytic origin analysed adapter proteins are functionally linked to HPK1. so far (HL60, U937, K562, NB4) appear to express low Since HPK1 is expressed in early progenitor cells, it (but upon prolonged exposure detectable) levels of could also be involved in the regulation of HPK1. It is currently unclear if these cell lines di€erentiation processes in the hematopietic system. accurately re¯ect the protein expression in native The related KHS kinase seems at present expressed in myeloid cells. Since the Bcr-Abl oncoprotein has been a quite di€erent pattern of cell types and to di€er reported to activate the JNK-pathway (Raitano et al., also in the subcellular localization of the protein pool 1995) and CRKL is a well documented binding protein which is capable to bind to the Crk family adapters of Bcr-Abl, HPK1 bound to Bcr-Abl via CRKL could eciently. Furthermore, KHS does not bind to the play a role in mediating JNK-activation by Bcr-Abl. It SH3 domains of Nck, Grb2 or the Grb2-related is also possible that JNK-activation by Crk or CRKL protein GRAP. can be mediated by HPK1 in normal hematopoietic The functional roles of the interactions of HPK1 cells. One candidate stimulus for this is TGF-b (Wang and KHS with small adapter proteins c-Crk, CRKL et al., 1997). In COS7 cells which lack detectable and Grb2 remain to be studied. One possibility is the amounts of endogenous HPK1, the observed JNK translocation of HPK1 to appropriate sites of action activation after forced expression of HPK1 may be through c-Crk or CRKL after the tyrosine phosphor- mediated by the signalling system normally utilized by ylation of relevant docking proteins like p120 c-Cbl, the endogenously present, HPK1-homologous kinase members of the growing p130Cas-protein family or other KHS. large docking proteins (IRS1, IRS2 etc.). A direct Future studies will aim to reveal which c-Crk- and association of HPK1 with receptor proteins is also CRKL-dependent signalling systems are regulated by possible, although the evidence for functionally HPK1 and KHS. Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1900 the same bu€er containing only 1% Triton X-100 to better preserve protein complexes. The subcellular fractionation of HuT78 and HEK293 cells was done according to a published protocol (Chen et al., 1992). E 16 mouse embryos were dissected from the uterus, chilled on ice and then lysed as a whole in RIPA bu€er. Most antisera and antibodies used for precipitations and Western blots were from commercial sources (anti-CRKL: sc-319, Santa Cruz; anti-c-Crk: C12620, Transduction Laboratories; anti-FLAG: M2, East- man Kodak; anti-HA: 12CA5, Boehringer Mannheim; anti- JNK1: sc-474, Santa Cruz) or have been previously described (anti-HPK1: Hu et al., 1996; anti-KHS: Tung and Blenis 1997). Unlabelled GST-fusion proteins were anity puri®ed, extensively dialysed against 5 mM Tris HCl pH 7.5 and analysed for quality by SDS ± PAGE and Coomassie-staining. Proteolytic fragmentation was minimized by comparing expression in di€erent host strains. Many GST-fusion protein constructs have been described elsewhere (Feller et al., 1994b, 1995a,b; Knudsen et al., 1994, 1995; Posern et al., 1998; Feng et al., 1996). The GST-fusion protein of the kinase domain of HPK1 (GST-HPK1-KD amino acids 1 ± 291) and the noncatalytic region of HPK1 (GST-HPK1-CD amino acids 292 ± 833) were obtained by cloning PCR- ampli®ed fragments (Hu et al., 1996) into a pGEX-vector. The GST-c-Jun (5 ± 89) construct was kindly provided by A Avots and E Ser¯ing (University WuÈ rzburg). The metabolic 35S-labelling of GST-fusion proteins and far Western blot technique has been described in detail elsewhere (Feller et al., Figure 8 CRKL overexpression can activate the stress kinase 1995b). JNK1 independently of HPK1 but does not appear to modulate Peptides were synthesized by the Fmoc (fluorenylmethox- the HPK1-mediated activation of JNK1. Cos7 cells were ycarbonyl)/t-butyl based solid phase peptide chemistry transiently transfected with 8 mg of pSRa3-3xHA-JNK1-DNA, method, and puri®ed to over 95% homogeneity by C - 10 mg of pSG5-CRKL, and 12 mg of pCIneo-FLAG-HPK1 per 18 10 cm é culture dish when indicated, or the respective control reversed phase HPLC. The puri®ed peptides were further vectors. The cells were lysed in TX-lysis bu€er and 250 mg of the analysed for accurate mass by mass spectrometry. The lysate from the transfected cells were precipitated with 1 mgof peptide competition assay was carried out as previously anti-HA mAb immobilized on beads for 2 h to precipitate HA- described in detail (Knudsen et al., 1995; Posern et al., 1998). tagged JNK1. The beads were then washed twice with 1% TX- Transfections of COS7 cells were done with the calcium lysis bu€er and once with kinase bu€er. 1/5 of the precipitate was chloride precipitation technique. The active DNAs were removed and used for Western blot detection of HA-JNK1 with compared with equal amounts of the appropriate control anti-JNK1 (middle panel). The remaining precipitated HA-JNK1 vectors. Protein expression levels were always controlled by was analysed for JNK1 kinase activity using 10 mg GST-c-Jun(5 ± Western blots. The mammalian expression vectors and the 89) as substrate (upper panel). The lower panel shows an anti- CRKL Western blot after SDS ± PAGE of 70 mg total cell lysate control vectors have been previously described: pCI-neo from transfected COS7 cells to document the accomplished (Promega), FLAG-tagged pCI-neo-HPK1(wt) and pCI-neo- overexpression of CRKL HPK1(M46) (Hu et al., 1996); pSG5 (Stratagene), pSG5- CRKL (ten Hoeve et al., 1994), pMEXneo and pMEXneoc-Crk II (Fajardo et al., 1993); pSRa3 and pSRa3-3xHA-JNK1 (gift of Audrey Minden, Columbia University, New York; Minden Materials and methods et al., 1995). Kinase assays for HPK1 and JNK1 were carried out as essentially described (Hu et al., 1996) but using the Many cell lines were obtained from the ATCC and mAb 12CA5 for precipitation of HA-tagged JNK1 and GST- maintained under the conditions speci®ed by the supplier in c-Jun (5 ± 89) or GST-c-Crk II (full length) and GST-CRKL the presence of antibiotics, or as previously described (Feller (full length) as substrates. The kinase bu€er contained 20 mM et al., 1994b, 1995a,b; Posern et al., 1998). Mitotic HeLa cells HEPES pH 7.5, 100 mM NaCl, 0.1% Tween 20, 1 mM dithiothreitol, 5 mM MnCl ,5mM MgCl ,20mM b- were generated by treating growing cells with 5 mg/ml 2 2 thymidine (Sigma) for 16 h then washing cells three times glycerophate, 5 mM NaF and 500 nM unlabelled ATP. Three 32 with PBS, releasing the cells into regular medium for 6 h, mCi P-g-ATP (3000 Ci/mmol) were used per reaction. followed by the addition of 0.04 mg/ml Nocodazole (Sigma) for 10 h. Mitotic cells (approximately 70% of total) were collected by blowing them gently o€ the plate with medium from a pipette. PC12 cells were di€erentiated by the addition of 50 ng/ml of nerve growth factor NGF (2.5S, Gibco ± BRL) added fresh daily. Acknowledgements Chicken embryo ®broblasts were prepared from chicken Many colleagues have very generously provided reagents as embryos following standard protocols. Cells were usually detailed in Materials and methods. We wish to thank lysed in RIPA 100 bu€er containing 20 mM TrisHCl pH 7.5, Sylvia PfraÈ nger for the art work, Guido Posern for

1mM Na2EDTA, 100 mM NaCl, 0.1% (w/v) SDS, 0.5% comments on the manuscript and Kerstin MuÈ ller for (w/v) sodium deoxycholate and 1% (v/v) Triton X-100, technical assistance. This work was supported by grants 100 mM sodium-orthovanadate, 100 mMM sodium molybdate, of the Deutsche Forschungsgemeinschaft (DFG-FE439/1 & and the following protease inhibitors: 0.5 mg/ml Leupeptin, SFB465) and the Wilhelm-Sander-Stiftung to SF and by 5 mg/ml Antipain, 0.7 mg/ml Pepstatin, 10 mg/ml Aprotinin NIH grants RO1-GM49875 and RO1-AI38649 to THT. and 200 mg/ml PMSF. When indicated, cells were lysed with WO was in part supported by an A VoÈ lsch fellowship. Crk and CRKL interactions with HPK1 and KHS WOehrlet al 1901 References

Ana® M, Kiefer F, Gish GD, Mbamalu G, Iscove NN, and Okada S and Pessin JE. (1996). J. Biol. Chem., 271, 25533 ± Pawson T. (1997). J. Biol. Chem., 272, 27804 ± 27811. 25538. Barber DL, Mason JM, Fukazawa T, Reedquist KA, Druker Osada S, Izawa M, Saito R, Mizuno K, Suzki A, Hirai SI and BJ, Band H and D'Andrea AD. (1997). Blood, 89, 3166 ± Ohno S. (1997). Oncogene 14, 2047 ± 2057. 3174. Pawson T. (1995). Nature 373, 573 ± 580. Birge RB and Hanafusa H. (1993). Science, 262, 1522 ± 1524. Pombo CM, Bonventre JV, Molnar A, Kyriakis J, Force T. Birge R B, Knudsen BS, Besser D and Hanafusa H. (1996). (1996). Embo J., 15, 4537 ± 4546. Genes to Cells, 1, 595 ± 613. PosernG,ZhengJ,KnudsenBS,KardinalC,MuÈ ller KB, Boussiotis VA, Freeman G J, Berezovskaya A, Barber DL Voss J, Shishido T, Cowburn D, Cheng G, Wang B, Kruh and Nadler LM. (1997). Science, 278, 124 128. GD, Burrell SK, Jacobson CA, Lenz DA, Zamborelli TJ, Chen R-H, Sarnecki C and Blenis J. (1992). Mol. Cell. Biol., Adermann K, Hanafusa H and Feller SM. (1998). 12, 915 Ð 927. Oncogene, 16, 1903 ± 1912. Cohen GB, Ren R and Baltimore D. (1995). Cell, 80, 237 Ð Raitano AB, Halpern JR, Hambuch TM and Sawyers CL. 248. (1995). Proc. Natl. Acad. Sci. USA., 92, 11746 ± 11750. Diener K, Wang XS, Chen C, Meyer CF, Keesler G, Zukwski Reichman CT, Mayer BJ, Keshav S and Hanafusa H. (1992). M, Tan T-H and Yao Z. (1997). Proc. Natl. Acad. Sci. Cell. Growth. Di€er., 3, 451 ± 460. USA., 94, 9687 ± 9692. Ren R, Ye ZS and Baltimore D. (1994). Genes. Dev., 8, 783 ± Evans EK, Lu W, Strum SL, Mayer BJ and Kornbluth S. 795. (1997). EMBO. J., 16, 230 ± 241. Ribon V and Saltiel AR. (1996). J. Biol. Chem., 271, 7375 ± Fajardo JE, Birge RB and Hanafusa H. (1993). Mol. Cell. 7380. Biol., 13, 7295 ± 7302. Sattler M, Salgia R, Okuda K, Uemura N, Durstin MA, Feng GS, Ouyang YB, Hu DP, Shi ZQ, Gentz R and Ni J. Pisick E, Xu G, Li JL, Prasad KV and Grin JD. (1996). (1996). J. Biol. Chem., 271, 12129 Ð 12132. Oncogene, 12, 839 ± 846. Feller SM, Ren R, Hanafusa H and Baltimore D. (1994a). Schumacher C, Knudsen BS, Ohuchi T, Di Fiore PP, Trends Biochem. Sci., 19, 453 Ð 458. Glassman RH and Hanafusa H. (1995). J. Biol. Chem., Feller SM, Knudsen BS and Hanafusa H. (1994b). EMBO J., 270, 15341 ± 15347. 13, 2341 ± 2351. Schwartz RH. (1996). J. Exp. Med., 184, 1Ð8. Feller SM, Knudsen BS and Hanafusa H. (1995a). Oncogene, Schwartz RH. (1997). Curr. Opin. Immunol., 9, 351 ± 357. 10, 1465 ± 1473. Senechal K, Halpern J and Sawyers CL. (1996). J. Biol. Feller SM, Knudsen BS, Wong TW and Hanafusa H. Chem., 271, 23255 ± 23261. (1995b). Methods Enzymol., 255, 369 ± 378. Smit L, van der Horst G and Borst J. (1996). Oncogene 13, Ford AM, Healy LE, Bennett CA, Navarro E, Spooncer E 381 ± 389. and Greaves MF. (1992). Blood, 15, 1962 ± 1971. Sparks AB, Rider JE, Ho€man NG, Fowlkes DM, Quillam Gotoh T, Hattori S, Nakamura S, Kitayama H, Noda M, LA and Kay BK. (1996). Proc. Natl. Acad. Sci. USA., 93, TakaiY,KaibuchiK,MatsuiH,HataseO,TakahashiH, 1540 ± 1544. Kurata T and Matsuda M. (1995). Mol. Cell. Biol., 15, Tanaka S, Morishita T, Hashimoto Y, Hattori S, Nakamura 6746 ± 6753. S, Shibuya M, Matuoka K, Takenawa T, Kurata T, Hasegawa H, Kiyokawa E, Tanaka S, Nagashima K, Gotoh Nagashima K and Matsuda M. (1994). Proc. Natl. Acad. N, Shibuya M, Kurata T and Matsuda M. (1996). Mol. Sci. USA., 91, 3443 ± 3447. Cell. Biol., 16, 1770 ± 1776. Tanaka S, Ouchi T and Hanafusa H. (1997). Proc. Natl. Hu MCT, Qui WR, Wang X, Meyer CF and Tan TH. (1996). Acad.Sci.USA.,94, 2356 ± 2361. Genes Dev., 10, 2251 ± 2264. Tari AM, Arlinghaus R and Lopez-Berestein G. (1997). Husson H, Mograbi B, Schmid-Antomarchi, Fischer S and Biochem. Biophys. Res. Commun., 235, 383 ± 388. Rossi B. (1997). Oncogene, 14, 2331 ± 2338. ten Hoeve J, Morris C, Heisterkamp N and Gro€en J. (1993). Ingham RJ, Krebs DL, Barbazuk SM, Turck CW, Hirai H, Oncogene, 8, 2469 ± 2474. Matsuda M and Gold MR. (1996). J. Biol. Chem., 271, ten Hoeve J, Kaartinen V, Fioretos T, Haatja L, Voncken W, 32306 ± 32314. Heisterkamp N and Gro€en J. (1994). Cancer Res., 54, Katz P, Whalen G and Kehrl JH. (1994). J. Biol. Chem., 269, 2563 ± 2567. 16802 ± 16809. Torres M and Bogenmann E. (1996). Oncogene, 12, 77 ± 86. Kiefer F, Tibbles LA, Ana® M, Janssen A, Zanke BW, Tung RM and Blenis J. (1997). Oncogene, 14, 653 ± 659. Lassam N, Pawson T, Woodgett JR and Iscove NN. van den Berghe N, Cool RH, Horn G and Wittinghofer A. (1996). EMBO J., 15, 7013 ± 7025. (1997). Oncogene, 15, 845 ± 850. Knudsen BS, Feller SM and Hanafusa H. (1994). J. Biol. VuoriK,HiraiH,AizawaSandRuoslahtiE.(1996).Molec. Chem., 269, 32781 ± 32787. Cell. Biol., 16, 2606 ± 2613. Knudsen BS, Zheng J, Feller SM, Mayer JP, Burrell SK, Wang B, Mysliwiec T, Feller SM, Knudsen B, Hanafusa H Cowburn D and Hanafusa H. (1995). EMBO J., 14, 2191 ± and Kruh GD. (1996). Oncogene, 13, 1379 ± 1385. 2198. Wang W, Zhou G, Hu MCT, Yao Z and Tan TH. (1997). J. Matsuda M, Ota S, Tanimura R, Nakamura H, Matuoka K, Biol. Chem., 272, 22771 ± 22775. Takenawa T, Nagashima K and Kurata T. (1996). J. Biol. Williamson MP. (1994). Biochem. J., 297, 249 ± 260. Chem., 271, 14468 ± 14472. WuX,KnudsenB,FellerSM,ZhengJ,SaliA,CowburnD, Minden A, Lin A, Claret FX, Abo A and Karin M. (1995). Hanafusa H and Kuriyan J. (1995). Structure, 3, 215 ± 226. Cell, 81, 1147 ± 1157. Yamaguchi K, Shirakabe K, Shibuya H, Irie K, Oishi I, Minegishi M, Tachibana K, Sato T, Iwata S, Nojima Y and Ueno N, Taniguchi T, Nishida E and Matsumoto K. Morimoto C. (1996). J. Exp. Med., 184, 1365 ± 1375. (1995). Science, 270, 2008 ± 2011.