This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Arnold, R. Articles by Kiefer, F. Search for Related Content PubMed PubMed Citation Articles by Arnold, R. Articles by Kiefer, F.

 Previous Article  |  Next Article 

Molecular and Cellular Biology, March 2005, p. 2364-2383, Vol. 25, No. 6
0270-7306/05/$08.00+0     doi:10.1128/MCB.25.6.2364-2383.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Activation of Hematopoietic Progenitor Kinase 1 Involves Relocation, Autophosphorylation, and Transphosphorylation by Protein Kinase D1

Max Planck Institute for Molecular Biomedicine, Münster,¹ Tumor Immunology Program, German Cancer Research Center (DKFZ), Heidelberg,² Max Planck Institute for Physiological and Clinical Research, Bad Nauheim Germany,³ Division of Biochemistry, Katholieke Universiteit Leuven, Leuven, Belgium,⁴ IRCM, Montreal, Quebec, Canada⁵

Received 14 June 2004/ Returned for modification 23 July 2004/ Accepted 16 December 2004

	ABSTRACT

Top Abstract Introduction Materials and Methods Results Discussion References

 
Adaptive immune signaling can be coupled to stress-activated protein kinase (SAPK)/c-Jun N-terminal kinase (JNK) and NF-B activation by the hematopoietic progenitor kinase 1 (HPK1), a mammalian hematopoiesis-specific Ste20 kinase. To gain insight into the regulation of leukocyte signal transduction, we investigated the molecular details of HPK1 activation. Here we demonstrate the capacity of the Src family kinase Lck and the SLP-76 family adaptor protein Clnk (cytokine-dependent hematopoietic cell linker) to induce HPK1 tyrosine phosphorylation and relocation to the plasma membrane, which in lymphocytes results in recruitment of HPK1 to the contact site of antigen-presenting cell (APC)-T-cell conjugates. Relocation and clustering of HPK1 cause its enzymatic activation, which is accompanied by phosphorylation of regulatory sites in the HPK1 kinase activation loop. We show that full activation of HPK1 is dependent on autophosphorylation of threonine 165 and phosphorylation of serine 171, which is a target site for protein kinase D (PKD) in vitro. Upon T-cell receptor stimulation, PKD robustly augments HPK1 kinase activity in Jurkat T cells and enhances HPK1-driven SAPK/JNK and NF-B activation; conversely, antisense down-regulation of PKD results in reduced HPK1 activity. Thus, activation of major lymphocyte signaling pathways via HPK1 involves (i) relocation, (ii) autophosphorylation, and (iii) transphosphorylation of HPK1 by PKD.

	INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

 
Innate and adaptive immune functions are governed by instructive signals from receptors that sense pathogens and toxins. Ligation of lymphocyte antigen receptors triggers a series of characteristic events that is initiated by activation of Src family tyrosine kinases followed by recruitment and activation of Syk and BTK family nonreceptor tyrosine kinases (18). The ensuing tyrosine phosphorylation provides docking sites for SH2 and PTB domains on adaptor proteins and enzymes, which assemble into a spatially highly defined complex that will ultimately cause lymphocyte activation. For instance, in T cells, the adaptor proteins SLP-76 and Gads (Grb2-related adaptor downstream of Shc) are recruited to the transmembranous adaptor LAT, forming a central scaffold for the activation of several effector molecules, such as phospholipase C (PLC), Ras, and Rho GTPases (15). These events are followed by the activation of a number of different serine/threonine kinases, which include members of the protein kinase B (PKB), PKC, and PKD families as well as mitogen-activated protein 3 (MAP3) and MAP4 kinases, such as mitogen-activated protein kinase kinase kinase 1 (MEKK1) and hematopoietic progenitor kinase 1 (HPK1). Further downstream, mitogen-activated protein kinases (MAPKs) which contribute to the immune response are efficiently triggered; for instance, stress-activated protein kinases (SAPKs)/c-Jun N-terminal kinases (JNKs) have been shown to regulate critical steps in T-helper-cell differentiation (8). Transcription factors of the NF-B family constitute an additional major pathway involved in the regulation of lymphocytes and execution of immune functions. NF-B transcription factors determine life and death decisions in developing lymphocytes and also provide important costimulatory signals in T cells (16).

Lymphocyte receptor signaling can be coupled to SAPK/JNK and NF-B activation by HPK1 (18). HPK1 is a mammalian hematopoiesis-specific Ste20 homologue belonging to the germinal center (GC) kinase family, which contains at least 20 kinases (6). HPK1 is characterized by an N-terminal kinase domain, followed by an intermediate region and a C-terminal Citron homology domain. Both the N-terminal kinase domain and the C-terminal Citron homology domain are highly conserved among related kinases of the GC family. In the GC family, HPK1 is most closely related to germinal center kinase (GCK), germinal center-like kinase (GLK), and germinal center-related kinase (GCKR), which together form subfamily I of GCKs (19). The biological function of these Ste20 homologues is largely elusive. GCK and GCKR have been implicated in tumor necrosis factor receptor signaling (36, 42). More recently, HPK1 has been shown to be a positive regulator of cell death in T cells (35).

HPK1 has been demonstrated to be a potent and highly specific activator of the SAPKs/JNKs in various cell types, a property shared with the related GCK family members (18, 19). HPK1 likely acts at the level of a MAP4 kinase, resulting in the activation of MAP3 kinases, such as MLK3 or MEKK1, which signal through the well-established MKK4/7 SAPK/JNK cascade (11, 17). Furthermore, HPK1 contributes to the activation of NF-B family transcription factors (2, 12). Forced ectopic expression of HPK1 in epithelial cells or murine fibroblasts results in constitutive activation. In lymphocytes, HPK1 kinase activity has been shown to be up-regulated in response to T- or B-cell receptor cross-linking (25). The kinetics of HPK1 activation is very rapid, arguing in favor of an immunoreceptor-proximal function; however, the molecular details of HPK1 kinase activation are poorly understood so far.

HPK1 tyrosine residue 379 is phosphorylated in lymphocytes after immunoreceptor stimulation, likely by kinases of the Src or Syk family. Tyrosine 379 provides a binding site for the SH2 domains of the adaptor proteins SLP-76, SLP-65 (BLNK/BASH), and cytokine-dependent hematopoietic cell linker (Clnk) (34, 38, 41). Phosphorylation of HPK1 on this tyrosine residue is a prerequisite for its enzymatic activation in response to immunoreceptor stimulation (26, 38). Furthermore, HPK1 activation crucially depends on three proline-rich motifs that have been shown to mediate constitutive association with SH3 domain-containing adaptor proteins, such as Grb2 (1), HS1 (30), Crk/CrkL (24, 31), Gads (Grb2-related adaptor downstream of Shc) (22, 26), Bam32/DAPP1 (10), or SH3P7/HIP-55 (9). Recently, it has been shown that HPK1 in association with SH3P7/HIP-55 is recruited to the immunological synapse and contributes to the down-regulation of T-cell receptor (TCR) signaling (20).

The serine/threonine kinase PKD1 is the most predominant PKD isoform expressed in mouse T cells (27). In T cells, B cells, and mast cells, PKDs are strongly activated by antigen receptor stimulation but not by costimulatory receptors, chemokines, or cytokines (29). Upon stimulation which depends on the action of protein kinases C (PKCs) and PLCs, PKD can shuttle between the plasma membrane and the Golgi apparatus, where it likely contributes to the regulation of intracellular transport (39). In T cells, PKD appears to fulfill distinct functions in lineage development and lymphocyte activation, depending on its cellular location and context (27).

Because HPK1 is potently activated by immunoreceptors and is a highly selective signal mediator on the road to SAPK and NF-B activation, HPK1 is an important candidate for the fine tuning and orchestration of immunoreceptor signaling in leukocytes. Despite the increasing evidence for a function of HPK1 in lymphocyte signaling, the molecular mechanism of its activation remains obscure. A more profound understanding of the details of HPK1 activation promises important insights into the regulation of leukocyte function. We show here that the Src family kinase Lck is capable of localizing HPK1 to the plasma membrane. After immunoreceptor stimulation, HPK1 relocates to the contact site of antigen-presenting cell (APC)-T-cell conjugates. Relocation is accompanied by tyrosine phosphorylation and adaptor protein binding and results in enzymatic activation of HPK1. Full activation depends on phosphorylation of key regulatory sites in the HPK1 kinase activation loop; these site include threonine 165 and serine 171. While threonine 165 probably represents an autophosphorylation site, serine 171 is efficiently phosphorylated by protein kinase D1 (PKD1). We provide evidence that transphosphorylation of serine 171 by PKD1 is prerequisite for full activation of HPK1. Furthermore, in Jurkat cells, PKD1 enhances HPK1 activity upon T-cell receptor engagement, while antisense down-regulation of PKD1 results in reduced HPK1 inducibility. This identifies PKD1 as a physiological regulator of HPK1 and hence of HPK1-driven SAPK/JNK and NF-B activation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

 
Cell culture and stimulation. Cos1 and 293T cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 5% fetal bovine serum (Sigma or Bethesda Research Laboratories), 100 µg of penicillin (Life Technologies, Inc.) per ml, 100 µg of streptomycin (Life Technologies, Inc.) per ml, and 20 mM L-glutamine (Life Technologies, Inc.).

Jurkat T cells, mouse T-cell hybridoma DC27.1, DO11.10 T cells, and A20 B cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (Gibco/BRL), 100 µg of penicillin (Life Technologies, Inc.) per ml, 100 µg of streptomycin (Life Technologies, Inc.) per ml, 20 mM L-glutamine (Life Technologies, Inc.), and 0.001% monothioglycerol (Sigma). For TCR stimulation, 10⁷ T cells were resuspended in phosphate-buffered saline (PBS) together with 2 µg of anti-CD3 antibody (OKT3 for Jurkat T cells and 145-2C11 [BD Bioscience] for DC27.1 T cells) per ml and 20 ng of soluble protein A per ml at 37°C for 3 min.

Transient transfections and NF-B reporter assays. Luciferase reporter assays were performed as previously described (2). All experiments were repeated at least three times. In the figures, bars depict the averages of duplicate transfection experiments. All values presented in a single diagram are derived from a single transfection experiment. Jurkat or DC27.1 T cells (10⁷) were transiently transfected by electroporation (250 V, 950 µF) in 500 µl of cell culture medium using 20 to 30 µg of total DNA. 293T cells were transfected by the Ca²⁺ phosphate coprecipitation method. For in vitro kinase assays, cells were harvested 48 h after transfection.

Immunoblotting, in vitro kinase assays, and SAPK activation. HPK1 and SAPK kinase assays were performed as described previously (17). Polyclonal rabbit sera for anti-HPK1 have been described previously (2). The hemagglutinin (HA) tag was detected using the mouse monoclonal antibody 12CA5 or sc-7392 (Santa Cruz Biotechnology). Phosphotyrosine was detected using mouse monoclonal antibody 4G10 (Upstate Biotechnology). PKD was detected using polyclonal antibody D-20 (Santa Cruz Biotechnology).

Immunofluorescence staining. Cos1 cells were seeded onto glass coverslips in six-well plates and transfected using the Ca²⁺ phosphate coprecipitation method with 2 to 5 µg of expression plasmid per well. Raft-like membrane patches were stained by 10-min incubation on ice with Alexa Fluor 594-conjugated cholera toxin subunit B (diluted 1:1,500 in PBS plus 0.1% bovine serum albumin [BSA] [Molecular Probes, Eugene, Oreg.]). After the cells were washed, they were fixed at room temperature for 30 min using PBS containing 4% formaldehyde and washed repeatedly.

For immunofluorescence staining, cells were washed, fixed in 2% paraformaldehyde, permeabilized with 0.5% Triton X-100, and blocked in PBS containing 5% BSA. Cells were then incubated with either specific antiserum for Lck or Clnk or a monoclonal antibody against the HA tag (Santa Cruz Biotechnology) for 1 h at 37°C and then washed. Proteins were visualized by incubation with appropriate fluorochrome-coupled secondary antibodies for 1 h at 37°C. Cell nuclei were counterstained with 2 µg of Hoechst 33258 (Sigma) per ml in PBS for 15 min on ice. Cells were embedded in fluorescence mounting medium (Mowiol) and observed by epifluorescence microscopy.

Cell conjugation analysis. A20 B cells were stained with 1 µM cell tracker green (Molecular Probes) and pulsed with OVA (ovalbumin and ovalbumin) peptide or left alone. A20 cells were then mixed with DO11.10 T cells in a 2:1 ratio and incubated for different times at 37°C. The conjugates were gently resuspended in 4% paraformaldehyde and fixed on poly-L-lysine-coated slides for 30 min. Cells were washed, permeabilized in 0.5% Triton X-100, blocked with PBS containing 2.5% BSA, and stained with anti-HPK1 and anti-CD3 antibodies, followed by anti-rabbit Cy3-labeled and anti-rat fluorescein isothiocyanate (FITC)-labeled secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, Pa.). Samples were embedded in Mowiol mounting medium and analyzed by confocal laser scanning microscopy (Leica TCS SP, Heidelberg, Germany).

Phosphoamino acid analysis and two-dimensional tryptic phosphopeptide mapping. Two-dimensional tryptic phosphopeptide mapping or phosphoamino acid analysis of phosphorylated HPK1 proteins using Hunter thin-layer electrophoresis system HTLE-7000 (C.B.C. Scientific Company, Inc.) was performed closely adhering to the published protocols (4). For two-dimensional peptide maps, peptides were electrophoresed in an acidic buffer system (pH 1.9) and subsequently separated orthogonally by ascending thin-layer chromatography (TLC) in phosphor chromatography buffer. Phosphopeptides were visualized by autoradiography.

Phosphopeptide spots subjected to phosphoamino acid analysis were marked on TLC plates, and the corresponding cellulose was scraped off using a pipette filter tip. The peptides were eluted sequentially with acidic buffer (pH 1.9) and pyridine diluted in acidic buffer (pH 1.9). After the peptides were hydrolyzed in 6 N HCl, the resulting amino acids were separated on TLC plates by two-dimensional electrophoresis using an acidic buffer system (pH 1.9 and 3.5) and exposed to X-ray film. The positions of the different phosphoamino acids were identified by overlay with a reference phosphoamino acid standard stained with ninhydrin (0.25% [wt/vol] in acetone).

In vivo labeling. Jurkat T cells were transfected with expression plasmids for HPK1 together with Clnk. After 24 h, the cells were washed and incubated in phosphate-free medium (Dulbecco's modified Eagle's medium without phosphate) with 0.5% fetal calf serum, 20 mM L-glutamine (Life Technologies, Inc.), 0.001% monothioglycerol (Sigma), and 93 MBq of ³²P_i for 7 h at 37°C and 7% CO₂. Cells were left unstimulated or stimulated with anti-CD3 antibody (5 µg/ml) and soluble protein A (20 ng/ml) for 3 min. Cells were washed, HPK1 was immunoprecipitated using anti-HA antibody, and analyzed by autoradiography after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

 
In Cos1 cells, HPK1 is localized to the cytoplasm and tends to accumulate in intensely fluorescing structures upon overexpression. To better understand the details of HPK1 activation, we investigated the intracellular localization of transiently expressed HA-tagged HPK1 (HPK1-HA) in Cos1 cells by immunofluorescence staining. In cells expressing low levels of HPK1, the tagged protein was localized to the cytoplasm and was excluded from the nucleus (Fig. 1a and b). At intermediate levels of expression of HPK1, the protein appeared to localize in a multitude of small fluorescent spots throughout the cytoplasm, while at high levels of expression, HPK1 was mainly confined to large, intensely fluorescing clusters that were often located perinuclearly (Fig. 1b and c). Interestingly, upon appearance of the intensely fluorescing structures, the remainder of the cytoplasm appeared completely clear of HPK1 (see Fig. 4B). These clusters may have consisted either of vesicles highly loaded with ectopically expressed protein or of protein aggregates forming as a consequence of ectopic overexpression. Antibody specificity was controlled by staining of nontransfected cells; furthermore, in transfected cell cultures, HPK1-expressing cells were unambiguously discerned from nontransfected cells (Fig. 1).

View larger version (59K): [in this window] [in a new window]   FIG. 1. Full-length HPK1 and the isolated HPK1 C terminus localize comparably in Cos1 cells. (Top left) Schematic representation of the domain structure of HPK1-HA and the HPK1 C terminus fused to EGFP (EGFP-HPK1291-827). In EGFP-HPK1291-827, the kinase domain (kinase dom.) has been replaced by EGFP. P1, P2, and P4 denote proline-rich SH3 domain-binding sites. Tyrosine 379 can be phosphorylated by Src/Syk family kinases and provides a docking site for the SH2 domain of SLP-76 family adaptors. The numbers are amino acid positions in HPK1. The epitope tag (HA) is indicated. HPK1-HA (a to f), EGFP (g and h), or the EGFP-HPK1291-827 hybrid protein (i to o) was transiently expressed in Cos1 cells grown on coverslips. For visualization of HPK1-HA, formaldehyde-fixed cells were stained with anti-HA antibodies and then with Alexa Fluor 488-labeled secondary antibodies. Cells were embedded in fluorescence mounting medium (Mowiol) and observed with an epifluorescence microscope using a 63x/1.25 oil immersion objective (b, c, and e to o; bars, 25 µm) or a 20x objective (a and d; bars, 100 µm). HPK1-HA (a to c) and EGFP-HPK1291-827 (g, i, k, and l) are either shown alone or as an overlay with nuclei counterstained by Hoechst 33258 (d, e, f, h, m, n, and o). (a to c) Cells expressing low (L), medium (M), or high (H) levels of HPK1-HA (arrowheads) are identified. Cells expressing low, medium, or high levels of EGFP-HPK1291-827 are shown in panel i, k, or l, respectively.

View larger version (16K): [in this window] [in a new window]   FIG. 4. Active Lck dominates HPK1 localization upon overexpression in Cos1 cells. Cos1 cells grown on coverslips were transfected with expression plasmids encoding either Clnk or Lck (A) or HPK1-HA (B), HPK1-HA plus Clnk (C), HPK1-HA plus Lck (D), or HPK1-HA plus Clnk and Lck (E) as described in the legend to Fig. 3. Twenty-four hours after transfection, cells were fixed and stained with antibodies directed against the HA tag, Clnk, or Lck. In each experimental group, successfully transfected cells were then classified as expressing low, middle, or high levels of HPK1 (Fig. 1). For at least 20 cells of each level of expression of HPK1, HPK1 fluorescence in the nucleus (N), cytoplasm (C), and spots in the cytoplasm (S) and at the plasma membrane (M) was graded. The level of fluorescence is indicated on the y axes as follows: weak (•), intermediate (••), and strong (•••). The average fluorescence intensity of the different cellular compartments was then blotted for cells expressing low, middle, or high levels of HPK1. When Clnk and/or Lck were coexpressed, intracellular distribution of these proteins was determined in an analogous manner; however, the levels of expression always refer to HPK1 expression.

View larger version (16K): [in this window] [in a new window]   FIG. 5. Intracellular localization of HPK1 is mediated exclusively by its C-terminal part in Cos1 cells. Cos1 cells grown on coverslips were transfected with expression plasmids encoding either EGFP (A) or EGFP-HPK1291-827 (B) alone or EGFP-HPK1291-827 with Clnk (C), Lck (D), or Clnk and Lck (E) as described in the legend to Fig. 3. Twenty-four hours after transfection, cells were fixed and stained with antibodies directed against Clnk or Lck. In each experimental group, successfully transfected cells were then classified as expressing low, middle, or high levels of EGFP or EGFP-HPK1291-827 as judged by their green fluorescence (Fig. 1). For at least 20 cells of each expression level, green fluorescence (GF) in the nucleus (N), cytoplasm (C), and spots in the cytoplasm (S) and at the plasma membrane (M) was graded. The level of fluorescence is indicated on the y axes as follows: weak (•), intermediate (••), and strong (•••). The average fluorescence intensity of the different cellular compartments was thenblotted for cells expressing low, middle, or high levels of HPK1. When Clnk and/or Lck were coexpressed, intracellular distribution of these proteins was determined in an analogous manner; however, levels of expression always refer to EGFP or EGFP-HPK1291-827 expression.

View larger version (11K): [in this window] [in a new window]   FIG. 2. The adaptor protein Clnk enhances phosphorylation of HPK1 on tyrosine in the presence of the active Src kinase Lck. (A) The EGFP-HPK1291-827 fusion protein is not subject to excessive proteolysis in Cos1 cells. Expression plasmids for EGFP, EGFP-HPK1291-827, or the HPK1 C-terminus HPK1291-827 were transfected into Cos1 cells as indicated. After 24 h, the cells were lysed, and the stability of the HPK1 hybrid protein was tested by anti-HPK1 or anti-EGFP Western blotting. An extraordinarily strong exposure of the EGFP Western blot is shown to visualize degradation products of the HPK1 hybrid protein. (B) Cos1 cells were transfected with expression plasmids encoding either wild-type HPK1 [HPK1(wt)], HPK1(wt) plus Lck, or HPK1(wt) plus Lck and Clnk. After immunoprecipitation, HPK1 tyrosine phosphorylation was assessed by antiphosphotyrosine (anti-pTyr) Western blotting. In parallel, HPK1 kinase activity was determined by an in vitro kinase assay. HPK1 expression was demonstrated by anti-HPK1 Western blotting.

So far, the intracellular localization of HPK1 in resting and activated cells has not been investigated. We decided to address this question and determine in particular the effects of proteins necessary for HPK1 activation in lymphocytes on its subcellular localization. To more precisely assess the effects of Lck and Clnk on the intracellular localization of HPK1, we coexpressed HPK1-HA with either Clnk or Lck or both in Cos1 cells. Using immunofluorescence microscopy, we classified transfectants as expressing low, medium, or high levels of HPK1 (in accordance with Fig. 1); for at least 20 cells of each class, we visually judged the fluorescence intensity in the nucleus, cytoplasm, and cytoplasmic aggregates and at the plasma membrane. Each intracellular location was attributed either a weak, intermediate, or strong fluorescence signal, and the average fluorescence at the different localizations was blotted for the different proteins analyzed. Throughout the analysis, staining specificity was controlled on nontransfected samples.

When Clnk was expressed alone, we found Clnk persistently localized to the plasma membrane and in small spots that were most prominent around the nucleus. Similarly, Lck was found at the plasma membrane but also in an intensely fluorescent area close to the nucleus (Fig. 3, top row, middle panel, and Fig. 4A). Interestingly, at low levels of expression, coexpression of HPK1 and Clnk redirected both proteins into cytoplasmic aggregates and cause a decrease in Clnk membrane localization (Fig. 4C). At intermediate and high levels of expression, HPK1 remained in cytoplasmic aggregates, and again membrane localization of Clnk was reduced (Fig. 3 and 4C). Taken together, Clnk alone was inefficient in localizing HPK1 to the plasma membrane, while HPK1 appeared to divert Clnk away from the plasma membrane. In contrast, coexpression with Lck resulted in HPK1 membrane recruitment even at low levels of expression; at intermediate and high levels of expression, HPK1 and Lck largely colocalized (Fig. 3 and 4D). Simultaneous expression of all three proteins resulted in significant colocalization at all levels of expression analyzed (Fig. 3 and 4E). Membrane localization of HPK1 was again most noticeable at low levels of expression. While cytoplasmic and membrane localization were balanced in cells expressing low levels of HPK1, increasing levels of expression of HPK1 resulted in a concentration of all three proteins in spots in the cytoplasm. Interestingly, in this group, a high proportion of HPK1-expressing cells had already undergone apoptosis 24 h after transfection. From our analysis, it appeared that at low levels of expression of HPK1, which best reflect natural HPK1 levels in lymphocytes, Lck was most efficient in directing HPK1 to the plasma membrane. In Cos1 cells, HPK1 was, at least to some degree, capable of directing the localization of Clnk, arguing in favor of an interaction between both proteins. Surprisingly, in Cos1 cells, localization of the HPK1(Y379F) mutant was similar to that of the wild-type protein (not shown).

View larger version (47K): [in this window] [in a new window]   FIG. 3. Active Lck and Clnk direct HPK1-HA or the EGFP-HPK1291-827 fusion protein to the plasma membrane. Cos1 cells grown on coverslips were transfected with expression plasmids encoding either HPK1-HA, Clnk, or Lck alone (top row) or HPK1-HA in combination with Lck or Clnk or Lck and Clnk or EGFP-HPK1291-827 in combination with Lck and Clnk. Twenty-four hours after transfection, cells were fixed with formaldehyde and stained with the respective antibodies. Hoechst 33258 was used to counterstain nuclei (middle column and top row). Specimens were observed with an epifluorescence microscope using a 63x/1.25 oil immersion objective, and micrographs were acquired using a Spot digital camera. Merging of green (anti-HA or anti-EGFP) and red (anti-Clnk or anti-Lck) fluorescence is depicted as overlay.

The virtually identical intracellular distributions of HA-tagged full-length HPK1 and EGFP-HPK1_291-827 fusion protein suggested that the intracellular localization of HPK1 is dominated by sequences located between amino acids 291 and 827. Apparently, Lck is more efficient in recruiting HPK1 to the plasma membrane than Clnk. Nevertheless, when these proteins are present simultaneously, these proteins colocalize, and through the concerted action of SLP-76 family adaptors and Src and/or Syk family kinases, HPK1 is recruited into a membrane-proximal position where it is phosphorylated on tyrosine.

HPK1 is recruited to the interface of B- and T-cell conjugates. Does immunoreceptor activation in lymphocytes cause translocation of HPK1 to the site of receptor clustering? To directly address this question, we generated conjugates of OVA peptide-pulsed A20 B cells and DO11.10 T-cell hybridoma cells. We then visualized the distribution of endogenous HPK1 30 and 60 min after conjugate formation. HPK1 polarization in pairs of nonpulsed A20 and DO11.10 cells was not discerned. In conjugates of OVA-pulsed A20 and DO11.10 cells, we observed recruitment of HPK1 to the B-cell-T-cell interface at 30 and 60 min after conjugate formation (Fig. 6). Somewhat unexpectedly, we detected significantly higher HPK1 levels of expression in nonpulsed A20 B cells than in DO11.10 T cells (Fig. 6, –Ova). Our analysis showed that HPK1 can be recruited in either cell type to the conjugate interface (Fig. 6, +Ova 30 min). Finally, we noted a strong decline in HPK1 immunoreactivity after conjugate initiation (compare staining intensities of the –Ova, +Ova 30 min, and +Ova 60 min groups), suggesting that HPK1 is efficiently cleared from activated lymphocytes.

View larger version (22K): [in this window] [in a new window]   FIG. 6. HPK1 is recruited to the contact site of APC-T-cell conjugates. A20 B cells labeled with cell tracker green were pulsed with OVA peptide and incubated with antigen-specific DO11.10 T hybridoma cells to induce conjugate formation. Conjugates were fixed and stained for HPK1 (red) and CD3 (green). In areas where HPK1 is recruited to the contact sites, the overlay of Cy3-stained HPK1 and FITC-stained CD3 and/or cell tracker appears yellow. The negative control shows A20 cells not pulsed with OVA peptide (– Ova); conjugate formation with OVA-pulsed A20 cells was allowed to proceed for 30 or 60 min. Representative conjugates analyzed by confocal laser scanning microscopy are shown.

Three amino acids in the kinase domain activation loop of HPK1 are indispensable for maximal HPK1 kinase activity. GCKs, including HPK1, are constitutively active upon expression in epithelial cells. HPK1 is immunopurified from Cos1 cells in an active conformation that readily undergoes autophosphorylation in vitro. Hence, activation of HPK1 appears to occur preferentially under conditions that are associated with local clustering of the kinase, due to immunoreceptor ligation or forced expression. We hypothesized that the actual activation process may involve auto- or transphosphorylation steps, which rely on local accumulation. In the catalytically active form, all protein kinases adopt a structurally similar conformation. A loop of typically 20 to 30 amino acids localized at the center of the protein kinase domain between subdomains VII and VIII is known to regulate entry of the substrate peptide to the active site and to provide important charge interactions with the active center. For many kinases in which the catalytic aspartate is preceded by an arginine (RD kinases), phosphorylation of the activation loop results in a stable open conformation and enzymatic activation (14). On the basis of a sequence comparison with related serine/threonine kinases, we identified three potentially activating phosphoacceptor sites in the HPK1 activation loop: threonine 165, serine 171, and threonine 175 (Fig. 7A). These sites were mutated into alanine either individually or in combination, and the activity of the resulting HPK1 mutant kinases was assessed in vitro after transient expression in Cos1 cells. We have previously shown that the SLP-76 family adaptor Clnk will bind HPK1 with phosphorylated tyrosines and will also copurify with HPK1 during immunoprecipitation and is efficiently phosphorylated by HPK1 in vitro (41). Therefore, we coexpressed Clnk to provide a substrate for HPK1 kinase which directly binds to HPK1. In addition, a bacterially expressed c-Jun fusion protein (glutathione S-transferase [GST]-c-Jun_N) was included in the assay as an additional surrogate substrate. Autophosphorylation of HPK1 and transphosphorylation of the substrate proteins were assessed.

View larger version (9K): [in this window] [in a new window]   FIG. 7. Mutation of potential phosphorylation sites in the HPK1 kinase domain activation loop impairs HPK1 kinase activity and generates HPK1 mutants refractory to TCR ligation-mediated activation. (A) Alignment of the amino acid sequences of activation loops of different serine/threonine kinases. Residues reported to be phosphorylated and/or implicated in the regulation of the respective kinase are shown in bold type. The alignment identified three potential phosphorylation sites in the HPK1 activation loop, which are shown as white letters on a black background. The signature motifs of the kinase subdomains VII (DFG) and VIII (APE) and a central serine/threonine and an invariant threonine residue are highlighted by light grey shading. (B) Cos1 cells were transiently transfected with plasmids encoding Clnk and either empty vector (control) or HPK1 [wild-type (wt)] or the ATP-binding site mutant HPK1(K46E) or various HPK1 constructs containing single or multiple activation loop mutations at position 165 (threonine to alanine [T165A]), position 171 (serine to alanine [S171A]), or position 175 (threonine to alanine [T175A]). After immunoprecipitation, the indicated variant proteins were tested for their abilities to autophosphorylate and to transphosphorylate a recombinant GST-c-JunN fusion protein in an in vitro kinase assay. Reaction products were separated by SDS-PAGE, and phosphoproteins were visualized by autoradiography (top panel). The migratory positions of HPK1 proteins, coimmunoprecipitated Clnk, and GST-c-JunN are indicated to the right of the gels. The levels of expression of HPK1 and Clnk were visualized by Western blotting (bottom panels). A representative experiment of four experiments is shown. (C) Jurkat T cells were transiently cotransfected with plasmids encoding Clnk and either empty vector (control), or HPK1 (wild type [wt]), or HPK1 constructs containing the indicated activation loop mutations. Forty-eight hours after transfection, cells were stimulated for 3 min using the anti-CD3 antibody OKT3. HPK1 proteins were immunoprecipitated and tested for their ability to autophosphorylate in an in vitro kinase assay. Reaction products were separated by SDS-PAGE; the migratory positions of HPK1 proteins and coimmunoprecipitating Clnk are indicated to the right of the gels. Expression of the HPK1 proteins was determined by anti-HA Western blotting. The positions of the two nonspecific signals are indicated by asterisks. The experiment was repeated twice with similar results.

For all mutants analyzed, the relative ratio of HPK1 autophosphorylation to Clnk and GST-c-Jun_N transphosphorylation remained constant, indicating that with respect to these phosphorylation targets, HPK1 was the only active kinase present in the assay.

HPK1 activation in response to TCR stimulation is regulated by the kinase domain activation loop. At this time, immunoreceptor stimulation is the best-characterized physiological stimulus for HPK1. If immunoreceptor-mediated activation of HPK1 occurred through phosphorylation of the activation loop, activation loop mutants should be refractory to stimulation by TCR cross-linking. To test this hypothesis, we ligated CD3 in Jurkat T cells and determined the kinase activity of wild-type or mutant HPK1 in vitro (Fig. 7C). Wild-type HPK1 kinase was markedly activated after CD3 cross-linking. Of the three single amino acid activation loop mutants, only the T165A mutant displayed a residual capacity for TCR-mediated activation. The S171A and T175A single mutants failed to become stimulated after TCR cross-linking. Concomitantly, substrate phosphorylation of the coexpressed adaptor protein Clnk was no longer detected.

We wondered whether activation loop phosphorylation of HPK1 occurred in vivo. To address this question, we transfected Jurkat T cells with either wild-type HPK1 or the T165A or S171A mutant. After metabolic labeling with ³²P_i, we ligated the TCR and immunopurified HPK1 proteins (not shown). We detected no significant difference in phosphate content between wild-type and mutant forms of HPK1. Even in nonstimulated cells, HPK1 was already phosphorylated, suggesting that HPK1 is phosphorylated on multiple acceptor sites in vivo. Some of these sites are likely located outside the regulatory loop of the kinase domain and are occupied independently of immunoreceptor stimulation.

The capacity of HPK1 kinase activation loop mutants to activate downstream effector pathways is strongly impaired. Next, we tested the capacity of the described HPK1 activation loop mutants to activate well-established HPK1 effector pathways, namely, the SAPK and NF-B pathways (Fig. 8). The indicated HPK1 mutant proteins were transiently expressed in Cos1 cells, and after immunopurification, the activation state of endogenous SAPK/JNK was determined by an in vitro kinase assay (Fig. 8A). All activation loop mutants tested failed to significantly activate SAPKs; only wild-type HPK1 elicited a robust enhancement of SAPK kinase activity.

View larger version (10K): [in this window] [in a new window]   FIG. 8. HPK1 activation loop mutants are severely impaired in mediating SAPK/JNK and NF-B activation. (A) Wild-type HPK1 (wt) or HPK1 proteins bearing the indicated activation loop mutations were transiently expressed in Cos1 cells together with p54-SAPKß/JNK. Forty-eight hours after transfection, coexpressed SAPK/JNK protein was immunopurified, and then its kinase activity towards bacterially expressed GST-c-JunN protein was determined. Phosphorylated reaction products were separated by SDS-PAGE, and relative GST-c-JunN phosphorylation levels were determined assessed using a phosphorimager. Expression of HPK1 proteins was demonstrated by anti-HPK1 Western blotting. A representative experiment of two experiments is shown. (B) Cos1 cells were transfected with increasing amounts of plasmids (1, 2, and 4 µg) encoding either wild-type HPK1 (wt) or HPK1 constructs containing single activation loop mutations as indicated or empty vector (–) in the presence of a NF-B double reporter system (2). After 48 h, NF-B-driven luciferase activity was determined and normalized against a ß-galactosidase transfection control using a chemiluminescence assay system. Relative activation of NF-B-driven luciferase activity normalized for transfection efficacy is shown. Bars depict the averages of duplicate samples; the experiment was repeated three times with similar results. Matching levels of expression of HPK1 proteins were visualized by anti-HPK1 Western blotting.

For GCK, it has been argued that activation of downstream effector pathways can occur in the absence of kinase activity, probably mediated by a scaffolding function of GCK aligning pathway components in a favorable conformation. In our assays, HPK1 activation loop mutants failed to activate downstream pathways. Even the T165A mutant, which had retained a small but significant degree of kinase activity in the autophosphorylation assay, elicited no relevant SAPK or NF-B activation. This demonstrates that in the absence of HPK1 kinase activity, scaffolding effects are irrelevant for SAPK/JNK or NF-B activation.

Our results demonstrate that in intact cells, amino acids T165, S171, and T175 in the HPK1 activation loop are necessary to render the molecule sufficiently active for phosphorylation of its downstream effector substrates. On the basis of these results, we reasoned that one or more of the mutated amino acids might be a substrate for protein kinase phosphorylation.

HPK1 autophosphorylates sites in its activation loop in vitro. Therefore, we attempted to define an in vitro system that would allow us to study phosphorylation events during HPK1 activation. Taking advantage of the fact that immunopurified HPK1 readily autophosphorylates in vitro, we asked whether HPK1 can function as a HPK1 activation loop kinase. Phosphotryptic peptide mapping of full-length in vitro autophosphorylated HPK1 revealed a large number of tryptic phosphopeptides (Fig. 9A). As we had already observed in vivo, in vitro HPK1 also autophosphorylates at multiple amino acids, most of which are located outside the activation loop, possibly even outside the kinase domain. To reduce the difficulties associated with the analysis of such a complex pattern of spots, we generated a truncation mutant of HPK1 containing the first 232 amino acids C terminally flanked by an HA tag. This truncation mutant, designated HPK1_1-232-HA, encompassed subdomains I to X of the HPK1 kinase domain. In subsequent experiments, HPK1_1-232-HA served as a phosphorylation target for full-length, kinase-active HPK1. We reasoned that elimination of potential phosphorylation sides in the central and C-terminal domains of HPK1 should result in a significantly simplified spot pattern.

View larger version (23K): [in this window] [in a new window]   FIG. 9. HPK1 proteins autophosphorylate in vitro on multiple amino acids. Autophosphorylation sites in the kinase domain are mostly threonine residues. (A) HPK1 protein was transiently expressed in Cos1 cells and subjected to immunoprecipitation and autophosphorylation in vitro. Reaction products were separated by SDS-PAGE and visualized by autoradiography. Full-length HPK1 phosphoprotein was eluted from the gel (white box in panel B) and subjected to tryptic digestion. Tryptic phosphopeptides were separated by two-dimensional TLC on a Hunter HLTE-7000 machine and visualized by autoradiography. The polarity of the electrophoretic separation and direction of the ascending chromatography are indicated. (B) Full-length HA-tagged HPK1 (wild type) was coexpressed in Cos1 cells with either empty vector (control) or with the HA-tagged HPK11-232 truncation mutant consisting of the first 232 amino acids of HPK1. The truncation mutant was either wild type at amino acid position 46 or harbored a K46E mutation which precludes ATP binding to the kinase domain. Anti-HA immunoprecipitation and a subsequent in vitro kinase reaction demonstrated transphosphorylation of the kinase-dead HPK11-232 fragment (left panel). Expression of all HPK1 proteins was visualized by anti-HA Western blotting (right panel). The white box depicts full-length HPK1 protein eluted for the generation of tryptic phosphopeptide maps as shown in panel A. The black box depicts the HPK11-232 fragment that was eluted for the subsequent phosphoamino acid analysis (C) and tryptic phosphopeptide maps (Fig. 10B). (C) HPK1 (wild type [wt]) or truncated HPK11-232(K46E) or HPK11-232(K46E) containing a triple activation loop mutation (TST165,171,175AAA) were transphosphorylated by full-length HPK1 in vitro as described in panel B. After SDS-PAGE, the depicted HPK1 phosphoproteins were eluted and subjected to phosphoamino acid analysis using a Hunter HTLE-7000 machine. The buffer system, polarity of the electrophoretic separations, and positions of reference phosphoamino acids are indicated. p-Ser, phosphoserine; p-Thr, phosphothreonine; p-Tyr, phosphotyrosine, Pi, free phosphate.

Phosphoamino acid analysis of autophosphorylated full-length HPK1 expressed in Cos1 cells predominantly revealed phosphorylation on threonine residues and a significantly lower level of serine phosphorylation (Fig. 9C). The HPK1_1-232-HA fragment contained phosphothreonine at a level comparable to that of wild-type HPK1 and, if at all, minute amounts of phosphoserine. Furthermore, threonine phosphorylation levels were significantly reduced in a triple mutant of the HPK1_1-232-HA fragment in which the three potential activation loop phosphorylation sites had been changed into alanine (TST165,171,175AAA). These results suggested that HPK1 autophosphorylates on multiple sites in the first 232 amino acids and that autophosphorylation takes place in the activation loop but is restricted to threonine phosphorylation.

Threonine 165 of the HPK1 kinase activation loop is an autophosphorylation target. To test for autophosphorylation of threonine 165, serine 171, and threonine 175, the respective alanine mutations were introduced into HPK1_1-232(K46E)-HA, and the resulting mutated kinase domain fragments were transiently coexpressed with full-length HPK1-HA in Cos1 cells. After transphosphorylation by full-length HPK1, phosphoproteins were separated by SDS-PAGE and visualized by autoradiography. Compared to the fragment harboring the wild-type activation loop, mutation of the three candidate phosphorylation sites resulted in a relative reduction in phosphorylation levels. However, even mutation of all three candidate sites did not completely abolish fragment phosphorylation, supporting the notion that additional amino acids of the kinase domain outside the activation loop are phosphorylated by full-length HPK1 (Fig. 10A).

View larger version (23K): [in this window] [in a new window]   FIG. 10. Phosphotryptic peptide mapping identifies threonine 165 in the HPK1 activation loop as an autophosphorylation site. (A) Wild-type HPK11-232(K46E)-HA (wt) or HPK11-232(K46E)-HA proteins containing the indicated activation loop mutations were transiently coexpressed with full-length kinase-competent HPK1-HA. After immunopurification using anti-HA antibodies, the truncated proteins were transphosphorylated by full-length HPK1 in vitro. Reaction products were separated by SDS-PAGE, phosphoproteins were visualized by autoradiography, and expression of the different truncated proteins was demonstrated by anti-HA Western blotting. The various mutations in the activation loop sequence are indicated over the gel. (B) The HPK11-232-HA proteins shown in panel A were recovered from the SDS-polyacrylamide gel and subjected to tryptic peptide mapping. Phosphopeptides were separated on a Hunter HTLE-7000 machine and visualized by autoradiography. The maps are oriented as in Fig. 9A. (C) HPK1 activation loop amino acid sequence. The potentially phosphorylated amino acids threonine 165, serine 171, and threonine 175 (asterisks) and trypsin cleavage sites (scissors) are indicated. The sequences of four possible tryptic peptides generated from this sequence are displayed. The charges of charged amino acids at pH 1.9 are denoted. (D) Phosphoamino acid analysis of the phosphopeptides 1 and 2 indicated in panel b of panel B. The spots corresponding to peptides 1 and 2 were identified by autoradiography and scraped off the TLC plate, and phosphopeptides were eluted. After digestion in 6 M HCl, phosphoamino acids were separated. The buffer system, polarity of the electrophoretic separations, and positions of reference phosphoamino acids are indicated as described in the legend to Fig. 9C.

Surprisingly, mutation of serine 171 to alanine (S171A) did not lead to the loss of different spots from the phosphotryptic peptide map but yielded a pattern identical to that of the T165A mutation (Fig. 10B, panel d). Possible explanations for this puzzling result are that either the amino acid serine 171 was indispensable for T165 to become phosphorylated or both amino acids were phosphorylated but the phosphorylation events were interdependent. In the latter case, phosphorylation of S171 could be a prerequisite or priming phosphorylation for phosphorylation of T165 or vice versa.

Due to the tandem arginine residues R168 and R169, tryptic digestion would produce two phosphopeptides containing serine 171, which are designated peptides 3 and 4 in Fig. 10C. Peptide 3 is predicted to be 21 amino acids long, carry a net electrical charge of +2 at pH 1.9, and be less hydrophobic than peptide 4, which is predicted to carry a charge of +1 at pH 1.9. Due to the very similar properties of the pairs of peptides (peptides 1 and 2 and peptides 3 and 4), spots 1 and 2 in Fig. 10B could also represent peptides 3 and 4, which contain S171.

As it was impossible to distinguish phosphopeptides containing phosphorylated T165 (pT165) or pS171 on the basis of their migratory behavior in phosphotryptic peptide maps, we eluted the peptides corresponding to spots 1 and 2 from the TLC carrier and performed a subsequent phosphoamino acid analysis. Spots 1 and 2 contained only phosphothreonine, confirming our initial assignment as pT165-containing peptides 1 and 2 (Fig. 10D). Furthermore, this analysis also demonstrated that serine 159 is not an autophosphorylation target.

Mutation of threonine 175 alone (T175A) or in combination with S171 (ST171,175AA) resulted in a general strong reduction of fragment phosphorylation. The observed residual weak phosphotryptic peptide pattern was reminiscent of the pattern generated by the wild-type fragment (Fig. 10B, panels e and f). This result suggested that mutation of T175 to alanine introduced a significant structural perturbation into the activation loop, which affected its binding as a substrate to the active site. Further support for this notion was provided by a mutant in which in addition to the three potential phosphorylation sites, the adjacent tyrosine 177 was mutated into alanine. The phosphopeptide map of this quadruple mutant (TSTY-AAAA) resembled the maps of T165A and S171A, suggesting that the Y177A mutation compensated for the T175A mutation. Possibly, the Y177A mutation released structural constraints introduced by T175A, and substrate binding and phosphorylation were again possible (Fig. 10B, panel g).

Serine 171 in the HPK1 activation loop is a PKD1 phosphorylation site. We were puzzled by the apparent interdependence of T165 and S171 phosphorylation. Intriguingly, mutation of S171 had a significantly stronger effect on overall HPK1 activity than mutation of T165 (Fig. 7B). A possible explanation is that phosphorylation of S171 is a prerequisite or priming phosphorylation for T165 autophosphorylation and that a kinase other than HPK1 is responsible for phosphorylation of S171. This phosphorylation event could take place during transient expression in intact cells and would consequently not be detected in subsequent in vitro kinase reactions. Therefore, we examined whether the sequence context of the mutated amino acid S171 was particularly favorable for any known serine/threonine kinase and found S171 to be a consensus phosphorylation site for PKD family serine/threonine kinases.

To test for phosphorylation of the HPK1 activation loop by PKD1, we made use of mutations in the PKD1 zinc fingers or PH domain which render this kinase constitutively active (13). Coexpression of wild-type HPK1_1-232(K46E)-HA with active PKD1 resulted in efficient phosphorylation of the HPK1 fragment (Fig. 11A). The HPK1 activation loop T165A and T175A mutants were phosphorylated at levels comparable to that of the wild-type fragment. In sharp contrast, a S171A mutant completely failed to become phosphorylated by active PKD1. This result demonstrated that PKD1 was able to phosphorylate the activation loop of HPK1.

View larger version (24K): [in this window] [in a new window]   FIG. 11. Serine 171 in the HPK1 activation loop can be transphosphorylated by PKD1. (A) The indicated HPK11-232(K46E)-HA truncation proteins were transiently coexpressed with activated versions of full-length PKD1-HA. Anti-HA immunopurified proteins were subjected to in vitro kinase reactions and separated by SDS-PAGE. The migratory position of the HPK11-232 fragment is indicated to the right of the gels. Expression of the different proteins was verified by Western blotting. wt, wild-type protein. (B) (Left) HPK11-232(K46E)-HA protein was transphosphorylated by the PKD1 (Zn-Mut) mutant in vitro. After separation of the reaction mixture by SDS-PAGE, the HPK1 fragments were recovered from the SDS-polyacrylamide gel and subjected to tryptic peptide mapping. Phosphopeptides were separated on a Hunter HTLE-7000 machine and visualized by autoradiography. The maps are oriented as in Fig. 9A. (Right) Schematic depiction of the phosphotryptic map on the left. Major spots which coincide with the presence of S171 in the activation loop peptide are shown in black; these spots are not present after phosphorylation of the HPK11-232(K46E,S171T)-HA fragment. Dark grey spots likely represent degradation products of the active kinase PKD1 (Zn-Mut). (C) Phosphoamino acid analysis of phosphopeptides 3 and 4 indicated in the right panel of panel B. The buffer system, polarity of the electrophoretic separations, and positions of reference phosphoamino acids are indicated as described in the legend to Fig. 9C. (D) Activation loop sequence alignment of subfamily I GCKs. Grey boxes highlight the amino acids T165 and S171 that are phosphorylated in and contribute to the activation of HPK1. A hydrophobic residue at position –5 relative to S171 (white letters on black background) is critical for phosphorylation by PKD1 and conserved in all family members.

In summary, our analysis identified T165 of the HPK1 activation loop as an autophosphorylation target, while S171 is phosphorylated by PKD. Furthermore, we obtained evidence of a priming function of phosphoserine 171 for T165 phosphorylation. Therefore, transphosphorylation by PKD1 and subsequent autophosphorylation appear to be prerequisite for full activation of HPK1.

PKD1 enhances activation of HPK1 after TCR stimulation. To analyze a possible effect of PKD1 on HPK1 activation, PKD1 and HPK1 were coexpressed in Jurkat T cells. The presence of either wild-type PKD1 or the constitutively active PKD1 (PH-Mut) and PKD1 (Zn-Mut) mutants augmented HPK1 kinase activity after TCR stimulation (Fig. 12A). In the presence of wild-type PKD1, HPK1 activity was mildly reduced in the basal state and slightly enhanced after TCR ligation, resulting in a clear increase in inducibility. Constitutively active forms of PKD1 caused a strong stimulation of HPK1 activity, regardless of TCR ligation. A kinase-deficient mutant, PKD1 (KD), had no influence on HPK1 activity, likely due to compensation by other PKD isoforms endogenously expressed in Jurkat cells.

View larger version (34K): [in this window] [in a new window]   FIG. 12. PKD influences HPK1 kinase activity in various cell types and augments HPK1 effector pathway signal strength. (A) HPK1-HA and wild-type PKD1 [PDK1(WT)], the dominant-active PKD1 variants PKD1 (PH-Mut) or PKD1 (Zn-Mut) or the kinase-dead mutant PKD1 (KD) were transiently expressed in Jurkat T cells. After 48 h, samples were split, and half of the cells were stimulated for 3 min by anti-CD3 (CD3) TCR ligation. Cells were lysed, and HPK1 protein was immunopurified and subjected to an in vitro kinase reaction. HPK1 activity was determined in the presence of a recombinant GST-c-JunN fusion protein as a pseudosubstrate. The activation values describe the increases in HPK1 kinase activity in the presence (+) or absence (–) of CD3 ligation. The experiment was repeated three times with similar results. (B) 293T cells were transfected with an HPK1-HA expression plasmid and either empty vector or a PKD1 antisense (PKD1 a.s.) plasmid at a 1:1 or 1:2 ratio. The activity of HPK1 protein was determined as described in the legend to Fig. 7B. Immunoprecipitated HPK1-HA protein (IP) was assessed by anti-HA Western blotting. PKD levels of expression and protein content of the lysates were visualized by anti-PKD and antitubulin Western blotting. We show one of three experiments; the three experiments had the same outcome. (C) Mouse DC27.1 T cells were transfected with empty vector or a PKD1 antisense (PKD1 a.s.) plasmid. Forty-eight hours after transfection, cells were stimulated for 3 min with anti-CD3 antibody (+) or left unstimulated (–). Endogenous HPK1 proteins were immunoprecipitated and tested for their ability to autophosphorylate in an in vitro kinase assay. HPK1 protein present in the immunoprecipitation (IP) was controlled by anti-HPK1 Western blotting. The levels of expression in the protein lysate were visualized by anti-PKD1 and antitubulin Western blotting. One of three experiments is shown; the three experiments had similar results. (D) HPK1, wild-type PKD1 [PKD1(WT)], or kinase-dead PKD1 (KD) were transiently expressed in Cos1 cells as indicated. The activity of coexpressed SAPK/JNK protein was determined as described in the legend to Fig. 8A. The levels of expression were visualized by anti-SAPK, anti-HPK1, or anti-PKD1 Western blotting. Bars represent the averages of duplicate transfection experiments. A representative experiment is depicted. The two experiments had similar results. (E) Cos1 cells were transfected with increasing amounts of HPK1 expression plasmid (1, 2, and 4 µg) and either a constant amount of constitutively active PKD1 (PH-Mut) or empty vector in the presence of a 8x NF-B double reporter system. Analysis was done as described in the legend to Fig. 8B. The levels of expression were visualized by anti-HPK1 or anti-PKD1 Western blotting. The experiment was repeated twice with identical results.

We then took advantage of the mouse T-cell hybridoma DC27.1 to analyze the role of PKD1 in the activation of endogenous HPK1 in T cells. TCR ligation mediated by anti-CD3 antibody reproducibly induced endogenous HPK1 activity, as determined by in vitro kinase assays (Fig. 12C, top blot). Again, down-modulation of PKD1 protein after transfection of the PKD1 antisense construct resulted in a clear reduction of HPK1 activation, providing independent confirmation for the importance of PKD1 in TCR-dependent activation of HPK1 (Fig. 12C).

Finally, we wanted to address possible consequences of the functional interaction between PKD1 and HPK1 for the activation of downstream signaling pathways. Upon forced expression in epithelial cells, we found that the constitutively active PKD1 (PH-Mut) mutant moderately activates an AP-1-dependent reporter gene, while expression of HPK1 alone had no effect on basal AP-1 activity. Coexpression of the PKD1 (PH-Mut) mutant and HPK1 did not result in enhanced PKD-induced AP-1 activation but instead led to mild repression (not shown).

In contrast, wild-type PKD1 profoundly stimulated HPK1-mediated SAPK activation, while the kinase-dead PKD1 (KD) mutant exerted a suppressive effect (Fig. 12D). The stimulatory effect of wild-type PKD1 was most likely mediated by HPK1, as PKD1 by itself failed to elicit any SAPK/JNK activation. Concomitantly, the constitutively active PKD1 (PH-Mut) mutant showed no NF-B activation, while HPK1-driven NF-B activation was augmented in the presence of the PKD1 (PH-Mut) mutant (Fig. 12E).

These results demonstrate a nonoverlapping pathway specificity for HPK1 and PKD1 downstream signaling when analyzed independently. However, if coexpressed, activation of HPK1 is significantly augmented by the presence of kinase-competent PKD1, leading to a stronger, more robust signal input into the effector SAPK/JNK and NF-B pathways.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

 
Activation of serine/threonine kinases is readily detected after immunoreceptor stimulation at various developmental stages. However, the biological consequences of serine/threonine kinase action have been less apparent, possibly due to the existence of multiple isoforms and overlapping expression patterns in the relevant kinase families that defy simple genetic strategies, such as the elimination of single genes. The biological function of the subfamily I GCKs, HPK1, GCK, GCKR/KHS, and GLK, has remained particularly enigmatic. Here we describe a study in which we investigated in precise detail the mechanism of HPK1 activation after TCR ligation.

Several SH2 and SH3 domain-containing adaptor proteins have been reported to be HPK1 interaction partners. Some of these molecules are recruited to the plasma membrane upon lymphocyte activation; however, the localization of HPK1 in resting and activated cells has not been described so far. We determined the intracellular localization of HA-tagged full-length HPK1 and an HPK1 fusion protein in which the kinase domain had been replaced by EGFP in Cos1 cells. Both proteins were located predominantly in the cytoplasm but accumulated in multiple clusters in cells expressing intermediate levels of protein that appeared to coalesce into large aggregates in cells expressing high levels of protein. The large overlap in localization of both HPK1 forms suggests that motifs defining HPK1 intracellular distribution reside outside the kinase domain. The strong clustering of HPK1 is likely a consequence of the forced overexpression and without a direct physiological counterpart, yet it is likely to be the basis for the constitutive activation observed under these conditions.

Under more physiological conditions in lymphocytes, HPK1 activation has been reported to depend on Src family kinases and SLP-76 family adaptors (38). Therefore, we investigated the effects of these proteins on HPK1 localization. Surprisingly, coexpression of Clnk had little effect, while Lck efficiently recruited HPK1 to the plasma membrane. When expressed together, all three proteins displayed a large degree of colocalization. Exactly how Lck recruited HPK1 to the membrane remains unclear; a possible mechanism could involve SH3 domain-containing adaptor proteins related to the Grb2 family or HS1 (18, 30). Our results indicate that the membrane-proximal localization is further stabilized by SLP-76 adaptors that are needed for efficient HPK1 tyrosine phosphorylation in epithelial cells and activation in lymphocytes. Of note, the localization of a HPK1(Y379F) mutant that is incapable of binding the Clnk SH2 domain was comparable to that of wild-type HPK1. This finding suggests that neither Lck-mediated localization nor a basal interaction with Clnk depends exclusively on the SH2 domain binding to HPK1.

Finally, we were able to demonstrate recruitment of HPK1 to the clustered immunoreceptor in APC-T-cell conjugates and Jurkat cells (I. M. Patzak, unpublished data). Given that immunoreceptor engagement also results in HPK1 activation, it was tempting to speculate that clustering of HPK1 is an essential prerequisite for its enzymatic activation inasmuch as it makes subsequent auto- and transphosphorylation reactions possible by providing a sufficient local concentration of substrate and enzyme.

Activation of a large number of RD kinases has been demonstrated to depend on phosphorylation of activation loop amino acids (14). Employing alanine mutants, we identified three amino acids in the HPK1 activation segment that are indispensable for full kinase activity and effector pathway stimulation. While mutation of threonine 175 apparently introduced a strong structural perturbation, threonine 165 and serine 171 were characterized as bona fide phosphorylation sites. Surprisingly, phosphorylation of these two sites was interdependent, phosphorylation of serine 171 appeared to be a prerequisite for phosphorylation of threonine 165 and HPK1 appeared not to be the predominant serine 171 kinase. On the basis of its sequence context, we identified serine 171 as a possible PKD phosphorylation site and demonstrated a strong positive influence of PKD1 on HPK1 kinase activity and downstream effector activation.

Therefore, we propose a multistep model for HPK1 activation. Initially, HPK1 is relocalized to the plasma membrane such that activation loop trans- and autophosphorylation steps can take place during the ensuing second phase of activation. Full activity necessitates phosphorylation of threonine 165 and serine 171, with phosphorylation of threonine 165 being dependent on phosphoserine 171. To our knowledge, this is the first identification of a priming phosphorylation in a RD kinase activation segment. We and others have previously reported the puzzling observation that forced expression in epithelial cells renders subfamily I GCKs constitutively active, a property that has significantly hindered the elucidation of upstream activators and biological function. Clustering induced by ectopic overexpression likely alleviates the need for plasma membrane relocalization, thereby allowing activation loop phosphorylation to occur in the absence of upstream input. Could this ectopic activation of HPK1 involve PKD1? A significant proportion of the observed HPK1 fusion protein clusters appeared near or at the Golgi complex (F. Kiefer and B. Neuhaus, unpublished data). PKD1 and related family members (PKD2 and PKD3) are ubiquitously expressed and are also present in epithelial cells. In addition, PKD1 has been demonstrated to localize to the Golgi complex where it is part of a vesicle-budding machinery that assures traffic through the trans-Golgi network (39). Hence, upon forced expression, PKD1 might encounter, colocalize, and transphosphorylate HPK1 at cellular locations other than the native environment of immunoreceptors.

A mechanism for activation of the MAP3 kinase MEKK1 by the Ste20 homologue GCK has been proposed in which membrane recruitment of GCK/MEKK1 heterodimers by TRAF2 creates a higher-order aggregate that activates MEKK1 (5). This mechanism implies GCK aggregation during MEKK1 activation, although GCK oligomerization per se has so far not been demonstrated. MEKK1 activation clearly involves activation loop phosphorylation; however, s activation loop phosphorylation may be entirely self catalyzed or may involve GCK transphosphorylation (5). A rather unusual mode of regulation has recently been reported for GCK itself (43). In this study, GCK was shown to be subject to constant turnover by the proteasome; however, Toll-like receptor agonists caused a TRAF6-dependent stabilization of constitutively active GCK. Therefore, the apparent receptor-mediated kinase activation was the result of protein accumulation rather than modulation of enzymatic activity (43). In contrast to the reported stabilization of GCK, we observed a distinct down-modulation of HPK1 protein after induction of synapse formation, suggesting that there may be very different or possibly overlapping modes of regulation of subfamily I GCKs.

Activating autophosphorylation sites for subfamily I GCKs have not been identified so far, but phosphorylation of the archetypical Saccharomyces cerevisiae Ste20p at a threonine residue equivalent to HPK1 threonine 175 has been described (40). Interestingly, a threonine residue is found at a corresponding position in many kinases (Fig. 7A), and alanine mutation results in a strong reduction of kinase activity, but phosphorylation of this most C-terminal activation loop threonine has been seen in only a few kinases, including cdc2, Rsk1, and MEKK1 (37). Indeed, for MEKK1, conflicting data have been reported by Deak and Templeton (7); they failed to detect phosphorylation of this residue and instead proposed a structural function as we do here for HPK1 threonine 175. Phosphorylation of a serine or threonine residue corresponding to HPK1 serine 171 has been reported for numerous RD kinases. Uniquely, we found that this site in HPK1 was not an autophosphorylation site but a transphosphorylation site that provides a priming phosphate for threonine 165 autophosphorylation. Somewhat reminiscent, serine 281 of the MAP3K MLK3 has been identified as a site of HPK1 transphosphorylation, while threonine 277 and serine 281 of MLK3 were found to be autophosphorylation sites (21). Furthermore, many kinases of the AGC family (cyclic AMP-dependent protein kinase/protein kinase G/protein kinase C extended family) depend on activation segment phosphorylation by the master kinase PDK1 (32). For instance, PKB/AKT is phosphorylated by PDK1 on threonine 309, which corresponds to HPK1 serine 171, and in this case, phosphorylation also depends on both kinases colocalizing to the cell membrane. Interestingly, monophosphorylated PKB/AKT acquires only 10% of maximal activity, and full activation depends on phosphorylation of a conserved C-terminal motif outside the kinase domain (3, 33). Whether phosphorylation of C-terminal residues contributes to kinase activation in HPK1 remains to be established.

The identified phosphoacceptor sites are conserved in the activation loops of all subfamily I GCKs. Interestingly, the PKD1 target site corresponding to S171 is conserved in all subfamily I GCKs (Fig. 11D); hence, PKDs could act as master regulators for this kinase family. We found that PKD1 enhanced the extent of HPK1 inducibility after CD3-mediated TCR ligation in Jurkat cells, while constitutively active forms of PKD1 generally augmented HPK1 kinase activity. As a result of the more robust signal input, we observed n enhanced stimulation of the effector SAPK and NF-B pathways. Coexpression of the kinase-dead PKD1 (KD) mutant did not inhibit HPK1 activation after TCR cross-linking. In coprecipitation studies, we found no evidence for a stable, direct interaction between HPK1 and PKD1 (I. M. Patzak, unpublished data), which does not contradict the notion that HPK1 is a PKD substrate, if both kinases colocalize with other molecules. Consequently, a kinase-dead version of PKD1 does not necessarily act in a dominant-negative fashion; instead, its function to provide a priming HPK1 phosphorylation could be taken over by other PKD family members expressed in Jurkat cells, such as PKD2. We obtained independent support for this assumption when we down-regulated endogenous PKD levels in 293T and DC27.1 cells, which resulted in reduced HPK1 activation levels in both cell