Direct Interaction in T-cells between θPKC and the Tyrosine Kinase p59fyn*

The protein kinase C (PKC) family has been clearly implicated in T-cell activation as have several nonreceptor protein-tyrosine kinases associated with the T-cell receptor, including p59fyn. This report demonstrates that θPKC and p59fyn specifically interact in vitro, in the yeast two-hybrid system, and in T-cells. Further indications of direct interaction are that p59fyn potentiates θPKC catalytic activity and that θPKC is a substrate for tyrosine phosphorylation by p59fyn. This interaction may account for the localization of θPKC following T-cell activation, pharmacological disruption of which results in specific cell-signaling defects. The demonstration of a physical interaction between a PKC and a protein-tyrosine kinase expands the class of PKC-anchoring proteins (receptors for activated C kinases (RACKs)) and demonstrates a direct connection between these two major T-cell-signaling pathways.

From Telik, Inc., South San Francisco, California 94080 The protein kinase C (PKC) family has been clearly implicated in T-cell activation as have several nonreceptor protein-tyrosine kinases associated with the T-cell receptor, including p59fyn. This report demonstrates that PKC and p59fyn specifically interact in vitro, in the yeast two-hybrid system, and in T-cells. Further indications of direct interaction are that p59fyn potentiates PKC catalytic activity and that PKC is a substrate for tyrosine phosphorylation by p59fyn. This interaction may account for the localization of PKC following Tcell activation, pharmacological disruption of which results in specific cell-signaling defects. The demonstration of a physical interaction between a PKC and a protein-tyrosine kinase expands the class of PKC-anchoring proteins (receptors for activated C kinases (RACKs)) and demonstrates a direct connection between these two major T-cell-signaling pathways.
Signal transduction networks comprise parallel pathways that achieve integrated responses at multiple levels including second messenger accumulation and activation of kinases and transcription factors. Two major pathways involve nonreceptor protein-tyrosine kinases (1) and the protein kinase C (PKC) 1 family of serine/threonine kinases (2). Direct interaction between the ⑀ (3) and (4) members of the PKC family and particular protein-tyrosine kinase signaling proteins have recently been described. These reports are of interest because precedents for direct integration of serine/threonine and tyrosine kinase pathways by protein-protein interactions are rare, with perhaps the most prominent being identification of ras as a mediator between receptor tyrosine kinases and the MAPK pathway of serine/threonine kinases (5).
T-cell receptor (TCR) stimulation leads to hydrolysis of phosphatidylinositol-4,5-bisphosphate to produce diacylglycerol (DAG) and inositol-3,4,5-trisphosphate (6). These result, respectively, in direct activation of PKC and in increased intracellular calcium, which can then influence PKC activation among its other effects. Use of activators and inhibitors has established that PKC activation in T-cells is necessary but not sufficient for proliferation and production of characteristic cytokines (7). Furthermore, in the murine thymoma line EL-4, transiently overexpressing PKC, the PKC activator phorbol-12-myristate-13-acetate (PMA) increased transcription of a reporter gene regulated by the IL-2 promoter (8). Although Tcells express 10 of the 11 PKC isozymes (9), PKC is of special interest because it is the most nearly T-cell-restricted isozyme in its expression (10). In addition, activation of T-cells results in PKC translocation to the zone of TCR clustering present at the contact between the T-cell and an antigen-presenting cell (APC) (11).
PKC-mediated events can operate in a manner largely independent of other signaling pathways (12). A prominent parallel pathway following stimulation of the TCR involves activation of a tyrosine kinase signaling cascade (6). Among the proteintyrosine kinases that have been described as reversibly associated with the TCR complex are src family members p59fyn and p56lck and syk family member ZAP-70 (6). p59fyn is notable for its low fractional stoichiometry of association with the TCR CD3 complex, which suggests a transient modulatory role, and for its associations with a wide range of other proteins implicated in T-cell signaling including ZAP-70 (13), c-cbl (14), fas (15), shc (16), IL-7 receptor (17), CD43 (18), ␣-tubulin (19), inositol-3,4,5-trisphosphate kinase (20) and the inositol-3,4,5trisphosphate receptor (21). These interactions, mediated primarily through SH2 and SH3 domains, can be short lived. In the case of ZAP-70, for example, the association with p59fyn in a T-cell hybridoma was found as early as 10 s after activation, peaked at 5 min, and was gone by 10 min (13). Although 10 min is much longer than the interaction between a typical enzyme and substrate, it is short compared with the time course for expression of mitogenic cytokines, implying that a variety of other interactions are also likely to be involved in the regulation of cytokine secretion.
T-cells express one of the two splicing variants of p59fyn, the other being found primarily in brain (22). The enzyme is posttranslationally modified by N-terminal myristylation or palmitoylation, which is thought to facilitate membrane localization (23). Although clearly implicated in T-cell signaling by biochemical and genetic experiments using transgenic mice (24), the precise physiological functions of p59fyn remain unclear. It is believed to be involved in TCR-induced calcium release from intracellular stores mediated by the inositol-3,4,5-trisphosphate receptor (25), a process thought to be regulated in part by PKC as well. In other hematopoietic cells, p59fyn is associated with Btk (26) and with WASp (27); mutations in each have been implicated in human immune disorders.
Recent progress in the understanding of PKC function has focused on the fact that individual isozymes translocate from one cell compartment to other subcellular sites following physiological stimulation, often manifested as a shift in distribution * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 415-567-6815; Fax: 415-567-6819; E-mail: kauvarLM@earthlink.net. 1 The abbreviations used are: PKC, protein kinase C; TCR, T-cell receptor; DAG, diacylglycerol; IL-2, interleukin 2; RACK, receptor for activated C kinase; PMA, phorbol-12-myristate-13-acetate; PS, phosphatidylserine; PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; MBP, maltose-binding protein; PBS, phosphatebuffered saline; GST, glutathione S-transferase; BSA, bovine serum albumin; TBST, Tris-buffered saline Tween; IFN, interferon. from the soluble fraction to the particulate fraction. This phenomenon is believed to reflect a conformational change in PKC leading to specific interaction with particular anchoring proteins, designated receptors for activated C kinase (RACKs) (28). Restricting localization to defined sites within the cell represents a general strategy (29) that serves to limit the substrates exposed to the catalytic site of each PKC isozyme, providing physiological specificity for a family of enzymes that can phosphorylate a wide range of proteins in vitro. Two such anchoring proteins have been described in detail, RACK1, specific for ␤PKC (30), and RACK2, specific for ⑀PKC (31). Each is found in the particulate fraction and shows saturable binding to its cognate PKC following stimulation. In addition to promoting localization, anchoring proteins can also potentiate PKC catalytic activity, presumably by stabilizing the active conformation (29). Both RACK1 and RACK2 have sequence homology to the WD-40 family of proteins previously implicated in signaling (32), and in both cases, the interaction is centered on the regulatory domain of PKC, which has the most extensive sequence variation within the PKC family. Notably, neither RACK serves as a substrate for PKC itself (30,31). However, other PKC-binding proteins, several of which are implicated in regulating membrane-cytoskeleton interactions, have been shown to be substrates (33). Of particular relevance to the present work, the pleckstrin homology domains of Tec family protein-tyrosine kinases Btk and Emt have been suggested as possible anchors (3). The more distantly related atypical class PKC has been shown to both bind and phosphorylate ZIP, a protein thought to provide a scaffold-linking PKC to cytokine receptor tyrosine kinases (4).
Given the pharmacologically established significance of PKC in general for the T-cell response (2) and the intracellular localization of PKC in close proximity to the T-cell receptor following activation (11), specific functions for endogenous anchoring proteins of PKC are likely. Support for this view comes from disruption of activation-induced translocation and anchoring of PKC in Jurkat T-cells by overexpression of either the human immunodeficiency virus protein Nef (34) or the signaling protein 14-3-3- (35). In the latter case, PKC-dependent activation of the IL-2 promoter was shown to be inhibited. Recent work using electroporated antibodies to disrupt PKC function in peripheral blood lymphocytes has further implicated PKC in early responses to T-cell activation, specifically up-regulation of the IL-2 receptor (36). In the present work, identification of p59fyn as an endogenous anchoring protein for PKC has provided a basis for experiments indicating a role for this interaction in the regulation of IL-4 in nontransformed T-cell lines, using both electroporated antibodies and a small organic compound as pharmacological probes.
Recombinant Proteins-Human PKC was PCR-amplified from Jurkat cell cDNA, cloned into the baculovirus transfer vector pBlue-BacHisB (Invitrogen), and confirmed by sequencing. The recombinant N-terminal His-tagged PKC was isolated from Sf-9 insect cells on a nickel chelation resin (Qiagen) according to the manufacturer's protocols. The resulting protein was 85% pure by silver-stained SDS-PAGE gel and was catalytically active for autophosphorylation and using histone IIIs as substrate.
PKC-V1 domain (amino acids 1-140) and the V1 domain of ␦PKC (amino acids 1-140) were PCR-amplified from Jurkat cell cDNA and expressed in XL1-Blue strain of Escherichia coli (Stratagene) using pQE-30 vector (Qiagen) with an N-terminal His tag for nickel chelation purification and modified to encode a C-terminal c-myc epitope for immunological detection.
Human p59fyn cDNA was PCR-amplified from human T-cell cDNA, cloned in-frame into pMAL-c2 expression vector to produce a fusion protein with maltose-binding protein (MBP), and the sequence was verified using an ABI373 sequencer (Applied Biosystems). Protein was purified on an amylose affinity column following the manufacturer's protocols (New England Biolabs). The resulting MBP-p59fyn fusion protein was 90% pure and was catalytically active as measured by autophosphorylation and by using enolase as substrate. Similarly, p59fyn cDNA was cloned into the pGEX-3X expression vector (Amersham Pharmacia Biotech) to generate GST fusion proteins and purified using the manufacturer's protocols. The resulting GST-p59fyn fusion protein was 90% pure and was also catalytically active.
In Vitro Kinase Assays-Recombinant MBP-p59fyn or MBP (2.5 g) was incubated with or without substrates, 2.5 g of PKC-V1, and/or 2 g of acid-denatured enolase, in 50 l of assay buffer (10 mM MnCl 2 , 40 mM Hepes buffer, pH 7.6, and 5 Ci of [␥-32 P]ATP) for 15 min at room temperature (37). Reactions were stopped by adding 12.5 l of 5ϫ SDS-PAGE sample buffer and boiling for 5 min. Samples were resolved by SDS-PAGE, transferred onto nitrocellulose, and exposed to x-ray film using an intensifying screen at Ϫ80°C. Parallel manipulations were used for PKC autophosphorylation or phosphorylation of histone using 50 ng of PKC in the absence or presence of 40 g/ml histone type IIIs (Sigma) with or without activating lipids (DAG and phosphatidylserine, 0.8 and 50 g/ml, respectively) and 10 g of GST-p59fyn, all in 20 mM MgCl 2 , 20 mM Tris-HCl, pH 7.5, 12 mM 2-mercaptoethanol, 20 M ATP, and 5 Ci [␥-32 P]ATP.
Microplate Binding Assay-Polystyrene 96-well plates (Corning) were coated with indicated amounts of purified MBP-p59fyn or MBP in PBS for 1 h at 37°C. The plates were washed with TBST buffer (20 mM Tris, pH 7.4, 150 mM NaCl, and 0.05% Tween 20), blocked with 1% BSA in PBS at room temperature for 1 h, and washed again with TBST. Recombinant PKC or V1 fragments of PKC or ␦PKC were added in a buffer containing 10 mM MnCl 2 , 150 mM NaCl, 40 mM Hepes, pH 7.6, 1 mM ATP, and 0.1% BSA and incubated 30 min with shaking at room temperature. The plates were then washed 5 times with TBST buffer followed by incubation with peroxidase-conjugated anti-PKC antibody in 1% BSA/PBS (Transduction Labs) for 1 h at room temperature. The wells were washed 5 times with TBST buffer and developed using the 3,3Ј,5,5Ј-tetramethylbenzidine peroxidase substrate system according to the manufacturer's directions (Kirkegaard and Perry Laboratories), followed by reading OD at 650 nm. Background binding of PKC constructs to MBP alone, Ͻ0.1 OD, was subtracted to yield the specific binding to MBP-p59fyn. No signal was seen in the absence of PKC.
Yeast Two-hybrid Analysis-The Saccharomyces cerevisiae strain L40 (MATa trp1 leu2 his3 LYS2::lexA HIS3 URA3::lexA-lacZ) and vectors were provided by S. Hollenberg (38). PKC and its V1 domain (amino acids 1-140) and the V1 domain of ␦PKC (amino acids 1-140) were PCR-amplified from the corresponding plasmid cDNA and cloned in-frame downstream of LexA DNA binding polypeptide in BTM116 vector to form the bait construct. DNA fragments encoding full-length and truncated variants of p59fyn were PCR-amplified from human T-cell cDNA and cloned in-frame downstream of VP16 transcription activation domain in pVP16 vector to form the prey construct. The full-length human p59fyn in which Lys-296 was mutated to Ala (K296A) was PCR-amplified from the wild type p59fyn cDNA. Fulllength, regulatory (amino acids 1-248), and kinase (amino acids 249 -509) domains of human p56lck were PCR-amplified from corresponding cDNA (39) and also cloned into pVP16 vector. Human (CD4ϩ T-cells, purified from the blood of normal donors) or murine (HT2 cell line) T-cell cDNA libraries were similarly prepared as prey, with Ͼ10 5 independent clones in each.
Interaction analysis between fusion bait and prey proteins in yeast was performed by first introducing the LexA fusion bait plasmids into L40 yeast using the standard lithium acetate method (40), testing for background ␤-galactosidase activity, and then transforming the resultant bait yeast strain with the prey to be tested. The transformants were plated to synthetic UTL medium lacking leucine and tryptophane to maintain selection for the prey and bait plasmids, respectively, and to synthetic THULL medium lacking histidine, leucine, tryptophane, uracil, and lysine to assay for protein-protein interaction. Colonies from the UTL plates were picked and plated on THULL medium grids to retest the interaction observed on primary THULL medium plates. About 25 independent transformants were analyzed for each bait-prey pair.
Yeast grown on THULL grids were assayed for ␤-galactosidase activity by a qualitative filter lift assay (40). For quantitative analysis, at least six independent transformants were pooled to generate mixed liquid cultures, which were grown in UTL medium overnight, then inoculated into fresh UTL or THULL media at 1:10 dilution and grown overnight at 30°C. A 600 was taken, and 100 -500 l of the culture was used for the assay, which was performed as described (40). Relative ␤-galactosidase activity was expressed in Miller units as A 420 ϫ 100)/ A 600 /min of incubation/(ml of yeast culture), with the very low background (no prey) subtracted. Absolute units varied with age of the yeast cultures and have not been normalized; each figure panel is from a single experiment.
Cells were generally cultured overnight in modified Yssel's media without IL-2 and IL-4 before stimulation. Typically, 5 ϫ 10 7 cells were washed with PBS, pelleted, and resuspended in 2 ϫ 5 ml warmed modified Yssel's media. Direct PKC stimulation was achieved by treatment with 10 nM PMA and the calcium ionophore ionomycin at 25 g/ml or by a combination of 20 nM PMA plus 1 g/ml phytohemagglutinin for 10 -15 min at 37°C followed by washing in 10 ml of PBS. Alternatively, cells were stimulated for 1 h using OKT3 (1 g/ml), an antibody to the CD3 complex of the TCR.
Immunoprecipitation Analyses-Cells were lysed by resolubilizing cell pellets in solubilization buffer ϩ 1% Triton X-100, followed by a 10-min incubation in an ice bath. Lysates were clarified by centrifugation at 12,000 ϫ g for 10 min at 4°C. Triton-soluble protein was quantified by BCA protein assay (Pierce). A minimum of 350 g of lysate protein was used for each immunoprecipitation, to which was added 2.5-5 g of monoclonal antibody to PKC (Transduction Laboratories), polyclonal antibody to p59fyn (Upstate Biotechnology), or control antibodies. After incubation overnight (16 h) at 4°C with end-overend rotation on a rotating platform, protein PLUS G/A-agarose (Oncogene) was added, and the mixture was incubated for an additional 20 min at 4°C with rotation. The agarose was then washed four times with immunoprecipitation buffer.
SDS-PAGE sample buffer was added to the washed immune complexes and boiled for 5 min. Samples were resolved on 8 or 10% SDS-PAGE gels and transferred to nitrocellulose. The membranes were incubated in blocking buffer (5% nonfat dry milk in PBS) for 1 h at room temperature, followed by overnight incubation at 4°C with primary antibodies, anti-PKC monoclonal antibodies (Transduction Laboratories), or anti-p59fyn polyclonal antibodies (Santa Cruz Biotechnology) diluted 1:500 in 1% BSA in PBS. Membranes were washed 3 times for 10 min in PBS, 0.05% Tween 20. Secondary antibody conjugated to peroxidase was added at a concentration of 1:1,000 in 1% BSA in PBS for 1 h at room temperature. The membranes were washed as above and developed for electrochemiluminescence detection as described by the manufacturer (Amersham Pharmacia Biotech).
Protein Expression Analyses-T-cells grown and stimulated as described above were incubated for 24 h in culture media, following which secreted cytokines were assayed using enzyme-linked immunosorbent assay reagents from R&D Systems or BIOSOURCE, following the manufacturer's protocols for detection of bound biotinylated antibodies with peroxidase-streptavidin (Pierce) using 3,3Ј,5,5Ј-tetramethylbenzidine substrate (Kirkegaard and Perry). Experimentation focused on cells that express IL-4 as well as ␥-IFN but not IL-2. Expression of surface antigens was measured 3-24 h after stimulation using fluorescently labeled antibodies (Pharmingen) and a FACSCalibur flow cytometer with CellQuest software (Becton Dickinson). Cells, ϳ5 ϫ 10 5 /assay, were suspended in PBS ϩ 1% FCS, 0.1% sodium azide and incubated with antibody at room temperature for 30 min. Cells were washed three times and read immediately or stored in 0.5% formaldehyde in the dark at 4°C for less than 5 days. An aliquot of TER14687 from frozen Me 2 SO stocks was added to the cells during stimulation. Antibodies to PKC and p59fyn (Santa Cruz Biotechnology) or to choline acetyl transferase (CAT) (Promega), as a control, were electroporated into the cells. A Bio-Rad Gene Pulser II with Capacitance Extender II was used at 350 V and 250 microfarads. 1 ϫ 10 7 cells were suspended in a sterile cuvette in ice-cold PBS with antibody for 10 min, then electroshocked and rested at room temperature for 15 min before stimulation. Under these conditions, Ͼ50% of cells were able to take up antibody from the media, without substantial cytotoxicity.

RESULTS
In Vitro Binding-The first variable region of novel class PKC isozymes had previously been shown to contain anchoring protein binding sites (41). Accordingly, initial experiments were designed using that domain as a biochemical bait to identify PKC-binding proteins. Sorbents were prepared from PKC-V1, comprising residues 1-140. Adsorbed and eluted proteins from Jurkat lymphoma cell extracts were characterized using antibodies to phosphotyrosine (Tyr(P)) and to known T-cell-signaling proteins. These results led to the preliminary identification of p59fyn as the most prominent adsorbed Tyr(P)-containing protein (Fig. 1). The other bands in lane 3 stained with anti-Tyr(P) antibody are attributable to leachedoff protein contaminants present in the PKC preparation used to make the sorbent (see lane 2) or to p59fyn degradation products (see lane 4). In control experiments using sorbents constructed from the V1 regions of ␦PKC and ⑀PKC, no band at the molecular weight of p59fyn was observed (not shown).
To further establish that PKC can bind to p59fyn, purified recombinant constructs were used in a 96-well microplate binding assay (Fig. 2). A solid phase capture format was used, with p59fyn immobilized as a fusion protein with MBP. Based on the concentration needed for half-maximal binding under these conditions, the affinity of the interaction was estimated to be 5 ϫ 10 -9 M -1 . Similar results were obtained using GST-p59fyn immobilized on nitrocellulose. An important feature of these results is that the binding of the PKC-V1 region was approximately equivalent to that seen with the full-length recombinant PKC, whereas the binding of the corresponding V1 region of the closely related novel class ␦PKC isozyme was substantially weaker (Fig. 2B). Binding was quite reproducible in immediate replicates, with interassay coefficient of variation below 10% and background values well below the specific binding. The only significant source of variability was dependence on the PKC-activating phospholipids, DAG and phosphatidylserine, possibly reflecting variation among the different batches of recombinant proteins in posttranslational modifications, e.g. phosphorylation or minor proteolysis, or in carryover of lipid activators from the baculovirus production system.
As described previously for other PKC isozymes in the presence of their anchoring proteins (28), the presence of p59fyn potentiated the kinase activity of PKC, increasing maximal phosphorylation of histone IIIs in an in vitro kinase assay (Fig.  3 lanes 7 and 8), with a more modest effect on autophosphoryl-ation (lanes 3 and 4). The figure shows representative results from three independent preparations of PKC. Increased phosphorylation in lanes 7 and 8 at the molecular weight of PKC is attributed to histone dimers.
Association in T cells-Following preliminary studies in Jurkat lymphoma cells, the binding of the two proteins was examined in nontransformed T-cells. As shown in Fig. 4A, an antibody against PKC co-precipitated p59fyn from a T-cell supernatant. Conversely, shown in Fig. 4B, an antibody to p59fyn co-precipitated PKC. Cell activation is known to result in segregation of some TCR-associated proteins into detergentinsoluble microdomains (29); p59fyn in particular was difficult to solubilize quantitatively, and therefore, no attempt was made to determine the stoichiometry of association in resting cells compared with activated cells. The specificity of the interaction between PKC and p59fyn in T-cells was established through a series of negative results. The interaction of ␤PKC with RACK1, originally described in cardiac myocytes (30), was confirmed in Jurkat cells by immunoprecipitation studies. By this technique, ␤PKC did not bind to p59fyn, and PKC did not bind to RACK1. Furthermore, neither inositol-3,4,5-trisphosphate kinase nor p56lck were immunoprecipitated with PKCspecific antibody under conditions where p59fyn was detected.
p59fyn Kinase Activity-When tested in vitro, PKC-V1 was a substrate for recombinant p59fyn fused to MBP (Fig. 5). Controls showed that the MBP portion was irrelevant to the activity (lanes [5][6][7][8]. Six tyrosines are present in PKC-V1, and one of these, Tyr-53, is conserved in ␦PKC-V1, which is also a substrate for MBP-p59fyn in vitro, although to a lesser extent than PKC-V1 (lanes 2 and 4). ␦PKC did not compete as strongly with the known p59fyn substrate, enolase, as did PKC-V1 (lanes 1 and 3 versus 9). Consistent with a role for phosphorylation in stabilizing the interaction, inclusion of ATP in the microplate binding assay improved the signal to noise ratio (not shown). Further, p59fyn catalytic activity was required for binding in the yeast two-hybrid constructs described below.
Yeast Two-hybrid Analysis-Additional evidence for a direct interaction, consistent with the immunoprecipitation results and in vitro recombinant protein binding data, was obtained using the yeast two-hybrid system. Fragments of PKC fused to lexA formed a DNA binding bait, and fragments of p59fyn fused to VP16 formed an RNA polymerase-activating prey, with ␤-galactosidase as the reporter gene turned on by bait binding to prey. Because the combination of full-length PKC and p59fyn was found to be toxic to the yeast, the experiments focused on the PKC-V1 region. The other major variable portion of the PKC regulatory domain, V3 comprising residues 282-385, produced a high background signal as bait even in the absence of a prey and was not analyzed further. Fig. 6 presents a summary of PKC-V1 binding to p59fyn.   autophosphorylation (lanes 1-4) and phosphorylation of histone IIIs (lanes [5][6][7][8] were assayed by incorporation of radiolabeled phosphate in the absence or the presence of GST-p59fyn, and in the absence or presence of DAG/phosphatidylserine (PS), as indicated. Equal amount of material was loaded in each SDS gel lane.
The strongest binding was to the kinase domain, although some binding to the regulatory domain was observed. When transfected into the same parental bait construct yeast strain, approximately equal amounts of the fusion proteins were produced for full-length p59fyn and for both the kinase and regulatory domains, as measured by Western blotting with anti-p59fyn antibodies. Thus, the reporter gene signal is attributable to variations in the affinity of the interacting bait and prey proteins.
Specificity of the interaction was explored in two respects. First, the V1 region of ␦PKC, which is the isozyme most closely related to PKC by sequence, showed no binding to p59fyn or its major fragments. Second, p56lck, which is closely related to p59fyn by sequence (39) and is also critical for the TCR-mediated signal transduction, showed reduced binding. The quantitative comparison of PKC-V1 binding to p59fyn and p56lck is shown in Fig. 6A. Unlike p59fyn, for which the kinase domain was as effective as the intact protein, the two major domains of p56lck were not nearly as effective as the whole protein. Furthermore, when p59fyn kinase and regulatory domain prey constructs were mixed into a murine T-cell cDNA library and screened using PKC-V1 as bait, the positive clones recovered were predominantly from p59fyn, consistent with the degree to which they had been spiked into the library. Likewise, screening a human cDNA library yielded several p59fyn clones but no p56lck clones.
Also shown in Fig. 6A is evidence that the catalytic function of p59fyn is necessary for binding to PKC-V1; the K296A p59fyn mutant, which has no kinase activity (42), did not bind to PKC-V1. Both wild type full-length p59fyn and the kinase domain constructs were catalytically active in the yeast cells as determined by anti-Tyr(P) Western blot analysis of total yeast proteins. In contrast, yeast cell extracts containing PKC-V1 bait alone or in combination with the K296A mutant had no detectable Tyr(P) by the same analysis.
The yeast two-hybrid system was further used to characterize TER14687, (Ϯ)-2-(N,N-dimethylaminomethyl)-1-indanone HCl, a compound identified by preliminary screening of an empirically diverse set of compounds (43) for activity in a Jurkat cell IL-2 production assay (35). In the yeast reporter assay, the compound was found to block the association between PKC-V1 and p59fyn (Fig. 6B). Because this compound undergoes a spontaneous elimination at neutral pH to form a more reactive compound, the active species probably results from reaction with a media component; the final active structure has not yet been identified. Whatever the ultimate structure of TER14687, however, its effects are apparently specific to the interaction of interest because it had no effect on a Tal1/E2A two-hybrid construct. In this control, an active transcription factor is reconstituted by noncovalent association of domains from two separate transcription factors. Moreover, when added to media at the 60 M concentration used in the two-hybrid experiments, TER14687 did not affect growth of normal yeast cells.
Physiological Role-Prior work had shown that disrupting PKC translocation by overexpressing a binding protein suppressed IL-2 production in Jurkats (35) and that PKC was one of the first PKC isozymes to be activated following T-cell stimulation (36). Under the conditions used here, PMA stimulation caused all isozymes studied to translocate from the soluble to particulate fraction in both Jurkat lymphoma cells and normal T-cell lines. These include closely related ␦PKC and ⑀PKC as well as the more distantly related ␤PKC. However, PKC was the only isozyme that translocated following physiological stimulation using the OKT3 antibody to CD3 (Fig. 7A). PKC translocation in nontransformed T-cells following OKT3 activation was blocked by TER14687 (Fig. 7B). No effect of the compound on other isozymes was observed in Jurkats or normal T-cells. Thus, a compound that had been found to specifically inhibit PKC binding to p59fyn was also found to cause a selective effect on PKC translocation. TER14687 therefore provides a useful tool for exploring the physiolgical consequences of PKC translocation.
OKT3 stimulation of normal human T-cells in the presence of 5 M TER14687 resulted in substantial suppression of the up-regulation of surface antigens normally associated with Tcell activation. The results shown in Fig. 8A for CD69 are representative of effects observed in parallel experiments examining CD25 and CD40L. Expression of the cytokines IL-4 and ␥-IFN was also measured for several cell lines that express these cytokines but not IL-2 upon stimulation. Culture supernatants collected 24 h after OKT3 stimulation were analyzed by enzyme-linked immunosorbent assay assays; tritiated thymidine uptake was used to assess the proliferation response. The representative example in Fig. 8 (panels B-D) shows that TER14687 caused a more pronounced suppression of IL-4 as compared with ␥-IFN. Considering all experiments together, there was a trend to reduced proliferation at higher doses of TER14687, possibly as a secondary consequence of the reduc- tion in secretion of IL-4, which acts as an autocrine growth factor. Reductions in ␥-IFN were correlated with reduced proliferation, whereas the more extreme reductions of IL-4 secretion were independent of effects on proliferation.
Specificity of the TER14687 results was enhanced by several negative results. At 5 M in the cell media, TER14687 had no effect on calcium flux, an early sequel to OKT3 activation that was blocked by 25 M piceatannol, a tyrosine kinase inhibitor. Cytotoxic effects of TER14687 were modest, below 5 M; the higher tolerance of yeast cells to TER14687 may be because of active export pumps, which are known to be readily inducible in yeast (44). At 50 M in vitro, the compound did not inhibit the kinase activity of either PKC or p59fyn.
Additional indications that the cytokine effects arose from the direct interaction between PKC and p59fyn were provided by exploratory experiments using antibodies to either PKC or p59fyn as pharmacological probes, introduced into the cells by electroporation. For these experiments, the cells were stimulated for 30 min using PMA and ionomycin, conditions that produced results comparable with OKT3 activation with regard to PKC translocation and cytokine secretion. Suppression of IL-4, with little effect on ␥-IFN, was seen with antibodies to either protein but not with a control antibody against choline acetyl transferase (CAT) (Fig. 9). As shown in the figure, these results were significantly more variable in replicate experiments than was the case with TER14687, possibly because of variable degradation of the antibody. Consistent with a progressive degradation of the antibody, it was found that doubling the length of time the cells were stimulated eliminated the IL-4 suppression effect of the antibodies electroporated into the cells before stimulation. OKT3 activation, which required 60 min of stimulation to increase cytokine secretion, was thus not feasible using antibodies as pharmacological probes. DISCUSSION The key experimental result that has emerged from the present study is that a specific protein-tyrosine kinase, p59fyn, interacts with a specific PKC isozyme, PKC, in a manner that goes well beyond the typical interaction of an enzyme with its substrate. Rather, the interaction is quite analogous to that previously described between other PKC isozymes and their anchoring RACKs (28). This result, together with other recent evidence expanding the range of direct PKC interactions with proteins involved in signaling pathways (3,4), thus has ramifications beyond T-cell biology.
Signal transduction networks are inherently nonlinear, replete with feedback loops, and convergent in several aspects. This complexity makes definitive proof for the role of any single component difficult to achieve and even more difficult for the interaction of two components. Accordingly, the credibility of p59fyn as an anchoring protein for PKC is fortified by being the common conclusion from three independent lines of experimentation. First, the two proteins interact in vitro, as measured by using affinity sorbents on cell extracts (Fig. 1) as well as by direct binding of recombinant proteins (Fig. 2). Furthermore, the presence of p59fyn potentiates the kinase activity of PKC on another substrate (Fig. 3), a property shared with other RACKs.
Second, the two proteins are associated in vivo, as detected by co-immunoprecipitation from cell extracts (Fig. 4). The use of nontransformed human T-cell lines, rather than immortalized tumor lines or genetically engineered overexpressing lines, enhances the credibility of this result. One possible divergence from the previous PKC/RACK literature concerns the dependence of the interaction on activating phospholipids. Lipid-derived second messengers, normally produced intracellularly upon cell activation, induce a change in conformation that both activates catalytic activity and enables RACK binding (2). The activation dependence observed for PKC binding to p59fyn was variable in the T-cells examined. The known segregation of TCR-associated proteins into detergent-insoluble domains following activation (29) precludes accurate quantitation of the interaction between PKC and p59fyn in Triton extracts, however. The issue could not be resolved from in vitro dependence on DAG for binding because removing cytokines from the cell media overnight consistently reduced the amount of particulate fraction PKC to some degree, suggesting that the cytokines needed to promote growth of the cells also caused a variable level of activation.
Third, the yeast two-hybrid system established that the interaction encompasses an extended region of p59fyn, with strong binding via the kinase domain of p59fyn (Fig. 6). This result is consistent with evidence that PKC is a substrate for tyrosine phosphorylation by p59fyn (Fig. 5). PKC obtained from T-cells by immunoprecipitation also contained phosphotyrosine, as determined by Western blot analysis using Tyr(P)specific antibody. The stable nature of the binding revealed by co-immunoprecipitation is unusual for enzyme interactions with substrates, although there is precedence for catalytic activity increasing stability of a binding event. Specifically, experiments with a kinase-dead mutant showed that p59fyn kinase activity is required for p59fyn binding to ZAP-70 (13), analogous to the present result showing that p59fyn must be catalytically active to interact with PKC-V1 in the yeast twohybrid format (Fig. 6A). Importantly, both two-hybrid and immunoprecipitation experiments showed strong specificity for the interaction compared with closely related proteins for both binding partners (␦PKC in place of PKC, and p56lck in place of p59fyn). These results do not preclude the possibility that additional cognate proteins may exist to which PKC binds, however.
It would be surprising if there were no physiological effects attributable to the interaction between these two proteins, each of which has individually been implicated in T-cell signaling. TER14687, which blocks their association in the yeast twohybrid system (Fig. 6B) and specifically prevents normal translocation of PKC in T-cells (Fig. 7) does indeed show specific suppressive effects on production of cytokines (Fig. 8B) as well as surface antigens characteristic of activated T-cells (Fig. 8A). Furthermore, a differential effect on expression of IL-4 compared with ␥-IFN was observed, extending prior work in Jurkat cells showing that overexpressed 14-3-3-protein is both a translocation inhibitor and IL-2 suppressor (35). The active species formed in cells from TER14687 is not known. The consistent findings using antibodies to either PKC or p59fyn as pharmacological probes (Fig. 9) strengthens the validity of the results, extending prior antibody electroporation work (36). Because the expression of ␥-IFN was 10-fold higher than IL-4 in the cell lines studied, the specificity may reflect a quantitative effect rather than a qualitative difference in the regulation of the two cytokines. Whatever the precise consequences are for the signal network, the available pharmacological data suggest that compounds disrupting the interaction of PKC and p59fyn could have therapeutic utility for suppression of IL-4, a key cytokine implicated in allergy (45).