DGKγ (diacylglycerol kinase γ) was reported to interact with β2-chimaerin, a GAP (GTPase-activating protein) for Rac, in response to epidermal growth factor. Here we found that PMA and H2O2 also induced the interaction of DGKγ with β2-chimaerin. It is noteworthy that simultaneous addition of PMA and H2O2 synergistically enhanced the interaction. In this case, PMA was replaceable by DAG (diacylglycerol). The β2-chimaerin translocation from the cytoplasm to the plasma membrane caused by PMA plus H2O2 was further enhanced by the expression of DGKγ. Moreover, DGKγ apparently enhanced the β2-chimaerin GAP activity upon cell stimulation with PMA. PMA was found to be mainly required for a conversion of β2-chimaerin into an active form. On the other hand, H2O2 was suggested to induce a release of Zn2+ from the C1 domain of β2-chimaerin. By stepwise deletion analysis, we demonstrated that the SH2 (Src homology 2) and C1 domains of β2-chimaerin interacted with the N-terminal half of catalytic region of DGKγ. Unexpectedly, the SH2 domain of β2-chimaerin contributes to the interaction independently of phosphotyrosine. Taken together, these results suggest that the functional link between DGKγ and β2-chimaerin has a broad significance in response to a wide range of cell stimuli. Our work offers a novel mechanism of protein–protein interaction, that is, the phosphotyrosine-independent interaction of the SH2 domain acting in co-operation with the C1 domain.
DGK (diacylglycerol kinase) phosphorylates DAG (diacylglycerol) to generate PA (phosphatidic acid) [1–3]. DAG is liberated from inositol phospholipids and other phospholipids by the action of phospholipase C and phospholipase D/PA phosphatase upon cell stimulation by growth factors, cytokines, lipopolysaccharide and other agonists [4–6]. DAG is well known to regulate a wide variety of cellular functions through binding to the cysteine-rich zinc-finger structures (the C1 domains) that were first found in PKC (protein kinase C) . In addition to conventional and novel PKCs, several signalling proteins have previously been found to be regulated by DAG, as exemplified by protein kinase D, Unc-13, the Rac-specific GAP (GTPase-activating protein) chimaerin and Ras guanyl-nucleotide-releasing protein, all of which contain one or two of the C1 domain(s) [5,7]. Moreover, PA, the reaction product of DGK, has also been reported to regulate a number of signalling proteins such as phosphatidylinositol-4-phosphate 5-kinase, Ras GAP, Raf-1 kinase, mTOR (mammalian target of rapamycin) and atypical PKC [1–3]. It is noteworthy that chimaerins are activated by, in addition to DAG, PA [8,9]. DGK is therefore thought to play roles not only in the downregulation of DAG signalling, but also in the production of another lipid mediator, PA. Because the cellular concentrations of these signalling lipids must be strictly regulated, their interconversion by DGK is likely to be one of the key processes in cellular signal transduction.
Mammalian DGK is known to exist as a large protein family consisting of ten isoenzymes classified into five subtypes according to their structural features [1–3]. These subfamilies can be characterized by the presence of a variety of regulatory domains of known and/or predicted functions. The type I DGKs presently consisting of α-, β- and γ-isozymes contain, in addition to a tandem repeat of the C1 domains, two sets of Ca2+-binding EF-hand motifs at their N-termini. The tissue- and cell-dependent expression patterns detected distinctively for these isozymes suggest that, even when they belong to the same subfamily, each member exerts differentiated functions in particular types of cells.
Rac1, together with RhoA and Cdc42, is a member of the Rho small GTPase family. The active form of Rac1 interacts with various effectors to initiate downstream signalling events that control cell morphology, actin dynamics, superoxide generation, migration, metastasis, gene expression, apoptosis and the cell cycle [10,11]. Rac GTPase is known to act as a molecular switch, which is regulated by the three groups of effector molecules: guanine nucleotide exchange factors, guanine nucleotide dissociation inhibitors and GAPs.
Although there is a significant redundancy in the specificities of Rho GAPs at least in vitro, chimaerin is known to be specific to Rac, and it does not accelerate GTP hydrolysis for either RhoA or Cdc42 in vitro and in vivo [7,12,13]. Chimaerin comprises a family of four isoforms (α1- or ‘n’-, α2-, β1- and β2-chimaerins), which are splice variants of the α- and β-chimaerin genes. The common structural feature among all chimaerins is, in addition to the GAP domain, the presence of the C1 domain [7,12]. Moreover, the main structural difference between the spliced variants is the presence of an SH2 (Src homology 2) domain at the N-termini of α2- and β2-chimaerins [14,15]. SH2 domains are generally presumed to be competent to bind phosphotyrosine-containing proteins , but physiological partners of the domain of β2-chimaerin remain unknown. Although these chimaerin Rac-GAPs are activated in response to several cell stimuli [7,12,13], their activation mechanisms are not fully understood.
We previously reported that DGKγ acted as an upstream suppressor of Rac1, resulting in the suppression of lamellipodium/membrane ruffle formation . Moreover, we have recently demonstrated that DGKγ transiently interacts with and activates β2-chimaerin in response to cell stimulation with EGF (epidermal growth factor) . To characterize the molecular interaction in more detail, we searched for other cell stimuli enhancing the interaction. In the present study, we found that PMA (or DAG) and H2O2 also induced the interaction of DGKγ with β2-chimaerin. It is noteworthy that simultaneous addition of PMA (or DAG) and H2O2 synergistically and persistently induced the interaction. Moreover, our results show that, in co-operation with the C1 domain, the SH2 domain of β2-chimaerin is responsible for interacting with DGKγ, whereas the interaction is unexpectedly independent of phosphotyrosine. Thus these proteins are suggested to interact with each other through a novel protein–protein interaction mechanism.
We generated anti-DGKγ polyclonal antibodies against the N-terminal portion (DGKγ-N)  and the C-terminal portion (DGKγ-C)  as described previously. Other antibodies were obtained from commercial sources as follows: anti-FLAG M2 antibody (Sigma–Aldrich, St. Louis, MO, U.S.A.), anti-c-Myc antibody (9E10; Roche, Indianapolis, IN, U.S.A.), anti-GFP [anti-(green fluorescent protein)] antibody (for Western blotting; B-2; Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.), anti-GFP polyclonal antibody (for immunoprecipitation, TaKaRa-Clontech, Tokyo, Japan) and anti-Rac1 antibody (23A8; Upstate Biotechnology, Lake Placid, NY, U.S.A.). These anti-GFP antibodies cross-react with CFP (cyan fluorescent protein). PMA and 1,2-dioctanoyl-sn-glycerol were purchased from Sigma–Aldrich. EGF and H2O2 were obtained from Wako Pure Chemical Industries (Osaka, Japan). The PKC inhibitor GF 109203X was purchased from Biomol (Plymouth Meeting, PA, U.S.A.). TPEN [N,N,N′,N′-tetrakis-(2-pyridylmethyl)ethylenediamine] and FluoZin-3 were obtained from Invitrogen (Tokyo, Japan).
Cell culture and transfection
COS-7 cells were maintained in Dulbecco's modified Eagle's medium (Sigma–Aldrich) containing 10% (v/v) fetal bovine serum at 37 °C in an atmosphere of 5% CO2. COS-7 cells (∼1×106 cells per 60-mm-diameter dish or ∼3×105 cells per 35-mm-diameter dish) were transiently transfected with plasmids using Effectene (Qiagen, Tokyo, Japan) according to the manufacturer's instructions. After 24 h of transfection, cells were serum-starved for 15 h in Dulbecco's modified Eagle's medium containing 0.1% BSA and used for further analysis.
pCMV-Tag3-DGKα, pCMV-Tag3-DGKγ, pEGFP-DGKγ, pEGFP-DGKγ-G494D, pECFP-α1-chimaerin, pECFP-α2-chimaerin, pECFP-β2-chimaerin, pDsRed-monomer-β2-chimaerin and p3xFLAG-β2-chimaerin were generated as described previously [18,21] (DsRed is a red fluorescent protein from the coral Discosoma and ECFP is enhanced CFP). cDNAs encoding amino acids 1–569, 1–396 and 1–259 of human DGKγ were amplified by PCR and inserted into the HindIII–XhoI site of pCMV-Tag3B (Stratagene, La Jolla, CA, U.S.A.) for N-terminal c-myc tagging. pEGFP-DGKγ-1-396 and pEGFP-DGKγ-425-791 were constructed by inserting PCR fragments encoding amino acids 1–396 and 425–791, respectively, of human DGKγ into the HindIII–SalI site of pEGFP-C3 (TaKaRa–Clontech) (EGFP is enhanced GFP). cDNAs encoding amino acids 145–468, 1–272, 1–205, 57–272, 57–144 and 206–272 of human β2-chimaerin were amplified by PCR and inserted into the EcoRI–SalI site of pECFP-C1 (TaKaRa–Clontech). Point mutants of β2-chimaerin (β2-chimaerin-L30A , β2-chimaerin-I132A , β2-chimaerin-R313G , β2-chimaerin-R66L  and β2-chimaerin-R83L ) were generated using the QuikChange® site-directed mutagenesis kit (Stratagene). The authenticity of the constructs was confirmed by DNA sequencing.
Immunoprecipitation and Western-blot analysis
COS-7 cells were transfected with plasmids as described in the Figure legends. After 15 h of serum starvation, cells were stimulated with PMA, H2O2 or a mixture of the two. COS-7 cells were lysed in 500 μl of ice-cold buffer A [50 mM Hepes/NaOH, pH 7.2, 150 mM NaCl, 5 mM MgCl2, 1 mM DTT (dithiothreitol), 1 mM PMSF, Complete™ protease-inhibitor mixture (Roche) and phosphatase inhibitor cocktail II (Sigma–Aldrich)] containing 1% (v/v) Nonidet P40. The mixture was centrifuged at 12000 g for 5 min at 4 °C to give cell lysates. The cell lysates were precleared by mixing with 1 μg of normal IgG and 20 μl of Protein A/G Plus–agarose (Santa Cruz Biotechnology) for 30 min at 4 °C. The precleared lysates were incubated for 1 h at 4 °C with 5 μg of anti-c-Myc, 2 μg of anti-FLAG, 2 μl of anti-GFP or 2 μg of anti-DGKγ-C antibody, followed by incubation with 20 μl of Protein A/G PLUS–agarose for further 1 h. The beads were washed four times with buffer containing 50 mM Hepes/NaOH, pH 7.2, 100 mM NaCl, 20 mM NaF, 5 mM MgCl2, 0.1% Triton X-100 and 10% (v/v) glycerol and then boiled in SDS sample buffer. Cell lysates and immunoprecipitants were separated by SDS/PAGE and transferred to a PVDF membrane (Bio-Rad Laboratories, Tokyo, Japan). The membrane was blocked with Block Ace (Dainippon Pharmaceutical, Tokyo, Japan) and incubated with anti-FLAG, anti-GFP, anti-c-Myc or anti-DGKγ-N antibody. The immunoreactive bands were visualized with horseradish-peroxidase-conjugated anti-mouse or rabbit IgG antibody (Jackson ImmunoResearch Laboratories, West Grove, PA, U.S.A.) and the ECL® (enhanced chemiluminescence) Western-blotting detection system (GE Healthcare Bio-Sciences, Piscataway, NJ, U.S.A.).
COS-7 cells grown on fibronectin-coated glass coverslips were transiently transfected with pDsRed-monomer-β2-chimaerin and either pEGFP or pEGFP-DGKγ. After 15 h of serum starvation, cells were treated with PMA, H2O2 or PMA plus H2O2. The cells were processed as described previously  and were examined using a confocal laser scanning microscope (Zeiss LSM 510; Carl Zeiss, Tokyo, Japan). To assess the extent of β2-chimaerin translocation to the plasma membrane, cells exhibiting translocation of DsRed–β2-chimaerin to the plasma membrane were counted. The extent of translocation was expressed as the percentage of cells exhibiting translocation.
COS-7 cells grown on 35-mm-diameter dishes were transfected with pECFP-β2-chimaerin and either pCMV-Tag3 or pCMV-Tag3-DGKγ. After 15 h of serum starvation, cells were incubated with PMA plus H2O2 for 30 min and then lysed by sonication (Branson sonifier 250; Branson Ultrasonics Co., Danbury, CT, U.S.A.; output control 2; duty cycle 50%; 1 min) in 200 μl of buffer A. Supernatant and pellet were separated by a centrifugation at 100000 g for 1 h at 4 °C. The soluble and particulate fractions represent supernatants and pellets resuspended in the original volume (200 μl) of the lysis buffer respectively. An equal volume of each fraction was subjected to SDS/PAGE. CFP–β2-chiamerin was detected using anti-GFP monoclonal antibody by Western blotting. The detected bands were quantified by densitometry using the software package ImageJ 1.34s (National Institutes of Health, Bethesda, MD, U.S.A.).
Affinity precipitation of activated Rac1
COS-7 cells grown on 60-mm-diameter dishes were transfected as described in the Figure legends. After 15 h of serum starvation, cells were preincubated with 5 μM GF 109203X, a pan-PKC inhibitor, for 30 min and then stimulated with 1 μM PMA for 30 min. The cells were lysed in 500 μl of ice-cold buffer B (25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 1% Nonidet P40 and 1 mM DTT) containing 5% glycerol, Complete™ protease-inhibitor mixture and 5 μg of GST (glutathione transferase)-fused PBD (p21-binding domain) of PAK (p21 activated kinase 1) (GST–PAK–PBD; Cytoskeleton, Denver, CO, U.S.A.). After centrifugation at 12000 g for 5 min at 4 °C, the supernatant was incubated with 20 μl of glutathione–Sepharose 4B beads (GE Healthcare Bio-Sciences) for 30 min at 4 °C. The beads were washed three times with 500 μl of ice-cold buffer B and then boiled in SDS sample buffer. Rac1 associated with GST–PAK–PBD was detected with anti-Rac1 antibody by Western blotting.
Expression and purification of GST-fusion proteins
Escherichia coli XL-1-Blue cells (Stratagene) transformed with pGEX-6P-1 (GE Healthcare Bio-Sciences) harbouring cDNA encoding β2-chimaerin or DGKγ were grown to an attenuance (D600) of 0.5. After addition of 0.1 mM isopropyl β-D-thiogalactoside, cells were cultured for 24 h at 20°C and then harvested by centrifugation. The cells were lysed by sonication in 50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, 1% Nonidet P40 and Complete™ protease-inhibitor mixture, and the debris was removed by centrifugation at 15000 g for 10 min at 4 °C. GST-fusion proteins in the supernatant were affinity-purified using glutathione–Sepharose 4B and dialysed against 10 mM Tris/HCl, pH 7.4, 150 mM NaCl for 15 h at 4 °C. The purified proteins were stored at −80°C until used.
In vitro determination of zinc concentration
The amounts of Zn2+ released from purified recombinant proteins were determined using FluoZin-3 as described previously . GST-tagged β2-chimaerin (1 μM) or GST-tagged DGKγ (0.5 μM) was incubated with 1 μM PMA, 2 mM H2O2 or 1 μM PMA plus 2 mM H2O2 in 100 μl of 10 mM Tris/HCl, pH 7.4 and 150 mM NaCl at 37 °C for 30 min, and FluoZin-3 was then added to the solution to a final concentration of 5 μM. A complete release of Zn2+ from proteins was achieved by boiling for 5 min in the presence of 0.1% SDS. The fluorescence of the Zn2+–FluoZin-3 complex was measured with excitation at 485 nm (bandwidth 20 nm) and emission at 528 nm (bandwidth 20 nm) using a plate reader (Synergy HT Multi-Detection Microplate Reader; Bio-Tek, Winooski, VT, U.S.A.). ZnCl2 (Katayama Chemical Co., Osaka, Japan) was used as standard.
PMA and H2O2 synergistically induce an interaction between DGKγ and β2-chimaerin
In the present study, to characterize the interaction between DGKγ and β2-chimaerin in more detail, we searched for cell stimuli other than EGF augmenting the interaction. PMA, a potent analogue of DAG, has been reported to bind to the C1 domain of β2-chimaerin  and to induce subcellular redistribution of DGKγ  and β2-chimaerin  in a C1-domain-dependent manner. Therefore, we first looked at the effect of PMA on the interaction of DGKγ with β2-chimaerin using COS-7 cells expressing these proteins. In cells stimulated with 1 μM PMA for 30 min, β2-chimaerin was clearly co-precipitated when DGKγ was immunoprecipitated (Figure 1A). The band intensity of β2-chimaerin in the PMA-stimulated cells (1 μM PMA, 30 min) was ∼6-fold higher than that induced by EGF stimulation (100 ng/ml, 2 min) (results not shown). DAG, instead of PMA, also enhanced their interaction (see Supplementary Figure 1 at http://www.BiochemJ.org/bj/409/bj4090095add.htm). Since the subcellular localization and enzymatic activities of some DGK isoforms have been reported to be regulated by oxidative stress [28,29], we next examined effects of H2O2 on the interaction. As shown in Figure 1(A), H2O2 (2 mM, 30 min) also induced a clear interaction of DGKγ with β2-chimaerin. Intriguingly, we further found that simultaneous addition of H2O2 and PMA synergistically and intensely induced the interaction between DGKγ and β2-chimaerin (Figure 1A). Similarly, when DAG, instead of PMA, was added in combination with H2O2, the interaction was synergistically enhanced (Supplementary Figure 1). As expected, PMA, which is a non-metabolizable analogue of DAG, enhanced the interaction between DGKγ and β2-chimaerin more strongly than DAG. These results thus raise the possibility that H2O2 and PMA (or DAG) co-operatively regulate the interaction of DGKγ with β2-chimaerin.
Interaction of DGKγ with β2-chimaerin in response to PMA and H2O2
The interaction induced by PMA and H2O2 depended on time and reached a maximum by 30 min (Supplementary Figure 2A at http://www.BiochemJ.org/bj/409/bj4090095add.htm). Thus the interaction induced by PMA and H2O2 is much more sustainable than that by EGF, which reaches a maximum by 2 min after the stimulation and returned to the basal level within 10 min . In the presence of H2O2, PMA enhanced the interaction in a dose-dependent manner (EC50 75 nM; Supplementary Figure 2B). Likewise, stimulation with H2O2 also showed dose-dependent enhancement of the interaction in the presence of PMA (EC50 0.20 mM; Supplementary Figure 2C). Although we thoroughly examined more than ten cell lines, our antibody against β2-chimaerin , which could successfully detect overexpressed β2-chimaerin protein, failed, unfortunately, to detect the endogenous β2-chimaerin protein. Thus we could not analyse the interaction between endogenous DGKγ and β2-chimaerin.
We next examined the selectivity of DGKγ and β2-chimaerin in their interaction in response to PMA plus H2O2. DGKα, one of the type I DGKs, failed to interact with β2-chimaerin, as opposed to DGKγ (Figure 1B). For some unknown reason, the expression level of DGKβ was noticeably low in cells transfected with the cDNA. We thus could not analyse the interaction between DGKβ and β2-chimaerin. Like β2-chimaerin , α1- and α2-chimaerins [24,32], but not β1-chimaerin , are also expressed in the brain, with distribution patterns at least in part overlapping with those of DGKγ. However, α1-chimaerin showed no detectable interaction with DGKγ (Figure 1C). Moreover, α2-chimaerin, a closely related isoform, bound to DGKγ to a much lesser extent when compared with β2-chimaerin (Figure 1D). Taken together, these results indicate that DGKγ selectively interacts with β2-chimaerin in response to PMA and H2O2, consistent with the case of EGF stimulation, as reported previously .
β2-chimaerin is recruited by and is co-localized with DGKγ in response to PMA and H2O2
To determine whether DGKγ is co-localized with β2-chimaerin in response to PMA and H2O2, we used COS-7 cells expressing EGFP-tagged DGKγ and DsRed-monomer-tagged β2-chimaerin and assessed their subcellular distribution by confocal microscopy. Both DGKγ and β2-chimaerin were diffusely distributed throughout the cytoplasm and the nucleus in cells unstimulated and treated with H2O2 alone (Figure 2). Stimulation with PMA clearly induced translocation of DGKγ to the plasma membrane, whereas β2-chimaerin was mainly translocated to the perinuclear region, as has already been reported [13,27]. Only when cells were simultaneously stimulated with PMA and H2O2, did DsRed–β2-chimaerin, but not DsRed alone, exhibit a distinct translocation to the plasma membrane and co-localization with DGKγ.
Co-localization of DGKγ with β2-chimaerin in response to PMA and H2O2
Next, to assess the possibility that DGKγ recruits β2-chimaerin to the plasma membrane in response to PMA and H2O2, as observed for EGF stimulation , we determined the effects of DGKγ expression on the redistribution of β2-chimaerin. Upon cell stimulation with PMA plus H2O2, the number of cells exhibiting the translocation of β2-chimaerin to the plasma membrane in DGKγ-co-expressing cells was about twofold higher than the number of control vector-expressing cells (Figure 3A). This result supports the notion that DGKγ is, at least in part, capable of recruiting β2-chimaerin to the plasma membrane in response to PMA and H2O2.
DGKγ enhances PMA/H2O2-induced translocation of β2-chimaerin to the plasma membrane
To confirm the apparent recruitment of β2-chimaerin to the plasma membrane by DGKγ, in addition to the cell-biological method (i.e. confocal microscopy), we further studied the phenomenon using a biochemical method, i.e. cell fractionation. We separated cell homogenates into soluble and particulate fractions and quantified β2-chimaerin in these fractions. Addition of H2O2 and PMA markedly enhanced translocation of β2-chimaerin to the particulate fraction containing the plasma membranes (Figure 3B). Interestingly, the co-expression of DGKγ clearly enhanced the β2-chimaerin translocation about twofold at both 0.1 and 1 μM PMA in the presence of H2O2. These results further suggest that DGKγ can recruit β2-chimaerin to the plasma membrane in response to PMA plus H2O2.
DGKγ apparently enhances GAP activity of β2-chimaerin partially through its catalytic action in phorbol-ester-stimulated cells
Since DGKγ was suggested to enhance β2-chimaerin GAP activity in EGF-stimulated cells , we investigated whether DGKγ affected the GAP activity of β2-chimaerin in response to PMA and H2O2. However, for unknown reasons, stimulation with H2O2 considerably decreased intracellular Rac-GTP, thereby limiting the cell stimulation in response to PMA alone for the GAP assay. Therefore, COS-7 cells expressing β2-chimaerin and/or DGKγ were stimulated with 1 μM PMA and then the endogenous Rac-GTP levels were determined by GST–PAK–PBD pull-down assay. To exclude effects of phorbol-ester-responsive PKC on the intracellular Rac-GTP levels, we preincubated cells with 5 μM GF 109203X, a pan-PKC inhibitor . We confirmed that GF 109203X did not interfere with the interaction of DGKγ with β2-chimaerin induced by PMA (results not shown). In cells expressing β2-chimaerin, PMA decreased the intracellular Rac-GTP level by ∼40% (Figure 4A). Co-expression of wild-type DGKγ with β2-chimaerin further decreased the Rac-GTP level (by ∼70%) in PMA-stimulated cells. However, a kinase-dead mutant of DGKγ (DGKγ-G494D) affected the Rac-GTP level much less effectively than did wild-type DGKγ. On the other hand, in unstimulated cells, the expression of DGKγ did not affect the Rac-GTP level . We confirmed that DGKγ-G494D had a pattern of subcellular distribution similar to that of wild-type DGKγ in both unstimulated and PMA-stimulated cells (results not shown). In addition, DGKγ-G494D was shown to have the same interaction intensity with β2-chimaerin when compared with wild-type DGKγ (results not shown), indicating that this mutation does not affect protein folding. Thus it is likely that the regulation of β2-chimaerin Rac-GAP activity by DGKγ is dependent on the catalytic activity of this isoenzyme in PMA-stimulated cells.
Effects of overexpression of DGKγ and β2-chimaerin on Rac1-GTP levels in cells stimulated with PMA
We next attempted to confirm that the negative effect of DGKγ on Rac activity is indeed mediated by β2-chimaerin. When β2-chimaerin-R313G, a GAP-inactive mutant , was expressed, PMA failed to affect the Rac activity (Figure 4B). Moreover, co-expression of wild-type DGKγ with β2-chimaerin-R313G did not decrease the Rac-GTP level in cells stimulated with PMA. Taken together, these results suggest that DGKγ further augmented PMA-dependent activation of β2-chimaerin, at least in part, through its catalytic action. Since the GAP activity can be directly assessed only using purified proteins in an in vitro GAP activity assay, we attempted to analyse the effects of DGKγ on β2-chimaerin GAP activity in vitro. However, because bacterially expressed DGKγ exhibited quite low catalytic activity, we could not assess the positive effects of DGKγ on the GAP activity in vitro.
Open-formed mutant of β2-chimaerin does not require PMA to interact with DGKγ
To analyse the mechanism of the interaction of DGKγ with β2-chimaerin we used two open-formed mutants of β2-chimaerin, namely L30A and I132A (L28 and I130 as denoted by Canagarajah et al.  are referred to as L30 and I132, respectively in the present paper because we obtained an N-terminally extended 468-amino acid β2-chimaerin clone  instead of the 466-amino acid species ). Canagarajah et al. demonstrated that impaired intramolecular interactions in β2-chimaerin–L30A and – I132A lock them in a conformation that exposes the C1, SH2 and GAP domains . In unstimulated cells, these mutants did not significantly interact with DGKγ (Figure 5A). However, when cells were stimulated with H2O2 alone, DGKγ interacted with the L30A and I132A mutants more markedly than with the wild-type β2-chimaerin. Moreover, the addition of PMA enhanced the H2O2-induced interaction of DGKγ with the L30A and I132A mutants much less effectively than that with wild-type β2-chimaerin. Furthermore, PMA alone (without H2O2) hardly affected the interaction (results not shown). These results strongly suggest that PMA mainly induces a conversion from a ‘closed’ into an ‘open’ form of β2-chimaerin, thereby enhancing the interaction between β2-chimaerin and DGKγ.
Effects of H2O2 and a selective zinc-chelating reagent, TPEN, on the interaction between DGKγ and open-formed mutants of β2-chimaerin
A selective zinc-chelating reagent, TPEN, apparently mimics the effect of H2O2
We next attempted to analyse the role of H2O2. Because H2O2 is known to induce activation of Src-family tyrosine kinases in some cells , we first examined effects of a Src-family kinase inhibitor, PP2, on the interaction between DGKγ and β2-chimaerin. However, the interaction was not inhibited by PP2 (see Supplementary Figure 3 at http://www.BiochemJ.org/bj/409/bj4090095add.htm), indicating that the interaction is not mediated by an Src-family kinase. Noteworthily, H2O2 (0.1–2 mM) was previously reported to release Zn2+ from the cysteine-rich zinc-finger structures (the C1 domains) of PKCα, PKCδ, PKCζ and cRaf1, probably due to thiol oxidation of cysteine residues in the domain(s) . Since DGKγ and β2-chimaerin also possess the C1 domain(s), we thus hypothesized that H2O2 directly causes Zn2+ release from the domain(s) of DGKγ and/or β2-chimaerin. If the hypothesis is correct, a potent and selective zinc-chelating reagent, TPEN, which was indeed reported to liberate Zn2+ from the C1 domain as did H2O2 , would mimic the effect of H2O2. Therefore, we tested effects of TPEN on the interaction. In this experiment, to avoid effects of PMA on the C1 domains, the open-formed mutant of β2-chimaerin (I132A), which does not require the addition of PMA for the enhancement of the interaction (Figure 5A), was used. As shown in Figure 5(B), TPEN substantially enhanced the interaction between DGKγ and β2-chimaerin-I132A, apparently mimicking the effect of H2O2. However, TPEN did not significantly enhance the interaction induced by H2O2. On the other hand, when TPEN was preincubated with ZnCl2, this chelator did not enhance the interaction, confirming that the effect of TPEN was Zn2+-dependent.
To determine whether H2O2 indeed induces Zn2+ release from the C1 domain(s) of DGKγ and/or β2-chimaerin, we purified bacterially expressed GST–DGKγ and GST–β2-chimaerin. As shown in Figure 6(A), H2O2 released Zn2+ from the C1 domain of β2-chimaerin in vitro, whereas PMA had no such effect. Moreover, H2O2, but not PMA, directly librated Zn2+ from the C1 domains of DGKγ (Figure 6B). Taken together, these results support the notion that H2O2 directly provokes the Zn2+-release from the C1 domain(s) of DGKγ and/or β2-chimaerin, leading to their interaction. To clarify the PMA- and H2O2-dependent mechanisms of the interaction in more detail, we sought to establish an in vitro system. However, for some unknown reason, bacterially expressed DGKγ non-specifically bound to IgG and protein tags. Therefore we could not further analyse the mechanisms in vitro.
Effects of PMA and H2O2 on the release of Zn2+ from β2-chimaerin and DGKγ in vitro
The SH2 and C1 domains of β2-chimaerin are co-operatively responsible for the interaction with DGKγ
We next attempted to identify the region(s) of β2-chimaerin responsible for the interaction with DGKγ. COS-7 cells co-expressing DGKγ and various deletion mutants of β2-chimaerin (Figure 7A) were stimulated with PMA plus H2O2, and then DGKγ in cell lysates was immunoprecipitated. Deletion of either the SH2 domain or the C1 domain of β2-chimaerin abolished the interaction of β2-chimaerin with DGKγ, whereas deletion of the GAP domain did not affect the interaction (Figure 7B). Moreover, the SH2 domain alone and the C1 domain alone failed to interact with DGKγ. These results collectively indicate that both the SH2 and C1 domains of β2-chimaerin are necessary for the interaction with DGKγ in cells stimulated with PMA plus H2O2.
Both the SH2 and C1 domains of β2-chimaerin are required for the interaction with DGKγ in response to PMA and H2O2
The interaction between DGKγ and β2-chimaerin through the SH2 domain is independent of phosphotyrosine
SH2 domains are generally thought to interact with phosphotyrosine . Moreover, H2O2 was reported to induce activation of Src-family tyrosine kinases in some cells . However, we failed to detect tyrosine phosphorylation of DGKγ in cells stimulated with PMA plus H2O2 (results not shown). In this context, we recently revealed that, on cell stimulation with PMA and H2O2, β2-chimaerin was markedly tyrosine-phosphorylated by Src-family kinases , raising the possibility that the tyrosine phosphorylation may control the interaction with DGKγ. However, the interaction between DGKγ and β2-chimaerin was not inhibited by the Src-family kinase inhibitor PP2 (see Supplementary Figure 3). A protein tyrosine phosphatase inhibitor, vanadate, also failed to affect the level of the interaction (results not shown). Arg66 and Arg83 in the SH2 domain of β2-chimaerin correspond to the highly conserved αA2 and invariant βB5 arginine residues essential for phosphotyrosine binding respectively . Indeed, replacement of the corresponding arginine residues in the SH2 domain of α2-chimaerin with leucine (R56L and R73L) abolished the interaction with phosphotyrosine . In the next experiment, Arg66 and Arg83 were thus replaced by leucine in the SH2 domain of β2-chimaerin. However, these mutations (R66L and R83L) failed to inhibit the interaction (see Supplementary Figure 4 at http://www.BiochemJ.org/bj/409/bj4090095add.htm). Taken together, these results suggest that the SH2 domain of β2-chimaerin contributes to the interaction with DGKγ in a phosphotyrosine-independent manner.
The N-terminal half of catalytic region of DGKγ is mainly responsible for the interaction with β2-chimaerin
We next examined which region(s) of DGKγ is (are) responsible for the interaction with β2-chimaerin. First, we co-transfected plasmids encoding 3×FLAG-tagged β2-chimaerin and either the GFP-tagged N- or C-terminal half of DGKγ into COS-7 cells (Figure 8A). After cell stimulation with PMA plus H2O2, β2-chimaerin was immunoprecipitated. The C-terminal half of DGKγ (DGKγ-425–791) interacted with β2-chimaerin as strongly as the full-length DGKγ (Figure 8B). On the other hand, the interaction of the N-terminal half of DGKγ (DGKγ-1–396) with β2-chimaerin was markedly weaker than those of the full-length and the C-terminal half of DGKγ, indicating that the C-terminal half of DGKγ is mainly responsible for the interaction with β2-chimaerin in PMA/H2O2-stimulated cells. To further narrow down the region involved in the interaction, we used plasmids encoding DGKγ mutants generated by a stepwise deletion (Figure 8A). Deletion of the C-terminal half of the DGKγ catalytic domain hardly affected the interaction, whereas a complete deletion of the DGKγ catalytic domain markedly impaired the interaction (Figure 8C). Moreover, the deletion of both C1 and catalytic domains of DGKγ completely abolished the interaction. These results collectively indicate that although the C1 domain of DGKγ is partly involved in the interaction, the N-terminal half of the catalytic domain (amino acid 425–569) mainly contributes to the interaction.
The N-terminal half of DGKγ catalytic domain is mainly responsible for the interaction with β2-chimaerin in response to PMA and H2O2
Because the N-terminal half of the DGKγ catalytic domain, which lacks the C1 domains, mainly contributes to the interaction with β2-chimaerin (Figure 8), it is possible that the interaction enhanced by H2O2 is accounted for mostly by the release of zinc from the C1 domain of β2-chimaerin, but not from the C1 domains of DGKγ. To confirm this possibility, we co-transfected into COS-7 cells plasmids encoding β2-chimaerin-I132A and either the full-length DGKγ or DGKγ-425–791, which lacks the C1 domains (see Figure 8A), and then analysed the effects of H2O2 on the interaction between them. In this experiment, to avoid effects of PMA on the C1 domain of β2-chimaerin, the open-formed mutant of β2-chimaerin (I132A) was used again. H2O2 markedly enhanced the interaction between β2-chimaerin-I132A and the full-length DGKγ (Figure 9). Similarly, the interaction of DGKγ-425–791 with β2-chimaerin-I132A was significantly augmented. Although we cannot rule out the possibility that the C1 domains of DGKγ partly contribute to the interaction, the result indicates that the interaction enhanced by H2O2 can be accounted for mainly by the release of Zn2+ from the C1 domain of β2-chimaerin.
H2O2 markedly enhanced the interaction between β2-chimaerin and DGKγ-425–791 lacking the C1 domains
We previously reported that DGKγ functions as an upstream suppressor of Rac1 . Moreover, DGKγ was revealed to interact with β2-chimaerin in response to EGF . To characterize the interaction in more detail, we thus searched for other cell stimuli that enhance the interaction. In the present study we found that simultaneous addition of PMA (or DAG) and H2O2 synergistically and selectively induced the interaction between DGKγ and β2-chimaerin. Since, unfortunately, our antibody against β2-chimaerin failed to detect the endogenous β2-chimaerin protein, we could not analyse the interaction between endogenous DGKγ and β2-chimaerin. However, the interaction between DGKγ and β2-chimaerin is highly selective (Figures 1B–1D), suggesting that the interaction is physiologically significant. Generally, cells are not considered to be exposed to sustained stimulation with both DAG and H2O2. However, such conditions are probably fulfilled, for instance, in an inflammatory lesion. During a pathological event, a variety of cells and tissues, including phagocytes (e.g., granulocytes and macrophages) are probably exposed to high concentrations of reactive oxygen compounds, including H2O2, which are externally released by phagocytes themselves . In addition, monocyte/macrophage-like cells internally generate DAG in response to inflammatory stimulants such as tumour necrosis factor-α and lipopolysaccharides [39,40]. We confirmed that both of the DGKγ  and β2-chimaerin (results not shown) mRNAs are expressed in macrophages differentiated from HL-60 leukemia cells. Of interest, some malignant tumors generate DAG  as well as reactive oxygen species including H2O2  much more than normal cells. Thus, it is possible to speculate that the functional linkage between DGKγ and β2-chimaerin has a broad significance in response to a wide range of stimuli, not only under physiological conditions (growth factor stimulation ), but also in pathological conditions such as inflammation and the growth of tumours.
Using this model system, which induces the strong interaction between DGKγ and β2-chimaerin in a PMA/H2O2-dependent manner, we revealed several novel and interesting features of the interaction as described below. First, PMA enhances the interaction between DGKγ and β2-chimaerin through inducing a conversion from a closed into an open form of the GAP. PMA is known to directly bind to the C1 domain of β2-chimaerin, converting the GAP into an open form . The results obtained in Figure 5 indeed demonstrated that PMA mainly contributes to the interaction between DGKγ and β2-chimaerin by inducing the open-form conversion of the GAP. In this case, the conformational change induced by PMA probably exposes the SH2 and C1 domains, both of which are necessary for the binding to DGKγ (Figure 7) and are masked by the GAP domain in a closed form. Although the C1 domain alone of DGKγ was reported to bind to phorbol ester , the full-length DGKγ did not show the phorbol-ester-binding activity . Thus a direct modification of DGKγ by PMA seems likely to occur to a limited extent.
Secondly, H2O2 is suggested to enhance the interaction between DGKγ and β2-chimaerin via the induction of the release of Zn2+ from the C1 domain of the GAP isoform. In addition to PMA, H2O2 is also required to accomplish the maximum interaction between DGKγ and β2-chimaerin. How does H2O2 enhance the interaction? Interestingly, H2O2 (0.1–2 mM) was previously reported to release Zn2+ from the C1 domains of PKCα, PKCδ, PKCζ and cRaf1 . Hence, we hypothesized that H2O2 directly induces Zn2+ release from the C1 domain(s) of DGKγ and/or β2-chimaerin. Indeed, the potent and selective zinc-chelating reagent TPEN also augmented the interaction of DGKγ with β2-chimaerin, mimicking apparently the effect of H2O2 (Figure 5B). Moreover, H2O2, but not PMA, directly librated Zn2+ from the C1 domain(s) of DGKγ and β2-chimaerin (Figure 6). Furthermore, experiments using the deletion mutant of DGKγ (DGKγ-425–791) lacking the C1 domains showed that the interaction enhanced by H2O2 can be accounted for mainly by the release of zinc from the C1 domain of β2-chimaerin (Figure 9). These results are consistent with the above hypothesis. However, further studies will be necessary to more convincingly determine what role H2O2 plays in modulating the interaction of DGKγ with β2-chimaerin.
Thirdly, the C1 domain of β2-chimaerin is responsible for the interaction with DGKγ. The C1 domain of β2-chimaerin has already been reported to function as the PMA/DAG binding site . In the present study, we demonstrated that, in addition to this function, the C1 domain serves as a part of the DGKγ-association site (Figure 7). Tmp21-I (21 kDa transmembrane trafficking protein or p23) was also reported to interact with β2-chimaerin via the C1 domain of the GAP in response to PMA stimulation . In this case, Tmp21-I recruits β2-chimaerin to the Golgi apparatus. By contrast, on cell stimulation with PMA and H2O2, DGKγ enhanced the translocation of β2-chimaerin to the plasma membrane, where active Rac1 is generally thought to exist (Figures 2 and 3). Taken together, it is thus possible to speculate that DGKγ and Tmp21-I competitively direct β2-chimaerin to be distributed in distinct subcellular compartments in response to different stimuli.
Fourthly, the SH2 domain of β2-chimaerin is necessary for interaction with DGKγ, whereas the interaction is unexpectedly independent of phosphotyrosine. In contrast with the C1 domain, there has been no report describing a binding partner of the SH2 domain of β2-chimaerin so far. We, for the first time, found that the SH2 domain interacts with DGKγ (Figure 7). Intriguingly, this interaction probably co-operates with the C1 domain. The SH2 domain of α2-chimaerin was previously reported to interact with collapsin response mediator protein-2 . Thus, it now emerges that the β2- and α2-isoforms of chimaerin can be associated with different partners through the same domain. Four experimental results described below indicate that phosphotyrosine residues are not involved in the interaction through the SH2 domain. First, tyrosine phosphorylation of DGKγ was not detectable in cells stimulated with PMA plus H2O2 (results not shown). Secondly, the Src-family-kinase inhibitor PP2 did not inhibit the interaction between DGKγ and β2-chimaerin (Supplementary Figure 3). Thirdly, a protein tyrosine phosphatase inhibitor, vanadate, did not affect the interaction (results not shown). Finally, the R66L and R83L mutants of β2-chimaerin, which abolish the interaction of SH2 domains with phosphotyrosine [24,37], failed to inhibit the interaction between DGKγ and β2-chimaerin (Supplementary Figure 4). Interestingly, the SH2 domain of α2-chimaerin was also associated with its partner protein in a phosphotyrosine-independent manner . However, besides α2-chimaerin, only a few reports (e.g. [45,46]) describing the phosphotyrosine-independent interaction of SH2 domains are available at present. Moreover, their interaction mechanisms remain unclear. Hence, it will be interesting to determine how the SH2 domain of β2-chimaerin interacts with DGKγ or indirectly contributes to the interaction.
Finally, the N-terminal half of the catalytic region of DGKγ is mainly responsible for the interaction with β2-chimaerin (Figure 8). As for the catalytic region of DGK, it was merely reported that the C-terminal half of the catalytic domain in DGKζ binds to PKCα and Ras guanyl-nucleotide-releasing protein [47,48]. Therefore we, for the first time, demonstrate that the N-terminal half of the catalytic domain of DGK serves as a binding site in addition to the catalytic region. The primary structure of the N-terminal half of the DGKγ catalytic domain is highly similar to that of another type I isoenzyme, namely DGKα, which, by contrast, does not bind to β2-chimaerin (Figure 1B). However, sequence comparison showed that three short stretches (amino acids 466–493, 501–511 and 549–554) in this region of DGKγ are relatively distinct from those of DGKα. Thus these short areas may confer the different affinities for β2-chimaerin.
Because, for unknown reasons, stimulation with H2O2 considerably decreased intracellular Rac-GTP, the effects of DGKγ on β2-chimaerin action in the presence of H2O2 could not be investigated. Therefore we demonstrated, using cells stimulated only with PMA, that DGKγ apparently enhanced the β2-chimaerin GAP activity (Figure 4), as observed for EGF-stimulated cells . The effect of DGKγ on Rac inactivation by β2-chimaerin in response to PMA treatment alone (Figure 4A) is quite dramatic, despite the low levels of co-localization between the proteins, as observed in Figure 2. Because active Rac has been reported to be localized at the plasma membrane , it is thought that only β2-chimaerin existing at the plasma membrane, but not that in the cytoplasm, exerts its GAP activity. In the presence of PMA, DGKγ was recruited to the plasma membrane (Figure 2), where the functional β2-chimaerin was present. Therefore a possible explanation for the striking effect of DGKγ on Rac inactivation by β2-chimaerin (Figure 4A) is that DGKγ effectively interacts with, and exclusively activates, the functional β2-chimaerin at the plasma membrane. It is thus reasonable to speculate that the apparent activation of β2-chimaerin by DGKγ occurs also in cells stimulated with PMA plus H2O2, a situation where these proteins are associated with each other more tightly when compared with cells stimulated by PMA alone. The apparent activation of β2-chimaerin was partly dependent on the catalytic action of DGK (Figure 4A). As PA was reported to activate β2-chimaerin in vitro , PA produced by DGKγ may directly activate β2-chimaerin. How is DAG, the substrate of DGK, supplied in cells stimulated with PMA? PKC, which is positively regulated by PMA , activates phospholipase D . Thus DAG may be supplied via the phospholipase D/PA phosphatase pathway evoked by PMA-dependent PKC action. Alternatively, it is possible that DGKγ utilizes a pre-existing pool of DAG, as does DGKα . Unlike the situation with EGF stimulation , the kinase-dead mutant of DGKγ did not completely blunt the negative effect of wild-type DGKγ on the Rac1-GTP level (Figure 4A). Thus, in PMA-stimulated cells, DGKγ may partly activate β2-chimaerin in a DGK-activity-independent manner (for example by serving as an adaptor or scaffold protein).
Does cell stimulation with EGF and PMA plus H2O2 share the same mechanism inducing the interaction between DGKγ and β2-chimaerin? Noteworthily there are several features in common between the EGF- and PMA/H2O2-dependent interactions. First, the interactions induced by EGF and PMA/H2O2 showed the same selectivity. In both cases, DGKα did not bind to β2-chimaerin and, moreover, α1- and α2-chimaerins failed to associate with DGKγ (; Figure 1). Secondly, the interactions induced by EGF and PMA/H2O2 are phosphotyrosine-independent. Upon cell stimulation with EGF or PMA/H2O2, β2-chimaerin was markedly tyrosine-phosphorylated by Src-family kinases . However, the Src-family kinase inhibitor PP2, and the R66L and R83L mutations of the SH2 domain, failed to inhibit the interactions between DGKγ and β2-chimaerin induced by EGF (results not shown) and PMA/H2O2 (Supplementary Figures 3 and 4). Thirdly, EGF and PMA/H2O2 induced redistribution of both DGKγ and β2-chimaerin from the cytoplasm to the plasma membrane (; Figure 2). We further tried to determine the interaction regions of DGKγ and β2-chimaerin on cell stimulation with EGF. However, because their deletion mutants showed weak non-specific interactions and hindered the precise determination of interaction intensities, we could not obtain conclusive results. The interaction induced by PMA/H2O2 depended on time and reached a maximum by 30 min (Supplementary Figure 2A). Thus the interaction induced by PMA and H2O2 is much more sustainable than that by EGF, which reaches a maximum by 2 min . With regard to this discrepancy, we speculate that the interaction between DGKγ and β2-chimaerin induced by EGF is tightly regulated, whereas PMA and H2O2 overcome the negative regulation conducted through unknown mechanisms. It is noteworthy that EGF stimulation is known to produce H2O2  in addition to DAG , supporting the notion that EGF induces the interaction between DGKγ and β2-chimaerin through the production of H2O2 and DAG. However, future studies are needed to clarify this issue in more detail.
This work was supported in part by grants from the the Japanese Government Ministry of Education, Culture, Sports, Science and Technology.
cyan fluorescent protein
epidermal growth factor
green fluorescent protein
p21 activated kinase 1
protein kinase C
Src homology 2
21 kDa transmembrane trafficking protein (p23)