Formation of Ternary Complex of Human Biliverdin Reductase-Protein Kinase Cδ-ERK2 Protein Is Essential for ERK2-mediated Activation of Elk1 Protein, Nuclear Factor-κB, and Inducible Nitric-oxidase Synthase (iNOS)*

Background: ERK2 activation by PKCδ relays cell growth signals. hBVR is a bridge/scaffold protein and nuclear transporter of ERK. Results: hBVR forms a ternary complex with PKCδ and ERK2; this requires specific hBVR sequences. Corresponding peptides inhibit PKCδ/ERK2 interaction. PKCδ/ERK-mediated transcriptional activation is hBVR-dependent. Conclusion: hBVR is essential for ERK2 activation by PKCδ and MEK1/2. Significance: hBVR-based peptides are useful in regulating PKCδ/ERK signaling. Growth factors, insulin, oxidative stress, and cytokines activate ERK1/2 by PKCδ and MEK1/2. Human biliverdin reductase (hBVR), a Ser/Thr/Tyr kinase and intracellular scaffold/bridge/anchor, is a nuclear transporter of MEK1/2-stimulated ERK1/2 (Lerner-Marmarosh, N., Miralem, T., Gibbs, P. E., and Maines, M. D. (2008) Proc. Natl. Acad. Sci. U.S.A. 105, 6870–6875). hBVR, PKCδ, and MEK1/2 overlap in their tissue expression profile and type of activators. Presently, we report on formation of an hBVR-PKCδ-ERK2 ternary complex that is essential for ERK2 signal transduction and activation of genes linked to cell proliferation and cancer. MEK1/2 and the protein phosphatase PP2A were also present in the complex. When cells were stimulated with insulin-like growth factor-1 (IGF-1), an increased interaction between hBVR and PKCδ was detected by FRET-fluorescence lifetime imaging microscopy. hBVR and ERK2 were phosphorylated by PKCδ; however, the PKC was not a substrate for either ERK2 or hBVR. IGF-1 and phorbol ester increased hBVR/PKCδ binding; hBVR was required for the activation of PKCδ and its interaction with ERK2. The C-terminal phenylalanine residues of PKCδ (Phe660, Phe663, and Phe665) were necessary for binding to ERK2 but not for hBVR binding. Formation of the hBVR-PKCδ-ERK2 complex required the hBVR docking site for ERK, FXFP (DEF, C-box) and D(δ)-box (ILXXLXL) motifs. The hBVR-based peptide KKRILHCLGLA inhibited PKC activation and PKCδ/ERK2 interaction. Phorbol ester- and TNF-α-dependent activation of the ERK-regulated transcription factors Elk1 and NF-κB and expression of the iNOS gene were suppressed by hBVR siRNA; those activities were rescued by hBVR. The findings reveal the direct input of hBVR in PKCδ/ERK signaling and identify hBVR-based peptide regulators of ERK-mediated gene activation.

Transduction of the extracellular stimuli to the nucleus for gene activation is primarily conducted through the type-dependent stimulation of kinases in the three MAPK subfamilies as follows: ERK, JNK, and p38. There is extensive cross-talk between the three subfamilies and their upstream activators that include protein kinase C␦ (PKC␦), a member of the novel subfamily of PKCs. A variety of functions, some of them controversial, has been attributed to PKC␦; for instance, both proand anti-apoptotic effects of the PKC have been reported. Generally it is agreed that the kinase is a regulator of cell growth, proliferation, and cell division arrest and plays a role in glucose signaling (1). ERK1/2 kinases are upstream activators of an estimated 50 nuclear factors and proteins that influence cell differentiation, proliferation, stress response, and promote tumor growth (2)(3)(4). The list includes Elk1, NF-B, iNOS, 2 c-Myc, and HSF-1 (5,6). ERK1/2, however, do not have either a functional nuclear localization signal or a nuclear export signal sequence and rely on human biliverdin reductase (hBVR), a soluble 36-kDa cytoplasmic polypeptide, for their nuclear import and export (7). Furthermore, activity of hBVR, as an intracellular scaffold/bridge/anchor protein, is required for placing ERK1/2 in proximity to its kinases, MEK1/2, in the cytoplasm and bringing Elk1 in contact with the activated ERK1/2 in the nucleus. Experiments with kinase-inactive hBVR have shown that this function of hBVR is independent of its kinase activity (7). The oxidative stress-responsive genes HO-1, c-Fos, c-Jun, and ATF-2/CREB, as well as the mitochondrial NADPH oxidase, the major source of oxygen radical production, are among the downstream targets of activated ERK1/2 (8 -10). ERK is also activated by PKC␦, which has been proposed to involve sequential activation of PKC␦/Raf/MEK/ERK (11).
Until recently, BVR was considered solely in the context of its reductase function in the heme catabolism pathway. In this pathway, the enzyme reduces the ␥-bridge of the open tetrapyrrole, biliverdin-IX␣, that is formed by cleavage of the heme (Fe 2ϩ -protoporphyrin-IX) molecule at the ␣-meso-bridge by the heme oxygenase isozymes, HO-1 and HO-2, to bilirubin-IX␣. However, within the past decade pleiotropic functions of hBVR have been uncovered (reviewed in Ref. 12). Functions include Ser/Thr/Tyr kinase activity, intracellular transport of signaling molecules, regulation of stress-responsive gene expression, and cytoplasm-cell membrane translocation of PKC␤II and PKC, members of the conventional and atypical PKCs, respectively, as well as cytoplasm-nuclear transport of hematin and, as noted above, ERK1/2 (7,(13)(14)(15). A summary of the findings and sequence of motifs in hBVR for which functions have been ascribed to date is shown in Fig. 1.
Several observations that included the primary and secondary structural features of hBVR, and consideration of the type of stimuli that activate the reductase, led us to hypothesize that the protein is a key component of PKC␦/ERK signal transduction. To elaborate, the C-terminal half of the protein includes a large 6-stranded ␤-sheet, a structure frequently associated with a site of protein/protein interaction (16). Within this structure are consensus motifs identified by MotifScan software with functions in insulin/IGF-1/PI3K/MAPK signaling pathways (12,17,18). Two sequences are particularly striking, 162 FGFPAF and 275 KKRILHCLGLA. The former contains the core (FXFP) of high affinity ERK-specific binding motif, also known as docking site for ERK, FXFP (DEF or C-box), identified by Jacobs et al. (19); and the latter is similar to the low affinity leucine-rich D(␦)-box motif, with the core consensus sequence of LX 2 LX 1 L, flanked upstream by a chain of positively charged residues (20). The D-box-like sequence is located in the C-terminal ␣-helix of hBVR. The D(␦)-box is specific to substrates and kinases in the MAPK signaling pathway (19,20). Notably, present at the C-terminal segment of PKC␦ is a conserved phenylalanine-rich hydrophobic motif, FAGFSFVN; this segment of PKC␦ functions as a docking site for regulatory molecules (21). Sandwiched between the last two phenylalanine residues is Ser 664 (numbering is based on the human PKC␦ sequence). This residue, together with Ser 645 , is essential for activation of PKC␦ (1). Binding of PKC␦ to regulatory molecules is dependent on phosphorylation of the Ser 664 (21).
Presently, with the noted information in mind, we tested the specificity of hBVR C-and D-box-like sequences for ERK and PKC␦ binding. In considering such interactions, we were also mindful of the tissue expression profile of hBVR, ERK and PKC␦, and we noted that there is an overlap in their profiles. Also, we noted that all three enzymes respond to similar types of activators. In addition to insulin, the list includes TNF-␣ and reactive oxygen species, which are associated with cell survival, proliferation, and apoptosis (2,12). We also sought and found support for the specificity of bindings from our previous observations. Specifically, we had noted hBVR participates as a scaffold/bridge in insulin/IGF-1/PI3K or insulin/IGF-1/MAPK signal transduction pathways and functions as an Src homology 2 (SH2) adaptor protein (22,23). In addition, hBVR is an activator of PKC␤II and -, by a mechanism that likely involves confor-mational change in the PKCs and exposure of their activation loops (21,24). In the case of PKC␤II, in vitro experiments suggest that Thr 500 in its activation loop is also phosphorylated by hBVR (14). This residue, which is conserved in other conventional PKCs, is one of the three phosphorylation sites necessary for PKC␤II activation (25).
In this study, in vitro and in cell approaches were used, and PMA and IGF-1 were utilized as the stimulants. We only examined ERK2 as there is a great deal of cross-reactivity among antibodies to ERK1 and -2. The study has revealed formation of a complex among hBVR, PKC␦, and ERK2 that is required for activation of ERK2. In this complex, hBVR functions as a bridge rather than a kinase, with the two specific sequences in the protein constituting the contact points between hBVR and its partners. We interpret findings of this study, together with previous findings (7), to suggest that hBVR is required for ERK signaling, independent of the activation pathway, and the type of stimulus. Furthermore, because in tumor cells proliferation is stimulated by deregulated ERK (26), the present observation that small hBVR-based peptides can effectively block activation of ERK2 by PKC␦ offers a potentially viable prospect for development of therapeutic agents to control tumor growth in a variety of malignancies.

EXPERIMENTAL PROCEDURES
Materials-PMA, recombinant human, untagged PKC␦, and TNF-␣ were obtained from Calbiochem. The specific PKC␦ peptide substrate, ARRKRKGSFFYGG, was from Biomol (Camarillo, CA). The hBVR-based peptides, FGFPAFSG and KKRILHCLGLA, both unmodified (for in vitro assays) and N-myristoylated (for in-cell studies) were synthesized by EZBiolab (Westfield, IN). Polyclonal antibodies to the N-or C-terminal segments of PKC␦ and antibody to the PP2A C-subunit were purchased from Cell Signaling (Beverly, MA). Antihuman BVR polyclonal antibodies were prepared as described before (27); all other antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). MEK1 and lipid activators were purchased from Millipore (Billerica, MA). PKC␦, p65, ERK1, MEK1, and MEK2 siRNA were purchased from Santa Cruz Biotechnology.
Plasmids and Mutant Constructs-The plasmid pcDNA-HA-hBVR was generated by fusing the open reading frame (ORF) of hBVR downstream of the HA epitope. The PKC␦ ORF was amplified by PCR from a human brain cDNA library and was cloned in the pcDNA3 expression plasmid. Site-directed mutagenesis of wild-type pcDNA-HA-hBVR (28) or pcDNA-PKC␦ was used to prepare all mutant constructs; mutations were verified by sequencing. The C-box (FGFPAF) mutant hBVR was generated by replacing Phe 162 with Val, Phe 164 with Ala, and Phe 167 with Val; the D-box (KKRILHCLGLA) hBVR mutant was made by replacing Ile 278 , Leu 279 , Leu 282 , and Leu 284 with Ala. The human PKC␦ kinaseinactive mutant was generated by replacing Lys 378 by Arg (29). The truncated mutant of the PKC, lacking the last 21 amino acids of the C-terminal tail, was generated by using a PCR primer to amplify the shorter open reading frame and to provide a termination codon. Phenylalanine residues in the C-terminal tail of PKC␦ were replaced with small hydrophobic amino acids, Phe 660 by Val, Phe 663 by Ala, and Phe 665 by Val, using site-directed mutagenesis of WT PKC␦. pEGFP-PKC␦ and pDsRed2-hBVR were prepared by cloning the ORFs into the appropriate vectors (Clontech). The NF-B reporter plasmid was from Stratagene (La Jolla, CA).
For expression of proteins in Escherichia coli, hBVR, PKC␦, and ERK2 cDNA were cloned as fusions with the GST open reading frame of pGEX4-T2 (GE Healthcare). Human ERK2 was amplified from a cDNA library, as above. The ERK2 plasmid was also subjected to site-directed mutagenesis of Lys 54 to Arg, to give a kinase-inactive form analogous to the Lys 52 mutation of the rat protein (30). Proteins were isolated from cultures of E. coli BL21 after induction with isopropyl 1-thio-␤-D-galactopyranoside and batch absorption to GST-agarose (GE FIGURE 1. Schematic presentation of consensus sequences of hBVR for which functions have been ascribed. The numbers indicated for each consensus sequence are those of the hBVR primary structure. The N-terminal segment of 99 residues is the catalytic domain of hBVR; it houses a sequence of four valines followed by the consensus for the ATP/adenine ring-binding site. The kinase activity of hBVR is responsible for its autophosphorylation (22,38). hBVR is a kinase for serine phosphorylation of IRS-1, the phosphorylation of which halts glucose uptake (60). hBVR is also a likely kinase for Thr 500 in the activating loop of PKC␤II (14); the PKC is a key component of cell growth and differentiation. The reductase domain catalyzes reduction of biliverdin to bilirubin, a component of cellular defense mechanisms protecting against reactive oxygen species (ROS) (61) and apoptosis (32). The sequences designated by one or two asterisks closely resemble sites in the primary sequence of repeats V (QAMLWDLNE) and VI (SIKIWDLE) of the receptor for activated C-kinase-1 (RACK1). RACK1 is a 36-kDa protein that is similar in size to hBVR (62). Activation of PKCs, including the ␤, ␦, and ⑀ isoforms (62), is associated with conformational change that exposes their RACK-binding sites. We predict that the presence of RACK1-like sequences in hBVR may allow its binding to PKCs. The binding would not require kinase activity of hBVR. The basic leucine zipper motif binds to 7-and 8-bp AP-1 and AP-2 sites. Stress-response genes are activated by AP-1, and cAMP-responsive genes are regulated by AP-2 regulatory elements. hBVR regulates expression of stress-responsive HO-1, c-Fos, c-Jun, and ATF2/CREB (14,31,32). Within this sequence is a motif that strongly resembles a conserved protein kinase motif (63). The high affinity ERK-binding site, known either as C-box or DEF (19), is the site of interaction of ERK1/2 and hBVR, positioning ERK in proximity to its kinase (7). Nuclear localization of hBVR is also critical for transport of the transcriptional regulators ERK1/2 and heme into the nucleus (7,13). Reentry of ERK into the cytoplasm requires the intact hBVR nuclear export signal (NES) (7). hBVR is directly phosphorylated by IRK upon activation by insulin or IGF-1 (22). The tyrosine in the SH2 recognition motif of hBVR, as with other SH2 recognition motifcontaining proteins, is predicted to form a platform for formation of signaling complexes (64). hBVR is phosphorylated by ERK, and MotifScan predicts serine in the SP sequence as the phosphorylation target site of ERK1/2. A second SH2 recognition motif follows the nuclear localization signal (NLS) and is involved in activation of PKC by TNF-␣ (15). The low affinity D-box-like sequence is the binding site for kinases and substrates in the MAPK signaling cascade. The C-terminal six residues are the Zn 2ϩ -binding domain of hBVR (28). Based on the reported role of Zn 2ϩ for plasma membrane translocation of PKCs and nuclear translocation of NF-B (65) we predict that the function of hBVR in translocation of PKCs -␤ and -to the cell membrane may involve its associated Zn. Notably, hBVR under resting conditions is found in the cytoplasm and membrane caveolae (66). The C-terminal lysine 296 is critical for the hBVRs catalytic activity (M. D. Maines, unpublished data.); although it lies in a disordered region of the BVR molecule (16), this does not preclude a catalytic function.
Healthcare). The proteins were either eluted with reduced glutathione or the fusion protein was cleaved with thrombin to release the desired protein.
Cell Culture, Infection, Transfection, and Co-immunoprecipitation-HEK293A cells were grown and transfected with plasmids as described elsewhere (31). Overexpression was confirmed by Western blotting. Transfected cells were routinely serum-starved (0.1% FBS) overnight prior to treatment with either 100 nM PMA or 20 ng/ml IGF-1 for 15 min, unless specifically indicated. Preparation and use of viruses expressing sihBVR or schBVR (randomized control for sihBVR) were as described previously (32). Transfection of cells with synthetic siRNA duplexes was performed as suggested by the manufacturer (Santa Cruz Biotechnology). Immunoprecipitates from cell lysates were prepared as described previously (32) and were separated by SDS-PAGE followed by transfer to nitrocellulose, and purified protein was included as a standard. The membrane was probed with antibody as detailed in the appropriate figures and legends.
Metabolic Labeling-Cells were synchronized in culture medium containing 0.1% serum for 24 h. Thereafter, cells were treated with carrier-free [ 32 P]H 3 PO 4 as described previously (15). 4 h later cells were treated with 100 nM PMA for 15 min. The cell lysate was immunoprecipitated with anti-PKC␦ antibody, and phosphorylated proteins were visualized by autoradiography after gel electrophoresis of the precipitate.
PKC␦ Kinase Activity Measurement-PKC␦ activity in vitro was measured using 0.1 ng/l human PKC␦ in a 50-l assay containing 50 mM HEPES, pH 7.4, 20 mM MgCl 2 , 2 mM EGTA, 0.2 mM DTT, sonicated lipid activators (1% phosphatidylserine and 0.1% diacylglycerol), and peptide substrate. The reaction was started by the addition of 50 M ATP labeled with 5 Ci of [␥-32 P]ATP. Incubation was carried out for 40 min at 30°C. Effector molecules, the WT or mutant hBVR proteins or peptides, were added 10 min prior to the addition of peptide substrate. The reaction product was detected either by gel electrophoresis and autoradiography or by the P81 filter assay as described before (22). For measurement of cellular PKC␦ activity, the kinase was immunoprecipitated from cell lysate with anti-PKC␦ antibody and treated with protein A/G-agarose. The agarose-bound immunoprecipitate was centrifuged and used for measurement of PKC␦ activity in a kinase reaction mixture containing 50 M PKC␦ peptide substrate; incorporation of [ 32 P] was assessed by the P81 method (22).
ERK Kinase Activity-GST-ERK2 (at 0.5 M) prepared from E. coli was incubated with an equimolar amount of kinase-inactive GST-PKC␦ in 50 mM HEPES, pH 7.5, 20 mM MgCl 2 , and 1 mM DTT. hBVR was included instead of PKC␦ in some reactions. The reaction was started by addition of [␥-32 P]ATP (1 Ci/mol) and continued for 1 h. Reaction products were detected by gel electrophoresis and autoradiography, as above.
hBVR Kinase Assay-Kinase activity of hBVR was assayed as described earlier (22). hBVR was incubated at 30°C in a 50-l reaction mixture containing 50 mM HEPES, pH 8.4, 30 mM MnCl 2 , 0.2 mM DTT, 10 M ATP labeled with 10 Ci of [␥-32 P]ATP and PKC␦ for 30 min. The reaction products were detected by gel electrophoresis and autoradiography.
Measurement of hBVR Reductase Activity-hBVR activity was measured at pH 6.7 using NADH as the cofactor, as described (33). The rate of reduction of biliverdin to bilirubin was determined by the increase in absorbance at 450 nm at 25°C. Specific activity is expressed as nanomoles of bilirubin/ min/mg of protein.
Confocal Microscopy-HeLa cells were maintained as described for HEK29A cells, and transfection was performed at ϳ80% confluency. One day after the transfection, cells were serum-starved for 24 h (0.1% FBS). Thereafter, cells were treated with 20 ng/ml IGF-1 for 10 min prior to collection of fluorescence images. Live cells were maintained at 37°C under an atmosphere of 5% CO 2 in the course of image collection using a Cell Observer SD (spinning disc) from Zeiss and a Plan Apochromat 63 ϫ 1.4 oil objective.
FLIM Experiments-HeLa cells were maintained, transfected, and starved as described above. The frequency domain FLIM images was taken in the presence or absence of 20 ng/ml IGF-1 that was added before image collection. FLIM experiments on transiently transfected HeLa cells were performed using a Zeiss Axiovert 200 M inverted wide field microscope and a Lambert Instruments Fluorescence Lifetime Attachment (Lambert Instruments, Roden, The Netherlands) (34). A lightemitting diode (Luxeon Emitter, max ϭ 470 nm) modulated at 40 MHz was used to excite GFP, and the emission of GFP was detected through a narrow emission filter (525/25 nm). FLIM measurements were calibrated with a 10 M solution of fluorescein as a lifetime standard, the lifetime of which was set to 4.02 ns. Fluorescence lifetimes were calculated from several regions of interest, defined to include multiple cells, and data are presented as histograms. The FL histograms were fitted to Gaussian functions, from which the centers of the distributions and the distribution widths were extracted; the errors reported are one-half of the distribution width.
Luciferase Reporter Assay-To assay Elk1-dependent transcriptional activity, cells were co-transfected with the reporters pFA2-Elk1 and pFR-Luc (Stratagene) and pCMV-␤-gal, together with plasmids encoding proteins of interest. After transfection and starvation, ERK signaling was stimulated by treatment with 100 nM PMA for 8 -10 h. Activation of NF-B was determined in cells transfected with a luciferase plasmid regulated by multiple NF-B recognition elements; its use with these cells has been described previously (35). NF-B was activated by treatment with 20 ng/ml TNF-␣ in DMEM containing 0.1% FBS; treatment was initiated 12 h after plasmid DNA was added to the cells and continued for an additional 12 h. In some experiments, cells were transfected with double-stranded siRNA for the NF-B p65 subunit 18 h before again being transfected with reporter plasmids. The iNOS promoter plasmid has been described elsewhere (35); it was co-transfected with pCMV-␤-gal. Details of treatments are given in the figure legends. Luciferase assays for promoter activity were normalized on ␤-galactosidase as detailed before (14,31).
RT-PCR-Primers for quantitative PCR to measure expression of iNOS and 18 S rRNA were obtained from IDT (Coralville, IA) and were designed using PrimerQuest software. Total RNA was isolated from cells using TRIzol reagent (Invitrogen) and used as a template for cDNA synthesis. The iNOS mRNA was assayed by quantitative PCR using the ⌬⌬C T method with 18 S rRNA as the internal standard.
Statistical Analysis-Experiments were repeated three times unless otherwise indicated. Data in bar graphs are the means Ϯ S.D. of three experiments, each with triplicate samples. Prism 3.0 software (GraphPad, San Diego) was used for one-way analysis of variance; statistical significance was determined by Student's t test for sample pairs.

RESULTS
Physical Interaction of PKC␦ with hBVR-The association of hBVR and PKC␦ in living cells was examined using confocal microscopy and FRET. Cells were transfected with pEGFP-PKC␦ and pDsRed2-hBVR and treated with IGF-1, as detailed under "Experimental Procedures." As noted in Fig. 2a, there is overlap in the cellular localization of PKC␦ and hBVR in IGF-1-treated cells. To examine intracellular association of EGFP-PKC␦ and DsRed2-hBVR, FRET analysis was employed, using FLIM; in this system, interaction of the two proteins should be reflected in a significantly decreased fluorescence lifetime of EGFP-PKC␦ in the presence of DsRed2-tagged hBVR. Fluorescence lifetime is indicated by false color images in Fig. 2b, the change in color toward blue is indicative of shortened lifetime. Integration of the signals, shown in Fig. 2c, confirms that there is a significant reduction in fluorescence lifetime of the donor EGFP in the presence of DsRed2-hBVR as follows: 2⅐31 Ϯ 0.14 compared with 2⅐64 Ϯ 0.09 ns for EGFP-PKC␦ alone. The reduction in lifetime is indicative of physical association of the proteins. Treatment with IGF-1 reduces the fluorescence lifetime further to 2⅐26 Ϯ 0.12 ns (Fig. 2c), suggesting increased association or more stable binding (Fig. 2a).
Presence of hBVR Increases the Autophosphorylation and Kinase Activity of PKC␦-To test whether a complex of hBVR with PKC␦ can be recovered from the cells, the following coimmunoprecipitation experiment was performed. Cells were transfected with expression plasmids for PKC␦ and HA-tagged hBVR, and PKC␦ was activated with IGF-1 or PMA. Cell lysates were immunoprecipitated with anti-HA antibodies and a Western blot of the immunoprecipitate was probed with antibodies to PKC␦ and hBVR. As shown in Fig. 3a, activation of PKC␦ by IGF-1 or PMA resulted in an increased binding of the PKC to hBVR. Because hBVR activates the conventional PKC␤II and atypical PKC (14, 15) by formation of complexes, we examined whether hBVR activates PKC␦, a member of the novel PKC subclass. Two experiments were carried out to test this. Using commercial, activated PKC␦ and a kinase assay system optimal for PKC␦ kinase activity, the effect of hBVR in vitro on PKC␦ autophosphorylation was examined. The assay conditions were not compatible with hBVR kinase activity. As noted in Fig. 3b, the presence of hBVR increased PKC␦ autophosphorylation in a dose-dependent manner, and at the same time hBVR was phosphorylated. The study was extended to examine whether in the cell hBVR/PKC␦ interaction increases the kinase activity of the PKC. For this, cells were co-transfected with WT or kinase-inactive PKC␦ (K378R) and hBVR and treated with PMA. Cell lysates were prepared, and immunoprecipitated with anti-PKC␦ antibodies. The immunoprecipitates were then assayed for PKC␦ activity. As noted in Fig. 3c, there was a significant increase in the incorporation of label into the substrate in preparations obtained from cells transfected with intact PKC␦ and hBVR, compared with those transfected with PKC␦ alone, with or without PMA treatment. However, the presence of hBVR did not influence activity of the kinase-inactive PKC␦. Because the increase in PKC␦ phosphorylation could conceivably be brought about as a result of kinase activity of hBVR, we examined this possibility. As noted in Fig. 3d, the FIGURE 2. Interaction between PKC␦ and hBVR is enhanced by extracellular stimulus. a, IGF-1 treatment promotes intracellular co-localization of PKC␦ and hBVR. HeLa cells were co-transfected with EGFP-PKC␦ and pDsRed2-hBVR and treated with IGF-1 as described in the text. The fluorescence images from live cells maintained at 37°C and 5% CO 2 were collected as described under "Experimental Procedures." b, stimulation with IGF-1 enhances PKC␦/hBVR binding. Cells expressing the above constructs were treated with IGF-1 as described in the text. Representative fluorescence intensity (panels i, iii, and v) and FLIM (panels ii, iv, and vi) images of EGFP-PKC␦ (panels i and ii), EGFP-PKC␦/DsRed2-hBVR (panels iii and iv), and EGFP-PKC␦/ DsRed2-hBVR ϩ IGF 1 (panels v and vi) in HeLa cells are shown. The scale bar ranges from 1 to 4 ns. c, fluorescence lifetime histograms of EGFP-PKC␦ (E), EGFP-PKC␦/DsRed2-hBVR (•), and EGFP-PKC␦/DsRed2-hBVRϩ IGF-1 (OE) in HeLa cells. Curves represent FLIM data recorded from ϳ25 cells per condition; frequency of events is shown in arbitrary units (a.u.).
increased kinase activity of PKC␦ was not a consequence of its phosphorylation by hBVR, even under reaction conditions optimized for hBVR kinase activity, and a minimal increase in PKC␦ phosphorylation was observed in the presence of hBVR, a finding that suggested that hBVR activated PKC␦ by inducing changes in secondary structure and folding of the PKC. To confirm that indeed hBVR is instrumental in enhancing PKC␦ phosphorylation, the effect of hBVR depletion on PKC␦ autophosphorylation was examined in cells that had been transfected with a PKC␦ expression plasmid and metabolically labeled with [ 32 P]orthophosphate. As expected, treatment with 100 nM PMA resulted in a significant increase in label incorporation into PKC␦ (Fig. 3e), which was attenuated in cells infected with sihBVR. This treatment did not affect the level of PKC␦, but, as established previously, it depletes cellular levels of hBVR (32,36). Accordingly, activation of PKC␦ in response to PMA to a great extent required hBVR. However, it might be argued that activation of PKC␦ in the complex with hBVR was not only a consequence of increased autophosphorylation of the PKC, but it may also have involved blocking recruitment of SH2-interactive phosphatases to the PKC, allowing accumulation of the phosphorylated form of the kinase. To test this possibility, the hBVR-PKC␦ complex was examined for the presence of the phosphatase PP2A C-subunit in cells overexpressing both PKC␦ and hBVR. This phosphatase has been demonstrated to dephosphorylate PKC␦ in vitro and to associate with the PKC in mouse cells (37). As shown in Fig. 3f, the phosphatase was indeed present in both the IGF-1-or PMAactivated complexes. Therefore, the increased PKC␦ phosphorylation in the complex likely reflects, at least in part, a more rapid cycling of phosphate on and off PKC␦.
Complex Is Formed between ERK2, PKC␦, and hBVR-Because hBVR was previously shown to form a complex with ERK1/2 (7) and presently found to bind PKC␦, we examined whether a complex is formed between the three proteins. For this, cells were co-transfected with PKC␦ and HA-tagged hBVR plasmids, starved, and treated with either IGF-1 or PMA. The cell lysate was immunoprecipitated with anti-HA antibodies, and the immunoprecipitate was subjected to Western blotting using anti-ERK2, anti-PKC␦, and anti-hBVR antibodies. Data shown in Fig. 4a revealed the presence of all three proteins in the immunoprecipitate subsequent to stimulation with IGF-1 or PMA, indicating formation of a ternary complex in response to the extracellular stimuli. We then explored the role of hBVR in complex formation between PKC␦ and ERK2. The following experiments revealed direct involvement of hBVR in ERK/ PKC␦ binding. In the first experiment, cells transfected with the PKC␦ expression plasmid were treated with either the hBVR siRNA virus or control scRNA. After stimulation with IGF-1, cells were lysed and immunoprecipitated with anti-ERK2 antibodies. An increased PKC␦/ERK2 binding was observed in stimulated cells in the presence of scRNA (Fig. 4b); the presence of sihBVR nearly eliminated the binding, a finding that points to a requirement for hBVR for the ERK/PKC␦ interaction. In the next experiment the formation of a complex among the three proteins was examined in untransfected cells (Fig. 4c). For this, the cell lysate was immunoprecipitated with anti-PKC␦ antibodies, and the precipitate was probed with antibodies to ERK, PKC␦, and hBVR. As shown, treatment with IGF-1 led to a detectable increase in the level of the three proteins in the complex; endogenous protein levels were sufficient for formation of the ternary complex. Previously, studies by this laboratory had indicated that hBVR and ERK2 form a cytoplasmic complex that also included MEK1 (7). To test whether the complex immunoprecipitated with antibodies to PKC␦ also contained FIGURE 3. PKC␦ autophosphorylation and activity are increased by hBVR. a, stimulation with IGF-1 or PMA enhances PKC␦/hBVR binding. HEK293A cells were transfected with pcDNA-HA-hBVR and pcDNA-PKC␦ plasmids and then treated with IGF-1 or PMA as described (see under "Experimental Procedures"). Cell lysates were prepared, immunoprecipitated (IP) with anti-HA antibodies, and followed by sequential immunoblotting with antibodies to the C-terminal domain of PKC␦ and anti-hBVR antibodies. WB, Western blot. b, hBVR increases PKC␦ autophosphorylation. Increasing concentrations (0, 0.06, and 0.3 M) of hBVR were preincubated with active recombinant human PKC␦ for 5 min at room temperature prior to the kinase assay. The PKC␦ assay was carried out as described under "Experimental Procedures." The reaction products were processed for autoradiography as described in the text. c, hBVR increases PKC␦ kinase activity. WT PKC␦ or its kinase-inactive mutant were overexpressed either alone or together with hBVR in cells. Lysates prepared from PMA-treated cells were immunoprecipitated with anti-PKC␦ antibodies, as in a. PKC␦ activity of the immunoprecipitates was then determined using the PKC␦ peptide substrate ARRKRKGSFFYGG, and the incorporated phosphate was measured by the P81 method. *, p Ͻ 0.001. d, PKC␦ is not a substrate for hBVR kinase activity. PKC␦ kinase activity was measured in the presence or absence of hBVR, under assay conditions optimal for hBVR kinase activity. Reaction products were analyzed as in b. e, hBVR is required for activation of PKC␦ in metabolically labeled cells. HEK293A cells transfected with pcDNA-PKC␦ were infected with virus expressing hBVR-siRNA for 24 h. Cells were starved (24 h) and metabolically labeled with [ 32 P]H 3 PO 4 for 4 h prior to treatment with PMA as in a. Cell lysates were immunoprecipitated with anti-PKC␦ antibodies. The immunoprecipitates were processed for autoradiography, as described in the text. After decay of radioactivity, the membrane was probed with anti-PKC␦ antibodies as a reference for loading. f, phosphatase PP2A is present in the hBVR-PKC␦ complex. Cells were co-transfected with pcDNA-HA-hBVR and pcDNA-PKC␦ and treated with IGF-1 or PMA. Cell lysates were processed for immunoprecipitation with anti-HA antibodies, and analyzed by Western blotting. The blot was probed sequentially with anti-PP2A C-subunit and anti-hBVR antibodies.
MEK1, cells were transfected with the PKC␦ expression plasmid and treated with either IGF-1 or PMA. Cell lysates were prepared and subjected to immunoprecipitation using anti-PKC␦ antibodies. The immunoprecipitate was Western blotted and probed sequentially with antibodies to hBVR, MEK1, and ERK2. Interestingly, all proteins were present in the immunoprecipitates, indicating that the complex is a collection of multiple kinases (Fig. 4d).
We next questioned whether the active configuration of PKC␦ was required for formation of the complex with ERK2. Cells were transfected with either WT or kinase-inactive PKC␦ and treated with IGF-1; the endogenous ERK and hBVR were relied upon. The presence of PKC␦ was demonstrated in the immunoprecipitate obtained with anti-ERK2 antibodies (Fig.  5a). Moreover, the intensity of the PKC␦ signal was increased subsequent to treatment with IGF-1. However, IGF-1 did not increase kinase-inactive PKC␦/ERK2 binding. The data suggest that binding was, to a large extent, dependent on kinase competency of PKC␦. Finding that a K54R mutant ERK2 (see under "Experimental Procedures") was phosphorylated by PKC␦ (Fig.  5b) suggests that conformational changes in the activated PKC render its ERK2-binding site accessible. This suggestion is consistent with the finding that ERK2 was not a kinase for PKC␦ (Fig. 5c). Kinase-inactive PKC␦, prepared as a GST fusion protein, was used as the substrate in this experiment. WT GST-ERK2 demonstrated robust autophosphorylation, whereas no phosphorylation of PKC␦ was observed. hBVR, however, was a substrate for ERK2 (Fig. 5c), and moreover, ERK2/hBVR interaction potentiated hBVR reductase activity (Fig. 5d). We had previously shown that phosphorylation and reductase activity are directly linked (38).
Interactions in Ternary Complex Are Mediated by hBVR ERK-docking Motif and D-Box-like Sequences-It was previously shown that hBVR activates ERK1/2 and that this activation depends on complex formation involving the hBVR ERKdocking motif (DEF) and D-box-like sequences (7). Presently, to examine the role of these sequences in hBVR/ERK2/PKC␦, the series of experiments shown in Fig. 6 was performed. The experiments used WT or mutant hBVR expression constructs and hBVR-based peptides corresponding to the two binding motifs. To test the role of the sequences in intact hBVR, constructs were overexpressed together with PKC␦, and cells were stimulated with either IGF-1 or PMA. Irrespective of the stimulus, each mutant showed significantly reduced ternary complex formation (Fig. 6a), indicating that both types of motifs are required for the complex formation. Next, the effect of hBVRbased peptides on autophosphorylation of PKC␦ was tested in vitro. The D-box-like peptide (KKRILHCLGLA) strongly suppressed PKC␦ autophosphorylation at a low dose of 0.2 M, FIGURE 4. hBVR, PKC␦, and ERK2 form a ternary complex. a, hBVR, PKC␦, and ERK2 co-immunoprecipitate. HEK293A cells were co-transfected with pcDNA-PKC␦-and pcDNA-HA-tagged hBVR and treated with either PMA or IGF-1 as described in the text. Immunoprecipitates (IP) obtained with anti-HA antibodies were subjected to Western blotting (WB). The membrane was sequentially probed with anti-PKC␦, anti-ERK2, and anti-hBVR antibodies. b, depletion of hBVR attenuates PKC␦ and ERK2 binding. Cells were transfected with pcDNA-PKC␦ and infected with the viral constructs described in the text to express either siRNA for hBVR or scRNA. Lysates prepared from IGF-1-treated cells were immunoprecipitated with anti-ERK2 antibodies and analyzed by Western blotting, using anti-PKC␦ antibodies. IgG was used as a loading reference. c, hBVR, PKC␦, and ERK2 form a ternary complex in the cell. HEK cells were serum-starved and treated with IGF-1 as described in the text. Cell lysates were immunoprecipitated with anti-ERK2 antibodies, and the immunoprecipitates were processed for Western blotting; the blot was sequentially probed with anti-hBVR, anti-PKC␦, and anti-ERK2 antibodies. d, MEK1 also associates with the hBVR-ERK-PKC␦ complex. Cells were transfected with pcDNA-PKC␦ and treated with PMA or IGF-1 as in a. Cell lysates were immunoprecipitated with anti-PKC␦ antibodies and analyzed by Western blotting. The blot was sequentially probed with anti-hBVR, anti-MEK1, and anti-ERK2 antibodies. ST, protein standards. FIGURE 5. Activities affected by protein/protein interaction in the complex. a, activation of PKC␦ is not required for ERK binding. Cells were transfected with WT-or kinase-inactive pcDNA-PKC␦, starved, and treated with IGF-1. Cell lysate was immunoprecipitated (IP) with anti-ERK2 antibodies, and the precipitate was analyzed by Western blotting (WB), using antibodies to the C-terminal domain of PKC␦. IgG served as a loading reference. b, PKC␦ phosphorylates ERK2 in vitro. Kinase-inactive GST-ERK2 was used as the substrate for PKC␦ as described in Fig. 3b. Control reactions contained only one of the proteins. Reaction products were detected by autoradiography. c, ERK2 does not phosphorylate PKC␦ in vitro. GST-ERK kinase activity was assessed using GST-PKC␦ kinase-inactive mutant or hBVR as substrates. Control reactions included either the GST-PKC␦ kinase-inactive mutant or hBVR. The reaction products were processed for autoradiography. d, hBVR is activated by ERK2. GST-hBVR was phosphorylated in vitro by GST-ERK2 as in c, using unlabeled ATP; the reaction product was then assayed for reductase activity. The basal activity was measured for GST-hBVR incubated with ATP in the absence of ERK2. *, p Ͻ 0.001 compared with basal activity. Experimental details are provided in the text.
whereas the FGFPAFSG peptide was ineffective. Indeed, D-boxlike peptide suppressed PKC␦ autophosphorylation. When the myristoylated D-box-like peptide was added to cells overexpressing PKC␦ 2 h prior to treatment with IGF-1, a markedly reduced recovery of PKC␦ in the anti-ERK2 immunoprecipitate was observed (Fig. 6c). This observation indicates that the peptide blocks interaction of ERK2 with PKC␦ and therefore suggests that this hBVR motif is the participant in the interaction.
PKC␦ Hydrophobic Motif Is Essential for Its Kinase Activity and ERK2 Binding but Not for Interaction with hBVR-A phenylalanine-rich hydrophobic sequence, 660 FAGFSF, that somewhat resembles the hBVR ERK-docking motif (DEF, C-box-like FGFPAF) is present in the C-terminal kinase domain of the human PKC␦. We hypothesized that this sequence might participate in protein/protein interaction with ERK and/or hBVR.
To test this hypothesis, the three phenylalanine residues Phe 660 , Phe 663 , and Phe 665 were replaced with Val, Ala, and Val, respectively. We also tested a C-terminal truncated PKC␦, designed to delete the 21 C-terminal residues of the protein.
The results of this series of experiments are shown in Fig. 7. Cells co-transfected with pcDNA-HA-hBVR and either WT PKC␦ or the hydrophobic site mutant plasmids were treated with IGF-1 or PMA. HA-hBVR was immunoprecipitated from cell lysates using anti-HA antibodies, and the immunoprecipitate was examined by Western blotting with antibodies to the N-terminal region of PKC␦. PKC␦ was clearly present in the immunoprecipitates (Fig. 7a), irrespective of whether the WT or mutant form was overexpressed, indicating that the hydrophobic motif is dispensable for interaction with hBVR. Moreover, as shown in Fig. 7b, when lysates obtained from cells cotransfected with pcDNA-HA-hBVR and WT or C-terminal deletion mutant pcDNA-PKC␦ plasmids were immunoprecipitated with anti-HA antibodies, both WT PKC␦ and the truncated protein were detected in the immunoprecipitate. This observation indicates that the C-terminal 21 residues of PKC␦ are not required for the hBVR/PKC␦ interaction. However, when cells transfected with plasmids expressing WT PKC␦ or the phenylalanine mutant were immunoprecipitated with anti-ERK2 antibody, only the WT PKC␦ was recovered in the precipitate (Fig. 7c), suggesting that the 660 FAGFSF sequence is required for PKC␦/ERK2 binding. The possibility that the hydrophobic site or C-terminal deletion mutants might disrupt downstream signaling by ERK2 was examined by co-transfection of cells with WT or mutant PKC␦ expression plasmids together with Elk1 reporter plasmids. As shown in Fig. 7d, overexpression of WT PKC␦ together with Elk1-dependent luciferase reporter plasmids was associated with a robust induction of reporter gene activity by PMA. The kinase-inactive mutant generated an attenuated response, indicating a significant role for the PKC␦ activity in ERK2/Elk1 signal transduction. Expression of the truncated and hydrophobic site mutant proteins resulted in a further reduction of the signal, although the difference between the effects of these two mutants was not significant. Collectively, the data support the importance of the PKC␦ kinase activity in interaction of the PKC with ERK2 and in signaling downstream of ERK2. This suggestion is consistent with the finding that the C-terminal deletion and hydrophobic site mutants were inactive in PKC␦ kinase reactions in vitro, whereas the WT PKC␦ was both autophosphorylated and active in phosphorylation of hBVR (data not shown). As expected, the K378R mutation was also completely inactive.
hBVR Is a Partner in ERK2-mediated Transcriptional Activity-To assess the role of hBVR in the activation of the ERK1/2/Elk1 pathway by upstream kinases, including PKC␦ and MEK1 and MEK2, experiments were carried out using the Elk1-dependent luciferase reporter assay. Cells were transfected with the luciferase reporters (as described under "Experimental Procedures"), together with the control scRNA or siRNAs against PKC␦, hBVR, MEK1, and MEK2. Results are shown in Fig. 8. Fig. 8a shows that depletion of PKC␦ from cells led to a marked decrease in PMA-induced activation of Elk1dependent transcriptional activity when compared with scRNA-treated cells. When hBVR siRNA was added together FIGURE 6. hBVR ERK docking motif and D-box-like sequence participate in PKC␦-ERK binding. a, two hBVR motifs are required for PKC␦ and ERK2 binding. Cells co-transfected with pcDNA-PKC␦ and plasmids expressing hBVR protein carrying mutations in either ERK docking motif (FGFPAF 3 VGAPAV) or D-box-like sequence (KKRILHCLGLA 3 KKRAAHCAGAA) were treated with either IGF-1 or PMA as in Fig. 3. Cell lysates were immunoprecipitated (IP) with anti-ERK2 antibodies. The immunoprecipitated proteins were subjected to Western blotting (WB) using anti-PKC␦ antibodies as the probe. IgG served as a loading reference. b, hBVR-based KKRILHCGLA peptide inhibits PKC␦ autophosphorylation in a concentration-dependent manner, although the peptide FGFPAFSG, corresponding to the hBVR ERK-docking sequence, does not. PKC␦ autophosphorylation was measured in vitro in the presence of increasing concentrations of hBVR-based D-box peptide (KKRILHCGLA). The hBVR-based DEF-like FGFPAFSG peptide was also tested at a concentration of 1.0 M. The reaction was carried as described in the text; products were detected by autoradiography. c, KKRILHCLGLA hBVR-based peptide inhibits ERK1/2 and PKC␦ binding. Cells transfected with PKC␦ were starved and loaded with myristoylated hBVR-based D-box-like peptide. After treatment with IGF-1, cell lysates were immunoprecipitated with anti-ERK2 antibody followed by Western blotting, using anti-PKC␦ antibody as the probe.
with siPKC␦, the Elk1 activity was further decreased. We reasoned that if reduction in the activity were due to the treatments, then replacement of the depleted mRNA should restore the response to PMA. To test the validity of this reasoning, cells were pretreated with siRNAs, and 18 h later they were co-transfected with reporter plasmids and the appropriate pcDNA expression plasmid, or with further siRNA. Results are shown in Fig. 8b. As noted, upon treatment with PMA control cells displayed a robust stimulation of the Elk promoter activity; this was enhanced by overexpression of PKC␦ or hBVR. In siRNA-treated cells, the reduced Elk1 promoter activity could be effectively restored by introduction of the corresponding expression plasmids. Next, the effect of depletion of MEK1 and MEK2 alone or together with hBVR on ERK2-driven Elk1-dependent luciferase activity was examined (Fig. 8c). As expected, the presence of sihBVR markedly reduced PMA-induced Elk1dependent luciferase activity. Also, in the presence of MEK1 and -2 siRNA, there were significant decreases in the Elk1-dependent activity. The inclusion of hBVR sihBVR together with those for MEK further reduced activity, to less than 20% that of the scRNA control. This finding is in agreement with the previously reported role of hBVR in MEK/ERK1/2/Elk1 signaling (7). The efficacy of treating cells with siRNA for PKC␦ was confirmed by the levels of PKC␦ in immunoprecipitates obtained from cells transfected with PKC␦ scRNA or siRNA. The Western blot analysis shown in Fig. 8d confirms the effectiveness of the siRNA in depleting PKC␦.
ERK2 activates some 50 transcription factors, including NF-B (5,6). To test whether hBVR and PKC␦ affect the ERK2 activation of gene expression is limited to Elk1 and stimulation by PMA, and the next two series of experiments were performed with promoters driven by NF-B, using TNF-␣ as the stimulus. The results of the first series are also shown in Fig. 8. For this experiment, a luciferase reporter plasmid driven by multiple NF-B elements was used in cells transfected with pcDNA-PKC␦ and infected with sihBVR or schBVR. As noted in Fig. 8e, TNF-␣ induction of luciferase activity was enhanced in the presence of overexpressed PKC␦. The increased activity was abolished by the presence of hBVR siRNA. In the second experiment (Fig. 8f), the NF-B-dependent luciferase activity depleted by hBVR siRNA treatment was restored by introduction of an hBVR expression plasmid.
This line of investigation was further pursued by testing the response of a native cellular promoter that responds to NF-B. For this, the responses of the luciferase reporter gene driven by the human iNOS promoter to changes in the cellular level of PKC␦ were examined. The results of these experiments are shown in Fig. 9. Fig. 9a shows that induction of the iNOS promoter-dependent luciferase activity by TNF-␣ required the presence of both PKC␦ and hBVR. Specifically, in cells treated with either sihBVR or PKC␦ siRNA, the iNOS promoter did not respond to the cytokine. A significant response was observed in cells transfected with expression plasmids for PKC␦ or hBVR.

FIGURE 7. Intact C-terminal hydrophobic motif (FXXF(S/T)(F/Y)) of PKC␦ is essential for binding to and signaling by ERK2
. a, C-terminal PKC␦ phenylalanine residues are not required for binding to hBVR. Cells were transfected with WT PKC␦ or its hydrophobic site mutant ( 660 FAGFSF 3 VAGASV) plasmids and treated with IGF-1 or PMA. Proteins immunoprecipitated (IP) from cell lysates by anti-HA antibodies were examined by Western blotting (WB); the blot was sequentially probed with antibodies to the N-terminal domain of PKC␦ and anti-hBVR antibodies. b, PKC␦ hydrophobic sequence is not required for binding to hBVR. Cells were co-transfected with pcDNA-HA-hBVR and WT PKC␦ or its C-terminal truncated mutant and treated with IGF-1 or PMA. Cell lysates were immunoprecipitated with anti-HA antibodies and subjected to Western blotting, which was sequentially probed with anti-PKC␦ N-terminal domain and anti-BVR antibodies. c, intact PKC␦ hydrophobic sequence is required for binding to ERK2. Cells overexpressing either WT or hydrophobic site mutant PKC␦ plasmids were treated as described in a. Cell lysates were immunoprecipitated with anti-ERK2 antibodies and subjected to Western blotting; the blot was probed with anti-PKC␦ N-terminal domain antibodies. Immunoprecipitate inputs were adjusted to correct for differential expression of the WT PKC␦ and PKC␦ hydrophobic mutant. PKC␦-ST ϭ PKC␦ standard. d, intact PKC␦ is required for full activation of Elk1. Cells were co-transfected with the Elk1 reporter, ␤-galactosidase, and either WT pcDNA-PKC␦ or its kinase-inactive (K378R), truncated (⌬656 -676) or hydrophobic site mutants. Cells were serum-starved for 12 h and treated with 100 nM PMA for 10 h, and the lysates were assayed for luciferase activity. The activity was normalized against that of ␤-galactosidase. Experimental details are provided in the text. *, p Ͻ 0. 01 compared with WT PKC␦.
In the next experiment, the response of the endogenous iNOS promoter to siRNA treatment was examined using quantitative RT-PCR. After treatment of sc-treated control cells with TNF-␣, a modest but significant increase in iNOS mRNA was observed (Fig. 9b). As with the first experiment, the TNF-␣-dependent increase in iNOS mRNA was not observed in cells treated with siRNAs against hBVR or PKC␦. Again, an elevated level of iNOS mRNA was observed after TNF-␣ treatment of cells when PKC␦ was overexpressed. The increases in iNOS mRNA due to TNF-␣ and PKC␦ could be blocked by either sihBVR or siP65, indicating that both are necessary for the NF-B-dependent response.

DISCUSSION
The essential role of hBVR in the MEK1/2-dependent activation of ERK1/2 and in ERK1/2/Elk1 signaling was previously demonstrated (7). In that path, hBVR participates as a scaffold/ bridge/anchor protein in the cytoplasm for positioning the ERK FIGURE 8. hBVR is a determinant in both PKC␦ and MEK1/2 ERK-1/2/Elk1 signaling. a, depletion of PKC␦ and hBVR suppresses PMA-dependent Elk1 activation. Cells were co-transfected with the Elk1 reporter system (as used in Fig. 7d) and PKC␦, either alone or together with expression viruses for hBVR siRNA. Controls were transfected with schBVR virus. The cells were treated with 100 nM PMA for 10 h, and the cell lysates were assayed for luciferase activity as in Fig.  7d. *, p Ͻ 0.001 compared with scRNA control. b, restoration of siRNA-depleted hBVR or PKC␦ recovers PMA-dependent induction of Elk1. Cells were transfected with PKC␦ siRNA or infected with hBVR siRNA expression virus as in a. 18 h later, the cells were co-transfected with the Elk1 reporter system together with either pcDNA-PKC␦ or pcDNA-hBVR. siRNA or scRNA treatment was continued in the cells that were not rescued. After 12 h, cells were treated with 100 nM PMA or vehicle for an additional 12 h. Harvested cells were lysed, and the lysates were assayed for luciferase activity. c, depletion of hBVR increases suppression of PMA-dependent Elk1 activation by MEK1/2 siRNAs. Cells were co-transfected with Elk1 luciferase reporter and, where indicated, with siRNAs for MEK1/2 and/or hBVR. Cells were processed as in a. d, siRNA for PKC␦ decreases PKC␦ level in cells. Cells were transfected with either siRNA for PKC␦ or its control scRNA, as described in the text, and cell lysates were subjected to immunoblotting with antibodies to the C-terminal segment of PKC␦ followed by anti-␤-actin antibodies. WB, Western blot. e, PKC␦-driven induction of NF-B promoter is blocked by depletion of hBVR. Cells co-transfected with NF-B and ␤-galactosidase reporters and pcDNA-PKC␦ were also transfected with siRNA for hBVR or with control scRNA, as indicated. 24 h after transfection, cells were harvested, and cell lysates were assayed for luciferase activity, as in a. f, replacement of hBVR or PKC␦ expression restores TNF-␣-stimulated NF-B activation. Cells were transfected with PKC␦ siRNA or infected with hBVR siRNA expression virus. 18 h later, the cells were co-transfected with the NF-B reporter and either pcDNA-PKC␦ or pcDNA-hBVR, as indicated. Where indicated, siRNA treatment was continued. After 12 h, the cells were treated with 20 ng/ml TNF-␣ or vehicle for an additional 12 h and then harvested, and cell lysates were assayed for expression of luciferase, as in a. *, p Ͻ 0.001 compared with scRNA control. activation loop in proximity to its activator kinase MEK; in the nucleus, the activated ERK1/2 is brought in contact with its substrate Elk1 by hBVR; those functions of the reductase are independent of its kinase activity. Presently, we report on a similar function of hBVR in PKC␦-mediated ERK2 activation and transcriptional activity. Finding that silencing hBVR effectively attenuated activation of both PKC␦/ERK2/Elk1 and MEK1,2/ERK2/Elk1 signal transduction (Fig. 8, a and c) leads us to propose that hBVR is a key component of ERK2/Elk1 signaling. By extension, because ERK controls multiple pathways that regulate cell proliferation, transformation, and sur-vival, an expansive input of the reductase in a host of cellular functions can be surmised. The presently identified Elk1 and NF-B activation and iNOS expression as downstream targets of PKC␦/ERK2 signaling that are influenced by hBVR are consistent with this suggestion. The reported dependence of migration and proliferation of vascular smooth muscle cells on regulation of ERK1/2 activity by PKC␦ (39) underscores the potential biological significance of the role of hBVR in regulation of PKC␦/ERK2 signal transduction. As reported (39), either up-or down-regulation of the PKC resulted in diminished activity of ERK1/2, and in either case, cell migration and proliferation were impaired. Those observations are consistent with participation of a third and limiting component in ERK1/2 activation by PKC␦. We propose that hBVR may well be the limiting factor. Moreover, we draw an analogy between PKC␦ and another member of the novel class of PKCs, PKC⑀; the PKC⑀-ERK complex that is formed in response to treatment with IGF-1 also contains Raf1 (40). The PKC⑀-ERK complex can be disrupted with ceramide, which inhibits PKC⑀ association with Raf1.
Collectively, data obtained in this study and in a previous investigation (7) have identified hBVR as a key component in ERK1/2 phosphorylation/activation, be it by MEK1/2 or via PKC␦; additionally, based on the available information, hBVR is the only vehicle for both nuclear import and export of ERK1/2, processes that are essential for ERK-dependent regulation of gene expression. ERK dimer requires a transporter with a functional nuclear localization signal for translocation into the nucleus (6) and a transporter with a functional nuclear export signal for reentry into the cytoplasm. Notably, Gab1 (growth factor response factor-1) has been reported to be capable of nuclear import of ERK1/2 (41); to our knowledge, to date, the nuclear export activity of Gab1 or any other protein has not been reported. Moreover, although there are reports that PKC␦ is found in the nucleus of insulin-treated cells (42) and that its nuclear localization is required for cellular apoptosis (43), no evidence has been offered for the role of PKC␦ in ERK1/2 nuclear transport. Therefore, hBVR maintains its role as a cytoplasm-nucleus-cytoplasm transporter of ERK, and most likely, it is essential for transport of PKC␦-activated ERK2 into the nucleus and its export subsequent to gene regulatory activities.
The observed failure of ERK2/PKC␦ binding in the presence of hBVR mutants in which isoleucine and leucine residues Ile 278 and Leu 279 , Leu 282 , and Leu 284 were replaced with alanine, or phenylalanine residues Phe 162 , Phe 164 , and Phe 167 were replaced with Val, Ala, and Val, respectively, indicate dependence of the ERK2/PKC␦ binding on the intact hBVR. Moreover, data obtained with hBVR-based peptides suggest that the leucine-rich D-box-like sequence (KKRILHCLGLA) is involved in the hBVR/PKC␦ interaction and, as shown before, in the hBVR/ERK2 low affinity binding (7). The finding that the hBVR-based D-box-like peptide inhibited PKC␦ autophosphorylation lends further support to the suggestion that this sequence of hBVR is the hBVR-PKC␦ interactive domain. Conformational distortion of the binding pocket and interference at the protein/protein interface could account for the inhibitory action of the peptide. The peptide also inhibited PKC␦/ERK2 interaction (Fig. 6c). The sequence was previously identified as FIGURE 9. hBVR regulates PKC␦-mediated NF-B and iNOS induction. a, iNOS promoter activity is restored by replenishing siRNA-depleted protein. Cells were pretreated with siRNAs for 18 h as in Fig. 8b. Thereafter, cells were co-transfected with the iNOS-luciferase reporter and either pcDNA-PKC␦ or pcDNA-hBVR. 12 h later, cells were treated with 20 ng/ml TNF-␣ for an additional 12 h, and cell lysates were assayed for luciferase activity. *, p Ͻ 0.001 for TNF-␣-treated samples compared with scRNA ϩ TNF-␣ control. b, hBVR and p65 are necessary for PKC␦-dependent iNOS induction by TNF-␣. Cells were transfected with the indicated siRNAs or expression plasmids. 24 h later, cells were treated with TNF-␣ for 3 h. iNOS mRNA was measured using quantitative PCR, relative to 18 S rRNA. *, p Ͻ 0.01 for bracketed values.
the site of hBVR interaction with MEK1/2 and Elk1 (7). The phenylalanine-rich C-box/DEF-like sequence (FGFPAFSG) was shown previously to be involved in the high affinity hBVR/ ERK1/2 interaction (7). The high affinity C-box motif (FXFP) is exclusive to ERK-binding kinases (19). We interpret the finding that in the presence of either hBVR C-or D-box-like mutant proteins, the immunoprecipitate obtained with anti-ERK2 antibodies contained a markedly reduced level of PKC␦ (Fig. 6a) as supporting the proposed participation of those specific sites in the scaffolding function of hBVR. Accordingly, it is plausible that this function of hBVR is as relevant to its role in transduction of signals for cell growth and differentiation as is its kinase activity. The data offer little support for the possibility of nonspecific protein/protein interactions or of hydrogen bonding at the ␤-sheet interface as the sole mode of interactions between hBVR and PKC␦ or hBVR and ERK2.
Because the phenylalanine mutated PKC␦ ( 660 FAGFSF 3 VAGASV) was not recovered in the immunoprecipitate using ERK2 antibodies (Fig. 6a), we suggest that the site is involved in PKC␦/ERK2 interaction. The replacement of phenylalanine with other hydrophobic residues, albeit smaller ones, should not have changed the hydrophobicity of the sequences; accordingly, the observation cannot be explained in terms of the hydrophobic profile of the sequence and points to possibility that replacement of phenylalanines might have caused other pleiotropic effects. One such effect may relate to the phosphorylation status of Ser 664 . Ser 664 , together with Thr 507 , is phosphorylated by a pathway that involves the mammalian TOR (44), and together with Ser 645 , it is one of the phosphorylation sites required for PKC␦ activation. Kinase inactivity of the phenylalanine mutant PKC␦ is consistent with the known importance of the Ser 664 to activation of the PKC (1). Unlike the conventional and atypical PKCs, phosphorylation of the activation loop Thr 507 , however, is not a prerequisite for PKC␦ activation (45). The absence of the phenylalanine residues could affect recognition of 660 FAGFSF sequence by TOR, PDK1, and/or other upstream kinases activated by IGF-1 that are downstream of the IRK/IGF-1 receptor. The FAGFSF sequence is predicted by MotifScan search as a PDK1-binding or -docking site.
Loss of phenylalanine residues could also alter the folding and secondary structure of the PKC. The homologous residues in the recently solved structure of rat PKC␤II (46) have their phenylalanine side chains oriented toward the interior of the protein; and accordingly, it is reasonable to suggest that replacing phenylalanine residues with smaller hydrophobic amino acids would destabilize/disrupt the secondary structure of PKC␦. The PKC␦/ERK interaction is likely to require the correct spatial arrangement with respect to the bulky phenylalanine residues.
However, neither mutation of the residues nor truncation of the C-terminal 21 residues of the PKC disrupted the hBVR/ PKC␦ interaction, which suggests that this interaction is independent of the phenylalanine residues and the kinase competency of PKC␦. With respect to the effect of replacement of phenylalanine residues on interaction of the mutant proteins with ERK, it was found that the phenylalanine mutant PKC␦ was neither kinase-active nor retained its ability to bind to ERK.
Notably, it has been reported that although single mutations in the homologous residues of PKC⑀ significantly inhibited its kinase activity, they did not inhibit formation of kinase-kinase complexes with PDK1 (47). Admittedly, we did not examine the effect of mutation of phenylalanine residues individually.
Taken together, the findings of the present investigation suggest the molecular interactions that result in the formation of the ternary complex of PKC␦-hBVR-ERK2 involve the high affinity ERK binding C-box-like sequence of hBVR and low affinity D-box-like sequences in the three proteins. The proposed mode of interactions that involve the high and low affinity binding sites is depicted in Fig. 10. The suggested interactions heavily rely on MotifScan search results at the high stringency setting and on published data. Although a number of residues in the proteins are phosphorylated subsequent to stimulation of cells by IGF-1, we have focused on those residues that we suspect are most likely to affect the interactions. According to the proposed scheme, in the stimulated cells hBVR and PKC␦ interact. The association is predicted to involve the sole ERK D-domain of PKC␦ ( 590 RLGVTGNIKIHPFFK) and the D-boxlike sequence of hBVR ( 275 KKRILHCLGLA). Stimulation of cells with IGF-1 also leads to phosphorylation by IRK of hBVR Tyr 198 and Tyr 228 in its SH2 binding domains ( 198 YMKM and 228 YLSF) (22) and phosphorylation of two tyrosine residues in PKC␦, Tyr 313 and Tyr 514 (48,49). In addition, the MotifScan Search identifies Tyr 316 in ERK2 CD-domain ( 313 LEQYYDPSDEPIAE) (50) as a potential IGF-1 phosphorylation target. IGF-1 also activates the mammalian TOR pathway, PDK1, and their downstream targets that include PKC␦ (44,51).
We hypothesize that two hBVR serine residues, Ser 149 and Ser 230 , are important to its interaction with PKC␦; Ser 230 and Ser 149 are likely phosphorylation targets for PKC␦. Finding that hBVR is phosphorylated by PKC␦ (Fig. 3b), together with the hBVR-based peptides LLKGSLLFTA (where S is serine 149) and GKRNRYSFHFK (where S is serine 230) that indicated their phosphorylation by PKC␦, are consistent with the hypothesis. Notably, Ser 149 is a known target of hBVR autophosphorylation (22) and a PKC phosphorylation target (15).
If so, then we predict the following events. The formation of a complex between hBVR and PKC␦ facilitates phosphorylation of Ser 230 and Ser 149 and induction of a conformational change in hBVR. The crystal structure of hBVR (PDB accession number 2H63) locates Ser 149 and Ser 230 in adjacent antiparallel strands of the ␤-sheet domain and they are very closely juxtaposed. Upon phosphorylation, the negative charges are expected to repel each other, leading to exposure of the C-box-like motif and allowing binding of ERK2, via a sequence that thus far has not been identified. As BVR is already bound to PKC␦, its binding to ERK2, predictably, should bring the latter into proximity of the ERK D-domain of PKC␦. The CD-domain of ERK is hypothesized to compete with the hBVR D-box-like motif for binding to the ERK D-domain of PKC␦, further stabilizing the ternary complex. hBVR Ser 211 (in 209 KKSPLSWIEEKGP) and ERK2 Ser 248 (in 242 GILGSP-SQEDLNIC) are predicted phosphorylation targets for ERK2 and PKC␦, respectively.
We note that the present findings do not allow assignment of the order or affinity for binding to the D-box-like motif sequences among the components of the complex. The assignment of the hBVR C-box-like sequence as the dedicated ERK-binding site is, we believe, valid. The proposed scheme does not depict the mode of interaction of MEK, which was also in the complex of hBVR-PKC␦-ERK2, with the other components. Based on our previous study (7), we suggest involvement of the same sequences in MEK/ hBVR/ERK interactions. Also, because we detected the phosphatase PP2A in the complex, clearly what transpires upon stimulation of the cell with IGF-1 is formation of a multisubunit complex. The data do not exclude the possibility that Raf is also present in the complex (11,40).
The presently found hBVR phosphorylation by PKC␦ (Fig.  3b) and the previously reported phosphorylation of the reductase by ERK1/2 (7) may well have direct relevance to cellular defense mechanisms. To elaborate, a previous study that examined autophosphorylation of hBVR revealed dependence of the enzyme's reductase activity on its phosphorylation state (38). As demonstrated in Fig. 5d, ERK2activated hBVR displays increased reductase activity; this observation confirms the link between hBVR phosphorylation and its enzyme activity. The reductase activity is essential for the conversion of biliverdin to bilirubin. Biliverdin, a product of heme catabolism, is a global inhibitor of protein kinases (23), and bilirubin is an effective scavenger of reactive nitrogen and oxygen species (52,53); and, its anti-nitrosative activity protects against the apoptosis that accompanies mitochondrial dysfunction caused by excess NO production (10). Accordingly, the outcome of hBVR activation/increased phosphorylation may positively influence the cellular response to oxidative stress and cell signaling. Moreover, when stimulated by oxidative stress-inducing factors such as H 2 O 2 , sodium arsenite, or TNF-␣, as well as by insulin or IGF-1, production of bilirubin is increased (54 -58). As with bilirubin, CO, also a product heme cleavage by the heme oxygenase enzymes, is a contributor to cellular defense mechanisms (55,59).
Considering the many functions of hBVR in the cell, hBVR phosphorylation by PKC␦ likely would have consequences beyond immediate changes in its scaffolding and reductase activities. For instance, nuclear transport of heme, the regulator of a number of genes including HO-1, BACH1, NRF2, and NO synthase, is dependent on kinase activity of hBVR (13).
The present observations defining hBVR as a scaffold/ bridge/anchor protein for activation of ERK by PKC␦ underscores the hBVR function in MAPK/ERK signal transduction; although inhibition of PKC␦/ERK signaling by hBVR-based peptides offers the intriguing possibility for development of a FIGURE 10. Proposed mechanism for hBVR, PKC␦, and ERK2 ternary complex formation. Sites of interaction between proteins are represented by boxed sequences. IGF-1 stimulates the activation and phosphorylation of PKC␦ by upstream kinases such as PDK1 and mammalian TOR; the activated PKC is then tightly associated with hBVR. Two sequence motifs are believed to be critical for association as follows: an ERK D-domain in PKC␦ and the D-box-like sequence of hBVR. Activation and binding of PKC␦ to hBVR allow PKC␦ to phosphorylate the reductase, at Ser 149 , which is in a consensus protein kinase motif (63), and Ser 230 in the SH2 domain that is activated as a consequence of stimulation of cells with insulin/IGF-1 (22). Because the two serine residues are closely juxtaposed on adjacent antiparallel strands of a ␤-sheet (Ref. 16 and also PDB code 2H63), electrostatic forces are predicted to trigger a conformational change resulting in exposure of the C-box-like sequence of hBVR (FGFPAFS, amino acids 162-168) and formation of a ternary complex, with hBVR functioning as the bridge between PKC␦ and ERK2. The noted sequence is the high affinity ERK-binding motif (19); the sequence motif in ERK2 responsible for this binding reaction is not known. hBVR/ERK2 interaction positions PKC␦ and ERK2 domains in close proximity, allowing their interaction. Because a peptide based on the hBVR D-box inhibits interaction between PKC␦ and ERK2, the scheme predicts displacement of hBVR from PKC␦ by ERK2. However, there is a distinct possibility that a combination of PKC␦/hBVR, hBVR/ERK, and PKC␦/ERK also occurs.
new generation of small molecule inhibitors of tumor cell growth and proliferation.