Biochemical Journal

Research article

The identification and characterization of novel PKCϵ phosphorylation sites provide evidence for functional cross-talk within the PKC superfamily

Joanne Durgan, Angus J. Cameron, Adrian T. Saurin, Sarah Hanrahan, Nick Totty, Robert O. Messing, Peter J. Parker


PKCϵ (protein kinase Cϵ) is a phospholipid-dependent serine/threonine kinase that has been implicated in a broad array of cellular processes, including proliferation, survival, migration, invasion and transformation. Here we demonstrate that, in vitro, PKCϵ undergoes autophosphorylation at three novel sites, Ser234, Ser316 and Ser368, each of which is unique to this PKC isoform and is evolutionarily conserved. We show that these sites are phosphorylated over a range of mammalian cell lines in response to a number of different stimuli. Unexpectedly, we find that, in a cellular context, these phosphorylation events can be mediated in-trans by cPKC (classical PKC) isoforms. The functional significance of this cross-talk is illustrated through the observation that the cPKC-mediated phosphorylation of PKCϵ at residue Ser368 controls an established PKCϵ scaffold interaction. Thus our current findings identify three new phosphorylation sites that contribute to the isoform-specific function of PKCϵ and highlight a novel and direct means of cross-talk between different members of the PKC superfamily.

  • activation marker
  • chemical genetics
  • phosphorylation
  • protein kinase Cϵ (PKCϵ)
  • T-loop


PKCs (protein kinases C) comprise a family of phospholipid-dependent serine/threonine kinases with a broad array of cellular substrates and a central role in signal transduction [1]. This enzyme activity is conserved from yeast to mammals and has been implicated in the regulation of diverse physiological and pathophysiological processes [2]. The mammalian PKC superfamily consists of 13 different isoforms that are divided into four subgroups on the basis of their structural differences and related cofactor requirements [3,4]: cPKC (classical PKC) isoforms (α, βI, βII and γ), which respond both to Ca2+ and DAG (diacylglycerol), nPKC (novel PKC) isoforms (δ, ϵ, θ and η), which are insensitive to Ca2+, but dependent on DAG, atypical PKCs (aPKCs, ι/λ, ζ), which are responsive to neither co-factor, but may be activated by other lipids and through protein–protein interactions [5], and the related PKN (protein kinase N) family (PKN1, PKN2 and PKN3), members of which are subject to regulation by small GTPases [6].

Different PKC isoforms have broadly overlapping substrate specificities, and an element of functional redundancy exists within the superfamily [4]. Nevertheless, there is substantial evidence for numerous isoform-specific functions [7]. Insights have been gained into the non-redundant roles of PKCϵ, the PKC isoform at the focus of the present study, through the characterization of PKCϵ-null mice. These mice exhibit defects in innate immunity [8], demonstrate hypersensitivity to ethanol and other allosteric modulators of GABAA (γ-aminobutyric acid A) receptors [9] and are impaired in their capacity for ischaemic preconditioning [10]. Additional studies have implicated PKCϵ in the control of cell proliferation [11], survival [12,13], migration [14,15] and invasion [16]. Furthermore, alterations in PKCϵ have been detected in certain thyroid cancers [17,18], and this enzyme has been associated with transformation in a variety of cancer cell lines, including glioma [12], melanoma [19], squamous cell carcinoma [20] and small lung cell carcinoma [13]. Although it remains to be determined whether the dysregulation of PKCϵ represents a cause or a consequence of malignancy, it is clear that, within the context of an established tumour, this kinase may play an important role [21]. For instance, PKCϵ expression has been correlated with an aggressive, metastatic phenotype in breast cancer and has been suggested to constitute a biomarker for this condition [16]. In the light of the wide physiological and pathophysiological significance of PKCϵ, an important goal is to gain a complete understanding of the mechanisms that control its regulation and function.

PKC activity is subject to a complex network of regulatory inputs, including co-factor binding [22,23], protein–protein interaction [5, 24], regulated degradation [25] and, of particular significance here, phosphorylation. In common with other members of the AGC kinase superfamily, PKCs are phosphorylated at three conserved priming sites: the activation loop (T-loop), an autophosphorylation [turn motif/‘TP’ (Thr-Pro)] site and the hydrophobic motif [‘FSY’ (Phe-Ser-Tyr)] [26]. These sites play important structural roles [27,28] and their phosphorylation is considered to be permissive for the subsequent allosteric activation of the enzyme [29]. Certain PKC isoforms have also been shown to undergo autophosphorylation at additional sites, including PKCα at Thr250 [30], PKCβII [31] and PKCδ [32] at multiple sites, and PKCθ at Thr219 [33]. Autophosphorylation can regulate various aspects of PKC function. For instance, priming-site autophosphorylation regulates PKC activity [26,29], structural integrity [34,35], down-regulation [36] and localization [37], whereas modification of Thr219 is required for the proper membrane recruitment of PKCθ in activated T-cells [33]. In addition, autophosphorylation represents a useful experimental marker of PKC activity, which can be exploited for screening the potential involvement of these enzymes in disease. Since PKC is subject to acute allosteric regulation, it is likely to be more informative to analyse its activation state than its protein concentration in this context [38]. Ng et al. [30] utilized the PKCα Thr250 autophosphorylation site as a marker in this way and were thereby able to determine the pattern of PKCα activation in breast cancer.

In the present study we set out to identify and characterize novel autophosphorylation sites within PKCϵ, with the dual aims of exploring its isoform-specific regulation and developing tools to enable the screening of tumour samples for the activated enzyme. In the present paper we provide evidence for the in vitro autophosphorylation of PKCϵ at residues Ser234, Ser316 and Ser368, each of which is unique to this isoform and is evolutionarily conserved. Using phosphospecific antibodies, we demonstrate that these sites can be phosphorylated in various cell lines and in response to diverse agonists, including PMA, PDGF (platelet-derived growth factor) and the purinergic agonist ATP. We establish, using chemical genetics, pharmacological inhibitors and RNAi (RNA interference), that, in a cellular context, these phosphorylation events can be catalysed in-trans by a cPKC activity. Significantly, we also determine that the cPKC-dependent phosphorylation of PKCϵ at Ser368 is required to support 14-3-3β binding in MEFs (mouse embryonic fibroblasts). Together these findings identify three new PKCϵ phosphorylation sites and describe a novel and direct means of cross-talk within the PKC superfamily.



All reagents were purchased from Sigma–Aldrich unless otherwise indicated. DNA primers were synthesized by Sigma-Genosys, restriction enzymes were obtained from New England Biolabs, and kits for the purification of DNA were from Qiagen. Tissue-culture media was provided by CR-UK (Cancer Research UK) Research Services, Optimem was obtained from Gibco, and foetal-calf serum was from PAA Laboratories (Yeovil, Somerset, U.K.).

In vitro kinase assays

Autophosphorylation studies were performed in a total volume of 50 μl using 100 ng–1μg of recombinant, human PKCϵ (purified from insect cells; Calbiochem) and a kinase assay buffer containing 20 mM Tris, pH 7.5, 5 mM MgCl2, 1μg/μl phosphatidylserine (Lipid Products, Nuthill, Surrey, U.K.), 0.2% Triton X-100, 1 ng/μl PMA, 50 μM ATP and 5 μCi of [γ-32P]ATP [GE Healthcare (formerly Amersham)]. trans-Phosphorylation assays were performed using 100 ng–500 ng of recombinant bovine PKCα (purified from insect cells; kindly provided by Mr Philip Whitehead, London Research Institute, CR-UK, London, U.K.), with the addition of 200μM CaCl2 and 1mM dithiothreitol and the exclusion of [γ-32P]ATP. Reactions were initiated by the addition of ATP and the reaction mixtures were incubated for 5–120 min at 30 °C with shaking and stopped with 25 μl of kinase assay sample buffer [3×LDS (lithium dodecyl sulfate) Sample Buffer (Invitrogen), 0.2 M dithiothreitol, 4% (v/v) 2-mercaptoethanol and 5 mM EDTA]. Samples were boiled at 100 °C for 5 min, resolved by SDS/PAGE and either stained with Coomassie Brilliant Blue and analysed using the Storm 860 Phosphorimaging System (Molecular Dynamics) or proteins were visualized by Western blotting.

MS and Edman degradation

A 1 μg sample of in-vitro-autophosphorylated recombinant PKCϵ was resolved by SDS/PAGE, stained with Coomassie Brilliant Blue and subjected to in-gel digestion with 200 ng of trypsin (Promega) for 4 h at 37 °C. The resulting peptides were separated by reversed-phase HPLC using an ABI 130A separation system with a Vydac C8 reversed-phase column (1 mm diameter×150 mm long) run at a flow rate of 50 μl/min. Radioactive fractions were detected by liquid-scintillation counting and analysed by both MALDI–TOF (matrix-assisted laser-desorption ionization–time-of-flight) MS and Edman degradation. MALDI–TOF MS was performed using an Applied Biosystems 4700 Proteomics Analyzer in MS reflector-positive mode, and peptide masses were searched against the NCBI (National Center for Biotechnology Information) non-redundant database using Protein Prospector MS-FIT software. Edman degradation was carried out using an ABI Procise Sequencer and Sequelon® acrylamine membranes (Millipore).

Sequence alignments

The PKCϵ amino acid sequences from a variety of species were retrieved from the NCBI Entrez-Protein Database using the following accession numbers: Homo sapiens (man) (NP_005391.1), Macaca mulata (rhesus monkey) (XP_001112763), Bos taurus (ox) (XP_583587), Canis familiaris (domestic dog) (XP_851861), Oryctolagus cuniculus (European rabbit) (KIRBCE), Rattus norvegicus (brown rat) (NP_058867.1), Mus musculus (house mouse) (NP_035234), Gallus gallus (chicken) (XP_419464), Tetraodon nigroviridis (green spotted pufferfish) (CAG00344) and Xenopus tropicalis (pipid frog) (AAI24569). Sequence alignments were performed using ClustalW software, with default settings, according to EMBL-EBI (European Molecular Biology Laboratory European Bioinfomatics Institute, Hinxton, Cambridge, U.K.) recommendations.

Preparation of phosphospecific antibodies

Rabbit polyclonal antisera were raised to recognize PKCϵ phosphorylated at Ser234, Ser316 and Ser368. The following phosphopeptides (where pS is phosphoserine) were prepared at the Peptide Synthesis Facility at CR-UK for use as immunizing antigens: pS234 (phosphoserine234), DEVGpSQRFS; pS316, KITNpSGQRR; pS368, RKALpSFDNR (a corresponding dephosphopeptide was also synthesized for each site). The phosphopeptides were coupled to keyhole-limpet haemocyanin using glutaraldehyde and immunized in triplicate (Harlan Sera-Lab, Belton, Loughborough, Leics., U.K.). The resulting sera were tested by Western blotting against in-vitro-phosphorylated recombinant PKCϵ in the presence or absence of blocking dephospho- or phospho-peptides; at least one site- and phospho-specific serum was thus obtained for each residue: pS234, PPA-501; pS316, PPA-503; pS368, PPA-505. The pS368 antiserum, PPA-505, was affinity-purified using an Affigel 10 matrix according to the manufacturer's (Bio-Rad) instructions. Briefly, the IgG fraction was isolated from the PPA-505 terminal bleed (5 ml) using Protein A (obtained from the Monoclonal Antibody Service at CR-UK). The IgG fraction/0.02% Tween 20 was then run over a 2 ml column comprising Affigel 10 coupled to the Ser368 dephosphopeptide in order to capture binding antibodies lacking phosphospecificity. The flowthrough from this column was applied to a second 2 ml column comprising the Ser368 phosphopeptide/Affigel 10 and washed thoroughly using PBS/0.02% Tween 20, PBS/0.5 M NaCl/0.02% Tween 20 and PBS/0.02% Tween 20 again. Bound phosphospecific antibodies were eluted using 0.1 M citrate (pH 2.5)/0.02% Tween 20 and mixed 2:1 with 1 M Tris/HCl, pH 8.8, before storage.

cDNA constructs and site-directed mutagenesis

The cloning of mouse wtPKCϵ (wild-type PKCϵ) into the vector pEGFP-C1 was described previously [14]. Compared with that with the accession number NM_011104, this cDNA has the following ‘silent’ mutations: T861C, T918C, T1089C, C1176T, G1236A, A1554G, C1578T and T2175C; notably, T2175C creates an EcoRI site (see below). A panel of PKCϵ phosphorylation-site mutants was generated by site-directed mutagenesis using a two-step PCR approach. Mouse PKCϵ in a modified pMT2 vector (XmaI/KpnI) was used as a template. The flanking primers 5′-gaacacacccgggatggtagtgttcaatggccttc-3′ and 5′-catctatggtaccggggcatcaggtcttcaccaa-3′ were employed in combination with the following complementary mutagenic primer pairs to generate the indicated mutations (underlined bases encode the altered amino acid): S234A, 5′-gacgaggtgggcgcccaacggttcagc-3′ and 5′-gctgaaccgttgggcgcccacctcgtc-3′; S234D, 5′-gacgaggtgggcgaccaacggttcagc-3′ and 5′-gctgaaccgttggtcgcccacctcgtc-3′; S316A, 5′-gacaaaatcaccaacgctggccaaaggagg-3′ and 5′-cctcctttggccagcgttggtgattttgtc-3′; S316D, 5′-gacaaaatcaccaacgatggccaaaggagg-3′ and 5′-cctcctttggccatcgttggtgattttgtc-3′; S368A, 5′-cggaaggccttggcatttgacaaccgagg-3′ and 5′-cctcggttgtcaaatgccaaggccttccg-3′; S368D, 5′-cggaaggccttggattttgacaaccgagg-3′ and 5′-cctcggttgtcaaaatccaaggccttccg-3′. In order to clone the resulting mutants into pEGFP-C1, a fragment containing the altered codon(s) was transferred into the pEGFP-C1 wtPKCϵ backbone by exploiting the unique flanking EcoRV (593–598) and EcoRI (2170–2175; see above) restriction sites. For retroviral infection, constructs were subcloned into a modified pBABE vector containing an EGFP (enhanced green fluorescent protein) tag (introduced using the BamHI site). PKCϵ was inserted between the EcoRI and SalI sites using the following PCR primers: forward, 5′-taataatacaattgatggtagtgttcaatggccttcttaag-3′ (creates an MfeI site compatible with EcoRI); and reverse, 5′-tatgtttcaggttcagggggag-3′ (amplifies from the template vector pEGFP-C1). A chemical-genetic mutant of mouse PKCϵ, namely PKCϵ M486A, was cloned using 5′-ccgcctcttcttcgtcgcggaatatgtaaatggtgg-3′ (forward) and 5′-ccaccatttacatattccgcgacgaagaagaggcgg-3′ (reverse) primers. All constructs generated by PCR were validated by DNA sequencing.

Cell culture

COS7 and HeLa cells were obtained from CR-UK Research Services and maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) foetal-calf serum and penicillin (100 units/ml)/streptomycin (100 mg/ml) at 37 °C in a humidified atmosphere containing 5% CO2. The preparation and culture of PKCϵ−/− (PKCϵ knock-out) MEFs have been described previously [14]. N-terminally GFP (green fluorescent protein)-tagged PKCϵ and PKCϵ M486A were introduced into PKCϵ−/− MEFs using the pBABE retroviral system. Briefly, constructs were transfected into the BOSC cell packaging line, and supernatants containing retrovirus were collected 48 h post-transfection. Retroviruses were filtered using a 0.44-μm-pore-size filter and applied to subconfluent MEFs in the presence of Polybrene® (4 μg/ml). Cells were selected with puromycin (2.5 μg/ml; Sigma) and expression was assessed by FACS analysis.

Transfection of siRNA and plasmid DNA

Control siRNA (small interefering RNA; aatcgaagtattccgcgtacg) and a PKCα targeting siRNA pool (Oligo 1, aaggcttccagtgccaagttt; Oligo 2, aagaggtgccatgaatttgtt; Oligo 3, aagacgagctatttcagtcta; Oligo 4, aagcccaaagtgtgtggcaaa) were obtained from Qiagen. Transfection of siRNA (20 nM) and plasmid DNA was performed using Lipofectamine® 2000 (Invitrogen) as recommended by the manufacturer.

Agonists and inhibitors

At 24 h after either plating (MEFs) or transfection (COS7, HeLa), cells were treated or not with 400 nM PMA or 400 nM PMA+100 nM calyculin A (Calbiochem) as indicated in the Figure legends. Before treatment of subconfluent MEFs with either 50 ng/ml PDGF-BB (PDGF B-chain homodimer; PeproTech, Rocky Hill, NJ, U.S.A.) or 100 ng/ml ATP, the cells were first serum-starved in 0.25% foetal-bovine serum/DMEM for 16 h. Where included, kinase inhibitors were routinely added 30 min prior to stimulation as follows: 2 μM or 10 μM BIMI (bisindolylmaleimide I), 1 μM Gö6976 [12-(2-cyanoethyl)-6,7,12,13-tetrahydro-13-methyl-5-oxo-5H-indolo(2,3-a)pyrrolo-(3,4-c)-carbazole], 6 μM Gö6983 {2-[1-(3-dimethylaminopropyl)-5-methoxyindol-3-yl]-3-(1H-indol-3-yl) maleimide}, 1 μM staurosporine or 2 μM 1-NaPP1 {4-amino-1-t-butyl-3-(1′-naphthyl)pyrazolo[3,4-d] pyrimidine} (all from Calbiochem). In each case, DMSO was used to prepare stocks and perform carrier controls.

Immunoprecipitations and GST (glutathione transferase)–14-3-3-β pull-downs

After transfection and/or treatment as indicated in Figure legends, cells were washed once in ice-cold PBS and harvested by scraping into pre-chilled lysis buffer [50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% Nonidet P40, 0.1 mM Na3VO4, 10 mM NaF, 10 nM calyculin A (Calbiochem) and 1× Protease Inhibitor Cocktail (Roche)]. After incubation on ice for 30 min, insoluble material was pelleted by centrifugation at 13000 g for 15 min at 4 °C. An aliquot of the supernatant was removed (total cell lysate) and the remainder was subjected to either immunoprecipitation or GST–14-3-3-β pull-down. For immunoprecipitation, lysates were tumbled with pre-equilibrated Protein A–Sepharose 4B beads and mouse anti-GFP clone 4E12/8 (CR-UK Monoclonal Antibody Facility) at 4 °C for 2 h. Immunocomplexes were recovered by brief centrifugation at 1000 g, washed three times for 10 min with lysis buffer and then boiled at 100 °C for 5 min in LDS sample buffer [2× LDS Sample Buffer (Invitrogen)/0.2 M dithiothreitol]. GST–14-3-3-β pull-downs were performed using a similar protocol, except that the lysis buffer contained 1% Triton X-100 instead of 0.5% Nonidet P40 and glutathione–Sepharose 4B beads, bearing GST–14-3-3β, were used in place of Protein A–Sepharose 4B.

Western blotting

PAGE was performed using the NuPAGE system with BisTris/4–12%-(w/v)-polyacrylamide gels and Mops/SDS Running Buffer according to the manufacturer's (Invitrogen) instructions; samples were resolved alongside RPN800 Full Range Rainbow™ Molecular Weight Markers (GE Healthcare). Proteins were transferred to methanol-soaked PVDF membranes (Millipore) using a Trans-Blot Cell (Bio-Rad) and transfer buffer [10 mM Tris/80 mM glycine/10% (v/v) methanol]. All subsequent incubations were performed in TBS-T (20 mM Tris, pH 7.5, 0.9% NaCl and 0.1% Tween-20) at room temperature (20°C) with shaking. Membranes were blocked for 1 h in 2% (w/v) BSA/TBS-T and then incubated for a further 1 h with one of the following antibodies: 1:1000 rabbit anti-(PKCϵ pS234), PPA-501; 1:1000 rabbit anti-(PKCϵ pS316), PPA-503; 1:1000 rabbit anti-(PKCϵ pS368), PPA-505; 1:1000 anti-(PKCϵ pT-loop), PPA-204 [39]; 1:5000 rabbit anti-PKCϵ, sc-214 (Santa Cruz), 1:1000 mouse anti-PKCα, Clone MC5 (CR-UK Monoclonal Antibody Facility) or Clone 3 (BD Transduction Laboratories); or 1:1000 mouse anti-GFP, clone 3E1 (CR-UK Monoclonal Antibody Facility). Where primary antibodies were phosphospecific, 1 μg/ml specific blocking dephosphopeptide was included in the primary incubations. The membranes were washed three times for 10 min with TBS-T, incubated with 1:5000 HRP (horseradish peroxidase)-conjugated secondary antibody (GE Healthcare) for 45 min and washed again three times for 10 min with TBS-T. Bound HRP-conjugated antibodies were visualized using the enhanced chemiluminescence (ECL®) detection system and Hyperfilm according to the manufacturer's (GE Healthcare) guidelines. Images were captured using an Epson Expression 1680 Pro scanner and Adobe® Photoshop® CS software; where indicated, band intensities were quantified using National Institutes of Health ImageJ software. The Western blots shown are representative of results obtained from at least three separate experiments.

Confocal microscopy

COS7 or HeLa cells were seeded on acid-washed glass coverslips and transfected with GFP–PKCϵ or GFP–PKCϵ S368A. At 24 h post-transfection the cells were treated or not with 400 nM PMA for 40 min. The cells were then fixed with 4% (w/v) paraformaldehyde/PBS for 10 min, quenched using 50 mM NH4Cl/PBS for 10 min and permeabilized and blocked using 0.2% Triton X-100/2% BSA/TBS (20 mM Tris, pH 7.5, and 0.9% NaCl) for 10 min, all at room temperature. The coverslips were incubated with affinity-purified anti-(PKCϵ pS368) [PPA-505; 1:100 in 2% (w/v) BSA/TBS+1μg/ml blocking dephosphopeptide] for 1 h at room temperature, washed three times for 10 min in TBS and then stained using Alexa Fluor 568 goat anti-rabbit secondary antibody for 1 h at room temperature (Molecular Probes; 1:500 in 2% BSA/TBS). Finally, the coverslips were washed three times for 10 min in TBS, once in water and mounted on to glass slides using a solution containing 10% (v/v) Mowiol 4-88 (Calbiochem), 25% (v/v/) glycerol and 100 mM Tris, pH 8.5. Images were acquired using an upright confocal laser scanning microscope (LSM 510; Carl Zeiss, Jena, Germany) equipped with a 63×-magnification 1.4-numerical-aperture Plan-Apochromat Ph3 oil-immersion objective and 488 nm/543 nm laser excitation lines. For each data set, laser settings were optimized for the ‘GFP–PKCϵ+PMA’ samples and then used consistently under all other conditions, such that the signal intensity could be compared between images. Individual channels were scanned sequentially, line by line, with averaging set at 8. Each image (12 bit, 1024 pixels) represents a single 1.0 μm ‘Z’ optical section with a pixel size of approx. 0.1 μm and is representative of data obtained from at least three separate experiments.


Identification of three novel in vitro PKCϵ autophosphorylation sites.

Preliminary in vitro kinase assays were performed using recombinant human PKCϵ and [γ-32P]ATP. Incorporation of 32P was detected over time by phosphorimaging (Figure 1a), indicating that PKCϵ undergoes autophosphorylation under these conditions. We found the in vitro phosphorylation of PKCϵ to be lipid-dependent and sensitive to the cPKC/nPKC inhibitor BIMI, consistent with an autocatalytic process (results not shown). In order to identify the phosphorylated residue(s), we carried out phosphopeptide mapping analyses. In-vitro-phosphorylated PKCϵ was digested with trypsin and fractionated by reverse-phased HPLC. The resulting fractions were counted for radioactivity to locate 32P-labelled phosphopeptides. As shown in Figure 1(b), four major peaks of 32P were detected after HPLC. Fractions 6/7 correspond to the unbound material (including free orthophosphate) and were not studied further, whereas fractions 18, 28 and 36 were all subsequently analysed by both MS, to identify the phosphorylated peptides, and Edman degradation, to determine which residue within each peptide is modified. Figure 1(c) represents the mass spectrum corresponding to Fraction 18. The peak at m/z 1452.6525 precisely matches the mass of a phosphorylated tryptic peptide containing human PKCϵ residues 225–236; the broader peak detected at m/z 1356.6982 is characteristic of the post-source decay of this phosphopeptide, which releases phosphoric acid. The unphosphorylated form of the peptide comprising residues 225–236 is not apparent in this spectrum and it is likely that it was eluted in a different HPLC fraction. Since there are two candidate serine/threonine residues within the sequence 225–236, namely Thr228 and Ser234, we used Edman degradation to obtain positional information. N-terminal amino acids were sequentially removed from the phosphorylated peptide and liquid-scintillation counting of radioactivity revealed that the 32P-labelled residue is located at position 10 and thus corresponds to Ser234 (Figure 1d). Using the same approach, we were able to identify Ser368 as an in-vitro-phosphorylation site (Figures 1e and 1f) from fraction 28 (and also from an alternative tryptic peptide, corresponding to residues 365–377, in fraction 27; results not shown) and Ser316 from fraction 36 (Figures 1g and 1h). In summary, these data identify three novel in vitro PKCϵ phosphorylation sites, namely Ser234, Ser316 and Ser368.

Figure 1 PKCϵ in vitro autophosphorylation and phosphopeptide mapping

(a) In vitro kinase assays were performed using 100 ng of recombinant PKCϵ and the assay mixtures were incubated for 0–120 min. Samples were resolved by SDS/PAGE and analysed by phosphorimaging. This image is representative of data obtained from three separate experiments. (b) In-vitro-phosphorylated PKCϵ was digested with trypsin and the resulting peptides were separated by reversed-phase HPLC. Liquid-scintillation counting of radioactivity was performed to identify fractions containing 32P-labelled phosphopeptides. Fractions 18 (c), 28 (e) and 36 (g) were then analysed by MALDI-TOF-MS in order to identify the phosphorylated peptides. The residue numbers indicated correspond to full-length human PKCϵ (NP_005391); phosphorylation increases peptide mass by 80 Da and post-source decay (PSD) of the phosphopeptide decreases this mass by 98 Da. The same fractions were then subjected to Edman degradation to identify the phosphorylated residue within the peptide (d, f, and h). Fractions representing individual residues were subjected to liquid-scintillation counting. The sequence of the corresponding peptide identified by MS is shown, and the major phosphorylated site is indicated with an arrow.

Cellular phosphorylation of PKCϵ at Ser234, Ser316 and Ser368 sites

In order to determine whether the in vitro PKCϵ phosphorylation sites Ser234, Ser316 and Ser368 are modified in mammalian cells, we raised phosphospecific polyclonal antibodies. To verify the specificity of these reagents, a panel of GFP-tagged PKCϵ constructs was cloned in which each candidate phosphorylation site was mutated to either a non-phosphorylatable alanine residue (S234A, S316A or S368A) or to a phosphomimetic aspartate moiety (S234D, S316D or S368D). These constructs were transfected into COS7 cells, the cultures were treated or not with the phorbol ester PMA, a cPKC/nPKC activator, and then processed for Western blotting with the pS234-, pS316- or pS368-specific sera (Figure 2a). Phosphorylation of each site could be detected in response to PMA in wtPKCϵ, whereas no immunoreaction was detected in the corresponding serine-to-alanine or serine-to-aspartic acid mutants. Notably, a low level of basal Ser368 phosphorylation was also observed. These results demonstrated that PKCϵ can be phosphorylated at residues Ser234, Ser316 and Ser368 in a cellular context and confirmed the specificity of each antiserum. In addition, we observed that the PMA-induced phosphorylation of these sites was significantly decreased in the presence of BIMI, an inhibitor of cPKCs/nPKCs (Figure 2b), consistent with a possible autophosphorylation event (see further below).

Figure 2 Stimulation of PKCϵ phosphorylation by PMA, PDGF and ATP in mammalian cells

(a and b) COS7 cells were seeded on six-well plates and transfected with GFP-PKCϵ [wt (‘WT’)] or one of a panel of GFP-PKCϵ mutants (S234A/D, S316A/D, S368A/D). At 24 h post-transfection, cells were treated or not with 1 μM BIMI for 30 min and then stimulated or not with 400 nM PMA for a further 45 min. Lysates were prepared and subjected to immunoprecipitation with anti-GFP 4E12/8. Samples were analysed by Western blotting using phospho-specific antibodies pS234, pS316 or pS368, or anti-GFP clone 3E1 for total protein. (c and d) GFP-PKCϵ expressing PKCϵ−/− MEFs were seeded on 6-cm-diameter dishes. After 24 h, the cells were serum-starved overnight and then treated or not with 50 ng/ml PDGF-BB or 100 ng/ml ATP for 0–30 min. Immunoprecipitations and Western blotting were performed as described above. These results are representative of data obtained from three separate experiments.

We next tested whether PKCϵ is phosphorylated in other cell types and in response to more physiologically relevant agonists. PKCϵ−/− MEFs, into which GFP–PKCϵ was stably reintroduced, were stimulated with either PDGF (Figure 2c) or ATP, a purinergic agonist (Figure 2d). No increase in either Ser234 or Ser316 phosphorylation was detected under these conditions (results not shown). However, as shown in Figure 2(c), PDGF was found to stimulate a transient increase in PKCϵ phosphorylation at Ser368, which peaks 2–10 mins after application (3.0±1.3-fold increase; n=3) and persists for less than 30 min. Similarly, as shown in Figure 2(d), treatment with ATP stimulates a 3.7±1.7-fold (n=3) increase in Ser368 phosphorylation approx. 2 min post-treatment, which, again, returns to basal levels within 30 min. Interestingly, we have also detected Ser368 phosphorylation in RAW 264.7 macrophages following LPS (lipopolysaccharide) treatment (A. Faisal and P.J. Parker, unpublished work). Together, these findings confirm that residues Ser234, Ser316 and Ser368 can be phosphorylated in mammalian cells in a manner dependent on PKC activity and reveal that residue Ser368 can be phosphorylated on PKCϵ activation in response to various agonists in a range of cell lines.

Localization of PKCϵ pS368

In order to investigate the subcellular localization of PKCϵ pS368, the anti-(PKCϵ pS368) antiserum was affinity-purified and used for immunofluorescence [the anti-(pS234 PKCϵ) and anti-(PKCϵ pS316) sera did not prove amenable to such studies under the conditions tested]. As shown in Figure 3, COS7 and HeLa cells were transfected with either GFP–wtPKCϵ or GFP–PKCϵ S368A, treated with PMA as indicated, and then analysed for pS368 staining by confocal microscopy. In unstimulated cells, we found both wtPKCϵ and PKCϵ S368A to be distributed throughout the cytosol, with some perinuclear accumulation; on the basis of published observations in other cell lines [40], it is likely that the latter compartment represents the Golgi. Interestingly, staining with the anti-(pS368 PKCϵ) antibody reveals that the perinuclear subpopulation of wtPKCϵ is detectably phosphorylated under basal conditions; this observation is most pronounced in HeLa cells, but a low-level signal was also reproducibly identified in COS7 cells. The pS368 signal detected under these conditions is clearly specific, since it is absent when the S368A mutant is expressed in either cell line. This finding is also consistent with the basal phosphorylation of pS368 observed by Western blotting (see Figure 2a). On stimulation with PMA, both wtPKCϵ and PKCϵ S368A are translocated to the plasma membrane. The anti-pS368 signal is clearly increased in cells expressing the wt, but not the mutant protein, indicating that PKCϵ undergoes Ser368 phosphorylation on its activation at the membrane. For each data set, we used consistent laser settings for all samples so that the signal intensities could be directly compared. We also verified that the phospho signal was lost in the presence of the Ser368 phosphopeptide and absent when the anti-rabbit secondary antibody was used alone (results not shown). As such we conclude that our anti-(pS368 PKCϵ) antibody can be utilized effectively to localize pS368 PKCϵ by immunofluorescence, and that it may prove a useful tool for detecting activated PKCϵ species.

Figure 3 Localization of Ser368-phosphorylated GFP–PKCϵ in COS7 and HeLa cells

(a) COS-7 or (b) HeLa cells were seeded on coverslips and tranfected with either GFP–PKCϵ or GFP–PKCϵ S368A. At 24 h post-transfection, the cells were treated or not with 400 nM PMA or a DMSO carrier control for 40 min. Cells were fixed, permeablized, blocked and stained for anti-(pS368 PKCϵ) (Alexa Fluor 568, red). Coverslips were mounted and analysed by confocal microscopy (LSM 510; Carl Zeiss). For each dataset, laser settings were used consistently, such that the signal intensity can be compared between images. All images comprise single 1.0 μm ‘Z’ optical sections and are representative of data obtained from at least three separate experiments. Scale bars represent 10 μm.

Chemical-genetic mechanistic analyses show that PKCϵ phosphorylation can occur in-trans

The results presented above demonstrate that PKCϵ phosphorylation can occur in a purified in vitro system and in cells on treatment with a PKC activator in a manner sensitive to cPKC/nPKC inhibition by BIMI. As such, we reasoned that the cellular phosphorylation of PKCϵ at residues Ser234, Ser316 and Ser368 is likely to proceed via an autocatalytic mechanism. In order to test this hypothesis, we employed a chemical-genetic approach of the kind pioneered by the Shokat laboratory [41]. This method involves the mutation of a single residue within the kinase of interest, which renders it uniquely sensitive to inhibition by 1-NaPP1, a modified (1-naphthyl) analogue of PP1 {4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine}. Other properties of the kinase, including its catalytic activity and sensitivity to other inhibitors, are predicted to remain unaltered, and 1-NaPP1 has not been reported to inhibit any endogenous kinases [42]. As such, the inhibition of the sensitized kinase can be studied in a highly specific manner.

A sensitized PKCϵ mutant, PKCϵ M486A, was cloned by Qi et al. [43] and shown to be selectively and effectively inhibited by 1-NaPP1. We established a system in which either GFP–PKCϵ M486A or a GFP–wtPKCϵ control were expressed in PKCϵ−/− MEFs such that the effect of 1-NaPP1 could be assessed against a knockout background, excluding any possible contribution from the endogenous kinase (Figure 4). Figure 4(a) demonstrates that, as we observed in other cell lines, phosphorylation of wtPKCϵ can be detected in MEFs in response to acute PMA treatment and that this process is inhibited by BIMI; as would be expected, 1-NaPP1 has no effect on the phosphorylation of the wt kinase. It is noteworthy that the phosphorylation of residues Ser234 and Ser316 is somewhat delayed relative to that of Ser368. This variation in kinetics could be indicative of a difference in catalytic mechanism or turnover. However, this is a matter that remains to be investigated further. We next performed the same experiment using the sensitized PKCϵ M486A mutant (Figure 4b). In common with the wt protein, PKCϵ M486A undergoes PMA-stimulated BIMI-sensitive phosphorylation at residues Ser234 and Ser316. Furthermore, since no decrease in the phosphorylation of these sites is detected in the presence of 1-NaPP1, we might conclude that, unexpectedly, their phosphorylation does not occur by autocatalysis but rather is mediated in-trans by another kinase. However, our interpretation of these data is confounded by additional observations made in this experiment. First, as shown in Figure 4(b), PKCϵ M486A is not detectably phosphorylated at residue Ser368 in response to PMA, indicating that there are secondary consequences associated with the M486A mutation which complicate this system. Secondly, and somewhat counterintuitively, we find that the phosphorylation of PKCϵ M486A at all three sites is actually stabilized in the presence of 1-NaPP1. In order to better understand these observations, we investigated additional properties of PKCϵ M486A. We found that, unlike the wt protein, this mutant is prone to undergo dephosphorylation at its activation loop (T-loop) in response to PMA (Figures 4a and 4b); this effect is blocked by all PKCϵ inhibitors tested (BIMI and Gö6983; results not shown), including 1-NaPP1. Since T-loop phosphorylation is critical for PKC activity [26], its instability in PMA-treated cells is likely to compromise the ability of PKCϵ M486A to autophosphorylate, limiting the utility of this mutant under these conditions. Although complicated by this issue, these data nevertheless provide some insight into the mechanism of PKCϵ phosphorylation. Notably, the fact that PKCϵ M486A is phosphorylated at residues Ser234 and Ser316, regardless of its T-loop dephosphorylation, is still consistent with a mechanism in which these sites are modified in-trans. However, the data regarding the PMA-induced phosphorylation of Ser368 cannot be interpreted unambiguously.

Figure 4 Chemical-genetic analyses of PKCϵ phosphorylation in MEFs

(a) GFP–PKCϵ or (b) GFP–PKCϵ M486A expressing PKCϵ−/− MEFs were seeded on six-well plates and serum-starved overnight. The cells were pretreated or not with 2 μM 1-NaPP1 or 10 μM BIMI for 20 min, and then stimulated with 400 nM PMA for a further 0–60 min. Lysates were prepared and subjected to immunoprecipitation with anti-GFP 4E12/8. The samples were then analysed by Western blotting using phosphospecific antibodies pS234, pS316, pS368 or pT-loop (PPA-204), or anti-GFP clone 3E1 for total protein. (c) As (a), except that cells were serum-starved overnight, pretreated or not with 2 μM 1-NaPP1 or 10 μM BIMI, and then stimulated or not with 50 ng/ml PDGF-BB for 0–10 min. (d) As (b), except that cells were serum-starved overnight, pre-treated or not with 2 μM 1-NaPP1 or 10μM BIMI, and then stimulated or not with 50 ng/ml PDGF-BB for 0–10 min. These results are representative of data obtained from three separate experiments.

Since residue Ser368 can be phosphorylated in response to additional stimuli, we repeated our chemical-genetic analyses using PDGF as an alternative agonist (Figures 4c and 4d). Under these conditions, we found the T-loop phosphorylation site to be stable in both the wt and the M486A mutant, such that the effects of 1-NaPP1 could be tested and interpreted straightforwardly. Figure 4(c) demonstrates that wtPKCϵ was transiently phosphorylated at Ser368 in response to PDGF in a BIMI-sensitive fashion, as would be expected. 1-NaPP1 did not block this phosphorylation event. PKCϵ M486A was also shown to be Ser368-phosphorylated in a PDGF-dependent BIMI-sensitive manner (Figure 4d). Significantly here, since 1-NaPP1 does not block phosphorylation under these conditions, we can conclude that Ser368 is modified in-trans by a distinct kinase activity.

In summary, these data suggest that, contrary to expectation, PKCϵ residues Ser234, Ser316 and Ser368 do not represent obligate autophosphorylation sites, but can be modified in-trans. This finding is significant with respect to the use of these phosphorylation sites as markers of PKCϵ activation (see the Discussion). Interestingly, it also indicates that PKCϵ is subject to regulation by another kinase activity.

PKCϵ can be trans-phosphorylated by cPKCs in MEFs, COS7 cells and in vitro

Since the phosphorylation of PKCϵ at residues Ser234, Ser316 and Ser368 was found to be sensitive to BIMI (a cPKC/nPKC inhibitor), we hypothesized that another PKC isoform may be responsible for catalysis. To test this prediction, we examined the effects of an extended panel of pharmacological inhibitors on the PMA-induced phosphorylation of wtPKCϵ in both MEFs (Figure 5a) and COS7 cells (Figure 5b). These compounds, Gö6976, BIMI and Gö6983, exhibit different, but overlapping, patterns of specificity and could thereby be used in parallel to implicate a particular PKC subclass in a given catalytic event (see Figure 5c and [38]). Consistent with earlier experiments, we found that PKCϵ is phosphorylated at residues Ser234, Ser316 and Ser368 in response to PMA treatment in both cell lines. In MEFs, this effect is even more striking in the presence of calyculin A, a Protein phosphatase 2A/protein phosphatase 1 inhibitor, presumably because dephosphorylation is blocked, allowing a greater differential to accumulate. As shown in Figures 5(a) and 5(b), we found that modification of each site is sensitive to all of the inhibitors tested; Gö6976 mediates a partial reduction in phosphorylation, while BIMI and Gö6983 both cause more pronounced blocks. Significantly, since Gö6976 is not able to inhibit nPKCϵ itself (see Figure 5c), this data confirms unequivocally that these sites can be modified in-trans. Furthermore, this pattern of pharmacology suggests that a cPKC activity may catalyse the trans-phosphorylation of PKCϵ. However, it is noteworthy that, since the broader specificity compounds BIM1 and Gö6983 mediate more robust inhibition, phosphorylation is unlikely to be mediated exclusively in a cPKC-dependent manner. Rather, it remains possible that additional PKC superfamily members also contribute to phosphorylation under these conditions.

Figure 5 cPKC-dependent phosphorylation of PKCϵ in MEFs, COS7 and in vitro

(a) GFP–PKCϵ expressing PKCϵ−/− MEFs were seeded on six-well dishes and serum-starved overnight or (b) COS7 cells were seeded on six-well plates, transfected with GFP–PKCϵ and incubated for 24 h. The cells were pre-treated or not for 20 min with 2 μM Gö6976, 6 μM Gö6983, 10 μM BIMI or 1 μM staurosporine (STS) and then stimulated or not with 400 nM PMA or 400 nM PMA+100 nM calyculin for a further 40 min. Lysates were prepared and subjected to immunoprecipitation with anti-GFP 4E12/8. The samples were then analysed by Western blotting using phosphospecific antibodies pS234, pS316 or pS368, or anti-GFP clone 3E1 for total protein. These results are representative of data obtained from three separate experiments. (c) Reported specificities of PKC kinase inhibitors (c, classical; n, novel; a, atypical PKC isoforms) [38]. Other kinases are indicated as inhibitor targets if their activity was reduced by >50% under the conditions tested by Davies et al. [44]. Abbreviations: AMPK, AMP-activated protein kinase; CHK2, checkpoint kinase 2; ERK, extracellular-signal-regulated kinase; PDK, 3′-phosphoinositide-dependent protein kinase; PHK, phosphorylase kinase; ROCK-II, Rho-associated kinase II. (d) COS7 cells were transfected or not with 20 nM control or PKCα-targeting siRNA. After 24 h, the cells were transfected with GFP–PKCϵ, incubated for a further 24 h and then treated or not with 400 nM PMA or 400 nM PMA+100 nM calyculin for 20 min. Lysates were prepared and subjected to immunoprecipitation as described above. Immunoprecipitates were analysed by Western blotting for pS368 or GFP; lysates were probed for PKCα (clone 3) or tubulin. Band intensities were quantified from three separate experiments and pS368 values are presented, corrected for total PKCϵ levels (results are means±S.D., n=3). (e) In vitro kinase assays were performed using 100 ng of recombinant PKCϵ in the presence or absence of 100 ng of recombinant PKCα. Reaction mixtures were incubated for 0–30 min and the reaction stopped by the addition of sample buffer and 5 μM EDTA. Samples were resolved by SDS/PAGE and analysed by Western blotting as described above; anti-PKCϵ and anti-PKCα were used to visualize total protein. These results are representative of data obtained from three separate experiments.

A limitation of the approach described above is that many pharmacological compounds can inhibit multiple kinases (see Figure 5c and [44]). Thus, although our data are consistent with the cPKC-mediated phosphorylation of PKCϵ, we cannot formally exclude the involvement of other enzymes; for instance, GSK (glycogen synthase kinase), RSK (p90 ribosomal S6 kinase), MSK (mitogen and stress-activated protein kinase) and S6K (p70 S6 kinase) family members have all been shown to be inhibited by both Gö6976 and BIMI [44]. As such, we sought to test the involvement of cPKCα in PKCϵ phosphorylation more directly using RNAi. COS7 cells were co-transfected with GFP–PKCϵ and either a control or a PKCα-targeting siRNA pool. The cells were then stimulated or not with PMA and analysed by Western blotting (Figure 5d); levels of Ser368 phosphorylation were quantified by densitometry, corrected for total PKCϵ loading (using GFP as a marker) and plotted graphically (we consistently observed a decrease in GFP–PKCϵ expression on PKCα depletion, which is accounted for in this analysis). As in previous experiments, we detected an increase in PKCϵ phosphorylation in the presence of PMA. Significantly, we also observed that the level of Ser368 phosphorylation was decreased from 442±111 to 187±62 (n=3) on depletion of cPKCα (initial experiments indicate that Ser234 and Ser316 phosphorylation is reduced in the same manner on PKCα knockdown; results not shown). These findings reinforce the observations described above and indicate that cPKCα is likely to represent the major Gö6976-sensitive kinase capable of mediating PKCϵ phosphorylation.

Finally, we returned to a purified system and tested whether recombinant cPKCα can promote the in vitro phosphorylation of recombinant nPKCϵ. The results presented in Figure 5(e) prove that this is the case. In the absence of cPKCα, a low level of PKCϵ phosphorylation at residues Ser234, Ser316 and Ser368 can be detected, consistent with the in-vitro-autophosphorylation results presented in Figure 1. However, the addition of cPKCα clearly increases the rate of phosphorylation, indicating that each modification can also be mediated by this PKC isoform in-trans.

Together, these data show that, in MEFs, COS7 cells and in vitro, PKCϵ can be phosphorylated in a cPKC-dependent manner. This finding points to a direct mechanism for cross-talk between different PKC isoforms, from within different subgroups of the superfamily, and reveals yet another level of complexity within the PKC signalling network.

cPKC–mediated phosphorylation of PKCϵ at S368 is required for 14-3-3β association in MEFs

Recent work in our laboratory has revealed that an inducible PKCϵ–14-3-3β complex can be assembled in a manner dependent on the phosphorylation of PKCϵ at residue Ser368 (A.T. Saurin, J. Durgan, A. J. Cameron, A. Faisal, M. S. Marber and P. J. Parken, unpublished work). We therefore sought to explore the functional significance of cPKC-mediated PKCϵ phosphorylation by testing whether this activity can facilitate 14-3-3β association. Consistent with our previous work, the interaction between PKCϵ and 14-3-3β is stimulated by treatment with PMA/calyculin A and dependent on the phosphorylation of residue Ser368; it is not influenced by the phosphorylation of either Ser234 or Ser316 (Figure 6a). As shown in Figure 6(b), complex assembly is found to correlate precisely with Ser368 phosphorylation. If phosphorylation is blocked, using either the cPKC inhibitor Gö6976, or BIMI, the interaction of PKCϵ with 14-3-3β is also prevented. It is concluded that the trans-phosphorylation of PKCϵ at residue Ser368 by a cPKC is required to support 14-3-3β binding and hence that this isoform relationship represents a functionally significant link.

Figure 6 cPKC-mediated phosphorylation of PKCϵ at Ser368 is required for 14-3-3β association

(a) COS7 cells were seeded on six-well plates and transfected with one of a panel of GFP–PKCϵ constructs [wt (‘WT’), S234A, S316A and S368A]. At 24 h post-transfection, cells were treated or not with 400 nM PMA/100nM calyculin for 45 min. Lysates were prepared and subjected to GST–14-3-3-β pulldown. Samples were analysed by Western blotting using anti-PKCϵ antibody; GST–14-3-3β was visualized by Coomassie Blue staining. (b) GFP–PKCϵ expressing PKCϵ−/− MEFs were seeded on six-well plates. The cells were pretreated or not with 1 μM Gö6976 or 10 μM BIMI for 20 min and then stimulated or not with 400 nM PMA/100nM calyculin for a further 20 min. Lysates were prepared and subjected to GST–14-3-3-β pulldown. Samples were analysed by Western blotting using anti-(PKCϵ pS368) and anti-GFP clone 3E1 for total protein; GST–14-3-3β was visualized by Coomassie Blue staining. The results are representative of data obtained from three separate experiments.


PKCϵ belongs to the novel subclass of the PKC superfamily, and is characterised as a DAG/phorbol-ester-dependent, but calcium-insensitive, serine/threonine kinase [45]. This enzyme has been implicated in the regulation of a diverse range of cellular processes and has been shown to play a protective role in cardiac ischemia and Alzheimer's disease, whereas its dysregulation has been associated with cancer and diabetes [46]. In order to gain further understanding of the regulation and function of this enzyme, we sought to identify and characterize novel PKCϵ autophosphorylation sites. Our expectation was that this approach might (1) provide insights into its isoform-specific regulation, and (2) may lead to the development of reagents specific for the activated enzyme, such that this species might be analysed selectively in pathological samples.

Using phosphopeptide mapping analyses, we have identified three novel PKCϵ in vitro autophosphorylation sites: Ser234, which lies between the tandem C1a and C1b DAG/phorbol-ester-binding domains, and Ser316 and Ser368, which are located in the variable hinge region (V3 domain) of the molecule (see Figure 7a). All three of these residues, as well as the sequences surrounding them, are conserved among a range of different vertebrates, from the frog Xenopus to humans (Figure 7b; equivalent sites were not identified in PKC sequences from the fruitfly Drosophila, the nematode worm Caenorhabditis, the sea slug Aplysia or yeast). This evolutionary conservation suggests that these sites may be of functional significance. Moreover, since these motifs are not found in any other members of the PKC superfamily, it is likely that they contribute to the isoform-specific regulation and function of PKCϵ.

Figure 7 Sequence conservation of novel PKCϵ phosphorylation sites Ser234, Ser316 and Ser368

(a) The domain structure of PKCϵ is represented diagrammatically, with the positioning of novel phosphorylation sites Ser234, Ser316 and Ser368 highlighted and the surrounding sequences shown (dots represent the intervening residues). Conserved (C1, C2-like, C3 and C4) and variable (V1, V2, V3 and V4) regions are also indicated. (b) Sequence alignments were performed using ClustalW software and PKCϵ amino acid sequences from a variety of species (residue numbers correspond to the human sequence; the organisms are more specifically defined with their Latin names in the text). Within the sequence text, a dash (−) creates a space to facilitate alignment, whereas a wavy line (∼) represents intervening residues. Below the alignment, the following symbols denote the degree of conservation:*, residues are identical; :, conserved substitutions; ., semi-conserved substitutions; a space indicates no conservation.

The addition of a phosphate moiety to a particular amino acid has an effect on both the local charge and structure of the protein and may mediate a range of effects. Phosphorylation has previously been shown to influence PKC conformation [27,28], activity [26,29], localization [33,37], down-regulation [36,47,48] and protein–protein interactions [49]. To date we have not observed any gross differences between wtPKCϵ and the S234A/D, S316A/D or S368A/D mutants with respect to their activity, localization or degradation. However, we have established that the phosphorylation of residue Ser368 is indispensable for the recruitment of 14-3-3β. Recent work in our laboratory has elucidated a critical role for a PKCϵ–14-3-3β complex in controlling the final stages of cytokinesis (A.T. Saurin, J. Durgan, A. J. Cameron, A. Faisal, M. S. Marber and P. J. Parker, unpublished work). As such, at least one of these newly identified phosphorylation sites does indeed play a key role in controlling the function of PKCϵ. It is also of note that Ser234, which sits between the two zinc-finger motifs of the C1 domain, is located in an equivalent position to the functionally important PKCθ autophosphorylation site, Thr219. Thr219 is required for the proper membrane recruitment of PKCθ in activated T-cells [33], and hence it will be pertinent to explore whether the modification of Ser234 influences PKCϵ translocation or residency at the membrane. In addition, Ser234 is situated in close proximity to the PKCϵ actin-binding motif LKKQET (residues 223–228 [50]) and immediately adjacent to an eight-residue sequence (QRFSVNMP, residues 235–242) critical for PKCϵ-mediated induction of neurite outgrowth [51]. As such, it will be of interest to study the effects of this site on cell morphology.

We demonstrate that, in mammalian cells, the phosphorylation of PKCϵ at residues Ser234, Ser316 and Ser368 can be stimulated on PKC activation with the phorbol ester PMA. Interestingly, the time course of Ser234 and Ser316 PMA-induced phosphorylation was delayed relative to that of Ser368; it thus appears that, in cells, these sites are either less efficiently phosphorylated or that they are more readily susceptible to phosphatase action. Significantly, we were also able to detect phosphorylation of Ser368 in MEFS in response to either PDGF or ATP, and in macrophages upon treatment with LPS (A. Faisal and P.J. Parker, unpublished work), suggesting that numerous signalling pathways may converge on the functionally significant phosphorylation of PKCϵ Ser368. It will be of particular interest to explore the latter observation more fully, because deficiency in LPS signalling is a characteristic of PKCϵ−/−-knockout mice [8].

Since Ser234, Ser316 and Ser368 were identified as in vitro autophosphorylation sites, and since their phosphorylation could be stimulated in cells by treatment with the PKC activator PMA and blocked by the cPKC/nPKC inhibitor BIMI, we reasoned that the cellular phosphorylation of these sites occurs via an autocatalytic mechanism. In order to test this hypothesis, we employed a chemical-genetic approach whereby a single PKCϵ residue was mutated (M486A) in order to confer sensitivity to inhibition by a modified PP1 analogue 1-NaPP1. In the absence of an isoform-specific inhibitor of catalytic activity, this technique afforded a unique opportunity to selectively regulate and study PKCϵ. To exclude any contribution from the endogenous kinase, we expressed the sensitized mutant against a PKCϵ null background. Using this system, we discovered that, surprisingly, these sites can be modified in-trans by another kinase. However, it is important to note that we detected a secondary consequence of the sensitizing mutation (PKCϵ M486A): unlike the wt protein, the mutant is prone to undergo T-loop dephosphorylation following PMA stimulation. Thus, although chemical genetics is an immensely powerful tool [41], our current findings emphasize the need to extensively characterize such mutants in order to interpret the results obtained in an informed manner.

In order to investigate the nature of the endogenous trans-acting kinase responsible for PKCϵ phosphorylation, we made use of a panel of pharmacological inhibitors, namely Gö6976, BIM1 and Gö6983, with different, but overlapping, specificities. In this way we deduced that the stimulated phophorylation of PKCϵ at Ser234, Ser316 and Ser368 may be catalysed by one or more of the cPKC isoforms (PKCα, PKCβI, PKCβII and PKCγ). This conclusion was reinforced by our observation that the RNAi-mediated knockdown of PKCα significantly decreases the level of PMA-induced PKCϵ phosphorylation in COS7 cells and, furthermore, by our finding that purified PKCα can contribute to the phosphorylation of recombinant PKCϵ in vitro. We would emphasize, however, that our findings do not exclude the possibility that PKCϵ is able to autophosphorylate at these sites under certain conditions. Although our chemical-genetic analyses revealed that autocatalysis does not contribute to the PDGF-stimulated phosphorylation of PKCϵ S368 in MEFs, our pharmacological data revealed that broader-specificity PKC inhibitors (BIMI and Gö6983) block PMA-induced phosphorylation more effectively than does the cPKC-specific Gö6976 in MEFs and COS7 cells. Thus, it seems likely that other c/nPKC isoforms, including PKCϵ itself, may catalyse phosphorylation in response to PMA. Thus, our current working model proposes that, upon cellular stimulation with PMA, cPKCs and nPKCs are co-recruited to the plasma membrane and, in this compartment, activated cPKCα is able to trans-phosphorylate nPKCϵ, and that PKCϵ may also undergo autophosphorylation. In light of these findings, it would be of interest to test whether other PKC ‘auto’-phosphorylation sites, such as PKCα Thr250 or PKCθ Thr219, may also be modified in-trans under certain conditions.

Our current findings describe a novel example of cross-talk between different PKC family members. This is of particular relevance in the light of recent work from Collazos et al. [52], who demonstrated that the recruitment of PKCϵ to cell–cell contacts in pituitary GH3B6 cells after thyrotropin-releasing-hormone treatment is dependent on the activity of cPKCα. Specifically, it would be intriguing to establish whether Ser234, Ser316 and/or Ser368 are modified under these conditions and whether the localization of the corresponding phospho-site mutants is disturbed. More broadly, the interplay between cPKCs and nPKCϵ suggests that an additional level of complexity exists within the PKC signalling network. The present study has verified that this relationship can be of regulatory significance, since cPKC activity was shown to be critical for the induced assembly of a PKCϵ–14-3-3 complex. It will be important to investigate further the functional outcomes of cPKC-catalysed PKCϵ phosphorylation, since this raises the possibility that certain PKCα functions may operate through the regulation of PKCϵ.

One of the original aims of the present study was to develop tools for the detection of activated PKCϵ with which to screen tumour samples for this species. Since PKCϵ has been consistently associated with a transformed phenotype [46], and has been postulated as a biomarker of aggressive breast cancer [16], it represents a compelling candidate for further investigation in this context. Also, given that PKC is an allosterically regulated enzyme, it is likely to be more meaningful to test for its activity than its expression level in a disease setting [30]. We hypothesized that antibodies directed against sites of autophosphorylation would represent useful markers for such screening purposes. Previously, the Thr250 autophosphorylation site of PKCα has been exploited in this way and used to identify activated PKCα in 11 out of 23 breast-tumour specimens [30]; notably, in that study, activation did not correlate with overexpression. Our efforts to develop direct activation markers for PKCϵ have been confounded somewhat by our discovery that the phosphorylation of residues Ser234, Ser316 and Ser368 does not occur exclusively through an autocatalytic mechanism. As such, phospho-antibodies raised against these sites do not constitute markers of inherent catalytic activity. Nevertheless, since PKCϵ phosphorylation at these sites increases upon its activation, presumably as a consequence of its membrane recruitment and conformational change, these antibodies may still be used as extractable markers of the subpopulation of PKCϵ which has sampled the membrane and been activated. As such, the tools developed in the present study may yet prove useful for screening purposes.

In conclusion, the present study describes the identification and characterization of three novel PKCϵ phosphorylation sites. We find that these residues can be phosphorylated in-trans by a cPKC activity and thus provide evidence for a novel and direct means of cross-talk between different members of the PKC superfamily.


We thank Mr James Dodds for his assistance with MS, Mr Philip Whitehead for providing purified PKCα, members of P. J. P.'s laboratory for helpful discussions and Professor Alan Hall (Memorial Sloan-Kettering Cancer Center, New York, NY, U.S.A.) for supporting this work. J. D. was generously funded by the Medical Research Council and CR-UK. R. O. M. acknowledges support of grant AA013588 from the U.S. Public Health Service and funds provided by the State of California for medical research on alcohol and substance abuse through the University of California at San Francisco.

Abbreviations: BIMI, bisindolylmaleimide I; cPKC, classical protein kinase C; CR-UK, Cancer Research UK; DAG, diacylglycerol; DMEM, Dulbecco's modified Eagle's medium; EGFP, enhanced green fluorescent protein; GFP, green fluorescent protein; GST, glutathione transferase; HRP, horseradish peroxidase; LDS, lithium dodecyl sulfate; LPS, lipopolysaccharide; MALDI-TOF, matrix-assisted laser-desorption ionization–time-of-flight; MEF, mouse embryonic fibroblast; NCBI, National Center for Biotechnology Information; nPKC, novel protein kinase C; PDGF, platelet-derived growth factor; PDGF-BB, PDGF B-chain homodimer; PKC, protein kinase C; PKN, protein kinase N; pS, (in single-letter amino acid sequences), phosphoserine; pS234, phosphoserine-234 (etc.); RNAi, RNA interference; siRNA, small interfering RNA; wtPKCϵ, wild-type PKCϵ


View Abstract