Protein Scaffolds Control Localized Protein Kinase Cζ Activity*

Atypical protein kinase C (aPKC) isozymes modulate insulin signaling and cell polarity, but how their activity is controlled in cells is not well understood. These enzymes are constitutively phosphorylated, insensitive to second messengers, and have relatively low activity. Here we show that protein scaffolds not only localize but also differentially control the catalytic activity of the aPKC PKCζ, thus promoting activity toward localized substrates and restricting activity toward global substrates. Using cellular substrate readouts and scaffolded activity reporters in live cell imaging, we show that PKCζ has highly localized and differentially controlled activity on the scaffolds p62 and Par6. Both scaffolds tether aPKC in an active conformation as assessed through pharmacological inhibition of basal activity, monitored using a genetically encoded reporter for PKC activity. However, binding to Par6 is of higher affinity and is more effective in locking PKCζ in an active conformation. FRET-based translocation assays reveal that insulin promotes the association of both p62 and aPKC with the insulin-regulated scaffold IRS-1. Using the aPKC substrate MARK2 as another readout for activity, we show that overexpression of IRS-1 reduces the phosphorylation of MARK2 and enhances its plasma membrane localization, indicating sequestration of aPKC by IRS-1 away from MARK2. These results are consistent with scaffolds serving as allosteric activators of aPKCs, tethering them in an active conformation near specific substrates. Thus, signaling of these intrinsically low activity kinases is kept at a minimum in the absence of scaffolding interactions, which position the enzymes for stoichiometric phosphorylation of substrates co-localized on the same protein scaffold.

Atypical protein kinase C (aPKC) isozymes modulate insulin signaling and cell polarity, but how their activity is controlled in cells is not well understood. These enzymes are constitutively phosphorylated, insensitive to second messengers, and have relatively low activity. Here we show that protein scaffolds not only localize but also differentially control the catalytic activity of the aPKC PKC, thus promoting activity toward localized substrates and restricting activity toward global substrates. Using cellular substrate readouts and scaffolded activity reporters in live cell imaging, we show that PKC has highly localized and differentially controlled activity on the scaffolds p62 and Par6. Both scaffolds tether aPKC in an active conformation as assessed through pharmacological inhibition of basal activity, monitored using a genetically encoded reporter for PKC activity. However, binding to Par6 is of higher affinity and is more effective in locking PKC in an active conformation. FRETbased translocation assays reveal that insulin promotes the association of both p62 and aPKC with the insulin-regulated scaffold IRS-1. Using the aPKC substrate MARK2 as another readout for activity, we show that overexpression of IRS-1 reduces the phosphorylation of MARK2 and enhances its plasma membrane localization, indicating sequestration of aPKC by IRS-1 away from MARK2. These results are consistent with scaffolds serving as allosteric activators of aPKCs, tethering them in an active conformation near specific substrates. Thus, signaling of these intrinsically low activity kinases is kept at a minimum in the absence of scaffolding interactions, which position the enzymes for stoichiometric phosphorylation of substrates co-localized on the same protein scaffold.
The coordination of signal transduction by protein scaffolds controls downstream signaling for a multitude of protein kinases (1). Protein scaffolds locally enrich the kinase, positioning it near or sequestering it away from substrates to allow specificity and fidelity in cell signaling. In addition, binding to scaffold proteins can impact the conformation of kinases, tuning their signaling output. The atypical PKC isozymes (aPKC) 3 are an example of a kinase whose interaction with a scaffold tethers them in a signaling-competent conformation.
aPKCs make up one of three classes of the PKC family of enzymes. These classes are the canonical diacylglycerol (DAG)regulated conventional (Ca 2ϩ -dependent; ␣, ␤, and ␥) and novel (Ca 2ϩ -independent but also DAG-regulated; ␦, ⑀, , and ) PKCs and the aPKCs ( and /), which are regulated by neither DAG nor Ca 2ϩ (2)(3)(4). aPKCs share the same general architecture of other family members, with a C-terminal kinase domain whose function is controlled by determinants in the N-terminal regulatory moiety. Specifically, all PKCs have an autoinhibitory pseudosubstrate segment immediately preceding a C1 domain that masks the kinase domain to maintain the enzyme in an autoinhibited conformation. In addition, all PKCs are constitutively phosphorylated following their biosynthesis (3,5). aPKCs differ from the DAG-regulated PKCs in that 1) they are not regulated by second messengers, 2) their catalytic activity is an order of magnitude lower than the DAG-regulated PKCs (5), and 3) they have a protein-protein interacting Phox and Bem1 (PB1) domain at their N terminus. Additionally, they have a type III PDZ ligand at the end of their C terminus (3). For the DAG-regulated PKCs, binding of second messengers results in release of the pseudosubstrate. In the case of aPKCs, binding to protein scaffolds promotes the release of the atypical pseudosubstrate.
Two scaffolds that regulate the activity of aPKCs are p62 (also known as sequestosome 1, SQSTM1) (6) and partitioning-defective protein 6 (Par6), a cell polarity regulator (7). Binding to these scaffolds promotes the open and active conformation of the aPKCs. In the case of p62, an acidic surface unique to its PB1 domain binds the basic pseudosubstrate of aPKCs, tethering it away from the substrate binding cavity to engage aPKC in an active conformation (6). Binding of aPKC to Par6 also displaces the pseudosubstrate to tether aPKC in an open and signalingcompetent conformation (7). Par6 is well characterized as a signaling platform to localize aPKC near substrates such as the cell polarity regulators Par3 and Lgl (8 -13). As a signaling hub with multiple interacting domains and partners (14,15), p62 may also function to localize aPKC near its targets, although specific substrates on this platform remain to be identified.
A key function of aPKC is the regulation of insulin-stimulated glucose transport (16 -21), yet the mechanism for how it transduces this signal is poorly understood. We have recently shown that PKC is not regulated by the insulin-induced PI3kinase pathway that activates Akt (5), a mechanism previously proposed to regulate aPKC activity (22)(23)(24). Rather, scaffold interactions may be the critical regulators of aPKC function in insulin signaling. In this regard, insulin promotes a multitude of phosphorylation-dependent protein interactions, many centralized around the insulin receptor substrate (IRS-1/2) (25). The p62 scaffold has recently been shown to interact with IRS-1 in an insulin-dependent manner (26,27) and is also known to have functions in metabolism as p62-deficient mice exhibit an insulin-resistant, obese, and diabetic phenotype (28,29). A metabolic phenotype is also observed in mice lacking microtubule affinity regulating kinase 2 (MARK2), an aPKC substrate (30,31). These animals are insulin-hypersensitive and resistant to weight gain when placed on a high fat diet (32,33). MARK2 has a well characterized role in mediating cell polarity, but the mechanism by which it regulates insulin signaling is less understood. How aPKCs control substrate phosphorylation and downstream signaling in response to insulin remains to be elucidated.
Here, we address the role of protein scaffolds in controlling the activity of aPKCs. Specifically, we examine the scaffoldassociated activity of full-length PKC or constructs lacking the pseudosubstrate, PB1, or the entire N-terminal regulatory moiety using the C kinase activity reporter (CKAR) (34) fused to the PB1 domain of p62 or Par6. Basal activity at each scaffold is assessed by monitoring the drop in activity following addition of a validated aPKC active site inhibitor, PZ09. Our data reveal that aPKC is differentially bound and activated on the two PB1 scaffolds; it binds with higher affinity to Par6, which tethers it in a fully active conformation, compared with binding to p62 which is of lower affinity and which tethers a partially active enzyme. Furthermore, we demonstrate the ability of IRS-1 to regulate the phosphorylation and localization of the aPKC substrate MARK2. Finally, a fluorescent resonance energy transfer (FRET) assay reveals that insulin promotes the association of PKC with p62 in addition to p62 and PKC with IRS-1. Our data support a model in which scaffolds regulate aPKC signaling by not only providing a signaling scaffold but by differentially tuning activity.
Cell Lines and Plasmid Constructs-Human PKC and PKC␣ cDNA were gifts from Dr. Tony Hunter. Human p62 was a gift from Dr. Jorge Moscat. Human MARK2 and human IRS-1 were purchased from Addgene (plasmid no. 66706 and plasmid no. 11025, respectively), whereas human Par6␣ was a gift from Dr. Sourav Ghosh, originally from Addgene (plasmid no. 15474). The protein CKAR construct was described previously (34,35). Human PKC, PKC␣, MARK2, Par6␣, IRS-1, and p62 constructs were cloned into the pDONR vector and subsequently recombined with various pDEST vectors constructed in-house to make fusion proteins with HA, mCherry, CKAR, CFP, or YFP tags at the N terminus in pCDNA3 vectors for mammalian cell expression using the Gateway cloning system (Invitrogen). Human PKM (residues 183-592) of PKC was inserted into pDONR for Gateway recombination. The PB1 domains of both human p62 (residues 3-102) and human Par6␣ (residues 1-101) were cloned into pDONR and inserted into a Gateway expression vector for expression with N-terminal CFP and CKAR tags. Domain deletion mutations were constructed in PKC as follows: ⌬PS (deleted pseudosubstrate, residues 113-130) and ⌬PB1 (deleted aPKC PB1 domain, residues 25-106). Point and domain mutations were made using QuikChange site-directed mutagenesis (Stratagene). The catalytically inactive spine mutant (V266F (5)) was used for the kinase-dead controls of PKC and PKM.
Live Cell Fluorescence Imaging-COS-7 cells were plated onto sterilized glass coverslips in 35-mm imaging dishes, co-transfected with the indicated constructs, and imaged in Hanks' balanced salt solution supplemented with 1 mM CaCl 2 ϳ24 h post-transfection using a ϫ40 objective. For experiments with insulin stimulation, cells were serumstarved overnight prior to imaging. For experiments with forskolin treatment, cells were treated ϳ24 h prior to imaging and transfected the day before. Kinase activity was monitored via intramolecular FRET of the activity reporters (CKAR or CKAR-PB1), and protein translocation was monitored via intermolecular FRET between the YFP-tagged protein and the CFP-tagged target using methods previously described (6,36).
Statistical Analysis-All statistical analyses were performed using GraphPad Prism software.

PZ09 Inhibits Atypical PKCs but Not Conventional PKCs in
Cells-The toolbox of inhibitors that selectively inhibit atypical PKCs is scant compared with the collection of compounds that can effectively inhibit conventional and novel PKCs for pharmacological studies (37). Additionally, certain compounds previously claimed to be inhibitors of aPKC from in vitro studies (ZIP and chelerythrine (38,39)) have been shown to be ineffective at inhibiting aPKC in cells (40). Given the need for an effective aPKC modulator to investigate its biochemical regulation, and the history within the field of in vitro efficacy not concurring with efficacy in cells, we set out to validate an active site inhibitor of aPKC, PZ09 (Fig. 1A), previously shown to inhibit  The trace for each cell imaged was normalized to its t ϭ 0-min baseline value, and normalized FRET ratios were combined from three independent experiments with analysis of 8 -12 selected cells each and plotted as mean Ϯ S.E. D, COS-7 cells were serum-starved overnight and treated with either DMSO, staurosporine (Staur) (1 M), or increasing concentrations of PZ09 for 30 min prior to stimulation with 100 nM insulin for 10 min before lysis. Immunoblots show endogenous substrate phosphorylation representing aPKC inhibition (p595 MARK2), PDK1 inhibition (insulin-stimulated p308 Akt), and cPKC inhibition (pSer sub, using an antibody for a PKC-specific serine substrate sequence). Quantification of phosphorylated protein substrate over total protein normalized to DMSO ϩ insulin control using n ϭ 5 separate experiments plotted as mean Ϯ S.E. versus inhibitor concentration (PZ09 or staurosporine) is shown. Phosphorylated PKC serine substrate was quantified as the intensity of the total band ensemble detected between 50 and 250 kDa divided by tubulin signal and normalized to DMSO ϩ insulin control.
aPKC in vitro (41) and already used for studies in vivo (42) for selective modulation of aPKC substrate readouts in cells. Using a genetically encoded FRET reporter, the C kinase activity reporter (CKAR) (34) previously validated for measurement of aPKC activity in real time in live cells (5,6,40), we examined inhibition of endogenous aPKC or mCherry-tagged constructs of aPKC by PZ09. The drop in activity following inhibitor addition serves as a measure of the constitutive activity of aPKC. Henceforth, this inhibitor-sensitive activity is referred to as "basal activity." Fig. 1B shows the effect of increasing concentrations of PZ09 on PKM, an alternate transcript of PKC preferentially expressed in brain tissue that comprises only the kinase domain. PKM was effectively inhibited by PZ09, with an IC 50 of ϳ3 M. The addition of inhibitor caused a small drop in FRET readout in mCherry-Vector (Vec)-transfected control cells, reflecting inhibition of endogenous aPKCs and background activity on CKAR by other kinases sensitive to PZ09 (Fig. 1B). The larger drop in FRET readout following equal PZ09 treatment for the PKM-transfected cells versus the Vectransfected cells is indicative of higher basal activity for the overexpressed PKM on CKAR than for the endogenous kinases on CKAR, as each trace is normalized to its baseline value before inhibitor treatment. As an additional control, a kinase-dead version of mCherry-PKM with a catalytically inactive spine mutation V266F (5) was tested on CKAR and displayed the same response as the mCherry-Vec control (data not shown). To examine the selectivity of PZ09 in cells for aPKC versus conventional PKCs, cells expressing CKAR and mCherry-PKC␣ were treated with 200 nM phorbol dibutyrate to stimulate cPKC activity and then treated with either 5 M PZ09 or 250 nM Gö6983, a potent inhibitor of cPKCs and novel PKCs (Fig. 1C). Gö6983 reversed stimulated PKC␣ activity (Fig.  1C, blue trace); in contrast 5 M PZ09 had no effect on PKC␣ activity in cells (Fig. 1C, red trace), confirming previous selectivity results performed in vitro (41). Thus, monitoring PZ09sensitive phosphorylation of CKAR is an effective tool to measure aPKC activity in cells.
Higher Concentrations of PZ09 Have Off-target Effects on Endogenous PDK1 in Cells-Although in vitro inhibition studies of PZ09 have demonstrated its selectivity for aPKC over other PKCs (41), PZ09 also inhibited several other non-PKC kinases, including PDK1, PKA, and p70 S6K when used at 10 M concentration. To examine the effects of increasing PZ09 concentrations on endogenous readouts of aPKC, PDK1, and cPKC, we pre-treated cells with PZ09 followed by stimulation with insulin prior to lysis to achieve optimal readout of PDK1 activity on the agonist-induced site p308 Akt (lane 2 versus lane 1, Fig. 1D). We then blotted for p595 MARK2 (aPKC substrate (30,31)) and p308 Akt (PDK1 substrate (43)) or phosphorylated PKC serine substrate using an antibody specific for the PKC substrate recognition sequence and quantified each phosphorylation relative to total protein. As a control for general kinase inhibition, we also treated cells with 1 M staurosporine (Fig.  1D, lane 3), an inhibitor we have previously used to examine aPKC activity in cells (40). Both aPKC and PDK1 were inhibited to the same extent by increasing concentrations of PZ09 up to 10 M (Fig. 1D, lanes 4 -6 compared with lane 2 control, p595, and p308 readouts, respectively), whereas cPKC experienced significantly less inhibition by PZ09 (Ser(P) PKC substrate readout). However, off-target effects of PZ09 on PDK1 and cPKC were considerably less than those of the general kinase inhibitor staurosporine used at 1 M concentration (quantification, Fig. 1D).
PKC Is Constitutively Active on CKAR Substrate Reporter Tethered to Interacting PB1 Domains of Scaffold Proteins p62 and Par6 -To investigate the role of scaffolds in regulating the localized activity of PKC, we fused CKAR to either the PB1 domain of p62 or the PB1 domain of Par6␣ ( Fig. 2A). We have previously used a fusion construct of CKAR tethered to fulllength p62 to measure aPKC activity (6). Because this reporter had a tendency to form large clusters of aggregated p62 within the cells, we fused CKAR to just the PB1 domain of p62 or the PB1 of Par6␣ (which also forms aggregates when expressed in full-length form (7)) as the PB1 domain of each scaffold is the principal surface for aPKC binding (44,45). Untethered CKAR was expressed throughout the cell; the CKAR-PB1 p62 reporter became excluded from the nucleus, localizing within the cytosol in the same regions as PKC, with minimally visible puncta ( Fig. 2A). The CKAR-PB1 Par6 construct was only partially excluded from the nucleus ( Fig. 2A). We next examined the PZ09-sensitive basal activity globally or on the p62 or Par6 scaffolds ( Cells overexpressing mCherry-PKC displayed much more significant drops in basal activity than for mCherry-Vec on each of the various reporters (compare Fig. 2, panel ii to panel i, same scale on each y axis), indicating aPKC-specific activity. The greatest inhibitor-sensitive activity of mCherry-PKC was also observed on the Par6 scaffold (CKAR-PB1 Par6 , Fig. 2, panel ii). Curiously, lower basal activity was detected on the CKAR-PB1 p62 reporter compared with CKAR alone, although there was a pronounced drop in activity compared with its vectortransfected control (Fig. 2, panel ii versus panel i). The pronounced difference in basal activity readout by the two PB1 domain-tagged reporters prompted us to investigate whether the PB1 domain of Par6 was binding more tightly to PKC, thus recruiting more of the overexpressed enzyme to phosphorylate CKAR. Co-immunoprecipitation (co-IP) experiments shown in Fig. 2C revealed that PKC pulled down significantly more CKAR-PB1 Par6 (lane 3) than CKAR-PB1 p62 (lane 2) (2.5 Ϯ 0.5fold, p ϭ 0.0187) as determined through quantification of anti-GFP signal (detecting CKAR) over anti-PKC signal from four independent experiments. PKC did not pull down the CKAR reporter alone (lane 1), confirming its non-interaction with the untethered substrate and reliance on the PB1 domain for scaffolding. These results reveal that aPKC is basally active at both the Par6 and p62 scaffolds, with significantly more effective binding and thus activity at the Par6 scaffold.
Negative Charge at Ser-24/Ala-30 in the PB1 Domains of p62 and Par6␣ Impairs Binding of PKC and Activity on CKAR-PB1 Par6 -AproteinkinaseA(PKA)-mediatedphosphorylation site present in the PB1 domain of p62, residue Ser-24, has recently been identified (46). Phosphorylation at this site is proposed to regulate aPKC binding to the PB1 domain of p62; co-IP studies comparing S24A and S24D mutants showed that negative charge at this site dismantled the ability of aPKC to bind the p62 scaffold (46). Intriguingly, the sequence surrounding the Ser-24 site of p62 is present in the PB1 domain of the Par6 isoforms, except that an Ala occupies the position of the Ser for Par6␣ (residue Ala-30) but not for Par6␤ and Par6␥ (Fig.  3A). To address whether negative charge at this position in the PB1 domain of Par6␣ would dismantle the capacity to bind PKC, we constructed an A30D mutant in our CKAR-PB1 Par6 reporter. Indeed, co-IP studies revealed that PKC binding to the A30D mutant was dramatically reduced compared with binding to wild-type Par6␣ PB1 (Fig. 3B; 95 Ϯ 1% reduction, p Ͻ 0.0001 from three independent experiments). Subsequently, we compared the activity of full-length mCherry-PKC on CKAR-PB1 Par6 in which the PB1 domain was wild-type or had the A30D mutation (Fig. 3C). In accordance with the co-IP results, the basal activity of PKC was significantly reduced on the A30D mutant compared with the wild-type (WT) version of the Par6␣ PB1 reporter and comparable with its vector-transfected control. We also examined whether mutating the Ser-24 site on the PB1 domain of p62 affected PKC binding and activity. Activity readout by the S24A CKAR-PB1 p62 reporter demonstrated no difference in overexpressed or endogenous PKC basal activity compared with the wild-type version (Fig. 3D) and did not differ in its ability to bind PKC in co-IP (data not shown). The readout of the S24D CKAR-PB1 p62 reporter for PKC basal activity also did not change from the wild-type version, although this mutation on p62 was previously shown to displace binding of aPKC in co-IP (46). To examine the effect of Ser-24 phosphorylation on PKC binding to p62, cells were treated with forskolin (24 h at 50 M) prior to lysis, previously shown to induce PKA-mediated phosphorylation of Ser-24 (46). Treatment with forskolin increased Ser(P)-24 on CFP-PB1 p62 and diminished binding to PKC (40 Ϯ 8% reduction, p ϭ 0.005) as demonstrated through co-IP from six independent experiments (Fig. 3E). Basal Activity of PKC Is Regulated by a Combination of Autoinhibition and Scaffold-regulated Localization or Sequestration-We next investigated the role of the N-terminal regulatory domains, namely the PB1 domain and the pseudosubstrate (PS), in regulating the scaffolded activity of PKC on the various CKAR reporters examined in Fig. 2. The alternative transcript PKM lacks the N-terminal regulatory domains (PB1, pseudosubstrate, and atypical C1 regions) and consists of the kinase domain, the C-terminal region, and the PDZ ligand (Fig. 4A). In addition to PKM, we constructed two deletion mutants, PKC ⌬PS (deleted pseudosubstrate) and PKC ⌬PB1 (deleted PB1 domain), to investigate the role of these domains in regulating the activity of PKC on scaffolds (Fig. 4A). Co-expressing these constructs first with non-tethered CKAR and measuring basal activity after treatment with PZ09 (Fig. 2B), PKM demonstrated the most pronounced basal activity on CKAR. The ⌬PS and ⌬PB1 constructs displayed equal basal activity to full-length PKC, which was significantly increased from the vector-transfected control. Next, we examined the basal activity of the PKC constructs on the CKAR-PB1 p62 reporter, this time showing equal activity between PKC ⌬PS and PKM that was more pronounced than full-length PKC and PKC ⌬PB1 (Fig. 4C). On CKAR-PB1 Par6 , PKC ⌬PS and PKM once again displayed equal basal activity but were slightly less active than full-length PKC and more active than PKC ⌬PB1 (Fig. 4D). To compare the binding of PKC versus PKM on the reporters, we immunoprecipitated HA-PKC or HA-PKM co-expressed with CKAR-PB1 or CKAR and examined co-IP of the reporters using the ␣GFP antibody for detection (Fig. 4E). Intriguingly, PKM was capable of binding the untethered version of CKAR, although PKC did not (Fig. 4E, 4th versus 3rd lane, note identities of bands in ␣GFP blot of the immunoprecipitate indicated by red arrows as the ␣GFP antibody also recognized HA-aPKC signal), in addition to binding CKAR-PB1 Par6 (2nd lane). This binding interaction was confirmed to be between PKM and the CFP/YFP fluorophores present in CKAR, as PKM also co-immunoprecipitated CFP alone but not Myc-PB1 Par6 (data not shown), thus confirming the expected non-interaction between PKM and PB1.
PKM Is More Sensitive to Dephosphorylation at the Activation Loop Thr-410 than PKC-We have previously used the general kinase inhibitor staurosporine to inhibit the basal activity of PKM on CKAR (40). The inhibitory action of staurosporine on PKM is mostly indirect through its potent active site inhibition of PDK1 thus reducing basal phosphorylation on the PDK1 site of PKM, the activation loop Thr-410 (40). Inhibition

FIGURE 3. Negative charge at Ser-24/Ala-30 in the PB1 domains of p62 and Par6␣ impairs binding of PKC and activity on CKAR-PB1 Par6 . A, diagram
showing consensus within the PB1 domain for the Ser-24 site of p62 versus the Ala-30 site of Par6␣ and corresponding Ser sites in Par6␤ and Par6␥. B, HA-PKC was co-expressed with either wild-type (WT) or A30D mutation of CKAR-PB1 Par6 in COS-7, immunoprecipitated from soluble lysates using anti-HA antibody, and blotted for co-IP of CKAR tag using anti-GFP antibody, and whole cell lysate was loaded at 10% input. C and D, basal kinase activity of mCherry-PKC versus mCherry-Vec on either WT or A30D CKAR-PB1 Par6 (C) or WT, S24A, or S24D CKAR-PB1 p62 (D) after treatment with 5 M PZ09 in live COS-7 cells. The trace for each cell imaged was normalized to its t ϭ 0-min baseline value and plotted as means Ϯ S.E. Normalized FRET ratios were combined from four independent experiments (all traces in C and CKAR-PB1 p62 S24A Vec trace in D), five independent experiments (CKAR-PB1 p62 WT traces and CKAR-PB1 p62 S24D Vec trace), or six independent experiments (CKAR-PB1 p62 S24A and CKAR-PB1 p62 S24D PKC traces) with each experiment analyzing 6 -13 selected cells. E, HA-PKC was co-expressed with WT CFP-PB1 p62 , treated for 24 h prior to lysis with either DMSO or 50 M forskolin (Fsk) in COS-7, immunoprecipitated from soluble lysates using anti-HA antibody, and blotted for co-IP of CKAR tag using anti-GFP antibody along with a phospho-specific antibody for Ser(P)-24 p62; whole cell lysate was loaded at 10% input.
of PDK1 permits conquest of PKM by opposing phosphatases acting basally on Thr-410, a phosphorylation site necessary for aPKC catalytic activity (5,23). To investigate the role of N-terminally interacting aPKC scaffolds in regulating dephosphorylation of PKC at Thr-410, we transfected COS-7 cells with either mCherry-PKM or mCherry-PKC and treated them with 1 M staurosporine for 2 h prior to lysis (Fig. 5A). Phosphorylation of Thr-410 was significantly reduced by 2-fold for PKM (Fig. 5A, lane 2   effect on p410 levels (Fig. 5A, lane 4 versus lane 3). Thus, p410 on full-length PKC is less sensitive to PDK1 inhibition than PKM, indicating a protective mechanism against dephosphorylation provided by the N-terminal regulatory domains of PKC. PKM but Not PKC Is Active on Global Substrates although Both Are Basally Active on MARK2 Substrate-PKC is known to phosphorylate and interact with microtubule affinity regulating kinase 2 (MARK2) (30,31), a phosphorylation event at the Thr-595 site of MARK2, which results in the translocation of phosphorylated, inactive MARK2 away from the plasma membrane and into the cytosol (30,31). However, the activity of aPKC on other non-interacting substrates that contain a PKC phosphorylation site is unclear. Using p595 MARK2 as a read-out for aPKC activity in cells, we compared the catalytic activities of overexpressed PKM versus full-length PKC on global PKC substrates, using an antibody for PKC serine substrate (Fig. 5B). Expression of PKM (Fig. 5B, lane 2) revealed an ensemble of induced serine substrate bands (including a higher molecular weight band previously seen with PKM (40)) that were not present in lysates transfected with PKC (lane 3), indicating catalytic activity of PKM on multiple substrates inaccessible to PKC. Both kinase constructs were fully active basally to the same extent on co-expressed mCherry-MARK2 as quantified in Fig. 5B, whereas the kinase-dead (KD) form of PKC showed no effect on p595 MARK2 (lane 4). When expressed alone, mCherry-MARK2 is partially localized at the plasma membrane, evident by the ruffled appearance of the cell edges (panel i, Fig. 5C). Both YFP-tagged PKC and PKM constructs revealed the same ability to induce translocation of MARK2 from the plasma membrane to the cytosol (panels ii and iv, Fig. 5C, smooth appearance of cell edges), a feature that was absent when co-expressed with KD versions of PKC and PKM (panels iii and v, Fig. 5C). Scaffold Proteins Differentially Regulate the Phosphorylation and Localization of the aPKC Substrate MARK2-Using the MARK2 co-expression system validated for PKC catalytic activity and MARK2 cellular localization from Fig. 5, we set out to examine the effects of PKC-regulating scaffolds on MARK2 phosphorylation and localization. In addition to the PB1 domains of p62 and Par6 that directly interact with PKC (44,45), we also examined the effects on MARK2 localization of IRS-1, a large scaffold protein known to interact with PKC through its agonist-induced interaction with p62 (26,27). Intriguingly, co-expression of IRS-1 with MARK2 resulted in a decrease of p595 MARK2 levels (Fig. 6A, compare lanes 1 and  3), contrasting with the increased p595 levels produced with PKC co-expression (Fig. 6A, compare lanes 1 and 2). Additionally, co-expression of the PB1 domain of p62 had no effect on p595 MARK2 (Fig. 6A, lane 4), although the PB1 domain of Par6 increased p595 levels (lane 5), similar to PKC effects (quantification in Fig. 6A). We also examined the concordant effects on MARK2 cellular localization for each of the scaffolds used in Fig. 6A. In agreement with the phosphorylation results, co-expression of PKC resulted in localization of MARK2 into the cytosol (Fig. 6B, panel ii). Expression of the PB1 domain of p62 did not change MARK2 localization (Fig. 6B, panel iv), whereas co-expression of IRS-1 increased the localization of MARK2 at the plasma membrane (Fig. 6B, panel iii). Fully localized plasma membrane staining is evident by the spindle-like appearance of MARK2 in these cells along with the absence of visible nuclei, as decreases in nuclear visibility correlate with increases in plasma membrane localization (47). Curiously, coexpression of the Par6 PB1 domain did not change MARK2 localization relative to the plasma membrane (Fig. 6B, panel v) even though elevated p595 levels were observed. Similar results The p595/total MARK2 ratios were quantified from five (PB1 Par6 ), six (IRS-1), or seven (Vec, PKC, and PB1 p62 ) independent experiments, normalized to Vec-transfected control, and plotted as means Ϯ S.E. Statistical analysis was performed using separate t tests between each co-expression (PKC, IRS-1, PB1 p62 , and PB1 Par6 ) and Vec control. Significance notated as ***, p Ͻ 0.001; **, p Ͻ 0.01; or n.s., not significant. B, images of live COS-7 expressing mCherry-MARK2 alone or co-expressed with YFP-tagged PKC, CFP-IRS-1, CFP-PB1 p62 , or CFP-PB1 Par6 .
were also seen when conducted in polarized epithelial MDCK cells (data not shown). Insulin Regulates the Localization of Scaffolded PKC-Recent studies have demonstrated through co-IP experiments that insulin promotes the binding of p62 to IRS-1 via an interaction between the PB1 domain of p62 and an insulin-induced YXXM phosphorylation motif on IRS-1 (26). Additionally, insulin stimulates increased binding of PKC to p62 (27) and subsequent PKC engagement to IRS-1 through p62, also shown through co-IP (48). To investigate these interactions in real time translocation assays, we examined combinations of various YFP, CFP, and mCherry-tagged versions of p62, PKC, and IRS-1 in CHO-IR cells serum-starved overnight. The YFP tag was placed on the translocating protein (p62 or PKC); the CFP tag was placed on the destination protein (IRS-1 or p62), and the mCherry tag was placed on the third protein to confirm expression (diagram shown in Fig. 7A). Analysis of FRET as a measure of translocation (described previously (6)) revealed that insulin stimulation caused translocation of both YFP-p62 and YFP-PKC to CFP-IRS-1, in addition to translocation of YFP-PKC to CFP-p62 (Fig. 7B). We note that the increase in FRET upon insulin stimulation was only observed in a sub-fraction of the CHO-IR cells imaged in each experiment; an increase was observed for 26% (n ϭ 94 cells), 40% (n ϭ 50 cells), and 52% (n ϭ 31 cells) of cells imaged for the FRET from PKC to IRS-1, PKC to p62, and p62-IRS-1, respectively.

Discussion
The lack of regulation by either phosphorylation or second messengers (3,49,50)), including the Akt activator, phosphatidylinositol 3,4,5-triphosphate (5,43,51)), has confounded the understanding of how aPKCs signal in cells. Here, we show that protein scaffolds relocalize aPKC in response to insulin and that the binding to specific scaffolds differentially tunes the activity of aPKCs. Using FRET-based technologies to study the real time dynamics of aPKC activity and location in live cells, we show that p62 and Par6 tether PKC in a partially active or fully active conformation, respectively, to enhance signaling on these PB1 domain scaffolds. Translocation experiments reveal that insulin enhances the interaction of aPKC to p62, in turn recruiting the complex to IRS-1. Additionally, we show that IRS-1 can sequester aPKC away from phosphorylating the membrane-localized substrate MARK2. As part of this study, we validate the use of a suitable, newly available aPKC-specific active site inhibitor, PZ09, for study of aPKC regulation in live cells with PKC-specific substrates, noting that its inhibition of other non-PKC kinases such as PDK1 may cloud its utility for studying aPKC-specific function in vivo. These findings underscore the importance of scaffold interactions in controlling the cellular function of aPKCs and are particularly relevant as the exceptionally low catalytic activity (5 mol of phosphate/min/ mol of PKC (5) compared with 200 mol of phosphate/min/mol of PKC for cPKCs (52)) would require localization of substrates in the vicinity of aPKC for effective phosphorylation. The PB1 domain mediates multiple protein interactions primarily through electrostatic interactions between acidic and basic residues of heterodimerizing PB1 domains (53). p62 and Par6 are the two most characterized PB1 binding partners of aPKC and are thought to regulate separate downstream pathways of aPKC signaling (42). Here, we demonstrate that the basal activity of PKC is differentially regulated by these two scaffolds, with maximal signaling when scaffolded to Par6␣ compared with p62, which tethers a conformation that relieves significant autoinhibition. In addition, we show that the PB1 domain of Par6␣ binds more tightly to PKC than the PB1 domain of p62, indicating that the increased population of PKC localized to CKAR-PB1 Par6 contributes to the considerably high basal activity on this scaffold. However, the attenuated activity response of PKC on bound CKAR-PB1 p62 compared with unbound CKAR may also reflect greater PKC resistance to active site inhibitors when scaffolded to p62 but not to Par6, a feature known to impede the efficacy of other active site inhibitors on scaffolded PKC (54).
Investigation of the enhanced binding of PKC to Par6 compared with p62 led us to identify another key regulatory residue on the PB1 domain; Ser versus Ala at the position corresponding to Ser-24 and Ala-30 on p62 versus Par6, respectively, tunes the affinity for the aPKC PB1 domain. This residue is within a consensus region present only in the PB1 domains of p62 and the Par6 isoforms and not other human PB1 domain proteins (53). Previous work by Christian et al. (46) demonstrated that Ser-24 on p62 is phosphorylated by PKA, a modification that disrupts the interaction of p62 with aPKC. Consistent with phosphorylation at this site controlling protein interactions, we show that introduction of a negative charge at the comparable position on Par6␣ (A30D mutation) disrupts the binding of PKC to this PB1 domain and significantly reduces the basal activity of PKC on CKAR-PB1 Par6 . The Ala residue is only present on the Par6␣ isoform and is Ser at the corresponding site on Par6␤ and Par6␥. Although previous work has shown that aPKC binds to all three Par6 isoforms, yeast two-hybrid screens demonstrated the most significant interaction of aPKC to Par6␣ (also known as Par6C) (45), in agreement with the Ala favoring aPKC binding. Curiously, mutation of Ser-24 to either Asp or Ala on the p62 PB1 domain did not affect PKC activity. One possibility is that Asp is a poor phosphomimetic for this site on p62. However, mutation to Asp has been shown to inhibit aPKC binding (46); thus, any drop in activity may be below the detection limit of our cellular activity assays. Importantly, extended forskolin stimulation to promote PKA phosphorylation at this site decreased the binding of PKC to PB1 p62. Taken together, these data support a model wherein the unique Ser at position 24 on p62 provides a consensus phosphorylation site that regulates the binding and activity of aPKC.
The deletion of the PB1 domain (⌬PB1), the pseudosubstrate (⌬PS), or both (PKM) allows dissection of how scaffold interactions relieve the autoinhibition of aPKC to tune its the level of localized basal activity. For the non-scaffolded CKAR, PKM demonstrates the highest basal activity of all the constructs as it is neither autoinhibited (no PS) nor sequestered on an endogenous scaffold (no PB1), and therefore unrestricted to phosphorylate a globally expressed substrate such as CKAR. Supporting the model of PKM's unrestricted access to CKAR, co-IP results reveal that PKM interacts with CKAR whereas PKC does not, suggesting sequestration of the full-length kinase on a scaffold that restricts its interaction with CKAR. Similarly, analysis with an antibody targeted to specific PKC phosphorylation sites reveals that PKM exerts highly enhanced phosphorylation of global endogenous PKC substrates compared with PKC. However, PKC still has the ability to phosphorylate interactive specific substrates such as MARK2 (30) with equal capacity as PKM and to regulate its localization from the plasma membrane into the cytosol. Deletion of either the PS or the PB1 domain did not significantly affect the basal activity of PKC on non-scaffolded CKAR. In contrast, activity at the protein scaffolds was sensitive to deletion of the PB1 domain and, depending on the scaffold, the pseudosubstrate. On CKAR-PB1 p62 , deletion of the pseudosubstrate significantly enhanced basal activity, revealing some autoinhibition by the pseudosubstrate when PKC is bound to p62. This is consistent with our previous study showing that full-length PKC displays ϳ25% of its maximal unrestrained activity on p62 as assessed with a p62scaffolded CKAR (6). Deletion of the pseudosubstrate had no significant effect on the activity on the Par6 scaffold, revealing that the majority of the PKC bound to Par6 is in the open conformation, as reported previously (7), with the pseudosubstrate tethered away from the substrate-binding cavity. Activity on both scaffolds was sensitive to deletion of the PB1 domain, consistent with release of the enzymes from the scaffolds. These results reveal that interaction with regulatory determinants autoinhibit PKC and that binding to protein scaffolds differentially relieves these inhibitory constraints.
Further supporting autoinhibitory constraints of PKC, we show that PKM is more sensitive to dephosphorylation at its activation loop phosphorylation site, Thr-410 (5, 23), compared with PKC (5, 23). Thus, similar to conventional PKCs (55), intramolecular autoinhibitory interactions with regulatory domains mask this phosphorylation site not only in vitro (5) but also in cells. Additionally, binding to scaffolds may sequester aPKCs from phosphatases, although localization of phosphatases to protein scaffolds also serves as a mechanism to control signaling output of kinases (56).
Finally, we show that scaffold interactions can regulate the sequestration or localization of PKC to restrict or enhance its phosphorylation of physiological substrates. This phenomenon is analogous to the well known differential targeting of the protein phosphatase 1 catalytic subunit (PP1c) by its diverse arsenal of regulatory subunits to a variety of subcellular locations, including glycogen, spliceosome, and cytoskeleton where activity toward specific localized substrates is enhanced (57)(58)(59)(60)(61)(62)(63). In our studies, overexpression of IRS-1 inhibited the aPKC-dependent phosphorylation of MARK2, thus preventing the release of this substrate from the plasma membrane. In contrast, overexpression of the PB1 domain of Par6 enhanced the phosphorylation on the aPKC-regulated site of MARK2, suggesting that this scaffold may localize aPKC near MARK2. Note that this Par6-induced MARK2 phosphorylation did not displace the substrate from the plasma membrane, likely because shuttling phosphorylated MARK2 away from the apical surface of polarized cells depends on a second step, binding to the polarity regulator Par5 (also known as 14-3-3) (31,64). Overexpression of Par6 may indeed localize more endogenous aPKC to phosphorylate MARK2 yet interfere with the Par5 ability to bind MARK2.
How insulin controls aPKC function has been difficult to reconcile with the inability of this agonist to alter the phosphorylation state or activity of aPKC (5). Live cell imaging studies reveal that insulin promotes the association of PKC to both p62 and PKC, with a half-time on the order of 5 min under the conditions of our assays. This time frame agrees with aPKC-dependent changes in functions such as glucose transport, observed within 5-15 min of insulin stimulation (17,18,21). The insulin-dependent association of aPKC with these scaffolds supports previous co-IP results (26,27). IRS-1 is a large docking hub for downstream insulin signaling events to occur (25) and possesses Ser/Thr phosphorylation sites previously shown to be regulated by aPKC (65)(66)(67)(68). Taken together, these results suggest that the major effect of insulin on controlling aPKC function is by re-localizing the kinase to cellular signaling hubs such as IRS-1, poising it next to relevant downstream substrates.
In summary, our data support a model in which aPKC is regulated by binding to specific protein scaffolds that differentially control its activity. When unscaffolded, aPKC has low basal activity because of efficient autoinhibition by its regulatory domains, such as the pseudosubstrate (Fig. 7C, green triangle) masking the kinase domain (cyan circle). Binding to Par6 tethers the kinase in an open conformation, with the pseudosubstrate removed from the substrate-binding cavity in the kinase domain to allow maximal activity. Lower affinity binding to p62 results in less effective tethering of the pseudosubstrate, so that the scaffolded enzyme has ϳ25% maximal activity (6). This sequestration on protein scaffolds can either enhance phosphorylation of co-localized substrates or suppress phosphorylation of other substrates (e.g. the inhibition of MARK2 phosphorylation upon sequestration of aPKC by IRS-1 overexpression). The particularly low catalytic activity of aPKCs ensures that signaling is kept at a minimum in the absence of regulated scaffold interactions, which poise the enzyme for stoichiometric phosphorylation of substrates recruited to the same signaling platform.
Author Contributions-I. S. T. and A. C. N. conceived the experiments and wrote the manuscript. I. S. T. performed and analyzed the experiments. All authors reviewed the results and approved the final version of the manuscript.