C-spine mutations of protein kinase C and Akt as a novel generalizable approach to create stable pseudokinases

Abstract

Protein kinases function not only through their catalytic phospho-transfer activity but also by noncatalytic scaffold mechanisms. Introduction of mutations to inactivate catalysis provides a tool to differentiate between the two; however, kinase-inactivating mutations may alter the structure of the kinase domain and perturb scaffold functions. Here, we developed a strategy that prevents ATP binding, thereby preventing catalysis, while stabilizing the active conformation of the kinase domain. This approach leverages the structural role of ATP in assembling the catalytic spine (C-spine), a hydrophobic core essential for the active conformation. Specifically, we substituted Val or Ala residues proximal to the binding position of the adenosine ring of ATP with Phe in three protein kinase C isozymes (PKCβII, γ, and θ) and Akt1. Structural modeling suggests that Phe substitutions at these positions are a surrogate for the adenosine ring of ATP to assemble the C-spine. Live-cell imaging using genetically encoded PKC and Akt activity reporters reveals that C-spine mutations abolish kinase activity. Furthermore, phosphorylation of the hydrophobic motif, an autophosphorylation site, is abolished in C-spine mutants of PKC family members and reduced in C-spine mutants of Akt1, independent of epidermal growth factor stimulation. In PKCβII, these C-spine mutations accelerate plasma membrane translocation, consistent with impaired autoinhibition due to the lack of hydrophobic motif phosphorylation. Despite adopting reduced autoinhibition, turnover experiments with PKCθ reveal C-spine mutants do not impair the stability of the full-length PKC. The generation of pseudokinases by C-spine mutations provides a generalizable strategy for elucidating noncatalytic kinase functions.

Keywords: Akt; C-spine; kinase-dead; protein kinase C; pseudokinase.

Figure 1

Hydrophobic spine assembly by conserved residues in the kinase domain positions the enzyme for catalysis.A–D, kinase domain of PKCβ (PDB: 3PFQ; rat) A, PKCθ (PDB: 5F9E; human) B, Akt1 (PDB: 4EKK; human) C, and PKA (PDB: 1ATP; mouse) D, showing assembly of the catalytic (C) spine (red) and regulatory (R) spine (yellow) when bound to ATP, an ATP analog, or an ATP-competitive inhibitor. E–H, ATP-binding pocket of PKCβ E, PKCθ F, Akt1 G, and PKA H, highlighting residues involved in ATP coordination. The adenine ring of ATP is bound by C-spine residues valine and alanine in the N-lobe, and methionine or leucine in the C-lobe (V57, A70, and L173 in PKA). I, sequence alignment of PKC isozymes, Akt isozymes, and PKA showing conserved residues in the kinase domain. Kinases highlighted in red were analyzed in this study. C-spine residues, R-spine leucine (RS3), the invariant lysine, and the β3-strand leucine are indicated and analyzed in this study. Akt, protein kinase B; PDB, Protein Data Bank; PKA, protein kinase A; PKCβ, protein kinase C β; PKCθ, protein kinase C θ.

Figure 2

C-spine, not R-spine, phenylalanine substitution in PKCβII renders the kinase inactive and unphosphorylated.A, left, PKC activity in COS7 cells expressing CKAR2 alone (black trace) or coexpressing CKAR2 and the indicated mCherry-tagged PKCβII constructs. At 3 min, cells were treated with 100 μM UTP followed by treatment with 200 nM PDBu at 9.7 min. Changes in FRET/CFP ratios were normalized to the first 3 min and plotted. Data are representative of 24 to 38 cells per condition from three independent experiments (mean ± SEM). A nonphosphorylatable reporter (CKAR2 TA; gray trace) was used as a control for any changes independent of PKC activity. Right, change in PKC activity measured as ΔFRET/CFP ratios between FRET/CFP ratios at 3 min and 15 min. Data were plotted and represent mean ± SEM. ns = nonsignificant, ∗∗∗∗p < 0.0001 by one-way ANOVA. B, PKC translocation in COS7 cells coexpressing MyrPalm-CFP and the indicated YFP-tagged PKCβII constructs treated with 200 nM PDBu at the indicated time. Changes in FRET/CFP ratios were normalized to the first 3 min and plotted as a percentage of max translocation. Data are representative of 21 to 43 cells per condition from three to five independent experiments (mean ± SEM). C, representative western blot of whole cell lysate from untransfected COS7 cells or COS7 cells expressing the indicated YFP-tagged PKCβII. D–F, quantification of PKCβII phosphorylation from panel C at the activation loop D, turn motif E, and hydrophobic motif F, normalized first to loading control then to total PKCβII protein. Data were plotted relative to the phosphorylation of PKCβII WT and represent mean ± SEM from three independent experiments. ns = not significant, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 by one-way ANOVA. CFP, cyan fluorescent protein; CKAR2, C kinase activity reporter 2; PDBu, phorbol 12, 13-dibutyrate; PKCβII, protein kinase C βII; YFP, yellow fluorescent protein.

Figure 3

PKCγ C-spine phenylalanine substitution, V365F, is also inactive and unphosphorylated.A, PKC activity in COS7 cells expressing CKAR2 alone (black trace) or coexpressing CKAR2 and mCherry-PKCγ WT or V365F. At 3 min, cells were treated with 200 nM PDBu. Changes in FRET/CFP ratios were normalized to the first 3 min and plotted. Data are representative of 15 to 20 cells per condition from three independent experiments (mean ± SEM). B, representative western blot of whole cell lysate from COS7 cells expressing YFP-empty vector or the indicated YFP-tagged PKCγ constructs. Cells were treated with 50 nM CalA or DMSO control for 10 min. C–E, quantification of PKCγ phosphorylation from panel B at the activation loop C, turn motif D, and hydrophobic motif E, normalized to total PKCγ protein and loading control. Data were normalized to the phosphorylation of DMSO-treated PKCγ WT and represent mean ± SEM from four independent experiments. ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 by two-way ANOVA and Tukey post hoc test. Relevant statistical comparisons are shown. CalA, calyculin A; CKAR2, C kinase activity reporter 2; DMSO, dimethyl sulfoxide; PDBu, phorbol 12, 13-dibutyrate; PKCγ, protein kinase C γ; YFP, yellow fluorescent protein.

Figure 4

PKCθ C-spine mutants are inactive and unphosphorylated but remain stable.A, PKC activity in COS7 cells expressing CKAR2 alone (black trace) or co-expressing CKAR2 and mCherry-PKCθ WT, V394F, or A407F. At 3 min, cells were treated with 200 nM PDBu followed by 1 μM of the PKCθ-specific inhibitor, C20, at 10 min. Changes in FRET/CFP ratios were normalized to the first 3 min and plotted. Data are representative of 33 to 50 cells per condition from three independent experiments (mean ± SEM). B, representative western blot of whole cell lysate from COS7 cells transfected with YFP-empty vector (EV) or the indicated YFP-tagged PKCθ constructs and treated with 200 nM PDBu or DMSO control for 15 min. PKCθ activity was assessed by probing for phospho-Ser PKC substrates. C, phospho-Ser PKC substrate signal from panel B was quantified by densitometry of the full lane excluding the 100 kDa band corresponding to PKCθ. Data were normalized first to loading control then to total PKCθ protein and plotted relative to untreated EV. Data represent mean ± SEM from three independent experiments. ns = not significant, ∗p < 0.05 by two-way ANOVA and Tukey post hoc test. Relevant statistical comparisons are shown. D, representative western blot of whole cell lysate from COS7 cells expressing YFP-EV or the indicated YFP-tagged PKCθ constructs. Corresponding phosphorylation site alanine mutants were included as a control for antibody specificity. Cells were treated with 50 nM CalA or DMSO control for 10 min. E–G, quantification of PKCθ phosphorylation from panel D at the activation loop E, turn motif F, and hydrophobic motif G, normalized first to loading control then to total PKCθ protein. Data were plotted relative to the phosphorylation of DMSO-treated PKCθ WT and represent mean ± SEM from three independent experiments. ∗∗∗∗p < 0.0001 by two-way ANOVA and Tukey post hoc test. Relevant statistical comparisons are shown. H, representative western blot of whole cell lysate from COS7 cells expressing the indicated YFP-tagged PKCθ constructs or HA-tagged PKCθ R145H; note the nature of the tag does not alter stability as HA-tagged PKCθ WT has a half-life of 37 ± 5 min (63) comparable to that of YFP-tagged PKCθ WT reported here (39 ± 4 min). Cells were treated with 355 μM CHX and lysed at the indicated times. I, quantification of PKCθ protein levels from panel H normalized to loading control and plotted relative to the 0 h time point. Data were fit to a first-order decay and depict mean ± SEM from three independent experiments. C20, compound 20; CalA, calyculin A; CHX, cycloheximide; CKAR2, C kinase activity reporter 2; DMSO, dimethyl sulfoxide; HA, hemagglutinin; PDBu, phorbol 12, 13-dibutyrate; PKCθ, protein kinase C θ; YFP, yellow fluorescent protein.

Figure 5

Akt1 C-spine mutants are inactive with reduced phosphorylation at the hydrophobic motif.A–B, Akt activity in COS7 cells expressing BKAR and empty vector (black trace) or coexpressing BKAR and mCherry-Akt1 WT, V164F, or A177F. Cells grown in 10% serum A or serum starved for 18 h before imaging B. At 3 min, cells were treated with A, 20 μM of the Akt inhibitor, GDC0068 alone, or B, 50 ng/ml EGF followed by 20 μM of GDC0068 at 10 min. Changes in FRET/CFP ratios were normalized to the first 3 min A and plotted or absolute changes in FRET/CFP ratios were plotted B. Data are representative of 36 to 46 cells per condition from three to four independent experiments (mean ± SEM). C, representative western blot of whole cell lysate from COS7 cells transfected with the indicated HA-tagged Akt1 constructs. Cells were serum starved for 15 h before treating with DMSO or 50 ng/ml EGF for 8 min. D–F, quantification of Akt1 at the activation loop D, turn motif E, and hydrophobic motif F, from panel C normalized to loading control and total Akt1 protein. Data were normalized to the phosphorylation of DMSO-treated Akt1 WT and represent mean ± SEM from six independent experiments. ns = nonsignificant, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 by two-way ANOVA and Tukey post hoc test. Relevant statistical comparisons are shown. Akt, protein kinase B; BKAR, B kinase activity reporter; CFP, cyan fluorescent protein; DMSO, dimethyl sulfoxide; EGF, epidermal growth factor; HA, hemagglutinin.

Figure 6

Phenylalanine mutation at C-spine residues mimics ATP binding.A–B, ATP-binding pocket of PKCθ with A407 (red) A, or V394 (green) B, substituted with phenylalanine, illustrating how the phenyl group is a surrogate for the adenine ring of ATP to assemble the C-spine. PKCθ, protein kinase C θ.

Figure 7

C-spine mutations inactivate both PKC and Akt but with different outcomes on their phosphorylation states.Top panel: in wild-type PKC and Akt, the adenosine ring of ATP (red) assembles the C-spine (pale red), permitting processing phosphorylations. For PKC, these are the mTORC2-catalyzed phosphorylation at the turn motif (orange circle), the PDK1-catalyzed phosphorylation of the activation loop (magenta circle), and the autophosphorylation of the hydrophobic motif (green); phosphorylation of the hydrophobic motif is necessary for PKC to adopt the autoinhibited conformation with the pseudosubstrate (dark gray) in the substrate-binding cavity. Acute activity is controlled by second messenger binding to relevant domains. Akt undergoes the same phosphorylations but differs in that the phosphorylation by PDK1 at the activation loop (magenta circle) and subsequent autophosphorylation at the hydrophobic motif (green circle) are agonist-evoked; this kinase does not have a pseudosubstrate and acute activity is controlled by phosphorylation. Bottom panel: for both kinases, phenylalanine (green) C-spine mutations are a surrogate for ATP to structure a correctly assembled kinase domain. However, they are catalytically inactive toward substrates and toward hydrophobic motif autophosphorylation. In the case of PKC (left), spine mutants have a properly folded kinase domain but are unable to autoinhibit because the hydrophobic motif is not phosphorylated. Without autoinhibition, the activation loop and turn motif sites are phosphatase labile and do not accumulate. This unphosphorylated PKC dimerizes and is dominant negative toward endogenous PKC. In the case of Akt (right), spine mutants retain phosphate at the turn motif and activation loop, because hydrophobic motif phosphorylation is not necessary to retain phosphates at these positions; they are not dominant negative. Akt, protein kinase B; PKC, protein kinase C.