Cotranslational cis-phosphorylation of the COOH-terminal tail is a key priming step in the maturation of cAMP-dependent protein kinase

Abstract

cAMP-dependent protein kinase A (PKA), ubiquitously expressed in mammalian cells, regulates a plethora of cellular processes through its ability to phosphorylate many protein substrates, including transcription factors, ion channels, apoptotic proteins, transporters, and metabolic enzymes. The PKA catalytic subunit has two phosphorylation sites, a well-studied site in the activation loop (Thr¹⁹⁷) and another site in the C-terminal tail (Ser³³⁸) for which the role of phosphorylation is unknown. We show here, using in vitro studies and experiments with S49 lymphoma cells, that cis-autophosphorylation of Ser³³⁸ occurs cotranslationally, when PKA is associated with ribosomes and precedes posttranslational phosphorylation of the activation loop Thr¹⁹⁷. Ser³³⁸ phoshorylation is not required for PKA activity or formation of the holoenzyme complex; however, it is critical for processing and maturation of PKA, and it is a prerequisite for phosphorylation of Thr¹⁹⁷. After Thr¹⁹⁷ and Ser³³⁸ are phosphorylated, both sites are remarkably resistant to phosphatases. Phosphatase resistance of the activation loop, a unique feature of both PKA and PKG, reflects the distinct way that signal transduction dynamics are controlled by cyclic nucleotide-dependent PKs.

Eukaryotic PKs (EPKs) regulate many cellular processes through their ability to phosphorylate and activate themselves and substrate proteins at serine, threonine, or tyrosine residues. Structurally, all EPKs consist of a conserved bilobal kinase domain, which was initially described for the catalytic (C) subunit of cAMP-dependent protein kinase A (PKA) (1). A small N-terminal lobe (N-lobe) and a larger carboxyl-terminal helical lobe (C-lobe) create a cleft between the two lobes where the ATP substrate binds. All EPKs have an activation loop in the C-lobe that typically harbors a conserved activation loop phosphorylation site, which is a major regulatory site (2–4). Other kinase-specific phosphorylation sites may be auto- or trans-phosphorylated by heterologous kinases. AGC kinases, a subfamily of Ser/Thr kinases, require multiple phosphorylations for complete activation (1).

The AGC subfamily includes PKA, cGMP-dependent protein kinase (PKG), PKC, PKB/Akt, p90 ribosomal protein S6 kinase 1, and p70 ribosomal protein S6 kinase 1 (S6K1). In addition to the activation loop site, the AGC kinases usually possess two other phosphorylation sites in their carboxyl-terminal tail (C-tail), a turn motif and a hydrophobic motif (HM) site (Fig. S1) (2). The C-tail is anchored to both the N- and C-lobes, and it is an essential, conserved feature of all AGC kinases, (Fig. 1A) (3). The turn motif and HM sites wrap around the N-lobe and position it for catalysis (3, 4). Most AGC kinases also have one or more N- or C-terminal domains associated with the kinase domain that regulate kinase activity and/or subcellular localization (1).

Fig. 1.

Structural representation of the COOH-terminal tail and characterization of the phosphorylation mutants of the C-subunit of PKA. (A) Alignment of the C-terminal tail of major AGC kinase (described in the color code) around the kinase core. (B) Overlay of the C-terminal tail of PKC_θ (black) and C-subunit (red) showing that PKC has two phosphorylation sites (circles) in the C-terminal tail, whereas the PKA C-subunit has only one. (C) Close-up view of the C-terminal tail of PKC_θ (modified from Protein Data Bank ID code 2JED) around the N-lobe showing the turn motif phosphate interacting with a conserved basic residue from the β2-strand in the glycine-rich loop. (D) Close-up view of the C-terminal tail of the C-subunit (modified from Protein Data Bank ID code 1FMO) showing the atypical turn motif phosphorylation making multiple interactions to stabilize the N-lobe. These interactions are distinct from the typical turn motif site, which in the C-subunit, is replaced by a glutamate interacting with arginine. (E) SDS/PAGE gel showing purification of the C-subunit and other mutants of the C-subunit, which was shown by Coumassie staining. (F) Immunoblot analysis of phosphorylation status of WT C-subunit and various C-subunit mutants. (G) Enzyme activity using in vitro kinase assay described in ref. 6 of WT C-subunit compared with phosphorylation mutants of the C-subunit. The final enzyme concentration was 20 nM, ATP was 200 μM, and the Kemptide substrate concentration was 1 mM.

PKA has several features that make it a unique member of the AGC family. The C-subunits of PKA do not have large regulatory domains associated with the kinase domain, but instead, they are assembled as an inactive holenzyme that contains a regulatory subunit dimer and two C-subunits. The dimeric R-subunits have an inhibitor site that occludes the active site of the C-subunit in the absence of cAMP and keeps it in a stable, inactive complex (5). At their C terminus, the R-subunits have two cyclic nucleotide binding domains. The N-terminal dimerization domain also serves as a docking site for scaffold proteins called A kinase-anchoring proteins (AKAPs), which localize the PKA holoenzyme in close proximity to dedicated substrates. Binding of cAMP to the R-subunits unleashes the C-subunits from the holoenzyme and induces R–C dissociation. The dissociated C-subunit has a phosphotransfer burst rate of 250–500 s⁻¹ and a steady state rate of 20 s⁻¹ (6, 7), which are faster than most other PKs. The C-subunit of PKA phosphorylates ≥1,000 proteins, including ion channels (8), transcription factors (e.g., cAMP response element-binding and NF-κB) (9, 10), metabolic enzymes, and apoptotic proteins, and thus, regulates a plethora of cellular processes (11). In addition, up to 50 AKAPs localize PKA to specific subcellular sites and bring it in close proximity to substrates (12). The mammalian C-subunit also has a unique N-terminal helix with a myristylation site that is absent in other AGC kinases.

Compared with other AGC kinases, the C-subunit of PKA is also unusual in terms of its phosphorylation sites. It has a conserved activation loop site (Thr¹⁹⁷) in which phosphorylation stabilizes and activates the protein (13). Instead of a conventional turn motif or HM sites (Fig. 1B), the C-subunit has a unique C-terminal tail phosphorylation site, Ser³³⁸, that lies between the turn motif and the HM site (2, 14) and can be thought of as an atypical turn motif. In the C-subunit, Glu³³³ replaces the conventional turn motif phosphate and positions the N-lobe for catalysis (Fig. 1 B–D) (4). Unlike most PKs, which are transiently activated by phosphorylation of the activation loop phosphate, the C-subunit of PKA is assembled as a stable, fully phosphorylated holoenzyme that is typically anchored to a scaffold and has activity dependent on the second messenger, cAMP.

When overexpressed in Escherichia coli, the C-subunit is constitutively autophosphorylated at the activation loop (Thr¹⁹⁷) and the C-terminal (Ser³³⁸) sites. The activation loop phosphate, which has been studied in many AGC kinases including PKA, has a global effect on structure and function (13, 15). Recent results for PKA show that removing the activation loop phosphate abolishes the rapid phosphotransfer burst kinetics without affecting the steady state kcat (16). In mammalian cells, the C-subunit is phosphorylated in trans at Thr¹⁹⁷ by either another C-subunit or phosphoinositide-dependent PK 1 (PDK1), a master kinase of the AGC subfamily (17, 18). In Schizosaccharomyces pombe, the activation loop site is phosphorylated only by PDK1 (19). In contrast, the role of Ser³³⁸ in regulating C-subunit activity is not well-understood. Replacement of Ser³³⁸ with Ala generates a very unstable protein (14); this C^S338A mutant is the most destabilizing mutation in the C-tail (20). Although Ser³³⁸ is proposed to be a cis-autophosphorylation site (21), the mechanism, timing, and physiological role of Ser³³⁸ phosphorylation are not known. Thus, it is critical to understand how Ser³³⁸ is phosphorylated in cells, the relationship between this phosphorylation and the phosphorylation of Thr¹⁹⁷ in the maturation and activation of the C-subunit for its downstream signaling, and whether and why these two phosphates resist hydrolysis after the fully phosphorylated enzyme is assembled. The current studies used in vitro approaches in parallel with studies in cultured cells (particularly using WT S49 cells and mutants that lack PKA activity) to address these questions and define key events involved in the maturation and activation of PKA. The results help to explain why PKA is unique among the AGC kinases and most other PKs.

Results

Abolishing Autophosphorylation Sites in PKA by Removing Key Elements of the Consensus Sequence.

RRXS/T is a minimal consensus motif in the PKA C-subunit for substrate recognition (22). The conserved Arg at the P-2 or P-3 position is critical for autophosphorylation. By mutating R¹⁹⁴ to Ala instead of the more common strategy of mutating the phosphorylation site itself to Ala, we generated a mutant, C^R194A, that is not phosphorylated at Thr¹⁹⁷, a result consistent with previous findings (13, 23). Similarly, the C^R336A mutant cannot be autophosphorylated at Ser³³⁸. Mutation of the catalytic lysine, C^K72H, generates a protein that is not phosphorylated at either Thr¹⁹⁷ or Ser³³⁸ (21). We expressed all three mutants in E. coli, purified them to homogeneity along with WT C-subunit, and then assessed their phosphorylation state and enzymatic activity (Fig. 1 E–G). We found that C^R194A and C^R336A have ∼50% activity compared with WT C-subunit and that C^K72H is not active (Fig. 1G). These results show that, although the C-subunit has activity in the absence of phosphorylation of Thr¹⁹⁷ and Ser³³⁸, phosphorylation of both sites is required for optimal activity. Moreover, if C^K72H is phosphorylated by PDK1 in vitro, it still does not have enzymatic activity, thus underscoring the idea that phosphorylation of Thr¹⁹⁷ is not sufficient to compensate for the absence of K⁷².

Distinguishing cis- and trans-Phosphorylation Events by in Vitro Phosphorylation of C^R194A, C^R336A, C^K72H, and the GST–C-Tail (299–350) of the C-Subunit.

C^R194A can be phosphorylated by PDK1 on Thr¹⁹⁷ in its activation loop (Fig. 2A), which is a mechanism for the activation of PKA in mammalian cells (17, 24). By contrast, although the catalytically dead C^K72H binds ATP and can be trans-phosphorylated by C-subunit or PDK1, it is not autophosphorylated when incubated with Mg-ATP (Fig. 2B). In the absence of the P-2 Arg, the C^R336A mutant is not trans-phosphorylated at Ser³³⁸ by either the C-subunit or PDK1 or autophosphorylated at Ser³³⁸ (Fig. 2C). To prove further that the Ser³³⁸ site is not phosphorylated by a trans-mechanism, we fused the PKA C-subunit C-tail (residues 299–350) to GST to use as a substrate in two different assays. Fig. 2D shows that, if the GST-tagged C-tail is incubated with WT C-subunit, no phosphorylation occurs for at least 45 min. An in vitro kinase assay with C-subunit and radiolabeled [³²P]-ATP showed no incorporation of phosphate into the GST–C-tail of PKA, whereas the RIIβ-subunit is readily phosphorylated in the presence of cAMP (Fig. 2E). Collectively, these results indicate that Thr¹⁹⁷ is trans-phosphorylated by the C-subunit and PDK1, whereas neither Ser³³⁸ nor the isolated C-tail are trans-phosphorylated by the C-subunit or PDK1, results that are consistent with previous findings (21).

Fig. 2.

In vitro phosphorylation of C-subunit mutants of PKA. The phosphorylation state was assessed using immunoblot analysis with phosphospecific antibodies against the Thr¹⁹⁷ and Ser³³⁸ sites. (A) Time course of C^R194A transphosphorylation at Thr¹⁹⁷ by PDK1 vs. WT C-subunit. The final concentration of PDK1 was 20 nM, and C^R194A was 5 μM. (B) C^K72H can be transphosphorylated by PDK1 and WT C-subunit but not autophosphorylated at Thr¹⁹⁷. The final concentration of PDK1 and WT C-subunit was 20 nM, and C(L72H) was 5 μM; reaction time was 15 min. (C) Immunoblot analysis using phosphospecific antibody against Ser³³⁸ showing that C^R336A cannot be transphosphorylated by either WT C-subunit or PDK1 at Ser³³⁸. The final concentration of PDK1 and WT C-subunit was 20 nM, and C^R336A was 5 μM. (D) GST–C-tail (299–350) of the C-subunit cannot be trans-phosphorylated by WT C-subunit at Ser³³⁸. The final concentration of WT C-subunit was 20 nM; GST–C-tail was 5 μM. (E) No phosphoincorporation was seen in GST–C-tail at Ser³³⁸ in trans by WT C-subunit using standard in vitro kinase assay (6). RIIβ phosphorylation without cAMP, as a negative control, and with cAMP is used as a positive control.

Ser³³⁸ Phosphorylation Precedes Thr¹⁹⁷ Phosphorylation and Is Cotranslational.

To evaluate the kinetics of Ser³³⁸ and Thr¹⁹⁷ phosphorylation, we used a cell-free translational system reconstituted from E. coli (25). Phosphorylation of Ser³³⁸ in the C-subunit in an in vitro transcription/translation (IVTT) assay occurs in ∼5 min, whereas phosphorylation of Thr¹⁹⁷ is much slower (Fig. 3A, Upper). Quantification of the extent of Ser³³⁸ and Thr¹⁹⁷ phosphorylation with respect to C-subunit synthesis revealed that Thr¹⁹⁷phosphorylation, but not phosphorylation of Ser³³⁸, depends on C-subunit concentration (Fig. 3A, Lower), suggesting that Ser³³⁸ phosphorylation occurs in cis, whereas Thr¹⁹⁷ phosphorylation occurs by a trans mechanism.

Fig. 3.

Ser³³⁸ phosphorylation precedes Thr¹⁹⁷ of the C-subunit of PKA and occurs on ribosomes. (A) Time course of the IVTT reaction showing that Ser³³⁸ phosphorylation occurs before Thr¹⁹⁷ phosphorylation, which was assessed by immunoblot analysis using phosphospecific antibodies. Lower shows quantification of the extent of phosphorylation of Ser³³⁸ and Thr¹⁹⁷ with the overall time course of C-subunit synthesis using ImagJ. Extent of phosphorylation of Ser³³⁸ is independent of C-subunit synthesis, whereas phosphorylation of Thr¹⁹⁷ is dependent on C-subunit synthesis. This finding indicates that Ser³³⁸ is phosphorylated in cis, whereas Thr¹⁹⁷ is phosphorylated in trans. (B) Ser³³⁸ phosphorylation is present in both the ribosomal and flow-through fractions, whereas Thr¹⁹⁷ phosphorylation was detected only in flow-through fractions as assessed by immunoblot analysis using phosphospecific antibodies. Elongation factor 1 (EF1) antibody was used as a positive control for active translation. (C) Purified ribosomes from IVTT at various time points show the absence of lag between Ser³³⁸ phosphorylation and C-subunit translation as assessed by immunoblot analysis. Thr¹⁹⁷ phosphorylation is not detected in the ribosome fractions.

Because Thr¹⁹⁷ is trans-phosphorylated and posttranslational, whereas Ser³³⁸ phosphorylation precedes Thr¹⁹⁷ phosphorylation, we reasoned that Ser³³⁸ might occur on ribosomes during translation. We, thus, carried out the IVTT reaction for 2 h, quenched the reaction with dry ice, isolated ribosomes using a sucrose bed method (26), and used immunoblotting to assess the phosphorylation state of Ser³³⁸ and Thr¹⁹⁷. As shown in Fig. 3B, Ser³³⁸ phosphorylation was present in both the ribosomal pellet and flow-through fractions, whereas Thr¹⁹⁷ phosphorylation was found only in the flow-through fraction and the phosphorylation of Ser³³⁸ in the ribosomal fraction paralleled the expression of C-subunit (Fig. 3C). These results are consistent with Ser³³⁸ phosphorylation occurring cotranslationally by a cis mechanism as a one-off event, meaning that it happens one time during translation. The order of Ser³³⁸ and Thr¹⁹⁷ phosphorylation that we observe differs from the order described in a previous report (21), but in that report, H-89, an inhibitor of C-subunit, was used to generate unphosphorylated C-subunit (21); additionally, the work by Iyer et al. (21) monitored the kinetics of phosphorylation at Thr¹⁹⁷ by PDK1 and Ser³³⁸ by autophosphorylation, which contrasts with our approach that involved assessment of the kinetics of phosphorylation in parallel with translation of the C-subunit.

Ser³³⁸ Phosphorylation Is Required for Solubilization of the C-Subunit and Subsequent Phosphorylation of Thr¹⁹⁷.

Because Ser³³⁸ phosphorylation is unique to PKA in the AGC subfamily (Fig. S1), we sought to determine its role in the regulation of the C-subunit. We, thus, overexpressed WT C-subunit and C^S338A in HEK293 cells and found that a minimal amount of the C-subunit appears in the insoluble pellet, which is in contrast to what occurs with the C^S338A mutant, where most of the protein is insoluble. Moreover and in contrast to what we observe for the E. coli-expressed protein, the C^S338A mutant also shows substantial reduction of phosphorylation of the activation loop (Fig. 4A). In contrast, the HA-tagged C^R336A mutant expressed in HEK cells is soluble but lacks phosphorylation of Ser³³⁸ or Thr¹⁹⁷ (Fig. 4B), although Thr¹⁹⁷ is phosphorylated when the C^R336A mutant is expressed in E. coli (Fig. 1C). This result suggests that Ser³³⁸ phosphorylation is required for solubility and subsequent phosphorylation of Thr¹⁹⁷; in the absence of Ser³³⁸ phosphorylation, the C-subunit is prone to aggregation. These findings are consistent with results that have shown that C^S338A is the most unstable mutant of the C-tail residues in the C-subunit of PKA (20).

Fig. 4.

Ser³³⁸ phosphorylation is required for solubility of C-subunit. (A) Immunoblot analysis of GFP-tagged C-subunit and GFP-tagged C^S338A overexpressed in 293 cells. GFP-tagged C-subunit is mostly soluble, whereas C^S338A mutant is mostly insoluble (row 1). C^S338A has significant reduction in Thr¹⁹⁷ phosphorylation. (B) Immunoblot analysis of HA-tagged C-subunit and HA-tagged C^R336A overexpressed in 293 cells. Both WT C-subunit and C^R336A are soluble, but C^R336A is not phosphorylated on Ser³³⁸. (C) Helical propensity of C-tail residues 329–350. The WT sequence has helical propensity (trace A). Mutating Arg³³⁶ to Ala results in loss of helical propensity (trace B). Mutating Ser³³⁸ to Ala results in an approximately twofold increase of helical propensity (trace C), whereas mutating Ser³³⁸ to Glu results in a significant decrease in helical propensity (trace D). The dashed line represents the threshold below which there is no helicity. (D) Absence of Ser³³⁸ phosphorylation does not affect formation of RI_α or RII_α holoenzymes as shown by colocalization of C^R336A with either RI_α- (Upper) or RII_α-subunit (Lower). (E) Immunoblot analysis revealed that the C-subunit in kin− S49 cells is insoluble and not phosphorylated at Ser³³⁸, whereas WT S49 cells have a soluble and Ser³³⁸-phosphorylated C-subunit.

To explore whether there are unique structural features of the C-tail of the C-subunit, we analyzed the helical propensity of the C-subunit based on its primary sequence (27, 28). Although we found that the most prominent helix is at the N terminus, residues 328–350 in the C-tail sequence also have some helical propensity. In silico mutation of Ser³³⁸ to Ala increased the helical propensity ∼2.5-fold, whereas mutating Ser³³⁸ to Glu decreased helical propensity. Mutation of Arg³³⁶ to Ala resulted in loss of helical propensity (Fig. 4C). This analysis suggests that the C-tail region can exhibit multiple conformations and that either phosphorylation of Ser³³⁸ or removal of Arg³³⁶ can disrupt the propensity of this segment to assume a given conformation. Such results may help explain why the cotranslational cis-phosphorylation of Ser³³⁸ is the priming step for assembly of the active kinase, which is then followed by posttranslational trans-phosphorylation of Ser³³⁸. We do not yet have a structure of the C-subunit in a completely dephosphorylated state, but we predict that it is likely to be converted into a more active-like conformation and positioned around the top of the N-lobe, leaving the activation loop poised for autophosphorylation.

To investigate whether the absence of Ser³³⁸ phosphorylation affects formation of the tetrameric R₂C₂ PKA holoenzyme, we separately overexpressed CFP-RII_α with the C^R336A-mCherry construct and GFP-RI_α with the C^R336A mCherry in HEK293 cells in the background of dual specificity d-AKAP1, which localizes PKA to mitochondria (29). As shown in Fig. 4D, C^R336A colocalizes with both RI_α and RII_α, suggesting that Ser³³⁸ phosphorylation does not prevent binding to the R-subunit in cells. This result is akin to evidence that deletion of the N terminus of the C-subunit (Δ1–14) generates an insoluble protein in E. coli, but it is soluble if coexpressed with the RI_α-subunit (30), suggesting that the RI_α-subunit can serve as a potential chaperone for stabilization of the C-subunit. The crystal structure of the C:RI_α heterodimer suggests that the binding interface between C- and R-subunits of PKA primarily involves the C-lobe of the C-subunit. Mg₂ATP is required for formation of the RI_α holoenzyme; therefore, the C-subunit assumes a fully closed conformation, but the phosphate on Ser³³⁸ has no direct contact with the R-subunit (5).

C-Subunit Is Insoluble and Is Not Phosphorylated at Ser³³⁸ in S49 Kin− Lymphoma Cells.

To explore the physiological role of Ser³³⁸ phosphorylation and evaluate this role in an in vivo setting, we used kin− S49 mouse lymphoma cells. WT S49 cells grow in suspension culture and undergo mitochondria-dependent apoptosis in response to increases in intracellular cAMP (31). Kin− S49 cells, which were selected based on their resistance to cAMP-promoted apoptosis, have no detectable C-subunit or PKA activity in the soluble fractions, except a small amount of C-subunit in the insoluble fraction (32, 33). Based on our biochemical data with purified proteins and cell culture studies, we hypothesized that, if Ser³³⁸ phosphorylation is required for C-subunit solubility, kin− cells (in which the C-subunit is insoluble) would lack phosphorylation of the C-subunit at Ser³³⁸. We, thus, assessed WT and kin− S49 cells for the presence of C-subunit in soluble and insoluble fractions and Ser³³⁸ phosphorylation. We confirmed that C-subunit is in the soluble fraction in WT S49 cells (32) but insoluble in kin− S49 cells. Moreover, we found that the C-subunit is phosphorylated on Ser³³⁸ in WT S49 cells but that Ser³³⁸ phosphorylation as well as the previously reported Thr¹⁹⁷ phosphorylation site of the C-subunit are not detected in kin− cells (Fig. 4E). This finding is consistent with our finding that cotranslational cis-phosphorylation of Ser³³⁸ is a key priming step, which renders the C-subunit soluble and available for posttranslational trans-phosphorylation of its Thr¹⁹⁷ activation loop site and formation of a mature C-subunit.

Dephosphorylation of the C-Subunit and C-Subunit Mutants by λ-Protein Phosphatase.

After its synthesis, maturation, and phosphorylation on Ser³³⁸ and Thr¹⁹⁷, the C-subunit gains stability and resistance to removal of the activation loop phosphate by phosphatases (34). To investigate if this phosphatase resistance applies to both phosphates and requires both phosphates, we tested C-subunit mutants for their resistance to dephosphorylation by λ-protein phosphatase (λPP), a potent nonspecific phosphatase (35–37) (Fig. 5A). Although all of these mutants have kinetic defects in their ability to phosphorylate peptide substrates, they were all phosphorylated on both Ser³³⁸ and Thr¹⁹⁷. Because oxidation of the reactive Cys¹⁹⁹ or its replacement with Ala facilitates the removal of the phosphate on Thr¹⁹⁷ by λPP (38), we used this mutant, C^C199A, as a positive control for assessing the impact of phosphatase activity. As shown in Fig. 5B, C-subunit, C^E208A, C^R280A, C^F327A, C^Y204A, and C^E230Q are all resistant to dephosphorylation at both Thr¹⁹⁷ and Ser³³⁸. Importantly, the C^R336A mutant is also resistant to dephosphorylation at Thr¹⁹⁷; however, the C^R194A mutant, which is only phosphorylated at Ser³³⁸, is no longer resistant to dephosphorylation at Ser³³⁸. Overall, these data suggest that all these C-subunit mutants are resistant to phosphatase treatment except for C^C199A and C^R194A, thus confirming that Thr¹⁹⁷ phosphorylation is important for maintaining the structural integrity of the kinase domain. In addition, these studies highlight the importance of the highly reactive cysteine (C¹⁹⁹) two residues downstream of the phosphorylation site (T¹⁹⁷). This Cys renders the C-subunit sensitive to reactive oxygen species (ROS) -induced inactivation; oxidation of this cysteine also abolishes the resistance to phosphatases (38).

Fig. 5.

Dephosphorylation of the native C-subunit of PKA and various C-subunit mutants. (A) Schematic representation of residues on the C-subunit that were mutated. The figure was made in Pymol using Protein Data Bank ID code 1ATP. (B) Immunoblot analyses show that C-subunit and mutants C^R336A, C^E208A, C^R280A, C^Y204A, C^F327A, and C^E230Q are resistant to λPP. C^R194A is susceptible to λPP. C^C199A is shown as a positive control. C-subunit concentration was 2 μM, and 5 μL λPP were added per reaction. The total time was 45 min.

Dephosphorylation of Ser³³⁸ and Thr¹⁹⁷ Is Protected by the PKA R-Subunits.

Cys¹⁹⁹ in the WT C-subunit is very reactive, and when cells are under oxidative stress, the dissociated C-subunit is readily dephosphorylated and inactivated as a consequence of oxidation of Cys¹⁹⁹ (38). In contrast, Cys¹⁹⁹ is protected from oxidation in the holoenzyme. To test if association of the C^199A mutant with R-subunits protects both sites in the C-subunit from dephosphorylation, we incubated the C^199A mutant with RIα, RIIα (90–402), RIIβ (107–402), and RIIβ. We found that all of the R-isoforms (full-length R-subunit as well as monomeric deletion mutants) resist dephosphorylation of the C^199A mutant by λPP (Fig. 6A). These results suggest that the C-subunit is not susceptible to phosphatases during oxidative stress, unless they occur in association with increased cellular cAMP levels. Moreover, the findings imply that, at basal cellular levels of cAMP, if the C-subunit is associated with the R-subunit, it will not be accessible to phosphatases or ROS.

Fig. 6.

Phosphatase sensitivity of other AGC kinases and protection from λPP action on C^C199A by the R-subunits of PKA. (A) Immunoblot analysis shows that formation of holoenzyme with either monomeric R-subunit or full-length R-subunit protects C^C199A from dephosphorylation by λPP. Holoenzyme concentration was 2 μM, and 5 μL λPP were added per reaction. The total time was 45 min. (B) Immunoblot analysis using phosphospecific antibody shows that PKG1α is resistant to dephosphorylation by λPP in the presence and absence of cGMP. PKG1α concentration was 2 μM, and 5 μL λPP were added per reaction. The total time was 45 min. (C) Akt/PKB is sensitive to dephosphorylation by λPP, which was shown by immunoblot analysis using pThr308 antibody. Akt/PKB concentration was 2 μM, and 5 μL λPP were added per reaction. The total time was 45 min. (D) S6K1 (kinase domain) is sensitive to dephosphorylation by λPP, which was shown by immunoblot analysis using pThr229 antibody. S6K1 (kinase domain) concentration was 2 μM, and 5 μL λPP were added per reaction. The total time was 45 min.

PKG Activation Loop Phosphate Is also Resistant to Phosphatases, but Akt/PKB and S6K1 Are Susceptible to Dephosphorylation by Phosphatase.

PKA and PKG are the two AGC kinases that are effectors of cyclic nucleotide signaling. Based on our results showing that Ser³³⁸ and Thr¹⁹⁷ in the WT C-subunit are resistant to phosphatase and that the phosphates play a structural role (Fig. 5), we tested if resistance to phosphatase also occurs with PKG. We found that the activation loop phosphate (Thr ⁵¹⁵) in PKG, both in the absence and presence of cGMP, is resistant to λPP treatment (Fig. 6B). Like PKA, PKG does not have an HM phosphorylation site and also lacks a turn motif phosphorylation site. However, unlike PKA, PKG does not have an equivalent of Ser³³⁸. This lack of equivalent may be because PKG has inhibitory cyclic-nucleotide binding (CNB) domains as part of the same polypeptide chain as the kinase domain, thus guaranteeing that it will leave the ribosome in an inactive state. We assessed two other AGC kinases, Akt/PKB and S6K1 (kinase domain), which are regulated by insulin and growth factor signaling and not by second messenger signaling for their susceptibility to λPP. Fig. 6 C and D shows that both Akt and S6K1 are susceptible to λPP treatment, suggesting that PKA and PKG may be unique in their resistance to phosphatases. Akin to PKG, Akt/PKB, S6K1, and indeed, most kinases are synthesized with inhibitory domains fused to their kinase domains. Regulation is then achieved by posttranslational phosphorylation of the activation loop, which integrates critical residues to facilitate catalysis.

Discussion

PKA plays a regulatory role in virtually every mammalian cell, and its expression is conserved in early eukaryotes such as fungi and eukaryotic pathogens such as Plasmodia (39). PKA regulates a plethora of biological processes. It serves as a critical on/off switch for a number of key metabolic enzymes and is a regulator of gene transcription (40, 41). For certain proteins (e.g., NF-κB), the function of PKA-mediated phosphorylation may be more akin to a rheostat (10, 42). Along with other AGC kinases, PKA is also regulated by phosphorylation events. For certain AGC PKs such as ribosomal protein S6 kinase, four or more heterologous kinases are involved in the activation process (43). We show here, however, that the C-subunit of PKA is unique among the AGC kinases in several ways. Importantly, we find that the synthesis of the active C-subunit is regulated by only two key phosphorylation events and that assembly of the active C-subunit begins cotranslationally with the initial cis-phosphorylation of Ser³³⁸ in the C-tail taking place on the ribosome. By contrast, trans-phosphorylation of the activation loop Thr¹⁹⁷ is posttranslational, occurring after the protein leaves the ribosome. After the fully phosphorylated C-subunit is assembled, the phosphates resist removal by phosphatases. The C-subunit is then packaged with one of four functionally nonredundant R-subunits and usually anchored to a macromolecular scaffold. In this way, its activity is controlled exclusively by cAMP (probably bursts of localized cAMP) and not by turnover of the activation loop phosphate, which is the case for most other kinases. In the holoenzyme state, the C-subunit is resistant to oxidative stress, but on activation of the holoenzyme, the C-subunit is sensitive to oxidation of Cys¹⁹⁹ and subsequently, its resistance to phosphatases (Fig. 7).

Fig. 7.

Maturation and activation of the C-subunit of PKA. (A) The C-subunit of PKA is assembled by two distinct phosphorylation events: an initial intramolecular, cotranslational autophosphorylation of Ser³³⁸ in the C-terminal tail followed by an intermolecular, posttranslational phosphorylation of Thr¹⁹⁷ in the activation loop. The C-subunit then assembles into a tetrameric holoenzyme with two R-subunits and is kept in an inactive conformation. The C-subunit with only Ser³³⁸ phosphorylated is susceptible to PPs, but C-subunit that is part of holoenzyme is not sensitive to either PP or ROS. (B) On cAMP stimulation, R- and C-subunits of PKA dissociate, and the C-subunit carries out phosphorylation of target proteins. The free C-subunit is then susceptible to PP only in the presence of ROS and high cAMP concentrations.

The assembly of the C-tails of AGC kinases is tightly regulated and coordinated closely with phosphorylation of the activation loop. Cotranslational phosphorylation of the C-tail is also observed in PKB/Akt, another AGC family member; however, this phosphorylation is not a cis-phosphorylation but instead, is mediated by the mTOR complex 2 that is associated with ribosomes (44). The C-terminal phosphorylation site in PKA is different from the turn motif site that is conserved in other AGC kinases and usually constitutively phosphorylated (45) (Fig. S1). The PKA site is not only critical for the assembly of the mature enzyme but is also tightly regulated, presumably to ensure that active C-subunits do not leave the ribosome. Examination of the N- and C-terminal tails of the C-subunit reveals that both wrap around the N-and C-lobes of the kinase core and are an integral part of the active enzyme, with the A-helix in the N terminus being unique to PKA and having a high propensity to form a helix (Fig. S2). Although nothing is known about the in vivo or in vitro folding of the C-subunit and refolding of the denatured phosphorylated C-subunit does not occur, we predict that the β-rich N-lobe and the helix-rich C-lobe will also fold cotranslationally. The hydrophobic F-helix in the C-lobe would not likely be exposed to solvent as the protein is folding. Given that Ser³³⁸ is cis-phosphorylated cotranslationally, the C-tail must initially be positioned at the active site cleft. The P-2 Arg (residue 336) will presumably dock to Glu²³⁰ on the F-helix and Glu¹⁷⁰ on the catalytic loop. After the C-tail is phosphorylated, it will be strongly repelled from the active site because of electrostatic forces. As a final step, the phosphorylated C-tail would then dock onto the N-lobe, priming it for catalysis through its FDDY motif (residues 327–330), which anchors ATP and along with the glycine-rich loop and its hydrophobic motif (residues 347–350), positions the αC-helix for catalysis.

Maturation on the ribosome involves cis-autophosphorylation of Ser³³⁸ and a folding or chaperone-mediated event that is present in WT but disrupted in kin− S49 lymphoma cells; RI_α is also absent in kin− S49 cells. RI_α helps assure that unregulated C-subunits are not expressed in cells, and its deletion leads to embryonic lethality (46). RI_α is typically up-regulated if excess C-subunit is expressed, and its lethality is abolished if C_α and C_β are also deleted (46). Assembly/maturation of the active and fully phosphorylated C-subunit, thus, likely involves RI_α at some stage. We show here that the C^R336A mutant can associate with RI- and RII-subunits, an association that helps stabilize the C-subunit. In the absence of Ser³³⁸ phosphorylation, the C-subunit is prone to aggregation, which is consistent with a folding defect. Removal of Arg³³⁶ not only prevents the cis-autophosphorylation of Ser³³⁸ but also prevents aggregation. The conformation and dynamic properties of the C-tail are, thus, highly regulated.

It is important to note the differences observed when one abolishes a phosphorylation site (e.g., by modifying a critical recognition site such as the P-2 and/or P-3 Arg in PKA) compared with replacing Ser/Thr with Ala (to mimic the dephosphorylated state) or Glu/Asp (to mimic the phosphorylation). These different approaches yield different properties for the C-tail of PKA C-subunit. For example, Arg³³⁶ seems to have multiple roles, which was revealed by various experimental approaches. It primes the C-tail for cis-autophosphorylation at Ser³³⁸ and forms ionic interactions with the Ser³³⁸ phosphate. Thus, single residues not only can dictate the properties of a kinase but also guide its transition between active and inactive conformations.

Our results also reveal cross-talk between the C-tail and the activation loop after both sites are phosphorylated. When Thr¹⁹⁷ is phosphorylated, Ser³³⁸ is resistant to dephosphorylation by phosphatases, but this resistance is absent if Thr¹⁹⁷ is not phosphorylated. Another posttranslational modification associated with the activation loop of AGC kinases is oxidation of a conserved reactive cysteine that is two residues distal from the phosphorylation site (38). Oxidation of Cys¹⁹⁹ is protected in the holoenzyme, but after the C-subunit is released, it can be readily inactivated by oxidation; the phosphates can then be readily hydrolyzed (38). Oxidation of Cys¹⁹⁹ in the C(R336A) mutant will, thus, facilitate dephosphorylation of Thr¹⁹⁷ in mammalian cells.

The resistance of two C-subunit phosphorylation sites to phosphatases is also striking. Similar observations have been made for PKC when it is in its inactive state (47). The current data and results from other studies suggest that all three second messenger-activated kinases (PKA, PKG and PKC) evolved with a selection pressure for resistance to phosphatases, even in their phosphophorylated state. This resistance evolved to protect the kinases in cellular locations, because they form complexes with phosphatases on scaffold proteins (12). These kinases are, thus, activated in specific subcellular locations only by second messenger stimuli. In the absence of such stimuli, the kinases are protected from dephosphorylation. This phosphatase resistance may help ensure that the kinases are sensitive to their cognate second messenger and not to nearby phosphatases, which would be detrimental to the phosphorylation of cellular targets. Thus, the role of phosphatases in these complexes is perhaps to regulate the phosphorylation of substrates and not the kinase themselves. PKA and PKG, but not PKC, are unique in being protected from dephosphorylation even in their activated state.

Experimental Procedures

Plasmids and Constructs.

GFP-PKA C-subunit, mCherry-PKA C-subunit, GFP-R1α, and CFP-R2α were gifts from Roger Tsien (University of California at San Diego, La Jolla, CA). GFP-C^T197A, GFP-C^S338A, and HA-C^R336A were generated from the WT construct using standard procedures for site-directed mutagenesis, and products were sequence-verified. For E. coli expression, mouse PKA-C was cloned into the pRSET vector; the mutants C^E208A, CR^280A, C^Y204A, C^F327A, C^E208Q, and C^C199A were made from the WT construct using standard procedures. The proteins were overexpressed in E. coli as previously described (36). Phosphorylation-deficient mutants C^R194A, C^R336A, and C^K72H were generated from C-subunits cloned in pET15b with a polyhistidine tag and purified as described (13). PKB/Akt baculovirus was gift from Alexandra C. Newton (University of California at San Diego, La Jolla, CA).

Protein Purification.

All C-subunit constructs were made in E. coli using standard procedures (36). WT C-subunit and untagged mutants made from it were purified using P-11 resin followed by ion exchange and gel filtration chromatography. The C^R194A, C^R336A, and C^K72H were polyhistidine-tagged and purified using Probond Nickel resin (Invitrogen) followed by gel filtration. The purified proteins were tested for phosphorylation status using immunoblot analyses with phosphospecific antibodies (13).

Phosphorylation Assay.

For the in vitro kinase assay, we incubated 5 μM C^R194A with 10 mM Mg²⁺, 500 μM ATP, and 5× kinase buffer (50 mM Mops, pH 7.25, 125 mM pyrophosphate, 1 mM EGTA, 5 mM sodium orthovanadate, 5 mM DTT) in a reaction volume of 100 μL. The reaction was initiated by adding either PDK1 (20–50 nM) or C-subunit (20–50 nM) at 30 °C; 10 μL were removed at particular time points and quenched by 10 μL 2× SDS Laemelli buffer. The samples were run on 10% SDS/PAGE gel and probed for phosphorylation at Thr197 by phosphospecific antibodies. Similar assays were used to assess C^R336A and C^K72H and probed for phosphorylation using immunoblot analysis. The time course of GST C-tail phosphorylation by C-subunit was monitored using the kinase assay above. The GST C-tail (residues 299–350) of C-subunit was used as a substrate of C-subunit in a ³²P incorporation assay (6); RIIβ-subunit was used as a positive control.

IVTT.

His₆PKAs were overexpressed in a continuous exchange cell-free (CF) system as previously described (25) with further optimization. In general, CF extracts were prepared from the E. coli strain A19 (25), and T7-RNA polymerase was expressed using the pT7-911Q plasmid (48) and purified as described (49). CF expression was conducted in dialysis mode using Mini Slide-A-Lyzers with 20 kDa MWCO (Thermo Scientific) with 200 μL reaction mixture and 1,000 μL feeding mixture. Mini Slide-A-Lyzers were placed in a suitable plastic box containing 24 chambers for the feeding mixture and incubated in a shaker (New Brunswick Scientific) for 5–120 min at 30 °C. At each time point, a 5-μL reaction was pipetted and quenched with 2× SDS Laemelli buffer and frozen on dry ice. Duplicate samples were later evaluated for phosphorylation status. The remaining reaction mix was frozen on dry ice, and ribosomes were isolated as described (26). Briefly, 50 μL reaction mixture were layered on 500 μL sucrose bed (1.5 M) and spun at 627,000 × g in a table-top Beckman ultracentrifuge for 13 min. The pellet was washed two times followed by addition of Laemelli buffer. The supernatant and wash fractions were collected and stored for analysis.

Cell Culture.

HEK293 cells were maintained in DMEM with 10% (vol/vol) FBS. The cells were transfected in 6 cm² for 24 h using Fugene. All PKA-C constructs were GFP-tagged. The cells were then harvested and lysed using RIPA buffer. Whole-cell lysates were spun at 18,000 rpm to separate the supernatant and insoluble fraction. The fractions were solubilized in 2× Lamelli buffer and analyzed for their expression and phosphorylation status by immunoblot analysis. WT and kin− S49 cells were grown in DMEM with 10% (vol/vol) horse serum. Cells were harvested and lysed in RIPA buffer, and soluble and insoluble fractions were separated by centrifugation at 18,000 rpm and assessed by immunoblot analysis.

Fluorescence Microscopy.

HeLa cells were grown in 3.5 cm² Met Tek dishes with DMEM plus 10% FBS. GFP-R1α or CFP-RIIα were cotransfected with mCherry-C^R336A and d-AKAP1 for mitochondrial localization. After 24 h, cells were washed with PBS, fixed using 4% paraformaldehyde in PBS, and imaged with Zeiss Axiovert 200 M in separate channels for GFP, CFP, and mCherry.

Helical Propensity of the C-Tail.

The C-tail peptide of the C-subunit (residues 327–350) was analyzed by an Agadir tool that predicts secondary structure (27, 28). Additionally, the above peptide was mutated in silico at key residues, including R336A, S338A, and S338E, and assessed for predicted secondary structure.

Phosphatase Treatment of C-Subunit.

All C-subunit mutants were purified to homogeneity; 2 μM each protein was incubated with 5 μL λPP (Invitrogen), and the phosphatase assay (45 min at 30 °C) was assayed per the manufacturer's protocol. The reaction was quenched with Laemelli buffer, and phosphorylation statuses before and after treatments were evaluated using the phosphospecific antibodies described above. To assess the protection from phosphatase action, the C^199A mutant was also incubated with different R-subunits and treated for 45 min with λPP. Similar assays were conducted with PKG1a, PKB/Akt, and S6K1.