Protein Kinase A Regulates MYC Protein through Transcriptional and Post-translational Mechanisms in a Catalytic Subunit Isoform-specific Manner*

Background: MYC is rapidly degraded in cells, and its accumulation is associated with many human malignancies. Results: Sequential phosphorylation of MYC by protein kinase A (PKA) and polo-like kinase 1 (PLK1) protects MYC from proteasome-mediated degradation. Conclusion: A MYC-PKA-PLK1 signaling loop exists in cells. Significance: We highlight the importance of considering possible regulatory feedback loops while targeting molecules occupying hub positions in signaling pathways. MYC levels are tightly regulated in cells, and deregulation is associated with many cancers. In this report, we describe the existence of a MYC-protein kinase A (PKA)-polo-like kinase 1 (PLK1) signaling loop in cells. We report that sequential MYC phosphorylation by PKA and PLK1 protects MYC from proteasome-mediated degradation. Interestingly, short term pan-PKA inhibition diminishes MYC level, whereas prolonged PKA catalytic subunit α (PKACα) knockdown, but not PKA catalytic subunit β (PKACβ) knockdown, increases MYC. We show that the short term effect of pan-PKA inhibition on MYC is post-translational and the PKACα-specific long term effect on MYC is transcriptional. These data also reveal distinct functional roles among PKA catalytic isoforms in MYC regulation. We attribute this effect to differential phosphorylation selectivity among PKA catalytic subunits, which we demonstrate for multiple substrates. Further, we also show that MYC up-regulates PKACβ, transcriptionally forming a proximate positive feedback loop. These results establish PKA as a regulator of MYC and highlight the distinct biological roles of the different PKA catalytic subunits.

MYC is a basic helix-loop-helix leucine zipper transcription factor that regulates a large number of target genes important in cell growth, metabolism, differentiation, proliferation, and apoptosis (1)(2)(3)(4)(5)(6). Complete loss of MYC function results in embryonic lethality, whereas its overexpression predisposes cells to malignant transformation (7)(8)(9). MYC overexpression is an early and consistent feature of many human malignancies, where it is suspected to regulate key events in tumorigenesis.
Because MYC drives the transcription of genes important in multiple cellular processes, precise temporal regulation of MYC is required. Tight regulation of MYC is achieved through multiple mechanisms at the transcriptional, post-transcriptional, translational, and post-translational levels (10 -14). Once translated, MYC is rapidly turned over in normal cells. The MYC steady-state level is regulated at the post-translational level through a series of exquisitely orchestrated phosphorylation events. Ras-mediated activation of MAPK stabilizes MYC by phosphorylation at Ser-62 within the evolutionarily conserved MYC box I region (11). Ser-62-phosphorylated MYC is recognized by GSK3␤, which phosphorylates Ser-62primed MYC at Thr-58. Peptidylprolyl isomerase 1 then catalyzes a cis to trans isomerization of the bond preceding Ser-62, thereby allowing the trans-specific protein phosphatase 2A to remove the stabilizing phosphorylation at Ser-62 (15,16). MYC phosphorylated at Thr-58, but not Ser-62, is recognized by the E3 ligase Fbw7, which ubiquitinates MYC at the N terminus and targets it for proteasome-dependent degradation (11,17). Recently, another E3 ligase, ␤trcp, was shown to interact through a previously unknown phospho-degron and oppose the Fbw7-mediated ubiquitination at the N terminus. PLK1 3 phosphorylation within the phospho-degron was reported to be critical for ␤trcp binding (18).
In this study, we demonstrate the existence of a MYC-PKA-PLK1 signaling loop and show that MYC is regulated both through transcriptional and post-translational mechanisms by PKA, in a PKA catalytic subunit isoform-specific manner. This work also highlights both the promise and potential pitfalls of global kinase inhibition and emphasizes the need to develop next generation therapeutic strategies capable of disrupting specific kinase-substrate interactions.
Expression and Purification of Recombinant Proteins-All proteins were cloned with an N-terminal His 6 tag and expressed in either E. coli or mammalian cells (COS cells or PC3 cells). Recombinant proteins were purified under native conditions using Ni-affinity chromatography. The purified proteins were further enriched by ion-exchange chromatography whenever necessary. The final buffer condition for all proteins used in in vitro kinase assay is 50 mM NaH 2 PO 4 , 300 mM NaCl, 250 mM imidazole, and 0.1% Tween 20.
Western Blotting and Immunoprecipitation-For Western blot analysis, mammalian cells were harvested 24 -48 h posttransfection and lysed using 150 mM NaCl, 50 mM Tris (pH 7.5), 0.5% Nonidet P-40, and Complete protease inhibitor (Roche Applied Science). Western blotting was performed as described previously (21). Total protein from mouse prostate was extracted during RNA extraction using the protocol described in the RNeasy Kit (Qiagen). Western blotting was performed as described (22). PKAC␤2 protein containing the N-terminal HA and His 6 tag was purified from transiently transfected COS and PC3 cells either by immunoprecipitation using an antibody against the HA tag or by Ni-affinity chromatography (further enriched by anion-exchange chromatography).
In Vitro Kinase Assay-In vitro kinase assay was performed using purified recombinant proteins as described previously (21,23). Kinase assay was carried out for 2 h at room temperature and stopped using 2ϫ Laemmli buffer. After in vitro kinase assay, the proteins were analyzed by SDS-PAGE, transferred to a PVDF membrane, and analyzed by autoradiography. The PVDF membrane was also stained with Coomassie Brilliant Blue to reveal protein loading.
Reverse In-gel Kinase Assay (RIKA)-50 mg/ml PKAC␣ in 8 M urea, 50 mM NaH 2 PO 4 , 300 mM NaCl, 250 mM imidazole, and 0.1% Tween 20 was co-polymerized in the gel, and RIKA was performed as described previously (23). RIKA was performed on a PVDF membrane for PLK1-priming experiments (onmembrane kinase assay). Purified recombinant MYC and MYC mutants were resolved by SDS-PAGE and transferred onto a PVDF membrane. The proteins on the membrane were refolded using the buffer conditions used in RIKA. The membrane was incubated with purified active recombinant PLK1 (PLK1 T210D ) for 2 h in a buffer containing 20 mM Tris (pH 7.6), 20 mM MgCl 2 , 50 mM DTT) and [␥-32 P]ATP. The membrane was washed with water overnight to remove nonspecifically bound ATP. Phosphorylation of MYC by PLK1 was detected by autoradiography.
Protein Identification and Phosphosite Mapping Mass Spectrometry-In-gel tryptic digestion was performed as described previously (23). Desalted tryptic peptides were analyzed by nano-LC-tandem MS on a linear ion-trap mass spectrometer (LTQ; Thermo Fisher). Acquired data were searched against a Homo sapiens protein database or phosphoproteome database using the TurboSEQUEST algorithm (Thermo Fisher).
Immunofluorescence-Cells were plated on poly-L-lysinetreated glass coverslips in 6-well plates. PC3 cells were grown on coverslips and transfected using lipoD293T (Signagen) according to the manufacturer's instructions. Transfected cells were fixed using 2% paraformaldehyde in PBS. Immunofluorescence analysis was performed as described previously (22).

PKA Activity Influences the Steady-state Levels of MYC-
During our efforts to identify PKA substrates in prostate cancer cells, we identified ␣-enolase as a PKAC␣ substrate (Table 1). An alternative in-frame internal translation initiating at ϩ291 of the ␣-enolase mRNA yields an alternate product termed MYC promoter-binding protein 1 (MBP1) (24). We confirmed both ENO1 and MBP1 to be substrates of PKAC␣ by in vitro kinase assay (Fig. 4E). Because MBP1 is a known transcriptional repressor of MYC, we sought to determine whether phosphorylation of MBP1 by PKA has an effect on MBP1 function and steady-state MYC level (25). We monitored changes in endogenous MYC level in prostate cancer cells treated with the PKAselective small molecule inhibitor H89 by Western blotting. Pan-PKA inhibition using H89 resulted in decreased MYC accumulation in prostate cancer cells (Fig. 1A). This was found to be consistent in HeLa and K562 cells as well (Fig. 1B). Inhibition of PKA using the PKA-specific peptide inhibitor PKI also destabilized MYC (Fig. 1C). Conversely, activation of PKA using forskolin increased MYC levels (Fig. 1D). Although these results were consistent with our initial hypothesis, the rapid kinetics suggested that the effect could be the result of a direct interaction between PKA and MYC. To determine whether the effect of PKA on MYC is transcriptional, we evaluated changes in MYC mRNA levels upon PKA inhibition by qRT-PCR. The MYC mRNA level did not change upon H89 treatment, suggesting the effect of PKA inhibition on MYC to be post-transcriptional (Fig. 1E). Because MYC is degraded in cells via the 26S proteasome, we determined whether inhibition of protea-some activity rescues the effect of PKA inhibition on MYC accumulation. MG132 rescued the effect of PKA inhibition, thereby demonstrating that PKA protects MYC from 26S proteasome-mediated degradation (Fig. 1F).
MYC Phosphorylation at Ser-279 by PKA Stabilizes MYC-To determine whether PKA can phosphorylate MYC, we performed in vitro kinase reactions using recombinant PKAC␣ and MYC in the presence of [␥-32 P]ATP. PKA efficiently phosphorylated MYC, and the phosphorylation was inhibited by H89 (Fig. 2, A and C). LC-MS 2 analysis identified Ser-279 as a   and E). Furthermore, the region spanning Ser-279 is predicted to be solvent-exposed and intrinsically disordered (data not shown). Such intrinsically disordered domains have been implicated in the regulation of protein stability (26).
To determine whether the site phosphorylated by PKA on MYC is accessible and able to be phosphorylated in vivo, we transfected His 6 -3ϫHA-tagged MYC in PC3 cells. The cells were harvested 36 h post-transfection, lysed (using 150 mM The corresponding amino acid in the peptide is also labeled in red. The mass difference between these ions is also indicated. A complete list of ions for the given peptide is listed in supplemental Table S1. Bottom panels, LC-MS 2 spectra of the peptide RSESGSPSAGGHSKPPHSPLVLK from PKA-phosphorylated recombinant MYC is shown. A region spanning Ser-279 is shown to reveal the phosphorylated state at Ser-279.The b ions corresponding to Ser-277 and phosphorylated Ser-279 (also designated by an asterisk) are labeled using a red arrow. The corresponding amino acid in the peptide is also labeled in red. The positions of the hypothetical Glu-278 and nonphosphorylated Ser-279 ion are indicated by the black arrow. The mass difference between these ions is also indicated. A mass increase of 80 Da was observed in the phosphorylated peptide, demonstrating that Ser-279 is phosphorylated. A complete list of ions for the given peptide is listed in supplemental Table S2. NaCl, 50 mM Tris (pH 7.5), 0.5% Nonidet P-40 and Complete protease inhibitor), and MYC was purified using Ni-affinity chromatography. The purified MYC was divided in half. One half was completely dephosphorylated by hydrogen fluoride (HF) treatment as described (27). The remaining half was left untreated and contained the pool of MYC molecules that retained their in vivo phosphorylation status. We performed a RIKA with PKA in the gel on the HF-treated and untreated purified MYC samples. If MYC is phosphorylated in vivo at the site phosphorylated by PKA in vitro (in a RIKA), then we would expect an increase in phosphorylation signal in the HF-treated samples (because these sites would be dephosphorylated upon HF treatment and become available to be phosphorylated during the RIKA). Total levels of MYC loaded in these samples were quantified by anti-HA Western blotting, and the change in phosphorylation signal was calculated after normalization as described previously (27). This process is illustrated in Fig. 3A. We observed ϳ50% increase in MYC phosphorylation upon HF treatment, suggesting that 50% of the in vivo MYC population in the cells was phosphorylated at the PKA site under the given culture conditions (Fig. 3B). Importantly, endogenous PKA and MYC co-localized to the nucleus (Fig. 3C).
To ascertain whether MYC phosphorylation at Ser-279 by PKA affected the MYC steady-state level, HA-tagged MYC WT and MYC S279A were transfected into PC3 cells, and the effect of PKA inhibition on their accumulation was analyzed. Whereas MYC WT was destabilized upon H89 treatment, the steady-state level of MYC S279A was unresponsive to PKA inhibition (Fig.  3D). Further, the MYC S279A steady-state level was considerably lower compared with MYC WT (Fig. 3D). These data demonstrate that phosphorylation at Ser-279 by PKA plays a role in stabilizing MYC in cells. No difference in subcellular localization between the endogenous MYC, HA-tagged MYC WT and HA-tagged MYC S279A mutant was observed (data not shown).
PKA Phosphorylation at Ser-279 Primes MYC for PLK1 Phosphorylation-A recent report showed the region spanning Ser-279 to constitute a phospho-degron (Fig. 4A). PLK1-mediated ␤trcp binding within this region was shown to stabilize MYC (18). However, neither the events preceding PLK1 phosphorylation nor the PLK1 phospho-acceptor site(s) on MYC is known. Because both PLK1 and PKA phosphorylate residues within this phospho-degron, we sought to clarify their roles. We found that the effect of combined inhibition of PKA and PLK1 on MYC stability is more profound compared with inhibiting either kinase alone (Fig. 4B). We hypothesized three possible scenarios: (i) PKA and PLK1 phosphorylate the same residue on MYC (i.e. Ser-279) and hence elicit a similar response; (ii) PKA and PLK1 phosphorylate mutually independent residues on MYC within the phospho-degron; and (iii) PKA phosphorylates MYC at Ser-279, which primes PLK1 to phosphorylate an adjacent residue within the phospho-degron by relieving the inhibitory effect of the polo-box domain or by making an adjacent site a more favorable PLK1 phospho-acceptor (Fig.  4C). To distinguish among these models, we performed an in vitro kinase reaction using recombinant PLK1 T210D and recombinant MYC, which was either prephosphorylated by PKA in vitro using nonradiolabeled ATP (primed) or not prephosphorylated by PKA (unprimed). His 6 -tagged MYC purified from E. coli was prephosphorylated by in vitro kinase assay using recombinant PKAC␣ and nonradiolabeled ATP. Nonphosphorylated MYC served as the unprimed control. Subsequently, H89 (30 M), was added to all reactions to inhibit PKA activity. The absence of radiolabeling upon further incubation (1 h at room temperature) of the reactions containing MYC, PKA and H89 with [␥-32 P]ATP confirmed complete inactivation of PKA by H89. Recombinant PLK1 T210D and [␥-32 P]ATP was then added to these tubes containing either primed or unprimed MYC. The in vitro kinase reaction was allowed to proceed for 1 h at room temperature. Radioactive labeling of MYC by PLK1 T210D was assessed by running the reaction on a SDS-polyacrylamide gel, transferring to a PVDF membrane followed by autoradiography. We used the PLK1 T210D mutant version of PLK1 because it was previously shown to be constitutively active. The mutation at Thr-210 mimics the phosphorylation of PLK1 at this site within the activation loop, and phosphorylation at Thr-210 is necessary for PLK1 activation (28). We observed that the phosphorylation signal of MYC by PLK1 T210D (using [␥-32 P]ATP) was higher when MYC was primed by PKA (using nonradiolabeled ATP) compared with unprimed MYC (Fig. 4D). This result suggested that PKA phosphorylation at Ser-279 on MYC primes subsequent MYC phosphorylation by PLK1. To confirm this, we generated the phospho-mimetic mutant at Ser-279 (MYC S279D ) and performed an in vitro kinase assay using PLK1 T210D . We observed that PLK1 phosphorylation of the MYC S279D mutant was dramatically higher compared with MYC WT and other MYC mutants (Fig. 4E). We also compared the ability of PLK1 T210D to phosphorylate other MYC substitution mutants in the region and found MYC S279D to be a more efficient PLK1 substrate compared with MYC WT and other mutants (Fig. 4F). These results showed that MYC phosphorylation at Ser-279 increases the efficiency of subsequent PLK1 T210D phosphorylation of MYC, possibly at an adjacent site. LC-MS 3 analysis of MYC S279D phosphorylated in vitro using PLK1 T210D revealed Ser-281 as the PLK1 phospho-acceptor site on MYC (Fig. 5). MYC S279D phosphorylation by PLK1 T210D was significantly compromised in the MYCS 279D:S281A double mutant (Fig. 4G), thus confirming Ser-281 as the major PLK1 phospho-acceptor site on MYC that is primed by Ser-279 phosphorylation.
PKA Isoform-specific Transcriptional Repression of MYC-Three different genes encode PKA catalytic subunits in humans: C␣, C␤, and C␥. The C␣ and C␤ isoforms have multiple splice variants. To validate the effect of PKA inhibition on MYC and to dissect the role of individual PKA catalytic subunit isoforms in stabilizing MYC, we knocked down either PKAC␣ or PKAC␤ using siRNA in PC3 cells. Analysis of siRNA-treated samples yielded an unexpected but intriguing result. Contrary to our expectation based on chemical PKA inhibition, knockdown of PKAC␣ increased MYC steady-state level in PC3 cells (Fig. 6A). In contrast, PKAC␤ knockdown caused a slight decrease in MYC levels. Because siRNA knockdown occurs over an extended time period, we hypothesized that the result of PKAC␣ knockdown on MYC could be through an indirect mechanism. Comparison of MYC mRNA after siRNAmediated knockdown of PKA catalytic subunits revealed that prolonged knockdown of PKAC␣, but not PKAC␤, increased the steady-state MYC mRNA level (Fig. 6B). These data reveal a link between PKAC␣ activity and MYC transcription and, for the first time, show that the different PKA catalytic subunits perform distinct biological functions.
PKA Catalytic Subunit Isoforms Have Both Overlapping and Distinct Substrates-The ability of PKA catalytic subunits to have a distinct effect on the transcription of MYC suggested exclusive, nonredundant roles for PKA isoforms. We hypothesized that the different PKA catalytic subunit isoforms have distinct substrate profiles and hence elicit different responses upon activation. To test this hypothesis, we profiled and compared the substrates of PKA catalytic subunit isoforms by RIKA (23). To identify PKAC␣ substrates, the PKAC␣ RIKA was first standardized (Fig. 7A). LNCaP cell lysate was fractionated by anion-exchange chromatography, and a one-dimensional PKAC␣ RIKA on the enriched fractions was performed to identify fractions containing PKA substrates (data not shown). Fractions containing PKAC␣ substrates were used for two-dimensional PKAC␣ RIKAs. The signal on RIKA gels was aligned with parallel silver-stained gels (Fig. 7B). PKA substrates were excised from the silver-stained gels and identified by mass spectrometry (Table 1). We validated a subset of the identified C␣ substrates by in vitro kinase reaction using purified recombinant proteins (Fig. 7C).
To address whether substrate diversity contributes to the observed functional nonredundancy among PKA catalytic subunits, we tested the ability of PKAC␤1 and PKAC␤2 isoforms to phosphorylate these identified PKAC␣ substrates in vitro. Cat- alytically active PKAC␤2 was expressed and purified from mammalian cells (COS cells or PC3 cells) because PKAC␤2 purified from E. coli was catalytically inactive (data not shown). We observed that whereas PKA␤1 phosphorylated all of the tested PKAC␣ substrates (Fig. 7D), PKAC␤2 phosphorylated only the telomerase-binding protein among the PKAC␣ substrates tested (Fig. 7E). Addition of the pan-PKA inhibitor H89 diminished the phosphorylation of the telomerase-binding protein by PKAC␤2 (Fig. 7F). These differences in substrate selectivity were not totally unexpected because PKAC␣ and PKAC␤1 share a very high sequence similarity (92%), whereas PKAC␤2 has an additional 62 amino acids at its N terminus not present in PKAC␤1 (Fig. 7G). In silico secondary structure prediction of PKAC␤2 N terminus revealed an amphipathic helix region within this unique N terminus (Fig. 7G). We believe that this region might play a role in the differential substrate selectivity of PKAC␤2.
Increased PKAC␤ Expression in MYC-overexpressing Prostate Epithelial Cells-PKAC␤ is known to be a direct transcriptional target of MYC in rat fibroblasts and lymphocytes (29). In these cells, it was further suggested that PKAC␤ plays an important role in MYC-mediated transformation (29). MYC overexpression is a consistent and key early event in prostate cancer, where it drives proliferation (30). Rapidly proliferating prostate epithelial cells have also been reported to have higher levels of the PKAC␤2 isoform (31). These results prompted us to determine whether MYC overexpression in prostate cancer cells influences the PKAC␤ level (and the PKAC␤2 splice variant), thereby forming a proximate positive feedback loop. Ectopic expression of MYC in PC3 cells resulted in increased PKAC␤ and PKAC␤2 expression (Fig. 8A). We then compared the levels of PKAC␤2 protein in the prostates of mice overexpressing MYC specifically in the prostate to their age-matched controls. We observed a positive correlation between MYC expression and PKC␤2 levels in the prostates of multiple mouse models in which MYC expression was controlled by distinct prostate-specific promoters (Fig. 8B). We could not compare levels of PKAC␤1 protein in these experiments due to the unavailability of an antibody that would allow us to differentiate between the PKAC␤1 and PKAC␣ subunits, which have the same apparent molecular mass (ϳ42 kDa). Oncomine analyses of previously published human prostate cancer microarray datasets revealed that both MYC and PKAC␤ are overexpressed in human prostate cancer cases (Fig. 8C) (32,33). Oncomine analyses of multicancer datasets also revealed elevated PKAC␤ in prostate cancer (data not shown) (34,35).

DISCUSSION
Diverse factors and signaling pathways influence MYC accumulation. In this study, we identified a role for PKA in MYC regulation and demonstrated the existence of a MYC-PKA-PLK1 signaling loop in cells. Our data show that MYC transcriptionally up-regulates PKAC␤, and PKA in turn protects MYC from proteasome-mediated degradation through phosphorylation at Ser-279, establishing a proximate positive feedback interaction. Ser-279 lies within a phospho-degron recently shown to be important in MYC stability (18). These studies demonstrated this region to mediate binding of the E3 ligase ␤trcp, which competes with Fbw7 (recruited through Thr-58 phosphorylation) for ubiquitination at the MYC N terminus (18). Phosphorylation within the degron was shown to be important for ␤trcp recruitment, and PLK1 was suggested to phosphorylate MYC within this region (18). Our data demonstrate that MYC phosphorylation at Ser-279 by PKA primes subsequent PLK1 phosphorylation at Ser-281. Thus, through sequential phosphorylation, PKA and PLK1 cooperatively stabilize MYC.
Interestingly, whereas brief pharmacologic pan-PKA inhibition diminished MYC level, prolonged PKAC␣ knockdown resulted in transcriptional up-regulation of MYC. These results provide an example of contrasting functional responses between short term and long term kinase inhibition. Furthermore, unlike in the case of PKAC␣ knockdown, we observed no change in MYC mRNA level upon PKAC␤ knockdown, demonstrating functionally distinct roles for PKA catalytic isoforms in the transcriptional regulation of MYC. We hypothesized this transcriptional response to be mediated through a PKAC␣-specific substrate, possibly a transcription factor that regulates MYC transcription. Although the PKAC␣ target(s) critical for FIGURE 6. PKA catalytic subunits regulate MYC differentially. A, long term knockdown of PKAC␣ by siRNA (48 -72 h) resulted in increased MYC protein level whereas knockdown of PKAC␤ subunit caused a slight decrease in MYC protein level. Western blotting using pan-PKAC antibody confirmed knockdown of PKAC␣. PKAC␤ knockdown was confirmed by semiquantitative PCR (Fig. 6B). The change in MYC level was quantified using ImageJ software and normalized to the corresponding ␤-actin level. B, the effect of PKAC␣ knockdown on MYC is transcriptional. Semiquantitative PCR and Northern blot analysis reveal an increase in MYC mRNA upon siRNA knockdown of PKAC␣ but not PKAC␤.
this response are unknown, CBP/p300 is a plausible candidate. Recruitment of CBP/p300 at the MYC promoter is known to down-regulate MYC transcription (36 -38). CBP/p300 is a known PKAC␣ substrate, but the effect of PKA phosphorylation on the DNA binding ability of CBP or recruitment to the MYC promoter region has not been investigated (39,40). In this FIGURE 7. PKA catalytic subunits have distinct substrate selectivity. A, two-dimensional PKAC␣ RIKA using an LNCaP whole-cell protein extract. Numbers on the top left and right corners indicate pH of the IPG strip. The autoradiograms from gel containing catalytically active PKAC␣ (top) and the kinase-dead PKAC␣ control (bottom) are shown. PKA substrates appear as signals on the autoradiograms. The signal on the control gel is likely due to autophosphorylation of endogenous kinases or phosphorylation of the co-polymerized kinase-dead PKAC␣ or co-migrating proteins by endogenous kinases. B, two-dimensional RIKA on anion exchange fraction. A parallel gel from the same fraction (containing no kinase) was silver-stained to enable substrate identification (bottom). Arrows and closed circles represent several substrates used for alignment. Red-dashed circle represents an abundant non-PKAC␣ substrate protein, which was used as an internal marker to align the gel. C, in vitro kinase assay using recombinant substrates and PKAC␣. Substrates identified by RIKA are phosphorylated by PKAC␣ in vitro. Arrowheads indicate phosphorylated substrates. D, PKAC␤1 has a substrate profile similar to that of PKAC␣. In vitro kinase assay was performed using recombinant PKAC␤1 and PKAC␣ substrates. PKAC␤1 phosphorylated all tested PKAC␣ substrates in vitro. Arrowheads indicate phosphorylated substrates. E, PKAC␤2 has a distinct substrate profile compared with PKAC␣ and PKAC␤1. F, H89 inhibits phosphorylation of the telomerase-binding protein (TEBP) by PKAC␤2. G, amino acid sequence alignment of N-terminal region PKAC␣ and PKAC␤2 shows the presence of an additional N-terminal arm (62 amino acids) in PKAC␤2 isoform. The Yaspin secondary structure prediction program was used to predict the ␣-helical region in the N-terminal 62 amino acids in PKAC␤2, and the helical wheel program available from the University of Virginia was used to reveal the amphipathic nature of the predicted helix region (indicated by purple helix over the corresponding sequence). regard, it will also be interesting to investigate the role of MBP-1 phosphorylation by PKA. During the course of this study we had identified MBP-1 as a PKAC␣ (but not PKAC␤2) substrate.
Comparison of substrate selectivity of the different PKA isoforms revealed PKAC␤2 to have distinct phosphorylation selectivity compared with PKAC␣ and PKAC␤1. We attribute this difference to a predicted amphipathic helix in the N terminus of PKAC␤2, which could mediate substrate recognition and binding. Except for this unique N terminus region, PKAC␤2 is identical to PKAC␤1, and PKAC␤1 was shown to phosphorylate all PKAC␣ substrates tested. Substrate docking domains distant from catalytic sites have been reported for multiple protein kinases (41). In addition, amphipathic helices have been reported to mediate several kinase-substrate interactions (42). We suggest that the solvent-exposed N-terminal amphipathic helix of PKAC␤2 acts as a substrate-docking site and hence accounts for the differences in its substrate selectivity. These studies reveal the possibility of antithetical roles for closely related protein kinase isoforms and highlight the risk of therapeutic strategies aimed at global kinase inhibition. PKAC␣ overexpression in LNCaP cells was reported to induce trans-differentiation of these cells into neuroendocrine-like cells, which divide slowly. It has also been reported that PKAC␤2 is overexpressed in rapidly proliferating prostate cancer cells. Such discordance in the phenotype of cells overexpressing the different PKA isoforms argued for a significant functional difference between these proteins. The data reported here provide a mechanistic premise for the distinct functional roles of PKA catalytic subunits in prostate cancer. Although PKAC␣ has been extensively studied and many of its substrates are known, the role of other PKA catalytic isoforms has received less attention. This report is the first detailed analysis of differences in substrate selectivity among PKA catalytic isoforms and the functional implications of these differences. Collectively, our data demonstrate that differential substrate selectivity and functional diversity among protein kinase isoforms are critical for modulation and precise signaling regulation in response to stimuli of varying intensity and perdurance.
Our model suggests that short PKA activity bursts in the cell stabilize MYC through a post-translational mechanism, whereas prolonged activation of PKAC␣ transcriptionally represses MYC (Fig. 8D). This establishes a regulatory loop and protects the cells from downstream effects of post-translational MYC stabilization as a result of prolonged elevated PKAC␣ activity. This mechanism could contribute to the slow proliferative rate of LNCaP cells trans-differentiated to a neuroendocrine-like phenotype by PKAC␣ overexpression. MYC accumulation in turn activates a positive feedback loop that transcriptionally activates PKAC␤. PKAC␤ protects MYC from proteasomal degradation, but does not repress MYC transcriptionally. This is consistent with the higher expression of PKAC␤ subunits in rapidly proliferating, MYC-overexpressing prostate cancer cells (31)(32)(33). Thus, through differential substrate selectivity, PKA catalytic isoforms regulate MYC differentially.
It is clear from the results presented here that the short and long term effects of PKA inhibition differ substantially. These observations underscore the need to establish kinase inhibition strategies that interfere with the ability of the kinase to interact with specific target substrates without impairing the function of the kinase in general. It may be possible to achieve this pharmacologically using peptides or peptidomimetics that interfere with interaction of protein kinases and specific substrates. Clearly, this strategy, which must be predicated upon an indepth understanding of the structural aspects of kinase-substrate interactions, could enable subtle, yet functionally significant pathway disruption with fewer side effects and reduced toxicity compared with global inhibition strategies.