A RSK(y) Relationship with Promiscuous PKA

See allHide authors and affiliations

Science's STKE  22 Aug 2006:
Vol. 2006, Issue 349, pp. pe32
DOI: 10.1126/stke.3492006pe32


Where, when, and with "whom" do molecules interact? Such relations in space and time are key concepts that currently engage investigators of cellular signaling processes. The notion of compartmentalized signaling grew out of studies of adenosine 3′,5′-monophosphate (cAMP) signaling processes, and this area continues to generate exciting new paradigms. Distinct clouds of cAMP are formed and shaped within cells by tethered cAMP phosphodiesterases (PDEs). AKAPosomes, formed from distinct subpopulations of cAMP-dependent protein kinase (PKA) tethered to anchoring proteins (AKAPs) together with specific substrate molecules, interpret these gradients to generate individualized responses. PKA activity is also regulated by the interaction of other proteins with the regulatory (R) or catalytic (C) subunits of PKA, and a mechanism has been uncovered in which ribosomal S6 kinase (RSK1) interacts with either PKA subunit, depending on whether RSK1 has been phosphorylated and activated by extracellular signal–regulated kinase (ERK). Thus, inactive RSK1 binds the RI subunit of PKA to sensitize it to activation, whereas activated RSK1 binds the C subunit to desensitize PKA to cAMP activation. Cross-talk between the key cAMP and ERK signaling pathways provides a mechanism that, along with distinct mechanisms of both positive and negative attenuation provided by Raf and PDE4 isoforms, can be tailored on a cell type–specific basis.

Relationships in space and time . . . these are key concepts that currently engage investigators of cellular signaling processes. It’s becoming increasingly apparent that precisely where processes occur within the three-dimensional matrix of the cell, and "who" the partners are, are crucial to allow for effective intracellular signaling. Delineating these fundamental processes will give insight into not only normal signaling function but also pathological changes and thus is likely to provide ways to generate novel therapeutics and diagnostics.

In this arena, studies of adenosine 3′,5′-monophosphate (cAMP) signaling have provided important models (1, 2). The recent study by Chaturvedi et al. (3) offers a novel and exciting twist in showing that ribosomal S6 kinase (RSK1) can interact with and modify the functioning and properties of cAMP-dependent protein kinase (PKA). The ramifications of this study for the signaling of both these important kinases have yet to be fully appreciated, so to put it in context, let’s set the scene.

Cyclic AMP Signaling Is Compartmentalized

The concept of compartmentalized signaling grew out of studies of cAMP signaling processes occurring in cardiac myocytes (4, 5). This work identified subpopulations of PKA that were selectively activated by distinct hormone receptors despite the fact that each of these receptors stimulated adenylyl cyclase activity by coupling to the heterotrimeric GTP-binding protein (G protein) Gs. These data implied the formation of gradients of cAMP within the cell and the targeting of discrete PKA subpopulations that were able to sample them selectively. Recently, the molecular basis of this system has been uncovered, giving detail to the cellular machinery involved in generating, detecting, and acting on the gradients of cAMP that occur within living cells, so as to generate compartmentalized responses within discrete spatial environments (1).

First, genetically encoded probes have unequivocally demonstrated the occurrence of cAMP gradients within cardiac myocytes and various other cells (6, 7). Second, the ablation of cAMP degradation by inhibiting cAMP phosphodiesterase (PDE) activity destroys compartmentalized cAMP signaling by allowing cells to fill uniformly with cAMP (4, 6). PDEs, which are encoded by multiple gene families (5, 8), underpin the generation of intracellular cAMP gradients (9). They achieve this by being targeted to specific intracellular sites by either protein-protein or protein-lipid interaction (1). Isoforms of the PDE4 family have provided the model for this, with their N-terminal portions shown to contain motifs acting as "ZIP codes" for precise intracellular targeting of specific isoforms. This allows the assembly of complexes that then orchestrate the formation and shaping of intracellular "clouds" and gradients of cAMP (10). Thus, individual PDE4 isoforms have specific and distinct functional roles (11). Third, a mechanism is needed to detect such gradients and generate discrete, spatially constrained responses. Two major systems serve to generate signals in response to binding cAMP. One is the relatively recently identified EPAC protein, which upon binding cAMP activates the monomeric G proteins Rap1 and Rap2 (12). The other is PKA, which in its inactive state is a heterotetramer consisting of two regulatory (R) subunits associated with two catalytic units (13). Activation of PKA occurs when two molecules of cAMP bind to each of the regulatory subunits, causing a conformational change that activates the associated catalytic units. The ability of PKA to sense intracellular gradients of cAMP is conferred by its tethering to specific sites within the cell. This is achieved through a family of scaffold proteins called AKAPs (A kinase anchoring proteins), whose common property is an ability to interact with the dimerization surface provided by the interacting R subunits of PKA (14, 15). AKAPs also sequester PKA substrates and are targeted to specific intracellular sites, thereby providing a molecular scaffold that allows PKA to detect cAMP gradients and produce a spatially confined response. Indeed, certain AKAPs can sequester specific PDE4 isoforms (2), which allows these enzymes to precisely and selectively control the activity of a localized pool of PKA and its associated substrates in a defined module, which could be described as an AKAPosome. Thus, spatial determinants underpinning cAMP signaling pivotally involve the targeting of PDE isoforms to specific intracellular sites or complexes, as clearly shown for PDE4 (1, 10), constraining specific types of G protein–coupled receptors and adenylyl cyclase isoforms to subdomains of the cell-surface plasma membrane (16) and providing spatially constrained cAMP detection systems involving PKA and EPACs (2, 12, 15).

PKA R subunits are found as two main classes, RI and RII. It is the RII isoforms that predominantly interact with AKAPs and, unlike RI isoforms, are almost completely sequestered to subcellular structures. However, there are isoforms of each of the RI, RII, and C (catalytic) subunits, the functional significance of which is as yet not fully appreciated.

PKA Is Conformationally Constrained

Like many protein kinases, the unstimulated PKA molecule itself is held in an inactive state (13). The PKA C unit is conformationally constrained through its interaction with the R subunit. This is apparent because recombinant PKA C subunit expressed alone is fully active. Furthermore, dissociation of the enzyme caused by incubation in vitro at high ionic strength produces active C subunit freed from the inhibitory R subunit. The normal process of PKA activation ensues as a consequence of two molecules of cAMP binding to each R subunit, thereby negating the ability of the R subunit to constrain the activity of the C subunit.

PKA Is Promiscuous

The original idea that the sole role of PKA R subunits was to constrain the activity of PKA C subunits was challenged with the discovery of AKAPs, which interact with R subunits and whose prime function is to target PKA subpopulations to distinct intracellular sites. However, there are growing indications that other proteins can interact with both PKA R and C subunits, with distinct functional consequences.

One of the first indicators of this was the proposal that IκB could replace the PKA R subunit in interacting with and inhibiting the PKA C subunit (17). The function of this complex was envisaged as allowing for the cAMP-independent phosphorylation and activation of nuclear factor κB (NF-κB) after the degradation of IκB through the proteosome pathway. Intriguingly, the small GTPase regulatory protein Rab13 is able to interact directly with, and thereby inhibit, the PKA C subunit when in its GTP-bound, active state (18). Through this mechanism, activated Rab13 serves to inhibit PKA-dependent phosphorylation of VASP, causing alterations in actin and tight junction arrangement. The PKA C subunit also binds to and phosphorylates p73, which is a structural and functional homolog of the tumor suppressor p53, thereby inhibiting the pro-apoptotic function of this interacting protein (19). However, it remains to be seen whether this interaction is influenced by either cAMP or R subunits in intact cells. Of course, the PKA C subunit also binds PKI, a small heat-stable protein that acts to inhibit PKA activity by presenting a pseudosubstrate inhibitor peptide to the active site, where it binds (13).

AKAPs are defined as a diverse family of proteins with no obvious sequence identity that sequester PKA R subunits by binding to the R-subunit dimerization interface (2, 14, 15). A bioinformatics approach has allowed for the identification of such AKAPs and demonstrated that RII-binding domains form structurally conserved amphipathic helices with unrelated sequences (20). AKAPs can be identified diagnostically by displacement from R subunits, using competing peptides that bind to the R-subunit dimerization interface. However, it now appears likely that R subunits can bind functionally to other proteins, although the AKAP-peptide displacement test has yet to be applied in many instances. Thus, the PKA R subunit has been shown to bind to and inhibit phosphorylase phosphatases (21); cAMP can allow RIa to bind to cytochrome c oxidase subunit Vb, thereby inhibiting cytochrome oxidase activity (22); PKA RI, but not RII, interacts with PAP7, a peripheral-type benzodiazepine receptor that is involved in cholesterol transport and so facilitates human chorionic gonadotropin–stimulated steroid generation in Leydig cells (22); and PKA RI subunits can bind to RFC40, the second subunit of the Replication Factor C complex, and influence the survival of certain cell types (23). The PKA RI subunit can also be recruited to the activated EGF receptor through an ability to bind Grb2 (24). This interaction involves the Src homology 3 (SH3) domain of Grb2, which is likely to interact with the extended and highly disordered linker region in the R subunit that has a putative SH3-binding motif (13).

RSK(y) Attachments

Ribosomal S6 kinases (RSKs) are a family of serine-threonine kinases (also known as p90rsk and MAPKAP-K1) that lie downstream of the Ras/ERK mitogen-activated protein (MAP) kinase pathway (25). These proteins contain two kinase domains where the C-terminal kinase activity, which is activated upon ERK phosphorylation, effects autophosphorylation so as to activate the N-terminal kinase, which can then phosphorylate substrate proteins. RSK1 can phosphorylate a number of the same substrates as PKA, such as CREB, BAD, ER81, GSK3, LKB1, and Nur77, implying a relationship between these two proteins. The recent intriguing work of Chaturvedi and colleagues (3) shows, however, just how intimate this relationship is by providing, for the first time, evidence of a protein, namely RSK1, that is able to interact with both subunits of PKA (Fig. 1). Although both the structural basis of such interactions and the specificity for RSK1 versus other RSK isoforms needs resolution, the functional consequences for cross-talk between the PKA and ERK pathways are intriguing indeed.

Fig. 1.

Interaction of RSK1 with PKA.

The authors demonstrated that inactive RSK1 bound the PKA RI subunit in either the presence or absence of an AKAP. Such binding of RSK1 appeared to attenuate the interaction between the RI and C subunits, which may sensitize PKA to cAMP or even elicit the cAMP-independent activation of PKA in certain situations. Factors influencing this potential interaction will be the availability of free, inactive RSK1 that is not sequestered to competing scaffolds, and the presence of any other proteins able to compete with RSK1 for binding RI subunits. However, it appears that ERK phosphorylation of RSK1 (pRSK1), which triggers autophosphorylation and activation, also switches RSK1 from binding PKA RI subunits in their inactive state to binding PKA C subunits in their active state. This engenders diametrically opposed functional consequences, enhancing RI-subunit interaction with C subunits, thereby decreasing the sensitivity of PKA in complex with pRSK1.

The ERK MAP kinase and PKA pathways interact with each other at various points, and undoubtedly such cross-talk (Fig. 2) provides important elements of key processes, such as learning and memory, inflammation, and cell growth and survival (26, 27). Thus, depending on the expression of Raf isoforms, the elevation of cAMP levels can provide either an inhibitory input (c-Raf) or a stimulatory input (B-Raf) to ERK activation. Cross-talk is also seen in cells expressing hematopoietic protein tyrosine phosphatase (HePTP), which dephosphorylates and thereby deactivates pERK (Fig. 2). However, to achieve this, HePTP needs to dock onto ERK through its KIM domain, which can be functionally inactivated through PKA phosphorylation, thereby increasing levels of activated pERK (28). Although RSK1 also needs a KIM domain to dock onto ERK in order to be phosphorylated and activated, unlike HePTP, the KIM domain of RSK1 does not contain a PKA phosphorylation site that would allow PKA to negate ERK binding and, thereby, the phosphorylation and activation of RSK1. Instead, abutting the KIM domain of RSK1 is an autophosphorylation site for the RSK1 N-terminal kinase whose function is to ablate ERK binding to RSK1 (29). Thus, ERK first docks to RSK1, thereby phosphorylating it and activating it, whereupon autophosphorylation of the ERK docking site on RSK1 triggers the displacement of bound ERK (Fig. 1). Thus, activated RSK1 will not deliver ERK to any complex with the PKA C subunit.

Fig. 2.

Cross-talk between the ERK and cAMP signaling pathways

A further point of cross-talk between these pathways (Fig. 2) is supplied by PDE4 cAMP phosphodiesterases, which also have KIM docking domains (an exposed beta hairpin loop) and FQF specificity motifs (an exposed alpha helix) on their catalytic units that allow for ERK docking and phosphorylation (10, 30). As with the action of RSK1 on PKA and as seen with c-Raf/B-Raf isoforms, switching of signaling is also seen as a result of ERK phosphorylation of PDE4 isoforms. Thus, long PDE4 isoforms are inhibited upon ERK phosphorylation and short isoforms are activated. Furthermore, long PDE4 isoforms are able to bind various AKAPs (31), where they can be regulated by both PKA and ERK phosphorylation, thereby controlling the discrete pool of associated PKA (31, 32). Indeed, ERK phosphorylation and inhibition of PDE4 long forms can cause activation of (localized) PKA, which subsequently also phosphorylates long PDE4 so as to negate the inhibitory effect of phosphorylation by ERK, causing activation. Therefore, the activation of ERK can cause a transient rise in cAMP levels in the locality of long PDE4 isoforms through their initial inhibition followed by activation (33). It appears likely that the incorporation of RSK1 into such a complex would substantially change the kinetics of such a localized cAMP response. ERK-phosphorylated RSK1, in binding to the PKA C subunit, might be expected to desensitize PKA to activation by the increased localized level of cAMP triggered by ERK inhibition of long PDE4. In so doing, pRSK1 would slow down the apparent "rescue" of ERK-inhibited PDE4 through PKA phosphorylation, sustaining PDE4 in an inhibited state and allowing cAMP levels to rise to higher levels before the activation of PKA, and thus of PDE4, could ensue.

Compartmentalization is key. The activation of RSK1 by ERK allows it to translocate to the nucleus, where it can phosphorylate various substrates such as c-Fos, Mit1, Bub1, and histone3. Chaturvedi and colleagues (3) suggest that this process depends on RSK1 association with a PKA/AKAP complex. Thus, the dissociation of PKA/AKAP complexes, which can easily be achieved by using disrupting peptides, allows active RSK1 to accumulate in the cytosol. This increases the level of phosphorylation of cytosolic substrates, such as BAD and TSC-2, one consequence of which is to protect against apoptosis.


This work by Chaturvedi and colleagues (3) is provocative and exciting. It identifies a previously unknown point of cross-talk between the cAMP and ERK signaling pathways and a potential further means of regulating PKA activity and functioning. It is now apparent that key regulatory proteins that are sequestered to PKA signaling complexes, namely PDE4 cAMP phosphodiesterases and RSK1, control the functioning of modules (AKAPosomes) formed from AKAP-tethered PKA. These new discoveries indicate new directions to explore in understanding spatial and temporal regulation of the ERK and PKA signaling pathways and emphasize how crucial it is to understand the molecular details that integrate these pathways in health and disease for therapeutic exploitation.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
View Abstract

Navigate This Article