Introduction

Top Introduction MAPK Signaling Pathways Cyclic AMP Signaling Cross-Talk between the cAMP... Conclusion References

A myriad of signaling networks control the host of processes necessary for the functioning of cells in health and disease. The range of specialized cell types and the requirements for growth and differentiation have caused ubiquitously expressed signal transduction systems to be adapted to meet the specific requirements of particular cells at particular times. The integration of various signaling systems demands that such processes communicate with each other in feedforward and feedback regulatory loops. To allow for this, cellular differentiation specifies an appropriate array of components whose signaling can be integrated in time and space. A common means of achieving this is for key components in one signaling system to be expressed as isoforms that differ in their ability to couple to other regulatory signaling systems. This may occur either by virtue of their being differentially susceptible to modification or because their expression is targeted to a specific intracellular location allowing for compartmentalized signaling.

The mitogen-activated protein kinase (MAPK) pathway seems to be one of the most ancient signal transduction modules conserved in all eukaryotes (Lewis et al., 1998; English et al., 1999; Schaeffer et al., 1999) and of similar versatility and importance as the cyclic AMP (cAMP) signaling system. When MAPK was cloned about a decade ago, the cDNAs revealed high homologies to two yeast kinases, Fus3 and Kss1, both known to function in a pathway that induces cell cycle arrest in response to mating factor. Taking account of this unexpected relationship and the never ending discovery of yet another external signal that activated MAPK, the enzyme was rechristened "extracellular signal-regulated kinase" (ERK). Since then molecular cloning has unveiled an ever growing family of related genes and made the once biochemical entity MAPK the founder of a large family of kinases that share common regulatory motifs, but seem to serve different functions. This is testimony to nature's economy of using a basic kinase module for an amazing variety of tasks, but unfortunately it also generated a bewildering nomenclature. Although MAPK is often used interchangeably to designate either the whole family or the ERK branch, in this review we will use MAPK to refer to the whole family and use the specific names, such as ERK, for the subgroups.

Cyclic AMP has served as a paradigm for an intracellular second messenger (Houslay and Milligan, 1997). It is involved in mediating the action of a host of processes in specialized cells, ranging from control of various metabolic events, muscle contraction, secretion, and memory, for example. In most cells cAMP serves to inhibit cell growth. However, to confound this, in certain cell types, such as those from the thyroid and pituitary, it can actually stimulate cell growth. Over the years it has proved extremely frustrating and problematic to try and resolve the role of cAMP in regulating cell growth and, perhaps crucially, the pathway controlling ERK activation. Undoubtedly, the interaction between these two signaling systems is of pivotal importance. However, the very complexity of these signaling systems means that it is going to be extremely challenging to resolve the connections that serve to integrate cellular responses through these pathways.

In this review we propose that specific biological effects are achieved on a combinatorial basis. Depending on cell- and situation-specific expression patterns and activity profiles, a small set of basic signaling modules can be linked in various ways to achieve a multitude of specific biological responses. The realization of any potential linkage pattern is determined by spatial and temporal components, like a game of chess where an infinite variety of combinations is spawned by the temporal and spatial positioning of an otherwise identical set of pieces. Thus, there are at least five layers of regulation: i) the expression of regulators, modulators, and targets; ii) the subcellular localization or compartmentalization of these molecules; iii) their activation status; iv) their mutual interactions; and v) the order and timing of changes in parameters i through iv. In this review we will try to illustrate this concept of dynamic signaling networks using the multiple layers of cross-talk between the cAMP and ERK signaling systems as a paradigm.

	MAPK Signaling Pathways

Top Introduction MAPK Signaling Pathways Cyclic AMP Signaling Cross-Talk between the cAMP... Conclusion References

The core module of MAPK pathways (Lewis et al., 1998; English et al., 1999; Schaeffer et al., 1999) consists of three kinases, MAPK, a serine/threonine-specific kinase, which is activated via phosphorylation by MAPK kinase (MAPKK). MAPKK in turn is activated by a MAPKK kinase (MAPKKK). Typically, MAPKKKs integrate the input signals, and hence often exhibit complex regulation, which usually involves interaction with small GTPases of the Ras or Rho families. MAPKKs appear to be highly specific for their respective MAPKs, and phosphorylate only a restricted subset of MAPKs. In contrast, MAPKs have multiple substrates and constitute the effector end of the kinase cascade. The phosphorylation steps between the kinases are facilitated by physical interactions between the components (English et al., 1999; Schaeffer et al., 1999). They not only serve to amplify the input signals, but also provide regulatory interfaces for tuning the activity (Yeung et al., 1999). In this review we will focus on the ERK pathway, which features Raf family kinases as MAPKKK, MEK-1 and -2 (MAPK/ERK kinase) as MAPKK, and ERK-1, -2 (often designated p44 and p42 MAPK) as MAPK.

The Raf kinase family comprises three isoforms, Raf-1, the cellular homolog of the v-raf oncogene that is ubiquitously expressed, as well as A-raf and B-raf, which show a more restricted expression pattern. Raf-1 is activated in response to a multitude of growth factors and tumor-promoting phorbol esters. A crucial step in Raf-1 activation is the binding to activated, GTP-loaded Ras, which recruits Raf-1 from the cytosol to the plasma membrane where activation takes place. The exact mechanism of activation is still incompletely understood, but involves changes in phosphorylation, the rearrangement of protein associations and binding to lipids. Although both A-raf and B-raf can bind to Ras/GTP and are responsive to activation by Ras, important differences in the activation mechanisms exist. Raf-1 and A-raf require secondary signals for full activation in addition to Ras, whereas B-raf is fully activated by Ras alone. All three Raf isoforms can activate MEK, but the different phenotypes of Raf-1, A-raf and B-raf knock-out mice suggests that the function of Raf family kinases is not redundant.

At present at least six MAPK subgroups can be distinguished in mammalian cells (Fig. 1). The classical ERK pathway is mainly responsive to mitogens and governs such fundamental processes as cell proliferation, neoplastic transformation, differentiation, and survival. The physiological role of the stress-activated JNK and p38 pathways is more obscure, but they seem to be more involved in the control of cell cycle checkpoints, apoptosis, and differentiation (Schaeffer et al., 1999).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Schematic for MAP kinases. This shows the various mammalian mitogen activated protein (MAP) kinase pathways that have been elucidated. Shown on the left is the generalized scheme of a stimulus serving to activate a mini-guanine nucleotide binding protein (G-protein) by allowing GTP to replace GDP in its active site. The GTP-bound G-protein then activates a kinase cascade comprising a MAPKKK, a MAPKK, and, finally, MAPK itself. The components of the growth factor-stimulated ras pathway to ERK activation are shown together with the two cytokine/stress-activated pathways.

Although the diverse MAPK pathways respond to different external cues and exert distinct biological functions, there seems to be considerable potential for an extensive overlap not only with regard to the activating stimuli, but also in regard to downstream targets. Although it is largely unknown how the specificity and fidelity of distinct biological responses is secured, recent proposals indicate that MAPK substrates contain docking sites that serve to define fidelity of interaction by enhancing the binding between specific MAPK and their substrates. Thus, while both ERK and JNK prefer PX(S/T)P as a consensus phosphorylation motif the presence of such a motif is insufficient in itself to define a protein as genuine substrate in vivo. Rather, it has been suggested that authentic substrates contain additional "docking" motifs that enhance interaction of potential substrate proteins with either or both ERK and JNK (Kallunki et al., 1996). Thus ERK substrates appear to be modular systems containing the target phosphorylation site and either one or two docking sites (Gavin and Nebreda, 1996; Yang et al., 1998; Jacobs et al., 1999; Saxena et al., 1999; Smith et al., 1999; Zuniga et al., 1999). Two such docking sites have been identified (Fig. 2a). One is provided by a FxF motif that is located some 5 to 30 amino acids carboxyl-terminal to the ERK phosphorylation site and where the intervening residue (x) is most often Q. In the case of the transcription factor LIM-1 an adjacent proline residue, as in the motif FQFP, also appears to be important. However, this proline is clearly not needed in all instances. The other docking site is the so-called kinase interaction motif (KIM) that is located approximately 120 to 150 amino acids N-terminal to the target for ERK phosphorylation. At the core of the KIM region is a LxL motif located some 3 to 6 amino acids C-terminal to a pair of basic residues (R/K, R/K). These are usually, but not in all instances, located two residues C-terminal to an aliphatic residue (L/V). Interestingly, some substrates, such as the transcription factor Elk-1 contain both such sites while others, such as the STEP and the PTP tyrosyl phosphatases, contain only the KIM site.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Docking motifs for ERK. a, this shows the consensus motif for the KIM and FQF docking sites and their spatial organization relative to the target site for phosphorylation. b, these show a range of proteins having KIM docking sites that either do or do not contain a consensus PKA phosphorylation site associated with the core basic residues found in their KIM docking site. The sequence associated with this core region is shown.

	Cyclic AMP Signaling

Top Introduction MAPK Signaling Pathways Cyclic AMP Signaling Cross-Talk between the cAMP... Conclusion References

cAMP provided the paradigm for the second messenger concept and spawned the field of intracellular signaling. Intriguingly, however, there remains a dearth of knowledge concerning the role of the many isoforms of components related to its generation, detection, and degradation (Houslay and Milligan, 1997).

Why, for example, are there at least 10 forms of adenylyl cyclase able to generate cAMP? These are not evidence of redundancy, but relate to requirements to define regulation and intracellular localization (Sunahara et al., 1996). Thus, while all these isoenzymes can be activated by the GTP-bound -subunit of G_s, they are differentially sensitive to regulation by Ca²⁺, by phosphorylation and by G-protein - and inhibitory, -subunits (Smit and Iyengar, 1998). And the AC-3 isoform is apparently restricted to specialized regions of the cell surface plasma membrane in olfactory neurons (Houslay et al., 1998; Schwencke et al., 1999).

The complexity of the cAMP synthetic machinery is, surprisingly, outdone by that which degrades cAMP (Conti and Jin, 1999; Soderling and Beavo, 2000). For, there are at least 15 genes encoding over 30 different phosphodiesterase (PDE) isoforms, each of which is able to hydrolyze cAMP! The importance of this diversity is clearly evident from the actions of selective inhibitors that have been and are being developed for a variety of therapeutic uses. Nascent analyses of this multitude of isoenzymes shows that they are poised to play a pivotal role in integrating cAMP signaling to that of other signaling systems. In addition, PDE enzymes can be found both as soluble cytosolic species and also targeted to interact with particular subcellular membranes and other proteins (Houslay and Milligan, 1997; Houslay et al., 1998), indicating that they play a fundamental role in defining compartmentalized signaling reactions.

Protein kinase A (PKA) is an effector system that responds functionally to changes in intracellular cAMP. This is a heterotetrameric protein that is, again, found as isoforms. Of particular note are the RI and RII forms of the regulatory (R) subunits. While the RI isoform is found in the cell cytosol, the RII isoform is predominately localized at discrete intracellular sites. This anchoring is achieved by the dimerization interface of the RII subunit binding to specific proteins called AKAPs (Colledge and Scott, 1999). Indeed, there is a large family of such AKAPs and these are, seemingly, expressed in a cell-type specific fashion. Thus the anchored RII form of PKA is able to sample gradients of cAMP that have been established through the action of either or both anchored PDE isoforms and adenylyl cyclase isoforms localized to discrete subdomains of the cell surface plasma membrane. This provides the cellular machinery that underpins compartmentalized cAMP signaling for which the notion was originally established convincingly in studies done on cardiac myocytes (Brunton et al., 1981).

It is now apparent, however, that cells have additional cAMP detection systems. Cyclic nucleotide-gated ion channels provide one of these. Such channels are found at specific intracellular sites and thus, like RII-PKA are able to sample intracellular gradients of cAMP. The other is a family of recently discovered cAMP-stimulated GTPase exchange proteins, called EPACs or cAMP-GEFs that, seemingly, serve to activate the small G-proteins Rap1a and Rap1b (de Rooij et al., 1998; Kawasaki et al., 1998). The functional significance of these GEFs remains to be fully established in terms of their intracellular localization and the processes that they regulate.

	Cross-Talk between the cAMP Second Messenger System and the ERK Pathway

Top Introduction MAPK Signaling Pathways Cyclic AMP Signaling Cross-Talk between the cAMP... Conclusion References

When the charting of the ERK pathway was successfully accomplished some 8 years ago, it was one of the first pathways that laid open a seamless connection between extracellular growth factors and nuclear transcription factors, and in addition revealed a plethora of novel regulatory motifs.

Inhibition of the Raf-1 Kinase by PKA. A seminal discovery was that the elevation of cAMP in fibroblasts and vascular smooth muscle cells induced a profound inhibition of ERK activation by growth factors (Burgering et al., 1993; Cook and McCormick, 1993; Graves et al., 1993; Sevetson et al., 1993; Wu et al., 1993; Hafner et al., 1994; Hordijk et al., 1994; Russel et al., 1994). This inhibition was mapped to occur at a point downstream of Ras and upstream of Raf-1 and to require the action of PKA. The target identified was Raf (Fig. 3).

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Potential sites of cross-talk between the ERK and cAMP signaling systems: I. This highlights the range of connections where cAMP exerts controlling effects on the ERK signaling pathway. It is crucial to note that these connections will be formulated in a cell-type specific fashion. Thus different cell types will express different complements of, for example, PDE4 isoforms, Raf isoforms, and ERK-regulating protein tyrosyl phosphatases (PTP).

Complex Interactions between B-raf and the cAMP Signaling System. Cyclic AMP, however, seemingly promotes ERK activity in several cell types, including 3T3-F442A preadipocytes, ovarian granulosa cells, melanoma, pituitary cells, and neuronal cells, where ERK is intimately involved in cAMP-regulated processes such as long-term potentiation and circadian gene expression.

Cyclic AMP and the Control of ERK Deactivation. KIM motifs appear to play a pivotal role in defining the ability of ERKs to interact with other proteins. Intriguingly, certain of these motifs (Fig. 2b) contain a putative site (RRxS/T) for phosphorylation by PKA. This offers the potential for PKA to modify the KIM docking site for ERK substrates by introducing a negative charge near to the pair of positive charges shown to be essential for ERK docking. Very recently this has been examined with regard to the PTP and PTP-SL tyrosyl phosphatases (Blanco-Aparicio et al., 1999; Saxena et al., 1999). These two enzymes show a highly restricted, cell-type specific distribution and both serve as negative regulators of ERK. Indeed PTP is a cytosolic tyrosyl phosphatase that specifically regulates ERK and not JNK. Such specificity is seemingly conferred because ERK can bind to these tyrosyl phosphatases through their KIM motifs. This allows these phosphatases to dephosphorylate the critical tyrosine residue that MEK phosphorylates in the activation loop of ERK, thus causing ERK deactivation. Such an interaction, and the ensuing dephosphorylation of ERK, prevents the nuclear translocation of ERK-2 (Blanco-Aparicio et al., 1999) and thus the delivery of activated ERK to the site where it can affect transcriptional factors.

Regulation of cAMP Levels Directed by ERK-2 Docking and Phosphorylation of PDE4 cAMP Phosphodiesterases. Very recently it has been shown that members of the multigene PDE4 cAMP-specific phosphodiesterases can serve as ERK-2 substrates and thus provide a direct means whereby ERK-2 activation can lead to changes in intracellular cAMP levels (Hoffmann et al., 1999; MacKenzie et al., 2000). PDE4 enzymes are closely related to the drosophila dunc PDE whose disruption causes memory and learning defects (Houslay et al., 1998). They are characterized by their ability to be inhibited by the compound rolipram, which serves as the paradigm for a PDE4 selective inhibitor. There is currently great interest in these enzymes because PDE4-selective inhibitors are currently being developed for treating inflammatory disease, in particular chronic obstructive pulmonary disease and asthma, and also have potential as antidepressant and antileukemic agents (Houslay et al., 1998; Rogers and Giembycz, 1998; Torphy et al., 1999).

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Regulation of PDE4 cAMP-specific phosphodiesterases. This schematically shows the long PDE4 isoenzymes that have both UCR1 and UCR2 domains and the short isoenzymes that lack UCR1. Studies on PDE4D gene products have shown that, in the long PDE4D3 and PDE4D5 isoenzymes, UCR1 and UCR2 form a regulatory module that orchestrates inhibition of PDE4 activity consequent to ERK phosphorylation of the enzyme catalytic unit. UCR1 serves to direct the regulatory effect of PKA phosphorylation by uncoupling the interaction between UCR1 and UCR2, causing enzyme activation and ablation of any ERK-mediated inhibition. In the short PDE4D1 isoenzyme, which has a lone, intact UCR2 region, this directs enzyme activation to occur consequent upon ERK phosphorylation. However, in the short PDE4D2 isoenzyme, which has a lone, N-terminally truncated UCR2 region, this directs enzyme inhibition consequent upon ERK phosphorylation. However, unlike with the long isoenzymes, ERK-mediated inhibition of PDE4D2 cannot be ablated by PKA as there is no UCR1 with its PKA phosphorylation motif.

ERK Activation Can Elevate cAMP Levels in Smooth Muscle Cells through an Autocrine Loop. Challenge of certain smooth muscle cells with growth factors such as platelet-derived growth factor and bradykinin has been shown to elicit ERK activation and an increase in cAMP levels without growth stimulation (Pyne et al., 1997; Bornfeldt and Krebs, 1999). The prime reason for the increased accumulation of cAMP in these cells is, seemingly, not driven primarily by any direct effect of ERK-2 on either PDE activity or the tyrosyl phosphorylation of G_s. Rather it appears to be an indirect autocrine action that is elicited via the activation of cytosolic phospholipase A₂ (PLA₂). This is caused through its phosphorylation elicited most probably by the action of ERK (Lin et al., 1993) although, as seen in thyroid cells, possibly also via PKC action (Ekokoski et al., 2000). Such activation of PLA₂ generates arachidonic acid, a precursor for prostaglandin synthesis through the cyclooxygenase (COX) pathway. Thus the generation and export of prostaglandin E₂ (PGE₂) causes, in an autocrine fashion, the receptor-stimulated activation of adenylyl cyclase activation and increased cAMP levels (Fig. 5). It is this increase in cAMP that leads to negative regulation of cell growth as treatment with the cyclooxygenase inhibitor, indomethacin, releases cells from this inhibition and allows growth factors to cause cell proliferation. This feedback system appears to be physiologically relevant for some cell types as evidenced by the observation that the ability of ERK to drive the proliferation of smooth muscle cells is subverted by the expression of COX-2. In COX-2-expressing cells, ERK leads to growth inhibition caused by PGE₂-mediated cAMP accumulation and activation of PKA (Bornfeldt and Krebs, 1999).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Potential sites of cross-talk between the ERK and cAMP signaling systems: II. This shows two further scenarios that integrate PKA and ERK signaling. The schematic on the left side of this figure highlights the proposed ability of Gs-coupled receptors, such as the 2-adrenoceptor, to switch to coupling to Gi and thereby causing ERK activation. Increased cAMP levels, causing PKA activation and phosphorylation of the 2-adrenoceptor, operate this switch. Shown on the right side of this schematic is the proposal that, in smooth muscle cells, ERK can cause PLA2 activation to generate an autocrine effect that leads to an increase in cAMP levels by PGE2 stimulation of adenylyl cyclase.

7-Transmembrane (7-TM) G-Protein-Linked Receptors That Mediate ERK Activation: Constitutive and PKA-Switched Controls. The provocative proposal has been made (Daaka et al., 1997; Maudsley et al., 2000) that the 7-TM receptors that normally serve to activate adenylyl cyclase can, under certain circumstances, be switched to elicit ERK activation (Fig. 4). It has long been established that the ₂-adrenergic receptor (₂AR), upon agonist occupancy, binds to the heterotrimeric G-protein, G_s. This causes the G_s -subunit to bind GTP and then dissociate from the complex to stimulate adenylyl cyclase. The subsequent hydrolysis of GTP to GDP provides a turn-off reaction, allowing reassociation of the GDP-bound G_s -subunit with the pair. The ₂AR is also subject to rapid desensitization due to the dual action of both receptor-specific kinases (G-protein receptor kinases) and PKA that phosphorylate the C-terminal tail of the receptor and prevent productive coupling to G_s. Long-term desensitization results from the internalization of the receptor through coated pits. However, it has been suggested that the PKA phosphorylated form of the ₂AR is not only unable to couple productively to G_s but switches its allegiance to allow coupling to the G_i family of proteins, some of whom serve to inhibit adenylyl cyclase (Daaka et al., 1997). G_i proteins are usually found in excess of G_s and their activation releases not only GTP-bound -G_i but also a large pool of -subunits. It is the release of such -subunits that is believed to lead to ERK activation (Fig. 4). The details of this process remain to be unequivocally defined. A possibility is a direct stimulatory effect on SRC kinase that allows it to couple with SOS, triggering Ras, and hence ERK, activation.

Tyrosyl Kinase Receptor-Stimulated Adenylyl Cyclase. Occupancy of the EGF receptor tyrosyl kinase leads to ERK activation. However, it has been suggested that in a highly restricted range of cell types it can also lead to adenylyl cyclase activation (Sun et al., 1997). This is believed to occur through the direct tyrosyl phosphorylation of the -subunit of G_s. However, the site of tyrosyl phosphorylation and the mechanism that underpins the proposed activation of G_s by this effect is unclear, as is why this has only been seen in a few instances (example and/or reference). It is certainly possible to discriminate between EGF effects on adenylyl cyclase and on PDE activity by either including (selective) PDE inhibitors in cell-based assays or by isolating membranes for adenylyl cyclase-based assays.

CREB: A Target for Both cAMP and ERK Signaling Pathways. CREB is a 43-kDa transcription factor that binds to the conserved cAMP response element (CRE) (Shaywitz and Greenberg, 1999). This consists of an 8-base pair palindromic sequence (TGACGTCA). There are two additional CREB family members, namely CREM and ATF-1. All contain a single consensus phosphorylation site for PKA and all bind to DNA as dimers through a basic carboxyl terminal leucine zipper (bZIP) motif. The single PKA site in CREB is found at Ser-133. To be active CREB requires the phosphorylation of serine 133 enabling it to recruit p300/CBP, CREB binding protein, an essential cofactor that interacts with the basic transcription machinery (Shaywitz and Greenberg, 1999). Current evidence suggests that PKA activation in the extranuclear region of the cell leads to the translocation of the free PKA catalytic unit to the nucleus, whereupon it can phosphorylate CREB family members. However, there is also evidence indicative of the presence of PKA binding proteins (AKAPs) and PKA regulatory (RII) subunits associated with the nuclear membrane and even within the nucleus itself (Colledge and Scott, 1999). This implies that compartmentalized effects involving both extra- and intranuclear changes in cAMP concentrations may regulate CREB phosphorylation.

	Conclusion

Top Introduction MAPK Signaling Pathways Cyclic AMP Signaling Cross-Talk between the cAMP... Conclusion References

No signaling pathway functions in isolation. However, we are only at the start of understanding the range of possible connections that allow integration between them. The cAMP and ERK pathways are ubiquitous and pivotal to cellular growth, development, and function. We have tried in this review to highlight some new and exciting processes that provide connections between these pathways. In doing so it is quite clear that the modular design of proteins and their expression patterns are pivotal to the wiring systems that are unique to each cell-type. Thus, in analyzing systems it is crucial to take account of the set of players involved, their isoforms and intracellular location. In trying to reconcile studies performed in different cell types it is thus crucial to appreciate the identify of the isoforms present and, in addition, whether the intracellular organization of components of specific signaling systems generates compartmentalized responses. In addition, as these can be coupled to feedback and feedforward loops then analyses must be done over appropriate temporal windows. Thus changes in individual components of signaling systems are transient and geared to trigger a defined cascade of responses over time as they pass over set activation thresholds. An understanding of the cross-talk processes that interlink signaling systems in defined cell types is crucial to our understanding of cell function in health and disease and thus in developing new diagnostic methodologies and new therapeutics.