Elsevier

Cellular Signalling

Volume 9, Issue 5, August 1997, Pages 337-351
Cellular Signalling

Review
Regulation of the ERK Subgroup of MAP Kinase Cascades Through G Protein-Coupled Receptors

https://doi.org/10.1016/S0898-6568(96)00191-XGet rights and content

Abstract

The extracellularly-responsive kinase (ERK) subfamily of mitogen-activated protein kinases (MAPKs) has been implicated in the regulation of cell growth and differentiation. Activation of ERKs involves a two-step protein kinase cascade lying upstream from ERK, in which the Raf family are the MAPK kinase kinases and the MEK1/MEK2 isoforms are the MAPK kinases. The linear sequence of Raf → MEK → ERK constitutes the ERK cascade. Although the ERK cascade is activated through growth factor-regulated receptor protein tyrosine kinases, they are also modulated through G protein-coupled receptors (GPCRs). All four G protein subfamilies (Gq/11, Gi/o, Gs and G12/13) influence the activation state of ERKs. In this review, we describe the ERK cascade and characteristics of its activation through GPCRs. We also discuss the identity of the intervening steps that may couple agonist binding at GPCRs to activation of the ERK cascade.

Introduction

Mitogen-activated protein kinases (MAPKs) are Ser-/Thr- kinases that have been strongly conserved through evolution emphasising their importance in intracellular signalling (for reviews, see 1, 2, 3, 4, 5, 6, 7, 8, 9). There are at least three MAPK subfamilies present in mammalian cells. These are: (i) the extracellularly-responsive (or extracellular signal-regulated) kinases (ERKs) which were initially known simply as MAPKs; the c-Jun N-terminal kinases (JNKs) or stress-activated protein kinases (SAPKs); and (iii) the p38-MAPKs. The ERK cascade is intimately connected with the regulation of cell growth and differentiation. The JNKs/SAPKs and p38-MAPKs are, in contrast, involved in the responses of cells to environmental stresses.

ERKs were originally identified as protein kinases which phosphorylated microtubule-associated protein 2 in response to activation of receptor protein tyrosine kinases (RPTKs) by insulin and other polypeptide growth factors. The ERK cascade (as for all MAPK cascades) consists of a linear sequence of three protein kinases, namely the Raf subfamily of MAPK kinase kinases (MAPKKKs), the MEK1/MEK2 subfamily of MAPK kinases (MAPKKs) and the ERK subfamily of MAPKs (Fig. 1). Signals converge at or above the level of Raf and information is transmitted through the amplifying ERK cascade by phosphorylation and activation of the MEKs and ERKs. Thereafter, signalling pathways diverge, with the ERKs phosphorylating a wide range of protein substrates. The involvement of the ERK cascade in growth and differentiation has been clearly demonstrated by experiments which showed that activation of ERK by transfection of constitutively-activated MEK mutants is sufficient to induce differentiation or transformation of cells 10, 11, 12.

It has become increasingly apparent that, like RPTKs, GPCRs and G proteins are involved in the regulation of cell growth and differentiation (reviewed in Ref. [13]). Many of these effects are mediated by the ERK cascade. In this review, our aim has been to summarize the current state of knowledge concerning the activation of the ERK cascade by G protein-coupled receptor (GPCR) agonists. We discuss how the stimulus initiated by agonist binding to GPCRs may be transmitted to the ERK cascade. Although the JNK/SAPK and p38-MAPK cascades can also be activated through GPCRs, we have omitted any discussion on these cascades.

Raf (reviewed in Refs. 14, 15) was originally recognized as a viral oncogene (v-raf), of which the 74 kDa c-Raf (also known as Raf-1) is the cellular protooncoprotein counterpart. Two further mammalian homologues (69 kDa A-Raf and 95 kDa B-Raf) were subsequently cloned. The biological reasons underlying Raf diversity are not understood but may be related to preferential activation of Raf isoforms by agonists 16, 17or preferential activation of MEK isoforms by individual Raf isoforms 18, 19, 20. Although activation of the ERK cascade is responsible for many of the consequences of Raf stimulation, Raf may have additional actions that are independent of ERK [21]. Equally, Raf may not be the only MAPKKK for the ERK cascade 22, 23, 24, 25, 26. It should be noted that the MAPKKK MEK kinase 1 (MEKK1), which was initially suggested to couple GPCRs to the ERK cascade [22], is now thought to be primarily responsible for the activation of the JNK/SAPK and p38-MAPK cascades 27, 28, 29, 30. However, the recently-cloned MEKK1 homologues MEKK2 and MEKK3 may activate both the ERK and JNK/SAPK cascades [26].

All mammalian Raf isoforms show considerable homology (reviewed in Ref. [14]). They contain N-terminal regulatory and C-terminal catalytic domains. Deletion of or mutation in the N-terminal domain can render Raf oncogenic. The N-terminal regulatory domain contains the following. First, one or more [31]regions which interact with the active (GTP-ligated) form of the farnesylated [32]membrane-bound small G protein p21 Ras are present (for reviews on Ras, see 33, 34, 35, 36, 37). Secondly, there is a cysteine-rich Zn2+ finger-like region. This may be involved in phosphatidylserine binding [38]and is necessary for activation [39]. Thirdly, it includes a Ser/Thr-rich region which may contain sites for phosphorylation [40]. The conserved C-terminal catalytic domain encompasses the protein kinase active site and the ATP- and protein substrate-binding domains. A site for binding of phosphatidic acid has recently been identified in this region of c-Raf. As we discuss below, this may have important implications for its regulation [41].

One important pathway of activation of c-Raf (reviewed in Refs. 9, 14, 15, 34, 42, 43) involves its interaction with Ras. GTP which is located on the inner face of the plasma membrane (Fig. 2). In its basal state, Ras exists as a GDP-ligated species (Ras. GDP). It is activated by the exchange of GDP for GTP in a process stimulated by guanine nucleotide exchange factors (GEFs) such as Sos (see Fig. 3 and described in more detail below). The exchange of guanine nucleotides increases the affinity of Ras for c-Raf (which is cytosolic in unstimulated cells) which translocates c-Raf to the membrane 44, 45, 46, 47, 48. This interaction may be reinforced by the binding of the membrane phospholipid phosphatidylserine to the c-Raf Cys-rich region [38]. It is not entirely clear whether activation of Ras necessarily translocates the other Raf isoforms [49]. Binding of Ras.GTP to c-Raf is not sufficient to activate it fully (at least in vitro). In the membrane, further events lead to full activation of c-Raf. These may include the phosphorylation of Tyr-340 and Try-341 39, 50by Src family non-receptor protein tyrosine kinases (PTKs) 48, 49with which c-Raf interacts [51]. Dephosphorylation and inactivation of c-Raf may involve GPCR-regulated protein tyrosine phosphatases [52]. Activating phosphorylation of analogous Try-residues (Tyr-301 and Tyr-302) may occur in A-Raf. However, this mechanism of activation is probably not applicable to B-Raf where the pair of Tyr-residues is replaced by an Asp-Asp sequence. This substitution of acidic residues is likely to simulate the tyrosine-phosphorylation in A-Raf or c-Raf, which may explain the high basal activity of B-Raf and the lack of dependence of its activation on tyrosine phosphorylation [49].

In addition to tyrosine phosphorylation sites, there are also sites for Ser-/Thr-phosphorylation which may be activating (40, 53and see below) or inhibitory. The inhibitory phosphorylation involves the phosphorylation of Ser-43 in c-Raf (which lies just N-terminal to the Ras interaction domain(s) in the regulatory region). This is achieved by the catalytic subunit of cyclic AMP-dependent protein kinase (PKA) which thus inhibits Ras. GTP binding to c-Raf and agonist-induced activation of the ERK cascade (54, 59and reviewed in Ref. [60]). This residue is not conserved in A-Raf or B-Raf and, although they are still inhibited to some degree 16, 56, 57, these isoforms may be less sensitive to inhibition by cyclic AMP and the catalytic subunit of PKA (see Ref. [16]for a comparison of A-Raf with c-Raf).

There is considerable interest in the interaction of Raf with 14-3-3 proteins 61, 62. 14-3-3 proteins are ubiquitous and exist as dimeric species in the native state (reviewed in Ref. [63]). They recognize phosphoserine-containing consensus sequences (Arg,-Ser-Xaa-Ser(P)-Xaa-Pro) in other proteins [64]. All mammalian Raf isoforms contain two such sequences which may be constitutively phosphorylated [40]and hence Raf may interact constitutively with 14-3-3 proteins. Though controversial 65, 66, the interaction between Raf and 14-3-3 proteins may assist in or lead to its activation 61, 67, 68, 69, 70, possibly by inhibiting Raf dephosphorylation [53]. Alternatively, as reviewed in Ref. [71], the multivalent interaction of c-Raf with 14-3-3 protein dimers may increase local concentrations of c-Raf promoting activation 72, 73. If this is correct, the function of Ras.GTP binding to c-Raf could equally be to increase its local concentration in the juxtamembrane region [71], rather than placing c-Raf adjacent to membrane PTKs. Equally, concen- tration-dependent activation could conceivably explain some of the Ras-independent pathways of activation of Raf [39].

Established substrates for Raf are MEK1 and MEK2, which contain Pro-rich sequences necessary for the interaction of MEK with Raf [74]. The MEKs are soluble enzymes [75]which are phosphorylated by Raf on two Ser-residues, phosphorylation of both residues being necessary for full activation 76, 77, 78, 79. Given that translocation of Raf to a membrane site is necessary for its activation, it is not clear whether activated Raf returns to the cytoplasm to phosphorylate MEK or whether MEK is phosphorylated at some juxtamembrane site. However, in vitro, c-Raf and A-Raf can clearly phosphorylate and activate MEK in the absence of membranes [16]. An inhibitor of MEK activation, PD98059 80, 81, is commercially available and has become a useful tool for implicating the ERK cascade in biological processes. Like all MAPKKs, MEKs are “dual-function” kinases that catalyze the ordered phosphorylation [82]of first the Tyr- and then the Thr- residue within a Tyr-Xaa-Thr MAPK activation motif (Thr-Glu-Tyr in the case of the ERK1 and ERK2 isoforms). Phosphorylation of both residues is necessary for activation of these ERKs. The Thr-Glu-Tyr sequence lies on the “phosphorylation lip” of ERK1 [83]and contains additional residues that determine the recognition of ERK1 by MEK1 and MEK2 [84]. Because most protein kinases phosphorylate either Ser-/Thr- or Tyr- residues in proteins, this unusual dual-phosphorylation may prevent “non-specific” activation of ERKs which could be potentially oncogenic.

Three ERKs (ERK1, ERK2, and ERK3) have been identified by molecular cloning 85, 86. The 44 kDa ERK1 (also known as p44-MAPK) and the 42 kDa ERK2 (or p42-MAPK) are the only well-established substrates of MEKs. Like the majority of MAPKs (ERK3 being a possible exception), ERK1 and ERK2 are Pro-directed Ser/Thr kinases that preferentially phosphorylate the consensus sequence Pro-Xaa-Ser/Thr-Pro or minimally Ser/Thr-Pro in substrate proteins (reviewed in Ref. [87]). ERK binding domains in the protein substrate and substrate binding domains in ERKs further influence the interactions of ERKs with their substrates. However, because ERKs share a common substrate phosphorylation sequence with JNKs/ SAPKs and the p38-MAPKs, there is potential for “cross-talk” between the cascades. The significance of such cross-talk in a physiological setting is unclear.

Substrates of ERK1 and ERK2 are numerous (see Fig. 1). They include factors and enzymes involved in the phosphorylation-dependent stimulation of transcription (e.g., the Elk-1 88, 89and TAL1 [90]transcription factors, RNA polymerase II [91]). In order to regulate transcriptional processes, ERK1/ERK2 (but not MEKs [92]) are translocated into the nucleus following their activation 6, 92, 93, 94. Mainly-cytoplasmic substrates include translation factors (e.g., PHAS-I/eIF4E-BP, 95, 96, 97, 98and reviewed in Ref. [99]), proteins involved in phosphorylation/dephosphorylation reactions (e.g., p90 ribosomal S6 kinase [100]and protein phosphatase 2C [101]), and other signalling enzymes (e.g., cytoplasmic phospholipase A2 [102]and tyrosine hydroxylase 103, 104, 105). These phosphorylations stimulate the processes involved. ERK1 and ERK2 also phosphorylate structural proteins such as myelin basic protein [106]and microtubule-associated protein 2 [107]. Phosphorylation of these proteins may be involved in the regulation of cell structure. In addition, ERKs phosphorylate the upstream kinases of the ERK cascade (Raf [108]and MEK 109, 110). Other proteins involved in the regulation of the cascade such as Sos (see below) may be phosphorylated in response to activation of the ERK cascade 111, 112, 113, 114, 115. These modifications are likely to be involved in negative feedback regulation 109, 112, 116.

The regulation of the 62 kDa ERK3 is not understood and it is not clear whether it should be classified as an ERK. Although showing about 50% homology with ERK1 and ERK2, the residues corresponding to the Thr-Glu-Tyr phosphorylation motif in ERK1/ERK2 are Ser-Glu-Gly (86, 117). Cheng et al. [118]have reported that ERK3 localizes constitutively to the nucleus and that the enzyme does not phosphorylate standard ERK1/ERK2 substrates, such as myelin basic protein. The Ser-residue is phosphorylated by a MAPKK distinct from MEK1 and MEK2 [117]. In contrast, others have reported that ERK3 phosphorylates myelin basic protein [119]. Its intracellular location in the unstimulated state is partially cytoplasmic and activity increases in cells overexpressing the diacylglycerol- (DG-) regulated protein kinase C, cPKCβ [119]. This implies that ERK3 may be regulated through GPCRs of the Gq/11PCR subfamily (which couple to the DG/inositol 1,4,5-triphosphate (InsP3) signaling pathway, see below), but it is not known whether this is the case.

Recent reviews have described GPCRs, G proteins and their interactions in detail 120, 121, 122. GPCRs constitute a large family of single polypeptide chain receptors, with seven transmembrane α-helices composed predominantly of hydrophobic residues looping between the extracellular and intracellular phases (Fig. 4). GPCRs couple to heterotrimeric G proteins (composed of α, β and the smaller γ subunits) to initiate a variety of intracellular signaling pathways. Most α subunits contain sites for myristoylation and/or reversible palmitoylation (reviewed in Ref. [32]). All γ subunits contain sites for isoprenylation (reviewed in Ref. [32]). These acylations target G proteins to the membrane compartment which is the principal site of G protein signaling. In the inactive G protein heterotrimer, α subunits exist as GDP-ligated forms. Binding of an extracellular agonist to a GPCR induces a conformational change in the receptor and its associated G protein. This decreases the affinity of the α subunit for GDP, and GTP displaces GDP (Fig. 4). The GPCR dissociates from the G protein heterotrimer which also dissociates into α. GTP and βγ dimers (the latter always remaining tightly associated). Both α.GTP and βγ dimers fulfill effector functions. The innate GTPase activity of the α subunit returns it to the GDP-ligated from and α.GDP reassociates with both βγ dimers and the GPCR.

Four G protein subfamilies (Gq/11, Gi/o, Gs and G412/13) have been identified based on a classification of α subunits, 23 of which have so far been recognized (reviewed in Ref. [122]). The β and γ subunits are similarly diverse (at least five β and eleven γ subunits, reviewed in Ref. [123]). As described in detail below, all G protein subfamilies influence the activity of the ERK cascade, and can be either stimulatory (Gq/11, Gi/o, G12/13) or, depending on the circumstances, stimulatory or inhibitory (Gs). Although not all interactions are favoured, the multiplicity of α, β and γ subunits allows formation of many heterotrimeric species which may confer increased specificity or flexibility of signaling.

The G protein families that have been studied in most detail with respect to the regulation of MAPK cascades are Gq/11 and Gi/o. Classically, αq/11 family members stimulate the hydrolysis of the membrane phospholipid phosphatidylinositol-4,5,bisphosphate (PtdInsP2) by activating phosphoinositide-specific phospholipase Cβ (PI-PLCβ) (Fig. 5). Four mammalian isoforms of PI-PLCβ have been characterized (reviewed in Refs. 123, 124, 125, 126) of which PI-PLCβ1, PI-PLCβ3 and possibly PI-PLCβ4 are preferentially activated by αq/11. The products of PtdInsP2 hydrolysis, DG and InsP3, regulate respectively phospholipid-dependent Ser-/Thr- protein kinase PKC and intracellular Ca2+ movements (reviewed in Refs. 127, 128). Only the “classical” and “novel” subfamilies (but not the “atypical” subfamily) of PKC are activated by DG (reviewed in Ref. [128]). The involvement of DG-regulated PKCs in biological processes can be inferred by the use of tumour-promoting phorbol esters such as 12-O-tetradecanoyl phorbol 13-acetate (TPA). These act as long-lived DG analogues and acutely (in < 30 s) activate the DG-regulated PKCs. In contrast, chronic stimulation (usually 24 h) with TPA depletes cells of DG-regulated PKCs. This manipulation has been used extensively to implicate DG-regulated PKCs in response to agonists.

Although dissociation of other G proteins releases βγ dimers, Gi/o is considered to be the primary source of βγ dimers in cells because of its high relative abundance. Indeed, as discussed in detail below, the major input of Gi/o into the ERK cascade involves Gi/o-derived βγ dimers rather than αi/o. βγ Dimers were originally viewed as passive partners in the G protein heterotrimer whose major role was to mediate association of α subunits with the membrane. They are now recognized to have signaling functions in their own right. They interact with “pleckstrin homology” (PH) domains in proteins, a consensus motif first identified in the eponymous cytoskeletal protein and subsequently identified in a wide variety of other proteins (reviewed in Refs. 129, 130, 131). These include phospholipases, protein kinases, cytoskeletal proteins, and proteins involved in stimulating the GTPase activity of G proteins or GTP/GDP exchange (for example, Sos). The α subunit -βγ dimer interaction and the PH domain-βγ dimer interactions are mutually exclusive. The role of PH domains is unclear (reviewed in Ref. [37]), but they interact with PtdInsP2 suggesting that they may be involved in the association of proteins with membranes 132, 133. Furthermore, deletion of the PH domain in the GEF Sos (see below) causes a loss of function [134]. βγ Dimers activate the PI-PLCβ3 isoform and, to a lesser extent, PI-PLCβ2 and thus may activate PtdInsP2 hydrolysis (reviewed in Ref. [123]), but the physiological significance of βγ-stimulated PtdInsP2 hydrolysis is difficult to assess. Involvement of Gi/o in biological processes is best implicated by the sensitivity of responses to pertussis toxin (PTX) pretreatment. PTX catalyses the ADP-ribosylation of αi and αo in Gi/o protein heterotrimers and inhibits Gi/o signaling function. It should be noted that not all Gi/o subfamily members are sensitive to PTX (for example, Gz is insensitive), so that a lack of sensitivity to PTX may not necessarily preclude involvement of the Gi/o subfamily.

αs and α12/13 may also regulate the ERK cascade [135], although the significance of these effects is currently hard to assess. The principal physiological function of αs is to stimulate the membrane-bound adenylyl cyclases and thus increase cyclic AMP concentrations. Cyclic AMP activates PKA by dissociating the holoenzyme into its regulatory and catalytic subunits. The stimulation of adenylyl cyclase by αs is counteracted by Gi-derived αi- or βγ-mediated inhibition of the cyclase. Subunits of α12/13 stimulate the amiloride-sensitive Na+/H+ exchanger 1 (NHE1) increasing intracellular pH or proton extrusion following acid loading 136, 137. The α12-induced NHE1 activation is dependent on DG- regulated PKCs [136]and thus the effect is almost certainly indirect. The agonists that stimulate the G12/13-coupled pathways are poorly-characterized although thrombin and thromboxane A2 signal through G12/13PCRs in platelets and astrocytoma cells 138, 139.

All four G protein subfamilies have been implicated in the activation of the ERK cascade. Although the ERKs were only partially-characterized in the later 1980s and early 1990s, it became clear that they were stimulated by phorbol esters (see, for example, 140, 141, 142, 143, 144, 145) as well as by growth factors acting through RPTKs. Therefore Gq/11PCR agonists (which stimulate PtdInsP2 hydrolysis and activate DG-regulated PKCs 146, 147, 148) should activate the ERKs in a DG-regulated PKC-dependent manner. This has been shown to be the case (see e.g., Refs. 148, 149, 150, 151, 152). It has been subsequently recognized that some GPCR agonists (e.g., the serum mitogen lysophosphatidic acid (LPA, reviewed in Ref. [153]) or thrombin) activate ERKs in a PTX-sensitive manner, indicating an involvement of Gi/o (154, 155, 156, 157, 158, 159, 160, 161and reviewed in Ref. [130]). The finding that cyclic AMP activates the ERK cascade in some cells 135, 162suggests the existence of Gs-mediated activation. Transfection of constitutively-activated αq and αs (and indeed α12/13) leads to activation of ERKs 135, 163. In contrast, αi is ineffective [135], the effector in the case of Gi being βγ dimers rather than α subunits 135, 164, 165, 166. Since all G protein subfamilies stimulate ERKs and because some GPCR agonists act through more than one G protein subfamily, signals converge on the ERK cascade from more than one G protein subfamily in the same cell. Although there is not universal agreement 151, 167, 168, activation of ERK by endothelin-1 (ET-1) may be an example of this convergence. Here, activation is dependent on both DG-regulated PKCs as shown by sensitivity to chronic TPA treatment (implicating Gq) or PTX (implicating Gi) 169, 170.

Increases in intracellular Ca2+ concentrations may also lead to activation of ERKs 171, 172. This has implications for GPCR-linked agonism. In cardiac myocytes, the positively chronotropic and inotropic β-adrenergic agonist isoprenaline weakly activates ERKs [173]. This activation is dependent on the influx of extracellular Ca2+ but independent of cyclic AMP [173]. The physiological role of these Ca2+-dependent pathways is unclear.

Whilst it may be perturbing that so many agonists and signaling pathways activate ERKs (it is almost impossible to identify an intervention which does not activate ERKs to some extent), in vivo responses are likely to be cell-specific. It is important when studying these kinases to assess the relative magnitude of the response by comparison with the effect of a known powerful agonist. The time-course of ERK activation also needs to be considered. Activation is often transient (e.g., [150]) and, in some cases, only prolonged activation succeeds in eliciting a biological response (e.g., differentiation as indicated by neurite outgrowth in the neuronal (phaeochromocytoma) PC12 cell line [94]). ERK activity may need to attain a threshold either in terms of absolute activity or in terms of activity integrated over time in order to stimulate a downstream response.

The regulation of Raf activity against its physiological substrates (MEK1 and MEK2) by GPCR agonists has only recently been explored. In the cardiac myocyte, A-Raf and c-Raf (B-Raf is absent) are activated strongly by TPA or ET-1, but less so by PE or carbachol [16]. Activation of c-Raf is more transient than the activation of A-Raf [16]. Differences in the time-course of activation of Raf isoforms have been noted by others [17]and these may have implications for the activation of the ERK cascade. Stimulation of A-Raf and c-Raf by ET-1 is attenuated by depletion of DG-regulated PKCs or by PTX pretreatment, implying participation of both Gq- and Gi-linked pathways [16]. GsPCRs (e.g., the β-adrenergic receptor) increase the concentration of cyclic AMP which inhibits growth [174]. Although controversial [175], the inhibition of the Ras.GTP-c-Raf interaction by phosphorylation of Ser-43 in c-Raf (see above) by the catalytic subunit of PKA may account for cyclic AMP-mediated growth arrest. However, activation of PKA is much less effective in inhibiting stimulation of A-Raf by ET-1 [16]or of B-Raf by serum [59]. These findings presumably reflect the absence of a residue corresponding to Ser-43 in c-Raf. The inability of cyclic AMP to inhibit GPCR-mediated activation of MEK and ERK in some cells 16, 176may be attributable to the ability of A-Raf and/or B-Raf to circumvent inhibition of c-Raf (but see also 57, 177).

The steps linking GPCRs to the ERK cascade are still unclear and a number of pathways may be involved. The mechanisms that couple RPTKs to the ERK cascade are much better characterized (reviewed in Refs. 178, 179, 180). Since some steps are probably common to both, the RPTK-coupled pathway is a useful starting point for discussion of the GPCR-linked pathway (Fig. 3). Binding of agonists to the extracellular domains of transmembrane single chain RPTKs induces conformational change and dimerization. This allows the transautophosphorylation of specific Tyr-residues in the intracellular domains of RPTKs. The phosphotyrosine and its neighboring C-terminal residues are recognized by the Src-homology 2 (SH2) domains (reviewed in Refs. 179, 180) in the adaptor protein Grb2 (or Grb2-like proteins such as Grap [181]). Grb2 additionally contains an SH3 domain which is probably constitutively associated with Pro-rich sequences present in the GEF, Sos (reviewed in Ref. [180]). Formation of the RPTK-Grb2-Sos complex in the plane of the membrane places Sos in proximity to inactive Ras.GDP. Sos stimulates the exchange of GDP for GTP on Ras, thereby producing the active form of Ras. As described above, the interaction of Ras.GTP and c-Raf translocates Raf to the membrane for activation (Fig. 2).

In some cases, members of a second adaptor protein family are associated with tyrosine-phosphorylated RPTKs (Fig. 3). These are the 46, 52 and 66 kDa members of the Shc family (reviewed in Ref. [182]). p46-Shc and p52-Shc contain an N-terminal phosphotyrosine binding (PTB) domain, a central Pro-/Gly-rich domain, and an N-terminal SH2 domain. p66-Shc contains an additional Pro-/Gly-rich domain at its N-terminus. Like SH2 domains, PTB domains bind to phosphotyrosine-containing sequences, but in contrast to SH2 domains, sequence specificity involves residues N-terminal to the phosphotyrosine residue (reviewed in Ref. [183]). Shc binds to phosphotyrosine sequences in activated RPTKs which then phosphorylate a Tyr-residue in the central Pro-/Gly-rich domain of Shc. This phosphorylation site is recognized by Grb2 SH2 domains. As in the direct interaction of Grb2 with a phosphorylated RPTK, the formation of the [phospho-RPTK.phospho-Shc.Grb2.Sos] complex stimulates Ras.-GTP formation, thence c-Raf activity.

In RPTK-linked signalling, the adaptor proteins (Grb2 or Shc) interact directly with the phospho-RPTK. This does not occur with GPCRs and the pathway to activation of ERK is more complex and less clear. An increasing number of GPCR agonists have been shown to increase tyrosine-phosphorylation of Shc and association of Shc with Grb2 (and, in some cases, Sos). They include angiotensin II 184, 185thyrotropin-releasing hormone [186], ET-1 187, 188, thrombin 188, 189, 190, LPA [188], bradykinin [191], carbachol 191, 190, N-formyl peptide chemoattractant [192], gastrin [193], βγ subunits 166, 194and the α2A-adrenoceptor [194]. There is disagreement about whether this process is [194]or is not [189]sensitive to PTX, but this may reflect cell-specificity or which G proteins participate in signal transduction. It is also unclear whether TPA does [191]or does not [189]stimulate Shc phosphorylation. In spite of these inconsistencies, it is likely that Shc is involved in at least some of the pathways linking GPCRs to the ERKs.

Although GPCR-linked stimulation of ERKs probably involves tyrosine phosphorylation, Shc, Ras.GTP loading and activation of Raf, the pathway has yet to be fully elucidated. In this section, we summarize potential mechanisms that may be involved in the activation of the ERK cascade by GPCR agonists.

1. GPCRs may act through RPTKs. In Rat 1 fibroblasts, the RPTK epidermal growth factor (EGF) receptor and the related p185neu oncoprotein become rapidly and transiently tyrosine-phosphorylated in response to ET-1, LPA and thrombin [188], with formation of a complex between the phospho-EGF receptor, 52 kDa phospho-Shc and Grb2. Activation of ERKs by these agonists was decreased by a transfected dominant negative EGF receptor mutant or by a PTK inhibitor with selectively for the EGF receptor-coupled pathway. These findings suggest that GPCR agonists somehow stimulate tyrosine phosphorylation of the EGF receptor, and activation of the EGF receptor then stimulates the ERK cascade. How the phosphorylation of the EGF receptor is brought about is not understood. Although this hypothesis has some attractive features (it unifies RTPK and GPCR-linked signaling), further studies in other cell types are necessary.

2. The non-receptor PTK PYK2. A search for GRB2 SH3 domain binding proteins led to the cloning of a novel PTK, PYK1 (for proline-rich tyrosine kinase) [191]. A second PTK, PYK2, was cloned from a human foetal brain library. PYK2 is highly homologous with two other recently-cloned mammalian PTKs, CAKβ [195]and RAFTK [196]. These PTKs show significant homology with the focal adhesion kinase, pp125FAK (reviewed in Ref. [197]). PYK2, CAKβ and RAFTK all contain Pro-rich domains (to which SH3 domain-containing proteins presumably bind) but lack transmembrane sequences, acylation sites, or SH2 and SH3 domains. The regulation of CAKβ and RAFTK has not been studied extensively but the regulation of PYK2 has been characterized in PC12 cells [191]. PYK2 becomes rapidly tyrosine-phosphorylated in response to interventions that raise intracellular Ca2+ concentrations (nicotinic acetylcholine receptor (AchR) agonism, membrane depolarization, bradykinin) or activate PKC (bradykinin, TPA). These manipulations also cause tyrosine-phosphorylation of Shc and activate ERKs. Overexpression of PYK2 results in increased interactions between PYK2, Shc, Grb2 and Sos mediated by SH2 and SH3 domains. In addition, dominant-negative PYK2 inhibits nicotinic AchR activation of ERK, implicating PYK2 in the regulation of the cascade. Lev et al. [191]suggest that activated (tyrosine-phosphorylated) PYK2 can directly recruit Grb2-Sos, or can indirectly recruit Grb2-Sos through Shc phosphorylation, although the intracellular location of the putative interaction is unclear. The implication is that formation of these complexes leads to activation of Ras and Raf.

3. Activation of the Src family of non-receptor PTKs by GPCR agonists. Non-receptor PTKs of the c-Src family were recognized as the cellular homologues of tumour virus oncoproteins. The src gene family (reviewed in Refs. 198, 199, 200, 201, 202) has nine known members. Alternative splicing leads to the existence of at least thirteen protein products, although the tissue distribution of many Src PTKs is restricted. Sites for myristoylation and palmitoylation lie close to the N-terminus and are concerned with partitioning of Src PTKs to membranes. There are two conserved Tyr-residues, one in the kinase catalytic (SH1) domain and the other in the C-terminal variable domain. Phosphorylation of the C-terminal Tyr-residue is inhibitory and is mediated through the non-receptor PTK Csk or Csk-relatives. Mutation of the C-terminal Tyr-residue to Phe stimulates the kinase activity of Src and renders it oncogenic. Dephosphorylation is mediated by poorly-characterized protein tyrosine phosphatases. The kinase activity of Src PTKs may be suppressed by intramolecular (or intermolecular) binding of the SH2 domain to the C-terminal phosphotyrosine and the SH3 domain to an unidentified region in Src. Dephosphorylation of the C-terminal phosphotyrosine disrupts these interactions and produces a more “open” structure. Autophosphorylation of the SH1 domain Tyr-residue follows and stimulates kinase activity (Fig. 2). There may also be other pathways of Src activation. For example, phosphotyrosine-containing sequences in RPTKs may compete with the Src PTK C-terminal phosphotyrosine for binding to the Src SH2 domain. This may open Src PTK structure and allow activation.

Although the mechanisms are unclear, Src family PTKs are activated by GPCR agonists such as platelet-activating factor [203], LPA [204], thrombin 205, 206, angiotensin II [207], N-formylpeptide [192], ET-1 [208]and bradykinin [209]. The ability of both GiPCRs and GqPCRs to activate the ERK cascade is reduced in cells depleted of Src family PTKs [210]. The targets of GPCR-stimulated Src PTKs are unknown, but could conceivably include Shc [192]and c-Raf 48, 49, 50. There is thus an emerging view that Src PTKs are involved in the activation of the ERK cascade, although the precise nature of the involvement is still unclear.

4. GqPCRs and PKC. Hydrolysis of PtdInsP2 by Gq-activated PI-PLCβ produce DG which stimulates the DG-regulated PKCs (Fig. 5), and activation of the ERK cascade follows. Down-regulation of DG-regulated PKCs inhibits the ability of GqPCR agonists to activate ERKs (see, for example, Refs. 150, 151, 165). This PKC-dependent pathway is utilized by, for example, α1B-adrenergic agonists in a number of cells 149, 151, 165. αq Subunits rather than Gq-derived βγ dimers are generally considered to be responsible for activation of PI-PLCβ. As mentioned earlier, αq stimulates PtdInsP2 hydrolysis (reviewed in Ref. [123]) and sequestration of βγ dimers by the PH domains of the C-terminal region of the β-adrenergic receptor kinase 1 (βARKCT) does not affect the stimulation of PtdInsP2 hydrolysis by α1B-adrenergic agonists [165]. The general characteristics of GqPCR-dependent signalling to the ERK cascade are shown in Table 1.

PKC is a large multi-enzyme family (reviewed in Ref. [128]). The classical PKC isoforms (cPKC-α, cPKC-β1/2 and cPKC-γ) and the novel PKC isoforms (nPKC-δ, nPKC-ϵ, nPKC-η, nPKC-θ and possibly novel-like PKC, mPKC-μ/PKD) contain consensus sequences for DG binding. Activation of cPKCs and nPKCs (as shown by their translocation from the soluble to the particulate fraction) by TPA and GPCR agonists such as ET-1 and PE has been clearly demonstrated (see, for example, Refs. 148, 161, 211, 212). TPA and ET-1 activate Raf, this activation being sensitive to down-regulation of DG-regulated PKC isoforms [16]. Furthermore, GqPCR-linked activation of the ERK cascade is prevented by inhibitory Raf constructs [165]. These experiments strongly implicate Raf in PKC-mediated activation of the ERK cascade. The steps leading form PKC activation to stimulation of Raf are poorly characterized. In some (but not all [213]) cells, GqPCR-linked agonism and activation of DG-regulated PKCs by phorbol esters causes Ras.GTP loading 185, 214, 215, 216. Ras.GTP loading in response to GqPCR-linked agonists and phorbol esters might be anticipated, given the likely importance of Ras.GTP in the activation of Raf. However, evidence has been presented showing that inhibitory Ras constructs do not prevent activation of ERK by GqPCR-linked agonism [165]and Ras-independent pathways for such agonists may exist. This is clearly an area where additional experiments are required.

There have been sporadic reports that PKC directly activates c-Raf by phosphorylation of a Ser-residue and this leads to stimulation of the ERK cascade 217, 218, 219, 220. The significance of these results is difficult to assess. The finding that phosphatidate directly activates c-Raf [41]could have important implications for PKC-dependent pathways of ERK activation. Phosphatidate is produced by the hydrolysis of phosphatidylcholine catalyzed by phospholipase D (reviewed in Ref. [221]). GPCR agonists and TPA activate phospholipase D by a mechanism dependent on DG-regulated PKCs. The significance of a GPCR → PI-PLCβ → DG → PKC → phospholipase D → phosphatidic acid → c-Raf pathway leading to activation of ERK merits further investigation.

5. The G-protein βγ subunit pathway. GiPCR-linked receptors such as the LPA and thrombin receptors, the α2- adrenoceptor and the m2 AChR activate ERKs in a PTX-sensitive manner (154, 155, 156, 157, 158, 159, 160and reviewed in Ref. [130]). Because overexpression of αi does not lead to activation of ERKs [135], the interpretation of these results is that βγ dimers link GiPCRs to the ERK cascade. More direct evidence of βγ participation includes the ability of overexpressed βγ dimers to activate ERK 135, 164, 222, the potency of activation correlating with the ability of the β and γ isoforms to form heterodimers [165]. Furthermore, sequestration of βγ dimers by binding to transfected α subunits of the G protein retinal transducin, or to the pH domains of overexpressed βARKCT (or other proteins containing PH domains) inhibits activation of ERK by GiPCR agonists 135, 164, 222, 223, 224. The characteristics of GiPCR- and βγ-linked signaling to the ERK cascade are shown in Table 1. GsPcR-linked signalling to the ERK cascade may also involve βγ dimers [225], although this does not explain the direct activation of ERK by cyclic AMP 135, 162. In GsPCR-linked signalling, the balance between the activating effects of Gs derived βγ dimers [225]and the inhibition of c-Raf by cyclic AMP-mediation of PKA (reviewed in Ref. [60]) will presumably be involved in determining whether Gs-linked agonism inhibits or activates the ERK cascade.

Although GiPCR agonists may stimulate PtdInsP2 hydrolysis slightly in a PTX- and βARKCT-sensitive manner, GiPCR and βγ signalling is independent of DG-regulated PKCs [165]. Thus βγ-linked activation of PI-PLCβ is presumably of little significance. One pathway of βγ signaling to ERKs involves Shc → (GRB2.Sos?) → Ras → Raf → (MEK?) → ERK. βγ dimers cause tyrosine-phosphorylation of p46- and p52-Shc 166, 194but the PTK activity responsible has not been characterized. The activation of ERK by βγ dimers is inhibited by the PTK inhibitor herbimycin 158, 165, a finding that has been interpreted as indicating participation of Src PTKs [165]. However, herbimycin may not be specific for Src PTKs and the effect may be attributable to inhibition of other PTKs or even of the PTK activity of MEK1/MEK2. As might be predicted from their inability to cause Shc phosphorylation 166, 188, 189, 190, 191, 192, 193, 194, GiPCR agonists and βγ dimers stimulate Ras.GTP loading 159, 223, 226, 227. The stimulation of the ERK cascade by βγ dimers for GiPCR agonism is prevented by inhibitory Ras constructs 156, 157, 159, 160, 164, 165, 223. The GEF Ras-GRF has been implicated in the βγ pathway [35]. Confusingly, Ras.GRF is activated by Ser-/Thr-phosphorylation [35]. Moving further downstream, inhibitory constructs of Raf prevent GiPCR-linked and βγ-mediated activation of ERK indicating that Raf is the MAPKKK responsible 157, 159, 165. Although βγ dimers interact with c-Raf [228], the effect of this interaction on c-Raf activity has not been examined. In any case, a direct effect of βγ dimers on c-Raf activity would not account for many of the experimental observations. Thus the conclusion from these experiments is that βγ dimers are capable of activating ERK probably through a Shc-dependent pathway but the events upstream from Shc are still unclear. The activation of Shc appears to involve protein phosphorylation. It would be gratifying if a βγ- activated PTK that phosphorylated Shc could be identified.

The involvement of βγ dimers may not be confined to G protein-linked signalling. Although activation of ERK by most RPTK agonists is insensitive to PTX, activation by the RPTK agonist insulin-like growth factor-1 (and, in some cells, insulin) is sensitive to PTX [229]. Further characterization showed that activation of the ERK cascade by IGF-1 is: (i) sensitive to the PTK inhibitor genistein; (ii) unaffected by depletion of DG-regulated PKCs; (iii) sensitive to inhibition by βARKCT; (iv) prevented by inhibitory Raf constructs [229]. This suggests that βγ subunits may be involved in signalling from some RPTKs although the nature of the involvement is unclear. Maybe the putative βγ-activated PTK involved in the activation of Shc (see above) is an RPTK 188, 229.

6. Go-linked pathways. Honda et al. [230]and van Biesen et al. [231]recently showed that the activation of ERK by platelet activating factor (which couples to Go, Gi and Gs) is sensitive to PTX implicating Gi/o. Unlike the Gi-linked βγ pathway, activation is insensitive to βARKCT and does not involve Ras [231]. Activation is sensitive to depletion of DG-regulated PKCs [231]. These data were interpreted as indicating that the platelet activating factor pathway signaling to ERK involves αo, and are supported by the finding that a PTX-resistant αo mutant rescues the PTX-inhibited response. The characteristics of the Go-linked pathway are shown in Table 1. The mechanism of αo coupling to PKC is unclear. αo may be more widely involved in GPCR signaling to the ERK cascade than previously realized. Thus, the characteristics of activation of ERK by the m1 AchR (generally considered to be a Gq/11PCR) were similar to those for platelet activating factor [231]. Furthermore, activation of c-Raf and A-Raf by ET-1 is dependent on DG-regulated PKCs and is sensitive to PTX [16]. Although this was interpreted as indicating involvement of both Gq/11 and Gi/o [16], it is equally explicable by signaling through αo.

7. Involvement of phosphatidylinositol 3-kinase in βγ signalling to the ERK cascade. Activation of Shc, Raf and ERK by GiPCr agonists or βγ dimers is attenuated by inhibitors of phosphatidylinositol 3-kinase (PI3K) such as wortmannin and LY294002 194, 232, 233. PI3K phosphorylates PtdInsP2 to form the putative membrane phospholipid signaling molecule phosphatidylinositol 3,4,5-triphosphate (PtdInsP3) and is strongly activated by RPTK- and GPCR-linked mitogens (reviewed in Ref. [234]). The targets for PtdInsP3 are unclear but it binds SH2 domains [235]potentially localizing SH2-containing proteins to the membrane. “Classical” PI3K consists of a regulatory 84 kDa subunit and a catalytic 110 kDa subunit of which two isoforms (p110α and p110β) have been identified. The precise sequence of events in PI3K-dependent signaling is unclear. PI3K associates with phosphotyrosine sequences in activated RPTKs through SH2 domains in its p85 subunits [234]and with the Src family PTK, Lyn [192]. The p110 catalytic subunit of PI3K associates with Ras which activates it in a GTP-dependent manner 236, 237. This suggests the PI3K lies downstream from Ras. In contrast, wortmannin and LY294002 have been reported to inhibit LPA-stimulated Ras.GTP loading, suggesting that Ras lies downstream from PI3K [233]. Further evidence of the involvement of PI3K in the regulation of the ERK cascade is provided by the ability of a dominant negative PI3K p85 subunit mutant to inhibit stimulation of ERK by LPA [233]. Irrespective of the relative positions of Ras and PI3K in activation of the ERK cascade, the involvement of PI3K in the regulation of the ERK cascade deserves further investigation.

In addition to the classical PI3Ks, there have been two reports of βγ-activated PI3Ks. A PI3K with an apparent molecular mass of 210–220 kDa has been partially purified [238]. βγ-activated PI3K is inhibited by wortmannin albeit with a lower potency than the classical PI3Ks. Stoyanov et al. have cloned a PI3K isoform (p110γ) with a calculated molecular mass of 120 kDa [239]. It shows about 30% homology with p110α and p110β and has a potential PH (βγ-dimer interaction?) domain close to its N-terminus. PI3K p110γ is activated by both α and βγ G protein subunits. The relationship between p110γ and the PI3K identified by Stephens et al. [238]is currently unclear, as is the relationship of βγ-activation of PI3Ks to the ERK cascade.

Section snippets

General conclusions

The elucidation and understanding of the ERK cascade has been one of the most active areas in biological research over the past few years. The cascade plays a major role in signal transduction from both GPCRs and RPTKs. Activation of the ERK cascade by RPTKs is relatively well understood. In RPTK-linked signaling, adaptor proteins such as Shc and Grb2 interact directly with the phosphorylated receptors to promote activation of Ras and Raf. Despite intense investigation, the coupling mechanisms

References (240)

  • C.J. Marshall

    Curr. Opin. Gen. Dev.

    (1994)
  • E. Cano et al.

    Trends Biochem. Sci.

    (1995)
  • J.S. Campbell et al.

    Rec. Prog. Horm. Res.

    (1995)
  • M.H. Cobb et al.

    J. Biol. Chem.

    (1995)
  • C.J. Marshall

    Cell

    (1995)
  • S. Cowley et al.

    Cell

    (1994)
  • R. Seger et al.

    J. Biol. Chem.

    (1994)
  • G. Daum et al.

    Trends Biochem. Sci.

    (1994)
  • M.A. Bogoyevitch et al.

    J. Biol. Chem.

    (1995)
  • V. Wixler et al.

    FEBS Lett.

    (1996)
  • X. Wu et al.

    J. Biol. Chem.

    (1996)
  • P. Lenormand et al.

    J. Biol. Chem.

    (1996)
  • C.W.M. Reuter et al.

    J. Biol. Chem.

    (1995)
  • J.L. Blank et al.

    J. Biol. Chem.

    (1996)
  • A. Minden et al.

    Cell

    (1995)
  • B.M.T. Burgering et al.

    Trends Biochem. Sci.

    (1995)
  • S. Ghosh et al.

    J. Biol. Chem.

    (1994)
  • D.K. Morrison et al.

    J. Biol. Chem.

    (1993)
  • S. Ghosh et al.

    J. Biol. Chem.

    (1996)
  • J. Avruch et al.

    Trends Biochem. Sci.

    (1994)
  • V. Cleghon et al.

    J. Biol. Chem.

    (1994)
  • P. Dent et al.

    J. Biol. Chem.

    (1996)
  • P. Peraldi et al.

    FEBS Lett.

    (1995)
  • A. Aitken

    Trends Biochem. Sci.

    (1995)
  • A.J. Muslin et al.

    Cell

    (1996)
  • B. Yamamori et al.

    J. Biol. Chem.

    (1995)
  • C.-F. Zheng et al.

    J. Biol. Chem.

    (1994)
  • M. Yan et al.

    J. Biol. Chem.

    (1994)
  • D.R. Alessi et al.

    J. Biol. Chem.

    (1995)
  • T.A. Haystead et al.

    FEBS Lett.

    (1992)
  • T.G. Boulton et al.

    Cell

    (1991)
  • R.J. Davis

    J. Biol. Chem.

    (1993)
  • R. Marais et al.

    Cell

    (1993)
  • T.A.J. Haystead et al.

    J. Biol. Chem.

    (1994)
  • D.J. Robins et al.

    Adv. Cancer Res.

    (1994)
  • K. Malarkey et al.

    Biochem. J.

    (1995)
  • R. Seger et al.

    FASEB J.

    (1995)
  • D.T. Denhardt

    Biochem. J.

    (1996)
  • S.J. Mansour et al.

    Science

    (1994)
  • G.R. Post et al.

    FASEB J.

    (1996)
  • D.K. Morrison

    Mol. Reprod. Dev.

    (1995)
  • T. Jelinek et al.

    Mol. Cell. Biol.

    (1994)
  • C.A. Pritchard et al.

    Mol. Cell. Biol.

    (1995)
  • C.A. Lange-Carter et al.

    Science

    (1993)
  • C. Patriotis et al.

    Proc. Natl. Acad. Sci. U.S.A.

    (1994)
  • A. Salmerón et al.

    EMBO J.

    (1996)
  • A. Minden et al.

    Science

    (1994)
  • M. Yan et al.

    Nature

    (1994)
  • A. Lin et al.

    Science

    (1995)
  • J.K. Drugan et al.

    J. Biol. Chem.

    (1996)
  • Cited by (0)

    View full text