---

The features of three distinct protein phosphorylation cascades in mammalian cells are becoming clear. These signalling pathways link receptor-mediated events at the cell surface or intracellular perturbations such as DNA damage to changes in cytoskeletal structure, vesicle transport and altered transcription factor activity. The best known pathway, the Ras Raf MEK ERK cascade [where ERK is extracellular-signal-regulated kinase and MEK is mitogen-activated protein (MAP) kinase/ERK kinase], is typically stimulated strongly by mitogens and growth factors. The other two pathways, stimulated primarily by assorted cytokines, hormones and various forms of stress, predominantly utilize p21 proteins of the Rho family (Rho, Rac and CDC42), although Ras can also participate. Diagnostic of each pathway is the MAP kinase component, which is phosphorylated by a unique dual-specificity kinase on both tyrosine and threonine in one of three motifs (Thr-Glu-Tyr, Thr-Phe-Tyr or Thr-Gly-Tyr), depending upon the pathway. In addition to activating one or more protein phosphorylation cascades, the initiating stimulus may also mobilize a variety of other signalling molecules (e.g. protein kinase C isoforms, phospholipid kinases, G-protein a and bg subunits, phospholipases, intracellular Ca²⁺). These various signals impact to a greater or lesser extent on multiple downstream effectors. Important concepts are that signal transmission often entails the targeted relocation of specific proteins in the cell, and the reversible formation of protein complexes by means of regulated protein phosphorylation. The signalling circuits may be completed by the phosphorylation of upstream effectors by downstream kinases, resulting in a modulation of the signal. Signalling is terminated and the components returned to the ground state largely by dephosphorylation. There is an indeterminant amount of cross-talk among the pathways, and many of the proteins in the pathways belong to families of closely related proteins. The potential for more than one signal to be conveyed down a pathway simultaneously (multiplex signalling) is discussed. The net effect of a given stimulus on the cell is the result of a complex intracellular integration of the intensity and duration of activation of the individual pathways. The specific outcome depends on the particular signalling molecules expressed by the target cells and on the dynamic balance among the pathways.

---

Abbreviations used: see Table 1 for the names and abbreviations of most of the signalling intermediates mentioned in this review; CREB, cAMP response element-binding protein; GAP, GTPase-activating protein; GEF, guanine nucleotide exchange factor; GSK-3, glycogen synthase kinase-3; Hsp90, heat-shock protein of 90 kDa; IL-1, interleukin-1; JAK, Janus kinase; LPS, lipopolysaccharide; MAPK, mitogen-activated protein kinase; NFkB, nuclear factor kB; PH domain, pleckstrin homology domain; PI3-K, phosphoinositide 3-kinase; PKA, protein kinase A; PKC, protein kinase C; PLC and PLD, phospholipases C and D respectively; PTB domain, phosphotyrosine-binding domain; SH domain, Src homology domain; SRF, serum response factor; STAT, signal transducer and activator of transcription; TNFa, tumour necrosis factor a.

---

INTRODUCTION    

The 21 kDa mammalian Ras proteins (p21) consist of 188 or 189 amino acids. They are integral to signal transduction pathways connecting events at many cell surface receptors to intracellular processes. Mammalian cells contain three very similar ras genes (H-ras, N-ras and K-ras, which has two splice variants, A and B, that differ in their C-terminal sequences). In this review, Ras is used to refer collectively to the products of these three genes, even though in different cell types one or the other may predominate. At the protein level Ras is present in all cells, with the highest levels in proliferating cells. At the mRNA level H-ras is highest in skin and muscle, K-ras in gut and thymus, and N-ras in thymus and testes. For signalling purposes, there are no known differences among these Ras proteins, though cell-specific differences in the transforming potential of the codon 12 Gly Asp mutants have been reported [1]. Recent reviews of the Ras-controlled signalling pathways contain references to background material that is assumed or summarized here [2–4].

These three ras genes are part of a large superfamily of genes encoding small GTP-binding proteins (G-proteins) [5]. Closely related to Ras are several species of Rap proteins found in granules in the Golgi and endoplasmic reticulum. Rap1A can antagonize the transforming ability of Ras, a function that led to its isolation as a suppressor, called Krev-1, of the K-ras oncogene. Also closely related to Ras are the Ral-A and Ral-B proteins, which appear to regulate the activity of exocytic and endocytic vesicles. This Ras/Rap/Ral group is important in growth and development, and certain members are implicated in exocytosis, anabolic processes and/or regulation of the oxidative burst. Less closely related to Ras are several other families in the Ras superfamily [5–10]. The Rho family, which includes Rho-A, -B and -C, Rac-1 and -2, CDC42, Rho-G and TC10, embraces small G-proteins that play dynamic roles in the regulation of the actin cytoskeleton and focal contacts, mediating formation of filopodia and lamellipodia. Rac also controls NADPH oxidase activity in phagocytes. Ran proteins are involved in the transport of RNA and proteins across the nuclear membrane. ARF/SAR proteins are important for vesicle formation and budding. Members of the large and extensively studied Rab/YPT family are involved both in regulating intracellular vesicle trafficking between donor and acceptor membrane-enclosed compartments and in controlling the exocytosis and endocytosis of different types of vesicles.

Figure 1 illustrates the structure of the GTP–Ras protein [11–13]. Presumably all members of the Ras superfamily have a very similar structure. Notable features include the highly conserved catalytic domain made up of amino acids 1–164, the GTP-binding site (constituted from amino acids 12–18, 57–63, 116–119 and 144–147) and the two 'switch' regions (amino acids Asp-30–Asp-38 and Gly-60–Glu-76). The switch regions of the Ras protein are close to the g-phosphate group of the activating GTP and exhibit different conformations depending upon whether GDP or GTP is bound. This N-terminal proximate region is the effector binding site (involving amino acids Tyr-32–Tyr-40) responsible in part for interactions with the GTPase-activating protein (GAP) and with downstream effectors of Ras action such as Raf, which competes with GAP for binding to Ras. The two Ras-GAPs, p120Ras^GAP and neurofibromin, appear to interact somewhat differently with Ras in that substitution of the amino acids Lys-Arg-Val with Leu-Ile-Arg at positions 101–103 had no effect on the intrinsic hydrolytic activity of Ras or its sensitivity to neurofibromin, but did reduce p120Ras^GAP-stimulated hydrolysis 2–3-fold [14]. The interaction of the guanine nucleotide exchange factor (GEF) with Ras is, in part, with the switch 2 region [15].

The Ras proteins are bound to the inner surface of the plasma membrane by several lipophilic interactions involving post-translational modifications of the C-terminus [5,12]. A cysteine residue very near to the C-terminus, in a CAAX (where A = aliphatic amino acid) motif, typically has a polyisoprenyl moiety attached to it, either C₁₅ (farnesyl) for Ras or C₂₀ (geranylgeranyl) for some Rap proteins and members of the Rho superfamily. (These modifications are determined by the specific C-terminal amino acid sequences and they may determine the specific cell membrane targeted by the protein.) Subsequently the three C-terminal amino acids are removed and the newly generated farnesylated Ras C-terminal cysteine is carboxymethylated. H-Ras, N-Ras and K-Ras(A) have in addition a palmitoyl group added to a cysteine residue in the hypervariable region (amino acids 165–185) near the C-terminus; K-Ras(B) does not have a cysteine residue that can be palmitoylated, but does have a polybasic domain that fulfils a similar function, presumably by forming an amphipathic helix.

Activation of Ras induces proliferation in many types of cells, and in its mutant oncogenic form Ras transforms many immortal cell lines, conferring on them a malignant phenotype [16]. Oncogenic Ras can also co-operate with an 'immortalizing oncogene' to transform primary cells. In some cell types, e.g. pheochromocytoma PC12 cells, Ras activation induces differentiation and neurite outgrowth. Oncogenic Ras bearing a mutation that alters the Ras structure so as to prolong the lifetime of the active GTP-bound form (in amino acids 12, 13, 59, 61 and 63 in the phosphate-binding region) generates a constitutive signal that is a major factor in many human cancers. Mutations that increase the guanine nucleotide exchange rate (amino acids 116, 117, 119 and 146) are in the base-binding region and also produce a transforming Ras. Ras is also responsible for transmitting the proliferative signal generated by a number of oncogenes, including the src, fms, and fes oncogenes.

ACTIVATION OF RAS- AND RHO-GTPase CASCADES

Ras and its relatives are activated in response to an extracellular or intracellular signal that generates the GTP-bound form and energizes the signal transducing ability. Typically the Ras–GTP level will increase 2–3-fold and remain high for at least 30 min after receptor activation. Hydrolysis of the bound GTP by an intrinsic GTPase activity relaxes the conformation and terminates the signal. Figure 2 illustrates the cycling of Ras between these two forms. Table 1 lists many of the proteins discussed in this review.

---

Table 1 Names, abbreviations and terminology

Receptors and growth factors
    TrkA: 'tropomyosin receptor kinase'; the receptor for NGF (nerve growth factor)
    TrkB and TrkC bind NGF-related neurotrophins
    EGF: epidermal growth factor
    PDGF: platelet-derived growth factor
    FGF: fibroblast growth factor
    IGF: insulin-like growth factor
    CSF: colony-stimulating factor

Small-molecular-mass p21 GTPase proteins
    Ras: rat sarcoma virus
    Rac: Ras-related C₃-botulinum toxin substrate
    Rho: Ras homologous
    Ral: Ras-related protein
    cdc42: homologous to yeast cell division cycle gene 42
    ARF: ADP-ribosylation factor
    SAR: secretion-associated and ras-superfamily-related gene
    YPT: yeast protein two

GAPs (GTPase-activating proteins)
    p120Ras^GAP: 120 kDa Ras-GAP
    NF1: neurofibromin; also a Ras-GAP
    p100GAP1^m: a rat Ras-GAP1
    GAP1^1P4BP: a human Ras-GAP1
    CDC42-GAP (Rho-GAP)
    p190Rho^GAP: 190 kDa protein; binds p120Ras^GAP
    BCR: product of breakpoint cluster region gene
    ABR: active BCR-related

GDS (guanine nucleotide dissociation stimulator)

GEFs (guanine nucleotide exchange factors)
    SOS: son of sevenless (a Drosophila gene product); SOS1 and 2 are mammalian homologues
    CDC25: mammalian homologue of yeast cdc25
    C3G: Crk SH3-binding GEF
    DBL: diffuse B cell lymphoma
    VAV: named from the sixth letter of the Hebrew alphabet
    OST: oncogene from rat osteosarcoma cells
    smgGDS: small G-protein GDP dissociation stimulator; possible GEF for K-Ras
    Ral-GDS: GDS protein active on Ral

Adaptor proteins
    GRB2: growth-factor-receptor-bound protein 2
    SHC: SH2-domain-containing a2-collagen-related
    NCK: a novel cytoplasmic protein

MAPKKKs (mitogen*-activated protein kinase kinase kinases)
    Raf-1, A-Raf, B-Raf
    PAK: p21-activated kinase
    MEKK: MEK kinase
    TAK: transforming-growth-factor-b-activated kinase

MAPKKs (mitogen*-activated protein kinase kinases)
    MEK1 (MKK1) and MEK2 (MKK2): MEK = MAPK/ERK kinase; MKK = MAP kinase kinase
    JNKK: JNK kinase
    SEK1: SAPK/ERK kinase 1; vastly prefers SAPK as substrate
    MKK3 and MKK 4: MAP kinase kinases 3 and 4
    RKK: RK kinase

MAPKs (mitogen*-activated protein kinases)
    ERK: extracellular-signal-regulated kinase; ERK1 = p44; ERK2 = p42
    FRK: Fos-related kinase
    JNK: Jun N-terminal kinase, also known as:
    SAPK: stress-activated protein kinase; several splice variants are known, producing proteins of about 46 and 54 kDa;
         thus there are p46 and p54 versions of SAPKa1, SAPKa2, SAPKb and SAPKg (p46SAPKa1 = JNK2)
    p38: RK (reactivating kinase)
    CSBP: cytokine-suppressive anti-inflammatory drug binding protein

TCFs (ternary-complex factors)
    Elk-1: Ets-like transcription factor
    SAP-1: SRF accessory protein 1 or stress-activated protein 1

MAPKAPK (mitogen-activated protein kinase activated protein kinase)
    MAPKAPK-1: RSK, the ribosomal S6 kinase p90^rsk
    MAPKAPK-2: 50 kDa protein kinase that phosphorylates Hsp25/Hsp27

Phosphatases
    SHP-1: also known as SHP, PTP1C, SHPTP-1 and HCP
    SHP-2: also known as SYP, PTP1D, SHPTP-2, SHPTP-3 and PTP2C
    SH-PTP: SH-domain-containing protein tyrosine phosphatase
    MKP-1: MAP kinase phosphatase-1 (also known as 3CH134 and CL100)
    PTP: protein tyrosine phosphatase
    PAC: phosphatase of activated cells
    PP1G: protein phosphatase-1 associated with hepatic glycogen

* Only some of these kinases are strongly activated by mitogens.

---

GAPs enhance the GTPase activity of normal Ras (but typically not oncogenic Ras) and thus shorten the lifetime of the signalling form. Presumably Ras–GTP continues to signal as long as the GTP remains intact, although association with GAP may alter the signal. The mammalian GAPs include the well studied p120Ras^GAP and the quite distinct and larger (290 kDa) neurofibromin, NF1, whose expression is largely restricted to neural tissues (including neural-crest-derived tissues such as Schwann cells) and gonadal tissues [17]. There are two members of the GAP1 family, p100GAP1^m and GAP1^1P4BP, which via a PH (pleckstrin homology) domain bind to and are stimulated by inositol 1,3,4,5-tetrakisphosphate [18,19]. GAPs active on Rho family members include Ral-BP1 (CDC42-GAP), p190Rho^GAP and BCR, which are respectively preferentially active on CDC42, Rho and Rac [10,20] (see Table 1 and Figure 3).

Studies with homozygous null mice ('knock-outs') have revealed the importance of some of these proteins. Animals unable to make p120Ras^GAP are embryonic lethals that exhibit increased Ras–GTP signalling and possess major defects in endothelial cell organization [21]. Mice unable to produce NF1 also die as embryos. It appears that one function of NF1 is to promote the death of certain neurons when an appropriate signal from certain nerve growth factors (neurotrophins) is not received [22].

GEFs [also known as guanine nucleotide releasing proteins, or guanine nucleotide dissociation stimulators (GDSs)], catalyse the release of GDP from inactive p21 proteins complexed with GDP; they are major players in signal activation [10,15,22]. Release of GDP allows GTP, present at much higher levels than GDP in the healthy cell, to bind and activate the p21 protein once again. SOS1 and SOS2, CDC25 and C3G are the four known mammalian GEFs with activity towards Ras–GDP; DBL, OST and smg-GDS are known GEFs acting to activate various Rho family members. DBL, isolated from a diffuse B-cell lymphoma, stimulates exchange activity for Rho-A, Rac-1 and CDC42 [15]. OST, isolated from an osteosarcoma, is an effector for Rac-1, acting as a GEF for Rho-A and CDC42, but not for Rho-B or Rho-C [23]. A domain found in DBL, known as the dibble homology domain, along with a PH domain, appears to be characteristic of the Rho-GEFs. To what extent different GEFs act differentially on members of the Ras and Rho (including Rac and CDC42) families in vivo, perhaps co-ordinately, remains to be established. Their importance is underscored by the fact that several known or putative Rho-GEFs are proto-oncoproteins (e.g. DBL, VAV, OST) that can be activated by N-terminal truncations. VAV is a 95 kDa protein expressed in haematopoietic cells that is tyrosine phosphorylated in response to the activation of a variety of receptors (e.g. the T-cell antigen receptor, the IgM receptor or c-Kit). It has a dibble domain, suggesting it may be a GEF, and it is capable of binding both to tubulin and to the nuclear protein Ku-70, the DNA-binding element of the DNA-dependent protein kinase [24,25].

Rho-GDI (guanine nucleotide dissociation inhibitor) binds equally well to both the GDP- and GTP-bound forms of CDC42, inhibiting both GDP dissociation and GTP hydrolysis [25a]. It inhibits the interaction of the p21Rac, Rho and CDC42 proteins with GAP and stimulates their release from cell membranes, perhaps to allow them to relocate to another membrane site [10].

The interactions among Ras, Ral, CDC42, Rac and Rho and both the downstream effectors of their actions and their diverse upstream activators, particularly physical and chemical stress, are poorly understood [9,10,15]. Via their respective and sometimes shared GEFs and GAPs, they have the potential to socialize as suggested in Figure 3, although which interactions are important is likely to depend on the abundance and location of relevant proteins in the cell in question. An important consequence of these shared interactions may be to place a particular GAP or GEF in a location where it can act more efficiently on its target p21 protein. p120Ras^GAP interacts with p190Rho^GAP, which can deactivate GTP-bound Rho, Rac and CDC42. The precise consequences of the interaction between these two GAP proteins is not known. BCR, which contains a serine/threonine kinase domain, also contains both a dibble domain that may function as a GEF for certain Rho family members (CDC42 > Rho-A > Rac in terms of relative activity) and a GAP domain targeted to CDC42 and Rac, but not to Rho-A [25b]. N-chimerin and B-chimerin are Rho-GAPs specific for Rac. Ral is a downstream effector of Ras, as shown by the ability of oncogenic Ha-Ras to activate the guanine nucleotide dissociation stimulator Ral-GDS, which (as a GEF) activates Ral-A and Ral-B; this appears to enhance the transforming activity of Ras (and Raf). Ral interacts with Ral-binding protein 1 [Ral-BP1; also known as Ral interacting protein 1 (RIP1)], which is a GAP protein for CDC42 [26–28]. Elegant microinjection experiments by Nobes and Hall [29] and Kozma et al. [30] have revealed that CDC42 can activate Rac, which in turn activates Rho. Each of these members of the Rho family can generate specific changes in cytoskeletal elements in response to specific extracellular factors (bradykinin for CDC42, PDGF or insulin for Rac and lysophosphatidic acid for Rho) [29,30]. In this so-called 'GTPase cascade', activation of CDC42 promotes the formation of shaft-like filopodia and microspikes, activated Rac promotes membrane ruffling and the formation of curtain-like lamellipodia, and activated Rho promotes the formation of stress fibres and focal adhesions. The formation of these different cytoskeletal structures presumably reflects the specific cytoskeletal components recruited in each case. The different ways in which these signalling elements can be arranged, determined by which proteins are actually present, generate specific cell surface structures and choreograph complex activities such as cell migration and neurite outgrowth [9].

Khosravi-Far et al. [31] have presented compelling evidence that oncogenic Ras, possibly acting via CDC42, requires both Rac-1 and Rho-A in order to establish a fully transformed state. They showed that dominant negative mutants of Rac-1 and Rho-A blocked Ras transformation, and that constitutively activated but weakly transforming Rac-1 and Rho-A could synergize with a weakly transforming Raf-1 to transform cells efficiently. Rho-A appeared to be downstream of Rac-1 on a pathway that had its major impact on the actin cytoskeleton, as contrasted with the action of the Raf-1 MEK ERK pathway on gene expression.

Signalling initiated by receptor tyrosine kinases

In one-pass receptor-initiated signal transduction pathways, activation of the cell surface receptor tyrosine kinase by its ligand generally stimulates receptor homo- or hetero-dimerization and/or a conformational change that typically results in autophosphorylation of multiple tyrosine residues in the cytoplasmic portion of the receptor [3,32]. Receptor dimerization can be accomplished either by dimeric ligands such as PDGF or by monomeric ligands such as EGF, presumably because there are two ligand-binding sites on the monomer. One of the first tyrosines to be autophosphorylated in the PDGF receptor is in the kinase activation domain of the receptor; this enhances the activity of the kinase. Different growth factor receptors (e.g. the PDGF, EGF, FGF and NGF receptors) present phosphorylated tyrosines in different constellations of amino acids. Thus each receptor interacts with its own characteristic set of proteins, endowing each receptor type with a unique composite signal. Cytokine receptors present a variation on this theme in that the receptor is not a tyrosine kinase itself, but rather upon activation stimulates an associated protein, a Janus kinase (JAK), to phosphorylate tyrosine residues on both itself and the receptor, thereby providing docking sites for various proteins including STATs (signal transducer and activator of transcription) and other signalling molecules such as SHC, SHP-2, phospholipase Cg (PLCg) and p85 [33].

Some of the proteins known to associate with particular tyrosine phosphates in the cytoplasmic domain of the dimerized and activated PDGF receptor are depicted in Figure 4 [34]. The proteins interact with the tyrosine phosphates via the SH2 (Src homology 2) domain or the more recently described PTB (phosphotyrosine-binding) sequence [35–37]. Structural studies have revealed a similarity between the PTB and PH domains that suggests a possible role for the PTB domain in membrane localization, a conclusion that is supported by its ability to bind acidic phospholipids [38]. Only a subset of those proteins potentially able to associate with the fully phosphorylated receptor are actually able to bind with any one particular activated receptor molecule. In some cases (e.g. binding of p85 to the PDGF receptor) more than one interaction may be involved (e.g. two SH2–phosphotyrosine interactions, or one SH2 and one PTB interaction), presumably reinforcing what is otherwise a fairly weak single interaction (K_d = 0.3–4.0 µM) [39]. In other cases two different proteins may compete for the same phosphotyrosine. For example, NCK can bind to pTyr-751; SHC, which has more relaxed binding site characteristics, competes for pTyr-579, -740, -751 and -771 [34]. Src, with only one SH2 site, appears able to bind to either Tyr-579 or, less strongly, Tyr-581. All these proteins contribute to the consequences of PDGF activation. Some of them have the potential to interact with each other, and other proteins, via SH3 domains, which recognize a left-handed polyproline type II helix domain in their partner [40]. Via such an interaction, both NCK and SHC (non-enzymic adaptor proteins) form a complex with SOS (SHC can also associate with SOS via GRB2) [41,42]. van der Geer et al. [3] have reviewed in detail the properties of the various proteins that can bind to tyrosine-phosphorylated receptors, and they suggest that activation of several distinct species of receptor-binding proteins is essential for a full response to any given stimulus.

The EGF receptor possesses five tyrosines located in a cluster spanning residues 992–1173 near the C-terminus that are strongly autophosphorylated after receptor activation and are thus assumed to be important in signal transmission. However, a mutant EGF receptor in which these tyrosines have been deleted or mutated to phenylalanine can still deliver mitogenic and transforming signals, albeit not as vigorously [43,44]. Although protein tyrosine phosphorylation in general was substantially reduced, SHC appeared to be fully phosphorylated after EGF stimulation of the mutant receptor, probably accounting for the activation of Ras and the ERKs. The process leading to SHC phosphorylation remains obscure, but could be due to interaction of the mutant EGF receptor, which still contained an active kinase domain, with susceptible target proteins (e.g. cErbB2). Alternatively, Tyr-845 may become phosphorylated by Src in an EGF-dependent manner, thus enabling the mutant protein to transmit a signal [45]. When Ras action is blocked by a dominant negative Ras mutant (p21^rasAsn-17), EGF-induced phosphorylation of ERK2 can still occur, possibly as a result either of PKC activation or of an elevation of the free intracellular Ca²⁺concentration, depending upon the cell type [46]. In contrast, ERK2 phosphorylation induced by insulin or PDGF was abrogated by expression of the dominant negative Ras mutant. The ERK proteins, reviewed below, are mitogen-activated protein kinases (MAPKs) that are responsible for the phosphorylation of many target proteins in response to mitogen stimulation.

Proteins that interact with protein tyrosine phosphates

Proteins with SH2-binding sites, such as Src, p85, SHP-2 ( = syp), PLCg, GAP, GRB2, SHC and NCK, bind via these domains only to those tyrosine phosphates that are found in the appropriate amino acid sequence context. This context is usually determined by the 3–6 amino acids on the C-terminal side of the phosphotyrosine. In contrast, the ability to bind the PTB sequence, which has been identified in nine proteins [37], appears to be determined largely by the 3–5 amino acids on the N-terminal side of the phosphotyrosine. Certain of the amino acids in the SH2 domains that are important in determining which phosphotyrosines will be most strongly bound have been identified and used to classify SH2 domains into groups which distinguish the various contexts in which the phosphotyrosine in found [47].

Like the phosphotyrosine-binding proteins, the protein tyrosine kinases (both receptor and non-receptor kinases) target different tyrosines depending upon the local amino acid context. Cytosolic protein tyrosine kinases prefer to phosphorylate tyrosines in binding motifs recognized by group I SH2 domains (e.g. Src, GRB2 and GAP), whereas receptor protein tyrosine kinases phosphorylate tyrosines in motifs recognized by group III SH2 domains (e.g. SHC, p85 and PLCg) [48]. (These groups are distinguished on the basis of the bD5 residue in the SH2 domain.) Proteins containing phosphotyrosine-binding sites (i.e. SH2 or PTB domains) compete with protein phosphotyrosine phosphatases for the tyrosine phosphate, and by doing so prolong the lifetime of those phosphotyrosine residues that are particularly strongly bound. Some of these SH2-containing proteins are enzymes [e.g. p120Ras^GAP, SHP-2, phosphoinositide 3-kinase (PI3-K), Src and PLCg] whose enzymic activity is enhanced, whereas others are adaptor proteins (GRB2, SHC and NCK) that bind a specific target protein and deliver it to the activated receptor.

Enzymes regulated by phosphotyrosine interactions

Association of p120Ras^GAP with tyrosine-phosphorylated receptors is mediated by its SH2 domains. By virtue of its ability to activate the Ras GTPase it is a negative regulator of Ras function [49]. p120Ras^GAP can also acquire a signalling function, mediated by the SH2 or SH3 domains in its N-terminal region, as a consequence of the interaction of its C-terminal region with Ras. This signalling may involve two phosphorylated proteins: p62, a membrane-associated RNA-binding protein, and p190Rho^GAP, a protein that possesses GTPase-activating activity towards proteins in the Rho/Rac family. As discussed above (see Figure 3), p120Ras^GAP may serve to co-ordinate the activity of the Ras and Rho signalling pathways [9]. The N-terminal domain of p120Ras^GAP can modify cellular cytoskeletal structure (disruption of the actin cytoskeleton and focal contacts), possibly as a result of its interaction with p190Rho^GAP [50]. NF1 may also have a signalling function separate from its GTPase-activating ability, since it can inhibit transformation by v-ras, even though the GTPase activity of v-Ras cannot be stimulated by NF1 [51]. Deficits in NF1-GAP activity have been detected in a number of human tumours, including von Recklinghausen's neurofibromatosis, consistent with an important signalling role.

SHP-2 (the recently proposed name for the independently isolated but identical SYP, PTP1D, SH-PTP2, and PTP2C protein tyrosine phosphatases [52]) is a protein tyrosine phosphatase that contains two N-terminal SH2 domains through which it can associate with tyrosine-phosphorylated receptors, e.g. the PDGF receptor, and become activated. Phosphorylation of a C-terminal tyrosine in SHP-2 produces a docking site for GRB2, thus facilitating activation of SOS [53–55]. SHP-2 can activate Src by dephosphorylating Tyr-527 in inactive Src. When phosphorylated, Tyr-527 inhibits Src activity by forming an intramolecular complex with the Src SH2 domain [55]; freed, the Src SH2 domain can, for example, interact with an appropriately phosphorylated PDGF receptor (see Figure 4). (Alternatively, interaction of the Src SH2 domain with the phosphorylated receptor, should it be able to compete, would expose the Tyr-527 phosphate to the action of SHP-2 [3].) GRB2 has been reported to bind tyrosine-phosphorylated focal adhesion kinase, thus providing a potential link between integrin-mediated signalling and Ras-mediated signalling. This may explain the activation of the p42 and p44 ERKs when integrins are engaged by their ligands [56,57].

PI3-K is a dimeric enzyme (subunits of 110 and 85 kDa) capable of phosphorylating suitable phosphoinositides in the 3-position; it is also a serine/threonine protein kinase able to phosphorylate various proteins, including its own regulatory 85 kDa subunit and the insulin receptor substrate-1 [58,59]. The relationship between the lipid- and peptide-binding sites is not known. p85 possesses one SH3 group, two SH2 groups and a domain that may have GAP activity towards an unidentified target. It probably activates the kinase holoenzyme by an allosteric effect and by fostering relocation of the p110 catalytic subunit to the membrane. Tyrosine phosphorylation of the p85 subunit down-regulates PI3-K activity. p110 can also interact directly with activated, membrane-bound, Ras and thereby itself be activated. One function of the 3-phosphorylated phosphatidylinositol phosphates is to participate with Rac in inducing membrane ruffling. PtdIns(3,4,5)P₃ interacts with the SH2 domains of some proteins, including p85 and pp60^c-src, competing directly with receptor protein tyrosine phosphates for binding to these proteins [60]. PtdIns(3,4,5)P₃ also directly activates protein kinase (PKC) z, thereby stimulating phosphorylation of serine/threonine residues in proteins targeted by this kinase in a calcium-and diacylglycerol-independent process [61,62].

Src is a protein tyrosine kinase that is attached to the plasma membrane by its N-terminal myristate modification and is essential for the mitogenic action of certain receptors, e.g. that for PDGF. When activated, it phosphorylates a number of cytoskeleton-associated proteins (focal adhesion protein tyrosine kinase, paxillin) and probably contributes to the activation of both Raf-1 and PI3-K [63]. v-Src has been found to activate phospholipase D (PLD) by inducing Ras, possibly by phosphorylation of SHC with subsequent activation of GRB2/SOS, to activate a Ral-GDS activity, the exchange factor responsible for the activation of Ral [64] (see Figure 3). PLD is found in a complex with Ral-A, and when it is activated as a consequence of the Ras–Ral interaction it generates several lipid second messengers, e.g. phosphatidic acid derived from the action of PLD on phosphatidylcholine. v-Src augments both the Ras/MEKK1/JNK and Ras/Raf-1/ERK pathways, which among other actions contribute to the activation of a cAMP response element in the mouse prostaglandin synthase-2 promoter to enhance transcription of the gene [65].

Adaptor proteins can mediate phosphotyrosine interactions

SOS is a cytoplasmic Ras-GEF that is typically constitutively associated with the adaptor protein GRB2 and can be stimulated to activate Ras in various ways [66]. When GRB2 interacts with a tyrosine-phosphorylated membrane receptor, it positions SOS at the plasma membrane where it can promote activation of Ras. Emphasizing the importance of the cellular location of an enzyme is the observation that SOS derivatives engineered to localize at the membrane are able to activate Ras in a receptor-independent manner and to transform cells. Cell transformation by SOS (achieved by targeting SOS to the cell membrane by providing it with a farnesylation or myristoylation signal) is accentuated by deletion of the C-terminal region, suggesting that this region can fold to inhibit the interaction of SOS with Ras [67]. Alternatively, instead of GRB2 binding directly to a tyrosine-phosphorylated receptor such as the activated EGF receptor (Figure 5a), the SOS–GRB2 complex may interact with a phosphotyrosine on SHC, which in turn can associate with a receptor phosphotyrosine, as illustrated for the NGF receptor TrkA (Figure 5b). The importance of the SHC–GRB2–SOS pathway in ErbB2 (p185, or HER-2/neu) signalling was demonstrated by the finding that GRB2 in which the N-terminal SH3 domain was deleted was able to block substantially the oncogenic signal from activated ErbB2; although still able to bind SHC, the DN-GRB2 could no longer bind SOS [68].

The SHC gene encodes several variant proteins that become tyrosine phosphorylated when various receptors are activated. SHC can itself transform fibroblasts and promote PC12 differentiation in a Ras-dependent fashion. It appears that SHC first binds to a tyrosine-phosphorylated receptor, which phosphorylates a tyrosine in SHC that can then serve as bait for GRB2 docking and consequent SOS activation. The FGF receptor-1 stimulates the GRB2–SOS complex not only via their mutual interactions with SHC but also via p89, another adaptor protein that is in addition membrane-associated [69]. Although the SH2 domain of SHC has been considered as its primary receptor interaction domain, a recently identified PTB domain that recognizes NPXY(P) may mediate this interaction instead [70–72]. By computer analysis, Bork and Margolis [36] detected this domain in a number of other proteins and designated it PID (for phosphotyrosine interaction domain).

Signalling via the heterotrimeric G-proteins

Activation of heterotrimeric G-proteins (consisting of a, b and g subunits), typically by stimulated serpentine (7-pass) membrane receptors [74], is effected by the replacement of GDP with GTP on the Ga component and the consequent dissociation of Ga from the membrane-bound Gbg heterodimer. Like Ras, Ga is active only when bound to GTP; it is inactivated by its intrinsic GTPase. Research during the past few years has revealed that signals from the heterotrimeric G-proteins, particularly those stimulated by pertussis toxin-inhibitible receptors (G_i), can impact on the Ras/MAPK signal transduction cascade [75–77]. The Gbg subunit, by virtue of its ability to interact with certain PH domains, may influence the activity of either SOS or Ras-GAP, both of which have PH domains, perhaps by attracting them to the plasma membrane. In COS-7 cells, agonists acting on G_s-, G_q- or G_i-coupled receptors stimulated p44^MAPK (ERK1) expressed as an epitope-tagged molecule in transiently transfected cells; both the Ga and Gbg subunits were implicated in the signalling process [78].

Several groups have transfected COS-7 cells with various expression vectors in order to elucidate the mechanisms of G-protein-coupled receptor action. Crespo et al. [79] found that Gbg generated by stimulation of a muscarinic m2 receptor activated ERK2 via a Ras-dependent pathway, possibly involving SHC. Touhara et al. [80] used anti-SHC antibodies to demonstrate directly that the pertussis toxin-sensitive G_i-coupled receptor (the a2-C10 adrenergic receptor) stimulated tyrosine phosphorylation of the 50 kDa SHC protein via an action of Gbg. These and other results established that the Gbg heterodimer, or at least certain species thereof [77], can stimulate phosphorylation of SHC, facilitating the association of the latter with GRB2–SOS, activating Ras and resulting in Raf-dependent ERK activation [81]. This same group also reported that, in Rat 1 fibroblasts, activation of the IGF receptor by IGF-1 also required input from Gbg subunits (e.g. derived from the G_i-coupled receptor for lysophosphatidic acid) in order for full ERK activation to occur [82]. Signalling from the b-adrenergic receptor, assessed by ERK activation, involves both a positive signal conveyed by Gbg acting on a Ras-dependent pathway and a negative signal resulting from activation of adenylate cyclase by Ga_s. The increase in cAMP activates protein kinase A (PKA), which suppresses activation of the Ras pathway by an inhibitory phosphorylation of Raf-1 [77,83].

The PH domain (named for its presence in the abundant platelet protein pleckstrin) is an approx. 120-amino-acid module with a characteristic structure; it is found in many signalling proteins and mediates protein–protein interactions and membrane attachment [84]. The membrane-bound Gbg complex is believed to interact with PH domains in target proteins, causing them to relocate to the cell membrane [76]. The PH domain has also been implicated in the binding of both PtdIns(4,5)P₂ [85–87] and PKC (both calcium-dependent and calcium-independent isoforms) [88]. The PH domain in PLCd₁, but not in certain other proteins, can bind to Ins(1,4,5)P₃ with stereospecific high affinity [89]. Lee and Rhee [90] have reviewed the significance of PtdIns(4,5)P₂ hydrolysis by PtdIns-specific PLC isoenzymes with regard to the role of PtdIns(4,5)P₂ in regulating PLC isoenzymes, activating PLD and controlling actin polymerization. There are four PLCb, two PLCg and four PLCd enzymes, all possessing a PH domain near the N-terminus. The PLCg isoenzymes are activated by receptor (and non-receptor) tyrosine kinases, which phosphorylate three tyrosines, one of which (Tyr-783) is essential for activating the phospholipase activity and also for association with the cytoskeleton. The PLCb isoenzymes are activated by GTP-bound a_q and bg subunits of receptor-activated Gabg. Activated PLC not only generates diacylglycerol and Ins(1,4,5)P₃ but also reduces the level of PtdIns(4,5)P₂ and consequently the number of membrane attachment sites for proteins with PH domains.

THE p21 SIGNALLING PATHWAYS

Ras is normally found attached to the inner face of the plasma membrane by the lipid interactions discussed above. Mobilized Ras–GTP is capable of interacting with and activating a number of proteins; known direct targets include p120Ras^GAP, the p110 subunit of PI3-K, Ral-GDS, PKCz, Raf-1, A-Raf, B-Raf, and possibly certain MEKKs [91,92]. Each of these proteins is likely to be activated by Ras–GTP via a unique interaction with a particular site on the p21 protein, and each contributes to the overall consequence of Ras activation. With regard to interaction sites, Joneson et al. [93] reported recently that the mutation Tyr-40 Cys abrogated ERK activation but not membrane ruffling, whereas the Thr-35 Ser mutation impaired membrane ruffling but not ERK activation when quiescent fibroblasts were transfected with the indicated mutant of Ha-Ras. This is consistent with the idea that distinct downstream consequences of Ras activation may be caused by different Ras effectors generating different signals.

Figure 6 shows a generic schematic of the Ras Raf MEK ERK pathway, along with what is known about two related pathways that for the most part operate independently of Ras. These pathways are not equally activated by any single stimulus. Growth factors and mitogens preferentially activate ERK1 and ERK2, whereas inflammatory cytokines and various forms of stress preferentially activate SAPK/JNK and/or p38/RK (see Table 1 for details of nomenclature). Different laboratories using different cells and different activating strategies have identified various different components in these cascades and have often given them different names. The 'take-home message' is that there are several parallel tracks down which a signal can be transmitted. These three phosphorylation cascades are activated to different extents depending upon the particular stimulus and the various other signalling components that are activated (e.g. PI3-K, PKC isotypes, non-receptor Src-family tyrosine kinases and phospholipases). The degree of activation of a particular signalling pathway depends upon the number of activated receptors and the specific upstream effectors that are activated. Phosphorylation of specific residues in some proteins may be inhibitory, thus allowing modulation of the signal intensity. As discussed later (see Figure 9), if different phosphorylated sites in a protein convey separate signals, for example to phosphorylate different kinases, then one has the potential for a pathway to transmit more than one signal simultaneously (multiplex signalling).

The categorizing of various 'levels' in the different pathways in Figure 6 more likely reflects a human desire than nature's intent to arrange things this systematically. A comparison of these pathways with the five or so known pathways in yeast suggests that additional signalling cascades remain to be defined in mammalian cells [94,95]. Consistent with this thought is the existence of several enzymes that might be parts of as-yet-undefined mammalian pathways (e.g. ERK3 [96] and MEK5 [97]). Although 'mitogen-activated protein kinase' has been used as a generic term to describe collectively all of the MAP kinases that are activated by dual phosphorylation in Figure 6, only a subset of them (the ERKs) are typically strongly activated by a proliferative stimulus.

The determination of which proteins are true in vivo substrates for the various kinases is not trivial. Conclusions drawn from experiments with purified enzymes and substrates can be extrapolated to the intracellular situation only with certain caveats, particularly concerning protein concentrations and macromolecular associations. Since the concentrations of a particular kinase and its substrate, and their mutual affinity, determine whether that substrate is phosphorylated, in vitro experiments should in principle be performed using in vivo concentrations, which of course cannot easily be known if the proteins are not homogeneously distributed in the cell. There is also the potential that consecutive elements in a phosphorylation cascade exist in a specific complex, thus providing exquisite specificity with regard to transmission of the signal. Furthermore, if the signalling molecules are normally organized in a complex (a 'signalsome') with a particular scaffold protein present in limiting quantities, then results obtained using transient transfection protocols may not mimic the real situation in vivo, thus further clouding the interpretation of various experiments [93]. Signals transmitted within a complex of proteins will not be amplified. Amplification does occur at the level of the downstream effectors, particularly the MAPKs, and also as the result of positive synergistic interactions of individual signalling pathways.

The 'MAPK4' level

This 'level' encompasses everything between the receptor and MAPK3, i.e. mitogen-activated protein kinase kinase kinase. In this group, which is not as well defined a level as those that follow, are the GAPs and GEFs, the PKC isoforms, PI3-K and p65^pak. One could argue that PKA and assorted phospholipases should be included also.

The serine/threonine PKC family is divided into three subgroups depending on whether both Ca²⁺ and diacylglycerol are required, whether only diacylglycerol is required, or whether some other lipid (e.g. ceramide) is required [59,62,98]. Second messengers generated from the breakdown of various membrane lipids, especially phospholipids, enhance PKC activity, for example by recruiting the cytosolic protein to the plasma membrane or stimulating proteolysis of the inhibitory pseudosubstrate. The tumour-promoting phorbol esters, analogues of the natural activator diacylglycerol, are potent activators of most PKC isoforms. PKC is regulated by phosphorylation, which is necessary for activity, and by lipids such as phosphoserine and diacylglycerol, the product along with Ins(1,4,5)P₃ of the action of PtdIns-specific PLC on PtdIns(4,5)P₂. This PLCg is activated by many receptor tyrosine kinases via its SH2-mediated association with the appropriate phosphotyrosine (Figure 4), and it may make a definite contribution to the mitogenic response. In vitro the PKC isoforms are rather promiscuous with regard to which substrates, typically Arg-rich proteins, they will phosphorylate. Although the critical targets of the various PKC isoforms in vivo remain for the most part to be discovered, it is clear that PKC is an important constituent of many signalling pathways.

PAK (p65^pak; p21-activated kinase) was discovered as a brain serine/threonine kinase that is a potential downstream target for Rac-1 and CDC42 signalling. PAK binds to the activated GTP-bound p21 proteins CDC42 and Rac-1, inhibiting their intrinsic GTPase activity and promoting its own phosphorylation [99]. Because PAK-I is active in Xenopus oocytes and in quiescent and serum-starved cells, and because (when microinjected) it can inhibit cleavage in blastomers, Jakobi et al. [100] have suggested that PAK-I may be involved in the regulation of cytostasis and the response of the cell to stress. PAK is activated in phagocytes as a result of its interaction with Rac–GTP, and it can phosphorylate the p47^phox NADPH oxidase in a Rac–GTP-dependent manner. Rac–GTP is itself generated via an interaction with a Ga protein activated, for example, by a chemoattractant receptor [101]. Coso et al. [102] have suggested that PAK might be an intermediate linking Rac-1 and CDC42 to MEKK1 on the basis of their observation that mutationally activated CDC42 or Rac-1 stimulated JNK/SAPK, but not ERK, activity. In vivo experiments with constitutively activated CDC42 and PAK3 revealed that PAK3 was activated by CDC42, and that JNK1 and p38 were in turn activated more efficiently than ERK2 [103]. Studies by Minden et al. [104] and Qiu et al. [105] using constitutively active or dominant interfering alleles of the various signalling intermediates placed Rac-1, possibly followed by PAK, downstream of Ras and upstream of MEKK1 in a signalling cascade resulting in JNKK and p38 activation; attempts to locate CDC42 in the pathway were inconclusive. Studies of interleukin-1 (IL-1) signalling suggested a pathway (Rac/CDC42) PAK (unidentified intermediates) leading to p38 and JNK activation [106]. The observation that a dominant negative CDC42 mutant inhibited IL-1 signalling was consistent with such a pathway [103]. Mutationally activated Rac-1 and CDC42 GTPases, and also the GEFs for these Rho family proteins, activate JNK/SAPK without affecting the ERKs [102] (see Figure 3). In addition to PAK, CDC42 may activate activated CDC42-associated kinase (ACK), a non-receptor tyrosine kinase that inhibits the GTPase activity of CDC42, and PI3-K [9,10].

The 'MAPK3' level

Although Ras can interact with a number of effector proteins in a GTP-dependent manner, it is likely that there will be competition among effectors, with certain interactions being preferred over other interactions. Each of these Ras effectors also has its own differing target protein affinities. Thus A-Raf can activate MEK1 but not MEK2 in EGF-stimulated HeLa cells [107]. Ras and Raf-1 form a signalling complex with MEK1 but not MEK2 [108], whereas the Ras–MEKK1 and –MEKK2 interactions preferentially activate JNKK [109]. MEKK1, 2 and 3 are all able to stimulate the ERK and SAPK/JNK pathways, although MEKK1 and 2 preferentially stimulate SAPK/JNK, whereas MEKK3 preferentially activates MEK/ERK [92]. MEKK1 is activated in resting PC12 cells by EGF, and to a lesser extent by NGF, phorbol 12-myristate 13-acetate and oncogenic Ras [110]. EGF-induced activation of MEKK1 and B-Raf can be inhibited by a dominant negative Ras and also by PKA activation (which inhibits Raf). Both the Raf-1 and MEKK1 pathways are stimulated in rat 3Y1 fibroblasts by hyperosmolar (0.5 M NaCl) shock, with consequent activation of ERK1 and ERK2 (p43/44^MAPK and p41/42^MAPK) and the SAPKs (p54 and p46) [111]. How osmotic shock activates these pathways, including that leading to p38 phosphorylation, and the extent to which Ras is involved, is not known.

Downstream elements in Ras-controlled signalling pathways may be activated by various forms of stress-initiated signals that do not function by activating Ras itself. However, Ras does appear to be an important sensor of oxidative stress. Lander et al. [112] observed in PC12 and Jurkat cells that inhibitors of Ras function (a dominant negative mutant p21 and an inhibitor of farnesyltransferase) blocked nuclear factor kB (NFkB) activation by various oxidative agents that modulate cellular redox stress (peroxide, haemin, Hg²⁺). The authors concluded from these and other experiments that free radicals can activate p21^ras and generate a nuclear signal. Other studies have found that the MEK/ERK pathway is activated in neutrophils exposed to an oxidant (diamide or peroxide with a catalase inhibitor) [113] and in HeLa cells exposed to lysophosphatidic acid, which is believed to elevate reactive oxygen species via lysophosphatidic acid-induced arachidonic acid release [114]. However, in these latter two studies the mechanism of MEK activation was not determined.

The interaction of MEKK1 with Ras, at least in the test tube, involves the kinase domain of MEKK1 and the effector domain of Ras. When MEKK1 is complexed with activated Ras it is capable of phosphorylating MEK1 [115]. Xu et al. [116] have shown that a recombinant fragment of MEKK1 containing the catalytic domain could phosphorylate MEK1 and MEK2, and that MEK1 and MEK2 could be activated when co-transfected with MEKK1. However, ERK2 was not activated to the full extent expected, possibly, the authors suggest, because of an inhibitory effect of the up-regulated MEKK1 activity upon ERK2 activity. The relevance of these in vitro observations to the in vivo situation remains unclear, since it is possible that the highly specific interactions of MEK1/2 with Raf-1 and with ERK1/2 are the consequence of the formation of macromolecular complexes. It remains unclear to what extent Ras can activate MEKK1 directly, or whether it normally proceeds indirectly via Rac and PAK, or even by a paracrine process involving a secreted extracellular mediator.

In proliferating cells, transforming growth factor-b is usually a negative regulator of proliferation. However, this factor can deliver a mitogenic signal to quiescent (confluent, serum-starved) 3T3 cells in the absence, surprisingly, of detectable tyrosine phosphorylation of the 41 or 43 kDa ERKs (ERK activation typically is associated with a proliferative stimulus) [117]. A kinase called TAK1 has recently been identified as a possible mediator of transforming growth factor-b and bone morphogenetic protein signalling; consistent with this possibility was its inability to activate the MEK ERK pathway and its ability (in vitro at least) to activate SEK1 [118]. Another recently identified MEKK, a proto-oncogene called Tpl-2, is of interest in that it has the potential to activate, independently of Ras and Raf, both MEK1 and SEK1 (but not p38/RK); how Tpl-2 is activated is not known [119]. The so-called mixed-lineage kinases have sequence motifs that are associated with both serine/threonine kinases and tyrosine kinases, and at least one of them (called MUK) has the properties of a MEKK that is a selective activator of the SAPK/JNK pathway [120,121].

Reuter et al. [122] have recently characterized several MEK activators, including the 95 kDa B-Raf and an unidentified 40–50 kDa MEK activator, and shown that they, rather than Raf-1, are the major serum-stimulated MEK activators in fibroblasts. (Lysophosphatidic acid, which stimulates G-protein-coupled receptors, may be the active serum component [122a].) Raf-1 does respond well to both PDGF and phorbol esters. In PC12 cells B-Raf was responsible for the NGF-stimulated, p21^ras-mediated, activation of MEK [123]. Pritchard et al. [124] expressed the protein kinase domains of Raf-1, A-Raf and B-Raf coupled to the oestrogen receptor and found that after oestradiol activation the B-Raf construct was more effective than the Raf-1 or A-Raf constructs in activating MEK and the ERKs. All three constructs could (after oestradiol activation) morphologically transform 3T3 cells, but only the A-Raf construct could cause quiescent (G_o) 3T3 cells to enter the cell cycle.

The interaction of Ras with Raf-1, illustrated in Figure 7, has been extensively studied [125]. Raf-1 is normally located in the cytosol in association with Hsp90 (heat-shock protein of 90 kDa) and Hsp50, which appear to stabilize the protein and facilitate a proper interaction with Ras [126]. Activation of the 74 kDa Raf-1 protein by Ras involves relocation to the cell membrane via direct association of two regions in the N-terminal conserved region 1 of Raf-1, a previously recognized Ras-binding domain and a more recently recognized cysteine-rich region, with the switch 1 and switch 2 domains respectively of Ras [127,128]. Full activation of Raf-1 requires both an interaction with the plasma membrane and phosphorylation of key amino acids [129–131]. Phosporylation events include tyrosine phosphorylation by membrane-bound Src [132] and serine/threonine phosphorylation by a PKC isoform [133] or by KSR (kinase suppressor of Ras) [134]. The latter is a protein kinase with some similarity to Raf-1 that has been implicated in Ras signalling either upstream of or in parallel to Raf-1. Unlike Raf-1, which requires Ras-induced membrane localization and tyrosine phosphorylation, B-Raf contains aspartic acids in place of the phosphorylatable tyrosines (at 340 and 341) and is not localized to the plasma membrane upon Ras activation [135].

Sites of serine phosphorylation on Raf-1 include Ser-43, Ser-259 and Ser-621; the latter two occur in a consensus sequence RSXXP, which is conserved among all Raf family members [136]. Phosphorylation of at least some of the serine/threonine residues [those just mentioned and also Ser-499 and Thr-268 (the site of in vitro autophosphorylation)], and Tyr-340 and Tyr-341 is essential for complete Raf-1 activation [137]. Ser-43 and Ser-621 are both constitutively phosphorylated, whereas Ser-259 is phosphorylated in response to PDGF. The conformational changes induced in Raf-1 activate its serine/threonine kinase activity, which is specific for Ser-218 and Ser-222 in human MEK1 and for comparable conserved residues in other MEK family kinases [138]. The specificity appears to be provided by a proline-rich sequence in MEK1 and MEK2 that is missing from the other MAPKKs (JNKK/SEK1/MKK4 and RKK/MKK3, described below) and appears essential for the interaction of the MEKs with Raf-1 and downstream signalling proteins [139]. Other possible in vivo Raf-1 substrates include the NFkB inhibitor IkB [140], p53 [141] and the dual-specificity phosphatase CDC25A [142]. Thus Raf-1 activation can directly stimulate the expression of genes under the control of NFkB and p53, and can initiate cell cycle progression by dephosphorylation and activation of the cyclin kinases. Two of the three CDC25 isoforms associate with Raf-1 and are phosphorylated when Ras is activated, perhaps accounting for the synergism between CDC25A or CDC25B and Ras in transforming primary cells [143]. CDC25 is a dual-specificity phosphatase that regulates CDC2 kinase, activating CDC2 by dephosphorylating adjacent pThr-14 and pTyr-15 residues.

Hafner et al. [144] showed in studies with purified proteins that phosphorylation of Raf-1 by PKA was inhibitory both to Ras binding and to the functioning of the kinase domain. However, in vivo the situation becomes more complex. For example, Raf-1 and the downstream signalling pathway are inhibited by cAMP in fibroblasts, whereas in PC12 cells cAMP synergizes, in the presence of serum, with NGF to enhance neurite outgrowth induced by B-Raf. This difference in sensitivity is the result of the presence only in Raf-1 of Ser-43 in a consensus site for phosphorylation by PKA [145]. Because Raf-1 is rapidly and negatively regulated by PKA phosphorylation in vivo, Burgering and Bos [77] have suggested that Raf-1 activation is a reiterative process and that Ras action is repeatedly required to maintain Raf-1 activity. Raf-1 is also subject to an activating phosphorylation on Thr-269 by a membrane-bound and proline-directed kinase that is activated by ceramide, the second messenger of the sphingomyelin pathway [146]. Both tumour necrosis factor a (TNFa) and IL-1b signal through this pathway, enhancing the activity of Raf-1 towards MEK; presumably Ras–GTP is required to localize Raf-1 to the cell membrane, although this was not investigated. If the different phosphorylations on Raf-1 send different signals downstream, differentially activating MEK1 or MEK2 for example, then Raf-1 would be an example of a multiplex signalling element.

Proteins known as 14-3-3 proteins are typically acidic, dimeric proteins of about 60 kDa that are found in a broad range of organisms. Several isoforms are known, some of which are simply phosphorylation variants. They are found to associate with, and often to modify the activity of, a number of other proteins (e.g. tyrosine hydroxylase, PKC), not unlike chaperones [147]. The catalytic p110 subunit of PI3-K binds to and is inhibited by 14-3-3 [148]. Various isoforms of 14-3-3 have different effects on the different species of PKC; both activation and inhibition of PKC isoforms by preparations of 14-3-3 have been reported [149]. Binding of 14-3-3 to target proteins is via a phosphoserine residue in the motif RSXSXP [150], identical to the conserved consensus sequence for serine phosphorylation in Raf-1 noted above [136]. 14-3-3 associates efficiently with Raf-1 (via the constitutively phosphorylated Ser-621), but does not activate its kinase activity in immunoprecipitates [151]. The association of 14-3-3 (b and z isoforms) with inactive Raf-1 may stabilize that conformation, and upon Raf-1 activation it may facilitate activation, perhaps via a specific interaction with the kinase domain [152,153]. As indicated in Figure 7, there are two sites of interaction of 14-3-3z with Raf-1, one of which can be competed for by activated Ras [154]. Dent et al. [137] showed that both 14-3-3z and Hsp90 could inhibit the inactivation of purified Raf-1 by either serine/threonine or tyrosine protein phosphatases. Because of the dimer structure of 14-3-3, the protein has the potential to foster interactions between proteins to which it can bind, and Braselmann and McCormick [154a] have demonstrated that 14-3-3 can indeed perform such an adaptor function, generating a complex with BCR and Raf, particularly the membrane-bound form. This may be important in facilitating the action of BCR on target proteins (see Figure 3).

Buscher et al. [155] investigated in the macrophage-like line BAC-1.2F5 the signal transduction pathways used by CSF-1 and lipopolysaccharide (LPS) to induce phosphorylation of the two ternary-complex factors Elk and SAP. Although CSF-1, a mitogen, stimulated Ras–GTP formation, Raf-1 activation curiously did not seem to depend on Ras activation, as judged by its insensitivity to a dominant negative Ras. LPS, a stimulator of macrophage differentiation, clearly did not utilize Ras at all, and instead activated Raf-1, MEK and ERK via a PKC and phosphatidylcholine-specific PLC pathway. Elk, but not SAP, activation was dependent upon ERK activation. The CSF-1 receptor appeared to activate Src, which in turn activated Raf-1, probably by tyrosine phosphorylation; full mobilization of the downstream ERKs and Elk was accomplished synergistically with other signals generated simultaneously by the CSF receptor. The LPS receptor CD14 was also able to induce activation of Src family kinases.

The 'MAPK2' level

The MAPKKs are typically activated by serine/threonine phosphorylation catalysed by one of the MAPK3s (Raf, MEKK). The MAPKKs (MEKs, MKKs, JNKKs, SEKs, RKK; see Figure 6 and Table 1) are a subset of protein kinases that possess unique dual (or mixed-lineage?) kinase specificities. When activated (e.g. by phosphorylation of Ser-222 in MKK1a [156]) the MAPKKs phosphorylate only very specific native MAPKs on both Thr and Tyr residues in a Thr-Xaa-Tyr motif, where Xaa is Glu, Pro or Gly.

One subgroup of the MAPKKs includes MEK1a, MEK1b and MEK2, which phosphorylate different ERKs [157] but nevertheless appear to elicit similar transcriptional and morphological responses [158]. Using dominant negative and constitutively active forms of MEK1a (MKK1a), Seger et al. [156] demonstrated that this signalling element controlled NIH-3T3 cell proliferation and morphology. Mansour et al. [159] have shown that NIH-3T3 cells expressing constitutively active forms of MEK were transformed, as judged from the formation of both foci in culture and tumours in nude mice. Cowley et al. [160] have shown that, in the appropriate cell context, activation of MEK is necessary and sufficient for growth-factor-induced proliferation (NIH-3T3 cells) or differentiation (PC12 cells).

A second subgroup of these dual-specificity kinases includes SEK/JNKK/MKK4. These are preferentially activated by MEKK1 and 2, but not by the Raf proteins, and their target proteins are the SAPKs/JNKs [161,162]. JNK was identified on the basis of its ability to phosphorylate the N-terminal region of c-Jun on Ser-63 and Ser-73; its activity is stimulated by inflammatory mediators (e.g. LPS, interferon, TNFa, IL-1) and UV light, and can be further augmented by oncogenic Ras [163]. The SAPKs were identified independently as protein kinases that are activated by stress (e.g. heat shock, inflammatory cytokines, inhibitors of protein synthesis, and DNA-damaging agents such as UV light, ionizing radiation, arabinosylcytosine, alkylating agents and topoisomerase inhibitors). For some of these the activating signal is conveyed by the c-Abl non-receptor tyrosine kinase, a nuclear and cytoplasmic protein that shares structural features with Src and activates SAPK via SEK1 [164]. The presence of both SAPK and SEK1 in the nucleus suggests that this cascade may be activated independently of strictly cytosolic or plasma-membrane-bound kinases [164]. Activated (e.g. by AraC) SAPK binds to the SH3 domain of GRB2, which in turn can bind to the (tyrosine phosphorylated) p85 subunit of PI3-K via its SH2 group, forming a SAPK–GRB2–PI3-K complex that effectively inhibits the lipid kinase and protein serine kinase activities of PI3-K [165]. The significance of this may be to favour the SEK SAPK pathway over the MEK ERK pathway.

The JNKs and SAPKs are the same set of proteins, and they are activated by SEK1, the same enzyme as JNKK [166]. Because of the greater efficiency with which MEKK1 activates JNKK/SEK/MKK4 as compared with MEK, it appears that (despite its name) these may be the normal substrates for MEKK1 in the cell, while the 'true' activators of MEK are the Rafs and possibly MEKK3. Since MEKK1 can at high concentrations activate MEK, it is not excluded that there are circumstances whereby MEK is a substrate for MEKK1 in vivo. Ras and the Ras-coupled agonist Raf-1 are inefficient activators of SEK and SAPK. Thus there appear to be two pathways under the influence of Ras (see Figure 6): (1) the Raf MEK/MKK ERK pathway activated by growth factors, mitogens and tumour promoters that is highly dependent upon Ras activation; and (2) the MEKK SEK/JNKK/MKK4 SAPK/JNK pathway that can be stimulated through Ras but is more strongly activated by various stressors and cytokines in a largely Ras-, Raf- and MEK-independent manner.

MKK3, along with MKK4, was first cloned from human brain tissue [167]. It, and the closely related MKK6, appear to be related to a yeast gene involved in responding to hyperosmotic conditions (pBS2), and are responsible for phosphorylating p38, which was cloned on the basis of its enhanced phosphorylation in a mouse pre-B cell line by LPS [168,169]. The sequence of p38 most closely resembles that of the yeast kinase HOG-1, which is involved in adaptation to osmotic stress and is the substrate for pBS2. IL-1 and hyperosmolar conditions also induce p38 phosphorylation. MKK3/6 preferentially phosphorylate p38, whereas MKK4 preferentially phosphorylates JNK [166,170]. p38, which is not significantly activated via Ras-controlled pathways, is efficiently activated by MKK3 and MKK6, and can in turn activate both ATF-2 and Elk-1 [171].

The MAPK level

The activated serine/threonine protein kinases collectively referred to here as 'MAPKs' constitute a superfamily of proteins that includes the ERKs, JNKs/SAPKs and p38/RK MAP kinases. These are uniquely identified by the Thr-Xaa-Tyr dual-phosphorylation motif, where Xaa is respectively Glu, Pro and Gly for the ERKs, JNKs/SAPKs and p38/RK/CSBP. Phosphorylation of both tyrosine and threonine residues, which are found in the activation segment of the kinase domain, is essential for full kinase activity of the MAPKs [172]. (In the literature, 'MAPK' may refer either specifically to the ERKs or more generally to this superfamily.) Upon activation, the activated MAPK may be translocated to the nucleus (if it is not already there), where it can phosphorylate targeted transcription factors [e.g. c-Jun, SAP-1, Elk-1, c-Myc, activating transcription factor (ATF)-2 and the cAMP response element-binding protein (CREB)]. [SAP-1 was cloned as the serum response factor (SRF) accessory protein [173]; it is not a 'stress-activated protein' and is not activated by JNK.] The MAPKs are proline-directed in the sense that they target only serine and threonine residues that are closely followed by one or more prolines in a motif recognized preferentially by a particular MAPK. Whereas the ERKs are strongly activated by one-pass receptor tyrosine kinases (e.g. the PDGF, EGF, CSF, NGF and FGF receptors), the JNK/SAPKs are potently activated by heterotrimeric-Gabg-protein-coupled receptors, such as the muscarinic acetylcholine receptor, in a Ras- and PKC-independent manner that also does not involve the ERKs [174].

ERK1 (p43/44^MAPK) and ERK2 (p41/42^MAPK) have been extensively studied; they are discriminating with regard to the target proteins they phosphorylate, recognizing a proline domain that identifies the substrate and producing (usually) multiple phosphorylations on a Ser/Thr-Pro motif. ERK1 targets Elk-1, a TCF that associates with the SRF on the serum response element, whereas ERK2 exhibits a preference for c-Myc [175]. Potential cytoplasmic targets of ERK1/2 phosphorylation include cytoskeletal elements (microtubule-associated proteins, tau), various kinases [ribosomal protein S6 kinase (both p70^s6k and p90^rsk), cytoplasmic phospholipase A2, SHP-2, glycogen synthase kinase-3 (GSK-3) and protamine kinase], and upstream signalling elements (e.g. the EGF receptor, SOS, Raf-1 and MEK) [157]. The extent to which phosphorylation of upstream signalling elements reflects significant negative-feedback regulation remains to be determined. Elk-1 and SAP-1 bind to an Ets domain in the serum response element in the c-fos promoter along with the SRF. Serum induction of c-fos transcription, stimulated by lysophosphatidic acid acting through a serpentine receptor and a heterotrimeric G-protein, is the result of ERK-induced phosphorylation of these serum-response-element-bound proteins (Elk-1 or SAP-1; SRF) [176]. Elk-1 is also phosphorylated by the JNKs, which are activated by MEKK1 in response to UV radiation, heat shock, protein synthesis inhibition and other forms of stress [177].

Mouse p38 (RK) is a kinase at the MAPK level with the unique Thr-Gly-Tyr motif as the site of Thr/Tyr phosphorylation, which is accomplished by the MKK3/RKK dual-specificity kinase [178,179]. Raingeaud et al. [170], working with HeLa and COS cells, reported that UV light, osmotic shock, LPS, TNFa and IL-1 activated JNK and p38, but not the ERKs, whereas EGF and phorbol 12-myristate 13-acetate gave maximal activation of the ERKs but only a modest activation of JNK or p38. p38 did not phosphorylate cytosolic phospholipase A2, c-Myc or c-Jun, whereas it did phosphorylate ATF-2 and small heat-shock proteins, and stimulated expression of inflammatory cytokines. There are two versions of the human p38 protein, CSBP1 and CSBP2, that are splice variants differing in an internal 25-amino-acid sequence; when expressed in yeast, the protein kinase activity of CSBP1, but not CSBP2, was increased under hyperosmolar conditions [180].

The outline of several parallel pathways leading to activation of different subsets of MAPKs is taking shape (Figure 6), although the extent to which there is cross-talk among these pathways is unknown [181–183]. In some cases apparent cross-talk can instead result from the induced secretion of a protein or hormone that acts back on the cell in a paracrine manner [184]. In U937 human leukaemic cells, transient expression of a constitutively active MEK1 stimulated SAPK as well as ERK activity by an apparent intracellular signalling process [185]. Whether or not this happens under normal signalling conditions remains to be determined. It is important to note that, usually, multiple elements are activated by a given stimulus or receptor, generating parallel signalling tracks that may or may not be self-reinforcing. One consequence of this is that loss of an individual component may have a relatively small effect. The extent to which the signal is amplified at each step is unclear. The signal is unlikely to be amplified as it passes through a multiprotein complex, for example a MAPKK–MAPK complex. Upstream (e.g. Ras, Rac) and downstream (e.g. ERKs, JNKs) elements almost certainly do amplify the signal.

The 'MAPKAPK' level

This group of MAPK-activated protein kinases consists of various kinases phosphorylated by the MAPKs. Included are upstream signalling elements whose phosphorylation may modulate the signal.

Ribosomal protein S6 kinase, p90^rsk (which apparently is capable of phosphorylating the ribosomal protein S6 only in vitro; the related p70^s6k does the job in vivo), can enter the nucleus, phosphorylate c-Fos and thus contribute to the control of transcription of genes under the control of AP-1 [157,186]. It, along with c-Myc and cytoplasmic phospholipase A2, are among the proteins phosphorylated (but not necessarily activated) by the Ras Raf MEK ERK pathway in response to the activation of the T-cell receptor [187]. Phospholipase A2, activated by a single serine phosphorylation, preferentially releases arachidonic acid, a precursor of prostaglandins and leukotrienes, from the 2-position of various phospholipids. Bohm et al. [188] reported that, in normal human melanocytes stimulated with various growth factors, p90^rsk was likely to be responsible for the activation of the transcription factor CREB by phosphorylating it. Ginty et al. [189] made somewhat similar observations in PC12 cells and primary rat cortical neurons stimulated with NGF, although in this case the Ras-dependent protein kinase responsible for activating CREB did not appear to be p90^rsk. The p90^rsk kinase has also been called MAPKAPK-1.

The authentic S6 kinase, p70^s6k, is strongly but indirectly activated by PI3-K and by PLCg1 [190]. It is considered to regulate a range of signalling pathways by phosphorylation of important target proteins in addition to the ribosomal protein S6. Stimulation of p70^s6k via PI3-K may be mediated by certain PKC isoforms or by p65^PAK; stimulation by PLC is also likely to be via PKC, a known activator of p70^s6k, via formation of diacylglycerol. Interestingly, there is an 85 kDa splice variant of p70^s6k that possesses an N-terminal addition with a nuclear localization signal that places this isoform in the nucleus. Activation of p70^s6k is blocked by the immunosuppressive drug rapamycin, which has a number of effects on the cell, including the suppression of translation of specific mRNAs and inhibition of NFkB activation. CDC42/Rac contributes to p70^s6k activation also, possibly via p65^PAK [190].

MAPKAPK-2 is a 50 kDa protein kinase that can be activated in various cells (KB, HeLa, PC12, monocytes and macrophages) by chemical or physical stress (sodium arsenite, heat shock, osmotic stress) and by certain cytokines and inflammatory mediators (IL-1 and LPS). It is responsible for phosphorylation of the small heat-shock proteins Hsp25/Hsp27. Phosphorylation of Hsp27 facilitates actin polymerization, probably contributing to the repair of stress-damaged actin microfilaments [191]. Activation, by phosphorylation, of any two of the three residues Thr-222, Ser-272 and Thr-334 is accomplished by the 38–40 kDa protein p38/RK, which is itself activated by an RKK [169,178,179,192]. Although in vitro the ERKs can phosphorylate MAPKAPK-2, this does not typically happen in vivo, in that activators of the Ras Raf MEK ERK pathway do not stimulate phosphorylation of MAPKAPK-2 and, conversely, strong activators of p38/RK do not typically cause significant phosphorylation of the ERKs. Exceptions to this include both LPS and hyperosmolality, each of which can activate both ERK and p38 [169]. MAPKAPK-3 has also been identified as a substrate for CSBP p38 that is activated by stress, LPS, IL-1 or TNFa [193].

GSK-3 has a larger role than simply that of regulating intermediary metabolism [194]. Its control of early Drosophila and Xenopus development suggests strongly a comparable role in mammalian development. Interestingly, it also has properties of a dual-specificity protein kinase in the MAPK group [195] and it functions to complete the activation of CREB previously phosphorylated by PKA on a nearby serine residue (hierarchical phosphorylation) [196]. It can also phosphorylate multiple metabolic enzymes, cytoskeletal proteins, and nuclear factors including c-Jun, CREB and p90^rsk [197]. GSK-3 itself is subject to multiple levels of control, including both serine/threonine and tyrosine phosphorylation, both activating and deactivating, by PKC and MAPKAPK-1 (p90^rsk) [198,199]. Thus GSK-3 is a candidate for a multiplex signalling element. The insulin-stimulated inhibition of GSK-3 is mediated by protein kinase B, which is also known as AKT/Rac, a serine/threonine protein kinase distantly related to the PKC family and regulated by D-3-phosphorylated phosphoinositides, products of PI3-K [200,201].

Some extracellular cytokines that do not typically activate Ras nevertheless activate downstream elements of Ras-mediated signal transduction pathways. Such cytokines include IL-1 [178], interferons [202] and TNFa [109]. For example, IL-1 activates SAPK/JNK, but not ERK1 or ERK2, by a pathway that does not require Ras, Raf-1 or MEK activation [203]. David et al. [204] discovered an interaction between ERK2 and the a subunit of the interferon a/b receptor, and suggested that ERK2 may regulate the interferon JAK STAT signalling cascade. TNFa stimulates a sphingomyelinase activity that generates the second messenger ceramide, which appears to act via MEK1 to activate the SAPKs/JNKs, and possibly NFkB [205,206]. IL-3 appears to require Ras to deliver certain signals (e.g. for apoptosis) but not certain other signals (e.g. for growth stimulation), at least in a mouse pro-B-cell line [207]. Interestingly, growth hormone appears to stimulate a signalling pathway mediated by JAK2 activation of ERK2 in a Ras- and Raf-dependent process [208]. Clearly, there is the potential for an immense variety of different signalling pathways.

SIGNAL DOWN-REGULATION AND TRANSCRIPTION UP-REGULATION

Receptor signalling can be terminated in several ways. Phosphorylation by serine/threonine kinases activated as a consequence of receptor activation is one (negative feedback, e.g. by PKA, PKC or one of the MAPKs), and dephosphorylation of key phosphotyrosines by phosphotyrosine phosphatases (e.g. SHP-1) is another. In contrast to this easily reversed down-regulation is the more permanent shut-off resulting from ligand-induced internalization of the receptor and its degradation in lysosomes. An interesting example of signal desensitization is provided by the EGF receptor vIII transforming mutant, which is deleted for much of the extracellular domain of the receptor and undergoes spontaneous homodimerization, autophosphorylation and constitutive activation. It binds much of the cell's GRB2 and strongly activates MEK; nevertheless, ERK1 and ERK2 are activated to a lesser extent and seem to resist further activation by various agonists, apparently because of the activation of a tyrosine phosphatase that negatively regulates their activity [209].

Phosphorylation of upstream signalling elements by activated downstream kinases is one mechanism for shutting down a signalling pathway. For example, phosphorylation of SOS (on serine/threonine residues in proline-rich motifs in the C-terminal domain) by ERK (p44/p42 MAPK) inhibits its ability to activate Ras and results in separation of an intact SOS–GRB2 complex from the activated receptor (e.g. SHC or the EGF receptor) [210,211]. Somewhat at odds with this are the reports that ERK activation, either by an insulin-stimulated Ras/MEK pathway or by osmotic shock acting via a MEK-dependent but Ras-independent pathway, led to phosphorylation of SOS and dissociation of the GRB2–SOS complex, terminating the ability of SOS to promote disengagement of GDP from Ras–GDP [212,212a]. These differences could, of course, reflect differences in the input signals and the regulatory circuits activated in the different cell types in these studies. Specific activation of JNK but not ERK by anisomycin, which inhibits protein synthesis and activates the stress-activated signalling pathways, did not cause GRB2–SOS dissociation, suggesting that SOS is not a substrate for JNK [213]. In cells with a wild-type Ras requiring GEF activity, this negative-feedback regulatory circuit attenuates the Ras-activated signal transduction pathway. The signal would be more persistent in cells containing a mutant oncogenic Ras that did not require GEF activity.

The length of time a signalling pathway remains active varies with the cell type and the receptor involved. For example, PC12 cells stimulated with FGF or NGF exhibited a persistent elevation of Ras–GTP levels and a sustained activation of the ERKs, which were translocated to the nucleus, resulting ultimately in differentiation as shown by neurite outgrowth. In the same cells stimulated with EGF, Ras–GTP levels increased only transiently and ERK activation was short-lived. Besides differences in the details of just which intracellular signalling molecules are controlled by each receptor, there is also the fact that the EGF receptor is more rapidly down-regulated than the NGF receptor through phosphorylation and internalization. Appropriate manipulation of EGF or NGF receptor levels can lead to either response, suggesting that the critical element is the intensity of ERK activation [181].

Many of the signalling elements are returned to their ground state by cytoplasmic and nuclear phosphatases (Figure 8) that remove the activating phosphates from the various intermediates [54]. A corollary of this is that inhibitors of phosphatase activity can mimic the effect of a kinase by prolonging the state of activation of a signalling intermediate. Protein serine/threonine phosphatases (e.g. PP1, PP4, PP5, PP2A, PP2B, PP2C) remove phosphates from protein phosphoserine and phosphothreonine residues [214]. Many of them are sensitive to okadaic acid. Barford [215] has reviewed structural studies of the protein phosphatases, noting their classification into four structurally distinct families and the fact that the depth of the catalytic site cleft determines whether the phosphatase will target a tyrosine phosphate or serine/threonine phosphate.

Various protein tyrosine phosphatase activities have been identified (e.g. CD45, SHP-1, SHP-2), and much current research is directed at discovering their specificities [216]. Because of the number of names given to the same protein tyrosine phosphatases, a uniform nomenclature has recently been proposed [52]. Thus SHP-1 (also known as SHP, PTP1C, SHPTP-1 and HCP) dephosphorylates many tyrosine-phosphorylated receptor proteins and has been implicated in attentuating the mitogenic response [216]. SHP-2 (also known as SYP, SHPTP-2, SHPTP-3, PTP2C and PTP1D) is a protein tyrosine phosphatase that appears to act immediately downstream of Ras as a positive mediator of growth factor signalling by a number of receptors (those for insulin, IGF-1 and EGF). Phosphatases (e.g. SHP-2) can also provide a positive signal by removing inhibitory phosphates, for example the phosphate in (unactivated) Src that obstructs an SH2 domain necessary for activity. It is not known whether SHP-2 phosphatase action itself is necessary for transmission of the signal from activated receptors such as the PDGF receptor, or whether its effect is because of its potential to associate with GRB2–SOS [217–219].

Dual-specificity protein phosphatases (e.g. MKP-1, PAC-1) are predominately nuclear proteins that dephosphorylate both phosphothreonine and phosphotyrosine residues. They exhibit considerable specificity for individual MAPKs [220]. Growth factor stimulation and many forms of physical or chemical trauma enhance the mRNA levels of these enzymes, suggesting an important role for PAC-1 and MKP-1 in modulating the signal transduction pathways initiated by mitogens and stress respectively.

In cells with wild-type Ras, activation of Ras in response to the binding of a ligand to an appropriate receptor is transient, and consequently the transmitted signal has the nature of a pulse. Importantly, the Ras-transmitted signal is also accompanied by parallel signals generated by other signalling elements mobilized by the receptor; some of these signals may serve in a negative-feedback capacity. As shown in Figure 6, the signal often results in the modification of one or more transcription factors, giving rise to changes in gene expression. These changes in gene expression are the consequence of a cell-specific integration of all the input signals, including both their intensity (i.e. number of activated receptors) and duration (which varies with the receptor activated), and although the complexity of the pathways and their cross-talk is bewildering, there do seem to be some consistent patterns emerging [221].

Oncogenic forms of Ras have been shown to activate all three of the protein phosphorylation cascades illustrated in Figure 6 [94]. In contrast to wild-type Ras, the signal from a mutant oncogenic Ras (a transformation signal) is continuous, is not accompanied by independently generated upstream signals and results in the perpetual stimulation of certain downstream signalling elements, probably giving rise to an altered relationship among the signalling intermediates and a permanent state of modification of particular downstream transcription factors [16]. As a consequence of the particular Ras mutation or the permanent stimulation of certain pathways, there may also be changes in the way Ras interacts with proteins such as Ral-GDS, PI3-K, and the GAPs, GEFs, Rafs and MEKKs. Many of the identified Ras response elements in promoters and enhancers involve either AP-1 (Fos/Jun) or Ets transcription factors, often in situations where the two binding sites overlap. The resulting changes include the increased expression of genes that are important in making cells tumorigenic and metastatic, e.g. proteinases and adhesion molecules [222,223]. It is interesting to note that in at least some cell types both Rac and Rho make important contributions to Ras transformation [224]. Finally, the fact that most of the studies that have led to our current picture of all these signalling pathways have been done with immortal cell lines must not be ignored. It remains to be determined how the immortalization process has impacted on these signalling pathways.

PERSPECTIVE: IMPLICATIONS OF MULTIPLE SIGNALLING CASCADES AND MULTIPLEX SIGNAL TRANSMISSION

When a receptor is activated, a number of signalling pathways are typically stimulated to various degrees. Receptor, non-receptor and receptor-associated protein tyrosine kinases phosphorylate themselves and specific target proteins. Other proteins are then activated by virtue of their association with tyrosine phosphate groups on these proteins, directly or via an adaptor. Heterotrimeric G-proteins coupled to serpentine receptors trigger signalling pathways through both their Ga and Gbg subunits. The resulting cascades of interactions provide a self-reinforcing stability to the signal; the contribution of any one element is important to, but may not be critical to, the successful delivery of the signal. The branching of the pathways permits a degree of signal diversification, while at the same time cross-talk among pathways permits one pathway to reinforce or attenuate the signal transmitted by a second pathway. Signals initiated by different receptors are distinguished by the different mix of pathways stimulated and the intensity of their stimulation. Thus NGF stimulation of 3T3 cells expressing the human TrkA receptor inhibits proliferation, possibly via the induction of the cyclin-dependent kinase inhibitor p21^Cip1/WAF1; this occurs despite the fact that NGF is more effective at activating the p42 and p44 ERKs than is EGF, which stimulates proliferation [225]. It may be instructive that EGF binds GRB2–SOS directly, whereas TrkA acts via SHC phosphorylation and the formation of SHC–GRB2–SOS complexes [181].

A very important principle to appreciate is that most of these pathways in the cell are likely to be functioning at a modest level in a healthy cell in normal (unperturbed) conditions. The balance among the pathways results in a homoeostatic state. Perturbations of different sorts upset the balance in different ways, leading for example to cell differentiation or proliferation on the one hand, or to growth inhibition or apoptosis on the other. For example, removal of NGF from PC12 cells reduces the stimulation of the Ras Raf MEK ERK1/2 pathway and tilts the balance towards net activation of the MEKK SEK JNK and MKK3 p38 pathways, resulting in apoptosis [226]. Similarly, activation of the B-cell surface IgM receptor activates the Ras Raf MEK ERK (but not the JNK) pathway, leading to cell activation, differentiation or apoptosis (depending upon other variables), whereas activation of the CD40 receptor (a member of the TNF receptor family that recognizes a ligand expressed on activated T-cells) activates JNK/SAPK, but not ERK, and inhibits the apoptotic response induced by anti-IgM [227]. Insulin promotes the survival of neurons in culture, perhaps by virtue of the fact that it negatively regulates p38 kinase activity [228].

The fact that some of the proteins in a pathway become multiply phosphorylated (e.g. Raf-1, MEK, GSK-3) raises the possibility that more than one signal may be transmitted simultaneously via the same pathway (Figure 9). This is known as multiplex signalling. For example, MEK is preferentially phosphorylated on Ser-218 by MEKK1, but on both Ser-218 and Ser-222 by Raf-1 [4]. Is it possible that the type of signal MEK delivers is different in these two cases? As illustrated by the hypothetical example in Figure 9, when amino acid 85 is phosphorylated it interacts with and passes the signal on to protein X, whereas when residue 139 is phosphorylated it interacts with and passes a signal on to protein Y; phosphorylation of both amino acids leads to both signals being transmitted. Selective mutation of one or other of the relevant amino acids followed by determination of the type of signal transmitted would give some indication of whether different signals are being sent or not. Activating and inhibiting phosphorylations of a particular element, e.g. a transcription factor, are a form of multiplex signalling only in the sense that the second signal is a Null signal. If multiplex signalling is occurring, it adds yet another layer of complexity to the signalling process, one that will be as fascinating as it will be difficult to sort out in future research.

The comments of various colleagues, especially Ann Chambers, Mike Kiledjian, Mike Ostrowski and Ren Ping Zhou, on earlier versions of the manuscript were much appreciated. I thank Kathleen Curtis for preparing the illustrations. Because of the vast number of publications in this field, references have for the most part been restricted to representative research reports and reviews published in the last 2 years; sincere apologies are offered to those whose work has not been cited.

REFERENCES    

1 Maher J., Baker, D. A., Manning, M., Dibb, N. J. and Roberts, I. A. G. (1995) Oncogene 11, 1639–1647
1st Citation

2 Khosravi-Far, R. and Der, C. J. (1994) Cancer Metastasis Rev. 13, 67–89
Medline    1st Citation

3 van der Geer, P., Hunter, T. and Lindberg, R. A. (1994) Annu. Rev. Cell Biol. 10, 251–337
Medline    1st Citation   2nd   3rd   4th

4 Malarkey, K., Belham, C. M., Paul, A., Graham, A., McLees, A., Scott, P. H. and Plevin, R. (1995) Biochem. J. 309, 361–375
Medline    1st Citation   2nd

5 Wagner, A. C. and Williams, J. A. (1994) Am. J. Physiol. 266, G1–G14
Medline    1st Citation   2nd   3rd

6 Nuoffer, C. and Balch, W. E. (1995) Annu. Rev. Biochem. 63, 949–990
Medline    1st Citation

7 Rothman, J. E. (1994) Nature (London) 372, 55–63
Medline    1st Citation

8 Vojtek, A. B. and Cooper, J. A. (1995) Cell 82, 527–529
Medline    1st Citation

9 Chant, J. and Stowers, L. (1995) Cell 81, 1–4
Medline    1st Citation   2nd   3rd   4th   5th

10 Ridley, A. J. (1995) Curr. Opin. Genet. Dev. 5, 24–30
Medline    1st Citation   2nd  3rd  4th   5th   6th   7th   8th

11 Kim, S.-H., Prive, G. G. and Milburn, M. V. (1993) Handb. Exp. Pharmacol. 108, 177–194
1st Citation  2nd

12 Lowy, D. R. and Willumsen, B. M. (1993) Annu. Rev. Biochem. 62, 851–891
Medline    1st Citation   2nd

13 Polakis, P. and McCormick, F. (1993) J. Biol. Chem. 268, 9157–9160
Medline    1st Citation

14 Yoder-Hill, J., Golubic, M. and Stacey, D. W. (1995) J. Biol. Chem. 270, 27615–27621
Medline   J. Biol. Chem.    1st Citation

15 Quilliam, L. A., Khosravi-Far, R., Huff, S. Y. and Der, C. J. (1995) BioEssays 17, 395–404
Medline    1st Citation   2nd   3rd   4th

16 Lowe, P. N. and Skinner, R. H. (1994) Cell. Signalling 6, 109–123
1st Citation   2nd

17 Gutmann, D. H., Geist, R. T., Wright, D. E. and Snider, W. D. (1995) Cell Growth Differ. 6, 315–323
Medline    1st Citation

18 Maekawa, M., Li, S., Iwamatsu, A., Morishita, T., Yokota, K., Imai, Y., Kohsaka, S., Nakamura, S. and Hattori, S. (1994) Mol. Cell. Biol. 14, 6879–6885
Medline   Mol. Cell. Biol.    1st Citation

19 Cullen, P. J., Hsuan, J. J., Truong, O., Letcher, A. J., Jackson, T. R., Dawson, A. P. and Irvine, R. F. (1995) Nature (London) 376, 527–530
Medline    1st Citation

20 Takai, Y., Sasaki, T., Tanaka, K. and Nakanishi, H. (1995) Trends Biochem. Sci. 20, 227–231
Medline    1st Citation

21 Henkemeyer, M., Rossi, D. J., Holmyard, D. P., Puri, M. C., Mbamalu, G., Harpal, K., Shih, T. S., Jacks, T. and Pawson, T. (1995) Nature (London) 377, 695–701
Medline    1st Citation

22 Vogel, K. S., Brannan, C. I., Jenkins, N. A., Copeland, N. G. and Parada, L. F. (1995) Cell 82, 733–742
Medline    1st Citation   2nd

23 Hori, Y., Beeler, J. F., Sakaguchi, K., Tachibana, M. and Miki, T. (1994) EMBO J. 13, 4776–4786
1st Citation   2nd

24 Huby, R. D. J., Carlile, G. W. and Ley, S. C. (1995) J. Biol. Chem. 270, 30241–30244
Medline   J. Biol. Chem.    1st Citation

25 Romero, F., Dargemont, C., Pozo, F., Reeves, W. H., Camonis, J., Gisselbrecht, S. and Fischer, S. (1995) Mol. Cell. Biol. 16, 37–44
Medline   Mol. Cell. Biol.    1st Citation

25a Nomanbhoy, T. K. and Cerione, R. A. (1996) J. Biol. Chem. 271, 10004–10009
Medline   J. Biol. Chem.    1st Citation

25b Chuang, T.-H., Xu, X., Kaartinen, V., Heisterkamp, N., Groffen, J. and Bokoch, G. M. (1995) Proc. Natl. Acad.Sci. U.S.A. 92, 10282–10286
1st Citation   2nd

26 Cantor, S. B., Urano, T. and Feig, L. A. (1995) Mol. Cell. Biol. 15, 4578–4584
Medline   Mol. Cell. Biol.    1st Citation

27 Urano, T., Emkey, R. and Feig, L. A. (1996) EMBO J. 15, 810–816
Medline    1st Citation

28 Park, S.-H. and Weinberg, R. A. (1995) Oncogene 11, 2349–2355
Medline    1st Citation

29 Nobes, C. D. and Hall, A. (1995) Cell 81, 53–62
Medline    1st Citation   2nd

30 Kozma, R., Ahmed, S., Best, A. and Lim, L. (1995) Mol. Cell. Biol. 15, 1942–1952
Medline   Mol. Cell. Biol.    1st Citation   2nd

31 Khosravi-Far, R., Solski, P. A., Clark, G. J., Kinch, M. S. and Der, C. J. (1995) Mol. Cell. Biol. 15, 6443–6453
Medline   Mol. Cell. Biol.    1st Citation

32 Heldin, C.-H. (1995) Cell 80, 213–223
Medline    1st Citation

33 Ihle, J. N. (1995) Nature (London) 377, 591–594
Medline    1st Citation

34 Claesson-Welsh, L. (1994) J. Biol. Chem. 269, 32023–32026
Medline    1st Citation   2nd

35 Pawson, T. (1995) Nature (London) 373, 573–580
Medline    1st Citation

36 Bork, P. and Margolis, B. (1995) Cell 80, 693–694
Medline    1st Citation   2nd

37 van der Geer, P. and Pawson, T. (1995) Trends Biochem. Sci. 20, 277–280
Medline    1st Citation   2nd

38 Zhou, M.-M., Ravichandran, K. S., Olejniczak, E. T., Petros, A. M., Meadows, R. P., Sattler, M., Harlan, J. E., Wade, W. S., Burakoff, S. J. and Fesik, S. W. (1995) Nature (London) 378, 584–592
Medline    1st Citation

39 Ladbury, J. E., Lemmon, M. A., Zhou, M., Green, J., Botfield, M. C. and Schlessinger, J. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 3199–3203
Medline    1st Citation

40 Feng, S., Chen, J. K., Yu, H., Simon, J. A. and Schreiber, S. L. (1994) Science 266, 1241–1247
Medline    1st Citation

41 Hu, Q., Milfay, D. and Williams, L. T. (1995) Mol. Cell. Biol. 15, 1169–1174
Medline   Mol. Cell. Biol.    1st Citation

42 de Vries-Smits, A. M. M., Pronk, G. J., Medema, J. P., Burgering, B. M. T. and Bos, J. L. (1995) Oncogene 10, 919–925
Medline    1st Citation

43 Decker, S. J. (1993) J. Biol. Chem. 268, 9176–9179
Medline    1st Citation

44 Gotoh, N., Tojo, A., Muroya, K., Hashimoto, Y., Hattori, S., Nakamura, S., Takenawa, T., Yazaki, Y. and Shibuya, M. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 167–171
Medline    1st Citation

45 Sato, K.-i., Sato, A., Aoto, M. and Fukami, Y. (1995) Biochem. Biophys. Res. Commun. 215, 1078–1087
Medline    1st Citation

46 Burgering, B. M. T., de Vries-Smits, A. M. M., Medema, R. H., van Weeren, P. C., Tertoolen, L. G. J. and Bos, J. L. (1993) Mol. Cell. Biol. 13, 7248–7256
Medline   Mol. Cell. Biol.    1st Citation

47 Songyang, Z., Gish, G., Mbamalu, G., Pawson, T. and Cantley, L. C. (1995) J. Biol. Chem. 270, 26029–26032
Medline   J. Biol. Chem.    1st Citation

48 Songyang, Z. and Cantley, L. C. (1995) Trends Biochem. Sci. 20, 470–475
1st Citation

49 Yao, R. and Cooper, G. M. (1995) Oncogene 11, 1607–1614
Medline    1st Citation

50 McGlade, J., Brunkhorst, B., Anderson, D., Mbamalu, G., Settleman, J., Dedhar, S., Rozakis-Adcock, M., Chen, L. B. and Pawson, T. (1993) EMBO J. 12, 3073–3081
Medline    1st Citation

51 Johnson, M. R., Declue, J. E., Felzmann, S., Vass, W. C., Xu, G., White, R. and Lowy, D. R. (1994) Mol. Cell. Biol. 14, 641–645
Medline   Mol. Cell. Biol.    1st Citation

52 Adachi, M., Fischer, E. H., Ihle, J., Imai, K., Jirik, F., Neel, B., Pawson, T., Shen, S.-H., Thomas, M., Ullrich, A. and Zhao, Z. (1996) Cell 85, 15
Medline    1st Citation   2nd

53 Li, S., Janosch, P., Tanji, M., Rosenfeld, G. C., Waymire, J. C., Mischak, H., Kolch, W. and Sedivy, J. M. (1995) EMBO J. 14, 685–696
Medline    1st Citation

54 Sun, H. and Tonks, N. K. (1994) Trends Biochem. Sci. 19, 480–485
Medline    1st Citation   2nd

55 Peng, Z.-Y. and Cartwright, C. A. (1995) Oncogene 11, 1955–1962
Medline    1st Citation   2nd

56 Chen, Q., Kinch, M. S., Lin, T. H., Burridge, K. and Juliano, R. L. (1994) J. Biol. Chem. 269, 26602–26605
Medline    1st Citation

57 Schlaepfer, D. D., Hanks, S. K., Hunter, T. and van der Geer, P. (1994) Nature (London) 372, 786–791
Medline    1st Citation

58 Hunter, T. (1995) Cell 83, 1–4
Medline    1st Citation

59 Divecha, N. and Irvine, R. F. (1995) Cell 80, 269–278
Medline    1st Citation   2nd

60 Rameh, L. E., Chen, S.-S. and Cantley, L. C. (1995) Cell 83, 821–830
Medline    1st Citation

61 Nakanishi, H., Brewer, K. A. and Exton, J. H. (1993) J. Biol. Chem. 268, 13–16
Medline    1st Citation

62 Liu, J.-P. (1996) Mol. Cell. Endocrinol. 116, 1–29
1st Citation   2nd

63 Erpel, T. and Courtneidge, S. A. (1995) Curr. Opin. Cell Biol. 7, 176–182
Medline    1st Citation

64 Jiang, H., Luo, J.-Q., Urano, T., Frankel, P., Lu, Z., Foster, D. A. and Feig, L. A. (1995) Nature (London) 378, 409–412
Medline    1st Citation

65 Xie, W. and Herschman, H. R. (1995) J. Biol. Chem. 270, 27622–27628
Medline   J. Biol. Chem.    1st Citation

66 Downward, J. (1994) FEBS Lett. 338, 113–117
Medline    1st Citation

67 Aronheim, A., Engelberg, D., Li, N., Al-Alawi, N., Schlessinger, J. and Karin, M. (1994) Cell 78, 949–961
Medline    1st Citation

68 Xie, Y., Pendergast, A. M. and Hung, M.-C. (1995) J. Biol. Chem. 270, 30717–30724
Medline   J. Biol. Chem.    1st Citation

69 Klint, P., Kanda, S. and Claesson-Welsh, L. (1995) J. Biol. Chem. 270, 23337–23344
Medline   J. Biol. Chem.    1st Citation

70 Kavanaugh, W. M. and Williams, L. T. (1994) Science 268, 1862–1865
1st Citation

71 Blaikie, P., Immanuel, D., Wu, J., Li, N., Yajnik, V. and Margolis, B. (1994) J. Biol. Chem. 269, 32031–32034
Medline    1st Citation

72 Prigent, S. A., Pillay, T. S., Ravichandran, K. S. and Gullick, W. J. (1995) J. Biol. Chem. 270, 22097–22100
Medline   J. Biol. Chem.    1st Citation

73 Reference deleted

74 Strader, C. D., Fong, T. M., Graziano, M. P. and Tota, M. R. (1995) FASEB J. 9, 745–754
Medline    1st Citation

75 DeVivo, M. and Iyengar, R. (1994) Mol. Cell. Endocrinol. 100, 65–70
Medline    1st Citation

76 Inglese, J., Koch, W. J., Touhara, K. and Lefkowitz, R. J. (1995) Trends Biochem. Sci. 20, 151–156
Medline    1st Citation   2nd

77 Burgering, B. M. T. and Bos, J. L. (1995) Trends Biochem. Sci. 20, 18–22
Medline    1st Citation   2nd   3rd   4th

78 Faure, M., Voyno-Yasenetskaya, T. A. and Bourne, H. R. (1994) J. Biol. Chem. 269, 7851–7854
Medline    1st Citation

79 Crespo, P., Xu, N., Simonds, W. F. and Gutkind, J. S. (1994) Nature (London) 369, 418–420
Medline    1st Citation

80 Touhara, K., Hawes, B. E., van Biesen, T. and Lefkowitz, R. J. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 9284–9287
Medline    1st Citation

81 van Biesen, T., Hawes, B. E., Luttrell, D. K., Krueger, K. M., Touhara, K., Porfiri, E., Sakaue, M., Luttrell, L. M. and Lefkowitz, R. J. (1995) Nature (London) 376, 781–784
Medline    1st Citation

82 Luttrell, L. M., van Biesen, T., Hawes, P. E., Koch, W. J., Touhara, K. and Lefkowitz, R. J. (1995) J. Biol. Chem. 270, 16495–16498
Medline   J. Biol. Chem.    1st Citation

83 Crespo, P., Cachero, T. G., Xu, N. and Gutkind, J. S. (1995) J. Biol. Chem. 270, 25259–25265
Medline   J. Biol. Chem.    1st Citation

84 Musacchio, A., Gibson, T., Rice, P., Thompson, J. and Saraste, M. (1993) Trends Biochem. Sci. 18, 343–348
Medline    1st Citation

85 Harlan, J. E., Hajduk, P. J., Yoon, H. S. and Fesik, S. W. (1994) Nature (London) 371, 168–170
Medline    1st Citation

86 Abrams, C. S., Wu, H., Zhao, W., Belmonte, E., White, D. and Brass, L. F. (1995) J. Biol. Chem. 270, 14485–14492
Medline   J. Biol. Chem.    1st Citation

87 Hyvonen, M., Macias, M. J., Nilges, M., Oschkinat, H., Saraste, M. and Wilmanns, M. (1995) EMBO J. 14, 4676–4685
Medline    1st Citation

88 Yao, L., Kawakami, Y. and Kawakami, T. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 9175–9179
Medline    1st Citation

89 Lemmon, M. A., Ferguson, K. M., O'Brien, R., Sigler, P. B. and Schlessinger, J. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 10472–10476
Medline    1st Citation

90 Lee, S. B. and Rhee, S. G. (1995) Curr. Opin. Cell Biol. 7, 183–189
Medline    1st Citation

91 Marshall, M. S. (1995) FASEB J. 9, 1311–1318
Medline    1st Citation

92 Blank, J. L., Gerwins, P., Elliott, E. M., Sather, S. and Johnson, G. L. (1996) J. Biol. Chem. 271, 5361–5368
Medline   J. Biol. Chem.    1st Citation   2nd

93 Joneson, T., White, M. A., Wigler, M. H. and Bar-Sagi, D. (1996) Science 271, 810–812
Medline    1st Citation   2nd

94 Waskiewicz, A. J. and Cooper, J. A. (1995) Curr. Opin. Cell Biol. 7, 798–805
Medline    1st Citation   2nd

95 Elion, E. A. (1995) Trends Cell Biol. 5, 322–327
1st Citation

96 Zhu, A. X., Zhao, Y., Moller, D. E. and Flier, J. S. (1994) Mol. Cell. Biol. 14, 8202–8211
Medline   Mol. Cell. Biol.    1st Citation

97 English, J. M., Vanderbilt, C. A., Xu, S., Marcus, S. and Cobb, M. H. (1995) J. Biol. Chem. 270, 28897–28902
Medline   J. Biol. Chem.    1st Citation

98 Newton, A. C. (1995) J. Biol. Chem. 270, 28495–28498
Medline   J. Biol. Chem.    1st Citation

99 Manser, E., Leung, T., Salihuddin, H., Zhao, Z.-S. and Lim, L. (1994) Nature (London) 367, 40–46
Medline    1st Citation

100 Jakobi, R., Chen, C.-J., Tuazon, P. T. and Traugh, J. A. (1996) J. Biol. Chem 271, 6206–6211
J. Biol. Chem.    1st Citation

101 Knaus, U. G., Morris, S., Dong, H.-J., Chernoff, J. and Bokoch, G. M. (1995) Science 269, 221–223
Medline    1st Citation

102 Coso, O. A., Chiariello, M., Yu, J.-C., Teramoto, H., Crespo, P., Xu, N., Miki, T. and Gutkind, J. S. (1995) Cell 81, 1137–1146
Medline    1st Citation   2nd

103 Bagrodia, S., Derijard, B., Davis, R. J. and Cerione, R. A. (1995) J. Biol. Chem. 270, 27995–27998
Medline   J. Biol. Chem.    1st Citation   2nd

104 Minden, A., Lin, A., Claret, F.-X., Abo, A. and Karin, M. (1995) Cell 81, 1147–1157
Medline    1st Citation

105 Qiu, R.-G., Chen, J., Kirn, D., McCormick, F. and Symons, M. (1995) Nature (London) 374, 457–459
Medline    1st Citation

106 Zhang, S., Han, J., Sells, M. A., Chernoff, J., Knaus, U. G., Ulevitch, R. J. and Bokoch, G. M. (1995) J. Biol. Chem. 270, 23934–23936
Medline   J. Biol. Chem.    1st Citation

107 Wu, X., Noh, S. J., Zhou, G., Dixon, J. E. and Guan, K.-L. (1996) J. Biol. Chem. 271, 3265–3271
Medline   J. Biol. Chem.    1st Citation

108 Jelinek, T., Catling, A. D., Reuter, C. W. M., Moodie, S. A., Wolfman, A. and Weber. M. J. (1994) Mol. Cell. Biol. 14, 8212–8218
Medline   Mol. Cell. Biol.    1st Citation

109 Minden, A., Lin, A., McMahon, M., Lange-Carter, C., Derijard, B., Davis, R. J., Johnson, G. L. and Karin, M. (1994) Science 266, 1719–1723
Medline    1st Citation   2nd

110 Lange-Carter, C. A. and Johnson, G. L. (1994) Science 265, 1458–1561
Medline    1st Citation

111 Matsuda, S., Kawasaki, H., Moriguchi, T., Gotoh, Y. and Nishida, E. (1995) J. Biol. Chem. 270, 12781–12786
Medline   J. Biol. Chem.    1st Citation

112 Lander, H. M., Ogiste, J. S., Teng, K. K. and Novogrodsky, A. (1995) J. Biol. Chem. 270, 21195–21198
Medline   J. Biol. Chem.    1st Citation

113 Fialkow, L., Chan, C. K., Rotin, D., Grinstein, S. and Downey, G. P. (1994) J. Biol. Chem. 269, 31234–31242
Medline    1st Citation

114 Chen, Q., Olashaw, N. and Wu, J. (1995) J. Biol. Chem. 270, 28499–28502
Medline   J. Biol. Chem.    1st Citation

115 Russell, M., Lange-Carter, C. A. and Johnson, G. L. (1995) J. Biol. Chem. 270, 11757–11760
Medline   J. Biol. Chem.    1st Citation

116 Xu, S., Robbins, D., Frost, J., Dang, A., Lange-Carter, C. and Cobb, M. H. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 6808–6812
Medline   1st Citation   2nd

117 Chatani, Y., Tanimura, S., Miyoshi, N., Hattori, A., Sato, M. and Kohno, M. (1995) J. Biol. Chem. 270, 30686–30692
Medline   J. Biol. Chem.    1st Citation

118 Yamaguchi, K., Shirakabe, K., Shibuya, H., Irie, K., Oishi, I., Ueno, N., Taniguchi, T., Nishida, E. and Matsumoto, K. (1995) Science 270, 2008–2011
Medline    1st Citation

119 Salmeron, A., Ahmad, T. B., Carlile, G. W., Pappin, D., Narsimhan, R. P. and Ley, S. C. (1996) EMBO J. 15, 817–826
Medline    1st Citation

120 Dorow, D. S., Devereux, L., Tu, G. F., Price, G., Nicholl, J. K., Sutherland, G. R. and Simpson, R. J. (1995) Eur. J. Biochem. 234, 492–500
Medline    1st Citation

121 Hirai, S., Izawa, M., Osada, S.-I., Spyrou, G. and Ohon, S. (1996) Oncogene 12, 641–650
Medline    1st Citation

122 Reuter, C. W. M., Catling, A. D., Jelinek, T. and Weber, M. J. (1995) J. Biol. Chem. 270, 7644–7655
Medline   J. Biol. Chem.    1st Citation

122a Hill, C. S. and Treisman, R. (1995) EMBO J. 14, 5037–5047
Medline    1st Citation

123 Jaiswal, R. K., Moodie, S. A., Wolfman, A. and Landreth, G. E. (1994) Mol. Cell. Biol. 14, 6944–6953
Medline   Mol. Cell. Biol.    1st Citation

124 Pritchard, C. A., Samuels, M. L., Bosch, E. and McMahon, M. (1995) Mol. Cell. Biol. 15, 6430–6442
Medline   Mol. Cell. Biol.    1st Citation

125 Daum, G., Eisenmann-Tappe, I., Fries, H.-W., Troppmair, J. and Rapp, U. R. (1994) Trends Biochem. Sci. 19, 474–480
Medline    1st Citation

126 Schulte, T. W., Blagosklonny, M. V., Ingui, C. and Necker, L. (1995) J. Biol. Chem. 270, 24585–24588
Medline   J. Biol. Chem.    1st Citation

127 Hu, C.-D., Kariya, K.-i., Tamada, M., Akasaka, K., Shirouzu, M., Yokoyama, S. and Kataoka, T. (1995) J. Biol. Chem. 270, 30274–30277
Medline   J. Biol. Chem.    1st Citation

128 Druggan, J. K., Khosravi-Far, R., White, M. A., Der, C. J., Sung, Y.-., Hwang, Y.-W. and Campbell, S. L. (1996) J. Biol. Chem. 271, 233–237
J. Biol. Chem.    1st Citation

129 Leevers, S. J., Paterson, H. F. and Marshall, C. J. (1994) Nature (London) 369, 411–414
Medline    1st Citation

130 Stokoe, D., Macdonald, S. G., Cadwallader, K., Symons, M. and Hancock, J. F. (1994) Science 264, 1463–1467
Medline    1st Citation

131 Shirouzu, M., Koide, H., Fujita-Yoshigaki, J., Oshiro, H., Toyama, Y., Yamasaki, K., Fuhrman, S. A., Villacranca, E., Kaziro, Y. and Yokoyama, S. (1994) Oncogene 9, 2153–2157
Medline    1st Citation

132 Marais, R., Light, Y., Paterson, H. F. and Marshall, C. J. (1995) EMBO J. 14, 3136–3145
Medline    1st Citation

133 Bjorkoy, G., Overvatn, A., Diaz-Meco, M. T., Moscat, J. and Johansen, T. (1995) J. Biol. Chem. 270, 21299–21306
Medline   J. Biol. Chem.    1st Citation

134 Therrien, M., Chang, H. C., Solomon, N. M., Karim, F. D., Wassarman, D. A. and Rubin, G. M. (1995) Cell 83, 879–888
Medline    1st Citation

135 Jelinek, T., Dent, P., Sturgill, T. W. and Weber, M. J. (1996) Mol. Cell. Biol. 16, 1027–1034
Medline   Mol. Cell. Biol.    1st Citation

136 Morrison, D. K., Heidecker, G., Rapp, U. R. and Copeland, T. D. (1993) J. Biol. Chem. 268, 17309–17316
Medline    1st Citation   2nd

137 Dent, P., Jelinek, T., Morrison, D. K., Weber, M. J. and Sturgill, T. W. (1995) Science 268, 1902–1906
Medline    1st Citation   2nd

138 Zheng, C.-F. and Guan, K.-L. (1994) EMBO J. 13, 1123–1131
Medline    1st Citation

139 Catling, A. D., Schaeffer, H.-J., Reuter, C. W. M., Reddy, G. R. and Weber, M. J. (1995) Mol. Cell. Biol. 15, 5214–5225
Medline   Mol. Cell. Biol.    1st Citation

140 Li, S. and Sedivy, J. M. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 9247–9251
Medline    1st Citation

141 Jamal, S. and Ziff, E. B. (1994) Oncogene 10, 2095–2101
Medline    1st Citation

142 Galaktionov, K., Jessus, C. and Beach, D. (1995) Genes Dev. 9, 1046–1058
Medline    1st Citation

143 Galaktionov, K., Lee, A. K., Eckstein, J., Draetta, G., Meckler, J., Loda, M. and Beach, D. (1995) Science 269, 1575–1577
Medline    1st Citation

144 Hafner, S., Adler, H. S., Mischak, H., Janosch, P., Heidecker, G., Wolfman, A., Pippig, S., Lohse, M., Ueffing, M. and Kolch, W. (1994) Mol. Cell. Biol. 14, 6696–6703
Medline   Mol. Cell. Biol.    1st Citation

145 Erhardt, P., Troppmair, J., Rapp, U. R. and Cooper, G. M. (1995) Mol. Cell. Biol. 15, 5524–5530
Medline   Mol. Cell. Biol.    1st Citation

146 Yao, B., Zhang, Y., Delikat, S., Mathias, S., Basu, S. and Kolesnick, R. (1995) Nature (London) 378, 307–310
Medline    1st Citation

147 Aitken, A. (1995) Trends Biochem. Sci. 20, 95–97
Medline    1st Citation

148 Bonnefoy-Berard, N., Liu, Y.-C., von Willebrand, M., Sung, A., Elly, C., Mustelin, T., Yoshida, H., Ishizaka, K. and Altman, A. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 10142–10146
Medline    1st Citation

149 Acs, P., Szallasi, Z., Kazanietz, M. G. and Blumberg, P. M. (1995) Biochem. Biophys. Res. Commun. 216, 103–109
Medline    1st Citation

150 Muslin, A. J., Tanner, J. W., Allen, P. M. and Shaw, A. S. (1996) Cell 84, 889–897
Medline    1st Citation

151 Suen, K.-L., Bustelo, X. R. and Barbacid, M. (1995) Oncogene 11, 825–831
Medline    1st Citation

152 Li, S., Janosch, P., Tanji, M., Rosenfeld, G. C., Waymire, J. C., Mischak, H., Kolch, W. and Sedivy, J. M. (1995) EMBO J. 14, 685–696
1st Citation

153 Michaud, N. R., Fabian, J. R., Mathes, K. D. and Morrison, D. K. (1995) Mol. Cell. Biol. 15, 3390–3397
Medline   Mol. Cell. Biol.    1st Citation

154 Rommel, C., Radziwill, G., Lovric, J., Noeldeke, J., Heinicke, T., Jones, D., Aitken, A. and Moelling, K. (1996) Oncogene 12, 609–619
Medline    1st Citation

154a Braselmann, S. and McCormick, F. (1995) EMBO J. 14, 4839–4848
Medline    1st Citation

155 Buscher, D., Hipskind, R. A., Krautwald, S., Reimann, T. and Caccarini, M. (1995) Mol. Cell. Biol. 15, 466–475
Medline    1st Citation

156 Seger, R., Seger, D., Reszka, A. A., Munar, E. S., Eldar-Finkelman, H., Dobrowolska, G., Jensen, A. M., Campbell, J. S., Fischer, E. H. and Krebs, E. G. (1994) J. Biol. Chem. 269, 25699–25709
Medline    1st Citation   2nd

157 Seger, R. and Krebs, E. G. (1995) FASEB J. 9, 726–735
Medline    1st Citation   2nd   3rd

158 Mansour, S. J., Candia, J. M., Gloor, K. K. and Ahn, N. G. (1996) Growth Differ. 7, 243–250
1st Citation

159 Mansour, S. J., Matten, W. T., Hermann, A. S., Candia, J. M., Rong, S., Fukasawa, K., Vande Woude, G. F. and Ahn, N. G. (1994) Science 265, 966–970
Medline    1st Citation

160 Cowley, S., Paterson, H., Kemp, P. and Marshall, C. J. (1994) Cell 77, 841–852
Medline    1st Citation

161 Sanchez, I., Hughes, R. T., Mayer, B. J., Yee, K., Woodgett, J. R., Avruch, J., Kyriakis, J. M. and Zon, L. I. (1994) Nature (London) 372, 794–798
Medline    1st Citation

162 Lin, A., Minden, A., Martinetto, H., Claret, F.-X., Lange-Carter, C., Mercurio, F., Johnson, G. L. and Karin, M. (1995) Science 268, 286–290
Medline    1st Citation

163 Derijard, B., Hibi, M., Wu, I.-H., Barrett, T., Su, B., Deng, T., Karin, M. and Davis, R. J. (1994) Cell 76, 1025–1037
Medline    1st Citation

164 Kharbanda, S., Pandey, P., Ren, R., Mayer, B., Zon, L. and Kufe, D. (1995) J. Biol. Chem. 270, 30278–30281
Medline   J. Biol. Chem.    1st Citation   2nd

165 Saleem, A., Datta, R., Yuan, Z.-M., Kharbanda, S. and Kufe, D. (1995) Cell Growth Differ. 6, 1651–1658
1st Citation

166 Yan, M., Dal, T., Deak, J. C., Kyriakis, J. M., Zon, L. I., Woodgett, J. R. and Templeton, D. J. (1994) Nature (London) 372, 798–800
Medline    1st Citation   2nd

167 Derijard, B., Raingeaud, J., Barrett, T., Wu, I.-H., Han, J., Ulevitch, R. J. and Davis, R. J. (1995) Science 267, 682–685
Medline    1st Citation

168 Han, J., Lee, J.-D., Jiang, Y., Li, Z., Feng, L. and Ulevitch, R. J. (1996) J. Biol. Chem. 271, 2886–2891
Medline   J. Biol. Chem.    1st Citation

169 Han, J., Lee, J.-D., Bibbs, L. and Ulevitch, R. J. (1994) Science 265, 808–811
Medline    1st Citation   2nd   3rd

170 Raingeaud, J., Gupta, S., Rogers, J. S., Dickens, M., Han, J., Ulevitch, R. J. and Davis, R. J. (1995) J. Biol. Chem. 270, 7420–7426
Medline   J. Biol. Chem.    1st Citation   2nd

171 Raingeaud, J., Whitmarsh, A. J., Barrett, T., Derijard, B. and Davis, R. J. (1996) Mol. Cell. Biol. 16, 1247–1255
Medline    1st Citation

172 Johnson, L. N., Noble, M. E. M. and Owen, D. J. (1996) Cell 85, 149–158
Medline    1st Citation

173 Coso, O. A., Chiariello, M., Kalinec, G., Kyriakis, J. M., Woodgett, J. and Gutkind, J. S. (1995) J. Biol. Chem. 270, 5620–5624
Medline   J. Biol. Chem.    1st Citation

174 Dalton, S. and Treisman, R. (1992) Cell 68, 597–612
Medline    1st Citation

175 Chuang, C.-F. and Ng, S.-Y. (1994) FEBS Lett. 346, 229–234
Medline    1st Citation

176 Treisman, R. (1995) EMBO J. 14, 4905–4913
Medline    1st Citation

177 Cavigelli, M., Dolfi, F., Claret, F.-X. and Karin, M. (1995) EMBO J. 14, 5957–5964
1st Citation

178 Freshney, N. W., Rawlinson, L., Guesdon, F., Jones, E., Cowley, S., Hsuan, J. and Saklatvala, J. (1994) Cell 78, 1039–1049
1st Citation   2nd   3rd

179 Rouse, J., Cohen, P., Trigon, S., Morange, M., Alonso-Llamazares, A., Zamanillo, D., Hunt, T. and Nebreda, A. R. (1994) Cell 78, 1027–1037
Medline    1st Citation   2nd

180 Kumar, S., McLaughlin, M. M., McDonnell, P. C., Lee, J. C., Livi, G. P. and Young, P. R. (1995) J. Biol. Chem. 270, 29043–29046
Medline   J. Biol. Chem.    1st Citation

181 Marshall, C. J. (1995) Cell 80, 179–185
Medline    1st Citation   2nd   3rd

182 Cobb, M. H. and Goldsmith, E. J. (1995) J. Biol. Chem. 270, 14843–14846
Medline   J. Biol. Chem.    1st Citation

183 Cano, E. and Mahadevan, L. C. (1995) Trends Biochem. Sci. 20, 117–122
Medline    1st Citation

184 McCarthy, S. A., Samuels, M. L., Pritchard, C. A., Abraham, J. A. and McMahon, M. (1995) Mol. Cell. Biol. 9, 1953–1964
1st Citation

185 Franklin, C. C. and Kraft, A. S. (1995) Oncogene 11, 2365–2374
Medline    1st Citation

186 Blenis, J. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 5889–5892
Medline    1st Citation

187 Franklin, R. A., Tordai, A., Patel, H., Gardner, A. M., Johnson, G. L. and Gelfand, E. W. (1994) J. Clin. Invest. 93, 2134–2140
Medline    1st Citation

188 Bohm, M., Moellmann, G., Cheng, E., Alvarez-Franco, M., Wagner, S., Sassone-Corsi, P. and Halaban, R. (1995) Cell Growth Differ. 6, 291–302
Medline    1st Citation

189 Ginty, D. D., Bonni, A. and Greenberg, M. E. (1994) Cell 77, 713–725
Medline    1st Citation

190 Chou, M. M. and Blenis, J. (1995) Curr. Opin. Cell Biol. 7, 806–814
Medline    1st Citation   2nd

191 Lavoie, J. N., Lambert, H., Hickey, E., Weber, L. A. and Landry, J. (1995) Mol. Cell. Biol. 15, 505–516
Medline   Mol. Cell. Biol.    1st Citation

192 Ben-Levy, R., Leighton, I. A., Doza, Y. N., Attwood, P., Morrice, N., Marshall, C. J. and Cohen, P. (1995) EMBO J. 14, 5920–5930
1st Citation

193 McLaughlin, M. M., Kumar, S., McDonell, P. C., Van Horn, S., Lee, J. C., Livi, G. P. and Young, P. R. (1996) J. Biol. Chem. 271, 8488–8492
Medline   J. Biol. Chem.    1st Citation

194 He, X., Saint-Jeannet, J.-P., Woodgett, J. R., Varmus, H. E. and Dawid, I. B. (1995) Nature (London) 374, 617–622
Medline    1st Citation

195 Wang, Q. M., Fiol. C. J., DePaoli-Roach, A. A. and Roach, P. J. (1994) J. Biol. Chem. 269, 14566–14574
Medline    1st Citation

196 Fiol, C. J., Williams, J. S., Chou, C.-H., Wang, Q. M., Roach, P. J. and Andrisani, O. M. (1994) J. Biol. Chem. 269, 32187–32193
Medline    1st Citation

197 Wang, Q. M., Vik, T. A., Ryder, J. W. and Roach, P. J. (1995) Biochem. Biophys. Res. Commun. 208, 796–801
Medline    1st Citation

198 Welsh, G. I., Foulstone, E. J., Young, S. W., Tavare, J. M. and Proud, C. G. (1994) Biochem. J. 303, 15–20
Medline    1st Citation

199 Eldar-Finkelman, H., Seger, R., Vandenheede, J. R. and Krebs, E. G. (1995) J. Biol. Chem. 270, 987–990
Medline   J. Biol. Chem.    1st Citation

200 Cross, D. A. E., Alessi, D. R., Cohen, P., Andjelkovich, M. and Hemmings, B. A. (1995) Nature (London) 378, 785–789
Medline    1st Citation

201 Franke, T. F., Yang, S.-I., Chan, T. O., Datta, K., Kazlauskas, A., Morrison, D. K., Kaplan, D. R. and Tsichlis, P. N. (1995) Cell 81, 727–736
Medline    1st Citation

202 Silvennoinen, O., Ihle, J. N., Schlessinger, J. and Levy, D. E. (1993) Nature (London) 366, 583–585
Medline    1st Citation

203 Bird, T. A., Kyriakis, J. M., Tyshler, L., Gayle, M., Milne, A. and Virca, G. D. (1994) J. Biol. Chem. 269, 31836–31844
Medline    1st Citation

204 David, M., Petricoin, III, E., Benjamin, C., Pine, R., Weber, M. J. and Larner, A. C. (1995) Science 269, 1721–1723
Medline    1st Citation

205 Winston, B. W., Remigio, L. K. and Tiches, D. W. H. (1995) J. Biol. Chem. 270, 27391–27394
Medline   J. Biol. Chem.    1st Citation

206 Westwick, J. K., Bielawska, A. E., Dbaibo, G., Hannun, Y. A. and Brenner, D. A. (1995) J. Biol. Chem. 270, 22689–22692
Medline   J. Biol. Chem.    1st Citation

207 Terada, K., Kaziro, Y. and Satoh, T. (1995) J. Biol. Chem. 270, 27880–27886
Medline   J. Biol. Chem.    1st Citation

208 Winston, L. A. and Hunter, T. (1995) J. Biol. Chem. 270, 30837–30840
Medline   J. Biol. Chem.    1st Citation

209 Montgomery, R. B., Moscatello, D. K., Wong, A. J., Cooper, J. A. and Stahl, W. L. (1995) J. Biol. Chem. 270, 30562–30566
Medline   J. Biol. Chem.    1st Citation

210 Buday, L., Warne, P. H. and Downward, J. (1995) Oncogene 11, 1327–1331
Medline    1st Citation

211 Rozakis-Adcock, M., van der Geer, P., Mbamalu, G. and Pawson, T. (1995) Oncogene 11, 1417–1426
Medline    1st Citation

212 Waters, S. B., Holt, K. H., Ross, S. E., Syu, L.-J., Guan, K.-L., Saltiel, A. R., Koretzky, G. A. and Pessin, J. E. (1995) J. Biol. Chem. 270, 20883–20886
J. Biol. Chem.    1st Citation

212a Cherniack, A. D., Klarlaund, J. K., Conway, B. R. and Czech, M. P. (1995) J. Biol. Chem. 270, 1485–1488
Medline   J. Biol. Chem.    1st Citation

213 Chen, D., Waters, S. B., Holt, K. H. and Pessin, J. E. (1996) J. Biol. Chem. 271, 6328–6332
J. Biol. Chem.    1st Citation

214 Wera, S. and Hemmings, B. A. (1995) Biochem. J. 311, 17–29
Medline   Biochem. J.   Biochem. J.    1st Citation

215 Barford, D. (1995) Curr. Opin. Struct. Biol. 5, 728–734
1st Citation

216 Hunter, T. (1995) Cell 80, 225–236
Medline    1st Citation   2nd

217 Milarski, K. L. and Saltiel, A. R. (1994) J. Biol. Chem. 269, 21239–21243
Medline    1st Citation

218 Xiao, S., Rose, D. W., Sasaoka, T., Maegawa, H., Burke, Jr., T. R., Roller, P. P., Shoelson, S. E. and Olefsky, J. M. (1994) J. Biol. Chem. 269, 21244–21248
Medline    1st Citation

219 Sawada, T., Milarski, K. L. and Saltiel, A. R. (1995) Biochem. Biophys. Res. Commun. 214, 737–743
Medline    1st Citation

220 Keyse, S. M. (1995) Biochim. Biophys. Acta 1265, 152–160
Medline    1st Citation

221 Wiesmuller, L. and Wittinghofer, F. (1994) Cell. Signalling 6, 247–267
1st Citation

222 Chambers, A. F. and Tuck, A. B. (1993) Crit. Rev. Oncogenesis 4, 95–114
1st Citation

223 Bortner, D. M., Langer, S. J. and Ostrowski, M. C. (1993) Crit. Rev. Oncogenesis 4, 137–160
1st Citation

224 Qiu, R.-G., Chen, J., McCormick, F. and Symons, M. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 11781–11785
Medline    1st Citation

225 Decker, S. J. (1995) J. Biol. Chem. 270, 30841–30844
Medline    1st Citation

226 Xia, Z., Dickens, M., Raingeaud, J., Davis, R. J. and Greenberg, M. E. (1995) Science 270, 1326–1331
Medline    1st Citation

227 Sakata, N., Patel, H. R., Terada, N., Aruffo, A., Johnson, G. L. and Gelfand, E. W. (1995) J. Biol. Chem. 270, 30823–30828
Medline   J. Biol. Chem.    1st Citation

228 Heidenreich, K. A. and Kummer, J. L. (1996) J. Biol. Chem. 271, 9891–9894
Medline   J. Biol. Chem.    1st Citation

---

The Biochemical Society, London © 1996