Canonical RTK-Ras-ERK signaling and related alternative pathways.

Receptor Tyrosine Kinase (RTK)-Ras-Extracellular signal-regulated kinase (ERK) signaling pathways control many aspects of C. elegans development and behavior. Studies in C. elegans helped elucidate the basic framework of the RTK-Ras-ERK pathway and continue to provide insights into its complex regulation, its biological roles, how it elicits cell-type appropriate responses, and how it interacts with other signaling pathways to do so. C. elegans studies have also revealed biological contexts in which alternative RTK- or Ras-dependent pathways are used instead of the canonical pathway.


Introduction
The small GTPase Ras is a central player in multiple different signaling pathways, but one of the most important and well conserved of these is the RTK-Ras-ERK pathway (Malumbres and Barbacid, 2002;Yoon and Seger, 2006;Karnoub and Weinberg, 2008;Lemmon and Schlessinger, 2010;Udell et al., 2011) (Figure 1A). RTK-Ras-ERK signaling is used repeatedly during animal development to control many different biological processes, and mutations affecting RTK-Ras-ERK signaling cause many human syndromes and diseases, including Noonan syndrome, Costello syndrome and many types of cancers (Karnoub and Weinberg, 2008;Tidyman and Rauen, 2009). This review summarizes work on RTK-Ras-ERK signaling in C. elegans. In an attempt to make this review accessible to a broad audience, I use the more familiar names of the mammalian proteins in conjunction with, or often in place of, the C. elegans nomenclature. For example, I refer to the C. elegans Ras protein as LET-60/Ras or simply Ras. See Figure  1 and Table 1 for corresponding names of core signaling proteins in C. elegans vs. mammals.

Biological roles of Ras signaling in C. elegans
Ras signaling is not required for cellular viability or proliferation in C. elegans; in a mosaic analysis, loss of let-60/Ras from large portions of the cell lineage was still compatible with continued cell division and organismal viability, albeit with specific developmental alterations (Yochem et al., 1997). Similarly, increased Ras activity does not promote somatic cell proliferation in C. elegans. The absence of cell-cycle related phenotypes has facilitated studies of Ras requirements in other specific developmental, behavioral and physiological processes.

Excretory duct cell fate and differentiation
LIN-3-LET-23 signaling via Ras-ERK promotes multiple aspects of excretory duct cell development, including excretory duct vs. excretory pore fate specification and other aspects of duct tube morphogenesis and differentiation (Yochem et al., 1997;Abdus-Saboor et al., 2011). The excretory duct cell is a tubular component of the worm's renal-like organ and is required for osmoregulation (Nelson et al., 1983;Nelson and Riddle, 1984). Therefore, let-60/Ras loss-of-function mutants, which lack the excretory duct cell, die as "rod-like" L1 larvae with a turgid, fluid-filled appearance ( Figure 3B). let-60/Ras gain-of-function mutants (which are viable) have two fused excretory duct cells (Yochem et al., 1997;Abdus-Saboor et al., 2011). The excretory duct cell was the sole lethal focus identified in a mosaic analysis of let-60 ras (Yochem et al., 1997).
A mosaic analysis of let-23/EGFR identified two lethal foci, one corresponding to the ABpl lineage (which generates the excretory duct) and another corresponding to the ABa lineage (Koga and Ohshima, 1995). Therefore, let-23 appears to have an additional, as yet unknown, essential role.

Hypodermal fluid homeostasis
LET-756-EGL-15 signaling via Ras-ERK also appears to have essential roles in fluid homeostasis, but in the hypodermis. egl-15/FGFR mutant lethality can be rescued by constitutive activation of LET-60/Ras in the hypodermis Bulow et al., 2004). The fact that mosaic analysis of let-60 (Yochem et al., 1997) did not identify a discrete cellular focus for this lethal phenotype is consistent with a requirement for signaling in hyp7, a syncitium that derives from multiple cell lineages (Sulston et al., 1983). The physiological basis for egl-15/FGFR lethality or for the related "scrawny" phenotype has not been determined, but it has been speculated to involve desiccation and thus be the converse of the Clr phenotype described below .
Hyperactivation of egl-15/FGFR or let-60/Ras, as seen in clr-1 mutants, can result in fluid accumulation and lethality with a "Clear" (Clr) appearance, and these defects are suppressed by loss of mpk-1 or other downstream genes (the Suppressor of Clr or "Soc" phenotype) (Kokel et al., 1998;Schutzman et al., 2001). Expression studies, tissue-specific transgenes, and mosaic experiments suggest that the Clr phenotype is caused by hypodermal defects rather than excretory system defects (Huang and Stern, 2004). Indeed, the hypodermis is one major site of osmosensitive gene regulation (Lamitina et al., 2006). EGL-15 apparently plays a physiological role in osmoregulation and not a developmental role, since hyperactivation of EGL-15 at any larval stage leads to the Clr phenotype.

Germline development
mpk-1/ERK has genetically separable roles in multiple aspects of germline development, as outlined below. mpk-1 acts cell autonomously in the germline for these functions, and the pattern of MPK-1 phosphorylation in the germline is dynamic, consistent with these multiple roles (Lee et al., 2007). The entire Ras-ERK cascade is involved in many cases, as indicated, but in other cases the evidence is incomplete. Note that the relevant receptors that promote LET-60/Ras activation in the germline are unknown, although requirements for an adaptor protein, ROG-1, suggest that RTK(s) are involved at least in part (Matsubara et al., 2007) ( Figure 1D). meiotic pachytene progression-Ras-ERK signaling promotes progression through the pachytene stage of meiosis I prophase (Church et al., 1995). Null mutants have germ cells arrested at mid-pachytene (Lee et al., 2007) (Figure 3B). pachytene cellular organization-Ras-ERK signaling promotes proper germ cell membrane organization (Lee et al., 2007;Lin and Reinke, 2008;Arur et al., 2011). Null mutants have arrested meiotic pachytene nuclei that are clumped together and lack the normal honeycomb pattern of membrane organization ( Figure 3B). germ cell sexual fate-Raf-ERK signaling promotes the male germ cell fate (Lee et al., 2007). lin-45/Raf, mek-2/MEK and mpk-1/ERK null mutants lack sperm and have a feminized germline in which all germ cells adopt a female or oocyte fate. However, these mutants fail to form normal oocytes due to the pachytene arrest and membrane disorganization phenotypes described above. Note that let-60/Ras and ksr-2 mutants do make sperm Ohmachi et al., 2002;Lee et al., 2007). This suggests a potential Rasindependent role of Raf-MEK-ERK, although maternal rescue cannot be ruled out.
oocyte growth and differentiation-ERK signaling promotes oocyte differentiation and restrains oocyte growth (Lee et al., 2007). Mutants with reduced ERK signaling have oocytes with abnormal positions, shapes and sizes and/or oocytes that have multiple or abnormally positioned nuclei. oocyte maturation-ERK signaling promotes sperm-dependent oocyte maturation (i.e., the transition from diakinesis to metaphase of meiosis I) (McCarter et al., 1999;Han et al., 2010;Kim et al., 2012). Maturation occurs just prior to ovulation (movement of oocytes into the spermatheca for fertilization). Mutants with reduced ERK signaling show a failure in maturation and ovulation or a significant delay in maturation, with ovulation following at the normal time (Lee et al., 2007). Note that while LIN-3-LET-23 do not function in maturation, they act by a distinct mechanism to promote ovulation (see below). germline apoptosis-Ras-ERK signaling appears to affect germline apoptosis in multiple ways. Null mutants do not undergo physiological germline apoptosis (Gumienny et al., 1999;Kritikou et al., 2006), but show meiotic arrest prior to the late pachytene stage, when apoptosis normally occurs (Lee et al., 2007). Increasing or partially reducing MPK-1 suggests roles both in promoting (Gumienny et al., 1999;Kritikou et al., 2006) and inhibiting (Arur et al., 2009) physiological apoptosis. MPK-1 also promotes DNA-damage induced apoptosis (Rutkowski et al., 2011). mitosis to meiosis transition-ERK signaling also appears to affect the transition from mitosis to meiosis in multiple ways. RNAi against mpk-1 leads to an expanded germline mitotic zone and a decreased early meiotic transition zone, suggesting that signaling promotes meiotic entry (Lee et al., 2006;Lee et al., 2007). On the other hand, mpk-1 null mutants have a slightly decreased mitotic zone and enhance glp-1/Notch defects (Lee et al., 2007). Note that phosphorylated MPK-1 has not been detected in the distal germline, so some of these effects could be indirect (Lee et al., 2007).

Vulval cell fates
LIN-3-LET-23 signaling via Ras cooperates with LIN-12/Notch and Wnt signaling to promote hermaphrodite vulval development (see Vulval development). Ras acts via the Raf-MEK-ERK cascade to promote the 1° vulval fate and acts via the RalGDS pathway as well as non-autonomously via Notch signaling to promote the 2° vulval fate (Simske and Kim, 1995;Chen and Greenwald, 2004;Zand et al., 2011). Reduced Ras signaling causes a vulvaless or Vul phenotype ( Figure 3B), whereas increased Ras signaling leads to a multivulva or Muv phenotype. Vulvaless animals cannot lay eggs, leading to a "bag of worms" phenotype in which progeny hatch internally.

Uterine uv1 cell fate
LIN-3-LET-23 signaling via Ras-ERK promotes the uterine uv1 vs. utse cell fate . uv1 is important for establishing a proper vulval-uterine connection, and in its absence, hermaphrodites cannot lay eggs.

Male spicule fates
LIN-3-LET-23 signaling via Ras-ERK promotes several different male spicule fates that derive from the B ectoblast (Chamberlin and Sternberg, 1994;Seah and Sternberg, 2009). B.a gives rise to four anterior-posterior pairs of cells that form equivalence groups . Reduced Ras signaling causes anterior-to-posterior fate transformations within each of these equivalence groups (Chamberlin and Sternberg, 1994). The resulting spicule defects prevent males from mating effectively.

Aging
Ras signaling promotes longevity, potentially via multiple mechanisms. First, as animals age, LIN-3-LET-23 signaling acts via Ras-ERK and EOR-1 to upregulate skr-5 and other genes involved in the ubiquitin proteasome system (UPS), thereby promoting proper protein homeostasis . This pathway appears to operate broadly in epithelial cells such as the intestine and hypodermis. Second, the Raf-MEK-ERK cascade stimulates the activity of the SKN-1 transcription factor to repress insulin-like peptide expression (Okuyama et al., 2010). Whether Ras acts upstream in this process is unknown. Third, EGL-17 and EGL-15A also promote longevity, but whether they do so through Ras is unknown (Chateau et al., 2010). Loss of egl-17/FGF or egl-15A/FGFR function or RNAi knockdown of lin-45/Raf, mek-2/MEK or mpk-1/ERK shortens adult lifespan, while RNAi of the MPK-1 phosphatase gene lip-1 increases lifespan (Chateau et al., 2010;Okuyama et al., 2010). LET-23 also plays Ras-independent roles in aging (Iwasa et al., 2010) (see below).

Sex myoblast migration
EGL-17-EGL-15 signaling via Ras-ERK in sex myoblasts helps to specify the proper endpoint of sex myoblast migration Sundaram et al., 1996;Burdine et al., 1997). Two different splice isoforms of EGL-15 mediate chemoattractive and chemorepulsive mechanisms that position the SMs with respect to the gonad and vulva (Stern and Horvitz, 1991;Branda and Stern, 2000;Lo et al., 2008). In egl-15 or let-60/Ras null mosaics, sex myoblasts adopt a broadened range of final positions, whereas in egl-15Aspecific mutants (which perturb chemoattraction), sex myoblasts are displaced more posteriorly (Sundaram et al., 1996;Lo et al., 2008). The resulting defects in sex muscle position can affect egg-laying.

Sex muscle differentiation
The sex myoblasts divide to generate the sex muscles, which attach to the vulva and control egg-laying (Sulston and Horvitz, 1977). Although reduced EGL-15-Ras signaling does not perturb sex muscle differentiation, hyperactive signaling by EGL-15A prevents terminal differentiation and thus causes a strong egg-laying defect (Sasson and Stern, 2004).

Body muscle maintenance
In addition to causing the Clr phenotype, hyperactivating EGL-15-Ras-ERK in adults induces body muscle wasting and failure to express a muscle gene reporter (Szewczyk et al., 2002;Szewczyk and Jacobson, 2003). This has been proposed to be a direct effect in muscle.

Axon outgrowth
LET-756-EGL-15 signaling via Ras in the hypodermis influences axon extension and pathfinding by certain ventral cord neurons that migrate over the hypodermis . Whereas in wild-type these neurons extend axons along one side of the ventral midline, in mutants the axons are shorter than normal and/or wander across the midline. Whether Ras acts through Raf-MEK-ERK or through another pathway has not been determined.

Body muscle membrane extension
LET-756-EGL-15 signaling via Ras in body muscles restrains the outgrowth of muscle membrane extensions at neuro-muscular junctions (NMJs) (Dixon et al., 2006). Reduced signaling leads to ectopic muscle extensions, while increased signaling leads to fewer muscle extensions. Whether Ras acts through Raf-MEK-ERK or through another pathway has not been determined.

Acetylcholine receptor (AchR) synaptic clustering
EGL-15-Ras signaling in body muscles promotes AchR clustering and/or maintenance at NMJs (Gottschalk et al., 2005). Reduced signaling leads to reduced AchR expression at NMJs and confers moderate resistance to AchR agonists such as nicotine and levamisole. The ligand involved, and whether Ras acts through Raf-MEK-ERK or through another pathway, has not been determined.

Olfaction
Ras-ERK signaling in the mature AWC neuron is required for sensitivity to volatile attractants (Hirotsu et al., 2000). Whereas wild-type C. elegans will chemotax toward volatile attractants such as isoamylalcohol and diacetyl, mutants with reduced Ras signaling fail to chemotax towards such attractants. The receptor mediating this response is unknown, but it appears to be a G-protein coupled receptor that acts through EGL-30/Gαq to stimulate EGL-8/PLCβ and DAG production, activating the Ras GEF RGEF-1  ( Figure 1F). 6. Ras-independent roles of LET-23/EGFR, EGL-15/FGFR or the Raf-MEK-

ERK cascade
As mentioned above, there are several contexts in which LET-23/EGFR acts via alternative Ras-independent signaling pathways ( Figure 2). There also appears to be a Ras-independent mechanism for activating Raf-MEK-ERK.

Ovulation
LIN-3 and LET-23 signal through the PLCγ-IP 3 pathway to promote ovulation (Clandinin et al., 1998;McCarter, 1998;Bui and Sternberg, 2002;Miller et al., 2003) (Figure 2A). LIN-3 is required in the oocyte while LET-23 is required in the somatic gonadal sheath and spermathecal cells for dilation of the distal spermatheca and contraction of the myoepithelial sheath cells necessary for ovulation. When lin-3 or let-23 activity is reduced or absent, oocytes mature but somatic gonadal function is abnormal, resulting in failed ovulation into the spermatheca, with the unfertilized oocytes undergoing endoreduplication (the Emo phenotype).

Healthy aging
LET-23 signaling through the PLCγ-IP 3 pathway also promotes healthy aging (Iwasa et al., 2010). As animals age, their locomotion and swimming rates decline; this age-related decline is delayed in animals with increased LET-23 signaling and accelerated in animals with decreased signaling. The site of action of this signaling event is still unclear.

Behavioral quiescence
Over-expression of lin-3/EGF induces behavioral quiescence, a reversible, sleep-like state in which feeding and movement cease (Van Buskirk and . This involves another Ras-independent pathway consisting of LET-23, PLC-3/PLCγ and diacylglycerol, which stimulates synaptic vesicle release in the ALA neuron ( Figure 2B). let-23/EGFR and plc-3/PLCγ mutants show reduced behavioral quiescence during lethargus.

Axon maintenance
EGL-15A has a kinase-and presumably Ras-independent function in maintaining axon positions after outgrowth . This function was revealed by the ability of truncated EGL-15 transgenes to rescue axon positioning defects of egl-15 mutants.

Screens used to identify RTK-Ras-ERK pathway components, targets and regulators
A major strength of C. elegans as a model system is the ease of conducting unbiased forward genetic screens to identify mutants and genes relevant to a biological process of interest. Many studies on RTK-Ras-ERK signaling have focused on the role of the LET-23/ EGFR pathway in promoting vulval development (see Vulval development), and many core components of the pathway were first identified through forward genetic screens for mutants with Vul or Muv mutant phenotypes (Ferguson and Horvitz, 1985). Some components of the pathway were originally identified based on other mutant phenotypes described above such as sex myoblast migration defects (Stern and Horvitz, 1991) or germline meiosis defects (Church et al., 1995;Hsu et al., 2002;Ohmachi et al., 2002). Many relevant regulators and targets of the signaling pathway were identified as genetic modifiers (suppressors or enhancers) of other mutant phenotypes in forward genetic screens or reverse genetic tests. The genes identified in these types of genetic screens have revealed many mechanisms that influence the efficiency and spatial dynamics of RTK-Ras-ERK signaling, or that mediate its downstream responses, as discussed in the following sections.
Note that for many genes described below, single mutants have relatively mild phenotypes, yet can confer strong phenotypes in certain sensitized genetic backgrounds (Table 1). Given the many detrimental effects of misregulating Ras signaling, it is perhaps not surprising that, for most steps of the signaling pathway, there are multiple layers of regulation that function semi-redundantly.

Regulators of growth factor ligand expression
Transcriptional regulation of growth factor ligand expression is an important determinant of when and where RTK signaling occurs in C. elegans ( Figure 4). lin-3/EGF, egl-17/FGF and let-756/FGF all show relatively restricted patterns of gene expression. For example, lin-3 is expressed in the gonadal anchor cell during vulval development (Hill and Sternberg, 1992), in primary vulval descendants during uterine development , and in the excretory canal cell during excretory duct development (Abdus-Saboor et al., 2011). egl-17 is expressed in the gonad and in the primary vulval precursor cell P6.p during sex myoblast migration Branda and Stern, 2000). let-756 is expressed in body muscles and some neurons . egl-17 expression in P6.p and lin-3 expression in P6.p descendants both depend on LET-23/EGFR and LET-60/Ras-dependent cell fate specification Cui and Han, 2003). Simple overexpression of lin-3(+) is sufficient to cause gain-of-function phenotypes (Katz et al., 1995), and multiple genes that restrict or promote lin-3 expression have been identified.

Syn Muv genes
The "synthetic multivulva" (SynMuv) genes encode three sets of nuclear factors that function redundantly to inhibit ectopic lin-3 expression (Ferguson and Horvitz, 1989;Fay and Yochem, 2007). One set, the SynMuv B genes, encodes orthologs of a conserved transcriptional repression cascade including LIN-35/Rb (Lu and Horvitz 1998), DPL-1/DP and EFL-1/E2F (Ceol and Horvitz, 2001), a NuRD-like histone deacetylase complex including HDA-1/HDAC1 (Solari and Ahringer, 2000), an H3K9 histone methyltransferase MET-2/SETDB1 (Andersen and Horvitz, 2007) and the chromodomain protein HPL-2/HP1 (Couteau et al., 2002). The SynMuv C genes encode components of a Tip60/NuA4-like histone acetyltransferase complex (Ceol and Horvitz, 2004), while the SynMuv A genes include several THAP-domain proteins that are thought to act as transcriptional repressors (Huang et al., 1994;Davison et al., 2005;Davison et al., 2011). A point mutation in the lin-3 gene promoter behaves like a SynMuv A mutation, suggesting that synMuv A proteins may act directly on the lin-3 promoter to repress ectopic lin-3 expression . It is not clear if the other SynMuv complexes act directly or indirectly on lin-3. Although single mutants for any given SynMuv gene have relatively normal levels of lin-3 expression, double mutants for a SynMuv A and a SynMuv B mutation have increased levels of lin-3 expression and several corresponding Ras(gf)-like phenotypes such as Muv . Single molecule RNA FISH demonstrated that SynMuv mutants have ectopic lin-3 mRNA in most cells of the body, but that elevation in lin-3 mRNA levels is quite modest on a per cell basis, highlighting the sensitivity of signaling to ligand levels . Functional analysis suggests that the Muv phenotype primarily results from derepression of lin-3 in hyp7 (Myers and Greenwald, 2005;Cui et al., 2006).

Transcription factors that promote lin-3 expression
In the anchor cell, lin-3 expression requires the E-protein/daughterless homolog HLH-2 and an unidentified nuclear hormone receptor (NHR); a 59 bp lin-3 enhancer element containing predicted E-box and NHR binding sites is both necessary and sufficient for anchor cell expression (Hwang and Sternberg, 2004). In the primary vulval lineage, lin-3 expression requires the Pax family transcription factor EGL-38 .

Regulators of growth factor ligand activity
Proteolytic processing and secretion of ligands and interaction of ligands with receptors are additional steps at which signaling can be regulated (Figure 4).

LIN-3/EGF proteolysis and signaling
Like other ligands of the EGF family, LIN-3 is synthesized initially as a transmembrane protein but may be subsequently cleaved proteolytically to generate a diffusible form (Hill and Sternberg, 1992;Blobel et al., 2009). There are at least four different splice isoforms of LIN-3 that differ in their juxta-membrane regions, potentially affecting proteolysis (Dutt et al., 2004;Van Buskirk and Sternberg, 2007). Biological activity of one specific isoform, LIN-3L, appears to require proteolysis by the Rhomboid ortholog ROM-1 . rom-1 mutants have only subtle phenotypes, suggesting either that other LIN-3 proteases exist or that other LIN-3 isoforms can function effectively in an unprocessed, transmembrane form.
In Drosophila, various extracellular factors influence EGF-like ligand diffusion and interactions with receptors (Klein et al., 2004;Miura et al., 2006;Pizette et al., 2009). There is emerging evidence for this type of regulation in C. elegans as well. The secreted proteins HPA-1 and HPA-2 limit LIN-3 signaling during adult aging, potentially by LIN-3 sequestration (Iwasa et al., 2010). The secreted lipocalin LPR-1 facilitates LIN-3 signaling during excretory duct cell development (C. Stone and M. Sundaram, in preparation).

FGF secretion and signaling
Both C. elegans FGF-like ligands utilize unusual pathways for secretion. EGL-17 is secreted via a mechanism that requires the lipoprotein receptor-related proteins LRP-1 and LRP-2 and the Disabled-related adaptor DAB-1 (Kamikura and Cooper, 2003). LET-756 lacks a classical signal peptide, and like mammalian FGF9, its secretion depends on an internal hydrophobic domain (Popovici et al., 2004). Also like FGF9, LET-756 shows partial localization to the nucleus (Popovici et al., 2006), though the significance of this localization remains unknown.
In C. elegans, as in other organisms, FGF signaling can be influenced by the extracellular matrix (Bulow and Hobert, 2006). Heparan sulfate polysaccharides facilitate EGL-17dependent axon outgrowth (Tornberg et al., 2011). The matrix component unc-52/perlecan genetically interacts with egl-17 to influence distal tip cell migration (Merz et al., 2003). The ability of EGL-17 vs. LET-756 to signal through EGL-15/FGFR also depends to some degree on alternative splicing of egl-15 (see below).

EGL-15/FGFR alternative splicing and glycosylation
EGL-15/FGFR activity is regulated in part via alternative splicing and other modifications to its extracellular domain that may affect ligand binding. The EGL-15(A) and EGL-15(B) splice isoforms differ in a C. elegans-specific region of the extracellular domain, and EGL-15(A) primarily mediates EGL-17/FGF-dependent signaling, while EGL-15(B) primarily mediates LET-756/FGF-dependent signaling Lo et al., 2008). However, these functional distinctions are not complete, and both ligands can act to some degree through both receptor isoforms Bulow et al., 2004;Dixon et al., 2006). At least five other alternative splice forms differ in the EGL-15 extreme C-terminus, potentially affecting interactions with downstream targets. EGL-15 is Nglycosylated in vivo and this modification appears to negatively regulate its activity, potentially by interfering with ligand binding or self-association (Polanska et al., 2009). EGL-15A activity may be positively regulated by two C. elegans orthlogs of Klotho, a mammalian FGFR binding protein and co-receptor (Chateau et al., 2010;Polanska et al., 2011).

Cytosolic phosphatases and kinases
The cytosolic tyrosine phosphatase PTP-2/Shp2 positively regulates LET-23 and EGL-15 RTK signaling Schutzman et al., 2001), and the cytosolic tyrosine kinase ARK-1 (Ack-related) negatively regulates LET-23 RTK signaling . The substrates of PTP-2 and ARK-1 are still unclear, but these proteins can bind to the adaptors SOC-1 and SEM-5, respectively, and appear to act at a step immediately downstream of the RTKs to influence the strength or kinetics of signaling. In the germline, PTP-2 antagonizes the activity of GAP-2 and GAP-3, suggesting these could be relevant PTP-2 substrates (Yang et al., 2010). Evidence from mammalian cells suggests that Ack tyrosine kinases promote EGFR internalization and degradation (Shen et al., 2007).

Regulators of LET-60/Ras activity or expression GEFs and GAPs
As is the case for other small GTPases, the activity of LET-60/Ras is controlled by Guanine Nucleotide Exchange Factors (GEFs), which activate Ras by stimulating conversion of Ras-GDP to Ras-GTP, and by GTPase activating proteins (GAPs), which inactivate Ras by stimulating conversion of Ras-GTP to Ras-GDP (Malumbres and Barbacid, 2002;Karnoub and Weinberg, 2008) (Figure 5). Gain-of-function (gf) mutations lock Ras in the active, GTP-bound state; the let-60(gf) allele n1046 (G13E) has been widely used for genetic analyses . Dominant-negative (dn) mutations lock Ras in the inactive, GDP-bound state, causing it to bind stably to and titrate out GEFs; a variety of let-60(dn) mutations have been described (Han and Sternberg, 1991).
The GEF SOS-1 is required for most Ras-mediated developmental events . A temperature-sensitive allele, sos-1(cs41), is a useful tool for conditional inactivation of Ras signaling . A different GEF, RGEF-1 (related to mammalian RasGRPs), regulates LET-60 Ras in AWC neurons during olfaction .

Post-translational modifications of Ras
In other systems, a variety of post-translational modifications can modulate Ras trafficking and localization (Ahearn et al., 2012). For example, farnesylation of the Ras C-terminus promotes plasma membrane attachment. Chemical farnesyltransferase inhibitors suppress the Muv phenotype of let-60 ras(gf) mutants, suggesting that farnesylation promotes LET-60/Ras activity (Hara and Han, 1995). A negative regulator of LET-60/Ras is SUR-5, a protein of unknown function that resembles acetyl coenzyme A synthetases .

miRNA-dependent regulation of Ras
The overall level of let-60 ras expression can also influence Ras signaling activity, as overexpression of wild-type LET-60 leads to a Muv phenotype (Sundaram and Han, 1995). In vulval precursor cells and hypodermal seam cells, normal levels of let-60 ras depend on negative regulation by the let-7 family microRNAs mir-84 and let-7 (Johnson et al., 2005). These microRNAs appear to act directly on the let-60 3′ UTR, and this mode of regulation is conserved in mammalian ras genes (Johnson et al., 2005).

Regulators of the Raf-MEK-ERK kinase cascade
Raf, MEK and ERK activities depend in large part on their phosphorylation status (Udell et al., 2011). For example, Raf must be de-phosphorylated at certain sites and phosphorylated at other sites in order to become active (Chong et al., 2001) ( Figure 6A). Transgenic, hyperactive forms of LIN-45/Raf, MEK-2/MEK and MPK-1/ERK were generated by mutating relevant sites to prevent or mimic phosphorylation; such transgenes have been useful for epistasis experiments Chong et al., 2001;Kao et al., 2004;Yoder et al., 2004;Rocheleau et al., 2005). Commercially available antibodies also recognize phosphorylated MPK-1 by Western analyses and (in some cases) immunofluorescence, and have been very useful for molecular epistasis (Hirotsu et al., 2000;Page et al., 2001;Ohmachi et al., 2002;Lee et al., 2007;Chen et al., 2011).

KSR and other scaffold proteins
Positive regulators of Raf-MEK-ERK include so-called "scaffold" proteins ( Figure 6A) -KSR-1 (Kornfeld et al., 1995b;Sundaram and Han, 1995), KSR-2 (Ohmachi et al., 2002), CNK-1 (Rocheleau et al., 2005) and SOC-2 (also known as SUR-8) Sieburth et al., 1998). These proteins have no obvious enzymatic activity, but bind to one or more core kinases in the cascade and are thought to bring different components together at the right place and time (Udell et al., 2011). KSR-1 and KSR-2 are required downstream of activating Raf phosphorylation but upstream of MEK-ERK (Ohmachi et al., 2002;Rocheleau et al., 2005). CNK-1 is required upstream of Raf activating phosphorylation (Rocheleau et al., 2005), while SOC-2 is required upstream of (or during) Raf inhibitory phosphate removal . These epistasis results fit well with the observed protein interactions of the mammalian scaffolds ( Figure 6B): Mammalian KSR binds constitutively to MEK and variably to Raf and ERK (Xing et al., 1997;Yu et al., 1998;Stewart et al., 1999), while CNK binds Raf (Yao et al., 2000;Lanigan et al., 2003), and SOC-2 binds to Ras and Raf (Li et al., 2000). In C. elegans, the KSR scaffold is a core pathway component but encoded by two paralogs: KSR-1 and KSR-2 are individually required for a few developmental processes, but redundantly required for many others (Ohmachi et al., 2002). CNK-1 and SOC-2 are not individually required under standard laboratory conditions, but help buffer signaling against other perturbations Sieburth et al., 1998;Rocheleau et al., 2005). Another positive regulator of Raf-MEK-ERK signaling is CGR-1, a CRAL/TRIO and GOLD domain protein that promotes signaling in the vulva, possibly by modulating Raf activation (Goldstein et al., 2006;Johnson and Kornfeld, 2010).

Regulators of the KSR scaffold
Several gene products appear to influence KSR scaffold activity ( Figure 6). The PAR-1 kinase inhibits Raf-MEK-ERK signaling Yoder et al., 2004) and is homologous to mammalian C-TAK1, which directly phosphorylates KSR to affect its localization (Ory et al., 2003). SUR-6, a PR55/B-type regulatory subunit of Protein Phosphatase 2A (PP2A-B), promotes Raf-MEK-ERK signaling  and has been suggested to dephosphorylate and activate both Raf and KSR (Ory et al., 2003;Kao et al., 2004;Yoder et al., 2004). Finally, the zinc transporter proteins CDF-1 and SUR-7 also promote Raf-MEK-ERK signaling in the vulva (Bruinsma et al., 2002;Yoder et al., 2004). High levels of zinc can influence KSR phosphorylation in mammalian cells, suggesting that zinc modulates the activity of a KSR kinase .

Cross-tissue interactions
There is evidence that some genes may influence Raf-MEK-ERK signaling indirectly, by affecting neighboring cells. Some of the clearest examples involve genes that function within gonadal sheath cells to affect gap-junction dependent inhibition of MPK-1/ERK phosphorylation in oocytes (Kim et al., 2012) (Figure 7). Also, the Gαq protein EGL-30 and the voltage-gated calcium channel EGL-19 function in neurons and muscle, respectively to promote Ras-mediated vulval induction . The G-protein coupled receptor SRA-13 and its Gα target GPA-5 inhibit Ras-mediated vulval induction, but it is unknown whether these genes function in vulval cells or in neurons . The zinc transporter CDF-1 can influence vulval induction when expressed in either the vulva or the intestine (Bruinsma et al., 2002). The mechanisms underlying these other cross-tissue effects are still unclear.

Targets of MPK-1/ERK, and other factors influencing downstream responses
The specific outcomes of RTK-Ras-ERK signaling in different tissues are thought to depend on the availability and use of different ERK substrates and on combinatorial action of ERK substrates with other tissue-specific or signal-regulated factors (Tan and Kim, 1999;Flores et al., 2000;Halfon et al., 2000;Barolo and Posakony, 2002;Arur et al., 2009). In C. elegans somatic cell fate decisions, the combinatorial mechanism appears quite relevant, since two identified MPK-1/ERK targets (LIN-1 and EOR-1) are widely expressed and required for many different outcomes (Beitel et al., 1995;Jacobs et al., 1998;Howard and Sundaram, 2002;Rocheleau et al., 2002;Howell et al., 2010) (Figure 1B; Table 2). LIN-1 and EOR-1 are both nuclear transcription factors, consistent with the idea that transcriptional control of gene expression is the critical mechanism for cell fate specification. The situation is different in the germline, where there are many identified targets of MPK-1, few of which are transcription factors and each of which affects only a small subset of MPK-1-dependent processes (Arur et al., 2009) (Figure 1D; Table 2). In both the soma and the germline, a common theme is redundancy among different MPK-1/ERK substrates.

LIN-1/Ets
The Ets domain transcription factor LIN-1 is arguably the most important MPK-1 substrate in the soma (Beitel et al., 1995;Jacobs et al., 1998) (Figure 1B; Table 2). LIN-1 is related to the mammalian ERK target Elk1 (Sharrocks, 2002;Yang et al., 2003). LIN-1 is a direct substrate of ERK in vitro, and studies of lin-1 gain-of-function mutations first identified the FXF ERK docking site, which confers high affinity binding between ERK and many of its substrates, including Elk1 (Jacobs et al., 1998;Jacobs et al., 1999;Sharrocks et al., 2000). lin-1 null mutants have phenotypes resembling those of let-60 ras gain-of-function mutants and are epistatic to mpk-1, suggesting that Ras-ERK signaling antagonizes LIN-1 activity (Beitel et al., 1995;Lackner and Kim, 1998). However, some evidence suggests that LIN-1 has an additional positive role in Ras signaling, since removal of lin-1 in certain genetic backgrounds can cause Ras loss-of-function phenotypes Tiensuu et al., 2005;Abdus-Saboor et al., 2011) (see below). The biochemical basis for these opposite roles is not yet clear. One simple model consistent with studies of Elk1 (Boyer et al., 1999;Yang et al., 2003) is that MPK-1/ERK phosphorylation converts LIN-1 from a transcriptional repressor to a transcriptional activator in order to turn on signaldependent gene expression ( Figure 1B). Alternatively, MPK-1/ERK phosphorylation may simply inactivate LIN-1, and unphosphorylated LIN-1 may repress or activate different targets. The dual positive and negative roles of LIN-1 could indicate that LIN-1 has some targets that have positive roles and some targets that have negative roles.
There is evidence for LIN-1 having repressor activity in cell culture and in vivo (Leight et al., 2005). The repressive role of LIN-1 requires sumoylation, which then allows a SUMOdependent interaction with MEP-1, a component of the C. elegans NuRD-like repressor complex (Leight et al., 2005). LIN-1 also physically associates with the NuRD-like component LET-418/Mi-2 (Guerry et al., 2007). Two directly repressed targets of LIN-1 in the vulva are the Hox gene lin-39 (Wagmaister et al., 2006b;Guerry et al., 2007) and the Notch pathway ligand gene lag-2 (Zhang and Greenwald, 2011).
Evidence for a LIN-1 activator function is more limited, but unphosphorylated LIN-1 appears to directly activate the Rho kinase gene let-502 during vulva development (Farooqui et al., 2012). Ras-dependent upregulation of lag-2 appears to occur solely through relief of LIN-1-dependent repression (Zhang and Greenwald, 2011).

The Mediator complex
Two widely important positive factors that act in conjunction with LIN-1/Ets are the Mediator subunit SUR-2/Med23 and its partner LIN-25 (Singh and Han, 1995;Tuck and Greenwald, 1995;Nilsson et al., 1998) (Figure 1B). The conserved Mediator complex links certain sequence-specific DNA binding proteins to the general RNA Polymerase II transcriptional machinery (Borggrefe and Yue, 2011). SUR-2 and LIN-25 function together downstream or parallel to MPK-1, and they have very specific Ras-related phenotypes including a partial Vul phenotype and low penetrance defects in excretory duct and P12 development (Singh and Han, 1995;Tuck and Greenwald, 1995;Nilsson et al., 1998;Nilsson et al., 2000). Mammalian Sur2 binds to ERK-phosphorylated Elk1 to recruit Mediator and allow for transcriptional activation of target genes (Boyer et al., 1999;Stevens et al., 2002). The genetics of C. elegans sur-2 and lin-1/Ets are compatible with such a model, though physical interactions between the worm proteins have not been demonstrated and no directly activated targets of phosphorylated LIN-1 or SUR-2 have been identified to date. Analyses of the Ras-responsive lag-2 promoter suggest that, in addition to (or instead of) acting with LIN-1, SUR-2 must function with other unidentified transcription factors (Zhang and Greenwald, 2011).
Another component of the Mediator complex, DPY-22/Med12, functions antagonistically to LET-23/EGFR-Ras-ERK during vulval development, though the precise point at which it acts has not been determined .

EOR-1 and EOR-2
Two other widely important positive factors downstream of MPK-1 are the BTB/Zinc finger protein EOR-1 and its binding partner EOR-2 Howell et al., 2010) (Figure 1B; Table 2). EOR-1 contains two "D-domain" type ERK docking sites and is an excellent substrate of ERK in vitro, and mutation of one of these docking sites or of several potential phosphoacceptors reduces EOR-1 function in vivo . EOR-1 is related to known transcriptional activators and repressors, including Drosophila GAGA factor and the mammalian oncoproteins PLZF and BCL-6 (Collins et al., 2001;Barna et al., 2002), but it is not known if it acts primarily as an activator or a repressor (or both) or how phosphorylation affects its activity. eor-1 and eor-2 mutants have incompletely penetrant let-60 ras-like defects in excretory duct and P12 development  and strong defects in age-dependent ubiquitination , as well as several phenotypes that are apparently unrelated to Ras signaling (Hoeppner et al., 2004). EOR-1 and EOR-2 act redundantly with SUR-2 and LIN-25, such that eor-1; sur-2 double mutants have multiple strong let-60 ras-like loss-of-function phenotypes . EOR-1 and EOR-2 also function redundantly with the cryptic positive function of LIN-1, such that lin-1 eor-1 double mutants also have strong let-60 ras-like lossof-function phenotypes Abdus-Saboor et al., 2011). Thus, together, LIN-1 and EOR-1 appear to mediate many of the somatic effects of Ras signaling.
Hundreds of potential EOR-1 targets have been identified by the modencode project using ChIP-Seq (Niu et al., 2010), but the biological relevance of these targets is still unclear.

LIN-31
The forkhead transcription factor LIN-31 functions downstream of MPK-1 during vulval development (Miller et al., 1993;Tan et al., 1998) and male spicule development ). LIN-31 appears to be a tissue-specific MPK-1 substrate since it contains a potential ERK docking site (Swathi Arur and Tim Schedl, pers. comm.) and mutation of putative ERK phosphorylation sites in LIN-31 alters function in transgenic assays (Tan et al., 1998). lin-31 mutants have a mixed Vul and Muv phenotype, suggesting that LIN-31 both promotes and inhibits vulval development (Miller et al., 1993). Unphosphorylated LIN-31 appears to bind LIN-1 and inhibit vulval development, while phosphorylated LIN-31 promotes vulval development (Tan et al., 1998). lin-31 has a relatively restricted expression pattern, and ectopic expression of LIN-31 can cause other tissues to adopt some vulval-like characteristics, supporting the model that tissue-specific availability of ERK substrates is one key determinant of tissue-specific responses (Tan et al., 1998;Tan and Kim, 1999).

Hox genes
A common theme among many LET-23/EGFR-Ras-ERK-dependent cell fates is a requirement for specific Hox genes ( Figure 1B). In C. elegans, as in other animals, different Hox genes are expressed along the anterior-posterior axis (Aboobaker and Blaxter, 2003). The Hox genes lin-39, egl-5, mab-5 and ceh-13 are upregulated by Ras signaling and/or are required for Ras-dependent induction of vulval, P12, male hook and male spicule fates, respectively (Clandinin et al., 1997;Eisenmann et al., 1998;Maloof and Kenyon, 1998;Wagmaister et al., 2006a;Seah and Sternberg, 2009;Yu et al., 2010). Furthermore, ectopic expression of some Hox genes can change how cells respond to Ras activation (Clandinin et al., 1997;Maloof and Kenyon, 1998;Yu et al., 2010). It therefore appears that the Hox code of a given cell is one important contextual factor that determines tissue-specific outcomes of Ras signaling in the soma.

Germline MPK-1 substrates
In the germline, MPK-1 utilizes a completely different and much broader set of substrates than those described above ( Figure 1D; Table 2). Arur et al., (2009) used known ERK docking site sequences (Sharrocks et al., 2000) to identify candidate MPK-1 substrates and then systematically tested their requirements in the germline by RNAi. These studies identified 30 likely MPK-1 substrates that control specific steps of germline development. In general, multiple substrates influenced each step examined, and depletion of individual substrates caused only mild phenotypes, suggesting a high level of substrate redundancy. Many germline MPK-1 substrates also influenced the spatial pattern of MPK-1 phosphorylation, suggesting that they participate in positive or negative feedback loops.

Interactions between the RTK-Ras-ERK pathway and other signaling pathways
The RTK-Ras-ERK pathway often interacts with other signaling pathways to control downstream responses.

Ras-Wnt cooperation
The LET-23/EGFR-Ras-ERK pathway cooperates with a Wnt pathway (see Wnt signaling) to specify vulval fates (Eisenmann et al., 1998;Gleason et al., 2002), P12 fates (Jiang and Sternberg, 1998), and male hook group (Yu et al., 2009) and spicule ) fates ( Figure 8A). The relative importance of Ras vs. Wnt signaling varies among these different tissues-for example, Ras is more important in the vulva and Wnt is more important in the male hook group-but hyperactivation of the "minor" pathway can often bypass the requirement for the "major" pathway, suggesting that both pathways converge on common targets (Gleason et al., 2002;Yu et al., 2009). Some of these common targets may include Hox genes. Interestingly, the relative importance of Ras vs. Wnt signaling also varies between different nematode species -in Pristionchus pacificus, vulval development depends primarily on Wnt signaling (Tian et al., 2008).

Ras-Notch sequential signaling
The LET-23/EGFR-Ras-ERK pathway acts sequentially with a Notch pathway (see  to promote different cell fates during development of the vulva (see Vulval development), uterus , male hook (Yu et al., 2009) andexcretory system (Abdus-Saboor et al., 2011). In the uterus, LIN-12/Notch signaling induces the fate of a progenitor cell (the π cell) Newman et al., 1995) whose descendants later require Ras signaling to adopt their final fate (uv1) . In several cases, signaling by one pathway controls the fate of a cell type that expresses a ligand for the other pathway. For example, Ras signaling induces the primary vulval cell type, which then expresses Notch pathway ligands to induce the secondary cell fate (Simske and Kim, 1995;Chen and Greenwald, 2004). LIN-12/Notch signaling inhibits generation of the anchor cell (Greenwald et al., 1983) and LIN-12 and GLP-1/Notch signaling promotes generation of the excretory canal cell (Lambie and Kimble, 1991;Moskowitz and Rothman, 1996), both of which express LIN-3/EGF to induce other cell fates (Hill and Sternberg, 1992;Abdus-Saboor et al., 2011). In these various cases, sequential signaling allows for the properly patterned generation of multiple cell types that make up a functional organ.

Coordination of related developmental events
In addition to the Ras-Notch sequential signaling events described above, there are several additional examples where different RTK-or Ras-dependent pathways act together to coordinate related developmental events. For example, after LET-23/EGFR-Ras-ERKdependent specification of the 1° vulval cell fate, 1° vulval descendants then express EGL-17/FGF and LIN-3/EGF to coordinate sex muscle positioning and uterine development with vulval development . ERK signaling in oocytes and LET-23-PLCγ-IP 3 signaling in somatic gonadal cells act together to coordinate oocyte maturation and ovulation (Kim et al., 2012).

Summary and relevance to human health
Studies in C. elegans have been instrumental in elucidating the basic framework of the RTK-Ras-ERK pathway and many aspects of its regulation and cross-talk with other pathways. It is now clear that each step of the RTK-Ras-ERK pathway is subject to multiple regulatory mechanisms that facilitate or dampen signaling. Many of these mechanisms appear redundant under standard laboratory conditions, such that loss of a single regulatory gene has few phenotypic consequences, and some vary in importance between tissue types. There is also significant redundancy and tissue-specificity among downstream ERK targets. Although much has been learned, many questions remain about how regulatory mechanisms actually work and how use of different ERK targets ultimately leads to the many different and interesting biological outcomes observed.
Importantly, many of the regulatory mechanisms and targets described thus far in C. elegans are widely conserved in other organisms, including humans (Tables 1, 2). Therefore, these represent possible points of intervention in the treatment of diseases that involve aberrant RTK-Ras-ERK signaling. For example, given that hyperactivation of RTKs or Ras is a leading cause of many cancers (Malumbres and Barbacid, 2002;Yoon and Seger, 2006;Karnoub and Weinberg, 2008;Lemmon and Schlessinger, 2010;Udell et al., 2011), drugs targeting positive regulators of the signaling pathway could be one promising approach for cancer therapy. Indeed, antisense oligonucleotides targeting KSR1 have been effective in treating mouse models of pancreatic cancer, effectively suppressing Ras-dependent tumorigenesis with minimal toxicity to wild-type cells (Xing et al., 2003). There are still many challenges to developing clinically-effective drugs, but studies in C. elegans have contributed to a long list of potentially useful targets.

Figure 1. The canonical RTK-Ras-ERK signaling pathway and alternative Ras-dependent pathways in C. elegans
A) The canonical pathway. Positively acting factors are shown in green; negatively acting factors are in red. Only the most widely required and conserved factors are shown. GF, growth factor. TX, transcription. B) LET-23/EGFR and C) EGL-15/FGFR both act in part through the canonical pathway, but they are activated by different ligands and control different sets of developmental processes. Included in B) are nuclear proteins, including two MPK-1/ERK substrates (LIN-1/ETS and EOR-1), that are jointly important for most LET-23-dependent somatic cell fates. Downstream targets of EGL-15-Ras-ERK signaling are not yet known. D) Ras-ERK functions to promote meiotic pachytene progression and various other aspects of germline development; the receptors that activate Ras-ERK for these functions are not currently known. MPK-1/ERK utilizes more than 30 different substrates in the germline. E) A G-protein coupled receptor (GPCR) may activate Ras-ERK signaling in some contexts. F) LET-60/Ras can signal through the alternative Ral pathway. See text for additional details and references.     (Chong et al., 2001;Udell et al., 2011). Epistasis experiments with LIN-45/Raf phosphorylation site variants demonstrated that SOC-2 is required upstream or during inhibitory phosphorylation removal , while CNK-1 is required downstream of inhibitory phosphate removal but upstream of activating phosphorylation addition (Rocheleau et al., 2005). KSR-1 and SUR-6 are required downstream of both events Yoder et al., 2004;Rocheleau et al., 2005). B) Protein interactions of related scaffold proteins in mammalian cells (Udell et al., 2011). C-TAK1 is a mammalian homolog of the PAR-1 kinase (Muller et al., 2001). See text for additional details and references.