|Home | About | Journals | Submit | Contact Us | Français|
The C. elegans vulva is patterned by epidermal growth factor (EGF) activation of Ras to control 1° fate, and 1° fate induces antagonistic Notch-dependent 2° fate. Furthermore, a spatial EGF gradient, in addition to inducing 1° fate, directly contributes to 2° fate via an unknown pathway. We find that in addition to its canonical effector, Raf, vulval Ras utilizes an exchange factor for the Ral small GTPase (RalGEF), such that Ras-RalGEF-Ral antagonizes Ras-Raf pro-1° fate activity. Consistent with its restricted expression pattern, Ral participates in EGF pro-2° activity. Thus, we have delineated a Ras effector-switching mechanism whereby position within the morphogen gradient dictates that Ras effector usage is switched to RalGEF from Raf to promote 2° instead of 1° fate. Our observations define the utility of Ras effector switching during normal development, and may provide a possible mechanistic basis for cell and cancer type differences in effector dependency and activation.
An emerging complexity of mammalian Ras signal transduction is the assortment of catalytically diverse effectors that may facilitate the elaborate biological activities of Ras in normal and neoplastic cells. The precise role that each effector serves, dynamic regulation of effector utilization, and interplay between effector networks are issues that remain poorly understood. Analysis of C. elegans vulval development has provided key insights into Ras signaling components and concepts conserved in mammalian cells. The vulval precursor cells (VPCs) are a developmental equivalence group of six ventral epithelial cells (P3.p–P8.p) (Sternberg, 2005) (Figure 1A). The nearby anchor cell (AC) induces VPCs to assume a highly reproducible 3°-3°-2°-1°-2°-3° pattern of fates. The AC-proximal VPC is induced to assume the 1° fate, flanking VPCs assume the 2° fate, and distal uninduced VPCs assume the non-vulval 3° fate.
Along with studies in other systems, analyses of the pro-1° AC inductive signal were instrumental in delineating the first signal transduction pathway connecting the cell surface to the nucleus (Egan and Weinberg, 1993). The AC secretes LIN-3/EGF (epidermal growth factor), which promotes LET-60/Ras activation of the LIN-45/Raf-MEK-ERK mitogen-activated protein kinase (MAPK) signaling cascade to regulate the LIN-1 (Ets) and LIN-31 (HNF) transcription factors, thereby inducing 1° fate (Sundaram, 2006) (Figure 1B). Analogously to human cancers, mutational activation of LET-60/Ras promotes ERK activation, leading to excess vulval induction (Figures 2A and 2B), while loss of pathway components results in vulval absence. All constituents of this signaling pathway, particularly LET-60/Ras, are strongly conserved (Figure S1C). This degree of conservation argued, prematurely, that our understanding of Ras effector signaling was complete. However, subsequent mammalian cell studies characterized additional Ras effectors, with now at least 10 distinct functional classes identified (Repasky et al., 2004). With many effectors expressed ubiquitously, an unresolved issue is how Ras effector utilization is orchestrated to facilitate the complex biological outputs of Ras.
Recent analyses have implicated the guanine nucleotide exchange factor for the Ral GTPase (RalGEF) as an effector of importance comparable to Raf in Ras-dependent human oncogenesis (Chien and White, 2003; Hamad et al., 2002). Like Ras, Ral functions as a GDP/GTP-regulated switch. Since RalGEF and Ral are conserved in C. elegans, EGF activation of Ras could involve the RalGEF-Ral pathway in regulation of vulval cell fate.
Two competing models have been proposed to illustrate the mechanisms of vulval fate patterning. The original “morphogen gradient model” posits that a LIN-3/EGF, AC-maximal concentration gradient differentially patterns VPCs dictated by proximity to the AC (Katz et al., 1995; Katz et al., 1996; Sternberg and Horvitz, 1986, 1989). This model posits that while strong EGF signal induces 1° fate, diminished EGF signal directly promotes 2° fate for more distal VPCs. Appropriately, an ERK-responsive 1° fate reporter was highly expressed in the presumptive 1° VPC, with transient low expression in presumptive 2° VPCs (Yoo et al., 2004), but further mechanistic support is lacking.
In contrast, the “sequential induction model” proposes that EGF induces only the most proximal VPC, which becomes 1°. Subsequently this presumptive 1° cell expresses DSL ligands that, via the LIN-12/Notch receptor, laterally induce neighboring VPCs to assume 2° fate (Chen and Greenwald, 2004). Accordingly, the LET-23/EGF receptor (EGFR) is necessary for 1° but not 2° fate induction (Koga and Ohshima, 1995; Simske and Kim, 1995), and pro-1° EGF and pro-2° Notch pathways together are necessary and sufficient to generate initial commitment to the 2°-1°-2° fate pattern (Greenwald, 2005; Sternberg, 2005). However, the “sequential induction model” and the “morphogen gradient model” have yet to be mechanistically reconciled.
Importantly, 1° and 2° fates are mutually antagonistic; via “quenching” mechanisms inappropriate pathway activities are reduced to minimize conflicting pro-1° and pro-2° signals in the same cell. Presumptive 1° cells enact programs that antagonize pro-2° signaling (Levitan and Greenwald, 1998; Yoo and Greenwald, 2005), and conversely presumptive 2° cells enact programs that antagonize pro-1° signaling (Berset et al., 2001; Berset et al., 2005; Yoo et al., 2004). For example, the LIP-1 ERK protein phosphatase is expressed in presumptive 2° lineages to quench ERK signaling. Thus, the developmental consequences of EGF activation of Ras-Raf signaling in 2° lineages may be minimal, and pro-2° EGF activity mediated through distinct effector pathways.
The most plausible composite model for robust vulval patterning would be one that reconciles these three evidence-based models: graded morphogen signaling, sequential induction, and pathway quenching. Yet nothing is known about the mechanism of the putative pro-2° EGF signal. Despite decades of research into how a single morphogen gradient can induce multiple cell fates, there are few instances in which the mechanism of such differential inductions is understood (Piddini and Vincent, 2009).
In this study, we identify a mechanism for EGF pro-2° signaling and thus reconcile the three features of vulval patterning into a unified model. We show that during vulval patterning Ras through Raf transduces a pro-1° signal, then through the RalGEF-Ral pathway transduces a pro-2° signal. Ral signaling antagonizes Raf and regulates the balance of 1° and 2° fates. Ral is necessary and sufficient to drive maximal Notch pro-2° activity, and the RalGEF-Ral pathway is quenched in presumptive 1° cells by restricted Ral expression. In summary, our study establishes that Ras effector utilization is controlled to signal for distinct cellular outcomes. Analogous mechanisms may therefore contribute to the distinct patterns of effector utilization that occur in different settings of mutant Ras-driven human cancers.
The C. elegans genome contains single RalGEF (rgl-1; F28B4.2) and Ral (ral-1; Y53G8AR.3) genes. rgl-1 encodes predicted splice variants producing proteins of 860 and 880 residues that share the identical domain architecture with human Ras-GTP Association (RA) domain-containing RalGEFs: an N-terminal Ras Exchange Motif (REM), a central CDC25 homology (RasGEF) catalytic domain, and a C-terminal RA domain (Figure S1A). ral-1 encodes a predicted protein of 213 residues consisting of a GTPase domain and C-terminal membrane-targeting sequence sharing strong sequence identity (61–65%) with human RalA and RalB (Figures 1C and S1B). The effector binding regions of human and C. elegans Ral GTPases share high identity, suggesting common effector utilization; Ral effector orthologs are also conserved in C. elegans. The strong conservation of the RalGEF effector pathway components suggests an important role in C. elegans LET-60/Ras function.
Genetic dissection of LET-60/Ras signaling in vulval development was instrumental in delineating the Raf-MEK-ERK pathway in mammalian cells. Consequently, we used multiple genetic approaches to dissect the role of RGL-1 and RAL-1 in Ras-directed vulval development. In a moderately activating gain-of-function (gf) LET-60 background (let-60 allele n1046; G13E mutation), we introduced rgl-1(RNAi) or ral-1(RNAi). Unexpectedly, we found increased, rather than decreased, vulval hyper-induction (Figure 2C). Negative control RNAi targeting gfp (green fluorescent protein) and lin-3 (encoding EGF, acting upstream) had no effect, while positive control RNAi targeting gap-1 (encoding RasGAP, a negative regulator of LET-60 activity) enhanced the let-60(gf) phenotype. rgl-1 or ral-1 knockdown in a wild-type (WT) background caused no defect (data not shown), suggesting a modulatory role for RGL-1-RAL-1 signaling.
A caveat is that the RGL-1/RAL-1-dependent phenotypes could be specific to let-60(n1046gf), for example if the in situ activating mutation conferred inappropriate LET-60/Ras utilization of RGL-1. We ruled out this concern by showing that rgl-1(RNAi) and ral-1(RNAi) enhanced vulval hyper-induction due to LIN-3/EGF over-expression, an activating LET-23/EGFR mutation, and a transgene expressing activated LIN-45/Raf, and suppressed the under-induced phenotype conferred by reduced function mutations (rf) in lin-3/EGF or let-23/EGFR (Figures S2A–E).
To corroborate these RNAi results with an independent methodology, we also analyzed a ral-1 deletion (Δ), tm2760, which removes a portion of intron 3 including the splice donor site. Presumably intron 3 splicing is blocked, resulting in strong loss of function. ral-1(Δ) enhanced the let-60(gf) hyper-induced phenotype (Figure 2D).
Dominant negative (dn) RAL-1(S31N), predicted to sequester and inactivate its GEF (Urano et al., 1996), should also enhance activated LET-60-driven pro-1° activity. We generated let-60(gf) animals harboring ral-1(dn) driven by the VPC-specific lin-31 promoter (Plin-31) (Tan et al., 1998). The hyper-induced vulval phenotype of let-60(gf) animals expressing RAL-1(dn) was enhanced compared to non-transgenic siblings (Figure 2E). In contrast, VPC-specific expression of RAL-1(gf) (gain-of-function, Q75L) suppressed the hyper-induced vulval phenotype of let-60(gf) animals relative to their non-transgenic siblings (Figure 2F), while RAL-1(gf) caused no defect in a WT background (data not shown). Control VPC-specific expression of WT RAL-1 did not alter the let-60(gf) phenotype, indicating that our VPC expression system is phenotypically neutral (data not shown).
Taken together, these data support four conclusions. First, RGL-1 and RAL-1 antagonize the canonical Ras-Raf-MEK-ERK pro-1° signal. Second, RGL-1 and RAL-1 function comparably in vulval patterning and likely comprise a RGL-1-RAL-1 signaling module. Third, RGL-1 and RAL-1 function cell autonomously in VPCs. Fourth, neither loss nor gain of RGL-1 or RAL-1 function in a WT background perturbed vulval patterning, suggesting that RGL-1 and RAL-1 are not part of core pro-1° or pro-2° induction pathways, but rather are modifiers of LET-60/Ras stimulated vulval patterning signals. Other comparable pathway modifiers, both positive and negative, have been identified in sensitized genetic screens. Perturbation of these genes caused no phenotype alone, but collectively they exert a profound influence on vulval patterning (Berset et al., 2001; Berset et al., 2005; Sundaram, 2006; Yoo et al., 2004; Yoo and Greenwald, 2005).
To evaluate directly whether disruption of ral-1 perturbs cell fate specification, we used a Pegl-17cfp-LacZ transgene as a reporter of 1° cell fate (Yoo et al., 2004). Notch-dependent lateral signal normally prevents formation of neighboring 1° cells. However, when the 1°–2° signaling balance is genetically disrupted to favor 1° fate, the frequency of neighboring 1° lineages increases (Berset et al., 2001; Berset et al., 2005; Yoo et al., 2004). In a let-60(gf) background, ral-1(RNAi) increased significantly the number of adjacent CFP-positive lineages (Figures 3A–C and S3A–C). Because gfp-directed RNAi inhibits CFP expression, daf-3(RNAi) rather than gfp(RNAi) was validated and used as a negative control (Figure S3D). We conclude that RAL-1 activity promotes 2° fate at the expense of 1° fate, either indirectly by antagonizing 1° fate, or directly by promoting 2° fate.
RGL-1 was identified previously in a yeast two-hybrid screen with activated LET-60 bait, but further characterization was not pursued (Shibatohge et al., 1998). We hypothesize that LET-60/Ras binds and activates RGL-1 to antagonize the Ras-Raf pro-1° signal, and thus under certain conditions loss of LET-60/Ras should be functionally equivalent to loss of RGL-1 and RAL-1. However, LET-60/Ras activation of LIN-45/Raf is an essential event during vulval induction, so we could not directly assess the necessity of LET-60 for RGL-1 activation. To address this question we used hyper-inducing mutations in downstream genes, thought to be independent of upstream Ras-Raf-MEK-ERK activity. In vulval induction the principal pro-1° transcription factors are inhibited by ERK. Consequently, the lin-31(n301) null mutation (“0”) and the lin-1(e1275ts) reduced function mutation (“rf”) confer moderate hyper-induction. RNAi targeting lin-45/Raf and mpk-1/ERK had no effect on hyper-induction, indicating that lin-31(0) and lin-1(rf) are Raf-MEK-ERK independent. lin-45/Raf and mpk-1/ERK RNAi were validated separately for activity (Figures S3D–F and not shown). In contrast RNAi of let-60, rgl-1 or ral-1 enhanced the lin-31(0) and lin-1(rf) hyper-induced phenotypes (Figures 3D and 3E), suggesting that LET-60/Ras functions similarly to RGL-1-RAL-1 in antagonizing Ras-Raf, and that this signal functions parallel to or downstream of LIN-31/LIN-1 transcription factors. We also repeated this result with mutations; partial loss of let-60/Ras but not lin-45/Raf enhanced lin-31(0) (Figure 3F). These data further suggest that in addition to the canonical pro-1° Ras signal through Raf-MEK-ERK, Ras also signals through RGL-1-RAL-1 to antagonize Ras-Raf.
LET-60 shares 73–77% identity with human Ras proteins, with 100% identity in the core effector-binding domain (32–40; Figure S1C). To assess whether Ras is sufficient to activate RGL-1, we used mutationally activated LET-60 (let-60(12V)) with missense mutations that result in differentially impaired effector binding. The E37G mutation retains effective interaction with RalGEF but not Raf or PI3K, whereas the T35S mutation retains Raf but not PI3K or RalGEF binding (Rodriguez-Viciana et al., 1997; White et al., 1995). We generated otherwise WT animals harboring transgenes driving VPC-specific expression of let-60(12V) (general gf) or let-60(12V,35S) (Raf gf). As expected, both transgenes caused a Raf-dependent hyper-induced phenotype (Figures S3E and S3F). In contrast, transgenic VPC-expressed let-60(12V,37G) (RalGEF gf) suppressed the hyper-induced phenotype of let-60(gf) animals compared to non-transgenic siblings (Figure 3G), equivalent to the effect of ral-1(gf) (above) and consistent with RalGEF-selective activity. Control VPC-specific expression of let-60(+) in the let-60(gf) background caused no phenotype (data not shown).
While the H-Ras(12V,37G) effector-binding mutant is impaired in its ability to activate PI3K and Raf, the mutant protein still retains the ability to bind other Ras-binding proteins in addition to RalGEF (Kelley et al., 2001). However, we showed that the LET-60(12V,37G) phenotype is entirely RGL-1- and RAL-1-dependent (Figure 3G), suggesting that the LET-60(12V,37G) phenotype is not due to signaling through other effectors. Thus, we conclude that Ras can signal through RalGEF in vulval fate specification.
In mammalian cells Ras is thought to be the exclusive small GTPase activator of RalGEFs, but in Drosophila the Rap family of Ras-related small GTPases has been implicated in RalGEF activation (Mirey et al., 2003; Rodriguez-Viciana et al., 2004). Our results argue that in the vulva, it is Ras that signals through RalGEF, but we also tested vulval function of characterized C. elegans Rap proteins. Injected rap-1(RNAi) or rap-2(RNAi) failed to enhance the let-60(gf) phenotype (not shown), suggesting that Raps are not required for the RGL-1-RAL-1 signal. Previously, RAL-1 was shown to function redundantly with the Rap ortholog RAP-1 in essential epithelial morphogenesis, while loss of both rap-1 and rap-2 is synthetic lethal (Frische et al., 2007; Pellis-van Berkel et al., 2005). For these reasons Raps were not more extensively analyzed, but multiple lines of evidence argue that Ras activates RGL-1-RAL-1 in vulval development as a bona fide vulval signaling module.
1° and 2° fates are mutually antagonistic, such that loss (or gain) of one pathway boosts (or suppresses) the other. In this context there are two interpretations of our data thus far. First, Ras-RalGEF-Ral could be an anti-1° pathway that impinges on the pro-1° pathway downstream of the LIN-1 and LIN-31 transcription factors. Second, Ras-RalGEF-Ral could function as an alternative signaling output of the EGF signal to promote 2° fate, thus finally providing a mechanism for the morphogen gradient model. This second model is consistent with the described mutual 1° vs. 2° antagonism in conjunction with our observations, as follows. First, loss of a pro-2° signal (e.g. loss of rgl-1 and ral-1) should enhance the pro-1° activity of let-60(gf). Second, activation of a pro-2° signal (e.g. ectopic activated RAL-1 or RalGEF-selective Ras), should suppress the pro-1° activity of let-60(gf). Third, loss of a pro-2° signal (e.g. loss of let-60, rgl-1 and ral-1) in the ERK-independent hyper-induced transcription factor mutants should enhance their hyper-induced phenotypes. Therefore, we hypothesize that Ras-RalGEFRal promotes 2° fate, but our experiments thus far have only examined Ras, RalGEF and Ral function in pro-1° assays.
To evaluate the second model we used sensitized dominant activated (d) LIN-12/Notch backgrounds. lin-12(n302d) and lin-12(n379d) mutant animals have two features critical for our study. First, they lack an AC, the source of EGF. Second, their activated pro-2° signal is relatively weak, and thus sensitive to further stimulation (Greenwald et al., 1983). Importantly, since there is no AC, the Notch activity assayed is likely to be LIN-3/EGF-independent. VPC-specific expression of activated ral-1(gf) significantly enhanced lin-12(d) excess 2° cell phenotypes (Figures 4A–C and S4A). Ectopic ral-1(gf) also enhanced glp-1(q35d)/+ (Figure S4B); glp-1 encodes the second C. elegans Notch receptor (Mango et al., 1991). RAL-1 is therefore sufficient to promote Notch pro-2° activity.
Additionally, expression of a RalGEF-selective LET-60(12V,37G) transgene, described above, in the lin-12(n302d) background was sufficient to drive 2° fates, and this activity was entirely rgl-1- and ral-1-dependent (Figure 4D). We note that as predicted from the 1° vs. 2° mutual antagonism, LET-60(12V,37G) transgenes confer opposite effects in let-60(n1046gf) pro-1° (Figure 3G) vs. lin-12(d) pro-2° backgrounds (Figure 4D). As an alternative test of this same principle, we constructed a strain containing lin-12(n379d), let-60(n1046gf) and the lin-45/Raf loss-of-function mutation, n2506. In this strain LET-60/Ras was active, but Raf pro-1° signaling was abrogated. We observed strong Ras-dependent enhancement of the lin-12(d) 2° induction, and this effect was significantly suppressed by ral-1(RNAi) (Figure 4E). Thus, Ras is sufficient to drive 2° fate, and Ras requires RalGEF and Ral for this activity.
To determine whether RAL-1 is also necessary for LIN-12/Notch function, we used ral-1(RNAi) in weakly activated lin-12(n302d) and lin-12(n379d) mutant animals, as well as the moderately activating lin-12(n676d) and strongly activating lin-12(n950d) and lin-12(n952d) mutant animals. In no case did we observe ral-1(RNAi) suppression of the lin-12(d) excess 2° phenotype (data not shown). This result is expected if RAL-1 is required only for the EGF pro-2° signal, and not the Notch pro-2° activity per se, and thus loss of RAL-1 had no consequences in the absence of EGF signal.
To evaluate this model further we determined RAL-1 necessity for LIN-12/Notch pro-2° function under EGF-dependent conditions. A lin-12/Notch hypomorphic allele (rf) confers mildly compromised 2° induction, yet the AC is still present and a single vulval invagination forms (Sundaram and Greenwald, 1993). In this background, ral-1(RNAi) caused modest but significant losses of 2° lineages and increased morphogenetic defects (Figure 4F). We propose that ral-1 is necessary for full LIN-12/Notch pro-2° activity, but only under EGF-dependent conditions. Together, these observations suggest that Ras-RalGEF-Ral mediates an EGF signal that promotes 2° fate, and we tested this hypothesis.
EGF was shown previously to be sufficient to induce 2° cells in the absence of neighboring 1° cells, arguing that there exists an EGF pro-2° signal (Katz et al., 1995; Katz et al., 1996; Sternberg and Horvitz, 1986, 1989). To examine a putative RAL-1 function in propagating the pro-2° EGF signal, we used two reagents to develop a robust EGF pro-2° signaling assay. First, lin-12(n379d), described above, is a weakly activating Notch mutation that abolishes AC development in ~90% of animals and weakly induces an ectopic 2° phenotype (Greenwald et al., 1983). Second, to titrate EGF levels with temperature, we added to the lin-12(d) background the temperature sensitive lin-15(n765ts) mutation, which at 15°C caused no ectopic 1° induction but at 25°C is strongly hyper-induced via ectopic EGF expression. lin-15 encodes components of a transcriptional regulatory complex that represses LIN-3/EGF expression outside of the AC, in the epithelia surrounding the VPCs (Cui et al., 2006).
We compared the lin-12(d); lin-15(ts) double mutant to lin-12(d) and lin-15(ts) single mutant strains from 15° to 18°C (Figure 5A). lin-12(d) is not temperature sensitive; at all temperatures ~10% of animals had a normal AC/vulva and animals averaged ~0.4 ectopic invaginations that we judged to be 2° based on morphological criteria (the distal 2° lineage cells adhere to the cuticle, while the proximal cells invaginate) (Katz et al., 1995). In lin-15(ts) single mutant animals, we observed no and rare ectopic vulval induction at 15°C and 16°C respectively; however, since all animals had an AC they formed normal vulvae. But lin-15(ts) ectopic induction greatly increased at 17°C and 18°C, and these invaginations contained combined 1° and 2° lineages typical for ectopic pro-1° signaling. To better contrast the lin-15(ts) single mutant ectopic pro-1° phenotype with the double mutant lin-12(d); lin-15(ts) pro-2° phenotype, below, we show only ectopic pseudovulvae induced in the lin-15(ts) single-mutant background, and excluded WT vulvae (Figure 5A, red).
Strikingly, in the double mutant strain we observed strong synergy at 15°C and 16°C (P<0.00001 for both), temperatures at which lin-15(ts) alone was not sufficient to induce ectopic 1° invaginations (Figure 5A). By morphology these excess invaginations were 2°, indicating that sub-threshold EGF activity in a sensitized background induced large numbers of 2° cells. To verify that the observed lin-15(ts) effect was EGF-dependent, we targeted lin-3/EGF with RNAi and observed suppression of lin-15(ts) synergistic phenotypes (Figure 5B). Including the mgIs21 lin-11 promoterGFP fusion transgene in the lin-12(d); lin-15(ts) strain as a reporter of 2° fate verified that these ectopically induced cells were mostly 2° cells (Figure 5E and 5F). Thus we demonstrated that the putative EGF pro-2° signal cooperates with the Notch pro-2° signal to specify 2° fate, a property predicted to increase fidelity of vulval patterning. Furthermore, we have precisely controlled EGF input into 2° fate induction.
A parsimonious working model posits that in presumptive 2° cells, EGF-activated Ras signals preferentially through RGL-1 rather than Raf. A prediction of this model is that LET-60/Ras and RAL-1 activity are necessary for full LIN-3/EGF pro-2° signaling output. Our system for studying this EGF signaling property allowed us to analyze sufficient numbers of animals to evaluate our model. We subjected lin-12(d); lin-15(ts) animals grown at 16°C to let-60- (Figure 5B) or ral-1-directed RNAi (Figure 5C). Importantly, loss of both let-60 and ral-1 suppressed the level of 2° hyper-induction, as did control loss of lin-12. As an internal control for lin-12(RNAi) efficacy, we observed dramatic suppression of the lin-12(d) absent-AC defect (from 13.2% with gfp(RNAi) to 76.5% with lin-12(RNAi)) (data not shown).
We also performed lin-45(RNAi) in this background, and found that loss of LIN-45/Raf enhanced 2° fate induction, consistent with the pro-1° function of Ras-Raf (Figure 5B). We hypothesize that naïve presumptive 2° VPCs activate both Raf and RalGEF in response to EGF signal, such that loss of the antagonistic Raf derepresses general pro-2° signaling.
To corroborate the lin-15 system we reproduced the lin-12(d); lin-15(ts) phenotype with an activating mutation in LET-23/EGFR. Previously it was shown that let-23(sa62gf)/+ animals induced 2° and 3° fate in isolated VPCs without an AC, while sa62/sa62 conferred mostly 1° fate (Katz et al., 1996). We found that let-23(sa62gf)/+; lin-12(n379d) had more 2° cells than lin-12(n379d) alone, and this effect was suppressed by let-60(RNAi) or ral-1(RNAi) (Figure 5D). Together, these data indicate that the Ras-RalGEF-Ral pathway mediates the EGF pro-2° signal.
Transgenic embryos harboring a ral-1 promoter-driven gfp fusion construct showed broad embryonic GFP expression. Post-embryonically, GFP was observed in excretory canals, a small number of neurons, and was expressed dynamically in vulval lineages. Prior to EGF induction, GFP was expressed in all VPCs, but at the time of induction GFP was restricted to P5.p, P6.p and P7.p, cells receiving the EGF signal (Figures 6A and 6B). Soon thereafter expression was extinguished in the presumptive 1° cell (P6.p), persisted strongly in presumptive 2°s (P5.p and P7.p), and was faintly restored in presumptive 3°s (Figure 6C). Further dynamic expression changes were seen in later vulval development, after fate specification (Figure S5).
We show that LET-60/Ras switches effectors from pro-1° LIN-45/Raf output in presumptive 1° cells to pro-2° RGL-1-RAL-1 output in presumptive 2° cells. We hypothesize that the mechanism of effector switching contains two components. First, RAL-1 expression following initial induction is quickly restricted mainly to presumptive 2°s, and therefore persistent Ras pro-2° signaling is limited to presumptive 2° cells. Second, concordant restriction to presumptive 2° cells of LIP-1/ERK phosphatase quenches the ERK signal. Thus, soon after initial induction the predominant Ras effector output in presumptive 2° cells is RGL-1-RAL-1.
The early vulval RAL-1 expression pattern mirrors that of LIP-1 (Berset et al., 2001), indicating that both RAL-1 and LIP-1 are precociously present to influence interpretation of the initial EGF inductive signal to naïve VPCs. If so, in presumptive 1° cells the Ras-RalGEF-Ral pro-2° response to EGF is expected to conflict with the Ras-Raf pro-1° response. Likewise, in presumptive 1° cells the Ras-Raf pro-1° response should be blunted by early LIP-1/ERK phosphatase expression. Therefore, rapid transcriptional exclusion of both LIP-1 and RAL-1 proteins from the presumptive 1° cell facilitates maximal Ras-Raf pro-1° activity. Supporting this model, rgl-1(RNAi) or ral-1(RNAi) suppressed the under-induced phenotype conferred by reduced function mutations (rf) in lin-3/EGF or let-23/EGFR (Figures S2D and S2E), and comparable suppression was observed with loss of LIP-1 (Berset et al., 2001). We argue that loss of RAL-1 or LIP-1 strengthens the initial pro-1° inductive event in presumptive 1° cells, and thus rescues compromised EGF pro-1° signaling. Hence, RAL-1 and LIP-1 cooperate as a programmed switch to toggle Ras output from Raf to RGL-1 in presumptive 2° cells.
The continually expanding number of functionally diverse effectors raises the issue of how Ras signaling output is controlled through dynamic spatial and temporal effector utilization to orchestrate its complex biology in normal and neoplastic cells. We describe mechanisms whereby a balance of redirected effector signal output and pathway quenching can bring two antagonistic pathways into harmony, with each faithfully promoting divergent fates in response to the same initial patterning signal. This general patterning reinforcement/fidelity mechanism may prove to be widespread in metazoan development.
The molecular mechanisms of EGF induction of 1° fate and consequent 1°-dependent Notch induction of 2° fate are well characterized. Additionally, a graded EGF receptor signal has been shown to exist, but direct EGF signaling from the AC is sufficient but not necessary for 2° fate induction (Koga and Ohshima, 1995; Simske and Kim, 1995). The mechanism by which the pro-2° EGF receptor signal is propagated was previously unknown.
We incorporate the insights from our study of Ral signaling into a new model in which we reconcile prior models of graded morphogen signaling, sequential induction, and signal quenching (Figure 7). We mechanistically validated the “graded morphogen model” and show that Ras effector switching serves to emphasize the antagonistic relationship between Ras and Notch. In the AC-proximal VPC (P6.p) EGF activates Ras and the ERK MAPK cascade to induce 1° fate, which by stimulating production of Notch ligands in turn induces 2° fate in neighboring VPCs. In presumptive 2° cells the Raf pro-1° signal is rapidly quenched by 2°-specific expression of LIP-1/ERK phosphatase (Berset et al., 2001) and other negative regulators (Berset et al., 2005; Yoo et al., 2004; Yoo and Greenwald, 2005). Instead, Ras signals through RGL-1 to promote 2° fate. Thus, the utilization of the RGL-1-RAL-1 signaling module is a critical feature of the differential response of cells across the EGF gradient. Such pathway interweaving may result in developmental fidelity and robustness of vulval patterning (Braendle and Felix, 2008).
Loss of LET-60/Ras or RAL-1 suppressed the induction of 2° cells by lin-12(d); lin-15(ts) at 16°C and the induction of 2° cells by let-23(sa62gf)/+; lin-12(d), but only partially. We therefore speculate that Ras-RGL-1-RAL-1 comprises only part of the pro-2° EGF signal. We note that additional EGF pro-2° signaling activity may be Ras-independent, and is perhaps engaged directly by activated EGF receptor.
Morphogen gradients have been studied for decades, yet there are still significant mysteries in differential interpretation of signals across gradients (Lawrence, 2001). In other systems a variety of gradient response mechanisms exist, from differential transcription of target genes to signal-induced reprogramming of signal response (Ibanes and Izpisua Belmonte, 2008; Piddini and Vincent, 2009), but correlation and causation are not always clear in these systems. In cultured human cells exposed to ectopic EGF or heregulin ligand, downstream pathway utilization varies dramatically by cell line, time of exposure, and ligand concentration (Chen et al., 2009). In vulval patterning EGF gradient input is superimposed on sequential EGF and Notch signals, and our results suggest that Ras effector switching comprises a significant portion of EGF gradient interpretation.
Our results indicate that Ras switches effector utilization between presumptive 1° and 2° cells by restricting RAL-1 expression to presumptive 2° cells. It is unknown how this 2°-specific RAL-1 transcriptional expression is patterned, but we hypothesize a combination of ERK pro-1° and Notch pro-2° transcriptional outputs. However, unlike the promoters of 2°-specific Notch-responsive lst genes, the promoter of ral-1 lacks concentrated conserved Notch-responsive sequence elements (not shown). If past studies are an indicator, there are likely to be multiple overlapping systems that cooperatively reprogram EGF output.
Clearly LIN-45/Raf effector pathway quenching is also critical, since there are significant consequences of loss of LIP-1/ERK phosphatase (Berset et al., 2001). Based on our GFP expression studies we speculate that a similar quenching phenomenon may exist for RAL-1 at the transcriptional level, since a ral-1 reporter is rapidly excluded from presumptive 1° cells after initial induction. Thus RAL-1 quenching occurs in presumptive 1° cells and is complementary to Ras-Raf-ERK quenching in presumptive 2° cells.
Theoretically, effector switching can also occur at the level of Ras effector binding. Loss of the C. elegans SOC-2/SUR-8 (human Shoc2) diminishes Ras-Raf signaling (Selfors et al., 1998; Sieburth et al., 1998). Shoc2 physically scaffolds Ras and Raf, thus regulating Ras-Raf association and pathway activation (Li et al., 2000). Although no such protein has been identified for Ras-RalGEF scaffolding, dynamic developmental regulation of such scaffolds could critically impact effector usage.
Previous studies of PC12 pheochromocytoma cell differentiation suggested that, via differential effector usage, divergent developmental outcomes arise from a particular signal. Nerve growth factor activation of Ras promotes Raf- and PI3K-dependent neuronal differentiation and growth cessation (Jackson et al., 1996; Sano and Kitajima, 1998). Conversely, Ras activation of RalGEF promotes proliferation and not differentiation (Goi et al., 1999). Thus, Ras has the potential to promote both pro-differentiation and anti-differentiation by engaging different effectors in the same cell type. In PC12 cells, RalGEF is speculated to be eventually uncoupled from Ras (Goi et al., 1999). Thus, while the potential for effector switching has been demonstrated in cell culture, mechanisms of pathway interaction are lacking.
The Ras-RalGEF parallelism to Ras-Raf was foreshadowed by a previous Drosophila study (Karim and Rubin, 1998). Also in Drosophila, activated Ras phenotypes were enhanced by a Ral dominant negative, consistent with the antagonistic Ras-Raf and Ras-RalGEF activities described here (Mirey et al., 2003). The use of multiple Ras effectors in parallel in the same developmental event may explain these puzzling results, or the relationship amongst Ras, Ral and Rap small GTPases may be different in different developmental contexts.
Interplay between Notch and Ras signaling is a common theme in developmental biology, and Notch and Ras interplay is also observed in mouse pancreatic cell differentiation and cancer development (Mysliwiec and Boucher, 2009; Sundaram, 2005). Whether this pancreatic Ras-Notch interplay depends on K-Ras activation of the RalGEF-Ral pathway is not known, but it is intriguing that RalGEF but not Raf is preferentially activated in pancreatic cancer cells and Ral activation is necessary for pancreatic cancer growth (Lim et al., 2005; Lim et al., 2006).
Efforts to develop anti-Ras inhibitors have focused on targeting effector signaling (Yeh and Der, 2007), and have been complicated by cell- and cancer-type differences in effector dependency and activation (Hamad et al., 2002; Lim et al., 2005; Rangarajan et al., 2004; Tuveson et al., 2004). How differential effector utilization and activation is achieved remains an unresolved issue. Our observations establish one mechanism for this phenomenon. Additional mechanisms may involve regulation of the subcellular localization of Ras to distinct membrane compartments, leading to spatial regulation of effector activation (Bivona et al., 2006; Onken et al., 2006).
In conclusion, we demonstrate a patterning role for Ras effector switching that has implications beyond developmental genetics. Studies in model genetic organisms in conjunction with mouse and cell culture studies were instrumental in developing our early understanding of key signal transduction pathways, including canonical EGF signaling through the Ras-Raf-MEK-ERK signaling module to regulate transcription. Recent studies in pathway quenching, and now effector switching, to promote alternative ligand outputs argue that C. elegans vulval patterning continues to yield important insights into diverse biological fields.
Strain handling and generation of constructs and transgenic lines used standard methods, see Supplemental Data.
The ral-1(tm2760) deletion, kindly provided by Shohei Mitani, removes nucleotides 418–996 (numbered from position +1 of the ral-1 initiating ATG codon), deleting part of the splice donor site. Single worm PCR of ral-1(tm2760) was performed as described (Williams et al., 1992). In the initial strain isolate, heterozygous tm2760 co-segregated with a sterile mutation. Outcrossing tm2760 and recombination in the daf-2-dpy-17 interval failed to separate the sterile mutation from tm2760. ral-1(RNAi) did not impact fertility, even when performed in the eri-1(mg366) RNAi hypersensitive background (Kennedy et al., 2004). Therefore, we hypothesize that sterility is conferred by a mutation closely linked to ral-1(tm2760). Sterile animals formed a functional vulva, so in the let-60(n1046gf) background we balanced ral-1(tm2760) with unc-93(e1500sd) dpy-17(e164). We used the semi-dominant unc-93(e1550sd) Unc phenotype to distinguish tm2760/tm2760 homozygotes from tm2760/unc-93 dpy-17 heterozygotes, and scored vulvas of non-Unc animals. Single worm PCR (Tm=53°, cycles=35) with the primers TZ23 (CAACAAGTCGTCCATAAAGTG), TZ24 (GGCGAAAAACGAGAAAAGAAC), and TZ25 (GAATTTTTCAGGCTTTCTGACG) confirmed the tm2760/tm2760 genotype of each scored animal.
Bacterially mediated RNAi was performed mostly as described (Fire et al., 1998; Kamath et al., 2001; Timmons et al., 2001). Each feeding RNAi (fRNAi) clone was sequenced to confirm identity. Eighty μl of bacteria was seeded on NGM agar plates containing 1 mM IPTG and 50 μg/ml carbenicillin. L4 larvae were added to the plates the following day. After 24 h, animals were transferred to new plates, and parents were removed after an additional 24 h. We consistently obtained stronger fRNAi phenotypes at 23°C, and thus all fRNAi experiments were performed at 23°C. gfp(RNAi) or daf-3(RNAi) was used as a control. pop-1(RNAi) was included in all experiments as a positive control for RNAi efficacy. Phenotypes were only scored if we observed 100% lethality on the pop-1(RNAi) plates.
Late L4 hermaphrodites were mounted as described previously in 5 mM sodium azide/M9 buffer on slides with agar pads. L4 vulval invaginations were visualized under DIC Nomarski optics (Sternberg and Horvitz, 1986; Sulston and Horvitz, 1977). Images were captured using a Nikon Eclipse E800 microscope with a Hamamatsu C2400-07 Newvicon camera controlled by Metamorph acquisition software (Molecular Devices). WT animals scored 3.0 (3 induced VPCs). Values greater than 3.0 indicated hyper-induction, less than 3.0 under-induction.
Live animals were mounted in 2 mg/ml tetramisole/M9 buffer on slides with agar pads and visualized using a Nikon Eclipse TE2000U microscope equipped with a DVC-1412 CCD camera (Digital Video Camera Company) controlled by the Hamamatsu SimplePCI acquisition software.
We thank V. González-Pérez for pilot 37G experiments, A. Reiner for organization, N. Dudley, A. Fire, S. Kim, Y. Kohara, and M. Maduro for plasmids, K. Caron, A. Fanning, and B. Goldstein at UNC-CH for microscope use, I. Greenwald, A. Hajnal, M. Han, J. Kimble, S. Mitani, P. Sternberg, and the C. elegans Genetics Center for worm strains, and A. Cox, R. Dubose, V. González-Pérez, A. Kidd III, K. Monahan, K. Pedone, C. Rocheleau and D. Vigil for discussions and critical reading of the manuscript. This work was supported by NIH grants GM085309 (DJR) and CA42978 (CJD).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.