sel-11 and cdc-42, Two Negative Modulators of LIN-12/Notch Activity in C. elegans

Background LIN-12/Notch signaling is important for cell-cell interactions during development, and mutations resulting in constitutive LIN-12/Notch signaling can cause cancer. Loss of negative regulators of lin-12/Notch activity has the potential for influencing cell fate decisions during development and the genesis or aggressiveness of cancer. Methodology/Principal Findings We describe two negative modulators of lin-12 activity in C. elegans. One gene, sel-11, was initially defined as a suppressor of a lin-12 hypomorphic allele; the other gene, cdc-42, is a well-studied Rho GTPase. Here, we show that SEL-11 corresponds to yeast Hrd1p and mammalian Synoviolin. We also show that cdc-42 has the genetic properties consistent with negative regulation of lin-12 activity during vulval precursor cell fate specification. Conclusions/Significance Our results underscore the multiplicity of negative regulatory mechanisms that impact on lin-12/Notch activity and suggest novel mechanisms by which constitutive lin-12/Notch activity might be exacerbated in cancer.


Introduction
LIN-12/Notch signaling plays multiple roles in vulval development in C. elegans. Canonical roles for this signaling system in vulval development are the specification of a single anchor cell (AC), which is required for vulval induction, and the specification of the 2u vulval precursor cell (VPC) type. Numerous feedback loops and functional redundancies make vulval development relatively insensitive to small perturbations in signaling pathways. Thus, an individual regulator often does not cause an overt phenotype when it is removed in an otherwise wild-type genetic background, but its function may be revealed when it is removed in a sensitized genetic background. In C. elegans, a few core components of LIN-12/Notch signaling have been identified in phenotype-based genetic screens, but most core components and modulators have been identified as suppressors or enhancers in sensitized backgrounds (reviewed in [1]).
Mutations resulting in constitutive or elevated LIN-12/Notch signaling can cause cancer (reviewed in [2,3]). In principle, abrogation of any system of negative regulation of lin-12/Notch activity has the potential for contributing to the development or aggressiveness of cancer. For example, there is some evidence that increased expression of the human SEL-1 ortholog, SEL1L, correlates with a decrease in tumor aggressiveness (reviewed in [4]), but how this correlation relates to effects on Notch activity is not clear. A compelling example is afforded by the ubiquitin ligase SEL-10/Fbw7, which targets LIN-12/Notch directly for protea-some-mediated degradation ( [5]; reviewed in [6]). Point mutations in the extracellular domain of NOTCH1 are one cause of T cell acute lymphoblastic leukemia (T-ALL); when patients with these mutations relapse after chemotherapy, they generally have mutations in FBW7. Patients presenting with NOTCH1 mutations that remove the Fbw7 binding site do not have Fbw7 mutations upon relapse. These observations have implicated the loss of the negative regulation of NOTCH1 in patients whose T-ALL tumors are drug-resistant after relapse [7].
Here, we describe two negative modulators of lin-12 activity. One gene, sel-11, was initially defined as a suppressor of a lin-12 hypomorphic allele [8]. The other, cdc-42, is a well-studied Rho GTPase that we find has an effect on lin-12 activity that appears to be distinct from other roles of cdc-42 in vulval development described previously [9]. Our results underscore the multiplicity of mechanisms that impact on lin-12/Notch activity and suggest novel mechanisms by which aberrations in lin-12/Notch activity might be exacerbated in cancer.

Results
The genetic analysis of modulators of lin-12 activity involves combinations with different lin-12 alleles. Such alleles reduce or elevate lin-12 activity, and their effects on hallmark cell fate decisions are shown in Fig. 1. We first describe the genetic analysis that establishes hrd-1 as sel-11. We then describe genetic interactions between cdc-42 and lin-12.
Demonstration that the negative regulator sel-11 encodes Hrd1p/Synoviolin Mutations in sel-11 were originally identified in a genetic screen for negative regulators of lin-12 activity [8]. This screen was based on suppression of the egg-laying defective (Egl) phenotype caused by the hypomorphic allele, lin-12(n676n930) [10]; the properties of this allele are described further below. Extensive genetic analysis of the interactions between sel-11 and multiple lin-12 alleles in different cellular contexts established sel-11 as a negative regulator of lin-12 activity, but its molecular identity was not determined [8].
We molecularly identified sel-11 serendipitously because it is adjacent to the microRNA gene mir-61. A previous study in our laboratory concluded that mir-61 is expressed in response to LIN-12 activation and targets VAV-1, a negative regulator of lin-12 activity [11]. Our view of the role of mir-61 in vulval development was challenged by Miska et al. [12], who reported that a deletion of mir-61, nDf59, does not cause overt vulval lineage defects. We independently confirmed that nDf59 exhibits overtly normal vulval development by examining canonical 1u, 2u, and 3u cell fate markers (Table 1). However, as described below, our further analysis of nDf59 revealed that it removed sel-11, a negative regulator of lin-12 activity, and it is therefore problematic to use nDf59 to draw inferences about mir-61/250 function.
The possibility that nDf59 might affect a negative regulator of lin-12 activity arose when we examined the genetic interactions between nDf59 and lin-12(n676n930), a hypomorphic allele of lin-12 in which activity is compromised but not eliminated [10]. When lin-12(n676n930) hermaphrodites are grown at 25uC, they exhibit defects caused by reduced lin-12 activity: a highly penetrant egg-laying defect; a partially penetrant 2 AC defect; and, in hermaphrodites with 1 AC, a partially penetrant lateral signaling defect (P5.p and/or P7.p lose expression of a 2u fate marker) ( [10]; Fig. 2B and 2C). At 15uC, lin-12(n676n930) retains In wild-type C. elegans, vulval development is initiated in the early L3 stage when the anchor cell (AC) signals to the underlying VPCs via an inductive EGF signal, LIN-3 (white arrows). As a result, P6.p adopts the 1u cell fate. P6.p, in turn, sends a lateral signal (black arrows) to its neighboring VPCs, P5.p and P7.p, activating LIN-12 and causing them to adopt the 2u fate. During the AC/VU fate decision in the somatic gonad, the VU fate is determined by high lin-12 signaling. In wild-type animals, this results in 1 AC. (B) When lin-12 activity is reduced, a VU is often not specified, resulting in 2 ACs; in addition, as a result of increased inductive signaling from 2 ACs and/or reduced lateral signaling due to lower lin-12 activity, ectopic VPCs adopt the 1u fate at the expense of the 2u fate. The lin-12(n676n930) allele shows both aspects of this phenotype at 25uC [10]. (C) When lin-12 activity is mildly elevated, the AC fate is lost, and a VU is specified in its place. Since the inductive signal is lost, the underlying VPCs all adopt the 3u fate. lin-12(n379) and lin-12(n676) are examples of weak hypermorphic alleles, as is the lin-12(n676n930) allele at 15uC [10,18]. (D) However, when lin-12 activity is strongly elevated, multiple VPCs are induced to adopt the 2u cell fate even though the AC is absent [18] mildly elevated lin-12 activity due to the presence of the n676 mutation. According to the proposed circuit, loss of mir-61 should decrease lin-12 activity; thus, we would expect that nDf59 might enhance phenotypes caused by lin-12(n676n930) at 25uC and suppress phenotypes of lin-12(n676n930) at 15uC. However, we obtained the opposite results, that nDf59 suppressed the loss-offunction phenotypes of lin-12(n676n930) at 25uC, and furthermore, enhanced the weak hypermorphic phenotype of this allele at 15uC ( Fig. 2B and 2C). These results indicate that nDf59 increases lin-12 activity.
nDf59 had been reported to remove 1143 bases encompassing mir-61 and another microRNA, mir-250 [12]. We independently sequenced the nDf59 allele and confirmed the deletion breakpoints reported in WormBase (www.wormbase.org). During this process, we realized that, in addition to removing mir-61/250, nDf59 also removes part of the adjacent protein-coding gene, hrd-1 ( Fig. 2A) (see MATERIALS AND METHODS). C. elegans hrd-1 [13] is the ortholog of yeast HRD1, which was identified in a genetic screen for genes that mediate HMG-CoA reductase degradation in Saccharomyces cerevisiae [14]. Another gene defined in the same  [11]. The thick black lines indicate genomic regions deleted by tm1743 or nDf59 (see RESULTS). The RING domain of hrd-1/sel-11 is indicated by a grey box. Molecular lesions were found in the RING domain of hrd-1/sel-11 for two sel-11 alleles, ar39 and ar84, as indicated by arrows. These two alleles were found to contain point mutations in the RING finger domain of hrd-1, mutating either a histidine to a tyrosine (ar84, H312Y), or a cysteine to a tyrosine (ar39, C295Y), thereby mutating amino acids potentially important for the ubiquitination function of the RING domain. (B, C) lin-12(n676n930) has reduced activity at 25uC and mildly elevated activity at 15uC. The VU fate in the somatic gonad is determined by high lin-12 activity, as is the 2u vulval cell fate. (B) nDf59 decreases the number of anchor cells (AC) in the lin-12(n676n930) background at both 25uC and 15uC, which indicates an enhancement of lin-12 activity. The number of ACs observed for each strain is as follows: a 49/49 1 AC. b  :gfp]; nDf59 (GS5161), unc-32(e189)lin-12(n676n930); arIs51 (GS4894), and unc-32(e189)lin-12(n676n930); arIs51; nDf59 (GS5015). (C) nDf59 increases the number of VPCs expressing a 2u marker in the lin-12(n676n930) background at both 25uC and 15uC. When scoring for effects in the VPCs, we scored only animals with 1 AC to avoid the influence of altered numbers of ACs on VPC fate in the lin-12(n676n930) background. In a wild-type background, P5.p and/or P7.p express the 2u fate marker nIs106[lin-11::gfp] ( Table 1, [11]). At 25uC, some lin-12(n676n930) animals have less than two VPCs that express a 2u fate marker; this phenotype is suppressed by nDf59. At 15uC, some lin-12(n676n930) animals have more than two; this phenotype is enhanced by nDf59. screen, HRD3, corresponds to the C. elegans gene sel-1, the first molecularly characterized negative regulator of lin-12 [8,15,16].
In view of the functional relationship between HRD1 and HRD3 in yeast, we wondered if the genetic enhancement of lin-12(n676n930) by nDf59 may be explained if hrd-1, like HRD3/sel-1, is a negative regulator of lin-12 activity. This inference was confirmed by showing that hrd-1(tm1743), an apparent molecular null allele, suppresses the egg-laying defect (Egl) of lin-12(n676n930) at 25uC (Fig. 2D).
The map position of hrd-1 suggested that it might correspond to sel-11, previously defined as a negative regulator of lin-12 [8]. We therefore sequenced the two available alleles of sel-11, and found that they contain point mutations in the RING finger domain of HRD-1 ( Fig. 2A). hrd-1 has been renamed sel-11, the first published name for the gene, in accordance with the accepted nomenclature convention in the field. The finding that two point mutations in the RING finger domain behave like a deletion of the gene suggests that the ubiquitin ligase activity of SEL-11 is important for negative regulation of lin-12 activity.
Genetic interactions implicate cdc-42 as a negative regulator of lin-12/Notch activity As mentioned above, nDf59 deletes the microRNA mir-250, which appears likely to be cotranscribed with mir-61 in response to LIN-12 activation. Using the microRNA Registry release 2.0 prediction of the mir-250 sequence [17], we identified potential target genes computationally, using the same criteria that we used for mir-61 targets: we required that 39 UTRs have at least 7 bases of perfect complementarity to the 59 end of mir-250, with this putative ''seed match'' conserved in the C. briggsae orthologs of the C. elegans genes [11]. Our interest in cdc-42, a candidate identified in this way, was stimulated when we found that cdc-42(RNAi) enhanced lin-12 activity in sensitized backgrounds.
The backgrounds we used were afforded by the mild hypermorphs lin-12(n379) and lin-12(n676) (Fig. 3A). These mild hypermorphs lack an anchor cell and their VPCs generally behave as wild-type VPCs do in the absence of the inductive signal, i.e. they all adopt the 3u fate (Fig. 1). If lin-12(n379) or lin-12(n676) activity is increased, some or all of the VPCs adopt the 2u fate, resulting in a ''Multivulva'' phenotype that is visible in the dissecting microscope [8,18,19] (Fig. 3A). We verified that the Multivulva phenotype of cdc-42(RNAi); lin-12(n676) reflects an increased number of VPCs adopting the 2u fate using a transgenic marker (Fig. 3B). These observations are consistent with a role of cdc-42 as a negative regulator of lin-12 activity in the VPCs.
We generated a transcriptional reporter containing the 59 flanking region of cdc-42. This reporter drives strong, uniform expression in all six VPCs, and the level of expression appears to increase in the daughters of P5.p, P6.p and P7.p (Fig. 3C). The presence of cdc-42 in VPCs and their descendants is consistent with the multiple roles of cdc-42 in VPC development and specification inferred by Welchman [9] and the genetic interactions with lin-12 described here.
These observations are consistent with a role of cdc-42 as a negative regulator of lin-12 activity in the VPCs. Phenotypes caused by the expression of dominant-negative versions of Drosophila ortholog of cdc-42 have also suggested a role as a negative regulator of Notch signaling in wing development [20]. However, we were unable to obtain evidence that mir-250 negatively regulates cdc-42 in any assay (data not shown). When a revised mir-250 sequence was published (miRBase 10.1, [21]), it was evident that mir-250 has two additional adenines at its 59 end that would preclude the appropriate seed match base pairing with cdc-42. Thus, although cdc-42 behaves genetically as a negative regulator of lin-12, it is not likely to be regulated by mir-250 in an analogous way to the LIN-12-mir-61-VAV-1 circuit.

Discussion
We have described here two negative regulators of lin-12/Notch activity, sel-11 and cdc-42. We discuss possible mechanisms by which these genes may influence lin-12/Notch activity in terms of our findings and relevant literature on their mammalian orthologs.
SEL-11/Hrd1p and a previously-described negative regulator, SEL-1/Hrd3p [15,16], are both involved in the Hrd1p pathway of ER-associated degradation (ERAD). The Hrd1p pathway specifically targets proteins with misfolded lumenal domains for degradation. SEL-11/Hrd1p is the central E3 ubiquitin ligase, and SEL-1/Hrd3p acts in the recognition of terminally misfolded substrates [22,23]. All lin-12 alleles that display genetic interactions with sel-1 and sel-11 [8,15,16] carry mutations that alter the extracellular domain, which would be positioned in the ER lumen during protein synthesis and folding [24,25]. Loss of the Hrd1p quality control system likely increases the stability or export of these mutant LIN-12 proteins.
There is some evidence that increased expression of SEL1L, the human SEL-1 ortholog, correlates with a decrease in tumor aggressiveness (reviewed in [4]), but if this correlation reflects effects on Notch activity is not clear. In addition, the human SEL-11 ortholog, Synoviolin, is essential for development in mice and has been implicated in hyperproliferation of synovial tissues in rheumatoid arthritis and in the degradation of p53 and IRE1 [26,27,28]; furthermore, there appears to be multiple modes of crosstalk between Notch and p53 [29]. These observations are consistent with the possibility that loss of sel-11 might lead to elevated LIN-12/Notch activity in cancer.
C. elegans CDC-42 is a member of the Cdc42 subfamily of Rho family GTPases. Cdc42 has been implicated in many different cellular processes including polarity, cytoskeleton reorganization, vesicle trafficking, and signal transduction [30,31]. The VPCs are polarized epithelial cells, with LIN-12 localized to the apical domain [32] and LET-23/EGFR localized to the basolateral domain [33]. Reciprocal negative regulation of endocytosis and trafficking of these receptors helps ensure proper pattern formation [11,34,35,36,37,38,39]. CDC-42 may contribute to this regulation through effects on VPC polarization and apical membrane organization, or through a general effect on endocytic traffic [40,41,42]. CDC-42 may affect LIN-12 signaling per se or through promoting the activity of the LET-23/EGFR-Ras-MAPK pathway: although genetic enhancement of lin-12(d) mutations was observed in the absence of the anchor cell, the normal source of inductive signal, it is possible that VAV-1 and CDC-42 affect a basal activity of LET-23 or another receptor tyrosine kinase that promotes Ras activity.
Cancer results from aberrations in cell-cell interactions and growth control, processes that are influenced by Notch activity. Depending on the cellular context, Notch can function as either a tumor suppressor (promoting differentiation) or proto-oncogene (promoting proliferation and/or suppressing apoptosis) [3]. In these roles, Notch crosstalks with p53 and Rho/CDC42 effectors [29,43]. Our results raise the possibility that Cdc42 has effects on tumorigenesis through effects on Notch activity.
As mentioned in RESULTS, all of the lin-12 alleles used in this study carry missense mutations that alter the extracellular domain and cause constitutive LIN-12 activity; in some cases, the lin-12 alleles are composites of such activating mutations combined with second-site revertants that lower lin-12 activity. The missense activating alleles alter a negative regulatory domain [24] now known to be revealed under normal conditions by ligand binding (reviewed in [44]). Equivalent missense mutations in human Notch1 cause T-ALL and perhaps other cancers [45]. Whether SEL-11/ Hrd1p/Synoviolin and CDC-42/Cdc42p act to promote the activity of wild-type LIN-12/Notch or only these missense mutant forms is not clear. Nevertheless, the clear effect on the LIN-12/ Notch missense mutant forms associated with cancer suggest that the reduction in the activity of these negative regulators may be associated with cancer formation or progression.
GE24 pha-1(e2123) was used as the recipient to create transgenes. All strains were grown using standard procedures at 20uC unless otherwise noted, except for strains with pha-1(+)-containing transgenes in the pha-1(e2123) background, which were maintained at 25uC. For strains that were scored at 25uC or 15uC, animals were maintained at the temperature of interest for at least two generations prior to scoring.

RNAi experiments
Feeding RNAi experiments were performed at 20uC as described [47,48]. Briefly, gravid adults were bleached and the eggs were placed on plates seeded with HT115 cells expressing the dsRNA of interest. T7 polymerase expression in the HT115 cells had been induced with 6 mM IPTG for at least four hours at room temperature before plating the eggs. To score the number of pseudovulvae at the adult stage, animals were scored three days after eggs were placed on plates. To score at the L3 Pn.pxx stage, animals were scored roughly 45 hours after eggs were plated.

Imaging
All microscopy done on live animals was performed on a Zeiss Axioplan2 microscope, with a consistent exposure time used for each marker assayed.