A novel mode of crosstalk between the EGFR-Ras-MAPK and LIN-12/Notch pathways occurs during the patterning of a row of vulval precursor cells(VPCs) in Caenorhabditis elegans: activation of the EGFR-Ras-MAPK pathway in the central VPC promotes endocytosis and degradation of LIN-12 protein. LIN-12 downregulation in the central VPC is a prerequisite for the activity of the lateral signal, which activates LIN-12 in neighboring VPCs. Here we characterize cis-acting targeting sequences in the LIN-12 intracellular domain and find that in addition to a di-leucine motif,serine/threonine residues are important for internalization and lysine residues are important for post-internalization trafficking and degradation. We also identify two trans-acting factors that are required for post-internalization trafficking and degradation: ALX-1, a homolog of yeast Bro1p and mammalian Alix and the WWP-1/Su(dx)/Itch ubiquitin ligase. By examining the effects of mutated forms of LIN-12 and reduced wwp-1or alx-1 activity on subcellular localization and activity of LIN-12,we provide evidence that the lateral signal-inhibiting activity of LIN-12 resides in the extracellular domain and occurs at the apical surface of the VPCs.
In Caenorhabditis elegans, six vulval precursor cells (VPCs),consecutively numbered P3.p-P8.p, have the potential to adopt one of three fates, termed 1°, 2° or 3° (reviewed by Sternberg, 2005). In wild-type hermaphrodites, cell-cell interactions result in an invariant pattern of fates termed 3°-3°-2°-1°-2°-3°(Fig. 1A). Descendants of the 1° and 2° cells will form the vulva, and descendants of the 3°cells will adopt a non-vulval fate. Two of the signaling events that specify the pattern of VPC fates are `inductive' and `lateral' signaling(Fig. 1A). An EGF-like inductive signal produced by the anchor cell of the gonad activates an EGF receptor-Ras-MAP kinase cascade in the underlying VPCs. Among other outputs,this cascade activates transcription via SUR-2/MDT-23, the MED23 subunit of the `Mediator' transcription activation complex(Singh and Han, 1995; Boyer et al., 1999; Stevens et al., 2002; Bourbon et al., 2004). The centralmost VPC, P6.p, has the highest level of EGFR-Ras-MAPK activation; it becomes the presumptive 1° VPC and produces the lateral signal. The lateral signal is composed of three proteins of the Delta/Serrate/LAG-2 (DSL)family (Chen and Greenwald,2004), which activate LIN-12 in the neighboring VPCs, P5.p and P7.p, to promote the 2° fate.
Crosstalk between the EGFR-Ras-MAPK and LIN-12/Notch pathways is important for proper VPC patterning. In P5.p and P7.p, the presumptive 2° VPCs,activation of LIN-12 results in the expression of multiple negative regulators of the EGFR-Ras-MAPK pathway, restricting the effects of the inductive signal to P6.p (Berset et al., 2001; Yoo et al., 2004). In P6.p,the presumptive 1° VPC, the EGFR-Ras-MAPK pathway leads to the transcription of lateral signal genes (Chen and Greenwald, 2004), but expression of these genes in P6.p is not sufficient to activate LIN-12 in neighboring VPCs. Instead, LIN-12 must be downregulated in P6.p in response to EGFR-Ras-MAPK activation in order for lateral signaling to occur: the LIN-12 intracellular domain has a`downregulation targeting sequence' (DTS; see Fig. 1B), and if LIN-12 is stabilized in P6.p, by removing the DTS, lateral signaling is compromised(Shaye and Greenwald,2002).
The DTS contains a di-leucine sorting motif, and we have proposed that endocytic trafficking of LIN-12 is altered upon EGFR-Ras-MAPK activation,leading to degradation. Indeed, we found that if the DTS is removed, LIN-12 is not internalized efficiently and instead accumulates on the apical surface of the VPCs. Furthermore, modified LIN-12 trafficking in response to EGFR-Ras-MAPK activation is likely to depend on the activity of at least one trans-acting factor transcribed in response to activation of Ras, as degradation of LIN-12 does not occur when the function of the SUR-2/Mediator transcription activator complex is removed(Shaye and Greenwald,2002).
Endocytic downregulation has been well studied as a mechanism for attenuating the activity of activated receptors (reviewed by Sorkin and Von Zastrow, 2002). Downregulation of LIN-12 is a novel variation on this theme, as LIN-12 is downregulated as a consequence of activating a different signaling pathway. LIN-12 is likely to undergo a generic trafficking process, involving a series of protein sorting decisions that occur at different points in the endocytic pathway (Fig. 1C) (reviewed by Katzmann et al., 2002; Sorkin and Von Zastrow, 2002; Bonifacino and Traub, 2003). At the plasma membrane, receptors can be retained at the cell surface or be sorted into invaginating endocytic vesicles for internalization. In early endosomes, receptors destined to be recycled to the plasma membrane are sorted from those marked for downregulation, which are trafficked to late endosomes. En route to, or at, late endosomes, receptors marked for downregulation undergo a crucial sorting step known as multivesicular endosome (MVE) sorting(reviewed by Katzmann et al.,2002; Gruenberg and Stenmark,2004). From late endosomes, downregulated receptors are trafficked to the lysosome, where they are degraded.
MVEs have lumenal vesicles that are formed by inward invagination and pinching of the outer endosomal membrane into the lumenal space (reviewed by Gruenberg and Stenmark, 2004). Transmembrane receptors targeted for degradation are typically transferred from the outer endosomal membrane into invaginating lumenal vesicles(Fig. 1C). This sorting step is usually dependent on ubiquitination of the receptor, and is achieved by three multi-protein ESCRT complexes, which recognize ubiquitinated receptors and de-ubiquitinate them as they are transferred into lumenal vesicles (reviewed by Katzmann et al., 2002). The MVE sorting step is crucial to degrade both the extracytosolic and cytosolic domains of transmembrane receptors; in the absence of this step, the cytosolic domain of receptors would remain exposed to the cytosol, potentially resulting in prolonged signaling function.
Here we address how and why LIN-12 is downregulated, through the analysis of mutations within and near the DTS and other engineered forms of LIN-12, and the identification and analysis of trans-acting factors that mediate downregulation. Our results suggest that LIN-12 is degraded via the MVE sorting pathway, that the extracellular domain of LIN-12 is the agent that inhibits lateral signaling, and that internalization without degradation is sufficient to relieve this inhibition.
Materials and methods
Strains, genetics and transgenes
Information about all mutations can be found in WormBase(http://www.wormbase.org). The arIs82 transgene, which expresses LIN-12(+)::GFP, is described in Shaye and Greenwald (Shaye and Greenwald,2002). Plasmids and transgenes generated in this study are shown in Table 1. Transgenes were generated by germline injection using standard methods(Mello and Fire, 1995).
|Plasmid (concentration injected) .||Protein expressed .||Lines analyzed .||Figure .||lin-12(0)rescue¶(number of lines/total) .|
|pLIN12GFP*||LIN-12(+)::GFP (wild type)||arEx541||2||6/7|
|Plasmid (concentration injected) .||Protein expressed .||Lines analyzed .||Figure .||lin-12(0)rescue¶(number of lines/total) .|
|pLIN12GFP*||LIN-12(+)::GFP (wild type)||arEx541||2||6/7|
Transgenes containing unc-4(+) were analyzed in an unc-4(e120) background. Transgenes containing pha-1(+) were analyzed in a pha-1(e2123ts)background. Constructs driven by lin-12 genomic sequences (Figs 2, 3) were analyzed at 20°C. Constructs driven by the egl-17 promoter(Fig. 6) were analyzed at 25°C.
DTS is underlined.
Mixed with unc-4(+) (30 μg/ml), ttx-3p::gfp (60 μg/ml) and pBluescript (Stratagene, USA)carrier (20 μg/ml).
Mixed with unc-4(+) (30 μg/ml), egl-17p::lacZ (20 μg/ml) and pBluescript carrier (30μg/ml).
Mixed with pha-1(+) (50 μg/ml pBX) (see Granato et al., 1994), ttx-3p::gfp (60 μg/ml) and egl-17p::lacZ (20 μg/ml).
Rescue was assessed in a lin-12(n941) background, and the effects on downregulation and lateral signaling were assessed in a lin-12(+) background (see Shaye and Greenwald,2002).
Essentially all hermaphrodites carrying this transgene display a Multivulva(Muv) phenotype, which suggests elevated lin-12 activity (see text).
The DTS, 957K through 971E, was deleted and replaced by four amino acids, MAAG. LIN-12(ΔEΔDTS)::GFP is highly toxic but causes a Muv phenotype when expressed using regulatory sequences from the sel-12 gene, which drives expression in all VPCs and many other cells(data not shown).
Yeast two-hybrid screen
Yeast strains, plasmids and methods are described in the ProQuest two-hybrid system manual (Invitrogen). The starting strain contained the bait plasmid, LIN-12(region 1) fused in frame to the Gal4 DNA binding domain in plasmid pDBLeu, in strain MaV203. Approximately 3.84×106clones of a commercially available C. elegans cDNA library(Invitrogen) were screened, and five clones containing cDNAs from four different genes were not self-activating and interacted with LIN-12(region1)upon retransformation. These clones were tested by RNAi for an effect on LIN-12 downregulation (see below). Only clone pB8-1 showed an effect. This clone carried an insert corresponding to most of the gene R10E12.1B (coding sequences for the first 37 residues are missing from pB8-1), which is now named alx-1 based on sequence homology (see Results).
For alx-1(RNAi), we cloned the cDNA insert from clone pB8-1 (see above) into the double-T7 promoter plasmid pPD129.36(Timmons et al., 2001) to generate plasmid p402. For wwp-1(RNAi), we cloned the cDNA insert from clone yk1104a10 (kindly provided by Y. Kohara) into pPD129.36 to generate plasmid p416. For lacZ(RNAi) we used plasmid pXK10(Karp and Greenwald, 2003). RNAi plasmids were transformed into HT115 bacteria and used for RNAi as described by Timmons and Fire (Timmons and Fire, 2001), except that IPTG was omitted from the overnight seed culture and the plates contained 60 μg/ml AMP and 6 mmol/l IPTG. Synchronized starved L1-stage arIs82hermaphrodites were placed on bacterial lawns and grown at 25° until the mid-L3 stage (about 18 hours). Larvae were then fixed and stained as described below.
Immunofluorescence and image analysis
For lines expressing mutant proteins from lin-12 genomic constructs (Figs 2, 3), eggs were collected and grown for 42 hours at 20°C. For lines carrying egl-17p-driven constructs (Fig. 6), eggs were collected and grown for 32 hours at 25°C. Fixation, staining and image acquisition were essentially as described(Shaye and Greenwald, 2002),except that the monoclonal antibody MH27 (Developmental Studies Hybridoma Bank, USA), which recognizes AJM-1, an apical component of the adherens junction (Priess and Hirsh,1986; Koppen et al.,2001), was used at a dilution of 1:600. For measurements of endocytic puncta in LIN-12(+)::GFP and LIN-12(KtoA)::GFP images, we analyzed single confocal sections from five individual animals. Measurements were done with ImageJ software (National Institutes of Health, USA).
In a previous study (Shaye and Greenwald, 2002), we showed that a 90-amino-acid juxtamembrane fragment of the LIN-12 intracellular domain, `region 1', is sufficient to promote downregulation in P6.p. When we deleted a 15-amino-acid sequence within this region, we blocked internalization and downregulation of LIN-12. We termed this 15-amino-acid sequence the `downregulation targeting sequence'(DTS). In this study, we will show that in addition to a di-leucine motif and serine/threonine residues within the DTS, lysine residues that flank the DTS are also crucial for downregulation. Region 1 and the DTS are indicated in Fig. 1B,D.
A di-leucine motif and nearby serine/threonine and lysine residues play distinct roles in LIN-12 trafficking
The DTS contains two adjacent leucine residues(Fig. 1B,D). Many di-leucine-based endocytic sorting signals that mediate internalization and post-internalization trafficking to endosomes/lysosomes have been characterized (reviewed by Bonifacino and Traub, 2003). Both leucines are required for all activities of the signal, although in some cases one leucine can be substituted by a valine,isoleucine or methionine. When these residues are mutated to alanine in the context of the full-length, GFP-tagged LIN-12 protein LIN-12(+)::GFP, the mutant protein [LIN-12(LLtoAA)::GFP] effectively rescues the sterility and lethality of lin-12(0) mutants(Table 1). However,LIN-12(LLtoAA)::GFP appears to accumulate mostly at the apical membrane with little accumulation in intracellular puncta(Fig. 2E,F), indicating that the di-leucine is crucial for LIN-12 internalization. It is possible that the di-leucine is also involved in other aspects of trafficking, but the strong internalization defect in the mutant precludes assessing these other potential roles.
Other amino acids contribute to the activity of characterized di-leucine motifs, including upstream negatively charged amino acids and serine/threonine residues that can be phosphorylated to modulate the activity of these signals(Bonifacino and Traub, 2003). We compared the sequence of region 1 from C. elegans LIN-12 with the corresponding sequence of GLP-1, the other C. elegans Notch protein,which can functionally substitute for LIN-12(Fitzgerald et al., 1993), and with LIN-12 and GLP-1 homologs for C. briggsae and C. remanei (Fig. 1D). In addition to the conserved leucine residues, we identified conserved serines and a non-conserved threonine as potential sites of phosphorylation, and conserved lysine residues as potential sites of ubiquitination. LIN-12(S/TtoA)::GFP and LIN-12(KtoA)::GFP mutants effectively rescued the sterility and lethality of lin-12(0) mutants(Table 1), but had different effects on downregulation.
LIN-12(S/TtoA)::GFP, like LIN-12(LLtoAA)::GFP, appears to accumulate mostly at the apical membrane (Fig. 2G,H), suggesting that these residues are also important in regulating internalization. Some LIN-12(S/TtoA)::GFP hermaphrodites display mislocalization to the basolateral plasma membrane(Fig. 2G, arrowheads),suggesting that the Ser/Thr residues within the DTS may also modulate post-internalization trafficking. Internalization is not affected by mutation of the lysine residues; instead, LIN-12(KtoA)::GFP showed a striking accumulation in large and irregularly shaped intracellular vesicles(Fig. 2I,J arrowheads). We measured the area of these vesicles (see Materials and methods), and found that LIN-12(+)::GFP accumulated in vesicles that measured on average 0.250μm2 (±0.078, n=66). By contrast,LIN-12(KtoA)::GFP accumulated in vesicles that measured on average 0.442μm2 (±0.227, n=122). This difference in size was highly significant (P=2.8×10–10, two-tailed,unpaired Student's t-test). Therefore, we believe that LIN-12(KtoA)::GFP accumulates in endocytic compartments distinct from those where internalized LIN-12(+)::GFP is seen. These large vesicles may be late endosomes and/or lysosomes, as these compartments tend to be significantly larger and more pleiomorphic than earlier compartments in the endocytic system(Gruenberg and Stenmark, 2004; Patton et al., 2005). Such accumulation of LIN-12(KtoA)::GFP would be consistent with a block in MVE sorting, which depends on ubiquitination and occurs at late steps in the endocytic pathway.
Mutations that may mimic phosphorylated Ser/Thr enhance internalization from the apical membrane and do not affect downregulation of LIN-12 in P6.p
LIN-12(S/TtoA)::GFP was inefficiently internalized and was not downregulated (Fig. 2G,H),suggesting that phosphorylation of some, or all, of these serine/threonine residues is required for internalization, and perhaps for post-internalization trafficking of LIN-12. If these residues serve as a substrate for a protein kinase, then mutation of these residues to aspartate might mimic phosphorylated forms. LIN-12(S/TtoD)::GFP rescues lin-12(0) defects(Table 1) and is efficiently downregulated in the 1° lineage (Fig. 3A,B,E), suggesting that the S/TtoD mutations are not detrimental to LIN-12 function, internalization and downregulation.
If these residues were phosphorylated during the process of internalization and downregulation, then we would expect to see an effect of these mutations on subcellular localization or the promotion of downregulation in cells where we usually do not see it. Indeed, the phospho-mimicking mutations appeared to enhance internalization of LIN-12 from the apical plasma membrane, as LIN-12(S/TtoD)::GFP appeared to accumulate mostly in internal vesicles(Fig. 3A,B, white arrowheads). Additionally, even though these mutations did not promote downregulation in cells other than P6.p, we observed conspicuous GFP accumulation at what appeared to be the basolateral, instead of the apical, plasma membrane in the daughters of P5.p/P7.p (Fig. 3A,B, yellow arrowheads; compare with Fig. 2A,B).
However, phospho-mimicking mutations are not sufficient to promote downregulation in VPCs where the EGFR-Ras-MAPK pathway and SUR-2/Mediator are not active. This observation could indicate that a Ser/Thr kinase that phosphorylates the LIN-12 DTS is not a target, or at least not the only target, of sur-2. Alternatively, the phospho-mimicking mutations may functionally bypass the requirement for sur-2, but the basolateral localization of this mutant protein in P5.p and P7.p precludes effective downregulation. We address these questions below.
LIN-12(S/TtoD)::GFP downregulation is dependent on sur-2
If phosphorylation of serine/threonine residues were the only step regulated by a target of the EGFR-Ras-MAPK pathway, then we would expect that the phospho-mimicking mutations could bypass the requirement for sur-2. However, LIN-12(S/TtoD)::GFP was not downregulated in sur-2(–) hermaphrodites(Fig. 3C,D,E). Thus, the EGFR-Ras-MAPK pathway does not solely regulate a kinase that modifies these residues to mediate LIN-12 downregulation. There may be at least one other factor regulated by sur-2 required for downregulation. Alternatively,the putative kinase that phosphorylates the LIN-12 DTS may also phosphorylate another trans-acing factor required for downregulation. We note that the LIN-12(S/TtoD)::GFP that persisted in the 1° lineage of sur-2(–) hermaphrodites accumulated mostly at the basolateral membrane (Fig. 3C,D, yellow arrowheads), as in P5.p and P7.p daughters in a wild-type background(Fig. 3A,B). This suggests that the failure to downregulate LIN-12(S/TtoD)::GFP in P5.p and P7.p in wild-type hermaphrodites is not due to the basolateral localization of this protein, but instead that this re-localization is a consequences of reduced or absent sur-2 activity in these cells.
ALX-1 is required for LIN-12 degradation, but not internalization
To identify potential trans-acting factors, we performed a yeast two-hybrid screen for proteins that interact with the 90-amino-acid juxtamembrane fragment of the LIN-12 intracellular domain (region 1) that is sufficient to promote downregulation of a membrane-tethered GFP reporter. We obtained four candidate interactors (Fig. 4A and data not shown), and analyzed all using RNAi to reduce their activity in hermaphrodites that carry a LIN-12(+)::GFP transgene,followed by immunofluorescence to assess downregulation (see Materials and methods). Only alx-1(RNAi) affected downregulation: significantly more alx-1(RNAi) hermaphrodites than lacZ(RNAi)hermaphrodites failed to downregulate LIN-12(+)::GFP(Fig. 5A,C,I; P=3.5× 10–3, two-tailed, unpaired Student's t-test). Additionally, LIN-12(+)::GFP was able to accumulate in intracellular puncta in alx-1(RNAi) hermaphrodites(Fig. 5C, arrowheads). These results implicate alx-1 in LIN-12 downregulation at a post-internalization step.
ALX-1 is the C. elegans ortholog of yeast Bro1p and mammalian Alix (Fig. 4B). Bro1p plays a role in MVE sorting at a late step, probably after the function of the ESCRT-III complex (Springael et al.,2002; Nikko et al.,2003; Odorizzi et al.,2003; Luhtala and Odorizzi,2004). Alix promotes MVE formation and interacts with various endocytic proteins, including components of the ESCRT-I and ESCRT-III complexes (Martin-Serrano et al.,2003; Strack et al.,2003; Matsuo et al.,2004). Our data are consistent with a role for alx-1 in MVE sorting, and imply that downregulation of LIN-12 is achieved via this endocytic pathway.
We note that alx-1(RNAi) hermaphrodites resemble sur-2mutant hermaphrodites, in that LIN-12 is internalized but not degraded(Shaye and Greenwald, 2002),raising the possibility that alx-1 is a target of sur-2. However, a GFP::ALX-1 translational reporter (kindly provided by B. Grant)appeared to be highly and ubiquitously expressed, including in all the VPCs and their descendants (data not shown), suggesting that alx-1 is not a transcriptional target of the EGFR-Ras-MAPK pathway. GFP::ALX-1 appeared to accumulate in very large and pleiomorphic vesicular structures in the VPCs,which seem likely to be C. elegans MVEs(Fig. 4C).
WWP-1 is required for LIN-12 degradation, but not internalization
Our analysis of the DTS suggests the existence of a kinase that can phosphorylate the serine/threonine residues of the DTS and a ubiquitin ligase that targets the DTS-flanking lysines. The yeast two-hybrid screen did not yield predicted serine/threonine kinases or ubiquitin ligases, so we considered a candidate gene approach.
Unfortunately, there are approximately 230 predicted serine/threonine kinases in C. elegans (WormBase release WS144, searched for Interpro motif IPR002290), and it would be prohibitively difficult to test them all for failure of downregulation. Although there are also many potential ubiquitin ligases, we focused on the Nedd4 family of E3 ubiquitin ligases(Ingham et al., 2004), because Suppressor of Deltex [Su(dx)], a Drosophila Nedd4-like protein, was known to be a negative regulator of Notch(Cornell et al., 1999), and the mouse Nedd4-like protein Itch was also known to ubiquitinate mouse Notch1(Qiu et al., 2000; McGill and McGlade, 2003). Analysis of the C. elegans genome for proteins with the domain structure of Nedd4-family proteins (Fig. 4D) identified Y65B4BR.4 (WWP-1) (see also Huang et al., 2000) as the apparent ortholog of Su(dx) and Itch (Fig. 4E).
We depleted wwp-1 activity by RNAi and examined the effect on LIN-12(+)::GFP accumulation (Materials and methods). About 62% of wwp-1(RNAi) hermaphrodites failed to downregulate LIN-12(+)::GFP(Fig. 5E,I), as opposed to 14%of lacZ(RNAi) hermaphrodites (Fig. 5A,I). This difference was highly statistically significant(P=3.3×10–4, two-tailed, unpaired Student's t-test), indicating that wwp-1 is required for LIN-12 downregulation. Additionally, LIN-12(+)::GFP was still able to accumulate in intracellular puncta in wwp-1(RNAi) hermaphrodites(Fig. 5E, arrowheads). These results lead us to conclude that wwp-1 is required for LIN-12 downregulation at a post-internalization step. A GFP::WWP-1 translational reporter, in which GFP is fused in frame to the amino terminus of WWP-1 in a genomic context, was widely expressed and appeared to accumulate in the cytoplasm of all VPCs and their descendants (data not shown), suggesting that wwp-1 is unlikely to be a transcriptional target of the EGFR-Ras-MAPK pathway.
Lateral signaling is normal in wwp-1(RNAi) and alx-1(RNAi) hermaphrodites: correlation between subcellular localization and lateral signal inhibition
If lateral signaling occurs, P5.p and P7.p undergo three rounds of division, and their descendants stain with the adherens junction markerαAJM-1 (see Fig. 5B). If lateral signaling fails, then P5.p and P7.p undergo one round of division and their daughters join the hypodermal syncytium, so they do not stain withαAJM-1 (see Fig. 5H). Deletion of the LIN-12 DTS results in a concomitant failure of downregulation and lateral signaling (Fig. 5G,H) (Shaye and Greenwald,2002). By contrast, wwp-1(RNAi) and alx-1(RNAi)hermaphrodites displayed significant persistence of LIN-12 without compromised lateral signaling (Fig. 5C-F,I). Internalization appeared to be normal, but post-internalization trafficking and degradation appeared to be affected,suggesting that the crucial event that allows lateral signaling to occur is clearance of LIN-12 from the surface. The experiments presented in subsequent sections address why persistent LIN-12 may inhibit lateral signaling at the cell surface, but not after internalization.
LIN-12 activity is not necessary, and the extracellular domain is sufficient, for lateral signal inhibition
We first considered the possibility that persistent LIN-12 at the cell surface of P6.p is activated for signal transduction, perhaps in an autocrine fashion by ligands produced by P6.p (Chen and Greenwald, 2004). Ectopic activation of LIN-12 in P6.p might extinguish the production of the lateral signal through reducing the activity of the EGFR-Ras-MAPK pathway or by another mechanism.
To test this model, we expressed a constitutively active form of LIN-12 that cannot be downregulated and assessed whether lateral signaling is compromised. We activated LIN-12 by truncating its ectodomain (see Struhl and Adachi, 2000) and stabilized it in P6.p by removing the DTS. We expressed this form,LIN-12(ΔEΔDTS)::GFP, under the control of the egl-17promoter, which is continuously expressed in P6.p after inductive signaling(Burdine et al., 1998; Shaye and Greenwald, 2002; Yoo et al., 2004). We found that LIN-12(ΔEΔDTS)::GFP was not downregulated(Fig. 6A,I), and did not cause a lateral signaling defect (Fig. 6B,I), indicating that inhibition of lateral signaling is not caused by activation of LIN-12 in P6.p. GFP staining was found mostly in the nucleus (Fig. 6A',A”),consistent with the expected constitutive transmembrane cleavage and signal transduction of this form. In addition, we found that LIN-12(ΔEΔDTS)::GFP did not appear to affect the 1° fate of P6.p (Fig. 6B), suggesting that a presumptive 1° cell is refractory to activated LIN-12.
As activated LIN-12 did not cause lateral signal inhibition, we considered the possibility that the extracellular domain of LIN-12, which was deleted from the activated form, inhibits the function of the DSL proteins expressed in P6.p. We made LIN-12(extra)::TM::GFP, which consists of the extracellular and transmembrane domains of LIN-12 and a cytosolic GFP tag(Fig. 6J). When this protein was expressed in P6.p, it was not downregulated(Fig. 6C,J), and appeared to accumulate mostly at the apical plasma membrane, as it was mostly restrained to the apical domain demarcated by AJM-1(Fig. 6C',C”). LIN-12(extra)::TM::GFP caused a highly penetrant lateral signaling defect(Fig. 6D,J). Therefore,persistence of the LIN-12 extracellular domain is sufficient to cause lateral signal inhibition. The addition of region 1 to LIN-12(extra)::TM::GFP restored downregulation and relieved lateral signal inhibition(Fig. 6E,F,K), indicating that the mechanism by which LIN-12(extra)::TM::GFP inhibits lateral signaling is related to the inhibition caused by persistence of full-length LIN-12.
Apical localization of the LIN-12 extracellular domain is necessary for lateral signal inhibition
We assessed whether the DTS is sufficient to cause downregulation by adding it to LIN-12(extra)::TM::GFP, and found that it was not(Fig. 6G,L). However, we noticed that this protein was mostly found outside the apical domain, and instead appeared to accumulate on the basolateral membrane(Fig. 6G',G”). Thus, the DTS is not sufficient to promote downregulation, but appears to promote basolateral re-localization or to inhibit apical localization. The basolaterally localized LIN-12(extra)::TM::DTS::GFP protein is not degraded,nor does it inhibit lateral signaling (Fig. 6H,L). These observations suggest that apical localization of the LIN-12 extracellular domain is crucial for lateral signal inhibition.
We note that the subcellular localization of LIN-12(extra)::TM::DTS::GFP was similar to that of LIN-12(S/TtoD)::GFP, the phospho-mimicking mutant described above, in cells where sur-2 was not active(Fig. 3). This similarity leads us to speculate that the DTS is phosphorylated in LIN-12(extra)::TM::DTS::GFP,but this protein lacks other determinants required for degradation, such as the conserved lysines that may mediate ubiquitination.
In a previous study (Shaye and Greenwald, 2002), we showed that ectopic stabilization of LIN-12 in P6.p abrogates lateral signaling, which is necessary to activate LIN-12/Notch in neighboring VPCs and specify their correct fates. This established LIN-12 downregulation in response to EGFR-Ras-MAPK activation as a crucial event during VPC patterning. We also showed that deletion of a downregulation targeting sequence (DTS) in the juxtamembrane region of the LIN-12 intracellular domain blocks both internalization and degradation,suggesting that degradation requires endocytic trafficking. Here we have characterized cis-acting targeting sequences and trans-acting factors that mediate internalization and degradation,and the relationship between endocytic trafficking and the ability of LIN-12 to inhibit the paracrine action of ligands for LIN-12/Notch proteins. We discuss the implications of our results for the regulation of internalization and degradation of LIN-12, the differences in the apparent mechanisms for internalization of LIN-12 and Drosophila Notch, and the mechanism of lateral signal inhibition by surface LIN-12.
Regulation of LIN-12 internalization
We have found that a di-leucine motif and serine/threonine residues in the DTS are required for internalization, and that lysine residues near the DTS are required for degradation (see below). A mutant form in which the DTS serine/threonine residues have been mutated to alanine shows diminished internalization of LIN-12 in all VPCs, suggesting that at least some of these serine/threonine residues are likely to be constitutively phosphorylated to promote basal internalization.
A mutant form in which the serine/threonine residues have been mutated to aspartate displayed enhanced internalization of LIN-12 from the apical plasma membrane in all VPCs and basolateral accumulation of LIN-12 in VPCs with reduced sur-2 activity, i.e. P5.p and P7.p in a wild-type background,or P5.p, P6.p and P7.p in a sur-2(–) background. We also found that the wild-type DTS was sufficient to cause basolateral accumulation when flanking sequences that mediate degradation were not present. This observation leads us to speculate that basolateral localization indicates a phosphorylated state of LIN-12 that would usually not be detected because such a protein is normally marked for degradation. Basolateral localization may represent an intermediate step in the normal downregulation process. If so, sur-2may regulate the activity of a basolaterally located factor required for degradation of phosphorylated LIN-12, and thus this factor would not be found in P5.p and P7p. Alternatively, basolateral accumulation may be an aberrant consequence of the inability to degrade a phosphorylated LIN-12 in cells where sur-2 is not active. At this point, we are unable to distinguish between these possibilities.
The mutant form of LIN-12 in which the DTS serine/threonine residues had been mutated to aspartate was downregulated in P6.p, but was not ectopically downregulated in other VPCs. Furthermore, this form did not bypass the requirement for sur-2, as this protein accumulated in the basolateral membrane of P6.p in a sur-2(–) background. These observations suggest that phosphorylation of LIN-12 is not the limiting step in downregulation, and thus that the EGFR-Ras-MAPK pathway does not simply lead to transcription of a kinase that promotes LIN-12 phosphorylation. However,the nature of the link between EGFR-Ras-MAPK activation remains to be determined.
Regulation of LIN-12 post-internalization trafficking: MVE sorting and degradation
LIN-12 appears to be downregulated via MVEs. Although mutation of the conserved lysines near the DTS does not affect internalization of LIN-12,degradation in P6.p is blocked, and in all VPCs this mutant form accumulates in large pleiomorphic internal vesicles. Furthermore, the ubiquitin ligase WWP-1 and the MVE-associated factor ALX-1 are required for LIN-12 degradation after internalization. As sur-2 mutants display a similar phenotype,transcriptional targets of the EGFR-Ras-MAPK pathway may be involved in directing LIN-12 to MVEs.
Mutating the conserved lysines near the DTS caused the `Multivulva'phenotype associated with constitutive LIN-12 activation(Table 1). This phenotype is consistent with an MVE sorting defect. If a transmembrane protein does not go through the MVE sorting step, then upon delivery to the lysosome its extracellular domain will be degraded whereas its intracellular domain will remain exposed to the cytosol. For LIN-12/Notch, the mechanism of signal transduction involves cleavage and release of the intracellular domain. Thus,if MVE sorting is disrupted, degradation of the extracellular domain in the lysosome could mimic ectodomain shedding, creating a substrate for Presenilin-dependent release of the intracellular domain of LIN-12, or perhaps would release the intracellular domain by an alternative mechanism.
Recent reports have described `ligand-independent' activation of Drosophila Notch in late endosomes. In these studies, overexpression of the protein Deltex was shown to promote internalization and accumulation of Notch in late endosomes, correlated with activation of Notch signaling. It was suggested that such endosomal activation of Notch might represent a novel and relevant mode of activating this pathway(Hori et al., 2004). However,our finding that an apparent block in MVE sorting can lead to LIN-12 activation suggests an alternative explanation for the effect of Deltex overexpression: the enhanced internalization and endosomal accumulation of Notch may saturate the MVE sorting machinery, so that some Notch is not correctly internalized into MVE lumenal vesicles, leading to degradation of the extracellular domain without concomitant degradation of the intracellular domain.
Different mechanisms for regulating trafficking of LIN-12 and Drosophila Notch, and possible evolutionary conservation in vertebrates
We have provided evidence that internalization of LIN-12 is mediated by the di-leucine motif and basal phosphorylation of flanking serine/threonine residues. By contrast, for Drosophila Notch, recent evidence suggests that ubiquitination by the dNedd4 ubiquitin ligase is required for Notch internalization (Sakata et al.,2004). Drosophila Notch does not have a di-leucine-based motif similar to the one we have described for LIN-12(Fig. 7). Conversely, LIN-12 does not have a C-terminal PPXY signal (data not shown), which in Drosophila Notch promotes interaction with dNedd4(Sakata et al., 2004). We suggest that C. elegans and Drosophila may utilize different mechanisms for targeting LIN-12/Notch for internalization.
Both of these mechanisms may be utilized in vertebrate Notch proteins. Sequence analysis of vertebrate Notch proteins shows an intriguing inverse correlation between the presence of a di-leucine based motif and a PPXY signal. The corresponding juxtamembrane regions of vertebrate Notch1 and Notch2 proteins have a segment that is strikingly similar to the LIN-12 DTS,including conserved flanking lysines (Fig. 7), but these proteins do not have a conserved PPXY signal at their C-termini (data not shown). By contrast, most vertebrate Notch3 proteins appear more divergent in this region (Fig. 7), but possess a PPXY signal at their C-termini (data not shown);it is curious that zebrafish Notch3 lacks the PPXY motif (data not shown), but has instead a canonical di-leucine motif(Fig. 7). These observations raise the possibility that the two modes of internalizing Notch proteins(di-leucine based versus ubiquitination via a PPXY motif) have been conserved in different vertebrate Notch proteins through evolution. Perhaps other modes exist as well, as vertebrate Notch4 does not seem to have either of these conserved motifs (data not shown). Mutational analysis of these potential internalization sequences in vertebrate Notch proteins will be necessary to test their roles.
Regulation of DSL ligand activity by LIN-12/Notch in the signaling cell
We previously proposed that the lateral signaling defect of sur-2was caused by the failure to downregulate LIN-12(Shaye and Greenwald, 2002). Subsequently, it was found that dsl gene transcription is regulated by sur-2 (Chen and Greenwald,2004), and here, we have found that internalized LIN-12, as is seen in sur-2 mutants (Shaye and Greenwald, 2002), did not appear to inhibit lateral signaling. Thus, persistence of LIN-12 does not appear to be the basis of the sur-2 lateral signaling defect; rather, loss of lateral signaling in sur-2 mutants is likely to result simply from the failure to transcribe the lateral signal.
We observed that expression of a constitutively active LIN-12 that could not be downregulated in P6.p did not affect the fate of this cell or its ability to signal laterally, implying that the principal role of LIN-12 downregulation is to permit DSL ligands to activate LIN-12 in P5.p and P7.p in a paracrine mode, rather than to prevent autocrine LIN-12 activation in P6.p. However, this result also indicates that activation of the EGFR-Ras-MAPK pathway in P6.p causes it to become refractory to activated LIN-12, suggesting another potentially novel mode of crosstalk between EGFR-Ras-MAPK and LIN-12/Notch signaling in P6.p.
In certain gain-of-function mutants or ectopic overexpression situations in Drosophila, Notch also appears to be able to inhibit the ability of Delta to signal laterally (Heitzler and Simpson, 1993; Jacobsen et al., 1998). These results were interpreted as suggesting the formation of a DSL-Notch inhibitory complex, but whether such interactions occur during normal Drosophila development is not clear. Our results establish that inhibition of ligand activity occurs at the surface of the signaling cell, consistent with a DSL-Notch inhibitory complex. Furthermore,our results suggests that relief of this inhibition by internalization of LIN-12 appears to be part of the normal mechanism for coordinating EGFR-Ras-MAPK-mediated inductive signal and LIN-12-mediated lateral signaling. Given that the ability of LIN-12/Notch proteins to inhibit DSL signaling activity appears to be conserved, and the presence of conserved endocytic sorting motifs in all Notch proteins, there may be other natural situations in which inhibition of DSL ligands by endogenous LIN-12/Notch proteins, and regulated relief of such inhibition, may be relevant to patterning cell fates.
We thank: Barth Grant (Rutgers University) for reagents and helpful discussions throughout the course of this work; Dorothy Kim for help with sequencing the alx-1 cDNA in pB8-1 and performing the initial RNAi experiments; Richard Ruiz and Xinlan Zhou for technical assistance; and Barth Grant and Gary Struhl for discussion and critical reading of the manuscript. D.S. was supported by NIH postdoctoral training grant (T32-DK07328) and was a Postdoctoral Associate of the Howard Hughes Medical Institute. I.G. is an Investigator with the Howard Hughes Medical Institute.