CDC-25.1 stability is regulated by distinct domains to restrict cell division during embryogenesis in C. elegans

Precise spatiotemporal coordination between developmental pathways and the cell cycle machinery is essential to ensure the timely and accurate growth and differentiation of cells and organs. In many cells, this is controlled at the G1-S phase transition, while in others this coordination occurs at the G2-M boundary. These transitions are modulated by the action of key cell cycle regulators that ultimately fine-tune the activity of cyclin/cyclin-dependent kinase (cyclin/CDK) complexes (Nigg,2001; Nurse et al.,1998; Pines,1995). This regulation can occur at multiple levels, including cell cycle phase-dependent transcriptional modulation of CDKs, cyclins and cyclin-dependent kinase inhibitors (CKIs) coupled with activating or inhibitory phosphorylation events, subcellular relocalization and/or ubiquitin-mediated degradation of key cell cycle regulators. Together, these regulatory controls ensure the integrity and the faithful segregation of the genome throughout successive division cycles(Ang and Harper, 2005; DeSalle and Pagano, 2001; Morgan, 1995; Nilsson and Hoffmann, 2000; Schafer, 1998).

During unperturbed cycles, the timely degradation of cell cycle regulators is coordinately controlled by two major ubiquitin E3 ligases, the anaphase-promoting complex/cyclosome (APC/C) and the Skp1/Cullin/F-box (SCF)E3 ligase that trigger irreversible progression into the next phase(Castro et al., 2005; Nakayama and Nakayama, 2006; Vodermaier, 2004). Interestingly, after incomplete replication or DNA damage, the cells rely on checkpoint pathways that impinge on the same E3 ubiquitin ligases to delay the cell cycle phase transition and to allow for repair of the defect, thus preventing genomic instability (Bartek et al., 2004; Sancar et al.,2004).

The Cdc25 phosphatase is a key positive regulator of cell cycle progression through its ability to remove inhibitory phosphates from cyclin/CDK complexes(Moreno et al., 1990; Sebastian et al., 1993). Multicellular organisms often contain multiple Cdc25 paralogues with overlapping, yet distinct, phase- and tissue-specific functions. In mammals,Cdc25A activates cyclinE(A)/Cdk2 for progression into S phase, while cooperating with Cdc25B and Cdc25C to promote G2/M transition(Boutros et al., 2006). The ubiquitin-mediated degradation of the different Cdc25 paralogues is triggered by the APC/C^Cdh1 (activator=Cdc20 homologue) at the end of mitosis through recognition of a KEN box, while during S and G2 phase the abundance of Cdc25A is regulated by the SCF^βTrCP (F-box=β-transducin repeat-containing protein) E3 ubiquitin ligase that recognizes a phosphorylated β-TrCP-like destruction motif (DSGX₄S)(Busino et al., 2003; Donzelli et al., 2002; Jin et al., 2003). This SCF^βTrCP-mediated degradation is accelerated after DNA damage during S and G2 phase in an ATM/Chk2-ATR/Chk1-dependent manner(Falck et al., 2001; Sorensen et al., 2003).

Cdc25 has also been shown to integrate spatiotemporal information to coordinate timely cell cycle progression with developmental processes. In Drosophila melanogaster, the two Cdc25 paralogues twine and string have mutually complementing and overlapping functions in the female germ line and in early embryos, respectively(Edgar and Datar, 1996). During embryogenesis the zygotic expression of string becomes the limiting factor for successive cell divisions in the mitotic domains of the fly embryo (Edgar and O'Farrell,1989; Edgar and O'Farrell,1990; Foe, 1989). Transcription of string is highly regulated through elements in it large, complex promoter region (Edgar et al., 1994; Lehman et al.,1999), while degradation of string protein is also developmentally controlled by an ubiquitin-mediated pathway involving tribbles (Mata et al.,2000).

The C. elegans genome encodes four cdc-25 paralogues(Ashcroft et al., 1998), among which cdc-25.1 shows the greatest homology to the human Cdc25A gene. cdc-25.1 is expressed in the hermaphrodite germ line where it sustains germ cell proliferation (Ashcroft et al., 1999). Maternal cdc-25.1 gene product is contributed to oocytes and early embryos and is required for completion of meiosis and to ensure timely mitotic divisions until it is degraded after the 28-cell stage (Ashcroft and Golden,2002). Consistent with its role in regulating proliferation, it has been considered a proto-oncogene in humans, and overexpression of cdc-25.1 results in unscheduled S phase initiation(Kostic and Roy, 2002). We and others previously reported that gain-of-function (gf) mutations in cdc-25.1 cause embryonic hyperplasia restricted to the intestinal (E)lineage (Clucas et al., 2002; Kostic and Roy, 2002). The gf mutant CDC-25.1(rr31) protein becomes stabilized and is still present in all embryonic cells past the 100-cell stage. Despite this abnormal perdurance, CDC-25.1(rr31) induces only a single supernumerary division in the intestinal precursor cells at a specific time after the embryonic E⁸-stage (embryo containing eight intestinal cells).

To determine whether the increased stability of the mutant protein and the hyperplasia are linked, we developed a means to test candidate genes that could influence the stability of CDC-25.1, which in turn would provide further clues to explain the tissue specificity of this mutation. One candidate, the C. elegans orthologue of the mammalian F-box protein β-TrCP,LIN-23 is required for the timely degradation of CDC-25.1 during early embryogenesis and its differential embryonic expression may shed some light on the tissue specificity of cdc-25.1(gf). In addition, we report the isolation and characterization of a viable loss-of-function intragenic suppressor of cdc-25.1(gf) that affects a novel site required for CDC-25.1 stability, suggesting that multiple controls contribute to the stabilization of this cell cycle regulator through independent domains to control its activity appropriately during embryogenesis.

All C. elegans strains were cultured at 20°C according to standard procedures (Brenner,1974) unless stated otherwise. The following alleles and transgenes were used for this study: wild-type Bristol (N2); cdc-25.1(rr31)I; cdc-25.1(ij48)I; cdc-25.1(rr31rr36)I; lin-23(ot1)II;lin-23(e1883)II; unc-119(ed3)III; rrIs01[elt-2::gfp; unc-119+]X; rrIs12[pie-1::gfp::lin-23; unc-119+]; wIs137[end-3::end-3(P202L)::gfp; rol-6D] (gift from M. Maduro); otEx731[lin-23::gfp; rol-6(su1006)]; otEx840[lin-23::lin-23::gfp;rol-6(su1006)]; rrEx128/129/130[end-3::gfp::cdc-25.1(wt); unc-119+];rrEx131/132/133[end-3::gfp::cdc-25.1(rr31); unc-119+]; rrEx155/156/157[end-3::gfp::cdc-25.1(rr31rr36); unc-119+]; rrEx212/213[end-3::gfp::cdc-25.1(rr36); unc-119+]; rrEx134/135/136/137[end-3::gfp::cdc-25.1(wt); elt-2::gfp];rrEx138/139/140[end-3::gfp::cdc-25.1(rr31); elt-2::gfp];rrEx161/162/163[end-3::gfp::cdc-25.1(rr31rr36); elt-2::gfp];rrEx148/149/150/206/207/208/209 [end-3::lin-23; inx-6::gfp] and rrEx152/153/154[end-3::lin-23(ΔSalI); inx-6::gfp].

pMR174 end-3::gfp was generated by cloning the 990 bp BamHI/SpeI GFP fragment from pJH4.52(Strome et al., 2001) between a 1455 bp HindIII/BamHI end-3 promoter fragment (no end-3 coding sequence) and a 270 bp SpeI/ApaI end-3 3′UTR fragment amplified from pMM446 end-3::end-3(P202L)::gfp (gift from M. Maduro) into pPD49.26ΔHindIII/ApaI. pMR513 end-3::gfp::cdc-25.1(wild-type), pMR514 end-3::gfp::cdc-25.1(rr31) and pMR556 end-3::gfp::cdc-25.1(rr31rr36) were generated by inserting a 1818 bp SpeI fragment of cdc-25.1 cDNA from wild type, cdc-25.1(rr31) and cdc-25.1(rr31rr36) mutants, respectively,into the SpeI site of pMR174. pMR579 end-3::gfp::cdc-25.1(rr36) was generated by PCR-mediated mutagenesis from pMR513. pMR562 end-3::lin-23 was constructed by inserting a wild-type 2.6 kb BglII/SpeI fragment of genomic lin-23 into pMR174ΔBamHI/SpeI. pMR572 end-3::lin-23(ΔSalI) was generated by removing the 1.9 kb SalI fragment from pMR562. pMR576 pie-1::gfp::lin-23 was constructed by inserting a wild-type 2.6 kb SpeI fragment of genomic lin-23 into pJH4.52Δ SpeI together with a rescuing unc-119+ fragment into the SacII site. pMR911 inx-6::gfp was created by introducing a 3.1 kb SphI/XbaI fragment of the inx-6 promoter into pPD95.77 and used as a transformation marker. Primer sequences are available upon request.

C. elegans strains were transformed as previously described(Mello et al., 1991). Plasmids were injected at a concentration of 25 ng/μl. At least three independent transgenic lines were analysed for each construct. pMR174, pMR513, pMR514,pMR556 and pMR579 were microinjected into unc-119(ed3) with the unc-119+ rescuing plasmid pDP#MM051(Maduro and Pilgrim, 1995) or into N2 with the intestinal-specific pJM67 elt-2::gfp construct(Fukushige et al., 1999). pMR562 was microinjected into rrIs01, (cdc-25.1(rr31);rrIs01) and rrEx134[end-3::gfp::cdc-25.1(wt); elt-2::gfp]animals with pMR911 as a marker. pMR572 was microinjected into rrEx134[end-3::gfp::cdc-25.1(wt); elt-2::gfp] animals with pMR911.

Germline expression of pie-1::gfp::lin-23 was achieved by bombarding pMR576 into unc-119(ed3) worms as previously described(Praitis et al., 2001). A similar expression pattern was detected with a complex array(Kelly et al., 1997).

lin-23(RNAi) was performed by injecting 1 μg/μl lin-23 dsRNA into young adults(Fire et al., 1998). Embryos were dissected from 36 and 48 hour post-injection hermaphrodites and mounted for microscopic observation. Feeding RNA interference experiments were carried out according to standard procedures(Kamath et al., 2003). Hypomorphic RNAi was achieved by shortening the time during which animals were fed on dsRNA producing bacteria.

MR142 cdc-25.1(rr31); rrIs01 animals were mutagenized at the L4 stage with 40 mM ethylmethanesulfonate (EMS)(Brenner, 1974). The F3 generation was screened under a fluorescent dissecting microscope for animals with reduced cdc-25.1(rr31)-dependent hyperplasia. A total of 5456 haploid genomes were analyzed and five suppressors isolated. Owing to its strong suppression and tight genetic linkage with cdc-25.1(rr31), we hypothesized rr36 represented an intragenic suppressor. Independent sequencing analyses of MR183 cdc-25.1(rr31rr36) identified the second mutation. rr36 is a typical G to A EMS-induced transition that affects nucleotide 817 of cdc-25.1 cDNA.

Light microscopy, image analysis and processing were performed essentially as previously described (Kostic and Roy,2002).

We previously described a G47D amino acid change in CDC-25.1(rr31)that stabilizes the protein in all embryonic cells(Kostic and Roy, 2002). In an independent study, a S46F substitution in CDC-25.1(ij48) was identified and found to cause a similar intestinal-specific hyperplasia(Clucas et al., 2002),indicating that this domain plays an important regulatory function. We therefore sought to better understand when and how these gf mutations affect CDC-25.1 stability, in addition to providing insights concerning the molecular mechanisms responsible for the tissue-specific cell cycle defect associated with both cdc-25.1(gf) mutations.

Both S46F and G47D mutations affect adjacent residues in the N-terminal regulatory domain of the phosphatase, and although the kinetics of CDC-25.1(ij48) turnover have not been described, both proteins appear to behave similarly at higher temperatures(Table 1). Sequence analysis of the affected region predicts a conserved DSGX₄S motif previously described as a phosphodegron recognized by the F-box protein β-TrCP when phosphorylated on both serines (Fig. 1) (Fuchs et al.,2004). It is therefore quite plausible that both mutations enable the corresponding CDC-25.1(gf) proteins to escape timely degradation during embryogenesis by independently affecting two highly conserved residues within a potential β-TrCP-like phosphodegron.

Although the identified CDC-25.1(gf) mutations affect the appropriate turnover of CDC-25.1, whether this enhanced stability is responsible for the supernumerary E divisions observed remains questionable. This is particularly intriguing as CDC-25.1(rr31) perdures in every embryonic cell beyond the 100-cell stage (corresponding to E⁸stage), whereas only the intestinal lineage is affected(Kostic and Roy, 2002). Therefore, to test whether an alteration of the putative β-TrCP-like phosphodegron stabilizes CDC-25.1 and whether this stabilization is the basis of the intestinal hyperplasia, we designed a degradation assay to assess the stability and the function of the wild-type and CDC-25.1(rr31)proteins during C. elegans embryogenesis. By taking advantage of the early intestinal-specific expression of end-3(Maduro et al., 2005), we monitored the kinetics of wild-type and gain-of-function GFP-tagged CDC-25.1 variants in the E lineage. In wild-type animals that express the wild-type variant GFP::CDC-25.1[WT], nuclear fluorescence was detected from the E² to the E⁸ stage with a rapid loss of signal within the first 20-25 minutes of E⁸ (n=30/30), consistent with the GFP::CDC-25.1[WT] fusion protein being degraded during this period(Fig. 2). The gain-of-function GFP::CDC-25.1[G47D] variant, however, was still detectable ∼120 minutes after the E⁸ stage before it completely disappeared(n=30/30). Furthermore, the GFP::CDC-25.1[G47D] variant consistently triggered an extra mitosis at the stage when cdc-25.1(rr31) mutant embryos display their supernumerary division, around 40-45 minutes into the E⁸ stage (∼200-cell stage)(Fig. 2).

By scoring intestinal cells in L1 hatchlings, we found that transgenic animals expressing GFP::CDC-25.1[WT] underwent normal intestinal development and had ∼20 intestinal cells (9/10 independent transgenic lines), whereas fewer than 3% (n=200) exhibited slight intestinal hyperplasia (<25 intestinal cells). One exception was observed (wild-type#4) where 84% of the animals displayed modestly increased intestinal cell counts (∼30 intestinal cells) owing to higher transgene copy numbers(Table 2 and data not shown). Conversely, 90% (n=200) of the animals expressing GFP::CDC-25.1[G47D]displayed severe intestinal hyperplasia with ∼40 intestinal cells (10/10 independent transgenic lines), similar to what is detected in both cdc-25.1(gf) mutants at 20°C (compare Table 1 with Table 2). All the transgenic lines that displayed intestinal hyperplasia (wild-type#4 and the gain-of-function CDC-25.1 variants) were suppressed by cdc-25.1(RNAi), confirming that the cell cycle defects observed in these animals were due to the cdc-25.1 gene product (data not shown). No intestinal defects were seen post-embryonically in any of the transgenic animals.

These results indicate that the degradation machinery that destabilizes CDC-25.1 is still functional at the beginning of the embryonic E⁸stage, while confirming that the G47D change caused by rr31 perturbs the timely turnover of the phosphatase, causing it to perdure beyond this stage. Our findings also reveal a window of sensitivity in the E⁸stage during which the intestinal cells can respond to inappropriate levels of CDC-25.1 activity, thus providing a conclusive link between the stabilization of the phosphatase and the observed cell cycle defect.

As the affected sequence in both cdc-25.1(gf) mutations matches a consensus β-TrCP destruction motif, we wondered whether the cell cycle defects observed in cdc-25.1(rr31) could be phenocopied by loss-of-function in the C. elegans β-TrCP orthologue lin-23. LIN-23 was first described as a general negative cell cycle regulator that restricts embryonic and post-embryonic cell divisions(Kipreos et al., 2000), while it is also implicated in proper axon growth and pathfinding(Mehta et al., 2004), and in postsynaptic glutamate receptor recycling through destabilization ofβ-catenin (Dreier et al.,2005). lin-23(RNAi) and lin-23(0) embryos arrest without undergoing morphogenesis and display severe embryonic hyperplasia(Kipreos et al., 2000). We quantified the intestinal cells in terminally arrested lin-23(RNAi)embryos and counted up to 100 cells instead of the typical 20 or 38-40 observed at similar stages in wild-type and cdc-25.1(gf) embryos,respectively (Fig. 3A),indicating that loss of lin-23 causes more severe intestinal hyperplasia than stabilization of CDC-25.1 alone.

Owing to the involvement of the Wnt pathway in the E cell fate specification (Goldstein,1992; Huang et al.,2007; Phillips et al.,2007; Rocheleau et al.,1997; Thorpe et al.,1997) and as LIN-23 is known to target at least one β-catenin orthologue (BAR-1) for degradation in C. elegans(Dreier et al., 2005), we asked whether the lin-23(RNAi)-associated hyperplasia originates from a cell cycle defect or a cell fate transformation in the early embryonic blastomeres that result in extra intestinal cells. Using a combination of intestinal-specific markers, we found that the specification of the E lineage occurs normally at the seven-cell stage in all lin-23(RNAi) embryos,but thereafter one or two ectopic E blastomeres were detected in the posterior cells of some embryos (Fig. 3B). Sixty percent of lin-23(RNAi) embryos ectopically express the E marker in the descendants of Cp, while around 5% of them also expressed GFP in MSp descendants (n=91). These cell fate transformations were skn-1 dependent(Bowerman et al., 1992), as none of the lin-23(RNAi); skn-1(RNAi) double mutant embryos that lack endodermal specification produced any E or ectopic E-like cells(n=50). This is consistent with lin-23 being at least partially required in Cp and MSp to inhibit the SKN-1-dependent endoderm specification pathway (Maduro et al.,2005; Maduro et al.,2001; Zhu et al.,1997). Thus, these multiple E lineages generate a variable source of intestinal hyperplasia in lin-23 compromised embryos.

To determine whether the observed lin-23(RNAi) hyperplasia is due exclusively to these cell fate transformations or whether it may also depend on stabilization of its putative cell cycle target CDC-25.1 in the E lineages,we performed lin-23(RNAi) in wild-type animals that express GFP::CDC-25.1[WT] and monitored the kinetics of its turnover after the E⁸-stage. In these lin-23(RNAi) embryos, the tagged phosphatase became stabilized compared with control embryos, causing it to perdure much like the GFP::CDC-25.1[G47D] variant. Furthermore, as in the gain-of-function mutants and their corresponding transgenic variants, these animals displayed supernumerary cell divisions in the E lineage(Fig. 2). To quantify the degree to which the E lineage contributes to the lin-23(RNAi)-associated hyperplasia, E blastomeres were isolated from lin-23(RNAi) embryos by laser ablation. We found that ∼40 intestinal cells originated from the E blastomere under these conditions,indicating that these cells undergo a single supernumerary division when lin-23 is downregulated (Fig. 3C). Thus, through these and reciprocal laser ablation experiments(E ablated) (data not shown), we found that ∼40% of the total lin-23(RNAi)-mediated intestinal hyperplasia originates from the E blastomere, while the remaining ∼60% arises from ectopic E-like cells.

To confirm whether indeed lin-23 and cdc-25.1 act cooperatively in a linear pathway to regulate the intestinal cell divisions,we quantified the severity of the intestinal hyperplasia of terminally arrested lin-23(RNAi); cdc-25.1(rr31) double mutants. In these animals, the intestinal defect was only slightly more pronounced than that of lin-23(RNAi) single mutants and non-additive with cdc-25.1(rr31) (Fig. 3C), suggesting that both lin-23(lf) and cdc-25.1(rr31) act in a linear pathway and consistent with lin-23 negatively regulating cdc-25.1, probably through protein destabilization.

Homozygous viable lin-23(ot1) mutants also display mild cdc-25.1-dependent intestinal hyperplasia that arises during embryogenesis (∼26 intestinal cells at hatching)(Fig. 4A). Partial depletion of the phosphatase by cdc-25.1(RNAi) entirely suppressed this lin-23(ot1)-dependent cell cycle defect, while cdc-25.1(rr31);lin-23(ot1) double mutants did not display stronger intestinal hyperplasia than cdc-25.1(rr31)(Fig. 4B). As in cdc-25.1(rr31) mutants, no other embryonic or post-embryonic cell cycle perturbation nor any evidence of cell fate changes was observed,suggesting that lin-23(ot1) represents a weak hypomorph.

In summary, we show that lin-23 compromise causes mild to severe intestinal hyperplasia originating from two independent sources: through MSp and/or Cp to E cell fate transformations that generate extra E-like cells during early embryogenesis and through stabilization of CDC-25.1 that probably contributes to the supernumerary cell divisions in the E descendants, much like that observed in both cdc-25.1(gf) mutants. Furthermore, that the lin-23(ot1) mutation affects both intestinal cell cycle progression and appropriate neurotransmission would argue against the possibility of separable cell cycle and neuronal-specific target recognition domains in the LIN-23 protein, as previously hypothesized(Mehta et al., 2004).

Our findings suggest that lin-23 mediates the timely degradation of CDC-25.1, but the importance of the putative phosphodegron sequence in this process remains unclear. We therefore questioned whether LIN-23 might be limiting and whether simply increasing LIN-23 levels could suppress the intestinal hyperplasia typical of cdc-25.1(rr31) mutants. To test this, we generated a transgene that expresses lin-23 specifically in the E lineage prior to the stage when cdc-25.1(rr31)-dependent supernumerary divisions occur. The transgene completely rescues the lin-23(ot1)-dependent hyperplasia (5/5 independent transgenic lines),yet it had no effect on the intestine when expressed in wild-type animals. When expressed in cdc-25.1(rr31) mutants, we observed a slight, but reproducible suppression of the intestinal hyperplasia(Table 3), consistent with a reduced capacity of LIN-23 to trigger efficiently the degradation of the phosphatase with the mutant β-TrCP phosphodegron.

To test the importance of the phosphodegron sequence in the degradation process, we used the transgenic line (wild-type#4) that has an intact phosphodegron and causes modest intestinal hyperplasia. Following overexpression of LIN-23, we noted that the majority of animals containing both arrays displayed wild-type intestinal cell counts and morphology(Table 4). This suppression required an intact F-box protein, as a functionally compromised LIN-23 variant that lacked the typical F-box WD repeats and the F-box domain that are required for interaction between substrate and the E3 ligase components,respectively, did not suppress the extra divisions(Table 4).

Because the GFP::CDC-25.1[wild-type#4]-associated defect is suppressed by overexpression of LIN-23, while the CDC-25.1(rr31)-dependent hyperplasia is only marginally ameliorated, it is likely that an intact phosphodegron/destruction signal is required for the timely and efficient degradation of CDC-25.1 by LIN-23.

Although CDC-25.1(rr31) perdures beyond the 100-cell stage in most cells in the embryo it eventually disappears(Kostic and Roy, 2002). We have shown that the GFP::CDC-25.1[G47D] fusion protein also gradually disappears in our transgenic assay compared with the control GFP(Fig. 2). This suggests that another pathway could control CDC-25.1 levels through a secondary mechanism that may be independent of β-TrCP/LIN-23 activity. Using genetic analysis to isolate suppressors of the cdc-25.1(rr31) intestinal hyperplasia,we identified rr36, a recessive, maternal-effect intragenic mutation in cdc-25.1(rr31) that restores intestinal cell counts to wild-type levels (Fig. 4B). rr36causes a L273F substitution adjacent to the rhodanese-like catalytic domain of the phosphatase (Fig. 1). cdc-25.1-null mutations and cdc-25.1(RNAi) are embryonic lethal(Ashcroft and Golden, 2002; Ashcroft et al., 1999), but incomplete depletion of CDC-25.1 by RNAi in cdc-25.1(rr31) animals suppresses their supernumerary intestinal divisions(Kostic and Roy, 2002). We noticed that rr36 is temperature-sensitive embryonic lethal, wherein 10% of the embryos die at 15°C or 20°C, but this lethality increases to 100% when shifted to 25°C. Furthermore, when young larvae were upshifted at 25°C, they developed into sterile adults and displayed phenotypes very similar to cdc-25.1(0) mutants or cdc-25.1(RNAi)-treated animals, with very few germ cells(Ashcroft and Golden, 2002; Ashcroft et al., 1999) (and data not shown), suggesting that rr36 represents a new,temperature-sensitive, hypomorphic cdc-25.1 allele.

Because the rr36 lesion is in close proximity to the catalytic domain, we wondered whether rr36 restores the intestinal cell divisions by compromising the catalytic efficiency of the phosphatase or whether it re-establishes the proper turnover of CDC-25.1(rr31). Wild-type animals that express a transgene that mimics this suppressor variant(GFP::CDC-25.1[G47D;L273F]) in the early E lineage displayed timely degradation of the fusion protein after the E⁸ stage, with only∼10 minutes delay compared with GFP::CDC-25.1[WT](Fig. 2), while most of these transgenic animals displayed wild-type intestinal division patterns that resulted in 20 intestinal cells at hatching(Table 2). Interestingly, a GFP::CDC-25.1[L273F] fusion protein that mimics the rr36 mutation alone, was less stable than the wild-type variant, and was degraded before the E⁸ stage (Fig. 2). This would not be expected if the rr36 mutation resulted in a catalytically compromised CDC-25.1. Thus, it is most likely that the domain affected by rr36 is required for the normal stability of CDC-25.1,rather than for its catalytic activity, although the latter cannot be entirely ruled out.

To determine whether the rr36 mutation destabilizes CDC-25.1 by enhancing the effects of LIN-23 or the SCF^LIN-23 complex, we tested whether this mutation could suppress the cdc-25.1-dependent intestinal hyperplasia typical of lin-23(RNAi) embryos. By quantifying the intestinal cells in cdc-25.1(rr31rr36) mutants treated with lin-23(RNAi), we noticed that rr36 reduced the lin-23(RNAi)-mediated intestinal hyperplasia by ∼70%(Fig. 3C), while allowing a significant number of embryos to develop beyond the embryonic arrest typical of lin-23(RNAi) (Table 5). A positive effect of destabilizing CDC-25.1 was also observed in cdc-25.1(rr31rr36); lin-23(ot1) double mutants, wherein the intragenic mutation completely restored the intestinal cell counts to wild-type (Fig. 4B). However,the frequency of the lin-23(RNAi)-dependent cell fate change was unaffected and still occurred in cdc-25.1(rr31rr36); lin-23(RNAi)embryos (data not shown).

To confirm that effects observed with rr36 were indeed LIN-23-independent, we performed lin-23(RNAi) in transgenic embryos expressing GFP::CDC-25.1[L273F] during the early intestinal divisions. The GFP fusion was detected until the middle of the E⁸-stage, similar to GFP::CDC-25.1[G47D;L273F], and unlike lin-23(RNAi)-treated GFP::CDC-25.1[WT] embryos, neither abnormal perdurance nor extra intestinal cell divisions were detected (Fig. 2).

These results reveal the possibility that independent and antagonistic signals may impinge upon distinct domains in CDC-25.1 to fine-tune its timely turnover during embryogenesis. Furthermore, and perhaps most surprising, the general hyperplasia characteristic of terminal lin-23(RNAi) embryos appears to be, at least in part, cdc-25.1 dependent, as some animals lacking lin-23, which normally arrest as overproliferated balls of cells, progressed to the hatch when CDC-25.1 was destabilized.

Although the stability of CDC-25.1 may depend on a LIN-23-dependent interaction with its N-terminal regulatory domain, the tissue specificity of the cdc-25.1(gf)-associated defect remains enigmatic. One means of achieving this tissue specificity could include the differential expression,stability or activity of the putative SCF^LIN-23 degradation complex in the developing embryo. By in situ hybridization, lin-23 mRNA was shown to be maternally provided to embryos, where it is ubiquitous(Kipreos et al., 2000). We confirmed the presence of maternal LIN-23 in all cells until the 100-cell stage (E⁸) in transgenic worms expressing GFP::LIN-23 in their germ line (Fig. 5A). LIN-23 was reported to be zygotically expressed in most embryonic cells(Mehta et al., 2004). Consistent with this, our analysis of zygotic LIN-23::GFP indicates that it is expressed at relatively high levels in most cells, but surprisingly, very low,if any, GFP signal was detectable in the developing intestine. This differential expression was apparent at the onset of GFP detection at approximately the 51-cell stage (E⁴-stage) and lasted until the end of embryogenesis (Fig. 5B). To discriminate whether the synthesis or the stability of LIN-23 may be differentially controlled in the E lineage, we analyzed the expression pattern of a lin-23 promoter fusion. The transcriptional GFP pattern looked indistinguishable from the translational fusion (data not shown), confirming that the zygotic transcription of lin-23 is differentially regulated such that LIN-23 levels are lower in the E lineage. Therefore, the downregulation of a putative SCF^LIN-23 E3 ligase in the developing intestine may account for some of the tissue specificity associated with the cdc-25.1(gf) mutation and may provide insight into why other tissues are refractory to CDC-25.1(gf).

The early cell divisions of the C. elegans embryo are precisely controlled by gene products that are provided from the maternal germ line. The isolation of two gain-of-function alleles of CDC-25.1 that demonstrate a strict maternal effect indicates that this regulation can be perturbed,resulting in supernumerary cell divisions specifically within the E lineage. How these mutations give rise to the extra divisions is unclear, although CDC-25.1(rr31) was observed to be more stable than its wild-type counterpart. By developing a GFP-based transgenic assay to assess the dynamics of protein degradation during early embryogenesis, we show that the stabilization of CDC-25.1 is mediated by a point mutation within a conserved DSGX₄S β-TrCP-like phosphodegron in both cdc-25.1(gf)mutants. This stabilizes the protein, resulting in the abnormal presence of CDC-25.1 during a short, yet crucial, window during early development, which is presumably the cause of the observed cell cycle defect.

The equivalent phosphodegron/destruction motif is recognized by β-TrCP F-box containing proteins in other systems(Fuchs et al., 2004). Both serines in the consensus site must be phosphorylated to constitute a phosphodegron trigger that is efficiently recognized by an SCF^β-TrCP E3-ligase(Winston et al., 1999). In mammals, the checkpoint protein Chk1 facilitates the SCF^β-TrCP-mediated degradation of Cdc25A during unperturbed S phases by phosphorylating several N-terminal regulatory residues, but the kinase that modifies the actual β-TrCP motif remains unidentified(Busino et al., 2003; Jin et al., 2003). Our findings suggest that in C. elegans, the equivalent DSGX₄S motif in CDC-25.1 is targeted by the β-TrCP F-box orthologue LIN-23, but in a developmentally controlled manner, so that cell divisions are coordinated with the ongoing cellular processes that normally occur during embryogenesis.

By using our transgenic assay to monitor the dynamics of CDC-25.1 degradation in the E lineage, we show that the wild-type protein is stabilized when lin-23 activity is reduced, while the embryonic E lineage overproliferates, much like in cdc-25.1(gf) mutants. We also confirmed that the turnover of CDC-25.1 is regulated in a linear pathway that includes lin-23. Our data suggest that a putative SCF^LIN-23 E3 ligase actively targets CDC-25.1 for destruction during early embryogenesis, while this targeted elimination requires an intact, functional β-TrCP destruction box for efficient degradation. In the cdc-25.1(gf) mutants, this degradation process is impaired because of the change in the β-TrCP motif, resulting in abnormal stabilization of this protein. The potential targeting kinase and its upstream regulators may also be crucial for the timely regulation of this process, as their activity would ultimately trigger the degradation cascade.

Interestingly, the intragenic suppressor cdc-25.1(rr31rr36)displays phenotypes indistinguishable from the cdc-25.1(0) mutants when grown at the restrictive temperature(Ashcroft and Golden, 2002; Ashcroft et al., 1999). This is consistent with either the compromise of cdc-25.1 catalytic activity or changes in the levels of the protein. Our data indicate that rr36destabilizes the protein in a manner independent of LIN-23 and/or its recognition site.

At present, it is not clear how the domain affected by rr36confers this function, but it may provide a second regulatory interface to ensure the timely elimination of the phosphatase in response to spatial and/or temporal developmental cues. Using various bioinformatic approaches, we were unable to identify any known motifs associated with post-translational modification around the L273F lesion, with the exception of a strong phosphoacceptor value attributed to Y₂₇₁. It is, thus, conceivable that a phosphorylation event near L₂₇₃ stabilizes the phosphatase,similar to what is seen after phosphorylation of key residues in p53, p21 or CDC25A (Buschmann et al., 2001; Chehab et al., 1999; Lavin and Gueven, 2006; Li et al., 2002; Mailand et al., 2002). Alternatively, the region affected by the rr36 lesion could modify the association between CDC-25.1 and small regulatory proteins such as 14-3-3 or Pin1, both of which affect the localization and/or stability of mammalian Cdc25A orthologues (Chen et al.,2003; Conklin et al.,1995; Crenshaw et al.,1998; Stukenberg and Kirschner, 2001). The fact that the mutated leucine is conserved among the four C. elegans cdc-25 paralogues, the two Drosophila Cdc25 orthologues and human CDC25A suggests that it could potentially be part of an important regulatory domain that would control protein stability.

The cell cycle phenotypes of cdc-25.1(gf) and lin-23(lf)are similar but not equivalent. Loss of lin-23 results in a more severe embryonic hyperplasia that does not seem to be restricted to the E lineage (Kipreos et al.,2000), similar to the one observed in cul-1(0) (cullin)or skr-1/2(0) (Skp-related) mutants(Kipreos et al., 1996; Nayak et al., 2002). Our observations indicate that a majority of lin-23(0) embryos undergo one or two cell fate transformations, giving rise to separate populations of intestinal-like cells that divide throughout embryogenesis. Interestingly,transformation toward the E fate has also been detected in blastomeres of skr-1/2(RNAi) embryos (Z. Bao, personal communication). These extra-intestinal phenotypes cannot be solely attributed to stabilized CDC-25.1. SCF^β-TrCP targets multiple proteins for degradation,including β-catenin, IκB, Emi1/2, Wee1A and CDC25A/B(Nakayama and Nakayama, 2005; Nakayama and Nakayama, 2006);thus, the pleiotropic effects seen in lin-23(RNAi) suggest that LIN-23 also controls distinct aspects of C. elegans embryogenesis through multiple targets.

Interestingly, most of the hyperplasia that occurs in lin-23(lf)embryos, even that outside the usual E lineage, also appears to be cdc-25.1-dependent. It is therefore tempting to associate the general lin-23(RNAi)-mediated embryonic hyperplasia to misregulated CDC-25.1 activity. Indeed, the hyperplasia and lethality typical of lin-23(RNAi) is partially suppressed by a cdc-25.1(rr31rr36)mutant variant that has no apparent effect on normal cell cycle progression per se. Alternatively, it is plausible that the loss of lin-23 alone can induce hyperplasia independently of changes in CDC-25.1 stability. Thus,the absence/reduction of CDC-25.1(rr31rr36) protein would prevent excessive cell divisions in lin-23-depleted embryos indirectly, as would be the case by eliminating any other crucial cell cycle regulator.

Conversely, the lin-23(lf)-dependent early cell fate transformations are cdc-25.1 independent and are more likely to be the result of inappropriate regulation of developmental signals downstream, or in parallel, to the skn-1-dependent endodermal specification pathway. Interestingly, embryos depleted of gsk-3, the GSK3β orthologue,display skn-1-dependent Cp to E cell fate transformations(Maduro et al., 2001; Schlesinger et al., 1999),similar to what we observe in lin-23(RNAi) animals. As the Wnt pathway plays multiple roles in early cell fate specification(Huang et al., 2007; Phillips et al., 2007; Rocheleau et al., 1997; Thorpe et al., 1997), the misregulation of wrm-1 or sys-1 (two embryonically expressedβ-catenin-like molecules) could affect MS, C, D or even other blastomeres to give rise to the extra intestinal cell populations observed in lin-23 mutant embryos. Interestingly, other mutants with such C-derived extra intestinal cells may indirectly affect the Wnt pathway(Shirayama et al., 2006).

The transgenic GFP::CDC-25.1[G47D] resists degradation and triggers supernumerary divisions of the E daughters ∼40-45 minutes into the E⁸-stage, very similar to what is observed in cdc-25.1(rr31) mutants, defining a crucial period of sensitivity midway into the E⁸-stage (∼200-cell stage embryo). This window of sensitivity may reflect a temporal transition where maternal cdc-25.1-dependent regulation of cell cycle progression is gradually replaced by zygotic control, possibly through expression of other cdc-25 paralogues or even other positive or negative S-phase regulators. Our data suggest that abnormally high levels of CDC-25.1 that surpass a crucial threshold during this transition period may be sufficient to trigger extra intestinal divisions before zygotic cell cycle control would limit further overproliferation.

During embryogenesis, cell cycle timing is regulated through the control of S phase progression (Edgar and McGhee,1988), wherein each lineage could potentially use distinct mechanisms of coordinating cell division timing with other concurrent developmental events. This maternal to zygotic transition may indeed be one such property that is particular to the E lineage, thus rendering it sensitive to changes in CDC-25.1 turnover.

The differential expression of zygotic lin-23 in embryos from the 51-cell stage until the end of embryogenesis may help to explain the intestinal specificity associated with the lin-23(ot1) and cdc-25.1(gf) mutations. The combination of maternal and zygotic LIN-23 in non-E tissues could indicate a higher catalytic efficiency of their SCF^LIN-23 E3 ligase compared with that acting in the E lineage. Consistent with this idea, lin-23(ot1) animals display weak cdc-25.1-dependent hyperplasia solely in the intestinal tissue, despite wild-type levels and pattern of LIN-23(ot1)expression (Mehta et al.,2004). Thus, LIN-23(ot1) probably still recognizes its embryonic targets, although with reduced efficiency. Accordingly, lower levels or reduced activity of SCF^LIN-23(ot1) in E would stabilize CDC-25.1 to varying degrees causing moderate intestinal hyperplasia, while the higher E3 ligase activity in other tissues would prevent any cell cycle perturbation. Albeit speculative, this could also explain the tissue specificity of both cdc-25.1(gf) alleles. Although CDC-25.1(rr31) perdures in all cells, it does not cause extra divisions in every tissue, suggesting that despite the impaired phosphodegron,its levels are still maintained below a crucial threshold outside the E lineage. However, overexpression of LIN-23 in the early E lineage can only reduce the cdc-25.1(rr31)-mediated hyperplasia by a small margin. Therefore, LIN-23 may not be the only limiting factor for the catalytic efficiency of SCF^LIN-23, inferring that other components of this complex may also be limiting in the developing intestine.

The reason for the zygotic LIN-23 expression pattern remains unclear and may reflect intrinsic differences in the requirement for lin-23-dependent pathways between lineages. For example, early expression of lin-23 in Cp and MSp would be required to restrict the intrinsic potential of these lineages to adopt endodermal fates, while downregulation of SCF^LIN-23 in the E lineage may be required for the Wnt-dependent migration of Ea and Ep(Lee et al., 2006), and possibly subsequent morphogenetic movements associated with gastrulation. From this perspective, the levels of CDC-25.1 in the developing gut may be held very close to a sensitive threshold and small increases resulting from defects in SCF^LIN-23-dependent degradation may result in intestinal hyperplasia.

We thank Dr Christian Rocheleau, Dr François Fagotto, Ivana Kostic and the members of the Roy laboratory for their help, comments and critical reading of the manuscript. We are also grateful to Dr Zhirong Bao for sharing results before publication. We thank Dr Morris Maduro and the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR), for C. elegans strains and reagents. This work was supported by a Research Award from the Canadian Cancer Society. R.R. is a CIHR New Investigator.