Organ-specific cell division abnormalities caused by mutation in a general cell cycle regulator in C. elegans

Cell proliferation is essential for many key processes that occur during development including organogenesis, tissue renewal and germline formation. (Bartkova et al., 1997; Clurman and Roberts, 1995; Pines, 1995; Sandhu and Slingerland, 2000). Therefore, the timing of cell division and differentiation must be precisely coordinated with signals that specify morphogenesis, patterning and growth in a temporal, positional and cell type-specific manner (reviewed by Vidwans and Su, 2001). This coordination is executed through regulating both positive and negative regulatory components of the basal cell cycle machinery.

The cell cycle machinery is well conserved among eukaryotes and complex mechanisms ensure that cell cycle progression occurs in a timely and precise sequence. Cyclin-dependent kinases (Cdks) drive progression through the different cell cycle phases (reviewed by Nigg, 2001). In yeasts, these catalytic subunits are regulated through their association with stage-specific cyclin regulatory subunits (Wittenberg et al., 1990; Forsburg and Nurse, 1991). However, in more complex multicellular organisms, larger families of Cdks and cyclins exist, and their elaborate regulation provides cell-type and functional diversity.

These individual Cdks are activated in a cell cycle stage-specific manner (reviewed by Sherr, 1994; Sherr, 1996; Tsai et al., 1993; Draetta and Beach, 1988). The activity of these cyclin/Cdk complexes is required to phosphorylate substrates necessary to drive cell cycle progression and are regulated by activating and/or inhibitory kinases, or phosphatases, such as those of the cdc25 family (Nilsson and Hoffmann, 2000; Nigg, 2001). Cdks can also be negatively regulated by cyclin-dependent kinase inhibitors (CKIs); small polypeptides that bind to and inhibit the catalytic activity of these kinases (Sherr and Roberts, 1999).

Among the various cell cycle transitions, the G₁/S transition represents an important regulatory milestone where extracellular signals are integrated resulting in the progression of cell division or, alternatively, cell cycle arrest in G₁ or G₀ (Pardee, 1989; Sherr, 1994). Coordination of cell cycle progression and arrest may depend on the function of the CKI p27^KIP1, while final growth arrest and differentiation may require the downregulation of positive cell cycle regulators (Koff and Polyak, 1995; Casaccia-Bonnefil et al., 1999).

In a multicellular organism, cell divisions must be coordinated with the developmental program to ensure the cellular integrity in all tissues of the organism. These developmental signals converge on many of the same key cell cycle components described above. Studies performed in Drosophila have shown that developmental signals impinge on the positive cell cycle regulator String, a homolog of the G₂/M-specific Cdc25 phosphatase, at several points during development (Foe, 1989; Edgar et al., 1994a; Edgar et al., 1994b; Edgar and O’Farrell, 1989). The G₁/S transition is also developmentally regulated in flies through the activity of CKIs and cyclin E and cyclin D levels (Cayirlioglu and Duronio, 2001; Moberg et al., 2001; de Nooij et al., 1996; Lane et al., 1996).

In addition to cell cycle regulators that act globally, the activity of some regulators is important for the proper proliferation of cells in tissues at specific times during development. For example, in Drosophila, Roughex (Rux), acts specifically in the eye and in the male germ line to arrest cells in G₁ phase (Thomas et al., 1994; Gonczy et al., 1994; Avedisov et al., 2000). Decapentaplegic, a TGFβ family member, is required for the establishment of G1 arrest before differentiation during Drosophila eye development (Horsfield et al., 1998), while it is also essential for proliferation in the wing and in the germline (Burke and Basler, 1998; Xie and Spradling, 1998). Therefore, the complexity of tissues and the regulated development of many multicellular organisms make it difficult to characterize precisely how cell divisions are controlled in a specific developmental context.

The invariant cell lineage of C. elegans provides an invaluable tool to study cell division abnormalities at single cell resolution (Brenner, 1974). As the timing and fate of every cell division has been documented in a lineage map, the analysis of the effects of various developmental regulators on the cell cycle at specific developmental points and/or in specific cell lineages is possible (Sulston and Horvitz, 1977; Sulston et al., 1983).

Several conserved developmental regulatory genes have been shown to control embryonic and postembryonic cell division, and often, the resulting daughter cell fates in C. elegans (Kimble and Simpson, 1997; Euling and Ambros, 1996; Rougvie and Ambros, 1995). Mutations of conserved negative regulators have also been described, where the number of cell divisions and exit to G₀ has been shown to be regulated through the degradation of G₁ cyclins (Kipreos et al., 1996). The C. elegans p27^KIP1 homolog, cki-1, has been shown to confer developmental G₁ cell cycle arrest and to be one of the downstream effectors of many developmental pathways (Hong et al., 1998). Loss of cki-1 results in extra cell divisions in numerous lineages causing abnormalities in the organogenesis of the vulva, the somatic gonad, the hypodermis, and intestine (Hong et al., 1998).

To understand the nature of the developmental signaling pathways that regulate cell division in specific lineages and during organogenesis, we designed a screen to isolate mutants that had altered cell division in specific organs without affecting overall cell division. To do this, we focused on mutants that phenocopy the loss of cki-1 in the intestinal lineage using a lineage-specific GFP reporter. The study of mutants with organ-specific cell cycle aberrations could serve to elucidate the important role of cki-1 or other upstream regulators in linking developmental signals with normal cell type-specific cell cycle dynamics, while providing further tools to identify factors that confer tissue specificity.

We report the identification and the characterization of a maternal-effect, gain-of-function allele of the proto-oncogene cdc-25.1, one of the four C. elegans cdc25 homologs, which has a conserved role in positively regulating the G₁/S transition (Galaktionov et al., 1995b; Ashcroft et al., 1998). This allele causes tissue-specific embryonic cell cycle abnormalities, which occur in the cells that form the C. elegans intestine.

In this study, we used the following strains and chromosome rearrangements: N2 wild-type Bristol, RW7000 wild-type Bergerac, CB4856 wild-type Hawaiian, VT765 (maIs103 [(rnr::GFP unc-36(+)]X) (Hong et al., 1998), KM32 (gvEx32 [cye-1::GFP; rol-6D]; a gift from M. Krause, KR1142 (hDf8/szT1(lon-2(e678)) I; +/szT1 X), JK1726 (qDf16/dpy-5(e61) unc-15 (e1402)), EU384 [dpy-11(e1180) mom-2(or42) V/ nT1 (let-?(m435)] (IV;V), MR136 (rrEx04 [elt-2::GFP]; a gift from J. McGhee), MR156 (rrIs01 [elt-2::GFP; unc-119(+)]), rrEx12 [hs::cdc-25.1(+); ttx-3::GFP], rrEx13 [hs::cdc-25.1(gf); ttx-3::GFP], MR142 (rr31; rrIs01), MR178 (maIs103; rrEx12), MR180 (maIs103; rrEx13), MR196 (rrEx12; gvEx32), and MR197 (rrEx13; gvEx32). Strains were cultured using standard techniques (Brenner, 1974).

rrls01 animals were mutagenized with 40 mM ethylmethanesulfonate (EMS) (Brenner, 1974). Mutagenized L4 hermaphrodites were picked to plates (25-30 per plate) and allowed to produce progeny at 25°C. F₁ animals in the L4 stage were transferred to 60 mm plates, five per plate, and the F₂ progeny were screened for mutants that have extra numbers of intestinal nuclei, a phenocopy of cki-1(RNAi) animals, scoring with a fluorescent dissecting microscope. Candidate mutants were recovered and transferred to separate plates, and their progeny were examined for the presence of extra intestinal nuclei. 10,320 haploid genomes were screened.

rr31 was mapped to the right arm of chromosome I using RW7000 and STS markers (Williams et al., 1992), SNIP-SNP mapping using CB4856 (Wicks et al., 2001), followed by three factor mapping to the dpy-5 unc-13 interval.

pMR405 and pMR409 were generated by inserting 2098 bp of the cdc-25.1 sequence amplified from rr31 [cdc-25.1(gf)] and wild-type animals, respectively, into the pGEM-T vector (Promega). pMR407 and pMR408 were generated by inserting a 7495 bp PCR product, including 5035 bp of upstream sequence and 366 bp 3′ to the translational stop site corresponding to the mutant (rr31) or the wild-type cdc-25.1(gf) gene, respectively, into pGEM-T (Promega). pMR410 and pMR411 were generated by inserting the wild-type cdc-25.1 genomic sequence into the NcoI/SacI sites of pPD49.78 and pPD49.83, respectively. pMR412 and pMR413 were generated by inserting the mutant cdc-25.1 genomic sequence into the NcoI/SacI sites of pPD49.78 and pPD49.83, respectively. For sequencing of the mutant or wild-type cDNA, polyA RNA was isolated from cdc-25.1(gf) or wild-type animals, and mutant and wild-type cDNA was amplified after reverse transcription. The corresponding PCR products were placed into pGEM-T to yield pMR421 and pMR418.

Worms were transformed by microinjection as previously described (Mello et al., 1991). A 7495 bp PCR product corresponding the cdc-25.1 gene was amplified from cdc-25.1(gf) or wild-type N2 genomic DNA, and injected rrls01 at the concentration of 17 ng/μl with the co-transformation marker pRF4 (rol-6D) at the concentration of 128 ng/μl. MR178 (maIs103; rrEx12) was constructed by injection of 20 ng/μl pMR410 and pMR411 into the rnr::GFP strain with 100 ng/μl ttx-3::GFP (Hobert et al., 1997). MR180 (maIs103; rrEx13) was constructed by injection of 20 ng/μl pMR412 and pMR413 into rnr::GFP with 100 ng/μl ttx-3::GFP.

pMR405, pMR409, pMR421 and pMR418 were sequenced and the sequences were compared with each other and with published genomic sequences available from Wormbase (www.wormbase.org).

The cki-1 dsRNA was produced and injected according to Hong et al. (Hong et al., 1998). cyd-1 and cye-1 dsRNA was produced according to Park and Krause (Park and Krause, 1999) and Fay and Han (Fay and Han, 2000), respectively. cdc-25.1 dsRNA was produced by restriction enzyme digestion of pMR409 with NdeI or SacI for the sense and antisense cdc-25.1 RNA. Gel-purified template (1 μg) was used for in vitro transcription reactions according to Fire et al. (Fire et al., 1998). Double stranded cdc-25.1 RNA was injected into rrls01 or rr31; rrls01 animals at a concentration of 1 mg/ml, and the injected animals were transferred daily to new plates, where the intestinal cell number of the F₁ progeny was scored.

Embryos dissected from gravid rrls01 or rr31; rrls01 hermaphrodites were placed on NGM pads and cell divisions were observed from the zygote stage onwards. For the cki-1(RNAi) lineage, F₁ embryos of cki-1 dsRNA-injected hermaphrodites were mounted on NGM pads and cell division timing was recorded by following E cell divisions using the elt-2::GFP reporter.

Animals carrying the mutant or wild type cdc-25.1 transgenes (MR178, MR179, MR181, MR196, MR197) driven by the hsp16-2 and hsp16-41 promoters, or the heat-shock constructs alone were placed in the cye-1::GFP and rnr::GFP background in order to assay the entry into S phase. Adult transformed and non-transformed hermaphrodites were placed at 33°C for 3 hours and then allowed to recover for 2 hours at room temperature. The hermaphrodites were then mounted on 2% agarose pads in 2 mM levamisole, and cye-1::GFP, or rnr::GFP expression was observed.

Antibody staining of embryos with anti-PHA-4 antibody or anti-CDC-25.1 antibody was performed according to Boxem et al. (Boxem et al., 1999) and Ashcroft et al. (Ashcroft et al., 1999), respectively. For immunostaining of larvae, animals were fixed in 3% formaldehyde and antibody staining was performed according to standard procedures (Finney and Ruvkun, 1990).

Images of live embryos, or animals anesthesized with 1 mM levamisole, were captured using the Leica DMR compound microscope equipped with a Hamamatsu C4742-95 digital camera. Image analysis, computational deconvolution and pseudocoloring were performed using Openlab 3.01 software from Improvision. Images were merged using Adobe Photoshop.

The phenotype associated with the loss of cki-1 activity through dsRNA-mediated interference (RNAi) has demonstrated a role of this CKI in the regulation of cell division timing (Hong et al., 1998). Embryos homozygous for a deficiency that uncovers cki-1 arrest with substantially more endodermal precursors implicating a gene mapping within this genetic interval in the embryonic control of cell divisions in the E lineage (M. Fukuyama, S. Gendreau and J Rothman, personal communication). Intestinal cell numbers are increased in adult cki-1(RNAi) animals which possess an average of 50 as compared to 30 intestinal nuclei in wild-type animals (Table 1). To isolate mutants that may regulate cki-1 expression or function in the gut, we screened for mutants that would phenocopy this extra intestinal cell phenotype, using the intestinal-specific reporter elt-2::GFP (Fukushige et al., 1998). 10,320 haploid genomes were screened and although several mutants with fewer intestinal cells were isolated, we have identified only one mutant with extra intestinal cells. This mutant, rr31, has 57±4 intestinal nuclei at the adult stage, or approximately twice the wild-type complement (Fig. 1; Table 1).

Other than the intestinal cell defect, rr31 mutants appear phenotypically normal. To test whether other cell types were affected by the rr31 mutation, we examined cell numbers in rr31 by DIC, DAPI staining and with anti-PHA-4 antibody, which marks mesodermal pharyngeal precursors (Horner et al., 1998). rr31 mutants and wild-type controls showed no differences in these cell lineages examined; however, the number of intestinal cells was markedly elevated judged by the increased number of elt-2::GFP expressing nuclei (data not shown). However, we cannot rule out that there may be other less obvious lineage defects that were not apparent from our examination of rr31 mutants.

During normal development, after a series of mitotic divisions that occur during embryogenesis, the posterior intestinal cells undergo a single nuclear division at the end of the L1 stage, producing binucleate intestinal cells. Therefore, the extra intestinal nuclei in rr31 mutants could be the result of additional mitotic divisions during embryogenesis, or alternatively, extra postembryonic nuclear divisions. To address this, we scored the number of intestinal cells in newly hatched wild-type, rr31 and cki-1(RNAi) L1 larvae. rr31 and cki-1(RNAi) L1s possess an average of 38 (±3) and 29 (±3) intestinal cells, respectively, compared with 20 in wild type. Therefore, we conclude that in rr31 mutants, like cki-1(RNAi), the extra cells in the intestine arise at a point during embryogenesis before hatching. Furthermore, rr31 and cki-1(RNAi) animals stained with the MH27 antibody, which stains the cell junctions of all epithelial cells (Priess and Hirsh, 1986; Waterston, 1988), display numerous extra cell borders in the intestine, indicating that there is an increase in the number of cells, rather than extra nuclear divisions (data not shown).

The extra cells in both rr31 and cki-1(RNAi) backgrounds could arise from additional divisions of intestinal cells during embryogenesis, or from a mis-specification of another cell type into intestinal cells. To further understand when and how the defects occur in these mutant backgrounds, we performed lineage analysis on rr31 animals and cki-1(RNAi) animals. In wild-type animals, the intestine is formed from the E (endoderm) blastomere. During embryogenesis, this founder cell divides four times to give rise to 16E cells, while four of these cells undergo a fifth division, giving rise to the 20 intestinal cells present at hatching (Sulston et al., 1983) (Fig. 2). At the end of the L1 stage, 14 of these cells undergo a nuclear division leading to the formation of binucleate intestinal cells, followed by endocycles that coincide with each larval molt (Sulston and Horvitz, 1977) (Table 1). rr31 mutants display an additional cell division after the 8E stage during embryogenesis, giving rise to 16 intestinal cells at this time instead of the wild-type 8E cells (Fig. 2). All 16 of these cells divide afterwards, as in wild-type animals, giving rise to 32 cells. The final number of intestinal cells at hatching (38±3) suggests that, as in wild type, only a subset of intestinal cells undergo a final mitotic division (in wild type, this results in 20 cells being formed from 16, while in rr31 mutants, the number increases from 32 to 38±3). The increase in the number of intestinal nuclei during postembryonic development in rr31 mutants (from 38±3 to 57±4) indicates that the L1-specific nuclear divisions also occur in rr31 mutants. Finally, the series of endocycles that occur following each larval molt also seem to be unaffected in rr31 mutants.

In cki-1(RNAi) animals, a similar supernumerary intestinal cell division occurs, but instead it occurs later, after four rounds of division, in cells that should normally have ceased dividing (Fig. 2). The difference in the timing of the lineage defect observed in rr31 and cki-1(RNAi) suggests that these two genes do not act in the same pathway that controls embryonic intestinal cell divisions.

To further strengthen this, cki-1 (RNAi) was performed in the rr31 genetic background. If these genes function in a common pathway, one would expect to observe some epistasis; however, if they act in parallel pathways, some enhancement should be apparent. Although both of the mutants had increased intestinal cell numbers at hatching [38±3 for rr31 mutants, and 29±3 for cki-1(RNAi) animals], the double mutant rr31; cki-1(RNAi) showed an increase in the number of intestinal nuclei at hatching compared with the single mutants (45±7), suggesting that the rr31 and cki-1 function in parallel pathways. Interestingly, the total number of intestinal cells at the adult stage was not significantly different in the single and double mutants [58±7 in rr31; cki-1(RNAi) animals and 57±4 in rr31 mutants] (Table 1), implying the presence of downstream components limiting the proliferative capacity of intestinal cells, which are common to both cki-1 and rr31.

To understand how rr31 functions at the molecular level, we mapped the mutant and then used a novel positional cloning strategy to characterize the rr31 allele molecularly. Genetic analysis showed that the rr31 mutation segregated in a dominant, maternal-effect manner. All the F₁ progeny of a hermaphrodite heterozygous for the rr31 mutation displayed the extra intestinal cell phenotype, including the homozygous +/+ larvae (Table 2), whereas when homozygous rr31 males were crossed into N2 hermaphrodites, none of the F₁ progeny had extra intestinal cells. To determine whether the dominant rr31 mutation was due to a gain-of-function mutation, or a loss of function in a haploinsufficient gene, we analyzed the effects of rr31 when hemizygous with either of two deficiencies that uncover this region (hDf8 and qDf16). Progeny of +/Df hemizygotes showed no evidence of extra intestinal cell divisions, whereas the progeny of rr31/+ heterozygotes were all affected, indicating that rr31 is not a loss-of-function mutation in a haploinsufficient gene. Furthermore, in the progeny of animals hemizygous for rr31 and qDf16 or hDf8, the extra intestinal cell phenotype was still present and fully penetrant. From these results, we conclude that the rr31 mutation is a dominant, gain-of-function mutation.

As the rr31 mutation is a dominant gain-of-function mutation, it was impossible to clone the gene using standard transformation rescue techniques. To circumvent this problem, we attempted to phenocopy the extra intestinal cell phenotype by injecting wild-type animals with PCR-amplified genomic regions from the rr31 mutant that corresponded to the predicted genes within the genetic interval where rr31 mapped. The injection of a 7.4 kb fragment corresponding to the cdc-25.1 gene resulted in the formation of extra intestinal cells in the transformed F₁ progeny, while other candidates had no effect (Table 3). This phenotype was incompletely penetrant and did not persist in subsequent generations, probably owing to a requirement for transgene expression in the germ line to provide maternally expressed products (Kelly et al., 1997).

Considering that the injection of the 7.4 kb PCR product amplified from rr31 mutant genomic DNA encoding the cdc-25.1 gene phenocopied the rr31 gain-of-function phenotype, we predicted that a gain-of-function mutation in cdc-25.1 could be responsible for the intestinal phenotype in the rr31 mutant. We performed cdc-25.1(RNAi) to test whether the rr31 phenotype could be suppressed by removing all cdc-25.1 gene activity (Fire et al., 1998). The injection of cdc-25.1 dsRNA into wild-type animals carrying the elt-2::GFP intestinal specific promoter, produced a variably penetrant embryonic lethal phenotype as previously reported, as well as ‘escapers’, which were later sterile or not affected (Ashcroft et al., 1999). Most of the adult F₁cdc-25.1(RNAi) progeny possessed a wild-type number of intestinal nuclei. Alternatively, when cdc-25.1 dsRNA was injected into rr31 mutant animals, the resulting F₁ progeny showed a marked reduction in the number of intestinal nuclei and the final intestinal cell count approached the wild-type complement of intestinal cells (Table 4). Progeny of uninjected rr31 animals showed no decrease in the number of intestinal nuclei. This indicated that cdc-25.1 is absolutely required for the extra intestinal cell divisions characteristic of the rr31 phenotype.

To verify whether the extra intestinal cell phenotype was indeed due to a mutation in cdc-25.1, we analyzed the genomic and cDNA sequence of the mutant and wild-type cdc-25.1 genes (Fig. 3). A single GC to AT transition was detected at the first nucleotide position of exon 2 in both rr31 genomic DNA and in the mutant cDNA, resulting in a G to D substitution at amino acid 47 in the N-terminal region of the protein (Fig. 3C). Initial structural predictions of the mutant CDC-25.1 protein imply that this substitution imparts a more flexible loop domain adjacent to a region of the polypeptide chain, which is strongly predicted to form a buried alpha helix.

To further understand the mechanism of action of the cdc-25.1(gf), we examined the RNAi phenotypes of potential candidate cell cycle regulators that may play an important role in the generation of extra intestinal cells during embryogenesis in cdc-25.1(gf) mutants. As the mammalian homolog of cdc-25.1, Cdc25A, is presumed to accelerate G₁/S by dephosphorylating CDK2, effectors that modulate CDK2 would be good candidates to investigate (Blomberg and Hoffmann, 1999). The mammalian cyclin E plays such a role through its association with CDK2, which, when activated through this association, triggers the initiation of S-phase (Tsai et al., 1993). Removal of the C. elegans cyclin E homolog by RNAi of the cye-1 gene causes embryonic lethality at the 100-cell stage in C. elegans (Fay and Han, 2000). When we performed RNAi with cye-1 dsRNA, 15-50% embryonic lethality was observed, while most of the other animals ‘escaped’ but arrested shortly after hatching. Because the cdc-25.1(gf) cell cycle defect occurs after the time of the terminal embryonic phenotype of cye-1(RNAi), we examined these escapers for suppression of the extra intestinal cell defects following cye-1(RNAi). cye-1(RNAi) animals had 20 intestinal cells on average at hatching, while when cye-1 function was removed in cdc-25.1(gf) animals, the extra intestinal cell phenotype was suppressed from 38 (±3) in cdc-25.1(gf) mutants alone, to 20 (±5) in cdc-25.1(gf); cye-1(RNAi) animals (Table 5). No effect on the suppression of the extra cell division could be detected following removal of cyclin D by RNAi of the cyd-1 gene in cdc-25.1(gf) mutants, despite a larval arrest phenotype and the inability to undergo postembryonic intestinal cell divisions (Table 5) (Park and Krause, 1999). This indicates that the positive cell cycle regulator cye-1 is required for the formation of extra intestinal cells in cdc-25.1(gf) mutants, while cyd-1 is not.

Although cdc25 genes have been shown to be important general cell cycle regulators important for the G₁/S or G₂/M transition, the cdc-25.1(gf) seems to only confer the ability to undergo an additional round of division to the intestinal cell lineage. This lineage restriction could be due to the presence of a factor in the E lineage that predisposes these cells to cdc-25.1(gf), or perhaps the lack of an activity present in other cells, which blocks such an effect.

The entire C. elegans endoderm is derived from one single blastomere, E, at the eight-cell stage of embryogenesis. The E-cell fate is specified through maternally provided factors, which are asymmetrically localized within the early embryo. These factors induce the E-cell fate through cell-cell interactions that are mediated mainly by the Wnt signaling pathway (Thorpe et al., 1997; Rocheleau et al., 1997).

To ascertain whether the cdc-25.1(gf) effect on the E lineage is dependent on Wnt signaling and/or subsequent E specification, or whether it may be due to other signals from surrounding blastomeres, we blocked Wnt signaling using a mom-2 background, which undergoes an E to MS cell fate transformation. (Thorpe et al., 1997; Rocheleau et al., 1997). If the E-to-MS transformed cell still overproliferates in mom-2;cdc-25.1(gf), then the cdc-25.1(gf) defect could be considered independent of E-cell fate specification by Wnt signaling and as such, more mesodermal cells should be present in mom-2;cdc-25.1(gf) compared with mom-2 single mutants. If this defect depends on Wnt signals and/or E specification, then the mom-2;cdc-25.1(gf) mutant should show the same number of (MS) mesodermal precursors as the mom-2 mutant alone. We found that the mom-2;cdc-25.1(gf) mutants did not form endoderm and produced the same number of mesodermal precursor cells as the mom-2 single mutants (Fig. 4). This suggests that a cell must be specified as endodermal (E) through Wnt-signaling to be sensitive to cdc-25.1(gf).

It is therefore plausible that the extra cell division in the E lineage is exclusively due to a cell-autonomous effect in cells of the E lineage. To confirm this prediction, we performed a reciprocal experiment with pop-1(RNAi) in the cdc-25.1(gf) mutant. pop-1 mutants have an MS to E transformation and produce extra intestinal cells at the expense of mesoderm (Lin et al., 1995). cdc-25.1(gf);pop-1(RNAi) embryos demonstrated a twofold increase in the number of intestinal cells compared with pop-1(RNAi) embryos alone (Fig. 5). This indicates that the MS blastomere, which was transformed to ‘E’ in pop-1 (RNAi) animals, is now also predisposed to cdc-25.1(gf) and, as a result, undergoes an additional round of division similar to its neighboring endogenous E blastomere.

Abnormal cell division timing can cause apparent cell fate transformations, and other blastomere fate transformations can occur under specific genetic circumstances giving rise to ‘ectopic’ E cells (Ambros, 1999; Maduro et al., 2001; Mello et al., 1992). To further confirm the cell-autonomous effect of cdc-25.1(gf), we ablated the E blastomere in five embryos immediately following its formation after division of EMS in cdc-25.1(gf) mutants. In wild-type animals, the ablation of E results in embryos that arrest embryonic development with no endoderm. In Fig. 5E we show that early E blastomere ablation in cdc-25.1(gf) animals results in the complete absence of intestinal cells, indicating that the extra intestinal cells in cdc-25.1(gf) animals result exclusively from the additional cell divisions of the E cell descendants.

The mammalian Cdc25 homologs function as dual-specificity phosphatases at different points in the cell cycle. Cdc25A plays a role at the G1/S transition, whereas, Cdc25B and Cdc25C promote the G2/M transition (Nilsson and Hoffmann, 2000). To determine whether cdc-25.1 acts at G₁/S or G₂/M, we ectopically expressed mutant or wild-type cdc-25.1 under the control of the heat-shock promoter in adult worms carrying the rnr::GFP or cye-1::GFP reporter constructs. Both rnr::GFP and cye-1::GFP are expressed strongly in cells which are entering S phase (Hong et al., 1998) (M. Krause, personal communication). In these animals, we assayed the reporter gene expression in order to see whether cdc-25.1 was able to induce S-phase entry in cells that should have normally ceased division. Overexpression of mutant or wild-type cdc-25.1 caused adult intestinal cells to enter S phase, but did not cause any apparent lineage or morphological abnormalities in other tissues when animals were heat-shocked during larval or adult stages. Heat-shock alone had no effect on reporter expression (Fig. 6). We conclude that cdc-25.1 can induce S phase in intestinal cells and thus acts as a positive regulator of the G₁/S transition. No divisions were observed in these cells.

To test whether there are any differences in localization of the CDC-25.1 wild-type or gain-of-function protein, which could provide insight into the mutant phenotype, we performed anti-CDC-25.1 antibody staining in wild-type or cdc-25.1(gf) embryos (a kind gift from Andy Golden and Neville Ashcroft). The wild-type CDC-25.1 protein product localized to oocytes, cortical membranes and ubiquitously in all nuclei of embryos up to the 28-cell stage (2E), as previously described (Ashcroft et al., 1999). After the 28-cell stage, there is no detectable staining in wild-type animals, as can be observed in Fig. 7A. In cdc-25.1(gf) embryos antibody staining was identical to wild type up to the 28-cell embryonic stage. After this point, we were able to detect nuclear CDC-25.1 staining up until the 100-cell stage (Fig. 7B,C). This suggests that the CDC-25.1(gf) protein perdures abnormally and may not be properly degraded in cdc-25.1(gf) mutants.

We have identified and characterized a mutant (rr31) that has increased numbers of intestinal cells, similar to cki-1(RNAi) animals. We mapped the mutation and through three independent methods (phenocopy, RNAi, sequence analysis) we demonstrated that the mutation that causes this defect occurs in cdc-25.1. From genetic analysis, we conclude that rr31 is a novel maternal effect, dominant, gain-of-function allele of this gene. The Cdc25 phosphatases are important regulators of the cell cycle and act as potential oncogenes that act downstream of the Ras and Myc oncogenes, particularly because of their role in activating Cdks (Galaktionov et al., 1995a; Galaktionov et al., 1995b; Galaktionov et al., 1996). In addition, Cdc25 phosphatases are principal players in the DNA damage and DNA replication checkpoints (Lopez-Girona et al., 1999; Mailand et al., 2000; Falck et al., 2001).

The C. elegans homolog of Cdc25A, cdc-25.1, belongs to a family of four cdc25 homologs in C. elegans, and plays an important role in the proper progression of meiosis prior to embryogenesis (Ashcroft et al., 1998; Ashcroft et al., 1999). Both mice and humans have three homologues Cdc25A, Cdc25B and Cdc25C, each of which show different spatial and temporal expression patterns (Wu and Wolgemuth, 1995; Hernandez et al., 2000; Hernandez et al., 2001). This may also be true for the cdc25 genes in C. elegans, suggesting a tissue-specific function for each of these cell cycle regulators (Ashcroft et al., 1998; Ashcroft et al., 1999).

The cdc-25.1(gf) mutation seems to cause cell division defects uniquely in the intestinal cell lineage, without an apparent effect on any other cell types examined, unlike cki-1(RNAi) animals, which display a diverse array of postembryonic cell division defects (Hong et al., 1998). Because cdc-25.1(gf) and cki-1(RNAi) display their respective defects at different stages of embryogenesis, we believe that they do not function in the same pathway.

The cdc-25.1(gf) allele segregates in a manner consistent with it being a maternal-effect, dominant mutation. As previously mentioned, the CDC-25.1 protein product is localized to all nuclei of embryos up to the 28-cell stage (Ashcroft et al., 1999). The finding that the CDC-25.1(gf) protein is present in nuclei of cdc-25.1(gf) embryos at later stages of development than in wild-type embryos, suggests that the mutant protein is able to perdure for a longer time. This would explain how the extra intestinal cell defect in cdc-25.1(gf) can occur much later in embryogenesis than when the wild-type protein is normally expressed. It is therefore possible that the point mutation in CDC-25.1 affects the stability of the protein.

Our genetic data supports the hypothesis that the gain-of-function mutation in cdc-25.1 probably does not give rise to a dominant negative product by antagonizing wild-type CDC-25.1 function. The highly conserved catalytic region of CDC-25.1 is located at the C terminus, whereas the less-conserved N-terminal domain plays a regulatory function, although little is known about how it imparts such control (Fauman et al., 1998). It has been shown that the phosphatase activity of the CDC25 family of proteins is regulated by extensive phosphorylation in this domain of the protein (Strausfeld et al., 1994; Hoffmann et al., 1994; Kumagai and Dunphy, 1992). The G47D substitution in the N-terminal region could therefore confer a more favorable site for phosphorylation on surrounding residues in the region of the mutation. Alternatively, the G47D substitution might itself mimic or impede a regulatory phosphorylation event that normally occurs on residues in this vicinity, through the increased charge that is due to the novel acidic residue. Therefore, the gain-of-function phosphatase could potentially escape normal negative controls permitting it to perdure, thereby conferring an extended period of activity to dephosphorylate typical or atypical substrates (such as a different Cdks), to promote the extra round of embryonic cell division.

The analysis of the interaction with the G₁/S-positive cell cycle regulator cyclin E, cye-1 supports these possibilities. CDK2 is normally inactivated by phosphorylation on highly conserved threonine and tyrosine residues (Gu et al., 1992). At the G₁/S transition, the Cdc25A phosphatase dephosphorylates these conserved residues, thus activating CDK2. Cdc25A can also act as a target of the CDK2/Cyclin E complex at the G₁/S transition, creating a positive autoregulatory feedback loop (Hoffmann et al., 1994; Blomberg and Hoffmann, 1999). The reduction of cye-1 activity in cdc-25.1(gf) mutants suppressed the extra intestinal cell phenotype, suggesting that in cdc-25.1(gf) mutants, cye-1 is required for the extra cell division in the intestinal lineage and that cdc-25.1(gf) could act through positive regulators of the G₁/S transition.

Ectopic expression of Cdc25A accelerates the G₁/S transition and prematurely activates Cdk2 (Blomberg and Hoffmann, 1999). Consistent with this function, we have shown using the S-phase-specific reporters rnr::GFP and cye-1::GFP, that when overexpressed in adults, C. elegans cdc-25.1 is capable of inducing S-phase entry in intestinal cells, and therefore resembles the Cdc25A family of phosphatases. Extra intestinal (or other) cell divisions (mitoses) were not observed after overexpression of CDC-25.1, despite S-phase entry, suggesting that these cells are G₂/M blocked by the limited activity of positive regulators, such as CDK1, B-type cyclins or Cdc25 phosphatases (reviewed by Nigg, 2001).

Why the mutant CDC-25.1 protein is capable of causing additional cell divisions in the intestinal cell lineage, despite the fact that it should indiscriminately dephosphorylate and activate CDK2 in all cells of the embryo is still unclear. What makes endodermal cells competent to respond to this gain-of-function phosphatase, or what negative cell cycle regulator is not expressed specifically in the intestine? These are major questions that may be answered through genetic modifier screens that are currently under way in our laboratory.

Noteworthy of mention, the expression of the wee-1.1 kinase, which inhibits the activity of the G₂/M cyclin-dependent kinase CDK1, is specifically restricted to the E blastomere and AB progeny early in the embryo, and its expression is downregulated after the first division of E (Lundgren et al., 1991; Wilson et al., 1999). However, the removal of wee-1.1 kinase activity through RNAi does not result in aberrant divisions of the endodermal cells, probably due to redundancy, leaving its E-specific expression and function unclear (Wilson et al., 1999) (I. K., unpublished).

It does appear, however, that E specification through Wnt signaling makes cells susceptible to the cdc-25.1(gf) mutation, although at present we cannot discern whether this is a direct or indirect effect. It has been shown that in other systems Wnt does affect cell division through effects on Cdc25 (Johnston and Edgar, 1998; Rimerman et al., 2000).

We suggest that the early embryo contains a pool of maternally supplied CyclinE/CDK2 that is non-limiting for most of the early divisions; however, much of it may be inactive because of inhibitory phosphorylations on CDK2. In the cdc-25.1(gf) mutant, the continued presence of the mutant protein might render a small portion of this maternal Cyclin E/Cdk2 pool active at a specific window during the formation of the intestine, thereby causing an extra round of cell division. For example, such a window might reflect a maternal to zygotic transition for a negative Cdk regulator (such as wee-1). The divisions of other cell types, as well as further divisions of the E lineage might be dependent on zygotic expression of positive regulators, which could later become controlled by cki-1. This would explain why the early divisions of the E lineage are unaffected by the loss of cki-1 activity, while the later divisions are.

The proper control of E lineage divisions might be especially important as the cell division of endodermal precursors are blocked by the onset of morphogenetic movements typical of gastrulation, which begins at the 28-cell stage. In Drosophila, CDC25/String proteolysis has been shown to be important for the proper coordination of gastrulation and ingression of the mesoderm anlage (Mata et al., 2000; Grosshans and Wieschaus, 2000). A similar mechanism might be acting in the coordination of C. elegans endodermal divisions, whereby correct division timing, with specification and function, is essential for gastrulation and ensuing embryogenesis.

Unlike the early embryonic cell cycles in Drosophila, which are synchronous, the C. elegans early blastomeres demonstrate distinct and invariant cell division timing. These divisions are coordinated by maternally supplied factors, and zygotic transcription is not required for cell cycling until the 100 cell stage (Powell-Coffman et al., 1996; Edgar et al., 1994c). Little is known about these maternally controlled early embryonic cell divisions, nor have the important regulators that drive these divisions been identified, but our work stresses the importance of the proper control of these regulators to ensure the correct execution of cell divisions characteristic of each lineage.

The important finding that a mutation in a general cell cycle regulator can cause overproliferation in a specific tissue is not unique. The intestine in C. elegans and in other organisms seems very sensitive to changes in cell cycle regulators and their upstream regulators (Boxem et al., 2001; Smits et al., 1999). Understanding what sensitizes tissues to changes in cell cycle regulators will help us gain insight into how different cell types alter their cell cycle programs independently to impart increased tissue diversity and corresponding developmental potential.

We thank Andy Golden, Neville Ashcroft, Darcy Birse and Iain Johnstone for sharing reagents and discussion; anonymous reviewers for useful comments; Mike Krause for the cye::GFP strain; Sue Mango for the anti-PHA4 antibody; Jim McGhee for the elt-2::GFP constructs; and Joel Rothman for reagents. We also thank Victor Ambros, Beat Suter and Monique Zetka for comments on the manuscript. This work was supported by a grant to RR from the National Cancer Institute of Canada with funds from the Terry Fox Run.