We have identified six protein kinases that belong to the family of cdc2-related kinases in Caenorhabditis elegans. Results from RNA interference experiments indicate that at least one of these kinases is required for cell-cycle progression during meiosis and mitosis. This kinase, encoded by the ncc-1 gene, is closely related to human Cdk1/Cdc2, Cdk2 and Cdk3 and yeast CDC28/cdc2+. We addressed whether ncc-1 acts to promote passage through a single transition or multiple transitions in the cell cycle, analogous to Cdks in vertebrates or yeasts, respectively. We isolated five recessive ncc-1 mutations in a genetic screen for mutants that resemble larval arrested ncc-1(RNAi) animals. Our results indicate that maternal ncc-1 product is sufficient for embryogenesis, and that zygotic expression is required for cell divisions during larval development. Cells that form the postembryonic lineages in wild-type animals do not enter mitosis in ncc-1 mutants, as indicated by lack of chromosome condensation and nuclear envelope breakdown. However, progression through G1 and S phase appears unaffected, as revealed by expression of ribonucleotide reductase, incorporation of BrdU and DNA quantitation. Our results indicate that C. elegans uses multiple Cdks to regulate cell-cycle transitions and that ncc-1 is the C. elegans ortholog of Cdk1/Cdc2 in other metazoans, required for M phase in meiotic as well as mitotic cell cycles.
A fundamental question in biology is how cell division, growth and differentiation are coordinately regulated during development of multicellular organisms. This coordination likely involves connections between developmental signals and the cell-cycle machinery. Cyclin-dependent kinases (Cdks) have been identified as key components of the cell-cycle machinery in all eukaryotes studied (Forsburg and Nurse, 1991; Nigg, 1995; Norbury and Nurse, 1992). In yeasts, progression through the cell cycle is controlled by a single Cdk, encoded by the CDC28 and cdc2+ genes in Saccharomyces cerevisiae and Schizosaccharomyces pombe, respectively. The CDC28/cdc2 catalytic subunit requires association with a cyclin regulatory partner for kinase activity, and different cyclins are involved in progression through G1, S and M phase. Regulation of cell-cycle progression is more complex in multicellular organisms; a large family of kinases related to CDC28/cdc2 has been identified in metazoans and several of these kinases have been implicated in cell-cycle regulation (reviewed by Nigg, 1995). In addition, various cyclins are expressed during distinct phases of the cell cycle in metazoans. Thus, progression through the cell cycle in higher eukaryotes is thought to require successive activation of different Cdks, regulated in part by transient associations with distinct cyclins. Other levels of regulation include association with kinase inhibitors and activating as well as inactivating phosphorylations.
When vertebrate cells enter a division cycle, the first CDKs to become activated are Cdk4 and/or Cdk6 in association with D-type cyclins (Sherr, 1994, 1996). Late in G1, Cdk2 is activated and can be found associated first with cyclin E and subsequently cyclin A (Tsai et al., 1993a). Kinase activity of Cdk1/Cdc2 in combination with A- and B-type cyclins peaks at the G2/M transition (Draetta and Beach, 1988; Tsai et al., 1993a). Several lines of evidence indicate that these Cdks regulate specific cell-cycle transitions. Extracellular signals, such as growth factors, influence expression levels of D-type cyclins and association with Cdk4 and Cdk6 (Matsushime et al., 1994, 1991). Ectopic expression of cyclin D, as well as cyclin E, shortens G1 phase and accelerates entry into S phase (Ohtsubo and Roberts, 1993; Quelle et al., 1993). Inhibition of either cyclin A activity or Cdk2 activity inhibits DNA replication (Fang and Newport, 1991; Girard et al., 1991; Pagano et al., 1993; Tsai et al., 1993a; van den Heuvel and Harlow, 1993), while inhibition of Cdk1/Cdc2 prevents entry into mitosis (Fang and Newport, 1991; Riabowol et al., 1989; Th’ng et al., 1990; van den Heuvel and Harlow, 1993). The function of several other cdc2-related kinases is currently either unknown or, for Cdk5, appears unrelated to cell-cycle progression (Chae et al., 1997; Gilmore et al., 1998; Nikolic et al., 1996). Thus far, genetic analysis of Cdk loss-of-function mutations has been limited in multicellular organisms. A family of cdc2-related kinases has been identified in Drosophila and one of these kinases, Dmcdc2, has been implicated in regulation of mitosis and inhibition of DNA replication (Hayashi, 1996; Sauer et al., 1996; Stern et al., 1993). A second kinase, Dmcdc2c, appears active during S phase (Lehner and O’Farrell, 1990).
The nematode C. elegans is uniquely suited as an animal model to study developmental control of cell division. The transparency of these animals allows monitoring of cell divisions in living animals. In addition, the near invariance of the cell lineage has allowed a precise description of the time of division for every somatic cell (Sulston and Horvitz, 1977; Sulston et al., 1983). Combined with effective genetics, these characteristics allow efficient isolation and detailed examination of cell-cycle mutants.
Genetic studies have already revealed several genetic pathways that regulate postembryonic cell divisions. For instance, pathways that involve TGF-β and insulin-related signaling cascades control entry into and exit from a developmentally arrested dauer stage (Massague, 1998; Riddle and Albert, 1997). In addition, a pathway of heterochronic genes controls the timing of cell division in several postembryonic lineages (reviewed by Slack and Ruvkun, 1997). Furthermore, a Notch-like signaling pathway that includes the lag-2 ligand and glp-1 receptor controls the switch between mitotic versus meiotic division in the germline (Francis et al., 1995; Kadyk and Kimble, 1998; Kimble and Simpson, 1997). The targets of such developmental signaling pathways likely include components of the general cell-cycle machinery, which may be regulated through intermediates such as cki-1, a developmentally regulated Cdk inhibitor of the CIP/KIP family (Hong et al., 1998).
To be able to explore connections between developmental signals and the cell-cycle machinery, we have initiated analysis of the cyclin-dependent kinases present in C. elegans. We have identified six putative kinases that share extensive homology with cdc2-related kinases in higher eukaryotes. RNA-interference of only one of these kinases caused complete arrest of cell division. A cDNA clone of this kinase, named ncc-1 for nematode cell cycle, was previously identified and shown to complement the CDC28-1N mutation in S. cerevisiae (Mori et al., 1994). Here we show that ncc-1 is required for progression through M phase in meiotic and mitotic cell cycles, but not for G1 and S phase progression. Thus, ncc-1 performs a role similar to Cdk1/Cdc2 in higher eukaryotes.
MATERIALS AND METHODS
Culture conditions and strains
Worm strains used were derived from the wild-type Bristol strain N2 and Bergerac strain RW7000. Worms were cultured using standard techniques as described by Brenner (1974). We used the following mutations, descriptions of which can be found in Riddle et al. (1997) or cited references:
LGIII: dpy-17(e164), unc-36(e251), dpy-19(e1259), unc-32(e189), glp-1(q158), ncc-1(he5, he6, he24, he25 and n3064) (this paper), unc-47(e307), unc-49(e382), tra-1(e1099) and dpy-18(e364). Rearrangements: qDp3(III,f) (Austin and Kimble, 1987), ctDp6(III,f) (Hunter and Wood, 1990), nDf40 (Graham and Kimble, 1993), qC1 dpy-19(e1259) glp-1(q339) (Edgley et al., 1995). Expression of an rnr::GFP reporter was examined in strains VT774 unc36(e251); maIs103[rnr::GFP unc-36(+)] (Hong et al., 1998) and SV61 ncc-1(n3064)/dpy-19(e1259) unc-47(e307); maIs103[rnr::GFP unc-36(+)].
Cloning of cdc2-related genes
To clone cdc2-related genes from C. elegans, we used RT-PCR amplification with degenerate oligonucleotide primers, low stringency hybridization of a cDNA library and antibody screening of an expression library. For the reverse transcriptase PCR approach, RNA was isolated from mixed stage C. elegans cultures grown in bulk (Wood, 1988). First strand cDNA was prepared from total RNA and amplified by PCR, using degenerate oligonucleotides derived from the highly conserved motifs EKIGEGTY and EGVPSTA as 5′ primers and DLKPQNL and WYRSPEV as 3′ primers (primer sequences available upon request). PCR fragments were cloned into pBluescript II SK (Stratagene) and sequenced by the dideoxy method. Partial clones were obtained this way from ncc-1, cdk5 and pct-1. To obtain full length cDNAs, a nematode cDNA library cloned in the Uni-ZAP vector (Stratagene) was probed with PCR fragments labeled with [α-32P]dCTP. Although low stringency conditions were used, we obtained only cDNA clones from ncc-1, cdk5 and pct-1. The same library was induced with IPTG and probed with a mouse monoclonal antibody recognizing the PSTAIRE motif (a kind gift from M. Yamashita). This resulted in identification of multiple cDNAs that were all derived from the ncc-1 gene.
We expect the predicted kinases to be full length based on the isolation of multiple cDNAs of similar size, homology in the N terminus with the human kinases and a nonsense codon preceding the ATG initiation codon in the pct-1 gene. The genomic localization was identified for each gene by probing YAC contigs, followed by PCR analysis of cosmids from the region. Cosmids T05G5, C07G1 and R11C7 were found to contain the ncc-1, pct-1 and cdk5 genes, respectively, and were used to determine the gene structures by Southern blotting and DNA sequencing. Subsequently, all sequences have been confirmed by the C. elegans sequencing project.
Plasmids containing full length ncc-1, cdk-5 and pct-1 cDNAs were used for in vitro transcription. For B0285.1, a genomic fragment was amplified by PCR using primers 5′-agaattcacaatgcttgatcaaattggc and 5′-aaggatccagattgtttttcttttccg. For, K03E5.3 a genomic fragment was amplified using primers 5′-aatccgcaattttgatgactccag and 5′-caaccgatatcgctccaggtaagc. These fragments were cloned into pBluescript SK. A plasmid containing cDNA yk492e2, derived from H06A10.1, was kindly provided by M. Park, M. Krause and Y. Kohara. Templates were linearized and transcribed in vitro with T3 or T7 RNA polymerase. RNA was phenol extracted, ethanol precipitated and dissolved in TE (10 mM Tris-HCl pH 8.0, 0.1 mM EDTA) to a final concentration of 1 mg/ml. A 1:1 mixture of the two strands was injected into young adult N2 hermaphrodites as described (Fire et al., 1998).
Screening for ncc-1 alleles
N2 animals were mutagenized with 50 mM ethylmethanesulfonate (EMS) as described by Brenner (1974). Mutagenized L4 hermaphrodites were picked to 100 mm plates, 30 per plate, and allowed to produce self progeny at 25°C. F1 animals in the fourth larval stage were picked to 60 mm plates, four per plate, and their progeny were examined for mutants that resemble larval-arrested ncc-1(RNAi) animals. Mutants that were sterile, uncoordinated, slightly dumpy and had relatively large head regions were examined by Nomarski microscopy to determine defects in cell division. Candidate mutations were recovered from siblings and were mapped initially by PCR, making use of primers based on polymorphic Sequence-Tagged Sites in the RW7000 Bergerac strain (Williams et al., 1992). ncc-1 mutations mapped to chromosome III and were placed by standard three factor crosses (Wood, 1988) between dpy-19(e1259) and unc-47(e307) or dpy-17(e164) and tra-1(e1099), about 1.06 map units to the right of dpy-19 at position 0.89. In addition, all mutations failed to complement nDf40 and allele n3064.
Antibodies were used in the following dilutions: mouse monoclonal antibody DM1A against α-tubulin (Sigma) diluted 1:1000 for western blot analysis and 1:500 for immunostaining, rabbit polyclonal antibody against phosphorylated histone H3 (Upstate Biotechnology) diluted 1:400, mouse monoclonal antibody BU-33 against BrdU (Sigma) diluted 1:60, mouse monoclonal antibody against the PSTAIRE region of Cdc2 (a kind gift from M. Yamashita), used 1:1000 diluted for immunoblotting and 1:250 diluted for immunostaining. A polyclonal antiserum raised against the C terminus of NCC-1 was diluted 1:1000 for immunostaining (a kind gift from J. Schumacher and A. Golden). Donkey-anti-mouse and donkey-anti-rabbit IgG secondary antibodies, rhodamine- or FITC-conjugated, were from Jackson labs and diluted 1:200.
Western blot analysis and Immunostaining
To obtain protein samples, synchronized wild-type animals were washed 4 times in M9 buffer and dissolved by boiling for 5 minutes in 1× Laemmli protein loading buffer (Harlow and Lane, 1988). Protein samples were separated by SDS-PAGE followed by coomassie staining, to determine protein concentration, or immunoblotted using standard procedures (Harlow and Lane, 1988).
For antibody staining of germlines and embryos, gravid wild-type hermaphrodites were dissected in a drop of M9 buffer on a gelatin-coated slide. A coverslip was applied and the slide frozen on dry ice. After 10 minutes the coverslip was cracked off and slides were incubated in methanol at −20°C for 20 minutes and acetone at −20°C for 20 minutes. For immunostaining of larvae, animals were fixed in Bouin’s solution and antibody staining was performed according to standard procedure (Harlow and Lane, 1988; Nonet et al., 1997). For staining of DNA only, animals were fixed overnight in Carnoys solution (60% ethanol, 30% glacial acetic acid, 10% chloroform) and slowly rehydrated in PBS + 0.1% Tween-20. In all procedures, DNA was stained with either propidium iodide (PI) 1 mg/ml, following treatment with 200 μg/ml RNAse A for 30 minutes at 37°C, or with 1mg/ml 4,6-diamidino-2-phenylindole (DAPI) (Sigma). Samples were mounted on slides in 10% PBS, 2.3% Dabco, glycerol.
Images were obtained using a Sensys cooled CCD camera (Photometrics, Tucson, AZ) and a Power Macintosh computer. Image analysis and computational deconvolution was performed with Openlabs software version 1.7.8 (Improvision, Boston, MA). Images were pseudocolored and merged using Adobe Photoshop.
Synchronized cultures were obtained from embryos isolated by hypochlorite treatment and hatched in the absence of food (Wood, 1988). The resulting early L1 arrested larvae were transferred to plates with bacteria, which triggers highly synchronized development. DNA synthesis was visualized by incorporation of BrdU. A solution of BrdU in S-medium was added to plates with synchronized larvae to a final concentration of 0.5 mg/ml. Animals were fixed in Bouin’s fixative and stained with anti-BrdU antibodies. The staining procedure was as described above with one addition; prior to preadsorbtion, DNA was denatured by incubating the fixed animals for 2 hours at room temperature in a 1:1 mixture of 4 M HCl and PBS + 1% BSA + 0.5% Triton X-100.
To quantitate DNA content, series of 32 Z-sections were taken of propidium iodide-stained animals with a confocal scanning laser microscope (Leica). Total pixel intensities were calculated for each section on a Macintosh computer using the public domain NIH Image program (developed at the U.S. National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image/). Pixel intensities of all 32 sections were added and recalculated to DNA content, using body wall muscle nuclei as a 2n DNA standard.
Identification of Cdc2 family members in C. elegans
Using several molecular approaches, we obtained cDNA clones from three C. elegans cdc2-related kinases (Materials and Methods and Fig. 1A). The kinase most closely related to human CDK1 (64% amino acid identity) was identified previously and named ncc-1 (Mori et al., 1994). The predicted NCC-1 protein of 332 amino acids diverges from other Cdk1 and Cdk2 kinases at two striking positions: it has an N-terminal extension of 18 amino acids and contains an Ile-to-Val substitution in the highly conserved PSTAIRE region of the protein that is involved in cyclin binding (Jeffrey et al., 1995). A second kinase gene encodes a member of the PCTAIRE subfamily and was therefore named pct-1. Within the catalytic core, the predicted PCT-1 protein of 577 amino acids shares ∼80% identity with any one of three human PCTAIRE kinases. In addition, PCT-1 has N-terminal and C-terminal extensions that are less conserved but similar in size to the human kinases. A third C. elegans kinase is a close homolog of human Cdk5 (74% amino-acid identity) and was named cdk-5. For each of these three kinases we sequenced the longest cDNAs and determined the genomic structure (Fig. 1A).
Three additional kinases that are more distantly related to Cdc2 were identified in BLAST homology searches (Table 1). The predicted gene H06A10.1 encodes a kinase that is most related to Cdk4 and Cdk6. K03E5.3 encodes a kinase that is 42%, 43% and 41% identical to human Cdk1, Cdk2 and Cdk3, respectively. Finally, the predicted product of B0285.1 shares 47% amino acid identity with human CHED kinase, and 41% with Cdk2 and Cdk3. As the sequencing of the C. elegans genome has been completed (The C. elegans Sequencing Consortium, 1998), we can conclude that these are the C. elegans kinases that are most closely related to members of the Cdk1/2/3, Cdk5, Cdk4/6 and PCTAIRE subfamilies of cdc2-related kinases.
ncc-1 is a candidate cell-cycle regulator
We used RNA interference (RNAi) to determine which of these kinases may be involved in regulation of the C. elegans cell cycle. RNAi has been shown to cause specific loss-of-function phenotypes for many different genes in C. elegans, although the mechanism of action is not understood (Guo and Kemphues, 1996; Tabara et al., 1998). Double-stranded RNA (dsRNA) is at least one order of magnitude more potent in interfering activity than either single strand (Fire et al., 1998). We transcribed both strands of each of the six kinases in vitro and injected dsRNAs into wild-type hermaphrodites. Injection of ncc-1 dsRNA caused a completely penetrant embryonic arrest at the one cell stage in all progeny produced after 12 hours post-injection. In addition, some of the earlier progeny arrested during larval development (31 larval mutants from 10 injected animals). These arrested larvae were uncoordinated, sterile, slightly dumpy and had disproportionally large head regions. Examination by Nomarski microscopy revealed that no or very few cells in the postembryonic lineages had divided. Most striking was the absence of any division of the blast cells Z1 to Z4 that normally form the somatic gonad and germ-line. These gonad precursor cells divide multiple times in several C. elegans cell-cycle mutants, including lin-5 and lin-6 mutants (our unpublished observations; Sulston and Horvitz, 1981). Both the embryonic and L1-arrest phenotypes implicated ncc-1 as an important cell-cycle regulator.
RNAi for two Cdc2 related genes, cdk-5 and pct-1, did not result in any apparent phenotypes (data not shown). RNAi for each of the remaining three cdc2-related genes caused obvious developmental defects but not a strict arrest of cell division. A clear but highly variable effect was observed following dsRNA injection of K03E5.3: the progeny arrested either as embryos (12%, n=69), early larvae (17%, n=98), late larvae (9%, n=50), sterile adults (39%, n=228) or developed to wild-type adults (24%, n=132). Progeny from B0285.1 dsRNA-injected animals arrested late in the third larval stage (L3) or early L4 stage. Most cell divisions have been completed at this stage with the exception of germ-line divisions. Finally, H06A10.1 RNAi resulted in slightly abnormal progeny but no apparent cell-cycle defects. Variable defects in cell division have been observed to result from H06A10.1 RNAi by M. Park and M. Krause (personal communication). Redundancy between different kinases might mask individual gene functions. However, injection of a mix of dsRNAs from K03E5.3, F18H3.5, B0285.1 and cdk-5 did not cause stronger defects: the phenotype observed closely mimicked that of K03E5.3 RNAi. Thus, based on these RNAi results, ncc-1 is the best candidate for a general cell-cycle kinase in C. elegans.
NCC-1 is closely related to yeast CDC28/cdc2 as well as mammalian Cdk1 and Cdk2. Therefore, ncc-1 could act analogously to CDC28/cdc2+ in yeasts and regulate multiple transitions in the cell cycle. Alternatively, ncc-1 could play a role similar to one of the human kinases, regulating either G1/S progression or mitosis. To distinguish between these possibilities, we determined the requirements of ncc-1 in the C. elegans cell cycle.
NCC-1 protein expression correlates with cell division and is abolished by RNAi
If ncc-1 is required for cell division in general, NCC-1 protein is expected to be present in dividing cells. NCC-1 protein levels were found to correlate with the number of cell divisions during various stages of development, as was shown by western blot analysis (Fig. 1B). The highest NCC-1 levels were detected in embryos and the lowest levels in L1 larvae that were arrested by food deprivation. Upon release from L1 arrest, and resumption of cell division, NCC-1 protein levels were found to increase.
We used two different antibodies to detect the localization of NCC-1: a monoclonal antibody that recognizes the PSTAIRE peptide and an antiserum raised against a C-terminal NCC-1 peptide. Identical staining patterns were obtained with both antibodies. NCC-1 was found in the nucleus of oocytes and embryonic cells and, in addition, diffusely localized in the cytoplasm (Fig. 2A-F). During growth of oocytes in the proximal gonad, NCC-1 levels gradually increase (Fig. 2A,H). Following nuclear membrane breakdown, staining was diffuse in mitotic cells (Fig. 2D). NCC-1 appeared excluded from the DNA in metaphase and, to a lesser extent, in anaphase cells. NCC-1 was not detectable in oocytes or embryos of RNAi-treated animals (Fig. 2G,I). This result confirms the staining specificity and supports that ncc-1(RNAi) results in strongly reduced NCC-1 protein levels.
ncc-1 is required for meiotic maturation
To determine the cause of arrest in ncc-1(RNAi) embryos, we analyzed gametogenesis and early embryogenesis in the injected hermaphrodites. In wild-type animals, germ-precursor nuclei are produced by mitotic divisions in the distal ends of the syncytial gonad (Kimble and White, 1981). As these nuclei move further from the distal tip cell, they exit the mitotic cycle, initiate meiosis and progress through pachytene of meiotic prophase I. Oocyte formation begins when germ nuclei reach the flexure in the U-shaped gonad; they exit from pachytene, cellularize, continue to enlarge and progress to diakinesis of Meiosis I. Shortly before fertilization, meiotic maturation and nuclear envelope degradation are initiated. Mature oocytes ovulate by entering the spermatheca at the proximal end of the gonad. Meiotic maturation is completed after fertilization and two polar bodies are expelled from the fertilized egg. The maternal pronucleus subsequently migrates towards the paternal pronucleus, while the egg undergoes pseudocleavage. The two pronuclei meet in the posterior of the egg, migrate to the middle and initiate a first mitotic division.
We examined the consequences of ncc-1 dsRNA injection on these maturation and fertilization events in live animals by Nomarski microscopy. Two types of abnormalities were apparent: at the distal ends of the gonad fewer germ cells were produced (see below) and at the proximal ends many of the events that follow ovulation and fertilization of oocytes were aberrant (Table 2). Gonads were examined at various time points following dsRNA injection by Nomarski microscopy of living animals and fluorescence microscopy of fixed worms stained with DAPI for analysis of DNA organization. Formation and growth of oocytes was normal. The morphology and progression from pachytene to late stage oocytes, with six fully condensed bivalents in typical diakinesis arrangement, was also normal (Fig. 2I). Based on the position of the most distal oocyte with fully condensed chromosomes, relative to the flexure in the gonad, progress through pachytene appeared slightly delayed in about half of the injected animals (data not shown).
The first apparent defect was abnormal breakdown of the nuclear membrane in the mature oocyte, which required about 20 minutes to complete instead of 5 minutes in the wild type. Cytological analysis demonstrated that the transition from diakinesis to metaphase of meiosis I failed to occur as chromosomes did not congress to form a metaphase plate. No polar bodies were observed consistent with a complete failure of the meiotic divisions (Fig. 2G). However, fertilization did trigger formation of an eggshell. The maternal pronucleus enlarged and became about twice the wild-type size. Migration of the maternal pronucleus toward the paternal pronucleus was initiated but meeting of the nuclei was delayed and pseudocleavage did not occur (Table 2). The chromosomes in the paternal pronucleus failed to decondense and this nucleus remained small (Fig. 2G). Meeting of the pronuclei was followed by stable arrest: the nuclear membranes did not break down and mitosis did not occur. By immunostaining with tubulin-specific antibodies, we detected a bipolar spindle around the maternal nucleus in about half of the embryos (data not shown). We cannot exclude that formation of this spindle is dependent on residual activity of ncc-1 in the ncc-1(RNAi) embryos (see Discussion).
The earliest defect seen in ncc-1(RNAi) animals is lack of meiotic maturation. Thus, these results suggest that ncc-1 is required to promote meiotic maturation. The subsequent abnormalities may be secondary consequences of the meiotic maturation defect.
ncc-1 is required for mitotic divisions in the germ-line
We noticed that ncc-1 dsRNA-injected animals produced fewer oocytes than wild-type hermaphrodites. Upon closer examination, the number of germ cells was found to decrease gradually over time in ncc-1 RNA-injected animals. This reduction was first apparent in the mitotic region of the gonad arms, starting at around 12 hours after injection, at 20°C. The gonad was essentially empty by day three to four. During the same period, control injected animals did not show a noticeable change in the number of germ cells or general appearance of the gonad.
The germ precursor nuclei that were present in the distal region at 20 hours after injection were enlarged (Fig. 3B). Mitotic cells were not observed in gonads fixed and stained with DAPI. As metaphase cells are not always unambiguously detected in wild-type gonads, we also examined the presence of a mitotic phosphorylated epitope. Chromosome condensation in mitosis has been described to coincide with phosphorylation of histone H3 on Ser-10, which forms part of a consensus Cdc2-phosphorylation site (Ajiro et al., 1996; Hendzel et al., 1997). A rabbit polyclonal antiserum raised against a phosphorylated H3 peptide also recognizes a mitotic phospho-epitope in C. elegans (Kadyk and Kimble, 1998; Lieb et al., 1998). We immunostained ncc-1 dsRNA-injected and control injected animals 20 hours after injection to examine the presence of the phosphorylated histone H3 epitope. Staining was absent in ncc-1 RNA-injected germlines whereas control injected animals stained similarly to wild type (Fig. 3C,D). NCC-1 could directly or indirectly cause phosphorylation of histone H3, or be required at an earlier step in germ cell division. In either case, when taken together, these observations indicate that ncc-1 is required for mitotic divisions in the distal part of the gonad.
Isolation of ncc-1 mutants based on the larval ncc-1 (RNAi) phenotype
To further characterize the role of ncc-1 in cell division, we isolated ncc-1 mutations in a genetic screen. The larval arrested ncc-1(RNAi) animals may have received functional ncc-1 maternal product yet fail zygotic expression. Based on this interpretation, we screened for uncoordinated, sterile mutants with disproportionally large head regions in the F2 progeny of ethylmethanesulfonate (EMS) treated animals, in order to isolate ncc-1 mutants.
In a semi-clonal screen of about 22000 haploid genomes, we obtained 21 independent recessive mutations that caused uncoordinated, sterile, large head phenotypes and defects in cell division during the first larval stage. Based on standard complementation assays and genetic mapping, 14 of these mutations likely affect the same gene. We characterized 4 of these mutations in detail. In addition, we also characterized n3064 which displays a very similar phenotype and was isolated in a previous screen for potential cell-cycle mutants (S. v.d.H. and H. R. Horvitz, unpublished results). All five of these mutations mapped in close proximity to ncc-1 and failed to complement nDf40, a deletion that removes the ncc-1 locus. In addition, DNA sequencing revealed mutations within the ncc-1 ORF in all five mutants (Fig. 4). The mutant alleles all contain G-C to A-T transitions, the type of mutation most commonly induced by EMS. The best candidate null allele is he5. This allele contains a deletion of 131bp, that removes the entire PSTAIRE domain, as well as a nonsense mutation that should cause a truncated product (Fig. 4). Allele he26 also contains a nonsense mutation and encodes a product that lacks the C-terminal 265 out of 332 kinase residues. The other three alleles, he24, n3064 and he25, contain missense mutations at positions 75, 191 and 214, respectively, replacing negatively charged glutamates by positively charged lysine residues. Each mutation affects a residue that is highly conserved within the Cdc2 family, in he24 and he25, or in all protein kinases, in n3064 (Fig. 4).
ncc-1 is required for postembryonic cell divisions
We characterized the nature of the ncc-1 alleles and cell-cycle phenotypes in more detail. As homozygous mutants are fully penetrant sterile, the ncc-1 mutations were maintained in heterozygous strains. Derived from heterozygous (ncc-1/+) mothers, ncc-1 mutant progeny completed embryogenesis, presumably as a result of maternal gene function, but displayed no cell divisions after hatching. The percentage mutant progeny was close to 25% for all mutations, confirming the recessive nature of the alleles (Table 3). At 25°C, homozygous mutants and trans-heterozygotes with nDf40 were phenotypically indistinguishable. Thus, all five mutations cause fully penetrant strong loss-of-function phenotypes at 25°C. Similar results were obtained at 15°C for all alleles except he25. Homozygous he25 mutants were of more wild-type appearance at 15°C, although shorter and fully penetrant sterile. Based on our genetic and molecular characterizations, 4 of the 5 alleles are strong loss-of-function or null mutations, whereas he25 causes incomplete loss of function at 15°C.
Nomarski microscopy observations indicated that all blast cells that give rise to the postembryonic lineages in wild-type animals are present in ncc-1 mutants. However, these cells did not initiate mitotic divisions, with the possible exception of a rare division of an intestinal nucleus. Examination of fixed and DNA-stained animals confirmed these findings (Table 3). We quantified the cell-cycle defects for each of the mutations in three postembryonic lineages, by examining the first divisions of the ventral cord precursor cells (P), the intestinal cells (I) and the precursor cells of the somatic gonad and germline (Z1, Z2, Z3 and Z4). In wild-type animals, these cells start to divide about halfway through L1 (P and Z1-4) or late in L1 stage (I). All five mutations caused very similar defects in cell division. Not a single division of cells in the gonad primordium or P1-10 was observed in 20 animals examined (Table 3). Several observations indicate that ncc-1 is directly required for cell division and that the lack of division was not caused by sickness or lethality. The ncc-1 mutants were viable and continued to grow to 80% of the wild-type size (the length of young adult ncc-1 and wild-type animals was 0.73±0.04 mm and 0.91±0.08 mm, respectively). In addition, the precursor cells of the ventral nerve cord migrated at the normal time from a more lateral to a ventral position (data not shown).
Requirement for ncc-1 is not restricted to the first postembryonic or L1 divisions, as was shown in he25ts animals. At 15°C, the P cells undergo extensive divisions during the L1 stage in he25ts mutants and the P3.p to P8.p (Pn.p) daughters divide in late L3 to form the vulva (Fig. 5C). All P cells failed to divide in he25 animals shifted to 25°C during the early L1 stage, thereby resembling ncc-1(0) animals (Table 3). In contrast, temperature shift during the L2 or L3 stages allowed the P cells to divide during L1 but prevented division of the Pn.p daughters (Fig. 5B).
These results are consistent with a role for ncc-1 as a general cell-cycle regulator, which is in agreement with the high degree of identity between NCC-1 and the Cdk1 and Cdk2 kinases in other species (Fig. 4).
ncc-1 is required for mitosis, but not for DNA replication
We observed that the precursor cells of the postembryonic lineages did not initiate chromosome condensation or nuclear envelope breakdown in ncc-1 mutants. Apparently, the cells arrest prior to entry into mitosis, yet this arrest could occur in G1, S or G2 phase. We examined whether cells in ncc-1 mutants initiate S phase to determine the point of arrest.
As a first assay, we used an S phase reporter that expresses the green fluorescent protein (GFP) under the control of a C. elegans ribonucleotide reductase (rnr) promoter. Expression of rnr tightly correlates with S phase in other eukaryotes (Fernandez-Sarabia and Fantes, 1990). In addition, rnr::GFP expression in transgenic worms that contain the integrated rnr::gfp reporter coincides with the previously reported time of S phase (Hong et al., 1998). We examined GFP expression in L1 larvae that were double homozygous for the ncc-1(n3064) mutation and integrated rnr::gfp. Wild-type animals and ncc-1 mutants were found to express rnr::GFP at the same time and in the same cell types (Fig. 6A). Thus, blast cells of the postembryonic lineages complete G1 and appear to become competent for S phase in ncc-1 mutants.
To directly address whether ncc-1 mutants synthesize DNA, we examined incorporation of the thymidine analog BrdU. Progeny from animals heterozygous for he5, he26 or n3064 were incubated with BrdU from 3 to 9 hours of L1 development at 25°C and subsequently stained with anti-BrdU antibodies. In wild-type animals as well as ncc-1 mutants, BrdU incorporation was detected in all cells that were previously reported to undergo S phase during this developmental phase (Hedgecock and White, 1985). For example, BrdU staining was detected in the ventral cord precursor cells W and P1-P12 (Fig. 6B), and these cells divided in the wild-type but not in ncc-1 mutants. Other cells that stained strongly with BrdU in wild-type as well as homozygous ncc-1 animals include the lateral ectoblasts H1,H2, V1-V6 and T and cells in the gonad primordium (data not shown). These results show that neither the timing of S-phase nor the specific cells that enter S-phase are affected by the ncc-1 mutations.
The complete genome is replicated in ncc-1 mutants
The results described above indicate that ncc-1 is strictly required for M phase and not for G1 progression or DNA replication. However, we wished to examine two alternative explanations. First, it remained possible that cells in ncc-1 mutants initiate DNA synthesis but fail during the elongation phase. Second, perdurance of maternal product could formally allow initiation of the first division cycle in the postembryonic lineages. To address these issues, we compared the amounts of DNA in intestinal nuclei of adult wild-type animals and ncc-1 mutants. Wild-type animals hatch with 20 diploid intestinal cells. In general, the 14 most posterior of these undergo one round of DNA synthesis followed by nuclear division in late L1 stage (Hedgecock and White, 1985; Sulston and Horvitz, 1977 and Table 2). In addition, during each larval stage all intestinal nuclei undergo S-phase without mitosis, leading to a 32n DNA content by the adult stage.
We fixed adult wild-type and ncc-1 animals and stained them with the DNA stain propidium iodide, and determined the DNA content of the intestinal cells by quantitative confocal analysis. Body wall muscles are diploid throughout postembryonic development in wild-type animals. Using body muscle nuclei as an internal 2n DNA standard, we found that the intestinal nuclei in ncc-1 mutants and wild-type animals were equally polyploid and contained close to 32n amounts of DNA (Figs 6C, 7). The posterior nuclei did not have higher amounts of DNA than the anterior nuclei, suggesting that the lack of mitosis in ncc-1 mutants prevents the second S phase that normally follows this division during L1 development. Together, these results show that cells in ncc-1 mutants can replicate their entire genome even late in development. Thus, ncc-1 appears to be required specifically for M phase.
The family of cdc2-related kinases in C. elegans
In this study, we have identified members of the family of cdc2-related kinases in C. elegans. In vertebrates, the kinases that are most closely related to cdc2, can be divided into 4 different groups (Meyerson et al., 1992), and a member of each of these subfamilies appears to be present in C. elegans.
The most extensively studied subfamily consists of Cdk1/Cdc2, Cdk2 and Cdk3. These kinases are closely related in sequence, and each of the human kinases can complement CDC28 mutations in S. cerevisiae. Cdk2 and Cdk1/Cdc2 are activated sequentially in the cell cycle and studies in a number of systems indicate distinct functions. Immunodepletion of the Xenopus homologs showed a specific requirement for Cdk2/Eg1 during S-phase and for Cdc2 during M phase (Fang and Newport, 1991). Antibody-injection studies and expression of dominant-negative mutants implicated Cdk2 and Cdk3 in G1-S control and Cdk1 in G2-M regulation in mammalian cells (Pagano et al., 1993; Riabowol et al., 1989; Tsai et al., 1993a; van den Heuvel and Harlow, 1993). Finally, a Cdk1 mutation has been identified as the genetic lesion in a mouse cell line with a temperature-sensitive G2 arrest phenotype (Th’ng et al., 1990). The C. elegans kinase ncc-1 belongs in this subfamily; it shares more than 60% identity with the three human kinases in this group and can complement the CDC28-1N mutation in S. cerevisiae (Mori et al., 1994). Our studies further define NCC-1 as the ortholog of Cdk1/Cdc2. At present, no homologs of Cdk2 or Cdk3 have been identified, although K03E5.3 and B0285.1 are predicted to encode candidates that are 42% and 41% identical to Cdk2 (see below).
Human Cdk4 and Cdk6 form a subgroup of two kinases that share 71% amino acid identity among each other. These kinases are more distantly related to the Cdk1/2/3 subfamily and they have not been observed to complement CDC28 mutations in yeast (Meyerson et al., 1992). Based on their association with D-type cyclins and timing of activation, Cdk4 and Cdk6 have been implicated in G1 control. Moreover, the kinase inhibitor p16 can bind to and inactivate Cdk4 and Cdk6 and overexpression of p16 can cause G1 arrest in human cells (Sherr, 1996; Sherr and Roberts, 1995). Results from studies in mammals suggest a pathway for G1 control in which the tumor suppressor p16 acts upstream of the oncogene products Cdk4, Cdk6 and Cyclin D, which negatively regulate the retinoblastoma protein, pRB (Sherr, 1996). At present, neither Cdk4 nor Cdk6 have been shown directly to have essential cell-cycle functions. In fact, deletion of the Drosophila cdk4/6 gene was found to cause reduced fertility but not a general cell-cycle arrest (C. Lehner, personal communication). The product of H06A10.1 is most related to this subgroup and is predicted to share 37% identity with human Cdk4 (33% with Cdk6). In RNAi experiments with this kinase, we and others have only identified very limited defects in cell division (M. Park and M. Krause, personal communication).
A third subfamily consists of three closely related PCTAIRE kinases in humans. C. elegans PCT-1 is ∼80% identical to any one of the three human PCTAIRE kinases within a 240 amino-acid fragment of the catalytic core. Cyclin association has not been described for these kinases. In fact, PCTAIRE kinases contain N-terminal and C-terminal extensions that could act as regulators in cis. Such N- and C-terminal extensions are also present in the C. elegans PCT-1 kinase, and measure 234 and 48 amino acids respectively. PCTAIRE message and protein are abundantly expressed in human and rodent brains (Hirose et al., 1997; Le Bouffant et al., 1998; Meyerson et al., 1992). By developmental northern blotting, we detected pct-1 message during all developmental stages in the worm (data not shown). The pct-1 mRNA levels were not found to correlate with mitotic divisions but were strongly elevated in young adults. We did not detect any abnormalities in pct-1 dsRNA-injected animals or their progeny.
A single kinase, Cdk5, that is highly conserved in metazoans defines a distinct subfamily. Cdk5 is closely related to the Cdk1/2/3 kinases and yet no cell-cycle related functions have been identified. By contrast, Cdk5 kinase activity can be detected specifically in post-mitotic neuronal cells (Tsai et al., 1993b). Cdk5 has been implicated in neurite outgrowth during neuronal differentiation (Nikolic et al., 1996), and mice lacking Cdk5 or its activator p35 display severe defects in cortical lamination (Chae et al., 1997; Gilmore et al., 1998). We did not observe obvious defects in neural functions in cdk-5(RNAi) animals.
Implications from RNAi studies
For a remarkable number of genes RNA interference has now been shown to cause strong loss-of-function phenotypes (e.g.: Fire et al., 1998; Tabara et al., 1998). However, absence of an RNAi phenotype is little informative, especially because the mechanism of interference is not understood. Two aspects of RNA interference have been particularly useful in this study: RNAi offers the possibility to examine the function of maternally contributed product and RNAi phenotypes can be used to direct the isolation of mutant alleles in genetic screens. The first aspect relates to a common problem in studying genes that are required in early as well as later development. A single wild-type copy from the mother generally appears sufficient to provide maternal function and, despite early gene requirements, loss-of-function phenotypes may first become detectable during late embryogenesis or early larval development (Ahnn and Fire, 1994; Storfer-Glazer and Wood, 1994). However, RNA interference is particularly efficient in blocking maternal contribution, and RNAi allowed us to examine ncc-1 gene function during early embryogenesis. Following RNAi, NCC-1 protein levels were reduced below the detectable level even in the germline of injected animals. This way, we were able to detect requirement for ncc-1 in mitotic and meiotic division in the germline.
Another application of RNAi is first demonstrated in this study: we used RNAi to predict the ncc-1 loss-of-function phenotype and subsequently identified ncc-1 mutations in animals with similar phenotypes. In an F2 screen, homozygous mutants are likely derived from mothers with one functional gene copy and hence display zygotic phenotypes. For three genes that are both maternally and zygotically required, lin-5, lin-6 and ncc-1, we have observed that some of the early brood after RNA injection closely mimic zygotic phenotypes. Such RNAi larval phenotypes may provide a powerful reverse-genetics approach for many other genes in C. elegans.
ncc-1 acts to promote entry into mitosis
NCC-1 is closely related to yeast CDC28/cdc2 as well as Cdk1/Cdc2, Cdk2 and Cdk3 kinases in vertebrates. A second kinase with such a high degree of homology was not detected. Does ncc-1 perform a role similar to CDC28/cdc2+, regulating multiple transitions in the cell cycle, or does it act to regulate progression through a specific phase of the cell cycle? The data described here show that ncc-1 is required specifically for meiotic and mitotic M phase, indicating that ncc-1 is the ortholog of Cdk1/Cdc2 in other metazoans.
The strongest support for this conclusion stems from the cell-cycle arrest caused by ncc-1 mutations. Several lines of evidence indicate that four of the ncc-1 mutations create strong or complete loss of function. Two of the alleles should produce truncated products that lack most of the conserved kinase domains: he5 contains a deletion and nonsense mutation and he26 contains a nonsense mutation at codon 68. In addition, the missense mutation identified in n3064 causes a Glu-to-Lys substitution of one of nine residues that are nearly invariant in protein kinases. Based on crystal structure studies, this glutamate residue is expected to contact an equally conserved and oppositely charged Arg in the C-terminal end of the catalytic core (De Bondt et al., 1993; Knighton et al., 1991). The missense mutation in he24 results in a Glu-to-Lys change next to the PSTAIRE motif. The corresponding Glu 57 in human Cdk2 is believed to form a hydrogen bond with Tyr 185 in cyclin A (Jeffrey et al., 1995). Binding of cyclin A to the PSTAIRE domain probably contributes to kinase activation by changing both the orientation of ATP and the position of a kinase domain that blocks the catalytic cleft. The he25 allele contains a missense mutation that affects a glutamate residue that is conserved in Cdk1/Cdc2 and Cdk2 but not in kinases outside the Cdc2 family. The phenotype caused by the he25 mutation is partly temperature sensitive, indicating that this mutation does not create a null allele.
The ncc-1 mutants developed normally through embryogenesis and failed all cell divisions after hatching. The first postembryonic cell cycles in ncc-1 mutants appear to be normal in timing of G1 and S phase, based on rnr::GFP expression and BrdU incorporation. However, chromosome condensation and nuclear envelope degradation do not occur. Thus, the cells likely arrest in G2 and ncc-1 appears required to promote progression into mitosis. This function is consistent with the role of Cdk1/Cdc2 kinases in other metazoans. One minor difference exists with Drosophila, in which inactivation of the mitotic kinase activity has been observed to result in polyploidy (Hayashi, 1996). This difference may reflect the fact that many cells in Drosophila undergo endoreduplication after final division.
We expect that ncc-1 is essential for all embryonic and postembryonic divisions. This is based on the observed ncc-1 requirement for cell divisions in all postembryonic lineages and the fact that homologous Cdks are required for all mitotic divisions in other species. As explained above, maternal ncc-1 product likely is sufficient for embryogenesis. However, this assumption lacks formal proof, as ncc-1(RNAi) embryos arrest prior to the zygotic division.
ncc-1 is required for meiosis I
Following ncc-1 RNAi, fertilized oocytes fail to complete meiotic maturation. At the time of injection, a large number of germ precursor cells are in pachytene of meiosis I. These nuclei progress through meiotic prophase I but do not initiate metaphase, indicating that ncc-1 is required to promote the transition from prophase to metaphase in meiosis. A role in meiosis is consistent with the function of Cdk1/Cdc2 in other eukaryotes. In fact, Xenopus p34cdc2 was discovered by the biochemical characterization of maturation promoting factor (MPF), a cytoplasmic factor that induces meiotic maturation when injected into immature oocytes (Masui and Markert, 1971). Completion of meiosis I and II is a two step process in amphibians. Progesterone, or injection of MPF, triggers oocytes that are arrested in diplotene of prophase I to progress through meiosis I. Mature oocytes will subsequently arrest in metaphase of meiosis II and can be triggered by fertilization to complete meiosis (Masui and Markert, 1971). Meiotic maturation in C. elegans is induced by a factor in sperm that is independent from the sperm’s function at fertilization (McCarter et al., 1999). In the absence of sperm, C. elegans oocytes arrest for prolonged periods in diakinesis (McCarter et al., 1999). Future studies may reveal whether progesterone and the ‘sperm factor’ activate similar pathways that induce p34cdc2 and ncc-1, respectively, and trigger progression to meiotic metaphase I.
Does progression through meiotic prophase require ncc-1 function? In the RNA-injected animals, oocyte development appears normal and chromosomes become fully condensed. We cannot exclude that inactivation of ncc-1 by RNAi was incomplete. However, several observations support effective inactivation. First, upon injection of ncc-1 dsRNA, NCC-1 protein levels were reduced below the level of detection (Fig. 2G,I). Secondly, ncc-1 RNAi interfered with cell division in the mitotic part of the germ-line and neutralized phosphorylation of histone H3 in mitotic germ cells as well as oocytes. Thus, a potent effect from RNAi was detected throughout the entire gonad. Finally, injection of dsRNA at a concentration of 0.01 μg/μl was sufficient to completely arrest meiotic maturation, whereas even a 100-fold higher concentration did not affect progression through prophase of meiosis I. If inactivation was complete, entry into meiotic development, meiotic prophase progression, condensation of chromosomes to diakinesis bivalents and oocyte development can all occur independent of ncc-1. If these processes do require ncc-1 activity, this must be less activity than required for germ cell proliferation and for the transition from diakinesis to meiosis I. As yet, formation of diakinesis bivalents of normal morphology in the absence of Cdk1 activity has not been described in any eukaryote. During spermatogenesis in Drosophila, DNA condensation can occur in the apparent absence of Dmcdc2 activity (Sigrist et al., 1995). However, male meiosis in Drosophila does not involve meiotic recombination and attachment of bivalents by chiasmata.
Nuclear envelope breakdown occurs slowly in mature ncc-1(RNAi) oocytes; by the time it is accomplished meiosis would normally have been completed. The fact that degradation still occurs could indicate incomplete inactivation of ncc-1, as Cdk1 is believed to phosphorylate nuclear lamins and to trigger nuclear envelope breakdown in mitosis (Peter et al., 1990). However, other kinases have also been implicated in nuclear lamin phosphorylation, including S6 kinase II (Ward and Kirschner, 1990) and protein kinase C (Hocevar et al., 1993). Such kinases may trigger nuclear enveloppe breakdown in the absence of ncc-1 activity.
How is progression through G1/S regulated in C. elegans?
Our results provide an interesting paradox: ncc-1 appears specifically required for the regulation of M phase, yet essential regulators of the G1-S transition have not been identified. Such regulators are especially important as regulation of cell division by developmental signals in most cells occurs during G1 phase. Ectopic expression of the Cdk inhibitor cki-1 causes cell division arrest prior to S phase in C. elegans (Hong et al., 1998), which is consistent with the assumption that G1-S progression is dependent on Cdk activity. In other metazoans, Cdk2 appears to promote G1/S transition, although genetic null phenotypes have not been described as yet. As the C. elegans genome has been completely sequenced, the C. elegans kinase that is most similar to Cdk2 is likely encoded by the K03E5.3 gene. RNAi for this gene caused severe but highly variable developmental and cell division defects. More severe defects where seen when the same RNA was injected into heterozygous ncc-1/+ animals (data not shown). We are currently exploring whether K03E5.3 and ncc-1 can act redundantly in G1/S progression.
We are indebted to Ed Harlow, in whose lab this work was initiated. We express our gratitude to the members of the Harlow-Dyson laboratories for stimulating discussions, and Li Huei Tsai for help with the sequence analysis of cdk-5. Anne Hart, Marc Vidal, Mako Saito and Nick Dyson are acknowledged for critically reading the manuscript, M. Park, M. Krause, Y. Kohara and Andy Golden for kind gifts of plasmids and antibodies, Alan Coulson for cosmids and Michael Hengartner and Bob Horvitz for help and advice. The Caenorhabditis Genetics Center, supported by the NIH National Center for Research Resources, provided several strains for this work. S. v.d.H. received grants from the Helen Hay Whitney Foundation and the Jessie B. Cox Charitable Trust and the Medical Foundation. D. G. S. was supported in part by a William A. Starr Fellowship and NIH, National Research Services Award CA09642-09.