The Function of a Spindle Checkpoint Gene bub-1 in C. elegans Development

Background The serine/threonine kinase BUB1 (Budding Uninhibited by Benzimidazole 1) was originally identified in yeast as a checkpoint protein, based on its mutant's incapacity of delaying the cell cycle in response to loss of microtubules. Our understanding of its function is primarily from studies carried out in yeast S. cerevisiae. It has been shown that it is a component of the mitotic spindle checkpoint and regulates the separation of sister chromatids through its downstream molecules. However, its roles in multi-cellular organisms remain unclear. Methods and Findings In nematode C. elegans, rapid cell divisions primarily occur in embryos and in germline of postembryonic larvae and adults. In addition, a select set of cells undergo a few rounds of cell division postembryonically. One common phenotype associated with impaired cell division is described as Stu (Sterile and Uncoordinated) [1], [2]. We conducted a genetic screen for zygotic mutants that displayed Stu phenotype in C. elegans. We isolated seven Stu mutants that fell into five complementation groups. We report here that two mutations, FanWang5 (fw5) and FanWang8 (fw8) affect the bub-1 gene, a homolog of yeast BUB1. Both mutant alleles of fw5 and fw8 exhibited variable behavioral defects, including developmental arrest, uncoordination and sterility. The number of postembryonically born neurons in the ventral cord decreased and their axon morphology was abnormal. Also, the decrease of neurons in the ventral cord phenotype could not be suppressed by a caspase-3 loss-of-function mutant. In addition, bub-1(fw5 and fw8) mutants showed widespread effects on postembryonic development in many cell lineages. We found that bub-1 functioned maternally in several developmental lineages at the embryonic stage in C. elegans. Studies in yeast have shown that BUB1 functions as a spindle checkpoint protein by regulating the anaphase promoting complex/cyclosome (APC/C). We performed double mutant analysis and observed that bub-1 genetically interacted with several downstream genes, including fzy-1/CDC20, mat-2/APC1 and emb-27/APC6. Conclusions Our results demonstrate a conserved role of bub-1 in cell-cycle regulation and reveal that C. elegans bub-1 is required both maternally and zygotically. Further, our genetic analysis is consistent with that the function of bub-1 in C. elegans is likely similar to its yeast and mammalian homologs.


Introduction
Precise chromosome segregation during cell division is controlled by a feedback mechanism [3]. During the mitotic cell cycle, the metaphase-to-anaphase transition occurs after all chromosomes have established precise bipolar attachments to the mitotic spindles [4]. The spindle checkpoint inhibits anaphase onset until kinetochores are properly bound with the spindle microtubules [5], [6]. Malfunction of the spindle checkpoint leads to precocious anaphase and chromosomal missegregation, and results in subsequent loss of genetic fidelity. Misregulation of the spindle checkpoint has been suggested as a major cause of fatality and cancer [7,8,9,10,11]. In 1990s, several groups have isolated a number of genes involved in the budding yeast spindle checkpoint, including MAD1 (Mitotic Arrest Deficient 1), MAD2, MAD3 [12], BUB1, BUB2, BUB3 [13], and MPS1 (Monopolarspindle 1) [14]. BUB1 is a serine/threonine kinase that regulates the separation of sister chromatids. Studies from yeast have also shown that BUB1 acts through APC/C, a large multi-subunit E3 ubiquitin ligase [15,16]. In addition, BUB1 localizes at the kinetochore during the very early stages of mitosis, and is required for kinetochore localization of MAD1 and MAD2, independent of its kinase activity [9]. Following the localization of BUB1, MAD1 then lowers the energy barrier of MAD2 and triggers MAD2 conformational change, allowing MAD2 binding to the APC/C activator CDC20. After the formation of the mitotic checkpoint complex (MCC), which contains BUBR1-BUB3-MAD2-CDC20, APC/C is inhibited by the complex [17]. This process results in the stabilization of securin, an inhibitor keeping separase inactive, and also hindrance of sister chromatids separation [18]. In mammalian cells, phosphorylation of CDC20 by BUB1 has also been shown to inhibit the function of CDC20 [19]. In C. elegans, components of the spindle checkpoint are functionally conserved [20,21].
C. elegans has a single homolog of BUB1, bub-1. Antibody staining at one-cell stage shows that BUB-1 is an essential component in the mitotic kinetochore [22], consistent with its function in spindle checkpoint. RNAi of bub-1 in wild type results in embryonic arrest, and partially restores mitotic timing at onecell stage in conditional embryonic-lethal apo-5(or358ts) mutant embryos with cytoskeletal abnormalities, suggesting that bub-1 may be associated with spindle checkpoint at the early embryonic stage [23]. Studies of putative downstream genes of bub-1: mdf-1/MAD1, mdf-2/MAD2, mdf-3/MAD3, and fzy-1/CDC20 have also shown that these genes function during spindle checkpoint process [3,24]. In a genetic screen for zygotic mutants that are likely associated with cell cycle defects, we isolated two bub-1 mutant alleles. Our analysis shows that bub-1 functions in multiple cell lineages and plays essential roles in the development of C. elegans.

New Stu mutant screen
In C. elegans, some of the cell cycle mutants show morphological and behavioral defects including Stu and Emb (Abnormal EMBryogenesis). Emb commonly leads to embryonic lethality, while Stu mutants are often associated with defects in the development of gonads (sterility) or neurons in the ventral nerve cord (uncoordination) [25,26,27]. Some Stu mutants survive through embryonic development, likely due to maternal deposit of normal gene products [25]. In an effort to identify new cell cycle related genes in C. elegans, we conducted a clonal screen for Stu mutants using a GFP marker juIs76 [Punc-25::GFP] that visualizes the D-type ventral cord motor neurons, which include embryon-ically born DD neurons and postembryonically born VD neurons [28]. We isolated seven Stu mutants from 3500 haploid genomes. By linkage group mapping and complementation tests, we found that these mutants fell into five complementation groups, of which one was a mcm-5 allele that we had reported previously [29]. Table 1 shows the remaining four mutant complementation groups and their phenotypes. All animals isolated showed uncoordination, larval arrest, sterility and vulva defects (either vulvaless or protruding vulva). These phenotypic defects are commonly observed in animals with abnormal postembryonic development [27].

All new Stu animals have motor neuron defects
The generation of adult ventral nerve cord involves a series of postembryonic cell division in late L1 larvae, resulting in a fixed number of neurons arranged in a stereotypic manner [30]. To evaluate the mutant phenotype, we counted the number of ventral cord motor neurons. In wild type animals, Punc-25::GFP visualizes 6 DD and 13 VD neurons in the ventral nerve cord [28]. The DD neurons are born at the embryonic stage, whereas VD neurons are born at the L1 larvae stage. All mutants had normal number of DD neurons in L1 larvae (data not shown). However, in later larvae (L2 or older) and adults, all mutants showed a general decrease in the number of GFP-expressing VD neurons ( Figure 1). To confirm our findings, we used a pan-neuronal marker evIs111 [31] and DAPI staining. The result showed that the mutants were missing many neurons, consistent with previous findings (data not shown). As reported previously, impairment in cell cycle often causes defects in cell morphology [26]. By examining the morphology of motor neurons in the mutants, we found that some VD neuron axons showed defective morphology in several mutants ( Figure 1C and Table 2).
Both fw5 and fw8 are mutations in bub-1 To identify the corresponding genes of the new Stu mutations, we performed snip-SNP mapping (see Materials and Methods) [32]. We mapped fw5 and fw8 to the same interval (between the SNP marker of B0041:6882 and VF39H2L: 3079) on the chromosome I. Further, fw5 and fw8 failed to complement. Both   fw5 and fw8 were balanced by dpy-5(e61) unc-29(e403) for stock keeping.
We tested a set of RNAi clones covering the interval, and found that RNAi escapers of bub-1 led to reduced number of D-type neurons as well as Emb (data not shown). We then sequenced fw5 and fw8, and identified nucleotide alterations in the bub-1 gene in both alleles (Figure 2A). In C. elegans, the bub-1 gene encodes a 987aa protein with a conserved kinase domain at its C-terminal ( Figure 2A). The mutations in fw5 and fw8 result in stop codon at W848 and W726, respectively, which produce truncated proteins lacking the kinase domain. We also obtained a deletion mutant, tm2815, which had an in-frame deletion of 105 amino acids from E473 to A586 in the middle of the protein, with unaffected kinase domain ( Figure 2A). Homozygous tm2815 animals displayed embryonic arrest, larval arrest and sterility. However, the phenotypes observed in the deletion allele were weaker than those of fw5 or fw8. We also generated tm2815/fw5 and tm2815/fw8 animals and found that a larger number of surviving adult stage animals compared to homozygous fw5 or fw8 (Table 1). This result indicates that tm2815 mutant behaves as a partial loss of function mutation, and fw5 and fw8 are more likely to be null mutations of bub-1.
We also performed transgenic rescue of the bub-1 mutant using a PCR product which encompasses the region from 1.40-kb upstream to 0.82-kb downstream of the bub-1 locus. We obtained two transgenic fw8 homozygous lines after injecting the PCR product to the dpy-5(e61) unc-29(e403)/fw8 animals. In both lines, the mutant phenotypes were fully rescued. Furthermore, expression of bub-1 driven by a pan-neuronal promoter (the promoter of unc-119) was also able to rescue the neuronal defect of fw8 as well. All three transgenic lines showed partial rescue of the loss of D type neurons (the t-test compared to control fw8; juIs76: P,0.001) ( Figure 2B). These results suggest that BUB-1 is responsible for the Stu phenotypes of fw8, and bub-1 functions in the nervous system in a cell-autonomous manner.
The neuronal defect of fw8 is unlikely caused by caspasedependent programmed cell death As mentioned earlier, the number of VD neurons was reduced in bub-1(fw8) mutants ( Figure 1B). The loss of neurons could be due to abnormal cell division [33] or enhanced apoptosis [34], [35]. To examine these possibilities, we constructed a bub-1(fw8); ced-3(n717) double mutant in which programmed cell death would be blocked due to the loss of CED-3 caspase activity [36]. We found that bub-1(fw8) and bub-1(fw8); ced-3(n717) resulted in similar numbers of D-type neurons [average 8.7 (n = 244) and 8.3 (n = 31), respectively] ( Table 3). This result indicates that caspasedependent apoptotic cell death is unlikely responsible for the loss of motor neuron in bub-1(fw8). However, we could not rule out the possibility of caspase-3-independent cell death in bub-1(fw8) mutant.

bub-1 is required both maternally and zygotically
The fact that bub-1 mutants caused only postembryonic-born VD neuron defects suggested two possible reasons: 1) bub-1 is a maternal gene and 2) bub-1 is specifically required at postembryonic stages. A bub-1 promoter driven GFP was widely expressed from early embryonic stages to three fold stage ( Figure 3A). Anti-BUB-1 antibody staining also showed BUB-1 was present at the one-cell stage [22], and during the late embryonic stage ( Figure 3B). To examine the roles of bub-1 in early embryos, we fed bub-1 RNAi to the eri-1(mg366); juIs76 animals, which sensitized the RNAi effect [37]. We found that approximately 90% of the progenies from the RNAi-fed parents showed the Emb phenotype (n = 779). To characterize at which stage the embryos arrested, we stained the Emb embryos with DAPI and found that approximately 1% of them arrested at the early embryonic stage (an average of twenty nuclei, n = 107), while about 94% arrested at late embryonic stage (an average of 100 nuclei, n = 107). Only a few of the embryos arrested at the comma stage (5.6%, n = 107). These observations indicate that BUB-1 is required maternally during embryogenesis, in addition to its zygotic roles in postembryonic development.

Effects of bub-1 in postembryonic development
In C. elegans, multiple types of tissues undergo several rounds of cell divisions during postembryonic development. Using a panel of markers, we examined the development of several tissues in bub-1 mutants as described below.
Intestinal nuclei division but not endoreduplication was defective. The transgenic GFP line rrIs1 was used to visualize the nuclei of the intestine cells ( Figure 4A). In wide type late L1 animals, the intestine cells have 30 to 34 diploid nuclei. All intestinal nuclei endoreduplicate their DNA prior to each of the four molts, thereby producing the 32 n DNA content nuclei in the adult intestine [39]. We found, however, about 24 intestinal nuclei in the bub-1(fw8) mutant L4 larvae (n = 17) ( Figure 4B). In addition, some of the intestinal nuclei were elongated and showed a thread structure, suggesting a defect in chromosomal segregation [25]. This observation was consistent with the DAPI staining experiment ( Figure 5B). Furthermore, we checked the DNA content of the intestinal nuclei in the bub-1(fw8) L4 or adults. Using body wall muscle nuclei as an internal 2 n control, we determined that the amount of DNA in the intestinal lineages was 24.3 n in the bub-1(fw8) mutant, while 28.4 n in the WT ( Figure 6). If the arrest of cell division prior to L4 stage and the lack of the last DNA replication before L4 to adult molt are taken into account, we tend to believe that the intestinal nuclei endoreduplication might not be affected by the loss of bub-1 function. However, the cell division may be affected.
Division of seam cells was severely disrupted. We used a transgenic GFP line wIs51 to visualize the nuclei of seam cells [40] ( Figure 4C). Ten seam cells aligned on each side of the body undergo stage-specific division patterns at each of four (L1-L4) postembryonic larval stages. From the L2 to L4 stage, the wild type animal has 16 seam cells [38]. In most of the bub-1(fw8) L4 mutants, only the two most anterior seam cells H0 were present ( Figure 4D). These H0 cells normally do not undergo postembryonic division [41]. These results indicate a severe failure in postembryonic division of seam cells.
Gonad development was severely impaired. The transgenic GFP line qIs56 allowed us to visualize the two distal tip cells (DTCs) of the U-shaped gonads [42] (Figure 7A and 7C). The gonad arms acquire their U-shape by directed migration of the DTC. The arm elongation begins at the L2 stage and continues until the L4 molt [38]. We observed that about half of bub-1(fw8) animals showed only one gonad arm, and most of them stopped development prematurely (n = 32) ( Figure 7B and 7D). Among the 48 gonad arms scored, 9 grew one quarter or less of the normal gonad length; 15 gonad arms grew less than one half of the normal length; and 10 gonad arms grew about three quarters of the normal length. Furthermore, the number of germ cells in bub-1(fw8) was decreased to about 117 per arm (n = 11) (compared to about 1000 in wild type). In the abnormal gonads, we did not observe any eggs. Sperms, however, formed only in 2 of the 9 bub-1 mutant animals observed by DAPI staining ( Figure 5D).
Ventral cord motor neurons. We used a transgenic GFP strain, juIs14 [33], to visualize the cholinergic DA, DB, VA, and bub-1 in C. elegans PLoS ONE | www.plosone.org VB neurons ( Figure 8A). We observed a decreased number of neurons expressing GFP in the bub-1(fw8) mutant. Normally, embryonic-born DAs and DBs have commissural projections to the dorsal cord, while postembryonic-born VAs and VBs do not [38]. We found that the number of commissural projections to the dorsal cord was unchanged in the bub-1(fw8) mutant, and the axons of these neurons did not show any morphological defects (data not shown). Therefore, embryonic-born DAs and DBs were not affected, while most postembryonic-born VAs and VBs were missing in the bub-1(fw8) mutant.

Genetic interaction analysis supports a role of BUB-1 in the spindle checkpoint pathway
Previous studies have shown that several components of the spindle assembly pathway are functionally conserved in nematodes and yeast [20,21]. For example, the loss-of-function of mdf-1/ MAD1 causes embryonic and larval arrest [3], similar to the yeast mutant. Further, the lethal phenotype of mdf-1/MAD1 is suppressed by the mutations in the downstream genes, such as fzy-1/CDC20 [20], and APC/C homologues, such as emb-30/ APC4 [43] and such-1/APC5-like [44]. To test if bub-1 acts in the spindle checkpoint pathway, we examined genetic interactions between bub-1(fw8) and several candidate downstream genes.

Discussion
Identification and characterization of loss-of-function mutations of C. elegans bub-1, a cell cycle spindle checkpoint gene Our conclusion that fw5 and fw8 are loss of function mutations in bub-1 is based on the following results: 1) they failed to complement with each other; and were mapped to the same genetic interval; 2) RNAi against bub-1 exhibited the same phenotypes as fw5 and fw8; 3) sequencing data showed that fw5 and fw8 both contained nonsense mutations in the bub-1 coding sequence; 4) an in-frame deletion mutant of bub-1 (tm2815) failed to complement with fw5 and fw8, and exhibited weaker phenotypes than fw5 and fw8; and 5) fw5 and fw8 could be rescued by bub-1 DNA and partially rescued by expression of bub-1 gene driven by a pan-neuronal promoter.

BUB-1 may have both kinase-dependent and kinaseindependent functions
Compared to our bub-1 mutant fw5 and fw8, the deletion mutant bub-1(tm2815) showed milder defects. This is likely due to an existing partial function of bub-1(tm2815). Based on sequence alignment among different species, Bub1 has a conserved kinase domain at the C-terminus. Both fw5 and fw8 have premature stop codon prior to the kinase domain, whereas bub-1(tm2815) has an in-frame deletion which leaves an intact kinase domain. This might explain why fw5 and fw8 have more severe defects than bub-1(tm2815). Furthermore, this difference might suggest that bub-1 functions beyond a kinase. In yeast, BUB1 is required for kinetochore localization of MAD1 and MAD2 independent of its kinase activity [9]. Further, mdf-1 (mitotic arrest defective) and mdf-   2 were identified as homologs of MAD1 and MAD2, and both exhibited conserved function in nematode and yeast [3]. Whether or not the kinase-independent function of bub-1 exists in C. elegans still needs to be investigated further.
Our studies demonstrate that the cell cycle control gene bub-1 functions widely in the development of C. elegans. The bub-1 null mutants exhibited defects in several developmental lineages, including seam cells, intestine nuclei, vulva, gonad, germ cells, and ventral cord neurons. Other postembryonic cell lineages we inspected were also defective in bub-1 mutants (data not shown). In bub-1(fw5, fw8) mutants, all of the neurons in the ventral cord developed at the embryonic stage were intact, such as DAs, DBs, and DDs; while most of the postembryonic-born neurons were missing, such as VAs, VBs, and VDs. Our RNAi experiment shows that bub-1 is a maternal gene and the maternal effect of bub-1 is strong enough to support embryonic development even to the adult stage in bub-1 mutants. In C. elegans, some cell cycle-related genes also show long lasting maternal function. For example, cye-1 Cyclin E deletion animals showed surprisingly normal development until the L3 stage, although RNAi resulted in embryonic lethality at nearly the hundred-cell stage [46,47].
The endoreduplication may not be affected by the loss of bub-1 function Metazoans have various types of cell cycles during development. Endoreduplication is a specific type of cell cycle that skips the M phase. In C. elegans, such endoreduplication type of cell cycle takes place in the intestine and hypodermis during development [39]. Intestinal nuclei go through an endoreduplication cycle before each molt, which results in adults with intestinal nuclei with a 32 n DNA content. In adult animals or L4 with bub-1(fw8) mutants, we found that the amount of DNA was not affected. This result suggests that bub-1 function is specifically required for the spindle checkpoint in the M phase, which is missing from the endoreduplication in the C. elegans intestinal cells.

Out-crossing, mapping and complementation testing
All of the mutants were out-crossed at least twice with N2. The mutants were mapped using standard snip-SNP assay [32] and the three-factor mapping technique [52]. The mutants mapped to similar genetic loci were tested. fw2 and fw3 were allelic, as were fw5 and fw8. For the complementation procedure, we used heterozygous bub-1(fw5)/+ males to cross with the balanced strain dpy-5(e61) unc-29(e403)/dpy-5(e61) bub-1(fw8). The progenies bub-1(fw5)/dpy-5(e61) bub-1(fw8) were sterile and uncoordinated, which was similar to the fw5 or fw8 homozygous mutants. Complementation tests with known genes were also performed. These genes were within the same loci and generated similar phenotypes.

Phenotypic quantification of Stu mutants
L4 Heterozygous balanced mutants, such as dpy-5(e61) unc-29(e403)/bub-1(fw8), were cultured at 20uC and transferred everyday to new plates to obtain synchronized progenies. From these plates, the uncoordinated F1 animals were transferred to new plates and cultured for about 5 days at 20uC to quantify the final phenotype. Larval arrest phenotype was quantified according to body size. The absence of fertilized eggs was scored as sterility. For the adult Stu animals, vulval morphology was quantified by mounting them in 2% agar pads and viewed under a stereoscope. Animals with protruding vulva were scored as Pvl, and others without vulva were scored as Vul. The D-type neuron phenotype of L1 stage animals were quantified 10 hours later after lysing the adult heterozygous mutants (+/2), using alkaline hypochlorite. A quarter of the population in these L1 animals become homozygous mutants (2/2).

Nomarski fluorescent microscope examination
Live animals were mounted to M9 solution in 2% agar pads and viewed under Leica and Zeiss microscopes. Images were captured using a Leica DC500 or a Zeiss AxioCam.

Molecular analysis of bub-1
To identify the mutations in fw5 and fw8, the sequences for the exons and exon-intron boundaries of bub-1 were amplified from homozygous mutant animals using the following primers: first pair (59gcgtcctttctactttga39, 59gcttttcccgagttattt39); second pair (59ttcaatgcgggttctaag39, 59ctggagggttaccatctt39); third pair (59tcgtcggatacaaagtct39, 59ggttggagcaacaaatac39); fourth pair (59tttcaaaccgtctcgtgg39, 59tcaggcgattccgcattt39); fifth pair (59gtcaaggtggatacgctaa39, 59actttcctgcaacaacga39); and sixth pair (59aatggctgtcgttgttgc39, 59 ttctaccgtgatgggtct39). The mutations were confirmed by sequencing from both directions (through two different reactions). To generate a bub-1 promoter-driven GFP construct, duplex PCR [53] was conducted to amplify the 1266 bps bub-1 upstream sequence from N2 genomic DNA using the following primer set. 59gattcccacaagtaggtc39 and 59agtcgacctg-caggcatgcaagcttcaaagtagaaaggacgcga39. The final Pbub-1::GFP DNA fragment (100 ng/ml) was injected into the N2 strain using a pRF4 plasmid (100 ng/ml) as co-injection marker. Two lines were obtained and both showed similar expression patterns.

Antibody staining
The freeze-crack method was used for permeabilization and fixation of the embryos [55]. The rabbit polyclonal antibody against BUB-1 (1:1000, a gift from Dr. Hyman [22]) was used, followed by the FITC conjugated mouse anti-rabbit secondary antibody (1:1000).

RNAi by feeding
RNAi clones were made by J. Ahringer's laboratory [56], and obtained from the MRC service (UK). The bacteria expressing dsRNA of appropriate genes were cultured at 37uC overnight and seeded onto the NGM plates (containing 50 mg/mL Amp, 1 mM IPTG). The plates were kept at room temperature for two days. Three L4 CZ5547 (mg366; juIs76) animals were transferred to the plates. Two days later, the animals were then transferred to a second plate with the same interfering bacteria. About 10 hours later, the animals were removed and the embryos were cultured for a period of several days in order to examine the phenotype. The results were scored from the second plate, which displayed a better representation of the gene's mutant phenotype.

DAPI staining
Approximately 30 mutant animals were placed into M9 on a microscope slide and covered with coverslip. The slide was quickly frozen in liquid nitrogen and put into a pre-cooled iron block. The coverslip was then quickly removed. The slide was sequentially placed in methanol and then acetone for 10 minutes each at 220uC. After air drying, animals were treated with 49,6diamidino-2-phenylindole dihydrochloride (DAPI) and covered with a coverslip [55].

DNA quantitation
To quantitate DNA content, nuclei images of DAPI-stained animals were taken with a Zeiss AxioCam, and images were analyzed with NIH ImageJ 1.40 g software. Using body wall muscle nuclei as a 2 n DNA standard, C values of intestinal nuclei were estimated by their DAPI-based densitometric quantifications [57,58].