Many Genomic Regions Are Required for Normal Embryonic Programmed Cell Death in Caenorhabditis elegans

Corresponding author: Joel H. Rothman, Department of MCD Biology and Neuroscience Research Institute, University of California, Santa Barbara, CA 93106., rothman{at}lifesci.ucsb.edu (E-mail)

Communicating editor: P. ANDERSON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To identify genes involved in programmed cell death (PCD) in Caenorhabditis elegans, we screened a comprehensive set of chromosomal deficiencies for alterations in the pattern of PCD throughout embryonic development. From a set of 58 deficiencies, which collectively remove 74% of the genome, four distinct classes were identified. In class I (20 deficiencies), no significant deviation from wild type in the temporal pattern of cell corpses was observed, indicating that much of the genome does not contain zygotic genes that perform conspicuous roles in embryonic PCD. The class II deficiencies (16 deficiencies defining at least 11 distinct genomic regions) led to no or fewer-than-normal cell corpses. Some of these cause premature cell division arrest, probably explaining the diminution in cell corpse number; however, others have little effect on cell proliferation, indicating that the reduced cell corpse number is not a direct result of premature embryonic arrest. In class III (18 deficiencies defining at least 16 unique regions), an excess of cell corpses was observed. The developmental stage at which the extra corpses were observed varied among the class III deficiencies, suggesting the existence of genes that perform temporal-specific functions in PCD. The four deficiencies in class IV (defining at least three unique regions), showed unusually large corpses that were, in some cases, attributable to extremely premature arrest in cell division without a concomitant block in PCD. Deficiencies in this last class suggest that the cell death program does not require normal embryonic cell proliferation to be activated and suggest that while some genes required for cell division might also be required for cell death, others are not. Most of the regions identified by these deficiencies do not contain previously identified zygotic cell death genes. There are, therefore, a substantial number of as yet unidentified genes required for normal PCD in C. elegans.

PROGRAMMED cell death (PCD, or apoptosis) is essential for normal development and homeostasis in most multicellular organisms (JACOBSON et al. 1997 ). For example PCD is crucial for morphogenesis during embryogenesis and for elimination of excess neurons during construction of the nervous system. PCD also performs critical defensive functions in pathogenesis and carcinogenesis.

The nematode Caenorhabditis elegans has proven to be a useful model organism for studying the core machinery involved in PCD. A unique advantage of C. elegans is that all cell deaths occur at precisely defined times and locations. Of the 1090 somatic nuclei made during its development, exactly 131 cells die by PCD (SULSTON and HORVITZ 1977 ; SULSTON et al. 1983 ). Genetic approaches have made it possible to identify large numbers of mutations that affect PCD in this animal. Genetic screens have identified four proteins (CED-3, CED-4, CED-9, and EGL-1) that are central to the execution of the death program (reviewed by METZSTEIN et al. 1998 ). CED-3, a member of the caspase protease family, induces PCD when activated (YUAN et al. 1993 ; XUE et al. 1996 ). CED-4 (which is similar to mammalian Apaf-1) activates the protease activity of CED-3 (CHINNAIYAN et al. 1997A ; SESHAGIRI and MILLER 1997 ; ZOU et al. 1997 ). In cells in which the death program is blocked, CED-9, the orthologue of mammalian Bcl-2 (HENGARTNER and HORVITZ 1994 ), binds to CED-4 and prevents it from activating CED-3 (CHINNAIYAN et al. 1997B ; XUE and HORVITZ 1997 ). EGL-1, a BH3 domain containing protein, binds to CED-9 causing it to release CED-4, thereby promoting CED-3-dependent cell death (CONRADT and HORVITZ 1998 ; DEL PESO et al. 1998 ). These core events in PCD appear to be conserved in all metazoans (reviewed in SONG and STELLER 1999 ).

Genetic approaches have also identified several genes that are involved in other aspects of PCD. Six genes, ced-1, -2, -5, -6, -7, and -10 are involved in engulfment of cell corpses once the death program has been activated (HEDGECOCK et al. 1983 ; ELLIS et al. 1991 ). ced-8 is required for the timely onset of PCD (STANFIELD and HORVITZ 2000 ). ces-1 and ces-2, both of which encode apparent transcription factors, activate PCD in a particular set of cells (METZSTEIN et al. 1996 ; METZSTEIN and HORVITZ 1999 ).

Although many factors involved in PCD have been identified, the way in which these functions are integrated with developmental processes, such as cell proliferation and cell-signaling events, are not well understood and it is clear that other components remain to be identified. For example, the comprehensive set of targets of the CED-3 protease is not known. Moreover, it is unclear how EGL-1, the most upstream component known in the core PCD pathway, is regulated. In addition, it is likely that additional ces-like genes exist that determine the death fate of subsets of cells. Finally, while the participation of some cell cycle regulators in apoptosis has been reported in other organisms (reviewed in GUO and HAY 1999 ), their roles in C. elegans PCD have not been examined.

Since previous screens for cell death mutants were biased against lethal mutations (HEDGECOCK et al. 1983 ; ELLIS and HORVITZ 1986 ), genes essential for both viability and normal PCD are likely to have been overlooked. To screen the genome for zygotic genes required for normal PCD without bias against lethal genes, we have performed a screen of genetic deficiencies (chromosomal deletions).

Deficiencies can be used to scan segments of the genome for zygotic gene functions. Animals homozygous for most deficiencies arrest during embryogenesis. Their phenotypes usually reflect the null phenotype of one or many zygotic genes. The primary advantage of a deficiency screen is that a large fraction of the genome can be efficiently screened. In addition, deficiencies can identify genes (e.g., very small genes or neighboring genes that perform redundant functions) that are not readily mutated by conventional point mutagenesis (e.g., WHITE et al. 1994 ; GRETHER et al. 1995 ; CHEN et al. 1996 ; ZHU et al. 1997 ). Most deficiencies do not prevent differentiation or cell division and it is possible to screen deficiency homozygotes for defects in a wide range of processes that occur during embryogenesis. Indeed, deficiency screens in Drosophila have been useful for identifying regions of the genome required for morphogenesis and patterning (MERRILL et al. 1988 ; WIESCHAUS and SWEETON 1988 ; BILDER and SCOTT 1995 ), DNA replication (SMITH et al. 1993 ), and PCD (WHITE et al. 1994 ) and in C. elegans to identify loci required for early cell division patterns (STORFER-GLAZER and WOOD 1994 ), embryonic tissue patterning (CHANAL and LABOUESSE 1997 ; LABOUESSE 1997 ; TERNS et al. 1997 ; ZHU et al. 1997 ), and muscle differentiation (AHNN and FIRE 1994 ).

In this study, we report the analysis of 58 chromosomal deficiencies that encompass 74% of the C. elegans genome. We find that these deficiencies can be categorized into four phenotypic classes: (1) normal PCD, (2) no or reduced PCD, (3) excess PCD corpses, and (4) large PCD corpses. We identified 28 new genomic regions involved in PCD. These analyses show that additional genes that participate in execution of PCD and cell corpse engulfment remain to be identified. Finally, our findings suggest that the developmental program for cell death can be uncoupled from cell proliferation, demonstrating that the normal number of cell division rounds is not essential for PCD.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Worm culture and strains:
Techniques for culturing C. elegans are described elsewhere (BRENNER 1974 ; LEWIS and FLEMING 1995 ). All strains were grown at 20°. The strains used as sources of deficiencies analyzed in this study are listed in Table 1. Most deficiencies used are balanced with a balancer chromosome (mnC1 II, sC1 III, and qC1 III) or a chromosome bearing two visible mutations that closely flank the deficiency; such deficiency strains were used directly for analyses by Nomarski microscopy. Some deficiencies were balanced by reciprocal translocations [eT1 (III;V), nT1 (IV; V), szT1 (I;X)], which segregate a high proportion of lethal embryos due to aneuploidy. Strains of this type were crossed with N2 wild-type males and the F₁ progeny that produced 25% arrested embryos were used as sources for deficiency embryos in the screen. eDf2, rhDf1, and mnDf1 are maintained with free duplications and deficiency homozygotes are generated by loss of the duplications during meiotic segregation. To obtain nullo-X embryos we used him-8(e1489), which produces a high fraction of nullo-X ova (HODGKIN et al. 1979 ). When him-8(e1489) hermaphrodites are crossed with wild-type XO males, most of the arrested embryos are nullo-X (AHNN and FIRE 1994 ).

The data in the 1999 Genetic Map of Caenorhabditis elegans (published by the Caenorhabditis Genetics Center) were used to estimate the number of mutationally defined genes collectively deleted by the deficiencies. The genes in the genetic map between the leftmost gene and the rightmost gene deleted by each deficiency (confirmed by complementation tests) were regarded as those deleted by the deficiency. Since only a subset of genes was examined by complementation tests, each deficiency is likely to delete more genes than this estimate and this calculation therefore provides a conservative estimate of the fraction of the genome deleted. Collectively, the deficiency collection we screened deleted a minimum of 1213 out of the 1644 genetically defined genes on the genetic map, or 74% of the genetically defined genome.

Nomarski analysis of deficiency embryos:
In general, gravid hermaphrodites heterozygous for each deficiency were dissected in M9 solution and young (<100-cell stage) embryos were collected for cell corpse analysis. Approximately 40 embryos were mounted in M9 solution on agar pads (containing 5% Difco Bacto agar in distilled water) and observed by Nomarski optics. Approximately one-fourth of the embryos are expected to be homozygous for the deficiency and therefore arrest during embryogenesis without hatching. In some strains, >25% of embryos did not hatch owing to haploinsufficiency of genes deleted by the deficiency. Usually such arrested heterozygous embryos were distinguishable from the deficiency homozygotes by comparing their terminal phenotypes; the heterozygotes generally showed a less severe arrest phenotype than the majority class comprising the 25% arrested homozygotes.

The number and appearance of cell corpses were analyzed by scoring fields of embryos every 30–60 min. Embryos were scored until the wild-type siblings that started at the same developmental stage reached at least the 2-fold stage. Cell corpses in the terminally arrested deficiency embryos were also scored >14 hr after the first cleavage. Corpses were counted at four stages during embryogenesis: stage I, ("lima bean" stage in wild type, 340 min after first cleavage); stage II, ("comma" stage in wild type, 380 min); stage III, ("1.5-fold" stage in wild type, 420 min); and stage IV, ("2-fold" stage in wild type, 460 min) (SULSTON et al. 1983 ). Since most deficiency homozygotes show morphological defects, we used the time after first cleavage, instead of morphological characteristics, to determine the age of each embryo. The appearance of the collected embryos varied; the approximate time of the first cleavage of each embryo was estimated retrospectively based on the number of cells at the beginning of the observation.

Quantifying cell corpses:
For each deficiency, the number of corpses at each stage in the homozygotes (Df/Df) was compared with that of their wild-type-appearing siblings (Df/+ and +/+) and N2 (wild type) using Student's t-test (two-tailed). Nonhomozygotes and N2 were also compared to assess whether the parental strain showed alterations in cell corpse numbers. For P values of <0.05, the difference in the means between the two groups was regarded as significant. For P values of <0.01, the difference was regarded as highly significant.

Obtaining nuclear counts:
The number of nuclei in homozygous deficiency embryos was counted by the squash method (GOSSETT and HECHT 1980 ) with some modification. Briefly, Df/+ adults were placed on NGM agar plates with Escherichia coli. After several hours, the adults were removed from the plates. The eggs laid on the plates were allowed to develop for >14 hr. The 5–10 deficiency homozygous embryos that remained unhatched were recovered from the plates with M9 solution and placed on polylysine-coated slides. A silicone-treated coverslip was laid over the embryos and the slide was inverted onto a tissue and pressed to squash the embryos. The slide was immediately frozen on dry ice. After removing the coverslip by prying off with a razor blade, the slide was fixed in methanol for 5 min and then acetone for 5 min at -20°. The slide was washed in PBS containing 0.5% Tween 20 and mounted in a 25 mg/ml DABCO solution in 90% glycerol:10% PBS with 2 µg/ml 4',6-diamidino-2-phenylindole (DAPI) and sealed with nail polish. The stained nuclei were viewed under a fluorescent microscope and photographed. The number of nuclei per embryo was counted from an image projected from the negative using a slide projector.

Scoring duration of cell corpses:
A four-dimensional (4-D) time-lapse recording system (SCHNABEL 1991 ; HIRD and WHITE 1993 ; MOSKOWITZ et al. 1994 ) was used to score the duration of cell corpse persistence. The young embryos from Df/+ adults were collected and mounted on a 5% agar pad as described above. Using the 4-D system, images of 25 focal planes at 1-min intervals were collected for 10 hr. From the recording, the duration of each corpse was scored. We scored all the corpses readily visible in the recording; however, the analysis of some corpses (particularly those located at the edge of the embryos, or in a deep focal plane in the recorded image) were excluded due to optical limitations.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Deficiency screen for genomic regions required for normal PCD:
Most previous screens for PCD defects in C. elegans have required embryonic viability of the mutant animals. We chose to perform an unbiased screen for zygotic genes involved in PCD by examining embryos homozygous for deficiencies that sample most of the genome. As PCD occurs in cells shortly after their birth, and the corpses are rapidly engulfed, little information can be obtained regarding alterations in PCD based on the appearance of terminally arrested embryos. To screen deficiency homozygotes for defects in PCD, we followed the distribution of cell corpses over the course of embryogenesis. Of the 671 cells born during embryogenesis, 113 undergo PCD (SULSTON et al. 1983 ). The first cell death occurs at 220 min after first cleavage (250-cell stage) and the last embryonic cell deaths at 630 min (560-cell stage). Most deaths (109/113) occur between 250 and 450 min of development, the period during which we focused our observations.

We scored cell corpses at four developmental stages: stage I (340 min after the first cleavage, bean morphology in the wild type), stage II (380 min, comma), stage III (420 min, 1.5-fold) and stage IV (460 min, 2-fold). In addition, we scored terminally arrested embryos for each deficiency homozygote at 800 min after first cleavage. In wild-type (N2) embryos, the number of corpses seen at each of these stages is relatively reproducible. The average number of cell corpses ± standard deviation was as follows: stage I, 2.2 ± 1.4 (n = 41); stage II, 4.4 ± 1.6 (n = 45); stage III, 3.6 ± 1.1 (n = 28); stage IV, 2.8 ± 1.1 (n = 41). Generally no corpses are visible at the time of hatching in wild type (800 min). The rapid engulfment and degradation of cell corpses (over a period of <20 min) are responsible for the low number of cell corpses visible at any one time.

To examine the profile of cell corpses in deficiency homozygotes, embryos were collected from hermaphrodite adults that were heterozygous for each deficiency. Approximately 25% of these embryos are deficiency homozygotes (Df/Df) and arrest during embryogenesis owing to the removal of essential embryonic genes; the remaining 75% (Df/+ and +/+) are viable. The numbers of corpses in homozygous deficiency embryos were compared with those of their siblings and of wild type and tested for significant deviations.

Fifty-eight deficiencies were examined and divided into four classes based on the number of corpses observed (Table 2). Class I deficiencies (20 total) showed normal cell corpse numbers at all stages. Class II deficiencies (16 total) showed no or significantly fewer cell corpses than wild type. Class III deficiencies (18 total) showed excess numbers of corpses. Class IV deficiencies (4 total) arrested early in embryonic proliferation and showed abnormally large corpses. The deficiencies that led to significant PCD phenotypes are described here.

Deficiencies that delete known cell death genes:
Several known genes that function in cell death are deleted by the deficiencies used in this screen. ced-3, ced-4, and egl-1 are required for PCD and loss-of-function mutations in these genes result in an absence of cell death throughout development (ELLIS and HORVITZ 1986 ; CONRADT and HORVITZ 1998 ). As expected, the deficiencies that delete ced-3 (nDf27 and sDf21) completely block PCD (Fig 1). The other region whose removal completely eliminates PCD is defined by the overlapping deficiencies itDf2 and nDf42 and excludes the region deleted by the neighboring deficiencies arDf1, yDf8, yDf4, and yDf6, which do not block PCD. These observations narrow the relevant region to a small interval on chromosome V. Following our identification of this region, it was reported that loss-of-function mutations in the egl-1 gene, which resides in this interval, prevent PCD (CONRADT and HORVITZ 1998 ). Thus, it is likely that the absence of cell death in itDf2 and nDf42 is caused by removal of egl-1. These two examples of class II deficiencies validate our screen as a means to identify genomic regions involved in PCD.

View larger version (49K): In this window In a new window Download PPT slide   Figure 1. Cell death profiles for class II deficiencies. Average number of cell corpses counted per embryo at each developmental stage is shown for each deficiency strain. Hatched bar, N2 (wild type); stippled bar, +/+ or +/Df embryos; solid bar, Df/Df embryos. (I) 340 min after first cleavage (bean stage in wild type); (II) 380 min (comma stage in wild type); (III) 420 min (1.5-fold in wild type); (IV) 460 min (2-fold in wild type); (T: terminally arrested embryo) >800 min (the approximate time at which wild-type embryos hatch). For the terminal stages, cell corpses were scored only in deficiency homozygotes. More than 30 embryos from Df/+ mothers were scored for each deficiency, and the average for each category [(+/+ or +/Df) and (Df/Df)] is shown. The average numbers of corpses between the two categories were compared by t-test. For data marked with one asterisk, 0.01 P < 0.05; for those with two asterisks, P < 0.01. For overlapping deficiencies that delete common relevant regions, only one representative deficiency is shown.

A limitation of the deficiency screen is that it does not detect maternally rescued genes, as exemplified by ced-4 and ced-9. Although stDf121 deletes ced-4, homozygous embryos, unlike ced-4(lf) mutants (ELLIS and HORVITZ 1986 ; YUAN and HORVITZ 1992 ) undergo PCD (Fig 2A). We found that this discrepancy is attributable to maternal rescue of the homozygotes, as revealed in the genetic experiments presented in Table 3. Similarly, while ced-9 is required to suppress cell death, the ced-9(lf) homozygotes produced by heterozygous mothers develop normally until late in postembryonic development, owing to maternal supply of the ced-9(+) product (HENGARTNER et al. 1992 ). Therefore, although nDf40 (see below), which deletes ced-9, leads to excess numbers of cell corpses, this phenotype is likely not attributable to removal of ced-9.

View larger version (35K): In this window In a new window Download PPT slide   Figure 2. Cell death profiles for class III deficiencies. (A) Class III-1: cell corpses accumulated toward the end of embryogenesis. (B) Class III-2: excess corpses were observed at specific developmental stages. Data presented as described in Fig 1.

All described engulfment mutants, with the exception of ced-1, are maternally rescued (ELLIS et al. 1991 ). Since there was no available deficiency that deletes ced-1, we were unable to evaluate the efficacy of our screen in identifying engulfment genes. However, we found that several of the deficiencies we analyzed are profoundly defective in cell corpse engulfment (see DISCUSSION).

Some class II deficiencies suggest a link between PCD and cell cycle progression:
As described above, deletion of the two genomic regions that contain ced-3 and egl-1 result in the complete absence of PCD. All other class II deficiencies cause a partial reduction of cell corpse number (Fig 1). For some of the deficiencies, a diminution in cell corpse number appeared to result from mitotic arrest. Four overlapping deficiencies (sDf45, sDf33, sDf28, and sDf42) cause a profound decrease in cell corpse number at all stages. All four caused embryos to arrest with <200 nuclei, compared to the 560 nuclei present in wild-type embryos at hatching (Table 2); PCD does not occur before the 200-cell stage in wild-type embryos. (The number of PCDs that have been generated by various stages in wild-type development are noted in Table 2). By contrast, three nearby deficiencies (sDf74, sDf26, and sDf50) did not result in decreased cell corpse number or reduced number of nuclei. Thus, the PCD and cell proliferation phenotypes are closely linked and are not separable by the deficiencies that we examined. One of the genes in the relevant region is emb-29, mutations in which result in arrest of embryos with 150–200 differentiated cells (HECHT et al. 1987 ), suggesting that the arrest in proliferation is due to loss of emb-29. To test whether the absence of PCD in these deficiencies could be explained by the mitotic block caused by loss of emb-29, we examined the pattern of cell corpses in emb-29 mutants. Embryos carrying any of three mutant alleles of emb-29 (s819, g52ts, and b262ts), invariably arrested with <200 cells and showed depressed cell corpse numbers comparable to that seen with the deficiencies that delete emb-29. Although most cells in wild-type 200-cell stage embryos do not show obvious differentiation, arrested emb-29 embryos and deficiencies that delete emb-29 contain a number of differentiated cell types (e.g., intestine, neurons, and muscles); thus, differentiation is not blocked in these embryos despite the early arrest of cell proliferation. Therefore, the diminished cell corpse number in these deficiencies is likely to be explained by the mitotic block. EMB-29 may be directly required both for mitotic progression and activation of the cell death program; alternatively, it may not be possible to initiate PCD from the particular stage in the cell cycle at which emb-29 mutants arrest.

Embryos homozygous for qDf3, eDf19, stDf5, and sDf23 showed fewer corpses at all stages. stDf5 maps near the ced-8 gene, mutations in which cause cell corpses to appear later than they did in wild type (STANFIELD and HORVITZ 2000 ). However, we found that ced-8(n1891) complements the cell death phenotype of stDf5, indicating that the reduced number of cell corpses is not due to deletion of ced-8. qDf3 and eDf19 homozygotes arrest with slightly fewer (420–430) nuclei than are present in fully developed embryos. This partial block in cell proliferation may contribute to the reduction in cell corpse number in these deficiencies. However, based on their morphology, these embryos undergo differentiation and the reduction of corpses cannot, therefore, be explained by a general failure to differentiate.

Four deficiencies (mnDf63, syDf1, hDf6, and tDf3) lead to stage-specific decreases in cell corpse number. Most notably, tDf3 embryos showed fewer corpses during stages II and III, while corpses accumulated in the arrested embryos. This apparent delay in the production of corpses may be caused by the delayed execution of cell death in these embryos, as seen in ced-8 mutant embryos.

Possible involvement of some, but not all, Ras pathway components in embryonic PCD:
Among the class II deficiencies that produce diminished numbers of corpses, two delete genes of the ras signaling pathway: stDf5 deletes ksr-1 (KORNFELD et al. 1995B ; SUNDARAM and HAN 1995 ), and tDf3 deletes mek-2 (CHURCH et al. 1995 ; KORNFELD et al. 1995A ; WU et al. 1995 ; Fig 3). The Ras signaling pathway has been implicated in the regulation of apoptotic cell death in cultured mammalian cells (reviewed by PRITCHARD and MCMAHON 1997 ), and an activated form of Ras causes hyperplasia and increased cell death in a process that is dependent on the mitogen-activated protein kinase (MAPK) cascade in Drosophila (KARIM and RUBIN 1998 ). In C. elegans, germline cell death requires activation of the Ras/MAPK pathway (GUMIENNY et al. 1999 ). Unlike somatic cells, germ cells are produced by indeterminant rounds of cell division and PCD in the germline is subject to different control mechanisms than in somatic cells. It is not known whether the ras pathway is involved in PCD of cells arising from a defined cell lineage.

View larger version (31K): In this window In a new window Download PPT slide   Figure 3. Map positions of deficiencies screened. Chromosomes are indicated by horizontal bars, the central gene clusters by black bars. The positions of known cell death genes and other genes relevant to this study are indicated above each chromosome. The approximate location of each deficiency is indicated below its corresponding chromosome. The pattern of the horizontal line representing each deficiency is coded according to the cell death phenotype it exhibited (see annotations in the figure). yDf4 and yDf6 are coextensive based on available map data, yet they exhibit different phenotypes; however, since the precise endpoints of these deficiencies have not been determined, it is certainly reasonable to suppose that the latter removes genes that the former does not.

To investigate the possible role of Ras pathway components in embryonic PCD, we scored the profile of cell corpses during embryogenesis in ksr-1 mutants. Mutations in ksr-1 suppress the multivulva phenotype caused by activated let-60 ras but have no apparent phenotype in a wild-type background (KORNFELD et al. 1995B ; SUNDARAM and HAN 1995 ). We found that 20% of the ksr-1 (ku68) embryos arrest between the 2-fold and early pretzel stages and these arrested embryos contained fewer cell corpses than did wild-type or the ksr-1 embryos that did not arrest (Fig 4). By contrast, we found that let-60 ras mutants did not show dramatically altered developmental cell corpse profiles. Thus, while some components of the ras pathway may affect embryonic PCD, we could not obtain evidence for involvement of the entire pathway; this is likely to reflect a fundamental difference between the regulation of PCD in the germline and in somatic cells.

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. Cell death profiles for Ras pathway mutants. For ksr-1(ku68), arrested embryos (solid bars) are compared with embryos that developed normally (stippled bars). For let-60(n2031), homozygotes are compared with +/+ or +/let-60 (n2031). Asterisks represent P values, as described in Fig 1.

Stage-specific elevation of cell corpse number in class III deficiencies:
Thirteen deficiencies result in abnormally elevated cell corpse numbers at one or more stages. We further subdivided these deficiencies into two subclasses based on their temporal effects. In class III-1 deficiencies (sDf26, nDf40, mnDf1, sDf121, and mnDf88), the number of corpses was elevated particularly in later and terminal embryos (Fig 2). Of these, sDf26 and nDf40 showed the strongest effect: the number of corpses at stage III and IV was approximately three times higher than that in wild type. By contrast, mnDf88 showed normal corpse numbers through stage IV, but a large number were seen in terminal embryos.

As mentioned earlier, nDf40 deletes ced-9, a cell death inhibitor. The maternal supply of CED-9 is sufficient to suppress PCD during embryogenesis, and homozygous ced-9(-) progeny of ced-9(-)/+ mothers live and become sterile adults (HENGARTNER et al. 1992 ). Therefore, the excess cell corpses observed in nDf40 embryos are unlikely to be caused by increased PCD resulting from the lack of ced-9 function. To examine further whether the elevated cell corpse numbers result from increased numbers of dying cells or accumulation of corpses due to a defect in their engulfment, we scored for the persistence of individual cell corpses in 4-D time-lapse video images of recorded embryos (SCHNABEL 1991 ; HIRD and WHITE 1993 ; MOSKOWITZ et al. 1994 ). In wild-type embryos, most corpses remain visible for <30 min (average 20.2 min). By contrast, in nDf40 embryos, many corpses persisted for >30 min (Fig 5). The engulfment defect in nDf40 homozygotes was more severe after stage II: while only one-third (11 out of 33) of the corpses persisted longer than 50 min prior to stage II, all 20 corpses remained for such an extended period after stage II. This observation explains why embryos homozygous for nDf40 contained a normal number of corpses up until stage II, with a striking increase in corpse number thereafter. The defect in cell corpse engulfment can at least partially explain the elevated corpse numbers seen during embryogenesis, further supporting the view that removal of ced-9 is not responsible for this phenotype.

View larger version (31K): In this window In a new window Download PPT slide   Figure 5. Analysis of cell corpse duration in nDf40. Embryos homozygous for the indicated deficiency were recorded throughout embryogenesis by 4-D time lapse. All visible corpses from the recorded images were scored for their duration and histograms of their persistence time are shown.

In class III-2 deficiencies (sDf74, nDf24, mnDf30, rhDf1, sDf35, ozDf2, qDf16, and eDf3), cell corpse numbers are transiently elevated, but decrease in terminal embryos (Fig 2). sDf74 shows the most drastic effect: sDf74 homozygotes contain more corpses than do wild type even at stage I, and the numbers progressively increase through stage IV, reaching a peak average of 18.2 corpses per embryo; the number then drops to only 1.4 corpses in terminal embryos. By contrast, nDf24 homozygotes show almost normal numbers until stage III; the number rapidly increased to an average of 14.5 corpses at stage IV, diminishing to an average of 8.3 in terminal embryos. In qDf16 and eDf3 embryos, there was a slight elevation in cell corpse number during early stages, but no corpses were observed in terminal embryos.

Possible maternal effects of class III' deficiencies:
Although the deficiency screen was designed to find regions required zygotically, we found three regions (yDf6, the overlapping region defined by nDf30 and hDf9, and the overlapping region defined by ctDf1 and arDf1) that, when deleted, show phenotypes suggesting a maternal requirement for normal PCD. In all three cases, embryos produced by heterozygous mothers show a higher number of corpses at each stage compared to wild type, even though only 25% were expected to be homozygous for the deficiency (Fig 6). In fact, arrested (Df/Df) and viable (Df/+ or +/+) embryos showed similar numbers of cell corpses. The elevated cell corpse numbers seen in all embryos derived from such heterozygotes may be attributable to haploinsufficiency of the gene(s) responsible for the cell death phenotype: heterozygous mothers may contribute insufficient levels of gene product to the zygote to provide for normal function. An alternative explanation, i.e., that the strains used in the screen were homozygous for an unknown mutation responsible for the phenotype, is unlikely, at least in the case of the regions defined by overlapping deficiencies: it is improbable that independently isolated deficiencies carry additional mutations that act similarly. Instead, it seems likely that these regions include genes that are maternally required in the cell death process.

View larger version (40K): In this window In a new window Download PPT slide   Figure 6. Cell death profiles for class III' deficiencies. Data presented as described in Fig 1.

Class IV deficiencies suggest that cell division and cell death can be uncoupled:
Embryos homozygous for eDf2, nDf41, ccDf5, and maDf4, defining at least three genomic regions, showed prominent corpses that were abnormally large (Fig 7). In all cases, the number of corpses was greatly reduced compared to wild type (Fig 8). In addition, the arrested embryos contained fewer nuclei than the 558 of fully developed wild-type embryos: eDf2 embryos arrested with 420 cells, nDf41 with 180, and ccDf5 and maDf4 with <140. In the case of at least the two overlapping deficiencies, ccDf5 and maDf4, the large cell corpse size is probably attributable to the programmed death of abnormally early (therefore larger) cells. Such a conclusion is consistent with the observation that ccDf5 and maDf4, which arrested with the lowest number of nuclei, exhibited exclusively large corpses. Corpses in eDf2 and nDf41 embryos, though generally larger than those in wild type, were occasionally normal in size. These latter two deficiencies result in multinucleate hypodermal cells, suggesting defects in cytokinesis (TERNS et al. 1997 ; M. FURUYA and A. SUGIMOTO, unpublished observation). Moreover, eDf2 deletes cyk-4, and nDf41 deletes zen-4; loss-of-function mutations in either of these genes lead to multinucleate cells and both are required for completion of cytokinesis (POWERS et al. 1998 ; RAICH et al. 1998 ; JANTSCH-PLUNGER et al. 2000 ; M. FURUYA and A. SUGIMOTO, unpublished data). Thus, the large corpses in eDf2 and nDf41 are likely to be caused by the death of these multinucleate cells. However, we did not observe multinucleate cells in embryos homozygous for ccDf5 and maDf4. The first PCD in wild-type C. elegans embryos occurs when there are 250 cells (SULSTON et al. 1983 ) and most arise after the 400-cell stage. Our finding that embryos containing as few as 140 nuclei can nonetheless give rise to several cell corpses implies that activation of the cell death program is not absolutely dependent on the normal number of rounds of mitosis during embryogenesis.

View larger version (116K): In this window In a new window Download PPT slide   Figure 7. Examples of cell death phenotypes seen in deficiency homozygotes. Nomarski image of a representative embryo from each class is shown. Cell corpses are indicated by arrowheads. (A) Wild-type embryo at the 1.5-fold stage. (B) Terminal itDf2 homozygote (class II). No cell corpses are visible. (C) Terminal stDf74 homozygote (class III). Many corpses are apparent. (D) Terminal nDf41 homozygote (class IV). Abnormally large corpses are visible.

View larger version (19K): In this window In a new window Download PPT slide   Figure 8. Cell death profiles for class IV deficiencies. Data presented as described in Fig 1.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

New regions required for embryonic PCD:
As an approach for identifying new genes required zygotically (and in some cases maternally), a major advantage of this deficiency screen is that extensive genomic regions, and therefore a large number of genes, can be screened simultaneously. From a screen of 58 deficiencies, covering an estimated 74% of the genome, we identified many genomic regions involved in embryonic PCD. At least 30 regions required for the normal number or appearance of embryonic PCDs were found, nearly all of which (28) correspond to genomic segments in which no known cell death genes reside. These results suggest that, although extensive screens for point mutations leading to abnormal PCD have been performed (e.g., ELLIS and HORVITZ 1986 ; ELLIS et al. 1991 ; HENGARTNER et al. 1992 ; METZSTEIN et al. 1996 ; CONRADT and HORVITZ 1998 ; METZSTEIN and HORVITZ 1999 ), many cell death genes have yet to be identified and characterized in C. elegans.

The failure of past genetic screens to identify the relevant genes in the regions revealed by this study might be attributable to the lower sensitivity of the assay for aberrant PCD and/or the bias toward viable mutations in the screens for point mutations. By characterizing cell corpse numbers at multiple times during embryogenesis, our assay was sensitive to transient alterations in PCD, enabling us to identify subtle abnormalities (such as the stage-specific reduction, or increase, in corpse number), which might be missed by previous screening methods. In addition, all PCD-related genes previously reported in C. elegans were originally identified from viable mutants (ELLIS and HORVITZ 1986 ; ELLIS et al. 1991 ; CONRADT and HORVITZ 1998 ). Even the ced-9 gene, for which loss-of-function alleles produce a maternal lethal phenotype, was originally identified from a viable gain-of function allele (HENGARTNER et al. 1992 ). The deficiency screen is not biased toward genes that mutate to viability and it is therefore likely that some of the relevant genes in the regions we identified may be essential for viability.

New genomic regions required for PCD:
Many of the regions identified in the screen may contain previously unidentified genes that are directly involved in controlling PCD.

Deletion of nine distinct genomic regions (12 deficiencies) not containing a known PCD gene resulted in reduction of cell corpse number (class II). The class IV deficiencies (three regions; 4 deficiencies) also resulted in fewer corpses. However, none of these deficiencies completely eliminated PCD; thus, in the genomic regions we investigated there appear to be no genes that are zygotically essential for all PCD. It is possible, therefore, that all genes that are zygotically essential for PCD in C. elegans have been identified.

The deficiencies that delete emb-29 (sDf45, sDf33, sDf28, and sDf42) and the class IV deficiencies (eDf2, ccDf5/maDf4, and nDf41) resulted in arrest with reduced numbers of nuclei; this premature arrest is likely to contribute to, but may not be entirely responsible for, the general decrease in corpse number. For those deficiencies that do not decrease nuclear numbers, the observed decrease in PCD might be the result of defects in mechanisms that determine the fate of cells destined for death. To date, only two genes, ces-1 and ces-2, are known to be involved in such cell death determination processes in C. elegans (METZSTEIN et al. 1996 ; METZSTEIN and HORVITZ 1999 ). Both genes regulate a small subset of PCDs. The genomic regions defined by the deficiencies that result in normal cell numbers and reduced PCD may provide a starting point to identify other ces-like genes that determine the death fate of particular cells.

The removal of 13 genomic regions results in the excess accumulation of corpses. Five deficiencies (mnDf88, sDf121, nDf40, sDf26, and mnDf1) cause excess accumulation of corpses that persisted to the end of embryogenesis, while eight (qDf16, nDf24, eDf3, mnDf30, rhDf1, sDf74, sDf35, and ozDf2) led to a transient excess during embryogenesis. The excess corpse number may be the result of increased numbers of dying cells [as in ced-9(lf) mutants] or of defects in engulfment of corpses (as in the many known engulfment-defective mutants; HEDGECOCK et al. 1983 ; ELLIS et al. 1991 ). Our temporal analysis revealed that nDf40 is defective in engulfment (Fig 5). Similar analysis of two other class III deficiencies, sDf74 and rhDf1, revealed that these also lead to a severe defect in engulfment (our unpublished observations), suggesting that most, and possibly all, of the class III deficiencies might be engulfment mutants. Additional analyses of cell corpse persistence with the other class III deficiencies will reveal whether the increased cell corpse number is caused by excessive PCD or defective engulfment.

Although the deficiencies described here exhibited PCD defects, we cannot exclude the possibility that the aberrant numbers of PCD corpses are caused by dramatic changes in the embryonic cell lineage. In such a case, the relevant genes might play no role in the cell death program per se. Analyses of cell lineages will be required to determine whether the altered PCD numbers are the result of direct defects in PCD or of dramatic transformations in embryonic cell fates.

Embryonic PCD and Ras signaling:
In wild-type C. elegans, female germ cells undergo PCD near the region where cells exit the pachytene stage. It has been shown that the members of the Ras signaling pathway [let-60 (ras), lin-45 (raf), mek-2 (MAPKK), and mpk-1 (MAPK)] are required for exit from pachytene arrest, as well as for PCD in the female germline (CHURCH et al. 1995 ). Thus, it was suggested that the Ras pathway might directly or indirectly regulate germline PCD. We have shown that a ksr-1 mutant, and deficiencies that delete ksr-1 or mek-2, depress PCD during embryogenesis. KSR-1 is a conserved Raf-related kinase and has been implicated as a positive regulator in the Ras signaling pathway, functioning in parallel to, or downstream of, Ras (KORNFELD et al. 1995B ; SUNDARAM and HAN 1995 ). Our results raise the possibility that this component of the Ras signaling pathway might participate not only in regulating PCD of female germ cells, but also in embryonic PCD. However, we were unable to find evidence that ras itself affects embryonic PCD, perhaps indicating that some, but not all, components in this pathway function in the embryo to regulate cell death.

Cell cycle and PCD:
A number of studies have suggested a link between the regulatory mechanisms controlling cell proliferation and PCD (for review, GUO and HAY 1999 ; KING and CIDLOWSKI 1998 ; LUNDBERG and WEINBERG 1999 ). For example, some cell cycle regulators also control apoptosis. The p53 tumor suppressor promotes either cell cycle arrest or apoptosis in response to DNA damage (LEVINE 1997 ). Survivin, an inhibitor of apoptosis (IAP) protein, is a cell cycle-regulated caspase inhibitor, suggesting that caspases may be active normally during particular phases of the cell cycle (LI et al. 1998 ). However, IAP homologues of C. elegans have not been implicated in PCD (FRASER et al. 1999 ) and the link between cell proliferation and PCD in C. elegans is not well understood.

Previous analysis of lin-5 mutants demonstrated that nuclear deaths occur even when cell division does not occur (ALBERTSON et al. 1978 ). In lin-5 mutants, nuclear and cell divisions are blocked during postembryonic development, although DNA replication occurs. Despite the block in cytokinesis, the correct timing of developmental processes is retained, and cells exhibit varying characteristics of the differentiated descendants they would normally generate, including PCD. In the ventral nerve cord of lin-5 larva, multinucleate cells were found to undergo programmed death. We have found that embryos homozygous for eDf2 and nDf41 similarly generate multinucleate cells that undergo PCD, probably accounting for the large corpse phenotype.

Our deficiency screen contributed two additional findings regarding the relationship between regulation of the cell cycle and PCD in C. elegans. First, we found that the program controlling the proper number of cell division rounds and that controlling PCD are separable. Two class IV deficiencies (ccDf5 and maDf4) result in arrest with far fewer than the normal number of nuclei, do not appear to be defective in cytokinesis, and yet do not block PCD. Presumably, the regulatory machinery that normally activates PCD in particular lineages during embryogenesis is able to function even when too few cell divisions have occurred. This might indicate that PCD-promoting regulators are activated in response to a clock that is dependent upon absolute time rather than on rounds of cell division. By contrast, we also found that some specific cell cycle defects can apparently prevent cell death. The emb-29 mutants and embryos homozygous for deficiencies that delete emb-29 arrest with <200 nuclei but, unlike class IV deficiencies, few or no corpses were observed in these mutants.

One possible explanation for these contrasting phenotypes might be a difference in the stage of the cell cycle at which arrest occurs. Cells in emb-29 mutants and in homozygous deficiencies that delete emb-29 arrest at M phase (HECHT et al. 1987 ; our unpublished data). By contrast, cells of the class IV deficiencies appeared not to arrest at M phase, as judged by DAPI staining of chromosomes (data not shown). These observations suggest that it might not be possible to initiate PCD at M phase. While it is apparent that PCD might be regulated in a cell cycle-specific manner, the embryos lacking emb-29 function differentiate various cell types, including neurons, muscles (twitching), and intestine, implying that M-phase arrest does not prevent all aspects of cellular differentiation. It will be interesting to identify the genes that simultaneously block both PCD and cell proliferation, in hopes of learning which cell cycle regulatory functions are also required for initiation of the apoptotic program.

	FOOTNOTES

¹ Present address: Department of Biochemistry and Biophysics, Graduate School of Science, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113 Japan and PRESTO, Japan Science and Technology Corporation.
² Present address: Department of Biochemistry and Biophysics, University of California, San Francisco, CA 94143.

	ACKNOWLEDGMENTS

We are grateful to the Caenorhabditis Genetics Center (Theresa Stiernagle) and to Ralph Hecht, David Baillie, and Andy Fire for strains. Some of the strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the National Center for Research Resources of the National Institutes of Health (NIH). We thank Richard Halberg for initial assistance with analyses of deficiencies. A.S. was supported by a Naito Foundation Fellowship and a Leukemia Society of America Special Fellowship. R.H. was supported by a Cell and Molecular Biology predoctoral training grant from the NIH. W.B.D. was supported by a Cancer Foundation of Santa Barbara postdoctoral fellowship. This work was supported by grants from the NIH (GM-48137 and AG-13736), a Searle Scholars Award from the Chicago Community Trust, and a Shaw Scientists Award from the Milwaukee Foundation to J.H.R.

Manuscript received October 31, 2000; Accepted for publication February 12, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AHNN, J. and A. FIRE, 1994  A screen for genetic loci required for body-wall muscle development during embryogenesis in Caenorhabditis elegans.. Genetics 137:483-498[Abstract].

ALBERTSON, D. G., J. E. SULSTON, and J. G. WHITE, 1978  Cell cycling and DNA replication in a mutant blocked in cell division in the nematode Caenorhabditis elegans. Dev. Biol. 63:165-178[Medline].

BILDER, D. and M. P. SCOTT, 1995  Genomic regions required for morphogenesis of the Drosophila embryonic midgut. Genetics 141:1087-1100[Abstract].

BRENNER, S., 1974  The genetics of Caenorhabditis elegans.. Genetics 77:71-94[Abstract/Free Full Text].

CHANAL, P. and M. LABOUESSE, 1997  A screen for genetic loci required for hypodermal cell and glial-like cell development during Caenorhabditis elegans embryogenesis. Genetics 146:207-226[Abstract].

CHEN, P., W. NORDSTROM, B. GISH, and J. M. ABRAMS, 1996  grim, a novel cell death gene in Drosophila. Genes Dev. 10:1773-1782[Abstract/Free Full Text].

CHINNAIYAN, A. M., D. CHAUDHARY, K. O'ROURKE, E. V. KOONIN, and V. M. DIXIT, 1997a  Role of CED-4 in the activation of CED-3. Nature 388:728-729[Medline].

CHINNAIYAN, A. M., K. O'ROURKE, B. R. LANE, and V. M. DIXIT, 1997b  Interaction of CED-4 with CED-3 and CED-9: a molecular framework for cell death. Science 275:1122-1126[Abstract/Free Full Text].

CHURCH, D. L., K. L. GUAN, and E. J. LAMBIE, 1995  Three genes of the MAP kinase cascade, mek-2, mpk-1/sur-1 and let-60 ras, are required for meiotic cell cycle progression in Caenorhabditis elegans. Development 121:2525-2535[Abstract].

CONRADT, B. and H. R. HORVITZ, 1998  The C. elegans protein EGL-1 is required for programmed cell death and interacts with the Bcl-2-like protein CED-9. Cell 93:519-529[Medline].

DEL PESO, L., V. M. GONZALEZ, and G. NUNEZ, 1998  Caenorhabditis elegans EGL-1 disrupts the interaction of CED-9 with CED-4 and promotes CED-3 activation. J. Biol. Chem. 273:33495-33500[Abstract/Free Full Text].

ELLIS, H. M. and H. R. HORVITZ, 1986  Genetic control of programmed cell death in the nematode C. elegans. Cell 44:817-829[Medline].

ELLIS, R. E., D. M. JACOBSON, and H. R. HORVITZ, 1991  Genes required for the engulfment of cell corpses during programmed cell death in Caenorhabditis elegans.. Genetics 129:79-94[Abstract].

FRASER, A. G., C. JAMES, G. I. EVAN, and M. O. HENGARTNER, 1999  Caenorhabditis elegans inhibitor of apoptosis protein (IAP) homologue BIR-1 plays a conserved role in cytokinesis. Curr. Biol. 9:292-301[Medline].

GOSSETT, L. A. and R. M. HECHT, 1980  A squash technique demonstrating embryonic nuclear cleavage of the nematode Caenorhabditis elegans. J. Histochem. Cytochem. 28:507-510[Abstract].

GRETHER, M. E., J. M. ABRAMS, J. AGAPITE, K. WHITE, and H. STELLER, 1995  The head involution defective gene of Drosophila melanogaster functions in programmed cell death. Genes Dev. 9:1694-1708[Abstract/Free Full Text].

GUMIENNY, T. L., E. LAMBIE, E. HARTWEIG, H. R. HORVITZ, and M. O. HENGARTNER, 1999  Genetic control of programmed cell death in the Caenorhabditis elegans hermaphrodite germline. Development 126:1011-1022[Abstract].

GUO, M. and B. A. HAY, 1999  Cell proliferation and apoptosis. Curr. Opin. Cell. Biol. 11:745-752[Medline].

HECHT, R. M., M. BERG-ZABELSHANSKY, P. N. RAO, and F. M. DAVIS, 1987  Conditional absence of mitosis-specific antigens in a temperature-sensitive embryonic-arrest mutant of Caenorhabditis elegans.. J. Cell Sci. 87:305-314[Abstract/Free Full Text].

HEDGECOCK, E. M., J. E. SULSTON, and J. N. THOMSON, 1983  Mutations affecting programmed cell deaths in the nematode Caenorhabditis elegans. Science 220:1277-1279[Abstract/Free Full Text].

HENGARTNER, M. O. and H. R. HORVITZ, 1994  C. elegans cell survival gene ced-9 encodes a functional homolog of the mammalian proto-oncogene bcl-2. Cell 76:665-676[Medline].

HENGARTNER, M. O., R. E. ELLIS, and H. R. HORVITZ, 1992  Caenorhabditis elegans gene ced-9 protects cells from programmed cell death. Nature 356:494-499[Medline].

HIRD, S. N. and J. G. WHITE, 1993  Cortical and cytoplasmic flow polarity in early embryonic cells of Caenorhabditis elegans. J. Cell Biol. 121:1343-1355[Abstract/Free Full Text].

HODGKIN, J., H. R. HORVITZ, and S. BRENNER, 1979  Nondisjunction mutants of the nematode C. elegans.. Genetics 91:67-94[Abstract/Free Full Text].

JACOBSON, M. D., M. WEIL, and M. C. RAFF, 1997  Programmed cell death in animal development. Cell 88:347-354[Medline].

JANTSCH-PLUNGER, V., P. GONCZY, A. ROMANO, H. SCHNABEL, and D. HAMILL et al., 2000  CYK-4: A Rho family gtpase activating protein (GAP) required for central spindle formation and cytokinesis. J. Cell Biol. 149:1391-1404[Abstract/Free Full Text].

KARIM, F. D. and G. M. RUBIN, 1998  Ectopic expression of activated Ras1 induces hyperplastic growth and increased cell death in Drosophila imaginal tissues. Development 125:1-9[Abstract].

KING, K. L. and J. A. CIDLOWSKI, 1998  Cell cycle regulation and apoptosis. Annu. Rev. Physiol. 60:601-617[Medline].

KORNFELD, K., K. L. GUAN, and H. R. HORVITZ, 1995a  The Caenorhabditis elegans gene mek-2 is required for vulval induction and encodes a protein similar to the protein kinase MEK. Genes Dev. 9:756-768[Abstract/Free Full Text].

KORNFELD, K., D. B. HOM, and H. R. HORVITZ, 1995b  The ksr-1 gene encodes a novel protein kinase involved in Ras-mediated signaling in C. elegans. Cell 83:903-913[Medline].

LABOUESSE, M., 1997  Deficiency screen based on the monoclonal antibody MH27 to identify genetic loci required for morphogenesis of the Caenorhabditis elegans embryo. Dev. Dyn. 210:19-32[Medline].

LEVINE, A. J., 1997  p53, the cellular gatekeeper for growth and division. Cell 88:323-331[Medline].

LEWIS, J. A., and J. T. FLEMING, 1995 Basic culture methods, pp. 3–29 in Caenorhabditis elegans: Modern Biological Analysis of an Organism, edited by H. F. EPSTEIN and D. C. SHAKES. Academic Press, San Diego.

LI, F., G. AMBROSINI, E. Y. CHU, J. PLESCIA, and S. TOGNIN et al., 1998  Control of apoptosis and mitotic spindle checkpoint by survivin. Nature 396:580-584[Medline].

LUNDBERG, A. S. and R. A. WEINBERG, 1999  Control of the cell cycle and apoptosis. Eur. J. Cancer 35:1886-1894.

MERRILL, P. T., D. SWEETON, and E. WIESCHAUS, 1988  Requirements for autosomal gene activity during precellular stages of Drosophila melanogaster. Development 104:495-509[Abstract/Free Full Text].

METZSTEIN, M. M. and H. R. HORVITZ, 1999  The C. elegans cell death specification gene ces-1 encodes a snail family zinc finger protein. Mol. Cell 4:309-319[Medline].

METZSTEIN, M. M., M. O. HENGARTNER, N. TSUNG, R. E. ELLIS, and H. R. HORVITZ, 1996  Transcriptional regulator of programmed cell death encoded by Caenorhabditis elegans gene ces-2. Nature 382:545-547[Medline].

METZSTEIN, M. M., G. M. STANFIELD, and H. R. HORVITZ, 1998  Genetics of programmed cell death in C. elegans: past, present and future. Trends Genet. 14:410-416[Medline].

MOSKOWITZ, I. P., S. B. GENDREAU, and J. H. ROTHMAN, 1994  Combinatorial specification of blastomere identity by glp-1-dependent cellular interactions in the nematode Caenorhabditis elegans. Development 120:3325-3338[Abstract].

POWERS, J., O. BOSSINGER, D. ROSE, S. STROME, and W. SAXTON, 1998  A nematode kinesin required for cleavage furrow advancement. Curr. Biol. 8(20):1133-1136[Medline].

PRITCHARD, C. and M. MCMAHON, 1997  Raf revealed in life-or-death decisions. Nat. Genet. 16:214-215[Medline].

RAICH, W. B., A. N. MORAN, J. H. ROTHMAN, and J. HARDIN, 1998  Cytokinesis and midzone microtubule organization in Caenorhabditis elegans require the kinesin-like protein ZEN-4. Mol. Biol. Cell 9(8):2037-2049[Abstract/Free Full Text].

SCHNABEL, R., 1991  Cellular interactions involved in the determination of the early C. elegans embryo. Mech. Dev. 34:85-99[Medline].

SESHAGIRI, S. and L. K. MILLER, 1997  Caenorhabditis elegans CED-4 stimulates CED-3 processing and CED-3-induced apoptosis. Curr. Biol. 7:455-460[Medline].

SMITH, A. V., J. A. KING, and T. L. ORR-WEAVER, 1993  Identification of genomic regions required for DNA replication during Drosophila embryogenesis. Genetics 135:817-829[Abstract].

SONG, Z. and H. STELLER, 1999  Death by design: mechanism and control of apoptosis. Trends Cell Biol. 9:M49-M52[Medline].

STANFIELD, G. M. and H. R. HORVITZ, 2000  The ced-8 gene controls the timing of programmed cell deaths in C. elegans. Mol. Cell 5:423-433[Medline].

STORFER-GLAZER, F. A. and W. B.