The genes egl-1, ced-9, ced-4, and ced-3 play major roles in programmed cell death in Caenorhabditis elegans. To identify genes that have more subtle activities, we sought mutations that confer strong cell-death defects in a genetically sensitized mutant background. Specifically, we screened for mutations that enhance the cell-death defects caused by a partial loss-of-function allele of the ced-3 caspase gene. We identified mutations in two genes not previously known to affect cell death, dpl-1 and mcd-1 (modifier of cell death). dpl-1 encodes the C. elegans homolog of DP, the human E2F-heterodimerization partner. By testing genes known to interact with dpl-1, we identified roles in cell death for four additional genes: efl-1 E2F, lin-35 Rb, lin-37 Mip40, and lin-52 dLin52. mcd-1 encodes a novel protein that contains one zinc finger and that is synthetically required with lin-35 Rb for animal viability. dpl-1 and mcd-1 act with efl-1 E2F and lin-35 Rb to promote programmed cell death and do so by regulating the killing process rather than by affecting the decision between survival and death. We propose that the DPL-1 DP, MCD-1 zinc finger, EFL-1 E2F, LIN-35 Rb, LIN-37 Mip40, and LIN-52 dLin52 proteins act together in transcriptional regulation to promote programmed cell death.
PROGRAMMED cell death is important for many aspects of animal development, including morphogenesis, homeostasis, and neuronal refinement (Glücksmann 1951; Saunders 1966; Jacobson et al. 1997). Studies of the mechanisms of programmed cell death in the nematode Caenorhabditis elegans have identified a pathway that is largely conserved in other organisms, including humans (Metzstein et al. 1998). Four genes—egl-1, ced-9, ced-4, and ced-3—regulate essentially all somatic programmed cell death and define the core metazoan cell-death execution machinery. EGL-1, which promotes cell death, is a BH3-only protein that binds to and inhibits the CED-9 protein (Conradt and Horvitz 1998). CED-9, which inhibits cell death (Hengartner et al. 1992), is similar to the human proto-oncoprotein BCL-2 (Hengartner and Horvitz 1994b; Yan et al. 2005) and localizes to mitochondria (Chen et al. 2000). CED-9 binds CED-4 (Spector et al. 1997), which localizes to mitochondria in a CED-9-dependent manner (Chen et al. 2000), promotes cell death (Ellis and Horvitz 1986), and is similar to the human pro-apoptotic protein APAF-1 (Yuan and Horvitz 1992; Zou et al. 1997). The expression of egl-1 or the loss of ced-9 function can trigger cell death and result in a change in the localization of CED-4 from mitochondria to the perinuclear region (Chen et al. 2000). CED-4 activates CED-3 (Shaham and Horvitz 1996b), which promotes cell death (Ellis and Horvitz 1986) and is a defining member of a family of cysteine proteases termed caspases (Yuan et al. 1993). CED-4 can interact directly with the CED-3 procaspase (Wu et al. 1997) and facilitate the processing of proCED-3 into active CED-3 (Chinnaiyan et al. 1997; Yang et al. 1998).
Other C. elegans genes appear to promote cell death more subtly. For example, the gene ced-8 XK affects the timing of programmed cell deaths in C. elegans and has a minor role in cell killing (Stanfield and Horvitz 2000). The gene ced-9, which inhibits programmed cell death, also can promote cell death (Hengartner and Horvitz 1994a). The gene cps-6 encodes a mitochondrial endonuclease G protein that likely promotes cell death (Parrish et al. 2001). The process of the engulfment of dying cells also promotes cell death (Hoeppner et al. 2001; Reddien et al. 2001). Additional genes that affect programmed cell death in C. elegans, such as those encoding proteins that mediate the ability of engulfment to promote cell death, that mediate the cell-killing activities of ced-8 or ced-9, or that act downstream of CED-3, remain to be identified. We reasoned that genes with subtle contributing roles in programmed cell death might be identified by screening for mutations that further decrease cell death in a genetic background in which cell death is partially impaired. Here we report the identification and characterization of genes we identified using this approach and present evidence that these genes control activities that promote programmed cell death.
MATERIALS AND METHODS
C. elegans was cultured at 20° on NGM agar with Escherichia coli OP50 as described (Brenner 1974). In general, the wild-type strain was N2. For mapping with polymorphisms, the wild-type strain RC301 was used. All mutations used have been described (Riddle et al. 1997), unless from this work or otherwise noted. The following mutations were used:
LGI: lin-61(n3446), dpy-5(e61), lin-35(n745), lin-35(n2239, n2242) (Lu and Horvitz 1998), unc-13(e1091), unc-29(e1072), lin-53(n833) (Lu and Horvitz 1998), unc-75(e950), ced-1(e1735, n3390, n3402) (Hedgecock et al. 1983), hT2 [qIs48] (Mathies et al. 2003), nIs128 (H. Schwartz and H. R. Horvitz, unpublished observations).
LGII: lin-8(n111), lin-8(n2731) (Thomas et al. 2003), eT1, dpy-10(e128), egl-27(n170), lin-56(n2728) (Thomas et al. 2003), rol-6(e187), dpl-1(n2994, n3316, n3380) (Ceol and Horvitz 2001), unc-4(e120), rol-1(e91), lin-38(n751), mcd-1(n3376, n4005) (this work), unc-52(e444), mIn1 [dpy-10(e128) mIs14] (Edgley and Riddle 2001), mnC1 [dpy-10 unc-52].
LGIII: qC1, unc-79(e1068), ced-4(n1162, n2273), dpy-17(e164), lin-37(n758), lin-36(n766), lin-9(n112), lin-13(n770), unc-32(e189), unc-16(e109), lin-52(n771), lin-52(n3718) (Thomas et al. 2003), ced-7(n3370, n3373, n3378, n3383, n3401, n3408, n3394) (this work), unc-69(e587), unc-50(e306), ced-9(n2812, n1950 n2161, n1653, n3377, n3400, n3407) (Desai et al. 1988; Hengartner et al. 1992; Hengartner and Horvitz 1994b; and this work), ced-4(n3379, n3392) (this work), hT2 [qIs48] (Mathies et al. 2003).
LGIV: ced-2(n3387) (this work), dpy-20(e1282), unc-30(e191), let-60(n1876), ced-3(n2427, n2447, n2452, n3374, n3375, n3384, n3403, n3406, n3411) (Shaham et al. 1999; and this work), dpy-4(e1166), nT1 [qIs51] (Mathies et al. 2003).
LGV: eT1, lin-40(s1593) (Solari et al. 1999), tam-1(cc567) (Hsieh et al. 1999), unc-46(e177), let-418(s1617), dpy-11(e224), mys-1(n3681) (Ceol and Horvitz 2004), egl-1(n1084 n3082) (Conradt and Horvitz 1998), hda-1(e1795) (Dufourcq et al. 2002), unc-76(e911), nIs96 (Reddien et al. 2001).
Isolation of n4005:
mcd-1(n4005) was isolated by screening a library of UV-trimethylpsoralen-induced deletions using PCR as described previously (Jansen et al. 1997). The mcd-1(n4005) deletion begins at base 1228 after the A of ATG within the second intron and extends through base 2325 within the third exon. If the mcd-1 mRNA in n4005 animals is spliced from the second to the fourth exon, the product is predicted to be out-of-frame and could generate a protein of 92 amino acids.
Four of 52 n3376/n3376 progeny from n3376+/+unc-52 heterozygous animals were heterozygous for unc-52. Zero of 15 Rol-1 non-Dpy-10 animals from dpy-10 rol-1+/++n3376 carried n3376. Twenty of 21 Rol-1 non-Unc-52 recombinants between rol-1 and unc-52 from rol-1+unc-52/+n3376+ animals carried n3376, and 1/21 did not. Zero of 93 F2 animals from n3380+/+rol-6 heterozygous animals recombined between n3380 and rol-6. From 18 recombinants selected between dpy-10+rol-1/+n3380+ animals, 5/18 were between dpy-10 and n3380 and 13/18 were between n3380 and rol-1. From 22 recombinants selected between dpy-10+unc-4/+n3380+ animals, 9/22 were between dpy-10 and n3380 and 13/22 were between n3380 and unc-4.
Polymorphism identification and mapping:
PCR-size polymorphisms on LGIIR between RC301 and N2 were identified by PCR amplifying ∼1-kb fragments from intergenic regions and analyzing fragment sizes by electrophoresis. nP89 is located in the region of the genome represented in cosmid F08G2; the RC301 PCR product amplified with primers PWR.G1 5′-GCCGAAGAAGCGATACTGAATG-3′ and PWR.G2 5′-AAGCCCCCTTGAAAAATGAGC-3′ is ∼1.1 kb; and N2,1.0 kb. nP90 is located in the region of the genome represented in YAC Y51H1A; the RC301 PCR product amplified with primers PWR.G11 5′-GTCATTGTGCGTTGATGGGAG-3′ and PWR.G12 5′-TTACCGAGTGCGTTCTGTGAATC-3′ is ∼1.3 kb; and N2, 1.2 kb. nP91 is located in the region of the genome represented in cosmid W02B8; the RC301 PCR product amplified with primers PWR.G21 5′-CCATCATTTGTCATTGGAGCG-3′ and PWR.G22 5′-AGGGTAGGGGCACGGTAGATAAAG-3′ is ∼1.2 kb; and N2, 1.1 kb. nP92 is located in the region of the genome represented in cosmid W07G1; the RC301 PCR product amplified with primers PWR.G31 5′-CATTGGTAGTTGTCGGCTTCCTG-3′ and PWR.G32 5′-CCTTTTCATTTTTGCGGTGTCC-3′ is ∼1.5 kb; and N2, 1.2 kb. Recombinants were generated from animals heterozygous for LGII between RC301 and N2: rol-1 n3376 unc-52 (N2)/+++ (RC301); ced-3(n2427); nIs106 animals. We isolated 20 Rol-1 non-Unc-52 recombinants and generated homozygous recombinant strains. Two recombinants were found to the right of nP89 and to the left of n3376, and one recombinant was found to the left of nP91 and to the right of n3376.
The mcd-1 cDNA was excised from the phage clone yk464e11 (Y. Kohara, personal communication). To determine the 5′ end of mcd-1 messages we used a 5′-rapid amplification of cDNA ends (RACE) system (GIBCO, Grand Island, NY). DNA sequences were determined using an automated ABI 373A DNA sequencer (Applied Biosystems, Foster City, CA). RNAi was performed by microinjection or feeding, as previously described (Fire et al. 1998; Timmons et al. 2001). RNAi of efl-1, efl-2, and lin-35 was performed using efl-1, efl-2, and lin-35 cDNAs as previously described (Lu and Horvitz 1998; Ceol and Horvitz 2001). RNAi of mcd-1 was performed from a template isolated by PCR amplification of a region of the third exon of mcd-1 (see Figure 1H), using primers carrying the T7 promoter, I.9, 5′-GATCGATAATACGACTCACTATAGGGCGGAAAATCCGCCAAAAAAAATCGG-3′, and I.10, 5′-GATCGATAATACGACTCACTATAGGGGATCACAGAGTCGATCCATTACAGG-3′. Similar methods were used to amplify and generate RNA from F08G2.7, Y51H1A.1, Y51H1A.2, Y51H1A.4, and Y51H1A.5.
Defects in programmed cell death in C. elegans have been efficiently quantified using two assays. First, a Plin-11 gfp reporter allows the deaths of specific cells in the ventral cord to be assessed (Reddien et al. 2001). The Plin-11 gfp reporter is expressed in the six VC motor neurons, P3–8.aap (P, P blast cell; a, anterior daughter; p, posterior daughter), of the ventral cord (Figure 1A). The cells W.ap (W, W blast cell), P1–2.aap, and P9–12.aap, which die in the wild type, survive in animals with defects in programmed cell death and express Plin-11 gfp (Reddien et al. 2001). Five of these cells, P2.aap and P9–12.aap, can be reliably and easily scored for cell survival using a fluorescence stereomicroscope.
A second assay for programmed cell death uses Nomarski optics to determine the number of cell nuclei in another specific region of the animal, the anterior pharynx (Hengartner et al. 1992; Shaham et al. 1999). Animals carrying strong loss-of-function alleles of the killer gene ced-3 have ∼12–13 extra cells in this region, whereas animals carrying weak alleles of ced-3 (for example, n2427) have ∼1–2 extra cells in this region (Shaham et al. 1999). We observed that the degree of the cell-death defect conferred by the weak loss-of-function allele ced-3(n2427) as determined by counting the number of extra cells in anterior pharynges did not substantially vary with genetic background (supplemental Table S1 at http://www.genetics.org/supplemental/), suggesting that mutations that caused alterations in cell number in ced-3(n2427) animals would reflect specific effects on cell death rather than nonspecific effects on animal health.
A genetic screen for enhancers of ced-3(n2427):
Using the Plin-11 gfp assay described above, we observed that ced-3(n2427) results in 4% of animals having all five VC-like cells (P2.aap and P9–12.aap, n = 50). We previously showed that mutations in engulfment genes enhance the cell-death defects conferred by ced-3(n2427) as assayed with Plin-11 gfp (Reddien et al. 2001). For example, 88% of ced-1(e1735); ced-3(n2427) animals had all five extra VC-like cells present (n = 50), whereas 0% of ced-1(e1735) animals (n = 50) had all five extra VC-like cells present (Reddien et al. 2001). ced-1 is necessary for engulfment and encodes a cell-corpse receptor (Hedgecock et al. 1983; Zhou et al. 2001). We therefore concluded that the Plin-11 gfp assay could be used for isolating mutations that enhance ced-3(n2427).
We mutagenized ced-3(n2427); Plin-11 gfp animals, selected individual F2 progeny with five extra VC-like cells, and determined if these animals generated progeny with enhanced cell-death defects (Figure 1B). We expected to obtain alleles in the cell-death execution genes ced-3, ced-4, and egl-1. Because loss-of-function mutations in the ced-9 gene enhance the cell-death defect caused by weak alleles of ced-3 (Hengartner and Horvitz 1994a), we anticipated that mutations in ced-9 would also be isolated. We isolated 6 strains that likely carry mutations in ced-3, 2 in ced-4, and 3 in ced-9 (Figure 1C). Because mutations in engulfment genes can enhance the cell-death defect caused by ced-3(n2427), we also expected to isolate alleles of engulfment genes. At least 10 of the strains we isolated have defects in the engulfment of dying cells. We determined, using complementation tests, that 2 of these strains carry a mutation in ced-1, 7 carry one in ced-7, and 1 carries one in ced-2 (Figure 1C). Many alleles of ced-7 might have been isolated because loss of ced-7 function strongly enhances weak cell-death defects (Reddien et al. 2001). Two mutations, n3376 and n3380, define new cell-death genes, as described below.
The mutations n3376 and n3380 define two new cell-death genes:
Animals carrying the mutations n3376 and n3380 had defects in cell death (Figure 1, D and E). Genetic mapping established that n3376 and n3380 reside at different locations on LGII (Figure 1F). Because no previously characterized C. elegans cell-death gene is located on LGII, these mutations define new cell-death genes. The gene defined by n3380 is a previously known gene, dpl-1 (Figure 1G, see below for details). We named the gene defined by n3376, mcd-1 (mcd, modifier of cell death). More than 60% of double-mutant animals carrying mcd-1(n3376) or dpl-1(n3380) with ced-3(n2427) had all five extra VC-like cells in their ventral cords, whereas only 4% of ced-3(n2427) single-mutant animals had all five extra VC-like cells present (Figure 1D). mcd-1(n3376) and dpl-1(n3380) also enhanced cell-death defects conferred by ced-3(n2427) in the anterior pharynx (Table 1 ), indicating that these mutations affect programmed cell death broadly rather than specifically in the ventral cord. Furthermore, mcd-1(n3376) and dpl-1(n3380) enhanced the cell-death defect associated not only with the CED-3 G474R substitution caused by the n2427 allele but also with the S446L substitution caused by the n2447 allele (supplemental Figure S2A at http://www.genetics.org/supplemental/). Therefore, mcd-1(n3376) and dpl-1(n3380) are not allele-specific enhancers of ced-3 and likely generally affect the ability of cells to die if programmed cell death is slightly impaired. We generated animals homozygous for mcd-1(n3376) and dpl-1(n3380) in a wild-type ced-3 background. These animals were viable with no obvious morphological abnormalities. mcd-1(n3376) and dpl-1(n3380) each conferred a weak block in programmed cell death in the presence of an intact core cell-death execution pathway, indicating that mcd-1 and dpl-1 have normal roles in promoting cell death (Figure 1E).
n3380 is an allele of dpl-1 DP:
We mapped n3380 between unc-4 and dpy-10 on LGII (Figure 1F, materials and methods). We found that existing mutations in a gene in this region, dpl-1, enhanced the cell-death defect conferred by ced-3(n2427) (Table 1). dpl-1 encodes the C. elegans ortholog of the mammalian E2F-heterodimerization partner DP (Ceol and Horvitz 2001). We performed complementation tests and determined that the dpl-1(n3316) mutation (Ceol and Horvitz 2001) failed to complement n3380 for the enhancement of ced-3(n2427) (Table 1 and data not shown). dpl-1 is a class B synthetic multivulva (synMuv) gene (Ceol and Horvitz 2001). The synMuv genes fall into three classes, A, B, and C. Animals carrying mutations in any two genes from different classes undergo ectopic vulval development, the result of normally nonvulval cells adopting a vulval fate; the resulting phenotype is termed multivulva (Muv) (Horvitz and Sulston 1980; Ferguson and Horvitz 1989; Ceol and Horvitz 2004). We found that dpl-1(n3380); lin-15A(n767) animals were Muv [n767 is a class A synMuv mutation (Ferguson and Horvitz 1989)], indicating that dpl-1(n3380) acts as a class B synMuv mutation, like other mutations in dpl-1. We determined the sequence of the dpl-1 open reading frame in dpl-1(n3380) animals and found a C-to-T mutation predicted to result in a Q486stop (TAA) alteration in the 595-amino-acid DPL-1 protein (Figure 1G). Together these results indicate that n3380 is an allele of the gene dpl-1. Because dpl-1 null alleles confer a sterile phenotype (Ceol and Horvitz 2001), dpl-1(n3380) is not a null allele of dpl-1. dpl-1(n3380) conferred cell-death defects similar to those of the dpl-1 allele n2994 and the dpl-1 null allele n3316 (Table 1).
n3376 is an allele of a gene predicted to encode a novel zinc-finger protein:
We mapped the mutation n3376 to a region of ∼75 kbp containing 13 predicted genes (Figure 1F, see materials and methods). We inhibited the function of candidate genes in this region by RNA-mediated genetic interference (RNAi) (Fire et al. 1998) and found that inhibition of Y51H1A.6, but not of F08G2.7, Y51H1A.1, Y51H1A.2, Y51H1A.4, or Y51H1A.5, caused enhancement of the cell-death defect of ced-3(n2427) animals (Table 1). We defined the structure of the Y51H1A.6 genomic locus by determining the sequence of a cDNA clone, yk464e11 (kindly provided by Y. Kohara) (Figure 1H). We determined the 5′ end of the Y51H1A.6 transcript using 5′ RACE and found Y51H1A.6 transcripts to carry the SL1 trans-spliced leader sequence, which is commonly found on C. elegans mRNAs (Krause and Hirsh 1987). We identified a C-to-T mutation predicted to cause an H277Y substitution in the Y51H1A.6 protein in mcd-1(n3376) animals (Figure 1H). We isolated a deletion allele of Y51H1A.6, n4005 (Figure 1H, materials and methods), and found that n4005 resulted in a phenotype similar to that of mcd-1(n3376) and failed to complement mcd-1(n3376) for the enhancement of ced-3 (Table 1). n4005 appears to be a strong loss-of-function allele of mcd-1. Together these findings indicate that Y51H1A.6 is the gene defined by n3376.
The mcd-1 gene is predicted to encode a highly acidic novel protein containing one candidate C2H2 zinc finger and has no other significant homology to known genes (Figure 1H, supplemental Figure S1 at http://www.genetics.org/supplemental/). This C2H2 zinc finger (FKCAECGDGFPVMDRLCDHMIKQH) is an exact match to canonical zinc-finger domains [Y/F-X-C-X2-4-C-X3-F-X5-L-X2-H-X3-5-H (Wolfe et al. 2000)]. The mcd-1(n3376) missense mutation changes the first C2H2 histidine to a tyrosine (Figure 1H), indicating that the MCD-1 zinc finger is important for the MCD-1 cell-death function. Zinc-finger domains can interact with DNA (Wolfe et al. 2000) and mediate protein–protein interactions (Mackay and Crossley 1998).
efl-1 E2F and lin-35 Rb promote cell death:
Given that dpl-1 acts in C. elegans development with the E2F-like gene efl-1 and the Rb-like gene lin-35 (Lu and Horvitz 1998; Ceol and Horvitz 2001; Page et al. 2001), we asked whether these and other dpl-1-interacting genes also affect programmed cell death. DP-1 from mammals can bind to and promote DNA-binding by E2F-1 (Girling et al. 1993; Helin et al. 1993). E2F proteins can affect DNA replication, cell-cycle progression, and development in mammals (Helin 1998; Bracken et al. 2004) and can interact with the tumor-suppressor protein pRb (Dyson 1998). E2F proteins can promote programmed cell death (Harbour and Dean 2000b). We found that, like dpl-1 DP, efl-1 E2F and lin-35 Rb promoted programmed cell death (Table 2A; supplemental Figure S2, B and C, at http://www.genetics.org/supplemental/). Because mutations in efl-1 confer sterility, we studied the role of efl-1 E2F in programmed cell death using RNAi. RNAi of efl-1 enhanced the weak cell-death defects conferred by ced-3(n2427). Perturbation of a second C. elegans E2F-like gene, efl-2, caused no observed cell-death defects (data not shown). We studied the role of lin-35 Rb in programmed cell death using the putative null allele n745. Enhancement of a weak cell-death defect was caused by lin-35(n745) (Table 2A), as well as by RNAi of lin-35 and by other putative lin-35 null alleles (supplemental Figure S2C and data not shown). Our data suggest that dpl-1, lin-35, and efl-1 act together in programmed cell death as they do in vulval development. The cell-death defects of dpl-1 mutants were stronger than those of efl-1(RNAi) animals or of animals carrying a putative lin-35 null allele, n745 (Lu and Horvitz 1998), for unknown reasons. Disruption of the class B synMuv genes lin-37 and lin-52 also enhanced ced-3(n2427) (Table 2A, supplemental Figure S2D). lin-52(n771) more strongly enhanced the cell-death defect conferred by ced-3(n2427) in the anterior pharynx than in the ventral cord (Table 2A, supplemental Figure S2D). lin-37 encodes a protein similar to the Myb-interacting Mip40 protein (Korenjak et al. 2004) and can physically interact with LIN-53 RbAP48, an Rb-interacting protein (Walhout et al. 2000). lin-52 encodes a novel protein with similarity to proteins of unknown function in humans and Drosophila (Thomas et al. 2003).
Disruption of class B synMuv genes—lin-9, lin-15B, lin-36, or lin-53—or any of the class A synMuv genes lin-8, lin-15A, lin-38, or lin-56 did not enhance ced-3(n2427) (Table 2B). Therefore, the synMuv genes likely do not control a single process that affects both vulval cell fate and cell death. Rb-associated proteins have been found in multiple complexes, including a nucleosome-remodeling and histone-deacetylase complex (NuRD) (Xue et al. 1998) and two complexes from Drosophila containing the transcription factor Myb and multiple class B synMuv-like proteins (Myb–MuvB and dREAM) (Lewis et al. 2004). We failed to detect a cell-death role for candidate C. elegans components of a NuRD complex (Table 2B): hda-1 (Lu and Horvitz 1998), let-418 (von Zelewsky et al. 2000), chd-3 (von Zelewsky et al. 2000), lin-40 (egr-1) (Solari et al. 1999; Chen and Han 2001b), and egl-27 (Solari et al. 1999). We also failed to detect a cell-death role for lin-9, which encodes a protein similar to a component of the Myb–MuvB and dREAM complexes (Table 2B). Together, our results suggest that DPL-1 DP, EFL-1 E2F, LIN-35 Rb, LIN-52, and LIN-37 Mip40 define a novel association of class B synMuv proteins that act to promote programmed cell death. Because of lethality associated with strong loss-of-function alleles of multiple C. elegans genes encoding candidate NuRD, Myb–MuvB, and dREAM complex components, we examined animals with incomplete loss of function for some genes (Table 2B). Roles for essential candidate NuRD, Myb–MuvB, and dREAM complex components in programmed cell death therefore remain possible.
Because the synMuv genes act antagonistically to let-60 Ras in vulval development (Ferguson et al. 1987; Beitel et al. 1990; Han and Sternberg 1990; Sternberg and Han 1998), we asked whether the dpl-1 gene acts antagonistically to let-60 Ras in cell death. Null alleles of let-60 confer maternally rescued larval lethality and zygotically suppress the synMuv phenotype conferred by dpl-1 mutants (Ceol and Horvitz 2001). We examined homozygous let-60(n1876) animals from heterozygous mothers. let-60(n1876) did not significantly affect the cell-death phenotype of ced-3(n2427) animals and did not suppress the enhanced cell-death defect of dpl-1(n3380); ced-3(n2427) animals (supplemental Figure S2E at http://www.genetics.org/supplemental/). These observations indicate that dpl-1 does not promote cell death by antagonizing Ras signaling.
dpl-1 and mcd-1 might act together and with lin-35 Rb and efl-1 E2F to promote cell death:
To ask whether dpl-1 and mcd-1 act together or separately with one another and with lin-35 and efl-1 to promote cell death, we assayed the number of extra cells conferred by ced-3(n2427) in the anterior pharynges of appropriate double- and triple-mutant animals. dpl-1, mcd-1, lin-35, and efl-1 did not appear to act additively (Table 3B, supplemental Table S2B at www.genetics.org/supplemental/). dpl-1(n3380) mcd-1(n3376) animals had slightly more extra cells in their ventral cords than did dpl-1(n3380) or mcd-1(n3376) single-mutant animals (supplemental Figure S2F at www.genetics.org/supplemental/) but not in their anterior pharynges (supplemental Table S2B). Furthermore, dpl-1(n3380) mcd-1(n3376); ced-3(n2427) and dpl-1(n3380) mcd-1(n4005); ced-3(n2427) animals did not have more extra cells in their anterior pharynges than did dpl-1(n3380); ced-3(n2427) and mcd-1(n3376); ced-3(n2427) or mcd-1(n4005); ced-3(n2427) animals (Table 3B). That dpl-1(n3380) and mcd-1(n3376) or mcd-1(n4005) can confer additive defects with other mutations that caused similar levels of cell-death defects (ced-1, ced-8, and ced-9, see below, Table 3C) indicates that changes in cell-killing efficiency are within the detectable range of this assay. Thus, dpl-1, mcd-1, lin-35, and efl-1 might act together to promote cell death.
dpl-1 and mcd-1 promote cell death in a manner independent of ced-1 and ced-8:
We asked whether dpl-1 and mcd-1 act together with other subtle cell-death regulators. Mutations in engulfment genes (e.g., ced-1) can decrease cell killing (Reddien et al. 2001). Mutations in ced-8 cause a delay in cell death, enhance weak alleles of ced-3, and cause very weak defects in cell killing on their own (Stanfield and Horvitz 2000). We observed that mutations in mcd-1 and dpl-1 enhanced cell-killing defects conferred by ced-1(e1735) or ced-8(n1891) (Table 3C). Because ced-1(e1735) and ced-8(n1891) are both null or strong loss-of-function alleles (Stanfield and Horvitz 2000; Zhou et al. 2001), our data indicate that mcd-1 and dpl-1 act independently from ced-8 and ced-1. Furthermore, we examined doubly mutant strains carrying a wild-type ced-3 locus and observed enhanced cell-death defects: ced-1(e1735); dpl-1(n3380) animals, ced-1(e1735); mcd-1(n3376) animals, dpl-1(n3380); ced-8(n1891) animals, and mcd-1(n3376); ced-8(n1891) animals had defects in cell killing greater than those seen in the single mutants (supplemental Table S2C and supplemental Figure S2F at www.genetics.org/supplemental/). In addition, n3376 and n3380 did not cause defects in engulfment. These results indicate that dpl-1 and mcd-1 control one of multiple parallel processes that redundantly and significantly contribute to cell death.
dpl-1 and mcd-1 do not promote cell death through regulation of ced-9:
The egl-1 gene is required for essentially all programmed cell death in C. elegans and can be transcriptionally upregulated to promote programmed cell death (Conradt and Horvitz 1998, 1999). Because DP, Rb, and E2F can act in transcriptional regulation and mcd-1 has a zinc-finger domain, we sought to determine whether these genes might affect egl-1 expression. egl-1 acts upstream of ced-9 (Conradt and Horvitz 1998). To determine whether dpl-1 and mcd-1 act downstream or upstream of the death-inhibiting role of ced-9, we asked whether mutations in these genes could enhance ced-3(n2427) in the absence of ced-9 function using the ced-9 null allele n2812 (Table 3C). ced-9 has both a death-inhibiting and a death-promoting role (Hengartner and Horvitz 1994a). We found that ced-3(n2427) animals had on average 1.8 extra cells in their anterior pharynges, while ced-9(n2812); ced-3(n2427) animals had 6.3 extra cells (Table 3A). Triple-mutant animals with ced-9(n2812); ced-3(n2427) in addition to mcd-1(n3376) or dpl-1(n3380) had an enhanced number of extra cells (9.4 and 10.6, respectively, Table 3C). By contrast, null alleles of egl-1 do not enhance the cell-killing defects of ced-9(n2812); ced-3(n2427) double mutants (Conradt and Horvitz 1998). These results indicate that dpl-1 and mcd-1 have cell-death-promoting activity downstream of or independent of the cell-death protective activity of ced-9. In addition, the data indicate that dpl-1 and mcd-1 act independently from the cell-death-promoting activity of ced-9. Because egl-1 acts to inhibit ced-9, dpl-1 and mcd-1 likely do not promote death by regulating egl-1 transcription.
dpl-1 and mcd-1 can affect cells fated to live:
Embryos lacking maternal and zygotic ced-9 undergo excessive programmed cell death and die; i.e., in such embryos cells that are normally fated to live instead die (Hengartner et al. 1992). This lethality can be suppressed by mutations in the killer genes ced-3 and ced-4 (Hengartner et al. 1992). To determine if perturbation of dpl-1 and mcd-1 can affect the ectopic programmed cell death of cells that normally live, we utilized the observation that animals carrying both the ced-9(n1653) temperature-sensitive allele and the ced-4(n2273) mutation confer a synthetic maternal-effect lethality caused by excessive programmed cell death (Shaham and Horvitz 1996a). The ced-4(n2273) mutation affects ced-4 splicing and affects production of the minor cell-death inhibitory CED-4L product; therefore, ced-9(n1653) ced-4(n2273) animals might confer synthetic lethality as a result of a reduction of both ced-9 death-inhibitory function and ced-4L death-inhibitory function (Shaham and Horvitz 1996a). ced-9(n1653) ced-4(n2273) synthetic maternal-effect lethality can be suppressed by mutations that otherwise cause a very weak cell-death defect (E. Speliotes and H. R. Horvitz, unpublished results), suggesting that cells that normally live are poised between life and death in these animals. We found that either dpl-1(n3380) or mcd-1(n3376) partially suppressed the synthetic lethality conferred by ced-9(n1653) ced-4(n2273) (Table 4). dpl-1(n3380) or mcd-1(4005) also partially suppressed the maternal-effect lethality conferred by the ced-9 partial loss-of-function mutant ced-9(n1950 n2161) (Table 4). Therefore, dpl-1 and mcd-1 likely are active in cells that normally live and can promote the ability of such cells to die when cell-death inhibitory activity is reduced. Cell deaths in C. elegans are normally invariant among individuals (Sulston and Horvitz 1977). Our observations indicate that dpl-1 and mcd-1 likely do not act to mediate the effects of lineage-regulated factors that specify cell death, because these genes can affect the ability of any cell to die. This hypothesis is supported by our direct observation of the cell-death process in mcd-1 and dpl-1 mutants using Nomarski optics (see below).
Cells can initiate the death process and recover in dpl-1(n3380) and mcd-1(n4005) animals:
We directly observed cell death within the P9, P10, and P11 neuroblast lineages in dpl-1(n3380) and mcd-1(n4005) animals. Animals also carried the Plin-11 gfp reporter, allowing us to determine whether any Pn.aap cell ultimately survived and differentiated. We observed no defect in cell division patterns [dpl-1(n3380), n = 8 and mcd-1(n4005), n = 6]. We observed that the cell-death process could initiate and be followed by episodic changes in the morphology of the dying cells (Figure 2, A and B). Some cells ultimately survived following these episodic morphological changes. Specifically, 4 of 24 Pn.aap cells failed to die in the time period observed in dpl-1(n3380); Plin-11 gfp animals, and all 4 of these cells expressed GFP. Four of 18 Pn.aap cells failed to die in the time period observed in mcd-1(n4005); Plin-11 gfp animals, and 3 of these cells expressed GFP. The initiation of cell condensation during cell death requires the CED-3 caspase (Reddien et al. 2001). Therefore, we suggest that cells lacking normal dpl-1 or mcd-1 function can occasionally recover following the activation of CED-3 and the initiation of morphological changes associated with the cell-death process (Figure 2C).
dpl-1 and mcd-1 do not affect CED-3 caspase-independent death:
Strong loss-of-function alleles of ced-3, including a deletion allele that completely lacks the CED-3 protease-encoding region, ced-3(n2452), do not completely block programmed cell death (Shaham et al. 1999). For example, ced-1(e1735); ced-3(n2452) animals generated a few apparent cell corpses: 7 of 50 animals had two or more corpses in the heads of L1 larvae, suggesting that a low level of cell death can occur independently of the ced-3 caspase. mcd-1 and dpl-1 had no effect on these ced-3-independent deaths: 8 of 50 L1 larvae had two or more corpses, and 6 of 50 animals had two or more corpses in ced-1(e1735); mcd-1(n3376); ced-3(n2452) and ced-1(e1735); dpl-1(n3380); ced-3(n2452) animals, respectively. We therefore suggest that dpl-1 and mcd-1 promote cell death by affecting a process controlled by the CED-3 caspase rather than by acting independently from and parallel to CED-3 activity.
Loss of mcd-1 function confers synthetic lethality with mutations in lin-35 Rb and other class B synMuv genes:
Reduction of mcd-1 function by deletion or RNAi did not cause notable growth defects as compared to the wild type. However, perturbation of mcd-1 function in combination with mutations in some class B synMuv genes conferred synthetic lethality (larval arrest) or slow growth (Table 5A). Specifically, RNAi of mcd-1 and mcd-1(n4005) synthetically caused 100% arrest during the first larval stage of lin-9(n112), lin-15B(n744), lin-35(n745), lin-37(n758), and lin-54(n2231) animals, slow larval growth of lin-53(n833) animals, and slow growth and some larval arrest of animals carrying mutations in dpl-1 and lin-13. mcd-1(n4005) also synthetically caused slow larval growth in animals carrying a mutation in the class C synMuv gene mys-1 (Table 5A). Additionally, an L1-arrest phenotype was observed in mcd-1(n4005) animals when the class B synMuv genes lin-54, dpl-1, lin-9, or lin-37 were inactivated by RNAi. mcd-1(n4005) and RNAi of mcd-1 did not cause synthetic lethality with the class B synMuv mutations tam-1(cc567), lin-36(n766), or lin-52(n771) (Table 5B) or with the class A synMuv mutations lin-8(n2731), lin-15A(n767), lin-38(n751), or lin-56(n2728) (Table 5C). lin-36(n766) and lin-52(n771) are nonnull alleles (Thomas et al. 2003). RNAi of mcd-1 did not cause a synMuv phenotype in combination with the class B mutations lin-36(n766) and lin-52(n771). These results indicate that mcd-1 acts redundantly with most but not all class B synMuv genes for growth and viability.
From a screen for mutations that enhanced the cell-death defect caused by a weak ced-3 allele we identified two new positive regulators of programmed cell death in C. elegans, mcd-1 and dpl-1. mcd-1 encodes a novel zinc-finger protein that acts together with the dpl-1 DP gene. Because DP is the dimerization partner for the E2F transcription factor in mammals (Girling et al. 1993; Helin et al. 1993), these genes likely affect C. elegans cell death via transcription. DP and E2F can act together with Rb and a number of chromatin regulators in transcriptional repression (Harbour and Dean 2000a). A large number of genes encoding proteins implicated in chromatin remodeling have been identified in C. elegans that act together with dpl-1 DP in regulating vulval development. These genes, called “synMuv” for their synthetic multivulva loss-of-function phenotype (Ferguson and Horvitz 1989), number at least 27. We found that a few of these genes promote cell death; these genes encode an E2F-like protein (EFL-1), an Rb-like protein (LIN-35), a Mip40-like protein (LIN-37), and one novel protein with similarity to proteins in humans and Drosophila (LIN-52). Therefore, a DP/E2F-like heterodimeric transcription factor probably acts together with an Rb-like protein, the MCD-1 zinc-finger protein, a Mip40-like protein, and the LIN-52 protein to promote cell death in C. elegans via transcriptional regulation.
E2F and Rb proteins have roles in cancer and effects on cell death in mammals (Harbour and Dean 2000b; Sherr and McCormick 2002). Similar to the case in C. elegans, loss of function of E2F-1 in mice leads to reduced cell death (Field et al. 1996; Yamasaki et al. 1996). However, in contrast to C. elegans, in the mouse loss of Rb function leads to increased cell death (Morgenbesser et al. 1994; Macleod et al. 1996). It is possible that Rb-like genes promote cell death in mammals but that this defect is obscured by cell-cycle abnormalities; cell-cycle abnormalities are known to trigger cell death in mammals (Evan and Littlewood 1998). In C. elegans, an E2F-like and an Rb-like protein act together, rather than antagonistically, to regulate vulval development (Ceol and Horvitz 2001), the cell cycle (Boxem and van den Heuvel 2002), and cell death (this work). Because E2F-4 and E2F-5 are known to act together with Rb to mediate transcriptional repression in mammals (Frolov and Dyson 2004), some of the mammalian Rb and E2F genes might act together to promote cell death, as observed in C. elegans.
How might dpl-1, mcd-1, lin-35, efl-1, lin-37, and lin-52 affect C. elegans cell death?
Rb and E2F/DP are cell-cycle regulators, and a misregulated cell cycle can trigger cell death in mammals (Evan and Littlewood 1998). However, our data are inconsistent with the hypothesis that cell-cycle abnormalities account for the enhanced cell-death defects we observed. For example, the P9–11 neuroblast lineages in dpl-1 and mcd-1 mutants were grossly normal (i.e., no cell-cycle arrest or aberrant division patterns were observed; dpl-1(n3380); nIs106, n = 8; mcd-1(n4005); nIs106, n = 6). Furthermore, cell-cycle roles have been identified in C. elegans for a subset of synMuv genes, including lin-35 Rb, dpl-1 DP, efl-1 E2F, lin-9 ALY, lin-15B, and lin-36 (Boxem and van den Heuvel 2001, 2002), but this subset does not match the subset of synMuv genes that affects cell death. Specifically, not all synMuv genes that affect the cell cycle affect cell death (lin-9, lin-15, lin-36), and a synMuv gene that does not affect the cell cycle can affect cell death (lin-37). Many class B synMuv genes probably have multiple functions in development, such as regulation of the cell cycle and regulation of cell fate. For example, many synMuv genes have no observed effect on the cell cycle and act together with synMuv genes like dpl-1, lin-35, and efl-1 to regulate vulval cell fates. We propose that dpl-1, mcd-1, efl-1, lin-35, lin-37, and lin-52 act together to regulate cell death via transcriptional regulation of specific targets that affect the cell-death process rather than act nonspecifically to affect cell death as a consequence of effects on the cell cycle. Mammalian Rb and E2F-like proteins probably also have non-cell-cycle roles (Landsberg et al. 2003; Frolov and Dyson 2004).
There are multiple mammalian candidate E2F/DP transcriptional targets that could interface with the cell-death pathway. For example, E2F can trigger cell death via transcriptional activation of the p53 tumor-suppressor gene (Harbour and Dean 2000b). However, we observed that a putative null allele of the C. elegans P53-like gene cep-1 (Derry et al. 2001; Schumacher et al. 2001) failed to cause enhancement of the cell-death defects in ced-3(n2427) animals (data not shown). E2F genes can also regulate transcription of caspase (Muller et al. 2001) and the Apaf-1 (Moroni et al. 2001) genes. No clear candidate EFL-1 binding sites are readily apparent in ced-3 or ced-4 regulatory regions that are evolutionarily conserved with the related species C. briggsae. The initiation of cell death in C. elegans is known to involve transcription (Metzstein et al. 1996; Metzstein and Horvitz 1998; Conradt and Horvitz 1999; Thellmann et al. 2003), raising the possibility that dpl-1, mcd-1, and interacting genes regulate the transcription of egl-1 to promote cell-death initiation. However, multiple lines of evidence are inconsistent with this hypothesis. First, egl-1 acts upstream of the cell-death inhibitory ced-9 gene (Conradt and Horvitz 1998), and dpl-1 and mcd-1 do not. Second, dpl-1 and mcd-1 do not act only in cells fated to die. Third, cells that fail to die in dpl-1 and mcd-1 mutants display morphological changes reflecting attempts at cell death rather than a complete failure to initiate cell death. dpl-1 and mcd-1 therefore likely control the death process rather than the life vs. death decision-making process. We do not know whether dpl-1 and mcd-1 act to control the activity of the core cell-death-execution pathway (CED-9, CED-4, CED-3) or act independently. Our data indicate dpl-1 and mcd-1 do not mediate the effects of engulfment on cell death and do not act together with ced-8; engulfment and ced-8 are other known cell-death contributing factors. Because loss of function of mcd-1 and dpl-1 did not perturb the low level of cell death that occurs independently of the CED-3 caspase, these genes likely function in a pathway with CED-3.
MCD-1 is a novel regulator of cell death and is synthetically required for viability with the LIN-35 Rb tumor suppressor:
MCD-1 has a single C2H2 zinc finger and no obvious homology to other domains or proteins in existing databases. Because caspases execute the death process, modulators of cell death are candidate caspase targets for proteolysis. However, MCD-1 does not contain candidate CED-3 cleavage sites conserved in the related species C. briggsae. Many uncharacterized zinc-finger proteins exist in mammals, one or more of which could share functional similarities with MCD-1. Given known functions for C2H2 zinc fingers, MCD-1 could be a DNA-binding protein or be involved in protein–protein interactions. mcd-1 acts with a number of synMuv genes to regulate cell death. Because mcd-1 is synthetically required for viability with lin-35 Rb and many other class B synMuv genes, mcd-1 must act redundantly with these genes in some developmental processes. This might be because different assemblages of transcriptional regulatory factors control distinct developmental events (see below). Because Rb is a known tumor-suppressor gene, mcd-1 and potential interacting genes could define an uncharacterized complex that acts redundantly with Rb as a tumor suppressor in mammals.
Subsets of synMuv genes control diverse aspects of biology:
Our finding that a subset of synMuv genes affects cell death and that a different subset of synMuv genes is synthetically required for animal viability adds to an increasing number of functional categorizations of synMuv genes. Subsets of synMuv genes control at least 14 diverse biological processes in C. elegans (Ferguson and Horvitz 1989; Boxem and van den Heuvel 2001, 2002; Chen and Han 2001a; Page et al. 2001; Dufourcq et al. 2002; Fay et al. 2002, 2003, 2004; Unhavaithaya et al. 2002; Bender et al. 2004; Ceol and Horvitz 2004; Cui et al. 2004; Garbe et al. 2004; Reddy and Villeneuve 2004; Wang et al. 2005). The set of synMuv genes that affects one process is often distinct from the set affecting others. Therefore, the diverse functions of LIN-35 Rb and associated synMuv proteins could be mediated by different assemblages of coregulators of transcription. Many of the class B synMuv genes encode proteins similar to those found in several chromatin-remodeling complexes in other organisms, including the NuRD, Myb-MuvB, and dREAM transcriptional repression complexes (Xue et al. 1998; Korenjak et al. 2004; Lewis et al. 2004). Our studies define a new candidate assemblage of synMuv proteins, containing DPL-1, EFL-1, LIN-35, MCD-1, LIN-37, and LIN-52 and regulating programmed cell death.
Our results with mcd-1 highlight the complex manner in which synMuv genes are utilized. For example, mutations in some of the class B synMuv genes that cause synthetic lethality with loss of mcd-1 cause no detectable defect in cell death (e.g., lin-9 and lin-15B), whereas others do cause defects in cell death (e.g., lin-35). By contrast, one class B synMuv mutation that causes a defect in cell death, lin-52(n771), is homozygous viable with mcd-1(n4005). The process that is affected by mcd-1 and synthetically required for viability with some class B synMuv genes is likely distinct from the process affected by mcd-1 during cell death, because different gene sets associate with these aspects of the mcd-1 loss-of-function phenotype. The hypothesis that mcd-1 acts in at least two processes might explain the observation that mcd-1 appears to act redundantly with lin-35 Rb, dpl-1 DP, and other class B synMuv genes for growth and viability but not for cell death. Continued genetic studies in C. elegans should allow for the systematic dissection of the functional associations among the synMuv genes. Given that some synMuv genes are similar to human tumor-suppressor proteins (e.g., lin-35 Rb), an understanding of these associations could have important implications for our understanding of genes regulating human cancer.
A candidate transcriptional regulatory complex controls one of multiple redundant activities that promote programmed cell death in C. elegans:
We have identified six new C. elegans cell-death genes: mcd-1, dpl-1, efl-1, lin-35, lin-37, and lin-52. These genes are not needed for most cell death; rather, they have contributing roles. The primary cell-death pathway components remain egl-1, ced-9, ced-4, and ced-3. When animals contain mutations in multiple genes that control independent, cell-death-contributing activities, robust cell-death defects are observed. For example, every animal we observed with mutations in dpl-1 and the cell-corpse engulfment gene ced-1 had abnormal cell death in the ventral cord. Therefore, together, cell-death modifiers have very significant roles in the cell-death process. Other independent cell-death-promoting activities might exist and be identified using the strategy presented in this article. An understanding of cell death, a process important for disease and development, must account for the many partially redundant and largely unexplained activities that have been identified (Figure 2D). The genes described in this article are candidates to define components of a transcriptional regulatory complex that contains DP, E2F, Rb, MCD-1, LIN-52, and LIN-37 proteins and that promotes programmed cell death. Given the involvement of misregulated programmed cell death in cancer and neurodegenerative disorders, we have identified a new category of genes as potential therapeutic targets for the manipulation of cell death to treat human disease.
We thank Yamini Jagganath for assistance with screening. We thank Rajesh Ranganthan for cocoordinating the design and construction of the deletion library, members of the Horvitz laboratory for efforts in deletion library construction, and Beth Castor for deletion library screening. We thank Na An for maintaining the Horvitz laboratory strain collection. We thank Brendan Galvin and Hillel Schwartz for comments concerning this manuscript. P.W.R. was supported by a National Science Foundation predoctoral fellowship. E.C.A. is an Anna Fuller Graduate Fellow. H.R.H. is the David H. Koch Professor of Biology at Massachusetts Institute of Technology and an Investigator of the Howard Hughes Medical Institute. This work was supported by National Institutes of Health grant GM24663 to H.R.H. and by the Howard Hughes Medical Institute.
↵1 Present address: Department of Biology, Massachusetts Institute of Technology, Whitehead Institute, 9 Cambridge Center, Cambridge, MA 02142.
↵2 Present address: Department of Neurosurgery, Pasquerilla Healthcare Center, 7th Floor, Georgetown University Hospital, 3800 Reservoir Rd., NW Washington, DC 20007.
Communicating editor: B. J. Meyer
- Received November 11, 2006.
- Accepted January 11, 2007.
- Copyright © 2007 by the Genetics Society of America