DICE1 (deleted in cancer 1), first identified in human lung carcinoma cell lines, is a candidate tumor suppressor, but the details of its activity remain largely unknown. We have found that RNA interference of its C. elegans homolog (DIC-1) produced inviable embryos with increased apoptosis, cavities in cells and abnormal morphogenesis. In the dic-1(RNAi) germ line, ced-3-dependent apoptosis increased, and cell cavities appeared at the late-pachytene/oogenic stage,leading to defective oogenesis. Immunofluorescence microscopy of DIC-1 revealed its ubiquitous expression in the form of cytoplasmic foci, and cryoelectron microscopy narrowed down the location of the foci to the inner membrane of mitochondria. After dic-1 RNAi, mitochondria had an irregular morphology and contained numerous internal vesicles. Homozygous embryos from a heterozygous dic-1 mother arrested at the L3 larval stage, in agreement with the essential role of DIC-1 in mitochondria. In summary, C. elegans DIC-1 plays a crucial role in the formation of normal morphology of the mitochondrial cristae/inner membrane. Our results suggest that human DICE1 may have several functions in multiple intracellular locations.
INTRODUCTION
DICE1 (deleted in cancer 1; INTS6 - Human Gene Nomenclature Database) was originally isolated as a gene responsible for nonsmall cell lung carcinoma resulting from loss of heterozygosity in human chromosome 13q14(Wieland et al., 1999). Although DICE1 mRNA was detected in all human tissues, it was downregulated or absent in cell lines and tissues of non-small cell lung carcinoma. Hypermethylation of its promoter region was associated with its downregulation in several cell lines derived from non-small cell carcinomas(Wieland et al., 2001) and prostate cancers (Röpke et al.,2005). Ectopic expression of DICE1 inhibited colony formation by cell lines of both non-small cell lung cancer and prostate cancer, and suppressed anchorage-independent growth of mouse fibroblasts transformed with the IGF1 receptor gene (Wieland et al.,2004). Overexpression of a mouse DICE1 homolog (DBI1) also suppressed induction of cell proliferation by IGF1(Hoff et al., 1998). DBI1 was located in the nuclear fraction (Hoff et al., 1998), in agreement with the nuclear localization of EGFP-tagged human DICE1 in COS cells(Wieland et al., 2001).
Although DICE1 homologs in humans and mice are thought to be suppressors of cell proliferation, the mode of action of the proteins is largely unknown,because they contain only one functional domain, as defined by current prediction algorithms: the N terminus of each DICE1 homolog has a conserved amino acid sequence encoding a von Willebrand factor type A domain (VWF A)(Fig. 1). This domain is known to be involved in cell-cell, cell-extracellular matrix and intracellular protein-protein interactions, as observed in more than 20 human proteins,including the von Willebrand blood clotting factor, some types of collagens,integrin subunits and leukocyte adhesion receptor (for reviews, see Colombatti and Bonaldo, 1991; Whittaker and Hynes, 2002). Although the presence of a DEAD-box in human DICE1 suggests a helicase function, no DEAD box is present in the C. elegans homolog, and a more extensive helicase motif is not found in any DICE1 homologs. The likelihood that DICE1 acts as a helicase is thus very low, and the conserved VWF A domain does not suggest any other potential role.
In this study, we set out to analyze the knockdown and knockout phenotypes of a C. elegans DICE1 homolog in order to elucidate the molecular function of human DICE1. The protein is essential for C. elegansoogenesis, late embryogenesis and larval growth, and its deficiency induces apoptosis. It is localized to the mitochondrial inner membrane, and abnormalities in both overall and inner mitochondrial morphologies result from its absence. The localization of DICE1 and the phenotypes caused by its deficiency suggest that it is involved in remodeling of the mitochondrial cristae in C. elegans. C. elegans DICE1 shows differences in its subcellular location and function from the human homolog, and our results imply either that they have distinct functions due to divergent evolution, or that the proteins have additional cryptic roles.
MATERIALS AND METHODS
Strains
Bristol N2, a standard wild-type strain, and ced-1(e1735),ced-3(n717), ced-4(n1162), cep-1(gk138),elo-5(ok683)/nT1[qIs51] and ajm-1::gfp (strain SU93) strains were obtained from the Caenorhabditis Genetics Center (Minneapolis,MN, USA) and maintained at 20°C according to standard procedures. dic-1(tm1615) was generated as part of the National Bioresource Project (Japan). Psnap-25:gfp was a gift from Dr Junho Lee(Seoul National University, Seoul, Korea) and unc-15(paramyocin)::gfpwas obtained from Dr Hiroaki Kagawa (Okayama University, Japan). The Pmyo-3::mito::gfp DNA construct was obtained from Dr Alexander M. Van der Bliek (University of California, Los Angeles, CA, USA),an EST clone of the dic-1 gene (yk293c3) from Dr Yuji Kohara(National Institute of Genetics, Japan) and pPD expression vectors from Andrew Fire (Stanford University, CA, USA).
Evolutionary conservation of DICE1. (A) Amino acid sequence alignment of DICE1 homologs in C. elegans, human, mouse and fly. The striped boxes represent von Willebrand factor type A motifs, and the gray and white boxes are homologous and nonconserved regions, respectively. The percentages refer to identical amino acids in the human homolog and the other proteins. (B) The amino acid sequences of the human and C. elegans DICE1 homologs were compared using the Vector NTI Advance program(Invitrogen). Dark and light gray backgrounds of letters signify identical and similar amino acids, respectively.
Evolutionary conservation of DICE1. (A) Amino acid sequence alignment of DICE1 homologs in C. elegans, human, mouse and fly. The striped boxes represent von Willebrand factor type A motifs, and the gray and white boxes are homologous and nonconserved regions, respectively. The percentages refer to identical amino acids in the human homolog and the other proteins. (B) The amino acid sequences of the human and C. elegans DICE1 homologs were compared using the Vector NTI Advance program(Invitrogen). Dark and light gray backgrounds of letters signify identical and similar amino acids, respectively.
Bacteria-mediated RNAi of dic-1
RNAi was performed by feeding C. elegans worms with E. coli HT115(DE3) cells expressing double-stranded RNA (dsRNA) for the dic-1 gene (Kamath et al.,2001). The yk293c3 EST clone of dic-1 (F08B4.1) was digested with HindIII enzyme, and the resulting 1.6 kb cDNA fragment was cloned into vector pPD129.36(L4440), which contains convergent T7 polymerase promoters separated by a multicloning site. The resulting construct was transformed into E. coli HT115(DE3)(Timmons and Fire, 1998). Transformants were grown in LB medium containing 100 μg/ml of ampicillin for 18 hours and seeded onto NGM plates supplemented with 100 μg/ml of ampicillin and 1 mM isopropyl-thio-β-D-galactoside (IPTG). To examine gonads and mitochondria of adult worms, RNAi was performed from the L1 larval stage. To observe the embryonic phenotypes induced by knockdown of dic-1, L4 stage worms were grown on NGM plates covered with E. coli cells producing dsRNA of dic-1 for 24 hours. Adult worms were transferred to new NGM plates, allowed to lay eggs for 6 hours, and then removed. Embryo hatching was scored 24 hour later.
Outcross of the dic-1(tm1615) mutant and rescue of larval arrest by wild type dic-1
The dic-1(tm1615) heterozygous strain from the National Bioresource Project was outcrossed with N2 strains four times to remove possible unrelated mutations. The mutant harbors a deletion of 588 bp,corresponding to nucleotides 32905 to 33492 in the cosmid clone F08B4(Fig. 7). Primers used for PCR reactions to probe the deletion were as follows: outer sense,5′-GGTATTTCGTATTCCATCCACTCC; outer antisense,5′-CCTTGAAATCTCCTGCATAAATAGG; inner sense,5′-ACGAAGCTTCTCAATTGTATCACC; and inner antisense,5′-CCTCAATATCAACTTCACTGTTTCC. To observe the phenotypes of dic-1(tm1615) homozygotes, worms were mounted on slides overlaid with 2% agar and observed with Nomarski optics (DMR HC, Leica). C. elegans dic-1(tm1615) IV/nT1[qIs51] (IV;V) heterozygotes were generated by crossing dic-1(tm1615) heterozygotes with nT1[qIs51]/+ males that were the progeny of a cross between elo-5(ok683)/nT1[qIs51]hermaphrodites and N2 males
To test for rescue of the homozygotes by the wild-type gene, a DNA fragment containing wild-type dic-1 gene from 1.9 kb upstream of the start codon to 1.3 kb downstream of the stop codon was amplified from the F08B4 cosmid genomic DNA clone using the Expand Long Template PCR system (Roche). The amplified DNA fragment was cloned into pCR2.1-TOPO vector (Invitrogen) and the recombinant plasmid DNA (2 μg/ml) was microinjected into the gonads of young adult C. elegans dic-1(tm1615)/nT1[qIs51] heterozygotes,together with pRF4 plasmid DNA (100 μg/ml)(Krammer et al., 1990). Heterozygous transgenic worms in the F1 progeny were selected and maintained by observing their rolling (rol-6) and gfp-positive phenotypes. Rescue of homozygotes by the wild-type dic-1 gene was judged by scoring gfp-negative rollers reaching the adult stage.
To test whether human DICE1 can replace C. elegans DIC-1, a 2.7 kb coding region of human DICE1 gene was amplified from cDNA clone MGC:48751 (Invitrogen) as described above. The amplified cDNA fragment was cloned into pPD49.78 plasmid with an hsp-16.2 promoter, followed by microinjection into the gonads of dic-1(tm1615)/nT1[qIs51]heterozygotes together with pRF4 plasmid DNA. Rolling heterozygotes were allowed to lay embryos and the progeny were treated at the L1 stage by heat shock at 30°C for 1 hour to induce expression of the human DICE1in their extrachromosomal arrays. The rescue of homozygotes by human DICE1 was judged as above, by scoring gfp-negative rollers reaching the adult stage. As a positive control for the rescue, the coding region of C. elegans DIC-1 was amplified from the yk293c3 cDNA clone and cloned into pPD49.78. A transgenic line was obtained and the rescue of larval arrest in homozygotes was confirmed after heat shock induction of the dic-1cDNA expression.
GFP-tagged DIC-1 expression in C. elegans
To visualize the subcellular location of C. elegans DIC-1, the protein was fused with GFP and expressed under the control of either the hsp-16.2 or myo-3 promoter. The 2.6 kb cDNA fragment of dic-1 was cloned into pPD118.26 or pPD114.98, and transgenic lines were generated by microinjecting recombinant plasmid DNA (20 μg/ml)together with pRF4 plasmid DNA (50 μg/ml) into adult N2 worms. Adult worms of the transgenic line Phsp-16.2::dic-1::gfp were heat shocked at 30°C for 1 hour and GFP expression was observed in the progeny embryos 3 hours later.
Counts of apoptotic cells
Cell corpses in the germ line were counted by mounting worms in a drop of M9 solution and observing the gonads with Nomarski optics (DMR HC, Leica). Because corpses are cellularized and more refractive than normal cells, they can be readily identified under high magnification at the late pachytene stage. We observed cell corpses in embryos after mounting 380 cell or bean cell stage embryos on slides coated with 0.1% poly-L-lysine. We also performed TUNEL staining to detect apoptotic cells in the embryos with an In Situ Cell Death Detection Kit (Roche) according to the manufacturer's instructions. Embryos were freeze-cracked, fixed and transferred to the wells of a polylysine-coated slide. TUNEL reaction mixture (10 μl of enzyme solution and 90 μl of label solution per well) was added to the fixed embryos and they were allowed to incubate in the dark for 2 hours at 37°C in a humidified atmosphere. After washing with PBS, the specimens were stained with DAPI (4,6-diamidino-2-phenylindole, 1 μg/ml) for 10 minutes and observed with a fluorescence microscope.
Visualizing mitochondria with mitotracker and using a C. elegans strain expressing GFP in mitochondria
To stain mitochondria in adult worms, L4 stage worms were transferred to NGM plates containing Mitotracker Red (Molecular Probes, 2 μg/ml) and incubated in the dark at 20°C for 12 hours. They were then transferred to new NGM plates to reduce nonspecific staining. After 10 to 20 minutes, they were mounted on a slide overlaid with 2% agar and observed with a fluorescence microscope. The Pmyo-3::mitochondrial signal sequence::gfptransgenic line, in which muscle cell mitochondria are marked by GFP, was used to observe changes in mitochondrial morphology after dic-1RNAi.
Abnormal morphogenesis and tissue-specific gene expression in dic-1 (RNAi) embryos. RNA interference was carried out from the L4 stage by feeding bacteria producing double-stranded RNA.(A) Time lapse microscopy of embryonic development using Nomarski optics. Developmental stage and time post fertilization (in minutes) are indicated for each wild-type embryo. dic-1(RNAi) embryos laid at the same time as the wild-type embryos were arrested after the bean cell stage but before the comma stage. (B) Embryos obtained after RNAi of dic-1 in wild-type N2 and ced-3(n717),after longer incubation than in A. Cavities are marked with arrowheads.(C) Wild-type embryos at the threefold stage and dic-1(RNAi) embryos at the same time post fertilization,showing Psnap-25::gfp expression in neurons, ajm-1::gfp expression in the junctions of intestinal cells and unc-15::gfp expression in muscle cells. Scale bars: 10 μm.
Abnormal morphogenesis and tissue-specific gene expression in dic-1 (RNAi) embryos. RNA interference was carried out from the L4 stage by feeding bacteria producing double-stranded RNA.(A) Time lapse microscopy of embryonic development using Nomarski optics. Developmental stage and time post fertilization (in minutes) are indicated for each wild-type embryo. dic-1(RNAi) embryos laid at the same time as the wild-type embryos were arrested after the bean cell stage but before the comma stage. (B) Embryos obtained after RNAi of dic-1 in wild-type N2 and ced-3(n717),after longer incubation than in A. Cavities are marked with arrowheads.(C) Wild-type embryos at the threefold stage and dic-1(RNAi) embryos at the same time post fertilization,showing Psnap-25::gfp expression in neurons, ajm-1::gfp expression in the junctions of intestinal cells and unc-15::gfp expression in muscle cells. Scale bars: 10 μm.
Antibody preparation
The yk293c3 EST clone of dic-1 (F08B4.1) was digested with EcoRI, and the resulting 1.3 kb cDNA fragment (amino acid residues 143-570) was cloned into pGEX-5X-2 expression vector (Amersham Biosciences). E. coli BL21 cells harboring pGEX-5X-2/dic-1 were cultured at 37°C to an optical density at 600 nm of 0.6, at which time IPTG was added to 1 mM. After incubation for 4 hours, the cells were harvested and sonicated in 10 ml of lysis solution (200 mM Tris-Cl, pH 8.0, 500 mM NaCl, 0.1 mM EDTA, 0.1% Triton X-100, 0.4 mM PMSF). The lysate was spun in a microcentrifuge at 11,000 g for 5 minutes and the supernatant electrophoresed on an 8% SDS polyacrylamide gel. The gel piece containing the fusion protein was excised, crushed in PBS, mixed with Freund's adjuvant and injected into Balb/c mice four times at 10 day intervals (200 μg of protein per injection).
Increased apoptosis in dic-1(RNAi) embryos.(A) Apoptotic cell corpses (arrowheads) were counted in 380 cell or bean cell stage embryos of wild type N2 strain with or without dic-1knockdown. This procedure was repeated with ced-3(n717) strain.(B) TUNEL-positive cells in wild-type and dic-1(RNAi)embryos were scored with a fluorescence microscope at the same embryonic stages as in A. Scale bars: 10 μm. Error bars indicate standard errors of the mean (s.e.m.).
Increased apoptosis in dic-1(RNAi) embryos.(A) Apoptotic cell corpses (arrowheads) were counted in 380 cell or bean cell stage embryos of wild type N2 strain with or without dic-1knockdown. This procedure was repeated with ced-3(n717) strain.(B) TUNEL-positive cells in wild-type and dic-1(RNAi)embryos were scored with a fluorescence microscope at the same embryonic stages as in A. Scale bars: 10 μm. Error bars indicate standard errors of the mean (s.e.m.).
Immunostaining
C. elegans embryos were immunostained by a slightly modified version of the procedure of Crittenden and Kimble(Crittenden and Kimble, 1999). Embryos were freeze-cracked, fixed, incubated with polyclonal mouse antiserum against DIC-1 (1:50 dilution), followed by FITC-conjugated goat anti-mouse immunoglobulin G (1:1000 dilution, Molecular Probes). After staining with DAPI(1 μg/ml), the specimens were observed with a fluorescence microscope. For staining, gonads were extruded from decapitated adult worms, fixed in 3%paraformaldehyde and 0.1 M K2HPO4 (pH 7.2) for 1 hour,post-fixed with cold methanol, and incubated with primary and secondary antibodies as described by Jones et al.(Jones et al., 1996). To double-stain DIC-1 and cytochrome c oxidase, gonads were incubated successively with polyclonal mouse antiserum against DIC-1, FITC-conjugated goat anti mouse immunoglobulin G, and Alex Fluor 594-conjugated mouse immunoglobulin G against bovine cytochrome c oxidase subunit 1 (Molecular Probes).
Cryo-electron microscopy
C. elegans worms were plunged into liquid propane at 180°C. The frozen samples were then transferred to substitution medium (acetone) and freeze-substituted at 80°C for 72 hours, after which the temperature was raised to 40°C at a rate of 6°C hr-1. The samples were washed three times in precooled pure acetone over a period of 1 hour, and then infiltrated with a spur resin (EMS). Infiltration was performed with acetone/resin mixtures (v/v) of 3:1, 1:1 and 1:3 for 1 hour each and with the pure resin for 3 hours. The resin was polymerized at 60°C for 48 hours. Sections about 60 nm thick were cut horizontal to the plane of the sample using an ultra-microtome (RMC MTXL) and mounted on nickel grids. They were stained with uranyl acetate and lead citrate and viewed at 120 kV with a Technai 12 electron microscope (Philips, Netherlands).
For immunogold labeling, the ultra-thin sections on nickel grids were etched with 10% hydrogen peroxide for 30 minutes, rinsed in deionized water and floated on 0.56 mM sodium metaperiodate. They were then rinsed with deionized water and floated for 1 hour in PBS containing 1% bovine serum albumin (PBS-BSA), followed by overnight incubation at 4°C in antimouse DIC-1 diluted to 1:200 with PBS-BSA. After rinsing several times with PBS-BSA,they were floated for 1 hour in anti-mouse IgG conjugated to 10 nm gold particles (British Biocell International) diluted to 1:50 with PBS-BSA. The gold-labeled sections were rinsed successively with PBS, PBST (containing 1%Triton X-100) and deionized water, air dried, stained and examined as above by electron microscopy.
RESULTS
dic-1(RNAi) embryos are defective in morphogenesis,and form cell cavities
A C. elegans DICE1 homolog (DIC-1) is encoded by the F08B4.1 open reading frame (Wormbase, http://www.wormbase.org)and has about 30% identity in its N-terminal, central and C-terminal domains with the corresponding regions of the human homolog, as depicted in Fig. 1. In order to determine the function of DIC-1 in C. elegans (DIC-1) during C. elegans development, we used RNA interference (RNAi) from the L1 or L4 stage. This was accomplished by feeding C. elegans worms with E. coli cells expressing the cognate double-stranded RNA (dsRNA). RNA interference from the L4 stage resulted in complete lethality of progeny embryos. Time-lapse microscopy using Nomarski optics showed that embryogenesis was normal at the early stage in the dic-1(RNAi) strain,whereas it was abnormal from the bean cell stage(Fig. 2A). Even after prolonged incubation, dic-1(RNAi) embryos never reached the normal comma stage. In addition to their blocked morphogenesis, they contained cell cavities after longer incubation, as shown in Fig. 2B. The formation of these cavities was independent of ced-3, suggesting that they were not generated by the major C. elegans apoptosis pathway.
Increased apoptosis and cell cavity formation in dic-1(RNAi) gonads. (A) Apoptotic cell corpses were counted in the germline with Nomarski optics after feeding C. elegans worms with dsRNA encoding dic-1 for 48 hours from the L4 stage. Germ cell corpses in the loop region of the gonad are indicated by the boxed area. The effects of apoptosis regulators ced-3(lf),ced-3(n717); ced-4(lf), ced-4(n1162); cep-1(lf) and cep-1(gk138) on the induction of apoptosis by dic-1(RNAi) are plotted in the graph. Error bars indicate standard errors. (B) RNAi was performed from the L1 stage and resulted in various phenotypes in the gonad, including undersized gonads,defective oocyte development and cavities. Cavities (brackets) were first observed in the young adult stage gonad but later expanded to the distal tip region of the gonad at the adult stage. ced-3(lf), ced-3(n717). V,vulva; Ds, distal tip region; Pr, proximal region. Scale bars: 25 μm.
Increased apoptosis and cell cavity formation in dic-1(RNAi) gonads. (A) Apoptotic cell corpses were counted in the germline with Nomarski optics after feeding C. elegans worms with dsRNA encoding dic-1 for 48 hours from the L4 stage. Germ cell corpses in the loop region of the gonad are indicated by the boxed area. The effects of apoptosis regulators ced-3(lf),ced-3(n717); ced-4(lf), ced-4(n1162); cep-1(lf) and cep-1(gk138) on the induction of apoptosis by dic-1(RNAi) are plotted in the graph. Error bars indicate standard errors. (B) RNAi was performed from the L1 stage and resulted in various phenotypes in the gonad, including undersized gonads,defective oocyte development and cavities. Cavities (brackets) were first observed in the young adult stage gonad but later expanded to the distal tip region of the gonad at the adult stage. ced-3(lf), ced-3(n717). V,vulva; Ds, distal tip region; Pr, proximal region. Scale bars: 25 μm.
We probed tissue-specific morphogenesis using the transgenic strains Psnap-25::gfp, ajm-1::gfp and unc-15(paramyocin)::gfp, which express GFP in neurons, in the adherens junctions of pharynx, intestine and hypodermis, and in muscle cells,respectively (Fig. 2C). SNAP-25 is a component of synaptosomes expressed in C. elegans neurons(Hwang and Lee, 2003). Fig. 2C illustrates tissue-specific gene expression in wild-type (threefold) and dic-1(RNAi) embryos at the same time post fertilization,showing similar overall intensity of GFP fluorescence. Although the dic-1(RNAi) embryos did not pass the comma stage morphologically, the extent of GFP expression suggests that tissue-specific differentiation progressed further; however, the differentiated cells were disorganized, in agreement with the morphogenic abnormalities seen in Fig. 2A.
ced-3-dependent apoptosis is induced in dic-1(RNAi) embryos and germ line
Although cell cavity formation in dic-1(RNAi) embryos did not depend on the major apoptosis pathway(Fig. 2B), we tested whether DIC-1 deficiency led to an increased frequency of apoptosis. As dic-1(RNAi) embryos arrested before the comma stage,apoptotic cells were scored at the 380 cell and bean cell stages either by direct observation of cell corpses under Nomarski optics(Fig. 3A) or after TUNEL staining (Fig. 3B). By both measures, ced-3(n717)-dependent apoptosis increased about twofold at the 380-cell stage and 1.6-fold at the comma stage because of knockdown of dic-1. Increased apoptosis was also observed in adult gonads upon RNAi of dic-1 from the L4 stage(Fig. 4A). This germline apoptosis was abrogated by the ced-3(n717) mutation, as in embryos,and also by the ced-4(n1162) mutation, indicating that it was induced via the major apoptosis pathway. It was not affected in the cep-1(gk138) background, which rules out DNA damage as the main cause of the apoptosis.
dic-1(RNAi) blocks oogenesis and generates cell cavities in the germ line
When RNAi of dic-1 was performed from the L1 stage,various phenotypes appeared, including undersized gonads, inhibition of oogenesis and cavities in the gonad (Fig. 4B). No apparent defects in germline development were observed under Nomarski optics from the mitotic to the pachytene region, apart from some reduction in the number of germ cells. Nevertheless, slightly disturbed immunolocalization was observed in dic-1(RNAi) gonads for syntaxin 4 in the cytoplasmic membrane, and for porin in the nuclear membrane(S.M.H., unpublished). Although meiotic prophase progression was detected by immunostaining of GLD-1 at the pachytene stage, synapsed chromosomal strands were not as clear in the dic-1(RNAi) gonad after DAPI staining as in the wild type (S.M.H., unpublished). As in embryos, gonadal DIC-1 deficiency had a marked effect on actively differentiating cells such as late pachytene stage cells and oocytes, but not on rapidly dividing cells. We observed GFP at the rim of the cavities in a transgenic derivative of dic-1(RNAi) strain expressing annexin V::GFP, which binds to phosphatidylserine in the outer layer of the cytoplasmic membrane of apoptotic and necrotic cells (S. Züllig and T.H.L., unpublished). This suggests that the cavities resulted from apoptotic or necrotic cell death and were not merely vacuoles.
DIC-1 protein is localized to the mitochondrial inner membrane
When C. elegans embryos were reacted with a polyclonal mouse antibody raised against DIC-1, the protein appeared as speckles in the cytoplasm throughout embryogenesis (Fig. 5A). It was present in germ cells and oocytes, and its particulate distribution in the cytoplasm was clearly observed(Fig. 5B,C). At the post-embryonic stage, DIC-1 was present in all somatic cells, although at a lower expression level than in germline cells (S.M.H., unpublished). The speckles of DIC-1 in mitotic germ cells co-localized with anti-cytochrome c oxidase subunit 1 antibody conjugated to Alexa Fluor 594(Fig. 5C), supporting a mitochondrial location of DIC-1. The mitochondrial localization was confirmed by expressing GFP-tagged DIC-1 under the control of the hsp16.2 or myo-3 promoter. The fusion protein localized in the cytoplasm of embryonic cells as foci, as well as in the mitochondria of muscle cells (see Fig. 6), like the GFP fused to a mitochondrial localization signal sequence in Fig. 7B. In agreement with these results, DIC-1 proved to be located in mitochondria by cryo-electron microscopy, more precisely on the inner membrane at cristae(Fig. 5D). The TMpred program(http://www.ch.embnet.org/software/TMPRED_form.html)predicts the presence of two transmembrane regions in DIC-1, encompassing amino acids 129-148 and 377-397, thus supporting the localization in the membrane.
Immunolocalization of DIC-1 in mitochondria. (A-C)Immunolocalization of DIC-1 in embryos and germline by fluorescence microscopy. (A) Wild-type and dic-1(RNAi) embryos.(B) In wild-type gonads, DIC-1 was restricted to the cytoplasm from the distal tip cells to the oocytes. (C) Co-localization of DIC-1 with cytochrome c oxidase subunit 1 within mitotic cells of the gonad. (D)Electron micrograph (left) showing immunogold particles (arrowheads) in mitochondria: the bracketed area is enlarged in the right micrograph. Scale bars: 10 μm in A; 50 μm in B; 25 μm in C; 0.2 μm in D.
Immunolocalization of DIC-1 in mitochondria. (A-C)Immunolocalization of DIC-1 in embryos and germline by fluorescence microscopy. (A) Wild-type and dic-1(RNAi) embryos.(B) In wild-type gonads, DIC-1 was restricted to the cytoplasm from the distal tip cells to the oocytes. (C) Co-localization of DIC-1 with cytochrome c oxidase subunit 1 within mitotic cells of the gonad. (D)Electron micrograph (left) showing immunogold particles (arrowheads) in mitochondria: the bracketed area is enlarged in the right micrograph. Scale bars: 10 μm in A; 50 μm in B; 25 μm in C; 0.2 μm in D.
dic-1 knockdown results in morphologically aberrant mitochondria containing numerous vesicles
When wild-type C. elegans adults were treated with Mitotracker,which stains active mitochondria, aligned tubular mitochondria were observed parallel to the myofilaments in the muscle cells of wild-type worms(Fig. 7A). By contrast, the muscle cells of dic-1(RNAi) worms frequently showed disorganized alignment of mitochondria of irregular morphology. The aberrant morphology and disorganized distribution of mitochondria were also observed in transgenic dic-1(RNAi) worms, in which GFP was targeted to the mitochondrial matrix (Fig. 7B). The more detailed substructure of the morphologically abnormal mitochondria was revealed by cryo-electron microscopy(Fig. 7C). The mitochondria were smaller and packed with numerous vesicles in dic-1(RNAi) muscle cells, in contrast to the tubular cristae of wild-type worms. The localization of DIC-1 in the inner mitochondrial membrane and the extensive reticular network of cristae between vesicles in dic-1(RNAi) mitochondria show that DIC-1 is essential for the formation of normal morphology of cristae/inner membrane.
Subcellular localization of GFP-tagged DIC-1 in embryonic and muscle cells by fluorescence microscopy. (A) Expression of GFP-tagged DIC-1 under the control of the hsp-16.2 promoter was induced by heat shock in N2 worms and spotted expression was observed in the cytoplasm of early embryos. (B) GFP-tagged DIC-1 was expressed under the control of the myo-3 promoter in N2 worms and expression was observed in the mitochondria of body-wall muscle cells. N, nucleus. Scale bars: 10 μm in A;5 μm in B.
Subcellular localization of GFP-tagged DIC-1 in embryonic and muscle cells by fluorescence microscopy. (A) Expression of GFP-tagged DIC-1 under the control of the hsp-16.2 promoter was induced by heat shock in N2 worms and spotted expression was observed in the cytoplasm of early embryos. (B) GFP-tagged DIC-1 was expressed under the control of the myo-3 promoter in N2 worms and expression was observed in the mitochondria of body-wall muscle cells. N, nucleus. Scale bars: 10 μm in A;5 μm in B.
A dic-1 deletion mutant is arrested at the L3 larval stage and can be rescued by the wild-type gene
The open reading frame of dic-1 is predicted by WormBase to consist of 14 exons, but we identified 16 exons when we sequenced the cDNA of the EST clone yk293c3 (Fig. 8A). The dic-1(tm1615) strain, obtained from the National Bioresource project for C. elegans, was generated by deleting a region of exon 4 and all of exons 5 and 6, and this deletion is very likely to cause premature termination of translation. The homozygous progeny of a heterozygous mother hatched, but they remained arrested at the L3 larval stage for several days, and eventually died. Under Nomarski optics, they were found to have cavities, similar to the cavities in dic-1(RNAi)embryos and gonads, over much of their bodies(Fig. 8B). In order to determine whether the mutant phenotype could be rescued by the wild-type gene,we generated a dic-1(tm1615)/nT1[qIs51] heterozygous strain containing extrachromosomal arrays of wild-type dic-1 gene and the rol-6 gene. Among the progeny of these heterozygotes, 24%(n=554) had a rolling phenotype and 12% of the rollers that reached the adult stage were dic-1 homozygotes (lacking nT1[qIs51]balancer chromosome for pharyngeal GFP expression). On the basis of complete rescue by the wild-type gene, we expect 33% of the adult progeny to be dic-1 homozygotes. The lower percentage of rescue indicates incomplete rescue, presumably due to mosaicism of the extra-chromosomal array in somatic cells. The rescued adult homozygotes produced almost no viable progeny, probably because the dic-1 gene in the extra-chromosomal array was silenced in the germ line. By contrast, a mere 0.3% of the non-roller adults (n=580) had a non-fluorescent pharynx, signifying dic-1 homozygosity.
L3 growth arrest has been observed in mutants defective in mitochondrial proteins (Tsang et al., 2001)and after ethidium bromide treatment, which inhibits mitochondrial DNA replication (Tsang and Lemire,2002). This L3 arrest is related to the energy requirement for entry into the L4 stage, and the fivefold increase in the number of mitochondrial DNA copies at this time(Tsang and Lemire, 2002). The L3 arrest of dic-1(tm1615) homozygotes is yet further evidence that DIC-1 is crucial for mitochondrial activity. When RNAi of dic-1 was performed from the L1 stage, a minor fraction of the worms arrested at the L4 or early adult stage, not at the L3 stage. The incompleteness of RNAi and the later disappearance of zygotic dic-1expression may have contributed to the much weaker post-embryonic somatic phenotype of the knockdown strain compared with the knockout mutant.
Expression of human DICE1 does not rescue the larval arrest of dic-1 homozygotes
In order to determine whether human DICE1 is a functional homolog of C. elegans DIC-1, we generated a dic-1(tm1615)/nT1[qIs51]heterozygous strains containing extrachromosomal arrays of human DICE1 cDNA (or C. elegans dic-1 cDNA as a positive control)linked to the hsp-16.2 promoter and the rol-6 gene. When heterozygotes transformed with C. elegans dic-1 cDNA were allowed to produce progeny, followed by heat shock at the L1 stage, 11% of the roller progeny (n=682) that reached adulthood were dic-1homozygotes. This showed that heat shock-induced dic-1 cDNA expression rescued the mutant phenotype as effectively as the wild-type dic-1 gene expression described above. By contrast, none of the adult roller progeny (n=203) was dic-1 homozygotes when the roller progeny were from heterozygotes transformed with human DICE1 cDNA. Thus, we were unable to obtain evidence that human DICE1 and C. elegans DIC-1 are functionally equivalent. Whether this failure reflects technical limitations with the expression or stability of the human protein in worms, the inability of human DICE1 to interact with essential C. elegans partners due to divergence, or a difference in function of the two proteins is not known.
DISCUSSION
DIC-1 is involved in remodeling of the mitochondrial cristae
DIC-1 is expressed ubiquitously as cytoplasmic speckles, and is at a somewhat higher level in oocytes and embryos (Figs 5, 6). We narrowed down its location to the mitochondrial inner membrane by immunogold electron microscopy(Fig. 5D). In agreement with this, dic-1(RNAi) mitochondria had an irregular morphology and disorganized distribution, and contained numerous vesicles in the inner compartment (Fig. 7). Similar structures with reticular cristae or intra-mitochondrial vesicles have been observed in the mitochondria of other species deficient in a mitochondrial fusion protein or overexpressing a fission protein(Griparic et al., 2004; Messerschmitt et al., 2003; Olichon et al., 2003). Fusion and fission of mitochondria occurs dynamically in the cell, via fusion and fission of its membranes (Pfanner et al.,2004; Youle and Karbowski,2005). In yeast, Mgm1 in the inner membrane, and Ugo1 and Fzo1 in the outer membrane are involved in the fusion process(Pfanner et al., 2004; Wong et al., 2003). Interestingly, when expression of the mammalian homolog of Mgm1, OPA1, was reduced in HeLa cells by siRNA, reticular cristae were formed(Griparic et al., 2004; Olichon et al., 2003). S. cerevisiae Mdm33 is involved in the opposite process, fission of the mitochondrial inner membrane, and its overexpression leads to inner vesicle formation, as in DIC-1 deficiency, and to partitioning of the matrix by membrane septa (Messerschmitt et al.,2003).
Disrupted morphology and inner structure of mitochondria in DIC-1(RNAi) muscle cells observed by fluorescence microscopy and cryo-electron microscopy. (A) The musculature of dic-1(RNAi) worms shows abnormal morphology of active mitochondria stained with Mitotracker Red. (B) The mitochondrial matrix visualized with a mitochondrial matrix-targeting sequence::GFP construct controlled by the myo-3 promoter. (C) Electron micrographs of mitochondria (enclosed by brackets) in wild-type and dic-1(RNAi) muscle cells. Scale bars: 10 μm in A; 5 μm in B; 0.2 μm in C.
Disrupted morphology and inner structure of mitochondria in DIC-1(RNAi) muscle cells observed by fluorescence microscopy and cryo-electron microscopy. (A) The musculature of dic-1(RNAi) worms shows abnormal morphology of active mitochondria stained with Mitotracker Red. (B) The mitochondrial matrix visualized with a mitochondrial matrix-targeting sequence::GFP construct controlled by the myo-3 promoter. (C) Electron micrographs of mitochondria (enclosed by brackets) in wild-type and dic-1(RNAi) muscle cells. Scale bars: 10 μm in A; 5 μm in B; 0.2 μm in C.
In addition to causing the formation of reticular cristae, downregulation of OPA1 in HeLa cells induced apoptosis(Lee et al., 2004; Olichon et al., 2003),implying a functional similarity between DIC-1 and OPA1. Nevertheless,knockdown of EAT-3, a C. elegans homolog of human OPA1 and of yeast Mgm1, resulted in mitochondrial fragmentation (T.H.L., unpublished), unlike the disorganized and interconnected mitochondrial distribution evident in Fig. 7B. It also resulted in a reduced number of morphologically normal but longer cristae in cryo-electron micrographs (J.Y.M., unpublished). Therefore, DIC-1 is not likely to be involved in fusion of the inner mitochondrial membrane. When FIS1 or DRP1,both of which are involved in fission of the mitochondrial outer membrane, was knocked down in HeLa cells, the mitochondria became elongated and apoptosis was reduced (Lee et al.,2004). The differences between these phenotypic effects and those due to DIC-1 deficiency exclude the possibility that DIC-1 participates in fission of the mitochondrial membrane. Besides fusion and fission proteins,such proteins as mitofilin and e and g subunits of mitochondrial ATP synthase control cristae/inner membrane morphology. In mitochondria of HeLa cells deficient in mitofilin, which normally forms a large multimeric protein complex in the intermembrane space, concentrically packed inner membranes were interconnected at numerous sites (John et al., 2005). This `onion-like' structure was also induced in S. cerevisiae cells by decreased expression of e or g subunit of the ATP synthase, both of which are involved not in ATP synthesis but in the oligomerization of the ATP synthase(Arselin et al., 2004). Like mitofilin and the ATP synthase e and g subunits, DIC-1 is thought to be involved rather in remodeling (as a comprehensive term) of the mitochondrial cristae/inner membrane, not in its fusion/fission.
L3 stage growth arrest of dic-1(tm1615). (A)Schematic representation of the dic-1 gene and the region deleted in the dic-1(tm1615) mutation. (B) A dic-1(tm1615) L3 larva with cavities in all tissues observed with Nomarski optics. Cavities are marked with arrowheads. Scale bar: 50 μm.
DIC-1 is more crucial at the differentiation stages of oogenesis and embryogenesis
RNA interference of dic-1 led to ced-3-mediated induction of apoptosis in the gonad and embryos. Cell cavities were also generated, but their formation was not dependent on ced-3, suggesting that they were generated by necrosis or a minor apoptosis pathway. Cavities were also found in somatic tissues during larval development of the dic-1(tm1615)mutant. It is intriguing that similar vacuolar structures were observed throughout the body of a transgenic line overexpressing mutated forms of NUO-1(NADH- and FMN-binding subunit of complex 1 of the respiratory chain)(Grad and Lemire, 2004). Apoptosis caused by deficiency of dic-1 was observed at late pachytene and midembryonic stages, the developmental stages in which physiological apoptosis normally occurs. Cavity formation was observed at the same or similar stages of germline development and embryogenesis, suggesting a crucial role for DIC-1 at these stages. During the late pachytene/oogenic and the mid-embryonic stages, cells undergo active differentiation to form oocytes and specific tissues, respectively. Efficient mitochondrial activity is probably more crucial in these differentiating cells than in actively dividing cells, as has been observed in a dif-1 mutant(Ahringer, 1995). DIF-1 is a homolog of mitochondrial carrier proteins involved in solute transport through the inner mitochondrial membrane and is essential during a 3-hour period of active embryonic morphogenesis.
Human DICE1 differs in function and location from its C. elegans homolog
In summary, the DICE1 homolog in C. elegans is essential for normal mitochondrial morphology and function, and its deficiency induces apoptosis and inhibits development at a number of stages. The major differences between C. elegans and mammalian DICE1 homologs are in their subcellular locations and effects on cell proliferation. Although the C. elegans protein localizes to mitochondria, the human and mouse homologs localize to nuclei (Hoff et al.,1998; Wieland et al.,2001). When their amino acid sequences were analyzed using the pTARGET program(http://bioinformatics.albany.edu/~ptarget/), C. elegans and human DICE1 are predicted to localize in mitochondria at the confidence levels of 93.9% and 75.1%, respectively. However, nuclear localization is predicted for both of the proteins by the pSORT program(http://psort.ims.u-tokyo.ac.jp/)at the probabilities of 73.9% (in C. elegans) and 82.6% (in humans). In agreement with this prediction, bipartite nuclear targeting sequences are identified by the PROSITE program(http://www.expasy.org./prosite/)in C. elegans (amino acids 677-693 and 693-709) and human DICE1(amino acids 643-659). Although an exclusive localization of DIC-1 in mitochondria has been observed in this study, it is possible that it translocates to the nucleus under certain conditions. Recently, p53 was found to translocate to mitochondria, as well as to nuclei upon activation by DNA damage or oxidative stress, and it was localized in the matrix and outer membrane of the mitochondria (Chipuk et al., 2004; Zhao et al.,2005).
In addition to the difference in subcellular localization, the C. elegans and mammalian DICE1 homologs have distinct effects on cell proliferation and differentiation. The C. elegans protein has an anti-apoptotic role and is more important in differentiating cells than in actively dividing ones, in contrast to the role of its mammalian homologs as suppressors of cell proliferation (Hoff et al., 1998; Wieland et al.,2004). When dic-1(tm1615) heterozygotes were transformed with a human DICE1 cDNA under the control of the C. elegans hsp-16.2 promoter and expression was induced by heat shock at the L1 stage, no rescue of L3 larval arrest was observed in the homozygous dic-1(tm1615) progeny. This was probably due to differences in the subcellular location and function of the human homolog and C. elegansDIC-1. However, it is also possible that a low level expression of the human DICE1 cDNA or instability of the expressed protein could have caused the failure. Human DICE1 was recently reported to be a component of the integrator complex involved in small nuclear RNA processing(Baillat et al., 2005), in agreement with its nuclear localization. Nonetheless, the level of DICE1 mRNA was found to be higher in dominant follicle cells of cow ovaries than in the subordinate follicle cells that subsequently underwent apoptosis(Evans et al., 2004). This result, obtained with cDNA microarrays, suggests an anti-apoptotic role for DICE1 in cows, as observed for its C. elegans homolog in the present study. In summary, C. elegans DIC-1 plays a crucial role in the formation of normal morphology of mitochondrial cristae/inner membrane, unlike the mammalian homologs that localize in nuclei and are involved in the suppression of cell proliferation. Our results on the C. elegansDIC-1 suggest that a re-evaluation and further investigation on mammalian DICE1 may elucidate its multiple functions and/or very complex intracellular dynamics.
Acknowledgements
C. elegans strains N2, ced-1(e1735),ced-3(n717), ced-4(n1162), cep-1(gk138),elo-5(ok683)/nT1[qIs51] and ajm-1::gfp(strain SU93) were obtained from the C. elegans Genetics Center (St Paul, MN, USA),which is supported by the National Center for Research Resources. The dic-1(tm1615) strain was kindly generated by the National Bioresource Project (Japan) at our request. We thank Dr Junho Lee (Seoul National University, Seoul, Korea) for Psnap-25:gfp, Dr Hiroaki Kagawa (Okayama University, Japan) for unc-15::gfp, Dr Alexander M. Van der Bliek (University of California, Los Angeles, USA) for the Pmyo-3::mito::gfp DNA construct, Dr Yuji Kohara (National Institute of Genetics, Japan) for the yk293c3 clone of dic-1, Dr Alan Coulson (Sanger Center, UK) for the F08B4 clone and Andrew Fire (Stanford University) for pPD expression vectors. Suk Huh (Yonsei University, Korea)made a few plasmid constructs for the phenotype rescue of dic-1homozygotes. This work was supported by a Molecular and Cellular BioDiscovery Research Program grant (M1-0311-00-0070) from the Ministry of Science and Technology (to H.S.K.), and also by grants from the Swiss National Science foundation, the Ernst Hadorn Foundation and the Josef Steiner Foundation (to M.O.H.).