1069RESEARCH ARTICLE

INTRODUCTIONApoptotic cells generated by programmed cell death are recognizedand cleared by neighboring cells or professional phagocytes.Efficient clearance of apoptotic cell is crucial for tissue homeostasisand the regulation of immune responses. Defects in this processcontribute to persistent inflammatory diseases and autoimmunedisorders (Savill et al., 2002; Savill and Fadok, 2000). Phagocytosisof apoptotic cell is regulated by mechanisms that are highlyconserved from the nematode C. elegans to humans (Fadeel, 2003).For example, cells that undergo programmed cell death in C. elegansare quickly removed by neighboring cells (Sulston and Horvitz,1977). Two partially redundant pathways have been identified inworm that regulate the cell corpse engulfment process, with ced-1,ced-6, ced-7 and dyn-1 functioning in one pathway and psr-1, ced-2, ced-5, ced-12 and ced-10 in the other (Reddien and Horvitz, 2004;Wang et al., 2003; Yu et al., 2006).

CED-1 is a transmembrane receptor that mediates the recognitionof the dying cell by engulfing cells, but the ligand of CED-1 has notbeen identified (Zhou et al., 2001b). CED-7, a homolog of themammalian ABC transporter, functions in both dying cells andengulfing cells to promote cell corpse engulfment and is required forthe recognition of the dying cell by CED-1 (Wu and Horvitz, 1998;Zhou et al., 2001b). CED-6 is an adaptor protein that contains aphosphotyrosine-binding (PTB) domain and may interact directlywith CED-1 to transduce the engulfment signal (Liu and Hengartner,1998; Su et al., 2002). Recently, C. elegans Dynamin 1 (DYN-1)was found to function downstream of CED-1, CED-7 and CED-6 topromote vesicle delivery to the phagocytic cup for the internalizationof the cell corpse (Yu et al., 2006).

In the other pathway, surface-exposed phosphatidylserine (PS),an ‘eat me’ signal, is recognized by PSR-1, the C. elegans homologof the human phosphatidylserine receptor (Fadok et al., 2000; Wang

et al., 2003), which transduces the signal through a ternary signalingcomplex consisting of CED-5/DOCK180 (DOCK1), CED-12/ELMO and CED-2/CRKII to activate CED-10/RAC1, a smallGTPase. The activation of CED-10 then triggers the reorganizationof cytoskeleton needed for engulfment of apoptotic cells (Reddienand Horvitz, 2004). A recent study suggests that CED-10 might alsoact downstream of CED-1, CED-6 and CED-7 to promote cellcorpse engulfment (Kinchen et al., 2005). Despite the identificationof these genes involved in cell corpse engulfment, many crucialcomponents required for this process are still missing, includinggenes involved in the degradation of cell corpses. In addition, howphagosomes form and mature, and how internalized cell corpses aredegraded remain unclear.

Phagocytosis is a receptor-mediated, actin-dependent processthat results in internalization of foreign particles or apoptoticcells. The internalized vesicle, the phagosome, matures throughinteraction with organelles of the endocytic pathway to generatethe phagolysosome, which is capable of degrading particles orapoptotic cells (Desjardins et al., 1994; Henry et al., 2004; Vieiraet al., 2002). In C. elegans, internalized cell corpses are enclosedby the phagosome, which may undergo a similar maturationprocess. Although phagosome composition and maturation havebeen extensively studied in mammalian cells using latex-bead-containing phagosomes (Garin et al., 2001; Stuart et al., 2007),the formation and maturation of the phagosome that lead tothe degradation of apoptotic cells in vivo, remain poorlyunderstood.

In the present study, we have identified C. elegans UNC-108 asa novel component involved in the degradation of apoptotic cells.Both loss-of-function by RNA interference (RNAi) and a gain-of-function mutant of unc-108, sm237, resulted in accumulation ofcell corpses. Furthermore, we showed that cell corpses persistingin the unc-108(sm237) mutant or unc-108(RNAi) animal areinternalized, but not degraded. UNC-108 co-localizes with theendolysosomal markers RAB-5, RAB-7 and LMP-1 to thephagosome in C. elegans embryos. We also present evidence thatunc-108 is required for endosomal trafficking, affecting thetransition from the early to the late endosome, the recyclingendosome and the maturation of lysosome. Our results suggestthat UNC-108 promotes cell corpse degradation, possibly by

C. elegans Rab GTPase 2 is required for the degradation ofapoptotic cellsQun Lu1,3,*, Yan Zhang2,3,*, Tianjing Hu3,*, Pengfei Guo3, Weida Li3 and Xiaochen Wang3,†

During apoptosis, the dying cell activates an intrinsic mechanism that quickly dismantles itself. The apoptotic cell corpses are thenrecognized and removed by neighboring cells or professional phagocytes. How dying cells are degraded after internalization ispoorly understood. Here, we report the identification and characterization of unc-108, the Caenorhabditis elegans homolog of thehuman Rab GTPase 2, as a novel component involved in the degradation of apoptotic cells. unc-108 is expressed and functions inthe engulfing cells and is likely to affect the degradation rather than the internalization of cell corpses. Similar to other RabGTPases, unc-108 also affects endocytosis, acting in the endosomal trafficking from early to late endosome and late endosome tolysosome. UNC-108 co-localizes with RAB-5, RAB-7 and LMP-1 to the phagosome and promotes cell corpse degradation, possibly bymediating phagosome maturation.

KEY WORDS: C. elegans, Apoptotic cell, Degradation, unc-108, Rab GTPase 2, Endocytosis, Phagosome

Development 135, 1069-1080 (2008) doi:10.1242/dev.016063

1College of Biological Sciences, China Agricultural University, Beijing 100094, China.2Graduate Program in Chinese Academy of Medical Sciences and Peking UnionMedical College, China. 3National Institute of Biological Sciences, No. 7 Science ParkRoad, Zhongguancun Life Science Park, Beijing, 102206, China.

*These authors contributed equally to this work†Author for correspondence (e-mail: wangxiaochen@nibs.ac.cn)

Accepted 18 January 2008 DEVELO

PMENT

1070

mediating phagosome maturation, and is a novel componentcrucial for the post-engulfment/cell corpse degradation process inC. elegans.

MATERIALS AND METHODSC. elegans strainsStrains of C. elegans were cultured at 20°C using standardprocedures (Brenner, 1974). The N2 Bristol strain was used as the wild-type strain, except for polymorphism mapping that used Hawaiian strainCB4856.

Mutations used are described in C. elegans II (Riddle et al., 1997) unlessotherwise indicated. Linkage group I (LGI): dpy-5(e61), unc-29(e403), unc-11(e47), ced-1(e1735), ced-12(n3261) (Zhou et al., 2001), n501, n777,hT2(bli-4(e937)let-?(q782)qIs48)/sep-1(e2406) and ok1246 (Wormbase:www.wormbase.org), sm237 (this study). LGIII: ced-6(n2095), ced-7(n2094), ced-4(n1162). LGIV: ced-3(n717), psr-1(tm469) (Wang et al.,2003), ced-2(n1994), ced-5(n1812), ced-10(n3246).

The following strains carrying integrated transgenes were kindly providedby Dr Hanna Fares: bIs34 (RME-8::GFP) (Zhang et al., 2001); cdIs73(RME-8::mRFP) (Treusch et al., 2004); cdIs40 (pcc1:GFP::CUP-5)(Treusch et al., 2004); cdIs97 (pcc1:mCHERRY::CUP-5); cdIs39[pcc1:RME-1(271a1)] (Poteryaev et al., 2007); bIs46 (GFP::RME-1; pRF4)(Grant et al., 2001); cdIs141 (pcc1:mCHERRY::RAB-7); and cdIs113(pcc1:mCHERRY::RAB-5).

Other endocytosis markers used were: arIs37 (Pmyo-3ssGFP) (Fares andGreenwald, 2001a); pwIs50 (LMP-1::GFP) (Treusch et al., 2004); andbIs1(VIT-2::GFP) (Grant and Hirsh, 1999).

Mapping and cloning of unc-108sm237 was mapped very close to unc-11 on the left arm of linkage group I.From unc-11 dpy-5/sm237 mothers, 60 of 77 Dpy non-Unc recombinantscontained sm237, whereas 0 of 44 Unc non-Dpy recombinants containedsm237. We then performed single nucleotide polymorphism (SNP) mappingto locate sm237 between SNP markers snp-Y47G6A (–3.21) and snp-R12E2(–1.64). Transformation rescue experiments showed that one fosmid in thisregion, WRM0636aD05, rescued the persistent cell corpse phenotype of thesm237 mutant. Several deletion clones of WRM0636aD05 were generatedand one subclone that contains the MluI-SacII fragment of WRM0636aD05possessed the rescue activity. Only one intact open reading frame, F54F10.4,was found in this region, which corresponds to a previously identified gene,unc-108. We determined the sequence of unc-108 in the sm237 mutant andidentified a missense mutation that caused the substitution of Gly18 withGlu.

Quantification of cell corpsesThe number of somatic cell corpses in the head region of living embryos orL1 larvae and the number of germ cell corpses in one gonad arm fromanimals at various adult ages were scored using Nomarski optics asdescribed (Gumienny et al., 1999; Wang et al., 2002).

unc-108 RNAiSense and antisense RNA were in vitro transcribed from the T7- and SP6-flanked PCR template (unc-108 cDNA nucleotides 5-599) using RiboMAXLarge Scale RNA Production System (Promega, USA). Double-strandedRNA (dsRNA) was generated by annealing the sense and antisense RNA for10 minutes at 65°C, followed by incubating at 37°C for 15 minutes. dsRNAof unc-108 (550 ng/�l) was injected into the gonad or the body cavity ofwild-type animals, which were then transferred to fresh OP50-seeded platesevery 12 hours. Embryos laid between 36 and 48 hours post-injection wereused for analyzing the somatic cell corpse phenotype. To determine theengulfment phenotype in germ cells, wild-type animals were transferred tothe fresh plates 24 hours post-injection. The F1 progeny were aged andscored at 12, 24, 36, 48 and 60 hours post L4/adult molt. For examining theendocytosis phenotypes, wild-type animals carrying different endocyticmarkers were injected with dsRNA of unc-108 and transferred to the freshplate 24 hours post-injection. The F1 progeny were examined at 24 to 48hours post L4/adult molt for endocytosis defects. We found unc-108 RNAisignificantly diminished the expression of unc-108 in C. elegans embryos:

45% of embryos (n=105) transgenic for Punc-108unc-108::gfp had bright GFPfluorescence before injection, but none of the embryos from the sametransgenic line (n=119) showed any visible GFP fluorescence 24 hours afterinjection.

Four-dimensional microscopyFour-dimensional (4D) microscopy analysis of cell corpse duration wasperformed as described (Wang et al., 2003) using a Zeiss Axioimager M1coupled with an AxioCam monochrome digital camera. Images wereprocessed and viewed using Axiovision Rel. 4.5 software.

Time-lapse fluorescence microscopyC. elegans embryos (300 minutes old) were mounted on slides with an agarpad in egg salt (118 mM NaCl and 48 mM KCl) and a cover slip was placedon top and sealed with beeswax and Vaseline (1:1). Images in a 24 �m z-series (1.2 �m/section) were captured every 3 or 3.5 minutes for 180 minutesusing a Zeiss LSM 5 Pascal inverted confocal microscope. Images wereprocessed and viewed using LSM Image Browser software.

Acridine Orange stainingAcridine Orange (AO; Sigma, USA) staining in embryos was performed asdescribed (Hersh et al., 2002) with a few modifications. Briefly, embryoswere collected from the gravid adults treated with 1.6 M NaOH/12%hypochlorite until dissolved. Embryos were washed several times in M9buffer and then incubated in 50 �g/ml AO in M9 buffer for 1 hour beforeobservation by epifluorescence microscopy. For AO staining in germ cells,aged adults were soaked in 50 �g/ml AO in M9 buffer for 2 hours andrecovered on OP50-seeded plates for 3 hours before observation.

Endocytosis assayIn vivo pulse-chase experiments were performed as described (Zhang et al.,2001). Briefly, Texas Red-conjugated BSA (TR-BSA; Sigma, USA) wasinjected at 1 mg/ml into the body cavity in the pharyngeal region. Injectedworms were transferred to a seeded NGM plate at room temperature and thecoelomocyte uptake was monitored at different time points (5, 10, 15, 20, 30and 60 minutes; 6, 12 and 24 hours). At each time point, the injected wormswere transferred to an ice-cold NGM plate to stop the intracellular traffickingof endocytosed molecules before examination by epifluorescencemicroscopy.

The apical uptake of fluid-phase material in the intestine was analyzed bysoaking L4/young adults in 1 mg/ml TR-BSA in M9 buffer for 8 hours in thedark at room temperature. Animals were recovered on a seeded NGM platefor 2 hours before observation. For examining the apical uptake of lipophilicdye in the intestine, L4/young adults were soaked in 40 �M FM4-64(Invitrogen, USA) for 30 minutes in the dark at room temperature andrecovered on a seeded NGM plate for 30 minutes before observation. Thebasolateral uptake in the intestine was analyzed by injecting 1 mg/ml TR-BSA or 40 �M FM4-64 into the body cavity. The injected worms weretransferred to a seeded NGM plate and recovered at room temperature for30 minutes before observation.

Plasmid constructionTo construct Punc-108unc-108::gfp and Punc-108unc-108::mcherry, we inserteda 4 kb fragment containing the genomic sequence of the unc-108 geneincluding 2 kb promoter region into the pPD95.77 or pPD95.77-mcherryvector (generated from pPD95.77 by replacing the gfp fragment withmcherry) via its SphI-BamHI sites. To construct Pced-1unc-108 and Pegl-1unc-108, the full-length unc-108 cDNA was amplified from a C. elegans cDNAlibrary (Invitrogen, USA) and cloned into the Pced-1 vector via its KpnI siteor into Pegl-1 through its NheI-NcoI sites. To generate Punc-108unc-108, wefirst amplified a fragment containing the 2 kb DNA region upstream of thestart codon of the unc-108 gene and cloned it into a pPD49.26 vector via itsSphI-BamHI sites to create the construct Punc-108. The full-length cDNA ofunc-108 was then cloned to Punc-108 at the NheI-EcoRV sites to generatePunc-108unc-108. To construct the endosomal and lysosomal markers drivenby the ced-1 promoter, the mcherry fragment was amplified from plasmidpPD95.77-mcherry and cloned into the pPD49.26 vector through its NheI-KpnI sites to yield pPD49.26-mcherry. The full-length genomic sequence ofthe rab-5 and rab-7 genes were then amplified using N2 genomic DNA as

RESEARCH ARTICLE Development 135 (6)

DEVELO

PMENT

template and cloned into the pPD49.26-mcherry at the KpnI-EcoRV sites toobtain the N-terminally tagged mCHERRY fusions. Finally, the 5 kbpromoter region of the ced-1 gene was ligated to the 5� end of mCHERRYat the BamHI site. To obtain Pced-1lmp-1mcherry, the ced-1 promoter wascloned to the pPD95.77-mcherry at the BamHI site followed by the ligationof full-length genomic sequence of the lmp-1 gene through the KpnI site. Toexpress unc-108 specifically in the coelomocyte, an 800 bp fragmentupstream of the start codon of the unc-122 gene was amplified (Fares andGreenwald, 2001b) and cloned into the pPD49.26-gfp or pPD49.26-mcherryvector through its SphI-BamHI sites, which was then ligated with the full-length unc-108 genomic sequence at the KpnI-EcoRV sites to producePunc-122gfp::unc-108 and Punc-122mcherry::unc-108. The N-terminallyGFP/mCHERRY-tagged UNC-108 fusions driven by the unc-108 promoter(Punc-108gfp::unc-108 and Punc-108mcherry::unc-108) were then generatedfrom these two constructs by replacing the promoter of unc-122 with that ofunc-108 through the SphI-BamHI sites. The full-length mouse Rab2 cDNAwas obtained by reverse transcription PCR from mouse liver and cloned intothe C. elegans heat-shock vectors pPD49.78 and pPD49.83 via their NheI-NcoI sites or into the Pced-1 vector at its KpnI site.

RESULTSsm237 is a new allele of unc-108 that containsmany persistent cell corpsesThe sm237 mutant was isolated from a psr-1 enhancer screen (X.W.and D. Xue, unpublished), but its phenotype is not dependent on orenhanced by the psr-1 deletion mutant tm469 (data not shown). Thesm237 animal contains many persistent cell corpses at lateembryonic stages that would normally have very few cell corpses(Fig. 1A). The appearance of cell corpses in the sm237 mutant istotally blocked by strong loss-of-function mutations in the ced-3 and

ced-4 genes that are required for almost all apoptosis in C. elegans,indicating that the persistent cell corpses observed in sm237 areindeed apoptotic cells (data not shown). We cloned the gene affectedby sm237 and found that it corresponds to a previously identifiedgene, unc-108 (see Materials and methods; see Fig. S1A in thesupplementary material). unc-108 encodes a small GTPase thatshares high sequence homology with human RAB2 (RAB2A) (87%sequence identity and 93% similarity; see Fig. S1B in thesupplementary material), a member of the Rab small GTPase family(Pereira-Leal and Seabra, 2000). We determined the sequence ofunc-108 in the sm237 mutant and identified a missense mutation thatresults in substitution of Gly18 with Glu (G18E), which affects aconserved nucleotide-binding motif (PM1) present in all membersof the Ras GTPase superfamily (see Fig. S1B in the supplementarymaterial) (Pereira-Leal and Seabra, 2000). Expression of the full-length unc-108 cDNA under the control of its own promoter(Punc-108unc-108) efficiently rescued the persistent cell corpsephenotype of the sm237 mutant (see Fig. S1A in the supplementarymaterial). Expression of mouse Rab2 cDNA driven by heat-shockpromoters (Phspmrab2) or the ced-1 promoter (Pced-1mrab2) alsorescued the corpse phenotype of the sm237 mutant (see Fig. S1A inthe supplementary material), indicating that mouse Rab2 cansubstitute for UNC-108 in removing apoptotic cells in C. elegans.

The unc-108(sm237) mutant is defective in cellcorpse removalTo determine whether accumulation of cell corpses in sm237animals is due to a defect in cell corpse clearance, we performed atime-course analysis of cell corpse appearance during development

1071RESEARCH ARTICLEunc-108 promotes cell corpse engulfment

Fig. 1. unc-108 is important for cellcorpse clearance in C. elegans. (A) Time-course analysis of cell corpse appearanceduring development in the wild type (N2,black), the unc-108(sm237) mutant (white)and in wild-type animals treated with unc-108 RNAi (gray). Cell corpses were scoredat the following embryonic or larval stages:bean/comma (comma), 1.5-fold (1.5F), 2-fold (2F), 2.5-fold (2.5F), 3-fold (3F), 4-fold(4F) and early L1 larvae (L1). The y-axisrepresents the mean number of cell corpsesscored at the head region of embryos orlarvae; at least 15 animals were scored ateach stage. Error bars indicate s.e.m.(B) unc-108(sm237) mutant containspersistent germ cell corpses. The number ofgerm corpses were scored every 12 hoursafter the L4/adult molt from one gonadarm in wild-type (N2, black), unc-108(sm237) mutant (white) and unc-108(RNAi) animals (gray). The y-axisrepresents the average number of germ cellcorpses. At least 15 animals were scored ateach time point. Error bars indicate s.e.m.In A and B, data derived from differentgenetic backgrounds at multipledevelopmental stages were compared by two-way analysis of variance. Post-hoc comparisons were by Fisher’s PLSD (protected least squaresdifferences). *P<0.05, **P<0.0001. All other points had P values>0.05. (C) Four-dimensional microscopy analysis of cell corpse duration in the unc-108(sm237) mutant. The duration of 33 cell corpses from wild-type (N2) embryos (n=3, black), unc-108(sm237) embryos (n=3, white) and unc-108(RNAi) embryos (n=3, gray) were followed. The numbers in parentheses indicate the average durations of cell corpses (±s.e.m.). The y-axisrepresents the number of cell corpses within a specific duration range as shown on the x-axis. The durations of four cell divisions in the MS celllineage from MS cell to MS.aaaa cell were also followed to ensure that the embryos scored had similar development rates. The average duration offour cell divisions is 93±6 minutes in N2 embryos, 96±1 minutes in unc-108(sm237) embryos and 103±3 minutes in RNAi-treated embryos. D

EVELO

PMENT

1072

(Wang et al., 2003). In both somatic and germ cells, significantlyhigher numbers of cell corpses were observed in the sm237 mutantthan in wild-type animals at all developmental stages (Fig. 1A,B).To confirm that the increase in cell corpses in sm237 is caused by adefect in cell corpse removal, we performed 4D microscopy analysisto measure the duration of embryonic cell corpses in sm237 animals(Wang et al., 2003). In wild-type animals, the majority of the cellcorpses persisted from 10 to 50 minutes, whereas in the unc-108(sm237) mutant most cell corpses persisted from 30 to 110minutes (Fig. 1C). On average, the duration of cell corpses in unc-108(sm237) embryos was 93% longer than in wild-type embryos(Fig. 1C), indicating that the removal of apoptotic cells is defectivein the unc-108(sm237) mutant.

sm237 represents a gain-of-function allele ofunc-108sm237 animals are viable but display a dominant Unc(uncoordinated) phenotype, which is consistent with the previouscharacterization of the unc-108 gene (Park and Horvitz, 1986).Different from its dominant Unc phenotype, we found that thepersistent cell corpse phenotype of sm237 is semi-dominant andshows a maternal effect: sm237 homozygous embryos produced bysm237/+ heterozygous mothers showed a weak Ced (cell deathabnormal) phenotype equivalent to that of the mother (sm237/+),which was weaker than that of the sm237/sm237 embryos producedby the homozygous mothers (Table 1). The weak Ced phenotypeobserved in sm237/+ embryos from the heterozygous mother couldbe explained by the gain-of-function nature of sm237 or haploidinsufficiency of sm237/+. To distinguish between these twopossibilities, we examined an unc-108 deletion mutant (ok1246),which contains a 2198 bp deletion that removes the whole genelocus and represents a null allele of unc-108 (Wormbase:www.wormbase.org; see Fig. S2A in the supplementary material).Most homozygous ok1246 embryos from the heterozygous mother(hT2/ok1246) appeared to develop normally during embryogenesis,but failed to hatch or were arrested at early larval stage, indicatingthat UNC-108 is essential for C. elegans development. However, noobvious Ced phenotype was observed either in hT2/ok1246 animalsor their ok1246 progeny, suggesting that sm237 is likely to be a gain-

of-function allele (Table 1). Furthermore, the Ced phenotype insm237/ok1246 embryos was weaker than that of sm237/sm237embryos, but stronger than that of sm237/+ embryos (Table 1),indicating that sm237 indeed represents a gain-of-function allele ofunc-108 and that the wild-type unc-108 activity antagonizes the unc-108(sm237) allele. This result is also consistent with the finding thatoverexpression of wild-type unc-108 was able to rescue thepersistent cell corpse phenotype of sm237 animals and that wild-type gene product contributed maternally was able to partially rescuethe Ced phenotype of the homozygous sm237 progeny (see Fig. S1Ain the supplementary material; Table 1).

To confirm this result, we overexpressed the UNC-108(G18E)mutant product in wild-type animals using C. elegans heat-shockpromoters [PhspUNC-108(G18E)] and found that UNC-108(G18E)resulted in a similar, albeit slightly weaker, corpse phenotype to thatof the sm237 mutant (see Fig. S2C in the supplementary material).Since ok1246 larvae do not survive, we could not examine theirprogeny for the Ced phenotype and therefore cannot rule out thepossibility that the maternal contribution of the wild-type allele issufficient to mediate the normal clearance of cell corpses that weobserved in ok1246 embryos. To determine whether a loss-of-function mutation in the unc-108 gene affects cell corpse removal,we treated wild-type animals with unc-108 RNAi and examined thepersistent cell corpse phenotype in the progeny. Indeed, we foundthat unc-108 RNAi caused similar phenotypes to that of the sm237mutant in both somatic and germ cells, indicating that the wild-typeunc-108 functions to promote cell corpse clearance (Fig. 1A-C).Two other alleles of unc-108 (n501 and n777) isolated previously bydominant Unc phenotype also displayed a weak Ced phenotype (seeFig. S2B in the supplementary material) (Park and Horvitz, 1986).

UNC-108 is expressed and functions in theengulfing cells to promote cell corpse removalTo examine the expression pattern of unc-108, we generated UNC-108 translational GFP fusions under the control of its own promoter[Punc-108unc-108::gfp (UNC-108::GFP) and Punc-108gfp::unc-108(GFP::UNC-108)], which partially rescued the persistent cell corpsephenotype of sm237 animals (see Fig. S1A in the supplementarymaterial). unc-108::gfp was ubiquitously expressed in the embryo,starting from the very early stage of 50 to 100 cells and throughoutthe larval and adult stages. The expression of unc-108::gfp wasobserved in engulfing cells, such as hypodermal cells, intestine cellsand gonadal sheath cells (see Fig. S3A in the supplementarymaterial; data not shown). unc-108::gfp was also seen in many headand tail neurons as well as ventral cord neurons (see Fig. S3A in thesupplementary material). Interestingly, unc-108 is also expressed inthe coelomocytes, the scavenger cells in C. elegans that constantlyuptake macromolecules from the body cavity (see Fig. S3A in thesupplementary material). This expression pattern is consistent withthe function of UNC-108 in endosomal trafficking (see below).Similar expression patterns with more-vesicular localizations wereobserved in animals expressing GFP::UNC-108 fusion protein (seeFig. S3B in the supplementary material).

To determine whether UNC-108 activity is required in theengulfing cells or dying cells for cell corpse removal, we expressedunc-108 under the control of the ced-1 promoter (Pced-1) or egl-1promoter (Pegl-1), which drives gene expression specifically in theengulfing cells or dying cells, respectively (Conradt and Horvitz,1998; Zhou et al., 2001b), and examined whether expression of unc-108 in these cells rescued the persistent cell corpse phenotype of thesm237 mutant. Expression of unc-108 in engulfing cells (Pced-1unc-108), but not in dying cells (Pegl-1unc-108), rescued the cell corpse

RESEARCH ARTICLE Development 135 (6)

Table 1. sm237 represents a gain-of-function allele of unc-108Maternal genotype Zygotic genotype No. of cell corpses at 4-fold

+/+ +/+ 0.2±0.1sm237/+ 1.5±0.2

null/+ 0.5±0.1

sm237/+ sm237/+ 4.4±0.5sm237/sm237 6.5±0.5

sm237/sm237 sm237/+ 5.3±0.4sm237/null 9.1±0.7

sm237/sm237 (cross) 14.6±0.7sm237/sm237 (self) 14.5±1.4

+/null +/null 0.4±0.2null/null 0.4±0.1

Cell corpses were scored in 4-fold stage embryos and are shown as means±s.e.m. Atleast 15 embryos were scored for each genotype. The complete maternal genotypesare (from top to bottom): unc-76, hT2/sm237, sm237 and hT2/ok1246. Thecomplete zygotic genotypes are (from top to bottom): hT2/+;unc-76/+, sm237/+smIs13 (Psur5:gfp)/+;unc-76/+, red non-green progeny of hT2/ok1246;qxEx58(Psur5:rfp) males crossed with unc-76, green progeny of hT2/sm237, non-greenprogeny of hT2/sm237, green progeny of hT2/+ males crossed with sm237, red non-green progeny of hT2/ok1246;qxEx58 males crossed with sm237, green progeny ofsm237/+ smIs13/+ males crossed with sm237, progeny of sm237, green progeny ofhT2/ok1246 and non-green progeny of hT2/ok1246. D

EVELO

PMENT

clearance defect of sm237 animals, indicating that unc-108 needs tofunction in the engulfing cells to promote cell corpse removal (seeFig. S1A in the supplementary material).

Abnormal endosomal compartments in unc-108(sm237) mutant coelomocytesThe involvement of human RAB2 in vesicular trafficking (Tisdale,1999; Tisdale and Balch, 1996; Tisdale et al., 1992) and thecoelomocyte localization of UNC-108 promoted us to investigatewhether unc-108 affects endocytosis in C. elegans. In adulthermaphrodites, there are six coelomocytes acting as scavenger cellsto take up macromolecules from pseudocoelom. We first examinedwhether coelomocyte uptake is affected in the unc-108(sm237)mutant using the Cup assay (coelomocyte uptake) (Fares andGreenwald, 2001a). We introduced arIs37 [pmyo-3::ssGFP] intothe unc-108(sm237) mutant and examined the uptake of secretedsoluble GFP (ssGFP) by the coelomocytes. Compared with efficientuptake of ssGFP by coelomocytes of wild-type animals, the initialuptake of ssGFP by the coelomocytes of unc-108(sm237) animalsdecreased and GFP accumulated in the body cavity (Fig. 2A).Moreover, unlike the GFP pattern in wild-type coelomocytes,internalized GFP was present in the enlarged vacuoles in unc-108(sm237) animals (Fig. 2A). To further investigate the possiblecause of this defect, we examined whether the coelomocytes in unc-

108(sm237) animals contained normal endosomal and lysosomalcompartments using different markers fused with GFP. RME-8::GFP marks endosome membrane, displaying many ring-likestructures representative of endosomes in wild-type coelomocyte(Zhang et al., 2001). We observed three different patterns of RME-8::GFP in the coelomocytes of sm237 animals. Twenty percent ofcoelomocytes in unc-108(sm237) animals contained normalendosomes that showed a similar RME-8::GFP pattern to that inwild type. Forty percent of coelomocytes contained enlargedvacuoles that were labeled by RME-8::GFP, half of which were solarge that they almost occupied the whole coelomocyte. Another40% of coelomocytes showed a punctate pattern of RME-8::GFPinstead of the normal ring structure, and this did not correlate withthe age of hermaphrodites (Fig. 2B; data not shown). Moreover, wesaw very few endosomes of normal morphology in this type ofcoelomocyte (Fig. 2B). These data suggest that sm237 animalscontain both damaged and enlarged endosomes.

To confirm the identity of the large vacuole, we introduced LMP-1::GFP, an early lysosome marker into sm237 animals (Treusch etal., 2004). To our surprise, these large vacuoles were also marked byLMP-1::GFP, suggesting that they might represent aberrant hybridsof endosome and lysosome (Fig. 2C). In addition, only a few normallysosomes with LMP-1::GFP were found in the coelomocytes ofsm237 mutant (Fig. 2C; Fig. 4D). Other endosomal and lysosomal

1073RESEARCH ARTICLEunc-108 promotes cell corpse engulfment

Fig. 2. The coelomocytes of sm237 mutantC. elegans contain aberrant endosomes.(Aa-f) The coelomocyte uptake is reduced in theunc-108(sm237) mutant. The uptake of ssGFP inthe coelomocyte of a wild-type animal (a,b), unc-108(sm237) mutant (c,d) and unc-108(RNAi)animal (e,f) carrying Pmyo-3ssgfp were examined byvisualizing the GFP accumulation in the body cavityand coelomocytes. White dashed line indicates theoutline of the coelomocyte and arrows point to thebody cavity with accumulated GFP. (Ba-Df) sm237animals contain abnormal endosomes. Thecoelomocytes of a wild-type animal (a,b), unc-108(sm237) mutant (Bc-Bf, c,d in C,D), unc-108(RNAi) animal (Bg-Bj, e,f in C,D) carryingdifferent endosome and lysosome markers wereexamined. In wild-type coelomocyte, RME-8::GFP(B) associates with early and late endosomes; LMP-1::GFP (C) mostly stains lysosomes and GFP::RME-1(D) marks recycling endosomes (arrows). Scalebars: 2.5 �m.

DEVELO

PMENT

1074

markers, such as RAB-7, which associates with late endosome andlysosome, and CUP-5, a lysosomal component, were also found tobe associated with the large vacuole (Poteryaev et al., 2007; Treuschet al., 2004) (see Fig. S4A,B in the supplementary material). Tofurther confirm this result, we introduced RME-8::mRFP and LMP-1::GFP or RME-8::GFP and mCHERRY::CUP-5 simultaneouslyinto the sm237 mutant and found that these markers co-localizedto the enlarged vacuoles, rather than localizing separately toendosomes or lysosomes as in the wild-type coelomocytes (see Fig.S4C,D in the supplementary material), indicating that the enlargedvacuoles in the sm237 mutant represent hybrids of endosome andlysosome.

Taken together, our data showed that sm237 animals containeddamaged endosomes and enlarged vacuoles with both endosomaland lysosomal components, suggesting that unc-108 is involved inboth an early step of endosomal trafficking and in lysosomeformation from late endosome. To further investigate whethersm237 affects an early step of endosomal transport, we examined thelocalization of GFP::RME-1, an EH-domain-containing ATPaseassociated with recycling endosomes (Lin et al., 2001). In wild-typecoelomocytes, RME-1 was mostly found in close proximity to theplasma membrane (Fig. 2D). By contrast, this pattern was disruptedin the coelomocytes of sm237 animals as GFP::RME-1 was oftenfound around the endosomes (Fig. 2D, arrow). We also checked thepattern of early endosome-associated RAB-5 (Pfeffer and Aivazian,2004; Poteryaev et al., 2007). Many coelomocytes of sm237 animalsshowed a normal GFP::RAB-5 pattern, except for those thatcontained no other compartments but one large vacuole that waslabeled by GFP::RAB-5 (data not shown). Therefore, we concludethat unc-108 functions in both early and late steps of endosomaltrafficking, affecting the transition from early to late endosome, therecycling endosomes and the late endosome to lysosome transition.Consistent with this result, using early endosome markermCHERRY::RAB-5 and lysosome-associated mCHERRY::CUP-5,we found that GFP::UNC-108 localized to both endosomes andlysosomes (Fig. 3).

Lysosome maturation is affected in the unc-108(sm237) mutantIn order to examine the endosomal trafficking defect of sm237animals with higher temporal resolution, we performed in vivopulse-chase analysis of endocytosis by injecting TR-BSA (TexasRed-conjugated BSA) into the body cavity of adult hermaphroditesand examined the uptake of TR-BSA into the coelomocytes in bothwild-type and sm237 animals carrying different endosomal/lysosomal markers. In wild-type animals, 5 minutes after injection,

TR-BSA started to appear in the endosomes labeled by RME-8::GFP. After 15 minutes, a significant amount of TR-BSA left theRME-8::GFP ring, and after 30 minutes most of the TR-BSA waspresent in the lysosomes lacking RME-8::GFP (Fig. 4A). In thesm237 mutant, however, TR-BSA appeared in the RME-8::GFP-labeled compartment 5 minutes after injection and stayed therethroughout the time-course of the experiment (Fig. 4B; see Materialsand methods; data not shown). We also monitored the uptake of TR-BSA using the early lysosomal marker LMP-1::GFP, and found thatTR-BSA started to accumulate in the compartments lacking LMP-1::GFP 5 minutes after injection. After 15 minutes, TR-BSAappeared in the lysosomes marked by LMP-1::GFP (Fig. 4C). Bycontrast, 5 minutes after injection, TR-BSA accumulated in thevacuole marked by LMP-1::GFP in the sm237 mutant (Fig. 4D).During the remainder of the time points, most TR-BSA stayedwithin the vacuole or enlarged endosomes that were labeled byLMP-1::GFP and failed to move out even at 24 hours post-injection(Fig. 4D; data not shown). Therefore, our pulse-chase experimentsshowed that lysosome biogenesis was severely affected in the sm237mutant, suggesting that UNC-108 is required for the formation oflysosome from late endosome.

Yolk protein trafficking and apical uptake in theintestine are blocked in unc-108(sm237) animalsIn C. elegans, yolk uptake by growing oocytes presents a typicalexample of receptor-mediated endocytosis (Grant and Hirsh, 1999).Using a VIT-2::GFP reporter (Grant and Hirsh, 1999), we examinedwhether sm237 affects yolk uptake by oocytes. We did not observeany defect of initial uptake of yolk protein in unc-108(sm237)oocytes (Fig. 5A). Consistently, the localization of GFP::RME-1was also normal in the oocytes of sm237 animals (data not shown).However, the redistribution of yolk protein to gut primordium in theembryo or to the intestine in larva was blocked in the mutant (Fig.5B,C; data not shown). These results indicate that UNC-108 is notrequired for the initial uptake step of receptor-mediated endocytosisin developing oocytes, but is involved in the resecretion andtrafficking of the yolk protein. A similar yolk redistribution defecthas been observed previously in rab-7(RNAi) animals and in thesand-1 mutant, which might suggest that the yolk needs to reach thelate endosomal compartment for its later resecretion (Grant andHirsh, 1999; Poteryaev et al., 2007). Therefore, the yolkredistribution defect that we observed in sm237 animals could bedue to the disruption of UNC-108 function in the late step ofendosomal trafficking. To test whether UNC-108 is required forendocytosis in the intestine, animals were fed with TR-BSA (fluid-phase material) or with the lipophilic dye FM4-64, and the apical

RESEARCH ARTICLE Development 135 (6)

Fig. 3. UNC-108 localizes to both endosomes andlysosomes. GFP::UNC-108 was specifically expressed in thecoelomocyte driven by unc-122 promoter in cdIs113, whichcarries integrated pcc1:mCHERRY::RAB-5 (A), or in cdIs97that contains integrated pcc1:mCHERRY::CUP-5 (B).GFP::UNC-108 was observed on endosomes where itoverlapped with mCHERRY::RAB-5 (A, arrows) and onlysosomes marked by mCHERRY::CUP-5 (B, arrows). Scalebars: 2.5 �m.

DEVELO

PMENT

(luminal) uptake of the dyes was assayed. Both TR-BSA and FM4-64 were quickly taken up from the lumen by the intestinal cells inwild-type animals (see Fig. S5 in the supplementary material).However, in sm237 animals, most of the TR-BSA or FM4-64

accumulated in the intestinal lumen, indicating that the apical uptakewas mostly blocked (see Fig. S5 in the supplementary material). Wedid not observe any obvious defect in sm237 animals when bothmarkers were delivered basolaterally (data not shown).

1075RESEARCH ARTICLEunc-108 promotes cell corpse engulfment

Fig. 4. Endocytic trafficking in the coelomocytes of the unc-108(sm237) mutant is blocked from late endosome to lysosome. (A-D) TR-BSA was injected into the body cavity and its transport through endocytic compartments is shown over time in wild-type (A,C) and unc-108(sm237)mutant (B,D) animals with endosomal marker RME-8::GFP (A,B) or lysosomal marker LMP-1::GFP (C,D). White arrows point to the compartmentsthat contain TR-BSA. The blue arrow in D indicates the normally sized lysosome that lacks TR-BSA. Scale bars: 2.5 �m. D

EVELO

PMENT

1076

Loss-of-function of unc-108 causes similarendocytosis defects to those of sm237The data shown above indicate that sm237 affects the transitionfrom early to late endosome, recycling endosomes as well aslysosome biogenesis in coelomocytes, and yolk protein traffickingand apical uptake in the intestine. Since sm237 represents a gain-of-function allele of unc-108, we determined whether loss-of-function of unc-108 caused by RNAi affects endocytosis. We foundthat treatment with unc-108 RNAi caused similar endocytosisdefects to those of the sm237 mutant. First, in 83% of animalstreated with unc-108 RNAi, the uptake of ssGFP was affected,among which 75% failed to uptake any ssGFP, a more severephenotype than that of the sm237 mutant (Fig. 2A). Second, theabnormal endosomal compartments were observed in 76% ofcoelomocytes after RNAi treatment, including enlarged vacuolescontaining both endosome and lysosome components as revealedby labeling with endolysosomal markers RME-8, LMP-1 and CUP-5 (60%), and damaged endosomes as indicated by the punctatepattern of RME-8::GFP (16%) (Fig. 2B,C; see Fig. S4C,D in thesupplementary material). These distorted endosomal compartmentswere also found in the coelomocytes of sm237 mutants, but atslightly different frequency (40% each). Third, we found similarmislocalization of RME-1::GFP around endosomes in animalstreated with unc-108 RNAi (Fig. 2D). Fourth, the endosomaltransport was also carefully examined in pulse-chase experimentsafter RNAi treatment. We found that most TR-BSA was trappedwithin the endosomes for up to 12 hours after injection, whereas inthe wild-type coelomocytes it was transported to lysosome within15 to 30 minutes post-injection (see Fig. S6 in the supplementarymaterial; Fig. 4C; data not shown). However, this blockage was notas complete as that in the sm237 mutant in which TR-BSA stayedinside the endosomes even at 24 hours post-injection. Fifth, theresecretion and trafficking of yolk protein was totally blocked in theembryos or larvae after RNAi treatment, whereas the uptake of yolkprotein was not affected (Fig. 5D; data not shown). Finally, the unc-

108 RNAi-treated animals showed similar apical uptake defects inthe intestine to the sm237 mutants (data not shown). Taken together,our data showed that loss of unc-108 function caused variousendocytosis defects that were similar to those of the gain-of-function allele, sm237, demonstrating that the wild-type unc-108activity is required for endocytosis and is likely to act in both earlyand late steps of endosomal trafficking.

unc-108 affects the degradation of cell corpsesOur data indicate that unc-108 plays an important role inendocytosis. We next examined the role of UNC-108 in cell corpseclearance. We first examined whether cell corpses accumulating inthe unc-108(sm237) mutant or in animals treated with unc-108RNAi were internalized, using Acridine Orange (AO), whichpreferentially stains engulfed apoptotic cells (Gumienny et al., 1999;Lettre et al., 2004). Similar to that in wild-type animals, bothpersistent somatic cell corpses and germ cell corpses in the sm237mutant or animals treated with unc-108 RNAi could be labeled byAO (Fig. 6A-C; data not shown). By contrast, the persistent cellcorpses in the ced-1(e1735) or ced-12(n3261) mutant failed to beinternalized and were not stained (Gumienny et al., 1999; Lettre etal., 2004; Zhou et al., 2001a; Zhou et al., 2001b) (Fig. 6D; data notshown). These data suggest that the persistent cell corpses in sm237mutant or unc-108(RNAi) animals were internalized but notdegraded. Thus, unc-108 is likely to affect the degradation ratherthan the internalization of cell corpses.

unc-108 functions downstream of the engulfmentpathway to promote cell corpse degradationThe cell corpse degradation process is compromised in the sm237mutant and in unc-108(RNAi) animals. Several genes have beendescribed previously that act in two partially redundant pathwaysto regulate cell corpse engulfment in C. elegans (Reddien andHorvitz, 2004; Wang et al., 2003; Yu et al., 2006). We analyzeddouble mutants between sm237 and strong loss-of-function

RESEARCH ARTICLE Development 135 (6)

Fig. 5. Yolk protein trafficking isblocked in the unc-108(sm237)mutant. (Aa-d) Yolk proteinuptake is normal in the developingoocytes of unc-108(sm237) mutantC. elegans. The accumulation ofVIT-2::GFP in the developingooctyes was examined in wild-type(bIs1) (a,b) and in the sm237mutant [unc-108(sm237); bIs1](c,d). (Ba-d) The yolk protein isredistributed to the intestineprimordium (arrow) in the wild-type 1.5-fold (a,b) and 4-fold (c,d)stage embryos. (Ca-Dd) Yolkprotein trafficking is blocked in theunc-108(sm237) mutant (C) orunc-108(RNAi) animal (D). 1.5-(a,b) and 4-fold (c,d) stage embryoswere examined for theredistribution of VIT-2::GFP fromthe anterior region to the gutprimordium. Abundant VIT-2::GFPsignal could be observed in theanterior region of the embryos(arrow). Scale bars: 1 �m in A;5 �m in B-D.

DEVELO

PMENT

mutations in several other genes acting in the two cell-corpseengulfment pathways (ced-1, ced-6, ced-7 in one pathway, and ced-2, ced-5, ced-10 and ced-12 in the other) and found that sm237 doesnot significantly affect or enhance the engulfment defect of mutantsin either pathway (data not shown). Similar results were obtainedwith unc-108 RNAi treatment (data not shown), suggesting thatunc-108 does not act in a specific pathway and might functiondownstream of both engulfment pathways to promote cell corpsedegradation.

UNC-108 co-localizes to phagosomes with RAB-5,RAB-7 and LMP-1UNC-108 is expressed and required in engulfing cells to promotecell corpse removal. To further investigate its function in cell corpsedegradation, we examined whether UNC-108 associates withphagosomes that contain internalized cell corpses. CED-1 is aphagocytic receptor and CED-1::GFP localizes to the extendingpseudopods and nascent phagosomes (Yu et al., 2006; Zhou et al.,2001b). In unc-108(sm237) mutant or unc-108(RNAi) animals, theclustering of CED-1::GFP around the cell corpse was not affected(data not shown), which is consistent with our finding that UNC-108is not required for the internalization of cell corpses. To find outwhether UNC-108 associates with phagosomes, we first checked ifit clusters around cell corpses and co-localizes with CED-1::GFP to

the extending pseudopods or nascent phagosomes. In wild-typeembryos carrying Punc-108unc-108::gfp, strong GFP signal was seensurrounding the cell corpses (Fig. 7A), indicating that UNC-108might associate with phagosomes. Similar phagosome localizationwas observed in embryos expressing N-terminally GFP-taggedUNC-108 (GFP::UNC-108) (see Fig. S7A in the supplementarymaterial).

We next examined embryos expressing both Punc-108unc-108::mcherry and Pced-1ced-1::gfp and found that both UNC-108::mCHERRY and CED-1::GFP clustered around cell corpses,but we could barely detect any co-localization of these two proteinsaround dying cells. As a phagocytic receptor, the localization ofCED-1 on phagosomes is transient and it disappears long before thecomplete degradation of cell corpses (Yu et al., 2006). Since UNC-108 is likely to be involved in the degradation of cell corpses, onepossible explanation is that UNC-108 is recruited to phagosomesafter CED-1 completes its task and disappears. To test thishypothesis, we followed the recruitment of CED-1 and UNC-108 tophagosomes in embryos expressing both Punc-108unc-108::mcherryand Pced-1ced-1::gfp by time-lapse recording. Consistent with ourhypothesis, we found that CED-1::GFP and UNC-108::mCHERRYwere recruited to the phagosomes at different times during theengulfment process. We set the time point as 0 min when a clearCED-1::GFP ring was seen. At +5 minutes, CED-1::GFP formed abright ring around the cell corpse, whereas UNC-108::mCHERRYwas not seen (Fig. 7Ba-c). At +8 minutes, CED-1::GFP becameweaker and the UNC-108::mCHERRY signal started to appear (Fig.7Bd-f). At +11 minutes, almost no CED-1::GFP could be detectedwhereas the UNC-108::mCHERRY formed a clear circle around thecell corpse (Fig. 7Bg-i). At +14 minutes, strong UNC-108::mCHERRY signal was seen, while CED-1::GFP completelydisappeared from the phagosome (Fig. 7Bj-l). The UNC-108::mCHERRY signal could still be detected at +26 minutes whenthe ‘button-like’ morphology of the cell corpse was lost (Fig. 7Bm-o). UNC-108::mCHERRY eventually disappeared at +29 minutes(data not shown). These data indicate that UNC-108 is recruited tothe same engulfment site as CED-1 and its association with thephagosome is preceded by that of CED-1 and lasts until thedegradation of cell corpses. Similar phagosome recruitment kineticswere observed with the N-terminally tagged UNC-108(mCHERRY::UNC-108) (see Fig. S7B in the supplementarymaterial).

To investigate the potential function of UNC-108 in phagosomematuration, we examined whether UNC-108 co-localizes withseveral other phagosome-associated proteins that function atdifferent phagosome maturation stages in mammals. Rab5 is anearly endosome marker and has been shown to be associated withthe phagosome and to play an important role in phagosomematuration in mammals and fruit flies (Desjardins et al., 1994;Henry et al., 2004; Stuart et al., 2007; Vieira et al., 2002). Rab7, alate endosome component, is recruited to the phagosome by Rab5and mediates the fusion of phagosome with lysosome (Henry et al.,2004; Vieira et al., 2003). Lysosomal protein LAMP1 (vertebrateortholog of C. elegans LMP-1) was also found to be associated withthe phagosome and functions in mediating phagosome maturation(Garin et al., 2001). In wild-type embryos transgenic forPced-1mcherry::rab-5 and Punc-108unc-108::gfp, we found thatmCHERRY::RAB-5 and UNC-108::GFP co-localized to thephagosome, forming a ring-like structure around the cell corpse (Fig.7Ca-d). Similar phagosome co-localization was observed inembryos expressing Pced-1mcherry::rab-7 and Punc-108unc-108::gfpor Pced-1lmp-1::mcherry and Punc-108unc-108::gfp, as well as in

1077RESEARCH ARTICLEunc-108 promotes cell corpse engulfment

Fig. 6. The persistent cell corpses in the unc-108(sm237) mutantare labeled by Acridine Orange. AO staining of a 1.5-fold stageembryo of wild type (A,B), and of a 4-fold stage embryo of unc-108(sm237) (C,D), unc-108(RNAi) (E,F) or ced-1(e1735) (G,H) thatcontain apoptotic cells (A,B) or persistent cell corpses (C-H). Bright AOstaining was observed in the dying cell of the wild-type embryo and thepersistent cell corpses in the unc-108(sm237) mutant and unc-108(RNAi) embryos, but not in the ced-1(e1735) mutant (arrows). Scalebars: 5 �m.

DEVELO

PMENT

1078

animals expressing N-terminally GFP-tagged UNC-108(GFP::UNC-108) (Fig. 7Ce-h,i-l; see Fig. S7C in the supplementarymaterial). Since Rab5 is recruited to the phagosome at a very earlystage and LAMP1 is likely to be involved in the late step ofgenerating the phagolysosome in mammals (Vieira et al., 2002), theco-localization of UNC-108 with both of these markers on thephagosome suggests that UNC-108 might function in both early andlate stages of phagosome maturation.

DISCUSSIONUNC-108 may promote phagosome maturationrequired for cell corpse degradationPhagosome maturation is a dynamic process that involves a seriesof interactions among endocytic compartments, which eventuallyfuse with lysosomes to generate phagolysosomes that possessdegradative properties (Henry et al., 2004; Vieira et al., 2002; Vieiraet al., 2003). In many ways, this maturation process resembles the

RESEARCH ARTICLE Development 135 (6)

Fig. 7. UNC-108 associates withphagosomes. (Aa,b) UNC-108::GFPclusters around cell corpses (arrow).DIC (a) and fluorescent confocal (b)images of a wild-type C. elegansembryo transgenic for Punc-108unc-108::gfp. (Ba-o) UNC-108::mCHERRY is recruited to thephagosome preceded by CED-1::GFP.DIC (a,d,g,j,m), confocal time-lapseimages of CED-1::GFP (b,e,h,k,n) andUNC-108::mCHERRY (c,f,i,l,o)around the same cell corpse in awild-type embryo. The time pointwas set as 0 minute when the CED-1::GFP ring was clearly seen. Imagesfrom five time points after that areshown. Arrows point to the cellcorpse and to the correspondingfluorescent signals. (Ca-l) UNC-108::GFP co-localizes withmCHERRY::RAB-5, mCHERRY::RAB-7and LMP-1::mCHERRY to thephagosome. DIC and fluorescentconfocal images of a wild-typeembryo transgenic for Punc-108unc-108::gfp and Pced-1mcherry::rab-5(a-d) or Punc-108unc-108::gfp andPced-1mcherry::rab-7 (e-h) orPunc-108unc-108::gfp and Pced-1lmp-1::mcherry (i-l). Arrows indicate theco-localization of UNC-108::GFPwith mCHERRY::RAB-5,mCHERRY::RAB-7 or LMP-1::mCHERRY on the phagosome.Scale bars: 5 �m.

DEVELO

PMENT

progression of endocytic compartments, which undergo a series offissions and fusions to modify membrane composition and acquirenew contents (Vieira et al., 2002). Our data indicate that unc-108 isrequired for both endocytosis and cell corpse degradation in C.elegans, suggesting that there might be an intrinsic connectionbetween these two processes. UNC-108 affects the degradation ofcell corpses, associates with the phagosomes containing internalizedcell corpses, and co-localizes with early endosome protein RAB-5,late endosomal component RAB-7, and lysosomal protein LMP-1.Since UNC-108 localizes to both endosomes and lysosomes andfunctions in both early and late steps of endosomal trafficking, it ispossible that UNC-108 is recruited to the phagosome during itsfusion with early endosome and regulates phagosome maturation.

The gain-of-function allele, sm237, has a missense mutation thatchanges Gly18 to Glu (G18E) within the PM1 motif (GxxxxGKs,mutation underlined) that is required for the binding of phosphateand Mg2+ and is conserved in all Ras small GTPase superfamilymembers (Valencia et al., 1991). Structural and biochemical studiesindicate that mutations in this motif may affect the catalytic activityof GTPase (Pai et al., 1989; Reinstein et al., 1990). Therefore, G18Emutant protein might possess less GTPase activity and stay in theactive GTP-bound form that binds to the effector protein. Thepersistent interaction of UNC-108(G18E) with downstreameffectors might block phagosome maturation at a certainintermediate stage and affect the degradation of cell corpses.Overexpression of wild-type UNC-108 might increase the chanceof interaction between wild-type UNC-108 and its effectors, whichwould promote normal degradation of apoptotic cells. Thiscompetition between wild-type UNC-108 and G18E mutant inbinding to effector proteins might explain the variable rescuingactivities we observed with different unc-108 transgenes, which arelikely to carry different copy numbers of wild-type unc-108. In linewith this competition model, we found that overexpression of theUNC-108(G18E) mutant in wild-type embryos indeed resulted in asimilar persistent cell corpse phenotype to that of the sm237 mutant.Further experiments need to be undertaken to understand thebiochemical features of the UNC-108(G18E) protein and to test theabove competition hypothesis.

UNC-108 regulates endosomal trafficking atdifferent steps in C. elegansHuman RAB2 has been implicated in Golgi-ER retrograde transport(Stenmark and Olkkonen, 2001), but the mechanism by whichRAB2 controls this transport is unknown. In addition, it is not clearwhether RAB2 is involved in other aspects of endocytosis or vesicletrafficking. In the present study, we identified a gain-of-functionallele of unc-108, sm237, that affects the uptake of ssGFP bycoelomocytes, transition from early to late endosomes, recyclingendosomes, lysosome formation, yolk protein trafficking and theapical uptake in the intestine. Importantly, inhibition of unc-108expression by RNAi caused similar endocytosis defects to those ofthe sm237 mutant, indicating that unc-108 is indeed required for thisprocess. For example, the majority of animals treated with unc-108RNAi failed to uptake ssGFP, a more severe phenotype than that inthe sm237 mutant, suggesting that unc-108 is required for theinternalization of fluid-phase material in coelomocytes, which mightbe partially affected by the UNC-108(G18E) protein. In addition,various endocytosis defects were observed in animals treated withunc-108 RNAi which were similar to those in the sm237 mutant,such as distorted endosomal compartments, mislocalized recyclingendosomes and defects in TR-BSA trafficking, yolk redistributionand apical uptake in the intestine. These data demonstrate that the

wild-type unc-108 activity is required for endocytosis and it is likelyto regulate endosomal trafficking at different steps, including theprogression from early to late endosome, cargo recycling andlysosome maturation. Identification of the downstream effector(s)or the regulatory proteins that act together with UNC-108 is neededto understand its exact function at these different steps ofendocytosis.

Rab GTPases function as important regulators inremoving apoptotic cellsRab proteins are small GTPases that constitute the largest branchof the Ras GTPase superfamily. Rabs have been implicated inalmost all types of membrane trafficking and have emerged ascentral regulators of vesicle budding, docking and fusion withspecific target organelles (Mukherjee et al., 1997; Vieira et al.,2002). Several Rabs have been found to associate with phagosomescontaining latex beads in mice and fruit flies including Rab1, Rab2,Rab3, Rab4, Rab5, Rab7, Rab11 and Rab14 (Garin et al., 2001;Stuart et al., 2007). However, the phagocytosis of foreign particlesand of apoptotic cells involve different phagocytic receptors andelicit different immune responses. Therefore, the involvement ofRabs in regulating apoptotic cell clearance was not firmlyestablished. Our identification and characterization of UNC-108 inmediating cell corpse degradation and the finding that mouse Rab2can substitute for its function in removing apoptotic cells indicatethat Rab proteins are potential regulators of apoptotic cell clearancein vivo and that this function is likely to be conserved in mammalsas well. In addition to UNC-108/Rab2, C. elegans RAB-5 andRAB-7 also localize to the phagosome and an increased number ofcell corpses was observed in rab-5(RNAi) or rab-7(RNAi) animals,suggesting that these two Rab GTPases might also be involved inthe clearance of apoptotic cells (Fig. 7C; data not shown).Consistent with our findings, recent studies showed thatoverexpression of Rab5 in NIH3T3 fibroblast cells or bonemarrow-derived macrophages promoted the uptake of apoptoticthymocytes, whereas the dominant-negative constructs inhibited it(Nakaya et al., 2006).

We thank Drs Hanna Fares for providing endocytosis markers and A. Fire forvectors; Dr Ding Xue for his critical reading of the manuscript and DrsChonglin Yang, Hong Zhang and members in our laboratory for helpfuldiscussion and suggestions. The sm237 mutant was isolated in Dr Ding Xue’slaboratory. Some strains used in this work were obtained from theCaenorhabditis Genetic Center (CGC), which is supported by a grant from theNIH. This work was supported by the National High Technology Project 863from the Ministry of Science and Technology.

Supplementary materialSupplementary material for this article is available athttp://dev.biologists.org/cgi/content/full/135/6/1069/DC1

ReferencesBrenner, S. (1974). The genetics of Caenorhabditis elegans. Genetics 77, 71-94.Conradt, B. and Horvitz, H. R. (1998). The C. elegans protein EGL-1 is required

for programmed cell death and interacts with the Bcl-2-like protein CED-9. Cell93, 519-529.

Desjardins, M., Huber, L. A., Parton, R. G. and Griffiths, G. (1994). Biogenesisof phagolysosomes proceeds through a sequential series of interactions with theendocytic apparatus. J. Cell Biol. 124, 677-688.

Fadeel, B. (2003). Programmed cell clearance. Cell. Mol. Life Sci. 60, 2575-2585.Fadok, V. A., Bratton, D. L., Rose, D. M., Pearson, A., Ezekewitz, R. A. and

Henson, P. M. (2000). A receptor for phosphatidylserine-specific clearance ofapoptotic cells. Nature 405, 85-90.

Fares, H. and Greenwald, I. (2001a). Genetic analysis of endocytosis inCaenorhabditis elegans: coelomocyte uptake defective mutants. Genetics 159,133-145.

Fares, H. and Greenwald, I. (2001b). Regulation of endocytosis by CUP-5, theCaenorhabditis elegans mucolipin-1 homolog. Nat. Genet. 28, 64-68.

1079RESEARCH ARTICLEunc-108 promotes cell corpse engulfment

DEVELO

PMENT

1080

Garin, J., Diez, R., Kieffer, S., Dermine, J. F., Duclos, S., Gagnon, E., Sadoul,R., Rondeau, C. and Desjardins, M. (2001). The phagosome proteome: insightinto phagosome functions. J. Cell Biol. 152, 165-180.

Grant, B. and Hirsh, D. (1999). Receptor-mediated endocytosis in theCaenorhabditis elegans oocyte. Mol. Biol. Cell 10, 4311-4326.

Grant, B., Zhang, Y., Paupard, M. C., Lin, S. X., Hall, D. H. and Hirsh, D.(2001). Evidence that RME-1, a conserved C. elegans EH-domain protein,functions in endocytic recycling. Nat. Cell. Biol. 3, 573-579.

Gumienny, T. L., Lambie, E., Hartwieg, E., Horvitz, H. R. and Hengartner, M.O. (1999). Genetic control of programmed cell death in the Caenorhabditiselegans hermaphrodite germline. Development 126, 1011-1022.

Henry, R. M., Hoppe, A. D., Joshi, N. and Swanson, J. A. (2004). The uniformityof phagosome maturation in macrophages. J. Cell Biol. 164, 185-194.

Hersh, B. M., Hartwieg, E. and Horvitz, H. R. (2002). The Caenorhabditiselegans mucolipin-like gene cup-5 is essential for viability and regulateslysosomes in multiple cell types. Proc. Natl. Acad. Sci. USA 99, 4355-4360.

Kinchen, J. M., Cabello, J., Klingele, D., Wong, K., Feichtinger, R., Schnabel,H., Schnabel, R. and Hengartner, M. O. (2005). Two pathways converge atCED-10 to mediate actin rearrangement and corpse removal in C. elegans.Nature 434, 93-99.

Lettre, G., Kritikou, E., Jaeggi, M., Calixto, A., Fraser, A., Kamath, R.,Ahringer, J. and Hengartner, M. (2004). Genome-wide RNAi identifies p53-dependent and independent regulators of germ cell apoptosis in C. elegans. CellDeath Differ. 11, 1198-1203.

Lin, S. X., Grant, B., Hirsh, D. and Maxfield, F. R. (2001). Rme-1 regulates thedistribution and function of the endocytic recycling compartment in mammaliancells. Nat. Cell Biol. 3, 567-572.

Liu, Q. A. and Hengartner, M. O. (1998). Candidate adaptor protein CED-6promotes the engulfment of apoptotic cells in C. elegans. Cell 93, 961-972.

Mukherjee, S., Ghosh, R. N. and Maxfield, F. R. (1997). Endocytosis. Physiol.Rev. 77, 759-803.

Nakaya, M., Tanaka, M., Okabe, Y., Hanayama, R. and Nagata, S. (2006).Opposite effects of rho family GTPases on engulfment of apoptotic cells bymacrophages. J. Biol. Chem. 281, 8836-8842.

Pai, E. F., Kabsch, W., Krengel, U., Holmes, K. C., John, J. and Wittinghofer,A. (1989). Structure of the guanine-nucleotide-binding domain of the Ha-rasoncogene product p21 in the triphosphate conformation. Nature 341, 209-214.

Park, E. C. and Horvitz, H. R. (1986). Mutations with dominant effects on thebehavior and morphology of the nematode Caenorhabditis elegans. Genetics113, 821-852.

Pereira-Leal, J. B. and Seabra, M. C. (2000). The mammalian Rab family of smallGTPases: definition of family and subfamily sequence motifs suggests amechanism for functional specificity in the Ras superfamily. J. Mol. Biol. 301,1077-1087.

Pfeffer, S. and Aivazian, D. (2004). Targeting Rab GTPases to distinct membranecompartments. Nat. Rev. Mol. Cell. Biol. 5, 886-896.

Poteryaev, D., Fares, H., Bowerman, B. and Spang, A. (2007). Caenorhabditiselegans SAND-1 is essential for RAB-7 function in endosomal traffic. EMBO J.26, 301-312.

Reddien, P. W. and Horvitz, H. R. (2004). The engulfment process ofprogrammed cell death in Caenorhabditis elegans. Annu. Rev. Cell Dev. Biol. 20,193-221.

Reinstein, J., Schlichting, I. and Wittinghofer, A. (1990). Structurally andcatalytically important residues in the phosphate binding loop of adenylatekinase of Escherichia coli. Biochemistry 29, 7451-7459.

Riddle, D. L., Blumenthal, T., Meyer, B. J. and Priess, J. R. (1997). C. elegans II.Plainview, New York: Cold Spring Harbor Laboratory Press.

Savill, J. and Fadok, V. (2000). Corpse clearance defines the meaning of celldeath. Nature 407, 784-788.

Savill, J., Dransfield, I., Gregory, C. and Haslett, C. (2002). A blast from thepast: clearance of apoptotic cells regulates immune responses. Nat. Rev.Immunol. 2, 965-975.

Stenmark, H. and Olkkonen, V. M. (2001). The Rab GTPase family. Genome Biol.2, REVIEWS3007.

Stuart, L. M., Boulais, J., Charriere, G. M., Hennessy, E. J., Brunet, S., Jutras,I., Goyette, G., Rondeau, C., Letarte, S., Huang, H. et al. (2007). A systemsbiology analysis of the Drosophila phagosome. Nature 445, 95-101.

Su, H. P., Nakada-Tsukui, K., Tosello-Trampont, A. C., Li, Y., Bu, G., Henson, P.M. and Ravichandran, K. S. (2002). Interaction of CED-6/GULP, an adapterprotein involved in engulfment of apoptotic cells with CED-1 and CD91/lowdensity lipoprotein receptor-related protein (LRP). J. Biol. Chem. 277, 11772-11779.

Sulston, J. E. and Horvitz, H. R. (1977). Post-embryonic cell lineages of thenematode, Caenorhabditis elegans. Dev. Biol. 56, 110-156.

Tisdale, E. J. (1999). A Rab2 mutant with impaired GTPase activity stimulatesvesicle formation from Pre-Golgi intermediates. Mol. Biol. Cell 10, 1837-1849.

Tisdale, E. J. and Balch, W. E. (1996). Rab2 is essential for the maturation of pre-Golgi intermediates. J. Biol. Chem. 271, 29372-29379.

Tisdale, E. J., Bourne, J. R., Khosravi-Far, R., Der, C. J. and Balch, W. E. (1992).GTP-binding mutants of rab1 and rab2 are potent inhibitors of vesiculartransport from the endoplasmic reticulum to the Golgi complex. J. Cell Biol. 119,749-761.

Treusch, S., Knuth, S., Slaugenhaupt, S. A., Goldin, E., Grant, B. D. andFares, H. (2004). Caenorhabditis elegans functional orthologue of humanprotein h-mucolipin-1 is required for lysosome biogenesis. Proc. Natl. Acad. Sci.USA 101, 4483-4488.

Valencia, A., Chardin, P., Wittinghofer, A. and Sander, C. (1991). The rasprotein family: evolutionary tree and role of conserved amino acids. Biochemistry30, 4637-4648.

Vieira, O. V., Botelho, R. J. and Grinstein, S. (2002). Phagosome maturation:aging gracefully. Biochem. J. 366, 689-704.

Vieira, O. V., Bucci, C., Harrison, R. E., Trimble, W. S., Lanzetti, L., Gruenberg,J., Schreiber, A. D., Stahl, P. D. and Grinstein, S. (2003). Modulation of Rab5and Rab7 recruitment to phagosomes by phosphatidylinositol 3-kinase. Mol.Cell. Biol. 23, 2501-2514.

Wang, X., Yang, C., Chai, J., Shi, Y. and Xue, D. (2002). Mechanisms of AIF-mediated apoptotic DNA degradation in Caenorhabditis elegans. Science 298,1587-1592.

Wang, X., Wu, Y. C., Fadok, V. A., Lee, M. C., Gengyo-Ando, K., Cheng, L. C.,Ledwich, D., Hsu, P. K., Chen, J. Y., Chou, B. K. et al. (2003). Cell corpseengulfment mediated by C. elegans phosphatidylserine receptor through CED-5and CED-12. Science 302, 1563-1566.

Wu, Y. C. and Horvitz, H. R. (1998). The C. elegans cell corpse engulfment geneced-7 encodes a protein similar to ABC transporters. Cell 93, 951-960.

Yu, X., Odera, S., Chuang, C. H., Lu, N. and Zhou, Z. (2006). C. elegansDynamin mediates the signaling of phagocytic receptor CED-1 for theengulfment and degradation of apoptotic cells. Dev. Cell 10, 743-757.

Zhang, Y., Grant, B. and Hirsh, D. (2001). RME-8, a conserved J-domain protein,is required for endocytosis in Caenorhabditis elegans. Mol. Biol. Cell 12, 2011-2021.

Zhou, Z., Caron, E., Hartwieg, E., Hall, A. and Horvitz, H. R. (2001a). The C.elegans PH domain protein CED-12 regulates cytoskeletal reorganization via aRho/Rac GTPase signaling pathway. Dev. Cell 1, 477-489.

Zhou, Z., Hartwieg, E. and Horvitz, H. R. (2001b). CED-1 is a transmembranereceptor that mediates cell corpse engulfment in C. elegans. Cell 104, 43-56.

RESEARCH ARTICLE Development 135 (6)

DEVELO

PMENT