Cathepsin L protease (CPL-1) is essential for yolk processing during embryogenesis in Caenorhabditis elegans

In most non-mammalian species, yolk proteins are the major source of nutrients for developing embryos. Yolk proteins or vitellogenins are normally synthesised in extra-ovarian tissue. In vertebrates, such as chicken or Xenopus, yolk synthesis occurs in the liver, in insects it occurs in the fat body, and in the nematode Caenorhabditis elegans, yolk is synthesised in the intestine (Kimble and Sharrock, 1983). Yolk proteins are then secreted into the body fluid and selectively taken up by developing oocytes by receptor-mediated endocytosis. Within developing oocytes and embryos, yolk proteins are stored in yolk platelets where they undergo slow, controlled degradation (Fagotto, 1995). Yolk proteins from most species are related, with sequence similarities being found between insect, vertebrate and nematode yolk proteins. Phylogenetic analysis shows that the nematode yolk proteins are more closely related to those from vertebrates than from insects, possibly reflecting the similarity between lipids associated with vertebrate and nematode yolk proteins (Winter et al., 1996; Chen et al., 1997). Related to their role as lipid carriers, yolk proteins show similarity to the human lipid binding protein apolipoprotein B-100 (Chen et al., 1997).

Yolk endocytosis in oocytes is thought to be similar to uptake of low-density lipoprotein (LDL) in somatic cells (Goldstein et al., 1985). Yolk receptors have been identified from several species, including C. elegans, and belong to the LDL receptor superfamily (Schneider, 1996; Grant and Hirsh, 1999). Yolk endocytosis occurs via clathrin-coated vesicles, which shed their clathrin coat and fuse with early endosomes. The acidic environment of the early endosomes is thought to release the ligand from its receptor (Davis et al., 1987) and the receptor is recycled to the cell surface. Studies in Xenopus have suggested that, as well as an acidic environment, aspartic protease activity is required for yolk-receptor dissociation (Opresko and Karf, 1987). By developing a yolk protein YP170::GFP reporter assay, Grant and Hirsh (Grant and Hirsh, 1999) identified components of the early endocytic pathway of yolk in C. elegans oocytes, and found that these are conserved in C. elegans and vertebrates. In C. elegans, RNA interference (RNAi) of components essential for yolk uptake, such as clathrin and adaptins, led to embryonic lethality. RNAi of the rab7 GTPase, which is thought to be involved in transport from early to late endosomes (Feng et al., 1995), did not affect uptake but led to accumulation of yolk in slightly enlarged vesicles, although embryos developed normally (Grant and Hirsh, 1999). From these studies, the early stages of LDL and yolk endocytosis are now reasonably well understood. However, less is known about the mechanisms involved in formation of yolk platelets or in yolk degradation. It is thought that early endosomes fuse with late endosomes, which mature into and/or fuse with yolk platelets. Yolk platelets are similar to lysosomes, being the terminal endocytic compartment, and platelet biogenesis may involve fusion events similar to those described for lysosome formation (Mullock et al., 1998). However, in contrast to lysosomes, yolk platelets show slow, controlled degradation of yolk proteins during development, possibly regulated by pH and enzyme latency (Fagotto, 1995).

Proteases have been implicated in yolk processing but their identity, location, regulation and precise role(s) have not been studied in detail. In the chicken, the aspartate protease cathepsin D has been identified as the major enzyme involved in yolk processing (Retzek et al., 1992) whereas in Xenopus and fish, cathepsin D and the cysteine proteases cathepsins L and B have been implicated (Yoshizaki and Yonezawa, 1998; Kwon et al., 2001; Carnevali et al., 1999). In insects, cathepsins L and B, rather than cathepsin D, appear to be the major enzymes involved in yolk degradation (Cho et al., 1999; Liu et al., 1996; Yamamoto and Takahashi, 1993). Studies with the respective purified enzymes have demonstrated limited in vitro degradation of yolk proteins similar to that observed in vivo.

We examined the roles of proteases during embryonic development of the free-living nematode C. elegans and have recently shown by RNAi that a C. elegans cathepsin L cysteine protease, Ce-CPL-1, is essential for early embryonic development (Hashmi et al., 2002). We have also shown that Ce-CPL-1 is functionally conserved in parasitic nematodes, identifying it as a potential target in the control of parasite development (Britton and Murray, 2002). In the present study, we characterise a C. elegans cpl-1 genetic mutant and demonstrate that the CPL-1 protease is provided maternally and plays an essential role in yolk processing within platelets during early embryogenesis. Loss of CPL-1 leads to accumulation of abnormally large yolk platelets that seem to form from fusion of smaller yolk vesicles. Recent studies in cathepsin L-deficient mice have also identified large and apparently fused lysosomes in cardiomyocytes, which lead to impaired heart function (Stypmann et al., 2002). Studies on the precise role of C. elegans CPL-1 and the cpl-1 mutant phenotype therefore have relevance, not only to yolk endocytosis and degradation in oviparous organisms, but also to platelet/lysosome function in general.

C. elegans strains were cultured by standard methods (Sulston and Hodgkin, 1988) at 20°C. Strains used in this study were the N2 Bristol wild type and DH1033 bIs1[vit-2::GFP, rol-6(su1006)]/sqt-1(sc103) (Grant and Hirsh, 1999), kindly provided by Barth Grant (Columbia University). LysoSensor Green DND-189 (Molecular Probes, Leiden, Netherlands) was incorporated into standard NGM plates at a final concentration of 2 μM.

cpl-1 allele ok360 was provided by the C. elegans Gene Knockout Consortium (Oklahoma Medical Research Foundation). This was backcrossed four times with the N2 strain and maintained as a heterozygote, as confirmed by single worm PCR using cpl-1 primers (T03E6.7.1L1, 5′-CAGTGATTGTTGCGGTTACG-3′ and T03E6.7.1R1, 5′-ATTCTTCTGGCACTGGTTGC-3′). These primers generate PCR products of 2.6 kb and 1.8 kb from the wild type and ok360 mutant, respectively, indicating a deletion of approximately 800 bp in the mutant gene. A wild-type cpl-1 rescue gene was constructed in PCRscript (Stratagene), which included 1.76 kb of cpl-1 upstream promoter sequence, 2.6 kb cpl-1 gene sequence and 0.5 kb of cpl-1 3′UTR amplified by PCR from N2 genomic DNA [T03E6.7F2 forward primer, 5′-ACAGCATGCTCCCGAAAAAAACTTCAATATTC-3′, corresponding to positions -1761 to -1736 relative to the ATG start codon (SphI site underlined); T03E6.7R2 reverse primer, 5′-CCAGAGCTCCCACCGAGATTTGAAC-3′, complementary to positions 512 to 496 relative to the stop codon (SacI site underlined)]. Rescue plasmid DNA (final concentration 12.5 μg/ml) was co-injected with marker plasmid dpy-7::GFP (final concentration 10 μg/ml) (Roberts et al., 2003). Stable, GFP-expressing lines were maintained. Rescued, homozygous cpl-1 mutant lines were identified by PCR on single worms with primers T03E6.7.1L1 and T03E6.7.1R1.

To identify the deletion in cpl-1(ok360) mutant worms, cpl-1 genomic DNA was amplified from four individual homozygous mutants using T03E6.7F3 forward primer, 5′-AATCTCGAGATTCATTCTTCTGGCACTG-3′ and T03E6.7R3 reverse primer, 5′-GTCTCCGTGCTCTGGGTCGGTTCC-3′. PCR products were cloned separately into TOPO 2.1 vector (Invitrogen) and sequenced on an ABI stretch automated sequencer with vector primers and cpl-1 specific primers T03E6.7F3, T03E6.7R3 and T03E6.7F4 (5′-AAGCTGTTCTCTGTCTCGTTCCACTATC-3′).

RNA interference (RNAi) of the rme-2 yolk receptor and rab7 GTPase was carried out by standard procedure using the bacterial feeding method (Timmons et al., 2001). An 820 bp fragment of rme-2 (C. elegans cosmid sequence T11F8.3) and 900 bp fragment of rab7 (cosmid sequence W03C9.3) were generated by PCR on N2 genomic DNA [rme-2 forward primer, 5′-GAACTCGAGTGCCCATGGCTCTGCGTTATTG-3′ (XhoI site underlined); rme-2 reverse primer, 5′-CTCTCTAGAAGCTGTCGTTCTCAATCGAATG-3′ (XbaI site underlined); rab7 forward primer, 5′-ATCGTCGACCCGATTGCTGTGTGCTGGC3′, (SalI site underlined); rab7 reverse primer, 5′-GTGGGATCCCAATTGCATCCCGAATTCTGC3′ (BamHI site underlined)]. The PCR products were cloned into the T7 double vector L4440 (kindly provided by Julie Ahringer, University of Cambridge) and transformed into E. coli strain HT115.

Embryos were cut from wild-type N2 or cpl-1 mutant gravid hermaphrodites and mounted on a 0.2% agarose pad on a glass slide in M9 or egg salts. The agarose pads were sealed with a ring of Vaseline and a coverslip placed on top. Embryos were viewed throughout development for approximately 12-15 hours using an Axioskop 2 microscope (Zeiss, Germany). Images were captured with an Orca camera (Hamamatsu, Bridgewater, NJ) using Openlab 2.0.2 imaging software (Improvision, Coventry, UK).

Immunofluorescence on C. elegans embryos and gravid hermaphrodites was carried out by standard procedure (Miller and Shakes, 1995) using the freeze-cracking method. Rat antibodies to C. elegans yolk proteins YP170 and YP88 (Sharrock, 1983; Sharrock, 1984) were kindly provided by Peg MacMorris (University of Colorado). Rabbit antibodies to C. elegans RME-2 yolk receptor (Grant and Hirsh, 1999) were kindly provided by Barth Grant (Columbia University). Rabbit anti-peptide antibodies to C. elegans CPL-1 were generated by CovalAb (Cambridge, UK). Immunising peptides were selected from the CPL-1 pro-region (IEKWDDYKEDFDKEY, corresponding to amino acids 29-43; and NGYRRLFGDSRIKNS, corresponding to amino acids 95-109 of CPL-1) (Hashmi et al., 2002) or from the mature region (DTEESYPYKGRDMKC, corresponding to amino acids 203-217; and NHCGVATKASYPLV, corresponding to amino acids 324-337 of CPL-1). Secondary antibodies conjugated to fluorescein isothiocyanate (FITC) or Texas Red were obtained from Molecular Probes and used at 1:500 dilution. Incubation in 4,6-diamidino-2-phenylindole (DAPI; final concentration 0.5 μg/ml) for 1 minute prior to viewing was carried out to identify nuclei.

C. elegans gravid hermaphrodites were permeabilised by cutting with a razor blade in fixative (0.5% glutaraldehyde/2% paraformaldehyde, pH 7.4) and incubated in fixative for 1 hour on ice. Following several washes in PBS, worms were embedded in 1.5% agarose, dehydrated through a graded series of alcohol and infiltrated with LR White resin at -20°C. Small clusters were UV-cured in gelatin capsules at room temperature for four days. Thin sections (80 μm) were collected on formvar-coated nickel mesh grids and incubated with anti-YP170, anti-RME-2 or anti-CPL-1 antibody. Gold-conjugated secondary antibody (10 nm or 15 nm gold) (Aurion, Wageningen, Netherlands) was used to detect antibody binding. Grids were contrast stained by standard procedure in uranyl acetate and lead citrate for EM viewing. Images were captured using Easy Vision software.

Adult worm extracts were prepared by washing hermaphrodites in M9 and heating to 100°C for 15 minutes in lysis buffer [50 mM Tris pH 7.5, 2 mM phenylmethylsulfonyl fluoride (PMSF; Sigma, Dorset, UK), 1 mM EDTA, 10 mM E64 (Sigma)] containing 5% SDS and 5% β-mercaptoethanol (Sigma). Before loading onto gels, samples were boiled in an equal volume of loading buffer (lysis buffer containing 10% glycerol and 0.1% Bromophenol blue) and boiled for 10 minutes. Soluble protein extracts were separated on 10% SDS-polyacrylamide gels using the BioRad mini-system and blotted onto PVDF membrane. Western blotting was carried out by standard procedures, using anti-CPL-1 antibody (1:3000 dilution). Alkaline phosphataseconjugated secondary antibodies were used routinely at 1:3000 (Sigma).

In a previous study, we showed that RNAi of the C. elegans cathepsin L protease gene (cpl-1) resulted in lethality in 95-100% of F1 progeny embryos from treated hermaphrodites (Hashmi et al., 2002). To allow a more detailed analysis of cpl-1, a stable genetic mutant was obtained from the C. elegans Gene Knockout Consortium (Oklahoma Medical Research Foundation). DNA sequencing of the mutant cpl-1 gene (allele ok360) identified a deletion of 857 bp between introns 2 and 3. This deletion removes all of exon 3 (amino acid residues 120-268) (Hashmi et al., 2002), which encodes the active site cysteine residue conserved in all cathepsin-like cysteine proteases. We therefore predict that cpl-1(ok360) is a null allele. C. elegans CPL-1 (Ce-CPL-1) shows between 72-90% identity with the mature cathepsin L from animal and plant parasitic nematodes (Britton and Murray, 2002) and 62-67% identity (76-82% similarity) with cathepsin L from insect and amphibian species, including Drosophila.

All F1 progeny of heterozygous cpl-1(ok360) hermaphrodites developed normally to adults with no observable defects in growth rate or moulting, showing that cpl-1 is not required zygotically. Mutant hermaphrodites also had a normal brood size. However, approximately 25% of the F1 adult hermaphrodites produced 100% dead embryos, demonstrating that cpl-1 is required maternally for embryonic development. This embryonic lethal phenotype could not be rescued by mating with N2 males and is therefore strictly maternal.

The cpl-1(ok360) phenotype could be rescued efficiently by extrachromosomal transformation with the wild-type Ce-cpl-1 gene. Stable, rescued mutants were identified by expression of the dpy-7::GFP co-injection marker gene. Western blot analysis with CPL-1 anti-peptide antibody confirmed expression of wild-type CPL-1 protein (with a molecular weight of ∼36 kDa) in rescued and wild-type hermaphrodite worms (Fig. 1, lanes 1 and 2). As expected, this 36 kDa protein was not detected in cpl-1(ok360) homozygous hermaphrodites, although mutant CPL-1 protein (∼22 kDa) was expressed in cpl-1 homozygous mutant rescued and non-rescued worms (Fig. 1, lanes 2 and 3). The efficient rescue observed with wild-type cpl-1 indicates that the mutant CPL-1 protein has no dominant negative activity.

To examine the possible role(s) of cpl-1 during C. elegans embryogenesis, time-lapse Nomarski microscopy was carried out on wild-type and cpl-1(ok360) embryos. No difference was observed in the orientation or timing of early cell divisions in the cpl-1 mutant compared to wild-type embryos and P granule segregation occurred normally (not shown). However large cytoplasmic vesicles appeared in cpl-1 mutant embryos, which were first detected at the two-cell stage in most mutants (Fig. 2E). There was also a noticeably higher rate of movement of cytoplasmic vesicles in mutant compared to wild-type embryos. Around the 6-8 cell stage of embryogenesis, the enlarged vesicles in cpl-1(ok360) embryos increased in size and number (Fig. 2F). At this stage, the rate of cell division in mutant embryos began to decrease relative to that of wild-type embryos, although the sequence and orientation of cell divisions was normal. Approximately 2 hours after the first cell division, ok360 mutant embryos had developed to the 28-cell stage (Fig. 2G), whereas wild-type embryos had approximately 100 cells (Fig. 2C). At this time, large vesicles filled most of the cytoplasm of mutant embryonic cells. Mutant embryos arrested with approximately 100-150 cells and failed to undergo morphogenesis (Fig. 2H,I). Differentiation of muscle and nerve tissue did occur as indicated by the twitching of mutant embryos. Expression of dpy-7::GFP (Britton and Murray, 2002) and elt-2::GFP (not shown) in mutant embryos also demonstrated differentiation of hypodermal and gut cells, respectively, but no morphogenesis.

To examine the nature of the enlarged cytoplasmic vesicles in mutant embryos, we used the fluorescent probe LysoSensor Green, which accumulates in acidic organelles because of protonation. Adult hermaphrodites were fed overnight on NGM plates containing LysoSensor Green, during which time the fluorescent probe was incorporated into oocytes and embryos. In wild-type and cpl-1(ok360) oocytes, the probe could be clearly seen in small punctate vesicles. In wild-type embryos the probe remained localised to these small punctate vesicles throughout embryogenesis (Fig. 3A-D) and could still be detected at a low level in the gut of L1 larvae. In one-cell stage cpl-1 mutant embryos, the probe localised predominantly to small punctate vesicles, although a small number of larger fluorescent vesicles could be observed (Fig. 3E). As embryogenesis progressed, the number and size of these fluorescent vesicles increased (Fig. 3F,G), similar to our findings with Nomarski microscopy. In later stage mutant embryos the fluorescent cytoplasmic vesicles were larger but there were fewer of them, suggesting that they formed from vesicle fusion (Fig. 3H). Localisation of LysoSensor Green to the large cytoplasmic vesicles indicates that they are acidic in nature suggesting that they are late endosomes or platelets.

Cathepsin L has been proposed to be involved in the proteolytic processing of yolk proteins in other oviparous organisms. We therefore examined whether CPL-1 plays a role in yolk processing in C. elegans. This was firstly carried out using C. elegans strain DH1033 which is transformed with a chromosomally integrated C. elegans yolk protein gene YP170 fused to a green fluorescent reporter gene (YP170::GFP). Previous characterisation of DH1033 showed that, like endogenous yolk protein, the YP170::GFP protein is transported from the hermaphrodite intestine via the pseudocoelom to the gonad for endocytosis by oocytes (Grant and Hirsh, 1999). If CPL-1 is required for processing prior to endocytosis, it might be expected that there would be no uptake of YP170::GFP into mutant oocytes and GFP would accumulate in the pseudocoelom. In contrast, if CPL-1 is involved in yolk degradation during embryogenesis we would predict that YP170::GFP accumulates in mutant embryos. Homozygous cpl-1 mutants carrying the YP170::GFP transgene were generated by standard genetic crossing and GFP levels and distribution were compared to that in wild-type DH1033. As expected, in wild-type DH1033 hermaphrodites, GFP was observed in the adult intestine, late-stage oocytes and embryos (Fig. 4A,E,F). In homozygous cpl-1 mutants, YP170::GFP was also observed in the adult intestine, in late oocytes and in embryos until around the 8-12 cell stage (Fig. 4C). The level and localisation of GFP in mutant oocytes and early embryos was similar to that in wild-type DH1033 and there was no apparent accumulation of GFP in the pseudocoelom. CPL-1 is, therefore, not essential for initial yolk uptake by oocytes. However, from the 12-cell stage onwards, the level of fluorescence decreased significantly in cpl-1 mutant embryos (Fig. 4C,H), and no fluorescence could be detected at all in late stage embryos (approximately 100 cells) (Fig. 4I). This suggested that CPL-1 may be involved in yolk processing following endocytosis and that aberrant processing and/or conformational changes lead to loss of fluorescence from the YP170::GFP fusion protein.

We confirmed that loss of fluorescence resulted from alteration of YP170::GFP rather than a loss of yolk protein, by carrying out immunofluorescence with an antibody to C. elegans YP170. Our findings were very similar to those with the LysoSensor probe. Small punctate yolk vesicles were observed in wild-type embryos throughout embryogenesis (Fig. 4G). Similar punctate localisation was observed in oocytes and early embryos of the cpl-1 mutant. However, in later stage mutant embryos (from the 6-8 cell stage onwards) anti-yolk antibody localised to large cytoplasmic vesicles (Fig. 4J). As observed with LysoSensor Green, in later stage cpl-1(ok360) embryos most of the cytoplasm was filled with a few large yolk vesicles. Identical findings were observed with immunofluorescence using an antibody to a second yolk protein, YP88 (not shown). Our findings suggest that C. elegans CPL-1 plays an essential role in the processing of yolk proteins during early embryogenesis and loss of this activity results in the accumulation of yolk in large cytoplasmic vesicles.

Accumulation of enlarged yolk vesicles was confirmed by immuno electron microscopy (IEM) on gravid hermaphrodites using anti-YP170 antibody. In oocytes and early stage wild-type and cpl-1(ok360) mutant embryos, YP170 antibody localised to small electron-dense vesicles which have been identified as yolk platelets in previous studies (Hall et al., 1999). In wild-type embryos, these small vesicles appeared to fuse occasionally (Fig. 5A,B) but did not form large aggregates as they did in cpl-1 mutant embryos (Fig. 5C,D). In later stage mutant embryos (approximately 8-cell stage onwards), these large yolk platelets filled the cytoplasm of most embryonic cells (Fig. 5D). Clear regions devoid of yolk antibody were observed in the interior of some of these enlarged granules. These may be autophagic vacuoles, which are usually only observed when protein digestion is inhibited or delayed (Furuno et al., 1982).

Double IEM labelling with antibodies to yolk protein and CPL-1 showed colocalisation to yolk platelets in wild-type embryos (Fig. 5E) and in the adult pseudocoelom (Fig. 5F). This colocalisation shows that both the protease and its putative substrate, yolk protein, are present in the same vesicles before and after uptake into oocytes. In addition, as this localisation was observed with an antibody to a region of the pro-enzyme that is removed following activation, it demonstrates that at least some of the protease is transported and taken up in an inactive form, possibly being activated autocatalytically during embryogenesis. This is consistent with Bombyx mori cathepsin L, which is converted from an inactive to an active form as embryos develop (Takahashi et al., 1993). Antibody to the mature region of CPL-1 did not react significantly in IEM, possibly because of the weaker reactivity of this antiserum. In immunofluorescence studies, the anti-mature CPL-1 antibody showed a similar, albeit weak, localisation pattern to the anti-pro CPL-1 antibody and, in addition, localised to coelomocytes of adult worms (not shown). These are scavenger cells involved in the non-specific endocytosis of molecules from the pseudocoelom.

Yolk proteins are internalised by cpl-1 mutant oocytes and enter the endocytic pathway. Our results show that yolk platelets form initially, but owing to lack of CPL-1 activity, these fuse to form large vesicles. Specific fusion events in the endocytic pathway are thought to be mediated by rab proteins that are small GTPases of the ras superfamily. The rab7 protein is associated with late endosomes and is thought to be important in transport from early to late endosomes (Feng et al., 1995). We speculated from the temporal appearance of the large cytoplasmic vesicles in mutant embryos that cpl-1 functions in a late step in the endocytic pathway. To confirm that cpl-1 acts in late endosomes or platelets, after rab7 mediated transport, we compared the localisation of YP170::GFP fusion protein in DH1033 rab7(RNAi) mutants and DH1033 rab7(RNAi)/cpl-1 double mutant embryos. In rab7(RNAi) mutants, uptake of YP170::GFP is not impaired and GFP can be detected in late oocytes and embryos. However, the GFP fusion protein is observed in cytoplasmic vesicles that are slightly larger than normal (Fig. 6A). These are predicted to be early endosomes (Grant and Hirsh, 1999). The size of these GFP vesicles does not increase as the embryos develop, but remains constant and GFP vesicles can still be observed in early L1 larvae (not shown). The presence of these slightly enlarged vesicles does not severely affect embryonic or post-embryonic development, although larval and adult progeny of rab7(RNAi) worms have been reported to be mildly `dumpy' (Grant and Hirsh, 1999). These same slightly enlarged YP170::GFP vesicles were also observed in oocytes and early embryos of rab7(RNAi)/cpl-1 double mutants (Fig. 6C), indicating that rab7 functions in early to late endosome transport, before cathepsin L is required. CPL-1, therefore, acts in the late endosome/platelet compartment. In contrast to rab7(RNAi) mutant embryos, the number of these enlarged GFP-positive vesicles decreased as embryogenesis progressed in the double mutants. Again, this loss of YP170::GFP fluorescence did not reflect a loss of yolk protein, but rather the accumulation of yolk in even larger yolk vesicles, as indicated by staining with anti-YP170 antibody (Fig. 6D). No hyper-enlarged YP170 vesicles were present in the single rab7(RNAi) mutant embryos (Fig. 6B).

Studies in Xenopus suggested that initial processing is necessary to remove yolk from its receptor and for subsequent transport to platelets The aspartate protease cathepsin D was implicated in this process (Opresko and Karpf, 1987). We examined whether loss of CPL-1 had any effect on release of C. elegans yolk protein YP170 from the RME-2 receptor. In immunofluorescence studies, an antibody to RME-2 localises in a punctate pattern at the surface of late stage oocytes consistent with its role in yolk uptake (Grant and Hirsh, 1999). In one- and two-cell stage embryos, RME-2 can be detected in intracellular vesicles; it rapidly disappears as embryos develop, presumably as it is no longer required and is degraded following its release from the yolk ligand (Fig. 7B,C). cpl-1(ok360) oocytes and early embryos showed a similar punctate pattern of anti-RME-2 reactivity to that in wild-type embryos. However, instead of rapidly disappearing, RME-2 could still be detected in later stage mutant embryos, in which it localised to the periphery of the large yolk vesicles (Fig. 7E,F). Localisation of RME-2 to the peripheral region and yolk to the lumen, indicated that although present in the same compartment, yolk and RME-2 are separated from one another. RME-2 localisation could be observed in terminal stage mutant embryos, although the level of receptor was slightly decreased (not shown). This suggested that recycling and/or degradation of the yolk receptor is significantly delayed, but not completely blocked, in cpl-1 mutant embryos. Anti-RME-2 reactivity was also observed in rab7(RNAi) mutant embryos consistent with the suggestion that rab7 is thought to be necessary for receptor recycling (Grant and Hirsh, 1999). However, RME-2 reactivity did not colocalise with the enlarged yolk vesicles in rab7(RNAi) embryos but appears to be in separate cytoplasmic vesicles, possibly recycling vesicles (not shown).

Accumulation of RME-2 in cpl-1 mutant embryos was also detected by western blotting (not shown) and by IEM. In wild-type embryos, anti-RME-2 antibody localised to small cytoplasmic vesicles, which are possibly sorting vesicles or early endosomes (Fig. 7G). There was no localisation of RME-2 antibody in the electron-dense yolk platelets in wild-type embryos. In contrast, in cpl-1(ok360) embryos, RME-2 antibody localised to the periphery of some of the large yolk granules, in regions less electron dense than the rest of the yolk vesicles (Fig. 7H).

We have speculated that lack of yolk processing in cpl-1 mutant embryos leads to accumulation of large vesicles. It is not clear however, whether the vesicles themselves cause embryonic death or whether this is a result of lack of nutrients from yolk breakdown. To examine this we carried out RNAi of the yolk receptor rme-2. This would reduce the amount of yolk taken up by oocytes (Grant and Hirsh, 1999). Post-embryonic RNAi of rme-2 in N2 worms resulted in a decrease in brood size and production of some small and some normal embryos, as described for rme-2 genetic mutants (Grant and Hirsh, 1999). RNAi of rme-2 in cpl-1(ok360) worms also resulted in a decrease in brood size, production of small embryos and, surprisingly, a small number (4-5%) of embryos that developed normally and hatched. No hatched larvae were observed with cpl-1 mutants fed on control L4440 vector plates. The majority of rme-2(RNAi)/cpl-1 mutant embryos did not complete development, but showed significantly fewer enlarged cytoplasmic vesicles compared to vector-only controls and also underwent morphogenesis (Fig. 8A), which did not occur with the single cpl-1 mutant embryos (Fig. 8B). This suggests that it is the accumulation of large yolk vesicles in the cytoplasm that interferes with cell division and development, rather than lack of nutrients from yolk degradation.

Proteases have been implicated in the processing and degradation of intra- and extracellular proteins but their precise roles in vivo are not well understood. We examined protease function in the nematode C. elegans and here we characterise a genetic mutant (allele ok360) of the C. elegans cathepsin L cysteine protease gene cpl-1 and show that CPL-1 is essential for embryonic development. Loss of CPL-1 activity leads to formation of enlarged cytoplasmic yolk vesicles and subsequent embryonic lethality. This phenotype is similar to the enlarged lysosomes observed in cathepsin L-deficient mice (Stypmann et al., 2002), indicating the importance of cathepsin L activity in the function and maintenance of the final endocytic compartment in diverse organisms. No post-embryonic effects were observed in C. elegans cpl-1(ok360) mutants, indicating that CPL-1 activity is not essential or is redundant in later developmental stages.

It has been speculated previously that cysteine and aspartate proteases are involved in yolk processing. Our study is the first to demonstrate in vivo the essential role of cathepsin L in yolk degradation. We show that cathepsin L is not required for initial uptake of yolk via the yolk receptor or for its transport through the early endocytic compartments. This suggests that during early endocytosis yolk is transported largely intact. Yolk undergoes normal trafficking in cpl-1(ok360) mutant oocytes and can be clearly seen in small punctate vesicles by immunofluorescence and YP::GFP fluorescence, similar to the fluorescence pattern in wild-type oocytes. IEM studies identified these vesicles as electron-dense yolk platelets. Our data suggest that, initially at least, yolk platelets form as normal in the absence of CPL-1. During embryogenesis, however, lack of CPL-1 activity leads to formation of abnormally large yolk platelets. The significant increase in size and decrease in number of these vesicles during development suggests that they form through the fusion of small yolk platelets.

Yolk platelets are thought to be a heterogenous population that normally undergo fusion and fission with one another and with late multivesicular endosomes. What causes these vesicles to undergo increased or abnormal fusion in cpl-1 mutant embryos? In Xenopus, recently formed light yolk platelets and late endosomes do not appear to fuse with pre-existing platelets, suggesting that fusion is dependent on the internal composition and/or density of the platelets. In addition, formation of the final dense yolk storage granules is a slow process involving condensation and crystallisation of yolk protein (Wall and Patel, 1987). Platelet acidification and enzyme processing are thought to be essential for yolk crystallisation (Fagotto and Maxfield, 1994). Our studies in C. elegans suggest that cathepsin L is required for the initial processing of yolk proteins, which is necessary for further condensation and crystallisation. We postulate that in the absence of CPL-1, yolk crystallisation and condensation do not occur. This results in accumulation of platelets of the same density, which, our data suggest, undergo increased homotypic fusion. We show that the accumulation of large yolk vesicles within the cell cytoplasm leads eventually to embryonic death.

In cathepsin L-deficient mice, large and apparently fused lysosomes containing electron-dense material accumulate in cardiomyocytes. It is thought that these enlarged lysosomes may indicate defective degradation of extracellular matrix proteins and that this leads to increasing fibrosis, which affects heart structure and function. Cathepsin L has also been identified as a potential cardiomyopathy gene in humans (Stypmann et al., 2002). The mechanism of the lysosome fusion in cathepsin L-deficient mice is not known but is likely to be similar to the abnormal platelet fusion we observe in the C. elegans cpl-1 mutant. Abnormal aggregation of lysosomes also occurs in Chediak-Higashi Syndrome (CHS) in humans and in the mouse model of the disease, beige. These enlarged lysosomes are thought to form because of either increased fusion or decreased fission of lysosomes (Stinchcombe et al., 2000). Studies in Dictyostelium have also shown that loss of LvsB, the protein most similar to that implicated in CHS and beige, results in enlarged lysosomes (Harris et al., 2002). This suggests that the mechanisms regulating lysosome fusion/fission are well conserved and that malformation of the terminal endocytic compartment leads to developmental defects and/or disease in a wide range of organisms.

It is perhaps surprising that cathepsin L, which is characterised as being involved in non-specific bulk proteolysis in lysosomes, should be required for an initial, limited processing event. However, there is evidence for cathepsin L in the limited proteolysis necessary for MHC class II-mediated antigen presentation, pro-hormone activation and tumour invasion (Nakagawa et al., 1998; Chapman et al., 1997). In addition, in vitro studies have previously shown that cathepsin L can partially degrade yolk in a manner similar to the partial processing that occurs in vivo. It is also possible that cathepsin L acts indirectly, activating other proteases, such as aspartate proteases or other cathepsins, involved in yolk processing. However, RNAi screens have not identified an essential role for aspartate proteases in C. elegans embryogenesis (Syntichaki et al., 2002; Kamath et al., 2003). This could be the result of functional redundancy between related genes; however, we have found that culturing C. elegans in the presence of pepstatin (10-100 μM), an aspartate protease inhibitor, also failed to indicate involvement of aspartate proteases in embryogenesis. In contrast, incubation with a cysteine protease inhibitor, Mu-Phe-homoPhe vinyl sulfonylbenzene (10-100 μM) (Palmer et al., 1995), led to embryonic lethality, with enlarged cytoplasmic vesicles being observed in the dying embryos (C.B. and L.M., unpublished), consistent with our findings with the cpl-1 mutant. RNAi screens have identified one cathepsin B gene required for C. elegans embryogenesis (cosmid F57F5.1) (Kamath et al., 2003). Analysis of the F57F5.1 RNAi embryonic lethal phenotype shows that silencing of this gene does not result in enlarged yolk platelets (C.B. and L.M., unpublished), suggesting that it does not act with cathepsin L in initial yolk processing. These observations suggest that cathepsin L may act directly on yolk protein, although we cannot rule out an indirect role at present.

Using IEM we showed that cathepsin L and yolk proteins colocalise within the lumen of yolk vesicles in embryos and during yolk transport in the pseudocoelom. We have previously shown, using promoter-reporter constructs, that Ce-cpl-1 is expressed at high levels in intestinal cells (Hashmi et al., 2002). Our results suggest that cathepsin L and yolk proteins are transported together from the intestine to the gonad and taken up into the same compartments within developing oocytes and embryos. We observed that at least some of the cathepsin L present in embryos is in the inactive pro-form and is possibly activated auto-catalytically as embryogenesis progresses. In future studies it will be important to examine the mechanism of CPL-1 activation. Chicken yolk receptor has been shown to recognise α-2 macroglobulin (Jacobsen et al., 1995) and it will be interesting to examine whether C. elegans yolk protein and CPL-1 are complexed with a similar proteinase binding protein to regulate enzymatic activity. Interestingly, we found that RNAi of the rab7 gene, which is involved in transport from early to late endosomes, did not prevent yolk processing or affect embryonic development. The most likely explanation for this is that cathepsin L and perhaps other yolk processing enzymes are present together with yolk proteins in the early endosomes and are active, perhaps because of autonomous acidification of the early endocytic compartments.

Our immunofluorescence and IEM studies identified abnormal accumulation of the C. elegans yolk receptor RME-2 in the enlarged yolk platelets present in cpl-1 mutant embryos. This suggests that cathepsin L is also involved either directly or indirectly in the degradation of the receptor during embryogenesis. However, the receptor and yolk proteins do not colocalise within these enlarged platelets, indicating that they are already dissociated. Cathepsin L is, therefore, not essential for release of yolk from the yolk receptor. A role for cathepsin D in yolk-receptor dissociation has previously been suggested from studies with Xenopus oocytes (Opresko and Karf, 1987), although not, so far, with C. elegans (Syntichaki et al., 2002; Kamath et al., 2003) (C.B. and L.M., unpublished).

C. elegans CPL-1 shows significant identity and a similar expression pattern to cathepsin L proteases in other oviparous species. Therefore, it seems likely that the essential role of cathepsin L in yolk degradation is conserved across diverse organisms. Consistent with this, Drosophila cathepsin L (CP1) mutants show a very similar phenotype to C. elegans cpl-1 mutants and produce a normal-sized brood of eggs, which then fail to hatch (Gray et al., 1998). This indicates that CP1 is essential for Drosophila embryogenesis. Drosophila yolk proteins are not closely related to C. elegans yolk proteins, but more closely resemble lipases. However, sequence analysis has identified segments of homology between human apoB-100, lipoprotein lipase and yolk proteins, suggesting that they may have arisen from a common precursor (Baker, 1988). In addition, the Drosophila yolk receptor shows significant similarity to yolk receptors of other species (Schonbaum et al., 1995). Therefore, cathepsin L is probably involved in processing these diverse yolk proteins at conserved regions, such as the receptor- or lipid-binding regions.

This study shows that loss of C. elegans CPL-1 results in the fusion and accumulation of large yolk platelets. The mechanism of this aggregation is likely to be similar to the lysosomal aggregation and subsequent cardiomyopathy observed in cathepsin L knockout mice. The ease of RNAi and genetic screening approaches for C. elegans will be useful in identifying other genes relevant to this process. In particular, the loss of YP170::GFP reporter activity observed in cpl-1 mutant embryos could be used as a screening approach to identify mutations that cause a similar phenotype to cpl-1(ok360) or those that reverse this effect. Detailed characterisation of the genes involved in vesicle fusion and platelet/lysosome biogenesis in C. elegans has the potential to provide further insight into lysosomal disorders such as cardiomyopathy, CHS and other lysosomal diseases.

We are grateful to Barth Grant (Department of Molecular Biology and Biochemistry, The State University of New Jersey), Peg MacMorris (Department of Biochemistry and Molecular Genetics, University of Colorado Health Sciences Center), the C. elegans Gene Knockout Consortium (Oklahoma Medical Research Foundation), the Sanger Centre and Iain Johnstone (University of Glasgow) for antibodies, reporter genes, cosmids and strains. We also thank Martyn Quinn, Margaret Mullin and Laurence Tetley (University of Glasgow) for help with the EM studies, Sandra Terry for help with DNA sequence analysis and Brian Shiels for helpful discussion and comments on the manuscript. This work was funded by The Wellcome Trust.