Caveolin-2 Is Required for Apical Lipid Trafficking and Suppresses Basolateral Recycling Defects in the Intestine of Caenorhabditis elegans

Abstract

Caveolins are plasma membrane–associated proteins that colocalize with, and stabilize caveolae. Their functions remain unclear although they are known to be involved in specific events in cell signaling and endocytosis. Caenorhabditis elegans encodes two caveolin genes, cav-1 and cav-2. We show that cav-2 is expressed in the intestine where it is localized to the apical membrane and in intracellular bodies. Using the styryl dye FM4-64 and BODIPY-labeled lactosylceramide, we show that the intestinal cells of cav-2 animals are defective in the apical uptake of lipid markers. These results suggest parallels with the function of caveolins in lipid homeostasis in mammals. We also show that CAV-2 depletion suppresses the abnormal accumulation of vacuoles that result from defective basolateral recycling in rme-1 and rab-10 mutants. Analysis of fluorescent markers of basolateral endocytosis and recycling suggest that endocytosis is normal in cav-2 mutants and thus, that the suppression of basolateral recycling defects in cav-2 mutants is due to changes in intracellular trafficking pathways. Finally, cav-2 mutants also have abnormal trafficking of yolk proteins. Taken together, these data indicate that caveolin-2 is an integral component of the trafficking network in the intestinal cells of C. elegans.

INTRODUCTION

Caveolae are 50–100-nm invaginations located on the plasma membrane of many cell types (Anderson, 1998). They are enriched in lipid raft–associated molecules: cholesterol, sphingolipids, glycosphingolipids, and many signaling and receptor proteins. Although identified more than 50 years ago, their functions have remained elusive. However, they are known to play roles in specific cell-signaling processes, and it is clear that endocytosis through caveolae is an important clathrin-independent endocytic pathway (Parton and Richards, 2003; Parton and Simons, 2007). A range of molecules including glycosphingolipids, glycosylphosphatidylinositol (GPI)-anchored proteins, cholera toxin B, and SV40 virus have been shown to traffic through caveolae-dependent endocytosis pathways (see Marsh and Helenius, 2006; Mayor and Pagano, 2007; Parton and Simons, 2007 for reviews).

Caveolae require caveolin proteins for formation so that depletion of caveolins results in loss of morphologically detectable caveolae (Drab et al., 2001; Galbiati et al., 2001). Furthermore, introduction of caveolin into cells that do not produce caveolae stimulates caveolar biogenesis (Lipardi et al., 1998). The majority of caveolin appears to traffic to the plasma membrane and is relatively immobile; however, internal caveolin-containing structures, named caveosomes, have been visualized (Parton and Simons, 2007) and appear to be an integral part of the caveolin-dependent endocytic machinery (Nichols, 2003). Mammalian cells have three caveolin subtypes: caveolin 1 is widely expressed, but is found at particularly high levels in adipocytes, endothelial cells, and fibroblasts. Caveolin 2 interacts with caveolin 1, whereas caveolin 3 is expressed in myocytes only. Caveolin knockout mice show a range of physiological defects (Le Lay and Kurzchalia, 2005). These include defects in lipid homeostasis (Cohen et al., 2004a; Le Lay and Kurzchalia, 2005; Martin and Parton, 2005; Parton and Simons, 2007). Central to the defects in lipid homeostasis is the role of caveolin-1 in endocytosis in adipocytes (Cohen et al., 2004b; Martin and Parton, 2005; Le Lay et al., 2006). Mutations in caveolin-1 have also been shown to be associated with the rare lipodystrophies: Berardinelli-Seip congenital lipodystrophy (OMIM:269700) and an atypical partial lipodystrophy and hypertriglyceridemia (Cao et al., 2008; Kim et al., 2008a). Caveolin-1 is also required for liver regeneration (Fernandez et al., 2006; Frank and Lisanti, 2007).

Caenorhabditis elegans has two caveolin genes; cav-1 and cav-2 (Tang et al., 1997). cav-1 has been implicated in meiotic progression in the germ line (Scheel et al., 1999) and in neurotransmission at the neuromuscular junction (Parker et al., 2007). cav-1 is widely expressed in embryos but gradually develops a more restricted pattern of expression so that in late larval and adult animals it is restricted to the neuromuscular system (Scheel et al., 1999; Parker et al., 2007) and germ line (Scheel et al., 1999). The behavior of CAV-1 in the germ line and embryos is highly dynamic (Sato et al., 2006). Phylogenetic analysis and expression studies of CAV-1 and CAV-1-mammalian caveolin hybrids in mammalian cells suggest that CAV-1 does not induce caveola biogenesis (Kirkham et al., 2008) and so may have a different role to mammalian caveolins. The function of CAV-2 has not previously been investigated.

In this work we show that cav-2 is expressed and functions in the intestine of C. elegans. The intestine of C. elegans consists of a single layer of 20 polarized epithelial cells arranged as a tube. The apical surface of the cell is shaped into microvilli and forms the barrier with the lumen of the gut, whereas the basolateral surface is in contact with the pseudocoelomic space (body cavity; McGhee, 2007). Recent work using genetic and in vivo approaches has led to the development of the C. elegans intestine as a system for the study of trafficking (Fares and Grant, 2002; Grant and Sato, 2006). Grant et al. (2001) showed that exposure of the apical surface to lipid markers, such as FM4-64, and fluid-phase markers, such as Texas Red-BSA, results in accumulation of the markers in gut granules. Application of FM4-64 to the basolateral membrane has the same result, but basolaterally applied fluid-phase markers, such as Texas Red-BSA or ssGFP (GFP secreted into the body cavity; Fares and Greenwald, 2001b), undergo recycling. It has been shown that both RME-1 and RAB-10 are required for this fluid-phase recycling in the C. elegans intestine (Grant et al., 2001; Chen et al., 2006). RME-1, an EH domain protein, is required in a range of tissues for proper endocytosis, but mutations in the small GTPase RAB-10 cause gut-specific defects in recycling. In both cases the result of these defects is the accumulation of enlarged endocytic vesicles in the intestinal cells (Grant et al., 2001; Chen et al., 2006).

Here we report that depletion of CAV-2 results in reduced transport of lipid markers from the apical side of the intestine. We show that depletion of cav-2 suppresses the vacuolar phenotype caused by defective basolateral fluid-phase recycling in rme-1 and rab-10 mutants. Curiously, cav-2 mutants do not have any apparent defects in the trafficking of a variety of markers for basolateral endocytosis, but cav-2 depletion does restore normal trafficking of these markers in rme-1 mutants.

MATERIALS AND METHODS

C. elegans Strains and General Phenotypic Analysis

Maintenance, culture, and genetic manipulation of C. elegans were performed using standard methods (Lewis and Fleming, 1995). A list of strains used in this work is given in Supplementary Table S1. Defecation cycle length was measured as described in Kwan et al. (2008). Ten worms were scored for each strain. For brood-size assays, NGM plates were seeded with a 1-cm “patch” of Escherichia coli OP50. An individual L4 was picked to a single plate and then picked to a fresh patch plate every 12 h for 100 h. The number of live eggs laid on each plate was counted. Ten worms were scored for each strain. To quantify gut vacuoles, worms were anesthetized in 0.02% tricaine, 0.001% levamisole. Gut vacuoles were scored using differential interference contrast (DIC) optics on a Zeiss Axioskop upright microscope (Welwyn Garden City, UK). Fifty worms of each strain were scored.

Plasmids, Fluorescent Protein Reporter Constructs, and Transgenic Strains

CAV-2::GFP fusion proteins were generated with green fluorescent protein (GFP) inserted close to the 5′ or 3′ end of the gene to give N and C terminal fusions, respectively. Constructs included ca. 4 kb of DNA upstream of the ATG start (the forward primer was SP650: cgatctactatgcctccaatggg) and 2 kb of DNA downstream of the stop codon (the reverse primer was SP653: cgagtgaaagcacgtgacgac). The cav-2::gfp transcriptional fusion was constructed with the same upstream DNA. The cav-2::yfp fusion used in colocalization studies contains an N terminal fusion of yellow fluorescent protein (YFP) to the full cav-2 gene and used the same upstream and downstream regions.

rme-1 and rme-8 fusions to cyan fluorescent protein (CFP) were generated using the PCR fusion technique (Hobert, 2002), enabling attachment of CFP to the C terminus. CFP was amplified with a 3′ untranslated region (UTR) from let-858 using plasmid pSP002, a derivative of pHAB200 (Baylis et al., 1999) containing CFP rather than GFP, as a template. rme-1 and rme-8 genes and their respective 5′ regulatory regions were amplified from wild-type genomic DNA, including ∼1 kb of upstream sequence. The upstream region was defined by the forward primers SP1062 gtccggtggaacttatgaacaactggc and SP1294 ccaagtaggtagcttgcaactcgc for rme-1 and rme-8, respectively. The reactions were mixed and amplified using nested primers to produce rme-1::cfp or rme-8::cfp fusions.

The rescuing construct containing cav-2 driven by its own promoter (pSP040) consisted of a genomic DNA fragment amplified using the upstream and downstream oligonucleotides SP650 and SP653 described above cloned into pGEM-T (Promega, Southampton, UK).

To construct intestine specific cav-2 clones, we used Gateway multisite technology (Invitrogen, Paisley, UK). Individual entry clones containing the vha-6 promoter and cav-2 cDNA were produced and then combined in a Gateway reaction into the destination vector pHP2 (H. Peterkin and H. Baylis, unpublished data). The vha-6 promoter clone contained a fragment, similar to that previously used by Grant and colleagues (Chen et al., 2006), amplified using the oligonucleotides ggggacaagtttgtacaaaaaagcaggctcgcgttcaccactcgaccaccgaac and ggggacaacttttgtatacaaagttgttttttatgggttttggtaggttttag. The cav-2 cDNA was amplified using the oligonucleotides ggggacaacttttctatacaaagttgaaaaatgactcgtcagaatacttccgaaag and ggggacaactttattatacaaagttgttaaacatgatgaatgtgtttttc.

Constructs were microinjected into N2 worms at 7 ng/μl, and stable lines were selected. Where appropriate, worms were analyzed for fluorescence using a Leica SP1 confocal microscope (Milton Keynes, UK).

rme-1 and cav-2 RNA Interference

RNA interference (RNAi) was performed by injection of double-strand RNA (dsRNA) into the animal's gonad (Fire et al., 1998; Montgomery and Fire, 1998). From cav-2 and rme-1, 2.4 and 1.2 kbp, respectively, were amplified using the primers SP652: atgctagcatggataaggaagatcatcac, and SP654: atgcggccgctgctccaactgagtagaagaatgg for cav-2 and SP1027: ccaatgatcctgctcgtcgggc and SP1028: ggctgaaggctcctcctctgacag for rme-1. PCR products were cloned into pGEM-T (Promega), and the resulting plasmids were used as a template for transcription from the SP6 and T7 promoters using a Megascript kit (Ambion, Austin, TX). dsRNA was produced by annealing RNA from the two preparations (Hull and Timmons, 2004). The E. coli chloramphenicol acetyltransferase gene (cat) was used as a control.

Analysis of Endocytosis

Experiments using FM4-64 and labeled BSA were conducted as described by Grant et al. (2001). To assess uptake of lactosylceramide, animals were incubated for 2 h in 5 μM BODIPY-tagged lactosylceramide (BODIPY FL C5 LacCer; Invitrogen, Paisley, UK).

For microscopic analysis worms were anesthetized in 0.02% tricaine, 0.001% levamisole and analyzed using a Leica SP1 laser scanning confocal microscope. For the analysis of endocytic markers and to avoid autofluorescence, intestinal GFP signals were detected at 507–513 nm (Chen et al., 2006). We confirmed that the contribution of autofluorescence at this wavelength in our equipment was negligible. Where appropriate (e.g., for BODIPY FL C5 LacCer staining) images were taken using identical settings, to allow direct comparison. Quantification was performed using Image J (http://rsb.info.nih.gov/ij/; NIH, Bethesda, Maryland, MD). Analysis used three regions of interest from six worms for each condition.

Immunohistochemistry

Antibodies were raised in rabbits by Covalab UK (Cambridge, United Kingdom) against two C. elegans CAV-2 peptides: CNTQRPPIPQYDTVD and CTPQRRSHPQYDNLD and affinity-purified. For immunocytochemistry worms were freeze-cracked and fixed in methanol and acetone (Duerr et al., 1999). Primary antibodies were applied at 15 ng/μl and incubated overnight. The secondary antibody was Alexa 488 mouse anti-rabbit IgM (Molecular Probes, Eugene, OR).

RESULTS

CAV-2 Is Expressed in the Intestine

To explore the function of caveolin 2 in C. elegans, we first established its location in the animal. We constructed two full-length cav-2::GFP fusions, with GFP inserted close to either the N or C terminal coding region of cav-2. A third transcriptional fusion drove GFP alone under the control of the cav-2 promoter. All three constructs were expressed in the intestine of hermaphrodite and male wild-type animals (Figure 1). We did not detect expression in any other tissues. This pattern was confirmed with an anti-CAV-2 antibody (Figure 1). Therefore, cav-2 has a distinct expression pattern from cav-1, which is expressed in embryos, neurons, and body-wall muscles (Scheel et al., 1999; Parker et al., 2007).

Figure 1.

Caveolin-2 is located in the apical membranes of intestinal cells and in intracellular CAV-2 bodies. (A, C, and D) Confocal images of C. elegans carrying a cav-2::gfp transgene. CAV-2::GFP is expressed in the intestine (I) and is localized to the apical membrane (AM), as shown in A and at higher magnification in C and D. A and D are overlaid with transmitted light images. I, intestine; AM, apical membrane; BM, basolateral membrane; L, lumen; P, pharynx. (B) Staining with an anti-CAV-2 antibody confirms expression in the intestine. L, lumen; P, pharynx. (E and F) CAV-2 shows a punctate distribution in the apical membrane. (E) Confocal images of a section of apical membrane in CAV-2::GFP–expressing worms and (F) worms stained with an anti-CAV-2 antibody. (G and H) CAV-2 is also localized in vesicular structures (CAV-2 bodies). Confocal images of CAV-2 bodies (CB) identified by CAV-2::GFP labeling (green). Autofluorescent gut granules (GG) are shown in red. Scale bars, 10 μm.

The C. elegans intestine consists of a tube of 20 polarized epithelial cells, with a dense layer of microvilli on their apical membrane. Both cav-2::gfp fusion proteins localized to the apical side of the intestinal cells, very close to or within the membrane. Expression was first detected in midstage embryos and continued thence forward. The expression at the apical membrane was highest and most evenly distributed in L1 and L2 animals. With maturation, levels of CAV-2 were highest at the anterior and posterior cells of the intestine (Figure 1). Interestingly, the labeling of the membrane was discontinuous (Figure 1). We also observed the presence of discrete, spherically shaped, CAV-2::GFP–positive vesicles (diameter, 1.6 μm; n = 30). These vesicles were most common at the posterior end of the gut and were distinct from autofluorescent gut granules (AFGGs), which are terminal endocytic lysosomes (Figure 1). We call these structures CAV-2 bodies, as their relationship to mammalian caveosomes is unknown.

cav-2 Animals Have Reduced Fertility

The intestine of C. elegans performs a range of important functions including the following: the uptake and transport of nutrients, the production and export of egg yolk proteins, and the encoding of an ultradian oscillator controlling the defecation cycle (Kimble and Sharrock, 1983; Dal Santo et al., 1999; McGhee, 2007). To analyze the function of cav-2 in the intestine, we used cav-2 RNAi and two strains carrying putative null mutations of the cav-2 gene: HB508, cav-2(tm394), and BA1090, cav-2(hc191). cav-2 mutants and cav-2(RNAi)-treated worms do not show any gross phenotypes. They grow at the normal rate and have normal life spans and normal defecation cycles (data not shown). However, both mutants have a slight but reproducible reduction in brood size (N2: 315 ± 8; cav-2(hc191): 261 ± 23; cav-2(tm394): 284 ± 28; mean ± SD). That the depletion of cav-2, presumably in the intestine, reduces brood size might be explained by deficiencies in nutrient uptake, trafficking, or yolk protein production.

cav-2 Mutants Exhibit Altered Apical Trafficking

To test whether CAV-2 depletion altered endocytosis or trafficking in the intestine, we used the fluorescent fluid and lipid-phase markers, Texas Red-BSA and FM4-64, respectively, as used by Grant and colleagues (Grant et al., 2001). To follow trafficking from either side of the cell, dyes were applied to the apical (luminal) side of the intestine by feeding or to the basolateral side by microinjection into the body cavity. The anterior intestinal cells were compared. AFGGs were also visualized. Wild-type animals exposed to FM4-64 by apical or basolateral delivery rapidly endocytosed the dye and directed it to the AFGGs (Figure 2). Basolateral uptake of FM4-64 in cav-2(tm394) animals was normal (Figure 2), but apical uptake of FM4-64 in these animals was perturbed. Uptake is reduced by more than 50% (Figure 2O), and no or very little of the FM4-64 reached the AFGGs (Figure 2K). Any FM4-64 that was delivered to the AFGG region only stained the periphery of the granule. In contrast, wild-type animals had staining throughout the AFGGs (Figure 2, G–I). This phenotype was rescued by reintroducing a full-length wild-type copy of cav-2 into cav-2(tm394) animals (HB572; Figure 2M). To confirm that this phenotype resulted from defective uptake in the intestine, we rescued cav-2 using an intestine specific promoter (vha-6p; Chen et al., 2006). Animals rescued in this way have normal uptake of FM4-64 (Figure 2, N and O). Thus, cav-2 mutants have limited ability for uptake of FM4-64 at apical membranes and process the small amount of internalized dye differently to N2s.

Figure 2.

cav-2 mutants show defective accumulation of the lipid trafficking indicator FM4-64. Autofluorescence, primarily associated with gut granules, is shown in green, and FM4-64 is shown in red. Merged images reveal colocalization. (A–C) Wild-type (N2) animals exposed to basolateral FM4-64 show colocalization of FM4-64 and autofluorescence. (D–F) cav-2(tm394) animals exposed to basolateral FM4-64 show normal uptake. (G–I) Wild-type animals exposed to apical FM4-64 show colocalization of FM4-64 and autofluorescence. (J–L) cav-2(tm394) animals exposed to apical FM4-64 take up very low levels of dye. FM4-64 is no longer apparent in autofluorescent gut granules. (M) cav-2(tm394) animals carrying a wild-type cav-2 transgene exposed to apical FM4-64 show normal uptake. (N) cav-2(tm394) animals carrying an intestine specific cav-2 cDNA transgene, vha-6p::cav-2 exposed to apical FM4-64 show normal uptake. Scale bars, (A–N) 10 μm. (O) Quantification of FM4-64 fluorescence. Error bars, SD. For each strain six animals were sampled in three regions.

To observe fluid-phase trafficking, we performed similar experiments using Texas Red-BSA. We did not detect any differences between wild-type and cav-2(lf) animals. To further explore fluid-phase uptake from the body cavity, we used a strain (GS1912) in which GFP (referred to as ssGFP) is secreted into the body cavity of worms from body wall muscles (Fares and Greenwald, 2001a). We observed no changes in distribution of ssGFP either in the intestine (see Figure 5) or in coelomocytes (Supplementary Figure S1), another cell type used in the study of endocytosis.

Figure 5.

cav-2 depletion restores the trafficking of endocytic markers in rme-1 mutants to normal. (A–D) Confocal images showing the distribution of secreted GFP (ssGFP). rme-1 mutants cause the accumulation of large ssGFP-filled vesicles in the intestine (C), but these are not observed when cav-2 is depleted in rme-1 animals (D). cav-2(RNAi) animals show normal ssGFP localization (B). (E–P) Distribution of hTAC::GFP (E–H), hTfR::GFP (I–L), and LMP-1::GFP (M–P). Scale bars,10 μm. (Q) Quantification of the number of endocytosis marker-GFP puncta. rme-1(b1045) animals were not quantified as the distribution of the marker is substantially different to the other samples. Error bars, SD. For each strain six animals were sampled in three regions. Each of the three markers has a characteristic pattern in the intestine of wild-type animals (E, I, and M). Depletion of cav-2 does not alter this pattern (F, J, N, and Q). However depletion of rme-1 produces characteristic changes in the distribution of each marker (Chen et al., 2006). These changes are suppressed by the depletion of cav-2 in the rme-1(b1045) animals (H, L, P, and Q).

To confirm that cav-2 mutants have defects in apical trafficking, we exposed animals to BODIPY-tagged lactosylceramide (BODIPY FL C5 LacCer, Invitrogen). It has previously been shown that glycosphingolipid analogues are internalized through a predominantly caveolin-dependent route in human cells (Singh et al., 2003; Mayor and Pagano, 2007; Singh et al., 2007). In wild-type animals exposed to 5 μM BODIPY-LacCer for 2 h, fluorescent LacCer accumulates in a punctate pattern in the intestine. These putative vesicles do not generally overlap with AFGGs (Figure 3). In cav-2(tm394) animals, BODIPY-LacCer accumulation is reduced by ∼70% (Figure 3). Thus cav-2 depletion causes a specific defect in the uptake of both FM4-64 and BODIPY LacCer from the apical side of the intestine.

Figure 3.

cav-2 mutants show defective accumulation of fluorescently labeled lactosylceramide. (A) Wild-type worms show substantial uptake of BODIPY FL C5 LacCer into small vesicles of an unknown nature. (B) Autofluorescence for comparison with A. (C) cav-2(tm394) animals show significantly reduced BODIPY FL C5 LacCer uptake. (D) Autofluorescence for comparison with C. Images were collected using identical confocal settings. Images are representative examples of >20 worms. Scale bars, 10 μm. (E) Quantification of BODIPY FL C5 LacCer fluorescence. Error bars, SD. For each strain six animals were sampled in three regions.

cav-2 Suppresses Trafficking Defects in Mutants with Basolateral Recycling Defects

Our results implicate cav-2 in trafficking events in the intestine. To investigate if cav-2 plays a wider role in intestinal trafficking, we tested for interactions with a known intestinal trafficking gene, rme-1. RME-1 is strongly implicated in endocytic recycling and rme-1 mutants develop large intestinal vacuoles (Grant et al., 2001). We tested whether rme-1 and cav-2 interact in the intestine. We found that cav-2 loss-of-function mutations resulted in a marked reduction in the number and size of vacuoles resulting from rme-1 RNAi (Figure 4). Similarly, we made a double homozygous strain carrying cav-2(tm394) and rme-1(b1045) (HB570), and these worms exhibited reduced vacuole numbers compared with rme-1(b1045) alone. Rescue of the cav-2 deficiency in cav-2(tm394); rme-1(b1045) double mutants, by expression of a wild-type cav-2 transgene, or intestine-specific expression of cav-2 from a vha-6p::cav-2(+) transgene resulted in an increase in vacuole numbers (Figure 4). Thus, depletion of cav-2 can significantly rescue the vacuole phenotype that results from rme-1 depletion. In contrast to cav-2 mutants, rme-1 mutants show normal accumulation of FM4-64 from both the apical and basolateral membrane (Grant et al., 2001). However, rme-1(b1045);cav-2(tm394) worms show similarly reduced levels of FM4-64 uptake from the apical membrane as do cav-2(tm394) worms (Figure 2; data not shown). So although cav-2 mutations rescue the rme-1 defect, rme-1 mutations do not rescue the cav-2 lipid-uptake defect. rme-1 mutants also have defects in yolk protein uptake by oocytes (Grant et al., 2001). Depletion of cav-2 by RNAi did not suppress this defect (Supplementary Figure S2).

Figure 4.

Reduction of cav-2 function alleviates the vacuolar phenotype associated with defects in trafficking caused by rme-1 and rab-10 reduction. (A) Knockdown or knockout of cav-2 reduces vacuole formation in rme-1 loss-of-function animals. The mean number of vacuoles in animals of the indicated genotype is shown. rme-1 function was reduced by mutation rme-1(b1045) or RNAi. Similarly, cav-2 function was reduced by mutation tm394 or hc191 or by RNAi. Results for rme-1 loss-of-function alone are shown in blue, whereas those for rme-1 and cav-2 loss of function are shown in green. tm394 or hc191 alone do not cause a phenotype. Expression of a wild-type cav-2 transgene (Ex cav-2) or an intestinal specific cav-2 transgene (vha6p::cav-2(+)) in the cav-2; rme-1 double mutants restores vacuole levels to the normal rme-1 level. Error bars, SEM. (B) Detailed breakdown of vacuolar number. rme-1 was knocked down by RNAi in strains that carried YP-170::GFP with or without a cav-2 mutation (HB523 and DH1033, respectively). (C and D) CAV-2 and RME-1 are present on different subcellular structures. Confocal images of a section of intestine (C) and a single intestinal cell (D) in animals carrying cav-2::yfp (green) and rme-1::cfp (red). CAV-2 labels the apical membrane (AM) and sparse CAV-2 bodies (CB), whereas RME-1 primarily labels the basolateral membrane (BM) and putative recycling endosomes (RE). Scale bars, 20 μm. (E) Reduction of cav-2 function suppresses the vacuolar phenotype caused by rab-10 loss of function. The mean number of vacuoles in animals of the indicated genotype is shown. Error bars, SEM.

Because cav-2 and rme-1 interact, we tested whether the two proteins colocalize by using rescuing rme-1::cfp (see Supplementary Figure S3) and cav-2::yfp constructs. RME-1::CFP was localized to the basolateral membrane and distinct internalized vesicles, which are clearly separate from CAV-2::YFP labeling (Figure 4, C and D). We cannot rule out low level or transient colocalization between these two proteins. The distribution of neither gene is changed by loss of the other (data not shown).

To test whether the ability of cav-2(lf) to suppress defects in basolateral recycling was specific to rme-1, we performed cav-2(RNAi) on rab-10(q373) mutant animals. rab-10 animals also have defects in basolateral endocytic recycling (Chen et al., 2006) and also accumulate large fluid filled vesicles. Unlike rme-1, rab-10 mutants only have observable defects in the intestine. We found that cav-2(RNAi) was also able to suppress the number of vacuoles in rab-10(q373) animals (Figure 4E) confirming that depletion of cav-2 suppresses the abnormal formation of enlarged basolateral endosomes.

cav-2 Animals Have Normal Trafficking of Basolateral Transmembrane Cargoes

The ability of cav-2(lf) to rescue the accumulation of abnormal recycling endosomes in rme-1 mutants might be explained by defects in basolateral endocytosis or by subsequent altered trafficking of endocytosed cargoes. We observed no changes in the distribution of FM4-64, Texas Red BSA, or ssGFP on exposure of the basolateral membrane of the intestine, although it should be noted that both Texas Red BSA and ssGFP are hard to detect internally in normal animals. We therefore tested whether cav-2 depletion altered trafficking of three cargo proteins fused to GFP that were previously developed and utilized by Grant and colleagues (Chen et al., 2006). They are human transferrin receptor (hTfR-GFP), which has been used as a marker of clathrin-dependent uptake and of subsequent rme-1–mediated recycling (Chen et al., 2006 and references therein); human IL-2 receptor (hTAC-GFP), a marker for clathrin-independent uptake and also subsequent rme-1–mediated recycling (Chen et al., 2006 and references therein); and C. elegans LMP-1 (LMP-1-GFP), which is the worm ortholog of CD63/LAMP that labels endocytic compartments within the intestine (Hermann et al., 2005; Chen et al., 2006). In each case we observed distributions similar to those reported by Chen et al. (2006) in wild-type animals (Figure 5). We then examined the distribution in animals in which cav-2 was depleted. The number of GFP puncta was quantified for the hTfR and hTAC cargoes (Figure 5Q). In each case the distribution and number of puncta was unaltered (Figure 5), suggesting that both clathrin-dependent and -independent basolateral uptake and recycling are unaltered in cav-2(lf) animals.

cav-2 Depletion Fully Suppresses Recycling Defects in rme-1 Mutants

Chen et al. (2006) observed redistribution of each of the markers used above in rme-1(b1045) animals compared with wild type. We extended our analysis to address the effect of cav-2(RNAi) on the distribution of the three cargo proteins in rme-1(b1045) animals. For each marker we observed a distribution similar to that observed by Chen et al. in the rme-1(b1045) background (Figure 5). However when we depleted cav-2 by RNAi, the distribution of each marker was returned to the wild-type pattern (Figure 5). We also determined the number of hTAC and hTfR GFP puncta in rme-1(b1045); cav-2(RNAi) animals and observed no significant change in comparison to wild-type animals. Thus, depletion of cav-2 suppresses the defects observed in rme-1–deficient animals.

cav-2 Mutants Have Altered Yolk Protein Distribution

Because cav-2 alters trafficking in the intestine, we asked whether it could alter other aspects of intestinal function. To observe the synthesis and movement of yolk proteins (vitellogenins), we used a yolk protein-GFP fusion, YP-170::GFP (also known as VIT-2::GFP (Grant and Hirsh, 1999). This construct allows visualization of yolk protein and its export into the pseudocoelomic space for ultimate delivery to the growing oocytes (Kimble and Sharrock, 1983; Hall et al., 1999). We crossed this marker into cav-2(tm394) worms to generate HB523. Analysis of HB523 revealed that cav-2(lf) resulted in an increased accumulation of YP-170::GFP in the pseudocoelom, a swollen tail phenotype, and punctate accumulations of YP-170::GFP (Figure 6). Levels of YP-170::GFP in the HB523 intestine appeared normal but less localized to the basolateral membrane compared with wild type. cav-2 RNAi has the same effect (not shown). The level of YP-170::GFP in embryos (Figure 6, C and D) and oocytes (Supplementary Figure S2) is unaltered in cav-2 mutants, suggesting that changes in the level of YP-170::GFP in the body cavity do not result from altered uptake into oocytes. Therefore it appears that yolk protein is being exported from the intestine in larger than normal amounts, again suggesting that CAV-2 is intimately involved in basolateral trafficking in intestinal cells.

Figure 6.

Trafficking of yolk protein YP-170::GFP is disrupted in cav-2 mutants. Confocal fluorescent and transmitted light images of animals carrying the YP-170::GFP reporter fusion (Grant and Hirsh, 1999). (A and B) cav-2(tm394) mutant animals show accumulation of YP-170::GFP in large puncta in the pseudocoelomic space. These tend to accumulate in the tail which is often swollen. Compare the swollen tail phenotype in B and D. (C and D) In wild-type animals the YP-170::GFP is present at lower levels and tends to be diffuse with low numbers of granules (C). (E and F) cav-2 mutants (E) and wild-type (F) animals show accumulation of YP-170::GFP in developing eggs. Scale bars, 10 μm.

DISCUSSION

The functions of caveolins remain incompletely understood. We have shown that in C. elegans, cav-2 is located in the apical membrane and in intracellular bodies in the intestine. Depletion of cav-2 disrupts apical lipid trafficking but has no apparent effect on basolateral transport. However cav-2 depletion is also able to suppress defects in basolateral recycling caused by depletion of either rme-1 or rab-10.

C. elegans has two caveolin genes; our work and previous work has shown that these two proteins have different and largely nonoverlapping expression patterns. CAV-2 appears to be solely or primarily expressed in the intestine. CAV-1 shows a wider expression pattern in embryos and early larvae (Scheel et al., 1999; Sato et al., 2006; Parker et al., 2007) and thus may overlap CAV-2 expression early in development. However by adulthood, cav-1 expression is confined to the neuromuscular system and germ line. That cav-2 also functions, primarily, in the intestine is supported by our observation of intestinal phenotypes that are rescued by intestinal expression and by the lack of any effect of cav-2 on endocytosis in coelomocytes or oocytes. Both CAV-1 and -2 appear to be predominantly located in the membrane; however, in both cases intracellular vesicles, CAV-1 and -2 bodies, respectively, are also observed (Sato et al., 2006). Sato et al. (2006) demonstrated that the smaller CAV-1 bodies observed in the germ line are mobile. The relationship of the CAV-2 bodies that we observed to mammalian caveosomes (Pelkmans et al., 2001) is unclear. We found no colocalization of RME-1 and -8 with CAV-2 bodies and Sato et al. (2006) found no colocalization of EAA-1 or RME-2 with CAV-1 bodies; thus, so far no overlap between these intracellular CAV bodies and other endocytic markers in C. elegans has been found.

It is not known if caveolae are present in C. elegans because no systematic analysis of cells for the presence of caveolae has been published. The ability of CAV-2 to induce the formation of caveolae has not been tested. Kirkham et al., (2008) have shown that C. elegans CAV-1 does not induce caveolae in mammalian cells and have used phylogenetic analysis to indentify amino acid changes that may underlie this difference (Kirkham et al., 2008). In particular they identified two key regions: one involving residues 56–59 and another involving residues S104, F12, and A129 in the human caveolin 1 sequence (Supplementary Figure S4), which may determine the ability of caveolins to form caveolae. In CAV-1 the equivalent residue to P56 is A and to F124 is S. However in CAV-2 these two residues are identical to those found in humans; thus, it may be that, unlike CAV-1, CAV-2 is able to induce the formation of caveolae. Interestingly we observed a punctuate distribution of CAV-2 in the membrane using both CAV-2::GFP fusion and anti-CAV-2 antibodies.

cav-2 is expressed in the intestine of C. elegans. The intestine consists of a single layered tube of polarized cells (McGhee, 2007). C. elegans has a relatively simple anatomy, and so the intestine plays a number of roles. First, it is the primary nutritional interface with the environment, with food passing rapidly through the intestine, while nutrients are absorbed and transported. Second, it is the site of yolk protein production. Third, it is the site of a calcium-based signaling system that controls defecation, an event that occurs every 50 s (Dal Santo et al., 1999). Finally, the intestine is also one of two tissues, along with the epidermis (hypodermis), that are used for fat storage (Ashrafi, 2007). Thus the intestine has large numbers of fat droplets and may be regarded as the C. elegans adipose tissue. We observed that cav-2 mutants have defects in lipid trafficking from the apical membrane: the uptake of the styryl dye FM4-64 and of the BODIPY LacCer were disrupted. These results suggest that cav-2 is required for normal lipid uptake in the intestine. This function is reminiscent of the importance of caveolin function to lipid metabolism in mammals (Cohen et al., 2004b; Le Lay and Kurzchalia, 2005; Martin and Parton, 2005; Le Lay et al., 2006; Parton and Simons, 2007; Garg and Agarwal, 2008; Heimerl et al., 2008; Kim et al., 2008b). Caveolae and CAV1 are highly concentrated in mammalian adipocytes, and it has become clear that caveolae are an important part of the trafficking system that regulates the homeostasis of lipids in adipocytes and the formation of lipid bodies. The recent discovery of mutations in CAV1 that underlie lipodystrophies emphasizes this (Cao et al., 2008; Kim et al., 2008a). Thus caveolins may play a conserved role in lipid homeostasis. We note that it is unlikely that CAV-2 provides the only route of lipid uptake because cav-2 mutants grow normally. Furthermore C. elegans is a cholesterol auxotroph, and cholesterol depletion has the same affect on locomotion, defecation, and fecundity and is equally lethal to cav-2 animals as it is to N2s (Parker, unpublished observation; Matyash et al., 2001; Entchev and Kurzchalia, 2005). Therefore, C. elegans should provide a useful genetic tool for the analysis of caveolin-based lipid transport.

Depletion of cav-2 also suppresses the vacuolated phenotype observed in rme-1 and rab-10 mutants. Mutations in both rme-1 and rab-10 result in the presence of large vesicular structures as a result of defects in basolateral recycling (Grant et al., 2001; Chen et al., 2006). Our understanding of endocytic recycling has advanced substantially in recent years (Maxfield and McGraw, 2004). Endocytosis at the plasma membrane causes markers (e.g., transferrin receptors) to be transported to sorting endosomes. From sorting endosomes, receptors or lipids that are to be recycled to the plasma membrane may either traffic directly to the plasma membrane or may progress to the endosome recycling compartment (ERC) from where they can traffic to the plasma membrane (Maxfield and McGraw, 2004). Genetic studies in C. elegans have enabled new insights into recycling mechanisms. rab-10 and rme-1 are components of the recycling pathway at the basolateral membrane of the intestine. Depletion of rme-1 results in a build-up of material in ERCs, whereas depletion of rab-10 results in increases in sorting endosome size and number; thus, the two proteins act at different steps in the pathway (Grant et al., 2001; Chen et al., 2006). cav-2 mutants suppress both rab-10 and rme-1. One explanation for this is that loss of cav-2 reduces endocytosis at the basolateral membrane; however, this seems unlikely as the distribution of hTfR::GFP, hTAC::GFP, and LMP-1::GFP are all unaltered in cav-2 mutants, suggesting that clathrin-dependent and some clathrin-independent uptake is normal. The ability of cav-2 depletion to suppress both rab-10 and rme-1 may be explained if it increases trafficking directly from sorting endosomes to the plasma membrane. This would reduce build up of material in the sorting endosomes and ERC. Interestingly, Nilsson et al. (2008) recently identified another gene, num-1, which is able to suppress the effect of rme-1 in the intestine. num-1, the C. elegans Numb ortholog is able to suppress rme-1 mutations but not rab-10 mutation. It also differs from cav-2 in that it is clearly located at the basolateral membrane.

It remains unclear how depletion of cav-2 might achieve these effects, particularly as cav-2 appears to be primarily located at the apical membrane. However our observation that the level of yolk protein in the body cavity is increased also indicates that cav-2 depletion has broad-ranging effects on intestinal trafficking. One possible mechanism, given our data above, is that cav-2 depletion causes a change in the lipid composition of the cell membranes and that this in turn causes changes in trafficking throughout the cell. It is also possible that changes in apical trafficking could alter basolateral endocytosis through other mechanisms that are as yet unknown. We know very little about both apical trafficking and transcytosis in these cells. However the apparently normal distribution of, for example, hTfR::GFP, in rme-1;cav-2 animals suggests that cav-2 does not suppress rme-1 through a mechanism that results in large scale relocation of endocytosed markers. Dissecting the mechanism by which cav-2 suppresses defects in basolateral recycling should provide new insights into the molecular control of endocytosis in polarized cells.