C elegans worms:.

Standard procedures were used for culturing C. elegans worms (Brenner 1974). All strains were grown at 20° unless otherwise stated. The wild-type C. elegans strain Bristol N2 and the following mutant alleles were used: LGI, rab-10(dx2) (Chen et al. 2006); LGII, rrf-3(pk1426) (Simmer et al. 2002) and unc-4(e120) (Brenner 1974); LGIII, unc-119(ed3) (Maduro and Pilgrim 1996); LGV, rme-1(b1045) (Grant et al. 2001), num-1(bc365) (this study), and num-1(ok433) (this study). The transgenes used were arIs37[P_myo-3∷ssGFP dpy-20(+)] I (Fares and Greenwald 2001a), bIs1[vit-2∷gfp rol-6(+)] X (Grant and Hirsh 1999), pwIs50[lmp-1∷gfp unc-119(+)] (Treusch et al. 2004), pwIs69[P_vha-6∷gfp∷rab-11 unc-119(+)] (Chen et al. 2006), pwIs72[P_vha-6∷gfp∷rab-5 unc-119(+)], pwIs87[P_vha-6∷gfp∷rme-1 unc-119(+)] (Hermann et al. 2005), pwIs90[P_vha-6∷hTfR∷gfp], pwIs112[P_vha-6∷hTAC∷gfp unc-119(+)], pwIs170[P_vha-6∷gfp∷rab-7 unc-119(+)] (Chen et al. 2006), svIs23[num-1(+) unc-4(+)], svIs24[num-1(+) unc-4(+)], and svIs27[num-1A∷gfp unc-4(+)] (this study).

Isolation of num-1(bc365):

A PCR-based screen was used to identify worms harboring the num-1(bc365) deletion mutation. The protocol used was developed by G. Moulder and B. Barstead (Edgley et al. 2002). The primers used for the first round of amplification had the sequences 5′-GGA GAC TGT GTA CAG GAC AGT G-3′ and 5′-TCG CCT CAA TAC CAT AAC CAT A-3′. Those for the second round had the sequences 5′-GGC TCG GTG CAA TGT GAT GGA A-3′ and 5′-CTC CTG CCA CAT TAT GAG TCG A-3′. The bc365 mutation was outcrossed with wild type (N2) six times before further analysis. To characterize the molecular nature of the bc365 deletion, the DNA sequence of a PCR product derived from bc365 genomic DNA was determined. To characterize the nature of the num-1A transcript present in num-1(bc365) worms, total RNA was subjected to RT–PCR using num-1A-specific primers. Determination of the DNA sequence of the product revealed that the mutant transcript contained exon 1 sequence spliced directly to exon 6 in the wrong reading frame.

Endocytosis assays:

To investigate endocytosis at the baso-lateral membrane of the intestine, adult hermaphrodites were microinjected with 0.4 mm FM4-64 (Molecular Probes, Eugene, OR) or 0.1 mg/ml rhodamine–dextran (Sigma, St. Louis) dissolved in egg salts (118 mm NaCl, 48 mm KCl, 2 mm MgCl₂, 2 mm CaCl₂, 10 mm HEPES, pH 7.4) into the pseudocoelom. To investigate endocytosis at the apical membrane, animals were soaked in the same solutions.

Molecular characterization of num-1 transcripts:

To examine the splicing patterns of num-1A and num-1B transcripts, we determined the sequence of the cDNA clones yk565f1 and yk269a11, respectively. The 5′-ends of the num-1 transcripts were characterized by determining the sequence of RT–PCR products generated with 5′ primers with sequences corresponding to those of the C. elegans spliced-leader 1 or 2 sequences (SL1 or SL2) (Krause and Hirsh 1987). For the num-1A transcript, the 3′ primers had the sequences 5′-CGG TGT GAT TCC GAG TCC CAG CGG A-3′ (for the first round) and 5′-GCT CAT TGT AAA GTT GAG TGC CGA-3′ (for the nested PCR). To characterize the 5′-end of the num-1B transcript, 5′ RACE was performed on poly(A)⁺ RNA isolated from a transgenic strain of the genotype svIs23, which harbors multiple copies of the entire num-1 locus.

Electron microscopy:

Animals were prepared for electron microscopic analysis as described previously (Weimer 2006). Briefly, young adult hermaphrodites were frozen with a high pressure apparatus (Bal-Tec HPM010) in Escherichia coli. Animals were freeze substituted in 0.1% tannic acid and 0.5% glutaraldehyde in acetone for 4 days at −90° and in 2% osmium tetroxide in acetone for 7 hr at −90° and for 16 hr at −25°. After raising the temperature to 4°, samples were washed in pure acetone, infiltrated, and embedded in Araldite (at 60° for 48 hr). Sections (55-nm thickness) were collected on formvar carbon-coated copper grids (Pelanne Instruments) and counterstained in 2% uranyl acetate in water for 10 min, followed by 10 min in Reynold's lead citrate. Sections were imaged at 20,000–87,000 magnification on a Tecnai transmission electron microscope (FEI Compagny Eindhoven, The Netherlands) fitted with a Gatan (Pleasanton, CA) 300W digital camera.

Transcripts from the num-1 locus:

A cDNA derived from the C. elegans Numb locus, num-1 (previously called cka-1; Zhang et al. 2001a), has been shown to encode two proteins of 75 and 64 kDa, both of which show strong sequence similarity to Numb proteins from both mice and Drosophila (Zhang et al. 2001a). These two proteins result from the use of two distinct initiator methionines at positions 1 and 44 in the open reading frame of the longer protein (Zhang et al. 2001a) (Figure 1A). Both contain the PTB domain found in Numb proteins from other species, which has been shown to mediate protein–protein interactions (Figure 1P) (Uhlik et al. 2005). Computer analysis of the num-1 locus predicted that it could also give rise to a shorter transcript that encodes a protein lacking the PTB domain (http://ws170.wormbase.org/). We sequenced cDNA clone yk269a11 (kindly provided by Y. Kohara) and confirmed the existence of the 1.2-kb transcript, which we have termed num-1B (Figure 1A). Analysis of this transcript by RT–PCR showed that, like the 3.1-kb transcript characterized by Zhang et al. (2001a), the num-1B transcript is transpliced to SL1. The B transcript is predicted to encode a protein of 401 amino acids that is not significantly similar in sequence to any other proteins in the public databases.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Organization of the num-1 locus and expression of num-1A and num-1B. (A) The organization of the exons and introns in num-1 is shown above the scale bar. Boxes represent exons. The lines connecting them show the splicing patterns of the primary transcripts determined by characterization of cDNAs and RT–PCR products. The region spanning parts of exons 3 and 4 of num-1A that encodes the PTB domain is shaded. The regions deleted in num-1(bc365) and num-1(ok433) are indicated by lines below the scale bar. (B) Transgenes used to study num-1 expression and function. Each line is a schematic of part of a plasmid construction used in this study. The plasmid pVB63LN contains an 8.4-kb XhoI–SpeI fragment that spans the entire num-1 locus. The other fragments encode different NUM-1∷GFP fusion proteins. In each case, the gfp gene has been inserted immediately prior to the stop codon of num-1. (C–H) Photomicrographs of hermaphrodite worms harboring a num-1A∷gfp transgene viewed with either Nomarski differential interference contrast (DIC) (C) or fluorescence (D–J) optics. The worms were viewed from either the left or the right. C–F show medial optical sections; for G, I, and J, the optical section was a plane just below the baso-lateral membrane of the intestine. The images in C–H were obtained with the aid of a Leica DMRB compound microscope. Images in I and J were obtained with the aid of Leica confocal microscope. The worms in C–I were of the genotype unc-4(e120); svIs27 [num-1A∷gfp unc-4 (+)]. The svIs27 array harbors the plasmid pVB68LN, which lacks sequences necessary for the expression of num-1B. The worm in J was wild-type N2 control. The arrows in C and D indicate the basal (b) and apical (a) cell membranes of the intestine. The arrowhead in D indicates the position of an adherens junction separating adjacent intestinal cells. The arrow in E indicates the intestine primordium in the embryo. The arrows in F indicate the pharynx (p) and nerve ring (n), respectively. The arrows in G indicate two coelomocytes (c). Note also the punctate pattern within the intestine in this focal plane. The arrow in H points to the adherens junctions separating the apical and basal domains of the seam cells. (I) Confocal micrograph of the intestine of an svIs27 hermaphrodite showing the punctate staining of NUM-1A∷GFP close to the baso-lateral membrane of the intestine. (J) A confocal micrograph of a wild-type hermaphrodite taken with identical settings. Bars in C–H are 20 μm; in I and J, 4 μm. (K–N) Photomicrographs of L4 hermaphrodite worms harboring a num-1B∷gfp transgene viewed with either Nomarski DIC (K and M) or fluorescence (L and N) optics. “can” denotes the excretory canal. The uv1 cells are four neurosecretory cells in the uterus; two of the four are seen in this focal plane. (O) Diagram showing the arrangement of the intestine, which is shaded. At left is shown a medial saggital section viewed from the left. A, anterior; P, posterior; D, dorsal; V, ventral. The baso-lateral and apical membranes of the intestine are indicated. (Right) A transverse section viewed from the anterior. (P) Schematic of C. elegans, Drosophila, and human Numb proteins. All three contain both a PTB domain and at least one DPF motif. Human and Drosophila Numb proteins also harbor the NPF motif.

In addition to the PTB domain, several other motifs have been identified in Drosophila and mouse NUMB that mediate binding with other proteins. These include SH3-binding sites (PXXP), which are required for binding to phosphotyrosine residues in target proteins; aspartic acid–proline–phenylalanine (DPF) motifs, which in mammalian NUMB proteins have been shown to constitute binding sites for α-adaptin (Santolini et al. 2000); and asparagine–proline–phenylalanine (NPF) motifs. In humans, NPF motifs have been shown to mediate binding to a number of proteins including the EH-domain protein, Eps15 (Salcini et al. 1997). Both SH3 domains and a DPF motif are found in the C terminus of the predicted C. elegans NUM-1A and NUM-1B proteins (Figure 1P). However, neither A nor B forms contain an NPF motif.

NUM-1A is localized to baso-lateral domains within polarized epithelia:

To help determine in which cells the proteins encoded by the num-1 locus are expressed, we generated transgenic worm strains harboring num-1∷gfp fusion genes. To examine the expression of num-1A, we generated transgenic worms harboring the plasmid pVB68LN, which encodes full-length NUM-1A fused to GFP (Figure 1B). The plasmid lacks the B-specific exon and the promoter for num-1B (see below). Three independent transgenic lines were established in which pVB68LN was present on an extrachromosomal array. From one of these, svEx109, five lines were generated (svIs25–svIs29) in which the transgene was integrated into the genome. The expression pattern seen was the same in all eight lines. Strong expression of GFP was first detected in the embryo at the E¹⁶ stage in the intestine primordium (Figure 1E). In later stage embryos, expression was also seen in other epithelial tissues. In larval worms, and throughout the adult stage, expression was seen in the intestine, pharynx, hypodermis, seam cells, spermatheca, and uterus (Figure 1, C–H). During postembryonic development, several nonepithelial cells also expressed the transgene, including many neurons in the nerve ring and coelomocytes, scavenger cells situated in the pseudocoelom (Figure 1, F and G). Within the nervous system, expression was not uniform but higher in a subset of neurons (supplemental Figure 1). At the subcellular level, in most cells the protein accumulated in punctate structures within the cytoplasm (Figure 1, G and I). Some protein was present very close to the plasma membrane and may be associated with it. In polarized epithelia, expression of the protein was predominantly on the baso-lateral parts of the cell and was largely absent from the apical cell membrane or the apical cytoplasm. In the intestine, seam cells, ventral hypodermis, pharynx, and other polarized epithelial cells, the most apical boundary of expression coincided with the adherens junctions that separate baso-lateral and apical cell membranes. Localization of the GFP fusion protein close to the junctions in the intestine and seam cells is seen in Figure 1, D and H, respectively.

The num-1B transcript is generated from a tissue-specific promoter that is active in the excretory cell and in uv1 cells:

To examine the expression of NUM-1B, we generated two constructs encoding NUM-1B fused to the N terminus of GFP but lacking the ability to encode NUM-1A (Figure 1B). The first, pVB69LN, contains a stop codon in exon 3 immediately prior to the PTB domain. The second, pVB72LN, lacks the region 5′ of the start of the A transcript and all sequences 5′ of intron 5 (Figure 1B). GFP expression in worms harboring either of these two transgenes was identical but the pattern of expression was very different from that seen with the A-specific gfp transgenes. In adult worms, NUM-1B∷GFP was expressed solely in the excretory cell (Figure 1, K and L), a large cell that constitutes part of the osmoregulatory system. The fusion protein gave rise to punctate GFP fluorescence both in the cell body and along the entire lengths of the excretory canals (Figure 1, K and L). In L4 worms, in addition to the excretory cell, expression was also seen in the four uv1 cells (Figure 1, M and N), which at this stage of development are part of the epithelium separating the uterus from the vulva. Expression was seen in no other cells at any stage of development. A 700-bp deletion allele generated by the C. elegans Gene Knockout Consortium, ok433, removes most of the B-specific exon (Figure 1A). A num-1∷gfp fusion construct containing the deletion, pVB141ÅT (Figure 1B), gave a pattern of expression in transgenic worms identical to that of NUM-1A∷GFP and failed to direct strong expression in the excretory cell (supplemental Figure 2). Therefore, ok433 is likely to cause reduction or elimination of num-1B activity but may not affect num-1A function. Worms homozygous for ok433 did not display any obvious mutant defects: they grew at the normal rate and had normal numbers of progeny. We noted, however, that in worms with multiple copies of num-1B, the outgrowth of the excretory canals was frequently abnormal (Figure 2, G–J). For 3/3 transgenic lines analyzed, more than half of worms displayed the defect.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

Defects associated with elevated levels of num-1 transcripts. (A and B) Micrographs of wild-type N2 (A) and svIs23[num-1A(++)] (B) young adult hermaphrodites viewed with Nomarski DIC optics. Vacuoles (indicated by large and small arrows) are present in the intestine of num-1A(++) worms. (C and D) Micrographs of an adult svIs27[num-1A∷gfp] hermaphrodite worm viewed with Nomarski DIC (C) or fluorescence (D) optics. Two large vacuoles, indicated by arrows, are present in the intestine. GFP fluorescence is seen around the perimeter of both. (E and F) Micrographs of the intestines of rme-1 and rab-10 mutant hermaphrodites for comparison. (G and H) Micrographs of an svIs30[num-1B∷gfp] adult worm viewed with Nomarski DIC (G) or fluorescence (H) optics. The tip of a canal running posteriorly is indicated by an arrow. In wild-type worms, the canal runs almost the entire length of the animal. In the animal shown, the canal extends only to the mid-body region. (I and J) The same worm at higher magnification.

Overexpression of NUM-1A causes morphological defects in the intestine similar to those seen in rme-1 and rab-10 mutants:

We noted that worms harboring multiple copies of wild-type NUM-1A displayed a striking phenotype in the intestine characterized by the accumulation of large vacuoles that were visible even under the dissecting microscope (Figure 2, A and B, and Table 1). RNAi against num-1A suppressed the defect. Furthermore, 4/4 independent integrated transgenes gave rise to the defect, indicating that it was caused by the transgene sequences and not by mutations that had arisen during integration of the arrays. More than 90% of individual worms carrying any of the transgene arrays were affected. Consistent with the fact that NUM-1B∷GFP expression is not seen in the intestine, worms harboring multiple copies of num-1B did not display the intestinal vacuolar defect. In worms with increased copies of num-1A, the intestinal vacuoles first became visible during the L4 stage; their number and size increased with age. Forty-eight hours after the last molt, svIs23 animals (which have multiple copies of wild-type num-1A) had, on average, 15 large vacuoles. Most had between 1 and 20 vacuoles but approximately one-third had >20 (Table 1). Animals containing the vacuoles were otherwise healthy although some had a reduced capacity to lay eggs. The vacuolar phenotype was also seen in animals with multiple copies of the transgene encoding NUM-1A∷GFP. In such worms, in addition to the plasma membrane, the protein also localized to the surface of the large vacuoles (Figure 2, C and D).

The vacuoles seen in the intestines of worms with elevated num-1A gene dosage [num-1A(++)] were similar in appearance to those seen in rme-1 or rab-10 mutants (Grant et al. 2001; Chen et al. 2006) (Figure 2, E and F). Analysis of the intestines of num-1A(++) worms by transmission electron microscopy (TEM) also indicated that, like those in rme-1 mutants, the vacuoles abutted the baso-lateral cell membrane (Grant et al. 2001) (Figure 3C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Electron microscope analysis of worms with aberrant num-1 expression. Electron micrographs of transverse sections of the intestine of wild-type, num-1(bc365), NUM-1A-overexpressing svIs23[num-1A(++)] and rme-1(b1045) animals. Whereas num-1(bc365) mutants display a normal intestinal organelle distribution when compared to wild type, NUM-1 overexpression in svIs23[num-1A(++)] animals leads to aberrant accumulation of large vacuoles, indicated by arrows, similar to those observed in rme-1(b1045) mutants. The apical and baso-lateral membranes of the intestine are indicated by black and white dotted lines, respectively. Bars, 5 μm.

Intestinal vacuoles in worms with elevated levels of num-1A receive input from endocytosis at the baso-lateral membrane:

The vacuoles in rme-1 or rab-10 mutants receive input from the endocytosis pathway at the baso-lateral membrane (Grant et al. 2001; Chen et al. 2006). To determine whether the vacuoles present in num-1A(++) worms also resulted from defects in endocytosis, we injected a fluid-phase marker for endocytosis, rhodamine–dextran, into the pseudocoelom (body cavity) and monitored its uptake into the vacuoles. This marker has been shown to be taken up into the intestine from the pseudocoelom by endocytosis at the baso-lateral cell membrane (Grant et al. 2001), which directly abuts the pseudocoelom (Figure 10). Rhodamine–dextran injected into the pseudocoelom of adult svIs23 hermaphrodites was detected in the vacuoles within 30 min of injection (Figure 4, A and B). In worms in which a secreted form of GFP, ssGFP (Fares and Greenwald 2001a), was present in the body cavity, GFP accumulated in the vacuoles (Figure 4, G and H). Together, these observations indicate that the vacuoles receive input from the baso-lateral endocytosis pathway. The vacuoles do not, however, appear to receive input via endocytosis at the apical membrane. When rhodamine–dextran was delivered to the apical membrane by feeding worms the dye, it did not accumulate in the vacuoles even after extended periods (Figure 4, E and F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

The intestinal vacuoles in num-1A(++) worms receive input from the baso-lateral endocytosis pathway. (A and B) Micrographs of an adult svIs23[num-1(++)] hermaphrodite into the body cavity (pseudocoelom) of which rhodamine–dextran, a fluid-phase marker for endocytosis, has been injected. In B, the worm is viewed with fluorescence optics 30 min post-injection. The dye has accumulated in the vacuoles, indicating that the vacuoles receive cargo from the baso-lateral endocytosis pathway. (C and D) A wild-type N2 worm injected as a control. The dye fails to accumulate in the intestine. (E and F) Micrographs of an adult svIs23[num-1(++)] hermaphrodite fed rhodamine–dextran. In F, the worm is viewed with fluorescence optics. The dye accumulates in the lumen of the intestine but not in the vacuoles, indicating that the vacuoles do not receive input from the apical endocytosis pathway. The worm shown had been fed rhodamine–dextran for 30 min. The vacuoles remain unlabeled even when worms are fed for extended periods of time (10 hr). (G and H) Micrographs of an svIs23[num-1(++)] worm carrying a transgenic array, arIs37, encoding a form of GFP, which is secreted from muscles into the pseudocoelom (Fares and Greenwald 2001a). The GFP accumulates in the intestinal vacuoles, indicated by arrows. (I and J) arIs37 hermaphrodite worm for comparison. GFP does not accumulate in the intestine.

The vacuoles in num-1A(++) animals are labeled by markers for the ERC:

The vacuoles present in rme-1 null mutants have been shown to result from enlargement of organelles in the ERC (Grant et al. 2001). To determine whether the vacuoles present in the num-1A overexpressing worms might also be the result of a defect in endocytic recycling, we first injected FM4-64 (a membrane intercalating dye) into the pseudocoelomic fluid and monitored uptake into the vacuoles present in animals with multiple copies of num-1A. FM4-64 preferentially labels late endosomes and lysosomes in C. elegans but does not label the ERC (Grant et al. 2001). Consistent with the notion that the vacuoles present in the num-1A overexpressing strains result from enlargement of an organelle on the recycling branch of the endocytosis pathway, FM4-64 delivered either by injection or by feeding failed to accumulate in the membranes surrounding the vacuoles present in these worms (supplemental Figure 3).

To determine more precisely the nature of the vacuoles present in the worms with elevated levels of NUM-1A, we first examined whether they were labeled with markers specific for different parts of the endosomal system. GFP∷RAB-5 and GFP∷RAB-10 fusion proteins have been shown to be markers for early endosomes in C. elegans (Chen et al. 2006). GFP∷RAB-7 is enriched on both early and late endosomes but not the ERC, GFP∷RAB-11 labels recycling endosomes in the apical cytoplasm and Golgi, and GFP∷RME-1 predominantly labels baso-lateral recycling endosomes (Grant et al. 2001; Chen et al. 2006). We found that the vacuoles present in worms with elevated levels of NUM-1A were labeled with GFP∷RME-1 (Figure 5B) but none of the other markers (Figure 5, F and H).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.—

The membranes of intestinal vacuoles present in num-1(++) animals are labeled by markers for recycling endosomes. Confocal fluorescence micrographs of the intestines of adult hermaphrodite worms harboring GFP markers for endosomal organelles. num-1(++) indicates svIs23[num-1(++)]. The transgenes used were pwIs87[P_vha-6∷gfp∷rme-1], pwIs112[P_vha-6∷hTAC∷gfp], pwIs249[P_vha-6∷gfp∷rab-5], and pwIs253[P_vha-6∷gfp∷rab-7] (Chen et al. 2006). The fusion proteins are under control of an intestine-specific promoter from the vha-6 gene (Chen et al. 2006).

To determine whether cargo from the baso-lateral cell membrane could be transported to the enlarged vacuoles present in num-1A(++) worms, we examined the expression of a human interleukin 2 receptor α-chain (hTAC)∷GFP fusion protein. hTAC∷GFP is a marker for clathrin-independent endocytosis and rme-1-dependent recycling both in mammalian cells (Caplan et al. 2002) and in C. elegans (Chen et al. 2006). Whereas in wild-type worms, this marker is almost exclusively found on the plasma membrane (Chen et al. 2006), in num-1A(++) worms it was found both on the cell membrane and on the periphery of the vacuoles (Figure 5D). Similar results were obtained with a second marker for recycling, human transferrin receptor∷GFP fusion protein (Chen et al. 2006) (supplemental Figure 4). Thus, as they are in rme-1 mutants, transmembrane proteins destined for recycling are improperly transported in worms with elevated num-1A activity. These observations indicate that the vacuoles observed are grossly enlarged baso-lateral recycling endosomes.

Loss of num-1A function bypasses the requirement for rme-1 for endocytic recycling:

To investigate further the function of num-1A in C. elegans, we isolated a deletion allele, bc365 (materials and methods). bc365 removes 1.9 kb spanning exons 2–5 (Figure 1A). In worms carrying the deletion, exon 1 is spliced directly to exon 6 but the exon 6 sequences are out of frame (materials and methods). bc365 is therefore predicted to encode a severely truncated protein that lacks the entire PTB domain and sequences C-terminal to it. A bc365∷gfp fusion construct, pVB301LN (Figure 1B), gave rise to a pattern of expression in transgenic worms identical to that of NUM-1B. No expression was seen in tissues that express NUM-1A. Therefore, bc365 is likely to cause a severe reduction in num-1A activity and may be a null allele of num-1A. Worms homozygous for bc365 were viable, grew at the normal rate, and had normal numbers of progeny (280, n = 20).

Inspection of the intestines of bc365 homozygous mutants by light or electron microscopy did not reveal any obvious abnormalities (Figure 3B). Since increasing num-1A gene copy number causes a phenotype similar to that conferred by loss-of-function mutations in rme-1, we examined the effect of reducing num-1A activity on the rme-1 mutant phenotype. Both num-1(bc365) and num-1(RNAi) caused a dramatic suppression of the rme-1 vacuolar defect (Table 1). Whereas all rme-1(b1045) single mutants have at least one large vacuole (and more than one-third have >20), most rme-1(b1045) num-1(bc365) double mutants have none (Table 1). On average, rme-1(b1045) mutants have 14 large vacuoles in the intestine; rme-1(b1045) num-1(bc365) double mutants or rme-1(b1045) num-1A(RNAi) worms have <2.

Loss of num-1A function does not appear to affect uptake of material by endocytosis at the plasma membrane:

Suppression of the rme-1 phenotype by num-1 is consistent with the possibility that num-1 functions either downstream of or in parallel to rme-1 to regulate recycling. However, two other possible explanations for the suppression observed are, first, that loss of num-1 function reduces the rate of endocytosis at the plasma membrane or, second, that it causes an alteration in trafficking between the sorting endosome such that a greater proportion of vesicles leaving this compartment dock with the late endosome rather than with the ERC. To examine in more detail how NUM-1A affects trafficking in the endosomal system, we tested whether reducing or eliminating num-1A function could suppress phenotypes caused by loss of rab-10 activity. Similar to elevated levels of num-1A, or to null alleles of rme-1, loss of rab-10 function causes large vacuoles to accumulate in the intestine, resulting from a defect in baso-lateral endocytic recycling (Chen et al. 2006). However, in rab-10 mutants the vacuoles do not result from enlargement of the ERC but instead from an enlargement of sorting endosomes that are negative for RME-1 (Chen et al. 2006). A mutation in dyn-1, which affects an early step in the endocytosis pathway, can suppress the vacuole formation caused by defects in recycling (Grant et al. 2001). Therefore if num-1A mutations affect an early step in the endocytosis pathway in the intestine, loss of num-1A function should also suppress the rab-10 phenotype. If num-1A affects later steps, however, rab-10 should be epistatic to num-1A(bc365). Consistent with the latter possibility, num-1(bc365) did not suppress the vacuolar phenotype caused by the rab-10 null allele dx2 (Chen et al. 2006) or by rab-10(RNAi) (Table 1). Mutations in rme-6, which encodes a RAB GTPase-activating protein, affect a very early step in the endocytosis pathway (Sato et al. 2005). In rme-6 mutants, many 110-nm vesicles that are defective in fusion with early endosomes accumulate close to the plasma membrane. Analysis of num-1(bc365) mutants by TEM revealed, however, that 100-nm vesicles were not present (Figure 3B).

One possible explanation for the fact that reducing num-1A function suppresses the rme-1 phenotype but not the rab-10 phenotype (and that increasing num-1 function causes enlargement of the ERC) is that num-1A could promote the docking of vesicles, leaving the sorting endosomes such that they fuse more often with the ERC than with late endosomes or lysosomes. A second possible explanation is that NUM-1A could inhibit trafficking between the ERC and the plasma membrane. In the first case, a reduction in num-1A activity should lead to an enlargement of late endosomes or lysosomes, and an increase in num-1A activity should lead to a decrease. The expression of GFP∷RAB-7, a marker for later endosomes (Chen et al. 2006), or LMP-1∷GFP, a marker for late endosomes and lysosomes (Kostich et al. 2000), was not different from that in wild type either in num-1(bc365) (data not shown) or num-1A(++) worms (Figure 5, G and H), arguing against a role for num-1A in directing trafficking away from late endosomes or lysosomes. Furthermore, as described above, when FM4-64, which preferentially labels late endosomes and lysosomes in C. elegans (Grant et al. 2001), was injected into the pseudocoelom of num-1A(++) worms, it did not label the vacuoles (supplemental Figure 3).

The effect of num-1(bc365) on the localization of markers for recycling cargo was also consistent with an involvement for NUM-1A in recycling. In rme-1(b1045) single-mutant hermaphrodites, the hTAC∷GFP marker labels the periphery of the large vacuoles present in this strain (Chen et al. 2006) (Figure 6B). In rme-1(b1045) num-1(bc365) double mutants, however, the marker is not trapped intracellularly in this way but is found at higher levels at the plasma membrane (Figure 6D). In contrast, num-1(bc365) did not significantly alter the distribution of the hTAC∷GFP marker in rab-10(dx2) hermaphrodites (Figure 6, E and F). In this mutant background, hTAC∷GFP accumulates intracellularly even in the absence of num-1A activity. Thus num-1A is not required for hTAC∷GFP to be internalized at the plasma membrane. Analysis of the hTfR∷GFP marker showed that num-1A is also not required for internalization of this marker (supplemental Figure 4). Together these observations suggest that loss of num-1A function suppresses the intracellular accumulation of hTAC∷GFP in rme-1 mutants by causing an increase in hTAC∷GFP recycling.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.—

num-1(bc365) affects recycling but not internalization of a transmembrane endocytosis marker. (A–F) Confocal fluorescence micrographs of the intestines of adult hermaphrodites expressing the hTAC∷GFP marker for endocytic recycling. In rme-1 or rab-10 mutants, the marker is present on the rim of intestinal vacuoles, indicated by arrows in B and E. In the rme-1 num-1A double mutant, there are no vacuoles but some hTAC∷GFP is present intracellularly. In the rab-10; num-1A double mutant, vacuoles are still present, and the hTAC∷GFP marker is clearly visible around the rims (indicated by arrows in F) as well as in other parts of the cytoplasm. The mutations used were rab-10(dx2), rme-1(b1045), and num-1(bc365).

Effects of num-1A misexpression on other tissues:

The observation that NUM-1A is widely expressed in C. elegans suggests that NUM-1A function is not restricted to the intestine. As described above, NUM-1A protein is expressed in coelomocytes, apolar cells in the pseudocoelom that are known to take up material very efficiently by bulk endocytosis (Fares and Greenwald 2001a). When endocytic uptake at the plasma membrane is disrupted in coelomocytes, ssGFP secreted from muscles fails to be taken up and instead accumulates to high levels in the pseudocoelom (Fares and Greenwald 2001a). Arguing against a model in which NUM-1A inhibits endocytosis at the cell membrane of coelomocytes, ssGFP did not accumulate to abnormally high levels in the pseudocoelom of num-1A(++) worms but instead was found at high levels in vacuoles in the coelomocytes (Figure 7, A and B). One possibility, therefore, is that NUM-1A inhibits recycling in coelomocytes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.—

Aberrant num-1A expression causes defects in other tissues. (A and B) Confocal fluorescence micrographs showing vacuoles in coelomocytes containing ssGFP. See the legend to Figure 4 for an explanation of ssGFP. The arrows indicate organelles in the coelomocytes containing GFP. GFP accumulates to higher levels within coelomocytes in animals with increased num-1A gene dosage (B) than in animals with normal num-1 expression (A). Bars, 10 μm. (C) Micrograph of a part of the hypodermis viewed with Nomarski DIC optics. The animal is an adult hermaphrodite with increased num-1A gene dosage. The presence of large vacuoles is indicated by arrows. Bar, 20 μm. (D–H) Electron micrographs of transverse sections of adult hermaphrodite worms showing parts of hypodermal cells, the overlying cuticle, and alae. D, F, G, I, and J show parts of the seam cells and the cuticle above. E, H, and K show part of hyp7 and the cuticle above. The top black dotted lines indicate the outer limit of the cuticle. The lower black dotted lines indicate the apical membrane of hypodermal cells. The asterisks in D, F, and J indicate the alae that run along the left and right. The alae and cuticle in the num-1(bc365) mutant (F) are not different from wild type (D). In contrast, the alae in num-1(++) animals (G and I) are flattened compared to wild type (D). (H) Part of the cuticle above hyp7 in num-1(++) worms; the cuticle is much thinner than in wild type (E) and lacks the typical three-layered structure. num-1(++) denotes the integrated array, svIs23[num-1(++)]. Bars, 1 μm.

In the hypodermis, vacuoles were found in num-1A(++) worms similar to those seen in the intestine (Figure 7C). Furthermore, transmission electron microscopy of hypodermal cells indicated that num-1A(++) worms have cuticle defects that may result from abnormal vesicle trafficking. In wild-type animals, the cuticle surrounding most of the worm has a typical three-layered structure (Figure 7, D and E). Running along the left and right sides of adults are raised specializations of the cuticle known as alae. Whereas the cuticle and alae are normal in num-1(bc365) and rme-1(b1045) mutants (Figure 7, F, J, and K), cuticular structure is disrupted in animals with increased num-1A gene dosage. In particular, the alae are flattened and the layer below is aberrant (Figure 7, G and H). Outside of the area from which the alae arise, the cuticle can be reduced to an extremely thin single layer (Figure 7I). The observation that the cuticle defects in num-1A(++) worms are different from those in rme-1 mutants suggests that NUM-1A can function independently of RME-1 in the hypodermis. We have examined the possibility that NUM-1A can directly affect secretion in this tissue by examining the expression in num-1A(++) animals of a marker for secretory organelles, VPH-5∷GFP (Liegeois et al. 2006). However, no differences were seen compared to wild type (data not shown). The expression of VPS-27∷GFP, a marker for apical early endosomes (Roudier et al. 2005), was also not different from wild type. It is possible that the cuticle defects observed in num-1A(++) mutants result from indirect effects on baso-lateral endocytosis. Alternatively, NUM-1A could directly modulate apical secretion. Arguing against the latter possibility, however, is the observation that NUM-1A is localized baso-laterally and that the cuticle is normal in animals lacking num-1A (Figure 7F).