Glycosylation Genes Expressed in Seam Cells Determine Complex Surface Properties and Bacterial Adhesion to the Cuticle of Caenorhabditis elegans

Abstract

The surface of the nematode Caenorhabditis elegans is poorly understood but critical for its interactions with the environment and with pathogens. We show here that six genes (bus-2, bus-4, and bus-12, together with the previously cloned srf-3, bus-8, and bus-17) encode proteins predicted to act in surface glycosylation, thereby affecting disease susceptibility, locomotory competence, and sexual recognition. Mutations in all six genes cause resistance to the bacterial pathogen Microbacterium nematophilum, and most of these mutations also affect bacterial adhesion and biofilm formation by Yersinia species, demonstrating that both infection and biofilm formation depend on interaction with complex surface carbohydrates. A new bacterial interaction, involving locomotory inhibition by a strain of Bacillus pumilus, reveals diversity in the surface properties of these mutants. Another biological property—contact recognition of hermaphrodites by males during mating—was also found to be impaired in mutants of all six genes. An important common feature is that all are expressed most strongly in seam cells, rather than in the main hypodermal syncytium, indicating that seam cells play the major role in secreting surface coat and consequently in determining environmental interactions. To test for possible redundancies in gene action, the 15 double mutants for this set of genes were constructed and examined, but no synthetic phenotypes were observed. Comparison of the six genes shows that each has distinctive properties, suggesting that they do not act in a linear pathway.

ALL eukaryotic genomes include many genes encoding enzymes and transport proteins that act in the synthesis of complex carbohydrates and the glycosylation of proteins and lipids. Therefore, the glycome is evidently important, but ascribing biological function to these genes is often difficult in view of the many possible roles for glycosylation in cell signaling, development, structure, physiology, and defense. The scale of this problem is evident in the genome of the nematode Caenorhabditis elegans, which is one of the smallest and most intensively studied of animal genomes, but still contains at least 100 genes that encode proteins predicted to act in carbohydrate synthesis and/or modification, few of which have any firmly established biological function (for general review, see Berninsone 2006). In some cases, the predicted biochemical activities for nematode glycoslation genes have been demonstrated in vitro or by expression in heterologous systems, but such data are rarely very informative as to what processes they actually control or execute in vivo.

Specific biological functions have been ascribed to some of the carbohydrate-modifying enzymes in the worm, notably in various aspects of cell migration, but for the most part RNA interference (RNAi) knockdown or deletional knockout of the corresponding genes has relatively minor effects. Two sets of glycosylation genes that have better defined and more important roles in C. elegans biology are the sqv and bre genes. The eight sqv genes are all involved in chondroitin biosynthesis, and mutants in any of these genes exhibit a distinctive abnormal vulval morphogenesis phenotype (“squashed vulva”) together with maternal-effect lethality caused by defects in the fluid-filled space between the embryo and eggshell (Herman and Horvitz 1999; Hwang et al. 2003). The five bre genes are required for synthesis of glycosphingolipids, and four of them encode glycosyltransferases that act consecutively to build up carbohydrate chains (Griffitts et al. 2001, 2003). Mutants in these genes are viable but resistant to killing by certain Cry toxins produced by Bacillus thuringiensis because the glycosphingolipid on the luminal surface of intestinal cells acts as the receptor for these toxins. The bre genes also affect developmental signaling by the Notch/LIN-12 pathway (Katic et al. 2005).

Earlier work has shown that some of the glycosylation genes in C. elegans play important roles in determining the surface properties of C. elegans and also affect susceptibility to certain bacterial pathogens. The first of these to be studied, srf-3, encodes a UDP-sugar transporter that is required for normal surface antigenicity (Höflich et al. 2004). Mutants of srf-3 are resistant to infection by the coryneform pathogen Microbacterium nematophilum, which is able to cause disease by adherence to rectal cuticle. M. nematophilum elicits a characteristic swollen tail, or Dar (Deformed Anal Region) phenotype, in infected worms, and the srf-3 mutants exhibit a Bus (Bacterially Un-Swollen) phenotype (Hodgkin et al. 2000). Similarly, srf-3 mutants fail to form a bacterial biofilm on the head when grown on lawns of Yersinia spp. bacteria (the Bah, or Biofilm Absent on Head, phenotype) (Darby et al. 2002, 2007). The same two bacterial resistance phenotypes were observed in mutants of bus-17, which encodes a predicted galactosyltransferase (Yook and Hodgkin 2007). This gene is required for cuticle integrity because mutants exhibit hypersensitivity to agents such as alkaline hypochlorite (bleach) (Gravato-Nobre et al. 2005).

Subsequently, we found that the bus-8 gene, encoding another predicted glycosytransferase, plays two essential roles in the life of the worm: it is required during embryogenesis to enable ventral enclosure movements, and it is also required postembryonically at each molt cycle to permit successful shedding and replacement of cuticle, as well as to maintain the physical integrity of the cuticle (Partridge et al. 2008). Both viable and lethal alleles of bus-8 have been identified; weak mutants of bus-8 such as e2698 are viable and resistant to M. nematophilum (Bus phenotype), but they still permit formation of Yersinia biofilms, in contrast to other mutants of this type. Another essential carbohydrate-modifying gene required for cuticle integrity is glf-1, encoding UDP-galactopyranose mutase (Novelli et al. 2009). Available mutants in this gene are barely viable, making it difficult to assess interactions with M. nematophilum or Yersinia, but a glf-1 hypomorph (made by partial rescue of a lethal glf-1 mutant with a Leishmania-derived transgene) is resistant to M. nematophilum infection, indicating that this gene, like bus-8, is essential both for surface barrier formation and for bacterial surface infection.

Selections and screens for resistance to M. nematophilum have therefore provided a powerful means for identifying C. elegans mutants with altered surfaces to which these bacteria are unable to adhere. Many bus genes have been defined by mutation, but have not yet been studied at the molecular level (Gravato-Nobre et al. 2005). It seemed likely that additional bus genes would be involved in carbohydrate biochemistry, and indeed we report here the identification and analysis of three further bus genes with predicted roles in surface glycosylation (bus-2, bus-4, and bus-12). We compare their properties with the previously identified srf-3, bus-8, and bus-17 and demonstrate a diversity of phenotypic effects on bacterial susceptibility, cuticle integrity, locomotion, and male/hermaphrodite mate recognition. This contrasts with the largely uniform phenotypes observed for the sqv and bre gene sets.

MATERIALS AND METHODS

Nematode growth conditions and strains:

Standard procedures for C. elegans culture and genetics were employed (Brenner 1974; Sulston and Hodgkin 1988). Methods for infecting C. elegans with M. nematophilum were as previously described (Gravato-Nobre et al. 2005). Mutations examined were the following: for bus-2—e2676, e2677, e2687, e2705, e2776, e2780, e2781; for bus-4—e2693, e2700, e2752, e2788, e2803, br4; for bus-12—e2740, e2974, e2975, e2976, e2977, e2978, e2979, br5; for bus-8—e2698; for bus-17—e2800, e2693; and for srf-3—yj10, e2689. Other alleles utilized were the following: sqv-7(n2839), rrf-3(pk1426) (LGII); unc-119(ed3) (LGIII); fem-1(hc17), dpy-20(e1282), unc-26(e205) (LGIV); and xol-1(y9) (LGX). Deletion alleles obtained from the C. elegans Gene Knockout Consortium were the following: ZK896.9 (ok3050), K06H6.3(ok3012), F44C8.7(ok2206), T21B6.5(ok3220), and C50F4.14(ok2860). Strain RB1727, homozygous for F44C8.7(ok2206), was found to exhibit temperature-sensitive lethality, but out-crossing showed that this lethality was caused by an unlinked background mutation rather than by the ok2206 deletion.

Noncomplementation screen:

Additional alleles of bus-12 were sought by means of a noncomplementation screen: wild-type (N2) males were mutagenized with 0.47 mm ethyl methanesulfonate according to a standard protocol (Brenner 1974) and mated with females of the strain CB6662, genotype fem-1(hc17) bus-12(e2740) unc-26(e205); xol-1(y9). Mated females were placed in sets on mixed bacterial lawns (Escherichia coli OP50/M. nematophilum CBX102), and progeny were screened for rare Bus animals. These were picked and allowed to self-fertilize; new alleles were isolated as homozygotes by utilizing the flanking wild-type alleles of fem-1 and unc-26.

Double-mutant construction:

The following 15 double mutants, defective in two different predicted surface glycosylation genes, were constructed and their phenotypes were examined: bus-2(e2687) bus-4(br4) (CB6605), bus-2(e2687) srf-3(yj10) (CB6857), bus-2(e2687) bus-12(e2977) (CB6868), bus-2(e2687), bus-8(e2698) (CB6824), bus-2(e2687); bus-17(e2800) (CB6848), bus-4 (br4) srf-3(yj10) (CB6493), bus-4 (e2700) bus-12(e2977) (CB6863), bus-4(e2700); bus-8(e2698) (CB6859), bus-4 (e2700); bus-17(e2800) (CB6832), srf-3(yj10) bus-12(e2977) (CB6858), srf-3(yj10); bus-8(e2698) (CB6831), srf-3(yj10); bus-17(e2800) (CB6860), bus-12(e2977); bus-8(e2698) (CB6864), bus-12(e2977); bus-17(e2800) (CB6866), bus-8(e2698) bus-17(e2800) (CB6865), and sqv-7(n2839); bus-12(br5) (CB6656). In addition, a double mutant of bus-12 and the related gene sqv-7 was constructed: [CB6656, sqv-7(n2839); bus-12(br5)].

Simple crosses followed by segregation and progeny testing were used for constructing most of these double mutants. Doubles between closely linked genes were made by using linked visible markers. For example, the double mutant srf-3(yj10) bus-12(e2977) was constructed by first making two different double mutants, srf-3(yj10) unc-26(e205) and dpy-20(e1282) bus-12(e2977). Males of the latter strain were crossed with hermaphrodites of the former strain to yield phenotypically wild-type srf-3 + + unc-26/+ dpy-20bus-12 + males. These males were crossed with dpy-20unc-26 hermaphrodites, and rare non-Dpy non-Unc cross-progeny were picked, most of which proved to have the genotype dpy-20unc-26 / srf-3bus-12. Unmarked homozygous srf-3bus-12 hermaphrodites were obtained from the next generation. Crosses with appropriate test males confirmed the genotypes of this and the 15 other strains.

Reporter constructs:

The following reporter constructs and corresponding transgenic lines were constructed: for bus-2, CB6616 = bus-2(e2687); eEx621[bus-2p∷bus-2∷GFP(pEntry)] (3.77 kb upstream, 2.5-kb coding region, 23-bp 3′ UTR) and CB6846 = unc-119; bus-2(e2677); eEx677[bus-2∷mCherry; unc-119(+)]; primers K08D12.5_forward (CATCAACTGATAAGTTGTTGATATTGTTGTAAA) and K08D12.5_reverse (GGTGAAAGTAGGATGAGACAGCGGCAAAAAAATCCATCAAGATCGCCACC). The following are operon constructs that were examined as multicopy transgenic arrays, so the expression level of the fluorescent reporter is unlikely to be affected by any inefficiency in productive splicing of the bus-2 transcript (see results): for bus-4, CB6612 = bus-4(e2693); eEx619[bus-4p∷bus-4∷GFP; rol-6(dm)] and CB6817 = bus-4(e2700); eEx683[bus-4(+)∷mCherry; sur-5∷GFP] (2.99 kb upstream, 3.38-kb coding region); primers T22B11.2_forward (CCAATGCACCAAAACTCCCAACC) and T22B11.2_reverse (GGTGAAAGTAGGATGAGACAGCGGATTGTTTTCTGCCCACCCTGTCG); for bus-12, CB6630 = bus-12(br5); eEx628[bus-12(+); sur-5p∷GFP] (4.5-kb JC8.12 rescuing fragment); CB6655 = unc-119(ed3); eEx640 [bus-12p∷bus-12∷dsRedII; unc-119(+)] (1.3 kb upstream, 2.85 kb coding region, 160-bp 3′ UTR); primers JC8.12P_forward (GTGAAGCTCTGGGAAGAGGACT) and JC8.12P_reverse (TAGACCGACTAGAATCAAAATCGG).

bus-2 RNA isolation and cDNA synthesis:

Total RNA from mixed stage wild-type worms was isolated with TRIzol (Invitrogen) according to the manufacturer's instructions. Samples were purified by chloroform-isoamyl alcohol extraction and ethanol precipitation and subsequently treated with DNaseI to remove any residual DNA. cDNA synthesis was performed in the first step, using 1 μg of total RNA per 20-μL reaction and primed with oligo(dT) and SuperScript III (Invitrogen) according to the manufacturer's protocol. In the second step, PCR was performed using bus-2-specific primers and Phusion High-Fidelity DNA polymerase (NEB). The following oligos were used: bus-2 cDNA_forward (ATAGTTTCCCGACGGATTCCAGTT) and bus-2 cDNA_reverse (ATACGGATCCCACCACCTGCATAG). No evidence was obtained for a variant isoform recently predicted by WormBase, BUS-2B, which lacks six amino acids in exon 1 (aa 34–39).

Microscopy:

Living animals were mounted on 2% agar pads containing 4% (v/v) propylene phenoxytol or 1 mm sodium azide in M9 and examined under Nomarski differential interference contrast microscopy (Zeiss Axioplan2 microscope with Axiocam). To visualize bacterial colonization in the rectum, worms were stained with the live-cell nucleic acid stain SYTO 13 (Molecular Probes) as previously described (Hodgkin et al. 2000).

Hurdles assay:

Assay plates were prepared by painting 5.5-cm NGM plates with three stripes of B. pumilus CBX120 and one stripe of E. coli OP50, using an inoculating loop, as shown schematically in Figure 7B. Plates were incubated for 48–72 hr at 25° to create sticky bands of B. pumilus, each ∼4 × 50 mm in extent. Twenty young adult worms of each genotype to be tested were placed at one end of the plate, opposite to the E. coli stripe and outside the first B. pumilus stripe. After 10–20 hr at room temperature, plates were examined and the worms that had accumulated in the E. coli stripe by swimming through the B. pumilus “hurdles” were counted. Worms placed directly in the E. coli stripe tended to remain there, indicating that this is their preferred location.

Figure 7.—

Locomotory inhibition by B. pumilus. (A and B) Wild-type hermaphrodite (A) and bus-4 hermaphrodites (B) on B. pumilus lawns. The wild-type animal is trapped in a cleared patch of the lawn, whereas the bus-4 mutants can move well, leaving normal sinusoidal tracks. (C) Schematic of hurdles assay: worms attempt to swim across three bands of B. pumilus to reach a band of E. coli. (D) Numbers (maximum score 20) of animals accumulating in E. coli band after ∼15 hr. Mean ± SD are shown for six to nine experiments with each genotype. Significant differences from wild type are indicated (two-tailed t-test).

Mating contact assay:

For each test, a 25-μl drop of a stationary E. coli OP50 suspension was applied to the center of a 5.5-cm NGM plate and incubated overnight, creating a 10-mm circular spot of bacteria. Four egg-laying adult hermaphrodites (wild type or mutant) then were placed on the spot, followed by eight wild-type (N2) males picked at 1 day post L4 molt, a stage at which male mating efficiency is maximal (Hodgkin and Doniach 1997). After equilibration for 1–3 hr at 22°, contact between hermaphrodites and males was scored at 1-min intervals over two sessions of 25 min, separated by 2–4 hr. Hermaphrodites were scored positive if a male was sliding his tail over a female's body or had achieved vulval location. A score of 0–4 was noted at each time point, giving a maximum score of 200 (= 100% occupancy) for the total of 50 min assayed. Each “snapshot” measurement required <5 sec and allowed seven or eight plates to be scored in parallel, thereby avoiding extensive and tedious behavioral monitoring. All the hermaphrodites and males usually remained in the spot of bacteria or close to it for the duration of the assays. Tests in which any worms wandered away from the spot for long periods were omitted from analysis. Under these conditions, wild-type hermaphrodites experience mating contacts for less than half the time (average 44% occupancy), indicating that the number of males is far from saturating.

RESULTS

Molecular identification and characterization of bus-2:

Mutants in the bus-2 complementation group were frequently recovered after selection for resistance to M. nematophilum and found to exhibit a non-infectable phenotype, as previously reported (Gravato-Nobre et al. 2005). No bus-2 mutants were recovered in corresponding screens for Bah phenotype mutants, which fail to support biofilm formation by Yersinia spp. but tests on existing bus-2 mutants revealed that they had a Bah phenotype as well as a Bus phenotype (Darby et al. 2007). Genetic mapping together with analysis of transposon-induced mutations of bus-2 established that bus-2 corresponds to K08D12.5, encoding a predicted galactosyltransferase enzyme (Palaima et al. 2010). Sequencing of EMS-induced alleles revealed missense changes in all cases (Figure 1). Rescue experiments, in which the Bus (M. nematophilum resistant) phenotype was converted back to the wild type (sensitive to M. nematophilum) by transgenes expressing K08D12.5, provided further confirmation (Figure 2C).

Figure 1.—

Structure and mutations of bus-2. (A) Exon structure and localization of mutations, plotted against genomic coordinates of chromosome IV. The exact structure of allele e2781 is uncertain (exons 1 and 2 could not be amplified, while other exons appeared normal). (B) Sequence alignment of BUS-2A with the glcosyltransferase BRE-2A.1, which is the most similar to BUS-2 of the multiple isoforms encoded by bre-2. Sequence alterations in bus-2 mutants are indicated. A minor isoform, BUS-2B, which lacks the hexapeptide GIIFNN at the end of exon 1, has been reported by WormBase.

Figure 2.—

Expression patterns for bus-2. (A–C) Expression from the rescuing operon construct bus-2∷GFP in a bus-2(e2687) background, strain CB6616. (D and E) Expression from the rescuing operon construct bus-2∷mCherry in a bus-2(e2687) background and corresponding mCherry construct, strain CB6846. (A) Whole adult worm, showing seam-cell expression. (B) Detail of oviduct, showing strong spermathecal expression. (C) Detail of tail after exposure to M. nematophilum, illustrating tail-swelling (arrowed) to demonstrate functional rescue of bus-2 activity; seam-cell expression is visible (white arrowheads). (D) Adult posterior region, showing intestinal expression. (E) Adult head region, showing head muscle expression. Scale bars: ∼25 μm.

Scrutiny of the possible coding sequence for bus-2/K08D12.5 revealed an anomaly, as noted elsewhere (Palaima et al. 2010): the protein sequence based on the gene model in WormBase (referential data freeze WS200) can be improved in terms of alignment with other glycosyltransferases, including the Caenorhabditis briggsae ortholog of bus-2, by replacing the 19-aa WormBase exon 3 with a different 43-aa exon 3. The region concerned (exon 2 to exon 5) was amplified from a C. elegans cDNA preparation, and sequencing demonstrated the presence of the 43-aa exon 3 and its predicted junction with exon 4. However, correct removal of intron 3 must have entailed use of a GG 5′ splice donor in contrast to the normal GU or (rarely) GC 5′ splice donor (Farrer et al. 2002), which is why WormBase and other gene predictions failed to include this exon. The corresponding genomic region was repeatedly sequenced during molecular identification of the various mutant alleles, and the GG donor sequence was confirmed. Use of GG as a 5′ intron signal has not previously been commented on in C. elegans, which is surprising; moreover, the C. briggsae ortholog of bus-2 has a conventional GU at this site. However, there appears to be at least one other C. elegans gene, B0244.4, which is predicted to contain a 5′ GG donor in its third intron, and in fact it contains an identical junction sequence to that in bus-2 (TAGgggagtttt). Such a junction is nevertheless likely to result in very inefficient splicing, and the bus-2 mRNA abundance is substantially lower than for comparable genes (see supporting information, Figure S1). Unusual post-transcriptional events have also been inferred for the predicted glycosyltransferase gene bus-8, which appears to depend on translational frameshifting for proper expression (Partridge et al. 2008).

Probable biochemical properties of BUS-2 are reported at greater length elsewhere (Palaima et al. 2010). It is predicted to be a galactosyltransferase, and plausible orthologs can be found in the genomes of all Caenorhabditis species so far sequenced (see WormBase). Alignment with one of the galactosyltransferases encoded by bre-2 is shown in Figure 1B.

The seven mutant alleles of bus-2 all have very similar phenotypes, and two of the transposon alleles are simple insertions of Tc1 into exons 1 and 5, respectively, which are expected to result in a null phenotype. The second of these, e2776, has therefore been used in an extensive characterization of altered glycosylation patterns in bus-2 worms (Palaima et al. 2010). The EMS alleles are all missense changes in conserved regions of the predicted protein; the existing bus-2 reference allele, e2687, causes a Cys-to-Tyr change in one of these conserved regions and has been used for most of this work as a presumed null or severe hypomorph.

Expression patterns for bus-2:

Expression patterns for bus-2 have been briefly reported elsewhere (Palaima et al. 2010) and are described in more detail here. The expression pattern of bus-2 was initially examined using an operon construct for bus-2, with a wild-type bus-2 gene upstream of a separate GFP cistron. This construct rescued bus-2 function, as assayed by the restoration of susceptibility to M. nematophilum, with concomitant tail swelling in the presence of this pathogen (Figure 2C). Strong expression was observed in larval and adult seam cells and also in the hermaphrodite spermatheca (Figure 2, A–C). Expression in the intestine was difficult to score owing to tissue autofluorescence, which interferes with detection of GFP. Therefore, an equivalent mCherry reporter was constructed and transformed into bus-2 worms. Examination of these animals confirmed the seam-cell expression and revealed significant expression in the posterior intestine, as well as in head muscles (Figure 2, D and E). The intestinal expression is consistent with detectable alterations in intestinal lectin staining, which are described elsewhere (Palaima et al. 2010).

Molecular identification and characterization of bus-4:

Mutations of bus-4 were recovered in screens for both Bus and Bah mutants (Gravato-Nobre et al. 2005; Darby et al. 2007). The genetic map position for bus-4 was determined using SNP markers (data not shown), and a plausible candidate locus, T22B11.2, was identified within the probable interval for this gene. PCR amplification across T22B11.2 demonstrated the presence of transposon insertions in both mut-7-induced alleles of bus-4, and subsequent sequencing of EMS and ENU alleles showed that these also carried lesions in T22B11.2. Transgene rescue experiments provided further confirmation of the gene identity. All six mutant alleles have similar phenotypes and are likely to be severely hypomorphic or null for gene activity on the basis of the mutational lesions (Figure 3).

Figure 3.—

Structure and mutations of bus-4. (A) Exon structure and localization of mutations, plotted against genomic coordinates of chromosome IV. (B) Sequence alignment of BUS-4 with the glycosyltransferase encoded by C38H2.2. Sequence alterations in bus-4 mutants are indicated.

T22B11.2/bus-4 encodes a predicted galactosyltransferase, but in a different subfamily from bus-2. Also in contrast to bus-2, which lacks any paralogs, there are nine fairly closely related homologs of bus-4 in the C. elegans genome, all of which encode predicted galactosyltransferases. The most similar of these, C16D9.6 (BLAST score 2e-66 relative to bus-4), has a genomic location very close to that predicted for the functionally similar gene bus-6, but it does not correspond to this gene (D. O'Rourke and J. Hodgkin, unpublished results). The more distantly related gene C38H2.2 (BLAST score 2e-34 relative to bus-4) has been investigated as the apparent ortholog of human T-synthase, which is responsible for synthesis of the common core 1 O-glycan structure by catalyzing addition of galactose to GalNAc GalNAcα1-Ser/Thr. Appropriate galactosyltransferase activity was demonstrated by expressing C38H2.2 in insect cells by Ju et al. (2006). Alignment of BUS-4 and C38H2.2 is shown in Figure 3B. Given the sequence similarity among these genes, it seems likely that bus-4 also encodes a galactosyltransferase. However, another related gene in this group (sqv-5) encodes the nematode chondroitin synthase, acting to transfer both glucuronic acid and N-acetylgalactosamine to build up chondroitin sulfate chains (Hwang et al. 2003); therefore, other possible enzymatic activities cannot be excluded for BUS-4.

Expression patterns for bus-4:

The expression pattern for bus-4, as determined by GFP and red fluorescent protein (RFP) reporter fusions, was similar to that of bus-2. Major expression was observed in the seam cells throughout development (Figure 4, A–C). However, expression appears to start earlier than for bus-2 because strong fluorescence is seen in the seam cells of late embryos (Figure 4A) whereas embryonic expression was not seen for bus-2. The reporters used were rescuing operon constructs that were able to confer susceptibility to infection by M. nematophilum (Figure 4C). As with bus-2, significant intestinal expression could be detected using an RFP reporter (Figure 4D), although this could not be convincingly seen with the GFP reporter. Minor expression was seen in some other tissues such as the anterior pharynx (Figure 4F) but not in the spermatheca, in contrast to bus-2.

Figure 4.—

Expression patterns for bus-4. (A–C) Expression from the rescuing operon construct bus-4∷GFP in a bus-4 background, strain CB6612; (D and E) Expression from the rescuing operon construct bus-4∷mCherry in a bus-4 background, strain CB6617. (A) Late embryo, illustrating strong seam-cell expression. (B) L2 larva. (C) L4 larva infected with M. nematophilum, showing tail swelling (black arrow) and seam-cell expression. (D) Tail region with intestinal expression. (E) Head region with anterior pharyngeal expression. Scale bar: ∼25 μm.

Molecular identification and characterization of bus-12:

Single mutations of bus-12 were recovered in screens for both Bus and Bah mutants (Gravato-Nobre et al. 2005; Darby et al. 2007). Previous work had defined the map position of bus-12 on LGIV at ∼ +8.5 cM; further crosses demonstrated very tight linkage to the gene unc-26 at position +8.52 cM. Cosmid and fosmid rescue experiments using clones in this region, together with sequence analysis of candidate genes, indicated that bus-12 corresponds to the gene JC8.12 (Figure 5).

Figure 5.—

Structure and mutations of bus-12 and ttr-45. (A) Exon structure and localization of mutations, plotted against genomic coordinates of chromosome IV. (B) Sequence alignment of BUS-12 with SQV-7. Sequence alterations in bus-12 mutants are indicated with arrows.

The first allele identified for bus-12, e2740, has a perceptibly weaker Bus phenotype than that of allele br5, and significant rectal accumulation of the pathogen can be seen in some infected bus-12(e2740) worms (Figure 6A), but never in bus-12(br5) worms (Figure 6B). PCR experiments and sequencing indicated the e2740 allele, induced by mut-7, is a duplication of JC8.12, with two defective copies. One contains a Tc1 transposon insertion, and the other an insertion of four nucleotides in the fifth exon, and it seems possible that one or both copies may still supply some residual activity. In contrast, the allele br5, induced by ENU and isolated on the basis of a Bah phenotype (Darby et al. 2007), was found to be a deletion affecting both bus-12 and the gene immediately adjacent to its left, ttr-45. The null phenotype of bus-12 was therefore uncertain, and additional alleles of bus-12 were therefore sought by means of a noncomplementation screen (see materials and methods). This yielded six new alleles of the gene, all of which had a Bus phenotype similar to that of the deletion allele br5. Sequencing showed that all six were different molecular lesions affecting bus-12 alone (Figure 5A). Moreover, expression of transgenes expressing wild-type bus-12, but not ttr-45, were capable of rescuing susceptibility to M. nematophilum (Figure 6C). Therefore, the loss of ttr-45 activity does not contribute to the bacterial resistance phenotypes of br5, which are due only to loss of bus-12. Of the new alleles, five are nonsense mutations, and one of these (e2977, Gln123Amber) was used for most of the subsequent characterization of the gene.

Figure 6.—

Infection and expression patterns for bus-12. (A and B) bus-12 mutants exposed to M. nematophilum and SYTO 13 stained to reveal bacteria; black arrowhead marks anus. (A) Allele e2740: rectal colonization, slight tail swelling. (B) Allele br5: no rectal staining, no tail swelling. (C) bus-12(br5) rescued with bus-12(+); sur-5p∷GFP (strain CB6630) and exposed to M. nematophilum. Black arrow marks tail swelling. (D–F) bus-12(e2977) rescued with bus-12∷dsRedII, strain CB6655; white arrowheads in D mark seam cells. (D) Head region, showing seam-cell expression. (E) Tail region, showing weak intestinal expression. (F) Head region, showing expression in a few head cells, of uncertain identity. Scale bar: ∼25 μm.

On the basis of transcriptome analysis (WormBase), JC8.12/bus-12 encodes two predicted isoforms of 315 and 304 amino acids. The shorter isoform appears to be generated by SL1 transplicing to the beginning of the second coding exon in the gene (Figure 5A) and would lack 11 N-terminal amino acids; it is not clear whether both isoforms are functional. The encoded proteins are predicted to be nucleotide-sugar transporters. The closest homolog of bus-12 in the C. elegans genome is sqv-7, which has been shown to encode a protein capable of transporting UDP-glucuronic acid, UDP-N-acetylgalactosamine, and UDP-galactose, when expressed in yeast cells, and UDP-galactose when expressed in canine cells (Berninsone et al. 2001). Whether BUS-12 has similar substrates remains to be determined. Alignment between BUS-12 and SQV-7 is shown in Figure 5B. The sqv-7 gene is expressed in seam cells, like bus-12 (see below), and might therefore affect surface glycosylation, but sqv-7 mutants were found to exhibit normal sensitivity to M. nematophilum. Possible redundancy between bus-12 and sqv-7 was nevertheless explored by constructing a double mutant sqv-7(n2839); bus-12(br5). This strain did not exhibit any unexpected phenotypes, which argues against any redundancy in the function of the two genes.

Current gene models for C. elegans predict >15 different sugar transporters (Caffaro et al. 2007, 2008), of which three (srf-3, sqv-7, and bus-12) have now been associated with specific biological functions by means of mutations. Deletion alleles have been generated for five of the other predicted sugar transporter genes [ZK896.9(ok3050), K06H6.3(ok3012), F44C8.7(ok2206), T21B6.5(ok3220), and C50F4.14(ok2860)]. Strains homozygous for these alleles were examined for gross phenotype and resistance to M. nematophilum, but all appeared grossly normal. Seven other transporter genes (C53B4.6, F15B10.1, ZK370.7/ugtp-1, ZC250.3, C03H5.2, B0212.4, F54E7.1/pst-2) were tested by RNAi for altered sensitivity to M. nematophilum or other effects, using both rrf-3 and bus-12(br5) mutant backgrounds, but no abnormalities or bacterial resistances were observed. Caffaro et al. (2007) have reported gonadal abnormalities associated with RNAi knockdown of C03H5.2, but only in a srf-3 mutant background.

Expression patterns for bus-12:

The expression pattern for bus-12 was examined using the same methods as for bus-2 and bus-4. A bus-12p∷GFP reporter construct was expressed only weakly (data not shown), but a rescuing bus-12 operon construct with dsRedII (materials and methods) exhibited strong expression in seam cells (Figure 6D), along with some gut expression mainly in the posterior intestine (Figure 6E) and in a few unidentified cells in the head (Figure 6F). Expression in the embryo or in the spermatheca was not detected. The major part of the expression pattern for bus-12 therefore overlaps with that of bus-2 and bus-4.

Growth comparisons between glycosylation gene mutants:

Growth and fertility of hermaphrodites of reference mutants for bus-2, bus-4, and bus-12 were compared with wild-type and reference mutants of the previously defined glycosylation genes srf-3, bus-8, and bus-17. No conspicuous differences were observed, with the exception of markedly slower larval growth in the bus-17 mutant e2800. At 20°, generation time for most of the bus mutants was slightly longer than for wild type (3–5 hr difference) but ∼15 hr longer for bus-17(e2800) mutants. An independent allele of bus-17, e2693, exhibited a similar growth delay as did e2800/e2693 heterozygotes. Both mutants are notably smaller than wild type in mature adult body size. The viable bus-8 allele used in this comparison, e2698, does not exhibit major growth retardation, but stronger viable alleles of bus-8 exhibit severe growth retardation, small size, and some larval lethality, as previously described (Partridge et al. 2008).

All the mutants examined were both self-fertile (as hermaphrodites) and cross-fertile (as males). Strong gene expression in the hermaphrodite spermatheca was observed for bus-2 and for srf-3 (Höflich et al. 2004), but mutants of these genes failed to show any great deficit in self-fertility, suggesting that spermathecal function is normal.

Effects on motility: interaction with B. pumilus:

To make progress when crawling on a surface, nematodes need to exert a frictional force on that surface. For deformable, hydrophilic surfaces like those encountered on top of an agar gel, which is the usual place for examining C. elegans in a laboratory environment, it is not obvious how traction is exerted. As already noted in characterizing bus-8, bus-17, and bus-18 mutants (Yook and Hodgkin 2007; Gravato-Nobre and Hodgkin 2008; Partridge et al. 2008), several of the surface-abnormal mutants identified by Bus phenotype also exhibit a “Skiddy” (Skd) locomotory mutant phenotype, in which the worm is able to generate a normal sinusoidal wave pattern but its body exhibits extensive slippage on agar and makes little forward progress. The three genes examined in detail in this work have only minor effects on this component of motility; bus-12 and bus-4 mutants exhibit a slight Skd phenotype, while bus-2 mutants move almost normally under normal laboratory conditions.

A different aspect of motility, which is significant because it differentiates among the different surface glycosylation mutants, was discovered during a survey of interactions between C. elegans and plant-pathogenic pathovars of Pseudomonas (G. Preston and J. Hodgkin, unpublished results). One of the stocks examined was contaminated with a different, non-pseudomonad bacterial strain, CBX120, which had a striking effect on the motility of worms. Wild-type animals had great difficulty in moving through lawns of this contaminant and make progress only with a jerky or ratchet-like movement, as if trying to crawl through an extremely sticky medium. Sometimes worms eventually create small clearings in the lawn by eating the bacilli, but remain corralled within the clearing (Figure 7A). The aberrant locomotion has some similarities to the effect produced when wild-type nematodes are placed on lawns of Yersinia spp. (Darby et al. 2007), which requires a bacterial exopolysaccharide (Tan and Darby 2004). The contaminant was isolated and identified as a strain of the ubiquitous and important Gram-positive species B. pumilus, which has attracted detailed study as a common contaminant of spacecraft, among other reasons (Kempf et al. 2005). A sample of one of the well-studied and fully sequenced strains of B. pumilus, SAFR-032, was obtained and found to lack the nematode stickiness observed in CBX120. Despite its extremely inhibitory effect on worm movement, pure lawns of CBX120 are little different from pure lawns of SAFR-032 in their ability to support good worm growth, and neither of these bacterial strains has much effect on fecundity or rate of maturation, as compared to E. coli OP50 (data not shown).

Microscopic examination of worms growing on CBX120 reveals that these bacteria do not stick tightly to the nematode surface, in contrast to the strong adhesion of infective M. nematophilum cells to the rectal and peri-anal cuticle. The impaired movement caused by this bacterial strain therefore seems to be due to extracellular secretions rather than to the bacteria themselves, but it is still likely to depend on specific interactions with the nematode surface. The movement of various surface mutants in CBX120 lawns was examined and found to be significantly different from wild type in a number of cases. Several mutant types were able to move much better than wild-type worms, with little jerkiness and immobility. In contrast, others appeared even more impaired in movement than wild type. To assess this interaction objectively, a “hurdles” assay was developed, in which worms were tested for the ability to crawl through three bands of CBX120 bacteria to reach a lawn of E. coli OP50, which they appear to prefer as an environment and food source (assay shown schematically in Figure 7C and described in more detail in materials and methods). As indicated in Figure 7D, wild-type worms perform poorly in this assay, with many worms remaining stuck at the first hurdle. In contrast, bus-4 mutants are able to crawl efficiently across the hurdles.

Mutants of the six glycosylation genes discussed in this article were examined for movement on B. pumilus and found to exhibit a diversity of phenotypes: srf-3, bus-4, and bus-17, and to a lesser extent, bus-8(weak) mutants were able to move through CBX120 fairly efficiently, whereas bus-2 and bus-12 mutants were more impaired than wild type, with almost all worms remaining stuck at the first hurdle. For some of these genes, independent mutant alleles were tested and found to exhibit the same phenotypes as for the reference alleles (data not shown). The ability to move on CBX120 is not correlated with “skiddiness” during standard movement (Yook and Hodgkin 2007), nor with competence to support biofilm formation by Yersinia spp. (Darby et al. 2007). Moreover, mutants that appear very similar in other aspects of their phenotype, such as bus-2 and bus-4, are strikingly different in this assay. bus-4 mutants move well on CBX120, whereas bus-2 mutants are impaired. The bus-2; bus-4 double mutant resembles bus-4 alone, not bus-2, indicating that bus-2 mutants are not intrinsically inhibited in movement by CBX120.

These results reveal a further dimension to surface properties in C. elegans and demonstrate that some surface mutants can have a significant locomotory advantage over wild type in some environments. Sticky bacteria of many kinds are commonly seen as chance contaminants during the routine laboratory culture of C. elegans, so it seems very likely that colonies of such bacteria will encounter worms in their natural environments.

Effects on hermaphrodite surface recognition by males:

A different kind of general alteration in surface properties in these glycosylation mutants was suggested by occasional difficulties encountered in crossing Srf or Bus hermaphrodites with phenotypically wild-type males. Cross-progeny were sometimes fewer than expected, or entirely absent, after crosses carried out under conditions in which mating is normally close to 100% successful. A possible explanation for this shortfall was that the males were less able to recognize the surface of the mutant hermaphrodites during mating and therefore less persistent in mating behavior. Male mating behavior (reviewed by Barr and Garcia 2006) is a complex, multi-step process involving long-range pheromone-mediated detection of the presence of hermaphrodites, recognition of the hermaphrodite surface by direct contact between the hermaphrodite and the sensory rays of the male tail, tail-sliding and turning by the male, location of the hermaphrodite vulva by the male hook sensillum, and insertion of the male copulatory spicules, followed by ejaculation of sperm. The second of these steps, the recognition of the hermaphrodite surface, appears to involve some degree of specificity because C. elegans males will slide their tails over adults of other Caenorhabditis species, but they fail to slide over adults of other nematode genera and never remain in contact with them for long (Baird et al. 1992).

Time spent in contact between wild-type males and glycosylation-defective Bus/Srf hermaphrodites was therefore measured using a simple assay (see materials and methods for details). Adult hermaphrodites to be tested were maintained on a small spot of bacteria, together with an excess number of adult wild-type males, and examined over time for mating contacts between males and hermaphrodites. Results of six independent tests are summarized in Table 1 and Figure S2. In all these tests, males remained in or very close to the bacterial spot with the hermaphrodites, indicating that the males continued to detect the presence of the hermaphrodites by diffusible chemical cues. Male leaving assays (Lipton et al. 2004) confirmed this (data not shown). Under the conditions used, each wild-type hermaphrodite experienced one or more males sliding over its body for 40–50% of the time. In contrast, the various Bus/Srf hermaphrodites all experienced significantly reduced contact times, suggesting an impaired recognition by males. The defective recognition is distinct from a failure to detect the hermaphrodite vulva, the Lov (Location Of Vulva) phenotype, described by Barr and Sternberg (1999), because it involves a prior sensory step and a different motor response (a “keep sliding” as opposed to a “stop sliding” response). Different sense organs may also be involved, as the hook sensillum appears to be the major vulva detector whereas the sensory rays may act as the main detectors in initial surface recognition.

TABLE 1.

Mating contact assay

Genotype	% occupancy (mean ± SD)	% occupancy (range)	P-value relative to wild type
Wild type	44.33 ± 9.24	33–55	—
bus-2	13.50 ± 4.76	9–20	<0.001
bus-4	13.00 ± 6.87	4–21	<0.001
bus-8	25.83 ± 15.17	13–55	0.033
bus-12	13.00 ± 6.26	6–22	<0.001
bus-17	22.83 ± 11.54	6–39	0.006
srf-3	18.33 ± 5.75	10–28	<0.001

Occupancy indicates the percentage of time that an adult hermaphrodite spends in mating contact with a male under standard conditions (see materials and methods). Data summarize six independent sets of tests. The statistical significance of each difference from wild type was calculated by a two-tailed t-test.

None of the single mutants examined in these tests was completely unrecognizable by wild-type males, nor were any of the double-mutant combinations described below, because all eventually yielded cross-progeny when mated with males. This suggests that the detection of an appropriate hermaphrodite surface involves a complex set of chemical and/or physical cues that are recognized by the male tail sensilla, with significant but incomplete redundancy. Correct surface glycosylation evidently makes a substantial contribution to this contact recognition.

Tests for interaction between glycosylation genes:

Multiple alleles of the three glycosylation genes that form the focus of this article (bus-2, bus-4, and bus-12) have been identified, some of which are likely to be null. The results presented above indicate that these mutants are altered in surface properties, in a variety of ways, but in other respects their gross properties and development appear fairly normal. The same is true for the previously analyzed srf-3, encoding a UDP-Gal transporter. In contrast, it has been shown that bus-8 is an essential gene, required both for signaling during embryonic development and for successful molting in postembryonic life (Partridge et al. 2008). Also, the predicted galactosyltransferase gene bus-17 is required for cuticle integrity, because severe bus-17 mutants are viable, but exhibit significant cuticle fragility and drug sensitivity (Gravato-Nobre et al. 2005), as well as the small size and slower larval growth rate reported here.

The relative mildness of the defects observed in the bus-2, bus-4, and bus-12 mutants could be due to redundancy with other glycosylation genes. A precedent for such redundancy has been provided by Caffaro et al. (2007), who found that RNAi knockdown of C03H5.2 (encoding a nucleotide sugar transporter related to SRF-3) had no obvious effect in wild-type or RNAi hypersensitive strains, but resulted in gonad abnormality in a srf-3 mutant background.

We therefore tested for possible redundancy or other interaction in the action of the three genes bus-2, bus-4 and bus-12, both inter se and with the previously characterized bus-8, bus-17, and srf-3. This was done by constructing all 15 possible double mutants of the six genes under discussion. For the most part, probable null or extreme hypomorph alleles were used for these constructions (see materials and methods), with the exception of bus-8, for which null alleles are inviable. The standard weak allele of bus-8, e2698, was therefore used instead for these constructions.

All double-mutant combinations proved to be viable, and no cases of unexpected phenotypes or mutual suppression (as tested by resistance to M. nematophilum) were encountered, apart from one anomalous and probably artifactual interaction, which is described below. In addition, the double mutant bus-8(e2698) bus-17(e2800) is more fragile, slower growing, and less healthy than either of its parents, but it is still viable. Since the bus-8 allele used is non-null, this phenotypic enhancement is unremarkable.

The anomalous interaction was observed during the construction of double mutants of the two genes encoding the predicted transporters srf-3 and bus-12. The double mutant srf-3(e2689) bus-12(br5) exhibited a severe uncoordinated (Unc) and egg-laying defective (Egl) phenotype, which is entirely different from that of either single parental mutant. However, alternative allelic combinations of these two genes (e2689e2977, yj10br5, and yj10e2977) showed no sign of this severe behavioral phenotype. As noted above, the br5 allele is a deletion that removes both bus-12 and the adjacent gene ttr-45, so it is probable that the synthetic phenotype is due to an interaction between ttr-45 and a mutation linked to srf-3(e2689), rather than to any allele-specific interaction between srf-3 and bus-12. Very little is known about the function of the ttr gene family in C. elegans, with the exception of a recent report (Wang et al. 2010) demonstrating that ttr-52 functions in the recognition of apoptotic cell corpses.

DISCUSSION

Previous work has implicated three C. elegans glycosylation genes in determining properties of the nematode surface, presumably by controlling the biochemistry of the glycocalyx, which is the outermost layer of the cuticle (Höflich et al. 2004; Yook and Hodgkin 2007; Partridge et al. 2008). The results presented here identify a three more genes (bus-2 and bus-4, which encode predicted galactosyltransferases belonging to different families, and bus-12, which encodes a predicted UDP-sugar transporter protein) that have more subtle but still important roles in controlling the biological properties of the nematode surface. Characterization and cloning of all six genes were made easier by a shared feature, the fact that mutants of these genes are resistant to infection by the pathogen M. nematophilum, which induces a conspicuous and easily scored tail swelling (the Dar phenotype) in susceptible worms. In this work, we have further defined the mutant phenotypes of bus-2, bus-4, and bus-12 and examined their expression patterns, as well as compared the properties of mutants of all six genes and defined new components of their phenotypes.

Definition of the exact biochemical alterations in bus-2, bus-4, and bus-12 mutants, or in vitro examination of the enzymatic properties of the corresponding proteins, lies outside the scope of this article, not least because elucidation of the biochemistry of these mutants is unlikely to be easy. It is possible to examine substrate specificities for sugar transporters by expression in heterologous systems, as has been done for SRF-3, which was found to use UDP-galactose and UDP-N-acetylglucosamine as substrates (Höflich et al. 2004). A similar approach could be taken with BUS-12. However, such transporters probably contribute to multiple glycosylation pathways, and indeed, srf-3 mutants were found to have reductions in both O-linked and N-linked glycoconjugates (Cipollo et al. 2004). A notable feature of the C. elegans sugar transporters, previously pointed out by Caffaro et al. (2007), is that the number of genes encoding putative transporters (18) appears more than sufficient to transport the seven different sugars in its glycoconjugates, especially as some of these transporters can deal with multiple substrates. Yet the two similar transporters SQV-7 and BUS-12, expressed in overlapping tissues such as the seam cells, have entirely different functions, as demonstrated previously for sqv-7 mutants (Herman and Horvitz 1999) and here for bus-12 mutants, We also find no evidence for redundant functions between the two in sqv-7; bus-12 double mutants. It seems possible that these transporters, and other glycosylation factors, may act in separate subcellular components dedicated to the production of different extracellular carbohydrates.

Expression in heterologous systems could provide information about the possible enzymatic activities of BUS-2, BUS-4, BUS-8, and BUS-17, but it may be more immediately informative to examine glycomic changes in the mutant worms. Such investigations, using high-performance mass spectrometry, have been carried out in parallel work on mutants of bus-2 (Palaima et al. 2010) and of bus-4 (R. M. Mizanur, E. Jankowska, D. O'Rourke, D. Stroud, J. Hodgkin and J. F. Cipollo, unpublished results), revealing significant although complex alterations in glycans. Both of types of mutant were found to exhibit reductions, although not complete absence, of core-1 type O-glycans as well as reduced staining with ABA (Agaricus bisporus) lectin, especially in the tail region.

In contrast to the sqv and bre genes, which have been shown to affect chondroitin and glycosphingolipid biosynthesis, respectively, the substrate macromolecules for the bus glycosylation genes are not known. Enyzmatic investigations in vitro may shed light on this question. A different approach may be provided by analysis of other bus and srf genes with mutant phenotypes comparable to the genes described here, because some of them may encode proteins that are core substrates for the relevant carbohydrate modifications. Recent advances in whole-genome sequencing (Sarin et al. 2008) have allowed us to identify a number of previously uncloned bus and srf genes, which may provide clues to possible substrates (D. O'Rourke, D. Stroud, M. J. Gravato-nobre and J. Hodgkin, unpublished results).

An important feature of this analysis is that each of the six genes exhibits a distinctive set of properties, as summarized in Table 2. This general result stands in contrast to two other realms of C. elegans biology in which glycosylation genes have been shown to have significant roles: vulval morphogenesis (affected by sqv genes) and susceptibility to Bacillus thuringiensis Cry toxins (affected by bre genes). As described in the Introduction, the five sqv genes all have similar mutant phenotypes (Herman and Horvitz 1999); likewise, the bre genes have similar mutant phenotypes (Marroquin et al. 2000). This is not surprising, if each set encode enzymes in a linear pathway for the production of a complex carbohydrate end-product. The situation with the six genes discussed in this article is different: null phenotypes for these genes range from subtle differences from wild type, detectable only by pathogen resistance or mating tests (as with bus-2), to complete lethality (as with bus-8). Moreover, assays such as mobility on B. pumilus reveal opposite alterations in some cases (increased mobility for bus-4, decreased mobility for bus-12), and these do not correlate with other kinds of mobility alteration such as “skiddiness” (maximal in bus-8 and bus-17 mutants). Both bus-8 and bus-17 affect cuticle integrity, and mutants exhibit increased drug sensitivity, but there are no obvious changes in cuticle integrity or permeability for mutants of the other four genes. Further diversity is revealed by surface lectin and antibody staining (Politz et al. 1990; Link et al. 1992; Gravato-Nobre et al. 2005).

TABLE 2.

Comparison of surface glycosylation genes

Gene	WT	bus-2	bus-4	bus-8	bus-12	bus-17	srf-3
Predicted activity	NA	Glycosyltransferase	Glycosyltransferase	Glycosyltransferase	UDP-sugar transporter	Glycosyltransferase	UDP-sugar transporter
Function	NA	Surface	Surface	Surface, embryo, molting	Surface	Surface	Surface
Growth on E. coli	Normal	Normal	Normal	Slow	Normal	Slow	Normal
Growth on M. nematophilum	Dar	Bus	Bus	Bus	Bus	Bus	Bus
Growth on Yersinia	Non-Bah	Bah	Bah	Non-Bah	Bah	Bah	Bah
Movement on E. coli	Normal	Normal	Weak Skd	Weak Skd	Weak Skd	Skd	Normal
Movement on B. pumilus	Poor	Very poor	Very good	Good	Very poor	Very good	Very good
Hermaphrodite/male recognition	Good	Poor	Poor	Defective	Poor	Defective	Defective
Cuticle fragility	Normal	Slight	Slight	Severe	Normal	Very severe	Severe

Phenotypes are those shown by reference alleles of the respective genes, which are putative nulls or severe hypomorphs, with the exception of bus-8, for which data refer to the weak viable allele e2688. The last row summarizes data on cuticle fragility as assayed by bleach sensitivity, previously published in Gravato-Nobre et al. (2005). Dar, Deformed Anal Region (M. nematophilum sensitive); Bus, Bacterially Un-Swollen (M. nematophilum resistant), Bah (Biofilm Absent on Head); Skd, Skiddy. NA, not applicable.

An additional phenotype—defective recognition of hermaphrodite surfaces by males—was revealed in this work. Hermaphrodites of all six mutant types exhibited a significant difference from wild type in tactile attractiveness to mating males. This was measured by the time spent in direct contact between tested hermaphrodites and wild-type males under standard conditions. The males remain in the vicinity of mutant hermaphrodites, indicating that the hermaphrodites continue to emit pheromonal cues, but contact time between the sexes is reduced and consequently successful matings are rarer. We suggest that the mutants are deficient in surface features that males can detect only by direct contact, most probably by the sensory rays in the male tail. This detection might be achieved by short-range chemoreceptive “tasting” of some chemical feature on the hermaphrodite surface, or possibly by mechanoreceptive recognition of a physical surface property. Since none of the single or double mutants was completely unrecognizable by males, it seems likely that multiple cues are involved. Mate recognition is not essential for reproduction of a hermaphroditic species such as C. elegans, but is much more important for obligate male/female species such as most of the other species in this genus.

Tests for redundant function among these genes did not reveal any unexpected phenotypes in the double mutants examined. Redundancy is still possible, however. Indeed, the biochemical alterations in glycans so far detected by analysis of srf-3 mutants (Cipollo et al. 2004), bus-2 mutants (Palaima et al. 2010), and bus-4 mutants (R. M. Mizanur, E. Jankowska, D. O'Rourke, D. Stroud, J. Hodgkin and J. F. Cipollo, unpublished results) appear to be quantitative rather than qualitative, in contrast to the more readily interpretable changes seen for sqv and bre mutants. Biochemical phenotypes may also be affected by compensatory changes in mutants if different modification enzymes are able to compete for the same substrate, and indeed, the biochemical analyses indicate increases in some some complex glycan types along with reductions in others.

A common feature of the three genes primarily discussed in this work (bus-2, bus-4, and bus-12) is that they all exhibit strong expression in seam cells, at least as inferred from the expression patterns of operonic rescuing constructs. Strong seam-cell expression has also been observed for srf-3 (Höflich et al. 2004), bus-8 (Partridge et al. 2008), and bus-17 (K. J. Yook and C. Darby, unpublished observations), as well as for glf-1 (Novelli et al. 2009). These observations are in accord with the belief that the seam cells, rather than the hypodermis, are responsible for secreting the surface coat. The bulk of the cuticle consists of collagens secreted directly by the hypodermal cells, which envelop most of the worm, but the seam cells occupy only a small part of the worm's surface. Nevertheless, surface coat extends over the entire exterior of the worm, and the mutants described in this work affect both general surface properties and more local features, such as the rectal surface, that are anatomically remote from seam cells. Presumably, seam cells [which are known to be highly active in secretion during the molting process, as described by Singh and Sulston (1978)] produce precursor material that can spread in the space between old and new cuticles during molting, and thereby coat the whole surface, but exactly how this works, how the glycocalyx bonds to the underlying collagens and other cuticle proteins, and whether material can be added or turned over during the intermolt and adult stages of the life cycle, are all unresolved questions. Contributions from the intestinal secretions must also be considered in view of the significant gut-cell expression observed for bus-2, bus-4, and bus-12. On the other hand, the excretory system, pharyngeal gland cells, amphids, and phasmids do not seem to be involved, although they have been proposed as sources of surface coat material both in C. elegans and in other nematodes (Nelson et al. 1983; Bird et al. 1988; Jones and Baillie 1995).

This work, together with previous genetic and biochemical analyses, demonstrates that nematode surface glycosylation is complex in terms of biochemistry and biological function, both in the normal life of the worm (affecting locomotion, cuticle permeability, and mate recognition) and as a major factor in determining susceptibility to pathogens. The nematode cuticle is one of the most important defining features of the phylum Nematoda, which includes a very large number of important plant and animal pathogens. A surface coat is found on the surface of all nematodes, as far as is known, and is the largest interface between parasite and host in parasitic nematodes. Genomic data indicate that most or all of the bus and srf genes discussed in this article have orthologs or homologs in both animal and plant parasites, so the investigation of these surface-modification genes in C. elegans offers excellent opportunities for investigating a crucial interface both between an animal and its environment and between parasites and their hosts.