srf-3, a Mutant of Caenorhabditis elegans, Resistant to Bacterial Infection and to Biofilm Binding, Is Deficient in Glycoconjugates*

srf-3 is a mutant of C. elegans that is resistant to infection by Microbacterium nematophilum and to binding of the biofilm produced by Yersinia pseudotuberculosis and Yersinia pestis. Recently, SRF-3 was characterized as a nucleotide sugar transporter of the Golgi apparatus occurring exclusively in hypodermal seam cells, pharyngeal cells, and spermatheca. Based on the above observations, we hypothesized that srf-3 may have altered glyconjugates that may enable the mutant nematode to grow unaffected in the presence of the above pathogenic bacteria. Following analyses of N- and O-linked glycoconjugates of srf-3 and wild type nematodes using a combination of enzymatic degradation, permethylation, and mass spectrometry, we found in srf-3 a 65% reduction of acidic O-linked glycoconjugates containing glucuronic acid and galactose as well as a reduction of N-linked glycoconjugates containing galactose and fucose. These results are consistent with the specificity of SRF-3 for UDP-galactose and strongly suggest that the above glycoconjugates play an important role in allowing adhesion of M. nematophilum or Y. pseudotuberculosis biofilm to wild type C. elegans. Furthermore, because seam cells as well as pharyngeal cells secrete their glycoconjugates to the cuticle and surrounding surfaces, the results also demonstrate the critical role of these cells and their secreted glycoproteins in nematode-bacteria interactions and offer a mechanistic basis for strategies to block such recognition processes.

Caenorhabditis elegans is a genetically and developmentally well characterized organism that has been used as a model to study host-pathogen interactions. Many of these involve carbohydrate recognition, suggesting that chemical analyses of carbohydrate components involved in these processes may be important toward an understanding of the molecular basis for these interactions.
Host-pathogen interactions require bidirectional recognition factors, and often at least one of the components is a glycoconjugate (10). Examples are bacterial toxins such as aerolysin from Aeromonas hydrophila, cholera from Vibrio cholerae, hemolysin from Escherichia coli, and the crystal proteins from Bacillus thuringiensis (11). Infection of C. elegans by the last bacterium leads to destruction of its intestine. C. elegans mutants resistant to this infection have been isolated and are defective in the bre gene family, one of which is a homolog of Drosophila melanogaster egghead (egh) and encodes a GDP-Man:␤Glc-Cer ␤1,4-mannosyltransferase (bre-3); another encodes a UDP-GalNAc:␤1,4N-acetylgalactosaminyltransferase (bre-4); and a third is a homolog of Drosophila brainiac (brn) and encodes a UDP-GlcNAc:Man N-acetylglucosaminyltransferase (bre-5) (12,13). Identification of these BRE activities of C. elegans provides striking evidence for the importance of carbohydrates in host-pathogen interactions of this nematode. The BRE proteins appear to be required for sensitivity to crystal proteins and offer a useful model to study invertebrate-host toxin interactions.
Several of the above mentioned pathogens infect C. elegans by interacting with the cuticule and surrounding surfaces. Thus, M. nematophilum infects the posterior cuticule surface near the anal opening and adjacent rectal tissues, leading to reduced generation times and constipation (4). Y. pestis and Y. pseudotuberculosis form a biofilm around the head region of C. elegans, preventing feeding and leading to starvation. C. elegans srf-2, -5, and -3 mutants, which have altered lectin binding affinity to the cuticule surface (2,3) suggesting changes of oligosaccharides, are resistant to infection by M. nematophilum, Y. pestis, and Y. pseudotuberculosis. srf-3 has recently been cloned and shown to encode a nucleotide sugar transporter for UDP-Gal and UDP-GlcNAc (14). SRF-3: GFP fusion protein shows tissue-specific expression at the spermatheca, hypodermal seam cells, and g1 and g2 glandular cells, with the last two cell types known to play a role in the biosynthesis of the cuticle and surrounding tissues (15). Therefore, one might predict that in srf-3 a subset of glycans are altered, leading to the above described phenotypes.

EXPERIMENTAL PROCEDURES
Isolation of N-Glycans-The glycoprotein-rich fraction was isolated from 10 -15 g of C. elegans as previously described (16). Briefly, following treatment of proteins with L-1-tosylamido-2-phenylethyl chloro-methyl ketone-treated trypsin, N-glycans were released with 8000 units/ml PNGase F (New England Biolabs) overnight at pH 8.5 and 37°C. Free glycans were separated from the tryptic peptides by solvent precipitation (17) with 50% methanol at pH 5.5 followed by centrifugation at 3500 ϫ g. The solutions, containing free glycans and some peptides, were subjected to rotary evaporation, and the resulting precipitate was suspended in distilled water and applied to Sep Pak C-18 cartridges. Glycans were collected by elution with distilled water. Adsorbed peptides were eluted with isopropyl alcohol, combined with those that had precipitated in 50% methanol, the mixture was dried, and the precipitate was suspended in mild base, pH 10.0. The pH was then adjusted to 5.5, and the mixture was digested with PNGase A. The above isolation procedure was then repeated to separate glycans from peptides. Free glycans were quantitated using the phenol sulfuric assay for neutral hexose standardized with mannose (18) except for those released by PNGase A, which were estimated by total ion intensity of MALDI 1 -TOF MS spectra compared with standardized samples.
Isolation of O-Glycans-Subsequent to sequential PNGase F and A release of N-glycans, the above peptide/glycopeptide mixture was subjected to ␤ elimination using standard protocols (19), except that acidic glycans were eluted from AG-1X2 columns with 100 mM ammonium acetate following elution of neutral glycans from the same column with water. Solutions of the glycans were evaporated in a Savant speed vacuum apparatus prior to hydrolysis.
Monosaccharide Analysis-Oligosaccharides were applied to porous graphitized carbon cartridges, washed with 3 ml of distilled water, and eluted with 30% acetonitrile, 0.1% trifluoroacetic acid. Monosaccharide analysis was performed on a Dionex high pH anion exchange chromatography system equipped with a MA-1 column, and content was determined using a linear calibration curve for each monosaccharide as previously described (19). Permethylation of Oligosaccharides-Permethylation was done using a slight modification of the method of Ciucanu and Kerek (20) as previously described (16).
Bruker IV mass spectrometer in positive reflectron mode. Between 20 and 50 pmol of sample was applied to the MALDI target with an equal volume of 2,5-dihydroxybenzoic acid (20 mg/ml) in 20% acetonitrile in 10 mM sodium acetate. Between 150 and 200 shots from a nitrogen laser (337 nm) were summed. The laser pulse was 3 ns. Each analysis was performed in duplicate. The intensities of each molecular ion were averaged, and the S.E. was calculated.
Collision-induced dissociation (CID) fragmentation data were collected using an Applied Biosystems/MDS-Sciex QStar Pulsar quadrupole/orthogonal acceleration TOF mass spectrometer (QoTOF) with nanospray or MALDI (UV laser; nitrogen, 337 nm) sources (Applied Biosystems Inc., Framingham, MA). The MALDI matrix was 2,5-dihydroxybenzoic acid, and typically 50 -200 laser shots were summed for each spectrum. The laser power used was 30 -33 J. Nanospray data was collected using 1-m nanospray tips, pulled with a Sutter model P-87 micropipette puller. Ion source voltage was 1200 -1400 V. Nitrogen was used as the collision gas for MS/MS experiments. The range of operator-controlled collision voltages was 12-50 V for electrospray and 35-90 V for MALDI. Nomenclature is that of Domon and Costello (21).

RESULTS
Rationale for Workup Strategy-Based on our knowledge that C. elegans srf-3 mutants are deficient in a Golgi apparatus transporter for UDP-Gal and UDP-GlcNAc, we decided to assess the structures of N-and O-glycans released from protein extracts of whole mixed stages of srf-3 and wild type, parental N2 Bristol strain nematodes. Many of the C. elegans N-and O-glycan structures have been reported, thereby providing the basis for the comparisons of this study (16,(22)(23)(24)(25)(26)(27)(28). N-Glycans from C. elegans can be released first with PNGase F, yielding most known N-glycans except those containing ␣1,3Fuc linked to the reducing end GlcNAc; thereafter, PNGase A treatment releases most remaining N-glycans. Although a subset of the N-glycans containing internally linked Gal are released with both of the above enzymes, the expectation, based on previous studies, was that these species should be enriched in the oligosaccharide fraction released by PNGase A (24,25). Therefore, significant changes in Gal-or Fuc-containing N-glycans should be observable with this strategy, since previous studies with C. elegans suggest that Fuc addition to glycans often requires the previous addition of Gal (24,25). N-Glycans containing Golgi type GlcNAc or late additions may also be seen in the above N-glycan fractions.
Previous studies showed that the major C. elegans O-linked glycans contain ␤1,3Gal bound to core Ser/Thr-linked GalNAc residues (23). Gal may be substituted by ␤1,2GlcA or ␤1-4,6Glc. The core GalNAc may also be substituted by ␤1,6Glc or ␤1,6Gal. Neutral and acidic O-glycans were therefore analyzed based on the prediction that in srf-3 mutants, all of the above O-glycan species may be affected qualitatively and quantitatively.
N-and O-glycans were permethylated prior to MS analysis. This renders oligosaccharides chemically equivalent, with peak intensities generated in the mass spectrometer detector being similar over a broad mass range (29). Using this approach, we found molecular ion intensities to vary within 0.7% S.E. of total spectral peak intensity when analyses were performed in duplicate. This reproducibility was maintained in samples originating from different nematode preparations, as shown below.
Independently obtained duplicates of N-glycans released with PNGase F, followed by permethylation, were analyzed by MALDI-TOF MS. Fourteen oligosaccharides, shown in Fig. 1E and Table I Fig. 1E. These two species are consistent with Fuc 1-2 Gal 2 Man 3 GlcNAc 2 , previously reported in C. elegans (24,25), and their decrease is in agreement with the reduction in the molar Gal content described above. A slight increase of Hex 7-8 HexNAc 2 and complex Hex 3-4 HexNAc 3 glycans was seen in srf-3 compared with wild type (Fig. 1E). The peaks at m/z 1905.9 and 1701.8 are most likely those of a glucose-containing polyhexose contaminant ( Fig. 1; see below).
When the above N-glycan mixtures were subjected to ␣1,2mannosidase digestion, prior to permethylation and MALDI-TOF MS analyses, Hex 7-9 HexNAc 2 decreased, whereas Hex 5 HexNAc 2 increased, in both wild type and srf-3 (Fig. 2, A  and B), demonstrating that abundant molecular ions in both strains are high mannose-containing glycans ( Fig. 1 and Table  I). Glycans with m/z 1754.0 and 1929.0, described above, were resistant to the above enzymatic treatment and most likely do not contain terminally linked ␣1,2-linked Man ( Fig. 2A). The majority of Hex 6 HexNAc 2 , m/z 1784.0 (Fig. 1, B and D), of both strains is resistant to ␣1,2-mannosidase, demonstrating that it is not Man 6 GlcNAc 2 , which contains one terminal ␣1,2Man (Fig. 2, A and B). However, when the mixtures were subjected to jack bean ␣-mannosidase digestion, the Hex 6 HexNAc 2 signal intensity was greatly diminished in both strains. High mannose glycans are expected to be digested with ␣-mannosidase to Man 1-2 GlcNAc 2 . Since, after this treatment, peak intensity for Hex 3 HexNAc 2 remains, it is likely that a significant amount of this compound is derived from the Hex 6 HexNAc 2 after loss of three mannose residues (Fig. 3).
PNGase F glycans released from wild type strains were subjected to reductive amination with 2-aminobenzamide followed by separation on a C-18 column. The Hex 6 HexNAc 2 peak was collected, identified as the major species by MALDI-TOF MS (Fig. 4A, m/z 1539.6), further purified by porous graphite chromatography, and subjected to hydrolysis for monosaccharide analysis. In this case, GlcN, derived from GlcNAc, is underrepresented as a result of derivatization during reductive amination with 2-aminobenzamide of the reducing end. The Hex 6 HexNAc 2 sample yielded Gal:Man in a 1:3 molar ratio (Fig. 4B). A small peak for GlcN was also seen, as well as a major one for Glc, the latter being most likely a contaminant from the polyhexoses (m/z 1133.5 and 1295.5) in Figs. 1 and 4A.
We observed that some of these oligosaccharides are methylated in vivo, leading to their dispersion into a series of minor peaks that cause difficulties in their purification. This may explain our failure to detect these Gal-containing oligosaccharides in our original analysis of C. elegans N-glycans when using high field NMR (16), an inherently insensitive technique. The abundance of Hex 3 HexNAc 3-4 was slightly increased in srf-3 compared with wild type, demonstrating that the addition of GlcNAc was similar ( Fig. 1 and Table II). Previous studies from our own and other laboratories had identified the other PNGase F-released glycans shown in Fig. 1E  PNGase A-released Glycans-Following the release of glycans by PNGase F, release by PNGase A yielded ϳ5% of the total N-glycans. The values observed for the molecular ions of these permethylated glycans are listed in Table I. As seen in Fig. 6C, oligosaccharides reported to contain Fuc and Gal (deoxy-Hex 1-4 Hex 5 HexNAc 2-3 , indicated by asterisks) (24,25) are decreased in srf-3 when compared with wild type, whereas high mannose and complex oligosaccharides (Hex 7-9 HexNAc 2 and Hex 3-4 HexNAc 4 -5 , indicated with crosses) are increased. Previous studies had shown the latter glycans to be Man 3-4 GlcNAc 4 -5 and Man 7-9 GlcNAc 2 (16,22,24,25,30). This shift, which is consistent with the reduced availability of UDP-Gal in these srf-3 mutants, was also observed previously, but less dramatically, in the PNGase F-released fraction (Fig. 1E). The major peaks observed in the PNGase A fraction in both wild type and srf-3 correspond to residual high mannose glycans Man 7-9 GlcNAc 2 that were not completely released by PNGase F. Overall, in the above studies, the ion intensity is shifted toward GlcNAc-modified glycans; therefore, although SRF-3 transports UDP-GlcNAc, it is unlikely that the availability of this nucleotide sugar was reduced for the biosynthesis of PNGase A-releasable glycans in srf-3 mutants.
Acidic O-Glycan Fraction-Acidic O-glycans were eluted from an AG-1X2 column with 100 mM ammonium acetate following neutral O-glycan elution with water. The total amount of acidic O-glycans of srf-3 was ϳ3-fold less than wild type (106 versus 298 nmol/g) as determined by a phenol sulfuric assay. Most molecular ions detected in this fraction indicated the presence of HexA (Fig. 7 and Table II), consistent with the composition of GlcA-containing oligosaccharides previously re-ported for C. elegans (23).
Monosaccharide analyses were performed on fractions from both strains and were normalized to Gal. The molar ratio of Gal/Fuc/GlcN/Man/GlcA/Glc was 1:0.45:0.57:0.07:0.20:5.94 in srf-3 glycans and 1:0.11:0.14:0.04:0.50:11.60 in wild type. Increases in Fuc and GlcN and decreases in GlcA and Glc were seen in srf-3 compared with wild-type. MALDI-TOF MS analysis of the permethylated glycans revealed that in srf-3, HexA 1 Hex 4 HexNAc [M ϩ Na] ϩ (m/z 1364.68) was below detection limits, and that of HexA 1 Hex 3 HexNAc [M ϩ Na] ϩ (m/z 1160.69) was reduced compared with wild type (Fig. 7). MALDI QoTOF MS/MS fragmentation analysis of the [M ϩ Na] ϩ m/z 1160.69 isobar provided evidence of four branch isomers (Fig.  8A). Abundant Y ions were seen for all structures due to loss of the terminal Hex, or GlcA substitutions. The presence of Structure I was indicated by the observation of Y 1␣ , Z 1␣ , B 2␣ , C 2␣ , and Z 1␤ product ions at m/z 520.31, 502.30, 663.32, 681.33, and 924.5, respectively. These ions define the core region and provide strong evidence that two branches substitute the reduced HexNAc, one containing a GlcA and two Hex residues, whereas the other contains a single Hex residue. The larger branch contains a 3,6-disubstituted Hex as indicated by the presence of the 2,4 A 2␣Ј at m/z 329. 16 Fig. 8B. The presence of a trisubstituted branch is supported by the presence of the B 2 and C 2 ions at m/z 867.47 and 885.39, the absence of B-type and C-type fragments at m/z 738.38 and 720.34, and also by the presence of secondary fragments B 2␣ / Y 2x and B 2␣ /Z 2x seen at m/z 649.29 and 667.32 (where x represents ␣Љ or ␣ٞ). The identical compound has been reported in C. elegans (23).
Based on the above results and previous studies, the above glycans have the Gal␤1,3GalNAc type I core and probably have the branching patterns shown in Fig. 9. Guerardel et al. (23) have shown that the core GalNAc can also be substituted with ␤1,6Gal. We are not certain of the identity of the terminal Hex substitution in Hex2 in structures IIA and B but have represented it here as Glc. However, it is possible that Gal or Man substitutes. Thus, the reduced overall amount of acidic Oglycans and the loss of more highly processed GlcA-containing glycoforms are consistent with the decreased availability of UDP-Gal in srf-3.
Neutral O-Glycan Fraction-Neutral O-linked glycans, released from srf-3, were 21 nmol/g, whereas those of wild type were 30. Preliminary monosaccharide analysis of the neutral O-glycan fraction of both strains was also performed. The molar ratio of Fuc/Gal/Man/Glc was 1:2.29:0.39:10.1 in srf-3 strain versus 1:1.69:0.39:10.7 in wild-type glycans. The profiles were similar, although srf-3 glycans were increased in Gal likely due to a shift in Gal/Glc containing branch isomers. Virtually no GalN-ol was detected from the material released from the reducing end, most likely the result of its breakdown during the preparation of monosaccharides. Fig. 10, A and B, and Table II  show the molecular ions detected, whereas the histogram in Fig. 10C shows that virtually no differences in O-linked glycans were found between wild type and srf-3 except for a slight loss of Fuc-containing oligosaccharides.

DISCUSSION
In this study, we have found that C. elegans srf-3 mutants that are defective in a Golgi apparatus transporter for UDP-Gal and UDP-GlcNAc have major reductions in O-and Nlinked glycans compared with wild type nematodes. Acidic O-linked glycans containing GalNAc at their reducing end and GlcA at the nonreducing end were decreased by 65%. The oligosaccharide structures observed here were consistent with previous studies by Guerardel et al. (23) showing acidic Olinked species such as GlcA 1 Glc 2 Gal 1 GalNAc 1 containing ␤1,3Gal linked to the reducing end of GalNAc in these nematodes. N-Glycans containing Fuc 1-4 Gal 2 Man 3 GlcNAc 2 were also decreased compared with wild type nematodes.
Two important conclusions can be drawn from the above results: 1) the glycoconjugate species decreased in these mutants are most likely involved in the previously described interactions between the surface of wild type nematodes and the pathogenic bacteria M. nematophilum and the secreted biofilms of Y. pseudotuberculosis and Y. pestis; and 2) because all of the above glycans contain Gal, our findings are consistent with the substrate specificity of the SRF-3 transporter, resulting in a decrease of UDP-Gal supply to the Golgi apparatus lumen.
These results also strongly imply an important biological role for oligosaccharides of glycoproteins secreted to the surface coat and cuticle by seam cells and pharyngeal g1 and g2 cells, where the SRF-3 is localized. These oligosaccharides most likely play a critical role in recognition between nematodes and pathogenic bacteria. The srf-3 mutant used in this study, srf-3(yj10), has a G to A transition that creates a stop codon in exon 5 (of six) and is expected to terminate the protein at amino acid 234 instead of 328 as in the wild type protein, and the variant protein is thereby lacking the putative three most carboxyl terminal transmembrane domains. Is this protein a functional null? Whereas we cannot predict this with complete certainty, we did observe that, when the mutation is placed in the context of a genetic deficiency, srf-3(yj10/sDf22), the resulting phenotype was no more severe than the homozygous srf-3(yj10),  suggesting that the allele is a functional null (14).
Why was there no obvious decrease of the Golgi apparatustype addition of GlcNAc to the glycoconjugates, although SRF-3 also transports UDP-GlcNAc? Whereas we do not have a conclusive answer, two possibilities arise. The C. elegans genome contains 16 putative nucleotide sugar transporters; we know that at least two other transporters can use both UDP-Gal/ UDP-GlcNAc as substrates. 2 Whereas their tissue location has not yet been established, it is possible that one or both of them is also required for the supply of UDP-GlcNAc in seam cells and g1 and g2 pharyngeal cells or other organs that may contribute to glycoprotein secretion to the surface coat and cuticle. It is also possible that different UDP-Gal/UDP-GlcNAc transporters have different relative substrate affinities and that a transporter other than SRF-3 has high affinity for UDP-GlcNAc and therefore allows an adequate supply of this nucleotide sugar into the Golgi lumen for the GlcNAc addition to oligosaccharides destined to the surface coat and cuticle.
The discussion of the above described glycans of srf-3, in comparison with wild type, as well as the absence of infection of srf-3 mutants by Yersinia (3) and M. nematophilum (4) suggests a critical role for these glycans in the recognition between C. elegans and the above pathogens. However, it is also possible that proteoglycans and glycolipids play an important role in the interactions between the above bacteria and C. elegans.
The underlying molecular mechanism for the aberrant wheat germ agglutinin binding phenotype of the srf mutants is not known. Two models have been advanced (31): 1) wheat germ agglutinin binding targets that are normally masked in wild-type worms are unmasked in the srf phenotype; 2) the defects are due to altered glycoprotein processing. Both of these models are consistent with our results. Unmasking of wheat germ agglutinin epitopes may be the result of the observed drastic decrease of O-glycans and the slight increase of GlcNAc-2 P. Berninsone and C. B. Hirschberg, unpublished observations. containing complex N-glycans resulting from altered glycoconjugate processing. In another study, it was found that a wildtype L1-specific O-glycanase-sensitive O-glycoprotein detected by M38 monoclonal antibody is heterochronically expressed in L2-L4 stages of srf(yj43) but not expressed in srf-3; the latter is epistatic to the former, since in srf-3-srf(yj43) binding of M38 is eliminated (32). These observations are consistent with the role of SRF-3 UDP-Gal/UDP-GlcNAc transport and the observed decrease of O-glycans in srf-3.
The ␣1,2Fuc-containing nonreducing ends of the fucosylated N-glycans (24,25) that are decreased in srf-3 mutants are similar to those of the Lewis b and Lewis y antigens. These have been shown to be targets for bacterial invasion processes including those between Helicobacter pylori BabA-Lewis b and SabA-Sialyl Lewis X-mediated gastric mucosa adhesion (33,34) as well as P. aeruginosa infection in cystic fibrosis and in chronic bronchitis (34,35).
The types of oligosaccharides described in these studies also allow some speculations regarding biosynthetic pathways with caveats that will be outlined further below. The occurrence of Gal 1 Man 5 GlcNAc 2 in srf-3, as shown in Figs. 1, B and D, and 4A, suggests that it may represent an intermediate in the biosynthesis of Fuc 1-4 Gal2Man 3 GlcNAc 2 observed in Fig 1,  A-D, and Fig. 6, A-C. It may be that the latter structures cannot be synthesized due to an inadequate supply of UDP-Gal by SRF-3. This may prevent the subsequent action of galactosyltransferases and fucosyltransferases that have been previously described in C. elegans (36).
Man 3-5 GlcNAc 3-5 , observed in Fig. 1, was not decreased in srf-3. Thus, Golgi type GlcNAc addition must have occurred, suggesting, as previously mentioned, that wild type cells express another transporter capable of UDP-GlcNAc transport or that some glycans are derived from cells other than those having SRF-3. An intriguing observation is that N-acetylglucosaminyltransferase I orthologues gly-12 and gly-13 are expressed in the hypodermis, where SRF-3 is normally expressed (28,37,38). This glycosyltransferase catalyzes the first committed step in complex N-glycan biosynthesis by adding GlcNAc in a ␤1,2-linkage to the lower arm ␣1,3Man of Man 5 GlcNAc 2 to yield Man 5 GlcNAc 3 . This is the same site of the Gal addition proposed here to yield Gal 1 Man 5 GlcNAc 2 . Therefore, N-acetylglucosaminyltransferase I and the galactosyltransferase may compete for the same Man 5 GlcNAc 2 substrate; this in turn might explain the increase in Man 3-5 GlcNAc 3-5 seen in the srf-3 mutant as the availability of UDP-Gal is diminished.
Nevertheless, as mentioned above, because the multiple cell types secrete different glycans and the possibility exists that they may have several nucleotide sugar transporters for the same substrate as well as multiple glycosyltransferases of similar substrate specificities, it is not possible at this time to have a full understanding of the biosynthesis of the described glycan.
The acidic O-glycans detected in C. elegans are novel in structure; this, together with their drastic decrease in srf-3 and resistance of the mutant toward adhesion of biofilm produced by Yersinia and infection by M. nematophilum, suggests a possibly important role for the above glycans in nematodebacterial recognition and biofilm adherence. The most studied biofilms are those produced by Pseudomonas, and among its major components are alginates, linear polymers containing acetylated and nonacetylated guluronic acid and mannuronic acid, both sugars that are similar to glucuronic acid (39). Biofilm gelation requires calcium-coordinated intermolecular bonds between the hexuronic acid residues (39). Y. pestis contains homologs of P. aeruginosa alg-D, -A, and -C, genes required for the formation of alginates, as well as a protein with weak homology to alg-G, which encodes a C-5 polymannurane epimerase. It is therefore conceivable that calcium coordination or other interactions occur between alginates of Yersinia and the glucuronic acid-containing glycans of C. elegans. Interestingly, Pseudomonas sp. NCIMB 2021 adhesion to an alginate hydrogel withstands greater shear stress than that formed on agarose, chitosan, or PVA-SbQ (39). Whereas we do not yet know the structure and biological relevance of biofilms produced by Yersinia, nor whether they contain alginates, further studies on this subject should be of major help in understanding the mechanism of nematode-biofilm adhesion.
Acknowledgments-We thank the members of the Boston University School of Medicine Mass Spectrometry Resource. We give special thanks to Shui-Yung Chan for helpful advice for optimizing chemistries used in this study.