The Fine Structure of Caenorhabditis elegans N-Glycans* 210

We report the fine structure of a nearly contiguous series of N-glycans from the soil nematodeCaenorhabditis elegans. Five major classes are revealed including high mannose, mammalian-type complex, hybrid, fuco-pausimannosidic (five mannose residues or fewer substituted with fucose), and phosphocholine oligosaccharides. The high mannose, complex, and hybrid N-glycan series show a high degree of conservation with the mammalian biosynthetic pathways. The fuco-pausimannosidic glycans contain a novel terminal fucose substitution of mannose. The phosphocholine oligosaccharides are high mannose type and are multiply substituted with phosphocholine. Although phosphocholine oligosaccharides are known immunomodulators in human nematode and trematode infections, C. elegans is unique as a non-parasitic nematode containing phosphocholineN-glycans. Therefore, studies in C. elegansshould aid in the elucidation of the biosynthetic pathway(s) of this class of biomedically relevant compounds. Results presented here show that C. elegans has a functional orthologue for nearly every known enzyme found to be deficient in congenital disorders of glycosylation types I and II. This nematode is well characterized genetically and developmentally. Therefore, elucidation of itsN-glycome, as shown in this report, may place it among the useful systems used to investigate human disorders of glycoconjugate synthesis such as the congenital disorders of glycosylation syndromes.

Much of the current understanding of glycosylation processes in eukaryotes has been derived from yeast, mammalian cell lines, and organ tissues. Although these systems have been useful, they are not suited for the analysis of these processes in multicellular organisms as they relate to genetic, developmental, or environmentally interactive processes.
Caenorhabditis elegans is a developmentally well characterized model organism. Its genome has been sequenced, and the organism can be genetically manipulated. Several gene ablation services are currently available, and these should increase the rate of genetic and biochemical progress. The wealth of genetic information has allowed the cloning of some glycosyltransferases through reverse genetics (1,2). However, the lack of C. elegans N-glycan structural information has made substrate specificity and enzyme activity difficult to identify. Recently, the major O-linked glycan structures were reported (3). The observation of a novel ␤1,6Glc substitution facilitated the identification of the elusive activity of the previously cloned enzyme (4). Therefore, the elucidation of the C. elegans Nglycan structures is warranted. Once this system is in place, the effect that aberrant glycan biosynthesis has at the multicellular level can be more clearly understood. Such an effect has been shown by the study of the sqv mutants where metabolic error in chondroitin biosynthesis leads to malformation of the nematode vulva (5,6).
There is a growing number of human genetic disorders wherein an error in biosynthesis or metabolism leads to dysregulation of glycoconjugate synthesis and developmental impairment. This group of diseases has been classified under the general heading congenital disorders of glycosylation (CDG) 1 (7,8). Yeast has facilitated the discovery of several causative gene lesions associated with these human disorders. However, because yeasts are simple eukaryotes, information provided by yeast systems is only inferential when extrapolated to the multicellular organism. C. elegans may aid in such investigations because the effects of errors in glycosylation manifest themselves throughout cell lineages and resultant tissue formation.
Here we present an overview of the N-glycans produced by C. elegans. From mixed stage worms we have characterized a nearly contiguous series of N-glycans that suggest the major N-glycosylation pathways in this nematode. Together with the sequenced genome and mapped cell fates of the organism, this information will facilitate understanding the role of glycosylation in development.

Materials
All chemicals were ACS grade or better. Phosphocholine calcium salt was from Fluka (Fluka Chemie GmbH, Industriestrasse 25, CH-9471 Buchs SG, Switzerland). TEPC-15 antibody and anti-mouse horseradish peroxidase conjugate were from Sigma, and the chemiluminescence detection kit was from PerkinElmer Life Sciences. The PNGase F used in this study was the kind gift of Dr. Thomas Plummer.

Western and Carbohydrate Blotting
SDS-PAGE and blotting to polyvinylidene difluoride membranes was performed as described previously (9). Wheat germ agglutinin was purchased from EY (EY Laboratories, Inc., San Mateo, CA) and * This work was supported by National Institute of Health Grant GM 30365 (to C. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Chemical Procedures
Hydrofluoric Acid Hydrolysis-Oligosaccharides were evaporated to dryness and resuspended in 48% hydrofluoric acid (HF) for 48 h. The reaction was stopped by neutralization with saturated LiOH solution. LiF salt precipitate was removed by centrifugation, and the hydrolyzed product was evaporated in a Savant Speed evacuation apparatus.
Permethylation of Polysaccharides-Approximately 0.10 -3.0 nmol of each polysaccharide sample was dried in a Savant speed evacuation apparatus and permethylated using a modification of the Me 2 SO/NaOH method of Ciucanu and Kerek (14). Briefly, ϳ200 mg of NaOH was added per ml of Me 2 SO. Dried polysaccharides were resuspended in the NaOH/Me 2 SO solution (100 l) and allowed to stand at room temperature for 1 h. One-half volume of iodomethane (Sigma) was added, and the reaction was allowed to proceed for 1 h, at which time an additional volume of NaOH/Me 2 SO and one-half volume of iodomethane were added. The reactions were allowed to proceed for an additional hour and then were terminated by addition of 2 reaction volumes of dH 2 O. An equal volume of chloroform was added, and the permethylated sugars, which partition into the hydrophobic phase, were extracted repeatedly against dH 2 O. The chloroform was evaporated and sample reconstituted in 70% methanol.
For the permethylation of HF-treated glycans, the reaction was ended after the 1st h and the chloroform/dH 2 O wash procedure was performed. The sample was evaporated by speed evacuation and again permethylated for an additional hour followed by reaction termination and processing as above.

Monosaccharide Analysis
Monosaccharide composition of glycopeptide/peptide and released glycans was determined by GC/MS on a Finnigan TRACE GC 2000 series coupled to the GCQ plus electron ionization ion trap mass spectrometer. Monosaccharides generated by acid hydrolysis of glycans were trimethylsialylated and analyzed.

Mass Spectrometry
MALDI-TOF MS was performed on permethylated glycans using either a Finnogan Vision 2000 or Bruker Reflex IV instrument fitted with a nitrogen laser (337 nm, 3-ns pulse width) operated in positive ion linear and/or reflectron mode(s). To acquire branch information on select glycan species, jack bean ␣-mannosidase digestions were performed using a method published previously (15) followed by permethylation and analysis by MALDI-TOF MS. Targets were spotted in a layered fashion allowing each layer to dry before adding the next. First, 1 l of 6-aza-2-thiothymine (15 mg/ml in 70% methanol) was spotted. Next, an equal volume of 20 mM NaOAc was layered over the top, and finally 15-30 pmol of the permethylated oligosaccharide mixture in 70% methanol was layered. The TOF MS mass scales were calibrated externally with a mixture of malto-oligosaccharides in the size range of Glc 4 -10 purchased from Sigma. Permethylated authentic Man 9 GlcNAc 2 (Sigma) was also used as external standard. Average masses were reported for spectra collected in linear mode, and monoisotopic values for the 12 C peak were reported for those collected in reflectron mode. Mass accuracy was Ϯ 0.1%.  (Cambridge Isotopes) with 5 mM acetone added as internal standard. All spectra were collected on a Bruker Avance 500 MHz spectrometer. Spectral width for one-dimensional experiments was 3008 Hz collected with 8 K points digitization. Total scans were 1024. The two-dimensional DQF-COSY experiments were performed with spectral widths of 1608 Hz in both t 1 and t 2 dimensions. Digitization was 512 and 4096 points in t 1 and t 2 , respectively. Line broadening of 3-5 Hz/Hz was used in both dimensions, and a sine bell squared adiposation function was applied in t 2 . One-dimensional spectra were collected at 300 and 318 K and internally referenced to acetone at 2.225 and 2.217 ppm, respectively. DQF-COSY spectra were collected at 300 K and internally referenced to acetone at 2.225 ppm. For resonance intensity integrations, FIDs were collected for ϳ5 times t 1 relaxation time.

Preparation of C. elegans N-Glycans for Structural Analysis
By using wheat germ agglutinin for detection, we found that C. elegans N-glycans are releasable with PNGase F. Only a trace amount of N-glycan was detectable by 1 H NMR analysis of PNGase A released material performed subsequent to PN-Gase F treatment, suggesting that C. elegans mixed stage N-glycans contain little ␣1,3-linked core Fuc (data not shown).
TEPC-15 IgA detects phosphorylcholine (Pc). A TEPC-15 blot of the glycoprotein of the nematode showed a loss of antibody binding subsequent to PNGase F treatment suggesting that a portion of C. elegans N-glycans contain the Pc functional group.
The overall goal of this work was to provide an overview of N-glycosylation in C. elegans. PNGase F-released glycans were separated by size exclusion Bio-Gel P-4, and detection was by the phenol-sulfuric assay. The glycans were first pooled widely (Fig. 1, pools a-d) and analyzed by MALDI-TOF MS to determine major and minor component glycan compositions (Table  I). Narrow pooling (Fig. 1, pools A-E) for fine structural deter-mination of major pool isomers was accomplished (Table II,  . Little was known about N-linked oligosaccharide composition in C. elegans prior to this study. Seven g of oligosaccharide from the included volume of the Bio-Gel P-4 eluate was analyzed by GC/MS. Fuc, Man, and GlcNAc were identified on the basis of their GC retention times and electron ionization mass spectra. It is noteworthy that no sialic acid was seen, a result consistent with the previous report for C. elegans (16).

Structural Characterization
Characterization of C. elegans N-Glycans-A major concern in this work was the detectability of both abundant and rare glycans. As our glycan source was from mixed stage worms, it was expected that glycans derived from the lower mass larval stages would be under-represented. To address this issue properly, an aliquot of Bio-Gel P-4 separated glycans was pooled widely to include all fractions of the included volume. The glycans were then derivatized by permethylation.
We have employed an HF hydrolysis and permethylation strategy in order to aid in providing an overview of glycoconjugates in this study. Permethylation tends to impart similar chemical properties over a wide range of oligosaccharides, which translates into similar detector response with added sensitivity in MALDI-TOF MS analysis. HF treatment has been used in combination with permethylation to aid in characterizing oligosaccharides in complex mixtures, where some components contain Pc (17,18). As C. elegans protein extracts bind TEPC-15, we anticipated that some glycoconjugates would contain Pc and therefore used oligosaccharide permethylation, with and without prior hydrolysis with HF, to aid glycan detection. Table I summarizes the MALDI-TOF MS spectra of permethylated and HF-treated and permethylated glycans and their compositions assigned on the basis of observed m/z values (Fig.  1, pools a-d). The compositions indicate the presence of five classes of N-glycan compounds, which include high mannose, mammalian-type hybrid and complex, fuco-pausimannosidic, and those containing Pc. Permethylation alone produced ions representative of all classes. Complex and hybrid glycans were more numerous than high mannose and fuco-pausimannosidic. HF treatment prior to permethylation produced a change in relative abundance of detected glycan types, where high mannose glycans were most abundant followed by fuco-pausimannosidic and hybrid glycans, and complex glycans were absent. The appearance of an ion at m/z 1768.4 was in agreement with Hex 5 GlcNAc 2 Pi 2 , which is consistent with loss of two choline groups and substitution with methyl groups during the per-methylation procedure, a modification that has been noted (19).
At this point in our investigation the following three lines of evidence pointed toward the presence of Pc oligosaccharides: a Western blot result that was positive for TEPC-15 reactive N-glycans; a change in glycan class detection upon HF treatment, presumably resulting from a change in surface activity due to release of phosphate esters; and appearance of the m/z 1768.4 ion described above. Further investigation of the Pc as well as the other classes of N-glycans is described below.
Characterization of High Abundance N-Glycans-For a more detailed structural analysis of the major glycans present, the Bio-Gel P-4 eluate was pooled more narrowly as shown in Fig.  1. Permethylation Ϯ HF treatment was performed, and the glycans were analyzed by MALDI-TOF MS to estimate pool complexity and relative abundance of glycan species. The nar-  Table I. Pools A-E were investigated for fine structure determination. See text for details. b Data listed for glycans using generic monosacchoride identifiers were collected using MALDI-TOF linear MS, and those studied by 1 H NMR are reported using specific monosaccharide identifiers and were collected using MALDI-TOF reflectron MS.
c Glycans seen in narrowly pooled pools A-E, which were analyzed by 1 H NMR, are identified by A-E, respectively. d Several glycans not seen by MS analysis of Bio-Gel P-4 pools were later resolved by HPAEC ϩ MS. e MS data were collected using the underivatized compound. f Component may have been a contaminant.
rowly pooled glycans (Fig. 1, pools A-E) contained the major N-glycans from mixed stage C. elegans (see Table I and Supplemental Material Fig. 2). One-dimensional and two-dimensional 1 H DQF-COSY NMR spectroscopy of each pool revealed the glycosidic linkage and sequence of the pools A-E and confirmed the monosaccharide components determined previously by GC/MS. Anomeric proton chemical shifts and their relative proton intensities, as integrated from one-dimensional spectra, are shown in Table  II. The two-dimensional 1 H DQF-COSY experiment allows C1-H/C2-H correlations to be well resolved which, in combination with the library of these chemical shifts and associated coupling constant data, allows the linkage arrangement of oligosaccharides in complex mixtures to be assigned accurately (20 -31). The DQF-COSY C1-H/C2-H region of pools A-E is shown in Fig. 8.
The C1-H/C2-H correlation of glycan constituents are monosaccharide-and linkage-dependent. The results in Fig. 8 confirm the identities and linkage arrangement of the monosaccharide constituents of the pool. Apportionment of the anomeric resonance intensity of the one-dimensional NMR spectra gave each pool the constituent isomers and relative abundance (see Table II).
HPAEC was used to isolate each pool component, and the compositions of individual glycan species were verified by linear mode MALDI-TOF MS analysis (HPAEC ϩ MS) (Figs. 2-5 and Fig. 7), which agreed well with the sequence and abundance of each isomer as determined by NMR analysis (see Tables II and III and Scheme II). In Scheme I four representative oligosaccharide structures, containing all residue types detected, are presented for the classes of N-glycans found in the high resolution study of narrowly pooled C. elegans glycans. Anomeric chemical shifts, linkage type, and the residue numbering scheme are provided in Scheme I. The NMR results will be discussed here along with HPAEC and MALDI-TOF MS results as these data are highly complementary. The isomers that were determined in each pool are shown in Scheme II.
Detailed Analysis of pool A Oligosaccharides-The majority (96%) of the pool A components were high mannose glycans. 70% were Man 9 GlcNAc 2 and were identical to the unprocessed form found in the ER prior to mannosidase trimming. Another 26% were Man 8 GlcNAc 2 and differed from the previous glycoform only by the absence of the middle arm ␣1,2Man residue after, presumably, being trimmed by ER ␣1,2-mannosidase. Three trace components were detected by HPAEC ϩ MS in-  cluding two isomers of Fuc 2 Man 5 GlcNAc 2 , eluting at separate positions by HPAEC, and one Man 3 GlcNAc 3 isomer. The trace components were not in sufficient abundance to determine their fine structure. The pool isomers are shown in Scheme II. Key NMR proton resonances used in isomer assignments are underlined in Table II. Mass spectrometry, HPAEC ϩ MS, exoglycosidase data, and NMR analyses that resulted isomer assignments are discussed in detail below.
The 0.70 mol of resonance at 5.05 ppm exactly matches the 0.70 mol of resonance present at 5.40 ppm, indicating that 70% of pool isomers contain the ␣1,3-linked Man residue 7 and its 2-O-substituent Man 10. Thus, to the Man 6 GlcNAc 2 partial SCHEME I. The representative structures of C. elegans N-glycan released oligosaccharides are shown. A, the conserved Man 9 GlcNAc 2 ; B, high mannose-type; C, hybrid-type; and D, pausi-mannosidic. Residue identification number and anomeric chemical shift are shown next to each residue. structure defined above is assigned the resonance intensity from residues 7 and 10 to be added yielding archetypal Man 9 GlcNAc 2 as isomer A1 and is 70% of pool isomers, which closely agrees with the HPAEC ϩ MS analysis of the pool that gave a 19.7-min peak with a relative abundance of ϳ61% and m/z consistent with the Man 9 GlcNAc 2 composition (see Fig. 2). This leaves 0.26 mol of intensity at 5.09 ppm and 0.02 mol of resonance each at 5.05, 5.12, and 4.87 ppm and 0.29 mol at 5.09 ppm. To the remaining 0.26 mol of the Man 7 GlcNAc 2 partial structure is added 0.26 mol of 5.09 ppm yielding Man 8 GlcNAc 2 isomer A2 as 26% of pool glycans, in good agreement with mass (m/z 2192.4) and relative abundance (ϳ34%) detected for the HPAEC ϩ MS peak eluting at 15.1 min. This oligosaccharide is identical to ER-processed Man 8 GlcNAc 2 . On balance 0.02 mol each of 5.05, 5.12, and 4.87 ppm and 0.03 mol of 5.09 ppm resonance remains to be assigned.
As stated above, the HPAEC ϩ MS analysis of pool A showed the presence of trace amounts of Fuc 2 Man 5 GlcNAc 2 and Man 3 GlcNAc 4 ranging from ϳ1 to 2% of total pool isomers. In addition to the above remaining resonance intensity, a trace amount of NAc resonance was seen at 2.09 ppm close to those of core GlcNAc residues 1 and 2, consistent with ␤1,2GlcNAc 13 (see Table II). A slight foot at the downfield edge of the resonance at 4.87 ppm was seen, which may be from terminal ␣1,6-linked residue 4 and ␣1,6-Fuc (33). As discussed below in pools B and C, anomeric resonance was seen centered at 5.33 for ␣1,2Fuc ppm, and thus a trace amount of this Fuc linkage may be present in this pool. A trace amount of Fuc methyl proton resonance was seen at 1.20 ppm. To the Man␤1,4GlcNAc␤1,4GlcNAc␣/␤ partial structure the remaining resonance is assigned. However, the remaining resonance does not allow linkage assignment with completeness and certainty. Trace resonance at 5.12 ppm hints that the Man 3 GlcNAc 4 likely contains residue 13 and is assigned here as A3, which is supported by the trace amount of m/z 1661.4 ion seen in the reflectron MALDI-TOF MS analysis of the pool A isomers (see Table I and Schemes I and II). The low intensity DQF-COSY 4.63 C1-H/3.85 C2-H resonance shows the presence of trace ␣1,6Fuc, which causes a spatially dependent downfield shift of GlcNAc residue 2 (34). The Fuc residue was likely to be present in the Fuc 2 Man 5 GlcNAc 2 and is assigned here as A4, which is consistent with the compositions detected by MS (m/z 1931.2 and 1930.3) of pool HPAEC peaks seen at 4.6 and 7.4 min, respectively . As seen in Table III, the glycan isomer compositions and relative abundance derived by 1 H NMR closely match those detected by HPAEC ϩ MS.
Detailed Analysis of Pool B Oligosaccharides-Pool B contained high mannose, hybrid, and fuco-pausimannosidic gly-SCHEME II. cans. The high mannose isomer, Man 7 GlcNAc 2 , was most abundant. This glycan was apparently trimmed by mannosidases with activities similar to mammalian ER and Golgi ␣1,2-mannosidases. The hybrid isomer was Fuc 1 Man 3 GlcNAc 3 and contained core-bisecting ␤1,4GlcNAc and core-linked ␣1,6Fuc. The fuco-pausimannosidic isomers retained the ER Man 9 GlcNAc 2 central or lower arm, which was terminally substituted with Fuc. The pool isomers are shown in Scheme II. Key NMR proton resonances used in isomer assignments are underlined in Table II. Mass spectrometry data, HPAEC ϩ MS, exoglycosidase data, and NMR analysis that resulted in isomer assignments are discussed in detail below.
Reflectron MALDI-TOF MS of the Bio-Gel P4 pool B glycans was performed (Table I) HPAEC ϩ MS analysis of pool B isomers produced four resolved peaks (see Fig. 3), which eluted at 3.5, 4.3, 11.4, and 12.3 min. The first peak eluted at 3.5 min in the void volume, and MS analysis of the permethylated sample produced no well resolved spectra. However, the 4.3-min peak was collected to contain the tail of the previous peak and contained a low abundance ion at m/z 1592.0, which is consistent with Fuc 1 Man 3 GlcNAc 3 as seen in the MALDI-TOF MS analysis of pool B described above. A second major peak detected in the 4.3-min pool was seen at m/z 1755.0, which is consistent with a Fuc 1 Man 5 GlcNAc 2 composition. A third peak was seen at m/z 1510.1, which is consistent with Fuc 1 Man 5 GlcNAc 1 , which was likely produced by loss of GlcNAc 1 from Fuc 1 Man 5 GlcNAc 2 during the permethylation procedure and hints that the Fuc residue is not linked to the core-reducing end GlcNAc in the Fuc 1 Man 5 GlcNAc 2 isomer (m/z 1755.0). The 11.4-min pool produced an sodium-adducted ion at m/z 1988.3, which is consistent with Man 7 GlcNAc 2 . Upon MALDI-TOF MS analysis the peak at 12.3 min produced a sodiated ion at m/z 1755.8, also consistent with Fuc 1 Man 5 GlcNAc 2 , indicating that two isomers with this composition exist in the pool B. Therefore, by HPAEC ϩ MS pool B contains Fuc 1 Man 3 GlcNAc 3 (13%), Man 7 GlcNAc 2 (56%), and two isomers of Fuc 1 Man 5 GlcNAc 2 (23 and 8%) as the four major components. . At 318 K residue 3 was present at 1.00 mol of integrated resonance intensity at 4.76 ppm but was obscured at 300 K by the HDO peak (see Table II). This gives a total integrated anomeric intensity of 8.33 mol. Residues 1-3 are present in all structures and will not be considered in the allocations below. In the DQF-COSY spectra of the pool, a strong cross-peak was seen at 5.34 C1-H/4.10C2-H and was from 2-O-substituted residue 5. A unique cross-peak was seen at 5.33 C1H/3.86 C2-H ppm, and the C2-H/C3-H cross-peak of the scalar coupled system was seen at 3.86/4.00 ppm. The measured coupling constants for the system were J 1,2 , 3.5 Hz and J 2,3 , 9.9 Hz, which are consistent with ␣-L-Fuc (35). The presence of 2-O-substituted ␣1,6-linked upper arm residue 6 assignment is confirmed by the presence of a strong DQF-COSY cross-peak at 5.14 C1H/4.02C2-H ppm. The cross-peak at 5.05 C1-H/4.07 C2-H ppm confirms the presence of terminal ␣1,2Man residues 8 -10. A low cut close to base line in the DQF-COSY spectrum of the pool revealed a cross-peak at 4.89 C1-H/3.78 C2-H ppm, coinciding with the 0.20 mol resonance at 4.89 ppm in the one-dimensional spectrum (boxed in Fig. 8). This is diagnostic of core-linked ␣1,6-Fuc (33). The cross-peak at 4.87 C1-H/4.13 C2-H ppm confirms the presence of 3,(Ϯ6)di-O-substituted Man residue 4 (25). The presence of core bisecting GlcNAc residue 14 is confirmed by DQF-COSY crosspeak 4.45 C1-H/3.65 C2-H ppm (36).
Residues 7 and 8 are present at only 0.15 and 0.12 mol in 2-O-substitued forms, as indicated by their relative resonance intensities at 5.41 and 5.31 ppm, respectively. Therefore, the major pool isomer, Man 7 GlcNAc 2 , detected in the Bio-Gel P4 pool by MS and subsequently by pool HPAEC ϩ MS analysis (ϳ56% peak area at 11.4 min) must contain 2-O-substituted residues 6 (at 5.14 ppm substituted by residue 9 at 5.05 ppm) and 5 (at 5.34 ppm substituted by terminal residue 8 at 5.05 ppm) in order to reach the requisite size. The only Fuc 1 Man 5 GlcNAc 2 that can contain 6 or 9 must be devoid of residue 5 in order to contain the required number of Man residues. The absence of 5 causes the C2-H/C3-H resonance of 3 to shift to 4.09/3.66 ppm (37). The moderate J 2,3 (ϳ3.5) and strong J 3,4 (ϳ10 Hz) coupling constants of mannose allow detection of low abundance resonance in the C2-H/C3-H region of the two-dimensional NMR spectrum. Because no resonance was seen at 4.09 C2-H/3.66 C3-H ppm, all pool glycans contain residue 5. Thus, due to size constraints the Fuc 1 Man 5 GlcNAc 2 components in the pool cannot contain 2-O-substituted 6 or its substituent residue 9. Therefore, all of the 0.53 mol of resonance at 5.14 ppm from 2-O-substituted 6 exists in the Man 7 GlcNAc 2 pool B isomer. The 56% pool abundance of the 11.4 min HPAEC ϩ MS detected Man 7 GlcNAc 2 peak supports the size and 53% relative abundance of this NMR assignment. Hence, the major pool isomer is Man␣1,2Man␣1,3(Man-␣1,2Man␣1,6(Man␣1,3)Man␣1,6)Man␤1,4-GlcNAc␤1,4Glc-NAc␣/␤ assigned here as isomer B1 (see Scheme II and Table  II). This assignment consumes 0.53 mol of intensity at 5.34, 5.14, 5.09, and 4.87 ppm for residues 5-7 and 4 and 1.06 mol at 5.05 ppm for residues 8 and 9. On balance this leaves the following intensities to assign: 0.12 mol at 5. We cannot say with absolute certainty where the ␣1,2Fuc is in B␣ and B␤ from the NMR data presented here. However, both contain complete arms of the ER Man 9 GlcNAc 2 with intact ␣1,2Man residues, and this observation led us to hypothesize the existence of a capping substitution of terminal ␣1,2Man by ␣1,2Fuc.
To investigate further the position of Fuc substitution, we performed jack bean ␣-mannosidase digestion of the Bio-Gel P-4 pool B glycans, and the products were permethylated and analyzed by linear mode MALDI-TOF MS. Sodiated molecular ions were seen at m/z 765.0, 1550.3, and 1754.2 (data not shown). The m/z 765.0 ion corresponds to Man 1 GlcNAc 2 from complete digestion of pool high mannose glycans. The detection of ions at m/z 1550.3 and 1754.4 is consistent with Fuc 1 Man 4 GlcNAc 2 and Fuc 1 Man 5 GlcNAc 2 that are ␣1,2Fuc substituted at ␣1,2Man as hypothesized. The assignment of 0.15 mol of ␣1,2Fuc as a substituent of residue 10 of partial structure B␤ gives isomer B3 and is 15% of the pool. The assignment of 0.12 mol of ␣1,2Fuc as substituting residue 11 gives isomer B4 and is 12% of pool glycans. These assignments complete the anomeric proton allocations of the pool. B4 has a terminal ␣1,6Man, which is accessible to jack bean ␣-mannosidase. Isomer B3 has terminal ␣1,3Man, which is slowly accessible to the enzyme. This situation would produce species that, upon permethylation, would exhibit [M ϩ Na] at both m/z 1550.3 and 1754.4, just as observed. We should note that the predicted products of pool glycans with ␤1,4GlcNAc and ␣1,6Fuc core substituents were absent from the MALDI-TOF MS spectra of the digestion product, but their presence in the pool was detected by composition and linkage signatures in the pool B MALDI-TOF MS and NMR spectra, respectively. The pool B isomer distribution and relative abundance assigned by NMR are in close agreement with the relative abundance and compositions detected by HPAEC ϩ MS (see Tables II and IV  and Scheme II).
Detailed Analysis of Pool C Oligosaccharides-Pool C contained high mannose, hybrid, and fuco-pausimannosidic oligosaccharides. The high mannose glycans were Man 5-6 GlcNAc 2 and were probably the result of mammalian-type ER and Golgi mannosidase trimming. The hybrid glycans were Man 2-4 -GlcNAc 3 and contained either lower arm ␣1,2GlcNAc or corebisecting ␣1,4GlcNAc substitutions. The fuco-pausimannosidic oligosaccharides were Fuc 1 Man 4 -5 GlcNAc 2 , which contained either the novel terminal ␣1,2Fuc or core ␣1,6Fuc additions. In the case of ␣1,2Fuc substitution, the isomers retained the ER Man 9 GlcNAc 2 central or lower arm, which was terminally substituted with Fuc. The ␣1,6-fucosylated isomer was incompletely trimmed and retained a lower arm ␣1,2Man residue, which suggested that a C. elegans ␣1,6-fucosyltransferase acts in the absence of lower arm ␤1,2GlcNAc, unlike that of mammalian systems (38). The pool isomers are shown in Scheme II. Key NMR proton resonances used in isomer assignments are underlined in Table II. Mass spectrometry, HPAEC ϩ MS, exoglycosidase data, and NMR analyses that resulted in isomer assignments are discussed in detail below.
MALDI reflectron TOF MS analysis of the permethylated Bio-Gel P4 pool C glycans contained monoisotopic ions at m/z 1783.4, 1579.3, 1549.0, 1538.3, 1416.3, and 1171.4 (see Table I and Supplemental Material Fig. 2), which are consistent with the following respective sodiated glycan compositions: Man 6 -GlcNAc 2 , Man 5 GlcNAc 2 Fuc 1 Man 4 GlcNAc 2 , Man 6 GlcNAc 1 , Man 3 GlcNAc 3 , and Man 3 GlcNAc 2 (see Table I). HF hydrolysis of pool C glycans prior to permethylation and MALDI reflectron TOF MS analysis gave a nearly identical natriated monoisotopic ion series except that Man 3 GlcNAc 2 was not seen. Judging from the Bio-Gel P-4 elution position and HPAEC ϩ MS analysis (Figs. 1 and 5), it was concluded that the Man 3 Glc-NAc 2 was produced during the permethylation procedure by hydrolysis of other pool glycans.
HPAEC ϩ MS analysis of pooled and permethylated glycans (  (34). The anomeric protons of residues 1-3 are present in all glycans, and their proton allocations will not be discussed further. The ␣1,6Fuc residue 12 was only 0.06 mol and its DQF-COSY cross-peak was not seen due to low abundance.
The assignment of Fuc-and GlcNAc-substituted partial structures C␣ (10%), C␤ (18%), and C␥ (4%) above can be subtracted from the 74% of isomers left to assign, revealing the remaining 42% as the high mannose isomers Man 6 GlcNAc 2 plus Man 5 GlcNAc 2 . By PAD detection Man 5 GlcNAc 2 is ϳ4% of the pool leaving 38% as Man 6 GlcNAc 2 . As described above this Man 6 GlcNAc 2 isomer differs from the C1 Man 6 GlcNAc 2 isomer only in the absence of residue 9, and the presence of residue 8, Of the remaining 0.06 mol of C␤, 0.04 is assigned without further residue addition yielding isomer C5 (see Scheme II). To the remaining 0.02 mol of C␤, 0.02 mol of terminal ␣1,3Man 7 is assigned giving isomer C6, which also necessitates that resonance of residue 4 be present at 4.87 ppm due to its 3-O-substitution (see Scheme I). Thus, 0.02 mol is consumed at 4.87 ppm. Assignment of C4, C5, and C6 completes the allocation of C␤ and leaves 0.05, 0.06, 0.13, 0.05, 0.14, and 0.09 mol of intensity at 5.41, 5.34, 5.09, 5.05, 4.92, and 4.87 ppm left to assign, respectively (see Scheme II for isomer structures). This completes the assignment of 86% of pool components and leaves the linkage assignment of HPAEC ϩ MS Man 5 GlcNAc 2 , Fuc 1 Man 4 Glc-NAc 2 , and Fuc 1 Man 5 GlcNAc 2 left to assign. The Man 5 GlcNAc 2 isomer comigrates with authentic Man ␣1,6(Man ␣1,3)Man ␣1,6(Man ␣1,3)Man␤1,4GlcNAc␤1,4GlcNAc␣/␤ generated by exhaustive digestion of Man 9 GlcNAc 2 with Aspergillus satoi ␣1,2-mannosidase (data not shown). Thus, assignment of isomer C7 as this structure as 4% of pool isomers consumes 0.04 mol each of the intensity at 4.92 and 4.87 ppm and 0.08 mol at 5.09 ppm, which leaves 0.05, 0.06, 0.05, 0.05, 0.10, and 0.05 mol left to assign at 5.41, 5.34, 5.09, 5.05, 4.92, and 4.87 ppm, respectively. Assignment of the Fuc 1 Man 4 GlcNAc 2 and Fuc 1 Man 5 GlcNAc 2 as 6 and 4% of pool isomers yields isomers C8 and C9 (see Table I and Scheme II), where the former contains core ␣1,6Fuc and the later contains the novel Fuc substitution on residue 10. These assignments consume the remainder of the resonance intensity.
As in pool B, jack bean ␣-mannosidase-digested Bio-Gel P4 pool C glycans were permethylated and analyzed by MALDI-TOF MS in linear mode (data not shown) to investigate further the nature of the novel fucosylated glycan determined by NMR analysis. Ions were detected at m/z 765.08, 1213.78, 1550.31, and 1754.59, which are consistent with the sodium adducts of Man 1 GlcNAc 2 , Man 2 GlcNAc 3 , Fuc 1 Man 4 GlcNAc 2 , and Fuc 1 Man 5 GlcNAc 2 . Man 1 GlcNAc 2 is produced by complete digestion of Man 5 GlcNAc 2 and Man 6 GlcNAc 2 pool isomers C1, C2, and C7 (68% of pool). Man 2 GlcNAc 3 is generated by complete digestion of Man 3 GlcNAc 3 and Man 4 GlcNAc 3 pool isomers C4, C5, and C6 (20% of pool). The Fuc 1 Man 4 GlcNAc 2 and Fuc 1 Man 5 GlcNAc 2 products were produced by the incomplete digestion of Fuc 1 Man 5 GlcNAc 2 pool isomer C9 (4% of pool). C9 is identical to B3 and contains ␣1,3Man, which is slowly accessible to the ␣-mannosidase and results in the appearance of both Fuc 1 Man 4 GlcNAc 2 and Fuc 1 Man 5 GlcNAc 2 ions. Although product ions from glycans with ␣1,6Fuc and ␤1,4GlcNAc core substituents were not detected, their compositions and unique substituent signatures were seen by MALDI-TOF MS and NMR analysis, respectively. These data strongly support the presence of terminally linked ␣1,2Fuc as well as high mannose, and lower arm GlcNAc-substituted glycans derived here by NMR above. For every NMR-derived structure there was a corresponding pool isomer of the same predicted mass and nearly identical pool abundance as detected by HPAEC ϩ MS (see Table III).
Detailed Analysis of Pool D Oligosaccharides-Pool D contained high mannose, hybrid, Pc-, and fuco-pausimannosidic glycans. The most abundant isomer was Man 5 GlcNAc 2 . It was fully trimmed at the central, upper, and lower arm ␣1,2Man residues, consistent with activities identical to mammalian ER and Golgi ␣1,2-mannosidases. The hybrid glycan was Man 3 -GlcNAc 3 and contained core-bisecting ␤1,4GlcNAc. The Pc glycan was identical to the Man 5 GlcNAc 2 isomer of the pool but was substituted with three Pc groups. Fuco-pausimannosidic glycans included Fuc 1 Man 5 GlcNAc 2 and Fuc 1 Man 3 GlcNAc 2 . The former was identical to the Man 5 GlcNAc 2 pool isomer but was substituted with core ␣1,6Fuc. The later retained the core mannotriose (Man␣1,3(Man␣1,6)Man␤1,4-) and contained core-linked ␣1,6Fuc. The pool isomers are shown in Scheme II. Key NMR proton resonances used in isomer assignments are underlined in Table II. Mass spectrometry, HPAEC ϩ MS, and NMR analysis that resulted in isomer assignments are discussed in detail below.
MALDI-TOF MS analysis of the Bio-Gel P4 pool D permethylated glycans revealed ions at m/z 1579.2, 1497.2, 1375.3, 1345.3, and 1171.4, which are consistent with the sodiumadducted ions of Man 5 GlcNAc 2 , Man 7 (see Table I (Fig. 5E). These results strongly suggested the presence of a Pc 3 Man 5 GlcNAc 2 pool glycan.
In the one-dimensional NMR spectrum in addition to 1 mol of core residues 1-3, resonance intensity for the following anomeric chemical shifts were observed: 5.10 ppm, 1.74 mol (residues 5 and 7), 4.92 ppm, 1.00 mol (residues 4 and 6), 4.89 ppm, 0.35 mol (residue 12), 4.87 ppm, 0.74 mol (residue 4), and 4.45 ppm, 0.03 mol (residue 14), and at 3.23 ppm 2.0 mol of Pc methyl proton intensity was detected (see Table II As described for pools B and C, no resonance was detected at 4.09 C2-H/3.66 C3-H for residue 3 indicating that all pool isomers have residue 5, which substitutes 3. Because there is no detectable resonance intensity at 5.34 or 5.12 ppm for ␣1,3linked lower arm substitution of residue 5 by residue 8 or 13, 5 is unsubstituted in all pool isomers. These observations allow the assignment of the partial structure Man␣1,3Man␤1,4Glc-NAc(Fuc␣1,6)GlcNAc␣/␤, which is present in 35% of pool isomers and is designated here as partial structure D␣. This assignment leaves 1.39, 1.00, 0.74, and 0.03 mol of resonance at 5.10, 4.92, 4.87, and 4.45 ppm left to assign, respectively.
The 0.03 mol of intensity at 4.45 ppm allows the assignment of partial structure Man␣1,3(GlcNAc␤1,4)Man␤1,4GlcNAcB1, 4GlcNAc␣/␤ in 3% of pool isomers and is designated as partial structure D␤. The partial assignment consumes the 4.45 ppm resonance and 0.03 mol of resonance at 5.10 ppm. This leaves 1.36, 1.00, and 0.74 mol of resonance at 5.10, 4.92, and 4.87 ppm left to assign, respectively.
In addition to a terminal ␣1,3-linked lower arm residue 5 as described above, the upper arm 6-O-linked ␣1,6Man 6 residue is also unsubstituted in all pool isomers as no resonance is detected at 5.14 or 4.54 ppm for residues 9 or upper arm ␣1,2-linked Man, respectively. Also, no resonance at 5.41 ppm for residue 10 was detected indicating that central arm residue 7 is also unsubstituted. These observations provide that the maximum high mannose structure that may exist in the pool is the Man 5 GlcNAc 2 isomer Man␣1,6(Man␣1,3)Man␣1,6(Man-␣1,3)Man␤1,4GlcNAc␤1,4GlcNAc␣/␤ and is designated as partial structure D␥. All non-mannose resonances are accounted for by D␣ and D␤ and total to 38% of pool isomers. By difference D␥ is representative of 62% of the pool. The assignment of D␥ consumes 0.62 mol of intensity at 4.92 and 4.87 ppm for terminal ␣1,6Man 6 and 3,6-di-O-substituted ␣1,6-linked Man 4. At 5.10 ppm 1.24 mol of intensity is consumed with the assignments of ␣1,3-linked residues 5 and 7. On balance this leaves 0.12 mol at 5.10 and 4.87 ppm and 0.38 mol at 4.92 ppm left to assign. PAD detection and HPAEC ϩ MS analysis of the pool gave 38% Fuc-substituted isomers (13% Fuc 1 Man 5 Glc-NAc 2 and 25% Fuc 1 Man 3 GlcNAc 2 ) and 62% high mannose (55% Man 5 GlcNAc 2 and 7% Pc 3 Man 5 GlcNAc 2 ), and HF-treated and permethylated pool glycans indicated a trace amount of Man 3 GlcNAc 3 . These compositions and relative abundance agree closely with the above distribution indicated by assign-ment of partial structures D␣ (35%, containing Fuc), D␤ (3%, containing bisecting GlcNAc), and D␥ (62%, containing a Man 5 GlcNAc 2 core without Fuc or GlcNAc). With the pool's HPAEC distribution of Man 5 GlcNAc 2 and Pc 3 Man 5 GlcNAc 2 and the relative abundance of Pc methyl proton intensity at 3.23 ppm (1.83 mol, see Table II), 55% of the pool is assigned as D1 and is structurally identical to D␥. On balance, the remaining 7% of D␥ is assigned 3 Pc moieties yielding D2. As there are 9 methyl protons on each Pc moiety, the 1.83 mol integrated closely matches this 7% assignment (0.07 ϫ 9 protons ϫ 3 Pc ϭ 1.89 mol). To confirm the presence of the Pc oligosaccharide, we performed one-dimensional and twodimensional DQF-COSY NMR experiments on 5 mM phosphocholine chloride calcium salt solution. DQF-COSY cross-peaks for methylene protons were seen at 4.16/3.57 and 3.73/3.63 ppm (Fig. 6A). Examination of the DQF-COSY spectrum of pool D revealed Pc methylene protons at 4.19/3.60 and 3.74/3.64 ppm, which are slightly deshielded compared with free Pc (compare Fig. 6, A and B). Along with the MALDI-TOF MS results of the pool above, these results are conclusive for the presence of Pc 3 Man 5 GlcNAc 2 isomer D2.
The HPAEC ϩ MS analysis of the pool showed that 25% of the pool was Fuc 1 Man 3 GlcNAc 2 by PAD detection. The structure of this compound is given by the assignment of 0.23 mol of the remaining resonance at 4.92 ppm from residue 4 to partial structure D␣. This yields glycan D4 (see Scheme II). The HPAEC ϩ MS analysis of pool isomers did not provide obvious evidence of any structure bearing bisecting ␣1,4GlcNAc. However, the HF-hydrolyzed and -permethylated glycans revealed low abundance ions at m/z 1417.0 and 1006.5 (see Table I), which are consistent with Man␣1,3(Man␣1,6)(GlcNAc␤1,4)-Man␤1,4GlcNAc␤1,4␣/␤ and the fragment of that structure that occurred during the chemical procedures, GlcNAc␤1,4-Man␤1,4GlcNAc␤1,4GlcNAc␣/b. The remaining 0.03 mol of 4.92 mol of resonance is assigned to the remaining 0.03 mol of D␣ yielding isomer D5 (see Scheme II) completing the anomeric proton resonance assignment of pool D. Glycan compositions and relative pool abundance derived here by NMR were in close agreement with those found by HPAEC ϩ MS (Table III). Detailed Analysis of Pool E Oligosaccharides-The major pool isomer (96%) was Man 3 GlcNAc 2 and contained the core mannotriose. A minor isomer was Fuc 1 Man 3 GlcNAc 2 , which was identical to the major pool glycan with the addition of core-linked ␣1,6Fuc. The pool isomers are shown in Scheme II. Key NMR proton resonances used in isomer assignments are underlined in Table II. Mass spectrometry, HPAEC ϩ MS, and NMR analysis that resulted in isomer assignments are discussed in detail below.
The Bio-Gel P-4 permethylated glycans of pool E were examined by MALDI-TOF in reflectron mode. The pool produced an ion at m/z 1172.7, consistent with sodium adduct of core-type Man 3 GlcNAc 2 (see Table I, Fig. 8, and Supplemental Material Fig. 2). With HF treatment prior to permethylation, MALDI-TOF MS spectra performed in a linear mode revealed a trace signal at m/z 1346.7, consistent with core-type Fuc 1 Man 3 -GlcNAc 2 ( Table I). The one-dimensional 1 H NMR spectra of the pool contained 1 mol of resonance intensity each at 5.09 ppm for residue 5, 4.91 ppm for residue 4, and 4.60 ppm for residue 2. The ␣and ␤-anomers of residue 1 were split 57/43 at 4.70 and 5.19 ppm, respectively, and totaled 1 mol. The DQF-COSY of the pool showed the respective cross-peaks of the above residues as predicted by one-dimensional anomeric proton peaks and are as follows (see Table II and Fig. 8 The presence of residue 3 was also observed at 4.76 ppm by analysis at 318 K. About 0.04 mol of resonance intensity was seen at 4.89 ppm and is from core ␣1,6Fuc revealing that 4% of the pool contains Fuc 1 Man 3 GlcNAc 2 , which confirms the trace amount of this compound seen by MS subsequent to HF treatment and permethylation. Also supporting the low abundance glycan is the low intensity DQF-COSY cross-peak seen at 4.64 C1-H/3.86 C2-H ppm due to the deshielding affect of residue 12 on core residue 2. The above assignments reveal that ϳ96% of pool E is Man 3 GlcNAc 2 (isomer E1), and the remaining 4% contains Fuc 1 Man 3 GlcNAc 2 (isomer E2). Dionex HPAEC produced a major peak at 4.68 min. The MALDI-TOF MS analysis of the permethylated pool produced an ion at m/z 1172.3 (Fig. 7), which is consistent with the sodiated ion of Man 3 GlcNAc 2 , the major pool glycan. It should be noted that a DQF-COSY peak was seen at 4.52/4.12 ppm. Pool MS were performed on native, permethylated, and HF-permethylated samples. Under no conditions were we able to detect any other glycoforms other than those discussed above. Because the Fuc 1 Man 3 GlcNAc 2 compound was only seen by MS after treatment with HF, it is tempting to conclude that the 4.52/4.12 ppm resonance may be due to an undefined phosphorylated functional group. However, we cannot present definitive evidence for such a substitution. DISCUSSION We have determined that C. elegans contains a nearly contiguous series of N-glycans. Five classes were observed including high mannose, complex, hybrid, fuco-paucimannosidic, and Pc glycans. The high mannose, complex, and hybrid glycans show a high degree of conservation with those of mammals, whereas the last two are unique to C. elegans. This series provides insight into the nature, conservation, and uniqueness of N-glycosylation in this model organism.
The mannan Man 3-9 GlcNAc 2 series suggest that C. elegans possesses mannosidase activities that are identical to mammalian ER ␣1,2-mannosidase and Golgi mannosidases I and II (41). High field NMR analysis revealed the presence of hybrid glycans with GNT I-and GNT III-type additions, the enzymes that add ␤1,2GlcNAc to the lower arm and bisecting ␤1,4Glc-NAc to the ␤1,4Man, respectively. Three C. elegans homologues of GNT I have been cloned and their encoded activities demonstrated (1).
Although chitobiose core-linked ␣1,6Fucose was detected, the C. elegans enzyme appears to have looser substrate specificity than that of mammalian systems, where the lower arm GlcNAc (residue 13, Scheme I) is required for fucosyltransferase activity (38). Here we have found a series of core-fucosylated N-chains lacking the lower arm GlcNAc substitution, bisecting GlcNAc, or no GlcNAc at all. However, we cannot rule out the existence in C. elegans of a highly active Golgi ␤-Nacetylglucosaminidase acting on nascent N-glycans in vivo as has been reported in a lepidopteran insect cell system (42).
A novel high mannose branch Fuc substitution was seen, which caps ␣1,2-linked Man residues. This addition is novel, as we were unable to find any example in nature where fucose terminally substitutes mannose and, thus, strongly suggests the existence of a novel fucosyltransferase. It is reasonable to predict that the novel fucosylation seen in C. elegans mixed stage N-glycans is developmentally regulated, and we are presently investigating this possibility. The trematode Schistosoma mansoni differentially expresses fucose-containing moieties on its N-glycans throughout development, where core ␣1,3Fuc is seen in eggs and miracidia; Lewis X glycans are seen in cercaria, and mainly core ␣1,6Fuc is seen in adult worms (43). The differential expression is developmentally regulated (44). Novel distal core and outer chain fucosylation patterns on Hemonchus contortus N-glycans have been reported, and their expressions are, apparently, developmentally regulated as well (18,45). By using two-dimensional chromatographic mapping, Natsuka et al. (46) have reported that a minor portion of C. elegans mixed stage N-glycans released from mixed stage worms by hydrazinolysis contain core-linked ␣1,3Fuc. Although, according to our results, mixed stage C. elegans worms contain ␣1,6and novel ␣1,2Fuc substitutions, we did not detect core ␣1,3-fucosylated N-glycans (Fig. 8).
Pc substituted Man 5 GlcNAc 2 was detected in C. elegans. These N-glycans have been reported in Acanthocheilonema viteae, Onchocerca volvulus, and Trichinella spiralis (17,18,47,48). The Pc-containing oligosaccharides are able to modulate B-lymphocyte activation through activation of protein tyrosine kinase and mitogen-activating protein kinase signal transduction pathways resulting in desensitization to downstream activation (49). It is a conserved modification across parasitic nematode and trematode species, and the biosynthetic components are now being pursued as novel drug targets (50). As C. elegans contains this class of modification, our results demonstrate that this organism may serve as a model system to study the biosynthesis of Pc oligosaccharides. This nematode is much more easily manipulated in laboratory settings and is generally well characterized compared with the parasitic species discussed above.
GNT II adds ␤1,2GlcNAc to the upper arm-linked ␣1,6Man and GNT IV adds ␤1,4GlcNAc to the lower arm ␣1,3Man, respectively, in preparation for tri-and tetra-antennary Nglycans (36,38). Although complex glycans were not present as the major glycans, here MALDI-TOF MS analysis of size exclusion chromatography pooled glycans (pools a-d, Fig. 1 and Table I) strongly suggests that C. elegans produces complex N-glycans with up to six HexNAc residues (Hex 3 HexNAc 6 ) and hybrid and/or complex chains with up to five HexNAc residues (Hex 5 HexNAc 5 ). These oligosaccharides suggest that GNT II and IV type activities are present. Also, LacdiNAc- (43,44) and chito-oligomer branches (17) have been reported in nematodes glycans. Therefore, it is possible that C. elegans produces these oligosaccharides.
Recently C. elegans gly-2 was found to possess GNT V activity in vitro (6-O-substitutes residue 4, see Scheme I). The gly-2 gene restored Phaseolus vulgaris (PHA) lectin-binding affinity in lec-4 cells. The PHA lectins can detect the complex glycan determinant Gal␤1,4GlcNAc␤1,2Man ␣1,6-. It is possible that C. elegans produces the galactosylated branch, as some low abundance glycans, such as the Hex 5 GlcNAc 5 glycan seen here by MALDI-TOF MS, may have contained it, but the Gal abundance was too low to detect by GC/MS.
Although novel carbohydrate modifications were seen, it is shown here that much of the N-glycan biosynthesis of C. elegans is consistent with that seen in mammalian systems. Addition of GlcNAc residues by activities similar or identical to GNT I, II, and III and likely other mammalian-type HexNAc transferases were seen. The N-glycan series defined here shows that there is a functional C. elegans orthologue for nearly every CDG type I and II studied to date. Currently, there is a growing constellation of human congenital disorders of glycosylation in which CDG type I and CDG type II classify cases involving N-glycan dolichol-linked precursor synthesis and transfer and subsequent processing, respectively. Presently, these include CDG types Ia-g and IIa-c for each of which the biochemical lesion has been identified (8,(51)(52)(53)(54). The affected proteins are as follows: Ia, phosphomannomutase-2; Ib, phosphomannomutase-1; Ic, dolichyl-P-Glc:Man 9 GlcNAc 2 -PP-dolichylglucosyltransferase; Id, dolichyl-P-Man:Man 5 GlcNAc 2 -PP-dolichylmannosyltransferase; Ie, Dol-P-Man synthase (catalytic subunit); If, MPDU1 (Man-P-Dol-dependent utilization); Ig, dolichyl-P-Man:Man7GlcNAc2-PP-dolichylmannosyltransferase IIa, UDP-GlcNAc:␣-6-D-mannoside-␤-1,2-N-acetylglucosaminyltransferase II; IIb, ER glucosidase I; IIc, GDPfucose transporter; and IId, UDP-galactose:N-acetylglucosamine-␤-1,4-galactosyltransferase I. In fact, when expressed in CDG type IIc fibroblasts (LAD-II), a C. elegans GDP-fucose transporter reestablished expression of fucosylated glycoconjugates with high efficiency, demonstrating the existence of some level of functional compatibility between C. elegans and human systems (55).
In this communication we have reported the most complete characterization of C. elegans N-glycans to date. We are currently characterizing stage-specific glycosylation patterns in the worm. The present series has revealed a high degree of conservation with mammalian systems as well as novel substitutions including Pc and Man-linked Fuc. This nematode is well characterized genetically and developmentally, and its genome has been completely sequenced. The work presented here has characterized the N-glycans of C. elegans, which, along with its capacity as a developmental, genetic, and

FIG. 8. C1-H/C2-H section of DQF-COSY 1 H NMR spectra of pools A-E.
A-E are C1-H/C2-H sections of pools A-E, respectively. The boxed region in B was taken from a lower slice in the two-dimensional spectrum. For the identity of the numbered C1-H/C2-H resonances see Scheme I and Table III. genomic model system, has revealed its suitability as a model system for the study of glycosylation processes such as those involved in CDG and Pc oligosaccharide biosynthetic pathways.