LEC14, a dominant Chinese hamster ovary glycosylation mutant expresses complex N-glycans with a new N-acetylglucosamine residue in the core region.

The Chinese hamster ovary cell (CHO) glycosylation mutant, LEC14, was previously selected for resistance to pea lectin (Pisum sativum agglutinin) and shown to behave dominantly. The lectin resistance properties of LEC14 cells are related to, but distinct from, those of LEC18, a dominant Chinese hamster ovary mutant that synthesizes complex N-glycans with a novel O-6-linked GlcNAc residue in the core region (Raju, T.S., Ray, M., and Stanley, P. (1995) J. Biol. Chem. 270, 30294-30302). Detailed structural studies of a complex N-glycan fraction from LEC14 cells have revealed yet another novel modification of the core region. [3H]Glc-labeled LEC14 cellular glycopeptides were desialylated, and the fraction that did not bind to concanavalin A-Sepharose was found to have an increased proportion of species that bound to tomato-agarose, and to ricin-agarose. 1H NMR spectroscopy and methylation linkage analysis of the tomato and ricin-bound fractions purified from approximately 10(10) LEC14 cells showed they were complex N-glycans containing a 2,3,6-trisubstituted core Man residue. To examine the core region more closely, these N-glycans were digested with mixtures of beta-D-galactosidases and N-acetyl-beta-D-glucosaminidases to obtain core glycopeptides. The latter were largely unbound by concanavalin A-Sepharose or pea lectin-agarose. 1H NMR spectroscopy and electrospray ionization-mass spectrometry showed that the LEC14 core glycopeptides contain a new GlcNAc residue that substitutes the core beta(1-4)-Man residue at O-2 to give the following novel, N-linked core structure. [structure: see text]

The core region of N-linked glycans is derived from the first five sugars (Man 3 GlcNAc 2 ) attached to dolichol-phosphate during the synthesis of the Glc 3 Man 9 GlcNAc 2 intermediate that is transferred to Asn-X-(Ser/Thr) residues in glycoproteins (1). In plants and some lower organisms, this simple core structure can be modified by the addition of Fuc 1 to the GlcNAc attached to Asn (Asn-GlcNAc) in ␣(1,6)and/or ␣(1,3)-linkage (2)(3)(4)(5)(6), or Xyl in ␤(1,2)-linkage to the ␤(1,4)-Man residue (7)(8)(9)(10)(11)(12)(13)(14)(15). In mammals, the core may be modified by the addition of Fuc to the Asn-GlcNAc in ␣(1,6)-linkage (16,17); by the addition of Glc-NAc to the ␤(1,4)-Man at the O-4 position to generate the bisecting GlcNAc (16,17); and/or by the addition of a GlcNAc at O-6 to the GlcNAc residue linked to the ␤(1,4)-Man residue (18). This latter type of core was discovered in the complex N-glycans unique to the LEC18 CHO mutant (18,19), which is characterized by a high level of resistance to pea lectin (PSA) and Lens culinaris agglutinin (LCA). The phenotype is dominant (19) and is the result of the de novo expression of a novel UDP-GlcNAc:GlcNAc glycosyltransferase that is present in LEC18 but not in parental CHO cell extracts. 2 LEC14 CHO cells have a phenotype similar to, but subtly distinct from, LEC18 cells (19). LEC14 cells are resistant to PSA and LCA, which bind core fucosylated N-glycans (20), but only 3-4-fold compared with the 15-40-fold level of resistance displayed by LEC18 cells (19). On this basis, LEC14 might be expected to be a weak version of the LEC18 phenotype. Mitigating against this conclusion, however, is the fact that LEC14 and LEC18 cells differ significantly in their respective sensitivities to the lectins L-PHA (agglutinin from Phaseolus vulgaris) and ricin. LEC18 cells are more sensitive than LEC14 and parental CHO cells to the toxicity of ricin, while LEC14 cells are more resistant than LEC18 and parental CHO cells to the toxicity of L-PHA (19). Qualitative differences in lectin resistance properties predict a different biochemical basis for each phenotype amongst the CHO glycosylation mutants (21). In this paper we show that the biochemical basis of the LEC14 dominant phenotype is related to that of LEC18 in that complex N-glycans from LEC14 cells possess an altered core structure. The modified core in LEC14 cells is, however, novel and has not previously been observed in glycoconjugates from any source. . Partial support was also obtained from Cancer Core Grant PO 13330. 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.
Cell Lines and Cell Culture-CHO cells were grown in suspension at 37°C in complete ␣ medium (Life Technologies, Inc.) containing 10% bovine calf serum (Life Technologies, Inc.). Independent LEC14 isolates Pro Ϫ LEC14.4A and Gat Ϫ LEC14.2A were obtained from a selection for mutants resistant to the toxicity of PSA (19).
Preparation of Radiolabeled Cell Surface Glycopeptides-Pro Ϫ 5 and Pro Ϫ LEC14.4A cell cultures were established at a density of 1.5 ϫ 10 5 cells/ml in 10 ml of complete medium and 100 Ci of [ 3 H]Glc was added. After 48 h at 37°C, glycopeptides were generated by exhaustive Pronase digestion as described previously (18).
Serial Lectin Affinity Chromatography of Radiolabeled Glyocpeptides-Glycopeptides were fractionated on ConA-Sepharose (0.5 cm ϫ 20 cm) and PSA-agarose (0.5 cm ϫ 20 cm) as described previously (18). Pooled glycopeptides were desalted on Bio-Gel P-2 (1.5 cm ϫ 70 cm). To remove glycosidically bound sialic acid residues, glycopeptides were hydrolyzed with 10 mM HCl at 100°C for 1 h and desalted on Bio-Gel P-2. Desialylated glycopeptides were applied to tomato-agarose (0.25 cm ϫ 20 cm) in phosphate-buffered saline containing 1 mM CaCl 2 , 1 mM MnCl 2 , and 0.02% azide, pH 7.4 (PBS 2ϩ ), and the column was washed with Ն8 column volumes of buffer. The bound fraction was eluted with 10 -15 ml of PBS 2ϩ containing N,NЈ,NЉ-triacetylchitotriose (ϳ10 mg/ml) and purified by ultrafiltration followed by gel filtration on a Bio-Gel P-2 column. Glycopeptides that did not bind to tomato-agarose were applied in PBS 2ϩ to a column of RCA I -agarose (0.5 cm ϫ 25 cm). After washing with at least 10 column volumes of PBS 2ϩ , bound glycopeptides were eluted with 100 -200 mM lactose in PBS 2ϩ . Alternatively, a linear gradient of 0 -100 mM lactose in PBS 2ϩ was used. Lectin affinity chromatography was carried out either at room temperature or at 4°C at flow rates of 6 -10 ml/h.
The desialylated glycopeptides were dissolved in 1 ml of PBS 2ϩ and applied to tomato-agarose (0.5 cm ϫ 20 cm). The column was washed with PBS 2ϩ (ϳ100 ml), and the tomato-bound fraction was eluted with 40 ml of N,NЈ,NЉ-triacetylchitotriose (10 mg/ml) in PBS 2ϩ and purified by ultrafiltration followed by repeated Bio-Gel P-2 column chromatography. The pure glycopeptides were used for 1 H NMR spectroscopy, monosaccharide analysis, and to isolate core glycopeptides by exoglycosidase digestion.
The tomato-unbound fraction was desalted on Bio-Gel P-2 (1.5 cm ϫ 70 cm), lyophilized, dissolved in 1 ml of PBS 2ϩ , and chromatographed on a column of RCA I -agarose (0.5 cm ϫ 20 cm). The column was eluted with ϳ200 ml PBS 2ϩ followed by 100 ml of 100 mM lactose in PBS 2ϩ . Pooled fractions were desalted on Bio-Gel P-2. A portion of each glycopeptide fraction was used for monosaccharide analysis. The remainder was used to record 1 H NMR spectroscopy, was subjected to methylation linkage analysis, or treated with exoglycosidases to isolate core glycopeptides.
Preparation and Characterization of Radiolabeled Core Glycopeptides-The tomato-bound glycopeptides of LEC14 and parent cells (ϳ15000 cpm each) were separately treated with a mixture of ␤-Dgalactosidase (bovine testis and/or D. pneumoniae, 25 milliunits) and N-acetyl-␤-D-glucosaminidase (jack bean or D. pneumoniae) in McIvine buffer, pH 4.0 -5.0 at 37°C for 24 h. The incubation was continued for 72-96 h with addition of enzymes every 24 h. The reaction was stopped by heating the mixture at 100°C for 5-10 min, and the resulting core glycopeptides (V o , excluded fraction; V e , included fraction) were separated from released monosaccharides on Bio-Gel P-2 (1.5 cm ϫ 70 cm). A portion of the core glycopeptides were fractionated on ConA-Sepharose (0.5 cm ϫ 20 cm) and/or PSA-agarose (0.5 cm ϫ 20 cm) as described earlier. Fractions of 0.5 ml were collected and counted for radioactivity. Another portion of the core glycopeptides was hydrolyzed with 2 M trifluoroacetic acid, and the hydrolyzate was analyzed by HPAEC-PAD for monosaccharides. Core glycopeptides from the RCA I -bound glyco- [ 3 H]Glc-labeled glycopeptides of LEC14 and parental cells, which did not bind to ConA-Sepharose, were desialylated and fractionated on tomato-agarose as described under "Experimental Procedures" (upper panel). The fractions bound to tomato-agarose were desalted and subjected to glycopeptide mapping by HPAEC-PAD. Elution was achieved using a gradient of 100 mM NaOH and 100 mM NaOH containing 1 M NaOAc as described previously (18). Fractions of 1 min were collected and counted for radioactivity.

FIG. 2. Affinity chromatography of [ 3 H]Glc-labeled glycopeptides on RCA I -agarose. [ 3 H]
Glc-labeled glycopeptides that passed through ConA-Sepharose were desialylated and fractionated on tomatoagarose as described in Fig. 1. The tomato-unbound glycopeptides were fractionated on RCA I -agarose. The RCA I -bound glycopeptides were eluted with a linear gradient of 0 -100 mM lactose in PBS 2ϩ as described under "Experimental Procedures." peptides were similarly prepared and fractionated on ConA-Sepharose and/or PSA-agarose and analyzed for monosaccharide composition.
Structural Analyses of LEC14 Core Glycopeptides-The tomatobound and RCA I -bound glycopeptides (ϳ250 -300 g) from LEC14 cells were treated, separately, with a mixture of ␤-D-galactosidase (bovine testis) and N-acetyl-␤-D-glucosaminidase (jack bean) at 37°C for 96 h followed by another treatment with a mixture of D. pneumoniae ␤-Dgalactosidase and N-acetyl-␤-D-glucosaminidase in McIvine buffer as described above. The enzyme digestion was stopped by heating the reaction mixture at 100°C for 5-10 min. Core glycopeptide fractions were purified by repeated Bio-Gel P-2 chromatography and ion-exchange chromatography with Chelex resin to remove paramagnetic material. Pure glycopeptides (yield ϳ80 g from RCA I -bound and ϳ35 g from tomato-bound) were analyzed by 1 H NMR spectroscopy, and electrospray ionization-mass spectrometry (ESI-MS) and for monosaccharide composition by HPAEC-PAD.
500 MHz 1 H NMR Spectroscopy-1 H NMR spectroscopy of glycopeptide samples was carried out using a 500 MHz Varian Spectrometer at 23, 30, or 42°C with a sweep width of 4000 Hz, a cycle delay time of 3 s, and 90°pulses. Acetone was used as internal standard, and chemical shifts were expressed in parts/million based on acetone being at 2.225 ppm. The samples were purified and prepared for 1 H NMR spectroscopy as described previously (18).
Methylation Linkage Analysis-Glycopeptides (ϳ100 g) were permethylated according to Hakomori (23). The permethylated product was purified by dialysis, hydrolyzed with 2 M trifluoroacetic acid, reduced with sodium borodeuteride, acetylated with pyridine-acetic anhydride, and analyzed by gas-liquid chromatography-mass spectrometry (GLC-MS) as described previously (24). Furthermore, the trifluoroacetic acid hydrolyzate was hydrolyzed with acetic acid/sulfuric acid to achieve complete hydrolysis of glycosidic bonds, reduced with sodium borodeuteride, acetylated with pyridine-acetic anhydride, and analyzed by GLC-MS. GLC-MS was performed with a Hewlett-Packard 5890 series II gas liquid chromatograph, using a DB-5 column (J&W Scientific), coupled to a Hewlett-Packard 5871 MSD mass spectrometer.
Mass Spectrometry-An API-III triple-quadrupole mass spectrometer (PE-SCIEX, Ontario, Canada) was used to perform electrospray ionization-mass spectrometry (ESI-MS) of glycopeptides. The SCIEX IonSpray interface with nitrogen as the nebulizer gas, an ion spray voltage of 3300 and the orifice at 70 V was used. Harvard Apparatus syringe pump was used to infuse the sample into the mass spectrometer at 2 l/min after diluting 1:1 with 50% acetonitrile/H 2 O containing 0.1% trifluoroacetic acid as described previously (18).

Unique Glycopeptide Fractions From LEC14 Cells-
The resistance of LEC14 cells to PSA and LCA (19) is presumed to reflect the altered expression of cell surface carbohydrates (21). Consistent with this, two independent LEC14 mutants exhibited a 2-3-fold reduction in 125 I-PSA binding when compared with parental CHO cells, over a broad range of PSA concentra- Approximately 300 g of the tomatobound fraction (Fig. 1) was prepared from ϳ10 10 LEC14 cells. The glycopeptides were extensively desalted on Bio-Gel P-2, passed through Chelex, exchanged with D 2 O, and subjected to 1 H NMR spectroscopy at 500 MHz. Spectra were recorded at 23 or 42°C.
tions. 3 To identify carbohydrates unique to LEC14 cells, metabolically labeled cellular glycopeptides were compared by serial lectin affinity chromatography. No significant differences in profile were observed when [ 3 H]Glc-labeled glycopeptides from parental or LEC14 CHO cells were fractionated on a ConA-Sepharose column (25); about 90% of the glycopeptides passed through ConA-Sepharose, ϳ5% eluted in the biantennary fraction, and ϳ5% eluted in the hybrid and oligomannosyl fraction. Glycopeptides that did not bind to ConA-Sepharose were subjected to further lectin affinity chromatography after removing terminal sialic acid residues by mild acid hydrolysis. On a tomato-agarose column (26), less parental CHO glycopeptides (ϳ15%) bound and were eluted with N, NЈ,NЉ-triacetylchitotriose than LEC14 glycopeptides (ϳ45%, Fig. 1). A further difference between parental and LEC14 glycopeptides was uncovered when the glycopeptides that did not bind to tomatoagarose were fractionated on RCA I -agarose (26,27). The fraction from LEC14, which bound to RCA I -agarose eluted at ϳ70 mM lactose, whereas the equivalent fraction from parental cells eluted at ϳ30 mM lactose (Fig. 2), indicating a structural difference between these glycopeptides. Also ϳ2-3-fold more LEC14 glycopeptides bound to RCA I -agarose compared with parental glycopeptides (Fig. 2). Glycopeptide mapping of toma- a Relative proportions of partially methylated alditol acetates of Fuc and Gal were obtained from the total ion-chromatogram of GLC-MS of a 2 M trifluoroacetic acid hydrolysate (see "Experimental Procedures"). Raw values were corrected with the relevant response factors (24). Analysis of spectra of this hydrolyzate indicated that the permethylated product underwent only partial hydrolysis. Hence, the trifluoroacetic acid hydrolyzate was subjected to further hydrolysis with acetic acid/ sulfuric acid to release all sugar residues. Relative proportions of partially methylated alditol acetates of GlcNAc and Man were obtained from the total ion-chromatogram of GLC-MS of the latter preparation and were also corrected with the relevant response factors (24). The data from both analyses are combined in the table. Approximately 250 g of the RCA I -bound fraction (Fig. 2) was prepared from ϳ10 10 LEC14 cells. They were extensively desalted on Bio-Gel P-2, passed through Chelex, exchanged with D 2 O, and subjected to 1 H NMR spectroscopy at 500 MHz. Spectra were recorded at 23 or 42°C.
to-bound (Fig. 1)  Analysis of LEC14 Glycopeptides by 1 H NMR Spectroscopy and Methylation Linkage Analysis-In order to characterize glycopeptides unique to LEC14 cells, the tomato-bound (Fig. 1) and RCA I -bound fractions (Fig. 2) were prepared from ϳ10 10 LEC14 cells. 1 H NMR spectra of LEC14 tomato-bound glycopeptides (Fig. 3) showed that they contain poly-N-acetyllactosamine chains (28) and a core Fuc residue (16,17). Chemical shifts at 23°C of 4.697, 4.469 and 4.149 ppm could be assigned to H-1 of ␤-GlcNAc, H-1 of ␤-Gal, and H-4 of Gal, respectively (17,28). The large resonance at 2.030 ppm from N-acetyl protons is typical of GlcNAc residues in poly-N-acetyllactosamine chains (28). The resonance at 1.198 ppm is due to CH 3 protons of a Fuc residue ␣(1,6)-linked to the Asn-GlcNAc (16,17). When spectra were acquired for longer times, resonances due to core Man residues at 5.119 and 4.869 ppm could be assigned to H-1 of ␣(1,3)-Man and ␣(1,6)-Man, respectively (Fig. 3, inset). Although spectra were recorded at two temperatures in order to uncover resonances obscured by the HOD signal, a resonance corresponding to H-1 of ␤(1,4)-Man, which normally occurs at 4.715-4.741 ppm for tri-and tetraantennary carbohydrates (16,17), was not observed, indicating that this residue was in an atypical core structure.
Additional evidence in support of a structural change affecting the ␤(1,4)-Man residue was obtained from 1 H NMR spectroscopy of RCA I -bound glycopeptides (Fig. 4). The RCA I -bound glycopeptides also contained poly-N-acetyllactosamine chains, as evidenced from the large resonances due to H-1 and H-4 of Gal, and H-1 of GlcNAc residues (Fig. 4), and the resonances due to core Man residues that are more prominent than in Fig.  3. Again there was no evidence for a ␤(1,4)-Man residue between 4.715 and 4.741 ppm. The complexity of the N-acetyl proton region in Fig. 4 compared with Fig. 3 shows that RCA Ibound glycopeptides were more branched.
When the chemical shifts of Figs. 3 and 4 were entered into Sugabase, the 1 H NMR data base (17), no structure was given, indicating that both LEC14 glycopeptide preparations contained unique species. To examine sugar linkages in these LEC14 glycopeptides, a portion of the RCA I -bound fraction was subjected to methylation linkage analysis. From the data in Table I, it is evident that the LEC14 glycopeptides have terminal Gal, GlcNAc, and Fuc residues and substituted Gal, Glc-NAc, and Man residues. A key finding from this analysis was the presence of 4-O-methylmannose indicating the existence of a 2,3,6-trisubstituted Man residue in the RCA I -bound glycopeptides. In some plants and lower animals, N-linked carbohydrates contain a 2,3,6-trisubstituted Man residue due to Xyl  (Fig. 2) were treated exhaustively with a mixture of bovine testis ␤-D-galactosidase and jack bean N-acetyl-␤-D-glucosaminidase, and the products were purified by Bio-Gel P-2 chromatography to obtain the V o and V e fractions (Fig. 3). These core glycopeptide fractions were hydrolyzed with 2 M trifluoroacetic acid at 100°C for 3-4 h, subjected to HPAEC-PAD, and fractions were counted as described under "Experimental Procedures." Similar results were obtained with V o and V e fractions prepared from tomato-bound glycopeptides.  Analysis of Core Glycopeptides By Lectin Affinity Chromatography-To investigate directly whether LEC14 glycopeptides have an altered core structure, [ 3 H]Glc-labeled tomatobound ( Fig. 1) and RCA I -bound (Fig. 2) glycopeptides of both parental and LEC14 cells were separately treated exhaustively with a mixture of bovine testis ␤-D-galactosidase and jack bean N-acetyl-␤-D-glucosaminidase. The digestion products were fractionated on Bio-Gel P-2 into two major glycopeptide pools (V o , excluded volume; V e , included volume; Fig. 5).
Fractionation on ConA-Sepharose revealed a significant difference between the V o glycopeptides of LEC14 and parent cells. Approximately, 65% of the parental species bound to ConA-Sepharose, whereas Ͻ12% of LEC14 V o glycopeptides bound to this affinity column (Fig. 6). The V e glycopeptides of LEC14 and parent cells also showed a similar difference on ConA-Sepharose (data not shown). When LEC14 V o glycopeptides were obtained by exhaustive digestion with a mixture of ␤-D-galactosidase and N-acetyl-␤-D-glucosaminidase from D. pneumoniae (29), a large proportion (ϳ85%) did not bind to ConA-Sepharose. The corresponding V e glycopeptides from LEC14 also did not bind to ConA-Sepharose. In addition, a large proportion of V o (and V e ) LEC14 core glycopeptides did not bind to PSA-agarose (Fig. 6), whereas fucosylated, trimannosyl core glycopeptides isolated from parent CHO cells bind to this column (18). Therefore both the V o and V e core glycopeptides from LEC14 contain a core structure that prevents binding to ConA-Sepharose and PSA-agarose.
Monosaccharide analysis of V o and V e fractions (Table II) showed that Gal residues had been essentially removed. The V o fractions from both cell lines were the products of partial digestions. However, the V e fraction from parent cores had only  Fig. 7. The chemical shifts of the most relevant cores, Man 3 GlcNAc 4 (Fuc)Asn (core A) and Man 3

New Core in N-Glycans of CHO Cells
approximately two GlcNAc equivalents, as predicted if glycosidase digestion was complete. By contrast, LEC14 V e core glycopeptides contained three GlcNAc equivalents, suggesting the presence of an additional GlcNAc residue that had resisted N-acetyl-␤-D-hexosaminidase digestion. The extra GlcNAc residue in the V e fraction of LEC14 cells was postulated to be the reason that most LEC14 core glycopeptides did not bind to ConA-Sepharose or PSA-agarose (Fig. 6).
Structural Analysis of LEC14 Core Glycopeptides-To investigate the structure of LEC14 N-glycan cores, the unlabeled glycopeptides shown in Figs. 3 and 4 were separately treated exhaustively with a mixture of ␤-D-galactosidases and N-acetyl-␤-D-glucosaminidases from D. pneumoniae, bovine testis, and Jack bean. The V o and V e fractions obtained by Bio-Gel P-2 chromatography (Fig. 5) were subjected to 1 H NMR spectroscopy at 500 MHz (Fig. 7). Spectra were recorded at 23 and 42°C to reveal resonances obscured by the HOD peak. The 1 H NMR spectra shown in Fig. 7 contained resonances due to five -NAc groups (Table III), showing that the V o glycopeptides contained five GlcNAc residues. The resonance at 1.199 ppm due to -CH 3 of Fuc (16,17) showed that the V o cores are fully fucosylated. Table III compares the chemical shifts of LEC14 V o glycopeptides and two closely related structures; a GlcNActerminating, fucosylated biantennary glycopeptide (core A) and the same glycopeptide carrying a ␤(1,4)-linked bisecting Glc-NAc residue (core B). It is apparent that the chemical shifts for the majority of reporter groups are different in each glycopeptide. Assignments of chemical shifts to individual protons have been made for the LEC14 V o glycopeptides on the basis of the similarity of all chemical shifts to resonances in core A or core B, except for the novel resonances that are tentatively assigned to the new GlcNAc residue. The effect of this ␤-GlcNAc (which on the basis of methylation linkage analysis (Table I) must be linked at O-2 to the ␤(1,4)-Man) on the chemical shifts of adjacent residues is profound, in a manner analogous to, but distinct from, the effects of the bisecting GlcNAc (16,30). Compared with Core A, the only reporter groups unchanged are the -NAc resonances and the -CH 3 resonance of Fuc. All other reporter groups in LEC14 V o glycopeptides are close to a related reporter group in the two previously assigned cores, and coupling constants, as well as intensities, are consistent with the assignments given in Table III. The chemical shifts obtained at 42°C differed by Ͻ0.006 ppm for each reporter group.
The composition of the V o glycopeptides was confirmed by ESI-MS (Fig. 8). This spectrum was obtained at 70 V, under conditions that cause fragmentation of core glycopeptides (18,31). The ion at atomic mass unit 1780.2 corresponds to a molecular ion (MH ϩ ) containing five HexNAc, three Hex, one deoxyHex, and one Asn residue. The monosodiated molecular ion (MNa ϩ ) at atomic mass unit 1803.0 confirms this composition. The major V o glycopeptide underwent fragmentation from both the nonreducing end and from the Asn end, to give fragment ions that are interpretable in structural terms (Scheme 1). The ions at atomic mass unit 904.8 and 891.2 provide strong evidence that the new GlcNAc residue is attached to the ␤(1,4)-Man residue. The methylation linkage analysis (Table I)  Although only small amounts of V e glycopeptides were available, it was possible to obtain chemical shift information from 1 H NMR spectra at 23 and 42°C (Table IV) and to perform ESI-MS on this fraction (Fig. 9). 1 H NMR chemical shifts could be assigned by comparison with related compounds containing Xyl-linked O-2 to the ␤(1,4)-Man or GlcNAc-linked O-6 to the core GlcNAc residue (Table IV). The spectrum at 23°C contains three resonances due to -NAc groups (Table IV), indicating the presence of three GlcNAc residues that were assigned to the Asn-GlcNAc, core GlcNAc, and the new GlcNAc attached to ␤(1,4)-Man at O-2. The resonance at 5.105 ppm was assigned to H-1 of ␣(1,3)-linked core Man based on a comparison with core C (Table IV). The resonance at 4.866 ppm was assigned to H-1 of the ␣(1,6)-linked core Man, because a similar type of shift for this residue is induced in Xyl-containing oligosaccharide (9). A resonance at 4.761 ppm was observed in the spectra recorded at 42°C, which is assigned to the H-1 of the ␤(1,4)-Man residue, based on its coupling constant of 2 Hz and its relative intensity. The H-2 resonances of the core Man residues are significantly  (Table IV), indicating the novelty of the core region of LEC14 V e glycopeptides. The H-1 resonance of the Asn-linked GlcNAc in V e glycopeptides is very similar to that of Core C. However, the H-1 resonance for the core GlcNAc residue was significantly different in LEC14 V e glycopeptides (Table IV)

DISCUSSION
The structural studies described in this paper have identified a novel core of complex N-glycans (Scheme 3). These glycans from the LEC14 CHO glycosylation mutant behave as a unique species on lectin affinity chromatography (Figs. 1 and 2). Methylation linkage analysis (Table I) revealed the presence of a 2,3,6-trisubstituted Man residue and a residue of terminal GlcNAc, suggesting that a GlcNAc residue may substitute the ␤(1,4)-Man core residue. This prediction was confirmed by 1 H NMR spectroscopy (Fig. 7, Tables III and IV) and ESI-MS of core glycopeptides (Figs. 8 and 9) obtained by exoglycosidase digestions.
The GlcNAc residue linked at O-2 to the ␤(1,4)-Man residue is in an analogous position to the ␤(1,2)-Xyl found in various plant and animal glycoproteins (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19). The presence of this GlcNAc in LEC14 glycopeptides has profound effects on the properties of the N-glycans to which it is attached. It causes branched structures with polylactosamine chains to interact more strongly with tomato and RCA I affinity columns (Figs. 1 and 2) as well as with RCA II -agarose. 2 It causes cores obtained after prolonged exoglycosidase digestion not to bind to ConA or PSA affinity columns (Fig. 6). It alters the chemical shifts of  1 H NMR spectral chemical shifts of LEC14 V e core glycopeptides with related core glycopeptides Assignments for LEC14 V e core glycopeptides are taken from spectra recorded at 500 MHz (see "Experimental Procedures"). The chemical shifts of the related cores Man 3 GlcNAc 2 (Fuc)Asn (core C), and Man 3 Xyl 1 GlcNAc 2 (core D) were obtained from Grey et al. (30) and D'Andrea et al. (9), respectively. E, Man; ᮀ, GlcNAc; è, Fuc; Xyl; Obs, Obscured; NA, not assigned.
FIG. 9. Electrospray ionization-mass spectrometry of LEC14 V e core glycopeptides. The V e core glycopeptide fraction (Fig. 5) was isolated from the LEC14 glycopeptides shown in Fig. 3 as described under "Experimental Procedures." A portion (ϳ5 g) was subjected to ESI-MS. The ESI-MS spectrum was recorded as described in Fig. 8. adjacent residues in a specific fashion (Tables III and IV). It affects the fragmentation pathways of ions generated during ESI-MS at 70 V (Schemes 1 and 2), since completely different fragmentation patterns were obtained with LEC18 core glycopeptides that have a GlcNAc-GlcNAc linkage in the core (18), and it appears to increase the relative resistance of ␤(1,2)linked arm GlcNAc-residues to cleavage by ␤-hexosaminidases, because after exoglycosidase digestion, the proportion of V o partial digestion products was higher in LEC14 compared with parental glycopeptides (Fig. 5). This meant that few V e glycopeptides were generated from LEC14 compared with parental glycopeptides (Fig. 5). Interestingly, the new GlcNAc residue, although terminal, appeared completely resistant to removal by our conditions of ␤-hexosaminidase treatment. The same resistance was observed with glycopeptides from LEC18 cells that contain a GlcNAc-linked O-6 to the core GlcNAc residue (18). Thus, the new GlcNAc linkages are quite resistant to the action of the tested ␤-hexosaminidases. Interestingly, the bisecting GlcNAc (␤(1,4)-linked GlcNAc to ␤(1,4)-Man) is reported to be sensitive to these enzymes (32).
The two structures synthesized by mammalian cells that are most related to the new core are shown in Table III. Each core gives a unique 1 H NMR spectrum that reflects profound con-formational differences between the three structures (Table  III). These chemical differences translate into biological phenomena at the surface of CHO cells. Thus LEC10 CHO cells, which carry the bisected GlcNAc on cell surface N-glycans, are 10 -20-fold more resistant to ricin and ϳ15-fold more sensitive to P. vulgaris erythroagglutinin (E-PHA) compared with parental CHO cells, which have N-glycans lacking a bisecting Glc-NAc (33). The presence of a bisecting GlcNAc affects the conformation of distal Gal residues, making them less accessible for recognition by ricin (34). By contrast, LEC14 cells are similar to parental CHO cells in their sensitivity to ricin and E-PHA, while LEC18 cells are slightly sensitive to ricin (19). However, the latter mutants are altered with respect to other lectins that recognize lactosamine units. LEC14 cells have glycopeptides that bind more to tomato and RCA I lectins, while LEC18 cells bind more D. stramonium lectin than parental CHO cells (18). The cores of LEC14 and LEC18 N-glycans lead to cellular resistance to PSA and LCA (19), whereas a bisecting GlcNAc in N-glycans does not confer resistance to these lectins (35). Therefore, each core generates a distinct lectin resistance phenotype that reflects differences in the corresponding cell surface carbohydrates.
The lectin resistance properties of the CHO glycosylation mutants signify either decreased binding (resistance) or increased binding (hypersensitivity) of toxic plant lectins to cell surface carbohydrates (36). The profound changes in lectin recognition that are characteristic of, and unique to, each CHO dominant mutant are generated by the de novo expression of a single glycosyltransferase in each case: LEC10 CHO mutants express N-acetylglucosaminyltransferase III (GlcNAc-TIII; Ref. 33); LEC11, LEC12, LEC29, and LEC30 CHO cells express distinct ␣(1,3)-fucosyltransferase activities (37-39); LEC14 and LEC18 CHO mutants also each possess a GlcNAc-transferase activity that is absent from parental cells. 2 GlcNAc-TIII and the ␣(1,3)-fucosyltransferases are coded for by developmentally regulated genes (40), and it is predicted that this will also be true for the new transferases of LEC14 and LEC18 cells. Considering the consequences of expressing one of these glycosyltransferases on the array of N-glycan structures present at the cell surface and their regulated expression in different tissues, it seems very likely that animal lectins exist that recognize the specific glycan changes resulting from the expression of each transferase. This paradigm may in fact operate in the case of the selectins that recognize sialyl-Lewis X determinants, as the latter may be generated by the regulated expression of an ␣(1,3)-fucosyltransferase (41). Analogous carbohydrate binding proteins that recognize lactosamine units in the context of different N-glycan core structures, or that recognize the various cores themselves (42), are predicted to exist and to mediate specific cell-cell adhesion events important in morphogenesis.

T. Shantha Raju and Pamela Stanley
-Acetylglucosamine Residue in the Core Region N -Glycans with a New N Complex