Eur J Biochem 271, 1172–1186 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04021.x Determination by electrospray mass spectrometry and 1H-NMR spectroscopy of primary structures of variously fucosylated neutral oligosaccharides based on the iso-lacto-N-octaose core Heide Kogelberg1, Vladimir E Piskarev2, Yibing Zhang1, Alexander M Lawson1 and Wengang Chai1 MRC Glycosciences Laboratory, Imperial College Faculty of Medicine, Northwick Park Institute for Medical Research, Harrow, Middlesex, UK; 2Nesmeyanov Institute of Organoelement Compounds, Russian Academy of Sciences, Moscow, Russia We have isolated a nonfucosylated and three variously fucosylated neutral oligosaccharides from human milk that are based on the iso-lacto-N-octaose core Their structures were characterized by the combined use of electrospray mass spectrometry (ES-MS) and NMR spectroscopy The branching pattern and blood group-related Lewis determinants, together with partial sequences and linkages of these oligosaccharides, were initially elucidated by highsensitivity ES-MS/MS analysis, and then their full structure assignment was completed by methylation analysis and H-NMR Three new structures were identified The nonfucosylated iso-lacto-N-octaose, Galb1–3GlcNAcb1– 3Galb1–4GlcNAcb1–6[Galb1–3GlcNAcb1–3]Galb1–4Glc, has not previously been reported as an individual oligosaccharide The monofucosylated and trifucosylated iso-lacto-N-octaose,Galb1–3GlcNAcb1–3Galb1–4(Fuca1–3) GlcNAcb1–6[Galb1–3GlcNAcb1–3]Galb1–4Glc and Galb1– 3(Fuca1–4)GlcNAcb1–3Galb1–4(Fuca1–3)GlcNAcb1– 6[Galb1–3(Fuca1–4)GlcNAcb1–3]Galb1–4Glc, both containing an internal Lex epitope, are also novel structures A role for carbohydrates in cellular events has long been hypothesized, although strong evidence for this has only emerged over the last two decades Awareness of the biological function of oligosaccharide chains in glycoproteins, glycolipids and proteoglycans has intensified as an increasing number of examples have been reported that reveal that carbohydrate structures participate in various biological events in addition to modifying protein function One of the early demonstrations of the role of carbohydrates in recognition was binding of the influenza virus to red blood cells via sialic acid [1], and later by work on the chemical basis of the antigenicity of polysaccharides and of the well-known ABO (H) blood-group system [2,3], in which specificity is determined by oligosaccharide sequences Carbohydrates are well placed to act in cellular recognition as many cells are surrounded by an oligosaccharide layer made from cell-associated glycoconjugates, which often overshadows protein and lipid components on the cell surface Specific oligosaccharide sequences, such as the type (Galb1–3GlcNAc)/type (Galb1–4GlcNAc) chains and the blood group-related antigens bearing the H (Fuca1–2Galb1–3/4GlcNAc), Lewisa [Lea, Galb1– 3(Fuca1–4)GlcNAc] and Lewisx [Lex, Galb1–4(Fuca1– 3)GlcNAc] determinants, occur naturally as structural elements of free oligosaccharides or on the carbohydrate chains of glycoproteins and glycolipids and comprise recognition motifs for cell–cell and cell–matrix interactions [4,5] Human milk is a unique source of diverse oligosaccharides, and more than 80 have been isolated and sequences assigned [6] Many of these structures are closely related to the carbohydrate chains of glycoproteins and glycolipids [7] These diverse oligosaccharide sequences may also serve as cell differentiation and tumour antigens [5] Milk oligosaccharides are considered to play a part in the inhibition of bacterial adhesion to epithelial surfaces, as they are able to mimic the binding epitope of the epithelial receptor [8] Also, milk contains oligosaccharides that resemble structures recognized by the cell–cell adhesion molecules, the selectins, suggesting a role in inflammatory processes [8,9] Human milk has also been used as a rich source of oligosaccharides to map the fine binding specificity of E-selectin [10] In contrast with oligonucleotides and peptides, oligosaccharides can be branched, and hence a relatively simple set of monosaccharides can form a huge number of complex structures A greater degree of structural complexity produced by branching is the norm for naturally occurring carbohydrates, and often a branched sequence carrying two or more recognition motifs is more potent [11,12] Free oligosaccharides from human (milk, urine and infant faeces) have a common lactose (Galb1–4Glc) core It can be extended, for example, at the 4-position of the Gal as a linear sequence or at its 3,6-positions as a branched sequence The linear and branched chains are often Correspondence to W Chai, MRC Glycosciences Laboratory, Imperial College Faculty of Medicine, Northwick Park Institute for Medical Research, Watford Road, Harrow, Middlesex HA1 3UJ, UK Fax: + 44 20 8869 3253, Tel.: + 44 20 8869 3252, E-mail: w.chai@imperial.ac.uk Abbreviations: CID, collision-induced dissociation; ES-MS, electrospray mass spectrometry; iLNO, iso-lacto-N-octaose; Lea, Lewis a; Lex, Lewis x; PMAA, partially methylated alditol acetate; rOe, rotating frame nuclear Overhauser enhancement (Received December 2003, accepted February 2004) Keywords: electrospray mass spectrometry; human milk; oligosaccharide; NMR Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1173 fucosylated to varying degrees to form several of the blood group-related antigens Methods for detailed characterization of these recognition motifs are important in modern structural cell biology to derive structure/function relationships, particularly in the postgenome era, in order to understand post-translational glycosylation and its function Expansion of our knowledge on the repertoire of carbohydrate structures, the ÔglycomeÕ, and investigation of oligosaccharide epitopes involved in carbohydrate–protein interactions require their detailed isolation and structural determination With small amounts of material (e.g a few picomoles), no single analytical technique is capable of the complete characterization of an oligosaccharide structure Structure elucidation is therefore usually achieved by using several different techniques, of which MS and NMR are two of the most powerful Previously, we demonstrated the distinction of chain type and blood-group type (such as Lea/x and Leb/y) of underivatized oligosaccharides by negative-ion electrospray mass spectrometry (ES-MS) with collision-induced dissociation (CID) and MS/MS scanning with low picomole sensitivity [13,14] Several characteristic fragmentations are useful for obtaining detailed structural information The double glycosidic D-type cleavage [13,14] is unique to 3-linked GlcNAc and Glc residues, whereas 0,2A-type cleavages [15] only occur with 4-linked GlcNAc and Glc, resulting in fragment ions, indicating type and type chains or blood group types For a 3-linked GlcNAc or Glc, the D-type cleavage occurs at the glycosidic bonds of both reducing and nonreducing sides of the residue, by combined C-type and Z-type cleavages (see Results for discussion and Figs 1–4 for illustration) For -3GlcNAc- without further substitution, a fragment at m/z 202 is obtained If a Fuc is present at the 4-position of the -3GlcNAc- (e.g in the case of an Lea determinant), a fragment at m/z 348 (202 +146) results, whereas if the 4-position is substituted by Gal (e.g in the case of an Lex determinant), a unique fragment at m/z 364 is produced Therefore, a D-fragment at m/z 202 indicates a Fig Product-ion spectra of iLNO with [M-H]– (A) and [M-2H]2– (B) as precursors The structure of iLNO is shown to indicate the proposed fragmentation The nomenclature used to define the cleavage is based on that introduced previously [15], and ions marked with ÔhÕ are fragments produced by dehydration of the major ions Ó FEBS 2004 1174 H Kogelberg et al (Eur J Biochem 271) Fig Product-ion spectra of MFiLNO with [M-H]– (A) and [M-2H]2– (B) as precursors The structure of MFiLNO is shown to indicate the proposed fragmentation The nomenclature used to define the cleavage is based on that introduced previously [15], and ions marked with ÔhÕ are fragments produced by dehydration of the major ions type chain, whereas an 0,2A-ion doublet at m/z 281/263 indicates a type chain D-ions at m/z 348 or m/z 364 are characteristic of either terminal Lea or Lex determinants, respectively [13] We also established a method for corebranching pattern analysis using CID MS/MS of singly and doubly charged molecular ions [14] These spectra give complementary structural information In the CID spectra of [M-H]–, fragment ions from the 6-linked branch are dominant, and those from the 3-linked branch are absent, whereas fragment ions from both branches occur in the product-ion spectra of [M-2H]2– This allows us to distinguish between fragment ions derived from either the 3- or the 6-branch and to deduce the branching pattern and also assign structural details of the 3- and 6-branches Although MS is the more sensitive method of structural analysis, NMR spectroscopy is the choice for more complete assignment of carbohydrate structure when sufficient material is available In this report, we demonstrate our strategy of the combined use of ES-MS and NMR for analysis of the core-branching pattern and full sequence assignment of four oligosaccharides isolated from human milk These are variously fucosylated structures based on the iso-lacto-N-octaose core, of which three are novel sequences Materials and methods Isolation and purification of oligosaccharides Oligosaccharides were isolated from human milk obtained from a healthy 25-year-old woman, blood group B secretor, Leb positive, giving negative reaction in hepatitis B and HIV tests Written consent was obtained from this volunteer for analysis of the milk sample Fat was removed by centrifugation at °C (5000 g, 30 min) and proteins by precipitation with cold 50% (v/v) acetone Oligosaccharides were separated from lactose on a Sephadex G-25 column (5 · 90 cm), and then neutral Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1175 Fig Product-ion spectra of DFiLNO with [M-H]– (A) and [M-2H]2– (B) as precursors The structure of DFiLNO is shown to indicate the proposed fragmentation The nomenclature used to define the cleavage is based on that introduced previously [15], and ions marked with ÔhÕ are fragments produced by dehydration of the major ions from acidic oligosaccharides on a Dowex (· 2; 100–200 mesh; acetate form) column Further gel-filtration chromatography was carried out on a Fractogel HW-40(S) column (5 · 90 cm) Oligosaccharides were eluted from the gel-filtration and ion-exchange columns with distilled water The partially resolved octasaccharide to undecasaccharide fractions were further fractionated by normalphase HPLC on a preparative Separon amino column (10 · 250 mm) by elution with 50% (v/v) acetonitrile to give octasaccharide (F8), nonasaccharide (F9), decasaccharide (F10) and undecasaccharide (F11) fractions Each subfraction was purified by reverse-phase HPLC on a Zorbax octadecyl column (10 · 250 mm) by elution with water The octasaccharide iso-lacto-N-octaose (iLNO) was obtained from F8, monofucosyl iLNO (MFiLNO) from F9, difucosyl iLNO (DFiLNO) from F10, and trifucosyl iLNO (TFiLNO) from F11 Repeated normal-phase HPLC was carried out to ensure the purity of each oligosaccharide fraction before their analyses by ES-MS and 1H-NMR spectroscopy Methylation analysis After initial reduction with NaBD4, oligosaccharides were methylated, hydrolysed, reduced, and acetylated as described previously [16] GC-MS analysis of the partially methylated alditol acetates was performed on a ThermoQuest Trace system using a 15-m RTX-5 capillary column The initial column temperature was 50 °C programmed to 100 °C at 25 °CỈmin)1, to 220 °C at °CỈmin)1 and to 310 °C at 10 °CỈmin)1 ES-MS Negative-ion ES-MS and CID MS/MS were carried out on a Micromass (Manchester, UK) Q-Tof mass spectrometer Nitrogen was used as desolvation and nebuliser gas at a flow rate of 250 LỈh)1 and 15 LỈh)1, respectively Source temperature was 80 °C, and the desolvation temperature 150 °C Typically, a cone voltage of 80 V was used for CID MS/MS of singly charged ions [M-H]– and 50 V for Ó FEBS 2004 1176 H Kogelberg et al (Eur J Biochem 271) Fig Product-ion spectrum of D2b-5 ion m/z 1183 of DFiLNO as precursor (A) and the D2b-5 ion m/z 1037 of a contaminant in DFiLNO as the precursor (B) doubly charged ions [M-2H]2– The capillary voltage was maintained at kV Product-ion spectra were obtained from CID using argon as the collision gas at a pressure of 0.17 MPa The collision energy was adjusted to 23–43 V for optimal fragmentation and, typically, 40–43 V was used for CID of [M-H]–, and 23–27 V for [M-H]2– A scan rate of 1.5 s per scan was used for both ES-MS and CID MS/MS experiments, and the acquired spectra were summed for presentation For analysis, oligosaccharides were dissolved in acetonitrile/water (1 : 1, v/v), typically at a concentration of 5–10 pmolỈlL)1, of which lL was loop-injected Solvent (acetonitrile/1 mM ammonium bicarbonate, : 1, v/v) was delivered by a Harvard syringe pump (Harvard Apparatus, Holliston, MA, USA) at a flow rate of 10 lLỈmin)1 Alternatively, 0.5–1 lL sample solution was placed in a capillary needle for the nanospray experiment NMR spectroscopy Oligosaccharides were coevaporated with 2H2O (99.9 atom% 2H2) twice by lyophilization and dissolved in 500 lL high-quality 2H2O (100.0 atom% 2H2), containing 0.1 lL acetone NMR spectra were acquired on Varian (Palo Alto, CA, USA) Unity plus-500 (500.07 MHz 1H), Unity-600 (599.89 MHz 1H) and Unity-800 (800.27 MHz 1H) spectrometers and processed with standard Varian software The observed 1H chemical shifts were reported relative to internal acetone (2.225 p.p.m.) The NMR spectra were recorded at 15 °C for iLNO, 10 °C for MFiLNO, 13 °C for DFiLNO, and 27 °C for TFiLNO The temperatures were chosen in order to place the H2O signals with minimal disturbance to carbohydrate protons For MFiLNO, for example, a temperature of 10 °C placed the H2O signal optimally downfield from the Fuc IX H5 proton For TFiLNO, the same temperature would have resulted in total overlap of the H2O signal with the Fuc X and XI H5 protons Therefore the temperature of 27 °C was chosen for TFiLNO This placed the H2O signal between Fuc IX H5 and the GlcNAc V and VII H1 protons; nevertheless, the Fuc IX H5 was slightly obscured 2D phase-sensitive TOCSY spectra were recorded at mixing times of 10, 30, 50, 70 and 140 ms A spectral width of 3500 Hz was used in both dimensions, with eight scans per increment 2D phase-sensitive ROESY experiments were performed with a mixing time of 300 ms, a spectral width of 8000 Hz in both dimensions, and 16 scans per increment The spectrum offset was set 1.5 p.p.m to lower field of the most downfield shifted proton, Glc1a, to minimize TOCSY transfer The raw data sets of the homonuclear 2D experiments typically consisted of 4K · 256 complex data points Results Determination of branching pattern and blood group determinants by ES-MS/MS Negative-ion CID ES-MS/MS product ion scanning with the strategy established previously [13,14] was used first for analysis of the branching patterns and blood group determinants with low (picomole) amounts of the underivatized oligosaccharides The significant aspects of this strategy are the combined use of CID MS/MS of singly and doubly charged molecular ions as precursors to deduce the Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1177 Table Linkage and monosaccharide composition assignment from methylation analysis of milk oligosaccharides PMAA, partially methylated alditol acetate Molar ratios are relative to 1,5-di-O-acetyl-2,3,4,6-tetra-O-methylgalactitol –, Not detected Molar ratio PMAA Linkage iLNO MFiLNO DFiLNO TFiLNO Fucitol 1,5-di-O-acetyl Fuc1- – 0.8 1.8 3.3 Glucitol 4-mono-O-acetyl -4Glcol 1.2 1.2 1.4 0.7 Galactitol 1,5-di-O-acetyl 1,2,5-tri-O-acetyl 1,3,5-tri-O-acetyl 1,2,3,5-tetra-O-acetyl 1,3,5,6-tetra-O-acetyl Gal1-2Gal1-3Gal1-2,3Gal1-3,6Gal1- 2.0 – 0.4 – 1.0 2.0 – 1.2 – 1.2 2.0 – 0.9 – 0.9 2.0 – 0.9 – 1.7 N-Acetylglucosaminatol 1,5-di-O-acetyl 1,4,5-tri-O-acetyl 1,3,5-tri-O-acetyl 1,3,4,5-tetra-O-acetyl GlcNAc1-4GlcNAc1-3GlcNAc1-3,4-GlcNAc1- – 0.9 1.4 – – – 1.8 0.7 – – 1.4 2.1 – – – 3.1 branching pattern, sequence, and partial linkages, and assign structural details of each branch (e.g the 3- and 6-branches) Blood group H, Lea/Lex and Leb/Ley determinants, together with type and type chains, can be determined from their unique fragment ions iLNO The singly charged molecular ion [M-H]– at m/z 1436 (Fig 1A) and doubly charged [M-2H]2– at m/z 717.8 (Fig 1B) are consistent with an octasaccharide of composition Hex5HexNAc3 The approximate relative proportions of partially methylated alditol acetates (PMAAs) from methylation analysis (Table 1) are in agreement with this and indicate that the monosaccharide residues include one reducing terminal 4-linked Glc, four Gal (two terminal, one internal 3-substituted and one 3,6-disubstituted) and three GlcNAc (one 4-linked and two 3-linked) As established previously, the product-ion spectrum of [M-H]– only shows the fragment ions on the 6-linked branch The C-type ions C1a, C2a, C3a, C4a and C5 (see the fragmentation scheme in Fig 1) are in agreement with a 6-linked branch GalGlcNAc-Gal-GlcNAc-, while ions D2b-5 and 0,3A5 at m/z 891 and 819, respectively, define a 3,6-branched core Gal [14] The mass difference between the ion C5 (m/z 1274) and [M-H]– (m/z 1436) indicates a Hex at the reducing terminus The characteristic 0,2A6 doublet (m/z 1376/1358) and 2,4A6 (m/z 1316) ions confirm this to be a -4Glc [13,14] Thus, a reducing terminal core 3,6-disubstituted -Gal-4Glc can be proposed, in agreement with methylation analysis Furthermore, 3-substituted GlcNAc next to the nonreducing terminal Gal can be deduced from a D1a-2a ion [13] at m/z 202 while the HexNAc linked to the core Gal is deduced to be a -4GlcNAc from the unique 0,2A fragmentation (ions at m/z 646/628) Together with the monosaccharide linkage analysis data (Table 1), this further defines the sequence to be Gal1–3GlcNAc1–3Gal-4GlcNAc-6(Gal-GlcNAc-3)Gal4Glc From the knowledge that the product ion spectrum of [M-2H]2– shows fragment ions from both branches, and as no additional fragment ions are apparent, it can be deduced that the disaccharide sequence on the 3-branch shares the same terminal sequence Gal1–3GlcNAc- Taken together, these features indicate iLNO to be: MFiLNO The singly charged molecular ion [M-H]– at m/z 1582 (Fig 2A) and doubly charged [M-2H]2– at m/z 790.8 (Fig 2B) are consistent with a nonasaccharide of composition dHex1Hex5HexNAc3 The approximate relative proportions of PMAAs from methylation analysis (Table 1) are in agreement with this and indicate that the monosaccharide residues include Fuc, in addition to one reducing terminal 4-linked Glc, four Gal (two terminal, one internal 3-linked and one 3,6-disubstituted) and three GlcNAc (one 3,4-disubstituted and two 3-linked) The location of the Fuc residue can be deduced from the CID MS/MS fragmentation In comparison with the spectrum of iLNO (Fig 1A), the D2b-5 ion has shifted 146 Da to m/z 1037 corresponding to the presence of a fucose on the 6-branch The fragment ions at m/z 161, 382, 544 and 202 as B1a, C2a, C3a and D1a-2a, respectively, are the same as those in the spectrum of iLNO (Fig 1A) However, C4a has shifted from m/z 747 for iLNO to m/z 893, indicating the Fuc is at the internal 3,4-disubstituted GlcNAc The absence of a 0,2A4a doublet and the presence of a D4f-4a ion at m/z 729 (see the fragmentation scheme in Fig 2) are in agreement with a Fuc at the 3-position of a GlcNAc, indicating a Lex determinant Again, no additional fragment ions were revealed in the product ion spectrum of the doubly charged precursor (Fig 2B), confirming that the 3-linked disaccharide branch is the Ó FEBS 2004 1178 H Kogelberg et al (Eur J Biochem 271) same as the terminal disaccharide sequence of the 6-linked branch Hence, this monofucosylated octasaccharide, MFiLNO, can be assigned as: and the minor component as: DFiLNO The singly charged molecular ion at m/z 1729 and the doubly charged ion at m/z 863.8 in the spectrum of DFiLNO (Fig 3) are consistent with a difucosylated octasaccharide (dHex2Hex5HexNAc3) in agreement with methylation analysis (Table 1), which also shows a monosaccharide composition corresponding to MFiLNO with an additional Fuc It is apparent that both fucose residues are located in the 6-linked branch, as evidenced by the unique and intense D2b-5 ion at m/z 1183 (see the fragmentation scheme in Fig 3), from cleavage of the core residue Gal (Fig 3A), and by 0,3A5 at m/z 1111 The C2a ion at m/z 528 shows that one Fuc is at the subterminal GlcNAc, and the characteristic D1a-2a ion [13] at m/z 348 is consistent with a Fuc 4-linked to the GlcNAc of a terminal Lea determinant As the C3a ion is at m/z 690, this excludes the possibility of the other Fuc being at the Gal next to the subterminal GlcNAc The position of the second Fuc 3-linked to the internal GlcNAc forming an internal Lex determinant is deduced from the characteristic double cleavage D4f-4a ion at m/z 875 The ions at m/z 1037, m/z 729 and m/z 544 are similar to those in the spectrum of MFiLNO (Fig 2), in which only one Fuc is in the 6-branch, and believed to be from a contaminant (see below), having the same molecular mass with one Fuc in each of the 3- and 6-branches Comparison of the product-ion spectra of the singly and doubly charged molecular ion precursor (Fig 3A,B) show the major additional ions in the doubly charged ion spectrum to be m/z 202 and m/z 382 The former derives from a D1b-2b cleavage consistent with a 3-linked GlcNAc, and the latter from a glycosidic C2b cleavage, which supports assignment of the unsubstituted disaccharide sequence on the 3-branch The ion at m/z 803 is the doubly charged 2,4A6 fragment, which appears as a singly charged ion (m/z 1608) in the product-ion spectrum of the singly charged precursor (Fig 3A) The product-ion spectra of the doubly charged precursors generally give more intense fragment ion peaks The ions at m/z 1037, m/z 729 and m/z 544 appearing in both spectra are similar to those in the spectra of MFiLNO (Fig 2), in which only one Fuc is in the 6-branch, and, as indicated above, are believed to be from a contaminant with the same molecular mass with one Fuc in each of the 3- and 6-branches This is confirmed by MS/MS scanning of the D2b-5 ions m/z 1183 (Fig 4A) and m/z 1037 (Fig 4B) as precursors, produced by cone voltage fragmentation [14] The fragment ions of m/z 729 and m/z 544 are both from the monofucosylated D2b-5 ion m/z 1037 and not from the difucosylated D2b-5 ion m/z 1183 Thus the major component, the difucosylated branched octasaccharide, can be assigned as: TFiLNO The singly charged molecular ion [M-H]– at m/z 1874 (Fig 5A) and doubly charged [M-2H]2– at m/z 937.0 (Fig 5B) are consistent with a trifucosylated octasaccharide with composition Fuc3Hex5HexNAc3 The approximate relative proportions of PMAAs from methylation analysis (Table 1) are in agreement with this and indicate that the monosaccharide residues include three terminal Fuc1-, two terminal Gal1-, one reducing terminal -4Glc, together with internal monosubstituted and disubstituted residues: one -3Gal1-, one -3,6Gal1-, three -3,4GlcNAc1- The branching pattern and the locations of the three fucose residues can be deduced by similar reasoning to that for the other fucosylated analogues The product-ion spectrum of the singly charged precursor (Fig 5A) is very similar to that of DFiLNO (Fig 3A), indicating a similarly difucosylated tetrasaccharide sequence on the 6-branch The remaining Fuc is on the 3-branch, as indicated by C5 at m/z 1712 and D2b-5 at m/z 1183 (Fig 5) It can be deduced that the Fuc is 4-linked to the -3GlcNAc-, forming a second terminal Lea determinant, as no additional fragment ions are apparent in the spectrum of doubly charged precursor Hence the trifucosylated octasaccharide contains two terminal Lea and one internal Lex determinants with the sequence: Completion of sequence assignment by 1H NMR Homonuclear NMR was carried out to verify the MS assignment of the iLNO core structures and its variously fucosylated analogues, and to determine their anomeric configurations and linkages The proton chemical shifts of Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1179 Fig Product-ion spectra of TFiLNO with [M-H]– (A) and [M-2H]2– (B) as precursors The structure of TFiLNO is shown to indicate the proposed fragmentation The nomenclature used to define the cleavage is based on that introduced previously [15], and ions marked with ÔhÕ are fragments produced by dehydration of the major ions individual monosaccharide residues were assigned with the aid of TOCSY spectra with increasing mixing times (data not shown) Oligosaccharide sequences were established from interresidue rotating frame nuclear Overhauser enhancements (rOes) in combination with chemical shifts iLNO The monosaccharide composition of iLNO was shown by methylation analysis to comprise one Glc (reducing end), four Gal (two terminal) and three GlcNAc residues (see above) Their 1H chemical-shift assignments, obtained from TOCSY spectra (Fig 6), are given in Table Anomeric proton chemical shifts of four Gal residues are present between 4.47 and 4.427 p.p.m., while those of GlcNAc are between 4.71 and 4.634 p.p.m The reducing terminal Glc I residue (see structure in Fig and below) is indicated by the respective chemical shifts of the a-anomers and b-anomers at 5.22 and 4.665 p.p.m (Table 2) All residues are in b-anomeric linkages, as deduced from H1,H2 coupling constants between 8.0 and 8.3 Hz (Table 2) Sequence assignment of iLNO was derived from interresidue rOes (Table 2) and confirmed the MS assignment The reducing Glc I is deduced to be substituted at position 4, as the b-anomer of Gal II H1 gives an rOe to this proton Gal II is a branching point and substituted at positions and 6, as GlcNAc III H1 gives an rOe to H6a and H6b of Gal II, and GlcNAc VII H1 gives an rOe to the H3 of this residue The 3-branch is terminated by Gal VIII which is linked to the 3-position of GlcNAc VII, as Gal VIII H1 gives an rOe to GlcNAcVII H3 In addition, the GalVIIIb1–3GlcNAcVII linkage is supported by an rOe between GalVIII H1 and NAc protons of GlcNAc VII Extension of the 6-branch is by a GalIVb1–4GlcNAcIII linkage, as GalIV H1 gives an rOe to H4 of GlcNAc III Gal IV is further extended by a GlcNAcVb1–3GalIV linkage, as GlcNAc V H1 gives an rOe to Gal IV H3 Finally, a terminal Gal VI is linked b1–3 to GlcNAc V, as Gal VI H1 gives an rOe to GlcNAc V H3, and Gal VI H1 gives an rOe to the NAc protons of GlcNAc V 1180 H Kogelberg et al (Eur J Biochem 271) Ó FEBS 2004 Fig 1D and 2D 1H NMR spectra (800 MHz) of iLNO, region 5.5–3.0 p.p.m., at 15 °C Upper trace, 1H NMR spectrum; top-left half, 300-ms ROESY spectrum and bottom-right half, 140-ms TOCSY spectrum The structure is shown at the top, depicting the residue labelling The structure of iLNO is further supported from comparisons of chemical shifts with those reported previously for similar oligosaccharides The proton chemical shifts of the Gal II and Glc I residues and residues on the 3-branch (VII and VIII) are almost identical with those reported previously for an iso-lacto-N-octaose derivative that is difucosylated on the 6-branch [17] Furthermore the proton chemical shifts of the structural reporter protons of GlcNAc III are almost identical with those reported for the GlcNAcb1–6 residue of lacto-N-hexaose [18] Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1181 Table 1H chemical shifts, H1,H2 coupling constants, and intermolecular rOes from NMR spectra of iLNO Chemical shifts from a 1D 1H 800-MHz spectrum recorded at 15 °C are given to three decimals Other chemical shifts were taken from 2D spectra Chemical shifts in p.p.m and (H1,H2 coupling constants in Hz) Residue Linkages I Glca I Glcb II Gal III GlcNAc 6,4 IV Gal V GlcNAc VII GlcNAc VI Gal VIII Gal 4,6,4 3,4,6,4 3,4 3,3,4,6,4 3,3,4 a H1 H2 H3 H4 H5 H6a/b NAc 5.22 (3.6) 4.665 (8.0) 4.427(8.0) 4.652a 4.634b(8.0) 4.47(7.9) 4.71(8.3) 3.59 3.292 3.58 3.75 3.63 3.73 3.72 3.61 4.149 3.72 3.61 3.84 3.62 3.944/3.80 3.99/3.84 3.99/3.84 2.058 3.58 3.91 3.74 3.82 4.158 3.58 3.72 3.48 3.75 3.91/3.80 2.027 4.45(8.0) 3.523 3.642 3.91 3.71 rOes (from H1) 3.61(H4,Ib) 3.84(H6b,II); 3.99(H6a,II) 3.72(H3,H5a; H4III) 3.74(H3II,IV) 3.82(H3V,VII) 2.02(NAc,V,VII) Intramolecular rOes originating from Gal IV H1 Taken together these data allow assignment of the iLNO structure as follows: TFiLNO From methylation analysis, it was deduced that TFiLNO comprises four Gal residues, three GlcNAc residues, one reducing Glc residue, and three fucose residues (see above) 1H chemical-shift assignments of these residues (Table 3) were made from TOCSY spectra (see Fig 8) with increasing mixing times (data not shown) The Gal and GlcNAc residues are all in b-anomeric linkages as deduced from H1,H2 coupling constants between 7.8 and 8.0 Hz, while the fucose residues are a-linked to their neighbouring residues apparent from H1,H2 coupling constants of 3.7 and 3.8 Hz (Table 3) The sequence of the oligosaccharide was derived from interresidue rOes (Table 3, Figs and 8) which confirmed the MS assignment The reducing Glc residue is substituted at position 4; rOes are observed between Gal II H1 and Glc I H4 in the b-anomer Gal II is a branching point and substituted at positions and 6, as GlcNAc III H1 gives an rOe to H6b of this residue, while GlcNAc VII H1 gives an rOe to H3 of Gal II The 3-branch is further extended by a terminal Gal VIII residue, as Gal VIII H1 gives an rOe to GlcNAc VII H3 The Gal VIIIb1–3GlcNAc VII linkage is further supported by the rOe between Gal VIII H1 and the NAc protons of GlcNAc VII GlcNAc VII is also substituted by Fuc XI, which is linked to the 4-position of this residue, as Fuc XI H1 gives an rOe to GlcNAc VII H4 The terminal trisaccharide, Gal VIIIb1–3(Fuc XIa1–4)GlcNAc VII, the Lea epitope, also shows remote interresidue rOes of Fuc XI H5 and Fuc XI CH3 to Gal VIII H2 This is a characteristic rOe pattern for this epitope, resulting from stacking interactions of the Gal VIII and Fuc XI residues [19,20] The GlcNAc III on the 6-branch is substituted at positions and 4, as Gal IV H1 gives an rOe to H4 of GlcNAc III, and Fuc IX H1 gives an rOe to H3 and to the NAc protons of this residue The trisaccharide, Gal IVb1– 4(Fuc IXa1–3)GlcNAc III, the Lex epitope, is further characterized by remote rOes between Fuc IX H5 and Gal IV H2 and between Fuc IX CH3 and Gal IV H2 The latter rOes are similar to those seen for the Lea epitope (see above), as the Lex epitope also shows stacking interaction between the Fuc IX and Gal IV residues, resulting from very similar conformational features ([21] and references therein) Gal IV on the 6-branch is further extended at the 3-position, as GlcNAc V shows an rOe to H3 of Gal IV GlcNAc V is substituted at the 3- and 4-position, as Gal VI H1 gives an rOe to GlcNAc V H3 and to the NAc protons of GlcNAc V, while Fuc X H1 gives an rOe to GlcNAc V H4 This second Lea epitope (see also 3-branch above) also shows remote rOes between H5 and CH3 of Fuc X and H2 of Gal VI The TFiLNO structural assignment is further supported by similar chemical shifts observed for the protons of residues on the 6-branch to those previously reported for an iso-lacto-N-octaose that contains this difucosylation on the 6-branch [17] Thus, TFiLNO has the following structure: MFiLNO The monosaccharide composition of MFiLNO was shown by methylation analysis to consist of four Gal residues, three GlcNAc residues, one reducing Glc residue, and one fucose residue (see above) 1H chemical-shift assignments of these residues (Table 4) were made from TOCSY spectra The Gal and GlcNAc residues are involved Ó FEBS 2004 1182 H Kogelberg et al (Eur J Biochem 271) Table 1H chemical shifts, H1,H2 coupling constants, and intermolecular rOes from NMR spectra of TFiLNO Chemical shifts from a 1D 1H 600MHz spectrum recorded at 27 °C are given to three decimals Other chemical shifts were taken from 2D spectra Chemical shifts in p.p.m and (H1,H2 coupling constants in Hz) Residue Linkages H1 H2 H3 H4 H5 H6a/b I Glca I Glcb II Gal III GlcNAc IV Gal 6,4 4,6,4 5.218(3.7) 4.665(8.0) 4.427(7.8) 4.639(7.8) 4.433(8.0) 3.59 3.287 3.58 3.91 3.52 3.65 3.71 3.91 3.71 3.61 4.138 3.91 4.098 3.61 3.83 3.60 3.59 3.94/3.80 3.98/3.83 3.71 V GlcNAc VII GlcNAc VI Gal VIII Gal 3,4,6,4 3,4 3,3,4,6,4 3,3,4 4.697(8.4) 3.96 4.08 3.77 3.55 3.88 4.515/4.503 (7.8) 3.49 3.620 3.89 3.58 IX Fuc 3,6,4 5.090(3.8) 3.70 3.88 3.775 4.806 1.150 X Fuc XI Fuc 4,3,4,6,4 4,3,4 5.030(3.8) 3.80 3.88 3.79 4.873 1.180 a NAc 2.049 2.030 rOes (from H1) 3.61(H4b,I) 3.83(H6b,II) 3.91(H4III) H4a:3.59 (H5), 3.71(H6a, 6b) 3.71(H3II,IV) 4.08(H3V,VII) 3.96(H2V,VII) 2.03(NAc, V,VII) 3.91(H3III) 2.05(NAc III) H5 : 3.52 (H2IV) CH3 : 3.52 (H2IV) 3.77(H4V,VII) H5 : 3.49 (H2VI, VIII) CH3 : 3.49 (H2VI, VIII) Intramolecular rOes originating from Gal IV H4 in b-anomeric linkages as deduced from H1,H2 coupling constants between 7.7 and 8.1 Hz, while the fucose residue is a-linked to its neighbouring residue, deduced from the H1,H2 coupling constant of 3.7 Hz (Table 4) The sequence of the oligosaccharide was derived from interresidue rOes (Table 4) and fully confirmed the MS assignment The reducing Glc residue is substituted at position 4, as Gal II H1 of the b-anomer shows an rOe to H4 of this residue Gal II is a branching point residue and substituted at positions and 6, as GlcNAc III H1 gives rOes to H6b of Gal II, while GlcNAc VII H1 gives an rOe to H3 of Gal II The 3-branch is further extended by a terminal Gal VIII, as Gal VIII H1 gives an rOe to GlcNAc VII H3 The GlcNAc III on the 6-branch is substituted at the 3- and 4-position, as Gal IV H1 gives an rOe to H4 of GlcNAc III, and Fuc IX H1 gives an rOe to H3 and to the NAc protons of this residue The trisaccharide Lex epitope, Gal IVb1–4(Fuc IXa1–3)GlcNAcIII, is also characterized by rOes between Fuc IX H5 and Gal IV H2 and between Fuc IX CH3 and Gal IV H2 (see also TFiLNO above) Gal IV is substituted at the 3-position, as GlcNAc V H1 gives an rOe to Gal IV H3, and the GlcNAc V residue is substituted by a terminal Gal VI residue, as Gal VI H1 gives an rOe to GlcNAc V H3 The MFiLNO assignment is further supported by chemical-shift comparisons with the nonfucosylated iLNO The two oligosaccharides have very similar chemical shifts for the reducing Gal IIb1–4Glc residues, for the 3-branch and for the GalVIb1–3GlcNAcV residues Chemical shifts for the residues forming the Lex epitope are very similar to those residues in the TFiLNO structure The structure of MFiLNO is thus assigned as: DFiLNO The methylation analysis for DFiLNO identified a decasaccharide The chemical shifts of anomeric protons of the reducing terminal Glc at 5.216 and 4.67 p.p.m for a and b anomers, respectively, four Gal residues at 4.42 (II), 4.43 (IV), 4.515 (VI) and 4.44 (VIII) p.p.m., three GlcNAc residues at 4.632 (III), 4.674 (V) and 4.717 (VII), and two Fuc residues at 5.094 (IX) and 5.035 (X) (data not shown), are very similar to those reported previously for DFiLNO [17,22] Hence the structure of the main component is as shown below and in accord with the MS assignment A minor component (see above) is not apparent from the NMR spectrum because of overlapping chemical shifts with the major component Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1183 Fig 1D and 2D 1H-NMR spectra (600 MHz) of TFiLNO at 27 °C Upper trace, the 5.5–0.5 p.p.m region of the 1H-NMR spectrum; bottom trace, the 5.5–0.5 by 2.5–0.5 p.p.m regions of the 300-ms ROESY spectrum The structure is shown to indicate the residue labelling Discussion We have isolated various fucosylated neutral oligosaccharides from human milk that are based on the iso-lactoN-octaose core, and fully characterized four structures by the combined use of ES-MS, methylation analysis and NMR spectroscopy The branching pattern and blood group-related Lewis determinants can be readily determined by high-sensitivity ES-MS/MS analysis as demonstrated here and also previously [13,14] Even a minor isomeric component present in a mixture can be detected, as exemplified by the analysis of DFiLNO NMR is currently the only technique that can provide the full sequence information, including linkage and a,b-anormeric configuration However, because of the complexity of overlapping signals of the same monosaccharide residues in a very similar chemical environment, it is difficult for NMR alone to assign the branching pattern The combined use of product-ion scanning of negative-ion ES-MS and NMR has proved a powerful strategy for complete assignment of the branched structures NMR 1184 H Kogelberg et al (Eur J Biochem 271) Ó FEBS 2004 Fig 1D and 2D 1H-NMR spectra (600 MHz) of TFiLNO, region 5.5–3.0 p.p.m., at 27 °C Upper trace, 1D 1H spectrum; Top-left half, 300-ms ROESY spectrum and bottom-right half, 140-ms TOCSY spectrum The structure is shown to indicate the residue labelling chemical shifts have been assigned for all compounds and this opens the way to their conformational analysis by NMR Several variously fucosylated neutral iso-lacto-N-octaose derivatives have been described previously, including two difucosylated [17], one trifucosylated [23], one tetrafucosyl- ated [24], and one pentafucosylated [24] oligosaccharides Acidic structures described include one monosialyl-monofucosylated [25], two monosialyl-difucosylated [25] and one monosialyl-trifucosylated [26] oligosaccharides In the present study, three structures have been isolated for the first time The nonfucosylated iso-lacto-N-octaose has not been Ó FEBS 2004 Sequence determination of oligosaccharides (Eur J Biochem 271) 1185 Table 1H chemical shifts, H1,H2 coupling constants, and intermolecular rOes from NMR spectra of MFiLNO IV Chemical shifts from a 1D 1H 500-MHz spectrum recorded at 10 °C are given to three decimals Other chemical shifts were taken from 2D spectra Chemical shifts in p.p.m and (H1,H2 coupling constants in Hz) Residue Linkages H1 H2 H3 H4 H5 H6a/b I Glca I Glcb II Gal 5.214(3.9) 4.663(8.1) 4.43(7.8) 3.60 3.285 3.57 3.64 3.72 3.61 4.15 3.64 3.82 3.95 3.99/3.84 III GlcNAc IV Gal 6,4 4,6,4 4.64(7.7) 4.44(7.8) 3.90 3.53 3.90 3.72 3.90 4.11 3.58 V GlcNAc VII GlcNAc VI Gal VIII Gal IX Fuc 3,4,6,4 3,4 3,3,4,6,4 3,3,4 3,6,4 4.708(7.8) 3.90 3.80 3.59 3.48 4.45(7.8) 3.51 3.64 3.91 5.091(3.7) 3.68 3.88 3.77 a NAc rOes (from H1) 2.049 3.61(H4b,I) H4a:3.99,3.84 (H6a,H6b) 3.84(H6b,II) 3.90(H4III) 2.02(NAc,V) 3.72(H3II, IV) 2.022 3.80(H3V, VII) 4.824 1.145 3.90(H3III) 2.05(NAc,III) H5 : 3.53(H2, IV) CH3 : 3.53(H2IV) Intramolecular rOes originating from Gal II H4 reported as an individual oligosaccharide, although it has been described previously as a core type [24] based on the identification of its fucosylated derivatives The monofucosylated iso-lacto-N-octaose, containing a Fuc at the first GlcNAc residue of the 6-branch forming an internal Lex epitope, and the trifucosylated iso-lacto-N-octaose, containing two terminal Lea epitopes at each of the branches and one internal Lex on the 6-branch are also novel structures Acknowledgements We thank Dr Colin Herbert for his technical support This work was supported by a UK Medical Research Council programme grant (G9601454) NMR experiments and analysis were carried out at the MRC Biomedical NMR centre, National Institute of Medical Research, London References Gottschalk, A (1954) The influenza virus enzyme and its mucoprotein substrate Yale J Biol Med 26, 352–364 Watkins, W.M (1972) Glycoproteins: their composition, structure and function In Blood-Group Specific Substances (Gottschalk, A., ed.), pp 830–899 Elsevier, Amsterdam Kabat, E.A (1982) Contributions of quantitative immunochemistry to knowledge of blood group A, B, H, Le, I and i antigens Am J Clin Pathol 78, 281–292 Feizi, T (2000) Carbohydrate-mediated recognition systems in innate immunity Immunol Rev 173, 79–88 Feizi, T (1985) Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are oncodevelopmental antigens Nature 314, 53–57 Newburg, D.S & Neubauer, S.H (1995) Carbohydrates in milks: analysis, quantities and significance Handbook of Milk Composition, pp 273–349 Academic Press, San Diego Yamashita, K., Mizuochi, T & Kobata, A (1982) Analysis of oligosaccharides by gel filtration Methods Enzymol 83, 105–126 Kunz, C., Rudloff, S., Baier, W., Klein, N & Strobel, S (2000) Oligosaccharides in human milk: structural, functional, and metabolic aspects Annu Rev Nutr 20, 699–722 Rudloff, S., Stefan, C., Pohlentz, G & Kunz, C (2002) Detection of ligands for selectins in the oligosaccharide fraction of human milk Eur J Nutr 41, 85–92 10 Martin, M.J., Feizi, T., Leteux, C., Pavlovic, D., Piskarev, V.E & Chai, W (2002) An investigation of the interactions of E-selectin with fuco-oligosaccharides of the blood group family Glycobiology 12, 829–835 11 DeFrees, S.A., Kosch, W., Way, W., Paulson, J.C., Sabesan, S., Halcomb, R.L., Huang, D.H., Ichikawa, Y & Wong, C.H (1995) Ligand recognition by E-selectin: synthesis, inhibitory activity, and conformational analysis of bivalent sialyl Lewis x analogs J Am Chem Soc 117, 66–79 12 Chai, W., Feizi, T., Yuen, C.-T & Lawson, A.M (1997) Nonreductive release of O-linked oligosaccharides from mucin glycoproteins for structure/function assignments as neoglycolipids: application in the detection of novel ligands for E-selectin Glycobiology 7, 861–872 13 Chai, W., Piskarev, V & Lawson, A.M (2001) Negative-ion electrospray mass spectrometry of neutral underivatized oligosaccharides Anal Chem 73, 651–657 14 Chai, W., Piskarev, V & Lawson, A.M (2002) Branching pattern and sequence analysis of underivatized oligosaccharides by combined MS/MS of singly and doubly charged molecular ions in negative-ion electrospray mass spectrometry J Am Soc Mass Spectrom 13, 670–679 15 Domon, B & Costello, C.E (1988) A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates Glycoconjugate J 5, 397–409 16 Chai, W., Hounsell, E.F., Cashmore, G.C., Rosankiewicz, J.R., Bauer, C.J., Feeney, J., Feizi, T & Lawson, A.M (1992) Neutral oligosaccharides of bovine submaxillary mucin: a combined mass spectrometry and 1H-NMR study Eur J Biochem 203, 257–268 17 Strecker, G., Wieruszeski, J.M., Michalski, J.C & Montreuil, J (1989) Primary structure of human milk nona- and decasaccharides determined by a combination of fast atom bombardment 1186 H Kogelberg et al (Eur J Biochem 271) 18 19 20 21 mass spectrometry and 1H-/13C-nuclear magnetic resonance spectroscopy Evidence for a new core structure, iso-lactoN-octaose Glycoconj J 6, 169–182 Dua, V.K., Goso, K., Dube, V.E & Bush, C.A (1985) Characterization of lacto-N-hexaose and two fucosylated derivatives from human milk by high-performance liquid chromatography and proton NMR spectroscopy J Chromatogr 328, 259–269 Lemieux, R., Bock, K., Delbaere, L., Koto, S & Rao, V (1980) The conformations of oligosaccharides related to the ABH and Lewis human blood group determinants Can J Chem 58, 631–653 Bechtel, B., Wand, A.J., Wroblewski, K., Koprowski, H & Thurin, J (1990) Conformational analysis of the tumor-associated carbohydrate antigen 19–9 and its Lea blood group antigen component as related to the specificity of monoclonal antibody CO19-9 J Biol Chem 265, 2028–2037 Kogelberg, H., Frenkiel, T.A., Homans, S.W., Lubineau, A & Feizi, T (1996) Conformational studies on the selectin and natural killer cell receptor ligands sulfo- and sialyl-lacto-N-fucopentaoses (SuLNFPII and SLNFPII) using NMR spectroscopy and molecular dynamics simulations Comparisons with the nonacidic parent molecule LNFPII Biochemistry 35, 1954–1964 Ó FEBS 2004 22 Sabharwal, H., Nilsson, B., Gronberg, G., Chester, M.A., Dakour, J., Sjoblad, S & Lundblad, A (1988) Oligosaccharides from feces of preterm infants fed on breast milk Arch Biochem Biophys 265, 390–406 23 Strecker, G., Fievre, S., Wieruszeski, J.M., Michalski, J.C & Montreuil, J (1992) Primary structure of four human milk octa-, nona-, and undeca-saccharides established by 1H- and 13C-nuclear magnetic resonance spectroscopy Carbohydr Res 226, 1–14 24 Haeuw-Fievre, S., Wieruszeski, J.M., Plancke, Y., Michalski, J.C., Montreuil, J & Strecker, G (1993) Primary structure of human milk octa-, dodeca- and tridecasaccharides determined by a combination of 1H-NMR spectroscopy and fast-atom-bombardment mass spectrometry Evidence for a new core structure, the para-lacto-N-octaose Eur J Biochem 215, 361–371 25 Kitagawa, H., Nakada, H., Fukui, S., Funakoshi, I., Kawasaki, T., Yamashina, I., Tate, S & Inagaki, F (1991) Novel oligosaccharides with the sialyl-Lea structure in human milk Biochemistry 30, 2869–2876 26 Kitagawa, H., Nakada, H., Fukui, S., Funakoshi, I., Kawasaki, T., Yamashina, I., Tate, S & Inagaki, F (1993) Novel oligosaccharides with the sialyl-Le (a) structure in human milk J Biochem (Tokyo) 114, 504–508  ... produced by cone voltage fragmentation [14] The fragment ions of m/z 729 and m/z 544 are both from the monofucosylated D2b-5 ion m/z 1037 and not from the difucosylated D2b-5 ion m/z 1183 Thus the. .. iLNO core structures and its variously fucosylated analogues, and to determine their anomeric configurations and linkages The proton chemical shifts of Ó FEBS 2004 Sequence determination of oligosaccharides. .. previously as a core type [24] based on the identification of its fucosylated derivatives The monofucosylated iso-lacto-N-octaose, containing a Fuc at the first GlcNAc residue of the 6-branch forming