Glycobiology vol. 21 no. 2 pp. 162174, 2011 doi:10.

1093/glycob/cwq143 Advance Access publication on October 6, 2010

The structural basis of blood group A-related glycolipids in an A3 red cell phenotype and a potential explanation to a serological phenomenon

Lola Svensson 1,2, Laura Bindila 3,4, Jonas ngstrm 5, Bo E Samuelsson 2, Michael E Breimer 5, Lennart Rydberg 2, and Stephen M Henry 6
2 Department of Clinical Chemistry and Transfusion Medicine, Institute of Biomedicine, The Sahlgrenska Academy at University of Gothenburg, Sahlgrenska University Hospital, Sahlgrenska Campus, 413 45 Gothenburg, Sweden; 3Institute for Medical Physics and Biophysics, University of Mnster, 48149 Mnster, Germany; 4Luxembourg Clinical Proteomics, Centre de Recherche Public, Strassen L-1445, Luxembourg; 5Department of Surgery Institute of Clinical Sciences, The Sahlgrenska Academy at University of Gothenburg, Sahlgrenska University Hospital, stra Campus, 416 85 Gothenburg, Sweden; and 6Biotechnology Research Institute, AUT University and KODE Biotech Ltd, Auckland 1142, New Zealand

Keywords: blood group A / glycolipid / immunochemistry / mass spectrometry / para-Forssman

Downloaded from http://glycob.oxfordjournals.org/ by guest on October 8, 2013

Introduction Blood group ABO antigens are present both as glycolipids and as glycoproteins on the red cell membrane. These antigens are quantitatively expressed in a serologically dened manner, with 80% of blood group A Europeans having very strong A antigen expression and 20% having moderate A antigen expression, representative of the A1 and A2 red cell phenotypes, respectively (Economidou et al. 1967; Schachter et al. 1973; Schenkel-Brunner 2000). A series of much lessfrequent weak A subgroup red cell phenotypes each with decreasing blood group A antigen from Aint, A3, Ax, Aend, Am, Ann, Abantu, Alae, Ay to Ael are also recognized. The N-acetylgalactosaminyl transferase encoded by the blood group A gene was mapped in 1990 (Yamamoto et al. 1990), and thereafter studies have established an enormous heterogeneity between and within the subgroups (Olsson and Chester 1996; Olsson et al. 2001). In most cases, subgroups are due to recognized genetic mutations (Olsson et al. 2001); however, not all subgroups can be adequately explained by mutations within the ABO gene. In particular, with the exception of the A3 sample in this paper, all other A3 phenotyping individuals reported to date do not have a recognized ABO gene mutation (Olsson et al. 2001). Unlike the relatively well-investigated qualitative and quantitative differences in A antigen expression between the common A1 and A2 red cell phenotypes (Schachter et al. 1973; Clausen, Levery, Nudelman et al. 1985; Svensson et al. 2009), the qualitative differences of the weak A subgroups are still largely unexplored. Studies with electron microscopy of immunogold-labelled red cells and immunochemistry of glycolipids indicate that the weak A subgroups differ from A1 and A2 both in the number of antigens present and by their chemical nature (Heier et al. 1994; Svensson et al. 2005). The weak A subgroup known as the A3 phenotype is one of the more unusual A subgroups as it is dened by the serological reaction referred to as mixed-eld agglutination: small agglutinates in a sea of free unaggutinated cells (Klein and Anstee 2005). One recent ow cytometry study also showed that there appears to be two populations of A cells in the A3 phenotype, although the A3 sample in this paper (albeit with 162

Received on July 23, 2010; revised on August 20, 2010; accepted on September 6, 2010

Glycolipids from the red cells of a rare blood group A subgroup individual, expressing the blood group A3 phenotype with the classical mixed-eld agglutination phenomenon, A2(539G>A)/O1 genotype, and an unusual blood group A glycolipid prole, were submitted to a comprehensive biochemical and structural analysis. To determine the nature of blood group A glycolipids in this A3 phenotype, structural determination was carried out with complementary techniques including proton nuclear magnetic resonance (1D and 2D), mass spectrometry (MS) (nano-electrospray ionization/quadrupole time-of-ight and tandem mass spectrometry) and thin layer chromatography with immunostaining detection. As expected, total blood group A structures were of low abundance, but contrary to expectations extended-A type 2 and A type 3 glycolipids were more dominant than A hexaglycosylceramides based on type 2 chain (A-6-2 glycolipids), which normally is the major A glycolipid. Several para-Forssman (GalNAc3GbO4) structures, including extended forms, were identied but surmised not to contribute to the classic mixed-eld agglutination of the A3 phenotype. It is proposed that the low level of A antigen combined with an absence of extended branched glycolipids may be the factor determining the mixed-eld agglutination phenomenon in this individual.

1 To whom correspondence should be addressed: Fax: +46-31-417631; e-mail: lola.svensson@gu.se

The Author 2010. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Glycolipid structures in an A3 red cell phenotype

a different genetic basis) in the same study was an exception with only a single population of A cells (Hult and Olsson 2010). Further evidence to support dual populations in the A3 phenotype comes from the observation that A glycosyltransferase activity could only be found in the stroma of A3 agglutinates (Nakamura et al. 1989). Interestingly, early studies of A3 mixed-eld reactions found that the unagglutinated cells were agglutinable when retested with anti-A (Cartron et al. 1977), and despite no genetic evidence for an aberrant glycosyltransferase, 13 A3 individuals were found to have an A glycosyltransferase with a different pH optima than that of the A1 glycosyltransferase (Nakamura et al. 1989). Determining the structural basis of unusual phenotypes, or the insight they may give of background structures not usually accessible for identication in the presence of normal and dominating antigens, is of importance to understanding the biology of blood group antigens. Subgroups provide a unique insight into the biosynthesis of ABO antigens, or at least alternative biosynthetic pathways which may operate under certain biological pressures. This paper describes a comprehensive biochemical and structural analysis of glycolipids isolated from one individual with a well-dened A3 phenotype. The structural analysis is based on complementary structural proling involving nuclear magnetic resonance (NMR), mass spectrometry (MS) and thin layer chromatography (TLC) techniques. The data corroborated a previous study of this individual (Aweak-II), which revealed an A3 phenotype, with a new genetic mutation (A304 in Blood Group Antigen Gene Mutation Database; Blumenfeld and Patnaik 2004), and an unusual total neutral glycolipid prole (Svensson et al. 2005). To distinguish the sample of this paper from the undened ABO genetic background A3 samples, hereafter it is termed the A304 A3 phenotype. Results Conrmation of mixed-eld agglutination phenomenon The red cells of the A304 A3 sample were subjected to extended differential serology to prove the nature of the mixed-eld agglutination (Figure 1). As expected, the A304 A3 sample (lane I) showed the classic mixed-eld reaction of both agglutinating and non-agglutinating cells. When the agglutinating (lane II) and non-agglutinating cells (lane III) were puried and then retested with new reagents, both again showed the mixed-eld agglutination reaction. Total glycolipid pattern by TLC and TLCenzyme immunoassay analysis Analysis of the total neutral glycolipids extracted from the six blood unit donations (Figure 2, lanes T) revealed a TLC prole qualitatively identical to the original single donation (Aweak-II) extract (Figure 2, lane T*). In comparison with the control A1 Le(ab) ABH-secretor total glycolipids (Figure 2, lane A1) both Aweak-II samples (T* and T) reacted similarly against anti-A 222, albeit quantitatively reduced in reactivity, despite being loaded at a 2.5 times higher concentration. However, against anti-A type 2 reactive monoclonal antibodies (MAbs) 2-24 and HH4, staining was unambiguously reduced. Anti-AB 2-41 was unreactive in the six-sugar region

Fig. 1. Mixed-eld agglutination serology. A304 A3 red cells are shown to have typical mixed-eld agglutination (a mixture of positive (+) agglutinated and unagglutinated negative () cells). Retesting of the agglutinated cells (lane II) and the unagglutinated cells (lane III) from the original reaction (lane I) again produced mixed-eld agglutination for both populations of cells.

but showed signicant reactivity in the extended glycolipid regions in agreement with previously reported data (Svensson et al. 2005). Reactivity against three anti-B reactive reagents conrmed this reactivity was not due to B structures (results not shown). The anti-A type 3 reactive MAb TH1 revealed no bands in the regions expected for A type 3 glycolipids. Reactivity against anti-H type 3 MAb (HH14) and anti-H type 2 (BE2) showed a pattern similar to a blood group A2 phenotype control (result not shown). Chemical staining of the medium-pressure liquid chromatography fractions The medium-pressure liquid chromatography (MPLC) separation of the total neutral glycolipids resulted in seven fractions, each containing mixtures of complex glycolipids of increasing size, and persistent contamination with globoside (GbO4) the major glycolipid component in red cells, yet were sufciently pure for MS and NMR analyses. Anisaldehyde semi-quantitative chemical staining (Figure 3, plate I) showed that fraction 1 included glycolipids with 46 sugar residues, whereas fractions 2 and 3 mainly contained 712 residues. The dominant structures in fractions 4 and 5 were glycolipids with 8 to approximately 12 sugar residues. Fractions 6 and 7, although poorly resolved, represented the more extended glycolipids with more than 10 sugar residues. Table I shows a summary of the representative glycolipids and weight of each of the seven fractions. TLCenzyme immunoassay of the MPLC fractions The immunochemical staining was repeated on the MPLC fractions at a loading concentration of 10 g per lane. 163

L Svensson et al.

Fig. 2. TLCEIA of total neutral glycolipids from the Aweak-II A3 phenotype donor (A304 A3) immunostained with selected blood group A reactive MAbs. The numbers to the right denote the approximate number of carbohydrate residues of the glycolipids in the corresponding band. Lane A1, common blood group A1 phenotype control, 20 g loading; lane T*, the original Aweak-II glycolipid preparation, 50 g loading; lane T, new Aweak-II glycolipid preparation from six pooled donations, 50 g loading. Individual plates were overlaid with either MAb anti-A 2-22, 2-24, anti-A type 2 HH4, anti-AB 2-41 or anti-A type 3 TH1, respectively, and then compiled to create the gure.

Fraction 1 (Figure 3, plate II, lane 1), containing the four to six-sugar-residue glycolipids, stained in the six-sugar region (relative migration, rm 6) with two monoclonal A antigen reactive reagents and the lectin Helix pomatia. With due consideration to the semi-quantitative nature of TLCenzyme immunoassay (EIA), there was signicant contrast in band intensity between the strong six-sugar bands with Lorne/ H. pomatia and the moderate band with anti-A 2-24 (Figure 3, II) and the absence of reactivity with anti-AB 241. Compared with the unfractionated total samples (Figure 2), the concentrated glycolipids in the MPLC fractions (Figure 3, II and III) could be more condently immunochemically dened, particularly the presence of A type 2 (Hakomori 1981) and A type 3 glycotopes (Clausen, Levery, Nudelman et al. 1985; Clausen et al. 1986) (Figure 3, III). As expected, MAb 2-22 was strongly reactive in all fractions (Figure 3, III) and reactivity could be seen indicative of blood group A with 8 (rm 3) and 10/12 (rm 1 0.5) sugar residues and probably para-Forssman (GalNAc3GbO4, rm 6 0.5) (Ando et al. 1982). Identity of these structures as A type 2 glycolipids was indicated by their reactivity with the anti-A type 2 reactive MAb (224). In addition, very weak reactions seen with anti-A type 3 MAb TH1 suggest the presence of some A-9-3 (rm 3.0) and A-11-3 (rm 0.9) glycolipids in fractions 4 and 5, respectively (Figure 3, III). Expected glycolipid reaction patterns (when compared with the A2 phenotype) were observed with anti-H type 2 and H type 3 reactive MAbs (results not shown). Structural analysis of A glycolipids by nano-electrospray ionization-quadrupole time-of-ight tandem mass spectrometry By nano-electrospray ionization (ESI)-quadrupole time-ofight (QTOF) MS screening, a number of blood group A glycolipid-related ions were detected in fractions 1, 4 and 5 as illustrated in Figures 46 and Table II. In total, 19 blood group A glycolipid-related ions based on compositional assignment 164

were detected, indicative of A-6, A-8, A-9, A-10 and A-11 glycolipids. The evaluation of glycolipid molecular ion patterns provided evidence that A-8 structures dominate in number and abundance. The applied ionization conditions prompted the formation of doubly charged ions with a minimal, if any, in-source decay of glycosidic bonds. The A-6 glycolipid (Table II) was exclusively detected in its singly charged form. All ion species detected in the MS1 prole and postulated based on the compositional assignment to correspond to A glycolipids (cf. also Table II) were conrmed by collision-induced dissociation (CID) tandem mass spectrometry (MS/MS) fragmentation analysis. Basically, A-8, A-9 and A-10 type glycolipids were found to dominate the mixture. Sensitivity of the analysis was sufciently high to allow the detection of very low abundant species, such as A-6 glycolipids. Nano-ESI-QTOF MS identication of A-6 was initially based on compositional analysis of corresponding singly charged ions at m/z 1711.11 assigned to FucHexNAc2 Hex3Cer(d18:1/24:0). Accurate structural elucidation of the ions at m/z 1711.11 was achieved by CID MS/MS experiments (Figure 4) in the second-stage analysis. The resultant fragmentation pattern is unambiguously consistent with the A-6 structure having d18:1/24:0 Cer (Figure 4, inset scheme). The presence of a terminal fucose (Fuc) residue is easily deducible from the immediate loss of 146 (atomic mass units) from the molecular ions, represented at m/z 1565.02. The fragmentation pathway of the ions at m/z 1565.02 follows a stepwise glycosidic bond cleavage, starting with the monosaccharide residues from the nonreducing end up to the Cer residue, and dening this way the sequence of the main HexNAc-Hex-HexNAc-Hex-Hex-Cer glycosidic backbone. The presence of the ions at m/z 1507.99 corresponding to a neutral loss of the HexNAc residue from the precursor ions clearly indicates the location of HexNAc at the nonreducing end terminus and, at the same time, that there exists a branching point for Fuc attachment within the remaining glycan chain. Given the presence of Y3 type ions at m/z 1199.83, which denes the

Glycolipid structures in an A3 red cell phenotype

Fig. 3. TLC analysis of the MPLC-puried glycolipid fractions. Lane numbers on plates equate with fraction numbers. Plate I, anisaldehyde chemical staining. Plates IIIII, immunostaining of fractions at a 10 g per lane loading but with A1 and A2 phenotype controls loaded at 20 g per lane due to the lower relative abundance of individual glycolipid species in total fractions. Plate IV, GalNAc3GbO4 activity mapping of MAbs 2-22, 2-24 and 2-41. Plate II shows a comparison of the common A1 and A2 phenotyping controls and fraction 1 (containing the hexaglycosylceramides) and their respective reactivity against MAb anti-A Lorne, anti-AB MAb 241, anti-A type 2 MAb 224 and the lectin H. pomatia. Plate III shows the presence of various A antigens in the MPLC fractions as detected with the broadly reactive anti-A MAb 2-22 and specic A type 2 MAb 2-24 and A type 3 (MAb TH1) reagents. The control lane (A1) to the right is blood group A1 glycolipids stained with MAb anti-A Lorne. Plate IV is GalNAc3GbO4 immunostained with selected anti-A and anti-AB MAbs. Lane A1, common blood group A1 phenotype control at 20 g per lane; lane p-F, puried GalNAc3GbO4 loaded at 6 g per lane. Plates are immunostained with the broadly reactive MAb anti-A 2-22, anti-A type 2 reactive 2-24 and anti-AB 2-41.

HexNAc-Hex-Hex-Cer(d18:1/24:0) structure, and its counterpart B2-type ions at m/z 534.17, the attachment of the Fuc at the terminal Gal residue is clearly revealed. Additionally, the B3-type ions at m/z 737.27, consistent with the HexNAc(Fuc)Hex-HexNAc structure, and B4-type ions at m/z 899.63,

along with their counterpart fragments, support further the structural architecture of the glycan chain as consistent with an A-6 glycolipid. The identity of the Cer moiety is directly revealed by the Y0-type ions at m/z 672.62 consistent with d18:1/24:0 and the series of Y1-Y3 type fragment ions, 165

L Svensson et al.

Table I. The weight and dominant range of glycolipids in the MPLC fractions MPLC fractions Weight (mg) Carbohydrate per glycolipida 46 79 812 812 1012 1012 12

1 2 3 4 5 6 7
a

15 2 1 0.5 0.7 <0.1b <0.1b

Approximate range of carbohydrate residues per dominant glycolipid (excluding GbO4 contamination). b Approximate weight (mg) of glycolipids after excluding GbO4 contamination.

whereas the N-type ions at m/z 264.28 correlate with the d18:1 long-chain base of the Cer part. Under the fragmentation conditions applied, a number of fragment ions arising from internal double- or triple-glycosidic-bond cleavages were detected, which underscore more of the glycan chain structure: the core disaccharide unit is supported by the ions

at m/z 347.12 (B5/B3) and the Y2-type ions at m/z 996.77; the (Fuc)HexHexNAcHex is indicated by the ions at m/z 696.33, which then undergoes a neutral loss of Fuc to deliver the fragment ions at m/z 550.18. All other blood group A glycolipid species (listed in Table II) rendered similar rich fragment ion patterns. An additional example is presented in Figure 5, where the fragmentation patterns of the doubly charged ions at m/z 1304.58 and the co-isolated ions at m/z 1305.60, respectively, are assigned to Fuc2Hex5HexNAc4Cer(d18:1/24:0,1) based on the theoretical calculation. The resulting fragmentation pattern is consistent with an A-11 structure with (d18:1/24:0) (Figure 6, I). Given the very low abundance of this species in the original mixture of fraction 5, the MS/MS pattern is remarkably rich and the fragment ions abundance is sufcient to identify the structure. Similar to the A-6 structure, the neutral loss of a HexNAc residue from the precursor ions delivering the fragment ions at m/z 1202.82 (Y8) indicates the HexNAc as a terminal residue at the nonreducing end. More prominent is the neutral loss of one and two Fuc residues, characterizing the degree of fucosylation of the molecule. Following the loss of HexNAc residue from the precursor ions, the series of Y-type ions starting with the ions at m/z 1048.74 arising from

Fig. 4. Nano-ESI-QTOF fragmentation spectrum of singly charged ions at m/z 1711.11 detected in fraction 5 consistent with FucHexNAc2Hex3Cer(d18:1/24:0) and assigned to the A-6 glycolipid. Co-isolated are the ions at m/z 1709.10 which correspond to A-6 Cer(d18:1/24:1), giving rise to Y-type fragment ions downshifted by 2 amu (due to the monounsaturated C24 fatty acid) compared with the ions derived from precursor at m/z 1711.11. Inset: Fragmentation scheme of the A-6 glycolipid with Cer(d18:1/24:0).

166

Glycolipid structures in an A3 red cell phenotype

Fig. 5. Nano-ESI-QTOF fragmentation spectrum of doubly charged ions at m/z 1304.58 detected in fraction 5 consistent with Fuc2HexNAc4Hex5Cer(d18:1/24:1) and assigned to the A-11 glycolipid. Co-isolated are the ions at m/z 1305.60 which corresponds to a glycolipid with 11 sugar residues and Cer(d18:1/24:1), giving rise to Y-type fragment ions up shifted by 2 amu compared with the ions derived from precursor at m/z 1304.58. Fragmentation patterns are also consistent with an isobaric structure, possibly an extended X2 glycolipid.

the subsequent neutral loss of FucHex, and ending with Y1-type ions, HexCer, essentially concludes the carbohydrate chain structure of an A-11 glycolipid. The HexNAc-HexHexNAc-Hex-Hex-Cer structure represented at m/z 1565.18 is extended by two HexNAc-(Fuc)Hex units, which is typical for an A-11-3 structure. The fragment ions at m/z 1873.30 along with the fragmentation of ions at m/z 1202.78 to 1048.74, respectively, clearly point to the attachment of two Fuc residues at the rst two Hex residues in the glycan chain at the nonreducing end. These structural elements along with the B-type ions, B2 (HexNAc-(Fuc)Hex) and B3 (HexNAc-(Fuc)Hex-HexNAc) and a considerable number of internal double- and triple-glycosidic-bond cleavages (Figure 6, I) represent a reliable set of ngerprint ions for diagnosing an A-11-3 structure. An interesting feature revealed by data interpretation from Figure 5 is the existence of structural elements corresponding to an overlapping, isobaric structure, other than A-11. The fragment ions at m/z 1150.33 related to a neutral loss of a FucHex residue from the precursor ions along with the ions derived by the subsequent neutral loss of the second Fuc residue at m/z 1077.25, and the ions at m/z 1768.09 supporting the HexNAc-HexNAc-Hex-HexNAc-Hex-Hex-Cer

fragment reveal the existence of another species, as depicted in Scheme II (Figure 6). Even though at very low abundance, these peaks depict a reliable isotopomer pattern for their identication and assignment. In this conguration (Scheme II in Figure 6), a repeating (Fuc)Hex-HexNAc motif at the nonreducing end terminus extends the HexNAcHex-HexNAc-Hex-Hex-Cer core structure at m/z 1565.18, indicating a seemingly extended X2-type glycolipid. The presence of X2 glycolipid was also detected by NMR (Table III), while structural elements suggestive for other members of extended X2 glycolipid series were also found to interfere with other blood group A glycolipid fragmentation patterns. A more detailed study of such an extended X2 glycolipid as well as of extended GalNAc3GbO4 series (data not shown) will be undertaken in the near future to comprehensively identify the number and type of the structures in the series. Structural analysis of A glycolipids by proton NMR The proton NMR (1H-NMR) spectrum of fraction 1 is shown in Figure 7 and identied structures and references (Clausen, Levery, McKibbin et al. 1985; Holgersson et al. 1990; Thorn

167

L Svensson et al.

Fig. 6. Fragmentation scheme of the ions at m/z 1304.58 for (I) the A-11 glycolipid with Cer(d18:1/24:01), the major species and (II) an isobaric structure, possibly extended X2 identied as low abundant species.

Table II. List of ion species corresponding to A-glycolipids detected by nano-ESI-QTOF MS in three fractions and their structural assignment Structural assignment 1 4 HexNAc(Fuc)HexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAc(Fuc)HexHexNAcHexHex HexNAc(Fuc)HexHexNAc(Fuc)HexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAcHexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAc(Fuc)HexHexNAcHexHexNAcHexHex HexNAc(Fuc)HexHexNAc(Fuc)HexHexNAcHexHexNAcHexHex Ceramide d18:1/24:1 d18:1/24:0 d18:1/18:0 d18:1/20:1 d18:1/20:0 d18:1/22:1 d18:1/22:0 d18:1/24:1 d18:1/24:0 d18:1/18:0 d18:1/22:0 d18:1/24:1 d18:1/24:0 d18:1/24:1 d18:1/24:0 d18:1/24:1 d18:1/24:0 d18:1/24:1 d18:1/24:0 Name A-6-2 A-6-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-8-2 A-9-3 A-9-3 A-10-2 A-10-2 A-11-3 A-11-3 m/z 1709.10+ 1711.11+ 1006.832+ 1019.822+ 1020.872+ 1034.362+ 1035.452+ 1048.502+ 1049.502+ 993.412+ 1035.452+ 1048.502+ 1049.502+ 1122.102+ 1123.112+ 1231.522+ 1231.532+ 1304.582+ 1305.602+

N.B. All ion species were subjected to CID MS/MS experiments for structure elucidation. Bold indicates the blood group A epitope.

et al. 1992; Teneberg et al. 1994; Angstrom et al. 2004; Miller-Podraza et al. 2005) are summarized in Table III. The dominating structures were as expected GbO4, H-5-2 and Lea-5 in the descending order. Smaller contributions from GalNAc3GbO4, X2, FucGgO4 and FucGbO5 are also evident as are small contributions from Lex-5 and A-6-2. A closer examination of the Fuc2 and GalNAc3 signals of the 168

A-6-2 revealed, however, an overlapping component which most likely stems from the corresponding signals of the A-4-6 glycolipid (GalNAc3(Fuc2)Gal4Glc1Cer) (Breimer et al. 1982; Bjork et al. 1987). An intermediate contribution attributed to neolactotetraosylceramide and/or neolactohexaosylceramide (nLc-4 and/or nLc-6) is also deduced from the intensity of the GlcNAc3 resonance at 4.65 ppm which

Glycolipid structures in an A3 red cell phenotype

cannot be due solely to the internal GlcNAc3 resonance of the X2 glycolipid (cf. the intensities of resonances EIII and EV). The P1 structure may also be present in small amounts but is in that case masked primarily by GbO4 and GalNAc3GbO4 resonances. 1D and 2D NMR evidence (not shown) for the presence of different structures in fractions 25 containing extended glycolipid chains is summarized in Table IV. Most noticeable is the dominating presence of GbO4 and H-7-2/H-9-2 in all four fractions whereas blood group A type 2 structures are evident in very small amounts having 8 or 10 sugar residues. All other structures are also present in very small amounts except for an extended form of P1, which is present approximately in the same amount as the H structures in fraction 4. Evidence for a second extended structure stems from the observed perturbations of the anomeric shift of the terminal GalNAc3 residue of GalNAc3GbO4 in fraction 2. However, the nature of this extension from an NMR point of view is unclear at present. As expected, the extension of H-5-2 by one or two lactosamine fragments yielding H-7-2 and H-9-2 is present as are the structures H-8-3 and H-10-3 which are precursors for the similarly found A-9-3 and A-11-3 structures. It should be noted that only the signature resonances stemming from the glycotope residues have been recognized for different A and H structures. No additional or informative spectra were obtained from fractions 6 and 7. Novel glycolipid structures not related to blood group A, such as series of extended GalNAc3GbO4, extended X2 and extended P1 glycolipids were detected by both proton NMR and MS (Table IV). Discussion The A3 phenotype is rare occurring with a frequency of <1:1000 (Klein and AD 2005) and was recognized many years ago as one of the more unusual subgroups of A, essentially because it is characterized not only by weak expression of A antigen but also a co-requirement of the serological phenomenon called mixed-eld agglutination. Over the last

decade, a genetic basis for many of the ABO subgroups has been resolved, but the A3 genotype is still undened with almost all individuals reported today having normal ABO genotypes. This suggests that genes outside the ABO locus maybe responsible for the phenotype of reduced A antigen and mixed-eld agglutination. However, the individual of this study is more unusual than other A3s in that it does have an ABO genetic mutation and by ow cytometry it does not show the dual population of cells seen in other A3s (Hult and Olsson 2010). These observations suggest that there may be two different mechanisms leading to the A3 serological phenotype. The rst is the more common A3 with an undened ABO genetic background and a dual population of cells, whereas the second is the A3 of this paper with a novel A304 allele (called here the A304 A3 phenotype) (Svensson et al. 2005) and only a single population of cells identied by ow cytometry. The glycolipids of this sample A304 A3 were originally studied and found to be unusual compared with other subgroups of A in that it appeared to lack the otherwise dominant A-6-2 glycolipid and an extended glycolipid, together with evidence of novel blood group A structure(s) (Svensson et al. 2005). So far glycolipids from undened ABO genetic background A3 individuals with normal ABO alleles have not been studied, but it is suspected that their glycolipid prole may be different from the A304 A3 individual. All the same, the A304 A3 sample provides an insight into an alternative biochemical explanation for the A3 phenotype and in particular mixed-eld agglutination. In an attempt to elucidate the biochemical nature of the A3 phenotype or at least that of the A304 A3 phenotype and in particular mixed-eld agglutination, glycolipids from approximately 1 kg of packed red cells were extracted and puried so as to be suitable for structural analysis. Unlike the study of blood group antigens in common phenotypes, the study of weak subgroups is severely compromised by virtue of the fact that the blood group molecules are present in only very low concentrations, estimated at less than 5% of normal. Consequently, obtaining sufcient material of adequate purity

Fig. 7. A 600 MHz proton NMR spectrum of MPLC fraction 1 recorded at 30C showing the anomeric region of the four to six-sugar range of glycolipids from the Aweak-II blood group. Anomeric resonances are labeled according to the structures summarized in Table III.

169

L Svensson et al.

Table III. Glycolipid structures identified by NMR in the four to six-sugar region in rare blood group A erythrocytes from a single individual Structure Trivial name A B C D E F G H I-1 I-2 J
a

VI

IV GalNAc3 4.511 Gal4 4.328 (Fuc4) 4.767 GalNAc3 4.46 Gal4 4.296 (Fuc3) 4.836 Gal3 4.34 GalNAc3 4.46 Gal4 4.383 Gal4a

III Gal4 4.796 GlcNAc3 4.627 GlcNAc3 4.737 Gal4 4.80 GlcNAc3 4.65 GlcNAc3 4.705 GalNAc4 4.472 Gal4 4.80 GlcNAc3 4.605

IIa

Ia Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Glc1 Cer Cer Cer Cer Cer Cer Cer Cer Cer Cer Cer

NMR reference Holgersson et al. (1990) Holgersson et al. (1990) Angstrom et al. (2004) J. ngstrm, unpublished Thorn et al. (1992) Angstrom et al. (2004) Teneberg et al. (1994) Holgersson et al. (1990) Clausen, Levery, McKibbin et al. (1985) Breimer et al. (1982) Miller-Podraza et al. (2005)

GbO4 H-5-2 Lea-5 GalNAc3GbO4 X2 Lex-5 FucGgO4 FucGbO5 A-6-2 A-4-6 nLc-4/nLc-6

Fuc2 5.034 Gal3 4.32 GalNAc3 4.525 GalNAc3 4.581 Gal4 4.306 Fuc2 4.95 Fuc2 4.95 Gal3 4.46 GalNAc3 4.911 (Fuc2) 5.137 GalNAc3 4.916 (Fuc2) 5.132 GlcNAc3 4.65 Gal4a

Gal4 Gal4 Gal4 Gal4 Gal4 Gal4 Gal4 Gal4 Gal4 Gal4 GlcNAc3 4.65 Gal4

The identity of the anomeric proton resonances are given in the NMR spectrum shown in Figure 7. Chemical shift values are not given due to severe resonance overlap but can be found in the literature references given above.

Table IV. Summary of glycolipid structures identified in the seven MPLC fractions Structure nLc-4 nLc-6 nLc-8 H-5-2 A-6-2 H-7-2 A-8-2 H-9-2 A-10-2 H-8-3 A-9-3 H-10-3 A-11-3 GbO4 GalNAc3GbO4 ext. GalNAc3GbO4 Lea-5/Lex-5 ext. Lea/Lex P1 ext. P1 X2 Gb3, nLc-3 Galactosyl-Ab FucGgO4 FucGbO4 A-4-6 Hex5HexNAc4Fuc2 Hex4HexNAc3Fuc Hex6HexNAc2Fuc Hex6HexNAc2 H-10-2 H-12-2 >H-12 Fraction(s) 1 1 1 1 2 3 3 3 4 4 4 4 4 4 4 4 4 1 1 1 4 1 1 1 4 4 4 5 5 5 5 5 5 6 6 6 6 + + + 5 6 2 4 7 7 5 5 5 5 5 5 5 5 5 5 5 7 + 6 2 2 4 4 5 + +a + + + + + + + + + + + + 7 TLCEIA MS + + + + +b + + + + + + + + + + + + + +b + + + + + + + + + + + NMR + + + + +b + + + + + + + + + + + + +b + + + + +

6 6 6 7 7

1 1 1

2 2

3 3

nLc-8, neolactooctaosylceramide; nLc-3, neolactotriaosylceramide. Blank space, not tested or no known Mab reagent available for detection; ext., extended. a Very weak to weak reactions. b Very low levels.

for a structural analysis, particularly NMR spectroscopy, is demanding. Unfortunately, confounding issues posed by the poor solubility of puried extended glycolipids severely compromises the ability to obtain optimal levels of purity, and 170

instead fractions containing mixtures of glycolipids have to be used. However, by using three independent techniques, complementary advantages were exploited: TLC with immunochemical detection dened the approximate size and type of the glycotope(s) with a variety of monoclonal reagents; MS analysis with high sensitivity determines the existence of lowabundance structures and provides the structural architecture of glycolipids; proton NMR including 2D methods allows the identication of specic sugars and linkage signatures. It should be noted that, by tandem MS analysis of underivatized A type glycolipids, the discrimination between Gal3 and Gal4 linkages is not achievable, as ring cleavage or differential fragment patterns are not readily obtainable. Derivatization methods such as permethylation (Domon et al. 1990) were not explored in this study due to the increased complexity this procedure would introduce to already highly heterogeneous mixtures. Chain type as either 2 or 3 was therefore determined by immunochemistry and NMR. Nevertheless, given the fact that A type 3 chains are doubly fucosylated, whereas A type 2 chains are monofucosylated (for species with the same number of monosaccharide residues), and experimental conditions were applied to preventing in-source defucosylation, the type of the A chain could be postulated from tandem MS as well. By combining the results of all methods, an accurate and condent identication of the red cell glycolipids of this A304 A3 phenotyping individual was possible even for complex glycolipid molecular species present in trace amounts. It was fortunate that this A304 A3 individual was an ABO non-secretor, thereby eliminating the presence of type 1 A structures acquired by red cells from plasma from the interpretations. There are ve signicant observations related to the glycolipids of this individual. First, relative to the A1/ A2 phenotype, there are very low levels of blood group A glycolipids, which is expected as this is a requisite of a weak A phenotype. Second is the below-detection limits of A-6-2 in the total glycolipid fraction even when compared with the Ax subgroup (Svensson et al. 2005). Only in the puried fractions could A-6-2 be conrmed by both MAb 2-24 (Figure 3, II) and structural analysis (nano-ESI-QTOF MS (CID MS/MS) and NMR). Third was the conrmed presence of GalNAc3GbO4, which stained with certain anti-A reagents

Glycolipid structures in an A3 red cell phenotype

at the six-sugar interval, was present in extended forms and was identied by structural MS analysis through specic fragment ions corresponding to HexNAcHexNAcHexHexHexCer sequence and its extensions. Initially, it was considered that the GalNAc3GbO4 antigen could be contributing to the A3 phenotype; however, since this structure has been found in pools of red cell membranes of all ABO types (Ando et al. 1982; Angstrom et al. 1986), it is unlikely to be the cause of the A3 phenotype. Surprisingly, when we immunostained total glycolipid fractions from blood group B and O individuals with the GalNAc3GbO4 reactive anti-A (MAb 222), we were also able to clearly identify GalNAc3GbO4 bands in 6/ 6 B samples but in only 1/5 O samples (results not shown). It is possible to speculate that this could be due to an alternative substrate/precursor activity of the B and A304 A3 glycosyltransferases, and in the case of some O samples, a nontruncated enzyme incapable of utilizing H but still active against globotetraosylceramide, but this remains highly speculative. On the basis of these observations and the literature, it seems evident that GalNAc3GbO4 antigen does not contribute to the A3 phenotype but remains an important antigen to consider when using anti-A reagents by very sensitive techniques such as ow cytometry and TLCEIA. The fourth observation was the presence of most of the expected extended structures, including A-8-2, A-10-2, A-9-3 and A-11-3. Why these extended structures should still be synthesized, when the usually more dominant A-6-2 glycolipid is poorly represented, is an interesting questionand one which remains unresolved. The fth was both the absence of an extended structure and no evidence for any branched blood group A glycolipids in the MS spectra, which raises the possibility that the absent extended structure may have been branched. After excluding recognized dual-cell population reasons for mixed-eld agglutination, namely chimerism, transfusion, transplantation, fetal maternal bleeds or disease, the diagnosis of the A3 phenotype is solely on the basis of mixed-eld anti-A serological agglutination of red cells with reduced A antigen expression. Intriguingly, ow cytometry has reasonably suggested that dual populations of cells may be an explanation for the mixed-eld reactions observed (Hult and Olsson 2010). This is also supported by ndings that the A glycosyltransferase is found only on agglutinated cells (Nakamura et al. 1989). Unfortunately, this dual population explanation for mixed-eld agglutination does not hold for this A304 A3 sample, as in the same ow cytometry study (Hult and Olsson 2010), it did not have a dual population. Additional complexity is suggested by early observations that A3 cells initially not agglutinated in the mixed-eld reaction will agglutinate when secondarily reacted with anti-A. We were also able to conrm this unusual phenomenon of A3 red cells with this A304 A3 sample and serological gel cards. Although the glycolipids of this A304 A3 individual may not be typical of all A3 phenotypes this individual unambiguously has A3 serology of reduced A antigen and mixed-eld agglutination. Consequently, an alternative explanation for mixed-eld agglutination, not dependent on a dual-cell population, is needed. This requires a population of cells to start with a low antigen count close to the positive/negative agglutination threshold. If that population of cells is then also missing some

or most of its branched structures, then the binding strength of antibodies to them would be compromised (a phenomenon well recognized in the serology of newborns; Clausen and Hakomori 1989). When external shear forces are then applied in serological reactions, it would be reasonable to expect that many of the cells less strongly bound in the agglutinates would shear away as free cells and result in a mixed-eld reaction. Such a postulate is also compatible with the observation that the free cells when recovered are agglutinable by antibody. Overall, we can condently state that the glycolipid anomalies of this A304 A3 individual was a very low abundance of A structures, particularly A-6-2 and an absence of branched A glycolipids. The contribution of the glycoproteins or polyglycosylceramides was not able to be considered. It is speculated that the low level of A antigen combined with an absence of extended branched glycolipids may be the factor determining the mixed-eld agglutination phenomenon, at least in this A304 A3 individual.

Materials and methods Blood donor Six blood donations totaling approximately 1.0 kg of concentrated red cells were collected over the period of 2 years from the original donor (Aweak-II) with the A3 (anti-A mixed eld), P1, Le(a+b) ABH non-secretor phenotype and the A2(539G>A)/O1 genotype as previously reported (Svensson et al. 2005). Red cell units at the supplying Blood Centre were leukocyte-depleted by ltration, washed free of plasma with buffered isotonic saline and stored frozen at 20C until ready for processing. Mixed-eld agglutination serology A304 A3 red cells were tested against Novaclone murine monoclonal anti-A (Galileo; Dominion Biologicals Limited, Dartmouth, Nova Scotia, Canada) in a saline Diamed gel card system (DiaMed, GmbH, Cressier FR, Switzerland) and separated by centrifugation according to normal serological procedures. Following centrifugal separation, the agglutinated (+) cells from the upper part of the gel were recovered from the unagglutinated () cells from the lower part of the gel. Cells were recovered from several gel cards, washed, concentration adjusted and then reacted again in the original system against new reagents. Blood group O and A2 cells were tested in parallel as controls (not shown). Neutral glycolipids The glycolipids from the thawed hemolyzed red cell pool were extracted via a system of solvents and silica/ion exchange chromatography as previously described (Svensson et al. 2005), resulting in 250 mg of total neutral glycolipids. It should be noted this method does not isolate polyglycosylceramides, i.e. those with >20 sugar residues. The total glycolipid mixture was subjected to TLC, chemical or immunochemical staining analyses, and 200 mg was further rened by column chromatography (open column as well as MPLC). 171

L Svensson et al.

The terminology used to describe glycolipids is as recommended by IUPAC-IUB (Chester 1998). The abbreviation, for example, A-6-2 stands for blood group (group A), the number of carbohydrate units (six saccharides) and the type of core structure (type 2) of the glycolipid (Holgersson et al. 1992). With respect to Cer, for example, d18:1/24:0 means a dihydroxy long chain base with 18 carbon atoms and one double bond combined with a saturated non-hydroxylated 24-carbon atom fatty acid. Medium-pressure liquid chromatography Two hundred milligrams of total neutral glycolipids were separated into 20 fractions by the Bchi Sepacore Flash system (Bchi Labortechnik AG, Switzerland) using a silica S60 (2050027, SDS, France) packed column. The ow rate was set to 12.5 mL/min. The rst two fractions were eluted with a solvent mix of chloroform:methanol:water (CMW) 80:20:1 v/v each after 15 min followed by four more fractions eluted with CMW 70:25:4 v/v at 10 min intervals. Ten fractions were then eluted with CMW 60:35:8 v/v at 5 min intervals and two further fractions at 10 min intervals. Two nal fractions were eluted with CMW 40:40:12 v/v at 20 and 10 min intervals, respectively. Following TLC analysis (chemical and immunochemical), the rst nine fractions were found to contain no signicant blood group glycolipids or only short chain compounds (less than ve residues) and were not analyzed further. The remaining fractions were pooled into seven fractions according to TLC mobility (Table I) and then subjected to extended immunochemical analysis and structural characterization by MS and proton NMR spectroscopy. No attempts to further purify these fractions were made due to the low solubility in the solvent systems used and the risk of irreversible loss of individual componentsa phenomenon previously reported during the purication of glycolipid components (Angstrom et al. 1986). Thin layer chromatography TLC plates (HPTLC Silica gel 60 aluminum sheets; Merck, Darmstadt, Germany) were loaded with 10, 20 or 50 g of glycolipids per lane and separated in a solvent system of CMW 60:35:8 v/v. The air-dried plates were either chemically visualized with anisaldehyde (Karlsson 1987) or by immunochemistry (TLCEIA) (Schnaar and Needham 1994; Svensson et al. 2005). TLCenzyme immunoassay TLCEIA involved staining TLC (CMW 60:35:8 v/ v)-separated glycolipids with characterized primary murine MAbs, followed by alkaline phosphatase-conjugated antimurine immunoglobulin (A0162; Sigma, St. Louis, MO) and BCIP/NBT chromogenic substrate (B5655; Sigma) (Schnaar and Needham 1994; Svensson et al. 2005). A similar protocol was used for immunostaining with biotinylated H. pomatia lectin, which detects GalNAc residues (L6512; Sigma) diluted at 1:1000 except the secondary detection was with alkaline phosphatase-conjugated streptavidin (S2890; Sigma). The blood group A, AB and H MAb reagents used for TLCEIA and their reported specicities are listed in Table V 172

(Young et al. 1981; Clausen, Levery, Nudelman et al. 1985; Hirohashi et al. 1985; Le Pendu and Henry 2002; Svensson et al. 2005). Blood group B MAb reagents used are anti-B (DAKO Corporation, Carpinteria), HEB-29 (GeneTex Inc., Irvine) and MAb 035 (4th International Workshop on Monoclonal Antibodies Against Red Cells and Related Aantigens). The specicity of the broadly reactive MAb 2-41 and the anti-A type 2 MAb 2-24 was further investigated, and it was found that both were unreactive with pure NMR-veried GalNAc3GbO4 pentaglycosylceramide (Ando et al. 1982), which migrates in the hexaglycosylceramide region on the TLC plate. In contrast, the other broadly reactive antibodies MAbs 2-22 and MAb Lorne (result not shown) were both strongly reactive with this GalNAc3GbO4 structure (Figure 3, plate IV). On this basis, if a glycosylceramide present in hexaglycosylceramide region was reactive with MAbs 2-22 or Lorne but unreactive with MAb 2-41, it was considered strongly indicative of the presence of GalNAc3GbO4. Rm scale Rm scales are included in all TLCEIA plates to indicate the relative position of various bands. The scale was set with 0 as the origin and rm 6 for blood group A hexaglycosylceramide (Svensson et al. 2005). Nano-ESI-QTOF MS and tandem MS MS experiments were conducted on an orthogonal time-ofight mass spectrometer (Micromass, UK) equipped with a nano-ESI ion source. For data acquisition and processing, the MassLynx software (Micromass) was used. The gas-phase ions were generated from solution using an in-house pulled glass capillary. The ESI voltage was applied onto a metal wire inserted in the glass capillary. All fractions were prepared to a stock concentration of 1 g/L in different solvent systems systematically
Table V. Monoclonal anti-A, anti-AB and anti-H reagents used for TLCEIA Identity 224a HH4 TH1 222
a

Clone NaM200-16C5 HH4 TH1 AY209 11H5 BH517 HH14 BE2

Specificity A type 2 A type 2 A type 3 A+ GalNAc3GbO4b A+ GalNAc3GbO4b A+B H type 3 H type 2

Source ETS, Pays de Loire, France H. Clausen, Copenhagen, Denmark H. Clausen, Copenhagen, Denmark ETS, Bretagne, France Lorne laboratories, Reading, UK ETS, Bretagne, France H. Clausen, Copenhagen, Denmark H. Clausen, Copenhagen, Denmark

1401 241 HH14

a

BE2

a MAbs from the 4th International Workshop on Monoclonal Antibodies Against Human Red Cells and Related Antigens. b GalNAc3GbO4 reactivity tested against pure GalNAc3GbO4 glycolipid (see Materials and methods).

Glycolipid structures in an A3 red cell phenotype

determined in accordance to the requirements for the solubility and ionization of individual fractions. Thus, fraction 1 was prepared in methanol:water 10:1 v/v, fraction 5 in methanol:water 4:1 v/v and fraction 4 in 100:5:2 methanol: chloroform:water. Fractions 1 and 5 were further diluted in methanol and fraction 4 in 10:1 methanol:water v/v solutions to a nal concentration of approximately 5 pmol/L (calculated considering the average molecular mass of 2500 Da for fractions 5 and 4 and 1500 Da for fraction 1). The mass spectrometer was operated in a positive-ion mode at electrospray potentials between 1000 and 1200 V. Sampling cone potential was ranged between 35 and 55 V for fractions 5 and 4 to enhance the ionization of large molecules and prevent in-source glycosidic bond cleavages, particularly the labile Fuc linkages. For fraction 1, the sampling cone potential was varied between 65 and 120 V given the presence of shorter glycolipids. For structural elucidation, CID tandem MS experiments were performed employing collision energy values between 45 and 160 eV depending on the ion type, to obtain a comprehensive fragment ion pattern for an accurate structural identication. The collision gas pressure was set to 15 psi for all MS/MS experiments. All mass spectra were externally calibrated using a NaI solution. The tandem MS data assignment followed the nomenclature established by Domon and Costello (1988).

Conict of interest statement None declared. Abbreviations Cer, ceramide; CID, collision-induced dissociation; CMW, chloroform:methanol:water; ESI-QTOF MS, electrospray ionization-quadrupole time-of-ight mass spectrometry; Ext., extended; Fuc, fucose; GalNAc3GbO4, para-Forssman; GbO4, globotetraosylceramide (globoside); Hex, hexose; HexNAc, hexosamine; 1H-NMR, proton NMR; MAb, monoclonal antibody; MPLC, medium-pressure liquid chromatography; MS, mass spectrometry; MS/MS, tandem mass spectrometry; nLc-4, neolactotetraosylceramide; nLc-6, neolactohexaosylceramide; NMR, nuclear magnetic resonance; rm, relative migration; TLC, thin layer chromatography; TLCEIA, thin layer chromatographyenzyme immunoassay. References
Ando S, Kon K, Nagai Y, Yamakawa T. 1982. A novel pentaglycosyl ceramide containing di-beta-N-acetylgalactos-aminyl residue (Para-Forssman glycolipid) isolated from human erythrocyte membrane. Adv Exp Med Biol. 152:7181. Angstrom J, Karlsson H, Karlsson KA, Larson G, Nilson K. 1986. GalNAc13 terminated glycosphingolipids of human erythrocytes. Arch Biochem Biophys. 251:440449. Angstrom J, Larsson T, Hansson GC, Karlsson KA, Henry S. 2004. Default biosynthesis pathway for blood group-related glycolipids in human small intestine as dened by structural identication of linear and branched glycosylceramides in a group O Le(ab) nonsecretor. Glycobiology. 14:112. Bjork S, Breimer ME, Hansson GC, Karlsson KA, Lefer H. 1987. Structures of blood group glycosphingolipids of human small intestine. A relation between the expression of fucolipids of epithelial cells and the ABO, Le and Se phenotype of the donor. J Biol Chem. 262:67586765. Blumenfeld OO, Patnaik SK. 2004. Allelic genes of blood group antigens: A source of human mutations and cSNPs documented in the Blood Group Antigen Gene Mutation Database. Hum Mutat. 23:816. Breimer ME, Hansson GC, Karlsson KA, Lefer H. 1982. Isolation and partial characterization of blood group A and H active glycosphingolipids of rat small intestine. J Biol Chem. 257:906912. Cartron JP, Reyes F, Gourdin MF, Garretta M, Salmon CH. 1977. Antigen site distribution among weak A red cell populations. A study of A3, Ax and Aend variants. Immunology. 32:233244. Chester MA. 1998. IUPAC-IUB Joint Commission on Biochemical Nomenclature (JCBN). Nomenclature of glycolipidsrecommendations 1997. Eur J Biochem. 257:293298. Clausen H, Hakomori S. 1989. ABH and related histo-blood group antigens; immunochemical differences in carrier isotypes and their distribution. Vox Sang. 56:120. Clausen H, Levery SB, Kannagi R, Hakomori S. 1986. Novel blood group H glycolipid antigens exclusively expressed in blood group A and AB erythrocytes (type 3 chain H). I. Isolation and chemical characterization. J Biol Chem. 261:13801387. Clausen H, Levery SB, McKibbin JM, Hakomori S. 1985. Blood group A determinants with mono- and difucosyl type 1 chain in human erythrocyte membranes. Biochemistry. 24:35783586. Clausen H, Levery SB, Nudelman E, Tsuchiya S, Hakomori S. 1985. Repetitive A epitope (type 3 chain A) dened by blood group A1-specic monoclonal antibody TH-1: Chemical basis of qualitative A1 and A2 distinction. Proc Natl Acad Sci USA. 82:11991203. Domon B, Costello CE. 1988. Structure elucidation of glycosphingolipids and gangliosides using high-performance tandem mass spectrometry. Biochemistry. 27:15341543. Domon B, Vath JE, Costello CE. 1990. Analysis of derivatized ceramides and neutral glycosphingolipids by high-performance tandem mass spectrometry. Anal Biochem. 184:151164.

Proton NMR spectroscopy Intact non-derivatized glycolipid samples were dissolved in dimethylsulfoxide:D2O, 98:2 (v:v) after deuterium exchange and spectra were acquired on a Varian 600 MHz spectrometer at 30C. Both 1D and 2D double quantum-ltered correlated spectroscopy (Marion and Wuthrich 1983) proton NMR were performed. An extensive library of reference chemical shift data at the temperature of choice (30C) for previously isolated pure glycosphingolipids representing practically all resonances of the structures in Table III was available for structural identication.

Funding This work was supported by Swedish Research Council/ Medicine (grant no. 11612) and Claes Hgman SAGMAN stipendium, Baxter Medical AB, Sweden. This project and L.S. was supported in part by the Biotechnology Research Institute of AUT University.

Acknowledgements The authors acknowledge the support and valuable advice provided throughout the study by Prof. Dr. Jasna Peter-Katalinic, University of Mnster, Germany. Professors Karl-Anders Karlsson and Susann Teneberg at the Institute of Biomedicine, Department of Medical Biochemistry and Cell Biology, University of Gothenburg, Sweden, are thanked for their valued discussions.

173

L Svensson et al.

Economidou J, Hughes-Jones NC, Gardner B. 1967. Quantitative measurements concerning A and B antigen sites. Vox Sang. 12:321328. Hakomori S. 1981. Blood group ABH and Ii antigens of human erythrocytes: Chemistry, polymorphism, and their developmental change. Semin Hematol. 18:3962. Heier HE, Namork E, Calkovska Z, Sandin R, Kornstad L. 1994. Expression of A antigens on erythrocytes of weak blood group A subgroups. Vox Sang. 66:231236. Hirohashi S, Clausen H, Yamada T, Shimosato Y, Hakomori S. 1985. Blood group A cross-reacting epitope dened by monoclonal antibodies NCC-LU-35 and -81 expressed in cancer of blood group O or B individuals: Its identication as Tn antigen. Proc Natl Acad Sci USA. 82:70397043. Holgersson J, Breimer ME, Samuelsson BE. 1992. Basic biochemistry of cell surface carbohydrates and aspects of the tissue distribution of histo-blood group ABH and related glycosphingolipids. APMIS Suppl. 27:1827. Holgersson J, Jovall PA, Samuelsson BE, Breimer ME. 1990. Structural characterization of non-acid glycosphingolipids in kidneys of single blood group O and A pigs. J Biochem. 108:766777. Hult AK, Olsson ML. 2010. Many genetically dened ABO subgroups exhibit characteristic ow cytometric patterns. Transfusion. 50:308323. Karlsson KA. 1987. Preparation of total nonacid glycolipids for overlay analysis of receptors for bacteria and viruses and for other studies. Methods Enzymol. 138:212220. Klein HG, Anstee D. 2005. Mollisons Blood Transfusion in Clinical Medicine. 11th ed. Oxford: Blackwell Publishing Ltd. Le Pendu J, Henry S. 2002. Section 2: Immunochemical, immunohistological and serological analysis of monoclonal antibodies with carbohydrates. Coordinators report. Transfus Clin Biol. 9:5560. Marion D, Wuthrich K. 1983. Application of phase sensitive two-dimensional correlated spectroscopy (COSY) for measurements of 1H-1H spin-spin coupling constants in proteins. Biochem Biophys Res Commun. 113:967974. Miller-Podraza H, Lanne B, Angstrom J, Teneberg S, Milh MA, Jovall PA, Karlsson H, Karlsson KA. 2005. Novel binding epitope for Helicobacter pylori found in neolacto carbohydrate chains: Structure and cross-binding properties. J Biol Chem. 280:1969519703.

Nakamura I, Takizawa H, Nishino K. 1989. A3 phenotype with A1 gene-specied enzyme character in serum. Exp Clin Immunogenet. 6:143149. Olsson ML, Chester MA. 1996. Polymorphisms at the ABO locus in subgroup A individuals. Transfusion. 36:309313. Olsson ML, Irshaid NM, Hosseini-Maaf B, Hellberg A, Moulds MK, Sareneva H, Chester MA. 2001. Genomic analysis of clinical samples with serologic ABO blood grouping discrepancies: identication of 15 novel A and B subgroup alleles. Blood. 98:15851593. Schachter H, Michaels MA, Tilley CA, Crookston MC, Crookston JH. 1973. Qualitative differences in the N-acetyl-D-galactosaminyltransferases produced by human A1 and A2 genes. Proc Natl Acad Sci USA. 70:220224. Schenkel-Brunner H. 2000. Human Blood Groups, Chemical and Biochemical Basis of Antigen Specicity. 2nd ed. Wien: Springer-Verlag. Schnaar RL, Needham LK. 1994. Thin-layer chromatography of glycosphingolipids. Methods Enzymol. 230:371389. Svensson L, Rydberg L, de Mattos LC, Henry SM. 2009. Blood group A and A revisited: An immunochemical analysis. Vox Sang. 96:5661. Svensson L, Rydberg L, Hellberg A, Gilliver LG, Olsson ML, Henry SM. 2005. Novel glycolipid variations revealed by monoclonal antibody immunochemical analysis of weak ABO subgroups of A. Vox Sang. 89:2738. Teneberg S, Jovall PA, Karlsson H, Sjogren HO, Brodin T. 1994. Isolation and characterization of a rat testis glycosphingolipid based on gangliotetraosylceramide and having a blood group B determinant extended with beta 3-linked N-acetylgalactosamine. J Biochem. 116:697703. Thorn JJ, Levery SB, Salyan ME, Stroud MR, Cedergren B, Nilsson B, Hakomori S, Clausen H. 1992. Structural characterization of x2 glycosphingolipid, its extended form, and its sialosyl derivatives: Accumulation associated with the rare blood group p phenotype. Biochemistry. 31:65096517. Yamamoto F, Marken J, Tsuji T, White T, Clausen H, Hakomori S. 1990. Cloning and characterization of DNA complementary to human UDP-GalNAc: Fuc alpha 12Gal alpha 13GalNAc transferase (histoblood group A transferase) mRNA. J Biol Chem. 265:11461151. Young WW, Jr., Portoukalian J, Hakomori S. 1981. Two monoclonal anticarbohydrate antibodies directed to glycosphingolipids with a lacto-N-glycosyl type II chain. J Biol Chem. 256:1096710972.

174