Occurrence of the human tumor-specific antigen structure Galβ1-3GalNAcα- (Thomsen–Friedenreich) and related structures on gut bacteria: Prevalence, immunochemical analysis and structural confirmation

Glycobiology, Oct 2011

The Thomsen–Friedenreich antigen (TF; CD176, Galβ1-3GalNAcα-) is a tumor-specific carbohydrate antigen and a promising therapeutic target. Antibodies that react with this antigen are frequently found in the sera of healthy adults and are assumed to play a role in cancer immunosurveillance. In this study, we examined the occurrence of α-anomeric TF (TFα) on a large variety of gastrointestinal bacteria using a novel panel of well-characterized monoclonal antibodies. Reactivity with at least one anti-TF antibody was found in 13% (16 of 122) of strains analyzed. A more in-depth analysis, using monoclonal antibodies specific for α- and β-anomeric TF in combination with periodate oxidation, revealed that only two novel Bacteroides ovatus strains (D-6 and F–1), isolated from the faeces of healthy persons by TF-immunoaffinity enrichment, possessed structures that are immunochemically identical to the true TFα antigen. The TF-positive capsular polysaccharide structure of strain D-6 was characterized by mass spectrometry, monosaccharide composition analysis, glycosidase treatments and immunoblot staining with TFα- and TFβ-specific antibodies. The active antigen was identified as Galβ1-3GalNAc-, which was α-anomerically linked as a branching structure within a heptasaccharide repeating unit. We conclude that structures immunochemically identical to TFα are extremely rare on the surface of human intestinal bacteria and may only be identifiable by binding of both antibodies, NM-TF1 and NM-TF2, which recognize a complete immunomolecular imprint of the TFα structure. The two novel B. ovatus strains isolated in this study may provide a basis for the development of TF-based anti-tumor vaccines.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:


Occurrence of the human tumor-specific antigen structure Galβ1-3GalNAcα- (Thomsen–Friedenreich) and related structures on gut bacteria: Prevalence, immunochemical analysis and structural confirmation

Gemma Henderson 1 Philippe Ulsemer 0 Ute Schber 0 2 Anja Lffler 0 Carl-Alfred Alpert 1 Martin Zimmermann- Kordmann 2 Werner Reutter 2 Uwe Karsten 0 Steffen Goletz 0 Michael Blaut 1 0 Glycotope GmbH , Robert-Rossle-Str. 10, D-13125 Berlin , Germany 1 Department of Gastrointestinal Microbiology, German Institute of Human Nutrition , Potsdam-Rehbrucke (DIfE), Arthur-Scheunert-Allee 114-116, D-14558 Nuthetal , Germany 2 Institute for Biochemistry and Molecular Biology, Charite - University Medicine Berlin , Arnimallee 22, D-14195 Berlin , Germany The Author 2011. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: - The ThomsenFriedenreich antigen (TF; CD176, Gal13GalNAc-) is a tumor-specific carbohydrate antigen and a promising therapeutic target. Antibodies that react with this antigen are frequently found in the sera of healthy adults and are assumed to play a role in cancer immunosurveillance. In this study, we examined the occurrence of -anomeric TF (TF) on a large variety of gastrointestinal bacteria using a novel panel of well-characterized monoclonal antibodies. Reactivity with at least one anti-TF antibody was found in 13% (16 of 122) of strains analyzed. A more in-depth analysis, using monoclonal antibodies specific for - and -anomeric TF in combination with periodate oxidation, revealed that only two novel Bacteroides ovatus strains (D-6 and F1), isolated from the faeces of healthy persons by TF-immunoaffinity enrichment, possessed structures that are immunochemically identical to the true TF antigen. The TF-positive capsular polysaccharide structure of strain D-6 was characterized by mass spectrometry, monosaccharide composition analysis, glycosidase treatments and immunoblot staining with TF- and TF-specific antibodies. The active antigen was identified as Gal1-3GalNAc-, which was -anomerically linked as a branching structure within a heptasaccharide repeating unit. We conclude that structures immunochemically identical to TF are extremely rare on the surface of human intestinal bacteria and may only be identifiable by binding of both antibodies, NM-TF1 and NM-TF2, which recognize a complete immunomolecular imprint of the TF structure. The two novel B. ovatus strains isolated in this study may provide a basis for the development of TF-based anti-tumor vaccines. Introduction Malignant transformation of cells is accompanied by alterations to the oligosaccharide chain profile of membrane glycoproteins and/or glycolipids (Feizi 1985; Hakomori 1989; Singhal and Hakomori 1990; Brockhausen 1999, 2006). Some of these altered structures have been categorized as tumor-associated or oncofetal carbohydrate antigens. Of these, the Thomsen Friedenreich antigen (TF, also known as T antigen or CD176) has received special attention due to its high tumor specificity. TF does not occur on normal adult tissues (Springer et al. 1976; Springer 1984; Cao et al. 1996; Goletz et al. 2003), has a high prevalence in many types of carcinoma (Springer 1997; Goletz et al. 2003), is involved in metastasis (SchlepperSchfer and Springer 1989; Cao et al. 1995; Shigeoka et al. 1999) and potentially plays a role in immunosurveillance and therapy (Springer 1997; Shigeoka et al. 1999; Irazoqui et al. 2001; Kurtenkov et al. 2007; Yu 2007). TF (Gal1-3GalNA-) is concealed within prolonged chains of O-glycans and therefore immunologically undetectable in normal adult tissues (Cao et al. 1996). However, in cancer tissues, TF is exposed, as changes in the glycosylation machinery result in native glycan structures being truncated (Hull and Carraway 1988; Lloyd et al. 1996; Brockhausen 1999). The characteristics of TF suggest that this antigen could be a preferred therapeutic target. Initial attempts to induce therapeutic anti-TF responses in humans were undertaken by Georg F. Springer. He successfully immunized breast cancer patients with a TF-carrying vaccine containing asialoglycophorin that was derived from red blood cell membranes. Both humoral and cellular responses were elicited (reviewed in Springer 1997). Unfortunately, only limited clinical effects were observed in more recent trials that tested synthetic glycoconjugate vaccines (Fung et al. 1990; MacLean et al. 1991; Adluri et al. 1995; Slovin et al. 2005). In contrast to Springers vaccine, synthetic TF glycoconjugates may not present the TF disaccharide in an appropriate environment or at a sufficient epitope density to induce an efficient immune response. Interestingly, natural anti-TF antibodies are frequently found in healthy adults (Springer et al. 1979; Butschak and Karsten 2002) and are thought to play a role in tumor immunosurveillance. Early experiments (Boccardi et al. 1974; Springer and Tegtmeyer 1981) indicated that these antibodies may Subject-isolate no. RAPD profile Antibody NM-TF1 NM-TF2 A68-B/A11 NM-TF1 NM-TF2 A68-B/A11 A-18 Bac17 A-17 Bac13 B-6 n.d. D-6 Bac14 F-1 (3, 4, 5, 20) Bac15 E-8 (11) Bac16 C-11 Bac12 H-1 Bif9 E-9 (15) Bif7 A-33 (34) Bif1 A-37 Bif3 B-3 (4, 7) Bif10 D-4 (5) Bif6 C-13 Bif11 G-9 Bif8 A-35 Bif2 D-7 Bif6 B-2 Bif10 A-15 Ca27 A-19 Ca28 G-11 n.d. D-3 Ec18 A-20 (23, 29) Ec19 G-2 (3, 4, 5, 8, 13, 14, 15, 16, 19) Ec20 F-8 n.d. E-1 Ent34 E-2 (28) Ent35 E-21 (25) Ent36 E-3 Ef24 E-18 Ef25 E-29 Ef26 F-2 Lac21 H-2 Lac23 F-7 Lac22 F-12 Lac22 E-5 Lac21 E-10 Lac21 E-16 Lac21 E-20 Lac23 E-7 (13, 17, 19, 22, 23, 24, 26) Lac22 F-6 (9, 10) Lac22 F-11 Lac21 G-10 n.d. G-17 n.d. G-1 n.d. C-10 Rum29 B-5 Rum30 E-4 Veill31 G-12 Veill31 E-6 Veill32 E-14 Veill33 +++ +++ + +/ +/ +/ +++ +++ + +++ + +/ +++ + +/ +/ +/ +/++ +/ +/ +/ + ++ + + ++ ++ + +++ + +/ +/++ +/ +/ + +/++ +/ +/ + +/ +/++ + ++ ++ Origin (if known) NM-TF1 NM-TF2 A68-B/A11 NM-TF1 NM-TF2 A68-B/A11 Atopobium Bacillus Bacteroides Butyribacterium Citrobacter Clostridium Enterococcus Helicobacter Lactobacillus rimae DSMZ 7090 subtilis DSMZ 10 acidifaciens DSMZ 15896 distasonis DSMZ 20701 fragilis DIfE 05 fragilis DSMZ 1396 fragilis DSMZ 2151 ovatus DSMZ 1896 thetaiotaomicron DSMZ 2255 thetaiotaomicron DSMZ 2079 vulgatus DSMZ 1447 adolescentis ATCC15703 angulatum ATCC 27535 animalis DSMZ 20104 catenulatum ATCC 27539 gallicum DSMZ 20093 infantis ATCC 15697 infantis ATCC 15702 longum DSMZ 20219 pseudocatenulatum ATCC 27919 suis ATCC 27533 methylotrophicum DSMZ 3468 freundii DSMZ 30039 asparagiforme DSMZ 15981 butyricum DSMZ 10702 celerecrescens DSMZ 5628 coccoides DSMZ 935 pasteurianum DSMZ 525 scindens DSMZ 5676 xylanolyticum DSMZ 6555 barkeri ATCC 25849 biforme DSMZ 3989 callanderi DSMZ 3662 cylindroides ATCC 27803 limosum CIP 104169 ventriosum ATCC 27560 coli DSMZ 613 coli DSMZ 8697 coli lac+ coli lac coli MG1655 coli Nissle 1917 coli W3110 hermannii ATCC 33650 faecium DIfE TC3 pylori NCTC 11637 acidophilus DSMZ 20079 brevis DSMZ 1268 brevis DSMZ 2647 brevis DSMZ 20054 casei DIfE GH1 cellobiosus DSMZ 20055 fermentum DSMZ 20052 intestinalis DSMZ 6629 murinus DSMZ 20452 plantarum DSMZ 20174 reuteri DSMZ 20015 reuteri DSMZ 20016 reuteri DSMZ 20056 rhamnosus GG ATCC 53103 zeae DIfE TC7 Human gingival crevice Mouse caecum Pleural fluid Appendix abscess Perforated appendix Human faeces Human faeces Adult intestine Human faeces +/ Rat faeces Human faeces Adult intestine Infant intestine Infant intestine Adult intestine Infant faeces +/ Pig faeces Sewage sludge digester Human faeces Pig intestine Cow manure Mouse faeces +/ Human faeces Decayed wood chips + Mud Human faeces Anaerobic digestor Human faeces Faeces EPEC strain, children Mouse ileum Mouse ileum Mutaflor Toe Human faeces Human gastric antrum Human Beer Silage Faeces Human faeces Saliva Fermented beets Rat intestine +/ Rat intestine Pickled cabbage Manure Adult intestine Rat faeces Human faeces Human faeces Origin (if known) Propionibacterium acnes acnes acnes granulosum Ruminococcus obeum Blautia producta Staphylococcus aureus warneri Streptococcus intermedius pneumoniae Culture contaminant Facial acne Culture contaminant NM-TF1 NM-TF2 A68-B/A11 NM-TF1 NM-TF2 A68-B/A11 be induced by gastrointestinal microorganisms that carry TF or TF-related structures. Antigens that cross-react with TF-binding antibodies or lectins have been reported to occur in Helicobacter pylori (Klaamas et al. 2002) and Enterobacteriaceae (Springer et al. 1979). Microbial TF structures are likely to be exposed and present at suitable densities on the cell surface and as such are probably more immunogenic than synthetic glycans. However, these studies did not analyze the microbial TF-carrying glycan structures in depth. Also, they did not differentiate between TF-related (crossreactive) and TF-identical structures and/or antibodies. As the likelihood of the human intestinal microbiota harboring further unknown, non-pathogenic TF-expressing microorganisms was high, we screened a wide variety of commensal bacteria for the occurrence of TF structures. This was done with a novel panel of antibodies consisting of two TF-specific monoclonal antibodies (NM-TF1 and NM-TF2) that recognize the TF disaccharide from two different angles, as well as a TF-specific antibody. In addition, mild periodate oxidation was used to ensure the carbohydrate specificity of antibody binding. Finally, we characterized the TF-carrying structures in one selected bacterial strain by means of mass spectrometry (MS). Results Identification of the isolates enriched from human faeces The identity of the isolates was determined with biochemical methods or by partial sequencing of 16S rRNA genes (Table I). Random amplification of polymorphic DNA (RAPD) profiling was used to further distinguish isolates that belonged to the same species down to the strain level. RAPD profiles were assigned based on band patterns obtained with five primers. Many isolates that were enriched from the faeces of any individual subject belonged to the same species and had identical RAPD profiles. This served as an indication that these isolates are presumably representatives of one and the same strain. For this reason, the TF-screening results for such isolates were summarized at the presumptive strain level for the remainder of the results section and presented in a single line in Table I. Binding of anti-TF antibodies We studied the occurrence of TF cell surface structures in commensal bacteria from culture collections as well as bacteria enriched and isolated from human faeces by means of an enzyme-linked immunosorbent assay (ELISA). Cellular structures of bacterial cell cultures that bound to at least one of the two TF-specific antibodies were found in 10 of 71 (14%) strains from culture collections and in 6 of 51 (12%) TF-affinity-enriched strains from faeces (Tables I and II, respectively). Interestingly, the intensity of antibody binding tended to be higher in the strains that were newly isolated. The monoclonal antibody NM-TF2 only bound to three strains enriched from faeces when periodate oxidation was not used: strongly to two Bacteroides ovatus strains and moderately to one Escherichia coli strain (Table I). With the exception of one culture collection strain, binding of the monoclonal antibody A68-B/A11 (anti-TF) was only seen in E. coli isolates. Of all bacteria examined, only two distinct B. ovatus isolates met the criteria set by the human cancer-associated TF structure, namely strong binding to anti-TF monoclonal antibodies NM-TF1 and NM-TF2 and low or no binding to the TF-specific monoclonal antibody A68-B/A11. No major variability in TF-specific antibody-binding patterns was observed, when bacteria with TF-related or immunochemically identical structures were cultured in different media or sampled during various growth phases (data not shown). Periodate oxidation to confirm the carbohydrate specificity of antibody binding Periodate oxidation preferably cleaves terminal sugar rings with vicinal hydroxyl groups. Therefore, a decreased intensity of antibody binding following mild (carbohydrate-specific) periodate oxidation can be used to confirm whether antigens are likely to be of carbohydrate nature (Woodward et al. 1985). In our study, this observation was only made for a single defined species of fecal isolates, namely two B. ovatus strains that were isolated from subjects D and F. In a considerable number of other (apparently TF-negative) microorganisms, periodate oxidation resulted in demasking of TF (Tables I and Fucose GlcNH2/GlcNAc Rhamnose GalNH2/GalNAc Galactose Glucose Mannose Galacturonic acid Glucuronic acid Not identified DSMZ 1896 GlcNH2, glucosamine; GalNH2, galactosamine; GlcNAc, N-acetylglucosamine; GalNAc, N-acetylgalactosamine; CPS, capsular polysaccharide; LPS, lipopolysaccharide. Gal and GalNac are the subunits composing the TF antigen. II). Such cryptic TF or TF-related structures were found in over 70% (54 of 71) of strains from culture collections and in 74% (38 of 51) of strains enriched from faeces. Monosaccharides in lipopolysaccharide and capsular polysaccharide extracts Crude lipopolysaccharide (LPS) and capsular polysaccharide (CPS) were extracted from cell pellets of B. ovatus D6 and B. ovatus-type strain DSMZ 1896. The TF structure was detected by ELISA in CPS and LPS extracts of B. ovatus D6, but not in extracts from B. ovatus DSMZ 1896 (data not shown), which was in agreement with the screening data (Tables I and II). The monosaccharide composition of crude LPS and CPS extracts was analyzed by high-pH anion-exchange chromatography and pulsed amperometric detection (HPAEC-PAD). The types of monosaccharides found in the crude LPS and CPS extracts of B. ovatus D6 and DSMZ 1896 were similar, but the amounts of individual monosaccharides differed between LPS and CPS extracts and between the two strains (Table III). The components of TF, GalNAc (GalNH2) and Gal occurred in approximately equimolar amounts in B. ovatus D6 CPS and LPS extracts, whereas relatively more Gal than GalNAc was found in B. ovatus DSMZ 1896. The highest equimolar amount of Gal and GalNAc was determined in the crude CPS extract of B. ovatus, which was therefore used for further analyses. A characteristic component of LPS, 2-keto-3-deoxyoctonic acid (KDO), can occasionally occur in CPS. KDO is ubiquitous in LPS and was used here as a marker for a potential LPS contamination of CPS extracts. KDO (11.2 pmol/g) was detected by reverse phase high-petroleum liquid chromatography (RP-HPLC) in crude CPS extracts and considered as indicating that these extracts were contaminated with LPS. Confirmation of the TF cross-reactive polysaccharide structure in B. ovatus D6 crude CPS extract by western blot The presence of TF in B. ovatus D6 capsule extracts was confirmed by western blot (Figure 1). Several carbohydrateFig. 1. SDSPAGE and western blot analyses of crude CPS extracts of B. ovatus AG6. (A) Western blot stained with the DIG-Glycan-Detection Kit and (B) immunoblot stained with the mAb NM-TF2. containing bands were detected in Alcian Blue-stained sodium dodecylsulfatepolyacrylamide gel electrophoresis (SDS PAGE) gels. Following the transfer of polysaccharides to nitrocellulose, a 26 and a 37 kDa polysaccharide band dominated (Figure 1A). Only the 37 kDa band could be shown to contain the TF structure by means of immunostaining with the monoclonal antibody NM-TF2 (Figure 1B). Purification of TF-positive oligosaccharides in CPS extracts by reverse phase (RP-HPLC) and anion exchange (DEAE-HPLC) chromatography As the crude B. ovatus D6 CPS extract also contained several TF-negative polysaccharides and possibly also LPS, it was further purified by C18 RPHPLC to increase the amount of TF-containing polysaccharides. TFcontaining polysaccharides were eluted with 29.029.4% propanol (fraction RP1) and 39.042.0% propanol (fraction RP2, data not shown). In contrast to TF-negative polysaccharide fractions, TF-positive polysaccharide fractions contained no detectable KDO (data not shown). The yield of TF-positive polysaccharides obtained by RP-HPLC was 30% of the crude extract. Following mild acid treatment with 5% acetic acid (100C, 1 h), fraction RP2 was further separated by means of DEAE-HPLC. TF cross-reactive fractions were eluted with 0 M NaCl (water, fraction D-1), 0.040.06 M NaCl (fraction D-2) and 0.090.17 M NaCl (fraction D-3, data not shown). Fraction D-1 was used for further analysis. This TF cross-reactive fraction contained increased amounts of Fuc, GalNAc/GalNH2 and Gal, which appeared to be the main components of the repeating unit of the TF-positive CPS. Structural analysis of the TF-positive polysaccharide by MS and HPAEC-PAD The structures of TF-containing CPS fragments could be identified by monosaccharide analysis in combination with m/z (M + H)+/(M + NH4+)//(M + Na+) Chemical composition HexNAc Hex MeDeoxyHex DeoxyHex2 HexNAc Hex HexNAc2 Hex-DeoxyHex Hex HexNAc MeDeoxyHex DeoxyHex HexNAc HexNAc2 Hex Hex-DeoxyHex3 Hex MeDeoxyHex DeoxyHex2 DeoxyHex MeHexNAc HesNAc DeoxyHex3 - HexNAc Me DeoxyHex DeoxyHex HexNac Hex DeoxyHex2 HexNAc Hex DeoxyHex HexNAc2 Hex HexNAc2 Hex2 DeoxyHex Me DeoxyHex HexNAc DeoxyHex- MeDeoxyHex - HexNAc Hex DeoxyHex HexNAc2 Hex MeDeoxyHex HexNAc2 Hex The purified TF-positive polysaccharide was fragmented by hydrolysis with 1% acetic acid for 1.5 h at 100C. DeoxyHex, desoxyhexose; HexNAc, N-acetylhexosamine; Hex, hexose; Me, methyl ester. Fig. 2. Proposed structure of the repeating unit of the TF-positive CPS of B. ovatus D-6. The structure was deduced from overlapping fragments determined by MS, exoglycosidase digestion and monosaccharide analyses. Asterisk denote the exact position of the methyl group could not be defined. MS. CPS-fragment masses (between m/z 50 and 900) and their chemical composition obtained by mild acetic acid hydrolysis (with 5% acetic acid) are depicted in Figure 3 and Table IV, respectively. The proposed structure of the repeating unit consists of seven monosaccharides belonging to three species: Gal, Fuc and GalNAc. The structure and the glycosidic bonds between the monosaccharides were identified by cleavage with the following specific glycosidases: 14-galactosidase, 13-galactosidase, 13/ 4-fucosidase, chondroitinase ABC (-1-2,3,4-HexNAcase) or 1-4-HexNAcase, as well as double digestion with 13galactosidase/1-4-HexNAcase, followed by mass spectrometric analysis and monosaccharide analysis of the generated fragments. The overlapping fragments revealed the structure in Figure 2, where the potential TF structure, Gal1-3GalNAc, is a branching component. The glycosidic linkage between Gal1-3GalNAc and the backbone GalNAc is -anomeric. This was corroborated by sequential enzymatic cleavage of an RP-purified TF-positive polysaccharide followed by monosaccharide analysis of the released residues. Initial cleavage with 1-3-galactosidase released galactose, and further cleavage with 1-4-HexNAcase released GalNAc. In contrast, GalNAc could not be identified following sequential cleavage with 1,3-galactosidase and chondroitinase ABC. This finding was supported by dot blot analyses with monoclonal antibodies specific for either the -anomer (NM-TF1, NM-TF2 and HH8) or the anomer (A68-E/A2 and A68-E/E3) of TF (data not shown). Thereby we were able to identify the human tumor-identical antigen TF as a branching structure of the CPS of B. ovatus D6. Further mass spectrometric analyses revealed that one of the monosaccharides is methylated. The exact position of the methyl group could not be resolved. Discussion The normal microbiota of the gastrointestinal tract is not only of nutritional, but also of immunological importance. Several publications have indicated that antibodies that cross-react with tumor antigens such as the TF are found in sera of healthy adult persons. These antibodies may be part of the immunosurveillance system that protects against the development and spread of tumors (Springer 1997; Irazoqui et al. 2001; Butschak and Karsten 2002; Kurtenkov et al. 2007; Kodar et al. 2009). However, the structures of these bacterial antigens have, in most cases, not been fully characterized. In this study, we analyzed a substantial number of intestinal bacterial strains from culture collections as well as newly isolated strains from human faeces for the expression of true (-anomeric) TF and for the expression of cryptic TF or TF-related antigens. The distinction between TF (identical to human tumor TF), TF and TF-related antigens was possible with the aid of a novel panel of monoclonal antibodies specific for TF. In order to identify structures immunochemically identical to TF, we used a combination of the antibodies NM-TF1 and NM-TF2. Both are TF specific and bind to tumor cells in immunohistochemistry, but recognize the TF disaccharide from different angles. We interpret the simultaneous binding of both NM-TF1 and NM-TF2 to indicate that the epitope recognized is not only the true TF structure, but that it is also sufficiently exposed to be antigenically or immunogenically similar to tumor TF. With respect to a complete description of their specificity, it should also be kept in mind that all anti-TF antibodies are of the IgM isotype and recognize in fact clusters of their respective epitopes. For NM-TF1 and NM-TF2, it has been determined that substantial binding required at least three TF moieties at a distance of 0.9 nm each (U Karsten et al., unpublished data). Additionally, we employed an antibody to TF, a related but different antigen that is expressed on several glycolipids. TF is not per se a tumor antigen. For confirmation of the carbohydrate nature of the epitope recognized by TF-specific antibodies, mild periodate oxidation was used. Periodate oxidation leads to the destruction of selected sugar rings and thereby to a strong reduction in the binding of the respective antibodies (Woodward et al. 1985). By applying these criteria, we have isolated two new truly TF-positive bacterial strains from faeces of healthy individuals by means of a TF-specific affinity enrichment technique using immunomagnetic beads coated with antibodies NM-TF1 or NM-TF2. Only TF-related or cryptic TF antigens were detected in intestinal bacteria obtained from culture collections. We were surprised to learn that microorganisms that express structures immunochemically identical to the tumor-associated TF are obviously extremely rare in gastrointestinal bacteria. In fact, only two strains of the species B. ovatus that carry this structure could definitely be identified in this study. In a number of gastrointestinal bacteria, periodate oxidation led to the de novo expression of otherwise cryptic TF or TF-related epitopes. This can be explained by the fact that periodate oxidation of polysaccharides leads to partial depolymerization and the formation of highly flexible hinges, thereby exposing parts of the molecule not accessible before (Kristiansen et al. 2010). At a first glance, the rareness of TF-expressing bacterial strains documented here contradicts previous publications that describe the expression of TF on several bacterial strains (Springer et al. 1979; Klaamas et al. 2002). However, when submitted to our rigid criteria, most bacterial strains did not express true tumor-specific TF structures. Instead, they expressed what we call TF-related or cryptic TF antigens. TF-related antigens either reacted with only one of our two anti-TF antibodies, indicating incomplete spatial access to the disaccharide, or revealed preferred binding to the antibody A68-B/A11, indicating the expression of TF instead of the tumor-specific -anomer. The cryptic TF antigen relates to TF (or a TF-related antigen) that only becomes accessible following periodate treatment. For example, in contrast to data from the literature (Klaamas et al. 2002), we only found cryptic TF on H. pylori. This was true even when the same strain was tested under the same culture conditions as used previously. This apparent discrepancy may be explained by either the antigen extraction method employed by Klaamas et al. (glycine buffer at pH 2.2) or the different set of antibodies used in this publication. In another paper, where the well-established anti-TF antibody HB-T1 was employed, TF-positivity of H. pylori was also not detected (Barresi et al. 1999). However, it is also known that the carbohydrate composition of H. pylori displays high variability (Nilsson et al. 2008). Interestingly, the expression of TF or TF-related epitopes seems to be highly strain-specific. For instance, monoclonal antibody NM-TF1, but not NM-TF2, bound to Propionibacterium acnes DIfE 89 (P. acnes is a later synonym for Corynebacterium parvum) suggesting, in agreement with previous findings (Springer et al. 1979), the presence of a TF-related epitope. However, none of the anti-TF monoclonal antibodies employed bound to other strains of the same species such as P. acnes DSMZ 1897 or DSMZ 16379. Also, contrary to the TF-expressing B. ovatus strains D-6 and F-1 that we isolated, B. ovatus DSMZ 1896 was not bound by either NM-TF1 or NM-TF2. Therefore, and because previous studies neither identified clinical isolates down to the strain level nor used a highly specific tool combination comparable to ours, we conclude that the prevalence of the real TF in bacteria has so far been overestimated and that our study is the first report on the expression of true tumorspecific TF on commensal bacteria. The fact that TF immunoenrichment of stool samples from different donors yielded only two closely related B. ovatus strains that carry the TF epitope suggests that this structure is rare on intestinal bacteria, at least in an exposed and immunocompetent form. As a next step, we characterized the TF-carrying structure on the surface of the B. ovatus strain D-6. Both LPS and CPS came into consideration as potential carriers of the TF disaccharide. Through monosaccharide analyses, we identified Gal and GalNAc, the components of TF, in crude CPS and LPS extracts of B. ovatus D6. However, western blot data and the determination of KDO were considered as indications that the crude CPS extract possibly contained LPS and other impurities. Therefore, the CPS extract was further purified by RP- and anion exchange-HPLC to separate CPS from LPS and TF-positive from TF-negative polysaccharides. TF-positive fractions obtained with RP-HPLC did not contain KDO, meaning that the TF structure was located within the CPS. The accumulation of Fuc, GalNH2/GalNAc and Gal in the purified TF-positive polysaccharides indicated that these monosaccharides might be components of the repeating units. Furthermore, the separation of typical capsular monosaccharides such as galacturonic acid from TF-positive and TF-negative fractions by chromatography indicated that the CPS extract of B. ovatus D6 might contain several polysaccharides, which could be TF-positive or TF-negative. This is not unusual; Bacteroides fragilis expresses at least eight CPSs (Krinos et al. 2001), the structures of only two of which have been chemically determined (Baumann et al. 1992; Tzianabos et al. 1992). It is also feasible that TF-positive polysaccharide fractions eluted with different propanol and NaCl concentrations were bound to other polysaccharides or differed in the number of repeating units. We identified the sequence of a CPS repeating unit that consisted of seven monosaccharides. This is in accordance with the six to nine monosaccharide repeat units found in the CPS of the related B. fragilis strain (Kasper et al. 1983). We localized the tumor-associated TF structure as a branching component of the repeating unit. This result is important because the exposure of the entire repetitive TF disaccharide along the CPS may be essential for the induction of an adequate immune response. Branching structures of repeating units are common among CPSs (Weintraub et al. 1985; Tzianabos et al. 1992). The galactose, Nacetylgalactosamine and especially the fucose content within the repeating unit was high in the CPS preparation from B. ovatus D-6. Previous work has shown that CPS often includes high fucose contents and repeats of these monosaccharides (Kasper et al. 1983; Pantosti et al. 1991). Our findings demonstrate the existence of true TF-expressing bacteria in the gastrointestinal microbiota of healthy human individuals. This supports the hypothesis that naturally occurring anti-TF antibodies identified in the sera of healthy humans originate from an immune response of the mucosal immune system to gastrointestinal bacterial antigens (Springer and Tegtmeyer 1981). The sera of healthy adults contain 0.5 mg anti-TF IgM and 0.05 mg anti-TF IgG per 100 mL of serum (Butschak and Karsten 2002). The assumed role of such natural anti-TF antibodies in immunosurveillance is complex and essentially based on indirect evidence. There are no prospective long-term studies that have examined correlations between the level of anti-TF antibodies in the serum and the risk of primary tumor development. There is substantial evidence that the expression of TF on carcinomas indicates unfavorable prognosis and an enhanced risk for the development of metastases in colorectal and lung cancer (Cao et al. 1995; Takanami 1999; Baldus et al. 2000). It is also known that cancer patients tend to have lower levels of natural anti-TF antibodies when compared with healthy persons (Chen et al. 1995; Desai et al. 1995; Kurtenkov et al. 1995). More important in this respect are reports indicating that cancer patients with higher anti-TF IgG levels show significantly improved survival (Kurtenkov et al. 2007). There is also recent evidence that differently glycosylated subpopulations of anti-TF antibodies may reveal different prognostic potencies (Kodar et al. 2009). Anti-TF antibodies may exert their anti-tumor effects in multiple ways, dependent e.g. on the isotype. First, they may bind to TF-expressing tumor cells and thereby prevent TF-mediated docking of migrating tumor cells (SchlepperSchfer and Springer 1989; Shigeoka et al. 1999). Second, they may inhibit proliferation (Irazoqui et al. 2001; Jeschke et al. 2006), induce apoptosis (Cao et al. 2008) or generate antibody-dependent cellular cytotoxicity. TF is a promising vaccination target (Fung et al. 1990; MacLean and Longenecker 1991; Adluri et al. 1995; Springer 1997; Slovin et al. 2005). In order to induce an effective immune response against the tumor-specific TF, it has to be presented in the context of a suitable density, conformation and surrounding, which may only be found under natural conditions, such as those found on tumor cell membranes or bacterial cell walls. An additional benefit of bacterial components is that they have adjuvant-like effects. Therefore, selected bacterial strains that carry true TF structures may be appropriate candidates for a tumor vaccine. Such bacteria should be further investigated with regard to their potency to induce an immune response against TF, as well as their ability to prevent tumor formation and distribution in vivo. In conclusion, we have isolated two commensal bacterial strains by means of a TF-affinity enrichment procedure, which display the true human tumor-specific carbohydrate antigen TF on their surface. They were identified as two novel but closely related strains of B. ovatus. The structure of the TF disaccharide was identified as side chain of a repeat of the CPS. Therefore, the TF epitope is obviously exposed and present at suitable densities on the bacterial surface in order to be immunogenic. This structure may provide the basis for further studies to develop a new, effective, non-toxic and easy to produce TF-specific anti-tumor vaccine. Materials and methods Origin and maintenance of bacterial strains Bacterial strains were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany), the American Type Culture Collection (ATCC, Manassas), the collection of the German Institute of Human Nutrition (DIfE, Table II) or were isolated from human faeces (Table I). The E. coli strains lac+ and lac were kindly provided by Markus Heimesaat and Stefan Bereswill (Institute for Microbiology and Hygiene, Charit - University Medicine Berlin). Selective enrichment and isolation of bacteria from human fecal samples For enrichment of TF-positive bacteria from human fecal samples, anti-TF mAb-coated magnetic beads were prepared according to the manufacturers recommendations (Dynal Biotech ASA, Oslo, Norway) with either the NM-TF1 or the NM-TF2 mAb (Table V, Glycotope GmbH, Berlin, Germany). Fecal samples from eight healthy human subjects (AH) who had not taken antibiotics during the last 3 months were Antigen detected Fine specificity Reference NM-TF1 Mouse IgM NM-TF2 Mouse IgM HH8a Mouse IgM A68-B/A11 Mouse IgM A68-E/E3 Mouse IgG1 A68-E/A2 Mouse IgG1 A63-B/C2 Mouse IgM aKindly provided by Henrik Clausen (Copenhagen) to the UK. bCore 2 = Gal (1-6GlcNAc) 1-3GalNAc1-. examined in the study. Samples were maintained under anaerobic conditions (Anaerogen, Oxoid, Basingstoke, UK) and stored at 4C for a maximum of 4 h before processing in an anaerobic chamber (Don Whitley Scientific, Shipley, England, or Coy Laboratory Products, Grass Lake). A 10-fold (w/v) dilution of the fecal samples was prepared in reduced phosphate-buffered saline (PBSred: 8.5 g NaCl, 0.3 g KH2PO4, 0.6 g Na2HPO4, 0.1 g peptone and 0.25 g cysteineHCl per liter, pH 7.0). Six sterile 3 mm diameter glass beads were added and the samples were homogenized by vortexing. The fecal suspensions were centrifuged (300 g, 1 min, 21C) to sediment debris. The supernatant containing the bacteria was aspirated and further diluted 10-fold. The bacteria were washed once (8000 g, 5 min, 21C) and re-suspended in PBSred. A 20 L volume of the bacterial suspension was added to 180 L of PBSred and 5 L of either NM-TF1- or NM-TF2-coated magnetic beads. This mixture incubated for 30 min at room temperature. Subsequently, the beads were suspended in 1 mL PBSred and washed three times. Aliquots (100 L) were spread-plated on de Man-Rogosa-Sharpe agar (Merck, Darmstadt, Germany), Bifidus Selective medium (Fluka, St Gallen, Switzerland) and non-selective media such as nutrient agar, Schaedler anaerobe agar, WilkinsChalgren agar (Oxoid), brain heart infusion agar (Biomrieux, Marcy lEtoile, France), blood agar (Biomrieux) and ST agar (as broth with 1.5 g/L agar; Kamlage et al. 2000) and were incubated for 48 h at 37 C in an anaerobic chamber. For subjects EH, this enrichment procedure was repeated twice: following the 48 h incubation, colonies were suspended in PBSred to McFarland turbidity standards 35 (Smibert and Krieg 1994), and a 20 L aliquot of this suspension was again added to NM-TF1- or NM-TF2-coated beads. Colonies were picked randomly from agar plates and re-streaked several times on non-selective media. Growth and fixation of bacteria for TF-specific ELISA Colonies were picked and inoculated into ST, WC or MRS broth and grown at 37C in an 80% N2 and 20% CO2 [v/v] atmosphere. These cultures were subcultured (diluted 1:100) and grown to a stationary phase. Cells were harvested (8000 g, 15 min, 4C) and suspended in one volume of PBSM (8 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4 and 0.24 g KH2PO4 per liter; Sambrook et al. 1989). Three volumes of 4 % paraformaldehyde (Hugenholtz et al. 2001) in PBSM were added to the cell suspension, which was fixed for 34 h at 4C. Fixed bacteria were washed with an equal volume of PBSM (8000 g, 15 min, 4C) and the pellets suspended in one volume of PBSM, followed by addition of an equal volume of ice-cold ethanol. Samples were stored at 20C until analysis. Identification of bacterial isolates Bacteria were characterized with the VITEK System; API 50 CHL strips (Biomrieux) were used for further description of lactobacilli. For the identification of interesting isolates or those for which the biochemical identification was ambiguous, the 16S rRNA genes were PCR-amplified and partially sequenced. DNA was extracted using the Invisorb Genomic DNA Kit III according to protocol III B (Invitek, Berlin, Germany) or with phenol/chloroform/isoamyl alcohol (Hanske et al. 2005). The 16S rRNA genes were amplified with the primers 27f (5-AGAGTTTGATCCTGGCTCAG) and 1492r (5-TACCTTGTTACGACTT; Kageyama et al. 1999). The 50 L PCR mixture contained 50 mM KCl, 20 mM TrisHCl, 2.5 mM MgCl2, 0.25 mM each dNTP, 0.1 M of each primer, 2.5 U of Taq-DNA polymerase (Invitrogen, Carlsbad, CA) and 1 L of template DNA. The PCR program was at 94C for 4 min, 25 cycles at 94C for 1 min, at 55C for 1 min and at 72C for 1 min and finally at 72C for 10 min. PCR products were purified (High-Pure PCR Product Purification Kit, Roche, Indianapolis) and the DNA concentration and product size estimated (Low DNA Mass Ladder, Invitrogen). For sequencing (DYEnamicTM ET Dye Terminator Cycle Sequencing Kit, Amersham Biosciences, Little Chalfont, England; AMODIA, Braunschweig, or MWG Biotech, Martinsried, Germany) of selected regions of the 16S rRNA genes, the primers 27F, 338R (5-GCTGCCTCC CGTAGGAGT; Amann et al. 1990), 338F (5-ACTCCT ACGGGAGGCAGC), 968F (5-AACGCGAAGAACCTTAC; Zoetendal et al. 1998), 968R (5-GTAAGGTTCGCGTT) or 1492R were used. Sequences were assembled and edited with ContigExpress (Vector NTI Suite 9.0.0, Invitrogen) and aligned with ClustalW (BioEdit, version 7.02.1, Tom Hall, Ibis Therapeutics, Carlsbad, CA), with similar sequences obtained with BLAST (Altschul et al. 1990) and Sequence Match (Ribosomal Database Project; Cole et al. 2005). Percentages of similarity were calculated from unambiguously aligned sequences (Sequence Identity Matrix, BioEdit and Similarity Matrix, Ribosomal Database Project II). Strains with sequence similarities 97% with a given type strain were classified as such. RAPD pattern and profile analysis The primers OPX14 (5-ACAGGTGCTG), OPL7 (5-AGGCG GGAAC), M13 (5-GAGGGTGGCGGTTCT), OPA16 (5-AG CCAGCGAA) and OPA18 (5-AGGTGACCGT) were used for amplification of template DNA. The 50 L PCR mixture contained: 50 mM KCl, 20 mM TrisHCl, 1 mM MgCl2, 1.5 mM Mg(OAc)2, 0.25 mM each dNTP, 2 M primer, 2.5 U of Taq-DNA polymerase and 1 L of template DNA. The PCR program was: at 95C for 5 min, 30 cycles of at 95C for 1 min, at 50C for 1 min and at 72C for 1 min and finally at 72 C for 6 min. Isolates with identical band patterns for all primers on 1% agarose gels were assigned the same RAPD profile. ELISA to screen for the occurrence of TF in bacteria Fixed bacterial cells were adjusted to an optical density (OD600) of 0.1 in PBSE (9 g NaCl, 0.528 g Na2HPO42H2O and 0.144 g KH2PO4 per liter, pH 7.4) and concentrated 10-fold. Approximately 5 108 cells in 50 L were applied in triplicate to the wells of a PolySorp microtitre plate (Nunc, Wiesbaden, Germany) and coated overnight at 37C. Prior to all further incubation steps, the plates were washed three times with 200 L Tris-buffered saline with Tween 20 (TTBS; 8.78 g NaCl 6.06 g Tris per liter and 0.05% [v/v] Tween 20, pH 7.6). Residual binding sites were blocked by incubating the wells with 200 L of 2% bovine serum albumin (BSA) in PBSE for 20 min. Primary antibodies (Table V) were applied in 50 L of 1% BSA containing PBSE and incubated for 1 h. The secondary antibody ( peroxidase-rabbit-anti-mouse IgG/IgM P0260, DAKO, Hamburg, Germany) was diluted 1/5000 in 1 % BSA in PBSE, and 50 L were applied per well and incubated for 1 h. Plates were developed for 20 min in the dark by adding 100 L of developing solution (1 mg/mL tetramethylbenzidine in 1% [v/v] DMSO in 50 M sodium acetate buffer, pH 4.5) to each well. Subsequently, 50 L of 2.5 M H2SO4 was added to stop the reaction, and the absorbance (E450/630) was measured in an ELISA Reader (Dynex Technologies Inc., Chantilly, VA). Asialoglycophorin and glycophorin (100 ng/well in PBSE, Sigma-Aldrich, Taufkirchen, Germany) served as positive and negative controls for TF, respectively. The background was determined with monoclonal antibody A63-B/C2 (Glycotope GmbH), which is specific for glycophorin. The assays were performed on at least two separate occasions, using several different batches of bacterial preparations for some strains, especially the B. ovatus strains. The binding behavior of antibodies toward the bacteria was determined semi-quantitatively. Mean absorbance values at least 3 times higher than the background were classified as weakly positive (+), 5 times higher as strongly positive (++) and 10 times higher as very strongly positive (+++). Absorbance values that were less than three times the background signal or an absorbance reading <0.3 were classified as negative (). Periodate oxidation to confirm carbohydrate nature of antigens After coating of the bacterial antigens (see ELISA), 50 L of 50 mM sodium acetate buffer ( pH 4.5) per well was added and incubated for 5 min. Next, 50 L of 10 mM sodium periodate in sodium acetate buffer was added and incubated for 1 h in the dark. Sodium acetate buffer without periodate was used as a control. Thereafter, all wells were incubated for 5 min with 50 L sodium acetate buffer. This was followed by a 30 min incubation step with 50 L of 50 mM borohydride in PBSE. After that, plates were washed five times with PBSE, and the ELISA was carried out as described in the section SDSPAGE and immunoblotting, beginning with the blocking step. Extraction of LPSs and CPSs Smooth LPS was extracted according to the hot phenol water method of Westphal and Jann (Apicella et al. 1994). Crude CPS was extracted with hot phenol water as described previously (Pantosti et al. 1991). Dialyzed extracts were frozen in liquid N2, lyophilized and stored at 20C. SDSPAGE and immunoblotting SDSPAGE was performed according to the method of Laemmli (1970) with 12% separating and 5% stacking gels. A pre-stained marker (BenchMark ladder, Invitrogen) was used to determine the molecular weight. To visualize polysaccharides, gels were stained with Alcian Blue as described (Karlyshev and Wren 2001). SDSPAGE-separated samples were electrophoretically transferred (transfer buffer: 25 mM Tris, 192 mM glycine, 20% methanol, pH 8.3) to nitrocellulose membranes using a semi-dry blotting device (Mini Trans-Blot, BioRad, Munich, Germany) as previously described (Towbin et al. 1979). For dot blots of fractions, 5 L were applied directly to nitrocellulose membranes and dried. Blotted carbohydrates were detected with the DIG Glycan Detection Kit (Roche) as described in the manufacturers protocol. TF was identified on the blot by staining with the TF-specific monoclonal antibodies NM-TF1, NM-TF2 and HH8 (Table V). The monoclonal antibodies A68E/E3 and A68E/A2, which are specific for the anomer of TF, were also included. For immunoblotting, blots were blocked in Tris-buffered saline (0.5 M Tris, 0.15 M NaCl, pH 7.6), containing 1% TTBS and 3% BSA for 2 h overnight at room temperature. Blots were washed three times with TTBS and the primary monoclonal antibodies were applied at a concentration of 0.1 g/mL in TTBS and incubated for 2 h at room temperature. Membranes were washed three times with TTBS and the secondary antibody [horseradish peroxidase-conjugated goat anti-mouse immunoglobulin (Sigma-Aldrich)], diluted 1:5000 in TTBS, was incubated for 90 min at room temperature. Binding was visualized with a Chemiluminescence Kit (Super Signal, Pierce, Rockford, IL). Different concentrations of asialoglycophorin were used as a TF-positive standard to determine the intensity of antibody binding. Monosaccharide analysis of CPS and LPS extracts The monosaccharide composition of crude LPS and CPS preparations from B. ovatus D6 and DSMZ 1896 was compared by HPAEC-PAD. LPS and CPS extracts (500 mg) were hydrolyzed with 2 M trifluoroacetic acid at 100C for 4 h. Separation was achieved with HPAEC-PAD on an ICS-3000 high-performance chromatograph (Dionex, Eggstein, Germany). The chromatograph was equipped with a CarboPak PA1 column (12 250 mm, 001390, Dionex) protected by a PA1 guard column (2 50 mm, 001414) maintained at 25C and an electrochemical cell (gold working electrode, AgCl reference electrode). Neutral polysaccharides were eluted isocratically with 15 mM KOH from the column at a flow rate of 0.25 mL/min. Acidic polysaccharides were eluted with the sodium acetate gradient (06 M sodium acetate in 100 mM NaOH). For quantification of sugars, fructose and 2desoxyribose (20 pmol/10 L) were used as internal standards. Glucose, fucose, galactose, mannose, rhamnose, galacturonic acid, glucuronic acid, Nacetylglucosamine, Nacetylgalactosamine (10 pmol/10 L), 2desoxy ribose and fructose (20 pmol/10 L; Sigma-Aldrich) were used as external standards. The proportion of monosaccharides was calculated from the relative mass of each monosaccharide in relation to all monosaccharides present (molar ratio). For the detection of KDO, hydrolyzed KDO was labeled fluorometrically with 1,2diamino-4,5methylenedioxybenzen (DMB) and separated by RP-HPLC (RP18 Hypersil ODS 3 m, Phenomenex, Aschaffenburg, Germany) with fluorometric detection, as described previously (Hara et al. 1989). The KDO content was quantified with external DMB-KDO standards. Accumulation of TF-positive polysaccharides by RP- and DEAE-HPLC For mass spectrometric analyses, CPS and LPS extracts (4 mg each) were separated and TF-positive polysaccharides accumulated by RP-HPLC (Synergi 4m Fusion-RP 80, 10 250 mm, Phenomenex), by applying a gradient of propanol and methanol with an elution rate of 2 mL/min as described (Hashimoto et al. 2001). Fractions of 2 mL were collected. Polysaccharide-containing fractions were identified by dot blot staining with a DIG Glycan Detection Kit as described in the section SDS-PAGE and immunoblotting. TF-containing polysaccharide fractions were pooled. They were fragmented by mild acid hydrolysis (5% acetic acid, 1 h, 100C) and purified by anion exchange-HPLC (DEAE-Sephacel 4.6 250 mm, Phenomenex) with a gradient of 00.5 M NaCl and an elution rate of 1 mL/min as described previously (Tzianabos et al. 1992). Structural analysis of the TF-positive polysaccharide by MS The structure of TF-positive polysaccharide was analyzed by matrix-assisted laser desorption/ionization time-of flight (MALDI-TOF) MS and by electrospray-ion-trap-MS (ESI-Ion-Trap-MS). Mass spectrometric analyses (MS and MS/MS) were carried out in the positive mode. For MS, the purified TF-containing polysaccharide was fragmented either by hydrolysis with 1% acetic acid (1.5 h, 100C) or by enzymatic digestion with chondroitinase ABC (Sigma-Aldrich) at 37C overnight. Aliquots of glycan fragments were labeled by the fluorophore 2-amino benzamide (Bigge et al. 1995) and separated by normal phase HPLC (Luna 3 m NH2 100 A, 4.6 250 mm, Phenomenex) with fluorescence detection. Unseparated and separated fragments were analyzed by MS. Mass spectra were measured using a Biflex-MALDI-TOF mass spectrometer (Bruker, Bremen, Germany) equipped with a 337 nm pulsed nitrogen laser in the reflectron and positive ionization mode. About 5 pmol of glycans were mixed with the matrix arabinosazon (10 mg/mL in 80% ethanol) in a 1:1 ratio. The molecular masses of unknown glycans were determined by calibrating the mass spectrometer with a mixture of reducing glucose oligomers from dextran on the same target. After external calibration, between 150 and 300 scans were averaged for each spectrum. For mass determination by electrospray ionization, an ESI-Ion-Trap mass spectrometer (Agilent, Bblingen, Germany) equipped with a nanoelectrospray ionization source was used. The spray voltage was 3.6 kV and the spray temperature was 325C. The masses were measured in a positive ionization mode. Masses between m/z 50 and 900 were determined. Five scans were averaged for each spectrum. Samples were diluted (1:50, 1:100) in 2.5 mM NH3 in 40% acetonitrile and 50 L were injected. For the verification of glycosidic bonds between the monosaccharides and the identification of monosaccharide components of the fragments, oligosaccharides were digested with or without prior acetic acid treatment with the specific exoglycosidases 1-3-galactosidase (from Xanthomonas manihotis, New England BioLabs, Frankfurt, Germany), 1-3,4-fucosidase (from almond meal), 1-4-galactosidase (from Streptococcus pneumoniae, ProZyme Europa Bioproducts, Cambridge, UK) and/or -1,4-HexNAcase (from chicken liver, Sigma-Aldrich). The digestions were run following manufacturers instructions (usually overnight at 37C). Undigested samples were used as negative controls. Successful digestion was controlled by MS and cleaved terminal monosaccharides were identified by HPAEC-PAD. For further analyses, the fragment Hex-desHex-desHexdesHexM-HexNAc-HexNAc was completely hydrolyzed with 2 M trifluoro acetic acid, and the monosaccharide composition analyzed by HPAEC-PAD. Funding This work was partially funded by the German Federal Ministry of Education and Research (BioProfile Nutrigenomik Project 1202-27). Acknowledgements We would like to dedicate this paper to our friend and colleague Martin Zimmermann-Kordmann, who sadly passed away in 2006. We are grateful to Jenny Merz, Anke Ghler and Marion Urbich for technical assistance, Martin Osterhoff for sequencing and Thomas Clavel for critical reading of the manuscript and helpful discussions. Conflict of interest None declared. Abbreviations ATCC, American Type Culture Collection; BSA, bovine serum albumin; CPS, capsular polysaccharide; DMB, 1,2diamino4,5methylenedioxybenzen; ELISA, enzyme-linked immunosorbent assay; ESI-Ion-Trap-MS, electrospray-ion-trap-mass spectrometry; HPAEC-PAD, high-pH anion-exchange chromatography and pulsed amperometric detection; HPLC, high-petroleum liquid chromatography; KDO, 2-keto-3deoxyoctonic acid; LPS, lipopolysaccharide; MALDI-TOF, matrix-assisted laser desorption/ionization time-of-flight; MS, mass spectrometry; PBS, phosphate-buffered saline; RAPD, randomly amplified polymorphic DNA; SDSPAGE, sodium dodecylsulfatepolyacrylamide gel electrophoresis; TF, ThomsenFriedenreich antigen; TTBS, Tris-buffered saline with Tween.

This is a preview of a remote PDF: http://glycob.oxfordjournals.org/content/21/10/1277.full.pdf

Gemma Henderson, Philippe Ulsemer, Ute Schöber, Anja Löffler, Carl-Alfred Alpert, Martin Zimmermann-Kordmann, Werner Reutter, Uwe Karsten, Steffen Goletz, Michael Blaut. Occurrence of the human tumor-specific antigen structure Galβ1-3GalNAcα- (Thomsen–Friedenreich) and related structures on gut bacteria: Prevalence, immunochemical analysis and structural confirmation, Glycobiology, 2011, 1277-1289, DOI: 10.1093/glycob/cwr058