a tetraantennary glycan with bisecting n-acetylglucosamine ... · glycans of bovine...

Glycobiology vol. 18 no. 1 pp. 42–52, 2008doi:10.1093/glycob/cwm113Advance Access publication on October 19, 2007

A tetraantennary glycan with bisecting N-acetylglucosamine and the Sda antigen is thepredominant N-glycan on bovine pregnancy-associated glycoproteins

Karl Klisch1,2,3, Evelyne Jeanrond3, Poh-Choo Pang4,Andreas Pich5, Gerhard Schuler6, Vibeke Dantzer7,Mariusz P Kowalewski8, and Anne Dell4

3Abteilung Neuroanatomie, Medizinische Hochschule Hannover, 30625Hannover, Germany; 4Division of Molecular Biosciences, Faculty of NaturalSciences, Imperial College, London SW7 2AZ, UK; 5Abteilung furToxikologie, Medizinische Hochschule Hannover, 30625 Hannover, Germany;6Klinik fur Geburtshilfe, Gynakologie und Andrologie der Groß- undKleintiere, Justus-Liebig-Universitat Giessen, 35392 Giessen, Germany;7Department of Basic Animal and Veterinary Sciences, Faculty of LifeSciences, University of Copenhagen, Denmark; and 8Institut furVeterinaranatomie, Justus-Liebig-Universitat Giessen, 35392 Giessen,Germany

Received on May 21, 2007; revised on September 14, 2007; accepted onOctober 1, 2007

Pregnancy-associated glycoproteins (PAGs) are major secre-tory proteins of trophoblast cells in ruminants. Binucleatetrophoblast giant cells (BNCs) store these proteins in secre-tory granules and release them into the maternal organismafter fusion with maternal uterine epithelial cells. By matrixassisted laser desorption ionisation-mass spectrometry(MALDI-MS) analysis and linkage analysis, we show thatby far, the most abundant N-glycan of PAGs in mid-pregnancy is a tetraantennary core-fucosylated structurewith a bisecting N-acetylglucosamine (GlcNAc). All fourantennae consist of the Sda-antigen (NeuAcα2-3[GalNAcβ1-4]Galβ1-4GlcNAc-). Immunohistochemistry with the mono-clonal antibody CT1, which recognizes the Sda-antigen,shows that BNC granules contain the Sda-antigen fromgestation day (gd) 32 until a few days before parturition.Lectin histochemistry with Maackia amurensis lectin(MAL), which binds to α2-3sialylated lactosamine, showsthat BNC granules are MAL-positive prior to gd 32 andalso at parturition. The observed tetraantennary glycan isa highly unusual structure, since during the synthesis ofN-glycans, the insertion of a bisecting GlcNAc inhibits theactivity of the GlcNAc-transferases that leads to tri- andtetraantennary glycans. The study defines the substantialchanges of PAG N-glycosylation in the course of pregnancy.This promotes the hypothesis that PAGs may have differ-ent carbohydrate-mediated functions at different stages ofpregnancy.

Keywords: cattle/gestation/glycosylation/massspectrometry/placenta

1Present address: School of Veterinary Medicine and Science, University ofNottingham, Loughborough LE11 5RD, UK2To whom correspondence should be addressed. Tel: +44 11595 16464Fax: +44 11595 16415; e-mail: [email protected]

Introduction

The fetal binucleate trophoblast giant cells (BNCs) in the ru-minant placenta produce several glycoproteins and store theseproteins in cytoplasmic granules. Mature BNCs fuse with mater-nal uterine epithelial cells, exocytose the granules at the basalmembrane of the uterine epithelium, and thereby deliver theproteins into the maternal organism (Wooding 1992). This pro-cess happens continuously from the onset of BNC-formation inthe third week of gestation until parturition. Among the cargo-proteins of the granules are pregnancy-associated glycoproteins(PAGs), which belong to the protein family of aspartic pro-teinases. In the evolution of ruminants, gene duplications led toa high number of PAG-genes (∼100) in cattle and sheep (Xieet al. 1997). The high ratio of nonsynonymous to synonymousmutations in the evolution of PAG sequences led to the sugges-tion that natural selection caused the diversification of ruminantPAGs (Xie et al. 1997). This suggests that the diversity of PAGsis of functional importance; but the function of PAGs is still notwell characterized.

Lectin histochemical studies show that BNC granules havea very specific glycosylation pattern (Munson et al. 1989;Lehmann et al. 1992; Jones et al. 1994; Nakano et al. 2002;Klisch and Leiser 2003). During most of the time of preg-nancy, the granules can be labeled with Phaseolus vulgarisleucoagglutinin (PHA-L), which recognizes branched tri- andtetraantennary glycans with ß1-6 linked GlcNAc, and withDolichos biflorus agglutinin (DBA), which binds to terminalN-acetylgalactosamine (GalNAc) residues. The main targetsof these lectins in BNCs are PAGs and the GalNAc-bindinglectins proved to be useful tools for affinity-chromatographicPAG purification (Klisch and Leiser 2003; Klisch et al. 2005).Before parturition, the glycosylation of PAG changes and ter-minal GalNAc residues are largely absent at term (Klisch et al.2006).

This very specific pattern of PAG-glycosylation and its tem-poral changes during gestation strongly indicate a functional roleof the carbohydrates during pregnancy. In the present study, wecharacterize the main N-glycans which are attached to PAGs andwe further characterize the changes of glycosylation that occurin early pregnancy and at term.

Results

MALDI-MS of native and desialylated PAG-glycansPurification and characterization of PAGs was done as previ-ously described (Klisch et al. 2005). From the purified PAGs,the N-glycans were released by digestion with PNGase-F. TheMALDI-time of flight (TOF) spectrum in native glycan prepa-rations (Figure 1) was dominated by a molecular ion at m/z4697.2. Treatment of the native samples with neuraminidase

C© The Author 2007. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 42

at Pennsylvania State University on February 22, 2013

http://glycob.oxfordjournals.org/D

ownloaded from

http://glycob.oxfordjournals.org/

Glycans of bovine pregnancy-associated glycoproteins

Fig. 1. MALDI-TOF mass spectra of native PAG N-glycans. (A) MS of native N-glycans released by PNGase F acquired in linear negative ion mode (M−H)−. Thelow resolution of the MS acquired in linear mode does not allow the 12C peak to be distinguished. Therefore, the mass indicated is the average mass. (B) MS ofnative N-glycans which were desialylated using neuraminidase from C. perfringens and data were acquired in linear positive ion mode (M+Na)+. The sugarsymbols are of those employed by the Consortium for Functional Glycomics for the representation of glycan structures.

reduced the value to m/z 3556.2, which corresponds to a loss offour molecules of N-acetylneuraminic acid (NeuAc). It shouldbe mentioned that we were able to desialylate the PAG-glycansonly with the Clostridium perfringens neuraminidase fromRoche (molecular weight (MW) 60 kDa), but not with C. per-fringens neuraminidase from New England Biolabs (MW 41kDa) (data not shown). The peak at 3847 in the neuraminidasetreated samples results from incomplete desialylation, with one

molecule of sialic acid left. These major compounds are corefucosylated tetraantennary structures with bisecting GlcNAcand terminal NeuAcα2-3[GalNAcβ1-4]Galβ1-4GlcNAc (Sda)-epitopes on all four antennae as explained in the subsequentsection. Minor compounds are the corresponding triantennarystructures (m/z 3837.7, 2988.0 in the native and desialylatedsample respectively) and their counterparts without core fuco-sylation (m/z 3690.8, 2841.3).

43



ownloaded from


K Klisch et al.

Fig. 2. MALDI-TOF mass spectrum of permethylated PAG N-glycans. The glycans were released by PNGase F, permethylated, and subsequently subjected toSep-Pak cleanup. Data from one of the fractions collected upon Sep-Pak cleanup are shown. The MS was acquired in reflectron positive ion mode (M+Na)+. Peakassignments are based on molecular mass composition of the 12C isotope together with the knowledge of the biosynthetic pathways. Sugar symbols are as in Figure1 and the structures were confirmed by MS/MS analyses. (A) The m/z 1071 and m/z 1093 components are protonated and sodiated forms, respectively, of A-typeions resulted from in-source fragmentation. The peak which is labeled with an “x” is due to undermethylation of the m/z 5812 component. The minor molecular ionat m/z 5900 is not assigned because unambiguous MS/MS data were not obtained. The mass of this ion is consistent with a composition which differs from the m/z5812 component by the loss of one sialic acid and addition of Hex and HexNAc. (B) MALDI-TOF/TOF tandem mass spectrum of the m/z 5812 component.

MALDI-MS and MS/MS of permethylated PAG N-glycansTo corroborate the results obtained from MS analysis of nativesamples, and to facilitate unambiguous sequencing by MS/MS,the PAG glycans were permethylated and analyzed by MALDI-TOF and MALDI-TOF/TOF. The MALDI-TOF spectrum(Figure 2A) was dominated by a molecular ion at m/z 5812,

which is the predicted value for the permethylated counterpartof m/z 4697 in the native sample. The second most abundant sig-nal in the spectrum is m/z 1071, which is a fragment ion derivedfrom the antennae (see annotation in Figure 2B). Several addi-tional glycans are present but they are of low abundance. Thus,the cluster of weak signals near m/z 2000 are molecular ions for

44



ownloaded from



Fig. 3. MALDI-TOF mass spectrum of permethylated N-glycans from gd 220-placenta (A, C) and from a placenta obtained approximately 1 day ante partum(B, D). The main differences between the two samples is the presence of Sda-capped glycans (m/z 4583, 4757, 5812) only in the gd 220 sample.

high mannose glycans (Man6−9 at m/z 1784, 1988, 2192, and2396, respectively), and the minor signals between m/z 3500and m/z 5000 correspond to bi- and triantennary glycans withvarying levels of core fucosylation. For example, m/z 3702 and4757 are bi- and triantennary analogues of m/z 5812, whilst m/z3528 and 4583 are their nonfucosylated counterparts.

The major component at m/z 5812 was subjected to collisionalactivation in a MALDI-TOF-TOF experiment (Figure 2B). Thefragmentation pattern is dominated by cleavage of the antennaeon the reducing side of GlcNAc to yield m/z 4743 (loss of theSda-epitope from the molecular ion) and m/z 1071 (the Sda-

epitope itself; m/z 1093 is a sodiated adduct). The signal at m/z5437 corresponds to loss of sialic acid from the molecular ion.

Glycomics analysis of placental tissuesPermethylated N- and O-glycans from unfractionated placen-tomal tissue of one mid-pregnant (gd 220) and one ante partal(approximately one day before term) cows were analyzed. Thisexperiment allows the comparison of the general glycosylationpattern with the glycosylation of PAGs and it also gives MS-dataabout the ante partal changes of glycosylation. The N-glycanspectra of gd 220 (Figure 3A and C) and preterm (Figure 3B

45



ownloaded from


K Klisch et al.

Fig. 4. MALDI-TOF mass spectrum of permethylated O-glycans from gd 220-placenta (A), from a placenta obtained approximately 1 day ante partum (B), andfrom purified PAG (C).

and D) tissues were both dominated by high-mannose glycansand by core fucosylated bi-(m/z 2652), tri- (m/z 2897, 3305,3550), and tetraantennary (m/z 3754, 3958, 4145) glycans withcapping gal–gal residues. The high mannose glycans were alsoobserved among the PAG N-glycans (see Figure 2), while the lat-ter structures were completely absent in the PAG-glycans. Themain difference between the two samples were the Sda-cappedglycans (m/z 4583, 4757, 5812), which were only observed inthe gd 220-sample, but not in the near-term sample. Also, theSda fragment ion (m/z 1071) was only observed in the gd 220sample.

The analysis of permethylated O-glycans also revealed dif-ferences mainly concerning the presences of Sda-epitopes. Am/z 1501 ion was observed in the gd 220 (Figure 4A) sampleand in the PAG-O-glycans (Figure 4C), but not in the prepartumsample (Figure 4B). The PAG-O-glycans were dominated bythe Sda fragment ion (m/z 1071).

Linkage analysis of permethylated PAG N-glycansA sample of permethylated PAG-glycan was converted to amixture of partially methylated alditol actetates which were sub-jected to GC-MS linkage analysis. The region of the total ionchromatogram that was especially informative is reproduced inFigure 5. One of the major peaks is 3,4-linked Gal which is de-rived from the Sda-epitope. The abundant 3,4,6-linked mannose(Man) peak provides evidence for bisecting GlcNAc, whilst the

major peaks for 2,4- and 2,6-linked Man confirm the presenceof tri- and/or tetraantennary glycans. The less abundant 3,6-Man peak is attributed to the minor high mannose population.The fact that terminal GalNAc is significantly more abundantthan terminal GlcNAc corroborates the assignment of the Sda-antennae. The presence of a major peak for 4-GlcNAc, andthe absence of 3-GlcNAc, defines type 2 antennae backbones(Gal1-4GalcNAc). Fucosylation of the core is confirmed by the4,6-GlcNAc peak.

Western analysisThe CT1 antibody reveals one major band at approximately67 kDa and a minor band at 75 kDa in the non-N-deglycosylatedsamples (Figure 6). These bands are not present in the PNGaseF treated samples. The PAG antiserum shows a generally moreintensive staining with several bands in the glycosylated anddeglycosylated samples. The two most intensive bands in thenon-N-deglycosylated sample are of the same MW as in theCT1 staining and are also missing in the deglycosylated samples.In the prolactin related protein-I (PRP-I) staining, the PNGaseF treatment results in a shift from approximately 37 kDa to25 kDa.

HistochemistryPAG positive BNCs were detected in all studied stages of preg-nancy (Figure 7). PHA-L is generally colocalized with PAG.

46



ownloaded from



Fig. 5. Total ion gas chromatogram of the monosaccharide linkage analysis of the permethylated PAG N-glycans. Permethylated N-glycans were converted to amixture of partially methylated alditol acetates, which were subjected to GC-MS linkage analysis. The peaks which are labeled with an “x” are contaminant peaks.

Fig. 6. Western-Blot analysis of placentomal homogenates of a late pregnant(gd 260) cow. The protein samples (10 µg/lane) were either enzymaticallydeglycosylated with PNGase F (PNGase +) or left untreated (PNGase −). Inthe untreated samples, the bands of CT1-binding glycoprotein (main band at67 kDa, minor band at 75 kDa, marked by arrows) are of identical size as thePAG-bands. The nondeglycosylated PRP-I migrates at approximately 37 kDaand only a very faint band (arrowhead) of this size might be seen in the CT1staining. The enzymatical deglycosylation leads to the disappearance of thebands (CT1) or to a shift toward lower molecular weights (PAG, PRP-I).

In gd 20 and 23, all PAG-positive BNCs are also labeled withMaackia amurensis lectin (MAL), but not with CT1 (Figure 8).The staining pattern substantially changes after gd 30: there isstrongly reduced MAL staining, but the granules become CT1-positive. The staining with DBA was mainly similar to thatof CT1 and is therefore not shown. This staining pattern withCT1-positive or DBA-positive granules continues until a fewdays before parturition. An intermediate staining was observedin the samples which were taken approximately five days beforeparturition. In the term samples, BNCs were mainly CT1/DBA-negative and MAL-positive.

PCRThe relative gene expression of the β4GalNAcT-II is shown inFigure 9. Due to the uneven distribution of the data obtained byReal Time RT-PCR, the Kruskal–Wallis Test (a nonparametricANOVA) was applied with the statistical software program,GraphPad3 (GraphPad Software, Inc., San Diego, CA,). TheKruskal–Wallis Test (nonparametric ANOVA) considered thedifferences between the different stages of pregnancy as notsignificant (p = 0.3430).

Discussion

This report shows that one carbohydrate structure is predominat-ing on bovine PAGs in midpregnancy. This is a core-fucosylatedtetraantennary glycan in which all antennae carry a terminal Sda-antigen (Figures 1–3 and 5). The occurrence of bisecting Glc-NAc in tetra- and triantennary glycans is a highly unusual fea-ture. Generally, the attachment of bisecting GlcNAc to the coreβ-mannosyl residue by the N-acetylglucosaminyltransferase-III(GlcNAc-TIII) blocks the initiation of a tri- and tetraantennarybranching pattern by GlcNAc-TIV and GlcNAc-TV (Schachter1986). This usually results in a reduced branching of N-glycansin cells which express GlcNAc-TIII. A strictly regulated lo-calization of these GlcNAc-transferases in the Golgi-cisternaecould circumvent this, and could thereby be a preconditionfor the synthesis of highly branched glycans with a bisectingGlcNAc (Sasai et al. 2003). Such a sublocalization, inwhich GlcNAc-TIV and -TV are localized in earlier Golgi-subcompartments than GlcNAc-TIII, should facilitate the initi-ation of a tetra- and tri-antennary branching pattern before thebisecting GlcNAc is attached. In addition, GlcNAc-TIII itself isinhibited by the addition of galactose to the growing antennaeby the ß-1,4-galactosyltransferase (Fukuta et al. 2000). Thisimplies that GlcNAc-TIII should also be spatially well sepa-rated from ß-1,4-galactosyltransferase or must be abundantlyexpressed.

47



ownloaded from


K Klisch et al.

Fig. 7. Immuno- and lectin histochemical staining of gd 23, gd 110, and termplacental tissue. At gd 23, the PAG-positive BNC granules (arrowheads) arelabeled with MAL- and PHA-lectins. At this stage of pregnancy, theCT1-antibody stains the BNC nuclei (probably unspecifically) and the basalmembrane, but not the BNC granules. At gd 110, the granules are labeled withCT1. With MAL, an intensive staining of material outside of the BNC can beseen, but the granules inside the cells (arrowheads) are negative. In the termplacenta, most of the PAG-positive BNCs bind MAL again. At this time, thevast majority of BNCs do not bind the CT1-antibody (arrowheads), and onlyvery few of the BNCs are CT1-positive (arrow) (magnification bar = 20 µm).

Fig. 8. Percentage of CT1- and MAL- positive BNCs at different stages ofpregnancy. Major changes of glycosylation occur in the early pregnancy(between gd 20–23 and gd 32–43) and in the last week of pregnancy (around5 days ante partum [d a.p.]).

Fig. 9. Relative gene expression (RGE) of β4GalNAcT-II-mRNA in bovineplacentomes at different stages of pregnancy.

A second unusual feature is the relative uniformity of thetetraantennary Sda-glycan on PAGs in BNCs. Since most otherglycoproteins show a large variety of attached glycans, withdominating bi- and triantennary structures, the dominance ofthe tetraantennary PAG glycans points at a highly regulatedglycosylation machinery in BNCs. An alternative explanationfor this phenomenon could be a bias in the analyzed PAG-sample, which might result from the final step of the purificationprocedure. The Vicia villosa agglutinin (VVA)-lectin chro-matography could selectively bind proteins with multiple Sda-groups and thereby enrich PAGs with a maximal number ofSda-groups on tetraantennary glycans. This posibility was testedby the analysis of N-glycans of the unfractionated proteins of agd 220 placenta. In this sample, the dominance of the tetraan-tennary glycan (m/z 5812) over its triantennary counterpart (m/z4757) is less accentuated than in the PAG-N-glycans, but stillvisible. This shows that the affinity chromatography probablyenriches PAGs which carry tetraanntennary glycans, but that thisstructure is nevertheless dominating.

The function of bisecting GlcNAc is still not well under-stood. One interesting finding is that an NK-cell sensitivehuman cell line (K562) lost its NK-sensitivity after transfectionwith GlcNAc-TIII, which increases the amount of bisectingGlcNAc on cell surface glycoproteins (Yoshimura et al. 1996).So bisecting GlcNAc seems to reduce target-cell susceptibilityfor NK-induced cell lysis. Since NK-cells are potentiallyhazardous for MHC-I-negative trophoblast cells in bovineplacentomes (Davies et al. 2000), a local NK-cell directedimmunosuppressive function of PAG-glycans could be of greatphysiological importance. Recently, the ultrastructurallocalization of BNC-derived PAGs in the maternal placentalconnective tissue strengthened such speculations about animmunosuppressive function (Wooding et al. 2005).

Several lectin histochemical studies dealt with the glycosyla-tion of BNC secretory granules and demonstrated that GalNAcbinding lectins can be used to label these granules with highspecificity (Lehmann et al. 1992; Jones et al. 1994; Nakanoet al. 2002; Klisch and Leiser 2003). In our present study, weshow that the Sda-antigen is the predominant carbohydrate onbovine PAGs in midpregnancy and thereby is the target of theGalNAc-binding lectins. Both, DBA and VVA, recognize theβ1,4-linked GalNAc of the Sda-antigen (Wu et al. 1998; JimenezBlanco et al. 2001), although these lectins are regarded as

48



ownloaded from



specific for α-linked GalNAc in some publications (Jones et al.1994; Nakano et al. 2002). The lectin histochemical changesof glycosylation are in accordance with earlier observations,which show that terminal GalNAc is absent in BNC-granulesbefore gd 30 (Lehmann et al. 1992) and at parturition (Klischet al. 2006). Due to steric hindrance by the ß1,4 linked GalNAc,MAL does not bind to the Sda-antigen (Jimenez Blanco et al.2001). A simple explanation for the mutually exclusive stain-ing of the BNC with either MAL or CT1 (see Figure 8) wouldbe the absence of ß1,4 linked GalNAc in otherwise unalteredglycans in early pregnancy and at parturition. For parturition,this possibility was ruled out by the analysis of the prepartalN-glycan sample (Figure 3). In this sample, only a few sialy-lated lactosamine-type glycans were observed. An explanationof the histochemical MAL-staining of BNC at term is difficult,since the mass spectra do not show upregulation of putativelyMAL-binding glycans (α2,3-sialylated lactosamine) at term.One explanation could be that bulky Sda-glycans mask theMAL-binding sites at midpregnancy, but not before gd 30 andat parturition. Another possibility would be that the highlysensitive MAL-histochemistry detects structures which arerepresented only by very small peaks in the N-glycan MALDImass spectra.

In the Western-blot, there is a much more intensive bind-ing of the anti-PAG serum to the glycosylated PAG, comparedto the deglycosylated sample. This indicates that the polyclonalserum partially recognizes the attached N-glycans. The Western-analysis also shows that in bovine BNCs, the Sda-antigen ispredominantly attached to PAGs and not or to a much lesserextent to another glycoprotein (PRP-I), which is colocalizedwith PAGs in the BNC granules. This suggests that at least oneof the glycosyltransferases, which is involved in the synthesis ofthe Sda-antigen, recognizes specific features of the protein. Thispossibility is also supported by the fact that the Sda-antigen hasonly been demonstrated in a very limited number of glycopro-teins, for example, on the N-glycans of Tamm–Horsfall protein(Van Rooijen et al. 1998; Easton et al. 2000a). The Sda-antigenwas also found on human and mouse Zona pellucida protein-3(Easton et al. 2000b) and on CD45 in activated murine cytotoxicT-cells (Lefrancois and Bevan 1985).

The functional relevance of the Sda-antigen remains largelyunknown. An involvement in the regulation of PAG serum half-life seems likely. Thereby, the changes of PAG glycosylation inthe course of pregnancy may explain changes of serum half-lifebetween the different stages of pregnancy (Klisch et al. 2006).In early pregnancy, the serum half-life of PAGs is approximately4–5 days (Szenci et al. 2003), while it is around 8–9 days afterparturition (Kiracofe et al. 1993). Since the Sda-antigen is aligand for the asialoglycoprotein receptor in mice (Mohlke etal. 1999), it might accelerate PAG-clearance from the maternalblood. The absence of Sda before gd 30 and at parturition mightthereby cause a higher serum half-life of these glycoforms andcould thus cause the small peak of PAG-serum levels in thefourth week of pregnancy and also the much more prominentpeak at parturition (Green et al. 2005).

We speculated that the absence of the Sda-antigen in the earlypregnancy and before parturition might be caused by down-regulation of the β4GalNAcT-II transcription at these stages.Therefore, we studied the relative gene expression of this en-zyme, but we found no significant differences between thegestational groups. This suggests that the generation of the

Sda-antigen is not regulated by the transcription of this en-zyme. An alternative explanation would be the regulation at anearlier stage of Sda-synthesis. This could be the downregulationof an α 2,3-sialyltransferase, which catalyzes the penultimatestep of Sda-synthesis. This would be consistent with the ab-sence of sialylated lactosamine residues in the antepartal MSspectrum (Figure 3).

Pregnancy associated changes of the abundance of theSda-antigen on red blood cells have been observed in humans(Morton et al. 1970; Spitalnik et al. 1982). The rate of Sda-negative subjects increases with the progression of pregnancy.The Sda-glycotope is absent on erythrocytes of 4% of non-pregnant individuals, while 22% and 36% pregnant women inthe first and third trimester are Sda-negative (Spitalnik et al.1982). In contrast to this, no quantitative changes of theSda-antigen were observed in the urine of pregnant women(Morton et al. 1970) and there were also no significant changesof the N-glycosylation of Tamm–Horsfall protein (uromodulin)in the course of pregnancy (Van Rooijen et al. 2001). This sug-gests that the synthesis of the Sda-glycotope is regulated ina tissue-specific way and could be under endocrine regulation.Oestradiol might be the regulatory agent for the changes of PAG-glycosylation in cattle. The BNCs express oestrogen receptor-β(Schuler et al. 2005) and there is a dramatic increase in theconcentration of oestrogens in the last week of pregnancy incattle (Robertson and King 1979). Thus, the rise of oestrogenconcentration occurs simultaneously with the disappearance ofthe Sda-antigen. In early pregnancy, the changes of oestrogenconcentrations are less dramatic, but there is a decline of theoestrogen level during the second and third week of pregnancy(Patel et al. 1999). This suggests that the absence of Sda inthe gd 20 and 23 samples might be caused by oestrogen. Itshould also be considered that the local oestrogen concentra-tions might differ from that in the maternal blood. Thus, itappears that the observed changes of PAG-glycosylation duringpregnancy are under endocrine control, but the mechanisms arestill unresolved. Our study shows that a singular carbohydratestructure is the predominant constituent of the N-glycans ofbovine PAGs. From a glycobiological view, this is very excitingsince both the tetraantennary structure in combination with abisecting GlcNAc and its homogeneity, are highly unusual. Inaddition, the elucidation of the glycan structure and its changesduring pregnancy give new starting points for functional studieswhich might lead to a better understanding of the function ofPAGs.

Materials and methods

Purification of PAGsPAGs were isolated from cotyledons of one pregnant cow(approximate gestation day (gd) 155) following a recentlypublished protocol (Klisch et al. 2005). The gd was estimatedfrom the crown-rump length of the fetus (Rexroad et al. 1974).Briefly, the cotyledons were finely minced and homogenized.After ammonium sulfate precipitation (fraction between 40 and80% saturation) the proteins were dialyzed against Tris-buffer(20 mM Tris-HCl, 2 mM EDTA, pH 7.6) and loaded onto aDEAE-cellulose column. Proteins were eluted with a continuousgradient of increasing NaCl concentrations (0–0.3 M) in Tris-buffer. The fractions were checked for PAG-immunoreactivity

49



ownloaded from


K Klisch et al.

by Western-blot with a rabbit polyclonal PAG-antiserum andimmunoreactive fractions were pooled. Aliquots of the pooledfractions were incubated with agarose bound Vicia villosalectin (Vector Laboratories, Burlingame, CA). The column waswashed with HEPES-buffer and the proteins were eluted with25 mM GalNAc in HEPES-buffer. The GalNAc was removedby washing with several volumes of 1 mM Tris-HCl, pH 7.5 onan Amicon Ultra, Centrifugal Filter Device (MWCO 10,000)(Millipore, Schwalbach, Germany) and aliquots of the purifiedprotein were stored at −20◦C or lyophilized. The characteriza-tion of purified proteins was done by SDS-PAGE and MALDI-MS as published earlier (Klisch et al. 2005) and revealed thesame results. Three bands of approximately 56, 66, and 75 kDawere seen in a Coomassie blue-stained gel. These bands werein-gel trypsinated as described (Klisch et al. 2005) and analysisby MALDI-MS identified the proteins as a mixture of PAG-1(66 kDa), PAG-6 and PAG-7 (75 kDa), and PAG-17 (56 kDa).

MALDI-MS of native PAG-glycansAn aliquot of the purified PAGs (100 µg) was heated (96◦C)for 10 min and then the N-glycans were released by di-gestion with PNGase-F (New England Biolabs, Frankfurtam Main, Germany) in phosphate buffer (pH 7.5) for 24 hat 37◦C. The glycans were extracted by graphitised carbon(Alltech Grom, Rottenburg, Germany) after a protocol fromPacker et al. (1998), eluted in 75% acetonitrile with 0.15%trifluoroacetic acid and dried in a vacuum centrifuge. Analiquot of the glycans was desialylated with neuraminidasefrom C. perfringens (Roche, Penzberg, Germany) in 50 mMsodium-acetate buffer, pH 5.0. Samples were analyzed ina MALDI-TOF/TOF mass spectrometer (Ultraflex BrukerDaltonics, Bremen, Germany) at 20 kV in the linear mode. Thenondesialylated samples were measured in the negative modewith 20 mg/mL 2′,4′,6′-trihydroxyacetophenone monohydrate,20 mM ammonium citrate in 50% acetonitrile as matrix. Thedesialylated samples were measured in the positive mode with10 mg/mL 2,5-dihydroxybenzoic acid (DHB) as matrix.

MALDI-MS and MS/MS analysis of permethylated PAGN-glycansA sample of approximately 50 µg of the purified PAG wasreduced for 1 h at 37◦C in 50 mM Tris-HCl buffer (pH 8.5)containing a fourfold excess of dithiothreitol and carboxymethy-lated with a twofold molar excess of iodoacetic acid for 1 h atroom temperature in the dark. Following dialysis at 4◦C for72 h against 4 × 4.5 litres of cold 50 mM ammonium bicar-bonate, pH 7.5, and lyophilization, the sample was digestedwith sequencing-grade trypsin (Promega, Mannheim, Germany)(1 µg in 50 mM ammonium bicarbonate, pH 8.5, for 18 h at37◦C). The reaction was stopped by adding a few drops ofacetic acid to the solution. The sample was lyophilized priorto its dissolution in 150 µL (5% (v/v)) acetic acid and puri-fied using a Sep-Pak cartridge C18 (Waters Corp, Eschborn,Germany), as previously described (Jang-Lee et al. 2006). Thepurified glycopeptides were digested with PNGase-F (RocheApplied Science) in 50 mM ammonium bicarbonate (pH 8.5)containing 10 units of enzyme at 37◦C over 18 h. The samplewas lyophilized, and the released N-glycans were purified us-ing a Sep-Pak cartridge C18 (Waters Corp). Permethylation andsample cleanup were performed using the sodium hydroxide

protocol, as described previously (Jang-Lee et al. 2006). Prepa-ration of partially methylated alditol acetates was performed asdescribed (Jang-Lee et al. 2006).

MALDI-TOF MS data on permethylated samples were ac-quired in positive ion mode (M+Na)+ using a Perseptive Biosys-tems Voyager DE-STRTM mass spectrometer in the reflectormode with delayed extraction. MS/MS data were acquired usinga 4800 MALDI-TOF/TOF (Applied Biosystems, Foster City,CA) mass spectrometer. The collision energy was set to 1 kV,and argon was used as collision gas. Samples were dissolved in10 µL of methanol and 1 µL was mixed at a 1:1 ratio (v/v) withDHB as matrix.

Glycomics analysis of placental tissuesPlacental tissues (1 g each) from two pregnant cows (gd 220and one day before birth) were homogenized in six volumesof homogenization buffer (10 mM HEPES, 150 mM NaCl,pH 7.5) and subsequently lyophilized. For N-glycan analysis,approximately 80 mg of the homogenized placental tissues weresubjected to the same procedures as for the permethylated PAGN-glycans. For O-glycan analysis, the same amount of homog-enized placental tissues was subjected to reductive eliminationby adding 400 mL of 1 M potassium borohydride (54 mg/mL in0.1 M potassium hydroxide) for 24 h incubation at 45◦C. Thereaction was terminated by a dropwise addition of glacial aceticacid followed by Dowex chromatography and borate removalusing 10% of methanolic acetic acid (Jang-Lee et al. 2006). Thepurified O-glycans were then permethylated and the mass spec-trometric analyses were carried out as described in the earliersection for permethylated PAG N-glycans.

Western analysisProteins from homogenized placentomal tissue of a late preg-nant cow (approximately gd 260) were either deglycosylatedwith PNGase F or left nondeglycosylated. The samples (10µg/lane) were separated by SDS-PAGE and transferred to aPVDF membrane as described earlier (Klisch et al. 2005).The membrane was probed with CT1-antibody (1:500, con-centrated cell-culture medium of the CT1-hybridoma, gift fromL Lefrancois, University of Connecticut, Farmington, CT), anti-PAG (1:10,000; rabbit antiserum R727, gift from JF Beckers,University Liege, Liege, Belgium), and PRP-I (1:20,000; giftfrom L Schuler, University of Wisconsin-Madison, Madison,WI). Secondary antibodies were peroxidase-labeled anti-IgMmouse (Sigma, Deisenhofen, Germany) and anti rabbit (GE-Healthcare, Freiburg, Germany). Blots were developed with en-hanced chemiluminescence (ECL Plus Western Blotting Detec-tion Reagents, GE-Healthcare) after washing in PBS.

Immuno- and lectin histochemistryUteri of early pregnant cows (gd 20, 23, 32, 37; each n = 1)were fixed by a perfusion with 4% paraformaldehyde in 0.15 Mphosphate buffer (pH 7.4). Pieces of the chorionic sac wereremoved from the uteri and further fixed by immersion in 3%glutaraldehyde and embedded in epon as described (Klisch andLeiser 2003). Bovine placentomal tissues of mid pregnancies(gd 94, 110, 154, 198) were collected at a slaughterhouse.Tissues of preterm (caesarean sections either 27 h aftera Prostaglandin F2-alpha analogon injection approximately5 days (approximately gd 275) before the expected end of

50



ownloaded from



pregnancy (n = 3) or after the prepartal decline of maternal pro-gesterone became obvious (approximately 1 day ante partum;n = 2) and term placentomes (approximately gd 280; n = 3)were obtained by caesarean sections as described earlier (Klischet al. 2006; Schuler et al. 2006). The tissues were fixed in 4%formaldehyde (v/v) in 0.1 M phosphate buffer (pH 7.3) for 24 hand embedded in paraffin.

Paraffin sections (7 µm) were dewaxed in xylol, rinsed inthree changes of ethanol, rehydrated in descending concentra-tions of ethanol, and rinsed in distilled water. Epon sections(0.5 µm) were deplasticized and rehydrated as described earlier(Klisch and Leiser 2003). The slides were rinsed in 0.05 MTRIS-buffered saline, pH 7.6, 1 mM CaCl2 (TBS), and incu-bated for 45 min in a humid chamber at 37◦C with 10 µg/mLbiotinylated lectin (DBA, Sigma; PHA-L, EY-Laboratories, SanMateo, CA; MAL-I, Vector Laboratories) in TBS. For doublestainings, the sections were incubated with the carbohydrate-binding reagent (biotinylated lectins or CT1-antibody) incombination with anti-PAG. Lectins were visualized withstreptavidin-Cy3, CT1 by Cy3 anti-mouse, and the polyclonalanti-PAG by Cy2-anti-rabbit. In controls, the lectin or pri-mary antibody were replaced by buffer. In additional con-trols, 0.2 M GalNAc was added to the buffer during incuba-tion with DBA. As a control for the CT1-antibody, an irrel-evant mouse IgM antibody (monoclonal antibody to single-stranded DNA [F7–26]; Alexis Biochemicals, Gruenberg,Germany) was used instead of the CT1-antibody. The BNCswere identified by the PAG-immunostaining and in the dou-blestainings, the fractions of MAL-, DBA-, PHA-L- and CT1-positive BNC were evaluated. For each lectin, and also forthe CT1 antibody, approximately 30 fields of vision (each0.173 mm2) were evaluated for each animal. Due to the smallsize of the sections of the earliest stages (gd 20; gd 23), all BNCswere evaluated in these specimens.

Quantitative PCR of β4GalNAcT-IIBased on a predicted sequence of the bovine β-4-N-acetylgalactosaminyltransferase (β4GalNAcT-II) mRNA(GenBank Accession Number XM_584835), which has 75%and 74% amino acid homology to the human and mouseβ4GalNAcT-II-mRNA respectively, we confirmed the exis-tence of the corresponding transcript in bovine placentomesby PCR. We amplified a 847 bp fragment (forward primer5′-AGG GTG GAT GTG GTG AGT CT-3′; reverse primer:5′-CAC ATT GGA GGT GGT TCT TG-3′) on bovine pla-centomal cDNA. This fragment was sequenced and submittedto genbank (GenBank Accession Number EF445547). Thissequence was used for the quantification of the mRNA withthe TaqMan system (forward primer 5′-GTG GCT GAT GACAGC AAG GA-3′: reverse primer 5′-GCC GTA GGG CATGGT GTA AT-3′; TaqMan Probe: 5′-CCC CTG AAA ATTAAT GAC AGC CAT GTG G-3′) as described (Kowalewskiet al. 2006). Relative mRNA levels for bovine β4GalNAcT−IIand for the housekeeping gene glycerinaldehyd-3-phosphat-dehydrogenase (GAPDH) were determined in one placentomeof each of the 14 cows assigned to four observational groups,representing the midgestation (day 100–200; n = 3), late ges-tation (day 200–280; n = 5), the prepartal decline in maternalprogesterone concentrations (approximately one day before thenormal onset of parturition, n = 3) and parturition (n = 3)(Schuler et al. 2006).

Primer Express software (version 2.0, Applied Biosystems)was used to design primers and TaqMan Probe (forward primer5′-GTG GCT GAT GAC AGC AAG GA-3′: reverse primer 5′-GCC GTA GGG CAT GGT GTA AT-3′; TaqMan Probe: 5′-CCCCTG AAA ATT AAT GAC AGC CAT GTG G-3′). The primerswere ordered from MWG Biotech AG, Ebersberg, Germany,the TaqMan probe was from Eurogentec, Seraing, Belgium.TaqMan probe was labeled at the 5′- end with reporter dye 6-carboxyfluorescein (FAM) and at the 3′- end with the quencherdye 6-carboxytetramethyl-rhodamine (TAMRA).

ABI PRISMR©

7000 Sequence Detection System (AppliedBiosystems, Darmstadt, Germany) was used and experimentswere performed according to our previously described protocol(Kowalewski et al. 2006). Briefly: 200 ng of total RNA wasDNase-treated and reverse-transcribed as in the routine RT-PCR.Samples were analyzed in duplicates; 25 µL of reaction mixturecontained 12.5 µL TaqMan

R©qPCR MasterMix (Eurogentec),

300 nM of each primer and 200 nM TaqMan Probe, and 5 µL ofcDNA. Amplification was carried out as follows: denaturationfor 10 min at 95◦C followed by 40 cycles at 95◦C for 15 s and60◦C for 60 s.

Relative quantification was done by normalizing theβ4GalNAcT-II signals with the GAPDH signal (as “housekeep-ing gene”) using the comparative CT method (��CT method)according to the instructions of the manufacturer of the ABIPRISMTM 7000 Sequence Detector. The threshold cycle (CT)represents the PCR cycle at which an increase in reporterfluorescence above a base line signal can first be detected.

Finally, relative gene expression (RGE) was calculated asexpression of the target gene relative to the reference gene(GAPDH) and normalized to the calibrator (sample with thelowest amounts of the respective target gene transcripts). Briefly,the analysis was performed as follows: For each mRNA, a dif-ference in CT values (�CT) was calculated by taking the meanCT of duplicate tubes and subtracting the mean CT of the du-plicate tubes for the reference RNA (GAPDH) measured in analiquot from the same RT reaction (�CT = CT test gene − CTGAPDH; treated sample). The �CT for the treated sample wasthen subtracted from the �CT for the calibrator to generate a��CT (��CT = �CT treated sample − �CT calibrator).The mean of these ��CT measurements was then used tocalculate expression of the test gene (2-��CT) relative tothe reference gene (GAPDH) and normalized to the calibrator(relative gene expression (RGE) = 2−��CT).

Funding

Deutsche Forschungsgemeinschaft (DFG) (KL 1835/1-1 toK.K.); Biotechnology and Biological Sciences ResearchCouncil; Wellcome Trust (to A.D.).

Acknowledgements

The authors would like to thank Ms Silke Fischer for excellentlaboratory assistance. The PAG-antiserum was a gift fromJ.-F. Beckers, University of Liege, Liege, Belgium and thePRP-I antiserum was obtained from L. Schuler, University ofWisconsin-Madison, Madison, USA. The CT1-hybridoma was agift from L. Lefrancois, University of Connecticut, Farmington,CT, USA. P.-C.P. is a recipient of Imperial College London stu-dentships and A.D. is a Biotechnology and Biological SciencesResearch Council (BBSRC) Professorial Fellow.

51



ownloaded from


K Klisch et al.

Conflict of interest statement

None declared.

Abbreviations

BNCs, binucleate trophoblast giant cells; d a.p., days ante par-tum; DBA, Dolichos biflorus agglutinin; DHB, 2,5-dihydroxy-benzoic acid; GalNAc-T, N-acetylgalactosaminyltransferase;GAPDH, glycerinaldehyd-3-phosphat-dehydrogenase; gd, ges-tation day; GlcNAc, N-acetylglucosamine; GlcNAc-T, N-acetylglucosaminyltransferase; MAL, Maackia amurensislectin; MALDI-MS, matrix assisted laser desorption ionisation-mass spectrometry; Man, mannose; MW, molecular weight;NeuAc, N-acetylneuraminic acid; PAG, pregnancy-associatedglycoprotein; PHA-L, Phaseolus vulgaris leucoagglutinin;PRP-I, prolactin related protein-I; TOF, time of flight; Sda,NeuAcα2–3[GalNAcβ1–4]Galβ1–4GlcNAc; VVA, Vicia vil-losa agglutinin.

References

Davies CJ, Fisher PJ, Schlafer DH. 2000. Temporal and regional regulationof major histocompatibility complex class I expression at the bovine uter-ine/placental interface. Placenta. 21:194–202.

Easton RL, Patankar MS, Clark GF, Morris HR, Dell A. 2000a. Pregnancy-associated changes in the glycosylation of tamm-horsfall glycoprotein. Ex-pression of sialyl Lewis(x) sequences on core 2 type O-glycans derived fromuromodulin. J Biol Chem. 275:21928–21938.

Easton RL, Patankar MS, Lattanzio FA, Leaven TH, Morris HR, Clark GF, DellA. 2000b. Structural analysis of murine zona pellucida glycans. Evidencefor the expression of core 2-type O-glycans and the Sd(a) antigen. J BiolChem. 275:7731–7742.

Fukuta K, Abe R, Yokomatsu T, Omae F, Asanagi M, Makino T. 2000. Con-trol of bisecting GlcNAc addition to N-linked sugar chains. J Biol Chem.275:23456–23461.

Green JA, Parks TE, Avalle MP, Telugu BP, McLain AL, Peterson AJ, McMillanW, Mathialagan N, Hook RR, Xie S, Roberts RM. 2005. The establishmentof an ELISA for the detection of pregnancy-associated glycoproteins (PAGs)in the serum of pregnant cows and heifers. Theriogenology. 63:1481–1503.

Jang-Lee J, North SJ, Sutton-Smith M, Goldberg D, Panico M, Morris H,Haslam S, Dell A. 2006. Glycomic profiling of cells and tissues by massspectrometry: fingerprinting and sequencing methodologies. Methods En-zymol. 415:59–86.

Jimenez Blanco JL, Haseley SR, Kamerling JP, Vliegenthart JF. 2001. Analysisof the interaction between lectins and tetra- and tri-saccharide mimics ofthe Sd(a) determinant by surface plasmon resonance detection. Biochimie.83:653–658.

Jones CJ, Koob B, Stoddart RW, Hoffmann B, Leiser R. 1994. Lectin-histochemical analysis of glycans in ovine and bovine near-term placentalbinucleate cells. Cell Tissue Res. 278:601–610.

Kiracofe GH, Wright JM, Schalles RR, Ruder CA, Parish S, Sasser RG. 1993.Pregnancy-specific protein B in serum of postpartum beef cows. J Anim Sci.71:2199–2205.

Klisch K, Boos A, Friedrich M, Herzog K, Feldmann M, Sousa N, Beckers J,Leiser R, Schuler G. 2006. The glycosylation of pregnancy-associated gly-coproteins and prolactin-related protein-I in bovine binucleate trophoblastgiant cells changes before parturition. Reproduction. 132:791–798.

Klisch K, De Sousa NM, Beckers JF, Leiser R, Pich A. 2005. Pregnancy associ-ated glycoprotein-1, -6, -7, and -17 are major products of bovine binucleatetrophoblast giant cells at midpregnancy. Mol Reprod Dev. 71:453–460.

Klisch K, Leiser R. 2003. In bovine binucleate trophoblast giant cells,pregnancy-associated glycoproteins and placental prolactin-related protein-Iare conjugated to asparagine-linked N-acetylgalactosaminyl glycans. His-tochem Cell Biol. 119:211–217.

Kowalewski MP, Schuler G, Taubert A, Engel E, Hoffmann B. 2006. Expressionof cyclooxygenase 1 and 2 in the canine corpus luteum during diestrus.Theriogenology. 66:1423–1430.

Lefrancois L, Bevan MJ. 1985. Functional modifications of cytotoxic T-lymphocyte T200 glycoprotein recognized by monoclonal antibodies. Na-ture. 314:449–452.

Lehmann M, Russe I, Sinowatz F. 1992. [Detection of lectin binding sites inthe trophoblast of cattle during early pregnancy]. Anat Histol Embryol.21:263–270.

Mohlke KL, Purkayastha AA, Westrick RJ, Smith PL, Petryniak B, Lowe JB,Ginsburg D. 1999. Mvwf, a dominant modifier of murine von Willebrandfactor, results from altered lineage-specific expression of a glycosyltrans-ferase. Cell. 96:111–120.

Morton JA, Pickles MM, Terry AM. 1970. The Sda blood group antigen intissues and body fluids. Vox Sang. 19:472–482.

Munson L, Kao JJ, Schlafer DH. 1989. Characterization of glycoconjugatesin the bovine endometrium and chorion by lectin histochemistry. J ReprodFertil. 87:509–517.

Nakano H, Shimada A, Imai K, Takahashi T, Hashizume K. 2002. Associa-tion of Dolichos biflorus lectin binding with full differentiation of bovinetrophoblast cells. Reproduction. 124:581–592.

Packer NH, Lawson MA, Jardine DR, Redmond JW. 1998. A general ap-proach to desalting oligosaccharides released from glycoproteins. GlycoconjJ. 15:737–747.

Patel OV, Takenouchi N, Takahashi T, Hirako M, Sasaki N, Domeki I. 1999.Plasma oestrone and oestradiol concentrations throughout gestation in cattle:relationship to stage of gestation and fetal number. Res Vet Sci. 66:129–133.

Rexroad CE, Casida LE, Tyler WJ. 1974. Crown-rump length of fetuses inpurebred Holstein -Frisian cows. Journal of Dairy Science. 57:346–347.

Robertson HA, King GJ. 1979. Conjugated and unconjugated oestrogens infetal and maternal fluids of the cow throughout pregnancy. J Reprod Fertil.55:463–470.

Sasai K, Ikeda Y, Ihara H, Honke K, Taniguchi N. 2003. Caveolin-1 regulatesthe functional localization of N-acetylglucosaminyltransferase III within thegolgi apparatus. J Biol Chem. 278:25295–25301.

Schachter H. 1986. Biosynthetic controls that determine the branching andmicroheterogeneity of protein-bound oligosaccharides. Biochem Cell Biol.64:163–181.

Schuler G, Teichmann U, Kowalewski MP, Hoffmann B, Madore E, Fortier MA,Klisch K. 2006. Expression of cyclooxygenase-II (COX-II) and 20alpha-hydroxysteroid dehydrogenase (20alpha-HSD)/prostaglandin F-synthase(PGFS) in bovine placentomes: implications for the initiation of parturi-tion in cattle. Placenta. 27:1022–1029.

Schuler G, Teichmann U, Taubert A, Failing K, Hoffmann B. 2005. Estrogenreceptor beta (ERbeta) is expressed differently from ERalpha in bovineplacentomes. Exp Clin Endocrinol Diabetes. 113:107–114.

Spitalnik S, Cox MT, Spennacchio J, Guenther R, Blumberg N. 1982. Theserology of Sda effects of transfusion and pregnancy. Vox Sang. 42:308–312.

Szenci O, Beckers JF, Sulon J, Bevers MM, Borzsonyi L, Fodor L, KovacsF, Taverne MA. 2003. Effect of induction of late embryonic mortality onplasma profiles of pregnancy associated glycoprotein 1 in heifers. Vet J.165:307–313.

Van Rooijen JJ, Hermentin P, Kamerling JP, Vliegenthart JF. 2001. The patternsof the complex- and oligomannose-type glycans of uromodulin (Tamm-Horsfall glycoprotein) in the course of pregnancy. Glycoconj J. 18:539–546.

van Rooijen JJ, Kamerling JP, Vliegenthart JF. 1998. The abundance of ad-ditional N-acetyllactosamine units in N-linked tetraantennary oligosaccha-rides of human Tamm-Horsfall glycoprotein is a donor-specific feature.Glycobiology. 8:1065–1075.

Wooding FB. 1992. Current topic: the synepitheliochorial placenta of ruminants:binucleate cell fusions and hormone production. Placenta. 13:101–113.

Wooding FB, Roberts RM, Green JA. 2005. Light and electron microscopeimmunocytochemical studies of the distribution of pregnancy associatedglycoproteins (PAGs) throughout pregnancy in the cow: possible functionalimplications. Placenta. 26:807–827.

Wu AM, Wu JH, Watkins WM, Chen CP, Song SC, Chen YY. 1998. Differentialbinding of human blood group Sd(a+) and Sd(a-) Tamm-Horsfall glycopro-teins with Dolichos biflorus and Vicia villosa-B4 agglutinins. FEBS Lett.429:323–326.

Xie S, Green J, Bixby JB, Szafranska B, DeMartini JC, Hecht S, RobertsRM. 1997. The diversity and evolutionary relationships of the pregnancy-associated glycoproteins, an aspartic proteinase subfamily consisting ofmany trophoblast-expressed genes. Proc Natl Acad Sci U S A. 94:12809–12816.

Yoshimura M, Ihara Y, Ohnishi A, Ijuhin N, Nishiura T, Kanakura Y, MatsuzawaY, Taniguchi N. 1996. Bisecting N-acetylglucosamine on K562 cells sup-presses natural killer cytotoxicity and promotes spleen colonization. CancerRes. 56:412–418.

52



ownloaded from


a tetraantennary glycan with bisecting n-acetylglucosamine ... · glycans of bovine...

Documents