HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 10, 7736-7751, March 8, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/10/7736    most recent M111690200v1 Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gerken, T. A. Articles by Zhang, J. Search for Related Content PubMed PubMed Citation Articles by Gerken, T. A. Articles by Zhang, J. Social Bookmarking                      What's this?

Determination of the Site-specific Oligosaccharide Distribution of the O-Glycans Attached to the Porcine Submaxillary Mucin Tandem Repeat

FURTHER EVIDENCE FOR THE MODULATION OF O-GLYCAN SIDE CHAIN STRUCTURES BY PEPTIDE SEQUENCE^*

Thomas A. Gerken, Marc Gilmore, and Jiexin Zhang

From the Departments of Pediatrics and Biochemistry, W. A. Bernbaum Center for Cystic Fibrosis Research, Case Western Reserve University School of Medicine, Cleveland, Ohio 44106

Received for publication, December 7, 2001, and in revised form, January 2, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Little is known of the degree that polypeptide sequence and the local environment modulate the structures of O-linked glycans. Toward this understanding, the site-specific mono- (GalNAc-O-), di- (-Gal-1,3--GalNAc-O-), and trisaccharide (-Fuc-1,2--Gal-1,3--GalNAc-O-) distributions have been determined for 29 of the 31 O-glycosylated Ser/Thr residues in the tandem repeat domains of blood group A-negative porcine submaxillary gland mucin. The glycosylation patterns obtained from three individual animals are in agreement with earlier incomplete determinations on a pooled mucin (Gerken, T. A., Owens, C. L., and Pasumarthy, M. (1997) J. Biol. Chem. 272, 9709-9719; Gerken, T. A., Owens, C. L., and Pasumarthy, M. (1998) J. Biol. Chem. 273, 26580-26588), confirming that the addition of the peptide-linked GalNAc and its substitution by -1,3-Gal are sensitive to local peptide sequence in a highly reproducible manner in vivo. The present data further support earlier suggestions of an inverse correlation of the density of hydroxyamino acid residues (and by inference the density of peptide GalNAc) with the extent of substitution of the peptide-linked GalNAc by -1,3-Gal. This effect is highly correlated for Ser-linked glycans but not for Thr-linked glycans. A similar correlation is observed with respect to the in vivo peptide GalNAc glycosylation pattern. In contrast, the addition of -1,2-Fuc to -Gal shows no apparent correlation with hydroxyamino acid density, although a marked elevation in the fucosylation of Ser-linked glycans compared with Thr-linked glycans is observed. The above effects may represent both steric and conformational factors acting to alter the relative accessibility and activity of the glycosyltransferases toward substrate. These results demonstrate that the porcine submaxillary gland core 1 3-galactosyltransferase and 2-fucosyltransferase exhibit unique peptide/glycopeptide sensitivities that may provide mechanisms for the modulation of O-linked side chain structures.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The O-glycosylation of serine and threonine residues with glycans linked through GalNAc is a common post-translational modification of a wide range of secreted and membrane-associated proteins. One class of highly O-glycosylated proteins are the mucus glycoproteins commonly called mucins (1). These glycoproteins are typically 20-30% Ser and Thr, are between 50 and 80% carbohydrate by weight, and commonly contain tandemly repeated peptide sequences. Mucins and glycoproteins containing mucin-like domains play a major role protecting epithelial cell surfaces and are thought to be involved in modulating many biological processes including the immune response, adhesion, inflammation, tumorigenesis, and perhaps development (2-9).

The first step in O-glycan synthesis is the transfer of GalNAc to Ser/Thr residues, by UDP-GalNAc:polypeptide -GalNAc transferase (ppGalNAc transferase).¹ To date, 10 ppGalNAc transferase isoforms have been described (9-16). Although each transferase has not been fully characterized, their peptide substrate specificities vary within the family (11, 17-21); many show sensitivity to prior glycosylation, and others require prior addition of GalNAc for activity (9, 14, 20, 22-24). The expression of ppGalNAc transferase isoforms with different peptide and/or glycopeptide specificities therefore represents the first step in the regulation of O-glycan structure by peptide sequence in vivo. Subsequent elongation of O-linked glycans proceeds by the stepwise addition of single sugar residues via a series of substrate-specific Golgi resident transferases (25). It is well accepted that Golgi localization, nucleotide sugar concentration, and competition among transferases are the major dictates of O-glycan structure and elongation (25-31). Depending on the initial and subsequent substitutions on the GalNAc residue, a wide range of O-linked core structures is possible (25). Little, however, is known of the effects of local peptide sequence on O-glycan elongation (23, 32-37), although a number of examples of site-specific elongation of N-linked glycans are known (30, 38-46). Factors affecting site-specific N-glycan structures include specific peptide sequence motifs, protein surface structures, and the folded state of the protein. The paucity of our understanding of the influence of the acceptor peptide structure on O-glycan elongation in mucin-like domains arises from the difficulty in determining the site-specific glycosylation pattern of mucins and other heavily O-glycosylated glycoproteins.

With the goal of understanding the influence of peptide sequence and structure on mucin O-glycosylation, our laboratory has previously reported the site-specific GalNAc-O-Ser/Thr and core 1 ([R1-,R2-]-Gal-1,3--GalNAc-O-Ser/Thr) glycosylation patterns of 29 of the 31 glycosylated residues in the 81-residue tandem repeat of the porcine submaxillary mucin (PSM) pooled from several different animals (34, 37) (see Fig. 1 for the glycan structures and tandem repeat sequence of PSM (47-52)). We found that the extent of GalNAc core glycosylation was site-specific and that the degree of core 1 glycosylation appeared to inversely correlate to the density of Ser and Thr residues in the polypeptide sequence. From the latter finding, we suggested that the oligosaccharide side chain length may also correlate with the density of glycosylated Ser and Thr residues. These results suggested the reasonable possibility of steric interactions of neighboring glycosylated Ser and Thr residues affecting glycan elongation.

In this article we have extended these studies to the characterization of the site-specific glycosylation pattern of the PSM tandem repeat isolated from blood group A-negative animals, obtaining the mono- (GalNAc-O-Ser/Thr), di- (-Gal-1,3--GalNAc-O-Ser/Thr), and trisaccharide (-Fuc-1,2--Gal-1,3--GalNAc-O-Ser/Thr) distributions at each individual glycosylation site. The analysis of the glycan distributions from PSM obtained from three different animals with similar oligosaccharide compositions gave highly reproducible site-specific glycosylation patterns with GalNAc and core 1 glycosylation patterns nearly identical to those described previously for pooled mucin (34, 37). The analysis of the complete glycosylation pattern of these mucins provides additional support for the notion that oligosaccharide side chain structures and side chain lengths may be modulated to a significant extent by the density of neighboring (presumably partially glycosylated) hydroxyamino acid residues in the polypeptide sequence and that this effect is primarily observed on Ser-linked glycans at the initial step of -Gal addition by the core 1 3-galactosyltransferase, forming the core 1 structures. In contrast, our results indicate that the porcine 2-fucosyltransferase in these animals prefers to add Fuc to Ser-linked glycans compared with Thr-linked glycans by a factor of ~1.6, while exhibiting no correlation to hydroxyamino acid density for either linkage residue. These results unambiguously show that the elongation of O-linked mucin-type glycans is significantly and reproducibly affected, whether directly or indirectly, by local peptide sequence. This work represents the most complete characterization to date of the in vivo O-linked glycosylation pattern of any highly O-glycosylated glycoprotein.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Mucin Isolation and Characterization-- Paired frozen porcine submaxillary glands from individual animals were obtained from Pel-Freez (Rogers, AR) and screened for the presence of blood group A determinants by the ability of gland extracts (47) to inhibit the agglutination of human A-positive erythrocytes with Phaseolus limensis lectin (Sigma). Mucin used in this work was purified from paired blood group A-negative glands. Oligomeric PSM tandem repeat glycosylated domains, called TR-PSM, were obtained after trypsinization and gel filtration chromatography of the reduced and carboxymethylated mucin as described previously (37). Tandem repeat domains were further purified from contaminating globular proteins by passage through SP-Sephadex (Amersham Biosciences) equilibrated with 50 mM formic acid, pH 4, followed by passage through octyl-Sepharose (Amersham Biosciences) equilibrated with 1 M (NH₄)₂SO₄, 100 mM NaH₂PO₄, pH 5. The oligosaccharide composition of the purified mucin was obtained by integration of the anomeric carbon resonances of the carbon-13 NMR spectrum (50), see below.

Mucin Glycan Modifications-- Mild, 0 °C, trifluoromethanesulfonic acid (TFMSA)/anisole treatments (37, 53) were used to trim O-linked glycans from intact or modified TR-PSM domains, leaving intact peptide-linked GalNAc residues. This treatment on intact TR-PSM gives TTR-PSM-containing GalNAc residues at all positions originally containing glycans, irrespective of glycan length or structure. We shall refer to glycopeptides derived by this approach as "Native GalNAc"-PSM tandem repeats.

Selective removal of C-3 unsubstituted GalNAc residues on TR-PSM by periodate oxidation (100 mM IO, 166 mM NaCl, 100 mM NaHOAc, pH 4.5, 0 °C) followed by NaBH₃ reduction (~0.67 M, in 0.5 M Na₂HPO₄, pH 8.0, 0 °C) and subsequent mild 0 °C TFMSA/anisole treatment was performed as described previously (34). This procedure gives TROTR-PSM, which contains GalNAc residues only at those sites originally containing core 1 glycans (i.e. di- and trisaccharides in A-negative blood type mucin). These derivatives will be referred to as "Core 1" PSM tandem repeats. Several changes were made to optimize further these procedures as follows: 1) by using a 10-20-fold molar excess periodate over oxidizable carbohydrate and an oxidation time of 5-6 h; 2) by destroying periodate by NaI/NaHCO₃/NaS₂O₃ (200, 200, and 800 mM) (34); 3) by using a maximum NaBH₃ reduction time of 15 min with a pH maximum of ~8.5; and 4) by treating with 0 °C TFMSA/anisole for no more than 4 h. Under these conditions the degradation of the peptide core was kept to a minimum while completely removing oxidized GalNAc. These preparations were further purified by gel filtration chromatography on Sephacryl S200 as described previously (34, 37).

The trisaccharide site-specific glycosylation pattern was determined by the enzymatic conversion of the disaccharide side chains to monosaccharide side chains followed by periodate oxidation and mild TFMSA. Unsubstituted -1,3-Gal residues of the disaccharide were selectively removed from 20 to 30 mg of TR-PSM in 2-4 ml of buffer by placing in dialysis tubing with 5-8 units of bovine testes -galactosidase (Sigma, or Glyco Inc., Novato CA) and dialyzing against 250 ml of 100 mM sodium citrate, pH 4.5, overnight at 33 °C. A prior treatment with 1 unit of Clostridium perfringens neuraminidase (Sigma) dialyzed against 250 ml of 100 mM sodium acetate, pH 5.0, for 4-8 h at 33 °C was performed to remove sialic acid which might interfere with the removal of -Gal. Glycosidase treatments were repeated until complete removal of -Gal and sialic acid was demonstrated by the complete loss of the -Gal and sialic acid anomeric carbon-13 NMR resonances at 105 and 101 ppm, respectively (50). These preparations after oxidation/reduction and TFMSA treatments (as described above) yield mucin containing GalNAc residues only at sites that originally contained fucosylated trisaccharide. These TROGNTR-PSM preparations (designated "Trisaccharide" PSM tandem repeat domains) were further purified by Sephacryl S200 chromatography (data not shown) giving a major high molecular weight excluded volume peak identical to those reported previously (34, 37).

To confirm the specificity of the chemical periodate oxidation/reduction approach for removing unsubstituted GalNAc residues, an enzymatic approach was also performed using the -N-acetylgalactosaminidase from chicken liver (Sigma). Briefly 5-10 mg of -galactosidase-neuraminidase-treated mucin was incubated with 5-10 units of -N-acetylgalactosaminidase and dialyzed against ~250 ml of 100 mM sodium citrate/phosphate buffer, pH 3.7, overnight at 42 °C. Glycosidase treatments were repeated until the complete removal of unsubstituted GalNAc was confirmed by carbon-13 NMR, by monitoring the loss of the unique monosaccharide GalNAc C-5, C-4, and C-3 resonances at 72.57, 69.92, and 69.03 ppm (50).

Fully deglycosylated apo-PSM was obtained from Native GalNAc PSM oligomeric tandem repeats (containing only peptide-linked GalNAc residues) as described earlier (53) by oxidation and alkaline elimination.

PSM Tandem Repeat Isolation-- The 81-residue tandem repeat glycopeptides were obtained from the above Native GAlNAc, Core 1, and Trisaccharide PSM preparations after trypsinolysis with 1% modified sequence grade trypsin (TRSEQZ, Worthington) in 50 mM (NH₄)₂CO₃, pH 8.5, for 7 h at 37 °C (37). Trypsin was inhibited by the addition of 1 mM phenylmethylsulfonyl fluoride. Tandem repeat glycopeptides were pooled after chromatography on Sephacryl S200. The C-terminal portion of the tandem repeat glycopeptide (residues 39-79) was obtained after N-terminal biotinylation and digested with protease GluC as described (34). The large C-terminal glycopeptide beginning at residue 39 of the tandem repeat was isolated from the N-terminal glycopeptide after passage through 1 ml of immobilized avidin column (Pierce) and subsequent Sephacryl S200 gel filtration.

Amino Acid Sequencing-- Pulsed liquid phase Edman degradation amino acid sequencing was performed on an Applied Biosystems Process 494 protein sequencer (PerkinElmer Life Sciences) using manufacturer recommended pulse liquid cycles as described previously (34, 37). Amino acid phenylthiohydantoin (PTH) derivatives were chromatographed on an ABI 5-µm C18 PTH column using the fast normal I gradient program. The PTH-Ser/Thr-O-GalNAc elute as two diastereotopic peaks in the chromatogram at relatively unique positions (37). Picomoles of PTH derivatives were obtained after eliminating long range cycle preview and lag for each PTH derivative by a simple base-line subtraction approach as described (37). Corrections for the overlap of the second eluting PTH-Thr-O-GalNAc diastereomer with PTH-Thr were made using the previously obtained area ratios for the two PTH-Thr-O-GalNAc diastereomers obtained from fully glycosylated glycopeptides (34, 37). The extent of glycosylation was determined by comparing the relative picomoles of nonglycosylated and glycosylated PTH derivatives after taking into account the relative recovery of the individual PTH species. The recovery values used in this work were adjusted to give constant summed Ser and Thr amino acid compositions (obtained by summing the non-base-line corrected picomole content over all cycles of the sequence determination, typically greater than 40 cycles) regardless of the extent of GalNAc glycosylation of the tandem repeat. With these values, all preparations, from apo- to highly glycosylated Native-GalNAc PSM, gave identical amino acid compositions regardless of their extent of glycosylation, with standard deviations of 1 mol % or less for all residues including Ser and Thr (utilizing all 17 full-length tandem repeat determinations performed in this study as well as 3 apomucin tandem repeat determinations). Using the N-terminal sequence of the tandem repeat, the summed Edman derived amino acid composition was found to be consistent with a 97.5% repetitive yield. The same set of response factors were utilized for the processing of all Edman sequencing data described in this work. Typically 2-8 nmol of glycopeptide were sequenced to ensure sequencing to at least 40 cycles. The reproducibility between sequencing runs is relatively good, with standard deviations typically in the range of less than 5 percentage values.

NMR Analysis-- Carbon-13 NMR spectra were obtained using Varian Inova 300 or Inova 600 spectrometers using D₂O as solvent. Integrals of the oligosaccharide anomeric carbons unique to each oligosaccharide side chain were used to obtain the oligosaccharide side chain distribution (50). The absence of the tetrasaccharide was indicated by the complete absence of the terminal GalNAc anomeric carbon resonance at 92 ppm. The integrals of the GalNAc C-1, C-5, and N-acetyl carbon resonances and the -carbon resonances of Ser, Thr, and Gly after the TFMSA treatments were used to approximate the relative extent of glycosylation (50, 53). Comparisons of the integrals between fully relaxed nuclear Overhauser effect suppressed spectra (5.27 s recycle time) and partially relaxed spectra with full nuclear Overhauser effect typically revealed no systematic changes in resonance areas utilized for the analysis.

Proton NMR spectra were obtained at 600 MHz in 99.99% D₂O. Spectra were obtained at 45 °C to eliminate overlap of the HDO resonance with the GalNAc anomeric protons (54). HDO resonance presaturation was not typically utilized to avoid alterations in the anomeric proton resonance areas. The areas of the GalNAc anomeric proton resonances at 4.85 ppm and the GalNAc H2 protons that show sensitivity to Ser and Thr linkages (4.13 and 4.06 ppm, respectively (54)) were used to determine the extent of glycosylation. Areas were normalized to the methyl resonances of Ile and Val at 0.9 ppm, and the extent of glycosylation was obtained by utilizing the expected amino acid composition of the 81-residue PSM tandem repeat.

Amino Acid Analysis-- Amino acid analyses were performed by the Protein/Peptide Core Facility of the Massachusetts General Hospital on a Beckman 6300 amino acid analyzer. To determine the extent of destruction of Ser and Thr, hydrolysis times in 6 N HCl at 110 °C of 6, 12, 18, and 24 h were obtained. No significant losses of Ser or Thr were detected as a function of hydrolysis time. Nevertheless, the 18- and 24-h time points were used to extrapolate a 0 time nanomole content for each amino acid residue. Quantitation of O-glycosylation was estimated by amino acid analysis of -eliminated (0.1 M NaOH, 0.6 mM NaBH₄, 7 h at 45 °C) and reduced (0.66 M NaBH₄ + 0.016 M PdCl₂ in 0.8 M HCl) samples as described by Downs et al. (55). The extent of glycosylation was calculated from the loss of (glycosylated) Ser and Thr relative to the non-reactive amino acid residues and by the gain in Ala in the case of Ser.

Sequence "Density" Determinations and Data Analysis-- Sequence weighted average Ser/Thr density values were obtained by performing a 7-residue center weighted running average along the tandem repeat, using arbitrary weights of 0.5, 0.75, 1.0, 1.0, 1.0, 0.75, and 0.5 at each position and a denominator of 5.5. Ser and Thr residues were valued as 1 and all other residues were valued as 0 (34).² The coefficients were arbitrarily chosen to more heavily weight the central residues and to provide smoothing to the function. In vivo GalNAc density values were obtained using an identical weighting scheme while using the actual in vivo site-specific Ser or Thr percent glycosylation values. Statistical analyses were performed using Pearson product moment correlation procedure in the Sigma Stat statistical software package (version 2.0) (SPSS Inc., Chicago). Correlations were deemed significant for p values less than 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

PSM Oligosaccharide Characterization-- The possible O-linked glycan side chain structures described previously for PSM (Fig. 1) range from the monosaccharide Tn antigen, GalNAc-O-Ser/Thr, to the tetrasaccharide, blood group A structure, -GalNAc-1,3[-Fuc-1,2]--Gal-1,3--GalNAc-O-Ser/Thr (47, 51, 52). Each mono- to tetrasaccharide structure may also have a sialic acid (in the form of N-glycolylneuraminic acid, NeuNGl) 2-6-linked to the peptide GalNAc residue, and thus up to 8 different oligosaccharide structures are possible. Earlier carbon-13 NMR studies from our laboratory (50) have revealed that the relative proportion of these glycans varies among pigs, ranging from animals having over 50% the blood group A tetrasaccharide to animals having nearly equal amounts of the di- and trisaccharide and no tetrasaccharide. In the present study we chose to determine the site-specific glycosylation pattern of the latter class of animals having nearly equal amounts of the mono-, di-, and trisaccharide and no tetrasaccharide (see Table I). It was anticipated that from the analysis of these mucins we would be able to determine the extent that the activity of the 2-fucosyltransferase may be modulated by local peptide sequence.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Porcine mucin O-glycan structures and tandem repeat amino acid sequence. A, possible O-linked glycan structures ranging from the monosaccharide (Tn determinant) to the tetrasaccharide (blood group A determinant) (47, 51, 52). For each oligosaccharide a NeuNGl, sialic acid, residue may be 2,6-linked to the peptide core GalNAc residue. Typically 40-50% of the oligosaccharides contain NeuNGl(50). B, sequence of the PSM 81-residue tryptic tandem repeat (48). The repeat contains 31 hydroxyamino acids, 20 Ser and 11 Thr, dispersed along the sequence. Native mucin contains about 100 contiguous repeats of this sequence each containing one trypsin-susceptible site (49).

View this table: [in this window] [in a new window]   Table I 13C NMR-derived glycan compositions of individual A PSM preparations and estimated Core 1 PSM and Trisaccharide PSM percent GalNAc-O-Ser/Thr compositions

Selective Deglycosylation of PSM-- The determination of a glycoprotein's site-specific glycosylation pattern relies on the ability of mild TFMSA/anisole treatment to trim O-linked glycans to the peptide-linked GalNAc residue, the ability of periodate to selectively oxidize peptide-linked GalNAc residues that contain no C-3 substituents, and the ability to distinguish GalNAc-O-Ser/Thr residues by automated Edman amino acid sequencing (37, 53). By using a combination of periodate oxidation followed by mild TFMSA/anisole treatment, giving the Core 1 PSM tandem repeat, those sites containing core 1 (or longer) side chains (i.e. those attached to GalNAc via C-3) can be identified and quantified by Edman sequencing (34). This approach can be extended to the characterization of longer core 1 glycan side chains by sequential glycosidase treatments prior to the periodate cleavage step. To reveal the positions of the PSM-fucosylated trisaccharide (-Fuc-1,2--Gal-1,3--GalNAc-O-Ser/Thr), bovine testis -galactosidase was used to remove the -Gal residues from the disaccharide (-Gal-1,3--GalNAc-O-Ser/Thr), leaving mucin containing only monosaccharide and trisaccharide glycans. Subsequent periodate oxidation and mild TFMSA/anisole treatment removed the monosaccharide GalNAc residues while trimming the trisaccharide to GalNAc. The determination and quantification of the sites of trisaccharide attachment were obtained by the Edman sequencing of the derived Trisaccharide PSM tandem repeat obtained after trypsinolysis. Thus, by sequencing a series of sequentially deglycosylated Native-GalNAc, Core 1, and Trisaccharide PSM tandem repeat preparations, the site-specific mono-, di-, and trisaccharide distributions can be obtained.

Fig. 2 shows the carbon-13 NMR spectra for several of the deglycosylated preparations PSM B10 isolated from a single animal. As discussed in the figure legend, the spectra clearly demonstrate the selective removal of carbohydrate by neuraminidase and -galactosidase (Fig. 2B) and the increasing removal of GalNAc between the Native GalNAc, Core 1, and Trisaccharide PSM tryptic tandem repeats (Fig. 2, D-F). Also included in Fig. 2 are the NMR spectra of fully deglycosylated apomucin (Fig. 2G) and of a preparation of -galactosidase PSM B10 that has had its monosaccharide GalNAc residues removed by -N-acetylgalactosaminidase (Fig. 2C). As discussed in the figure legend and below, the latter spectrum shows that -N-acetylgalactosaminidase removes nearly the same amount of unsubstituted GalNAc as does the periodate oxidation, reduction, and TFMSA procedure (Fig. 2F).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Carbon-13 NMR spectra of native and selectively deglycosylated porcine mucin B10. Spectra A-C are for native and enzymatically deglycosylated mucin, and spectra D-F are for the series of Native, Core 1, and Trisaccharide mucin tryptic tandem repeats containing only peptide-linked GalNAc residues. The glycosylation patterns of the latter were determined by Edman amino acid sequencing (see Fig. 3 and Table III). A, TR-PSM containing intact oligosaccharide side chains. B, TR-PSM containing only the mono- and trisaccharide side chains, obtained after treatment with neuraminidase and -galactosidase. C, TR-PSM containing only trisaccharide side chains, obtained after the treatments in B followed by -N-acetylgalactosaminidase. Selective oligosaccharide side chains removal is shown by the anomeric carbon assignments in the 90-110 ppm region. Resonances from NeuNGl are labeled, N2-11, in A. Glycosylated (SOG and TOG) and nonglycosylated (SOH and TOH) Ser and Thr -carbons resonances, respectively, are labeled in spectra B and C in the 52-60 ppm region. Note the change in relative areas of the glycosylated and nonglycosylated -Ser and -Thr resonances after removal of GalNAc in spectrum C. D, Native GalNAc tryptic tandem repeat glycopeptide obtained after trimming all glycans to GalNAc by mild TFMSA. Peptide GalNAc carbons are labeled C1-6, and NAc. Note the resonance intensities for the glycosylated and nonglycosylated Ser and Thr -carbons in D are identical to those in A and B indicating no losses of peptide-linked GalNAc after TFMSA treatment. E, Core 1 tryptic tandem repeat glycopeptide obtained after removal of non-C-3-substituted GalNAc residues by periodate oxidation followed by mild TFMSA. This preparation contains peptide GalNAc residues only at positions originally substituted by core 1 (di and tri) oligosaccharides. Note the decrease in the anomeric carbon resonances of Ser- and Thr-linked GalNAc (~98 ppm) and the changes in the Ser and Thr -carbon resonances both indicating the partial deglycosylation of the mucin (i.e. decrease in TOG and SOG resonances and increase in TOH and SOH resonances). F, Trisaccharide tryptic tandem repeat glycopeptide obtained after neuraminidase and -galactosidase treatment (spectrum B) followed by periodate and mild TFMSA. This preparation contains peptide GalNAc residue only at positions originally substituted by the trisaccharide. Note the further loss of carbohydrate with respect to spectrum E and the similarity of the Ser and Thr -carbon resonance patterns to the enzymatically deglycosylated derivative containing intact trisaccharide, side chains, spectrum C. G, fully deglycosylated apoPSM tandem repeat domain showing the complete loss of carbohydrate. The -carbons of nonglycosylated Ser and Thr are labeled as are the -carbons of Ala and Gly. All spectra were plotted with a nearly constant height for the Gly -carbon resonance, 43 ppm. Sample sizes ranged from ~25 to ~3 mg.

The approximate extent of glycosylation determined from the NMR resonance areas of selected GalNAc and peptide carbons and protons (data not shown) is given in Table II. (Note that the average extent of glycosylation from the NMR data may not fully agree with the individual Ser and Thr glycosylations given because different sets of resonances were used for each determination; see Table II legend). For the most part the NMR-derived values agree reasonably well with the expected extent of glycosylation, although as discussed below the net Trisaccharide glycosylation is somewhat higher than expected. Both the proton and carbon-13 NMR analyses of the Trisaccharide derivatives indicate that Ser residues are more highly glycosylated than Thr residues (Table II). This is readily apparent by the relative areas of the glycosylated (SOG and TOG) and nonglycosylated (SOH and TOH) -carbon resonances of Thr and Ser in the 54-60 ppm range in spectrum F of Fig. 2 and by the relative areas of the GalNAc H-2 proton resonances of similar preparations (data not shown). Nearly identical differential glycosylation of Ser over Thr is also observed in the Trisaccharide-like preparation where -N-acetylgalactosaminidase was used to remove monosaccharide GalNAc residues rather than periodate and TFMSA (Fig. 2C and Table II). Therefore, the differences in Ser and Thr glycosylation are not likely artifacts of the chemical deglycosylation approach.

Amino Acid Sequencing and Glycosylation Pattern Determination-- Edman amino acid sequencing was performed on the differentially deglycosylated TFMSA-treated, 81-residue Native GalNAc, Core 1, and Trisaccharide tryptic tandem repeat glycopeptides obtained from each of the three PSM preparations described above. The 40-residue C-terminal glycopeptides resulting from Glu-C digestion were also sequenced to complete the analysis of the C-terminal portion of the tandem repeat. Representative Ser and Thr sequence profiles are given in Fig. 3, and the average residue-specific glycosylations for each PSM preparation (and the average of all three) are summarized in Table III. These data are further summarized in Table II and are in good agreement with the proton and carbon-13 NMR-derived glycosylation values. As with the NMR analysis, the Edman sequence analysis reveals a higher extent of glycosylation of Ser residues relative to Thr in the Trisaccharide PSM derivatives. These differences in Ser/Thr glycosylation are also detected when the extent of Ser and Thr O-glycosylation is estimated by the amino acid analysis of the -eliminated and reduced samples (Table II). Therefore, with three different PSM preparations using two different NMR approaches (¹³C and ¹H NMR), two different analytical chemical approaches (Edman sequencing and amino acid analysis after -elimination and reduction), and two different methods for the removal of monosaccharide GalNAc residues (periodate oxidation and -N-acetylgalactosaminidase), it is consistently found that Ser residues are more highly glycosylated by the trisaccharide than Thr residues. Combining all the analytical data in Table II, the average trisaccharide enhancement of Ser residues over Thr residues is ~1.6-fold.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Amino acid sequencing profiles of selectively deglycosylated tandem repeats from PSM preparation B10. A and D represent Ser and Thr profiles for Native mucin tandem repeats; B and E represent Ser and Thr profiles for Core 1 tandem repeats; C and F represent Ser and Thr profiles for Trisaccharide tandem repeats. GalNAc glycosylated (black lines, open square) and nonglycosylated (gray line, closed diamond) Ser and Thr residues are indicated. Data were base-line corrected as described previously (34, 37).

Edman sequencing derived values for the oligosaccharide distribution and percent GalNAc glycosylation for each of the three different PSM preparation are given in parentheses in Table I for comparison to the expected values. The Edman sequencing derived monosaccharide and summed di- + trisaccharide percentages and the Core 1 percent GalNAc glycosylation values (columns 1, 4, and 5, respectively) are in excellent agreement with the expected values, even maintaining the relative differences between each preparation. On the other hand, values for the percent glycosylation of the Trisaccharide (far right column) are about 10 percentage values higher than expected (29-37% versus an expected 20-29%), representing a 25-45% elevation in glycosylation over the expected values. Similar elevations in Trisaccharide percent glycosylation are observed by NMR and amino acid analysis (Table II). The elevated glycosylation of the Trisaccharide derivatives leads to a 30-40% increase in the calculated trisaccharide distribution at the expense of the disaccharide distribution over expected (columns 2 and 3, Table I). Nevertheless, the general differences among PSM preparations are preserved, with PSM B10 containing more disaccharide and less tetrasaccharide than PSM 12 or B7. After this discrepancy was noted, a series of studies were performed to determine the origins of the elevated glycosylation of the Trisaccharide derivatives.

First, it was necessary to determine whether the observed enhanced glycosylation of Ser could be the source of the increased total glycosylation because of specific destruction of nonglycosylated Ser as a result of the periodate oxidation or TFMSA treatments. This does not appear to be the case, because using -N-acetylgalactosaminidase to remove monosaccharide GalNAc residues from -galactosidase-treated mucin also reveals the enhanced glycosylation of Ser over Thr (¹³C NMR data Fig. 2, C and F, and Table II). Furthermore, standard amino acid analysis of the B7 and B10 derivatives show no significant loss of Ser or Thr among the differently deglycosylated derivatives (i.e. 22.5, 23.6, and 25.6% Ser and 13.8, 13.8, and 13.0% Thr for the Native GalNAc, Core 1, and Trisaccharide derivatives of B10, respectively; for comparison the expected values of Ser and Thr are 24.7 and 13.6% respectively). In addition, a control oxidation and reduction procedure performed on fully deglycosylated apoPSM also failed to alter its amino acid composition (data not shown).

That the amino acid compositions are constant regardless of the extent of glycosylation suggests the possibility that one or more steps in the procedure may be selectively degrading less highly glycosylated mucin tandem repeats or even intact mucin molecules. Several lines of evidence support this possibility. First, immobilized lectin chromatography of native PSM isolated from a single animal with intact oligosaccharide side chains can be fractionated into mucin species with different oligosaccharide distributions.³ These results suggest that isolated mucin is made up of a distribution of differently glycosylated molecules whose glycosylation presumably reflects the biosynthetic range of oligosaccharide structures, reflecting their different endoplasmic reticulum/Golgi localizations at the time of harvesting. It is also possible that heterogeneity may arise at the cellular level, with different cells producing differently glycosylated mucins. Regardless, the glycosylation patterns of individual mucin molecules would range from nascent-like poorly glycosylated peptide cores, containing mostly short oligosaccharides, to more heavily glycosylated mature mucin peptide cores, containing more fully elongated oligosaccharides. It is expected that those mucin molecules and tandem repeats with fewer and shorter oligosaccharide side chains would be more susceptible to degradation and loss. The large material losses (50-30% relative to theory) observed after a second TFMSA treatment of apoPSM and mucin containing only oxidized/reduced GalNAc side chains (i.e. mucin previously treated with TFMSA and subsequently oxidized and reduced), compared with the 100% recovery of mucin containing intact GalNAc side chains (i.e. mucin previously treated with TFMSA only) after the same TFMSA treatment, further suggest that deglycosylated apomucin and mucin containing only oxidized and reduced GalNAc residues are less stable toward TFMSA than mucins containing intact GalNAc residues. Intact GalNAc residues and longer oligosaccharide side chains therefore appear to protect the peptide core from TFMSA degradation, whereas oxidized-reduced GalNAc residues and regions of deglycosylated peptide do not. Other studies in our laboratory reveal an extreme sensitivity of apomucin to proteolytic degradation.³ Therefore, the higher than expected extent of glycosylation observed in the Trisaccharide PSM derivatives reflects the nonspecific destruction by TFMSA of poorly glycosylated tandem repeat regions of the mucin molecule and the loss of a population of immature mucin molecules containing short glycan side chains, both of which become more susceptible to TFMSA after -galactosidase and periodate treatment.⁴ We estimate that a 20-30% loss of poorly glycosylated mucin could account for the observed results. This results in an increase in the trisaccharide at the expense of the disaccharide over what would be expected (Table I) but no change in the core 1 or monosaccharide compositions. Therefore, the obtained site-specific glycosylation patterns are weighted toward the mucin fractions containing the more mature and more heavily glycosylated mucin molecules. These glycosylation patterns are expected to reasonably reflect the overall trends in site-specific glycosylation of the intact mucin.

PSM Tandem Repeat Site-specific Glycosylation Pattern-- Using the data in Table III we have calculated the apparent site-specific mono-, di-, and trisaccharide content for nearly every Ser and Thr in the tandem repeat. The results are given graphically in Fig. 4 for each of the three PSM preparations where the black, white, and gray bars represent, respectively, the mono-, di-, and trisaccharide content. The oligosaccharide distribution is given sequentially by residue number in the left panels of Fig. 4 (A-D) and grouped separately by Ser and Thr residue type in the right panels (E-H). The average site-specific distribution for all three preparations is shown at the bottom of the figure, D and H. (Note that the data in Fig. 4 are displayed so that the sum of the mono-, di-, and trisaccharides represents the extent of GalNAc glycosylation for each Ser or Thr residue, further indicating the degree of Ser or Thr glycosylation at each site). Also plotted in each panel is the derived Ser/Thr density function (gray lines), which is a measure of the relative density of Ser and Thr residues in the tandem repeat sequence (34). As discussed below, the Ser/Thr density appears to correlate with various aspects of the oligosaccharide distribution and with the in vivo peptide GalNAc density. Plots evaluating the correlation of the glycosylation patterns shown in Fig. 4 with Ser/Thr density are given in Fig. 5.

View larger version (71K): [in this window] [in a new window]   Fig. 4.   Sequence-specific mono-, di-, and trisaccharide glycosylation patterns of the PSM tryptic tandem repeats from three blood group A-negative animals and their average. Each residue's mono-, di-, and trisaccharide substitution is indicated by the black, white, and gray bars, respectively. The sum of the bars represents the total glycosylation of each residue by GalNAc. A-D display the data in sequential order along the tandem repeat and the E-H group display the data separately by Ser and Thr. A and E, PSM 12; B and F, PSM B7; C and G, PSM B10; and D and H the average of all three. The Ser/Thr density function is plotted as the open squares connected by the gray line (34). Data were derived from Table III. Where necessary values from Table III were adjusted to eliminate inconsistent (negative) values by increasing the Native GalNAc values or by lowering the Trisaccharide values given in Table III as necessary. Of the 116 glycosylation values listed in Table III, only 7 were adjusted by 1-3 percentage values, and 4 were adjusted by 4-7 percentage values. These changes are well within the estimated errors of the determinations. No other values in Table III were modified.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   PSM tandem repeat site-specific glycosylation patterns versus tandem repeat Ser/Thr (S/T) density (34). Left panels, A-E, represent data for Ser-linked glycans, whereas the right panels, F-J, represent data for Thr-linked glycans. The lines in each plot represent the linear least square fit to the data, omitting data for Ser-2 (open symbols) for the reasons discussed in the text. Note that individual data points representing each of the three PSM preparations are plotted in each panel and that for only the Ser-linked glycans are the trends with respect Ser/Thr density statistically significant (p < 0.05). A and F, percent monosaccharide versus Ser/Thr density (Ser, r2 = 0.28, p < 0.0001; Thr, r2 = 0.08 p = 0.35). B and G, percent core 1 substitution (sum of di- and trisaccharide) versus Ser/Thr density (Ser, r2 = 0.51, p < 0.0001; Thr, r2 = 0.15 p = 0.026). C and H, percent disaccharide versus Ser/Thr density (Ser, r2 = 0.15, p = 0.002; Thr, r2 = 0.11 p = 0.08). D and I, percent trisaccharide versus Ser/Thr density (Ser, r2 = 0.47, p < 0.0001; Thr, r2 = 0.03 p = 0.11). E and J, percent peptide-linked GalNAc on Ser/Thr versus Ser/Thr density (Ser, r2 = 0.32, p < 0.0001; Thr, r2 = 0.10 p = 0.02). Note that r2 and p values were obtained omitting data for Ser-2. Data points were derived from Table III (9 values out of 87 were adjusted as described in Fig. 4 to eliminate inconsistencies).

A detailed examination of Fig. 4 reveals that the glycosylation patterns of the three PSM derivatives are remarkably similar although, as discussed above, the trisaccharide content is greater than expected due to losses of poorly glycosylated tandem repeat. The distributions otherwise reflect each preparations original relative oligosaccharide distribution given in Table I; PSM 12 and B7 have very similar distributions, whereas PSM B10 has a distribution elevated in the disaccharide and lower in the mono- and trisaccharide. In general, the monosaccharide values agree well with the values given in Table I, whereas the trisaccharide values are increased at the expense of the disaccharide. The appropriate monosaccharide values and consistent trends observed in the di- and trisaccharide values between the different mucin preparations are taken as evidence that the Edman sequencing derived values are indeed representative of the site-specific oligosaccharide distribution of the mucins under study.

An analysis of the left panels in Fig. 4 reveal several trends. First there appears to be three regions where the monosaccharide content appears elevated in all three PSM preparations, in the range of residues 29-33, 49-50, and 57-64. These data confirm what we have shown previously on pooled mucin (34) that the monosaccharide (black bars, Fig. 4) and core 1 substitution patterns (sum of white and gray bars, Fig. 4) are not uniformly distributed along the tandem repeat and that their distribution appears to be directly and indirectly associated, respectively, to the Ser/Thr density plot (gray line, plots showing these correlations are shown in Fig. 5 and are discussed below). Overall glycosylation and oligosaccharide distribution also appear to be Ser- and Thr-specific as shown by the right panels, E-H, in Fig. 4. As shown earlier (34), Ser residues are more variably glycosylated by GalNAc than Thr (total height of bars), whereas the monosaccharide content of Thr residues is somewhat elevated relative to Ser residues (black bars). As discussed above, the plots also show more trisaccharide side chains attached to Ser residues compared with Thr residues (gray bars).

The differences between the glycosylation of Ser and Thr are also revealed by the Ser/Thr density plots in Fig. 5. Plots of the percent monosaccharide versus Ser/Thr density show a statistical correlation with Ser/Thr density for Ser residues (Fig. 5A, p < 0.0001, r² = 0.28) but not for Thr (Fig. 5F, p = 0.35, r² = 0.08). Inverse correlations with the core 1, di-, and trisaccharide contents with Ser/Thr density are also shown for Ser (Fig. 5B, p < 0.001, r² = 0.51; Fig. 5C, p = 0.002, r² = 0.15; Fig. 5D, p < 0.0001, r² = 0.47 respectively) but not for Thr (Fig. 5G, p = 0.026, r² = 0.15; Fig. 5, p = 0.079, r² = 0.11; Fig. 5I, p = 0.11, r² = 0.034, respectively). Even the plots of percent GalNAc on Ser/Thr are strongly inversely correlated with Ser/Thr density for Ser residues (Fig. 5E, p < 0.0001, r² = 0.33) and only barely inversely correlated for Thr residues (Fig. 5J, p = 0.022, r² = 0.097). The lack of a strong correlation for Thr may be the result of the relatively high and more uniform glycosylation of Thr residues by GalNAc relative to Ser. This is consistent with most in vitro studies showing that Thr residues are typically an order of magnitude more readily glycosylated compared with Ser (11, 17-21). Note that in Fig. 5, data from Ser-2 were excluded from the statistical analysis (open diamonds). This was justified because of the very low level of Native glycosylation observed for Ser-2 (~10%) which leads to very large errors in determining its glycosylation pattern by difference. In addition, Ser-2 is clearly an outlier on many of the plots. Unusual local sequence effects, beyond Ser/Thr density, presumably dominate the glycosylation of Ser-2 (34, 37).

Several trends are predicted by the Ser/Thr density plots in Fig. 5. At high Ser/Thr densities (i.e. ~1) the plots suggest that Ser residues will only be ~30% glycosylated by the monosaccharide and will contain no di- or trisaccharide. In contrast at very low Ser/Thr densities (i.e. ~0), Ser residues are predicted to be nearly fully elongated. They would be expected to contain ~20% and ~80% of the di- and trisaccharide side chains, respectively, while containing few monosaccharide side chains. For Thr, there are few statistically significant trends with respect to Ser/Thr density, although it would be expected that Thr residues would be relatively uniformly glycosylated at ~80% regardless of Ser/Thr density and would have more di- and fewer trisaccharide side chains.

The normalized core 1 substitution pattern (i.e. %Gal on GalNAc) is given in Fig. 6, A and E (where the white and gray bars represent the di- and trisaccharide proportions, respectively). This plot provides the relative site-specific propensity of the core 1, UDP-galactose:-N-acetylgalactosamine -1,3-galactosyltransferase (core 1 3-Gal transferase) to add -Gal to GalNAc irrespective of extent of Ser or Thr glycosylation by GalNAc. Note that for ease of presentation, averaged values obtained from the three PSM preparations have been plotted in Fig. 6. As we have shown previously on pooled mucin (34), these data clearly demonstrate that the propensity to add Gal varies uniquely along the peptide sequence, ranging from ~20 to nearly 100%. In addition, the plot suggests that the core 1 propensities appear to be indirectly associated with Ser/Thr density (gray line). Plots showing these correlations are given in Fig. 7, A and F, for Ser and Thr, respectively. Again, the propensity for core 1 glycosylation is highly correlated with Ser/Thr density for Ser (p < 0.0001, r² = 0.49) but not for Thr (p = 0.19, r² = 0.09). Note that individual data points from each PSM preparation are plotted in Fig. 7, and data for Ser 2 have been omitted in the correlation calculations for the reasons stated above.

View larger version (71K): [in this window] [in a new window]   Fig. 6.   Average normalized sequence-specific glycosylation patterns of the PSM tryptic tandem repeat from blood group A-negative animals. A-D display the data in sequential order along the tandem repeat, and E-H group the data separately by Ser and Thr. A and E, percent Gal on peptide GalNAc (i.e. percent core 1), percent disaccharide white bars, percent trisaccharide gray bars. B and F, percent Fuc on Gal. C and G average side chain length normalized to Ser or Thr. D and H average side chain length normalized to peptide-linked GalNAc. The Ser/Thr density function is plotted as the open squares connected by the gray line (34). Data were obtained from the average glycosylation values given in Table III (1 value out of 29 was adjusted as described in Fig. 4 to eliminate inconsistencies).

View larger version (27K): [in this window] [in a new window]   Fig. 7.   PSM tandem repeat normalized site-specific glycosylation patterns versus tandem repeat Ser/Thr density (34). Left panels, A-E, represent data for Ser-linked glycans, and the right panels, F-J, represent data for Thr-linked glycans. The lines in each plot represent the linear least square fit to the data, omitting data for Ser-2 (open symbols) for the reasons discussed in the text. Individual data points representing each of the three PSM preparations are plotted in each panel. A and F, percent Gal (i.e. core 1 glycans) on peptide GalNAc versus Ser/Thr density (Ser, r2 = 0.49, p < 0.0001; Thr, r2 = 0.09 p = 0.19). B and G, percent Fuc on Gal versus Ser/Thr density (Ser, r2 = 0.04, p = 0.055; Thr, r2 = 0.003, p = 0.59). C and H, hydroxyamino acid normalized side chain length versus Ser/Thr density (Ser, r2 = 0.47, p < 0.0001; Thr, r2 = 0.17, p = 0.005). D and I, peptide-linked GalNAc normalized side chain length versus Ser/Thr density (Ser, r2 = 0.46, p < 0.0001; Thr, r2 = 0.06 p = 0.31). E and J, in vivo GalNAc glycosylation (OG) density versus Ser/Thr density (Ser, r2 = 0.69, p < 0.0001; Thr, r2 = 0.86 p < 0.0001). Excluding the percent Fuc on Gal versus Ser/Thr density plot (B) the trends with respect Ser/Thr density are statistically significant (p < 0.05) for Ser-linked glycans only. Note that r2 and p values were obtained omitting data for Ser-2. Data points were derived from Table III (9 values out of 87 were adjusted as described in Fig. 4 to eliminate inconsistencies).

Fig. 6B shows the normalized site-specific fucosylation pattern (i.e. % Fuc on Gal) which reports on the propensity of the porcine GDP-fucose:galactose 1,3-1,2-fucosyltransferase (2-Fuc transferase) to transfer Fuc to -Gal. The plots suggest that the fucosyltransferase adds sugar to the -Gal residue in a nonuniform, site-specific manner, ranging from 35% to nearly 100%. However, when data for Ser- and Thr-linked glycans are plotted separately (Fig. 6F), it is evident that within each residue type the normalized extent of Fuc substitution is relatively uniform, ~80% for Ser and ~45% for Thr. The site-specific variations in fucosylation in Fig. 6B seem to be primarily due to Ser/Thr linkage type rather than any other unique local property of the peptide sequence. As confirmation, plots of normalized percent fucosylation versus Ser/Thr density, Fig. 7, B and G, do not reveal significant correlations with Ser/Thr density (p = 0.06, r² = 0.04 and p = 0.59, r² = 0.003, for Ser and Thr, respectively). These results suggest the porcine salivary gland 2-Fuc transferase recognizes differences in glycan Ser/Thr linkages but is otherwise relatively insensitive to polypeptide sequence or Ser/Thr density.

Note that in Fig. 5D the trisaccharide content on Ser residues appears inversely proportional to the Ser/Thr density of the residue, and in Fig. 7B the normalized propensity for adding Fuc to -Gal is not correlated with Ser/Thr density. This apparent contradiction arises at the previous step in the biosynthesis of the trisaccharide by the core 1 3-Gal transferase that produces the acceptor substrate for the 2-Fuc transferase. It is the relative synthesis of the disaccharide by the core 1 3-Gal transferase, shown to be sensitive to Ser/Thr density (for Ser-linked glycans, Figs. 5B and 7A), that is responsible for the inverse correlation between trisaccharide content and Ser/Thr density.

From the obtained site-specific oligosaccharide distribution, the site-specific glycan side chain lengths can be obtained. Side chain length can be calculated in two different ways as follows: one that includes the non-glycosylated proportion of Ser or Thr residues in the calculation, normalizing to Ser/Thr, and the other that excludes the non-glycosylated proportion of Ser and Thr, normalizing to peptide GalNAc. The former value provides an indication of the full extent of glycosylation of an individual Ser or Thr residue, whereas the latter provides information on the average length of the oligosaccharide side chains that are actually attached to a given residue regardless of the extent that the Ser or Thr is O-glycosylated. These different side chain length distributions, averaged over the three PSM preparations for the ease of presentation, are shown in Fig. 6, C, D, G, and H. These plots also show that the side chain lengths are not uniformly distributed along the peptide core (ranging from ~0.2 to ~2.2 and ~1.5 to ~2.8 residues, respectively) and that there also appears to be an inverse association of side chain length with Ser/Thr density (gray line). This is borne out for Ser-linked glycans (Fig. 7, C and D, p < 0.0001, r² = 0.48 and p < 0.0001, r² = 0.46) and less so for Thr-linked glycans (Fig. 7, H and I, p = 0.0053, r² = 0.17 and p = 0.32, r² = 0.06). No significant difference between the average side chain length of Ser-linked and Thr-linked glycan side chains is observed when normalized to Ser or Thr (1.4 and 1.5 residues, respectively, Fig. 6G), although Ser residues show greater variability ranging from ~0.2 to ~2.2 residues versus ~1.2 to ~1.8 residues for Thr. Only after normalizing to peptide GalNAc do differences in the average side chain lengths become evident, with Ser residues on average having slightly longer side chains than Thr (2.4 versus 2.0 residues, respectively, Fig. 6H). The variability between Ser- and Thr-linked glycans is similar, with Ser ranging from ~1.4 to ~2.7 residues and Thr ranging from ~1.4 to ~2.4 residues. Least square fits of the Ser side chain length data in Fig. 7, C and D, suggests that at very low Ser/Thr densities, the side chain lengths would be in the range of 2.5-3 residues. At high Ser/Thr densities, lengths in the range of ~0 and ~ 1 residues would be expected. Plots for Thr-linked glycans (Fig. 7, H and I) are not statistically correlated with Ser/Thr density.

Fig. 7, E and J, show the Ser- and Thr-specific plots of Ser/Thr density versus overall peptide GalNAc density derived from the Native glycosylation patterns given in Table III. From these plots, it is clear that the Ser/Thr density values are highly correlated for both Ser and Thr residues with the in vivo density of GalNAc glycosylation (Fig. 7E, p < 0.0001, r² = 0.69, Fig. 7J, p < 0.0001, r² = 0.86). These high correlations indicate that Ser/Thr density may be taken as a reasonable representation of a site's local GalNAc glycosylation density. On this basis we propose that the trends observed in Figs. 5 and 7 with respect to Ser/Thr density, and particularly with Ser-linked glycans, can be explained by inhibitory effects of neighboring GalNAc glycosylation due to steric interactions of neighboring GalNAc residue with transferase. Thus at high Ser/Thr densities, and by inference high neighboring GalNAc densities, the extent of site-specific Ser/Thr glycosylation by GalNAc is decreased due to the reduced access of the ppGalNAc transferase to the peptide core (Fig. 5E).⁵ Likewise, the accessibility of the core 1 3-Gal transferase to a given GalNAc may also be reduced at high Ser/Thr density (i.e. high GalNAc density) due to steric interactions. At high Ser/Thr GalNAc densities it is found that the extent of core 1 substitution is decreased (Fig. 5, B-D, and Fig. 7A), whereas the monosaccharide content is increased (Fig. 5A). These trends cumulatively result in the observed inverse behavior of side chain length with Ser/Thr GalNAc density shown in Fig. 7, C and D. It is intuitively very reasonable to expect that ppGalNAc transferase and core 1 3-Gal transferase would be sensitive to local GalNAc glycosylation density, whereas 2-Fuc transferase, whose acceptor substrate is 2 residues removed from the peptide core, would not. Attempts to relate the observed site-specific glycosylation patterns to other properties of the PSM mucin sequence such as predicted secondary structure have thus far been uninformative.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The goal of this work was to determine further the extent that O-glycosylation patterns may be affected by local peptide sequence by extending our studies to the characterization of the site-specific distribution of the blood group H trisaccharide, -Fuc-1,2--Gal-1,3--GalNAc-O-Ser/Thr, on the PSM tandem repeat. This work represents the most complete determination of the in vivo site-specific oligosaccharide distribution of any heavily O-glycosylated mucin type glycoprotein.⁶

The present work was undertaken using mucin isolated from three individual blood group A-negative animals, with nearly identical overall oligosaccharide side chain distributions (Table I). This was done to eliminate potential difficulties arising from characterizing mucin pooled from individuals with widely varying oligosaccharide side chain distributions. The obtained glycosylation patterns were found to be very similar among the different animals (Fig. 4), and in keeping with the peptide-linked GalNAc and core 1 glycosylation patterns reported earlier on a pooled mucin preparation composed of a mixture of blood group A-positive and -negative mucins (34, 37). These findings suggest that in the porcine salivary gland, the activities of the initiating ppGalNAc transferases and core 1 3-Gal transferase that synthesize the core and core 1 glycans remain relatively constant between animals regardless of individual blood group activities. Individual to individual variability in the PSM oligosaccharide structures arises predominantly at the stage of tri- and tetrasaccharide biosynthesis.

As discussed under "Results," the procedure for obtaining the trisaccharide distribution leads to the loss of a fraction (20-30%) of poorly glycosylated mucin, either by the specific loss of nascent poorly glycosylated mucin molecules or by nonspecific loss of poorly glycosylated tandem repeats. As a result, the obtained glycosylation patterns are representative of a more mature, more highly glycosylated fraction of mucin molecules.

This work has confirmed our earlier findings (34) that the porcine core 1 3-Gal transferase is sensitive to peptide sequence and that a significant extent of its activity, particularly for Ser-linked glycans, appears to be inversely correlated to local hydroxyamino acid density (Figs. 5B and 7A). This effect was proposed to be due to inhibition of the transferase by local steric effects of neighboring glycosylated Ser and Thr residues (34). The highly correlated relationship of the Ser/Thr density function with the similarly derived in vivo peptide GalNAc density function, shown in Fig. 7, E and J, supports this interpretation. As it is observed in vitro, it is likely that in vivo Thr residues are more rapidly glycosylated by GalNAc than Ser residues; therefore, the initially glycosylated Thr residues may interfere with the subsequent glycosylation and elongation of neighboring Ser residues. In contrast, the more rapidly glycosylated Thr residues are likely to initially have poorly glycosylated neighbors, hence their initial rates of core 1 substitution would be less likely to be affected by neighboring site glycosylation. We may be observing the differential in Ser and Thr glycosylation kinetics, which may explain why the glycosylation of Ser residues is more highly inversely correlated to Ser/Thr density than the glycosylation of Thr residues. The above findings are also consistent with earlier observations from in vitro glycosylation studies on model glycopeptides where the specificities of both the ppGalNAc and core 1 transferases were proposed to be affected by steric effects of neighboring glycosylated residues (22, 23, 32, 33, 58, 59).

Other presently unknown sequence or conformational factors associated with Ser/Thr density may be the origin of the observed correlations. The termination of monosaccharide elongation by 2-6 sialylation of GalNAc is not a likely alternative explanation, because previous studies of the PSM oligosaccharide alditols (51) indicate that only about ~25% of the monosaccharide GalNAc residues are substituted by NeuNGl, whereas ~50% of the core 1-substituted glycans were NeuNGl-substituted. One would expect higher levels of monosaccharide sialylation if sialylation were playing a significant role in terminating GalNAc elongation. Work is in progress determining the site-specific sialylation patterns of the PSM monosaccharide side chains to confirm this conclusion. Peptide substrate specificities for the sialyltransferases (ST6GalNAc I and II) that specifically add 2-6 sialic acid to unsubstituted peptide-linked GalNAc have yet to be characterized (56, 57).

In this work we have also shown that the extent of site-specific glycosylation of Ser residues by peptide-linked GalNAc appears to be inversely correlated with its Ser/Thr density (Fig. 5E), suggesting that in vivo one or more of the porcine salivary gland ppGalNAc transferase(s) may be affected by steric interactions of neighboring glycosylated residues. Indeed, our in vitro glycosylation studies on apoPSM with ppGalNAc T1 tend to support this conclusion.³

For the three PSM preparations studied, the tandem repeat trisaccharide distribution is shown to vary in a relatively reproducible sequence-dependent manner (Figs. 4 and 6B). However, after grouping the Ser and Thr residues separately, it was shown that these differences were predominantly due to the glycan's Ser/Thr glycosidic linkage (Fig. 6F) rather than to differences in neighboring peptide sequence or to Ser/Thr density (Fig. 7, B and G). These results suggest that the porcine salivary gland 2-Fuc transferase, responsible for forming the trisaccharide, is particularly sensitive to Ser/Thr glycosidic linkage and less so to peptide sequence or Ser/Thr density. Possible origins for these differences are discussed below. The total trisaccharide content at a given site appears to be inversely correlated to Ser/Thr density on Ser-linked glycans (Fig. 5D). As discussed under "Results," this is due to the apparent sensitivity of the core 1 3-Gal transferase that synthesizes the precursor disaccharide to Ser/Thr density (Figs. 5B and 7A). Ser-linked glycan side chain lengths are also correlated with Ser/Thr density (Fig. 7, C and D) for similar reasons.

The porcine 2-Fuc transferase responsible for forming the trisaccharide, in contrast to the core 1 3-Gal transferase, shows no apparent sensitivity to adjacent peptide sequence (Fig. 6F) or to local Ser/Thr density (Fig. 7, B and G). This suggests that the terminal -galactose acceptor may project far enough away from the peptide core that neighboring peptide and carbohydrate side chains fail to influence significantly transferase activity and/or accessibility. However, a pronounced difference in 2-Fuc transferase activity against glycans attached to Ser and Thr is observed, with Ser-linked glycans containing ~1.6-fold more Fuc than Thr-linked glycans (Fig. 6F). This unusual finding suggests that there may be differences in the conformation or dynamics of the acceptor disaccharide. These differences may influence the binding or accessibility of the acceptor to the transferase, or the transferase substrate-binding site may specifically recognize the peptide glycosidic linkage. Threonine O-glycosidic linkages are more restricted and less flexible than Ser linkages and may have different conformations (50, 60-63). Differential 2-6 sialylation by ST6GalNAc I or II of Ser or Thr glycans may also play a role in directing the core 1 glycan fucosylation toward Ser-linked glycans. The Ser/Thr peptide substrate specificities of the ST6GalNAc transferases active against the peptide-linked disaccharide T antigen remain unexplored.

Our ¹³C NMR analysis of the oligosaccharide structures on PSM isolated from a series of blood group A-negative glands reveals what appears to be two general classes of mucins with different fucosylation patterns. One class of mucins, whose site-specific patterns were studied here, contains typically 1/3 each of the mono-, di-, and trisaccharides, and the second class contains typically ~1/3 the monosaccharide and 1/2 to 2/3 the trisaccharide and very little disaccharide. It is possible that differences in the expression or activity of the porcine FUT2 (Se enzyme) 2-fucosyltransferase may be responsible for these differences among blood group A -negative animals (64, 65). Because mucin substrates containing -Gal-1,3--GalNAc-O-Ser/Thr can be fully fucosylated by excess porcine salivary gland 2-Fuc transferase (presumably the Se enzyme) in vitro (66, 67), a simple difference in copy number of the enzyme could readily account for the observed different glycosylation patterns. This explanation is in keeping with the generally observed lower activity (lower V_max/K_m values) of the Se enzyme compared with the H blood group 2 fucosyltransferase, FUT1 (68-71). On this basis, the observed preference for the fucosylation of Ser-linked glycans over Thr residues, in the less highly fucosylated mucins, could arise as a result of a low copy number of the Se enzyme combined with a moderate kinetic preference for Ser-linked glycans over Thr-linked glycans for the transferase. Recently it has been shown that a polymorphism in porcine FUT1 leads to protection against edema and post-weaning diarrhea in swine infected with toxigenic Escherichia coli (64). The expression of FUT1 in the porcine salivary gland is unknown, but it is conceivable that potential differences in the expression of FUT1 and its alleles may further account for the variable fucosylation observed among individual swine and between Ser- and Thr-linked glycans.

It is interesting to compare the glycosylation patterns of individual residues. For example the three least glycosylated residues of Fig. 4D, Ser-2 Ser-54, and Ser-62 (9-32% glycosylated), all appear to have relatively high core 1 substitutions (54-99%, Fig. 6A) and high levels of fucosylation (84-95%, Fig. 6B), and therefore relatively long average side chain lengths (2.1-2.9 sugar residues per GalNAc, Fig. 6D). In contrast, Ser-32 and Ser-63, which are 45 and 39% glycosylated on average by GalNAc, respectively, are poorly core 1-substituted (21%, Fig. 6A), but these are highly fucosylated (100 and 90%, Fig. 6B). These residues therefore have somewhat shorter average side chain lengths (~1.4 sugars per GalNAc, Fig. 6D). The glycans on Thr-29, Thr-30, Thr-49, and Thr-50 also have shorter average side chain lengths (1.4-1.8 sugars per GalNAc, Fig. 6D), but this is due to both lower core 1 glycosylation (30-55%, Fig. 6A) and lower fucosylation (33-45%, Fig. 6B). Thr-37, Thr-39, and Thr-52 have somewhat longer average side chain lengths (2.2-2.4 sugars per GalNAc, Fig. 6D) compared with the other Thr residues, due to both higher core 1 substitution (78-93%, Fig. 6A) and higher fucosylation (48-72%, Fig. 6B). Those residues with shorter side chains, Ser-32, Ser-63, Thr-29, Thr-30, Thr-49, and Thr-50, appear near peaks in the Ser/Thr density function (Fig. 6D). Also note that the residues with the lowest and highest Ser/Thr densities, Ser-43 and Ser-63, have nearly the longest and shortest average side chain lengths, 2.7 and 1.4 residues per GalNAc, respectively (Fig. 6D). Placing the longer side chains in less densely glycosylated regions of the peptide may help render the peptide core more uniformly protected against proteolytic degradation.

These studies clearly show that the mucin polypeptide sequence either directly or indirectly, affects the distribution of O-glycan structures in vivo in a reproducible, site-specific manner. Apparent correlations of the density of hydroxyamino acids with the distribution of Ser-linked glycan structures have led us to propose that steric interactions of neighboring glycosylated residues may play a significant role in the initial glycosylation and core 1 elongation of the porcine mucin, where presumably most Ser and Thr residues would be in sequences and conformations optimized for relatively efficient O-glycosylation. We further suggest that the observed differences in glycosylation sensitivity of Ser and Thr residues to Ser/Thr density may arise because of their different intrinsic rates of glycosylation by GalNAc. Because there are reproducible deviations in the correlations with Ser/Thr density, additional factors such as local peptide sequence or conformation must also contribute to the modulation of the initial O-glycosylation of the mucin. It is also likely that these latter factors may play more significant roles for substrates that may not be as optimized for O-glycosylation or that are less densely glycosylated than the mucins. In contrast, the distribution of the trisaccharide suggests that the addition of the Fuc residue by the porcine 2-fucosyltransferase is relatively insensitive to local peptide sequence but instead appears to be sensitive to a presently unknown property that differs between the Ser and Thr peptide glycosidic linkages. These findings indicate that the initial elongation of O-glycans is not an entirely passive or random process and that considerable work needs to be done before we have an understanding of the molecular basis of its regulation.

	ACKNOWLEDGEMENTS

We acknowledge the technical assistance of Cheryl Owens of the Molecular Biology Core Laboratory of the Case Western Reserve University Cancer Center, Case Western Reserve University Center for AIDS Research, and Skin Diseases Research Center (funded in part by National Institutes of Health Awards P30-CA43703, P30-AI36219, and P30-AR39750). We also thank Ashok Kahatri of the Protein/Peptide Core Facility Massachusetts General Hospital for the amino acid analysis and for helpful discussions. The assistance of Jessica Levine is also acknowledged.

	FOOTNOTES

^* This work was supported by NCI Grant RO1-CA-78834 from the National Institutes of Health and by the Cystic Fibrosis Foundation.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed: Dept. of Pediatrics, Case Western Reserve University School of Medicine, BRB, 2109 Adelbert Rd., Cleveland, OH 44106-4948. Tel.: 216-368-4556; Fax: 216-368-4223; E-mail: txg2@po.cwru.edu.

Published, JBC Papers in Press, January 2, 2002, DOI 10.1074/jbc.M111690200

² Note that the weighting coefficients reported here differ from the incorrectly given coefficients in Ref. 34. The numerical values listed for the Ser-Thr densities of the PSM tandem repeat in Ref. 34, however, are correct and are actually based on the coefficients given in this work.

³ T. A. Gerken, M. Gilmore, and J. Zhang, unpublished results.

⁴ The differences in overall glycosylation of the -N-acetylgalactosaminidase derived Trisaccharide-PSM B10 compared with the derivative obtained via oxidation and reduction after TFMSA treatment (~44 versus ~30%, Table II) suggest that the presence of oxidized and reduced GalNAc residues afford some protection from degradation by TFMSA. Therefore, the oxidation reduction-TFMSA approach for selectively eliminating unsubstituted GalNAc residues is at present the best available approach for these studies.

⁵ It should be noted that it is not inconsistent to have Ser/Thr density inversely correlated with the site-specific glycosylation of Ser residues by GalNAc while also having Ser/Thr density directly correlated with GalNAc density. In the former comparison a specific site is compared with Ser/Thr density, whereas in the latter the net properties of several neighboring residues are compared with Ser/Thr density.

⁶ All that remains in order to complete the characterization of the PSM tandem repeat glycosylation pattern is the determination of its site-specific sialylation pattern, which is currently in progress.

	ABBREVIATIONS

The abbreviations used are: ppGalNAc transferase, UDP-GalNAc:polypeptide -GalNAc transferase; 2-Fuc transferase, GDP-fucose: galactose -1,3-1,2-fucosyltransferase; core 1 3-Gal transferase, UDP-galactose:-N-acetylgalactosamine -1,3-galactosyltransferase; ST6GalNAc, GalNAc 2,6-sialyltransferase; Fuc, fucose; NeuNGl, N-glycoloylneuraminic acid or sialic acid; PSM, porcine submaxillary mucin; TR, tandem repeat; PTH, phenylthiohydantoin; Native GalNAc or (T)TTTR-PSM, Core 1 or (T)TROTR-PSM, Trisaccharide, or (T)TROGNTR-PSM, a series of differentially deglycosylated preparations of the PSM multiple tandem repeat domain described under "Materials and Methods" and under "Results," where (T) represents the 81-residue tandem repeat domain obtained after trypsinolysis; TFMSA, trifluoromethanesulfonic acid; SOG, glycosylated Ser; SOH, nonglycosylated Ser; TOG, glycosylated Thr; TOH, nonglycosylated Thr.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				C. Perrine, T. Ju, R. D Cummings, and T. A Gerken Systematic determination of the peptide acceptor preferences for the human UDP-Gal:glycoprotein-{alpha}-GalNAc {beta}3 galactosyltranferase (T-synthase) Glycobiology, March 1, 2009; 19(3): 321 - 328. [Abstract] [Full Text] [PDF]

				E. B. Maryon, S. A. Molloy, and J. H. Kaplan O-Linked Glycosylation at Threonine 27 Protects the Copper Transporter hCTR1 from Proteolytic Cleavage in Mammalian Cells J. Biol. Chem., July 13, 2007; 282(28): 20376 - 20387. [Abstract] [Full Text] [PDF]

				T. A. Fritz, J. Raman, and L. A. Tabak Dynamic Association between the Catalytic and Lectin Domains of Human UDP-GalNAc:Polypeptide {alpha}-N-Acetylgalactosaminyltransferase-2 J. Biol. Chem., March 31, 2006; 281(13): 8613 - 8619. [Abstract] [Full Text] [PDF]

				B. A. Potter, R. P. Hughey, and O. A. Weisz Role of N- and O-glycans in polarized biosynthetic sorting Am J Physiol Cell Physiol, January 1, 2006; 290(1): C1 - C10. [Abstract] [Full Text] [PDF]

				T. Freire, S. Bay, S. von Mensdorff-Pouilly, and E. Osinaga Molecular Basis of Incomplete O-Glycan Synthesis in MCF-7 Breast Cancer Cells: Putative Role of MUC6 in Tn Antigen Expression Cancer Res., September 1, 2005; 65(17): 7880 - 7887. [Abstract] [Full Text] [PDF]

				L. L. W. Cooling, K. Kelly, J. Barton, D. Hwang, T. A. W. Koerner, and J. D. Olson Determinants of ABH expression on human blood platelets Blood, April 15, 2005; 105(8): 3356 - 3364. [Abstract] [Full Text] [PDF]

				C. J. Schultz, K. L. Ferguson, J. Lahnstein, and A. Bacic Post-translational Modifications of Arabinogalactan-peptides of Arabidopsis thaliana: ENDOPLASMIC RETICULUM AND GLYCOSYLPHOSPHATIDYLINOSITOL-ANCHOR SIGNAL CLEAVAGE SITES AND HYDROXYLATION OF PROLINE J. Biol. Chem., October 29, 2004; 279(44): 45503 - 45511. [Abstract] [Full Text] [PDF]

				K. G. Ten Hagen, D. T. Tran, T. A. Gerken, D. S. Stein, and Z. Zhang Functional Characterization and Expression Analysis of Members of the UDP-GalNAc:Polypeptide N-Acetylgalactosaminyltransferase Family from Drosophila melanogaster J. Biol. Chem., September 12, 2003; 278(37): 35039 - 35048. [Abstract] [Full Text] [PDF]

				K. G. Ten Hagen, T. A. Fritz, and L. A. Tabak All in the family: the UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferases Glycobiology, January 1, 2003; 13(1): 1R - 16R. [Abstract] [Full Text] [PDF]

				T. A. Gerken, J. Zhang, J. Levine, and A. Elhammer Mucin Core O-Glycosylation Is Modulated by Neighboring Residue Glycosylation Status. KINETIC MODELING OF THE SITE-SPECIFIC GLYCOSYLATION OF THE APO-PORCINE SUBMAXILLARY MUCIN TANDEM REPEAT BY UDP-GalNAc:POLYPEPTIDE N-ACETYLGALACTOSAMINYLTRANSFERASES T1 AND T2 J. Biol. Chem., December 13, 2002; 277(51): 49850 - 49862. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 277/10/7736    most recent M111690200v1 Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gerken, T. A. Articles by Zhang, J. Search for Related Content PubMed PubMed Citation Articles by Gerken, T. A. Articles by Zhang, J. Social Bookmarking                      What's this?