A tandem mass spectrometric approach to determination of chondroitin/dermatan sulfate oligosaccharide glycoforms

Abstract

Dermatan sulfate (DS) chains are variants of chondroitin sulfate (CS) that are expressed in mammalian extracellular matrices and are particularly prevalent in skin. DS has been implicated in varied biological processes including wound repair, infection, cardiovascular disease, tumorigenesis, and fibrosis. The biological activities of DS have been attributed to its high content of IdoA(α1-3)GalNAc4S(β1-4) disaccharide units. Mature CS/DS chains consist of blocks with high and low GlcA/IdoA ratios, and sulfation may occur at the 4- and/or 6-position of GalNAc and 2-position of IdoA. Traditional methods for analysis of CS/DS chains involve differential digestion with specific chondroitinases followed by steps of chromatographic isolation of the products and disaccharide analysis on the individual fraction. This work reports the use of tandem mass spectrometry to determine patterns of sulfation and epimerization of CS/DS oligosaccharides in a single step. The approach is first validated, and then applied to a series of skin DS samples and to decorins from three different tissues. DS samples ranged from 74-99% of CSB-like repeats, using this approach. Decorin samples ranged from 30% CSB-like repeats for that from articular cartilage to 75% for that from sclera. These values agree with known levels of glucuronyl C5-epimerase in these tissues.

Introduction

Glycosaminoglycan (GAG) glycoforms are of particular interest because of their varied biological properties. PGs/GAGs are predominantly found on cellular membranes and in the extracellular matrix of mammalian tissues. They interact with numerous proteins, growth factors, and cell surface receptors, and thus serve a variety of important roles in cell proliferation, differentiation, adhesion, migration and recruitment of neutrophils (Bernfield, M., Gotte, M., et al. 1999, Hocking, A.M., Shinomura, T., et al. 1998, Trowbridge, J.M. and Gallo, R.L. 2002). GAGs are linear, sulfated polysaccharides that are attached by a linker to a serine residue of the core protein of a proteoglycan; the structures of GAGs bound to a given core protein are known to vary depending on the environment (tissue, developmental disease or cytokine state) (Clement, A.M., Nadanaka, S., et al. 1998, Faissner, A., Clement, A., et al. 1994, Kitagawa, H., Tsutsumi, K., et al. 1997, Tiedemann, K., Olander, B., et al. 2005). The dynamic nature and biological importance of GAG glycoforms drives the need for a rapid and sensitive means of determination of their sulfation and epimerization patterns. Dermatan sulfate (DS) is a CS variant that exhibits anticoagulant activity in both animals and humans, and has been incorporated into antithrombic therapies (Cofrancesco, E., Boschetti, C., et al. 1994, Di Carlo, V., Agnelli, G., et al. 1999, Maimone, M.M. and Tollefsen, D.M. 1990). In contrast to other antithrombic GAGs, DS is effective on both free and fibrin-bound thrombin (Bendayan, P., Boccalon, H., et al. 1994, Liaw, P.C., Becker, D.L., et al. 2001) and has low haemorrhagic complications (Cofrancesco, E., Boschetti, C., et al. 1994, Di Carlo, V., Agnelli, G., et al. 1999). Scleral proteoglycans interact via their DS chains, demonstrating the multifacited potentials of GAGs in promoting multifaceted binding with other macromolecules (Fransson, L.A., Coster, L., et al. 1982).

DS interactions with fibroblast growth factors FGF-2 (Penc, S.F., Pomahac, B., et al. 1998) and FGF-7 (Trowbridge, J.M., Rudisill, J.A., et al. 2002) have been shown with respect to cellular proliferation and wound repair, respectively. Interactions with hepatic growth factor/scatter factor (HGF/SF) establish that DS plays roles in tumorigenesis and metastasis (Lyon, M., Deakin, J.A., et al. 2002). Oversulfated DS chains play roles in promoting neurite ourgrowth in embryonic brain (Hikino, M., Mikami, T., et al. 2003). Decorin, a DS-containing proteoglycan predominant in skin, acts as an inflammation regulator through interaction with C1q (Krumdieck, R., Hook, M., et al. 1992) and transforming growth factor beta (Yamaguchi, Y., Mann, D.M., et al. 1990) and may play an important role in the inflammatory process of Lyme disease (Guo, B.P., Norris, S.J., et al. 1995).

Although the IdoA(α1-3)GalNAc4S(β1-4) disaccharide unit is the most highly expressed in mammalian DS, analyses of several preparations reveal heterogeneity within the polysaccharide chain as well as tissue origin (Poblacion, C.A. and Michelacci, Y.M. 1986). This is due to the biosynthetic modifying reactions that generate DS from nascent chondroitin chains of composition [GlcA(β1-3)GalNAc(β1-4)]_n after the initial polymerization; epimerization of GlcA to IdoA occurs prior to sulfation (Malmstrom, A. 1984, Malmstrom, A. and Aberg, L. 1982). Mammalian DS consists of domains of high percentage of IdoA(α1-3)GalNAc4S(β1-4) and those with high GlcA(β1-3)GalNAc4S(β1-4), respectively. Variable quantities of GlcA(β1-3)GalNAc6S(β1-4) are also present (Poblacion, C.A. and Michelacci, Y.M. 1986). DS chains from the body of an ascidian Ascidia nigra have a repeat unit of IdoA2S(α1-3)GalNAc6S(β1-4) (Pavao, M.S., Mourao, P.A., et al. 1995) and lack the anticoagulant activity of 4-sulfated DS (Vicente, C.P., Zancan, P., et al. 2001). Monoclonal antibodies generated against Ascidia DS have demonstrated the presence of IdoA2S(α1-3)GalNAc6S(β1-4) repeats in mouse brain (Bao, X., Pavao, M.S., et al. 2005).

Since the ability of GAGs to bind protein receptors largely depends on their structure, the number of disaccharide repeats (size), and sulfation pattern, analytical techniques to determine GAG fine structure are highly valuable. Total uronic acid content for GAG preparations may be determined using the carbazole reaction (Bitter, T. and Muir, H.M. 1962, Meyer, K. and Rapport, M.M. 1950, Rapport, M., Meyer, K., et al. 1951). Chemical measurement of IdoA and GlcA content entails reaction of uronate carboxyl groups with a carbodiimide, followed by reduction and acid hydrolysis. The monosaccharides products, differing for GlcA and IdoA, are then per-O-benzoylated, separated and quantified using chromatography (Karamanos, N.K., Hjerpe, A., et al. 1988). The IdoA content of CS/DS chains may be determined by differential chondroitinase digestion followed by chromatographic separation (Coster, L., Malmstrom, A., et al. 1975). The differential chondroitinase digestion approach has been used in conjunction with subsequent size exclusion chromatography (Malmstrom, A. and Fransson, L.A. 1975), reversed phase ion pairing HPLC (Karamanos, N.K., Hjerpe, A., et al. 1988), or slab gel electrophoresis (Cheng, F., Heinegard, D., et al. 1994, Fransson, L.A., Havsmark, B., et al. 1990). Oligosaccharide products of chondroitinase reactions isolated by RPIP-HPLC or other techniques may be subjected to additional chemical or enzymatic degradative steps (Theocharis, D.A., Papageorgacopoulou, N., et al. 2001) followed by monosaccharide or disaccharide analysis. Highly sensitive chromatographic (Toyoda, H., Kinoshita-Toyoda, A., et al. 2000, Toyoda, H., Yamamoto, H., et al. 1999), capillary electrophoretic (Karamanos, N.K., Axelsson, S., et al. 1995, Lamari, F.N., Militsopoulou, M., et al. 2002, Mitropoulou, T.N., Lamari, F., et al. 2001), fluorescence-assisted carbohydrate electrophoresis (Calabro, A., Benavides, M., et al. 2000, Calabro, A., Hascall, V.C., et al. 2000) and mass spectrometric (Desaire, H. and Leary, J.A. 2000) methods are available for CS/DS disaccharide analysis. In order to determine uronic acid epimers from disaccharide analysis, a chemical degradation method may be used. The intact GAG chains are de-acetylated by hydrazinolysis and subjected to de-aminative cleavage using nitrous acid at pH 4. The resulting disaccharides, each containing dehydromannose at the reducing end and an unmodified uronic acid residue, may be reduced with sodium borotritide or tagged with a chromophore or fluorophore prior to chromographic or electrophoretic separation. (Guo, Y.C. and Conrad, H.E. 1989).

Measurement of the mass of a CS/DS oligosaccharide determines the composition with respect to the number of uronic acid, N-acetylhexosamine, and sulfate units (Carr, S.A. and Reinhold, V.N. 1984, Sugahara, K., Takemura, Y., et al. 1994, Takagaki, K., Kojima, K., et al. 1992). Tandem MS determines the number of sulfate groups per disaccharide unit and provides a measure of IdoA(α1-3)GalNAc4S(β1-4) content in the target oligosaccharide (Zaia, J., Li, X.Q., et al. 2003, Zaia, J., McClellan, J.E., et al. 2001) based on product ion abundances.

We have used a liquid chromatography-MS platform for determining CS/DS glycoform distribution (Hitchcock, A.M., Costello, C.E., et al. 2006). The present work describes the principles behind this approach and extends the analysis to longer oligosaccharides. A series of dermal DS samples isolated under different conditions are analyzed, in addition to a series of decorins from different tissues. The CS/DS chains are partially de-polymerized using chondroitinase ABC and the resultant oligosaccharides analyzed using tandem mass spectrometry (MS). This approach is used to compare the expression of oligosaccharide glycoforms among the different samples. Such comparisons would be significantly more laborious using traditional techniques. These experiments demonstrate the principle and validity of a new method for glycoform analysis of CS/DS chains, one that will be useful in the emerging field of glycomics.

Results

Partial depolymerization of CS/DS samples

Oligosaccharides were generated from CS/DS samples by stopping chondroitinase ABC digestions at a time point found to produce 232 nm absorbance of 30% of its maximum value (see Materials and Methods for details). The oligosaccharides were fractionated using high performance size exclusion chromatography with 232 nm detection. Oligosaccharide length will be given as degree of polymerization (dp) and the number of monosaccharide units. Peak area percentages for dp2, dp4 and dp6 for each of the samples are given in Table I. The size exclusion chromatograms displayed no defined peaks greater than dp6 and the 232 nm baseline was elevated before the elution of this oligomer. Thus, the reported values include the area of this baseline elevation, listed as “>dp6”.

Table I.

Relative chromatographic peak areas for CS/DS samples partially digested with chondroitinase ABC. Chondroitinase ABC digestions were stopped at 30% completion, and separated using high performance SEC, as described in the Materials and Methods section. The relative peak areas for Δ-unsaturated dp2, dp4 and dp6 oligosaccharides are given.

	dp2	dp4	dp6	>dp6
CSA	48.35	9.16	9.98	32.5
CSB	38.92	18.27	18.53	24.3
CSC	55.30	19.19	9.48	16.0
CS6	58.02	21.38	9.13	11.5
DS18	35.80	14.48	16.59	33.1
DS36	37.93	10.91	11.53	39.6
DS50	43.13	15.24	12.71	28.9
ACD	39.80	15.72	-	44.5
CD	55.16	29.19	-	15.7
SD	42.69	23.03	-	34.3

Mass spectrometry of CS/DS oligosaccharides

The compositions of oligosaccharides with a 4,5-unsaturated uronic acid residue at the non-reducing end will be given preceded with the Δ symbol. Figure 1 is a representative electrospray mass spectrum of size exclusion chromatography fractions corresponding to dp4 (A) and dp6 (B), respectively. The ion at m/z 458 corresponds to [M-2H]^2- for Δdp4 (A) and [M-3H]^3- for Δdp6 (B). Because this ion is isobaric for Δ-unsaturated CS/DS oligosaccharides of composition Δ(HexA)_n(GalNAc)_n(SO₃)_n, careful isolation of the individual oligomers was done using high performance size-exclusion chromatography. The isotope pattern of the ion at m/z 458 (A, inset) shows peaks separated by 0.50 u, that signify a charge state of 2-, the neutral mass of which (918.10 Da, Table II) matches a composition of Δ(HexA)₂(GalNAc)₂(SO₃)₂. The inset in (B) shows peaks separated by 0.33 u, signifying a charge state of 3- , the neutral mass of which (1377.24 Da, Table II) is consistent with a composition of Δ(HexA)₃(GalNAc)₃(SO₃)₃. The absence of an ion at m/z 458.38 from (A) shows that the dp4 fraction is not contaminated with dp6. The absence of an ion at m/z 458.55 from (B) shows that there is no dp4 in the dp6 fraction. These negative electrospray ionization patterns produced from CS/DS oligosaccharides are consistent with the previous reports (Chai, W., Beeson, J.G., et al. 2002, Desaire, H. and Leary, J.A. 2000, Zaia, J. and Costello, C.E. 2001, Zaia, J., Li, X.Q., et al. 2003, Zaia, J., McClellan, J.E., et al. 2001, Zamfir, A., Seidler, D.G., et al. 2003).

Figure 1.

Representative ESI mass spectra of partially depolymerized and purified CS/DS size exclusion chromatography fractions in the negative mode, (A) the spectrum obtained from the dp4 fraction and (B) that from the dp6 fraction.

Table II.

Mass spectral ion assignments for CS/DS Δdp4 and Δdp6 oligosaccharides. The Δ symbol is used to signify oligosaccharides with a 4,5-unsaturated uronic acid at the non-reducing terminus.

	m/z	Ion	Description	Composition	Calcd mass	Obsvd mass
dp4	305.06	[M-3H]3-	Δ(HexA)2(HexNAc)2(SO3)2	C28H41N2O28S2	918.13	918.21
	458.05	[M-2H]2-	Δ(HexA)2(HexNAc)2(SO3)2	C28H40N2O28S2	918.13	918.14
	498.03	[M-2H]2-	Δ(HexA)2(HexNAc)2(SO3)3	C28H40N2O31S3	998.07	998.08
	917.01	[M-H]-	Δ(HexA)2(HexNAc)2(SO3)2	C28H41N2O28S2	918.13	918.02
dp6	343.31	[M-4H]4-	Δ(HexA)3(HexNAc)3(SO3)3	C42H59N3O42S3	1377.20	1377.28
	458.07	[M-3H]3-	Δ(HexA)3(HexNAc)3(SO3)3	C42H60N3O42S3	1377.20	1377.24
	464.08	[M-3H]3-	(HexA)3(HexNAc)3(SO3)3	C42H62N3O43S3	1395.21	1395.27
	484.71	[M-3H]3-	Δ(HexA)3(HexNAc)3(SO3)4	C42H60N3O45S4	1457.16	1457.16
	687.55	[M-2H]2-	Δ(HexA)3(HexNAc)3(SO3)3	C42H61N3O42S3	1377.20	1377.12
	696.51	[M-2H]2-	(HexA)3(HexNAc)3(SO3)3	C42H63N3O43S3	1395.21	1395.27
	727.52	[M-2H]2-	Δ(HexA)3(HexNAc)3(SO3)4	C42H61N3O45S4	1457.16	1457.06
	899.01	[M-H]-	Δ(HexA)2(HexNAc)2(SO3)2-H2O	C28H40N2O27S2	900.12	900.01

Assignments for all ions observed in the mass spectra of CS/DS dp4 and dp6 fractions are listed in Table II. The ions at m/z 498 in Figure 1A corresponds to Δ(HexA)₂(GalNAc)₂(SO₃)₃ and m/z 484 in 1B to Δ(HexA)₃(GalNAc)₃(SO₃)₄, see Table II. These CS/DS oligosaccharides contain more than one sulfate group per disaccharide repeat and are termed oversulfated. In addition, an [M-2H]^2- ion corresponding to a saturated oligosaccharide of composition of (HexA)₃(GalNAc)₃(SO₃)₃ was detected at m/z 696.5 in Figure 1B. This composition matches that expected for an oligosaccharide derived from the non-reducing terminus of the intact CS/DS chain, by virtue of the fact that the neutral mass is consistent with a saturated structure. The ion at m/z 464 in 1B corresponds to [M-3H]^3- for the same composition. The ion at m/z 899 in 1B corresponds to the composition [Δ(HexA)₂(GalNAc)₂(SO₃)₂-H₂O] (Table II). Gentle ionization conditions were used and no losses of SO₃ or other signs of in-source fragmentation were observed in Fig. 1. It is therefore likely that the species giving rise to m/z 899 exists in solution and is not an artifact of the mass spectrometric measurement. Although the oligosaccharide composition may be determined from the molecular weight information obtained from the ESI mass spectra, detailed structural information such as sulfation positions and epimerization state are not produced.

Tandem mass spectrometry of CS/DS Δdp4 glycoforms

To determine positions of sulfation and epimerization, we compared abundances of the fragment ions formed from DS oligosaccharides with those produced from commercial CS/DS standards with known epimerization and GalNAc sulfate positions. It was established previously that tandem mass spectrometric product ion abundances of CS oligosaccharides reflect sulfation position at GalNAc residues (McClellan, J.E., Costello, C.E., et al. 2002, Zaia, J., McClellan, J.E., et al. 2001), and epimerization of HexA residues (Zaia, J., Li, X.Q., et al. 2003). Thus, three commercial standards were used, each of which is approximately 90% pure. These were CSA (GlcA(β1-3)GalNAc4S(β1-4))_n, CSB (IdoA(α1-3)GalNAc4S(β1-4))_n, and CSC (GlcA(β1-3)GalNAc6S(β1-4))_n. Furthermore, to focus the analysis on a single uronic acid that is unaltered by lyase digestion, the Δdp4 series was first investigated.

Figure 2 show the tandem mass spectra of Δdp4 oligosaccharides generated from CS standard preparations, CSA (A), CSB (B), and CSC (C). The doubly charged precursor ion is typically present as the most abundant charge state and product ions resulting from glycosidic bond cleavages are abundant while those from losses of SO₃ are in low abundance. Although the three CS/DS Δdp4 isomers dissociate to form ions with identical m/z values, differences in abundances of the fragment ions characterize each. Specifically, there are six signature ions, the abundances of which differentiate the three isomers. The Δdp4 oligosaccharides derived from CSA are characterized by high abundance of Y₁^1- (m/z 300) and B₃^1- ions (m/z 616) relative to those observed for the other two isomers (Figure 2A). CSB Δdp4 oligosaccharides are characterized by the high abundance of ^0,2X₃^2- (m/z 400) and Y₃^2- ions (m/z 379) relative to the other two isomers. Abundant [M-SO₃]^2- (m/z 418) and C₃^2- ions (m/z 316) characterize CSC Δdp4 oligosaccharides. These fragment ions are summarized in Table III. In low energy CID experiments, product ion abundances reflect the lability of the covalent bonds that are cleaved during the tandem mass spectrometric dissociation. Thus, the sulfation and epimerization positions influence the lability of certain bonds in the oligosaccharide ions that are reflected by the observed ion abundances.

Figure 2.

Tandem mass spectra of Δ-unsaturated dp4 derived from CSA (A), CSB (B), and CSC (C). The structure of Δ-unsaturated dp4 from CSA is shown in (D) with product ion assignments.

Table III.

Tandem mass spectrometric signature ions for CS/DS Δdp4 and Δdp6 oligosaccharides.

	Ion	Composition	Calcd m/z	Obsvd m/z
Δdp4	[M-2H]2-	C28H40N2O28S2	458.06	458.05
	Y11-	C8H14NO9S	300.04	300.06
	B31-	C20H26NO19S	616.08	616.04
	0,2X32-	C24H36O24N2S2	400.05	400.06
	Y32-	C22H35O23N2S2	379.05	379.06
	C32-	C20H27O20NS	316.54	316.57
	[M-SO3-2H]2-	C28H40O25N2S	418.08	418.09
Δdp6	[M-3H]3-	C42H60N3O42S3	458.06	458.05
	Y11-	C8H14NO9S	300.04	300.05
	B52-	C34H46N2O33S2	537.07	537.06
	0,2X53-	C38H56N3O38S3	419.39	419.39
	Y53-	C36H54N3O37S3	405.38	405.39
	C52-	C34H48N2O34S2	546.08	546.03
	[M-SO3-3H]3-	C42H60N3O39S2	431.41	431.41

The Y₁^1- ion contains the sulfate group on the reducing terminal GalNAc residue of the precursor oligosaccharide ion. Abundance of this ion correlates directly with the presence of an oligosaccharide with a CSA-like GlcA(β1-3)GalNAc4S(β1-4) repeat. Similarly, the B₃^1- ion which contains the internal sulfated GalNAc residue, is produced by cleavage of the same glycosidic bond and its abundance correlates with the presence of the same repeat. It is likely that the abundance of both the Y₁^1- and the B₃^1- ions reflect the position of sulfation on the reducing end GlcA(β1-3)GalNAc4S disaccharide. The [M-2H]^2- ion of Δdp4 derived from CSC dissociates to produce abundant [M-SO₃]^2- and C₃^2- ions. The C₃^2- ion contains the internal sulfated GalNAc residue, but the ion abundance of the glycosidic bond cleavage could be influenced by the sulfate position of the reducing GlcA(β1-3)GalNAc6S disaccharide during the collision-induced dissociation (CID) process. Loss of sulfate, [M-SO₃]^2-, is non-specific as to the originating GlcA(β1-3)GalNAc6S disaccharide, but is indicative of CSC-like structures. The ^0,2X₃^2- and Y₃^2- ion abundances are enhanced when CSB-like IdoA(α1-3)GalNAc4S(β1-4) repeats are present.

Standards for Δdp4 and Δdp6 oligosaccharides containing mixed GalNAc 4- and 6-sulfation patterns are not available. However, for Δdp4, the abundances of the CSA signature ions, Y₁^1- and B₃^1-, reflect the position of sulfation of the reducing end GalNAc residue. It is therefore likely that all tetrasaccharides with GalNAc-4-sulfate on the reducing end will produce abundant CSA-like signature ions. CSC-like signature ions correspond to C₃^2- and [M-SO₃]^2-. The abundance of the C₃^2- ion varies according to the position of sulfation of the reducing terminal GalNAc residue. That of the [M-SO₃]^2- ion according to the presence of GalNAc-6-sulfate at either position in the tetramer. Thus, it would be expected that a mixed tetramer with GalNAc-6-sulfate on the reducing end would produce an abundant C₃^2- ion, but that of the [M-SO₃]^2- ion would be intermediate.

Signature ion patterns

The use of signature ions to distinguish the three CS isomeric oligosaccharides can best be seen when their percent total ion abundances (Table IV) are plotted. The Y₁^1- and B₃^1- ions are most abundant, expressed as a percent of the sum of all product ion abundances (Figure 3A), for CSA Δdp4. The Y₃^2- and ^0,2X₃^2- ions are most abundant for Δdp4 derived from CSB. The C₃^2- and [M-SO₃]^2- ions are most abundant for Δdp4 derived from CSC. The data are re-plotted in Figure 3B to show how each CS type produces a characteristic pattern of these diagnostic fragment ions. Thus, a given CS sample produces a pattern that is a linear combination of CSA-like, CSB-like and CSC-like signature ions. These ions will be used to characterize oligosaccharides derived from CS/DS samples of biological interest in a later section.

Table IV.

Percent total ion abundances for signature ions for Δdp4 derived from CSA, CSB and CSC. The data are plotted as a bar graph in Figure 3.

Ion	CSA	CSB	CSC
Y11-	33.91 ± 0.61	10.14 ± 0.56	19.88 ± 0.48
B31-	11.22 ± 0.33	2.92 ± 0.73	7.74 ± 0.52
Y32-	1.63 ± 0.27	3.18 ± 0.15	0.68 ± 0.05
0,2X32-	0.66 ± 0.05	2.41 ± 0.31	0.44 ± 0.04
C32-	0.96 ± 0.03	0.78 ± 0.08	2.06 ± 0.18
[M-SO3]2-	0.89 ± 0.09	0.72 ± 0.64	2.23 ± 0.05

Figure 3.

Signature ion patterns for Δ-unsaturated dp4 derived from standard preparations of CSA, CSB, and CSC. The abundances of selected product ions are compared in (A). The data are re-plotted in (B) to show how the ion abundances vary according to the type of CS from which the Δ-unsaturated dp4 was derived.

Tandem mass spectrometry of CS/DS Δdp6 glycoforms

As with the CS/DS Δdp4, signature ions characterize the Δdp6 oligosaccharides derived from standard CSA, CSB, and CSC. A summary of these ions is shown in Table III. The percent total ion abundances for the Y₁^1- (m/z 300) and B₅^2- ions (m/z 537) are highest for CSA-derived Δdp6 oligosaccharides (Figure 4A). Although these ions are abundant for Δdp6 oligosaccharides derived from CSC, the abundances as a percent of the total is higher for Δdp6 derived from CSA due to the lower abundances of other product ions. The ^0,2X₅^3- (m/z 419) and Y₅^3- (m/z 405) ions have the highest percent total ion abundances for Δdp6 derived from CSB (Figure 4B), while [M-SO₃]^3- (m/z 431) and C₅^2- (m/z 546) ions are observed in greatest abundance for those from CSC (Figure 4C). Figure 5A shows how the abundances of the signature ions vary for different types of CS. The data are re-plotted in Figure 5B to show how the different forms of CS Δdp6 produce characteristic ion patterns.

Figure 4.

Tandem mass spectra of the Δ-unsaturated dp6 derived from standard preparations of CSA (A), CSB (B), and CSC (C). The fragment ion assignments are shown in (D) for Δ-unsaturated dp6 derived from CSA.

Figure 5.

Signature ion patterns for Δ-unsaturated dp6 derived from standard preparations of CSA, CSB, and CSC. The abundances of selected product ions are compared in (A). The data are re-plotted in (B) to show how the ion abundances vary according to the type of CS from which the Δ-unsaturated dp6 was derived.

Analysis of varied DS samples and decorin

To demonstrate the utility of this method, we analyzed several CS/DS samples, the Δdp4 signature ion profiles of which are shown in Figure 6. The CS6 sample was purified from horse nasal septum and several DS preparations, namely DS18, DS36, and DS50 were purified from pig skin. From the signature ion patterns it can easily be confirmed that Δdp4 oligosaccharides derived from DS18, DS36 and DS50 display signature ion patterns similar to those of standard CSB. This indicates that a high percentage of the Δdp4 oligosaccharides in each of these samples contain the IdoA(α1-3)GalNAc4S(β1-4) sequence. Those from CS6, by contrast, display signature ion patterns consistent with a high percentage of GlcA(β1-3)GalNAc6S(β1-4) repeats. Although the signature ion patterns of these samples qualitatively resembles those of the standards, the heights of the bars differ. For instance, samples DS18, DS36, and DS50 show an apparent increase in the height of CSA-like ions, suggesting varying GlcA(β1-3)GalNAc4S(β1-4) content of DS18, DS36 and DS50. Also, the middle bar, which represents CSB-like ions, decreases going from DS18 to DS50, which reflects the decreasing IdoA(α1-3)GalNAc4S(β1-4) content in the DS samples. However, the CSC-like signature ions are observed consistently in low abundances among the DS samples, indicating a minimal content of GlcA(β1-3)GalNAc6S(β1-4)repeats. CS6 Δdp4 oligosaccharides have significantly more abundant CSC-like signature ions than do those derived from the commercial CSC standard, suggesting that they contain a higher percentage of GlcA(β1-3)GalNAc6S(β1-4) repeats.

Figure 6.

Signature ion profile for Δ-unsaturated dp4 oligosaccharides derived from CS and DS samples. CS6 is from horse nasal septum. DS18, DS36, and DS50 are from pig skin.

The signature ion pattern for the decorin samples is shown in Figure 7. All three decorins tested — bovine articular cartilage (ACD), human cervix (CD), and bovine sclera (SD) exhibit a CSB-like signature ion pattern, consistent with a high content of IdoA(α1-3)GalNAc4S(β1-4) in the Δdp4 oligosaccharides generated there from. A decreasing CSA-like signature ion contribution is indicated, proceeding from ACD to CD to SD, which again suggests that the content of GlcA(β1-3)GalNAc4S(β1-4) sequences varies among Δdp4 generated from the three decorins. Moreover, the CSB-like ions increase from ACD to CD to SD, suggesting that Δdp4 generated from SD has the greatest IdoA(α1-3)GalNAc4S(β1-4) content of the three decorins. However, the CSC-like ions have the greatest contribution for ACD compared to almost insignificant contribution on CD and SD, suggesting that ACD contains the greatest GlcA(β1-3)GalNAc6S(β1-4) percentage among the three decorins. For the Δdp6 series, the signature ion profile of DS samples and decorin is similar to that of the Δdp4 series.

Figure 7.

Signature ion profile for Δ-unsaturated dp4 oligosaccharides derived from decorin samples. Articular cartilage decorin (ACD), cervix decorin (CD) and sclera decorin (SD).

Quantitative Analysis of each isomeric contribution

As was pointed out earlier, the signature ion abundances vary directly with the abundances of CSA-, CSB-, and CSC-like disaccharide repeats in each sample. Since the signature ion abundances for a given sample are a linear combination of the three repeat-types, the percentages of CSA-, CSB-and CSC-like ions required to produce the observed patterns can be calculated. Tandem mass spectrometric ion abundances has been used to quantitate CS and heparin disaccharides (Desaire, H. and Leary, J.A. 2000, Saad, O.M. and Leary, J.A. 2003). Using similar principles applied to the Δdp4 oligosaccharide signature ion abundances, the following equations were applied to calculate the percent contribution of CSA, CSB and CSC:

$$\begin{array}{cc}\hfill & {\mathrm{AY}}_1+{\mathrm{BY}}_1+{\mathrm{CY}}_1={\mathrm{DY}}_1\hfill \\ \hfill & {\mathrm{AY}}_3+{\mathrm{BY}}_3+{\mathrm{CY}}_3={\mathrm{DY}}_3\hfill \\ \hfill & \mathrm{A}(\mathrm{M}-{\mathrm{SO}}_3)+\mathrm{B}(\mathrm{M}-{\mathrm{SO}}_3)+\mathrm{C}(\mathrm{M}-{\mathrm{SO}}_3)=\mathrm{D}(\mathrm{M}-{\mathrm{SO}}_3)\hfill \end{array}$$

where Y₁, Y₃ and M-SO₃ are the signature ions for CSA-like, CSB-like, and CSC-like respectively; A, B and C are the unknown contributions of CSA, CSB, and CSC isomers; and D is the contribution from the unknown sample. The percent total ion abundance of each signature ion from the standards is substituted on the left side of the equations and those from the sample on the right side, producing three equations with three unknowns. A summary of the calculated percent CSA, CSB, and CSC for DS18, DS36 and DS50 is shown in Table V. The percent of CSA-like sequences (GlcA(β1-3)GalNAc4S(β1-4)) is observed to increase, going from DS18 to DS36 to DS50, and that for CSB (IdoA(α1-3)GalNAc4S(β1-4)) to decrease. The absence of CSC-like sequences (GlcA(β1-3)GalNAc6S(β1-4)) is observed.

Table V.

Percent compositions for glycoforms of CS/DS Δdp4, calculated from tandem mass spectrometric signature ion abundances. CS6 was used as the CSC-like standard.

	1	2	3
	CSA-like GlcAGalNAc4S	CSB-like IdoAGalNAc4S	CSC-like GlcAGalNAc6S
CSC	41.81 ± 1.72	0 ± 0.33	58.18 ± 0.22
DS18	0.85 ± 1.42	99.15 ± 0.35	0 ± 0.14
DS36	11.21 ± 1.31	88.79 ± 0.41	0 ± 0.13
DS50	25.66 ± 1.22	74.35 ± 0.35	0 ± 0.15
ACD	50.94 ± 1.35	31.13 ± 0.23	17.92 ± 0.41
CD	35.92 ± 1.22	64.08 ± 0.28	0 ± 0.47
SD	21.56 ± 1.73	74.51 ± 0.47	3.92 ± 0.36

While the level of CSC-like sequences for the CSC Δdp4 oligosaccharides in Table V is 58.18 %, that of the level of ΔHexAGalNAc6S disaccharides in Table VI is 78.88%. The mass spectral method compares ion signatures for a given sample against those of the standard CS preparations. The method thus serves as a means of comparing a series of samples, but does not provide absolute quantification of the levels CS repeats. The CSC sample contains significantly lower content of Δ HexAGalNAc6S (78.88%) than CS6 (93.97%). CS6 is therefore used as a mass spectral standard against which CSC is compared. It is important to emphasize that the trend is the same for the two sets of data. Measurement of relative changes of glycoform expression between a given sample and the CS standards is the most appropriate use of the mass spectral method.

Table VI.

CE-LIF percentage compositions of ΔHexAGalNAc4S, ΔHexAGalNAc6S, ΔHexA2SGalNAc4S and ΔHexA2SGalNAc6S after an exhaustive digestion of the whole DS polymer chain. Percentage Δ-disaccharide compositions were calculated using CE fluorescence peak areas normalized to the area of the internal standard.

	1	2	3	4
	ΔHexAGalNAc4S	ΔHexAGalNAc6S	ΔHexA2SGalNAc4S	ΔHexA2SGalNAc6S
CSA	94.65 ± 0.29	5.34 ± 0.02	0	0
CSB	91.10 ± 0.39	7.10 ± 0.03	1.79 ± 0.04	0
CSC	17.45 ± 0.17	78.88 ± 0.66	0	3.66 ± 0.02
CS6	6.03 ± 0.07	93.97 ± 0.7	0	0
DS18	95.51 ± 1.00	2.65 ± 0.13	1.84 ± 0.02	0
DS36	94.39 ± 1.21	1.98 ± 0.09	3.69 ± 0.02	0
DS50	84.16 ± 2.03	13.69 ± 0.27	2.14 ± 0.03	0
ACD	60.77 ± 0.25	39.23 ± 0.16	0	0
CD	86.75 ± 0.08	9.64 ± 0.01	3.61 ± 0.01	0
SD	79.39 ± 0.01	15.59 ± 0.01	5.01 ± 0.06	0

Disaccharide Analysis

Disaccharide analysis of CS/DS standards and unknown samples was achieved by exhaustive digestion using chondroitinase ABC, subsequent fluorescent derivatization using 2-aminoacridone (AMAC) and capillary electrophoresis with laser-induced fluorescence (CE-LIF) detection to analyze the disaccharides (Militsopoulou, M., Lamari, F.N., et al. 2002). Migration times of Δ-disaccharides generated from standard CS preparations CSA, CSB, CSC and those from DS samples and decorin were verified against those of AMAC-derivatized commercial disaccharide standards. An AMAC-derivatized trisulfated disaccharide derived from heparin (HS-IS), was used as an internal standard for each run, as it has a unique migration time compared to the CS/DS samples analyzed. Table VI and Table VII list the Δ–disaccharide composition of the whole polymer chain and the dp6 fraction, respectively. Percentage Δ-disaccharide compositions were calculated using the CE fluorescence peak areas, normalized to the area of the internal standard.

Table VII.

CE-LIF percent compositions of CS/DS dp6 fraction after an exhaustive digestion using chondroitinase ABC. Percentage Δ-disaccharide compositions were calculated using CE fluorescence peak areas normalized to the area of the internal standard.

	1	2	3	4
	ΔHexAGalNAc4S	ΔHexAGalNAc6S	ΔHexA2SGalNAc4S	ΔHexA2SGalNAc6S
CSA	93.79 ± 0.11	6.20 ± 0.00	0	0
CSB	93.52 ± 0.19	5.43 ± 0.01	1.05 ± 0.01	0
CSC	20.07 ± 0.21	76.42 ± 0.87	0	3.51 ± 0.03
CS6	10.64 ± 0.01	89.36 ± 0.07	0	0
DS18	95.95 ± 0.76	0.85 ± 0.02	3.21 ± 0.01	0
DS36	93.07 ± 0.24	3.02 ± 0.03	3.90 ± 0.02	0
DS50	91.67 ± 0.17	6.84 ± 0.04	1.49 ± 0.03	0

Partial chondroitinase ABC digestion was used to produce oligosaccharide mixtures for subsequent mass spectral analysis. Although the distribution of oligosaccharides was dependent to some extent on the sample, the dp4 and dp6 fractions are representative of the glycoform distribution of the entire CS/DS chain. This can be seen from the good agreement in disaccharide analysis values obtained from isolated dp6 fractions versus those of the entire CS/DS chain from which they were derived (Table VI and Table VII).

For the GlcA-containing standards, CSA is shown to contain 94.65% ΔHexAGalNAc4S and 5.34% ΔHexA-GalNAc6S while CSC is a mixture of 17.45% ΔHexAGalNAc4S, 78.88% ΔHexAGalNAc6S and 3.66% of the 2-sulfated disaccharide, ΔHexA2SGalNAc6S (Table VI). The CS6 sample contains 93.97% ΔHexAGalNAc6S, significantly greater than the value observed for the commercial CSC, a trend that was observed in the signature ion plots for the Δdp4 and Δdp6 oligosaccharides (Fig. 6). Thus, the higher abundance of GlcA(β1-3)GalNAc6S(β1-4) in CS6 compared to that in the CSC standard from the MS ion signature data are supported by the CE-based Δ-disaccharide analyses of the abundances of the ΔHexAGalNAc6S disaccharides. Likewise, lower abundances of GlcA(β1-3)GalNAc4S(β1-4) signature ions in CS6 relative to commercial CSC is reflected by the lower percentage of ΔHexAGalNAc4S in the CE-LIF data

Discussion

The use of MS ion signatures to quantify CSA-like, CSB-like and CSC-like repeats in CS/DS oligosaccharides allows comparison of sulfation pattern and epimerization state among samples derived from different tissue contexts or purification schemes. Because the mass spectral ion signatures of each is measured against the same standard CS/DS preparations, it is possible to compare glycoform distributions among a series of GAG or PG samples derived from different environments. The method entails digestion with a single non-specific enzyme, a single stage of chromatography followed by mass spectral analysis. We have also used this approach in an LC-tandem MS platform (Hitchcock, A.M., Costello, C.E., et al. 2006). This platform is limited to analysis of tetrasaccharides, and the present work extends the analysis to hexasaccharides.

Decorins with GAG chains containing high DS content have been shown to be expressed in articular cartilage (Cheng, F., Heinegard, D., et al. 1994, Rosenberg, L.C., Choi, H.U., et al. 1985, Sampaio, L.d.O., Bayliss, M.T., et al. 1988). The activity of D-glucuronyl C5-epimerase is high in fibroblasts, intermediate in articular cartilage, and low in nasal cartilage (Tiedemann, K., Larsson, T., et al. 2001); the results of this work (Table V), showing that levels of CSB-like repeats are generally higher for the dermal DS than for articular cartilage decorins, are consistent with this trend in enzyme activities. The results for all decorin samples tested indicate significant abundances of CSB-like (IdoA(α1-3)GalNAc4S(β1-4)) repeats, from 31.1 to 74.5%, Table V.

Dermal DS samples show a consistent decrease in total CSB-like repeats, going from DS18 to DS50. DS chains are purified from skin, and the associated numeral refers to the percent ethanol at which the chains precipitated (Davidson, E., Hoffman, P., et al. 1956, Fransson, L.A. and Roden, L. 1967). Previous reports have indicated that DS samples contain high levels of IdoA-GalNAc4S repeats (Fransson, L.A. and Roden, L. 1967, Hoffman, P., Linker, A., et al. 1957), with lower levels of IdoA2S-GalNAc4S (Coster, L., Malmstrom, A., et al. 1975) and GlcA(β1-3)GalNAc6S(β1-4) repeats (Fransson, L.A. 1968). The disaccharide analysis of the present DS samples shows that ΔHexAGalNAc4S comprises 84.1-95.5% of the disaccharides formed by chondroitinase ABC digestion, but that a significant percentage of ΔHexAGalNAc6S is also detected (Table VI). The MS ion signatures indicate 74.3-99.1% of the oligosaccharides have CSB-like IdoA(α1-3)-GalNAc4S(β1-4) character (Table V, column 2) and that CSC-like GlcA(β1-3)GalNAc6S(β1-4) repeats are not detected.

It must be emphasized that the mass spectrometric measurements determine the degree of similarity of oligosaccharides from an unknown sample with those from standard CS/DS preparations. Thus, each unknown sample is compared against the same standard compounds and, by extrapolation, to other unknowns. Because only CSA, CSB and CSC were used as standards, it remains possible that low levels of other glycoforms present in the samples remain undetected. For instance, low levels of IdoA(α1-3)GalNAc6S(β1-4) repeats, known to be present in mammalian tissue (Bao, X., Pavao, M.S., et al. 2005), are possibly detected as CSB-like ions. Although it has 6-sulfation, MS detects the signature ion for epimerization more than the sulfation position. At this juncture, this proposal remains to be proven until IdoA(α1-3)GalNAc6S(β1-4) standard becomes available. Moreover, oligosaccharides containing both 6- and 4-sulfation on the same chain would also need an appropriate standard to accurately identify and quantify them.

Glycans released from biological sources are typically complex mixtures of glycoforms. Because the purification is difficult and time consuming, methods are needed to characterize mixtures directly. By determining the compositions of glycoforms, it is possible to profile expressed glycans as a function of changes to the biological context. The present work applies tandem mass spectrometric analysis to CS/DS glycoforms. Using this method, product ion profiles obtained from unknown CS/DS samples, consisting of mixtures of glycoforms, are compared against those obtained from purified standard preparations. Thus, the degree of similarity of the CS/DS oligosaccharides to the standards are quantified. In this manner, information on sulfation position and uronic acid epimerization are produced in a single measurement. In future work, these concepts will be developed into an LC/MS platform for glycomics profiling so as to streamline the workflow. Thus, partially depolymerized CS/DS oligosaccharides will be separated using high performance size exclusion chromatography with on-line tandem mass spectrometric detection. In principle, the concept demonstrated for CS/DS oligosaccharides should be applicable to any class of glycans. The nature of the separation system will change depending on the chemistry of the glycan class in question, and the structural detail produced will depend on the nature of the standards against which unknown glycoconjugate samples are compared.

Materials and Methods

CS type A (GlcA, GalNAc-4-sulfate), CSB (IdoA, GalNAc-4-sulfate) CSC (GlcA, GalNAc-6-sulfate), and chondroitinase ABC were obtained from Seikagaku America/Associates of Cape Cod (Falmouth, MA). Purified CS Δ-disaccharide standards were purchased from V-labs. Bovine articular cartilage decorin was purchased from Sigma Chemical (St. Louis, MO). DS samples were prepared by precipitation from calcium acetate/acetic acid solution with ethanol. 2-aminoacridone (AMAC) was purchased from Fluka Chemica (Milwaukee, WI), sodium cyanoborohydride was from Aldrich Chemical Co. (Milwaukee, WI) and cellulose spin columns were purchased from Harvard Apparatus (Holliston, MA).

Partial digestion of chondroitin and dermatan sulfate

The method for digestion was described previously (Zaia, J., Li, X.Q., et al. 2003). Briefly, 100 ug of CS/DS was mixed with 10 uL water, 2 uL 1M tris-HCl buffer (pH 7.4), 1 uL 1M NH₄OAc and 4 mU of chondroitinase ABC and was digested at 37°C. After 8 minutes (30% digestion) the digestion was terminated by boiling for 2 minutes. Oligosaccharides were fractionated using a Superdex Peptide 3.2/30 column (Amersham Biosciences), which was equilibrated in 10% aceteonitrile, 0.1M ammonium acetate at 100 uL/min, with detection at 232 nm.

Electrospray mass spectrometry

Mass spectra were acquired in the negative mode using an Applied Biosystems/MDS-Sciex API QSTAR™ Pulsar i quadrupole orthogonal time-of-flight mass spectrometer fitted with a nanospray ion source. Samples were dissolved in 10% isopropanol and diluted with 10% isopropanol and 0.1% NH₄OH to achieve a 1- 5 pmol/uL solution. Aliquots of 5 uL were sprayed into the mass spectrometer using 1 μm orifice nanospray tips pulled in-house using a capillary puller (Sutter Instrument P 80/PC micropipette puller, San Rafael, CA). An ionization potential of −1150 V produced a steady ion signal, and all spectra were calibrated externally. The collision energies were set so that the precursor ion remained the most abundant. For CID these were best obtained at collision energy of −18V and −16V for the Δdp4 and Δdp6 series, respectively.

Exhaustive digestion of CS/DS

To 100 ug of CS/DS was mixed 10 uL water, 2 uL 1M tris-HCl buffer (pH 7.4), 1 uL 1M NH₄OAc and 20 mU of chondroitinase ABC and digestion was allowed to proceed at 37°C. After 12 hours, an additional 20 mU of chondroitinase ABC was added. The mixture was allowed to digest for a total of 24 hours after which the reaction was stopped by boiling the mixture for 2 minutes. The disaccharide collected after size exclusion chromatography was derivatized with AMAC.

AMAC derivatization

Derivatization with AMAC was performed following the procedure described by Militsopoulou (Militsopoulou, M., Lamari, F.N., et al. 2002). To the lyophilized disaccharide CS/DS (at least 100 pmol) was added 5 uL 0.1M AMAC solution in acetic acid:DMSO (3:17, v/v) and 5 uL of a freshly prepared 1M NaBH₃CN solution in water. The mixture was vortexed for 3 minutes and was incubated at 45°C for 4 hours after which 50 uL of DMSO was added. Excess reagent was removed using cellulose spin columns.

CE Analysis

The AMAC derivatized DS samples were analyzed using a Beckman Coulter (Fullerton, CA) P/ACE MDQ capillary electrophoresis instrument. The uncoated fused silica capillary tube (75 μm ID, 60 cm total length) was regenerated with 0.1M HCl, 0.1M NaOH, and HPLC grade water before each run and analyses were performed using 50mM NaH₂PO₄ buffer, pH 3.5, at 30 kV at reverse polarity and detection using the AMAC chromophore at 254 nm. Distinct separation of the CS/DS isomers were obtained in the reverse polarity and reproducibility was also obtained when fresh buffer solution was used for each run. A trisulfated heparin (HSIS) was used as an internal standard for each run as it has a unique migration time compared to the CS/DS samples.

Abbreviations

AMAC

2-aminoacridone

CE

capillary electrophoresis

CID

Collision-induced dissociation

CS

chondroitin sulfate

CSA

chondroitin sulfate type A

CSB

chondroitin sulfate type B

CSC

chondroitin sulfate type C

Δ

4,5-unsaturated uronic acid

DS

dermatan sulfate

dp

degree of polymerization

FGF

fibroblastic growth factor

GAG

glycosaminoglycan

LIF

laser-induced fluorescence

MS

mass spectrometry

PG

proteoglycan