Introduction

Congenital disorders of glycosylation (CDG) are a large family of genetic diseases resulting from defects in the synthesis of glycans and in the attachment of glycans to lipids and proteins. These disorders cause a wide range of clinically relevant human diseases with different, sometimes diffuse, clinical pictures. There are currently 12 defined N-, seven O- and three N- and O-glycosylation defects due to defects in proteins with glycosyltransferase or transport function already identified [1]. For a growing group of patients with unidentified defects, but typical and atypical clinical presentations in CDG, there is a high need for an increase in the speed and accuracy of diagnostics in order to design rational therapeutic suggestions in as short a time span as possible.

The alteration of glycan patterns related to CDG defects is reflected in the patients’ serum, which can be analyzed for glycosylation patterns of glycoproteins, according to proteomics-based protocols [2]. However, urine as a rich source of metabolic structures renders an excellent opportunity for noninvasive sample diagnostics. Carbohydrate patterns of CDG patients’ urine were investigated by us using distinct glycomics-based strategies to explore the feasibility of protocols and their potential for unique structure discovery and clinical applications [3–6]. Components of these highly complex mixtures depict different levels of modifications by sialylation and fucosylation. A high number of intact and truncated N- and O-glycan structures were revealed by high resolution mass spectrometry and correlated to the structures deposited in the databases [4]. Using a novel approach of ion mobility mass spectrometry (IMS-MS), overlapping of the carbohydrate mass range landscape was shown to be highly reduced, allowing development of strategies for more efficient identification by sequencing the glycan precursor ions, following their separation by their mobility, or drift time, through the traveling wave IMS cell [3]. The fragmentation data obtained clearly demonstrate that carbohydrate epitopes can be de novo identified by this type of bioanalysis rendering an efficient novel tool for glycomics to be introduced to the CDG research and be applied to clinical samples. In the following investigation, monosialylated glycoforms from urine and their terminal epitopes were identified by IMS and automatic data assignment to be correlated with those described in carbohydrate databases for comparison of their possible function.

Materials and Methods

Materials and Sample Preparation

Materials

Methanol was obtained from Merck (Darmstadt, Germany) and used without further purification. Graphitized carbon powder for the graphitized carbon cartridge preparation was collected by dismantling commercially available Active Charcoal MicroTip Columns 25-100 µl (Harvard Apparatus Inc). AG50 (H+) resin was purchased from Bio-Rad (Richmond, CA, USA). Distilled and deionized water (Mili-Q water systems Millipore, Bedford, MA, USA) was used for the preparation of the sample solutions.

CDG urine sample

The sample investigated in this study was a native glycoconjugate mixture from the urine of patient K.L. suffering from symptoms assigned to the congenital disorder of glycosylation (CDG). For isolation of components, the patient’s urine was filtered and submitted to a first gel filtration chromatography step on Biogel P2 followed by several chromatography stages as described previously [4, 5]. The fraction M3 from the patient K.L. was used in the present study.

Desalting of glycoconjugate mixtures

Prior to analysis by mass spectrometry, the fraction M3 containing a mixture of glycoconjugates was purified by homemade graphitized carbon cartridges as described previously [7] and desalted by AG50 (H+) resin.

Mass spectrometry

All measurements were performed on a Synapt high-definition mass spectrometry (HDMS) system (Waters, Manchester, UK). The Synapt HDMS is a hybrid quadrupole-triwave IMS-oa-TOF device. Gas-phase ions were generated by nano-electrospray ionization in the negative ion mode, passed through a quadrupole into the triwave cell where the ions were separated according to their mobility, and finally transferred to the oa-TOF mass analyzer [3]. Ion source conditions were optimized to minimize in-source fragmentation. DriftScope™ Mobility Environment software was used for ion mobility data visualization and manipulation. Instrumental control and mass spectra acquisition were operated by Waters MassLynx 4.1 software.

Collision-induced dissociation (CID) structural analysis of selected ionic species was performed by transfer T-Wave fragmentation [3]. The fragmentation pattern of sequenced glycoconjugates was assigned according to nomenclature introduced by Domon and Costello [8].

Data interpretation

The computer algorithm for mass spectrometry data interpretation was developed by using Borland Delphi 7 for Windows (Borland) [3]. All computational assignments have been performed within the mass accuracy range of 0–25 ppm.

Glycan database

Consortium for Functional Glycomics (http://www.functionalglycomics.org, 2008) and Glycosuite DB (https://tmat.proteomesystems.com/glycosuite, 2004) glycan databases has been used for correlation with proposed compositions by searching of structural information, such as the type and the number of monosaccharide building blocks. The list of previously described oligosaccharide structures related to the imported composition was generated as the output, providing information about the sample source, methods of structural elucidation, and the corresponding references.

Results

IMS-MS Glycoscreening of the KLM3 CDG Sample

Mapping and sequencing of monosialylated free and amino acid-linked glycans in the urine of the CDG patient K.L. has been probed in the fraction KLM3 to identify the type of glycans involved, their branching patterns, and sialylation attachment. Analyzed in the negative ion MS mode using a nanoESI oa-TOF mass spectrometer more than 100 singly, doubly, and triply charged ions has been observed (Fig. 1). Three ubiquitous singly charged ion species at m/z 632.21, 673.22, and 876.30 assigned previously to NeuAcHex2, NeuAcHexHexNAc, and NeuAcHexHexNAc2, respectively, were present at high abundance 40–100%, but the majority of the ionic species, with highly overlapping isotopic envelopes, were at or below 15%, substantially hindering the tandem MS identification by mapping and sequencing. As already reported, the introduction of IMS as an additional dimension prior to mass analysis allows effective separation of molecular ions in the gas phase according to their charge, shape, and size [3]. Briefly, the ions are separated by their mobility through the dense gas-filled traveling wave and, subsequently, mass analyzed by the oa-TOF. Plotting the ions drift time versus their m/z values, it can be observed that the signal is distributed into three distinct nested areas, A, B, and C, based upon their m/z and drift time relationship. These nested sets correspond to the singly, doubly, and triply charge ions, respectively (Fig. 1, inset). Selection of these regions and maintaining their m/z and drift times produce mass spectra containing predominantly singly, doubly, and triply charged ion populations. These ions previously overlapped in the TOF MS acquisition are now separated and localized into the regions A, B, and C (Fig. 2). The ions with the signal-to-noise ratio significant for the determination of isotopic distribution are selected and submitted for automated compositional analysis calculation.

Fig. 1
figure 1

Negative ion mode nanoESI oa-TOF MS of the fraction M3 obtained from urine of the patient KL after gel permeation chromatography. Spectrum was acquired from the TIC chromatogram over all mobility drift times. Inset: plot of the drift time vs. m/z values for the negative ion mode nanoESI oa-IMS-TOF MS. Ions have been distributed onto areas of predominantly singly (A), doubly (B), and triply (C) charged ionic species (selected by dashed lines). XIC chromatograms A, B, and C were generated from the corresponding area

Full size image
Fig. 2
figure 2

Negative ion mode nanoESI oa-IMS-TOF MS of the sample KLM3 acquired over: a XIC chromatogram A, b XIC chromatogram B, and c XIC chromatogram C (Fig. 1, Inset)

Full size image

The singly charged ion population obtained by selecting region A (Fig. 1a) exposed the presence of at least 62 distinct ionic species, mostly below 15% abundance. Close inspection of the mass spectrum acquired upon IMS separation revealed a significant improvement in the signal-to-noise ratio, as illustrated for the mass range 1500–2000 Da. The data that acquired overall mobility drift times show only three ions at m/z 1565.57, 1727.60, and 1862.88 (Fig. 1), where in the same m/z range of the singly charged spectrum acquired from region A, six additional species at m/z 1548.70, 1768.64, 1846.84, 1878.82, 1884.87, and 1930.65, respectively, can be easily detected (Fig. 2a). Consideration of the established N-glycan biosynthetic pathways allows the composition of ions at m/z 1565.70, 1727.60, 1768.64, and 1930.65 to be assigned as [NeuAcHex4HexNAc3-H]−, [NeuAcHex5HexNAc3-H]−, [NeuAcHex4HexNAc4-H]−, and [NeuAcHex5HexNAc4-H]−, respectively and proposed to represent free N-glycans without Asn attachment at the reducing end. Computational analysis of all singly charged species in the region A ion population allowed the compositions of ~67% of the selected ions to be accomplished. The majority of the species were proposed to represent free oligosaccharides, except 11 ionic species identified by calculation as O-glycans linked either to Ser, Thr, ThrPro, or SerThr (Table 1).

Table 1 Computer-assisted assignment of the major peaks detected in KLM3 sample by the negative ion nanoESI IMS oa-TOF MS
Full size table

All doubly charged ions obtained by averaging over region B were low abundant at the level of relative intensity below ~12% (Fig. 2b). According to the compositional analysis, 15 species out of 58 were assigned to free oligosaccharides (Table 1). Three doubly charged ions at m/z 758.25, 839.29, and 1094.86 were proposed to represent N-glycans linked to Asn and assigned as [NeuAcHex3HexNAc3Asn-2H]2−, [NeuAcHex4HexNAc3Asn-2H]2−, and [NeuAcdHexHex5HexNAc4Asn-2H]2−, respectively. Four doubly charged O-glycan species were proposed according to the compositional analysis to be linked at their respective reducing termini to Thr, Ser, Ser-Thr, or Thr-Pro. These are the ionic species detected at m/z 436.16, 562.19, 575.69, and 974.88 assigned to [dHexHexNAc3Thr-2H]2−, [NeuAcHex2HexNAc2Ser-2H]2−, [NeuAc2HexHexNAcSerThr-2H]2−, and [NeuAcdHexHex3HexNAc4ThrPro-2H]2−, respectively. In the doubly charged pool, the ions detected in the abundance range of 8–12% at m/z 469.15, 624.26, 930.89, and 1139.94 could not be assigned to any glycoconjugate structures; but, according to their CID MS/MS data, they could be assigned as collagen peptides containing a common Gly-Leu/Ile-Pro repeat (data not shown).

The data from region C produced 18 triply charged species with distinct isotopic distribution that were utilized for compositional analysis (Fig. 2c) and showed only six ionic species representing glycoconjugates (Table 1). Ionic species at m/z 675.57 and 797.27 were assigned to [NeuAcdHexHex4HexNAc4Asn-3H]3− and [NeuAcdHexHex5HexNAc5Asn-3H]3−, respectively. On the other hand, triply charged ions at m/z 791.94, 845.96, 913.66, and 967.67 could be simultaneously assigned to free oligosaccharides such as [NeuAcHex4HexNAc7–3H]3−, [NeuAcHex5HexNAc7–3H]3−, [NeuAcHex5HexNAc8–3H]3−, [NeuAcHex6HexNAc8–3H]3−, or Asn-linked N-glycans, such as [NeuAcdHex2Hex4HexNAc5Asn-3H]3−, [NeuAcdHex2Hex5HexNAc5Asn-3H]3−, [NeuAcdHex2Hex5HexNAc6Asn-3H]3−, and [NeuAcdHex2Hex6HexNAc6Asn-3H]3−, respectively (Table 1). The most intensive triply charged ions detected at m/z 759.66 below 2% abundance were assigned to a collagen peptide with Gly-Leu/Ile-Pro repeat (data not shown).

Thus, in total 138 ionic species with distinct isotopic distribution have been detected by IMS oa-TOF MS approach, and 47 singly and multiply charged species that previously overlapped became well separated using this approach.

IMS CID MS/MS Analysis

For the investigation of glycan structure by tandem MS analysis, the overlapping of precursor ions represents a significant obstacle since a mixture of fragment ions originating from ions at the same or similar m/z values does not deliver useful data for assignment. In the case of overlapping ion envelopes of the singly charged species at m/z 1200.42 and the doubly charged at m/z 1199.93 assigned to [NeuAcHex3HexNAc2-H]− and [NeuAcdHexHex7HexNAc4–2H]2−, respectively, the isotopic pattern of the doubly charged ions is obscured by that of the singly charged one (Fig. 3a), obstructing both the determination of the monoisotopic m/z value and selection of the appropriate ion selection for fragmentation. Upon separation of the overlapping precursor ions by ion mobility, CID fragmentation using a transfer traveling wave gas cell, which is located after the IMS gas cell, generates distinct sets of fragments, as illustrated for ions at m/z 1200.42 and 1199.93 (Fig. 3b and c). The CID fragmentation of these IMS separated precursor ions generates series of fragment ions maintaining the same mobility as their parent ions (Fig. 3d). These distinct drift time areas can be used to generate individual chromatograms and, subsequently, mass spectra (A and B, respectively), as shown in Fig. 3e.

Fig. 3
figure 3

a Expansion of the mass range at m/z 1196–1208 of the negative ion mode nanoESI oa-TOF MS of the fraction KLM3, acquired from TIC chromatogram over all mobility drift times; b expansion of the mass range at m/z 1196–1208 of the spectrum acquired from XIC chromatogram of singly charged distributed area after IM separation; c Expansion of the mass range at m/z 1196–1208 of the spectrum acquired from XIC chromatogram of doubly charged distributed area after IM separation; d Plot of the drift time vs. m/z values for fragment ions obtained by CID of overlapped precursor ions at m/z 1199.93 (left) and 1200.42 (right); e Total ion current chromatogram with retained drift time of overlapped precursor ions at m/z 1199.93 (left) and 1200.42 (right). Selected areas indicate extracted ion current chromatogram A for the precursor ions at m/z 1200.42 and chromatogram B for the precursor ions at m/z 1199.93

Full size image

The fragmentation spectrum representing the glycan sequence of the singly charged precursor ions at m/z 1200.42 assigned to [NeuAcHex3HexNAc2-H]− obtained from drift time region A (Fig. 3e) is presented in Fig. 4. According to the evidence of the 0,2A6 cross-ring fragment ions at m/z 1099.36, the free HexNAc residue at the reducing end can be proposed [5]. The observed loss of Δm = 131u and 161u from the parent ions allows assignment of m/z 1069.42 and m/z 1039.35 as 2,4A6 and 0,3A6 cross-ring fragment ions, respectively, and proposes further extension of the reducing end only at the “C4” position of the HexNAc. The loss of Δm = 60u, detected at m/z 937.32 from the C5 fragment at m/z 997.32, indicates that the next structural element elongating the reducing HexNAc is the Hex and can be assigned to 0,2A5 fragment ions. Further inspection of the spectrum revealed the presence of an ion series, indicating the existence of a NeuAcLacNAc motif in the structure. Ions at m/z 290.09, 452.14, 655.22, and 673.23 compositionally proposed as NeuAc(-H2O), NeuAcHex, NeuAcHexHexNAc(-H2O), and NeuAcHexHexNAc could be assigned to B1, B2, B3, and C3 glycosidic fragments. Taking into account the presence of Z2, Y2, B4, and C4 at m/z 364.13, 382.14, 817.27, and 835.28, respectively, as well as the rules of glycan assembly, the potential structural candidate was proposed as a truncated sialylated N-glycan NeuAc(α2–6)Gal(β1–4)GlcNAc(β1–2)Man(α1–3/6)Man(β1–4)GlcNAc with only one antenna, where the α2–6 linkage between NeuAc and Gal was proposed according to the diagnostic 0,4A2-CO2 cross-ring fragment being present at m/z 306.12 [9]. Searching the glycan database (http://www.functionalglycomics.org, 2008; https://tmat.proteomesystems.com/glycosuite, 2004) allowed correlation of the proposed structure with the one previously found in the urine of a patient suffering from infantile sialidosis, type II (dysmorphic) [10, 11]. On the other hand, a number of ions which cannot originate from the proposed structure, such as NeuAcHexHexNAc2 at m/z 876.35 or NeuAcHex2 at m/z 632.21, were also found in the spectrum, indicating the existence of an additional structural isomer with the same composition.

Fig. 4
figure 4

Fragmentation pattern of the singly charged precursor ions at m/z 1200.42. The spectrum was obtained by averaging over XIC chromatogram A (Fig. 3e) after IMS separation. Inset: proposed structure—truncated sialylated biantennary N-glycan

Full size image

The fragmentation pattern of the doubly charged precursor at m/z 1199.93 separated by IMS and obtained by selecting the data shown in region B (Fig. 3e) can be assigned to [NeuAcdHexHex7HexNAc4–2H]2−(Fig. 5a). Inspection of the MS/MS spectrum revealed the presence of fragment ions corresponding to NeuAc(-H2O) at m/z 290.10, NeuAcHex at m/z 470.17, and NeuAcHexHexNAc at m/z 673.26. Besides the NeuAcLacNAc antennae terminal, a NeuAcLea/x antennae structure was present according to the presence of the fragment ion at m/z 819.35. Thus, taking into account the biosynthetic rules of N-glycan antennae assembly as well as the evidence of the existence of truncated glycans in the urine, the structure of this species has been proposed, as summarized in Fig. 5b.

Fig. 5
figure 5

a Negative ion mode nanoESI IMS oa-TOF MS/MS of the doubly charged precursor ions at m/z 1199.93. Fragmentation spectrum obtained by averaging over XIC chromatogram B (Fig. 3e) after IM separation. b Structural candidates proposed according to the rule of glycan biosynthesis and by the presence of diagnostic fragment ions

Full size image

Ions at m/z 809.30 assigned to Hex5(-H2O) and m/z 1038.43 assigned to NeuAcHex2HexNAc2 were considered as “D” type and \({\text{C}}_{4\alpha } \) fragments, respectively, originating from the potential hybrid-type N-glycan isomer (Structure B, Fig. 5b). The double loss of a HexNAc unit from the parent ions detected at m/z 1994.72 indicates the absence of the core Fuc at the reducing end, and together with ions, indicating the presence of NeuAcLea/x antennae can support structural isomer D (Fig. 5b). Ions at m/z 1143.54 assigned to NeuAcdHexHex2HexNAc can be considered as “D+H2O” type fragment of the structural isomer, where the NeuAcLea/x antennae is distributed at Man6 of the N-glycan core region (Structure C, Fig. 5b). Among the proposed structures depicted, only the structure A (Fig. 5b), supported by the presence of \({\text{C}}_{4\alpha } \) at m/z 1200.46 has been found in the glycan database with a nonspecified source of origin (http://www.functionalglycomics.org, 2008).

The doubly charged precursor ions at m/z 1055.85 assigned to [NeuAcdHexHex5HexNAc4-H+Cl]2− represent another example of sialylated and fucosylated oligosaccharide found in the urine of CDG patient. To obtain a distinct series of fragment ions related only to the doubly charged precursors, IMS T-Wave transfer fragmentation analysis has been performed, and the spectrum acquired over the drift time range from 4.0 to 7.5 ms of the TIC chromatogram (Fig. 6a, inset). According to biosynthetic rules of glycan assembly and taking into account the ratio of monosaccharide units in the proposed composition, potential structural candidates could be proposed as monosialylated N-glycans with one peripheral or one core fucosylation. The presence of a NeuAcLacNAc antenna has been monitored by the diagnostic fragment ions assigned to NeuAcHexHexNAc detected at m/z 673.27. Fragment ions detected at m/z 819.29 and assigned to NeuAcHexHexNAc indicate a potential NeuAcLea/x antennae structure, supporting the assumption of peripheral fucosylation. Thus, several structural isomers or isobars could be related to the precursor ions (Fig. 6b). The presence of diagnostic C4 and “D” type fragment ions [12] allows to determine the distribution of antennae at the pentasaccharide core of potential structural candidates. Fragment ions at m/z 688.23 and 835.28 assigned to Hex2HexNAc and NeuAcHex2HexNAc has been considered as “D” and \({\text{C}}_{4\alpha } \), respectively, of the monosialylated biantennary N-glycan with core Fuc, where NeuAcLacNAc antenna is attached to Man3 of the core region (Structure A, Fig. 6b). The attachment of NeuAcLacNAc antenna at Man6 of the core region has been specified by the presence of “D+H2O” and \({\text{C}}_{4\beta } \) fragment ions at m/z 997.34 and 544.19, respectively (Structure B, Fig. 6b). On the other hand, fragment ions at m/z 544.19 can also be originated as \({\text{C}}_{4\beta } \) from the Structure C (Fig. 6b), supported by “D” and “D+H2O” at m/z 1125.39 and 1143.39, respectively. The fourth structural candidate, isobaric with previous ones, could be proposed and assigned according to glycan biosynthesis rules to a hybrid type monosialylated N-glycan (Structure D, Fig. 6b). \({\text{C}}_{4\alpha } \) fragment ions detected at m/z 1184.42 and assigned to NeuAcdHexHex2Hex2 support this assumption. A correlation of the Structure A with that previously discovered in the urine of a male donor by monosaccharide analysis and 1H NMR [13] has been done by glycan database search (http://www.functionalglycomics.org, 2008).

Fig. 6
figure 6

a Negative ion mode nanoESI IMS oa-TOF MS/MS of the doubly charged precursor ions at m/z 1055.85. Fragmentation spectrum obtained by averaging over XIC chromatogram (Inset, dashed line); b structural candidates proposed according to the rule of glycan biosynthesis and by the presence of diagnostic fragment ions

Full size image

Thus, according to the analysis performed in this study, the present status of the KLM3 sample composition reveals 30 proposed structures which were found to be associated with detected ionic species. Upon classification of these 30 glycoconjugate structures, the expression of SiaLea/x and Sialyl T terminal epitopes in both N- and O-linked glycan components could be postulated (Table 2). These epitopes have been previously observed in malignant tissues in different organs of human bodies, where glycosylation changes caused by cancer and/or inflammation diseases are the probable cause for their overexpression [14].

Table 2 Proposed structures carrying Sialyl T and Sialyl Lea/x associated epitopes found in KLM3 urinary sample
Full size table

Since the significantly increased degree of core fucosylation of biantennary N-glycans as compared to normal has been reported in all so far known subtypes of CDG I [15], the validation of fucosylated oligosaccharides in the urine of CDG patient represents a potential direction for biomarker search.

Conclusions

Using the novel approach of ion mobility, separation coupled with tandem mass spectrometry intact and truncated N- and O-glycan structures from CDG patients’ urine can be de novo identified by mapping and sequencing. By automatic assignment, the composition of single species from the IMS-separated spectra, with reduced complexity, were shown to be accessible for structure correlations to those in databases. Through the use of IMS, the overlapping carbohydrate m/z landscape was highly reduced in given window, allowing the development of a highly efficient identification procedure by sequencing precursor ions, using transfer CID fragmentation after separation by IMS. Combining it with a previously proposed automated calculation procedure provides an efficient novel tool for glycomics to be introduced, and this can readily be applied to CDG research and clinical samples. Accordingly, different levels of sialylation and fucosylation can be analyzed and assigned to truncated and untruncated glycan structures containing possible clinically relevant epitopes.