|Home | About | Journals | Submit | Contact Us | Français|
Thirteen high mannose isomers have been structurally characterized within three glycomers, Man5GlcNAc2, Man7GlcNAc2, and Man8GlcNAc2 released from bovine ribonuclease B, six previously unreported. The study was carried out with a single ion trap instrument involving no chromatography. Three previously characterized isomers from Man7 and Man8 (three each) have been identified plus one unreported Man7 isomer. Incomplete α-glucosidase activity on the Man6 and Man7 glycoproteins appears to account for two additional isomeric glycans. The preeminence of ion traps for detail analysis was further demonstrated by resolving three new isomers within the Man5 glycomer summing to the six previously unreported structures in this glycoprotein. All reported structures represent a distribution of Golgi processing remnants that fall within the Man9GlcNAc2 footprint. Topologies were defined by ion composition pathways while linkage and branching by spectral identity in a small oligomer fragment library. The protocols described bring multiple advantages for unraveling glycan complexity, (antennal domain selectivity, isomer exposure, enhanced detectability with renormalized product spectra, effective energy enhancement for greater structural detail), that portend this instrument to be ideal for sequencing.
Lacking a definite endpoint for defining a carbohydrate sequence and coupled with the analytical difficulties of achieving an exacting structure, a tradition of reporting has developed without established standards. As a consequence, almost any structural inference appears publishable, including presumed biological insight. But, as researchers slowly unravel the details of function, it is becoming apparent that this relaxed form of reporting is inappropriate and the goals of exacting structural detail need to be relentlessly pursued and corroborated. Advances in the determination of structure are directly related to instrumental and technological developments, and for biopolymer sequencing, the mass spectrometer has played a major role. There are numerous adjunct techniques that complement a glycome analysis; however, this array of protocols often complicates a comparative structural evaluation. This can be further confused by applications on a wide assortment of sample types. Such diversity of technologies and applications needs to be focused, contrasted, compared, and brought into general acceptance before we can advance to the more challenging problems of biological function. Since many instrumental components are expensive and technically demanding, comparative evaluations from one laboratory are understandably lacking and it is quite usual to have a single laboratory position an instrumental approach energetically. With that inevitability, an alternative way to evaluate instrument advances or new protocols would be with a series of standard samples which can assess progress and potentially discover new structural features. Recent reports [1-3] have recognized this opportunity and represented protocols with a structural study of the glycans obtained from bovine ribonuclease B (RNase B). Although all glycan samples render the inherent problems of linkage and branching, these glycans exhibit an additional complexity of multiple isomers that are often transparent to MS characterization. This observation of isomers is not new, but its prevalence in common structures  and extensive presence in a diversity of malignant tissues  strongly suggests this component of structure needs greater attention. For a comprehensive glycan analysis, it may be appropriate to reason all possible isomers exist unless proven otherwise.
For a number of years selective strategies and instrumental approaches have been slowly coalescing to better understand oligosaccharide structures. Since the first commercial introduction of ITMS (ion trap mass spectrometer) equipment in 1995, (Finnigan's LCQ and GCQ and Bruker-Franzen's ESQUIRE), we have considered this approach as having the greatest promise for carbohydrate sequencing, and assembled around this technology a number of simple principles and understandings that are summarized here. Foremost: (i.) Samples prepared by methylation and reduction position fragments in a glycan array, thus, termini, extending, and branched components of structure can be distinctly placed; (ii.) Selected pathways of disassembly can define domains of structure providing antennal specificity; (iii.) metal ions (Na+), enhance sensitivity, fragmentation, and adduct stereo-specifically providing an opportunity to evaluate monomer stereochemistry; (iv.) Ion pathways that fail to define a single topology are an indicator of structural isomers, which can, in turn, be isolated and characterized; (v.) A pathway of disassembly preferentially releases labile residues providing renormalized product spectra with more comparable stability and enhanced detection; (vi.) Collision energies are effectively enhanced with disassembly, providing greater structural detail upon progression to fewer oligomers, (oscillators); and, (vii) a searchable library of fragments anchors the details of disassembly.
Thus, using ion trap MSn, a relational set of fragment ion compositions are corner-stones to structural understanding where each can be simply defined as the sum of methylated monomers plus or minus any mass change due to bond rupture at, or near, a former linkage. We refer to these small mass identifiers as scars, e.g., an open hydroxyl as in C- and Y-ions, or a (1-2)-pyranene residue characteristic of a B-ion. The ability to profile all ions in a sample glycome coupled with the facility (and need) to dig deep into product fragments and inquire about structural details generates a wealth of data that requires integrating for sequential meaning. Thus, bioinformatic tools are being built around this understanding, and such tools remain fundamental to handle the copious data. To approach this problem two strategies are being utilized, a spectral library of fragments  coupled with software tools to furnish a fragment ion's composition [unpublished]. A second tool  also couples changes in ion composition with continuity relationships for each step of an MSn disassembly pathway. Both tools and the fragment library are guides that focus output to provide discrete structures from multiple possibilities without invoking biological insight, or any other analytical data. In that sense these approach derives data de novo with a backup option to confirm and search spectral products with the fragment library. We do not have an exacting measure of how comprehensive or generally applicable such tools and ion trap MSn will become, but with that goal in mind this report is an attempt to contrast recent reports using an accepted biological reference standard, the glycans from bovine RNase B.
N-linked glycans from 2 mg of bovine ribonuclease B (Sigma Aldrich, St. Louis, MO) were released enzymatically following directions from the supplier and the isolated glycans were reduced with sodium borohydride. Methylation was carried out as described  and the dried samples re-suspended in aqueous methanol (75% (v/v) MS analysis.
Sequential mass spectra were obtained and contrasted on two ion trap instruments, a high capacity Paul trap (HCTultra PTM Discovery System, Bruker Daltonics, Billerica, MA) and from a linear ion trap (LTQ Thermo Fisher Scientific, Waltham, MA). The Thermo instrument was additionally equipped with a TriVersa Nanomate® nanoelectrospray ion source (Advion, Ithaca, NY). Signal averaging was accomplished by adjusting the number of microscans within each scan, generally ranging between 3-20 microscans. Collision parameters were left at default values with normalized collision energy set to 35% or to a value leaving a minimal precursor ion peak. Activation Q was set at 0.25, and activation time for 30 ms.
Initial topology and isomer assignments were made manually. Analyst-selected spectra were matched against a spectral library of permethylated fragments to confirm fragment ion assignments . Ion fragment pathways were selected manually by the analyst and then entered into OSCAR (Oligosaccharide Subtree Constraint Algorithm) to ascertain all isomeric structures consistent with the fragment pathway .
The nomenclature used to identify these high mannose glycans has been cumbersome and a more specific notation is sorely needed. The system used by Fu, et al, 1994  was based on antennal identification introduced by Vliegenthart, et al, 1983 . The Man7 isomers were identified as occupying positions on that template, D1, D2, and D3, (terminal positions extending each antennae on Man6 from bottom to top, respectively). This nomenclature was recently extended  to include the three Man8 isomers by supplementing double labels for the occupied sites, D1D2, D1D3, and D2D3. Different nomenclature has been introduced to include the two isomer groups, M7-1, M7-2, M7-3 and M8-1, M8-2, and M8-3 . Extension of either notation to represent all processing isomers would not be possible. Thus, introduced here is an alphanumeric notation to specifically identify all processing products, first by ascribing a glycomer number, and second, by occupancy at the six antennal positions beyond the N-linked core, two columns, (A and B) in three rows as subscripts. With this shorthand notation the Man5GlcNAc2 would be identified as 5A1,2, while the three isomers of Man7 would be identified as 7A1,2,3B3, 7A1,2,3B2, and 7A1,2,3B1. Assuming a uniform biological precursor all isomer products of processing could be classified by this simple scheme, (Fig. 1).
The most abundant glycomer to be released from bovine ribonuclease B has been that of Man5GlcNAc2 and this structure has been purported to be a single isomer. Sequential MS analysis of this ion, m/z 1595.8, (prepared as the methyl-reduced derivative), was investigated as a routine check on this readily available sample. The acquired data supported the canonical structure, and these MSn steps are presented in Figure 2, with pathway summaries in Table 2a, (MS2-6; m/z 1595.9, 1302.6, 1084.4, 866.4, and 648.2, spectra, Fig. 2a-e). Adjacent to each spectrum is a proposed structure that accounts for the major ion fragments.
When using the above pathway as data input, the output from Composition Finder indicated a consecutive monomer loss of GlcNAcol, (Fig. 2a), and three terminal (fully methylated) mannose residues, (Fig. 2c-e), as a consequence of simple glycosidic cleavage, (Table 2a). Also, the compositions increased by a single scar for each neutral loss indicating all were branched residues. Loss from a linear topology would show no scar increments. The last pathway ion, m/z 648.2, (Fig. 2e), possessed a sodiated Hex2GlcNAc composition with four scars. One scar would be expected from the first neutral loss in the pathway, (-293 amu, GlcNAcol), while the other three must be attributed to the former mannose sites. This fixes the product ion with three hydroxyl groups, matching the sodiated mass of m/z 648.2. Topology and linkage of this ion was supported by MS6, (Fig. 2e). This spectrum provided B-ions at m/z 213 and 403, positioning double, and single branching (two and one hydroxyl groups; e.g., scars) on the respective monomers, while the 3,5An-cross-ring fragments at m/z 301.3/273.2, (n = 2) and 491.2 (n = 3), identifies a sequence involving the 6-, and 4-position in the core Man-GlcNAc. For each disassembly step, facile losses provide a normalized product ion spectrum with more comparable collision stability and a different metal binding and ionization cross section. These factors, coupled with the smaller number of monomers (oscillators) for energy dissipation opens the possibility of unique fragments that are most likely inaccessible by tandem MS. A measure of glycan topology (monomer connectivity) can be accessed by selecting ion pathways that rupture at critical branch junctions (central core mannose) that delineate antennae composition. These details support an isomer with 5A1,2 topology. The familiar m/z 458 ion (Fig. 2e) represents a common methylated core disaccharide fragment, (HO)2ManGlcNAc, used to identify core branching. As seen below many of these small oligomer fragments reoccur in samples and their spectra are cautiously used as “standards” in a growing fragment library.
The spectral conclusions summarized for the canonical Man5GlcNAc2 (5A1,2) structure can be supported by following alternative disassembly pathways and product ion duplication provides assurance for specific structures (Table 2b-c). But equally as noteworthy has been the observation that sequential disassembly exposes isomers when product ions are inconsistent with a presumed precursor . This proved to be the case during disassembly of 5A1,2, where the MS4 spectrum (Fig. 2c) showed low abundant fragments, m/z 445.2, 533.3 and 880.4 that could not arise from the (5A1,2) canonical structure. Such fragments, would be possible from a structure possessing terminal disaccharides as B2-ions, (m/z 445.2), and a very common corollary, an adjacent cross-ring cleavage 3,5A3-ions, (m/z 533.3). Additionally, the fragment, m/z 880.4, (204.2 amu loss from the precursor m/z 1084.4, Fig. 2c), with the loss of a terminal and penultimate hexose (-218, and -204 amu) could not be a fragment product from 5A1,2. To evaluate this aberrant fragment (m/z 880.4), the ion was selected and disassembled, MS5-7, (Fig. 3b-e) which uncovered a series of product ions indicative of T2 topology with disaccharide antennae linked from both the 3-, and 6-linked mannose positions (Fig. 1). A series of neutral loss products, (m/z 662.2, 458.1, 268.1), were supported by the identical bond rupture but exhibiting sodium retention as B-ions, (m/z 445.1, 635.2), (Fig. 3b). This latter fragment provided a composition of Man3 with two scars which could be accounted for as GlcNAc and tMan neutral losses yielding the respective B3- and Y4-ions. Such fragments indicate a tMan-Man-(HO)Man-GlcNAc sequence. Moreover, linkage between each of the monomers was indicated with a 4,0A-ion series, (m/z 301.1, 505.3, 723.4), defining 6-, 6-, 4-interresidue linkages. The ions in this spectrum (Fig. 3b) suggest a composite of three tetrasaccharide fragments and their proposed structures are presented to the right of the figure. Two of these fragments retain a 6-link as suggested by the O,4A-ion series mentioned above, Man(1-6)Man(1-4)GlcNAc, and the second would be identical with the exception of the terminal linkage, Man(1-2)Man(1-4)GlcNAc. Thus, the respective precursors must originate from structures with T2 topology possessing linkage isomers 5A1,3, 5A2,3 (Fig. 1). Isomer disassembly using the m/z 880.4 fragment filters away two other topologies (T1, T3, Table 1) and captures the core remnants of two isomeric structures, 5A1,3, and 5A2,3.
A disassembly pathway including the fragments m/z 533.3, and 547.2, (3,5A3, B3-3,5X3, respectively), provided an opportunity to study the linkage specific details of the respective 6-, and 3-linked antenna (Fig.3d-e). The disaccharide pendant on the 3-linked core mannose was isolated with the fragment m/z 547 (Fig. 3e, pathway insert) which provided a C2-ion spectrum of the 3-linked core disaccharide (m/z 463.3).
The 6-linked antenna contributes two linkage isomers with T2 topology. They were characterized through the ion m/z 533.3 (MS6) following the pathway (Table 2o) which provided the product spectrum (Fig. 3d). The fragments indicate the spectrum to posses two disaccharide isomers, tMan(1-6)Man-, and tMan(1-3)Man- which would be the respective 6-linked antenna core components of the isomers 5A1,3 and 5A2,3. On the basis of ions present, the core 3-linked disaccharide appears to be a single tMan(1-2)Man- structure, (Table 2p, Fig. 3e). Their contrasting spectra are discussed below. Exacting confirmation could not be assured because C-ions are not in our fragment library at this time and further disassembly to B-ions could not be pursued due to a lack of signal.
In the canonical Man5GlcNAc2 structure an additional point of isomerism could be within the trisaccharide antenna 6-linked to the core mannose. To check this possibility the profiled ion (m/z 1595.8) was isolated, fragmented in two steps to release labile residues (Fig. 2c) and the 6-linked antenna was specifically isolated by selecting the 3,5A-ion cross-ring cleavage of the central core mannose (MS5, m/z 737.3, Fig. 4b). In the renormalized spectrum the C3-ion was the base ion and, as expected, retained the branched C2α-domain with characteristic ions indicating 6- and 3-linkages to a central mannose. However, there was also evidence for a linear isomer within this trisaccharide fragment with B2-, and C2-ions (m/z 445.2 and 463.2). To confirm these signals the base ion was isolated (MS6, m/z 667.3) and disassembled which provided a full set of abundant fragments allowing a better characterization of this precursor ion, (Fig. 4c). The fragments observed in this spectrum were characteristic of C-ions, with extensive rupture to the reducing-end monomer, (0,4A-, 1,4A-, 2,4A-ions; cf. Fig. 4c). This triple set of losses, (74, 104, 148 amu) was observed at m/z 593.3, 563.3, and 519.1, and again following the loss of the non-reducing terminal hexose (218 amu) at m/z 449.2, (m/z 375.2, 345.2, 301.1). These combined losses support a linear trisaccharide and the absence of a 6-linkage, (expected for the C2α-ion from 5A1,2). Such data argues for a linear T3 topology within the m/z 667.3 ion. Equally as convincing were the disaccharide fragments, m/z 445.0, and 463.2. With this understanding the 3-linked disaccharide was isolated (m/z 563.3) and disassembled (MS7, Fig. 4d) which provided access to the linkage details of this C-ion fragment and the isomer 5A2B2, Manα(1-2)Man, (MS8, m/z 463.2, Fig. 4e).
To our knowledge the isomers of Man5GlcNAc2 have not been reported, and it was the objective of this report to contrast this non-chromatographic isomer characterization with more conventional studies of Man7GlcNAc2 and Man8GlcNAc2 [1-3,9]. Using an approach identical to that described for the Man5GlcNAc2 glycomers, five Man7GlcNAc2 isomers were detected, two previously unreported. One turned out to posses a new topology (T3) within the Man7GlcNAc2 series (7A1,3B1,3, Fig. 6, Table 3h-i), and the other Glc1Man6GlcNAc2, could be a byproduct of incomplete α-glucosidase II Golgi processing, (7A1,3B3C3, Fig. 1 and Table 3q-t). A similar capped 3-linked antenna was detected in the isomers of Man8GlcNAc2. These were assumed to be glucose residues, but this has not been confirmed. The four Man7GlcNAc2 isomers are represented by three topologies, two of which were identical due to branching in the 6-linked antenna, (7A1,2,3B1; and 7A1,2,3B2).
Sequential disassembly of the methyl-reduced Man7GlcNAc2 precursor (m/z 1013.52+) following the pathway (Table 3a, 3b) identified the structure 7A1,2,3B3 (MS9, Fig. 5). These results were evident from the sequential product ion compositions (m/z 1710.7, 1492.6, 1288.6, 1084.4) and their mass intervals (-293, -218, -204, -204). In detail, the first ion, neutral loss of GlcNAcol, had a sodiated mass composition equal to Man7GlcNAc with one scar, the second ion, a neutral loss of a fully methylated mannose, tMan, had a sodiated mass composition equal to Man6GlcNAc+2 scars. This increment in scars indicates the loss originated from a different terminus on the precursor ion. This was not the case for two subsequent losses of the third and fourth residues, (m/z 1288.6, 1084.4, Fig. 5b). This linear trisaccharide could only originate from a 3-linked antenna and signify either 7A1,2,3B3 or 7A1,3B1,3 structure (Table 1). Confirming these results were the single scarred B-ions, m/z 445.2, and 649.4, (di- and trisaccharide, resp.), and the absence of any branched fragments, (doubly scarred) (Fig. 5b). Resolving which of the two possible structures (7A1,2,3B3, or 7A1,3B1,3) required disassembly of the 6-linked antenna. Thus, the 3,5A4-ion, (m/z 737.4), was isolated and disassembled following the pathway, (m/z 667.3, 449.4, 259.1, Table 3b). The first product ion and spectra supported a C2α-ion (m/z 667.3) structure with a tMan loss and an absence of any disaccharide B2-, or C2-ions. The second ion (m/z 449.4) provided a common library spectral match with fragments supporting a tMan B1-, C1-ions and an O,4A1-fragment (m/z 301.1) indicating a terminal mannose in a 1-6 linkage, Man(1-6)(HO)Man-OH. Isolation and analysis of the last ion, m/z 259, indicated it to be the C1-ion of mannose. These data confirm the presence of an isomer 7A1,2,3B3. (Fig. 5, Table 3a, 3b). Spectral identity was observed with small oligomers within the library originating from different biological sources and pathways providing reassurance of structure. Some spectra have been derived from synthetic standards; however, the majority of library files originate from well known biological sources.
In devising MSn pathways, an evaluation of precursor ions and their relationships provide clues of connectivity (topology). Monomer distribution between 6-, and 3-linked antenna reflected topology and these variations can be assessed in the core marker ions (3,5An-, 3,5Xm-ions, resp., n = 4, 5; m = 1, 2). Such fragments are more apparent following neutral loss and renormalization, e.g., GlcNAcol, tMan, (cf. Fig. 2a-c), and, as detailed with the previous glycomers, complexity can be resolved by including such specific ions in disassembly pathways. This provides a non-chromatographic filter that insures product-precursor relationships, and when presumed ion structures fail to match a single topology or linkage pattern this foretells isomers, (e.g., Man5GlcNAc2 isomers). Examples of this were the B2-, and C2-ions (m/z 445.2, 463.2), when none would be expected (Fig. 3), and the same ions in the Man7GlcNAc2 spectra (even after the sequential loss of the 3-linked linear trisaccharide, m/z 1084.4, Fig. 5b). An additional clue was the detection of a small m/z 941.6 fragment (Fig. 5b), a 6-linked antenna marker fragment 204 amu (3,5An-ion, n = 5) greater than that observed for the structures, 7A1,2,3B3 and 5A1,2, indicating a tetrasaccharide at that position. But, adding a hexose monomer to the 6-linked antenna while still releasing a linear 3-linked trisaccharide summed incorrectly and could only mean the presence of two additional isomers. Isolation of this fragment (m/z 941.6) and collision analysis (MS6, Fig. 6a) exposed a non-core marker fragment, a 3-linked trisaccharide, 3,5X3-ion, m/z 547.3, (Fig. 6a, cf. Fig. 3b), and this terminal disaccharide was isolated following the pathway, m/z 871.5, 547.3, 463.2 (Fig. 6b-c). This C2-ion, m/z 463.2, was a library match for the structure, Man(1-2)Man (Fig. 6c), identifying this Man7GlcNAc2 isomer as 7A1,2,3B2.
This alternative extension of the 6-linked branch isomer (above, 7A1,2,3B2) would have an identical topology, but a different structure as a consequence of linkage isomerism. This isomer was characterized by combining two antennae specific marker ions in the disassembly pathway, (m/z 941.5, 3,5A5-ion and m/z 505.2, O,4A3-ion, Table 3n) which would isolate the terminal upper arm disaccharide. The product of that pathway, (Fig. 7a) proved to be identical with the library B2-ion standard, Man(1-2)Man (cf. 8A1,2,3B1,3).
This unreported isomer had symmetrical topology with trisaccharide antennae extending from the mannose core at the 6-, and 3-positions. This was indicated from the respective antennal marker fragments, (m/z 737.4, 3,5A4-ion, and Bn-,3,5X3-ion, m/z 751.4). The topology was homologous with the 5A1,3 isomer differing by one monomer (204 amu) on each antenna, (Fig. 3b). Confirmation was pursued by isolating each marker fragment and disassembling the respective antennae (Fig. 6d, Table 3h-i). Analysis of the 6-linked antenna, (m/z 737.4) provided a spectrum identical with the same mass from isomer 5A2B2, (cf. MS5, Fig. 4b and MS7, Fig. 6d). Further analysis of this C-ion trisaccharide by selecting the base ion (m/z 667.3) for disassembly, however, provided a significantly different spectrum (Fig. 6e) than the comparable ion product, (Fig. 4c). This difference must be a reflection of the structure, Man(1-2)Man(1-6)Man (Fig. 6e) in contrast to the isomer Man(1-2)Man(1-3)Man (Fig. 4c). Further characterization of this trisaccharide product (Table 3g-i) showed the disaccharide non-reducing terminus (m/z 463.2) to be identical. These products define the isomer to be 7A1,3B1,3.
Three Man8GlcNAc2 isomers have been reported [1,3-6], and these structures differ from a Man9GlcNAc2 glycomer by having one missing α-2mannose residue on the termini of each antenna (Fig. 1). Since several antennae have been previously characterized in this report, the detailing of subsequent structures for the Man8 isomers would be a matter of revisiting those ions. The basic approach was to expose topology through the core mannose marker ions followed by probing the linkage details in the selected domains. However, the 8A1,2,3B1,2 isomer could not be approached in this manner, because of the greater lability of the mannose termini which exceeded that of the 3,5A6 marker ion. This required an alternative marker ion.
As with the Man7 glycomers, branching within the 6-linked antenna introduced a pair of isomeric structures that comprised one topological grouping, (T1, 8A1,2,3B1,3; 8A1,2,3B2,3), and the second topology was represented by a single structure with a shortened 3-linked antenna, (T2, 8A1,2,3B1,2). No additional topologies could be envisioned within the footprint of the Man8 processed glycomers. Thus, the structural isomers of Man8 were approached in the usual manner by releasing the terminal GlcNAcol, a 3-linked linear antenna, and renormalizing the product spectra, (Fig. 7 and Fig. 8a-d, Table 4e). The linear trisaccharide loss while maintaining the 3,5A5-ion composition (m/z 941.5) confirmed the T1 topology and the location of the isomeric linkages. These product spectra were analogous to the isomer 7A1,2,3B1, (cf. Fig. 6a-b and Fig. 7 resp.), and this spectral identity characterized the Man8 isomer as 8A1,2,3B1,3 with a matching B-ion product, Man(1-2)Man, (MS9, m/z 445.2).
The isomeric structure in the 6-linked domain representing the isomer 8A1,2,3B2,3 was approached in a similar manner using the 3,5An pathway fragment m/z 941.5. But identical to the Man7 isomer (7A1,2,3B2) a second marker ion, 3,5X3, m/z 547.3 was included in the pathway (Table 4i-j) to isolate the 3-linked disaccharide structure. The spectral identity at this position for the C2-ion Man(1-2)Man confirmed the 8A1,2,3B2,3 isomer, (Fig. 8).
This last isomer proved to be quite different in its activation stability, and alternative fragment ions had to be considered to define the structural details. In the previous glycomers we had relied on the marker ions 3,5An and 3,5Xn-ions to sort topology in the 6-, and 3-linked antenna. These fragments and their correlated mass shifts help identify markers, but when neutral loss fragments dominant the spectra this identity is lost. Thus, an alternative approach is to search for fragments that would expose elements of structure in a product spectrum. Trimming labile GlcNAcol, a disaccharide, and tMan residues provided a series of B-ions that defined a backbone sequence that could only be understood as an 8A1,2,3B1,2 isomer (Fig. 9, Table 4o-p). Particularly defining were the ion series m/z 505.3; 695.3; 913.4; indicating a sequence tMan-Man(1-6)(HO)Man(1-6)(HO)Man(1-4)GlcNAc (m/z 1070.5, Fig. c), and its precursor ion, with an incrementing terminal disaccharide, (Fig. 9b, m/z 1274.5). The only isomer that would fit these fragments would be 8A1,2,3B1,2.
As mentioned above for the Man7GlcNAc2 isomers, a glucose capped, 3-linked linear tetrasaccharide (Hex1Man6GlcNAc2) was also detected in the isomers of Man8GlcNAc2, (8A1,3B1,3C3, Fig. 1 and pathways Table 4l- nm). The stereochemistry of this glucose residue was not specifically determined and only assigned on the basis of a 3-linked linear tetrasaccharide antenna vs trisaccharide in the high-mannose footprint. This was coupled with the requisite 3,5An-marker ions, where n = 3 for Man7, (m/z 533.6), and n = 4 for Man8 (m/z 941.5). For both structures, the sequential loss of a tMan with a product composition of one scar, followed by three 204 amu losses, with no increment in scars, provided the information for concluding glucose capping.
Protocols to understand carbohydrate structure include a large list of mass spectral instruments coupled with numerous ionization techniques, and interfaced with various chromatographic methods. These provide improved component and mass resolution with differing strategies to induce fragmentation. Although some glycan structures may be better understood with improved mass accuracy, the transparency of linkage, branching, and stereochemical features brings new challenges beyond ion mass resolution. Combinations of internal and external source fragmentation techniques are striking and represent the cutting-edge of fixed energy deposition, and it may be too early to comment on the advantages such strategies bring to glycan analysis. But, as the energy of excitation becomes more selective the general features of activation of all linkage types may be missed.
In this report we have discussed how ion trap mass spectrometry can couple precursor isolation with sequential disassembly to define glycan topology and expose isomers in a non-chromatographic understanding of structure. This ion control is a salient feature for carbohydrate sequencing that allows a focus on glycan domains (antenna) and subsequently to specific inter-residue linkages. Disassembly with the ion trap also carries a different and variable set of physical parameters to induce fragmentation, and this takes place with each precursor-product step. Another aspect that plays most importantly into stepwise disassembly is the decreasing number of oscillators for energy dissipation thereby providing effective energy enhancement . Thus, structural details inherent to understanding linkages are maximally displayed in smaller oligomers, (cf. Fig. 4a and 4e). These attributes have been delineated in this study with RNase B glycans, a paradigm sample used by many. In repeating this application, it is hoped that the more comprehensive data presented here would provide greater insight into the subtleties and value of the ion trap as a carbohydrate sequenator. For the Man5GlcNAc2 glycomer three topologies that include four linkage isomers are detailed, three of which appear to be unreported. Selected ions indicative of isomeric structures have been listed in MS2 spectra of Man5GlcNAc2 where they appear to be assigned incorrectly , and even partially resolved in an extracted ion chromatogram, although such heterogeneity was not discussed . The three previously characterized isomers of Man7 and Man8 have been also identified plus one additional unreported Man7 isomer. Two additional products have been recognized that suggest incomplete α-glucosidase activity in the processing pathway of the Man7 and Man8 glycoproteome structures. All isomers have been identified with a single instrument without any ancillary techniques of ionization, chromatographic interfacing, or ion activation. The lack of detecting sensitivity must account for the inability of NMR to identify these low abundant isomers , esp., the Man5GlcNAc2 structures, while in contrast, nanospray static infusion (TriVersa Nanomate®) and signal averaging for extended periods exposes unique opportunities to detect trace sample amounts.
In describing the specific spectra, the stereochemistry of monomers were assumed on the basis of their susceptibility to endoglycosidase release, molecular weights, and position in the high mannose motif, Man9GlcNAc2. Where the 3-linked antennae were capped with an additional hexose, it was understood to be glucose and a product of incomplete α-glucosidase II activity. No confirmation by GLC-MS was performed. For the Man5GlcNAc2 glycomer of RNase B, four linkage isomers have been detected which fall into three unique branching topologies. This word topology has been adapted from the field of mathematics and applied to delineate glycan connectivity and separate from identification of linkage isomers. For the glycomers of Man7GlcNAc2 and Man8GlcNAc2 the three linkage isomers all have a unique topology. The abundance of individual isomers within the Man7 and Man8 glycomers has been summarized and reported [1,9,10]. Providing such information on the basis of fragment abundance would require information not available and was not considered. The ability to identify minor components in a field dominated by other isomeric structures has important implications for the detection and tracking of altered cellular processes. The trace levels initiating function, coupled with the dynamic and temporal aspects of glycosylation will be the bona fide structural challenge when contrasted to the tissue glycome. Under these circumstances, we anticipate ITMS instrumentation to be alone in approaching such problems.
A comprehensive evaluation of a sample glycome must entail checking for all possible isomers and a low mass scanning range that includes disaccharides, where unique structural details could exist. Rarely are glycoproteome glycans single structures, but appear to represent a distribution of multiple topologies and isomers. This would strongly suggest that tandem MS alone to be quite inappropriate for providing a comprehensive structural evaluation. As one of many examples, the MS2 C2α fragment of Man5GlcNAc2, m/z 667.3 (Fig. 2a) was shown to exist as a linear trisaccharide, T3 (Fig. 4). Although small, such fragments are well within the dynamic range necessary for further evaluation, and when free from more facile ruptures are renormalized for subsequent MSn steps, (cf. Fig. 3a and 3b). Directing disassembly along pathways that include 6-, and 3-linked antennal fragments also insures specificity, and this was the strategy and pathway that provided the two isomers within the C2α-ion 3-linked antennal domain of the putative Man5GlcNAc2 structure (Fig. 3a-e). These spectra show prominent B2 and C2 fragments and the expected 6-linked, 4,OA2, m/z 301.1, and 3,4X2, m/z 375.2, and 3-linked 2,4A2, m/z 315.0 fragments consistent with the branched Man1-6(Man1-3)Man topology. But importantly, the fragments, m/z 227.0, 463.2, 519.1, 533.1 and 563.3, (Y2, C2 2,4A3, 3,5A3, and 1,4A3, respectively), are separately and collectively indicating a linear C3 fragment topology (Fig. 3a). The fragment m/z 563.3 specifies the linear extension is at the 3-position and isolation of that ion, MS6, provided a spectrum exhibiting glycosidic cleavage at m/z 259.0, 327.2, 345.1, 445.0, and 463.2 (C1, Z2, Y2, B2, and C2, respectively), (Fig. 3b). To focus specifically on the terminal linkage within this 3-linked antenna, the C2 fragment, m/z 463.2, was isolated which provided the MS7 spectrum (Fig. 3c). Here again, as in the terminal disaccharide fragments of the D2 topology, linkage multiplicity reigns. Fragment ions m/z 227.0, 245.2, and 259.1 (Z1, Y, and C1, respectively), define the glycosidic ruptures, while the cross-ring fragments for the 3-, and 6-linkage can be easily observed, (m/z 315.1, 389.1, 329.1, 301.1). This provides detailed evidence of linkage isomers at that position with Man5GlcNAc2, in the T3 topology (Fig. 1 and Fig. 3).
This diversity of RNase B glycans was long suspected. The glycobiology pioneer Professor Akira Kobata, following data published almost 30 years ago reported, “…the structures of the sugar chains of ribonuclease B indicated that the removal of Manα1-2 residues in the processing step occurs randomly”, . In 1991 YC Lee's group (Baltimore) using HPAE chromatography and PMPMP derivatization demonstrated extensive column heterogeneity beyond the classical Man5-9GlcNAc glycomers . Resolving such complexity was projected to invite insurmountable difficulties, e.g., “…the Isomer Barrier, a persistent technological barrier to the development of a single analytical method for the absolute characterization of carbohydrates…” . More specifically, “…the stereochemical blindness of mass spectrometry, productions must first be separated based upon a physical principle that is not dependent upon m/z prior to fragmentation.” . Such skepticism was based on “……a serious problem because they (isomers) yield sets of substructures after every round of dissociation where subsequent fragmentation of any given isolated ion m/z furnishes identical product ion m/z values” . Clearly, the attributes of ITMS were not considered when these comments were drafted, and although this report may challenge such reasoning, we claim no absolutes in structural characterization as we move ahead. However, it does appear that different isomeric structures will eventually provide an altered fragmentation pattern somewhere during disassembly. Anomeric differences in spectra have been observed, but a lack of chemically synthesized standards makes extensive claims inappropriate. Duplication of end-product spectra pursued by different disassembly routes, spectral identity from different samples, preparations, and animal species brings some degree of confidence for a sequencing strategy that may be comprehensive.
An unfortunate aspect of much carbohydrate MS reporting is the conclusion that if you can not observe a difference MS/MS spectrum alternative structures are not present with the corollary that a tandem MS approach supplies a comprehensive answer to structure; “….assignable to unambiguous sequences.” Most wish carbohydrate sequencing would be that easy.
The authors acknowledge funding support from NIH-NIGMS RO1-GM54045, and NCRR-RR-16459.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.