Mass Spectrometry Imaging with Trapped Ion Mobility Spectrometry Enables Spatially Resolved Chondroitin, Dermatan, and Hyaluronan Glycosaminoglycan Oligosaccharide Analysis In Situ

Glycosaminoglycans (GAGs) are ubiquitous, vital polysaccharides (1) which possess numerous important roles, ranging from tissue structure and stability (2,3) to cell signaling, (4,5) which in turn modulates numerous biological processes including cell growth and proliferation, (6) cell adhesion, (7) wound repair, (8,9) and pathogenic invasion. (10−13) GAGs possess complex structures which are characterized by, with the exception of hyaluronic acid (HA), levels and types of sulfation. GAGs also have a high pharmaceutical potential, (13−17) which is likely due to their promiscuous binding activity, which is driven by sulfate composition. (18) Despite their importance in human health and disease and the treatment thereof, little is known about the states of natural GAGs within the local cellular environment. (19)

Mass spectrometry imaging (MSI) is a well-established technique which enables the profiling of various molecules in a spatially resolved manner. (20) To date, no direct attempt at analyzing the sulfation pattern of GAG oligosaccharides with MSI has been reported. Native fragments, attributed to sulfated HexNAc-HexA (N-acetyl hexosamine-Hexuronic acid) polymers which may correspond to GAGs have been detected with MALDI-FTICR-MSI (21) (matrix assisted laser desorption ionisation-Fourier transform ion cyclotron resonance) but were not characterized further. Clift et al. (22) utilized chondroitinase (CHase) digestion to improve downstream protease and PNGase F treatments, which enabled them to detect di- and tetra-saccharides (degree of polymerization; DP2 and 4) of chondroitin sulfate (CS). HA has also been imaged in human skin using hyaluronidase (H1136) digestion. (23) No MSI methodologies have been reported for the analysis of sulfate composition, most likely because the majority of sulfated of di- and oligosaccharides are isomers.

Usually, the sulfate composition of GAGs is analyzed using in-line separation (either chromatographic or electrophoretic). (24) In-line separation per pixel – while possible – is difficult to achieve, owing to long run times (minutes to hours per pixel depending on the analyte studied). (25) Here, ion mobility spectrometry (IMS), a technique which separates ions based on their different collisional cross sections (CCSs), more colloquially, by their apparent size and shape in 3D space (26) is utilized instead. Trapped IMS (TIMS), a variant of IMS developed by Bruker, (27) offers improved resolving power and ion transmission in a small footprint compared to other IMS techniques (28) and has separated derivatized heparin/heparan sulfate (HS) disaccharides (29) and, CS and heparan sulfate DP4 and DP6 isomers previously. (30)

Here, we demonstrate the ionization, identification, and analysis of mass, mobility, and spatially resolved CS, the sister molecule of CS, dermatan sulfate (DS), and HA ions generated through in situ tissue digests.

Methods

Nomenclature

We describe ions based on disaccharide and sulfate composition, where D indicates a single, unsulfated, unsaturated HexA-HexNAc disaccharide (C₁₄H₂₁NO₁₁). OS indicates an O-sulfate modification and corresponds to the addition of SO₃ with regard to mass (Figure 1Di). Consequently, a singly sulfated disaccharide is represented as D(OS) (Figure 1Di), while a tetrasaccharide with one sulfate is represented as 2D(1OS) (Figure 1Dii). Generic chain lengths regardless of the sulfation state are referred to by their degree of polymerization (DPX), where X is the number of saccharide residues; e.g., a hexasaccharide (6-mer) is a DP6. The number of sulfates per disaccharide residue (degree of substitution; DOS) can be easily calculated where, for XD(NOS), the DOS is N/X. For fragments, Domon and Costello nomenclature (31) is utilized.

Anal. Chem. 2024, 96, 45, 17969–17977: Figure 1. Chondroitin, dermatan, and hyaluronan structures. A) CS (GlcA-GalNAc; Glcuronic acid-Galactosamine) structure. B) DS (IdoA-GalNAc; iduronic acid) structure. C) HA (GlcA-GlcNAc; glucosamine) structure. D) Examples of the structures and their corresponding ion nomenclature. i) examples of disaccharides (D(xOS)). For D(xOS), D indicates a single disaccharide of mass C14H21NO11 (red and gold or black), while a single OS indicates the addition of an SO3 by mass (blue or purple). A chondroitin disaccharide containing only one sulfate at C-4 of GlcNAc (purple) is a CSA disaccharide and at C-6 of GalNAc (blue) is a CSC disaccharide, and that containing two at both C-4 and C-6 is a CSE disaccharide. Other CS disaccharides exist. The reducing end forms an equilibrium of different anomers in solution: α- (gold), β- (black), and open chain (not shown here); ii) example of a singly sulfated tetrasaccharide: 2D(OS).

Mass Spectrometry

MS analysis was undertaken using a tims-TOF fleX (Bruker Daltonics, Bremen, DE) instrument in negative ion mode. The MALDI matrix, 15 mg·ml^–1 9-aminoacridine (92817; Sigma-Aldrich, St. Louis, MO), 0.2% formic acid (Z0797502 21; Sigma-Aldrich) in 69.9% methanol (14262; Fisher Scientific, Hampton, NH), and 29.9% LC grade water (7732-18-5; Fisher Scientific) was used. The matrix was applied using a TM HTX3 sprayer (nozzle temperature = 80 °C, gas pressure = 10 psi, flow rate = 100 μL/min, velocity = 1200 mm/min, track spacing = 3 mm, number of passes = 16 in a criss-cross pattern with 2 s drying time). The instrument was mass and mobility calibrated using direct infusion ESI of the tune mix (G2431; Agilent Technologies, Santa Clara, CA, USA). The TIMS buffer gas flow was allowed to equilibrate for at least 30 min after loading the MALDI plate and then set to 2.85–2.87 mbar.

The ion source parameters were set as follows: for MALDI, MALDI offset = 50 V and shots = 200, rate = 10 kHz. For direct infusion ESI, an end plate offset of 500 V and capillary voltage of 3000 V were used, with the nebulizer set to 0.3 bar, the dry gas at 3.5 L/min, and the dry temperature at 200 °C. The instrument parameters were as follows (for a mass range of 50–2000 m/z, using a quadrupole mass filter at 200 m/z with 850 Vpp collision RF): deflection 1 delta = −70 V, funnel 1 RF = 500 Vpp, funnel 2 RF = 200 Vpp, multipole RF = 200 Vpp, isCID = 0v, ion energy = 5 V, CID = 5 V. The TOF transfer time was 110 μs, and the pre-pulse storage was 10 μs. TIMS was run with N₂ as the buffer gas, and the parameters were the following: accumulation time = 20 ms (corresponding to 200 shots when using MALDI), ramp time = 200 ms, ramp start (1/k₀start; voltage) = 0.60 V·s/cm²; 242.8 V, ramp end (1/k^₀end; voltage) = 1.80 V·s/cm²; 46.5 V. TIMS offsets set with IMEX were Δt1 = 20 V, Δt2 = 120 V, Δt3 = −70 V, Δt4 = −150 V, Δt5 0 V, Δt6 = −150 V, and collision cell in = −300 V.

Imaging parameters such as pixel size and x,y coordinates were set using Flex Imaging (Bruker Daltonics) software. The majority of images were acquired at a 20 μm pixel size, while one set of Chase AC images was acquired at a 40 μm pixel size.

MS/MS experiments were performed for ions with sufficient signal and sodiation to yield fragments (D(1-2OS), 2D(0-2OS), and 3D(0-2OS)) using CID (25–55 eV). Comparison of sample TIMS profiles to those of the CS standards was also used.

Results and Discussion

Spatial Resolution of GAG Oligosaccharides

MALDI MSI analysis of the primate retina after enzyme digestion was able to spatially resolve the identified sulfated oligosaccharide ions (Figure 2) and is shown alongside the histological identification of the retinal layers (Figure 3). H&E staining struggles to highlight all the regions of the sections in which GAGs are present. The vitreous humor, as it is 99% water, is often washed away during staining and hence is not indicated histologically, (34) however, optical images of the sections after matrix application indicate regions which likely correspond to the vitreous humor (Figure S6ii). The protective outer layer is also poorly stained, but the outside edge can be observed in H&Estained sections. This is most clear in later sections (Figure 4G). Extensions beyond these regions may have been due to diffusion during embedding or delocalization during enzyme treatment. The vitreous humor and outer protective layers are indicated in ion images where applicable with dashed and dotted lines respectively and were determined from optical images of the sections after either matrix application or H&E staining (Figure S6).

Anal. Chem. 2024, 96, 45, 17969–17977: Figure 3. MALDI-TOF MS images of the GAG oligosaccharides. A) Stacked ion images of a primate retina after CHase ABC treatment shown for DP2 (i) and DP4 (ii), and (iii) the labeled H&E stained section. B) Ion images of a primate retina after CHase AC treatment shown for DP2 (i), DP4 (ii), and DP6 (iii) ions and (iv) the labeled H&E stained section. White/red dotted lines indicate the edge of the protective outer layer and the pink dashed line the edge of the vitreous humor, both of which do not stain well with H&E. Single ion images for all GAG ions can be found in Figures S7 and S9.

Disaccharides yielded from CHase ABC digestion were located across the entire section (Figure 3Ai; blue and green). D(OS) is localized within the retinal layers, choroid, and sclera (Figure 3Ai; blue), while D(2OS) is localized to only the choroid and sclera (Figure 3Ai; green) and also forms a halo around the outside of the retinal layers and sclera, which is likely part of the vitreous humor and outer protective layer. Low levels of DP4 (Figure S7B) – specifically 2D(2OS) – were also identified (Figure 3Aii), appearing with localization comparable to that of D(2OS). Other DP4 and DP6 saccharides were detected in situ in other comparable sections and could be localized but had generally poorer signals (Figure S8). When treated with CHase AC, D(OS) and D(2OS) appear diffuse and, unlike CHase ABC treatment, share considerable overlap (Figure 3B). Both locate to the sclera, protective outer layer, and into the vitreous humor.

As the residue number increases, i.e. from a DP2 to a DP4, the localization becomes more specific (Figure 3Aii,Bii) such that spatial differentiation can be observed for oligosaccharides with different degrees of sulfate substitution (DOS) (Figure 3Bii). When analyzing DP6s (Figure 3Biii), improved localization is observed with unsulfated DP6s (3D(0OS); red), while DP6s with DOS ≥ 1 (3D(3–4OS); blue) localize similarly to DP4s with the same DOS. The distinction between these two sulfate levels becomes more apparent in DP6s than that in DP4s, with less sulfated DP6s (0 < DOS < 1; 3D(1-2OS)) localizing to the edges of the different tissue layers, while the most sulfated DP6s localize to the tissue stroma.

The observation of a difference between the protective layer and the stroma can be made using either enzyme, either as a halo of DP2s with DOS > 1 when using CHase ABC (Figure 3A) or as a halo of oligosaccharides with DOS < 1 when using CHase AC (Figure 3Bii,iii green). This is important to note, as both demonstrate that these regions have different CS profiles, but the way in which this is observed is dependent on the specificity of the enzymes used to probe them. CHase ABC will digest both CS and DS, yielding GlcA- and IdoA-containing sequences, while CHase AC will digest primarily GlcA containing CS chains, yielding IdoA-containing oligosaccharides. This explains why the same distinct tissue regions are observed through highly or lowly sulfated saccharide sequences. Furthermore, unsulfated DP2s are not detected, even with standards, using this matrix system, thus, the lowly sulfated information observed in CHase AC oligosaccharides is lost when examining only disaccharides. To investigate the halo further, we treated another section with CHase B (which targets IdoA-containing sequences; DS). Essentially no DP2s or DP4s were yielded in the retinal layers, choroid, or sclera and the signal was found primarily in the vitreous humor and protective outer layer (Figure S10), corresponding to the D(2OS) and lowly sulfated CS halos observed when treated with CHase ABC and AC respectively, suggesting that this halo contains DS chains.

Conclusion

The analysis of CS/DS in tissue traditionally includes tissue homogenization and subsequent extraction of CS from whole organs or animals, followed by disaccharide or oligosaccharide analysis involving separation mostly by LC. Such workflows average the oligosaccharide distributions for the entire tissue or organism. Here, we demonstrate that, using MSI coupled with TIMS, we are able to detect, identify, and profile GAG oligosaccharides in situ (Figure 2). This includes the spatial localization of mass-separated (either by degree of polymerization or degree of sulfation) oligosaccharides (Figure 3) and – using TIMS profiles – localization of mobility-separated species (Figure 4). The detection and localization of ions with masses and mobilities that correspond to CS disaccharides and oligosaccharides, and DS and HA oligosaccharides (Figures 2 and 3) demonstrate the spatial resolution of multiple GAGs for the first time. This information has never been available before, hence linking it directly to biological function will be the focus of future studies.

The structures within each TIMS peak of the GAG oligosaccharides still need to be elucidated. CID is a poor way to characterize GAG oligosaccharide sequences, as it primarily yields isomeric B and Y fragments and regularly results in sulfate loss. (24) This is further compounded due to a difficulty in sourcing pure standards that have not been derivatized (i.e., do not have natural mobilities due to linkers utilized for synthesis).

The presence of undersulfated oligosaccharides may be overrepresented due to the MALDI process, however, we do observe highly sulfated oligosaccharides (Figure 2Eiii, Table S1). It also seems that oligosaccharides analyzed from tissue are more heavily sodiated and hence may be more resistant to thermolabile sulfate loss. (35) Furthermore, if sulfate loss was a significant problem, then the spatial resolution of mass-separated species would not be achievable as shown in Figure 3A,B. We are able to observe distinct tissue regions that correlate to different GAG types and sequences, all of which can be observed in multiple ways through treatment with different enzymes. Changes in ion source either through the use of different matrices which are more sulfate protective or that require less laser power, or by moving to a softer ionization method (for GAGs) such as desorption electrospray ionization (DESI) (36) may yield a higher quantity of oligosaccharides with a higher DOS. This work in no way claims to accurately portray the entire sulfate composition of CS, DS, and HA in primate retina, but rather demonstrates the ability to release, detect, and profile CS, DS, and HA ions based on mass and mobility from tissue sections in a spatially resolved manner. Further work would be required to accurately and confidently characterize the sulfate composition of this and indeed any other tissue.