Evaluation of Cleavable Crosslinking for Characterization of Proteoform Structural Differences by Top-Down Mass Spectrometry

Abstract

Cleavable crosslinking has traditionally been employed in bottom-up mass spectrometry to elucidate protein structure and protein-protein interactions through identification of peptides bearing characteristic mass adducts. Here, we demonstrate the application of cleavable crosslinking in top-down mass spectrometry to enhance fragmentation efficiency and enable precise localization of crosslink sites. We first validated this approach using cytochrome c, a well-characterized model protein. Subsequently, we extended top-down cleavable crosslinking to transthyretin, a natively homotetrameric protein exhibiting extensive proteoform heterogeneity, to investigate whether proteoform variations induce structural changes detectable by this method. Our results confirm that cleavable crosslinks can be detected and characterized by top-down mass spectrometry, with crosslinker cleavage under collisional activation significantly enhancing fragmentation. Application to transthyretin (intramolecular crosslinks) yielded complex crosslinking patterns that precluded complete identification of crosslinks. However, the crosslinking data provided valuable information on solvent-accessible residues, functioning effectively as a covalent labeling strategy. This work establishes cleavable crosslinking as a viable chemical crosslinking approach for top-down mass spectrometry applications.

Graphical Abstract

Introduction

Mass spectrometry coupled with electrospray ionization (ESI) is a powerful tool for protein analysis. While bottom-up mass spectrometry relies on enzymatic digestion to study protein sequences, top-down mass spectrometry enables analysis of intact proteins, providing direct information about primary structure.^1–3 A key advantage of the top-down approach is the direct readout of the stoichiometries and identities of post-translational modifications for each proteoform, facilitating the characterization of multiple proteoforms.^{4, 5} Proteoforms arise from, for example, point mutations, post-translational modifications (e.g., phosphorylation, acetylation, glycosylation), and other products of the same canonical gene sequence.

Chemical crosslinking is a well-established technique in mass spectrometry for mapping tertiary and quaternary structure. Crosslinks are identified via tandem mass spectrometry of crosslinked peptides.⁶ Commonly employed crosslinkers include bis(sulfosuccinimidyl)suberate (BS3) and bis(sulfosuccinimidyl)glutarate (BS2G), which target lysine residues.⁷ A top-down workflow for crosslinking mass spectrometry^8–11 is advantageous as crosslinkers, and thus structure, can be related to function even when PTMs are distal in primary sequence to the crosslinks.¹² However, crosslinking in the top-down context can significantly suppress fragmentation between crosslinked residues, because even though backbone cleavage occurs, fragment ions remain bound to each other through the crosslinker, reducing the observed sequence coverage.¹³ Cleavable crosslinkers offer a potential solution to mitigate this fragmentation suppression in top-down mass spectrometry.

Mass spectrometry-cleavable crosslinking has been extensively utilized in bottom-up proteomics,^14–16 where it allows for greater crosslink peptide fragment ion coverage by abolishing the link after a crosslinked peptide pair is selected for tandem MS. Characteristic mass signatures and fragmentation patterns based on the reagent and cleavage site allow for the crosslinked sites to be identified using a workflow similar to PTM profiling. This approach reveals both intramolecular contacts and intermolecular protein-protein interactions with greater specificity.^17–19 Several cleavable crosslinkers are available, including disuccinimidyl sulfoxide (DSSO) and disuccinimidyl dibutyric urea (DSBU),²⁰ which cleave into asymmetric fragments with distinct masses for unambiguous identification. The fragmentation method required for crosslinker cleavage varies by reagent,^{21, 22} with collision-induced dissociation (CID) and higher-energy collisional dissociation (HCD) being most common, though electron transfer dissociation (ETD)-cleavable crosslinkers are also available.^{16, 19}

Combining cleavable crosslinking with top-down mass spectrometry integrates the benefits of chemical crosslinking and covalent labeling in a single experiment. When fragment ions are ambiguous with respect to localizing single crosslinks (e.g., when there are positional isomers of crosslinks, resulting in a mixture of cleaved crosslinks), reacted sites must be on the surface of the protein, generating information similar to covalent labeling experiments.^{23, 24} Crosslinker cleavage can be affected by adjusting collisional energy, potentially enhancing fragmentation.²⁵ This approach should enable more accurate localization of crosslink sites and identification of reactive residues that may be obscured by using non-cleavable reagents.

Cytochrome c and transthyretin were selected as model proteins for method development. Cytochrome c is a well-characterized model protein with extensive structural documentation, useful for validating the methodology. Transthyretin was examined due to its myriad mutations and post-translational modifications that alter its binding affinity and structural stability, allowing us to examine top-down cleavable crosslinking in a biologically-relevant use case.^{26, 27} While crystal structures of various transthyretin proteoforms reveal minimal structural differences,²⁸ NMR studies suggest subtle but functionally relevant conformational changes.²⁹ Herein, we show the capabilities of MS-cleavable crosslinks as both proximity mass and probes of solvent accessibility. Cleavable crosslinks ultimately reveal additional crosslinking identifications hidden by traditional non-cleavable linkers.

Materials and Methods

Materials

Cytochrome c (equine heart), transthyretin (human plasma), ubiquitin (bovine erythrocytes), myoglobin (equine heart), and phosphate-buffered saline (PBS) were purchased from Sigma-Aldrich (St. Louis, MO, USA). LC-MS grade dimethyl sulfoxide (DMSO) was obtained from Thermo Scientific (Waltham, MA, USA). Ammonium acetate, formic acid, and Optima LC-MS grade water and methanol were purchased from Fisher Scientific (Fair Lawn, NJ, USA). Disuccinimidyl dibutyric urea (DSBU) was obtained from CF Plus Chemicals (Brno-Řečkovice, Czech Republic). Bio-Spin columns containing Bio-Gel P-6 in Tris buffer (6 kDa molecular weight cutoff) for buffer exchange were purchased from Bio-Rad (Hercules, CA, USA).

Sample Preparation

Stock solutions of cytochrome c, myoglobin, and ubiquitin were prepared at 1 mg/mL in Optima LC-MS grade water. Transthyretin stock solution was prepared at 1 mg/mL in Optima LC-MS grade water. Protein stocks were stored at −20 °C. DSBU stock solution was prepared at 25 mM in DMSO. Ammonium acetate buffer (pH 7.4) was prepared at 250 mM concentration. PBS was prepared at 1X concentration in ultrapure water. For crosslinking reactions, DSBU was diluted to 1 mM. Cytochrome c was diluted to 10 μM in 1X PBS, and DSBU was added to a final concentration of 250 μM. The reaction was incubated at room temperature for 1 h. Transthyretin was diluted to 15 μM in 1X PBS, and DSBU was added to a final concentration of 75 μM. This reaction was incubated for 2 h at room temperature. These reaction conditions were chosen to minimize multiple crosslinker additions.

Crosslinking reactions were quenched by buffer exchange. The cytochrome c reaction was buffer-exchanged into 250 mM ammonium acetate (pH 7.0) using Bio-Spin columns. The transthyretin reaction was buffer-exchanged into Optima LC-MS grade water. Following buffer exchange, samples were diluted 1:1 (v/v) with methanol containing 0.1% formic acid to ensure complete quenching. Samples were subjected to two rounds of buffer exchange using spin columns for desalting.

Mass Spectrometry

Top-Down Analysis of Crosslinked Cytochrome c

Crosslinked cytochrome c samples were introduced by nanoelectrospray ionization through a NanoLockSpray source into a Waters Synapt G2-Si ion mobility-mass spectrometer (Waters Technologies, Wilmslow, UK) equipped with an electron capture dissociation (ECD) cell (Agilent Technologies, Santa Clara, CA, USA). Samples were infused through pulled borosilicate capillaries. Fragmentation was achieved using 60 V trap collision energy and 45 V transfer collision energy in combination with ECD to cleave the crosslinker and enhance sequence coverage.³⁰ Data were collected in triplicate on separate days under identical conditions, and fragment ions were analyzed using ExD Viewer software (Agilent Technologies) with 20 ppm maximum m/z error and all other parameters set to their default settings. P-scores and sequence coverage were determined by importing the theoretical fragment monoisotopic masses into Prosight Lite.³¹

Top-Down Analysis of Crosslinked Transthyretin.

Transthyretin samples were analyzed using a Waters ACQUITY M-Class UPLC system coupled to a Thermo Scientific Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). A 10 μL aliquot of 7.5 μM sample was injected onto a Waters nanoEase M/Z BEH C4 trapping column (300 Å pore size, 5 μm particle size, 300 μm × 50 mm) and separated on a nanoEase analytical column (1.7 μm particle size, 300 μm × 150 mm). Mobile phase A consisted of water/acetonitrile (95:5, v/v) with 0.1% formic acid, and mobile phase B consisted of acetonitrile with 0.1% formic acid. Samples were loaded at 10 μL/min, followed by an isocratic hold at 15% B for 2 min at 4 μL/min. A linear gradient from 25% to 55% B over 25 min was applied at 4 μL/min.

The instrument was operated in data-dependent MS/MS mode to acquire tandem mass spectra for each proteoform. Individual data acquisitions targeted the 12–14+ charge states of each proteoform. MS1 scans were acquired at 120,000 resolution (at m/z 200) and MS2 scans at 60,000 resolution. The m/z scan range was set to 350–3000 to capture the full range of fragment ions. The intensity threshold for precursor selection was 1 × 10⁴. Precursor ions were isolated in the quadrupole and fragmented using electron-transfer/higher-energy collision dissociation (EThcD). Electron-transfer dissociation (ETD) reagent ion injection and reaction times were determined by calibrated charge-dependent parameters, with 15% supplemental HCD applied to facilitate crosslinker cleavage, as determined by preliminary optimization experiments. Again, experiments were conducted in triplicate.

Distance Restraint Measurements

To validate the feasibility of this approach and confirm crosslinking assignments, Cα–Cα distances were measured between potentially reactive lysine residues identified in the cytochrome c crosslinking data. The crystal structure of cytochrome c (PDB code: 1HRC)³². was obtained from the Protein Data Bank. Inter-lysine distances were measured using PyMOL (Schrödinger, LLC, New York, NY, USA). To account for lysine side chain length on each end of the crosslinker, 6.4 Å was added to both termini of the crosslinker spacer arm length, based on previous studies.^{13, 33}

Results and Discussion

Top-Down Cleavable Crosslinking of Cytochrome c

Top-down cleavable crosslinking analysis involves two complementary approaches: characterization of the intact crosslinker, analogous to traditional crosslinking experiments, and identification of the two asymmetric cleavage products. DSBU yields two characteristic mass adducts upon cleavage: 85 Da and 111 Da, corresponding to fragmentation at the urea bond (Figure 1).

Figure 1:

Structure of DSBU showing cleavage sites (blue dashes) that generate 85 Da and 111 Da fragments.

The rationale for applying cleavable crosslinking in top-down mass spectrometry is to enhance sequence coverage relative to traditional crosslinkers. Upon crosslinker cleavage, the constraint imposed by the intact linker is released, facilitating more confident localization of crosslink sites.

The 8+ charge state of cytochrome c was selected for analysis based on its highest abundance in the mass spectrum. Figure 2 shows both the unmodified and crosslinking protein following the crosslinking reaction, confirmed by the exact mass addition of the intact crosslinker (196.08 Da). In Figure 2, c- and z-type fragment ions are indicated by red markers, while b- and y-type ions are indicated by blue markers. Light blue highlighting denotes crosslinker location: Figure 2B shows intact crosslinks between lysine pairs, whereas Figures 2C and 2D highlight individual lysine residues bearing the 111 Da and 85 Da cleaved crosslinker moieties, respectively.

Figure 2.

(A) Mass spectrum of DSBU-crosslinked cytochrome c (8+ charge state). (B) Fragmentation map showing intact DSBU crosslinks (196 Da). The sequence coverage is 43% and the p-score is 3.2e-209. (C, D) Fragmentation maps showing cleaved DSBU moieties of 111 Da and 85 Da, respectively. For the 111 Da fragment ions, the sequence coverage is 7% and the p-score is 6.3e-21. For the 85 Da fragment ions, the sequence coverage is 26% and the p-score is 1.2e-92. Within (B-D), c- and z-ions are represented by red markers and b- and y-ions are represented by blue markers. Light blue highlights indicate the presence of a crosslinker, with (B) showing intact crosslinking pairs and (C) and (D) showing individual residues modified by the 111 Da and 85 Da fragments, respectively.

Collisional activation energy was systematically optimized to maximize crosslinker cleavage. Fragment ions bearing the 85 Da and 111 Da modifications were manually validated to ensure correct assignment and exclude false positives. Collisional energy was applied in both the trap and transfer collision cells to induce protein backbone fragmentation and crosslinker bond cleavage. Energy was increased in 5 V increments, and fragmentation patterns were evaluated using ExDViewer until no further improvement was observed.

Analysis of intact crosslinks identified two confident assignments: K5–K7 and K88–K99. Annotation of the spectra with cleaved crosslinker moieties revealed four additional potential crosslinks: K27–K39, K27–K99, K39–K53, and K53–K99 (Figures 2C and 2D). These assignments are supported by the requirement that cleavage of one crosslinker moiety (85 Da) necessitates the presence of its complementary fragment (111 Da). However, due to the inherent ambiguity in pairing cleaved crosslinker halves, all plausible pairings are reported. Crosslink localization followed established methodologies that utilize both modified and unmodified fragments to constrain crosslinker placement.¹³

The first fragment ions - closest to the termini - that included the mass addition of the intact or cleaved linker were used to identify locations of crosslinking. Top-down sequencing does not allow for more internal sites to be determined based on sequence coverage alone in the presence of a mixture of reactive sites (e.g., an N-terminal crosslink will obscure the location of a more internal link since the fragments that explain the internal link are ambiguous, supporting both the internal link and the N-terminal link³⁴). As an example of how sites were identified, with the K5-K7 crosslink location, both the b₈ and b₇ ions were detected with an intact crosslinker mass addition. Therefore, the location was determined to be K5-K7 vs. K5-K8 due to the presence of the b₇ ion and the presence of the N-terminal acetylation. As an additional example, the b₃₂ ion in Figure 2C indicates that there is a 111 Da mass addition to that ion. Through the ion assignment process, K27 was assigned as the modified site since it is the lysine closest and N-terminal to the modified fragment ion. Therefore, by including the cleaved linker locations with the intact linkers, additional crosslinking sites were identified that cannot be determined using non-cleavable linkers (i.e., K27, K39, K53, K99).

A full list of labeled ions used for crosslinker site identification, and sequence maps for unlabeled ions, have been included in the Supporting Information (Table S1 and Figure S1, respectively). Unmodified ions are detected throughout the entire sequence. There are unmodified ions throughout the sequence (60% sequence coverage), indicating that there must be a mixture of crosslinking site isomers - although the stoichiometry for the m/z isolated precursor ion is still 1:1 protein:linker - and thus, cannot be used to identify crosslinking sites.

Distance Restraint Validation and Structural Interpretation

To validate crosslink assignments, inter-residue distances were measured using the cytochrome c crystal structure (PDB: 1HRC). Cα–Cα distances were extracted for all identified lysine pairs using PyMOL. The maximum allowable distance for a DSBU crosslink is 25.3 Å, calculated as the sum of the crosslinker spacer arm length (12.5 Å) and the combined lysine side chain contributions (2 × 6.4 Å = 12.8 Å).³³ All intact crosslinks (K5–K7 and K88–K99) exhibited Cα–Cα distances below this threshold, confirming their structural feasibility. For cleaved crosslinker moieties, solvent accessibility of the modified lysine residues was evaluated. All lysines identified in Figures 2C and 2D are surface exposed in the crystal structure, consistent with their reactivity toward the aqueous crosslinking reagent.

Given the multiplicity of lysine residues bearing cleaved crosslinker moieties, unambiguous pairing of the 85 Da and 111 Da fragments to reconstruct intact crosslinks was not possible. Consequently, cleaved crosslinker modifications are best interpreted as covalent labels that report on solvent-accessible lysine residues rather than definitive distance restraints. This labeling information nonetheless provides valuable structural insights by identifying reactive surface residues. In addition to identifying reactive surface residues, the cleaved labels also assist in improving the sequence coverage when paired with the intact crosslink data that is present. Whenever the crosslinker is cleaved, the additional cleaved labels can provide more information about the structural conformation of the protein currently being studied.

Top-Down Cleavable Crosslinking of Transthyretin Proteoforms

Transthyretin exhibits multiple proteoforms in human plasma, including point mutations and PTMs.³⁵ Proteoform variation is implicated in transthyretin amyloidosis, which results in pathological protein aggregation in tissues.³⁶ Two clinically relevant proteoforms were selected for analysis: cysteinylated transthyretin (CysTTR) and the C10G point mutation (cysteine-to-glycine substitution at position 10). The theoretical monomer masses are 13,752.89 Da (wild-type TTR), 13,871.89 Da (CysTTR), and 13,706.90 Da (C10G TTR) (Table 1). These species are evident in the deconvoluted mass spectra (Figure 3A). Although these proteoforms differ by single modifications, they may adopt subtly different structures that can contribute to amyloidosis,^{37, 38} making structural characterization by top-down cleavable crosslinking particularly valuable. Proteoform identities were confirmed prior to crosslinking. As shown in Table 1, experimental masses agree with literature values within 5.0 ppm. Proteoform identifications of C10G TTR and CysTTR were further validated by MS/MS fragmentation (Figures S2 and S3). Oxidation products, especially of the wild-type TTR, were identified both before (Figure 3A) and after (Figure 3B) crosslinking. Using a top-down approach allows specific interrogation of the unoxidized form or any of the oxidized form, showing the enhanced structural selectivity obtainable by this method.

Table 1:

Calculated and measured masses of transthyretin proteoforms, with mass measurement accuracy, and the ratio of intensities of uncrosslinked to crosslinked transthyretin for each proteoform. Data in the table are extracted from the deconvoluted mass spectra in Figure 3.

	Calculated Mass (Da)	Measured Mass (Da)	PPM Error	Ratio of Uncrosslinked to Crosslinked Peak Intensity
WT TTR	13,752.8877	13,752.9089	1.54	38:1
sC10/pS52 TTR	13,832.8446/13,832.8541	13,832.8924	3.46/2.77	15:1
C10G TTR	13,706.9000	13,706.8835	1.20	22:1
CysTTR	13,871.8918	13,871.8953	0.25	31:1

Figure 3.

(A) Deconvoluted mass spectrum showing wild-type, C10G, cysteinylated, and S-sulfonylated or phosphorylated transthyretin proteoforms. (B) Deconvoluted mass spectrum of DSBU-crosslinked products (oxidation and small-molecule loss products not annotated). (C) Ratio of intensities of uncrosslinked to crosslinked transthyretin for each proteoform.

Intact DSBU crosslinks (196.08 Da) were identified in both CysTTR and C10G TTR (Figure 3 B). The WT TTR and CysTTR had the highest ratio of uncrosslinked to crosslinked peak intensity (Table 1), which may suggest that these proteoforms are less reactive and/or more extended. The phosphorylated or sulfonated proteoform(s) as well as the C10G TTR mutant showed more efficient crosslinking, indicating greater reactivity and/or more compact structures. Upon fragmentation, the characteristic 85 Da and 111 Da cleavage products were observed and localized. Electron-transfer/higher-energy collision dissociation (EThcD) was employed to simultaneously cleave the crosslinker and fragment the protein backbone. Fragmentation was monitored using ExDViewer to identify intact crosslinks (196 Da) and cleaved moieties (85 Da and 111 Da). Crosslink assignment followed the methodology established for cytochrome c, with identifications confirmed or inferred from the presence of 85 Da and 111 Da fragments. Both proteoforms were analyzed in the 12–14+ charge states (fragment identifications combined) due to their high abundance. Notably, crosslinks were unique for each proteoform (Table 2). Complete fragmentation maps and crosslink/label assignments are provided in the Supporting Information (Figures S4–S5). While the cytochrome c data showed a preference for the 85 Da mass addition compared to the 111 Da mass addition in terms of ion abundance, that pattern is not repeated for transthyretin as there is no evidence of a preferential mass addition.

Table 2:

Crosslinking assignments identified in TTR proteoforms.

C10G TTR, Uncleaved	C10G TTR, Cleaved	Cysteinylated TTR, Uncleaved	Cysteinylated TTR, Cleaved
Nt-K9	K9-K35	Nt-K15	K15-K70
K70-K126	K9-K126	K76-K126	K15-K126
	K35-K126

Crosslink Assignment and Structural Interpretation

Unlike cytochrome c, assignment of cleaved crosslinker fragments to specific intact crosslinks in transthyretin was more straightforward due to reduced ambiguity. For C10G TTR, the smallest N-terminal fragment ion bearing intact DSBU (196 Da) resulted from backbone cleavage N-terminal to L12, indicating an intact crosslink N-terminal to this position. The only feasible assignment is an N-terminus to K9 crosslink. Similarly, the smallest C-terminal fragment ion with intact DSBU resulted from cleavage C-terminal to G67, indicating a crosslink involving residues C-terminal to this site, consistent with K70–K126. Analysis of cleaved crosslinker fragments provided complementary information. K126 was identified in both 85 Da and 111 Da fragment maps (via C-terminal fragments generated by cleavage C-terminal to T96 and S100, respectively). Since a lysine residue cannot crosslink to itself, K126 must pair with the other identified sites bearing the complementary cleaved moiety. This supports crosslinks between K9–K35, K9–K126, and K35–K126. Analogous analysis of CysTTR yielded intact crosslink assignments of N-terminus–K15 and K76–K126, while cleaved DSBU data suggested K15–K70 and K15–K126.

The AlphaFold database model AF-P02766-F1-v6^{39, 40} of transthyretin shows that residues 1–9 are estimated with a low confidence level, suggesting the presence of a dynamic or disordered region.⁴¹ There is a crosslink present specifically for the cysteinylated proteoform of transthyretin between the N-terminus and K15, suggesting that the first 9 residues may interact more closely with the first portion of the protein that is well-folded. This change can suggest that differences in structure are related to the local microenvironment of this region between the cysteinylated C10 and C10G mutant, reflecting their different side chain chemistries. For the C10G TTR, the presence of crosslinks between K9 and K35/K126 suggests interactions between the putatively disordered N-terminal region and both a distal internal region and the C-terminus instead of with the more folded region as in the cysteinylated proteoform. The linking of K35 to K126 is also unique for the C10G mutant which can suggest a subpopulation with a potentially flexible C-terminus, as K35 is not predicted to be proximal to K126. The distinct crosslinking patterns between C10G TT and CysTTR are shown in Table 2, indicating potentially subtle structural differences between these proteoforms.

Conclusions

This study demonstrates the successful integration of cleavable crosslinking with top-down mass spectrometry. We found that cleavage allowed identification of crosslinks not observed in the intact form. The approach was validated using cytochrome c as a model protein and subsequently applied to transthyretin, a biologically relevant protein exhibiting multiple endogenous proteoforms. Distance restraint analysis using PyMOL confirmed that crosslink assignments in cytochrome c were consistent with published distances, supporting the feasibility of this methodology. Application to transthyretin revealed distinct crosslinking patterns between proteoforms, suggesting subtle structural differences. The ambiguity inherent in pairing cleaved crosslinker halves, particularly when multiple modification sites are identified, can potentially limit the utility of cleaved DSBU fragments as unambiguous distance restraints. However, this limitation does not diminish the structural value of the data. Cleaved crosslinker modifications function effectively as covalent labels that report on solvent-accessible lysine residues, since reactivity with the crosslinking reagent inherently indicates surface exposure. Thus, even when definitive pairing of 85 Da and 111 Da moieties is not possible, as was the case for cytochrome c, cleavable crosslinking provides complementary structural information by mapping the solvent-accessible surface of proteins analyzed by top-down mass spectrometry.

Figure 4:

The figure above shows the structure of transthyretin from the AlphaFold database. Crosslinks listed in table 2 are shown within this figure, with red lines representing the crosslinks present in C10G TTR (both uncleaved and cleaved). The green dashed lines represent the crosslinks present in CysTTR (both uncleaved and cleaved).

Highlights.

• Cleavable crosslinking reveals unique crosslinks in top-down mass spectrometry.

• Cytochrome c validation shows added crosslinks and better sequence coverage.

• Transthyretin proteoforms show distinct crosslink patterns and structures.

• Crosslinking detects subtle differences missed by collision cross-section tests.

• Cleaved crosslinkers act as covalent labels for solvent-accessible residues.

Appendix A. Supplementary data

The following is the Supplementary material related to this article. (S1) Sequence map of unreacted transthyretin with C10G point mutation for charge states 12–14+. (S2): Sequence map of unreacted transthyretin with cysteinylation at glycine 10 for charge states 12–14+. (S3): Sequence map and crosslink location identification of crosslinked transthyretin with C10G point mutation. (S4): Sequence map and crosslink location identification of crosslinked transthyretin with cysteinylation at glycine 10.