On the Reproducibility of Label-Free Quantitative Cross-Linking/Mass Spectrometry
Quantitative cross-linking/mass spectrometry (QCLMS) is an emerging approach to study conformational changes of proteins and multi-subunit complexes. Distinguishing protein conformations requires reproducibly identifying and quantifying cross-linked peptides. Here we analyzed the variation between multiple cross-linking reactions using bis[sulfosuccinimidyl] suberate (BS3)-cross-linked human serum albumin (HSA) and evaluated how reproducible cross-linked peptides can be identified and quantified by LC-MS analysis. To make QCLMS accessible to a broader research community, we developed a workflow that integrates the established software tools MaxQuant for spectra preprocessing, Xi for cross-linked peptide identification, and finally Skyline for quantification (MS1 filtering). Out of the 221 unique residue pairs identified in our sample, 124 were subsequently quantified across 10 analyses with coefficient of variation (CV) values of 14% (injection replica) and 32% (reaction replica). Thus our results demonstrate that the reproducibility of QCLMS is in line with the reproducibility of general quantitative proteomics and we establish a robust workflow for MS1-based quantitation of cross-linked peptides.
KeywordsQuantitation Cross-linking Human serum albumin Label-free Mass spectrometry Reproducibility
Cross-linking/mass spectrometry (CLMS) has become a powerful tool aiding the structural analysis of proteins and their complexes [1, 2, 3, 4, 5] since its onset almost two decades ago [6, 7]. Reaction with a cross-linker converts 3D proximity of amino acid residues into covalent bonds. The bridgeable distance between residues depends on the cross-linker used. Bis[sulfosuccinimidyl] suberate (BS3), one of the most commonly used reagents, links residues up to 25–30 Å apart (Cα-Cα distance) . Following proteolytic digestion of the proteins, cross-linked peptides are identified using liquid chromatography-mass spectrometry (LC-MS) and database search.
Previous studies have used CLMS to investigate the structures of single proteins , multi-protein complexes , and protein–protein interaction networks [10, 11]. The proteins in these studies are often undergoing dynamic conformational changes, which are difficult to determine and visualize by knowing only the sites of cross-linking. For this, understanding the dynamics through relative abundances of certain cross-linked residue pairs is required by adding quantitation to CLMS pipelines. In mass spectrometry-based proteomics there are two broad quantitative strategies, label-free and labeled approaches, both of which are suitable for CLMS. A previous study by Huang 2006  using an 18O labeling-based QCLMS approach had several drawbacks that prevented widespread use of this approach, including incomplete labeling and inadequate software for data analysis. Fischer et al. 2013  overcame these hurdles by using an isotope-labeled cross-linker and developing the software tool XiQ, which combined the accuracy of manual peak validation with the convenience of automated quantitation. Since then, several software packages became available to analyze QCLMS data [14, 15]. Although isotope labeling-based QCLMS has been used successfully in several studies [14, 16, 17, 18, 19], it suffers from the usual limitations that often come with the experimental design of labeling approaches: cost of isotope-labeled reagents (which can be expensive), complex sample preparation, and reduced data coverage [20, 21]. In contrast, label-free quantitation can avoid these pitfalls and there are no limits to the numbers of samples that can be compared. Advantages of label-free quantitation were presented recently with an MS2-based QCLMS workflow using Skyline . A general caveat of label-free approaches is that samples are processed separately, which can result in technical biases during sample preparation . As the sample preparation procedure of cross-linking is more elaborate than in normal proteomics, one might expect a larger variance.
Here we investigate the reproducibility of label-free QCLMS. We determined the variation introduced during sample preparation and contrast this with the variation between multiple injections during LC-MS acquisition. As a model system, we cross-linked human serum albumin (HSA) using bis[sulfosuccinimidyl] suberate (BS3) and we adapted Skyline into a workflow for semi-automated label-free QCLMS.
HSA was purchased from Sigma Aldrich (St. Louis, MO, USA). The cross-linker BS3 was purchased from Thermo Scientific Pierce (Rockford, IL, USA).
Ten cross-linking reactions were performed in parallel as follows: purified human serum albumin (40 μg; 2 μg/μL) in cross-linking buffer (20 mM HEPES-KOH, pH 7.5, 20 mM NaCl, 5 mM MgCl2,) was mixed with BS3 (160 μg, 30 μg/μL in cross-linking buffer) and cross-linking buffer (14.6 μL), to a total reaction volume of 40 μL (1 μg/μL protein concentration) with a protein to cross-linker mass ratio of 1:4. After 1.5 h incubation on ice, the reaction was stopped using 5 μL saturated ammonium bicarbonate (~2.5 M) for 30 min at room temperature. Forty μg of cross-linked HSA from each reaction were subjected to SDS-PAGE and protein bands were visualized using Coomassie staining. Cross-linked HSA monomer bands were excised for digestion.
Sample Preparation for Mass Spectrometric Analysis
LC-Mass Spectrometric Analysis
LC-MS/MS analysis was performed using Orbitrap Fusion Lumos (Thermo Fisher Scientific, CA, USA) with a “high/high” acquisition strategy. The peptide separation was carried out on an EASY-Spray column (50 cm × 75 μm i.d., PepMap C18, 2 μm particles, 100 Å pore size, Thermo Fisher Scientific, Germany). Mobile phase A consisted of water and 0.1% v/v FA and mobile phase B consisted of 80% v/v ACN and 0.1% v/v FA. Peptides were loaded onto the column with 2% buffer B at 0.3 μL/min flow rate and eluted at 0.25 μL/min flow rate with following gradient: 150 min linear increase from 2% to 40% mobile phase B followed by 11 min increase from 40% to 95% mobile phase B. Eluted peptides were sprayed directly into the mass spectrometer and analyzed using a data-dependent acquisition (DDA) mode. In each 3 s acquisition cycle, precursor ions were detected in the Orbitrap with resolution 120,000 and m/z range 400–1600. Ions with charge states from 3+ to 7+ were selected for fragmentation. The selection priority was set to first lowest charge and then highest intensity. Selected ions were isolated in the quadrupole with a window size of m/z 2. The isolated ions were fragmented by high energy collision dissociation (HCD) and analyzed with resolution 30.000 in Orbitrap. Dynamic exclusion was enabled with the exclusion duration set to 60 s and exclusion mass tolerance was set to 10 ppm.
Identification of Cross-Linked Peptides
The raw mass spectrometric data files were processed into peak lists using MaxQuant  (v. 22.214.171.124). “FTMS top peaks per 100 Da” was set to 20, “FTMS de-isotoping” box was unticked, and all other parameters were set to default (Figure 1b). The subsequent database search was conducted using Xi  against the sequence of HSA (UniProt ID: P02768) with the reversed HSA sequence as decoy. The following search parameters were used: MS accuracy: 6 ppm, MS/MS accuracy: 20 ppm, enzyme: trypsin, missed cleavages: 4, cross-linker: BS3, fixed modification: carbamidomethylation on cysteine, variable modification: oxidation of methionine and modification by BS3 with the second NHS ester hydrolyzed or aminated. The BS3 reaction specificity was assumed to be at lysine, serine, threonine, tyrosine, and the N-termini of proteins. The data have been deposited to the ProteomeXchange  Consortium via the PRIDE  partner repository with the data set identifier PXD007250 For all identified cross-links that were auto-validated by Xi Software, the Cɑ-Cɑ distance between cross-linked residue pairs was measured in the crystal structure of HSA (PDB: 1AO6 chain A). Residue pairs with distance ≥ 30 Å and cross-links matched to decoys were excluded from subsequent quantitation using Skyline.
Creation of Spectral Library for Autovalidated Cross-Links and Quantitation Using Skyline
Results and Discussion
To assess the reproducibility of quantitative CLMS in a label-free experiment, we measured cross-linked HSA, a well-studied model protein for CLMS , to monitor reproducibility of 10 cross-linking reactions and 10 LC-MS injections of the same sample. HSA was cross-linked in solution using BS3 and digested in gel using trypsin. Unfractionated peptides were analyzed by LC-MS using a “high-high” (Orbitrap MS1 and MS2) acquisition strategy and data-dependent acquisition (DDA).
Identification of Cross-Linked Peptides by Xi
Label-Free Quantification of Cross-Linked Peptides by Skyline
For label-free quantification, we used Skyline . In short, we prepared a spectral library comprising all 196 identified residue pairs (1064 spectra) from the reaction replica experiment and all 180 identified residue pairs (885 spectra) from the injection replica experiment. For each experiment, every identified cross-linked peptide can in principle be quantified across 10 replicas even if it was not identified in all of them by DDA.
Prior to quantitation in Skyline, we created a Skyline input file (.ssl file) for each experiment using an in-house script. The .ssl file contains the following information for each identified cross-linked peptides: the assigned mzML file, the scan number, charge state, sequence (including modifications), score type, and score. In this file, the sequence of each cross-linked peptide has been converted into a linear representation with an additional modified lysine residue connecting two linked peptides A and B (Kxlink, Figure 2d) , giving rise to an identical mass to the original cross-linked form (Figure 2e). Skyline used the .ssl file and the assigned mzML files to create a spectral library using BiblioSpec. Peptide and transition settings had to be defined to explore the library and import peptides that matched the filter settings or the library into the quantitation worksheet.
To assess the reproducibility of peak area after quantitation on unique residue pairs, we calculated the coefficient of variation (CV) of a residue pair as the median CV values of all corresponding cross-linked peptide features. The CV values of quantified features are calculated in Skyline, representing the mean variation between peak areas of all replicas. The higher the value the more variation exists between the peak areas over all replicas. As expected, injection replica resulted in higher reproducibility (CV 14%) than reaction replica (CV 32%) (Figure 5c). Simply injecting 10 times from the same tube carries higher reproducibility than starting 10 cross-link reactions in parallel. There is no general consensus on what CV constitutes a good basis for quantitative statements. However, the results fit into variations observed in other studies [35, 37, 38, 39, 40, 41, 42, 43]. Perrin et al. 2013  assessed quantitative label-free approaches of linear peptides using cerebrospinal fluid in terms of injection reproducibility and inter-individual variation. Most of the quantified proteins showed a very low coefficient of variation (<5%) for injection replica, which is remarkably low, and a much higher variance across samples from different individuals (48%). Our lower injection reproducibility might be explained by having many modified cross-linked peptides (methionine oxidation and alternative cross-link products) and early eluting peptides, all of these being a source of technical variability. Kramer et al. 2015  reported an injection variance of 10% and an inter-assay variability of 16% using label-free quantification of proteins and data-independent acquisition (DIA). Lai et al. 2015  suggested to use a CV of 30% as threshold for injection replica to get reproducible quantifications using label-free approach and data-dependent acquisition (DDA) strategy.
Finally we investigated reproducibility (CV) in relation to median peak area of residue pairs (Figure 5d). Quantitation reproducibility is linked inversely with peak intensity, as one would expect. Reaction replicas show less reproducibility than injection replicas, but the intensity dependence of reproducibility remains present. Lowering abundance increases variation and reduces reproducibility of quantitation. One should therefore inject as much material as feasible. In summary, the reproducibility of quantitative CLMS and studies with linear peptides are very comparable.
In this study, we demonstrate that cross-linked residue pairs are identified with reproducibility and saturation characteristics that resembles random sampling in standard shotgun proteomics . Additional injections improve the number of identifications but also increase variability between runs caused by random sampling. Hence, a reliable quantitation procedure when seeking quantitative information is needed. We described a quantitative cross-linking workflow based on DDA and label-free quantitation in Skyline. This allows leveraging of information from multiple injections due to matching features and identifications between runs. We observe that label-free quantitation in cross-linking is in line with the reproducibility of studies using linear peptides. Quantitative cross-linking has already proven its potential for structural and mechanistic studies of proteins and reliable label-free quantitation of cross-linked residue pairs now offers a set of new avenues and experimental designs.
This work was supported by the Wellcome Trust (103139, 108504) and the DFG (RA 2365/4-1). The Wellcome Trust Centre for Cell Biology is supported by core funding from the Wellcome Trust (203149).
Compliance with ethical standards
The Authors declare no competing financial interest.
- 7.Young, M.M., Tang, N., Hempel, J.C., Oshiro, C.M., Taylor, E.W., Kuntz, I.D., Gibson, B.W., Dollinger, G.: High throughput protein fold identification by using experimental constraints derived from intramolecular cross-links and mass spectrometry. Proc. Natl. Acad. Sci. USA. 97, 5802–5806 (2000)CrossRefGoogle Scholar
- 8.Belsom, A., Schneider, M., Fischer, L., Brock, O., Rappsilber, J.: Serum albumin domain structures in human blood serum by mass spectrometry and computational biology. Mol. Cell. Proteom. 25, 663–682 (2015)Google Scholar
- 9.Chen, Z.A., Jawhari, A., Fischer, L., Buchen, C., Tahir, S., Kamenski, T., Rasmussen, M., Lariviere, L., Bukowski-Wills, J.-C., Nilges, M., Cramer, P., Rappsilber, J.: Architecture of the RNA polymerase II-TFIIF complex revealed by cross-linking and mass spectrometry. EMBO J. 29, 717–726 (2010)CrossRefGoogle Scholar
- 10.Herzog, F., Kahraman, A., Boehringer, D., Mak, R., Bracher, A., Walzthoeni, T., Leitner, A., Beck, M., Hartl, F.-U., Ban, N., Malmstrom, L., Aebersold, R.: Structural probing of a protein phosphatase 2A network by chemical cross-linking and mass spectrometry. Science. 337, 1348–1352 (2012)CrossRefGoogle Scholar
- 14.Walzthoeni, T., Joachimiak, L.A., Rosenberger, G., Röst, H.L., Malmström, L., Leitner, A., Frydman, J., Aebersold, R.: xTract: software for characterizing conformational changes of protein complexes by quantitative cross-linking mass spectrometry. Nat. Methods. 12, 1185–1190 (2015)CrossRefGoogle Scholar
- 15.Chen, Z.A., Fischer, L., Cox, J., Rappsilber, J.: Quantitative cross-linking/mass spectrometry using isotope-labeled cross-linkers and MaxQuant, http://biorxiv.org/content/early/2016/05/30/055970, (2016)
- 18.Ioannou, M.S., McPherson, P.S.: Rab13 and the regulation of cancer cell behavior. J. Biol. Chem. 3, jbc.R116.715193 (2016)Google Scholar
- 21.Megger, D.A., Pott, L.L., Ahrens, M., Padden, J., Bracht, T., Kuhlmann, K., Eisenacher, M., Meyer, H.E., Sitek, B.: Comparison of label-free and label-based strategies for proteome analysis of hepatoma cell lines. Bioch. Biophys. Acta – Proteins and Proteomics. 1844, 967–976 (2014)CrossRefGoogle Scholar
- 22.Chavez, J.D., Eng, J.K., Schweppe, D.K., Cilia, M., Rivera, K., Zhong, X., Wu, X., Allen, T., Khurgel, M., Kumar, A., Lampropoulos, A., Larsson, M., Maity, S., Morozov, Y., Pathmasiri, W., Perez-Neut, M., Pineyro-Ruiz, C., Polina, E., Post, S., Rider, M., Tokmina-Roszyk, D., Tyson, K., Vieira Parrine Sant’Ana, D., Bruce, J.E.: A general method for targeted quantitative cross-linking mass spectrometry. PLoS One. 11, e0167547 (2016)CrossRefGoogle Scholar
- 27.Vizcaíno, J.A., Deutsch, E.W., Wang, R., Csordas, A., Reisinger, F., Ríos, D., Dianes, J.A., Sun, Z., Farrah, T., Bandeira, N., Binz, P.-A., Xenarios, I., Eisenacher, M., Mayer, G., Gatto, L., Campos, A., Chalkley, R.J., Kraus, H.-J., Albar, J.P., Martinez-Bartolomé, S., Apweiler, R., Omenn, G.S., Martens, L., Jones, A.R., Hermjakob, H.: ProteomeXchange provides globally coordinated proteomics data submission and dissemination. Nat. Biotechnol. 32, 223–226 (2014)CrossRefGoogle Scholar
- 32.Holman, J.D., Tabb, D.L., Mallick, P.: Employing ProteoWizard to convert raw mass spectrometry data. Curr. Protoc. Bioinformatics 1–9 (2014)Google Scholar
- 35.Tabb, D.D.L., Vega-Montoto, L., Rudnick, P.A., Variyath, A.M., Ham, A.-J.L., Bunk, D.M., Kilpatrick, L.E., Billheimer, D.D., Blackman, R.K., Cardasis, H.L., Carr, S.A., Clauser, K.R., Jaffe, J.D., Kowalski, K.A., Neubert, T.A., Regnier, F.E., Schilling, B., Tegeler, T.J., Wang, M., Wang, P., Whiteaker, J.R., Zimmerman, L.J., Fisher, S.J., Gibson, B.W., Kinsinger, C.R., Mesri, M., Rodriguez, H., Stein, S.E., Tempst, P., Paulovich, A.G., Liebler, D.C., Spiegelman, C.: Repeatability and reproducibility in proteomic identifications by liquid chromatography-tandem mass spectrometry. J. Proteome Res. 9, 761–776 (2010)CrossRefGoogle Scholar
- 41.Perrin, R.J., Payton, J.E., Malone, J.P., Gilmore, P., Davis, A.E., Xiong, C., Fagan, A.M., Townsend, R.R., Holtzman, D.M.: Quantitative label-free proteomics for discovery of biomarkers in cerebrospinal fluid: assessment of technical and inter-individual variation. PLoS One. 8, (2013). https://doi.org/10.1371/journal.pone.0064314
- 42.Kramer, G., Woolerton, Y., Van Straalen, J.P., Vissers, J.P.C., Dekker, N., Langridge, J.I., Beynon, R.J., Speijer, D., Sturk, A., Aerts, J.M.F.G.: Accuracy and reproducibility in quantification of plasma protein concentrations by mass spectrometry without the use of isotopic standards. PLoS One. 10, 1–22 (2015)Google Scholar
- 43.Lai, X., Wang, L., Tang, H., Witzmann, F.A.: A novel alignment method and multiple filters for exclusion of unqualified peptides to enhance label-free quantification using peptide intensity in LC-MS/MS. J of Proteome Research. 19, 161–169 (2015)Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.