assay · deep · partly written
Crosslinking mass spectrometry (XL-MS)
Crosslinking mass spectrometry glues together protein residues that sit within reach of a chemical linker of known length, then uses a mass spectrometer to read which two residues were joined.
Also called: XL-MS, CLMS, chemical crosslinking MS
The short version
XL-MS glues two residues together with a molecule of known length, then reads which two got glued. A hit says these two residues were within reach, about 30 A between alpha-carbons for BS3. The hard part is trust: a hit needs two correct peptides, so errors multiply and grow as spectra are merged into residue pairs (5% became 8.4% in one yeast dataset). 123
Reported, not yet replicated
The ruler (chemistry) more
BS3 has a sulfo-NHS ester at each end, each reacting with a primary amine at pH 7.0 to 9.0, with an 11.4 A, 8-atom spacer (water soluble, not cleavable). The usable limit is longer than the spacer because lysine side chains are flexible: Merkley and colleagues mined molecular dynamics of 807 proteins and concluded a crosslink fits a structure if the lysine nitrogen distance is under the linker length plus 11.3 A, a 30.4 A alpha-carbon distance for DSS or BS3 (abstract only). The ruler is an upper bound with tolerance, never a measurement. 32
Reported, not yet replicated
The error bar (decoys) more
The software also searches a deliberately false database; matches to it are false by construction and estimate the random matches among real hits. Crosslinks break the simple version three ways. (1) Two peptides must both be right, so the random space is quadratic and target-target, target-decoy and decoy-decoy matches are counted. (2) True hits repeat while false ones stay alone, so the false share rises as spectra are merged: 5% at spectrum level gave 5.8% at peptide pairs and 8.4% at residue pairs. (3) Self-links and between-protein links need separate error rates: random matches were 10.6 times likelier among heteromeric links in E. coli lysate. 415
Reported, not yet replicated
Where it breaks more
Validating only against known structures hides error: random false hits usually map to one end of a structure and are ignored, and three K562 datasets of very different quality looked the same by that test (P > 0.85), while the fraction from known interactions was 55.5% for a stringent set against 4.4% at a 10% FDR. In a synthetic library with known answers (95 peptides, 426 theoretical crosslinks), true error at a nominal 5% ranged from 2.4% to 32% for one search engine depending on the validation strategy, and reached 44% for another engine with a cleavable linker and no extra score cut-offs. Proteins seen only in heteromeric links had lower median abundance and were enriched among false matches. 675
Reported, not yet replicated
Price and route deep
The only price found is a university core: $420 for 3 injections or $630 for 5, as of 2026-10-28, purified complexes only. No CRO or cloud-lab price was found, so none is given. 8
Reported, not yet replicated
Question answered
Readout physical
Mass of peptide pairs. Crosslinked proteins are cut with trypsin; two peptides joined by a linker travel together as one heavier ion. The instrument fragments it and software searches a sequence database for the pair of peptides and the linkage position that explain the spectrum. Each answer is a crosslink-spectrum match (CSM), later merged into peptide pairs, residue pairs and protein pairs. 4
Reported, not yet replicated
Not measured
Binding, affinity or kinetics. A residue pair says two residues were within linker reach at some moment, which is an upper bound with a tolerance and not a distance measurement. Dynamics are invisible, and a link longer than the tolerance is not automatically false. Self-links within one protein and links between copies of the same protein are usually indistinguishable without special design. 24
Reported, not yet replicated
Setting
Where in cascade
Sample prep
A purified complex, a lysate or cells are treated with the linker (BS3, whose two sulfo-NHS ester ends react with primary amines at pH 7.0 to 9.0), the reaction is quenched, proteins are digested with trypsin and the peptides are read by tandem mass spectrometry. BS3 is water soluble and not cleavable; MS-cleavable linkers such as DSSO break inside the instrument. 39
Reported, not yet replicated
Conditions
Linker length sets the ruler: BS3 has an 11.4 A spacer, but the accepted limit between alpha-carbons is about 30 A (30.4 A from molecular dynamics of 807 proteins) because lysine side chains are flexible. Abundant proteins are favoured, and low-abundance proteins seen only in heteromeric links were enriched among false matches in E. coli lysate. Which linker and which FDR level is reported changes the error. 325
Reported, not yet replicated
Throughput
No throughput figure was found in an opened source. Class is set to low as an inference: the one price found is per purified-complex sample of 3 to 5 injections, and lysate-scale studies run as a whole experiment, not as a plate. 8
Our inference
Cost
Per sample
420 to 630 USD per purified-complex sample (3 to 5 injections) (University of Texas Southwestern proteomics core rate card for crosslinking MS: $420 for 3 injections, $630 for 5; DSS, DMTMM, BS3 and DSSO only; purified complexes only. Academic core rate, not a CRO or commercial price. Other universities and companies pay a higher tier at that core.; as of 2026-10-28)8
Engineerable variants
MS-cleavable linkers (for example DSSO)
The linker breaks inside the mass spectrometer. In E. coli lysate at 5% CSM-level FDR, DSSO gave 7,316 heteromeric CSMs against 3,308 for BS3; the authors attribute the gain to better peptide backbone fragmentation. Treating DSSO data as non-cleavable cut heteromeric identifications by 74%, so cleavability must be modelled. 9
Reported, not yet replicated
Separate error rates for self and heteromeric links
A random decoy match was 10.6 times more likely among heteromeric links than self-links in E. coli lysate, so a single pooled FDR over-credits the between-protein links that matter for interactions. 5
Reported, not yet replicated
Use cases
E. coli lysate interaction network (Lenz 2021)
Crosslinking MS of E. coli lysate, with error estimated per level, gave a network of 590 protein-protein interactions at 1% FDR. A naive 5% CSM-level FDR would have given about 35 to 36% false PPIs; separating heteromeric links gave about 15 to 16%, and merging CSMs into PPIs before FDR gave 6.6% (BS3) and 4.9% (DSSO). This is a methods result, not a drug programme. 5
Reported, not yet replicated
Yeast RNA polymerase II crosslinks (Fischer and Rappsilber 2017)
With BS3 on yeast Pol II, a nominal 5% FDR at spectrum level became 5.8% at peptide-pair level and 8.4% at residue-pair level. Filtering earlier (6% and 10.5% cut-offs before a 5% residue-pair cut-off) gave 144 residue pairs instead of 108 (+33%), and 33 of the 36 extra pairs matched the expected distance in PDB 1WCM. Methods result. 1
Reported, not yet replicated
Data availability
UT Southwestern proteomics core (university rate card) cro
Crosslinking MS on purified complexes with DSS, DMTMM, BS3 or DSSO: $420 for 3 injections, $630 for 5. 8
Caveats. Purified complexes only, so it does not cover lysate or in-cell studies. University core tiers: other universities and companies pay more. No commercial CRO price was found. Rate is as of 2026-10-28.
Price. 420 to 630 USD per sample (3 to 5 injections) (UTSW core rate card, as of 2026-10-28; as of 2026-10-28)8
Reported, not yet replicated
PRIDE and jPOST proteomics repositories public
The community white paper names PRIDE and jPOST as the partner repositories for crosslinking MS data; the Fischer and Rappsilber yeast Pol II data are described there as deposited as PXD004749. 41
Caveats. The repository accession and licence were not opened in this run, so reuse terms are unchecked.
Unverified: from our own report, source not opened
Teach this
In one sentence
XL-MS chemically staples nearby residues together and reads the staples by mass spectrometry, so each hit is a ruler reading of about 30 A with an error bar that depends on the level you report at. 321
Reported, not yet replicated
The five-minute version
Add a linker of known length (BS3 is 11.4 A) to a complex, lysate or cells. It joins amines that are close. Digest, run the mass spectrometer, and search for pairs of peptides that explain each spectrum. Control error with a decoy database. Because two peptides must both be right, and because false hits stay alone while true ones repeat, the false share rises when you merge spectra into residue pairs and protein pairs (5% to 8.4% in one dataset), so always ask at which level the FDR was set. 413
Reported, not yet replicated
What a listener will ask
If the linker is 11.4 A, why is 30 A the limit?
Does a crosslink mean two proteins bind?
It means two residues were within reach at some moment, in a lysate or cell; it carries no affinity or kinetics, and a false-match rate still applies.
Why not just check against known structures?
That check only scores links with both ends in a structure and drastically underestimates error; ground-truth peptide libraries or unrelated-proteome decoys do better.
The common wrong idea
A 5% FDR means 5% of the reported residue pairs are wrong.
5% set at spectrum level became 8.4% at residue-pair level in one yeast dataset, because false hits stay alone while true hits repeat. 412376598
Reported, not yet replicated
The one citation to hand over: 1
Still open here
- How often do in-cell XL-MS links come from crosslinking artefacts (proteins that were not neighbours)? No source was opened.
- The decoy FDR formulas (target-target, target-decoy, decoy-decoy counts) were images in the source and could not be read; the logic is described, the equations are not given.
- Kao 2011 (the MS-cleavable DSSO linker) was not opened, and figures for integrative models of the nuclear pore (crosslink counts, PDB-IHM entries) were search-summary only and are left out.
- The 24.4 A alpha-carbon cut-off sometimes quoted for BS3 was not found in an opened source and is not used.
- No before-and-after (breakthrough watch) is written: no opened primary source supports the four steps. No capital entries.
- No throughput or time per run was found in an opened source; throughput class is an inference from the core price rule.
- No CRO or cloud-lab price for XL-MS was found; the only price is one university core.
Not yet written: Time per run, Before after, Capital, ML role.
Linked from
- Genetic interaction and epistasis screens
- Isothermal titration calorimetry (ITC)
- Structural evidence of complexes (crystallography and cryo-EM)
- Surface plasmon resonance (SPR)
- Pathways and interaction partners
- What would a CRO or cloud lab charge for protein-interaction assays?
- A target with no known partners
Sources
- Fischer L, Rappsilber J. Quirks of Error Estimation in Cross-Linking/Mass Spectrometry. Analytical Chemistry (2017). paper; opened 2026-11-17 all records citing this
- Merkley ED, Rysavy S, Kahraman A, et al.. Distance restraints from crosslinking mass spectrometry: mining a molecular dynamics simulation database to evaluate lysine-lysine distances. Protein Science (2014). paper; opened 2026-11-17 all records citing this
- BS3 (bis(sulfosuccinimidyl)suberate) product page. Thermo Fisher Scientific. vendor-page, not peer reviewed; opened 2026-11-17 all records citing this
- Leitner A, et al.. Towards Increased Reliability, Transparency and Accessibility in Crosslinking Mass Spectrometry. arXiv (2020). preprint, not peer reviewed; opened 2026-11-17 all records citing this
- Lenz S, Sinn LR, O'Reilly FJ, Fischer L, Wegner F, Rappsilber J. Reliable identification of protein-protein interactions by crosslinking mass spectrometry. Nature Communications (2021). paper; opened 2026-11-17 all records citing this
- Yugandhar K, Wang TY, Wierbowski SD, Shayhidin EE, Yu H. Structure-based validation can drastically underestimate error rate in proteome-wide cross-linking mass spectrometry studies. Nature Methods (2020). paper; opened 2026-11-17 all records citing this
- Beveridge R, Stadlmann J, Penninger JM, Mechtler K. A synthetic peptide library for benchmarking crosslinking-mass spectrometry search engines for proteins and protein complexes. Nature Communications (2020). paper; opened 2026-11-17 all records citing this
- UT Southwestern Proteomics Core: rates. UT Southwestern. organisation-page, not peer reviewed; opened 2026-11-17 all records citing this
- Kolbowski L, Lenz S, Fischer L, et al.. Improved Peptide Backbone Fragmentation Is the Primary Advantage of MS-Cleavable Crosslinkers. Analytical Chemistry (2022). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (1)
This record has been changed 1 time, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.
- Throughput class
- low
- Emerging
- no
- Status
- partly written
- Last updated
- 2026-11-17
- Record
- content/assays/crosslinking-mass-spectrometry.json