Post-translational modifications need extra evidence

Allowing phosphorylation in a search explains a mass shift of about eighty daltons. It does not show that a kinase put the phosphate there. The same gap sits between every variable modification and a biological sentence. This glossary is about that gap: fixed versus variable modifications, neutral loss, site probability, and the extra evidence a residue claim needs. How the unmodified peptide is identified at all is bottom-up proteomics in plain language. How the tube chemistry creates some of these masses is preparing peptides for mass spectrometry.

Fixed rules, variable guesses, and a larger search

A fixed modification is applied to every instance of a residue. Carbamidomethyl on cysteine is the usual example after iodoacetamide alkylation. The search does not consider unmodified cysteine. That is correct only when the alkylation actually went to completion. If it did not, the cysteine peptides are missing rather than reported as half-modified, unless you also allow the other form at a cost in search space.

A variable modification may be present or absent. Oxidation of methionine, phosphorylation of serine, threonine and tyrosine, acetylation of protein N-termini or of lysine, and deamidation of asparagine and glutamine are the common requests. Each one asks the engine to try the peptide with and without the extra mass, and with the mass on each allowed residue. Two variable modifications on one peptide are not twice the work in a casual sense. They multiply the candidates. Three of them, unrestricted, can dominate the search and push true unmodified matches down the score list. The false-discovery rate, estimated with decoys in that same enlarged space, often looks worse or becomes miscalibrated if the tool was not designed for the expansion. Add modifications in a set you can justify.

The Proteomics Standards Initiative discusses how modification identifications ought to be reported. A report that says "variable modifications: default" is not that reporting. Name each mass and each residue.

Fragments, neutral loss, and site probability

Localisation is a fragment problem. The precursor mass says the peptide carries the extra mass somewhere. The fragment ions have to say which residue. If two serines sit close together and the fragments do not include an ion between them, both sites remain possible. A localisation probability, sometimes called a site probability, compresses that fragment evidence into a number. Set the minimum you will accept before you look at the hits. Report the modified peptide without a residue number when the probability fails. Do not print the higher-scoring serine as if the spectrum had chosen it cleanly.

Neutral loss is a fragment behaviour, not a second modification. Phosphoserine and phosphothreonine often lose phosphoric acid in the mass spectrometer, a drop of about ninety-eight daltons from the fragment. Seeing that loss supports the idea that a phosphate, rather than some other eighty-dalton shift, is present. It still may not say which serine lost it. Sulphur-containing side reactions and other modifications have their own losses. Match the loss to the chemistry you searched. Do not treat every neutral loss as phosphorylation.

Sequence context lives in UniProt and the NCBI Protein database. A site that is annotated there is a hypothesis you may test. It is not confirmation of your spectrum. PeptideAtlas can show whether modified forms of a peptide have been observed elsewhere. Observation elsewhere is still not observation in your tube.

Shifts the tube is good at making

Methionine oxidation increases when peptides wait in air, in warm solvent, or through a long digestion. Search it so the oxidised and unoxidised forms are not both invisible, and keep the biological claim empty unless an experiment was built to measure oxidation with controls for handling.

Deamidation of asparagine, especially in asparagine-glycine pairs, accumulates with time, alkaline pH and heat. It is a common artefact of preparation and of storage. It can also be biological. The spectrum does not know which. Age-matched handling of the comparison groups matters more than the mass itself.

Carbamylation appears when urea has been warm enough to form isocyanate. That is a digest artefact. If you did not use urea, do not add carbamylation to tidy unexplained masses. Find the real shift.

N-terminal acetylation and removal of the initiator methionine are frequent protein-processing events, and they are also easy to over-call if every peptide N-terminus is allowed to be acetylated. Restrict acetylation to protein N-termini unless you have a reason to search lysine acetylation, and then expect to need enrichment or an orthogonal check for the lysine claim.

Phosphorylation is often biological and often scarce. Phosphatases remove it if the lysis did not stop them. Enrichment with a metal-affinity or titanium-dioxide class of chemistry is a common way to see it at all. Enrichment changes the sample. Report it. A phosphopeptide found without enrichment in a whole lysate is possible for an abundant site and surprising for a rare one. Either way, localisation still needs fragments.

Artefactual phosphorylation is less common than artefactual oxidation, and it is not impossible. A buffer that carried phosphate, or a wrong mass assignment, can imitate the shift. The fragment series and the neutral loss are the checks. A single precursor mass is not.

When the extra evidence is still missing

A common over-call is a phosphosite copied from the highest-scoring arrangement in a two-serine peptide. Keep the peptide. Drop the residue number. Another over-call is a long list of deamidation events interpreted as a regulatory programme. Check whether both conditions waited the same number of hours. If the treated samples sat in the fridge while controls were digested fresh, the modification table is the waiting time.

Enrichment failure looks like biology too. A titanium or metal enrichment that was not selective returns unmodified peptides and a few acidic ones. Judge selectivity from the fraction of modified matches, and compare with a blank enrichment. If the fraction is poor, do not tell a kinase story.

Multiple modifications on one scarce peptide, each with a weak localisation, are the place the false-discovery rate is most stressed. Prefer a second search that allows one modification class. Concordance between the two searches is more persuasive than a single permissive run.

Research limits

Modification calls are research observations. They are not a diagnosis and not a map of signalling by themselves. A kinase motif on a sequence is a reason to look, not a result. Biosafety of the cells is an institutional matter. Stopping phosphatases with an inhibitor does not change the biosafety level of the culture.

Heat, delay, and a phosphatase that kept working

Warm urea, a digest that cooled and reheated after a power cut, and a lysate that waited on a humid bench all push oxidation and deamidation. Snap-freeze or start the inhibitor and denaturant steps on the schedule the protocol gives. If the freezer warmed overnight, do not search the recovered lysates for "stress oxidation" and treat the hits as physiology. For phosphorylation, a lysis delay without phosphatase inhibitors is enough to empty real sites. The extra evidence then includes the clock, not only the spectrum. Write the delay next to the site table so a later reader can see it.

What a modification question needs

Say which modification class you mean, whether enrichment is required, which residues are allowed, and what localisation rule you will accept. Say whether oxidation and deamidation are being searched only so that peptides are found, or because they are the claim. Those are different enquiries.

The protein identification by LC-MS/MS reference covers naming the modified peptide. The shotgun discovery proteomics reference is the survey in which an unrestricted modification list does the most harm. The differential abundance reference is the comparison you must not build on an artefact that tracks the handling. Describe the modification and the evidence you will require through the quote request. The pages frame that discussion. They do not mean an enrichment has already been run.