Tandem Mass Spectrometry (MS/MS) Explained: How Peptides Are Fragmented and Sequenced

Tandem mass spectrometry, written MS/MS, turns a mass spectrometer from a scale into a sequencer. A single mass measurement tells you a peptide weighs what it should; MS/MS isolates that one ion, breaks it apart and weighs the pieces, so the order of the amino acids can be read from the fragment masses.

This guide covers how an MS/MS experiment works, how peptides are fragmented, how b and y ions are named and calculated (with a worked example on a real peptide), de novo versus database interpretation, and what MS/MS cannot settle.

What tandem mass spectrometry is

A mass spectrometer measures ions by their mass-to-charge ratio. Tandem mass spectrometry chains two such measurements with a deliberate reaction in the middle. The first stage (MS1) selects one ion, the precursor; the precursor is fragmented; the second stage (MS2) records the product ions. The spectrum belongs to one molecule rather than to a whole mixture.

In 1978 Richard Yost and Christie Enke described selected-ion fragmentation on a tandem quadrupole, fragmenting the isolated ion in a radio-frequency-only quadrupole. Fred McLafferty’s 1981 Science review argued that two spectrometers in series could pick a target compound out of a complex mixture with high specificity. In 1986 Donald Hunt’s group at the University of Virginia applied the approach to protein sequencing, analysing HPLC fractions holding roughly 10 to 15 peptides each without further purification.

That selectivity is the point for synthetic peptides too: MS/MS can ask what a minor impurity peak is, not just how large it is. The single-stage measurement is covered in reading a peptide mass spectrum, and the chromatography that feeds it in LC-MS explained.

Instruments and scan modes

Tandem in space uses separate analysers in a row, such as the triple quadrupole or the quadrupole/time-of-flight (Q-TOF) hybrid. Tandem in time traps ions and performs selection, fragmentation and analysis in sequence in one device, as ion traps and Fourier-transform instruments do; repeating the cycle gives MS³ and beyond.

There is genuine Canadian history here. A 1997 paper on rapid de novo peptide sequencing with a Q-TOF was a collaboration between the European Molecular Biology Laboratory, Kenneth Standing’s group in physics at the University of Manitoba, and PE SCIEX in Concord, Ontario. It reported tandem spectra with better than 50 ppm accuracy and resolution above 5,000, and sequenced a previously unknown protein. See also our history of Canadian peptide science.

How peptides are fragmented

Collisions: CID and HCD

In collision-induced dissociation the selected ion is accelerated into a neutral gas, and collisions convert kinetic energy into internal energy until a bond breaks. For protonated peptides that is usually a backbone amide bond, which is why CID spectra are dominated by b and y ions. The accepted framework is the “mobile proton” model, refined by Wysocki and colleagues in 2000, which also explains enhanced cleavage next to aspartic acid or protonated histidine. A 2005 review by Paizs and Suhai described fragmentation as several “pathways in competition”, whose energetics decide which peaks appear.

The consequence is that peak heights are uneven, and a spectrum can lack rungs for chemical reasons rather than because the sequence is wrong. Higher-energy C-trap dissociation (HCD), introduced in 2007 on Orbitrap instruments for peptide modification analysis, is also collisional, and its peptide spectra are likewise read mainly through b and y ions.

Electrons: ECD and ETD

Electron capture dissociation was reported in 1998 by Zubarev, Kelleher and McLafferty as a nonergodic fragmentation of multiply charged protein cations. Electron transfer dissociation, published in 2004 by Syka, Coon, Schroeder, Shabanowitz and Hunt, brought similar chemistry to a linear ion trap by transferring an electron from anthracene anions to multiply protonated peptides. Both produce c and z ions, and both need precursors carrying more than one charge.

The methods are complementary: a 2008 study that chose CID or ETD for each precursor by its charge state and m/z identified 53,055 peptides, against 38,293 with CID alone and 39,507 with ETD alone.

b and y ions, and c and z

The names come from a 1984 proposal by Peter Roepstorff and Jan Fohlman, refined by Klaus Biemann in 1990. If the charge stays on the N-terminal piece, the fragment is an a, b or c ion; if on the C-terminal piece, an x, y or z ion. The subscript counts residues from that end, so b₃ holds the first three residues and y₃ the last three.

Two smaller peak types help. Immonium ions flag residues present: Matrix Science’s reference lists strong ones for phenylalanine (m/z 120), histidine (110), tyrosine (136) and tryptophan (159). Neutral-loss satellites sit 17 or 18 units below fragments containing residues that lose ammonia (R, K, N, Q) or water (S, T, E, D). Residue codes are in our amino acid codes reference, and peptide molecular weight shows how the parent mass is built.

Reading a sequence: a worked example

SNAP-8 is a good teaching case: short, with two terminal modifications that change the arithmetic. Its sequence is Ac-Glu-Glu-Met-Gln-Arg-Arg-Ala-Asp-NH₂ (notation explained in peptide sequence notation), and its monoisotopic [M+H]⁺ is m/z 1075.49. N-terminal acetylation adds 42.01057 to every b ion, and C-terminal amidation lowers every y ion by 0.98402.

Three rules make the ladder readable. Gaps are residues: b₃ minus b₂ is 131.04 (methionine), and y₄ minus y₃ is 156.10 (arginine). Complementary pairs add up: the b and y ions from one bond sum to the neutral mass plus two protons, 1076.50 here. Modifications move blocks: if the methionine is oxidised (+15.99492), b₃ to b₇ and y₆ to y₇ shift together while the rest do not, which localises the oxidation.

Real spectra are messier. With two arginines the precursor is often doubly charged, so fragments may also appear at about half the m/z. Glutamine (128.05858) and lysine (128.09496) differ by only 0.036, so mass accuracy decides whether a 128 gap can be assigned. Interpretation is testing a hypothesis against evidence.

De novo sequencing vs database search

In a database search, software takes the precursor mass, retrieves database sequences that fit it, predicts their fragments and scores the match. SEQUEST (Eng, McCormack and Yates, 1994) searched within about ±1 u of the precursor and scored by cross-correlation; Mascot (Perkins and colleagues, 1999) introduced probability-based scoring with a simple significance threshold. Because a search always returns a top hit, Elias and Gygi’s 2007 target-decoy strategy estimates false matches by searching deliberately false sequences alongside real ones.

De novo sequencing derives the sequence from mass differences alone. PEAKS (2003) is a Canadian example, from Western University in London, Ontario, the University of Waterloo and Bioinformatics Solutions in Waterloo, built for samples whose genomes are not in any database.

For a synthetic peptide there is one expected sequence, so the question is whether the spectrum fits it and rules out plausible alternatives such as synthesis deletion products (see how research peptides are made). Non-standard residues need care: GHRP-2 contains 3-(2-naphthyl)alanine, absent from the 20 standard amino acids, so default search settings miss it unless the residue (C₁₃H₁₁NO, 197.08) is defined by hand.

What MS/MS can and cannot prove

A complete, well-assigned ladder is strong evidence that the residues are present in the stated order. Anything that leaves residue mass and fragmentation unchanged is invisible to it.

Leucine and isoleucine are discussed in LC-MS explained; Matrix Science notes they can be separated through side-chain satellite ions formed under high-energy conditions, which a routine scan does not provide. LL-37 contains several of each.

D-amino acids make a peptide an epimer of its all-L form, with the same mass and often only subtly different fragmentation. Tao, Quebbemann and Julian (2012) used radical-directed dissociation to discriminate D-serine, D-alanine and D-aspartate in eight peptides, with clearer chiral discrimination than CID for all but one of them. GHRP-6 contains D-tryptophan and D-phenylalanine, and an ordinary ladder reads the same masses either way; see D-amino acids explained.

Isoaspartate forms from asparagine deamidation or aspartate isomerisation via a succinimide (see the peptide stability guide). Cournoyer and colleagues (2005) showed ECD gives two diagnostic ions, reported as c+58 and z−57, only from isoaspartyl peptides. Finally, MS/MS does not measure purity; that is a chromatographic result, explained in our HPLC testing guide and peptide purity explained.

What’s established and what isn’t

Using MS/MS evidence when sourcing peptides in Canada

Research-peptide certificates commonly report an HPLC purity figure and a mass-spectrometry identity result. When comparing research peptides in Canada, know what kind of mass evidence a report contains.

• Intact mass or fragment ladder? A matching [M+H]⁺ or deconvoluted mass is a composition check; a tandem mass spectrometry result is a sequence check.
• Mass basis and deviation. Look for monoisotopic or average, and the difference in ppm or Da. The peptide COA guide covers the rest.
• D-residues or non-standard residues. Neither intact mass nor routine MS/MS confirms configuration.
• Match the batch. A report describes only the lot that was tested, so compare its batch number with the vial; see the Lab Testing page.
• Protect the sequence. Oxidation and deamidation change mass after delivery; see the peptide storage guide.

SNAP-8The worked b and y example above: acetylated, amidated, one methionineView product GHRP-6Contains D-Trp and D-Phe, which an ordinary MS/MS ladder cannot tell from L formsView product GHRP-2Contains 2-naphthylalanine, a non-standard residue outside default search settingsView product LL-3737 residues with 11 lysines and arginines, so it forms highly charged precursorsView product

Frequently asked questions