Graduate BCHM613 ⭐ 3 Credits Semester 2

Proteomics and Mass Spectrometry

A thorough course on mass spectrometry-based proteomics covering instrument platforms, ionization strategies, peptide sequencing, quantitative proteomics, post-translational modification analysis, and emerging structural proteomics methods. Emphasis on applying proteomics to peptide drug research, quality control, and biomarker discovery.

📅 Studied by Skipper on 2026-10-02

📖 Core Concepts

Claude Opus 5.5

Core Concepts: Proteomics and Mass Spectrometry

1. What This Subject Is Really About

Skipper, you have spent many courses learning what proteins are: their primary sequences (BIOL201), folds (BCHM603), active sites (BCHM605), and three-dimensional structures by crystallography, NMR, and cryo-EM (BCHM604). Proteomics asks a different set of questions. Of all the proteins a genome could encode, which ones are actually present in this cell, at this moment? How much of each is there? Which forms are present, meaning which sequence variants, which cleavage products, and which of the many possible combinations of phosphate, glycan, acetyl, and ubiquitin decorations?

The genome is a parts catalogue. The proteome is the factory floor during a shift, with machines being switched on and off, retooled, tagged for disposal, and moved between rooms. Proteomics tries to take an inventory of that floor while it is running.

The instrument that made this possible is the mass spectrometer. A mass spectrometer reduces chemistry to one observable: how a charged particle moves in an electric or magnetic field. Under the right conditions, that motion depends only on the ratio of mass to charge. Everything else in this course (identification, quantification, PTM mapping, structural proteomics, peptide drug QC) is clever experimental design built on top of that single measurement.

Aebersold and Mann framed the field in their 2003 review with an unusually candid title: proteomics is a "technology driven and technology limited science." Their point was that the biological questions were mostly obvious. What limited progress was the ability to measure.

"The ability of mass spectrometry to identify and, increasingly, to precisely quantify thousands of proteins from complex samples can be expected to impact broadly on biology and medicine." [Aebersold & Mann, 2003]

Keep that framing in mind all semester. Almost every concept below exists because someone hit a measurement wall and engineered around it:

  • Proteins are non-volatile, so ESI and MALDI were invented to get them into the gas phase.
  • Spectra are too complex to read directly, so tandem MS and database search were developed.
  • Signal intensity does not report abundance, so stable isotope labeling was introduced.
  • PTMs are substoichiometric, so enrichment chemistries were designed to concentrate them.

One more framing idea, which you met in CHEM302 and CHEM501: a mass spectrometer is a measurement device with error bars, a dynamic range, and a sampling bias. Proteomics data are not a photograph of the proteome. They are a statistically filtered, ionization-biased, stochastically sampled projection of it. Doing proteomics well mostly means knowing exactly how that projection distorts the underlying reality.


2. Concept One: The Mass Spectrometer Measures m/z, Not Mass

Every analyzer you will meet (quadrupole, TOF, ion trap, Orbitrap, FT-ICR) responds to the mass-to-charge ratio, m/z. For a peptide or protein ionized by protonation in positive mode, the observed m/z of an ion carrying z protons is:

         M + z · m_p
m/z  =  ─────────────          where m_p = 1.007276 Da (proton mass)
              z                      M   = neutral monoisotopic mass

Note that the proton mass is used, not the hydrogen atom mass (1.007825 Da). The difference is one electron, 0.000549 Da. At the ppm accuracy of modern instruments, that difference matters.

Charge states and why they are a gift

For a singly charged ion, m/z is essentially the mass. In electrospray, however, a 25 kDa protein might carry 15 to 30 charges, so it appears as a series of peaks between roughly m/z 800 and 1700. This seems like a complication, but it has two benefits.

  1. It brings enormous molecules into the m/z range of analyzers designed for small molecules. Fenn called this giving "electrospray wings for molecular elephants" in his 2002 Nobel lecture.
  2. Each peak in the charge envelope is an independent measurement of the same neutral mass. Averaging across them improves accuracy.

From any two adjacent charge-state peaks you can recover both z and M. Let peak 1 (lower m/z, value m₁) carry charge n, and peak 2 (higher m/z, value m₂) carry charge n−1:

  n · m₁       = M + n · m_p
  (n − 1) · m₂ = M + (n − 1) · m_p

  Subtracting:     n · (m₁ − m₂) + m₂ = m_p

                    m₂ − m_p
  ⇒   n  =  ───────────────
                    m₂ − m₁

  then  M = n · (m₁ − m_p)

This is the algebraic core of charge-state deconvolution, which Fenn and colleagues used in their 1989 paper [Fenn et al., 1989]. Modern algorithms such as maximum-entropy deconvolution (MaxEnt) and Bayesian approaches like UniDec generalize the idea. They fit the whole envelope rather than pairs of peaks, which makes them robust to noise and to overlapping species.

Reading charge from isotope spacing

There is a second, independent way to determine z, and it is the one you will use most often for peptides. The isotopes of an ion are separated by about 1.00335 Da, which is the ¹³C–¹²C mass difference. On the m/z axis, adjacent isotope peaks therefore appear at:

  Δ(m/z)  ≈  1.00335 / z

  z = 1  →  peaks ~1.003 apart
  z = 2  →  peaks ~0.502 apart
  z = 3  →  peaks ~0.334 apart

If you zoom in on a peak and see isotopes half a unit apart, you are looking at a 2+ ion. This only works if your resolving power can separate those peaks, which is the first of many places where instrument specifications become biochemical capabilities.


3. Concept Two: Isotope Patterns Are Probability Distributions

Here MATH201 becomes directly useful. Natural carbon is about 98.93% ¹²C and 1.07% ¹³C. A peptide with n carbon atoms is effectively a set of n independent trials, each with probability p ≈ 0.0107 of being "heavy." The number of ¹³C atoms follows a binomial distribution, which for large n is well approximated by a Poisson distribution with mean λ = np:

  P(k heavy atoms)  =  C(n,k) · p^k · (1−p)^(n−k)   ≈   e^(−λ) · λ^k / k!

  Relative intensity of M+1 to M (carbon only):  ≈ λ = n · 0.0107

Hydrogen (²H), nitrogen (¹⁵N), oxygen (¹⁷O, ¹⁸O) and sulfur (³³S, ³⁴S) contribute smaller corrections. Senko, Beu, and McLafferty (1995) defined a hypothetical "averagine" residue with average elemental composition of about C₄.₉₄H₇.₇₆N₁.₃₆O₁.₄₈S₀.₀₄ per 111.1 Da. It lets you predict a typical peptide's isotope envelope from its mass alone.

The consequences are worth memorizing:

  • ~1,000 Da peptide (~44 C): the monoisotopic peak (all ¹²C) is the tallest, and M+1 is roughly half its height.
  • ~1,800 Da: M and M+1 are about equal. Above this mass, the monoisotopic peak is not the most intense.
  • ~20 kDa protein (~900 C): λ ≈ 9.5, so P(k=0) ≈ e−9.5 ≈ 7 × 10⁻⁵. The monoisotopic peak is essentially invisible, and the envelope maximum sits about 9–10 Da above it.

This is why intact protein masses are usually reported as average masses (the centroid of the envelope), while peptide masses are reported as monoisotopic masses. Mixing the two is one of the most common reporting errors in peptide QC. For a 4 kDa peptide, the two differ by about 2.5 Da.

The isotope pattern is also an elemental fingerprint. Sulfur gives a distinctive M+2 contribution from ³⁴S (about 4.2% per S atom). Chlorine and bromine produce unmistakable M+2 doublets, and metals have their own patterns. Software that fits observed against theoretical envelopes (for example, isotope-pattern scoring in MaxQuant) uses this information to reject false matches.


4. Concept Three: Getting Proteins Into the Gas Phase (ESI and MALDI)

Before 1988–89, mass spectrometry of proteins was largely impossible. Electron ionization requires volatile analytes. Proteins are non-volatile and decompose before they evaporate, much as a steak chars long before it boils. Two "soft" ionization methods solved this almost simultaneously, and it is worth appreciating how different their physics is.

Electrospray ionization (ESI)

A solution is pushed through a capillary held at a few kilovolts. The liquid forms a Taylor cone that emits a fine jet of charged droplets. As solvent evaporates, charge density on the droplet surface rises until electrostatic repulsion matches surface tension. That point is the Rayleigh limit:

  q_R  =  8π · √(ε₀ · γ · r³)

  q_R = maximum charge on a droplet of radius r
  γ   = surface tension
  ε₀  = vacuum permittivity

At the Rayleigh limit the droplet undergoes Coulombic fission, ejecting a jet of much smaller, highly charged offspring droplets. This cascade repeats until analyte ions are released. How the last step happens depends on the analyte [reviewed in Konermann et al., 2013]:

  • Ion evaporation model (IEM) [Iribarne & Thomson, 1976]: small ions, including most peptides, are field-desorbed directly from the surface of a nanometer-sized droplet.
  • Charged residue model (CRM) [Dole et al., 1968]: large globular proteins are released when the droplet evaporates to dryness around them. The protein keeps the charge the droplet carried, which is close to the Rayleigh charge of a protein-sized droplet. This explains a useful empirical rule for native proteins, z ≈ 0.078·√M [Fernández de la Mora, 2000].
  • Chain ejection model (CEM): unfolded, hydrophobic-exposed chains are expelled from the droplet end-first and pick up protons along the way. This explains why denatured proteins carry far more charge than native ones.

The difference between CRM and CEM is the physical basis of native MS. Spray a protein from aqueous ammonium acetate at neutral pH and you see a narrow envelope of low charge states, because the compact fold is retained and noncovalent complexes survive. Spray the same protein from 50% acetonitrile with 0.1% formic acid and you see a broad envelope of high charge states. The charge state distribution is therefore a crude conformational probe, which links this course to BCHM603. You could, in principle, follow an unfolding transition by watching the envelope shift.

Nano-ESI [Wilm & Mann, 1996] reduces flow to tens to hundreds of nL/min using pulled emitters with micrometer-scale tips. The initial droplets are much smaller, so fewer fission generations are needed, ionization efficiency rises, and ion suppression from salts falls. ESI behaves as a concentration-sensitive technique: the signal depends on analyte concentration in the spray, not on total flow. Shrinking the flow therefore wastes less sample without losing signal. This is why essentially all discovery proteomics runs on nanoLC–nanoESI.

Matrix-assisted laser desorption/ionization (MALDI)

MALDI solves the same problem very differently. The analyte is co-crystallized with a large molar excess (around 10³–10⁴:1) of a small, UV-absorbing organic acid called the matrix. A pulsed UV laser (337 nm N₂ or 355 nm Nd:YAG) deposits energy into the matrix, not the analyte. The matrix ablates in a plume that carries intact analyte molecules into the gas phase. Proton transfer, mostly from matrix-derived ions in the plume, ionizes them.

Karas and Hillenkamp showed that this approach could desorb intact proteins above 100,000 Da [Karas & Hillenkamp, 1988]. Tanaka and colleagues reported protein desorption the same year using an ultrafine cobalt powder in glycerol. Tanaka shared the 2002 Nobel Prize with Fenn, an allocation that remains a classic case study in how scientific credit is assigned (relevant to CHEM504).

Matrix choice is chemistry, not superstition:

  • CHCA (α-cyano-4-hydroxycinnamic acid) is a "hot" matrix. It forms small, homogeneous crystals and is the standard for peptides below about 4 kDa. It causes more in-source decay.
  • SA (sinapinic acid) is a "cooler" matrix for intact proteins above about 10 kDa.
  • DHB (2,5-dihydroxybenzoic acid) is "cool," tolerant of salts and detergents, and preferred for glycans and labile species. It forms heterogeneous needle crystals, which create "sweet spots."

The key contrast is that MALDI produces predominantly singly charged ions, while ESI produces multiply charged ones. MALDI spectra are therefore easy to read: one peak per species. That simplicity is why MALDI-TOF dominates rapid identity checks of synthetic peptides and peptide mass fingerprinting. The cost is that MALDI couples poorly to online LC, and fragmenting 1+ ions is less informative.


5. Concept Four: Mass Analyzers as Applied Physics

Every analyzer turns m/z into something measurable, such as a time, a frequency, or a stability condition. Recognizing which physical quantity is being measured tells you each analyzer's strengths and limits. You derived most of these relationships in PHYS201/202 without knowing they would come back.

Time-of-flight (TOF)

Ions accelerated through potential V gain the same kinetic energy per charge, so lighter ions fly faster:

  z·e·V  =  ½ · m · v²     ⇒     v = √(2zeV / m)

  Flight time over drift length L:

  t  =  L · √( m / (2 z e V) )        ⇒      m/z  ∝  t²

In principle TOF has unlimited mass range, since a heavy ion simply arrives later. Its resolution is limited by the initial spread of kinetic energies and positions. The reflectron (an ion mirror) fixes this elegantly. Faster ions of a given m/z penetrate deeper into the retarding field and travel a longer path, so they arrive at the detector together with their slower siblings. Delayed extraction provides further focusing.

Reflector mode gives high resolution and ppm-level accuracy for peptides. Linear mode is needed for large proteins, which tend to fragment metastably during the longer reflectron flight, and it gives lower resolution.

Quadrupole

Four parallel rods carry a combined DC voltage (U) and RF voltage (V cos Ωt). Ion trajectories obey the Mathieu equation, with dimensionless stability parameters:

  a_u  =  8 z e U / (m r₀² Ω²)         q_u  =  4 z e V / (m r₀² Ω²)

Only ions whose (a, q) values fall inside the stability region pass through. Scanning U and V together at a fixed ratio transmits one m/z at a time. Quadrupoles are mass filters: robust, fast, and unmatched for targeted work. In a triple quadrupole, Q1 selects the precursor, q2 fragments it, and Q3 selects the fragment. This SRM/MRM arrangement is the workhorse of clinical LC-MS/MS (Week 13).

Ion traps

Ion traps use the same Mathieu physics in three dimensions, or in a linear trap, to store ions and then eject them in m/z order. They are sensitive and fast, and they can perform multiple rounds of fragmentation (MSⁿ).

Their notable limitation is the low-mass cutoff. During resonance-excitation CID, fragments below about one third of the precursor m/z are not stably trapped and are lost. This "one-third rule" matters later, because isobaric tag reporter ions sit at m/z 114–135, far below most precursors.

Orbitrap and FT-ICR: measuring frequencies

The Orbitrap [Makarov, 2000] traps ions in an electrostatic field between a spindle-shaped central electrode and an outer barrel. Ions orbit the spindle while oscillating harmonically along its axis. As in the mass-on-a-spring problem from PHYS201, the axial frequency depends only on m/z:

  ω_z  =  √( k · (z / m) )         k = field curvature constant

The oscillating ions induce an image current in the outer electrodes. That time-domain signal, called the transient, is Fourier-transformed (MATH301 and CHEM303) to give a frequency spectrum, which converts directly to m/z.

Because frequency can be measured very precisely, mass accuracy is typically below 3 ppm and resolving power can exceed 100,000. Resolution grows with transient length, so higher resolution costs scan speed. Orbitrap resolving power also falls with m/z as (m/z)−1/2. FT-ICR, which measures cyclotron frequency in a strong magnet (ω = zeB/m), falls off faster, as (m/z)−1, but reaches even higher absolute resolving power at large field strengths.

Mass accuracy and resolving power: two different things

                       m_measured − m_theoretical
  Mass error (ppm)  =  ─────────────────────────── × 10⁶
                            m_theoretical

  Resolving power   R = m / Δm        (Δm = peak width at half maximum, FWHM)

Accuracy is how close the centroid lies to the truth. Resolving power is whether two nearby peaks can be told apart. An instrument can be precise and wrong, which is a calibration problem, or accurate on one peak but unable to separate two overlapping ones, which is a resolution problem.

A worked example shows why this matters biochemically. Lysine trimethylation adds C₃H₆ (+42.0470 Da), and acetylation adds C₂H₂O (+42.0106 Da). The difference is 0.0364 Da. On a 1,500 Da peptide that is about 24 ppm, so resolving the two requires R of roughly 40,000 or more, plus accuracy well under 24 ppm. An older ion trap simply cannot tell an epigenetic activation mark from a repression context. An Orbitrap can.


6. Concept Five: Tandem MS: Breaking Peptides Predictably

An accurate precursor mass narrows the candidates, but many peptides share a mass. To read a sequence, we isolate one precursor, fragment it, and measure the pieces. This is MS/MS.

The key observation is that peptides break preferentially at the amide bonds of the backbone. The fragments therefore form ladders whose rungs are spaced by residue masses. Reading a ladder means reading the sequence.

Nomenclature

Following Roepstorff and Fohlman (1984), later refined by Biemann, cleavage at each of the three backbone bond types gives complementary N-terminal and C-terminal fragments:

         a │ b │ c
  H₂N─CHR─C─│─N─│─CHR'─ ...
          │  O ‖  H
         x │ y │ z

  Cα–C(O) bond  →  a / x ions
  C(O)–N  bond  →  b / y ions      (amide bond — dominant in CID/HCD)
  N–Cα    bond  →  c / z ions      (dominant in ETD/ECD)

For singly charged fragments of an unmodified peptide:

  b_i  =  Σ (residue masses 1..i)  +  1.00728
  y_j  =  Σ (residue masses of last j)  +  18.01056 (H₂O)  +  1.00728
  a_i  =  b_i  −  27.99491 (CO)

  Complementarity:   b_i + y_(n−i)  =  [M+H]⁺ + 1.00728

That last identity is a powerful sanity check. Complementary fragment pairs must add up to the precursor plus one proton. If you find two peaks that sum correctly, you have strong evidence that they come from the same cleavage.

CID and the mobile proton model

In collision-induced dissociation, ions are accelerated into an inert gas. Kinetic energy converts to internal (vibrational) energy over many collisions. The process is ergodic: energy redistributes across all vibrational modes before bond cleavage, as RRKM theory from CHEM303 describes. The weakest, most kinetically accessible pathways therefore dominate.

Which pathways are those? The mobile proton model [Wysocki et al., 2000] gives the answer. A proton initially sits on the most basic site, such as an Arg or Lys side chain or the N-terminus. Collisional heating lets it migrate to a backbone amide nitrogen. Protonating the amide weakens the C–N bond, and the neighboring carbonyl oxygen attacks to form an oxazolone b ion. Charge-directed cleavage of this kind produces fairly uniform b/y ladders.

Sequence effects follow directly from the mechanism, which is one of the satisfying parts of this subject:

  • Proline effect: cleavage N-terminal to Pro is strongly enhanced. The tertiary amide nitrogen of Pro is unusually basic, and the ring constrains the conformation in a way that favors this cleavage. Expect an intense y ion beginning with Pro.
  • Aspartic acid effect: when the proton is not mobile (for example, a 1+ peptide containing Arg, where the proton is sequestered on the guanidinium), the Asp side-chain carboxylic acid can supply the proton locally. Cleavage C-terminal to Asp then dominates.
  • His can enhance cleavage C-terminal to itself, depending on charge state, because its imidazole participates in proton transfer.
  • Labile PTMs fall off first. Phosphoserine and phosphothreonine readily lose H₃PO₄ (−97.977 Da) by β-elimination, often giving a dominant neutral-loss peak and few sequence ions. This is the central problem in phosphosite localization.

A subtle point: because b ions are cyclic oxazolones, larger b ions can open and re-close at a different position. This sequence scrambling produces fragments that do not correspond to the linear sequence [Harrison et al., 2006]. It is one reason y ions are often trusted more than b ions in tryptic peptides.

ETD and ECD: radical chemistry that does not equilibrate

In electron capture dissociation [Zubarev, Kelleher & McLafferty, 1998] or electron transfer dissociation [Syka et al., 2004], a multiply charged cation captures an electron, either a free low-energy electron (ECD) or one transferred from a radical anion such as fluoranthene (ETD). The resulting odd-electron species cleaves the N–Cα bond, producing c and z• ions.

The surprising and beautiful feature is that this cleavage is generally considered non-ergodic, or at least much faster than vibrational energy redistribution across the molecule. The backbone breaks before the energy reaches the weakly bonded phosphate or glycan. Labile modifications therefore stay attached to the fragments, and you can see exactly which residue carries them.

Practical consequences:

  • Electron-based methods are excellent for PTM localization (phospho, O-GlcNAc, glycans) and for the high charge states of intact proteins in top-down work.
  • They require charge ≥ 2+, because one charge is neutralized. They work poorly on 2+ tryptic peptides.
  • No cleavage N-terminal to Pro is observed. Breaking N–Cα in proline does not separate the fragments, because the pyrrolidine ring still holds them together.
  • Disulfide bonds are preferentially cleaved, which is useful for disulfide mapping (Week 14).

HCD and the hybrid strategies

Higher-energy collisional dissociation is beam-type CID performed in a dedicated multipole cell, with fragments then sent to the Orbitrap. The chemistry is still vibrational (b/y ions), but there is no low-mass cutoff, and secondary fragmentation produces diagnostic immonium ions and the low-mass reporter ions of TMT and iTRAQ.

EThcD combines ETD with supplemental HCD activation of all ions, giving b, y, c, and z ions together. It is particularly powerful for glycopeptides and for distinguishing Leu from Ile through side-chain w ions.

UVPD (193 nm photons) is a further option. It deposits high energy in a single photon absorption event and accesses a/x ions and many others, producing very high sequence coverage in top-down experiments.

Think of the methods as tools with complementary blind spots. CID/HCD is reliable for unmodified peptides but strips labile PTMs. ETD/ECD preserves PTMs but needs high charge and skips proline. Matching the fragmentation method to the question is a large part of experimental design.


7. Concept Six: Identification Is Statistical Inference

In bottom-up (shotgun) proteomics, proteins are digested, usually with trypsin. Trypsin cleaves C-terminal to Lys and Arg, generally not before Pro. That gives peptides of convenient length (about 7–25 residues) with a basic C-terminus. Such peptides tend to carry 2+ charge (N-terminal amine plus C-terminal Lys/Arg) and give strong y-ion series.

Other proteases provide complementary coverage. Lys-C cleaves after Lys only. Glu-C cleaves after Glu, and after Asp in phosphate buffer. Asp-N cleaves N-terminal to Asp.

The resulting peptides are separated by nanoLC and analyzed in data-dependent acquisition (DDA). The instrument takes a survey scan, picks the top N most intense precursors, fragments each one, excludes them dynamically for a while, and repeats.

Each MS/MS spectrum is compared against theoretical spectra predicted from an in-silico digest of a protein database. The comparison is constrained by:

  • precursor mass tolerance (for example, ±10 ppm)
  • fragment tolerance
  • enzyme specificity and number of missed cleavages
  • fixed modifications (for example, carbamidomethyl Cys, +57.021)
  • variable modifications (for example, Met oxidation, +15.995)

Search engines differ mainly in how they score matches:

  • SEQUEST [Eng, McCormack & Yates, 1994] uses cross-correlation (XCorr).
  • Mascot [Perkins et al., 1999] uses a probability-based score.
  • Andromeda in MaxQuant uses a binomial-probability score.

Every additional variable modification enlarges the search space combinatorially. A larger search space raises the chance of random high-scoring matches, so you pay for flexibility with statistical power.

Target–decoy and FDR

How do you know a match is real? The elegant answer [Elias & Gygi, 2007] is to search against the real ("target") database concatenated with an equally sized "decoy" database of reversed or shuffled sequences. No genuine peptide should match a decoy, so decoy hits sample the distribution of false matches. Above a given score threshold:

                 # decoy hits above threshold
  FDR   ≈   ─────────────────────────────────
                 # target hits above threshold

The threshold is set so that FDR equals 1%. The q-value of a match is the minimum FDR at which that match would be accepted. This is the multiple-testing logic from MATH201 applied to tens of thousands of spectra. Here proteomics is closer to genomics than to classical analytical chemistry: you never prove any single identification, you control the error rate of the whole list.

FDR at different levels

FDR does not carry over unchanged from peptides to proteins. True proteins tend to accumulate many peptide matches, while false matches scatter randomly across the database, usually one per protein. As a result, a 1% PSM-level FDR can inflate to a much higher protein-level FDR in large datasets. Protein-level FDR must be controlled separately.

The related protein inference problem also matters. Shared peptides may come from several isoforms or paralogs, so the "protein" reported is often a protein group.

De novo sequencing

When no database entry exists, for example for novel antibodies, venom peptides, or unusual synthetic analogs, algorithms such as PEAKS or Novor read the residue spacing directly from the spectrum. This makes the isobaric limits of MS concrete. Leu and Ile are identical in mass (113.08406 Da). Gln (128.05858) and Lys (128.09496) differ by only 0.036 Da. A Gly–Gly dipeptide has the same elemental formula as a single Asn residue (C₄H₆N₂O₂, 114.04293 Da).


8. Concept Seven: MS Is Not Inherently Quantitative. Isotopes Make It So.

This is perhaps the most important conceptual point in the course. Signal intensity in ESI-MS is not proportional to abundance across different analytes. Ionization efficiency depends on the following factors:

  • surface activity and hydrophobicity (the IEM favors ions that crowd the droplet surface)
  • basicity
  • co-eluting competitors that cause ion suppression
  • solvent composition at the moment of elution

Two equimolar peptides can differ in signal by 100-fold.

The solution is the same idea you met in CHEM302 as an internal standard, taken to its logical limit. Compare each peptide with itself, using a version that is chemically identical but heavier:

  • SILAC [Ong et al., 2002]: cells are grown with ¹³C₆¹⁵N₂-Lys (+8.014 Da) and ¹³C₆¹⁵N₄-Arg (+10.008 Da). After tryptic digestion, every peptide except the protein C-terminal peptide carries exactly one label. Light and heavy samples are mixed before any processing, so all downstream variability cancels. The heavy/light ratio of co-eluting peaks gives relative abundance.
  • Isobaric tags (TMT [Thompson et al., 2003]; iTRAQ [Ross et al., 2004]): amine-reactive tags of identical total mass are distributed differently between a reporter group and a balancer group. Peptides from 6, 10, 16, or more samples co-elute and co-isolate as one precursor. Only upon HCD fragmentation do the reporters separate, at m/z 126–135. This allows high multiplexing, but co-isolated interfering peptides contribute reporters too, which causes ratio compression toward 1:1. MS3-based methods such as SPS-MS3 address this.
  • AQUA [Gerber et al., 2003]: a synthetic heavy-isotope peptide is spiked at known concentration. Monitored by SRM/MRM on a triple quadrupole or by PRM on an Orbitrap, it gives absolute quantification. This is where your peptide synthesis training (BCHM602) becomes a quantitative-biology tool: the accuracy of the assay rests on the purity and concentration accuracy of a peptide that someone synthesized.

Label-free approaches (spectral counting, MS1 peak areas, MaxLFQ) compare the same peptide across runs. This sidesteps differences in ionization efficiency between peptides, but it demands excellent chromatographic reproducibility and careful normalization.

Data-independent acquisition (DIA), introduced as SWATH [Gillet et al., 2012], fragments everything within sequential wide m/z windows instead of picking the top N precursors. DDA is stochastic: a low-abundance peptide may be picked in one run and missed in the next, which leaves "missing values." DIA is systematic, so every run records fragment information for every precursor. The price is highly multiplexed spectra that must be deconvolved computationally, using spectral libraries or library-free tools such as DIA-NN and Spectronaut.


9. Concept Eight: PTMs Are Mass Shifts, and Enrichment Is Necessary

Mann and Jensen's review [Mann & Jensen, 2003] made the defining observation of PTM proteomics. Nearly every PTM changes mass, so MS can in principle detect all of them. In practice, modified forms are usually substoichiometric: a phosphosite might be 1% occupied, and it is buried in a 10⁴-fold excess of unmodified peptides. Detection is a dynamic-range problem first and a mass-measurement problem second.

The masses worth knowing by heart:

  Modification                    Δ mass (Da, monoisotopic)    Notes
  ─────────────────────────────────────────────────────────────────────────
  Phosphorylation (HPO₃)              +79.96633         neutral loss −97.977 (CID)
  Acetylation (C₂H₂O)                 +42.01057         vs. trimethyl: Δ 0.036
  Trimethylation (C₃H₆)               +42.04695
  Ubiquitin remnant Gly-Gly           +114.04293        after trypsin; same formula as Asn
  Oxidation (Met)                     +15.99491         often artifactual
  Deamidation (N→D, Q→E)              +0.98402          vs. ¹³C spacing 1.00335!
  Carbamidomethyl (Cys, IAA)          +57.02146         identical to a Gly residue
  HexNAc (O-GlcNAc / GalNAc)          +203.07937        oxonium ion m/z 204.087
  Succinylation                       +100.01604
  Disulfide formation                  −2.01565  (per bond)

Enrichment strategies exploit each modification's chemistry:

  • Phosphate: its affinity for Lewis-acidic metal oxides (TiO₂) or chelated metals (Fe³⁺/Ga³⁺ IMAC).
  • Ubiquitination: trypsin leaves a Gly-Gly remnant on the modified lysine, which a specific anti-K-ε-GG antibody recognizes.
  • Glycopeptides: their hydrophilicity (HILIC) or carbohydrate recognition (lectins).

The design principle is always to find a physical or chemical property that is unique to the modification and use it to pull modified peptides out of the unmodified majority.

Once a modified peptide is identified, site localization is a separate statistical question. If a peptide contains three Ser/Thr residues and one phosphate, which residue carries it? Only fragment ions that fall between candidate sites can discriminate. Scoring tools such as PhosphoRS, Ascore, and the Mascot Delta Score turn this into a probability. A peptide can be identified with high confidence while its site assignment remains ambiguous.


10. Concept Nine: MS as a Structural Probe

Mass spectrometry measures mass, yet it can report on three-dimensional structure, which links it directly to BCHM603 and BCHM604.

HDX-MS uses the fact that backbone amide hydrogens exchange with D₂O at rates set by local structure. Hydrogen-bonded or buried amides exchange slowly. Exchange is minimized near pH 2.5 and 0 °C, so the reaction can be "frozen" by quenching under those conditions. The protein is then digested with pepsin, which remains active at that acidic pH, and the deuterium uptake of each peptide is measured. In the EX2 regime:

  k_obs  =  K_op · k_int           Protection factor   P = k_int / k_obs = 1 / K_op
  ΔG_op  =  −RT ln K_op  =  RT ln P

This is the same local-unfolding thermodynamics you met in BCHM603, now read peptide by peptide. Differential HDX, comparing with and without a ligand or antibody, maps epitopes and allosteric changes.

XL-MS covalently links residues that are close in space. Examples are DSS between lysines (spacer about 11.4 Å, Cα–Cα constraint roughly ≤ 30 Å) and EDC zero-length links between carboxylates and amines. Identifying the cross-linked peptide pairs gives distance restraints for integrative modeling of complexes that are too large or heterogeneous for crystallography.


11. Connections to Peptide Chemistry

Skipper, this is where the course meets your training in BCHM602 and BCHM611. For a peptide chemist, the mass spectrometer is effectively part of the synthesis workflow: every resin cleavage ends with an LC-MS run. Learn to read these spectra as diagnostic reports on your chemistry.

Reading SPPS failure modes as mass shifts

  • Deletion sequences (incomplete coupling): [M − residue mass]. A −57 peak is a missing Gly, −71 a missing Ala, and so on. Their pattern tells you which coupling step failed.
  • Truncations (capped after failed coupling): a ladder of N-terminally acetylated shorter species if acetic anhydride capping was used.
  • Incomplete side-chain deprotection: +56.06 (tBu), +252.10 (Pbf on Arg), +242.11 (Trt), +100.05 (Boc). These point directly to insufficient cleavage time or scavenger problems.
  • Aspartimide formation at Asp-Gly and Asp-Ser motifs: −18.011 (loss of H₂O). It is often followed by piperidine ring-opening (+67 relative to the aspartimide, i.e., +49 relative to the target) or hydrolysis back to α- and β-Asp. That hydrolysis product is isobaric with the target, so MS alone cannot see it.
  • Met oxidation (+16) and Trp modification by cleavage cations, which are signs of scavenger failure.
  • Double insertions (+ residue mass), often from premature Fmoc loss or excess activated amino acid.
  • TFA adducts (+113.99) appear in ESI and are easily confused with a Gly-Gly remnant (+114.04). The 0.05 Da difference is about 50 ppm on a 1,000 Da peptide, which is trivial on an Orbitrap and invisible on a low-resolution instrument.

What MS cannot see, and why that matters for peptide drugs

Racemization (an L→D epimer at one residue) is isobaric, as is the formation of isoAsp after aspartimide hydrolysis. Both often co-elute with the main peak on fast gradients. MS confirms identity, not stereochemical purity. That requires chiral analysis or orthogonal chromatography, and for isoAsp specifically, ETD gives diagnostic c+57 and z−57 fragments. Regulatory characterization of peptide APIs therefore pairs MS with orthogonal HPLC methods.

Peptide therapeutics: where this course leads

  • Intact mass of lipidated peptides. Semaglutide (average MW about 4113.6 Da) carries a C18 diacid–γGlu–OEG–OEG side chain on Lys26. ESI shows a 3+/4+/5+ envelope, and deconvolution confirms both the backbone and the full lipid moiety.
  • PEGylated peptides. PEG is polydisperse, so instead of one mass you see a distribution of peaks spaced 44.026 Da apart (C₂H₄O) on every charge state. The overlapping envelopes make deconvolution a real challenge (Week 14). Charge-reducing agents such as triethylamine are often used.
  • Disulfide mapping for cyclic and multi-disulfide peptides (insulin analogs, conotoxins, linaclotide). Comparing non-reduced and reduced digests, with ETD to cleave S–S bonds preferentially, establishes connectivity. Each disulfide is −2.016 Da relative to the free thiols.
  • Metabolite identification: tracking proteolytic clipping and lipid side-chain β-oxidation of GLP-1 analogs in plasma and urine, using the same fragment-ladder logic as bottom-up proteomics.
  • Pharmacology through phosphoproteomics: mapping signaling downstream of GLP-1R and MC4R (Week 9) tells you not just that a peptide binds its receptor, but what that binding does to the kinase network.

Also note how the same chemistry appears in both settings. The carbamidomethyl Cys mass (+57.021) is identical to a Gly residue because iodoacetamide adds C₂H₃NO, which is a glycyl unit. In proteomics, over-alkylation at peptide N-termini can therefore masquerade as an extra Gly. Everything you learned about side reactions in synthesis also applies to sample preparation for proteomics.


12. Common Misconceptions to Watch Out For

  1. "Mass spectrometers measure mass." They measure m/z. A peak at m/z 1000 could be a 999 Da 1+ ion or a 1998 Da 2+ ion. Always determine charge, from isotope spacing or adjacent charge states, before interpreting a peak.
  2. "The tallest isotope peak is the monoisotopic mass." This holds only below about 1,800 Da. For larger peptides and proteins, picking the wrong isotope peak gives a +1 Da error. Software "monoisotopic peak misassignment" is a known source of false identifications, sometimes reported as spurious deamidation.
  3. "A +1 Da shift means deamidation." Deamidation is +0.984 Da, and a ¹³C isotope is +1.003 Da. They differ by 0.019 Da, about 19 ppm on a 1,000 Da peptide. Without high resolution and correct isotope-envelope fitting, the two are easily confused.
  4. "Bigger peak means more of that compound." Not across different analytes. Ionization efficiencies vary widely, and ion suppression depends on context. A 2% impurity in a MALDI spectrum may be 0.2% or 20% by mass. Quantitative purity requires UV-HPLC, or MS with matched isotope standards.
  5. "Not detected means not present." DDA samples stochastically, and dynamic range is limited, to about 4–5 orders per run against more than 10 orders in plasma [Anderson & Anderson, 2002]. Absence of evidence is not evidence of absence. This is one of the main motivations for DIA and targeted methods.
  6. "1% FDR means 1% of my proteins are wrong." Only if FDR was controlled at the protein level. PSM-level FDR propagates non-linearly upward.
  7. "MS can sequence anything." It cannot distinguish Leu from Ile (except via w ions in ETD/EThcD), D- from L-amino acids, α- from β-Asp without special fragmentation, or glycan linkage isomers without additional methods. Keep a mental list of isobaric pairs.
  8. "Higher resolution is always better." Resolution costs transient time, which costs scan speed, which costs the number of peptides sampled per LC peak. Good methods balance resolving power against duty cycle.
  9. "Native MS shows the solution structure." It preserves much of the solution structure: stoichiometry, noncovalent complexes, and approximate compactness. But desolvation removes the water that stabilizes some interactions. Gas-phase ions can strengthen electrostatic contacts and weaken hydrophobic ones. Treat native MS as strongly correlated with solution structure, not identical to it.
  10. "Identified a phosphopeptide means I know the site." Identification and localization are separate statistical claims. A phosphopeptide can be confidently identified while the site is ambiguous among adjacent Ser/Thr/Tyr residues. Always report localization probabilities.
  11. "Isobaric tags give accurate ratios." Co-isolation interference systematically compresses ratios toward 1. A measured 3-fold change may reflect a true 5-fold change.

A Closing Thought

What makes this field intellectually satisfying is how directly physics becomes biology. A 0.036 Da mass difference, about one sixtieth of a neutron's worth of nuclear binding energy, separates a lysine that is acetylated from one that is trimethylated. Those two marks can have different downstream readout in chromatin biology. A non-ergodic bond cleavage on a femtosecond-to-picosecond timescale is what lets you say which serine a kinase phosphorylated. A binomial distribution in carbon isotopes tells you whether your 4 kDa peptide drug carries its lipid side chain.

For the rest of the semester, practice moving between those scales. When you see a spectrum, ask what physical process produced each peak. Then ask what that process implies about the molecule, and finally what it implies about the cell or the medicine.

Primary sources for this section: Aebersold & Mann, Nature 2003 (DOI: 10.1038/nature01511); Fenn et al., Science 1989 (DOI: 10.1126/science.2675315); Karas & Hillenkamp, Anal. Chem. 1988 (DOI: 10.1021/ac00171a028); Mann & Jensen, Nat. Biotechnol. 2003 (DOI: 10.1038/nbt0203-255); Liebler, Introduction to Proteomics. Additional cited works (Makarov 2000; Zubarev et al. 1998; Syka et al. 2004; Elias & Gygi 2007; Ong et al. 2002; Gerber et al. 2003; Gillet et al. 2012; Wysocki et al. 2000; Konermann et al. 2013) should be read in the original as they come up in the weekly readings.

⚙️ Mechanisms & Applications

GPT-6 Astra

Mechanisms & Applications: From Molecules to Defensible Biological Claims

Skipper’s central organizing idea: a mass spectrometer measures ions, not “proteins” or “pathways.” Protein identities, abundances, modification sites, and structural changes are conclusions reconstructed from those ion measurements. Every experimental step determines which molecules become visible—and which interpretations remain justified.

Biological sample
  → extraction and preservation
  → separation, digestion, and/or enrichment
  → ion formation
  → separation by mass-to-charge ratio (m/z)
  → optional precursor isolation and fragmentation
  → ion detection
  → identification and statistical validation
  → quantification
  → biological interpretation

This extends a principle from analytical chemistry: the measurement is inseparable from the sample preparation and calibration model. A beautiful spectrum cannot rescue a degraded sample, and a statistically confident peptide identification does not automatically establish a protein’s biological function. Aebersold and Mann’s foundational account of mass-spectrometry-based proteomics is best read through this lens: technological power and technological selectivity develop together.[1]

1. What Mass Spectrometry Actually Measures

An electric or magnetic field acts on an ion according to its charge as well as its mass. Consequently, the horizontal coordinate of a mass spectrum is m/z, not neutral molecular mass. For a molecule protonated at z sites:

M + zH⁺ → [M + zH]ᶻ⁺

m/z = (M + z mH⁺)/z

M = z(m/z) − z mH⁺

mH⁺ ≈ 1.007276 Da

Here, M is the neutral molecular mass and z is the positive charge number. A neutral 1,000.0000 Da peptide appearing as a doubly protonated ion is therefore observed at approximately m/z 501.0073, not 500.0000. Sodium, potassium, and other adducts require different mass accounting.

Isotope patterns provide an independent clue to charge. Replacing one carbon-12 atom with carbon-13 increases mass by approximately 1.003355 Da. Thus adjacent peaks in a peptide’s isotope envelope are separated by approximately:

Isotope spacing in m/z ≈ 1.003355/z

z = 1: approximately 1.0034
z = 2: approximately 0.5017
z = 3: approximately 0.3345

This is a particularly elegant inference: closely spaced isotope peaks reveal a property—charge—that is needed to recover the neutral mass. For large proteins, the monoisotopic peak may be extremely weak or unobserved; software may report an average mass or a model-derived monoisotopic mass. Those conventions must not be mixed.

Mass error (ppm)
  = 10⁶ × [(m/z)observed − (m/z)theoretical] / (m/z)theoretical

Resolving power
  R = (m/z) / Δ(m/z)

Δ(m/z) is commonly measured at full width at half maximum.

At m/z 500, a 5 ppm error corresponds to 0.0025 in m/z. Accuracy and resolving power are different: an instrument can separate neighboring peaks while placing both slightly off their correct positions. Calibration improves the coordinate system; resolution improves discrimination between nearby signals. Neither alone distinguishes all structural isomers.

2. Electrospray: Transferring Biomolecules into the Gas Phase

Electrospray ionization, or ESI, solves a seemingly contradictory problem: how can a large, nonvolatile protein enter the gas phase without first being thermally destroyed? Instead of boiling the protein, ESI transfers charge-bearing material out of tiny liquid droplets. Fenn and colleagues’ 1989 demonstration made this approach transformative for large biomolecules.[2]

  1. Create an electrified liquid surface. A potential difference at a capillary tip produces a Taylor cone and a spray of charged droplets.
  2. Remove solvent. Evaporation shrinks droplets while concentrating their charge.
  3. Destabilize highly charged droplets. Electrostatic repulsion eventually competes with surface tension, leading to Coulomb fission.
  4. Generate gas-phase analyte ions. Depending on analyte and conditions, ion evaporation, charged-residue, and chain-ejection mechanisms contribute.
  5. Desolvate and transmit the ions. Residual solvent and adducts are reduced as ions pass into the instrument’s vacuum system.
Rayleigh charge limit for an ideal conducting spherical droplet:

QR = 8π(ε₀γr³)½

ε₀ = vacuum permittivity
γ  = surface tension
r  = droplet radius

As r decreases, the charge that the droplet can stably carry decreases.

The final ion-formation step is not governed by one universal mechanism. Small ions may escape a droplet directly; a compact protein may remain as the last solvent evaporates; an unfolded chain may be progressively expelled from a droplet. Keeping these models distinct prevents the misleading picture that every protein simply “evaporates” from solution.

Why multiple charging matters: a 30 kDa protein carrying 20 positive charges appears near m/z 1,501. Multiple charging brings large molecules into accessible analyzer ranges. Deconvolution software recognizes several charge states of the same species and reconstructs their common neutral mass. Overlapping envelopes, adducts, and insufficiently resolved isotopes can make that reconstruction ambiguous.

  • Nano-ESI: low flow rates, commonly tens to hundreds of nanoliters per minute, favor small initial droplets and efficient ion production. This improves mass sensitivity but increases demands on emitter cleanliness, fluidics, and reproducibility.
  • Denaturing ESI: acidic, organic-containing solutions often unfold proteins and produce higher charge states. This is useful for intact-mass analysis and fragmentation.
  • Native MS: carefully chosen, usually volatile aqueous buffers and gentle transfer conditions can preserve noncovalent assemblies sufficiently to examine stoichiometry and ligand binding. “Native” is an operational term: gas-phase survival does not prove that every solution interaction or conformation is unchanged.
  • APCI: atmospheric-pressure chemical ionization first vaporizes the sample and then uses reagent ions generated by a corona discharge. It is valuable for many smaller, moderately polar molecules, but is generally unsuitable for intact, nonvolatile proteins.

Connection to protein folding: charge-state distributions can respond to protein compactness, but they also depend on solvent, protonation chemistry, and instrument settings. They are structural evidence, not an unambiguous folding assay.

3. MALDI: A Matrix Makes Laser Desorption Useful

In matrix-assisted laser desorption/ionization, or MALDI, the analyte is embedded in a large excess of a UV-absorbing matrix. Karas and Hillenkamp’s 1988 protein work demonstrated that laser desorption could be extended to proteins above 10 kDa—a major step toward routine biomolecular mass measurement.[3]

  1. Mix analyte and matrix, deposit them on a target, and allow co-crystallization.
  2. Irradiate the spot with a short laser pulse absorbed strongly by the matrix.
  3. Produce an expanding plume containing matrix and analyte species.
  4. Generate analyte ions through coupled desorption and ionization processes, including proton-transfer chemistry.
  5. Accelerate ions into a mass analyzer, often a time-of-flight instrument.

CHCA is commonly effective for peptides, sinapinic acid for larger proteins, and DHB for many glycans and glycopeptides. These are practical starting points, not universal rules. Crystal morphology, salts, and local analyte distribution can create substantial spot-to-spot variation. MALDI commonly gives predominantly singly charged peptide ions, simplifying spectra.

Idealized time-of-flight relationship:

qV = ½mv²
t = L/v

Therefore:
t = L[m/(2qV)]½

With fixed acceleration voltage and flight length:
t ∝ √(m/z)

A reflectron uses an electrostatic field to compensate for differences in ion kinetic energy: faster ions penetrate farther and travel a longer path. This sharpens arrival-time distributions. Linear mode often offers advantages for very large ions, whereas reflectron mode commonly improves peptide resolution.

Peptide-industry application: MALDI can rapidly check whether a synthesis or purification fraction contains material near the expected peptide mass. It does not establish purity. Impurities may ionize poorly, be suppressed, overlap the desired mass, or differ only in stereochemistry. A correct mass is an identity constraint, not a certificate of chemical homogeneity.

4. Analyzers, Detectors, and the Trade-Off Between Speed and Information

  • Quadrupole: combined radiofrequency and direct-current fields permit stable transmission of a selected range of m/z. It can act as a scanning mass filter or as a precursor selector.
  • Ion trap: stores ions in a time-varying electric field, then selectively ejects or fragments them. Repeated isolation and fragmentation enable MSn. Excess stored charge creates space-charge effects that compromise performance.
  • TOF: determines m/z from flight time and can acquire spectra rapidly.
  • Orbitrap: confines ions electrostatically and measures image currents generated by their motion. Fourier transformation converts the time-domain signal into frequencies related to m/z.

Ion detection may involve electron multipliers or microchannel plates, which amplify the impact of incoming ions, or image-current detection, which senses coherent ion motion without requiring impact. Dynamic range is constrained by ion statistics, detector response, ion-storage capacity, and interference from abundant species.

The experimental bargain: longer ion accumulation can improve sensitivity; longer Orbitrap transients can improve resolving power. Both consume time. During liquid chromatography, peaks continue to elute whether the instrument is ready or not. Maximizing resolving power on every scan can therefore reduce the number of peptides adequately sampled.

This is the instrumental-analysis version of an old lesson from statistics: a more precise measurement of fewer objects is not automatically a better experiment.

5. Turning a Peptide into Sequence Evidence: Tandem MS

MS/MS deliberately fragments an isolated precursor ion. The masses of the products reveal where the original molecule could have broken. In peptide analysis, a ladder of fragments can encode residue order.

Peptide backbone fragment families:

N-terminal fragments: a, b, c
C-terminal fragments: x, y, z

Common tendencies:
CID/HCD → predominantly b and y ions
ECD/ETD → predominantly c and z-type ions

For a singly charged, unmodified linear peptide:

m(bₙ) = Σ(N-terminal n residue masses) + mH⁺
m(yₙ) = Σ(C-terminal n residue masses) + mH₂O + mH⁺

Adjacent fragment difference:
m(bₙ₊₁) − m(bₙ) = residue mass of residue n+1

Residue masses are amino-acid masses after accounting for incorporation into a peptide chain. Comparing like charge states, a 71.0371 Da difference suggests alanine. Routine backbone-fragment mass measurements cannot distinguish leucine from isoleucine because their residue masses are identical.

  1. Select a precursor. A finite isolation window admits the intended ion and potentially neighboring ions.
  2. Deposit energy or transfer an electron. This activates specific dissociation pathways.
  3. Measure product ions. The resulting spectrum contains sequence fragments, possible neutral losses, diagnostic ions, and sometimes fragments from co-isolated molecules.
  4. Compare competing sequences or modification sites. Interpretation depends on fragments that discriminate among alternatives, not merely on the number of peaks.

CID and HCD: collisions with gas convert translational energy into internal energy. Proton mobility and local sequence affect which peptide bonds break. Cleavage N-terminal to proline can be favored, while histidine and other basic residues alter proton localization. HCD is beam-type collisional activation, commonly implemented in hybrid Orbitrap systems; it is not electron-based fragmentation. It is especially useful for broad product-ion detection, including low-mass reporter and diagnostic ions.

ECD and ETD: electron capture or transfer to multiply charged cations initiates radical-driven backbone cleavage. These approaches often retain labile modifications while generating sequence information. ECD emerged from the work of Zubarev, Kelleher, and McLafferty in 1998; Syka and colleagues introduced ETD in 2004, making related chemistry accessible in ion-trap platforms.[4], [5] Neither method guarantees complete fragmentation: charge density and peptide structure matter.

EThcD supplements ETD with collisional activation, combining complementary fragments. UVPD deposits photon energy and can generate extensive, diverse fragmentation, particularly useful for intact proteins.

Localization example: a peptide containing two serines has one +79.9663 Da modification consistent with phosphorylation. Precursor mass establishes neither which serine is modified nor, by itself, an unequivocal chemical assignment. Fragments separating the candidate sites provide localization evidence. A phosphate neutral loss can support recognition of a phosphopeptide but usually does not identify the modified residue.

6. Bottom-Up Identification: Digestion Creates Both Power and Ambiguity

Bottom-up proteomics converts proteins into peptides because peptides are easier to separate, ionize, and fragment reproducibly. This is a controlled loss of molecular context: the experiment becomes more tractable, but information about which peptides originally belonged to the same intact molecule is partly destroyed.

  1. Extract and stabilize proteins. Use conditions compatible with the biological question; suppress unwanted proteolysis and PTM turnover.
  2. Reduce and alkylate when appropriate. Reduction opens disulfides, and alkylation limits their re-formation. Iodoacetamide typically produces a +57.0215 Da carbamidomethyl modification on cysteine.
  3. Digest. Trypsin cleaves mainly after Lys or Arg, usually not when followed by Pro. Lys-C, Asp-N, and Glu-C generate alternative peptide maps.
  4. Separate by reversed-phase LC. Peptides partition differently between the stationary phase and a changing mobile phase.
  5. Acquire MS/MS spectra. Match spectra to candidate peptides, then infer proteins or protein groups.

Database-search tools such as SEQUEST, Mascot, and Andromeda compare measured spectra with predictions derived from protein sequences. Eng, McCormack, and Yates’s 1994 SEQUEST work was historically pivotal: it turned tandem spectra into a scalable database-matching problem.[6]

Search settings encode chemical assumptions: protease specificity, missed cleavages, precursor and fragment tolerances, fixed modifications, variable modifications, and sequence variants. Allowing every imaginable modification is not “being thorough.” It expands the candidate space and increases the burden of separating true matches from chance matches.

A common target–decoy estimate after competition:

FDR ≈ D/T

D = accepted decoy matches
T = accepted target matches

Some procedures use a correction such as (D + 1)/T.
The estimator must match the database and competition design.

Decoys are deliberately incorrect sequences that approximate the behavior of false target matches. Elias and Gygi helped establish target–decoy searching as a practical framework for large-scale confidence assessment.[7]

  • A 1% peptide-spectrum-match FDR is not automatically a 1% distinct-peptide FDR.
  • A 1% peptide FDR is not automatically a 1% protein-group FDR.
  • FDR is an expected false fraction within a selected set, not an exact error probability for every member.
  • Shared peptides support protein groups unless additional evidence distinguishes their source proteins.

De novo sequencing instead infers sequence directly from fragment patterns. It is valuable for organisms with incomplete databases, unexpected variants, and endogenous peptides, but incomplete ladders and isobaric residues can leave ambiguity. Peptide mass fingerprinting, often performed by MALDI-TOF on isolated protein digests, matches a collection of peptide masses to predicted digests; it is much less suited to highly complex mixtures than tandem-MS identification.

7. Top-Down and Middle-Down: Preserving Molecular Context

A proteoform is a specific molecular form of a protein arising from sequence variation, processing, and modifications. Bottom-up data may identify several modifications across a protein without revealing whether they coexist on the same molecule.

Bottom-up:
Protein → many short peptides
Excellent analytical accessibility; reduced long-range connectivity.

Middle-down:
Protein → fewer, larger fragments
Intermediate accessibility and retained connectivity.

Top-down:
Intact protein → intact mass + gas-phase fragmentation
Greatest opportunity to connect sequence and modifications within one proteoform.

The surprising point is that measuring more peptides does not necessarily recover the connectivity lost during digestion. If phosphorylation at site A and acetylation at site B occur on different peptides, their co-occurrence remains unresolved without additional evidence.

Top-down analysis addresses this problem, but intact-protein separation, charge-state overlap, proteoform heterogeneity, and incomplete fragmentation are substantial challenges. Middle-down analysis can offer a useful compromise, especially for histones and other heavily modified proteins. Applications include therapeutic protein characterization, endogenous peptide processing, and distinguishing combinations of mutations and PTMs.

8. Quantification: Ion Signal Becomes Abundance Only Through a Model

A useful first approximation is:

Measured signal for peptide i:

Iᵢ ≈ nᵢ × Rᵢ × Yᵢ + background

nᵢ = amount of peptide
Rᵢ = ionization, transmission, and detection response
Yᵢ = cumulative preparation and recovery yield

Different peptides have different response factors. Therefore, twice the signal from peptide A compared with peptide B does not generally mean twice as many molecules. Comparisons of the same peptide across controlled conditions are more defensible, although matrix-dependent suppression can still interfere.

SILAC: cells incorporate stable-isotope-labeled Lys or Arg into proteins during growth. Samples are mixed, then processed together. Corresponding light and heavy peptides have different precursor masses but closely matched chemistry. Ong and colleagues established SILAC as a practical proteomic labeling strategy in 2002.[8]

For matched light/heavy isotopologues:

Abundance ratio ≈ integrated heavy-ion area / integrated light-ion area

Check:
label incorporation, mixing ratio, isotope overlap,
metabolic label conversion, and peptide interference.

TMT and iTRAQ: chemical tags label peptides from different samples. Reporter and balance groups make the tags isobaric, so corresponding labeled peptides combine at the precursor level. Fragmentation releases channel-specific reporter ions. TMT was introduced by Thompson and colleagues in 2003; later tag designs greatly expanded multiplexing.[9]

The strength of shared precursor measurement is also a vulnerability: co-isolated peptides contribute their own reporter ions. This commonly pulls measured ratios toward the mixture average—ratio compression. Narrow isolation windows, improved fractionation, and suitable acquisition methods such as SPS-MS3 can reduce interference, but introduce sensitivity or sampling trade-offs. Reporter isotope impurities also require correction.

AQUA and isotope dilution: add a known amount of a stable-isotope-labeled peptide and compare it with its endogenous counterpart. Gerber and colleagues formalized this targeted strategy in 2003.[10]

Under validated, approximately equal response:

nendogenous ≈ nstandard × (Aendogenous / Astandard)

For protein inference:
account for peptide stoichiometry, digestion yield,
sample recovery, and modifications affecting the target peptide.

A heavy peptide added after digestion controls much of the downstream measurement, but cannot report protein lost during extraction or peptide never generated by incomplete digestion. A labeled protein added earlier can control more steps, though its processing must still resemble that of the endogenous analyte.

SRM/MRM typically uses a triple quadrupole: Q1 selects a precursor, Q2 fragments it, and Q3 selects a product ion. Each precursor–product pair is a transition. PRM isolates a precursor and measures a high-resolution product-ion spectrum. Both require retention-time agreement, appropriate fragment patterns, interference testing, and calibration—not merely a peak at an expected mass.

9. Label-Free Quantification, DDA, and DIA

Label-free quantification usually compares integrated chromatographic precursor-ion areas across runs. Spectral counting is another approach, but is more indirect and depends strongly on sampling, protein size, and abundance. Reliable area-based LFQ requires consistent preparation, retention-time alignment, signal normalization, and attention to missing values.

DDA: data-dependent acquisition
Survey scan → choose selected precursors → fragment individually.

DIA: data-independent acquisition
Step through predefined isolation windows
  → fragment all transmitted precursors in each window
  → computationally resolve overlapping peptide signals.

DDA is like interviewing a few prominent people in a crowd. DIA records overlapping conversations across systematically defined sections of the room, then uses expected voices and timing to separate them. DIA does not eliminate selectivity or interference; it redistributes the problem from precursor selection toward computational deconvolution.

SWATH-style DIA became influential through work such as Gillet and colleagues’ 2012 study.[11] Spectral libraries describe expected fragments and retention behavior. Modern workflows can also use predicted libraries or library-free strategies; a separate experimental DDA library is not always required. DIA-NN exemplifies modern methods for detecting precursors and reducing interference in complex DIA data.[12]

  • Narrower windows reduce mixture complexity but require more windows to cover the same mass range.
  • Cycle time must provide enough measurements across each chromatographic peak for reliable integration.
  • Missingness may depend on abundance and detectability. Replacing every missing value with a small number can manufacture apparent biological changes.
  • Normalization makes assumptions. Median normalization can erase a genuine global shift if most proteins change in the same direction.
  • Biological replicates measure biological variability; repeated injections primarily measure technical variability.

Connect this to statistics: identification FDR and multiple-testing control for differential abundance are separate problems. A confidently identified protein can still have an unreliable treatment-effect estimate.

10. Phosphoproteomics: Measuring a Dynamic Chemical State

Phosphorylation changes molecular charge, binding interactions, and sometimes catalytic activity, often without changing total protein abundance. Phosphoproteomics therefore studies a regulated molecular state rather than simply another list of proteins. Mann and Jensen’s review remains a useful conceptual foundation for PTM analysis.[13]

  1. Preserve the state. Use rapid, consistent sampling and conditions that suppress phosphatase activity.
  2. Digest proteins. Include a total-proteome measurement where possible.
  3. Enrich phosphopeptides. TiO₂ and immobilized metal affinity chromatography exploit phosphate interactions with metal-containing surfaces or immobilized Fe³⁺/Ga³⁺. Anti-phosphotyrosine enrichment addresses the relatively low abundance of phosphotyrosine.
  4. Acquire informative fragmentation. Choose HCD, ETD, or combined methods according to peptide properties and localization needs.
  5. Score localization separately from identity. Tools such as phosphoRS compare evidence for competing site assignments.
  6. Interpret changes against protein abundance and biological timing.
Simplified phosphopeptide signal:

Iphospho ∝ P × θ × Y × R

P = abundance of the parent protein
θ = occupancy of the measured phosphorylation state
Y = preparation/enrichment recovery
R = instrumental response

A twofold phosphopeptide increase could reflect increased occupancy, increased protein abundance, or both. Even protein-normalized phosphopeptide ratios do not generally establish absolute site occupancy. Occupancy estimation requires suitable measurement of relevant modified and unmodified forms, including their unequal responses.

Application to GLP-1R and MC4R signaling: these receptors can engage cAMP-linked signaling networks, but a phosphosite change after receptor stimulation does not prove direct phosphorylation by PKA or another particular kinase. Motifs, kinetics, selective perturbations, and orthogonal biochemical experiments are needed to infer mechanism.

Important timing distinction: rapid phosphorylation can occur within seconds to minutes. Pulsed-SILAC tracks synthesis of new protein and is especially useful for synthesis and turnover studies; it is not, by itself, a direct clock for phosphate addition. Rapid signaling experiments generally need carefully timed stimulation and quenching.

11. Glycoproteomics and Other PTMs: Two Layers of Identity

An intact glycopeptide contains two linked questions: which peptide and site carry the modification, and which glycan is attached? Glycan heterogeneity creates many molecular forms from one peptide sequence, while structural isomers may share the same mass.

  1. Enrich. HILIC favors hydrophilic species, including many glycopeptides. Lectins recognize selected glycan motifs and therefore introduce motif-specific sampling bias.
  2. Choose released-glycan or intact-glycopeptide analysis. Released glycans simplify glycan profiling but lose site connectivity unless the design preserves it separately.
  3. Use complementary fragmentation. HCD often generates glycan fragments and diagnostic oxonium ions; ETD-based methods can preserve glycan attachment during backbone cleavage.
  4. Assign composition, site, and structure at their actual evidence levels. A glycan composition is not a fully determined linkage and branching structure.

PNGase F releases many N-linked glycans and converts the glycosylated Asn to Asp, producing an approximately +0.9840 Da peptide mass shift. Spontaneous deamidation can produce the same nominal chemical change, so the shift alone is not proof of prior glycosylation. PNGase F also has substrate limitations, including certain core-fucosylated glycans.

O-GalNAc glycosylation is especially difficult because Ser/Thr-rich regions can contain several candidate sites and there is no single broadly applicable enzyme equivalent to PNGase F for all O-glycans. For therapeutic glycoproteins, site-specific profiles help connect manufacturing conditions with receptor interactions, circulating lifetime, and product consistency.

Ubiquitin-remnant profiling: tryptic digestion of ubiquitinated proteins often leaves a diglycine remnant on the modified lysine. Anti-K-ε-GG enrichment captures peptides bearing this +114.0429 Da remnant. However, related modifiers such as NEDD8 and ISG15 can produce the same remnant. Nor does a substrate remnant alone establish ubiquitin-chain length or linkage topology.

Lysine acylations: antibody enrichment enables acetylome and related studies, subject to antibody bias. High mass accuracy helps distinguish acetylation (+42.0106 Da) from trimethylation (+42.0470 Da). For densely modified histones, middle-down or top-down methods are especially useful because separate bottom-up peptides cannot establish distant PTM co-occurrence.

12. HDX-MS: Reading Protein Dynamics Through Solvent Exchange

Hydrogen–deuterium exchange MS connects directly to protein folding thermodynamics. Backbone amide hydrogens exchange with solvent through chemistry that depends on pH, temperature, sequence, hydrogen bonding, and structural opening. A protected amide is not simply “buried”; it may be unable to access an exchange-competent conformation.

Minimal exchange model:

Closed ⇌ Open → Exchanged
       kop  kcl   kch

EX2 limit: kcl ≫ kch
kobs ≈ Kop × kch

Protection factor:
PF = kint / kobs

Under suitable EX2 assumptions:
ΔGopening ≈ RT ln(PF)

Here, kint is the intrinsic exchange rate of an appropriate unprotected reference state. The free-energy interpretation requires appropriate kinetic assumptions; it is not valid for every peptide uptake curve. In the EX1 limit, exchange during an opening event is faster than closing, and opening kinetics can dominate the observed behavior.

  1. Expose the protein to D₂O-containing buffer for a defined labeling interval.
  2. Quench exchange using low temperature and pH near 2.5.
  3. Digest rapidly with an acid-active protease, commonly pepsin.
  4. Separate peptides under cold, low-pH conditions.
  5. Measure isotope-envelope shifts and compare uptake over time between states.

Routine peptide-resolved HDX typically quantifies mass shifts in MS1; peptide identities can be established separately. MS/MS localization of deuterium requires special care because collisional activation can redistribute deuterium through scrambling. Suitable electron-based methods may help, but must be validated.

Applications: antibody epitope mapping, ligand-binding studies, biosimilar comparisons, and detection of allosteric changes. Decreased uptake after antibody binding can indicate direct contact or a remote stabilization. It is therefore a regional dynamic constraint, not an automatic atomic contact map.

Back-exchange during handling removes some incorporated deuterium. Matched handling, controls, replication, peptide coverage, and transparent reporting are essential; community recommendations emphasize precisely these points.[14]

13. Crosslinking MS: Capturing Proximity, Not a Static Photograph

Crosslinking MS creates covalent connections between residues that come within the chemical reach of a reagent. It complements structural biology by providing sparse restraints in solution, large assemblies, or cells.

  1. React under controlled conditions. DSS uses NHS esters to target accessible primary amines, commonly lysine side chains and protein N-termini. EDC promotes coupling between carboxyl and amino groups without a retained spacer.
  2. Quench and digest. The digest contains linear peptides, monolinks, intrapeptide links, and interpeptide links.
  3. Enrich and fragment crosslinked peptides. Identification must account for two connected peptide sequences and the linker chemistry.
  4. Control identification errors. The paired-peptide search space is large; link-level and interaction-level confidence need explicit assessment.
  5. Integrate restraints with models. Use crosslinks to test arrangements rather than forcing all links into one rigid structure.

A linker’s spacer length is not its maximum Cα–Cα distance. Side-chain lengths, conformational flexibility, attachment geometry, and solvent-accessible paths also matter. Photo-crosslinkers can trap transient proximities after irradiation, but do not make every captured contact a functionally important interaction.

The conceptual connection to cryo-EM and NMR is powerful: XL-MS supplies incomplete but complementary constraints. A surprising crosslink can indicate an incorrect model, a false identification, an alternative assembly, or a transient conformation. Structural inference must distinguish among those possibilities.

14. Clinical Proteomics: Discovery Is Not Yet a Diagnostic Test

Plasma proteomics is difficult because a few abundant proteins dominate the sample while many informative signaling or disease-associated proteins occur at much lower concentrations. Depletion and fractionation improve access to low-abundance species, but depletion can also remove proteins associated with the targeted abundant carriers.

Clinical translation:

Discovery
  → candidate prioritization
  → targeted analytical verification
  → independent clinical validation
  → assessment of clinical utility
  • Analytical validity: does the assay measure the intended analyte accurately and reproducibly? Examine precision, bias, selectivity, calibration, matrix effects, stability, and limits of detection and quantification.
  • Clinical validity: does the result discriminate or predict the relevant clinical state in the intended population?
  • Clinical utility: does using the result improve a clinical decision or patient outcome?

SRM and PRM are valuable for verifying biomarker candidates in larger sample sets. Orthogonal samples, pre-specified analysis plans, and control of collection and storage conditions matter as much as instrumental sensitivity. Hemolysis, clotting time, medication use, and batch structure can create convincing but noncausal “biomarkers.”

Positive predictive value:

PPV = [sensitivity × prevalence] /
      [sensitivity × prevalence
       + (1 − specificity) × (1 − prevalence)]

Example:
Sensitivity = 0.90
Specificity = 0.95
Prevalence  = 0.01

PPV ≈ 0.154, or about 15%.

This is why impressive case–control discrimination does not automatically support population screening.

Regulatory precision: LC-MS/MS is widely used clinically for vitamin D metabolites, testosterone, and immunosuppressant drugs, but these are largely small-molecule assays rather than proteomics. Regulatory authorization applies to a specific assay, platform, jurisdiction, and intended use; one must not assume that every clinical LC-MS/MS method is FDA-cleared. Many are laboratory-developed tests.

15. Peptide and Pharma Applications: Distinguishing Identity, Purity, and Mechanism

For a therapeutic peptide, the question is not merely “Is the expected mass present?” It is “What molecular species are present, in what amounts, and could they change potency, stability, exposure, or safety?”

  1. Confirm intact mass. Account for terminal amidation, cyclization, disulfides, lipidation, and other intended modifications.
  2. Map sequence by MS/MS. Seek coverage that distinguishes the intended sequence from plausible synthesis impurities.
  3. Profile impurities chromatographically. Examine deletion products, incomplete deprotection, oxidation, deamidation, truncation, and conjugation heterogeneity.
  4. Use orthogonal methods for invisible distinctions. Ordinary mass measurement cannot distinguish epimers; isobaric sequence variants and some positional isomers also need additional evidence.
  5. Quantify using validated response relationships. LC-MS ion-area percentages are not automatically mass-percent purity because impurities can have different ionization responses.
Useful peptide mass changes:

Oxidation, addition of one O:          +15.9949 Da
Asn/Gln deamidation:                    +0.9840 Da
Disulfide formation from two thiols:    −2.0157 Da
C-terminal acid → amide:                −0.9840 Da

These shifts constrain hypotheses; they do not locate changes by themselves.

Disulfide mapping: digest a non-reduced sample to retain disulfide-linked peptide pairs, then compare with a reduced sample in which those peptides separate. The intact mass can support the expected number of disulfides but cannot determine their connectivity. Free-thiol blocking and carefully controlled preparation help prevent artificial disulfide scrambling.

Peptide-drug metabolism: compare biological incubations or post-dose samples with appropriate controls. LC-MS/MS can detect proteolytic truncations, oxidation, deamidation, or loss of a lipid-containing group. Accurate precursor mass suggests a transformation; fragments locate the change. Authentic standards and retention-time matching strengthen assignments, and uncalibrated ion areas do not establish metabolite molar yields.

PEGylation and lipidation: intact mass can confirm the overall modification, but site assignment requires mapping. Conventional PEG often creates a distribution of chain lengths, with approximately 44.0262 Da between successive ethylene oxide repeat units. A broad envelope may therefore represent genuine product heterogeneity rather than poor instrument performance.

Connection to peptide synthesis and protein engineering: MS is most useful when the impurity hypotheses are chemically informed. Knowing the coupling chemistry, protecting groups, oxidation sensitivity, and conjugation strategy tells us which alternative structures the experiment must discriminate.

16. Historical Perspective and the Habit to Carry Forward

Early mass analysis—including J. J. Thomson’s positive-ray work and Aston’s mass spectrograph—helped establish isotopes and accurate atomic masses. Modern proteomics required another conceptual leap: making large biological molecules available as ions. MALDI and ESI supplied complementary solutions; database searching turned spectra into scalable identifications; electron-based fragmentation preserved vulnerable molecular features; isotope labeling and DIA expanded quantitative comparisons.

The history is not simply a march toward “more sensitivity.” Each advance changed which biological questions could be asked:

  • Soft ionization: can an intact biomolecule be measured at all?
  • Tandem MS and searching: which sequence best explains the ions?
  • Quantitative standards: how much analyte was present?
  • PTM and proteoform analysis: which molecular state was present?
  • HDX and crosslinking: how was the molecule organized and how did it respond?

Skipper’s final working rule: for every conclusion, name the measured observable, the chemical mechanism connecting it to the claim, and the strongest alternative explanation. That habit turns mass spectrometry from a peak-generating technology into rigorous biochemistry.

References

Bibliographic titles and DOI links below use the published records; these correct discrepancies in several supplied reading-list entries.

  1. Aebersold, R., and Mann, M. (2003). Mass spectrometry-based proteomics. Nature 422, 198–207. doi:10.1038/nature01511.
  2. Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., and Whitehouse, C. M. (1989). Electrospray ionization for mass spectrometry of large biomolecules. Science 246, 64–71. doi:10.1126/science.2675315.
  3. Karas, M., and Hillenkamp, F. (1988). Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Analytical Chemistry 60, 2299–2301. doi:10.1021/ac00171a028.
  4. Zubarev, R. A., Kelleher, N. L., and McLafferty, F. W. (1998). Electron capture dissociation of multiply charged protein cations. A nonergodic process. Journal of the American Chemical Society 120, 3265–3266. doi:10.1021/ja973478k.
  5. Syka, J. E. P., Coon, J. J., Schroeder, M. J., Shabanowitz, J., and Hunt, D. F. (2004). Peptide and protein sequence analysis by electron transfer dissociation mass spectrometry. PNAS 101, 9528–9533. doi:10.1073/pnas.0402700101.
  6. Eng, J. K., McCormack, A. L., and Yates, J. R. III (1994). An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. Journal of the American Society for Mass Spectrometry 5, 976–989. doi:10.1016/1044-0305(94)80016-2.
  7. Elias, J. E., and Gygi, S. P. (2007). Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry. Nature Methods 4, 207–214. doi:10.1038/nmeth1019.
  8. Ong, S.-E., et al. (2002). Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Molecular & Cellular Proteomics 1, 376–386. doi:10.1074/mcp.M200025-MCP200.
  9. Thompson, A., et al. (2003). Tandem mass tags: A novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS. Analytical Chemistry 75, 1895–1904. doi:10.1021/ac0262560.
  10. Gerber, S. A., Rush, J., Stemman, O., Kirschner, M. W., and Gygi, S. P. (2003). Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. PNAS 100, 6940–6945. doi:10.1073/pnas.0832254100.
  11. Gillet, L. C., et al. (2012). Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: A new concept for consistent and accurate proteome analysis. Molecular & Cellular Proteomics 11, O111.016717. doi:10.1074/mcp.O111.016717.
  12. Demichev, V., Messner, C. B., Vernardis, S. I., Lilley, K. S., and Ralser, M. (2020). DIA-NN: Neural networks and interference correction enable deep proteome coverage in high throughput. Nature Methods 17, 41–44. doi:10.1038/s41592-019-0638-x.
  13. Mann, M., and Jensen, O. N. (2003). Proteomic analysis of post-translational modifications. Nature Biotechnology 21, 255–261. doi:10.1038/nbt0303-255.
  14. Masson, G. R., et al. (2019). Recommendations for performing, interpreting and reporting hydrogen deuterium exchange mass spectrometry (HDX-MS) experiments. Nature Methods 16, 595–602. doi:10.1038/s41592-019-0459-y.

🪬 Skipper's Reflection

Gemini 3.1 Pro

What Surprised Me Most

If you had asked me before this course how we measure the mass of a large biomolecule, I would have pictured something blunt and destructive. In organic chemistry, electron ionization shatters small molecules into pieces. The idea of doing mass spectrometry on a 150 kDa monoclonal antibody or a heavily modified peptide drug seemed impossible—wouldn't it just vaporize into atomic dust? The greatest surprise of BCHM613 was the exquisite gentleness of "soft ionization" techniques. Reading Fenn et al. (1989) on Electrospray Ionization (ESI) and Karas & Hillenkamp (1988) on MALDI completely flipped my perspective.

The physics of ESI is downright poetic. We force a liquid through a capillary at high voltage, creating a Taylor cone that emits charged droplets. As the solvent evaporates, charge repulsion overcomes surface tension (the Rayleigh limit), leading to a "Coulombic explosion." It’s basically controlled droplet fission until a naked, multiply-charged peptide is gently lofted into the vacuum of the mass spectrometer. It blew my mind that we can transition massive, fragile biomolecules from the liquid phase to the gas phase without breaking a single covalent bond. Biology operates in water, but we can literally weigh its components in a vacuum.

Conceptual Breakthrough Moments

My biggest "aha!" moment occurred during Week 4, studying tandem MS (MS/MS) and the generation of b- and y-ion series. Before this, sequencing a peptide felt like a wet-lab chore (like Edman degradation). But Collision-Induced Dissociation (CID) turns biochemistry into a pure mathematical cryptography puzzle. Because CID preferentially breaks the lowest-energy bonds—the peptide amide bonds—the resulting fragmentation spectrum is a ladder. Seeing a spectrum where the mass difference between two adjacent peaks is exactly 129.04 Da (a Glutamic acid residue) or 147.07 Da (a Phenylalanine residue) was a beautiful moment of clarity. The molecule literally writes its own sequence in the mass spectrum.

A second breakthrough was Hydrogen-Deuterium Exchange Mass Spectrometry (HDX-MS) in Week 12. I had always thought of mass spec as a tool for identity (what is this molecule?). HDX-MS uses mass spec for structure and dynamics. By placing a protein in D₂O and watching its mass increase over time as deuterium exchanges with backbone amide hydrogens, we get a time-lapse map of protein "breathing." Unstructured loops gain mass instantly; buried alpha-helices gain mass slowly. It connects the static structures I learned in structural biology directly to solution-phase thermodynamics.

Connecting the Dots to the Peptide World

This course completely transformed how I view the Quality Control and structural verification of the therapeutic peptides I work with.

Take Tirzepatide and Retatrutide. These aren't just simple linear amino acid chains; they are heavily engineered with fatty acid diacids attached via complex linkers (like gamma-Glu and PEG spacers). How do we prove the manufacturer actually attached the lipid to the correct Lysine residue, and not a different one? Bottom-up MS/MS is the answer. By digesting the peptide and running it through the Orbitrap, we can look for the specific shifted y-ion or b-ion that carries the massive lipid modification. This directly relates to Mann and Jensen’s (2003) work on localizing Post-Translational Modifications (PTMs).

For something like BPC-157 or TB-500, MALDI-TOF (Week 3) is a rapid lifesaver for crude synthesis QC. If I synthesize BPC-157 and the MALDI spectrum shows a peak exactly 56 Da higher than the theoretical mass, I instantly know I have incomplete removal of a tert-butyl protecting group from my solid-phase synthesis. Furthermore, the Week 14 material on disulfide mapping is critical for cyclic peptides. Digesting a peptide with and without a reducing agent (like DTT) and watching the mass shifts tells us exactly if the disulfide bridges formed correctly, or if we accidentally created a scrambled polymer.

Questions for Further Investigation

1. The Deep Learning Revolution in Data-Independent Acquisition (DIA): We touched on SWATH-MS and DIA-NN software. In DIA, we fragment everything at once, creating massively complex, overlapping MS/MS spectra. How far can neural networks go in deconvoluting this? Can AI predict peptide fragmentation patterns so perfectly that we won't even need experimental spectral libraries anymore?

2. Top-Down Proteomics and the "Combinatorial Explosion" of PTMs: We learned that top-down MS (analyzing intact proteins) is great for preserving the context of multiple PTMs. But if a protein has 10 potential phosphorylation sites and 5 acetylation sites, the number of possible proteoforms is astronomical. How does the mass spectrometer's dynamic range not completely fail when the ion current is divided among thousands of different proteoform envelopes?

Confidence Rating

Most Confident:
I feel incredibly solid on MALDI/ESI mechanisms, MS/MS fragmentation nomenclature (b/y ions, CID vs. ETD), and peptide drug QC (Week 14). Because I have a strong background in organic synthesis and biochemistry, the logic of how peptides break apart and how to spot impurities maps perfectly onto my existing knowledge.

Least Confident:
I am still wrapping my head around the deep physics of Orbitrap mass analyzers and advanced multiplexed quantitative proteomics (TMT 16-plex). The Orbitrap uses a central spindle electrode to trap ions in orbital motion, and the mass is derived by applying a Fourier Transform to the image current generated by their axial oscillations. I understand it conceptually, but the underlying electrodynamics feels like dark magic. Furthermore, normalizing data across 16 different samples in a TMT multiplex experiment involves statistical corrections for "ratio compression" that I need to practice hands-on to truly master.

🗺️ Visual Aids & Key Takeaways

Claude Opus 5.5

Mass Spectrometry Ion Source Comparison: ESI vs. MALDI

╔══════════════════════════════════════════════════════════════════════════════════════╗
║              ION SOURCE COMPARISON: ESI vs. MALDI                                  ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

┌─────────────────────────────────────┐    ┌─────────────────────────────────────────┐
│  ELECTROSPRAY IONIZATION (ESI)      │    │  MATRIX-ASSISTED LASER DESORPTION/      │
│                                     │    │  IONIZATION (MALDI)                     │
├─────────────────────────────────────┤    ├─────────────────────────────────────────┤
│                                     │    │                                         │
│  SOLUTION ──→ TAYLOR CONE           │    │  LASER PULSE (337 nm UV)                │
│                  │                  │    │       ↓                                 │
│                  ↓                  │    │  MATRIX CRYSTALS (CHCA, DHB, SA)        │
│          CHARGED DROPLETS           │    │  co-crystallized with analyte           │
│                  │                  │    │       ↓                                 │
│    solvent evaporates; droplets     │    │  ENERGY ABSORBED BY MATRIX              │
│    shrink until Coulombic repulsion │    │  matrix shields analyte from            │
│    overcomes surface tension        │    │  direct photodamage                     │
│                  │                  │    │       ↓                                 │
│          COULOMBIC EXPLOSION        │    │  DESORPTION + PROTON TRANSFER           │
│                  │                  │    │  to analyte molecule                    │
│                  ↓                  │    │       ↓                                 │
│         GAS-PHASE IONS              │    │  GAS-PHASE IONS                         │
│                                     │    │                                         │
├─────────────────────────────────────┤    ├─────────────────────────────────────────┤
│  CHARGE STATE:  MULTIPLY CHARGED    │    │  CHARGE STATE:  SINGLY CHARGED (+1)     │
│  e.g., [M+10H]¹⁰⁺ for a protein    │    │  rarely doubly charged                  │
│                                     │    │                                         │
│  SAMPLE FORM:   solution (LC eluent)│    │  SAMPLE FORM:   solid co-crystal        │
│                                     │    │                                         │
│  COUPLING:      online LC-MS ✓✓✓   │    │  COUPLING:      offline; spotted plate  │
│                                     │    │                                         │
│  BEST FOR:      peptides, proteins, │    │  BEST FOR:      peptide mass            │
│                 intact native MS,   │    │                 fingerprinting, QC of   │
│                 PTM analysis        │    │                 synthetic peptides,      │
│                                     │    │                 large proteins (linear) │
│  MASS RANGE:    unlimited in theory │    │  MASS RANGE:    up to >1,000,000 Da     │
│                 (deconvolution req) │    │                 in linear mode           │
└─────────────────────────────────────┘    └─────────────────────────────────────────┘

         KEY DISTINCTION:
         ESI produces multiply charged ions → m/z shifted into detector range
         MALDI produces singly charged ions → m/z = true mass (approx.)

ESI and MALDI differ fundamentally in how they transfer molecules into the gas phase as ions — ESI from solution via droplet desolvation, MALDI from a solid crystal via pulsed laser energy — with profound consequences for charge state, coupling to chromatography, and analytical application.

Bottom-Up Proteomics Workflow: From Cell to Protein Identification

╔══════════════════════════════════════════════════════════════════════════════════════╗
║          BOTTOM-UP PROTEOMICS: COMPLETE ANALYTICAL WORKFLOW                        ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

  ┌──────────────────┐
  │   CELL / TISSUE  │  ← starting material: lysate, plasma, organelle fraction
  └────────┬─────────┘
           │ lysis, denaturation (urea, SDS), reduction + alkylation
           ↓
  ┌──────────────────┐
  │  PROTEIN MIXTURE │  ← thousands of proteins, wide dynamic range (~10⁶ in cell)
  └────────┬─────────┘
           │ enzymatic digestion
           │   TRYPSIN  → cleaves after Lys, Arg (except before Pro)
           │   Lys-C    → cleaves after Lys only
           │   Asp-N    → cleaves before Asp
           │   Glu-C    → cleaves after Glu (or Asp, pH dependent)
           ↓
  ┌──────────────────┐
  │ PEPTIDE MIXTURE  │  ← 10,000–100,000+ peptides; mostly 6–20 residues
  └────────┬─────────┘
           │ optional enrichment (phospho: TiO₂ / IMAC; glyco: HILIC; Ub: anti-GG Ab)
           │ optional fractionation (SCX, high-pH RPLC, SDS-PAGE)
           ↓
  ┌──────────────────────────────────┐
  │  NANO-LC SEPARATION              │  ← reversed-phase C18, 75 µm column
  │  (75–120 min gradient)           │     flow rate ~200–400 nL/min (nano-ESI)
  └────────────────┬─────────────────┘
                   │ online coupling
                   ↓
  ┌──────────────────────────────────┐
  │  MASS SPECTROMETER               │
  │                                  │
  │  MS1 SCAN  → intact peptide m/z  │  ← full scan; identifies precursor ions
  │      │       (charge states)     │
  │      │ precursor selection (DDA) │
  │      ↓                           │
  │  MS2 SCAN  → fragmentation       │  ← CID, HCD, ETD
  │             b-ions / y-ions      │     generates sequence-informative fragments
  └────────────────┬─────────────────┘
                   │ raw spectra
                   ↓
  ┌──────────────────────────────────┐
  │  DATABASE SEARCH                 │
  │  (Mascot / SEQUEST / MaxQuant)   │  ← in silico digestion of reference proteome
  │                                  │     match observed MS2 to theoretical spectra
  │  TARGET-DECOY FDR CONTROL        │  ← reversed/shuffled decoy database
  │  peptide FDR ≤ 1%, protein ≤ 1% │
  └────────────────┬─────────────────┘
                   ↓
  ┌──────────────────────────────────┐
  │  PROTEIN IDENTIFICATIONS         │  → gene ontology, pathway, network analysis
  │  + QUANTIFICATION (if labeled    │
  │    or LFQ)                       │
  └──────────────────────────────────┘

The bottom-up proteomics workflow converts complex protein mixtures into identifiable peptides through enzymatic digestion, separates them by nano-LC, and identifies them by matching MS/MS fragmentation spectra against a theoretical protein database with statistical FDR control at each level.

Peptide Fragmentation: b-Ion and y-Ion Series Nomenclature

╔══════════════════════════════════════════════════════════════════════════════════════╗
║         PEPTIDE FRAGMENTATION: b/y ION NOMENCLATURE AND SERIES                    ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

  Example peptide:   A  -  L  -  K  -  E  -  G  -  V  -  R
                    Ala  Leu  Lys  Glu  Gly  Val  Arg

  ┌──────────────────────────────────────────────────────────────────────────┐
  │                                                                          │
  │    H₂N─[ALA]─CO─NH─[LEU]─CO─NH─[LYS]─CO─NH─[GLU]─CO─NH─[GLY]─CO─NH─[VAL]─CO─NH─[ARG]─COOH
  │                                                                          │
  │         ↑      ↑          ↑          ↑          ↑          ↑            │
  │        b₁     b₂         b₃         b₄         b₅         b₆           │
  │    (N-terminal fragment ions retain charge on N-terminal side)           │
  │                                                                          │
  │                  ↑          ↑          ↑          ↑          ↑     ↑    │
  │                 y₆         y₅         y₄         y₃         y₂   y₁    │
  │    (C-terminal fragment ions retain charge on C-terminal side)           │
  │                                                                          │
  └──────────────────────────────────────────────────────────────────────────┘

  FRAGMENTATION RULES:
  ┌─────────────────────────────────────────────────────────────────────────┐
  │                                                                         │
  │   b-ions:  N-terminal fragments                                         │
  │            mass = sum of residue masses from N-term + 1 (H)            │
  │            b₁ is rarely observed (usually lost as immonium)             │
  │                                                                         │
  │   y-ions:  C-terminal fragments                                         │
  │            mass = sum of residue masses from C-term + 18 (H₂O) + 1 (H) │
  │            tryptic peptides end in K or R → y-ions often intense        │
  │                                                                         │
  │   RELATIONSHIP:   bₙ + y(total−n) = precursor mass + H₂O               │
  │                                                                         │
  │   a-ions:  b-ions minus CO (−28 Da); less common                        │
  │   c-ions:  N-terminal + NH₃ (+17 Da); produced by ETD/ECD              │
  │   z-ions:  C-terminal − NH₂ (−16 Da); produced by ETD/ECD              │
  │                                                                         │
  └─────────────────────────────────────────────────────────────────────────┘

  DISSOCIATION METHOD → ION TYPES PRODUCED:
  ┌────────────────┬──────────────────┬───────────────────────────────────┐
  │  METHOD        │  PRIMARY IONS    │  BEST APPLICATION                 │
  ├────────────────┼──────────────────┼───────────────────────────────────┤
  │  CID / HCD     │  b and y ions    │  standard peptide sequencing      │
  │  ETD / ECD     │  c and z ions    │  PTM localization (phospho, glyco)│
  │  UVPD          │  a,b,c,x,y,z     │  top-down, complex PTMs           │
  │  EThcD         │  b,y,c,z (mixed) │  glycopeptides, charge-state flex │
  └────────────────┴──────────────────┴───────────────────────────────────┘

When a peptide is fragmented in the mass spectrometer, the amide backbone cleaves to produce b-ions (N-terminal) and y-ions (C-terminal) whose mass differences directly encode the amino acid sequence, while the choice of fragmentation method determines which ion series dominate and how well labile modifications are preserved.

Mass Analyzer Comparison: Quadrupole, TOF, Ion Trap, and Orbitrap

╔══════════════════════════════════════════════════════════════════════════════════════╗
║            MASS ANALYZER PROPERTIES: A PRACTICAL COMPARISON                       ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

                        QUADRUPOLE     TOF          ION TRAP      ORBITRAP
                       ───────────   ────────      ──────────    ──────────
  RESOLVING POWER        ~2,000       ~20,000       ~5,000        >100,000
  (R = m/Δm)

  MASS ACCURACY          ~100 ppm     ~5–10 ppm     ~100 ppm      <2 ppm
  (external calibration)              (reflector)                (internal cal)

  SCAN SPEED             fast         very fast     fast          moderate
                         (µs filter)  (parallel)    (ms–s)        (ms per scan)

  MASS RANGE             ~4,000 Da    >1,000,000 Da ~4,000 Da     ~4,000 Da

  MS/MS CAPABLE?         yes (triple  yes (Q-TOF    yes (MSⁿ      yes (via HCD
                         quad: SRM)   or TOF/TOF)   in trap)      or ion trap)

  QUANTIFICATION         excellent    good          good          excellent
  (dynamic range)        (SRM/MRM)    (DIA)                       (survey scan)

  KEY STRENGTH           selectivity  speed +       MSⁿ           high resolving
                         for SRM      mass range    capability    power + accuracy

  KEY WEAKNESS           low          limited       space charge  cost; slow
                         resolution   MS/MS         effects       scan rate

  ─────────────────────────────────────────────────────────────────────────────

  COMMON INSTRUMENT CONFIGURATIONS IN PROTEOMICS:

    TRIPLE QUADRUPOLE  (Q1 ─ q2 collision cell ─ Q3)
         → SRM/MRM quantification; clinical assays; targeted proteomics

    Q-TOF              (Q1 ─ collision cell ─ TOF)
         → accurate mass MS/MS; peptide ID with good mass accuracy

    ORBITRAP FUSION / ECLIPSE  (Orbitrap survey + ion trap or HCD MS/MS)
         → gold standard for discovery proteomics; >10,000 spectra/hr

    QTOF (SWATH / DIA)
         → data-independent acquisition; complete peptide coverage

Each mass analyzer makes different tradeoffs between resolving power, mass accuracy, scan speed, and mass range — the Orbitrap dominates discovery proteomics for its sub-2 ppm mass accuracy, while the triple quadrupole remains the gold standard for targeted quantification via SRM/MRM.

Quantitative Proteomics Strategies: Isotope Labeling and Label-Free

╔══════════════════════════════════════════════════════════════════════════════════════╗
║         QUANTITATIVE PROTEOMICS: STRATEGY COMPARISON                              ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

  STRATEGY 1: SILAC (Stable Isotope Labeling by Amino Acids in Cell Culture)
  ─────────────────────────────────────────────────────────────────────────────
  CONDITION A               CONDITION B               CONDITION C
  (light Lys/Arg)           (medium Lys⁴, Arg⁶)       (heavy Lys⁸, Arg¹⁰)
       │                         │                         │
       └─────────────────────────┴─────────────────────────┘
                                 │ mix all three; digest together
                                 ↓
                     CO-ELUTING ISOTOPE PAIRS
                     quantified at MS1 level by
                     peak area ratios (ratio accuracy ~5%)
                     multiplexing: up to 3-plex in vivo

  STRATEGY 2: TMT / iTRAQ (Isobaric Chemical Labeling — post-digestion)
  ─────────────────────────────────────────────────────────────────────────────
  Sample 1   Sample 2 ... Sample 16 (TMT 16-plex)
     │            │               │
     ↓            ↓               ↓
  TMT-126     TMT-127         TMT-134  ← different mass reporter groups
     │            │               │
     └────────────┴───────────────┘
              mix; analyze together
              │
              ↓  MS1: all channels identical m/z (isobaric!)
              ↓  MS2/MS3: reporter ions 126–134 Da quantified
                           (ratio ~10–30% CV; ratio compression issue in MS2)

  STRATEGY 3: LABEL-FREE QUANTIFICATION (LFQ)
  ─────────────────────────────────────────────────────────────────────────────
  Sample A      Sample B      Sample C
  (run 1)       (run 2)       (run 3)
     │              │              │
     ↓              ↓              ↓
  PEAK AREA     PEAK AREA     PEAK AREA
  integration   integration   integration
     │              │              │
     └──────────────┴──────────────┘
         alignment + normalization (MaxQuant LFQ algorithm)
         requires high LC reproducibility (CV < 5% retention time)

  STRATEGY 4: TARGETED SRM/MRM (Absolute Quantification with AQUA Peptides)
  ─────────────────────────────────────────────────────────────────────────────
  Heavy-labeled synthetic peptide (known amount, AQUA)
  spiked into digest at known concentration
            ↓
  Triple quadrupole: Q1 selects precursor → q2 fragments → Q3 selects product ion
  ratio of endogenous signal / AQUA signal = absolute amount (attomole sensitivity)

  ─────────────────────────────────────────────────────────────────────────────
  COMPARISON SUMMARY:
  ┌──────────────┬──────────────┬───────────────┬────────────┬──────────────┐
  │  METHOD      │  MULTIPLEXING│  QUANTIFIED   │  ACCURACY  │  LIMITATION  │
  ├──────────────┼──────────────┼───────────────┼────────────┼──────────────┤
  │  SILAC       │  2–3 samples │  MS1 level    │  good      │  cell culture│
  │  TMT/iTRAQ   │  6–16 samples│  MS2/MS3      │  moderate  │  ratio compr.│
  │  LFQ         │  unlimited   │  MS1 level    │  moderate  │  LC stability│
  │  SRM (AQUA)  │  ~100 targets│  MS2 level    │  excellent │  pre-specify │
  └──────────────┴──────────────┴───────────────┴────────────┴──────────────┘

Quantitative proteomics offers four fundamentally different strategies — SILAC metabolically incorporates mass-shifted amino acids, TMT/iTRAQ uses isobaric chemical tags that are distinguished only at the MS2 reporter level, LFQ compares peak areas across separate LC-MS runs, and SRM with AQUA spike-ins achieves absolute quantification — each balancing multiplexing capacity, accuracy, and experimental feasibility differently.

Key Takeaways

  1. ESI produces multiply charged ions; MALDI produces predominantly singly charged ions. In ESI, a protein of mass M acquires n protons to give [M+nH]n+, observed at m/z = (M + n·1.0073)/n — well within the detector range of any instrument. This is not a limitation but a design advantage: a 50 kDa protein with 40 charges appears at m/z ~1,250. MALDI's singly charged ions mean the instrument must measure the true mass directly, which is why MALDI-TOF in linear mode can detect proteins >100,000 Da (Karas and Hillenkamp, 1988).
  2. The Orbitrap measures ion frequency in a harmonic electrostatic field, achieving resolving power >100,000 and mass accuracy <2 ppm. Ions orbit around a central electrode and oscillate axially at a frequency ω = (k/m·z)½ — directly proportional to (z/m)½. A Fourier transform converts the image current signal into a mass spectrum. This sub-2 ppm accuracy is sufficient to distinguish elemental compositions, resolve isotope envelopes, and detect mass shifts from PTMs with confidence impossible on ion traps or single quadrupoles (Aebersold and Mann, 2003).
  3. Trypsin is the workhorse protease of bottom-up proteomics because it cleaves C-terminal to Lys and Arg (but not before Pro), generating peptides of 6–20 residues that are ideally suited for LC-MS/MS. These peptides carry a basic residue at the C-terminus, which directs fragmentation to produce long, informative y-ion series — critical for database matching. Tryptic peptides are also the right size to elute over a broad RPLC gradient and produce multiply charged precursor ions under ESI conditions (Liebler, 2002).
  4. CID/HCD produces b-ions and y-ions; ETD/ECD produces c-ions and z-ions — and the choice of fragmentation method fundamentally determines PTM localization ability. In CID, the labile phosphate group on Ser/Thr is frequently lost as a neutral loss (−98 Da, H3PO4) before backbone cleavage, obscuring the modification site. ETD/ECD fragment the backbone via radical mechanisms that leave labile PTMs intact, making them the methods of choice for phosphoproteomics and glycoproteomics (Mann and Jensen, 2003).
  5. The target-decoy strategy is the standard method for controlling false discovery rate (FDR) in database searches. A decoy database of reversed or randomized protein sequences is searched simultaneously with the target database. Peptide spectrum matches (PSMs) to decoys are false positives by definition; the ratio of decoy hits to target hits at any score threshold estimates the FDR. A 1% FDR at the peptide level means ~1% of accepted PSMs are incorrect — the standard threshold for publication-quality proteomics data (Liebler, 2002; Aebersold and Mann, 2003).
  6. SILAC (Stable Isotope Labeling with Amino Acids in Cell Culture) encodes quantitative information at the metabolic level, before any sample preparation error is introduced. Cells grown in heavy Lys (¹³C₆¹⁵N₂, +8 Da) and heavy Arg (¹³C₆¹⁵N₄, +10 Da) incorporate these into all newly synthesized proteins. After mixing with light-labeled control cells and digesting together, co-eluting peptide pairs are distinguished in MS1 by their mass shift and quantified by peak area ratios. Because the samples are mixed before digestion, variability from sample handling is minimized — a fundamental conceptual advantage over post-digestion labeling (Aebersold and Mann, 2003).
  7. TMT isobaric tags allow up to 16-plex quantification but suffer from ratio compression in MS2 mode due to co-isolation of interfering precursors. Because all TMT-labeled samples have identical precursor m/z (isobaric), they cannot be distinguished at MS1; reporter ions at 126–134 Da are only released and measured at MS2. Co-isolated contaminant peptides from adjacent m/z contribute reporter signal from all channels, artificially compressing measured ratios toward 1:1. MS3 (SPS-MS3 on Orbitrap Fusion instruments) largely solves this by selecting only the desired product ions for a second round of fragmentation before reporter ion quantification.
  8. Phosphopeptide enrichment by TiO2 or IMAC is essential because phosphopeptides are stoichiometrically substoichiometric and ionization-suppressed in complex mixtures. TiO2 exploits bidentate coordination of phosphate to Ti4+ under acidic conditions; selectivity is improved by using 2,5-dihydroxybenzoic acid (DHB) or glutamic acid as competitor ligands to displace acidic non-phosphorylated peptides. Without enrichment, phosphopeptides are typically invisible in a standard LC-MS/MS run of a whole-cell digest — not because they are absent, but because the signal is overwhelmed by the ~10,000-fold more abundant unmodified peptides (Mann and Jensen, 2003).
  9. Hydrogen-deuterium exchange mass spectrometry (HDX-MS) reports on protein backbone amide dynamics by measuring the rate of exchange of backbone NH protons for deuterium from D2O solvent. Amides in stable secondary structure (hydrogen-bonded α-helices and β-sheets) exchange slowly; exposed, disordered, or solvent-accessible regions exchange rapidly. After a defined deuterium incubation, the reaction is quenched at pH 2.5 and 0°C (conditions that minimize back-exchange), the protein is digested with pepsin (active at low pH), and the peptides are analyzed by LC-MS immediately. The mass shift of each peptide reports the average deuterium incorporation — a direct readout of local dynamics and solvent accessibility (Bhattacharya et al.).
  10. Data-Independent Acquisition (DIA/SWATH) systematically fragments all precursor ions in predefined m/z windows, generating comprehensive fragment ion data regardless of which peptides are present — fundamentally different from the stochastic sampling of data-dependent acquisition (DDA). In DDA, the instrument selects only the most abundant precursors for MS2 in each cycle; less abundant peptides are frequently not selected and thus go unidentified across runs. In DIA, every m/z window (e.g., 25 Da wide windows stepped across 400–1200 Da) is fragmented every cycle, producing chimeric MS2 spectra that are subsequently deconvolved against a spectral library using software such as DIA-NN or Spectronaut. This yields reproducible, near-complete peptide quantification across large sample cohorts — critical for biomarker studies (Aebersold and Mann, 2003).
  11. Crosslinking mass spectrometry (XL-MS) provides distance constraints between residues in protein complexes by covalently linking proximal amino acids with bifunctional reagents and identifying crosslinked peptides by their non-linear mass. The most widely used crosslinker, DSS (disuccinimidyl suberate), spans ~11.4 Å between NHS-ester-reactive Lys residues. A crosslinked peptide is the mass sum of two peptides minus the crosslinker's leaving groups — detectable as a higher-mass species that fails tryptic mass prediction. Identified crosslinks act as distance restraints (Cα-Cα ≤ ~25–30 Å for DSS) that can dock crystal structures of individual subunits into the architecture of large assemblies (Altschuh and Bhattacharya).
  12. De novo sequencing — reading peptide sequence directly from MS/MS spectra without a database — is essential for organisms with unsequenced genomes, antibody sequencing, and discovery of novel PTMs. The mass differences between consecutive y-ions (or b-ions) directly spell out the residue mass of each amino acid along the sequence: yn − yn−1 = residue mass. Software tools like PEAKS use de novo scores as seed sequences for homology database searching, enabling identification even when the exact protein sequence is not in any database. This is the only valid strategy for sequencing novel endogenous peptides or therapeutic antibodies whose sequence is proprietary (Liebler, 2002).
  13. The ubiquitin remnant profiling approach exploits the fact that trypsin cleavage of poly-ubiquitinated proteins leaves a diagnostic diglycine tag (GG, +114.04 Da) on the modified Lys residue. This mass shift is stable, unique, and sufficient to identify the modified site by MS/MS. Anti-K-ε-GG antibodies (Cell Signaling Technology UbiScan) immunoprecipitate GG-modified peptides from tryptic digests, enabling global, site-specific mapping of the ubiquitylome across thousands of sites in a single experiment. The elegance is that the ubiquitin tag itself — after trypsin digestion — becomes the enrichment handle and the mass-spectrometric signature simultaneously (Mann and Jensen, 2003).
  14. Intact mass analysis of peptide drugs by ESI-MS or MALDI-TOF confirms molecular weight, detects sequence errors, and verifies chemical modifications such as lipidation, PEGylation, or disulfide bond formation. A correctly synthesized peptide will give a monoisotopic mass matching the theoretical value within 5 ppm on a high-resolution instrument. Any mass discrepancy exceeding this indicates a synthetic error (deletion, insertion, oxidation) or unexpected modification. For disulfide-bonded peptides, comparison of reduced versus non-reduced intact masses reveals the disulfide bond count: each disulfide reduces the mass by 2.016 Da (loss of 2H). This non-reduced/reduced digestion strategy extends to mapping which specific cysteines are paired (Liebler, 2002).

References

  1. Aebersold, R. and Mann, M., 'Mass spectrometry-based proteomics,' Nature, 422, 2003, pp. 198–207. DOI: 10.1038/nature01511
  2. Altschuh, D. and Bhattacharya, A., Mass Spectrometry-Based Protein Interaction Studies, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2019.
  3. Bhattacharya, A. et al., Proteomics: A Cold Spring Harbor Perspectives in Biology Collection, 1st edn., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2020.
  4. Fenn, J.B., Mann, M., Meng, C.K., Wong, S.F. and Whitehouse, C.M., 'Electrospray ionization for mass spectrometry of large biomolecules,' Science, 246(4926), 1989, pp. 64–71. DOI: 10.1126/science.2675315
  5. Karas, M. and Hillenkamp, F., 'Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons,' Analytical Chemistry, 60(20), 1988, pp. 2299–2301. DOI: 10.1021/ac00171a028
  6. Liebler, D.C., Introduction to Proteomics: Tools for the New Biology, 1st edn., Humana Press, Totowa, NJ, 2002.
  7. Mann, M. and Jensen, O.N., 'Proteomic analysis of post-translational modifications,' Nature Biotechnology, 21(3), 2003, pp. 255–261. DOI: 10.1038/nbt0203-255