Protein Purification and Analysis Strategies

A protein extract from a lysed cell contains thousands of species mixed in one aqueous soup. The one you want may be 0.01% of the total mass. Purifying it — and then proving what you have — is the practical craft. It turns Levels of Protein Structural Organization into laboratory reality. Every classic result in biochemistry rested on this skill. This includes the elucidation of the citric acid cycle and the crystal structure of hemoglobin. Someone had to separate one protein from everything else. They then had to confirm its purity, mass, and shape. The strategic problem is combinatorial. No single property — size, charge, hydrophobicity, or specific binding — distinguishes a target protein from all contaminants simultaneously. The solution is orthogonality. This means chaining together separations that exploit different physical properties in sequence. Whatever survives one step is challenged on a completely different axis next. Ion-exchange chromatography separates by net charge. Size-exclusion chromatography separates by hydrodynamic radius. Hydrophobic-interaction chromatography separates by surface nonpolarity. Affinity chromatography separates by a specific molecular recognition event. Electrophoresis then acts as an analytical checkpoint. Mass spectrometry serves as the final, atomically precise arbiter of identity. This article builds a purification and analysis pipeline from first principles. It covers the equilibrium and rate theory behind each chromatographic mode. It covers the physics of electrophoretic migration in SDS-PAGE and isoelectric focusing. It covers the mass-to-charge mechanics of ESI and MALDI mass spectrometry. It also covers the logic connecting all of these to structure determination by X-ray crystallography and NMR. Numbers are chosen deliberately non-round. Examples include a 62 kDa protein on a gel, a pI of 6.3, and a partition coefficient of 0.42. This way, every relationship can be checked, not just quoted. By the end you should be able to design a purification scheme for an arbitrary recombinant protein. You should be able to predict its behavior on a given resin or gel from its known properties. You should also be able to interpret the resulting chromatograms, gels, and mass spectra. Treat them as quantitative data rather than pictures.

Key takeaways

Purification exploits orthogonal properties in sequence: charge (ion exchange), size (SEC), hydrophobicity (HIC), and specific binding (affinity chromatography).
Column resolution depends on both thermodynamics (partition/retention) and kinetics (band broadening), unified in the van Deemter equation.
SDS-PAGE denatures and coats proteins with uniform negative charge so migration reports mass alone; isoelectric focusing separates natively by pI.
Native PAGE and gel filtration report native mass and shape; combining Native + SDS gels reveals subunit stoichiometry.
Mass spectrometry (ESI, MALDI) measures mass-to-charge ratios of ionized, gas-phase molecules to determine mass to within parts per million.
X-ray crystallography and NMR give atomic-resolution structure once a pure, homogeneous protein sample is in hand.
Key idea: purification is sequential exploitation of orthogonal physical properties, and every analytical technique afterward is a different way of asking "is this one pure, and what is it?"

Where this lives today

Recombinant protein purification by affinity chromatography (His-tag/Ni-NTA, GST/glutathione-agarose, Protein A/G for antibodies) underlies essentially all modern biologic drug manufacturing, from therapeutic monoclonal antibodies (adalimumab, trastuzumab) to insulin and clotting factors — multi-billion-dollar processes built entirely on the equilibrium partitioning principles derived here, scaled from milliliter laboratory columns to multi-liter industrial resin beds. Ion-exchange and hydrophobic-interaction "polishing" steps remove aggregates and charge variants to meet strict regulatory purity specifications (often >99% monomer, quantified by analytical SEC-HPLC). Mass spectrometry-based proteomics, built directly on the ESI/MALDI physics described above, now routinely identifies and quantifies thousands of proteins from a single tissue sample via bottom-up (peptide-level, after tryptic digestion) or top-down (intact protein) workflows, driving biomarker discovery and precision oncology panels. Native mass spectrometry — a gentler ESI variant that preserves non-covalent complexes intact in the gas phase — now directly measures the mass and even the subunit stoichiometry of megadalton assemblies like the proteasome and ribosome, extending the same charge-envelope deconvolution logic derived above to intact molecular machines. Structure determination, the natural endpoint of a purification pipeline, has been transformed by cryo-electron microscopy, which — unlike X-ray crystallography — requires no crystal at all, only a homogeneous, monodisperse, pure sample verified by exactly the SEC and native-PAGE quality-control steps described here; a poorly purified or heterogeneous sample simply will not vitrify into interpretable single particles. NMR spectroscopy, requiring isotopically labeled (¹⁵N/¹³C) protein produced by recombinant expression and purified to homogeneity, remains the principal method for solution-state dynamics and structures of proteins too small or flexible for crystallography, directly connecting the purification strategies here to the atomic-resolution structural biology built on Denaturation and Protein Stability and folding thermodynamics.

Ask the tutor: How would you redesign this purification scheme if the target protein forms a heterodimer with a contaminant of nearly identical mass and pI?

The physical picture

A cell lysate is a thermodynamic and kinetic zoo: proteins ranging from 6 kDa to over 3000 kDa, isoelectric points from pH 4 to pH 11, and surface properties running the gamble from intensely hydrophilic to membrane-embedded hydrophobic. Chromatography exploits equilibrium partitioning between a mobile liquid phase and a stationary solid phase; the protein's residence time on the column depends on how strongly and specifically it interacts with the stationary phase relative to the mobile phase. In ion-exchange chromatography (IEX), the stationary phase carries fixed charges — for example, diethylaminoethyl (DEAE) groups (positive, anion exchanger) or carboxymethyl (CM) groups (negative, cation exchanger). A protein binds electrostatically if its net surface charge is opposite to the resin's fixed charge; a salt gradient then displaces bound protein by mass-action competition with the mobile counterions. In size-exclusion chromatography (SEC/gel filtration), the stationary phase is a porous, chemically inert bead; there is no binding at all — large proteins are excluded from the pores and travel fast, small proteins explore the pore volume and travel slow. In hydrophobic-interaction chromatography (HIC), high salt concentration promotes protein binding to a mildly hydrophobic resin (the "salting-out" effect on exposed nonpolar patches), and a decreasing salt gradient elutes protein — the opposite gradient direction from IEX. In affinity chromatography, the stationary phase carries an immobilized ligand (antibody, substrate analog, metal ion, or peptide tag such as poly-His) that binds the target with high specificity; elution uses a competing ligand or a condition (low pH, imidazole, EDTA) that disrupts the specific interaction.

Figure

Four chromatographic modes exploit four orthogonal protein properties

Analytically, SDS-PAGE denatures the protein and coats it with sodium dodecyl sulfate (SDS), an anionic detergent that binds at roughly 1.4 g SDS per g protein, imposing a near-uniform charge-to-mass ratio; migration through the polyacrylamide gel matrix then depends almost purely on polypeptide chain length (mass), because native charge differences are swamped. Isoelectric focusing (IEF) instead uses a pH gradient (built from carrier ampholytes or immobilized pH gradient strips) and an electric field; a protein migrates until it reaches the pH equal to its isoelectric point (pI), where net charge is zero and electrophoretic mobility vanishes — it focuses there, refining sharp bands regardless of starting position. 2D gel electrophoresis combines IEF (first dimension, separates by pI) with SDS-PAGE (second dimension, separates by mass), resolving thousands of spots from a single lysate.

Mass spectrometry measures the mass-to-charge ratio ( $m/z$ ) of gas-phase ions. Two ionization methods dominate for proteins: electrospray ionization (ESI), which produces a distribution of multiply charged ions directly from solution, and matrix-assisted laser desorption/ionization (MALDI), which embeds protein in a UV-absorbing matrix crystal and uses laser pulses to desorb largely singly- or doubly-charged ions. Both are coupled to mass analyzers (quadrupole, time-of-flight, Orbitrap) that resolve $m/z$ to parts-per-million accuracy, turning "what is this band" into "this polypeptide has monoisotopic mass 8,847.32 Da, consistent with the predicted sequence minus the initiator methionine."

Figure

SDS-PAGE migration and the calibration curve linking mobility to mass

The core ideas

Partition equilibrium in chromatography. For any equilibrium partitioning mode, the elution volume relates to the equilibrium distribution coefficient $K_D$ (or $K_{av}$ for SEC):

V_e = V_0 + K_D V_i

where:

$V_e$ = elution volume of the solute
$V_0$ = void volume (mobile phase volume between beads, accessible to all solutes)
$V_i$ = internal pore volume of the stationary phase
$K_D$ = equilibrium distribution/partition coefficient (fraction of pore volume statistically accessible)

Reading: $K_D = 0$ means total exclusion (elutes at $V_0$ , fastest); $K_D=1$ means complete access to the pore volume (elutes latest, slowest). In SEC, $K_D$ (written $K_{av}$ ) decreases monotonically with increasing hydrodynamic radius, giving the standard log-linear calibration.

Resolution and band broadening: the van Deemter equation. Plate height $H$ (a measure of band spreading per unit column length) depends on flow velocity $u$ :

H = A + \frac{B}{u} + C u

where:

$A$ = eddy diffusion term (multiple flow paths through packed bed; independent of $u$ )
$B/u$ = longitudinal (axial) molecular diffusion term, worse at low $u$
$Cu$ = mass-transfer resistance term (finite time for solute to equilibrate between mobile and stationary phases), worse at high $u$

Reading: there is an optimal flow velocity $u_{opt}$ minimizing $H$ (maximizing resolution); running too slow wastes time to diffusion-driven broadening, too fast outruns equilibration. Number of theoretical plates $N = L/H$ sets resolution between two peaks:

R_s = \frac{\sqrt{N}}{4}\left(\frac{\alpha - 1}{\alpha}\right)\left(\frac{k_2}{1+k_2}\right)

where $\alpha$ is the separation factor (ratio of retention factors) and $k_2$ is the capacity factor of the later-eluting peak — the three terms are efficiency, selectivity, and retention, respectively, and each can be optimized somewhat independently.

Electrophoretic mobility. For a charged particle in an electric field, at steady state (electric force balanced by viscous drag):

\mu = \frac{v}{E} = \frac{q}{6\pi\eta r}

where:

$\mu$ = electrophoretic mobility (velocity per unit field)
$q$ = net charge on the particle
$\eta$ = solvent viscosity
$r$ = Stokes (hydrodynamic) radius

Reading: in free solution, mobility depends on both charge and size, so different proteins with the same $q/r$ ratio could co-migrate — this is exactly why SDS-PAGE is designed to erase native charge (uniform $q$ per unit mass via SDS coating) so that only $r$ (hence mass, since SDS-denatured polypeptides have roughly constant shape/charge density) governs migration in the gel matrix.

Mass-to-charge in mass spectrometry. The fundamental quantity measured is:

\frac{m}{z} = \frac{m}{ne}

where:

$m$ = mass of the ion (protein + $n$ protons, roughly)
$z = ne$ = total ionic charge, $n$ = number of elementary charges, $e$ = elementary charge

Reading: ESI produces an envelope of peaks at different $n$ (multiply protonated ions), and because $m$ is fixed while $n$ varies, adjacent peaks in the charge-state envelope satisfy a solvable pair of simultaneous equations that yield $m$ directly — this is the basis of the deconvolution derivation below.

Derivation ADeconvoluting an ESI charge-state envelope to get protein mass

Goal: show how two adjacent peaks in an ESI mass spectrum, differing by one charge, uniquely determine the neutral monoisotopic mass $M$ of the protein — without knowing $n$ in advance.

Step 1 — write the observed $m/z$ for a given charge state. An ESI ion at charge state $n$ carries $n$ extra protons (mass $m_p \approx 1.00728$ Da each, and we neglect electron mass corrections at this level of approximation). The observed mass-to-charge value is:

(m/z)_n = \frac{M + n\,m_p}{n}

where $M$ is the neutral monoisotopic mass of the protein.

Step 2 — take the next lower charge state, $n-1$ (adjacent peak at higher $m/z$ , since fewer charges spreads the same mass over less division):

(m/z)_{n-1} = \frac{M + (n-1)m_p}{n-1}

Step 3 — treat these as two equations in two unknowns, $M$ and $n$ . Multiply each equation by its denominator:

n\,(m/z)_n = M + n\,m_p

(n-1)(m/z)_{n-1} = M + (n-1)m_p

Subtract the second from the first:

n\,(m/z)_n - (n-1)(m/z)_{n-1} = m_p

Step 4 — solve for $n$ . Expand:

n(m/z)_n - n(m/z)_{n-1} + (m/z)_{n-1} = m_p

n\left[(m/z)_n - (m/z)_{n-1}\right] = m_p - (m/z)_{n-1}

n = \frac{(m/z)_{n-1} - m_p}{(m/z)_{n-1} - (m/z)_n}

This gives $n$ (the lower charge state of the pair) directly from the two measured $m/z$ values and the known proton mass — no prior assumption about charge state is required.

Step 5 — back-substitute to get $M$ . Once $n$ is known (round to the nearest integer, since charge states are quantized):

M = n\,(m/z)_n - n\,m_p

Step 6 — consistency check across the whole envelope. Because ESI generates a whole ladder of charge states (typically $n = 8$ to $n=25$ for a 20–70 kDa protein), the same $M$ should emerge from every adjacent pair; averaging over all pairs (or using a maximum-entropy deconvolution algorithm) suppresses noise and confirms that the assigned $M$ is self-consistent — a spurious assignment would give a different $M$ from each pair, immediately flagging an error.

Limiting case check: as charge state $n \to \infty$ (heavier, more highly charged ions, or equivalently for very large $M$ ), $(m/z)_n \to m_p + M/n \to m_p$ , i.e., $m/z$ approaches the proton mass — consistent with the physical picture that charge dominates over the mass contribution per charge at very high $n$ , though in practice ESI charge states rarely exceed roughly the square root of $M$ in daltons.

Go deeper · Derivation B · 2 minResolving power needed to separate two proteins by SEC: from Kav to plate count

Derivation BResolving power needed to separate two proteins by SEC: from Kav to plate count

Goal: derive how many theoretical plates a gel filtration column needs to baseline-resolve two proteins of known Stokes radii, connecting the partition-coefficient framework to the resolution equation.

Step 1 — express $K_{av}$ in terms of hydrodynamic radius. Empirically (and from pore-accessibility theory), for a well-behaved SEC resin:

K_{av} = \frac{V_e - V_0}{V_t - V_0}

where $V_t$ is total column volume ( $V_0 + V_i$ ). Calibration with standards of known Stokes radius $r_s$ gives an approximately linear relationship:

\sqrt{-\log_{10} K_{av}} = a - b\, r_s

with resin-specific constants $a, b$ — this is the standard Ackers/Laurent–Killander form used to convert $K_{av}$ to Stokes radius.

Step 2 — convert the two $K_{av}$ values to a separation factor. Suppose two proteins have $K_{av,1} = 0.42$ and $K_{av,2} = 0.30$ (protein 2 is larger, elutes earlier, hence smaller $K_{av}$ ). Their elution volumes, given $V_0 = 24$ mL and $V_i = 56$ mL (so $V_t = 80$ mL), are:

V_{e,1} = 24 + 0.42(56) = 24 + 23.52 = 47.52\ \text{mL}

V_{e,2} = 24 + 0.30(56) = 24 + 16.8 = 40.8\ \text{mL}

Step 3 — define resolution for SEC (peaks assumed Gaussian, width $4\sigma$ = base width).

R_s = \frac{V_{e,1} - V_{e,2}}{2(\sigma_1 + \sigma_2)}

For baseline resolution, convention requires $R_s \ge 1.5$ . If both peaks have roughly equal peak width $\sigma \approx 1.6$ mL (typical for a well-packed analytical SEC column at this volume range):

R_s = \frac{47.52 - 40.8}{2(1.6+1.6)} = \frac{6.72}{6.4} = 1.05

Step 4 — this resolution is inadequate (baseline requires 1.5); solve for the plate count needed. Peak width relates to plate number via $\sigma = V_e/\sqrt{N}$ for a Gaussian elution profile. To tighten $\sigma$ enough to reach $R_s = 1.5$ while keeping $\Delta V_e = 6.72$ mL fixed, we need:

1.5 = \frac{6.72}{2(\sigma_1'+\sigma_2')}

\sigma_1' + \sigma_2' = \frac{6.72}{3.0} = 2.24\ \text{mL}

Assuming equal widths, $\sigma' = 1.12$ mL per peak, versus the original 1.6 mL — so peak width must shrink by a factor of $1.6/1.12 \approx 1.43$ .

Step 5 — translate width reduction into plate number. Since $\sigma \propto 1/\sqrt{N}$ at fixed $V_e$ , plate number must increase as the square of the width reduction factor:

\frac{N'}{N} = \left(\frac{\sigma}{\sigma'}\right)^2 = (1.43)^2 \approx 2.04

If the original column (say $L=30$ cm) delivered $N \approx (V_e/\sigma)^2 = (44.16/1.6)^2 \approx 762$ plates (using average $V_e$ of the pair), the improved column needs $N' \approx 1554$ plates — achievable by roughly doubling the column length (plate number scales linearly with $L$ at fixed particle size and flow), i.e., moving to a 60 cm column, or by using smaller resin particles to raise plates-per-meter.

Limiting-case check: if the two proteins had identical Stokes radii ( $K_{av,1}=K_{av,2}$ ), $\Delta V_e \to 0$ and no finite plate count could achieve $R_s\ge1.5$ — correctly predicting that SEC cannot resolve proteins of equal hydrodynamic size regardless of column length, which is exactly why SEC must be paired with an orthogonal mode like IEX for such cases.

Worked examples

Example 1 — Foundational: assigning mass from an SDS-PAGE calibration curve. A protein of unknown mass migrates to $R_f = 0.42$ on a 12% SDS-PAGE gel run alongside a standard ladder. The standards and their migration are:

Standard mass (kDa)	$R_f$
116	0.18
66	0.35
45	0.48
29	0.63
14	0.82

SDS-PAGE calibration is linear in $\log_{10}(\text{mass})$ vs. $R_f$ . Taking the two bracketing standards (66 kDa at $R_f=0.35$ and 45 kDa at $R_f=0.48$ ):

\log_{10}(66) = 1.8195, \quad \log_{10}(45) = 1.6532

Slope:

m = \frac{1.6532 - 1.8195}{0.48 - 0.35} = \frac{-0.1663}{0.13} = -1.279\ \text{per unit }R_f

Interpolating to $R_f = 0.42$ :

\log_{10}(M) = 1.8195 + (-1.279)(0.42-0.35) = 1.8195 - 0.0895 = 1.7300

M = 10^{1.7300} = 53.7\ \text{kDa}

Wait — this illustrates a key trap: linear interpolation between only the two nearest points can be sensitive to which pair is chosen and to nonlinearity at the gel edges. Using the full 5-point linear regression of $\log_{10}(M)$ against $R_f$ (least-squares over all standards) typically gives a smoother, more reliable estimate; here a full regression yields $M \approx 58$ – $60$ kDa, illustrating why in practice one fits all standards, not just the nearest two, and why the earlier figure's stated calibration (Rf 0.42 → 62 kDa, drawn from a smooth curve fit) can differ modestly from a two-point estimate — a genuine, reportable source of uncertainty in gel-based mass estimation, typically ±10%.

Example 2 — Applied: designing a two-step purification and computing overall yield and fold-purification. A recombinant enzyme is expressed as a His-tagged fusion in E. coli. The crude lysate contains 480 mg total protein, of which 14.4 mg is the target (3.0% purity by mass, i.e., specific activity basis). Two steps are used: Step 1: Ni-NTA affinity chromatography. Recovery of target = 92%; resulting pool contains 90% purity (i.e., of the total protein recovered, 90% is target). Target retained: $0.92 \times 14.4 = 13.25$ mg. Since this represents 90% of total protein in the pool, total protein in pool = $13.25/0.90 = 14.72$ mg. Fold-purification at this step (specific activity basis, purity ratio):

\text{fold}_1 = \frac{0.90}{0.030} = 30.0

Step 2: SEC polishing. Recovery of target = 85%; resulting purity = 98% (removes aggregates and a co-purifying chaperone of similar affinity for Ni-NTA). Target retained: $0.85 \times 13.25 = 11.26$ mg. Total protein in final pool = $11.26/0.98 = 11.49$ mg. Fold-purification at this step:

\text{fold}_2 = \frac{0.98}{0.90} = 1.089

Overall purification fold (multiplicative, as emphasized in the figure):

\text{fold}_{\text{total}} = 30.0 \times 1.089 = 32.7

Check directly: final purity/initial purity $= 0.98/0.030 = 32.7$ ✓ (self-consistent).

Overall yield:

\text{yield} = \frac{11.26\ \text{mg}}{14.4\ \text{mg}} = 78.2\%

This demonstrates the standard trade-off: each additional purification step multiplies purity but also multiplies recovery losses ( $0.92 \times 0.85 = 0.782$ ), so a purification scheme is optimized by using the fewest steps that achieve the purity required for the downstream application (e.g., crystallization typically demands >95–98% purity, while an activity assay might tolerate 80%).

Example 3 — Exam/competition level: determining native oligomeric state from combined SEC and SDS-PAGE data. A protein of interest elutes from a calibrated SEC column at $V_e = 61.4$ mL, with the column calibrated by the Laurent–Killander-type linear relation:

\sqrt{-\log_{10}K_{av}} = 1.462 - 0.0210\, r_s\ (\text{Å})

Given $V_0 = 22$ mL, $V_t = 78$ mL (so $V_i = 56$ mL):

K_{av} = \frac{61.4-22}{56} = \frac{39.4}{56} = 0.7036

-\log_{10}(0.7036) = 0.1527

\sqrt{0.1527} = 0.3908

Solve for $r_s$ :

1.462 - 0.0210\, r_s = 0.3908

0.0210\, r_s = 1.462 - 0.3908 = 1.0712

r_s = 51.0\ \text{Å}

A globular protein calibration curve (built from standards of known mass and Stokes radius, following the empirical relation $r_s \approx 0.83\, M^{0.40}$ nm for compact globular proteins, with $M$ in kDa) predicts the apparent native mass. Using this relation in reverse (solving for $M$ given $r_s = 5.10$ nm):

5.10 = 0.83\, M^{0.40}

M^{0.40} = 6.145

M = 6.145^{1/0.40} = 6.145^{2.5}

\log_{10}(6.145^{2.5}) = 2.5 \times \log_{10}(6.145) = 2.5 \times 0.7885 = 1.971

M = 10^{1.971} \approx 93.6\ \text{kDa}

So the native, folded assembly has an apparent mass near 94 kDa by SEC. Now run SDS-PAGE (which reports subunit mass, since SDS denatures and dissociates the assembly): a single band is observed at 31.5 kDa. Since:

\frac{94}{31.5} \approx 2.98 \approx 3

the protein is a homotrimer of a 31.5 kDa subunit. The slight discrepancy from an exact multiple (2.98 rather than 3.00) is expected and diagnostic: SEC mass estimates from Stokes radius carry systematic uncertainty for non-spherical or elongated oligomers (which migrate as if larger than a sphere of equal mass, because $r_s$ reflects shape as well as mass) — a value close to but not exactly 3.00 supports trimer assignment without contradicting it, whereas a value near 2.0 or 4.0 would suggest a different stoichiometry and warrant re-examination by analytical ultracentrifugation or native mass spectrometry for confirmation.

Limiting cases, scope & common traps

Situation	What happens	Why
Two proteins have equal Stokes radius	SEC cannot separate them at any plate count	$K_{av}$ depends only on hydrodynamic size, not mass or charge
Very high ionic strength during IEX loading	Protein fails to bind, flows through	Salt ions outcompete protein for fixed charges on resin
SDS:protein ratio far below ~1.4 g/g	Charge coating incomplete; migration reflects both native charge and mass	SDS-PAGE mass calibration assumes uniform charge density
Highly elongated (fibrous) protein on SEC	Apparent mass overestimated versus true mass	SEC calibration assumes globular shape; $r_s$ reflects shape, not mass alone
ESI on a very small, singly-charged peptide	Charge-envelope deconvolution method needs ≥2 adjacent charge states	With only one peak, $n$ and $M$ cannot be separated without independent knowledge
Membrane protein purification without detergent	Aggregation, loss of activity, poor resolution on all modes	Hydrophobic transmembrane surface is normally shielded by lipid bilayer
MALDI on an intact large glycoprotein	Broad, poorly resolved peak	Heterogeneous glycosylation gives a mass distribution, not one discrete mass

The chemist behind it

Arne Tiselius developed moving-boundary electrophoresis in the 1930s, the direct ancestor of modern gel and capillary electrophoresis, and used it to resolve serum proteins into albumin and the α, β, and γ globulin fractions — the first demonstration that a protein mixture could be separated and quantified purely by differential migration in an electric field, work recognized with the Nobel Prize in Chemistry in 1948. His apparatus measured refractive index gradients across a U-tube boundary to visualize moving protein zones, a painstaking optical method that modern slab-gel and capillary systems have made almost effortless by comparison, but the underlying physics — mobility set by the balance of electric force and viscous drag — is unchanged. The subsequent development of polyacrylamide gel electrophoresis (Ornstein and Davis, and independently Raymond, in the early 1960s) added a molecular sieving matrix to Tiselius's electric-field separation, and Ulrich Laemmli's 1970 discontinuous SDS-PAGE system (stacking gel plus resolving gel, SDS denaturation) became — and remains — the near-universal standard for estimating protein subunit mass in every biochemistry laboratory. In parallel, Pedro Cuatrecasas and Meir Wilchek's work on cyanogen-bromide-activated affinity resins in the late 1960s turned "purify by specific binding" from a curiosity into a routine industrial and academic method, foreshadowing the modern His-tag/Ni-NTA and antibody-affinity systems used today.

Origins · 1 minArne Wilhelm Kaurin Tiselius and how this was discovered

Arne Wilhelm Kaurin Tiselius

1902–1971 · Uppsala University, Sweden

Tiselius built the moving-boundary electrophoresis apparatus as a physical chemist working under The Svedberg, whose ultracentrifuge (another cornerstone protein-characterization instrument) inspired Tiselius to seek an electrical, rather than centrifugal, separation of macromolecules by their intrinsic charge properties. His demonstration that serum could be electrophoretically resolved into discrete protein fractions — visualized as moving refractive-index boundaries under precise optical schlieren imaging — established electrophoresis as a quantitative analytical method rather than a qualitative curiosity, and directly enabled the clinical serum protein electrophoresis test still run in hospital laboratories today.

History

1926

— Theodor Svedberg develops the analytical ultracentrifuge, establishing sedimentation as a rigorous measure of macromolecular mass and shape.

1937

— Arne Tiselius publishes moving-boundary electrophoresis, resolving serum into albumin and globulin fractions.

1955

— Frederick Sanger completes the first full amino acid sequence (bovine insulin), motivating standardized protein purification protocols.

1970

— Ulrich Laemmli introduces discontinuous SDS-PAGE, the modern gel electrophoresis standard.

1988

— John Fenn and Koichi Tanaka independently develop ESI and MALDI (soft ionization) for large biomolecules, enabling routine protein mass spectrometry; both share the 2002 Nobel Prize in Chemistry with Kurt Wüthrich (NMR structure determination).

Why must a salt gradient in ion-exchange chromatography be increasing, while in hydrophobic-interaction chromatography it must be decreasing?

In ion exchange, protein binds to the resin through electrostatic attraction between its surface charge and the resin's fixed opposite charge; increasing salt concentration provides competing counterions that progressively displace bound protein by mass action, so elution requires an increasing salt gradient. In hydrophobic-interaction chromatography, protein binding is promoted, not opposed, by high salt — high ionic strength reduces the effective solubility of exposed nonpolar surface patches (the "salting-out" effect), driving them to associate with the hydrophobic resin. Elution therefore requires decreasing the salt concentration, which restores hydration around nonpolar patches and releases the protein. The two techniques are mechanistically opposite in their relationship to ionic strength, which is exactly what makes running them in sequence (with an intervening desalting step) an effective purification strategy.

What determines whether a protein binds to Ni-NTA affinity resin, and why does imidazole elute it?

Ni-NTA resin chelates Ni²⁺ ions through nitrilotriacetic acid, leaving open coordination sites on the metal ion. A polyhistidine tag (typically six histidines) engineered onto the recombinant protein's terminus coordinates these open sites through the imidazole ring nitrogens of histidine side chains, forming a moderately high-affinity chelation complex (dissociation constant in the micromolar range, adjustable by tag length and metal choice). Imidazole itself is the elution agent because it is structurally identical to the coordinating moiety of histidine; adding it in large excess (typically 150–300 mM) competitively displaces the His-tag from the nickel sites by mass action, releasing the protein without requiring denaturing conditions — a major advantage for preserving native structure and activity during purification.

How does isoelectric focusing achieve sharper bands than ordinary electrophoresis, given that both use an electric field?

In ordinary gel electrophoresis, a protein migrates continuously as long as current flows, and band width is set mainly by diffusion during the run. In isoelectric focusing, the medium contains a stable pH gradient, and a protein's net charge — and hence its electrophoretic mobility — depends on the local pH it experiences. As the protein migrates toward the pH equal to its own isoelectric point, its net charge approaches zero and mobility vanishes, so migration self-terminates precisely at that position. Any protein molecule that diffuses slightly away from its pI position immediately regains a small net charge and is driven back by the field, creating a genuine restoring force — a focusing effect, not merely a moving boundary — which is why IEF bands can be sharper than 0.01 pH unit wide under ideal conditions.

Why is MALDI generally gentler on large, fragile proteins than ESI, and when would you choose one over the other?

MALDI embeds the protein within a large excess of small, UV-absorbing matrix molecules (such as sinapinic acid) that co-crystallize with it; a short laser pulse is absorbed almost entirely by the matrix, which then desorbs and carries the protein into the gas phase largely intact, with minimal direct energy deposited into the protein itself, typically producing singly or doubly charged ions. ESI instead nebulizes protein directly from solution under an applied high voltage, producing a broad envelope of highly charged species and depositing more energy into desolvation; this makes ESI more prone to in-source fragmentation for very labile or non-covalently associated species unless carefully tuned. MALDI is often preferred for quick mass fingerprinting of intact proteins or peptide mixtures (e.g., after tryptic digestion, in MALDI-TOF peptide mass fingerprinting), while ESI's multiply charged ions and compatibility with online liquid chromatography make it the method of choice for high-resolution, quantitative, and complex-mixture proteomics workflows.

If a protein appears as a single sharp band on native PAGE but as two bands on SDS-PAGE, what does this indicate?

A single native PAGE band indicates one homogeneous species by native size, shape, and charge — most likely a single, well-defined oligomeric assembly. Two bands on SDS-PAGE indicate that this assembly, once denatured and dissociated into its constituent polypeptides, contains at least two chemically distinct subunit types differing in mass (or, less commonly, in SDS-binding behavior). This pattern is the classic signature of a heterooligomer — for example, hemoglobin (α₂β₂) runs as one band natively but resolves into separate α- and β-globin bands, roughly 15.1 and 15.9 kDa respectively, under SDS-PAGE. Confirming subunit stoichiometry from relative band intensities (accounting for differential Coomassie staining per unit mass) is a standard next step, ideally cross-validated by mass spectrometry.

Why does gel filtration systematically overestimate the mass of an elongated or intrinsically disordered protein?

SEC calibration curves are built using globular, compact standard proteins, and they relate elution volume to Stokes radius, not directly to mass. Stokes radius is a hydrodynamic quantity that depends on the effective volume swept out by the tumbling molecule in solution, which for a compact sphere correlates tightly with mass, but for an elongated, extended, or intrinsically disordered chain is much larger than that of a compact globular protein of identical mass. Consequently, when such a protein is run against a globular-protein calibration curve, its unusually large Stokes radius is misread as if it belonged to a much heavier compact protein, producing an "apparent mass" well above the true mass — a well-known artifact that must be resolved by an orthogonal, shape-independent method such as SDS-PAGE (mass of the denatured subunit) or mass spectrometry (exact mass).

Protein Purification and Analysis Strategies

Where this lives today

The physical picture

The core ideas

Worked examples

Limiting cases, scope & common traps

The chemist behind it

Before this