Calibrate an SEC column, and why apparent MW can mislead

A size exclusion column can be calibrated with proteins of known mass, and an unknown can then be given a molecular weight from where it elutes. That number is useful, but it is an apparent MW: it says the unknown elutes where a globular standard of that mass would. This tutorial builds the calibration step by step, then shows the cases where the apparent MW and the real mass part ways.

1. Put elution on a common scale: K_av

A raw elution volume V_e depends on the column size and packing. Calibration uses a fraction instead:

K_av = (V_e − V₀) / (V_t − V₀)

• V_e is the volume from injection to the centre of the peak.
• V₀, the void volume, is the liquid between the beads. A molecule too large to enter any pore leaves the column after this volume, with K_av = 0. It is measured with a large marker, typically Blue Dextran 2000 (about 2,000 kDa), run on its own because its peak is broad. Take V₀ from its first peak.
• The upper reference is where it gets confusing, because two conventions are in use. Vendor calibration instructions divide by the geometric bed volume V_c, the packed bed’s volume worked out from its dimensions, and reserve V_t for the total liquid volume. Others divide by a measured total volume: the elution volume of a small molecule that enters every pore and does not stick to the resin. The total liquid volume is smaller than V_c, because the bead matrix itself takes up space.

Either choice works if you use it for standards and unknowns alike. Changing the denominator multiplies every K_av by the same constant. The fitted intercept and slope scale by that constant too, so the apparent MW and R² come out identical. (This is our algebra. The test suite for the shared calculation checks it.) Two things do change: the K_av values themselves, so you cannot compare them across labs without knowing the convention, and K_av-based warning thresholds such as “near void” or “near total volume”. Write down which one you used. The SEC technique page and its simulator use V_c.

If you record time rather than volume, K_av is the same as long as the flow is constant, because the flow cancels in the ratio.

2. Why the line is straight only in the middle

Plot K_av of globular standards against the logarithm of their mass and you get an S-shaped curve. Above the resin’s fractionation range, everything is excluded and sits at K_av ≈ 0, whatever its mass. Below it, everything permeates fully and sits at the top plateau. Between the two, the curve is close to a straight line, and that middle part is what a calibration uses. The calibration kit data file shows this in practice: a standard at the edge of a column’s range, such as aprotinin on a Sephacryl™ S-300 column or thyroglobulin on a Superose™ 6 Increase column, may need to be dropped from the fit because it no longer falls on the line.

So choose standards that bracket the unknown and lie inside the column’s range. An unknown outside the calibrated K_av range is an extrapolation, and near V₀ or the top plateau a small shift in V_e means a large change in apparent MW.

You can see the plateaus in the simulator below. Use the larger-pore resin and move Protein A above 600 kDa, then to the slider maximum: its K_av stops falling and the peak parks just after V₀. This simulator uses the bed volume V_c as its upper reference.

3. Worked example: five standards, one unknown

An illustrative column with a 24 mL bed and V₀ = 8.0 mL, using V_c as the upper reference. Five globular standards elute at 19.15, 17.95, 16.30, 14.35 and 12.30 mL. These volumes are invented for teaching; the masses are the values the calibration kit lists for carbonic anhydrase (29 kDa), ovalbumin (43), conalbumin (75), aldolase (158) and ferritin (440).

1. K_av for carbonic anhydrase is (19.15 − 8.0) / (24 − 8.0) = 0.697. The others give 0.622, 0.519, 0.397 and 0.269.
2. A least-squares line through K_av against log₁₀(MW in Da) gives a = 2.305 and b = −0.3637, with R² = 0.992.
3. An unknown elutes at 15.2 mL, so K_av = 0.450. Then log₁₀ MW = (0.450 − 2.305) / (−0.3637) = 5.10, an apparent MW of about 126 kDa.

R² looks excellent, but look at the residuals in the table: positive at both ends and negative in the middle. That pattern means curvature, an early sign that the outer standards are reaching the bends of the S-curve. Set the unknown to 9 mL and the result is flagged as outside the calibrated range.

4. Why apparent MW can mislead

The calibration assumes that the unknown relates mass to size the same way the standards do. The kit instructions say this directly: the standards must have the same relationship between molecular weight and molecular size as the sample. Each of the following breaks that assumption.

Shape: SEC sorts by hydrodynamic size, not by mass

A gel filtration column separates molecules by how they move through liquid, expressed as the Stokes radius R_s: the radius of a sphere with the same frictional coefficient. For compact globular proteins, mass and R_s rise together, which is why a mass calibration works for them. An elongated or partly unfolded protein has a larger R_s than a globular protein of the same mass. It can elute where a globular protein of about twice its mass would. Read the other way: an apparent dimer can be an elongated monomer.

So a protein’s R_s can be read from a column calibrated with standards of known R_s, and the kit instructions suggest this for non-globular proteins. Mass needs a second measurement. The classic route combines R_s from gel filtration with the sedimentation coefficient from density gradient centrifugation.

Glycosylation

The kit instructions warn that glycoproteins, lipoproteins and non-globular proteins may not correlate well with a calibration made with globular protein standards. A glycoprotein need not fall on that curve, and its SEC number need not match either its polypeptide mass or its total mass. Light scattering combined with UV and refractive-index detection can separate the two. It gives the polypeptide mass of a glycoprotein, independent of where the protein elutes.

Detergent micelles around membrane proteins

A solubilized membrane protein elutes as a protein–detergent complex. The amount of detergent and lipid bound is usually not known, so its elution reflects the whole complex, not the protein alone. Its apparent MW cannot tell you the oligomeric state. Static light scattering and sedimentation equilibrium can measure the protein mass in the complex without assuming how much detergent is bound or what shape the complex has.

Interactions with the resin

Ideal SEC assumes molecules only diffuse in and out of pores. Real resins carry some charge and some hydrophobic surface. If the protein and the resin carry the same charge, the protein can be kept out of the pores (ion exclusion). It then elutes earlier and looks larger. Attractive ionic or hydrophobic interactions retain the protein instead, so it elutes later and looks smaller, and some may not be recovered. The mobile phase controls this: salt suppresses electrostatic effects, and other additives such as amino acids or a little organic solvent are used to reduce adsorption. A peak after the total liquid volume cannot be explained by size alone and points to interaction. If the apparent MW shifts when you change the salt concentration, suspect interaction, or a salt-dependent change in the protein itself.

Self-association that depends on concentration

The kit instructions note that a sample of 0.5% of the bed volume is diluted about 5 to 15 times as it passes through the column. For a protein in a fast monomer–oligomer equilibrium, that dilution shifts the equilibrium during the run. Our inference from mass action is that its apparent MW can then fall between monomer and oligomer and change with the loaded concentration. Run two or three loads. If the peak position moves with concentration, the apparent MW is not a fixed property of the protein. SEC with multi-angle light scattering measures mass across the peak. That shows whether it holds a fixed distribution of species or a dynamic equilibrium.

When you need the real mass: SEC-MALS

Standard SEC is a relative method: elution volume compared with standards. Multi-angle light scattering (MALS) after the column, together with a concentration detector, gives the molar mass at each elution volume from first principles. It needs no calibration curve and does not depend on where the protein elutes. Use it, or analytical ultracentrifugation, when the decision depends on the oligomeric state or the mass of a complex.

5. Reporting %HMW, monomer and LMW

For size variants, the usual report is the area of each class as a percentage of the total. HMW means high-molecular-weight species such as aggregates, eluting before the main peak. LMW means fragments, eluting after it. These percentages carry assumptions:

• Equal response per mass. Area percentages assume equal detector response per mass (for UV, similar extinction coefficients) and baseline-resolved peaks. With UV, the area is proportional to mass times the extinction coefficient. Different species need the same coefficient, which is reasonable for an aggregate of the same protein and less so for an unrelated contaminant. Light scattering from particles can inflate UV signal at low wavelengths.
• Resolution. When a shoulder overlaps the main peak, where you drop the boundary decides the split. Report how regions were set, and do not compare percentages across columns or methods with different resolution. Overloading the column lowers resolution.
• Everything elutes. Aggregates or proteins that adsorb to the resin, or are filtered out before injection, are missing from the total. Recovery losses on the column are a known concern for aggregate analysis.
• The sample does not change during the run. Dilution can dissociate reversible aggregates, so %HMW measured by SEC can be lower than in the original sample. This is our inference from the dilution described above.

Do this with your own data in Chromatobench

Chromatobench uses the same calculation as the exercise above. Everything runs in your browser.

1. Import the standards run and, if you have them, the void-marker and sample runs. Volume (mL) or time (min) on the x axis both work, as long as standards and sample use the same unit.
2. Create regions. Under “Multiple peaks & regions”, save a region around each standard peak.
3. Open “SEC calibration & size variants”. Enter V0 and Vt, with the same convention throughout.
4. Enter the standards. Use “Add standard” and type each mass in kDa or Da. Fill V_e from the apex of a saved region with “Apex of…”. The panel shows K_av, residuals, the fit and R².
5. Read the sample. Switch to the sample run, save its peaks as regions and assign each to HMW, Monomer, LMW or Excluded. The panel lists the apparent MW per region, flags extrapolation, and gives the area percentages. Export the SEC CSV to keep the fit, the standards and the notes with the result.

Sources and scope

This is an editorial tutorial. The worked example and the exercise use illustrative elution volumes, not measured data for any column; the masses are the kit’s listed values. The statement that the K_avconvention does not change the apparent MW, and the self-association consequence of on-column dilution, are our derivations and are labelled as such above.

• GE Healthcare, Gel Filtration Calibration Kits LMW and HMW, instructions 28-4038-41PL Rev AD (2009; copy hosted by a university group, read 2026-11-07). Supports K_av with V_c, the term V_t as total liquid volume, Blue Dextran 2000 for V₀ (run alone, first peak), V_e measured to the peak centre, the sigmoid K_av vs log M_r relation, and dropping standards that fall off the line. It also supports the requirement that standards share the sample’s mass–size relation, the caveats for glycoproteins, lipoproteins and non-globular proteins, R_s-based plots, and 5–15-fold on-column dilution. Use your current column and kit instructions for operating values.
• Cytiva™, Gel Filtration Calibration Kit LMW and HMW data file, CY13572-07Aug20-DF (2020; distributor copy, read 2026-11-07). Supports the standard masses used in the example, Blue Dextran 2000 as the void marker, the sigmoid relation, and the examples of aprotinin on a HiPrep™ Sephacryl S-300 HR column and thyroglobulin on a Superose 6 Increase column falling off the line.
• Erickson (2009), Size and shape of protein molecules at the nanometer level determined by sedimentation, gel filtration, and electron microscopy, Biological Procedures Online 11, 32–51. Review: gel filtration separates by frictional coefficient (Stokes radius), not mass; an elongated protein can elute like a globular protein of about twice its mass; and gel filtration combined with sedimentation gives mass. Open access, CC BY 2.0.
• Siegel and Monty (1966), Determination of molecular weights and frictional ratios of proteins in impure systems by use of gel filtration and density gradient centrifugation, Biochimica et Biophysica Acta 112, 346–362. The primary paper for combining R_s from gel filtration with sedimentation. Cited from its Crossref record and Erickson’s account of the method; the full text was not read for this page.
• La Verde, Dominici and Astegno (2017), Determination of hydrodynamic radius of proteins by size exclusion chromatography, Bio-protocol 7(8), e2230. SEC separates by size and shape; the quantity derived for proteins is R_s. Cited from the abstract.
• Hong, Koza and Bouvier (2012), A review: size-exclusion chromatography for the analysis of protein biotherapeutics and their aggregates, Journal of Liquid Chromatography & Related Technologies 35, 2923–2950. Separation by Stokes radius; ion exclusion causing earlier elution; hydrophobic interactions; salt in the mobile phase; light scattering at low UV wavelengths; recovery losses; resolution lost with overloading. Open access; authors from Waters Corporation.
• Arakawa, Ejima, Li and Philo (2010), The critical role of mobile phase composition in size exclusion chromatography of protein pharmaceuticals, Journal of Pharmaceutical Sciences 99, 1674–1692. Protein adsorption to SEC resins and its suppression by salt, amino acids and organic solvents. Cited from the abstract.
• Wen, Arakawa and Philo (1996), Size-exclusion chromatography with on-line light-scattering, absorbance, and refractive index detectors for studying proteins and their interactions, Analytical Biochemistry 240, 155–166. Molecular weight independent of elution position, and polypeptide mass of glycoproteins without the glycan contribution. Cited from the abstract.
• Slotboom, Duurkens, Olieman and Erkens (2008), Static light scattering to characterize membrane proteins in detergent solution, Methods 46, 73–82. Bound detergent is usually unknown; light scattering and sedimentation equilibrium give the protein mass without assuming detergent binding or shape. Cited from the abstract.
• Some, Amartely, Tsadok and Lebendiker (2019), Characterization of proteins by size-exclusion chromatography coupled to multi-angle light scattering (SEC-MALS), Journal of Visualized Experiments (148), e59615. SEC is relative and fails for non-globular or interacting proteins; MALS is absolute and distinguishes a fixed distribution from a dynamic equilibrium. Covers glycoproteins and detergent-bound membrane proteins. Cited from the abstract.

The exercise and Chromatobench both use one implementation of the K_av fit, apparent MW and flags. Nothing you enter on this page is saved or sent.