Analytical Ultra Centrifugation
How it works
AUC places the sample directly in a rotor spinning at up to roughly 60,000 rpm, generating centrifugal fields hundreds of thousands of times gravity, while an optical detection system (absorbance, interference or fluorescence) scans the cell and records the concentration profile as a function of radial position, repeatedly, over the course of the run. In sedimentation velocity, molecules move outward under the centrifugal field and are opposed by frictional drag and diffusion; the shape of the resulting moving boundary is analysed to yield a distribution of sedimentation coefficients (c(s)), from which molecular shape, size heterogeneity and oligomeric state are derived without any assumption about how many species are present. In sedimentation equilibrium, a lower speed is used until sedimentation and diffusion exactly balance and the concentration profile stops changing with time; fitting that equilibrium profile gives the absolute molar mass directly, together with self-association or hetero-association constants when applicable.
What you measure
Sedimentation velocity resolves a sample into a distribution of sedimentation coefficients (in Svedberg units, S), from which apparent molecular weight, frictional ratio (a measure of shape — how far a molecule departs from a compact sphere) and the relative abundance of each species can be read off directly, making it the method of choice for detecting and quantifying aggregates, fragments or oligomers as a percentage of total sample. Sedimentation equilibrium instead returns an absolute molar mass with no calibration or standard required, plus, when the sample self-associates, the equilibrium constants that describe the monomer–dimer, dimer–tetramer or other association scheme — including whether that association is reversible.
A typical experiment
AUC needs no column, chip or immobilization: the sample is loaded into a two-sector cell (sample and reference buffer) and spun. Sedimentation velocity runs typically take several hours at high speed, with scans collected every few minutes throughout; sedimentation equilibrium runs are longer, often a day or more per speed, since the sample must reach a true, unchanging equilibrium profile before data can be fitted, and is commonly repeated at two or three speeds and concentrations to fit shared parameters with more confidence. Sample volumes are modest and concentrations are chosen so the optical system stays within its linear range — for absorbance detection this generally rules out very high concentrations, which is one reason interference or fluorescence detection is used for concentrated antibody formulations.
Applications
Antibody & ADC characterization. AUC is used to confirm monomer content and detect low-level aggregates, fragments or oligomers in antibodies and antibody–drug conjugates — a level of detail on minor species that chromatography-based sizing methods can miss.
Biopharmaceutical QC & comparability. Because it needs no calibration standard, AUC provides an independent, first-principles molar mass and purity measurement used in quality control and in comparability studies for biosimilars and reformulated biologics.
Self-association & complex formation. Sedimentation equilibrium quantifies reversible self-association (monomer–dimer–oligomer equilibria) and hetero-complex formation, informing both formulation work and mechanistic studies of signalling complexes.
Viral vectors & nanoparticles. AUC characterizes the size distribution and empty/full ratio of viral vectors (e.g. AAV) and other nanoparticle-based drug delivery systems, where accurate particle counting matters for both efficacy and safety.
Strengths & limitations
Because it works in free solution with essentially no interaction between the sample and any stationary phase or surface, AUC avoids a whole class of artefacts that column- or chip-based methods can introduce — but that comes at the cost of longer run times (hours for velocity, often a day or more for equilibrium) and lower throughput than techniques like DLS. Standard absorbance and interference optics detect the whole sample and cannot distinguish two co-sedimenting species by identity alone; where that matters, fluorescence-detected AUC can follow a labelled species selectively, at the cost of requiring an extrinsic fluorophore. Because the results depend on precise rotor, temperature and optical calibration, AUC is generally run and interpreted by an experienced operator rather than as a walk-up assay — but for confirming oligomeric state, purity and self-association behaviour directly in solution, with no assumptions borrowed from a calibration standard, it remains a reference method that other techniques are validated against.
Frequently asked questions
How much sample do I need?
A sedimentation velocity cell takes 400 µl in the sample sector and 400 µl of matched buffer in the reference sector. Concentration is set by the optics: with absorbance detection you want an optical density between 0.1 and 1.2 at the chosen wavelength, and 0.5-0.8 is comfortable.
At 280 nm that corresponds to roughly 0.1-1 mg/ml for a typical protein, so 40-400 µg per cell. Concentration dependence of the sedimentation coefficient is real and often the point of the experiment, so a proper run uses three loading concentrations spanning a factor of about ten — plan for 1-1.5 ml of the most concentrated sample plus 2-3 ml of the dialysis buffer.
Weakly absorbing samples can be read at 230 nm, where the peptide bond absorbs about ten times more strongly, provided the buffer is transparent there. That drops the requirement to a few µg per cell.
What does AUC tell me that SEC or DLS does not?
A first-principles hydrodynamic description in free solution, with no column, no matrix and no calibration against standards.
Sedimentation velocity gives the distribution of sedimentation coefficients c(s), converted to s20,w so that results from different buffers and temperatures are comparable. From s and the diffusion coefficient obtained from boundary spreading you get the molar mass of each species independently, plus the frictional ratio f/f0, which says whether the particle is compact or elongated. A monomer at 3.5 S with f/f0 = 1.25 and the same mass at 3.5 S with f/f0 = 1.8 are two very different molecules, and no chromatographic method distinguishes them.
Sedimentation equilibrium gives molar mass and association constants without any shape assumption at all, because at equilibrium the concentration gradient depends only on buoyant mass.
The other advantage is quantitative composition: the areas under the c(s) peaks are true mass fractions, so 2% dimer is reported as 2%, without the recovery losses and on-column dissociation that make SEC quantification unreliable for weak complexes.
Which buffer should the sample be in?
Almost anything, as long as it is transparent at the detection wavelength and you send the exact dialysate. The reference sector is filled with that dialysate, and the whole analysis is a difference measurement.
Watch absorbance at 280 nm from imidazole, nucleotides, ATP and DTT, and at 230 nm from Tris, chloride, HEPES, DTT and glycerol. Reducing agents can usually be swapped for 0.5 mM TCEP. If the buffer is opaque where you need to look, we either change the wavelength or move to interference optics.
Density and viscosity enter the calculation directly, so glycerol, sucrose, high salt and D2O have to be declared and are computed with SEDNTERP rather than guessed. Ten percent glycerol changes viscosity by about 30% and moves every s value accordingly.
How long does an AUC run take?
A sedimentation velocity run is 6-12 hours for a typical protein and is usually started in the afternoon and analysed the next morning. Small proteins need higher speed and longer runs; large complexes and viral particles sediment in an hour or two.
Sedimentation equilibrium is a different scale. Equilibrium takes 24-48 hours per speed, and a proper experiment uses three speeds and three concentrations, so a week per sample set is normal.
Add rotor equilibration at temperature, which takes two to three hours and cannot be skipped without corrupting the early scans. This is not a technique with same-day turnaround, and the analysis afterwards is a real piece of work rather than a button press.
Do you need to label or modify anything?
No. Absorbance and interference detection see the molecule as it is, and the sample is recovered at the end of the run.
Interference optics is useful when the sample has no useful chromophore — a peptide, a detergent, a polysaccharide — because it reads refractive index rather than absorbance. Fluorescence detection exists on some machines for work at picomolar concentrations or in crowded backgrounds; ask us before planning an experiment around it, because it depends on how the instrument is equipped.
Can you measure the affinity of a self-association?
That is one of the things AUC does better than most alternatives. A monomer-dimer equilibrium that dissociates on a SEC column stays intact in the ultracentrifuge, because there is no dilution and no matrix.
The experiment is a concentration series: for a reversible self-association the whole c(s) distribution shifts with loading concentration, and the weight-average s as a function of concentration is fitted to an association model. Practical range is roughly 100 nM to 100 µM for a monomer-dimer, set at the low end by detection and at the high end by non-ideality.
An irreversible mixture behaves differently — the peaks keep their positions and only their relative areas change. Distinguishing the two cases is often the real question, and the concentration series answers it.
Can you run membrane proteins or detergent complexes?
Yes, with a caveat that has to be understood before the experiment is designed. A detergent-solubilised membrane protein sediments as a protein-detergent complex whose buoyant mass includes the bound detergent, and micelles of free detergent sediment as their own species. Interpreting s in terms of protein mass requires knowing how much detergent is bound, which you generally do not.
The standard way around it is density matching: repeating the run in H2O/D2O mixtures until the solvent density equals the partial specific density of the detergent, so the detergent becomes invisible and the buoyant mass reflects protein alone. It works for detergents whose density falls between about 1.0 and 1.1 g/ml, which excludes several common ones — fluorinated and some glucoside detergents fall outside.
Nanodiscs, liposomes and lipid nanoparticles are all measurable and are among the applications where AUC has few real competitors. Send the empty-particle control as well as the loaded sample.
How do I know the data are good?
Look at three things in the report.
The fit residuals, shown as a bitmap over radius and time, should be structureless noise. Diagonal stripes mean the model is missing a species; horizontal bands mean a problem with the meniscus or the time-invariant noise.
The RMSD of the fit should be comparable to the noise of the optical system, around 0.005-0.01 absorbance units for the absorbance optics.
The mass balance: the total signal in c(s) should account for the loading concentration you declared. A large deficit means material sedimented before the first scan — big aggregates — or stuck to the cell windows.
Sample-side warning signs are a boundary that never clears the meniscus, a broad continuum rather than peaks, and s values that drift between replicate cells.
What do you need to know about my sample before the run?
The molecular weight of the monomer, the extinction coefficient at the wavelength we will use, the exact buffer composition including any glycerol or sugar, the concentration and how it was measured, and the temperature at which the sample is stable.
The partial specific volume is calculated from the sequence for a protein, so send the sequence. For glycoproteins, nucleic acid complexes, detergent complexes and conjugates the calculated value is wrong and we need the composition to estimate it properly — a 10% error in partial specific volume produces roughly a 25% error in molar mass, which is the single largest source of nonsense in AUC results.
Send 2-3 ml of the dialysis buffer, and say whether the sample has been frozen.
Do I need to be present, and can I be trained?
No need to be present. Runs are long and largely unattended, and you receive the raw scans, the c(s) distributions, the fitted parameters and their interpretation.
Training is a bigger commitment than for the other instruments here. Cell assembly is a manual skill — window alignment, torque, leak testing — and a badly assembled cell leaks at 50 000 rpm and ruins the run for every cell in the rotor. The analysis with SEDFIT and SEDPHAT takes longer to learn than the operation. People who intend to make AUC part of their routine work are welcome to learn both; for occasional needs the service route is more sensible.
Instruments
Beckman XLA