Protein Concentration from A280

The equation has exactly two inputs

Beer-Lambert says A = epsilon x c x l. Absorbance equals the extinction coefficient times the concentration times the path length. Rearranged for what you want, c = A / (epsilon x l).

Path length is usually 1 cm and can be ignored, with two exceptions that bite often enough to name. A nanodrop-style pedestal instrument uses a much shorter path — typically 1 mm or less — and either reports a path-normalised value or does not, so check which. And a plate reader's path length depends on how much liquid is in the well, which is why plate-based A280 needs a path-length correction or a standard curve.

That leaves epsilon, the extinction coefficient at 280 nm, as the only thing you have to look up or compute. It comes in two flavours and the units matter: the molar coefficient (M-1 cm-1) gives you a molar concentration, and the A(0.1%) value, also written E1% divided by ten or as the mass extinction coefficient, gives you mg/mL directly. The second is what most people actually want at the bench, and it is just the molar coefficient divided by the molecular weight.

Where the number comes from: count three amino acids

Proteins absorb at 280 nm almost entirely because of tryptophan and tyrosine, with a small contribution from disulfide-bonded cysteine pairs. Everything else in a polypeptide is effectively transparent there. So the extinction coefficient of a folded protein in water is a sum of three counts times three constants, and the constants have been the same since Pace and colleagues measured them in 1995:

• Tryptophan: 5,500 M-1 cm-1 each.
• Tyrosine: 1,490 M-1 cm-1 each.
• Cystine (a disulfide-bonded pair of cysteines, so one per two Cys): 125 M-1 cm-1 each.

Reduced or cystine: which of the two numbers to use

Because the cysteine contribution only exists when two cysteines are joined in a disulfide, any sequence-based calculator reports two values. The reduced value counts Trp and Tyr only. The cystine value adds 125 for each disulfide the protein could form, assuming all cysteines pair up.

Use the cystine value for a native, correctly folded, oxidised protein — anything you have purified and expect to be functional, and anything with a known disulfide count. Use the reduced value for a protein in a reducing buffer (DTT, TCEP, beta-mercaptoethanol), for anything you have denatured in guanidine or urea, and for a cytoplasmically expressed protein that has no disulfides to begin with.

The gap is usually small, and it is worth knowing how small so you do not agonise over it for proteins where it does not matter. It is 125 M-1 cm-1 per disulfide against a total that is normally tens of thousands. It becomes worth caring about only for a cysteine-rich protein with few aromatics, which is exactly the case where A280 is already shaky for other reasons.

Worked: hen lysozyme, and checking it against the literature

Hen egg-white lysozyme is the standard teaching example because its coefficient is in every protein methods book, which makes it a way to check that you are doing the calculation the same way everyone else does.

The UniProt entry P00698 is 147 residues, but the first 18 are the signal peptide — the mature protein you actually buy in a bottle is residues 19 to 147, 129 amino acids. Feeding that mature chain to the Protein Properties tool gives 6 tryptophans, 3 tyrosines and 8 cysteines, and a molecular weight of 14,313.14 Da.

So the reduced coefficient is 6 x 5,500 + 3 x 1,490 = 33,000 + 4,470 = 37,470 M-1 cm-1. The eight cysteines form four disulfides, adding 4 x 125 = 500, for a cystine coefficient of 37,970 M-1 cm-1. Dividing each by the molecular weight gives A(0.1%) values of 2.618 reduced and 2.653 with the disulfides.

Native lysozyme has all four disulfides, so 2.653 is the number to use, and it lands on the commonly quoted literature value of about 2.64 to 2.65. That agreement is the point of the exercise: if your own calculation on a known protein does not reproduce the published figure, you have used the wrong chain, the wrong constants, or the wrong one of the two values.

Note also that the difference between the two figures here is 1.3 percent — well inside the error of pipetting a dilution. For a protein with four disulfides and nine aromatics, the reduced-versus-cystine choice is not what will make your number wrong.

Worked: GFP, and why A280 is a bad idea for it

Green fluorescent protein makes the opposite point. The Aequorea victoria sequence (UniProt P42212) is 238 residues with a molecular weight of 26,886.32 Da, and it contains exactly one tryptophan and eleven tyrosines.

That gives 1 x 5,500 + 11 x 1,490 = 21,890 M-1 cm-1, and an A(0.1%) of 0.814. Read that number carefully: a 1 mg/mL solution of GFP has an A280 of 0.81. Lysozyme at the same concentration reads 2.65, more than three times higher.

The consequence is practical. To get an A280 of 1.0 — comfortably in the linear range of most instruments — you need 1.2 mg/mL of GFP against 0.38 mg/mL of lysozyme. A GFP prep at 0.1 mg/mL gives you an A280 of 0.08, which is down in the region where cuvette scratches, buffer mismatch and baseline drift are the same size as the signal.

A low A(0.1%) is the warning sign to look for, and you get it from the sequence before you ever touch the spectrophotometer. Anything much below 0.5 means A280 will be a noisy measurement at working concentrations. GFP has a way out that most proteins do not — it absorbs strongly at 488 nm and you can quantify it there — but the general lesson is to check the coefficient first and pick the assay second.

Proteins with no aromatics at all

A protein with zero tryptophans and zero tyrosines has an extinction coefficient at 280 nm of zero, or near enough. Small peptides, some structural proteins, many designed sequences and a surprising number of tags and linkers fall here. The calculator will tell you the coefficient is 0, which is not a bug and is not a number you can divide by.

For those, A280 is simply not available and the answer is a colourimetric assay: Bradford, BCA, or an amino-acid analysis if you need real accuracy. It is worth running the sequence through a properties calculator before designing the purification, because discovering this after you have a tube of protein and a spectrophotometer is a bad moment.

A related case: a protein whose only aromatic contribution is from tyrosines has a different absorbance spectrum shape than one dominated by tryptophan, and the 280 nm peak is not where tyrosine peaks. The coefficient arithmetic still holds — it was measured for exactly this — but the measurement is less forgiving of a wavelength that is slightly off.

What else is absorbing at 280

The coefficient assumes that all the absorbance at 280 nm is your protein. Two common contaminants make that false.

Nucleic acid is the big one. DNA and RNA absorb strongly at 260 nm and still substantially at 280, so a prep carrying host nucleic acid reads high — sometimes very high. The standard check is the A260/A280 ratio: pure protein sits around 0.57, pure nucleic acid around 2.0, and anything well above 0.6 means you are measuring something other than protein. A ratio near 1.0 or above means the A280 number is not usable at all, and the fix is a cleanup rather than a correction factor.

Scattering is the quieter one. Aggregates and particulates scatter light across the spectrum, which the instrument reports as absorbance. The tell is absorbance at 320 or 340 nm, where nothing in a clean protein sample should absorb: if A320 is not close to zero, your sample is turbid and every reading including A280 is inflated. Spin it or filter it, and re-read, rather than subtracting a guess.

Buffer components matter too. Imidazole, DTT above a few millimolar, some detergents and anything with an aromatic ring will contribute. Always blank against the exact buffer the protein is in, not against water.

When to use a different assay

A280 is fast, non-destructive, needs no standard curve and uses no reagent, which is why it is the default. It is the right choice for a reasonably pure protein at a reasonable concentration with a decent aromatic content, and it is the only one of the common methods that gives you an absolute answer from first principles rather than against a standard.

Switch to Bradford or BCA when the extinction coefficient is low or zero, when the sample is a crude lysate where the sequence-based coefficient is meaningless because you do not know what is in there, or when you need to compare across proteins and only care about total protein. Both are relative methods calibrated against a standard — usually BSA, whose response is not the same as your protein's — so they trade the absolute accuracy of A280 for robustness.

The honest summary is that A280 on a pure protein with a computed coefficient is the most accurate of the routine options, and A280 on anything else is the least. Knowing which situation you are in takes one pass of the sequence through a properties calculator.

Related references

Related tools

Related guides

• Hydrophobicity Plots for Transmembrane Domains →
• In Silico Peptide Mass Fingerprinting Guide →
• Six-Frame Translation, and When You Need All Six →