Two calculators give two different isoelectric points for the same sequence, and neither is wrong. The disagreement comes from which set of pK values they use, and that choice matters more for a short synthetic peptide than for a protein. This page sets out the arithmetic, the pK values appropriate to an unstructured peptide, and — just as usefully — the limits of what a charge calculation can tell you.
An ionisable group is not simply "charged" or "uncharged". At any pH it is a population, and the Henderson–Hasselbalch relation gives the fraction in each state. For a basic group — lysine, arginine, histidine, the N-terminus — the fraction carrying a positive charge is 1/(1 + 10(pH − pK)). For an acidic group — aspartate, glutamate, cysteine, tyrosine, the C-terminal carboxyl — the fraction carrying a negative charge is 1/(1 + 10(pK − pH)). Net charge is the sum of the positive fractions minus the sum of the negative ones, and the isoelectric point is simply the pH at which that sum crosses zero.
That is the whole calculation. Everything that makes one calculator disagree with another is in the pK values fed into it.
Most reference tables quote values measured either in free amino acids or in folded proteins. Neither is quite right for a short synthetic peptide, which is usually unstructured in solution. The closest match is the set measured in uncharged alanine-based model pentapeptides, which Grimsley, Scholtz and Pace tabulated alongside their survey of folded proteins1.
| Group | Model pentapeptide | Average in folded proteins | Charge change as pH rises |
|---|---|---|---|
| Asp side chain | 3.9 | 3.5 ± 1.2 | neutral → negative |
| Glu side chain | 4.3 | 4.2 ± 0.9 | neutral → negative |
| His side chain | 6.5 | 6.6 ± 1.0 | positive → neutral |
| Cys side chain | 8.6 | 6.8 ± 2.7 | neutral → negative |
| Tyr side chain | 9.8 | 10.3 ± 1.2 | neutral → negative |
| Lys side chain | 10.4 | 10.5 ± 1.1 | positive → neutral |
| C-terminus | 3.7 | 3.3 ± 0.8 | neutral → negative |
| N-terminus | 8.0 | 7.7 ± 0.5 | positive → neutral |
Two things in that table are worth dwelling on. First, the standard deviations: aspartate side chains in folded proteins have been measured across a range from about 0.5 to 9.2, because the local environment shifts them2. A number in a table is a starting point, not a measurement of your peptide. Second, cysteine is the group where the model-peptide and folded-protein values diverge most, and it is also the group whose ionisation state governs whether a free thiol will react — worth remembering before assuming a cysteine is available at neutral pH.
Arginine has no measured range in that survey, but the same table gives its pK as approximately 12.3. For any working pH it can therefore be treated as fully protonated, which is why arginine-rich peptides are positively charged essentially everywhere; only above pH 11 or so does the guanidinium group begin to titrate, which matters if a calculated curve is extended that far.1
A pI is one point on a curve, and reporting it alone discards the useful information. Three things are visible in the plot that a single number is not.
Terminal modifications shift the whole curve. C-terminal amidation removes a negative charge; N-terminal acetylation removes a positive one. A peptide and its acetylated, amidated counterpart are different molecules electrostatically, and comparing them in an assay without accounting for that is comparing two variables at once. See acetylation and amidation for what else those caps change.
Net charge is a good predictor of a few things and a poor predictor of others, and most disappointment comes from expecting it to cover the second group.
It predicts well: the direction of electrostatic interaction with a charged surface or partner; behaviour on an ion exchange column; the sign of the charge at a given pH, and therefore which solvent is a sensible first attempt at dissolution; roughly where solubility will be at its worst, since a molecule with no net charge has least reason to stay in water.
It predicts poorly: whether a peptide will dissolve at all. Charge is one term among several. A sequence dominated by hydrophobic residues, or carrying a run of them, can be stubbornly insoluble while carrying a healthy net charge, because the driver is burial of hydrophobic surface rather than electrostatics. Sequences prone to β-sheet formation aggregate for structural reasons that a charge calculation cannot see at all. And a calculation says nothing about kinetics — peptides that eventually dissolve and peptides that dissolve immediately look identical on this plot.
For the practical side of getting a peptide into solution, our existing pages cover it directly: how to dissolve peptides, predicting peptide solubility, what to do when a peptide will not dissolve, and the special cases of free cysteines and disulfide-bonded peptides.
If two tools report different pI values for the same sequence, the difference is almost always one of these.
None of these makes a calculator wrong. They make the number conditional, which is why the honest way to report a pI is alongside the pK set and terminal assumptions used to get it — as the figure on this page does.
| Positive fraction | 1 / (1 + 10(pH − pK)) for Lys, Arg, His, N-terminus |
| Negative fraction | 1 / (1 + 10(pK − pH)) for Asp, Glu, Cys, Tyr, C-terminus |
| pI | the pH where the two sums cancel |
| Best pK set for a peptide | model pentapeptide values, since short peptides are usually unstructured |
| Biggest hidden variable | whether the termini are free, acetylated or amidated |
If a peptide needs to be soluble at a particular pH, that is a design constraint and it is cheaper to handle before synthesis than after. LifeTein will review a sequence before quoting and flag charge and hydrophobicity problems, along with options such as adding a solubilising tag, changing the terminal caps, or supplying a different counterion.