Isothermal Titration Calorimetry
How it works
An ITC experiment titrates one binding partner (the ligand, in the syringe) into the other (the macromolecule, in the cell) in a series of small injections, at constant temperature. Because the reaction chamber is thermally isolated, the instrument measures the differential power needed to keep the cell and a reference cell at the same temperature as heat is released or absorbed by binding — this raw signal is integrated, injection by injection, into a titration curve of heat per mole of injectant against the molar ratio of ligand to macromolecule. Fitting a binding model — usually a simple 1:1 model, but multi-site or sequential models where appropriate — to that curve in a single experiment yields the dissociation constant (KD), the reaction stoichiometry (n), and the binding enthalpy (ΔH); the binding free energy (ΔG = −RT ln Ka) and entropy (ΔS = (ΔH − ΔG)/T) follow directly. No fluorescent, radioactive or chemical labelling is needed, because the heat of binding itself is the signal — which also means the choice of buffer matters: both partners must be dialyzed into the same solution, since a mismatched buffer produces heats of dilution that swamp the binding signal.
What you measure
A single ITC run gives the complete thermodynamic profile of an interaction rather than affinity alone. The dissociation constant KD sets the concentration scale of the interaction; the stoichiometry n confirms how many ligand molecules bind per macromolecule (and flags multivalent or unexpected binding); and the enthalpy ΔH and entropy ΔS reveal the physical driving force behind the affinity — a large favourable ΔH typically reflects hydrogen bonding and van der Waals contacts formed on binding, while a favourable ΔS points instead to the hydrophobic effect or the release of ordered water from the binding interface. Two ligands with identical KD can have very different ΔH/ΔS balances, information invisible to affinity-only assays but that guides which chemical series is worth optimizing further.
A typical experiment
Both macromolecule and ligand are prepared in the same buffer, filtered and degassed. The titrant (syringe) is typically formulated at around ten times the concentration of the titrate (cell); for an interaction of unknown affinity, starting points of roughly 10 µM macromolecule in the cell and 120 µM ligand in the syringe are a reasonable default. The instrument then performs a programmed series of injections — commonly 15 to 25 — automatically, with the whole experiment running in well under an hour and multiple experiments queued to run unattended overnight. The single most important design parameter is the Wiseman c-value (c = Ka × [macromolecule]), which should fall roughly between 10 and 100: too low and the titration curve is too shallow to fit KD reliably; too high and the curve becomes a step function from which KD can no longer be extracted, even though stoichiometry still can.
Applications
Drug discovery & fragment screening. ITC ranks candidate ligands by affinity and, more informatively, by binding mechanism during hit validation, fragment-based screening and lead optimization — distinguishing enthalpy-driven binders, which tend to make specific, optimizable contacts, from entropy-driven ones.
Structure-based design. Titrating a compound in the presence of a second ligand reveals competitive, cooperative or independent binding, helping to map which intermolecular contacts a structure-based design campaign should target next.
Protein–protein & signalling complexes. ITC quantifies the affinity and stoichiometry of complexes relevant to cell signalling and disease biology, including interactions between disease-associated proteins and the small-molecule or peptide binders studied in oncology and neurodegeneration research.
Natural products & nanocarriers. Beyond classical small-molecule ligands, ITC is increasingly used to characterize binding to biomimetic nanocarriers — liposomes, solid lipid nanoparticles and extracellular vesicles — relevant to drug delivery.
Strengths & limitations
ITC’s practical window sits around dissociation constants from roughly 10 nM to 100 µM: tighter interactions need macromolecule concentrations so low that the resulting heat signal disappears into the noise, while weaker interactions demand concentrations high enough that aggregation or non-specific binding can distort the titration curve. Sample purity and accurate concentration determination matter more here than in most biophysical techniques, since a 10–20% error in titrant concentration propagates directly into the fitted KD, ΔH and n. Compared with SPR, ITC needs more material and cannot resolve on- and off-rates — but unlike SPR it requires no immobilization or labelling of either partner, measures the interaction free in solution, and is the only common technique that returns enthalpy and entropy directly rather than as a temperature-dependent derivative.
Frequently asked questions
How much protein do I need?
On the PEAQ-ITC the cell holds 200 µl and you load about 280 µl; the syringe holds 40 µl and you load 75 µl. The cell is usually run at 10-50 µM and the syringe at 10-20 times that. For a 25 kDa protein at 20 µM this comes to roughly 150 µg per titration. Plan for three titrations — a scouting run, the real one, and the controls — so 0.5-1 mg is a comfortable amount to send.
The VP-ITC has a 1.4 ml cell and needs close to ten times more material. We use it only when the heat per injection is too small for the 200 µl instruments.
The concentration is set by the affinity, not by a fixed rule: the cell concentration should be about 10 to 100 times the Kd. If you have no estimate of the Kd, send enough for a scouting titration and we adjust from there.
What buffer should my samples be in?
Both partners have to be in the same buffer, and "same" means the same batch of liquid, not the same recipe written twice. Dialyse the protein overnight and use the final dialysate to dissolve or dilute the ligand. A 1 mM difference in salt or 0.05 pH unit produces dilution heats that can exceed the binding signal.
If ΔH matters to you, keep away from buffers with a large ionisation enthalpy. Tris is around 47 kJ/mol, HEPES 20, phosphate 5. When binding takes up or releases a proton, the buffer enthalpy is added to what you measure. Running the same titration in two buffers with different ΔHion is how you find out whether protonation is involved.
DTT oxidises and gives a drifting baseline; 0.2-0.5 mM TCEP behaves much better. Glycerol below 5% and strictly identical on both sides.
What do I get out of an ITC run, and how precise is it?
One titration gives Kd, the binding enthalpy ΔH and the stoichiometry n directly from the heat; ΔG and ΔS follow. No model of the molecule enters the measurement.
ΔH is typically reproducible to 5-10% when the concentrations are known. Kd usually agrees within a factor of two between independent protein preparations. The soft spot is always concentration: an error in the active protein concentration lands in n, an error in ligand concentration lands in ΔH. Tell us how each concentration was determined — A280 with a calculated extinction coefficient, amino acid analysis, weighing — because that is what sets the real error bar.
Is there an affinity range where ITC stops working?
The window is set by the c value, c = n·[M]/Kd, with [M] the concentration in the cell. Between roughly 5 and 500 the isotherm is sigmoidal with a clear inflection and all three parameters come out. Between 1 and 5 the curve is shallow: Kd is still fittable if n is fixed and the concentrations are solid. Above 1000 the isotherm is essentially a step — you get n and ΔH, and only a lower limit on affinity.
In practice, Kd from about 10 nM to about 100 µM is direct. Tighter binding is handled by displacement: saturate the protein with a weaker competitor of known Kd and titrate the strong ligand into that mixture. Kd weaker than about 1 mM would need concentrations most proteins will not survive.
Do you need to label or immobilise my protein?
Nothing is attached to anything. The measurement happens in free solution; the only requirement is that both partners stay soluble and folded at the working concentration for the ninety minutes the titration lasts.
This is usually why people come to ITC after a fluorescence or SPR result — there is no surface, no tag and no fluorophore that could be shifting the number.
Can you work with membrane proteins, nucleic acids or crude samples?
Nucleic acids and small molecules are routine, and so are protein-protein and protein-peptide systems when the concentrations are reachable.
Membrane proteins in detergent work if the detergent is above its CMC and at exactly the same concentration in cell and syringe. Micelle dilution heats are large and will bury a modest binding signal. Nanodiscs and SMALPs have been measured, with the same requirement that the empty-particle background be matched.
Lysates, serum and conditioned media do not work. Anything in the tube that changes state on mixing contributes heat, so the sample has to be a defined two-component system.
My titration came out flat. What went wrong?
A flat trace has a short list of usual causes. The protein may already be saturated by a ligand carried over from purification. The interaction may be real but nearly athermal at that temperature, with ΔH close to zero; repeating at 15 °C and at 35 °C normally settles it, since ΔCp moves ΔH with temperature. The active fraction may be well below the nominal concentration. Or there is genuinely no binding.
A trace with a constant non-zero heat from first to last injection is usually a buffer mismatch, not a weak interaction. We run ligand-into-buffer and buffer-into-protein as a matter of course so that the two can be told apart.
How long does an experiment take?
A 13-19 injection titration on the PEAQ-ITC runs 60-90 minutes, plus equilibration and washing between runs. The VP-ITC is closer to 2.5 hours. A working day fits six to eight titrations including controls.
What stretches a project is the number of conditions rather than the machine time: a temperature series, a pH series, a competition experiment or a mutant panel each multiply the run count. The delivery date depends on the queue and on the number of conditions agreed in the quotation.
What should I send together with the samples?
The dialysed protein, the ligand, and at least 15-20 ml of the final dialysis buffer in a separate tube. The buffer is not an extra — it is what the blanks and any dilution are made with.
On paper we need: buffer composition with pH and reducing agent, the concentration of each sample and the method used to measure it, molecular weights, extinction coefficients, the solubility limit of the ligand, and an expected Kd if you have one, even as an order of magnitude. If the ligand is in DMSO, give the stock concentration and the final percentage.
Tell us how the protein is stored, whether it tolerates freeze-thaw, and ship accordingly on dry ice or wet ice.
Do I have to be there, and can I learn to run the instrument myself?
Presence is not required. Most orders run as a service: the samples arrive, we measure, you receive the raw thermograms, the fitted parameters and a short report.
Coming for the first titration is worth the trip when the system is unusual or when you expect to do a lot of ITC afterwards. Hands-on training is possible for people who will use the instrument regularly; count on two or three supervised sessions before working alone. For a single measurement it rarely pays off.
Instruments
ITC200
PEAQ-ITC
VP-ITC