Differential Scanning Calorimetry
How it works
PEAQ-DSC scans two matched cells — one with the protein sample, one with matching buffer as reference — through a temperature ramp, and measures the differential power needed to keep both cells at the same temperature at every point in the scan. As the protein unfolds, it absorbs heat (an endothermic transition), and the extra power required to keep pace with the reference cell traces out a peak in heat capacity against temperature — the melting temperature Tm is the peak of that transition, where half the protein population is folded and half unfolded, and the area under the peak is the enthalpy of unfolding ΔH. Because no dye, probe or label is added, DSC reads the protein's own thermal behaviour directly, and a scan rate slow enough (typically well under 1 °C/min) for the sample to stay close to equilibrium throughout is essential for the recorded Tm to reflect a true thermodynamic transition rather than an artefact of scan speed.
What you measure
A DSC thermogram gives Tm (thermal stability), ΔH (the energetic cost of unfolding, related to how cooperatively the structure unfolds) and ΔCp (the change in heat capacity on unfolding, related to how much hydrophobic surface becomes exposed to solvent). Multi-domain proteins and antibodies typically show two or three overlapping or separate transitions rather than one — for an antibody, the Fab and Fc domains commonly unfold at different temperatures, so a single scan resolves the stability of each domain independently, which a global measurement (like the apparent Tm from nanoDSF's single fluorescence-derived transition) may not distinguish as clearly.
A typical experiment
Sample and reference buffer are degassed and loaded into matched cells, and the scan runs from a low starting temperature, typically close to the storage or formulation temperature, up through and beyond the unfolding transition(s) — often to 100 °C or higher. Protein concentrations of roughly 0.5–2 mg/mL and sample volumes of a few hundred microlitres per run are typical for microcalorimeters like the PEAQ-DSC; a full scan generally takes on the order of an hour depending on the temperature range and scan rate, and a rescan of the same sample after cooling checks whether the unfolding was reversible.
Applications
Biosimilar comparability. DSC is widely accepted by regulators as an orthogonal method for demonstrating that a biosimilar shares the same higher-order structure and thermal stability profile as its originator molecule.
Antibody & fragment developability. Resolving separate Fab and Fc unfolding transitions helps rank antibody candidates and engineered fragments by domain-level stability during early developability assessment.
Formulation & stress-stability studies. Comparing thermograms across candidate buffers, excipients and stress conditions (freeze–thaw, agitation) identifies which formulation best preserves native structure.
Structural integrity & folding studies. Because DSC reads the protein directly, it detects subtle changes in folding or domain integrity that can result from mutation, glycosylation differences or manufacturing changes.
Strengths & limitations
Many proteins, especially at the higher concentrations DSC requires relative to nanoDSF, aggregate or precipitate once unfolded rather than refolding cleanly on cooling — this irreversibility means the recorded transition is kinetically, not just thermodynamically, controlled, and the apparent Tm can shift measurably with scan rate, which is why scan-rate dependence is itself used as a diagnostic for irreversible unfolding. DSC also needs more material and more time per sample than nanoDSF, making it a poor fit for screening large numbers of buffer or formulation conditions — but because it measures heat directly rather than an indirect fluorescence proxy, it remains the reference, regulator-recognized method against which faster screening techniques are validated, particularly for higher-order-structure comparability studies.
Frequently asked questions
How much protein do I need, and at what concentration?
The PEAQ-DSC has a 130 µl capillary cell, but the autosampler draws 300 µl from the well, so send at least 400 µl per condition. Standard working range is 0.5-1 mg/ml; well-behaved proteins give usable thermograms down to 0.1 mg/ml, occasionally 0.05.
For one melting curve with its buffer baselines that means roughly 0.3-0.5 mg of protein. A formulation screen of ten conditions multiplies that, which is normally the point where nanoDSF becomes the more sensible choice.
Which buffer, and how should it be prepared?
The reference cell is filled with your buffer, so the buffer has to be the dialysate of the sample, not a fresh preparation from the same protocol. Send 10-15 ml of it.
Watch the temperature dependence of the pKa. Tris moves by about -0.028 pH unit per °C, so a buffer at pH 8.0 on the bench is near pH 6.6 at 50 °C, and a Tm measured in Tris is not a Tm measured at pH 8. Phosphate (-0.0028) and citrate are much flatter; HEPES sits in between at about -0.014.
Samples and buffer are degassed before loading, and both go into the instrument bubble-free — a bubble crossing the capillary during the scan produces a spike that cannot be corrected afterwards. Avoid DTT, which oxidises across the scan and distorts the baseline; TCEP at 0.5 mM is stable.
What does a DSC scan tell me?
The direct output is excess heat capacity against temperature. From it come the transition midpoint Tm, the calorimetric enthalpy ΔHcal from the peak area, the van 't Hoff enthalpy ΔHvH from the peak shape, and the change in heat capacity ΔCp between folded and unfolded states.
A multi-domain protein gives several peaks, and DSC is one of the few methods that resolves them and assigns an enthalpy to each. The ratio ΔHvH/ΔHcal is the usual diagnostic: close to 1 means a two-state transition of a single cooperative unit; well below 1 means intermediates or independent domains; well above 1 points to an oligomer unfolding as one unit.
Tm is reproducible to about 0.1-0.2 °C on the same sample; ΔHcal carries the concentration error directly, so 5-10% is realistic.
How is this different from nanoDSF, and which should I choose?
Both give you a Tm. Only DSC gives you the enthalpy, the heat capacity change and the number of thermodynamic domains, because it measures heat rather than a spectroscopic proxy for structure.
The price is material and time: 400 µl at 0.5 mg/ml per condition against 10 µl at similar concentration for a capillary on the Prometheus, and one scan an hour against 48 in parallel.
The usual split is nanoDSF for screening — buffers, additives, ligand hits, batch comparison — and DSC on the two or three conditions that matter, when you need the thermodynamics or when the protein has no tryptophan.
My protein aggregates when it unfolds. Is DSC still useful?
Yes, but read the numbers for what they are. Most protein unfolding above 50 °C is followed by aggregation, which makes the transition irreversible. An irreversible Tm is an apparent Tm: it shifts with scan rate and it cannot be used for equilibrium thermodynamics. ΔHcal from an irreversible peak is a lower bound, since part of the signal is lost to the exotherm of aggregation.
The standard test is a rescan of the same sample. If the second scan reproduces the first, the transition is reversible and full thermodynamic analysis applies. If the second scan is flat, you have an apparent Tm — still perfectly good for ranking constructs, buffers or ligands, which is what most people need it for.
Heavy aggregation also blocks the capillary. If your protein precipitates visibly by 60 °C, tell us in advance so we can lower the concentration or stop the scan early.
Can I see ligand binding with DSC?
A ligand that binds the folded state raises Tm, and the shift grows with concentration and affinity. Comparing apo and saturated samples is a clean way to confirm engagement, and for a tight binder the shift can be 5-15 °C.
Extracting a Kd from ΔTm is possible but demanding: it needs ΔHcal, ΔCp, the ligand concentration and an assumption that binding is to the folded state only, and the Kd you obtain refers to the temperature of the transition, not to 25 °C. If you want a room-temperature Kd, ITC or SPR is the direct route and DSC is the confirmation.
What kinds of samples can you run?
Soluble proteins, antibodies and antibody fragments, nucleic acid duplexes and structured RNA, protein-ligand and protein-protein complexes, and formulated biologics.
Antibodies are a common case and give the familiar three-peak thermogram, with CH2 lowest, then Fab, then CH3, which is useful for comparing formulations or batches.
Membrane proteins in detergent can be scanned, though the micelle contributes to the baseline and the reference has to contain exactly the same detergent at the same concentration. Liposomes and lipid phases give their own strong transitions, which is either the signal you want or the noise that hides your protein. Lysates and other undefined mixtures are not suitable.
How long does a run take, and how many samples fit in a day?
A 20-100 °C scan at the standard 60 °C/h takes about 80 minutes, plus the buffer-buffer baselines that bracket the series and the cleaning cycle between samples. That is six to eight scans in a day.
Raising the rate to 180-240 °C/h roughly triples throughput and is fine for Tm ranking, but a fast scan pushes an irreversible Tm upward and distorts peak shape, so it is not the setting to use when you want ΔHcal. For anything kinetically controlled we run at least two rates so the scan-rate dependence is visible rather than hidden.
What should I send with the sample?
Sample and its dialysis buffer, plus the composition of that buffer, the concentration and how it was measured, the molecular weight, and the temperature range you care about.
Say whether the protein is known to precipitate, whether it has been through freeze-thaw, and whether any ligand or cofactor is present. A cofactor that dissociates during the scan produces a second transition that is easy to misread as a domain.
Do I need to come, or can I be trained on the instrument?
Samples can be sent and measured without you. You get the raw thermograms, the baseline-subtracted and concentration-normalised curves, the fitted parameters and the analysis.
Training makes sense for groups running regular stability work — formulation, construct comparison, batch release. The instrument itself is not hard to operate; the part that takes practice is baseline handling and deciding what an irreversible transition allows you to claim, and that is what the sessions concentrate on.
Instruments
PEAQ-DSC