Before You Trust That Screening Hit, What Is DMSO Doing to Your Protein?

If a compound gives you a nice thermal shift you're obviously going to go and follow that up.

Before we get too attached to it, though, I'd check what the same amount of DMSO does without the compound.

Questions of this type are cropping up all the time in customer meetings. The more I get to know assay development the more I'm looking at the conditions under which a result is obtained. DMSO is used to solubilise the screening compounds but it can also have an effect on the protein and the interaction that you're measuring.

Check the solvent before ranking the compounds

My starting point would be a protein only experiment over the DMSO range as part of the screen with all other conditions being identical.

Differential scanning fluorimetry (DSF) monitors the fluorescence as the protein is heated. The apparent melting temperature (Tm) is the temperature at which the unfolding transition is observed.

In a 2024 study by Feoli and colleagues, adding 1% DMSO produced quite different results for two enzymes:

So "we normally use 1%" wouldn't be enough for me. I'd want to know how that particular protein behaves.

Once you've established your working DMSO concentration, keep that constant throughout the compound dilution series and the controls. Otherwise, the compound and solvent concentrations will be going up and down together.

There are a few checks I’d keep in the experiment:

• Protein alone (no compound) at the same DMSO concentration.
• Compound containing buffer (no protein) to look for any added fluorescence or background.
• A known binder, where available, to confirm that the assay still gives the expected response.

For an enzyme target, I'd also look at activity. A readable unfolding curve doesn't necessarily mean the enzyme is active at the temperature you're screening at.

What does the result actually mean?

Intrinsic DSF uses the protein's own fluorescence (no reporter dye is added) so that removes one element from the assay. Compounds can still fluoresce or absorb the light used in the measurement though. The nanoDSF review from November 2026 looks at those issues.

I'd look at the starting signal, the curve shape and the repeats before I'd go any further and group everything by Tm. A protein free blank is of use for background checks but it won't rule out every possible effect on protein fluorescence.

Then there's the binding question. A larger thermal shift doesn't in any way imply tighter binding and gives you absolutely no information at all about the rate of binding or the rate of dissociation.

Surface plasmon resonance (SPR) monitors the interactions at the sensor surface over time and can give you both binding kinetics and affinity information if the surface assay is appropriate.

The solvent does make a difference. The Feoli SPR work used reference subtraction and solvent correction, but the binding rates were different in DMSO. Correcting the optical response in no way cancels out the physical effect on the interaction.

A separate galectin-3 study actually does go some way to explaining why that distinction is important. Increasing the DMSO did slow the ligand association (as you'd expect given the increased solvent viscosity) but the affinity change was very small.

That's something to keep in mind when two conditions lead to the same affinity value. The interaction could still be very different.

Where I see the instruments helping

For this kind of condition screening the SUPR-DSF format is absolutely the right one. Nicoya can run up to 384 samples per run with 10 µL sample volumes. You can of course allocate space for solvent conditions, blanks and replicates in the experiment and still keep the individual sample volumes very small.

It also collects full fluorescence spectra. That gives you the spectral information to look at in conjunction with the fitted melting point if a condition gives you an odd result.

There's a published example using SUPR-DSF: Cohrs and co-workers found that the DMSO conditions they tested had no effect on the background or the unfolding curves for hen egg-white lysozyme in their buffer. A useful result for that experiment, with the same target specificity qualification as required for the next protein.

For the next binding question, Revo could provide the complementary kinetics and affinities.

Before we go for the best looking compound we'd want a solvent matched result that we can explain. That gives the next experiment a hell of a lot better basis to start from.

Desmond Gilmour

Email: dgilmour@nicoyalife.com

Phone: +1 519-546-9876

I work at Nicoya Lifesciences. The views expressed here are my own and do not necessarily reflect those of Nicoya. This is a personal publication, not an official company communication.