You attach a drug to an antibody and the binding curves look almost the same. Would you carry the parent antibody's formulation straight over to the conjugate? I'd want to see what happens when the two samples heat up, cool back down and take some stress.
In a 2026 Nicoya poster, Michael Piazza compared sacituzumab with its antibody–drug conjugate (ADC), sacituzumab tirumotecan. Their target is a cell-surface protein called TROP2. The ADC carries about seven to eight copies of a topoisomerase I inhibitor, attached to cysteines freed by breaking the antibody's interchain disulfide bonds. An IgG1 has only four of those bonds, so at this loading most of the covalent links holding the chains together are gone.
How well did the conjugate bind TROP2?
Using surface plasmon resonance (SPR) on Alto, the team followed TROP2 as it bound to each sample and then let go. The dissociation constant ((K_D)), a measure of binding strength, was 1.51 nM for the parent antibody and 1.42 nM for the ADC. Their binding and release rates were within about 6% of each other. In this assay, the conjugate retained its affinity for TROP2.
Cheng and colleagues saw similar binding and uptake into cells for the same pair. For clinical benefit, the ADC now has phase 3 results in breast and lung cancer.
Still, an ADC's (K_D) is an average across molecules carrying different amounts of drug. In another Nicoya study, a biosimilar of the bispecific antibody mosunetuzumab matched the original almost exactly in bulk (13.4 nM vs 12.9 nM). One of its charge fractions, though, bound about nine times more weakly. If binding were my main reason to clear a conjugate, I'd split it by drug load and check the fractions too.
Which sample handled heat better?
On SUPR-DSF, the team used intrinsic differential scanning fluorimetry to follow the samples' natural fluorescence as they heated up. Changes in that signal can reveal unfolding. They also used modulated scanning fluorimetry to ask a different question. After each trip to a higher temperature, how well did the signal recover when the sample cooled?
| Sample | Unfolding onset ((T_{onset})) | First unfolding midpoint ((T_{m1})) | Onset of incomplete recovery ((T_{nr})) |
|---|---|---|---|
| Parent antibody | 63.6 °C | 71.5 °C | 65.3 °C |
| Parent, after oxidative stress | 59.6 °C | 66.7 °C | not reported |
| ADC | 58.5 °C | 66.0 °C | 73.8 °C |
| ADC, after oxidative stress | 46.7 °C | 63.3 °C | not reported |
A single spreadsheet column labelled "stability" could rank either sample first. The ADC started unfolding earlier, yet its fluorescence recovered after hotter cycles. The parent began losing recovery about 6 °C before its own (T_{m1}). The ADC held on until about 8 °C past its own.
I'd give (T_{nr}) more weight than (T_{m1}). In a study of 13 antibodies, Svilenov and colleagues found that (T_{nr}) tracked the onset of aggregation closely (R = 0.93), while (T_{m1}) barely did (R = 0.20). Those were unconjugated antibodies, though, and (T_{nr}) doesn't measure aggregation directly. To see whether heating made the molecules stick together, I'd still compare heated and unheated samples by size-exclusion chromatography.
What happened under oxidative stress?
This is the part I'd show a formulation team first. After oxidative stress, the parent's unfolding onset fell 4.0 °C. The ADC's fell 11.7 °C. Yet the ADC's (T_{m1}) moved less than the parent's (2.7 °C vs 4.8 °C). If you only tracked (T_{m1}), you'd think the ADC handled oxidation better. The onset suggests part of the stressed ADC starts unfolding well before the rest.
The poster doesn't describe the stress conditions or report (T_{nr}) for the stressed samples, so I'd read these numbers as a direction, not a final answer. My next run would put both molecules, stressed and unstressed, through modulated scanning fluorimetry on the same plate. It uses 10 µL per well and runs overnight.
What changed about the antibody's fold?
On the Chirascan V100, both samples gave the far-ultraviolet circular dichroism (CD) pattern of an antibody rich in beta sheets, where stretches of the protein chain lie side by side. The overall fold looked familiar. But when the team scaled each curve to the same total area and compared their shapes, all four ADC measurements scored outside the parent antibody's reference range. They weren't borderline either, scoring 4.7 to 9.1 against a limit of 2.
Before calling that a change in shape, I'd rule out two simpler explanations. Disulfide bonds add to the far-UV signal near 230 nm, so losing the interchain bonds could shift the curve on its own. And some drug payloads absorb light at wavelengths used for near-UV CD and fluorescence, so I'd check where the linker–payload itself absorbs and emits light. If the difference holds up after those checks, hydrogen–deuterium exchange on FastHDX could show which part of the antibody moved.
So can the parent formulation carry over?
Not on this evidence. Binding said nothing changed. (T_{m1}) said the ADC was weaker, and (T_{nr}) said it was tougher. Oxidative stress exposed a fragile fraction, and CD said its shape was reproducibly different. Any one of those numbers alone could have led to a confident call. Together they say the conjugate is a different molecule from the one the formulation was built for.
I'd treat conjugation as a formulation event, not just a chemistry step, and earn the formulation again on the conjugate itself. If the workflow includes a room-temperature hold, I'd also compare binding in fresh and held samples of both molecules before deciding anything.
Which part of your parent-versus-conjugate comparison remains unresolved for you: binding, thermal behaviour or structure? If you'd like to talk through how to set up that experiment on your own pair, reach out.
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.
