Reducing interchain disulfides
An IgG1 has four interchain disulfides and eight cysteines available on full reduction. Conjugating to them means reducing some and leaving the rest, and the reduction step sets the DAR distribution before a single equivalent of payload is added. Everything downstream inherits whatever happens here.
Partial reduction is a stoichiometry problem with a kinetic wrapper
The four interchain bonds do not reduce at equal rates. The two hinge disulfides between the heavy chains are more solvent accessible and go first. The heavy-light bonds follow. This is why a DAR 2 material is enriched in hinge-conjugated species rather than being a random draw from eight sites.
Equivalents of reductant matter more than time, provided the reaction is allowed to reach a steady state. Two equivalents of TCEP per antibody gives roughly DAR 4 after conjugation, and one to 1.5 gives roughly DAR 2, but these are starting points rather than constants. Lot-to-lot variation in the antibody, the amount of free thiol already present, and the buffer all move it.
Temperature and time buy you control at the margin. Running at 4 degrees rather than 37 slows everything and widens the window in which you can stop, which matters at scale where a large batch cannot be quenched instantly.
TCEP and DTT are not interchangeable
TCEP is a phosphine and does not carry a thiol. That is the practical difference. It does not compete with the antibody cysteines for maleimide, so it does not need removal before conjugation, and single-pot reduce-then-conjugate is possible. It also works across a wide pH range and does not need to be freshly made every time.
DTT carries two thiols and will be alkylated by any maleimide present, consuming payload and producing a small-molecule adduct that then has to be cleared. It must be removed before conjugation, usually by desalting or diafiltration, and that removal step reoxidizes some of what you just reduced.
TCEP has its own failure mode. At high excess it can reduce the intrachain disulfides, which destabilizes the domains and shows up later as aggregation or loss of binding. Intrachain reduction is slow and buried, so it is usually a sign the reaction ran too long or too hot rather than too concentrated.
Reoxidation is real and it happens on your timescale
Free thiols reoxidize in air, catalyzed by trace copper and iron. A reduced antibody left on the bench for two hours is not the same material you measured. The reduction is partially reversed, the distribution shifts, and the conjugation you run against it gives a lower DAR than the reductant stoichiometry predicted.
This is the most common reason a process that worked at small scale drifts at larger scale, where hold times between steps are longer by necessity.
Two defenses. Chelate the metals with EDTA at low millimolar, which is cheap and effective. Keep the hold between reduction and conjugation short and defined, and write it into the process rather than leaving it to whoever is running the bench that day.
Measure the thiols, do not infer them
Free thiol content after reduction is measurable in minutes by Ellman assay against a cysteine standard. Run it. The number tells you whether the reduction went where you thought, and it is the input to the payload stoichiometry.
A PEG-maleimide shift assay is more informative and only slightly slower. Alkylate with a five kilodalton PEG-maleimide and run reducing SDS-PAGE. Each conjugated thiol adds a visible mass shift, so the ladder shows you the distribution of occupied sites rather than a single averaged number.
Where the antibody arrives with unexpected free thiol already present, and many do, that changes the equivalents you need. Measure it on the starting material as well as after reduction. A therapeutic-grade antibody with half a free thiol per molecule is not unusual, and treating it as fully oxidized will overshoot the target DAR every time.