Is freezing a reconstituted peptide solution ever acceptable, and what exactly does an ice crystal do to it?

Accepted answer

Your reasoning fails at one specific point: freezing does not stop chemistry, it concentrates it. The mechanical shear story is largely a myth, and the real mechanisms are cryoconcentration, interfacial denaturation at the ice front, and buffer-driven pH shift. Take them in turn.

1. Cryoconcentration is the main event

When an aqueous solution freezes, it does not freeze as a homogeneous block. Pure ice crystallises out first, and everything that is not water — peptide, buffer salts, preservative, bulking agent — is excluded from the growing crystal and pushed into a shrinking unfrozen phase between the crystals. That freeze-concentrated liquid can reach solute concentrations many times the starting value before it eventually vitrifies.

Consider what that does to your 1.5 mL at 5 mg/mL. As freezing proceeds, the residual liquid phase can transiently be at tens of milligrams per millilitre. Aggregation is a nucleation-limited, higher-order process in concentration, so a transient tenfold concentration increase is not a tenfold increase in aggregation rate — it is worse than that. And this happens on both the way in and the way out, because thawing passes back through the same concentrated regime.

So the answer to "does chemistry stop" is no. Rate constants drop with temperature, but concentrations rise, and for the concentration-dependent failure modes the second effect can dominate right through the phase change.

2. The ice-water interface, not mechanical shear

The ice front is a large, cold, structured surface. Peptides adsorb to it, and adsorption to any surface promotes partial unfolding and subsequent association — the same mechanism as air-liquid interface aggregation, which is why shaking a vial is harmful. Freezing 1.5 mL creates an enormous interfacial area compared with the air-liquid surface of the same solution sitting quietly in a fridge. That is where the damage happens.

Nothing is being sheared. A 4 kDa peptide is far too small for hydrodynamic shear at these scales to matter; it is not a mammalian cell and it is not a shear-sensitive high-molecular-weight polymer. Drop the shear framing entirely — it leads you to the wrong conclusions about rate, because it suggests fast freezing is more violent when the opposite is true.

3. pH shift on freezing

This one surprises people and it is well documented. In sodium phosphate buffer systems, the dibasic salt crystallises out preferentially as freezing proceeds, which strips base from the residual liquid and drives its pH down — shifts of two to three pH units on freezing are reported for sodium phosphate. Your peptide, sitting in that shrinking liquid phase, is briefly in a solution nothing like the one you made.

The practical consequence is that a cake formulated with a phosphate buffer is more freeze-sensitive in solution than one formulated with histidine or citrate, and you generally do not know which you have. Note that this also means the intuition "buffered is safer" reverses on freezing: the buffer is the thing that shifts.

4. Freezing rate

Rate matters and the direction is the opposite of the shear intuition. Fast is better. Rapid freezing produces many small ice crystals and spends less time in the freeze-concentrated regime; slow freezing produces fewer, larger crystals and, more importantly, holds the solution in the damaging partially-frozen state for far longer. A domestic freezer taking 40 minutes to freeze 1.5 mL is close to the worst available protocol. Snap-freezing in liquid nitrogen is far gentler, which is exactly backwards from how people expect it to work.

The same asymmetry applies to thawing: fast, gentle warming to the target temperature beats a slow ramp, because again you are minimising residence time in the concentrated phase. Not in hot water — you would then trade one problem for a thermal one — but a controlled warm-hand or room-air thaw beats leaving a vial in the fridge overnight to come up slowly.

5. Cycles

Damage is roughly linear in cycles for the interfacial component, because each cycle creates a fresh ice front, but with an important non-linearity: aggregates already present act as nucleation sites, so cycle five on a solution that already has soluble oligomers does more than cycle one did on a clean solution. The practical form of this is that one cycle is usually survivable and five is a different question.

6. Why protein labs do it anyway

Because they aliquot to eliminate cycles, and because their material is formulated for it. A research-grade protein stock that gets frozen routinely typically contains a cryoprotectant — glycerol, sucrose, trehalose — and a surfactant, and it is aliquoted so each tube is thawed exactly once. Under those conditions freezing is genuinely the best storage option available.

Your reconstituted vial has none of that. It is unbuffered or unknown-buffered, has no cryoprotectant, no surfactant, and it is one container you would thaw repeatedly. That is the version of the practice that does not work.

So what should you do about six weeks?

Neither option in your framing. Do not reconstitute more than you will use. If you knew you would use 0.5 mL, that was the reconstitution volume. Six weeks of a 4 °C hold and a freeze-thaw of an unprotected solution are both bad answers to a question you should not have been asked, and the fix is upstream: reconstitute small, keep the rest as dry cake, where the storage problem is nearly free.

edited 15 Jan 2025 by sian_llewellyn — added a caveat about sampling