Why low-pH elution is the wrong tool for sensitive modalities

Protein A affinity chromatography has run on the same elution chemistry for forty years: bind at neutral pH, elute at pH 3.0–3.5. The platform was built around full-length IgG, which tolerates a brief acid hold without losing structure or activity. For monoclonal antibodies destined for first-generation biologics pipelines, that worked — and still works.

The pipelines have changed. Bispecific antibodies, antibody-drug conjugates, viral vectors and cell-derived products now make up a growing share of clinical development. These formats are not built like a robust IgG. The assumption that an acid hold is a free operation is the assumption that quietly fails in the downstream — usually as aggregation in the eluate, sometimes as outright loss of activity, occasionally as a comparability problem that surfaces only in late-stage development.

What low-pH elution actually does

Protein A binds the Fc through a largely hydrophobic interface at the CH2–CH3 junction, with a small number of histidines sitting in it. Histidine has a side-chain pKa near 6, which is why the interaction is pH-sensitive at all — but protonation alone does not release the antibody. Full release around pH 3.0–3.5 also needs the net positive charge and the local conformational change that the lower pH brings, which is precisely why the step cannot be made gentle by buffer chemistry alone.

What the mechanism does not control is what happens to the rest of the molecule. At pH 3.5 the entire antibody is exposed to conditions far from its native environment. Surface charges flip, internal salt bridges weaken, hydrophobic patches that the native fold keeps buried become accessible. For a thermodynamically stable IgG1, the molecule re-folds when neutralised. For less stable formats, parts of the journey are not fully reversible.

Where it breaks

Bispecific antibodies

In an asymmetric format the heterodimer interface is the least stable part of the molecule, and it unfolds first. The downstream signature is a monomer peak that loses area and an aggregate peak that grows. What the acid step does not do is create mispaired species — those come from assembly in the cell — but it does make the correctly paired molecule less likely to survive the step intact.

Antibody-drug conjugates

Some linker chemistries are genuinely acid-labile — the hydrazone class was designed to be cleaved by exactly this pH. Most linkers in current clinical use, such as valine-citrulline-PAB, tetrapeptide and non-cleavable thioether linkers, are not, and the maleimide failure mode is retro-Michael exchange at physiological pH rather than acid cleavage. For an ADC the acid step matters less through the linker than through the conjugate itself: payload loading lowers the conformational stability of the antibody, so a conjugated molecule can aggregate at a pH its naked parent tolerates.

AAV and lentiviral vectors

Viral vector capsids are not antibodies, and they are not built for an acid environment. AAV capsids in particular show conformational changes below pH 4 — VP1/VP2/VP3 protein arrangement loosens, surface antigenic structure shifts, and at the low end of the typical Protein A elution range the capsid integrity itself becomes uncertain. For an AAV polishing step that aims to separate full and empty capsids, an upstream acid hold that has already compromised the full population works against the entire premise of the separation.

Fragile fusion proteins and cell-derived products

The pattern extends beyond named modality classes. Fc-fusion proteins where the fused partner is a flexible peptide or a small folded domain often unfold at pH 3.5 even when the Fc portion does not. Cell-derived therapeutics with surface-bound antibody fragments lose viability under acid exposure. The common thread is that classical Protein A elution is a chemistry designed around one molecule — robust full-length IgG — being applied to molecules it was never validated on.

The downstream cost

When acid elution drives aggregation, the downstream pays in three ways. The aggregate-removal polishing step works harder — typically a cation-exchange or HIC pass that now has to clear a larger and more heterogeneous aggregate fraction, with lower recovery. Yield drops, often by several percent per step. And for programmes that reach late-stage development, the acid-induced micro-heterogeneity becomes a comparability question: does the process at scale produce the same product as the process at bench? That question is harder to answer than to ask.

None of this shows up in a development summary table that only lists Protein A capture yield. The cost is paid by every step that follows.

What classical workarounds offer — and miss

The full set of levers is also larger than the elution buffer: the product concentration in the eluate pool, the speed and the path of the neutralisation — the pool crosses its isoelectric range on the way back — the buffer species at identical pH, the ionic strength and the temperature. A team that tunes all of them can bring a marginal molecule through an acid step. The point is not that it is impossible; it is that every one of those levers is a constraint the release chemistry imposes on the rest of the process.

All of these levers — milder ligands, amino-acid and arginine buffers, glycols, salt and calcium systems, detergents, alternative ligands, mixed-mode capture and simply shortening the acid contact — are set out side by side, with what each one costs and which of them genuinely remove the acid rather than reduce it, in how to elute from Protein A without acid. That page is the place to start if you are choosing between them; this one is about why the problem exists in the first place.

For a robust IgG, the optimised acid step is good enough. For a bispecific where the heterodimer interface starts to relax at pH 4.5, raising elution pH from 3.5 to 4.0 reduces — but does not eliminate — the aggregation. The structural problem is that affinity is being released by changing the chemistry of the buffer, and the molecule cannot tell which part of the chemistry the change is supposed to act on.

A structural alternative: control by light, not chemistry

PureLight® is Lumatix's approach to that structural problem. The affinity ligand is built around a photoswitchable core: a small azobiaryl group integrated into a 3-helix-bundle scaffold that carries the target-binding interface. The chemistry of the elution buffer no longer controls release. A light pulse does.

Red light (630 nm) keeps the ligand in its high-affinity conformation. The product binds at native pH and ionic strength — the same buffer the harvest arrives in. Blue light (480 nm) triggers a reversible cis/trans isomerisation in the photoswitch; the ligand changes conformation, the affinity collapses, and the product elutes in whatever buffer you choose. No acid hold, no high-salt step, no chaotrope. The release event is optical, not chemical.

Solaris® Protein A runs on the same modified cellulose monolith as MonoCore™, with the same 15 µm channels and the same capsule format — the matrix does not change, only the ligand and the way it lets go.

When PureLight matters — and when it does not

Two boundaries belong here before anything else. First, in a licensed Fc process the longest acid exposure is not the elution — it is the low-pH viral inactivation hold, around pH 3.5 for 30 to 60 minutes. Eluting under light takes the elution out of the acid budget; it does not take out the inactivation step, which has to be solved separately where a product cannot take it. Second, Solaris® Protein A binds Fc. Affinity capture of AAV does not run on Protein A but on camelid VHH ligands, which elute in a similar pH range, and lentiviral capture usually runs on anion exchange — so for viral vectors the acid problem is real but the ligand does not exist yet on this platform. That is the discovery programme, not a product.

PureLight is not positioned as a replacement for classical Protein A on every process. For a robust full-length IgG running at commercial scale under cost-of-goods pressure, the optimised acid step is well understood, well validated, and economically hard to beat. Switching elution mechanisms on a working process buys little.

The fit is on sensitivity-critical modalities — bispecifics, ADCs, viral vectors, fragile fusion proteins, cell-derived products — where the acid step is the unit operation that quietly limits yield and complicates comparability. On those processes, neutral-pH elution is not a convenience. It is the difference between a downstream that works at scale and one that fights itself at every polishing step.

Outlook

Protein A is the first ligand on the PureLight platform; it is not intended to be the last. The 3-helix-bundle scaffold is designed to host different target-binding interfaces while keeping the photoswitchable release mechanism intact. The longer-term direction is a discovery platform for light-controlled affinity ligands against targets where no good affinity tool currently exists — which is most of the post-IgG biologics pipeline.

For now, the practical question is narrower. If your process has a step where the acid hold is the part you would change if you could, that is the step PureLight is built for.

Further reading

For the matrix layer that sits underneath both MonoCore™ and Solaris®, see our comparison of monolith and membrane chromatography. The PureLight® technology page covers the photoswitch mechanism in more detail and links directly to the Solaris® Protein A Beta application.

For the route as a whole, and where the elution step sits within it, see antibody purification: how the standard route works.

Frequently asked questions

Which antibody formats are most sensitive to low-pH elution?

Bispecific antibodies with an asymmetric heterodimer interface, antibody-drug conjugates with acid-labile linker chemistry, Fc-fusion proteins whose fused partner is a flexible peptide or small folded domain, and viral vectors, whose capsids are not built for an acid environment. Full-length IgG1 tolerates the standard acid step well, which is exactly why the platform was built around it — and why the assumption travels badly to newer formats.

Is it the elution or the virus-inactivation hold that damages the product?

Usually the hold, because it lasts far longer. The elution passes in minutes; a low-pH hold typically runs 30 to 60 minutes at pH 3.6 or below. The exception is a process where the eluate stands before neutralisation, or where the elution runs at pH 3.0 against a hold at pH 3.6 — then the elution is the more acidic event. Take a sample directly from the eluate and another after the hold and compare by size exclusion: that single experiment tells you which step to work on, and it is common for the capture step to be blamed for what the hold did.

Does a shorter residence time help?

It reduces the column part of the exposure, and not the part that usually matters more: the eluate waiting in a collection vessel before neutralisation is acid exposure too, whatever the column format. For acid-catalysed cleavage of a linker the benefit is proportional to the time saved and rarely decisive on its own. The trade-offs of every route away from the acid are compared in how to elute from Protein A without acid.

Why does neutralisation speed matter as much as elution pH?

Because the damage is a product of how low the pH goes and how long it lasts, and the pool spends longer in acid than the column does. A pool that collects over several minutes and is neutralised at the end sits at elution pH the whole time, and it crosses its own isoelectric range on the way back — which is where precipitation happens. Neutralising each fraction as it lands addresses the time term without touching the elution chemistry at all.

Is Solaris® Protein A available for this today?

Not as a validated process option. Solaris® Protein A is in Beta, with access by application, and what has been demonstrated is the release mechanism with IgG — the sensitive formats this page is about are exactly the ones still to be run on it. That is what the Beta programme is for.