Finding the Weakest Caustic That Cleans a Column Using a Sodium Hydroxide Gradient to Set the Regeneration Dose

This application is about cleaning the resin in a hydrophobic interaction chromatography column, a purification step used when making antibody drugs. The title and abstract describe regenerating such a column with an alkaline solution in a stated pH and conductivity range, but claim 1 is narrower and different: it is a method of finding the right strength of alkaline solution by running a ramp of sodium hydroxide up to 1N and back down, and spotting where the stuck material comes off. Regeneron Pharmaceuticals, Inc. filed it as a divisional of Australian application 2020352547 and names Robert Stairs, James Reilly, John Mattila and Samantha Wadsworth as inventors.

The Problem

Hydrophobic interaction chromatography separates proteins by how water-repelling their surfaces are. A load with a lot of salt is run onto the resin, which encourages the protein to stick, and then solutions with less salt release it. In biologic drug manufacture it is used to strip out host cell proteins, lipases and other contaminants that would otherwise travel with the product.

The resin is expensive and is reused for many cycles. Each cycle leaves something behind, and the specification lists host cell proteins, aggregated proteins, lipids, polypeptide fragments and nucleic acids. If it is not cleaned off, it can contaminate the next batch and shorten the life of the media. A lipase left behind can attack the polysorbate 20 in the final formulation, and the first example uses free fatty acids from that breakdown as an alarm signal.

The conventional answer in the specification is a multi-step strip, such as pure water, then 1N sodium hydroxide, then water again, then 20 per cent ethanol, with guanidine hydrochloride at 6N as a harsh final check. That works but is slow, uses strong reagents and does not tell you whether a gentler solution would do.

What This Invention Does

Claim 1 is a method of identifying a concentration of an alkaline solution for regenerating a hydrophobic interaction column that has had a load applied. The steps are to pass a first solution of water and alkali whose concentration rises from about 0N to a maximum of 1N, then pass a second solution that falls from that maximum back to about 0N, and to identify the portion of either solution that removes material bound to the media. In other words, the column is the instrument. The absorbance trace shows exactly where, along the gradient, the residue elutes.

The other independent claim, claim 16, specifies sodium hydroxide, a first solution with total dissolved salt of 50 mM or less at its maximum concentration, and identification of the sodium hydroxide concentration that removes bound material. Claim 14 then regenerates a column using a concentration found by the claim 1 method. The dependent claims set the likely answer: pH of about 10 to 14, conductivity of about 0.5 to 10 mS/cm, 1 to 20 column volumes, phenyl, butyl or octyl ligands at about 20 to 30 micromole per millilitre of media, and less than about 1.0 per cent of the load mass left behind.

The examples show what the gradient finds. In a run with used Capto Phenyl (High Sub) media, a single peak of residue came off when about 5 mM sodium hydroxide had passed through the column, and a second gradient produced no further peaks. The area under that peak was 1,160 L*mAU, against 13,305 L*mAU for a 6N guanidine strip in a control procedure, which the specification reports as a 91.3 per cent reduction. A set of single-step runs at 1000, 500, 100, 50, 25, 10 and 5 mM sodium hydroxide, each followed by 6N guanidine, found that 5 mM removed the most material and left the least for the guanidine to remove. The specification’s explanation is that higher pH drives residue off while higher conductivity strengthens its grip on the media.

Salt is not a substitute. In Example 14 a 3 mM sodium hydroxide solution at pH 11.49 and 0.74 mS/cm gave a regeneration peak area of 29,671, and left 501 for the guanidine to remove. Sodium chloride solutions of 8 mM and 11 mM gave regeneration peaks of 146 and 226, and left 34,114 and 32,852 for guanidine. The chromatograms were also similar across six different monoclonal antibodies at differing pH and citrate loads, which the specification takes to mean the effect is not tied to one product.

Key Features

• The column as its own sensor. A ramp of sodium hydroxide up and down, read on the absorbance trace, shows the concentration at which bound material leaves.
• A low-salt limit. Claim 16 requires total dissolved salt of 50 mM or less at the top of the ramp, and claim 6 puts conductivity at 0.5 to 10 mS/cm.
• Wide choice of alkali. Claim 2 lists sodium, potassium, calcium and magnesium hydroxide, or tris.
• A clean-up target. After regeneration, less than about 1.0 per cent of the load mass is left bound, per claim 4 and claim 22.
• Named product context. Several dependent claims name dupilumab, or an anti-interleukin 4 receptor antibody, as the load or target molecule.
• Storage step. Claim 15 follows regeneration with storage in sodium hydroxide at about 0.05M to 0.15M. As printed, claim 15 refers to “the method of claim 10”, which is a claim about dupilumab loads and looks like a drafting slip for claim 14.

Who Is Behind It

The applicant is Regeneron Pharmaceuticals, Inc., a New York biotechnology company. Its own site at regeneron.com describes it as a science-driven medicine company and cites 16 approved medicines developed in-house. The Wikipedia article on Regeneron records its founding in 1988 by Leonard Schleifer and George Yancopoulos and its base at Tarrytown, New York. Dupilumab, the antibody named in the claims, is sold as Dupixent and was developed jointly with Sanofi.

On the family position, the specification states that it is a divisional of Australian application 2020352547, and that it claims priority to US provisional applications 62/905,033, filed 24 September 2019, and 62/958,899, filed 9 January 2020. The page headers show it as the Australian entry of PCT/US2020/052243, published as WO 2021/061790. The priority country is the United States.

Why It Matters

Purification columns are a bottleneck in making monoclonal antibodies. The media is costly, batch records require evidence that one cycle does not carry into the next, and each regeneration step has to be validated. A method that turns the choice of cleaning solution into a measurement, rather than a copy of what the last process used, shortens that development. The specification itself ties this to a plate-based screening approach and reports 98 per cent agreement between small-scale well results and large columns.

The results also point towards gentler chemistry. In the single-step test, a few millimolar of sodium hydroxide, with low conductivity, removed more bound material than any of the stronger solutions up to 1000 mM, which points to less caustic handling and possibly less wear on the resin. The specification is cautious on one point worth noting: it reports that 6N guanidine added little after a 5 mM sodium hydroxide solution, but does not say how many cycles of reuse the resin tolerates beyond the 49-cycle run it describes.

Related Concepts

• Hydrophobic interaction chromatography – the separation method whose resin is being regenerated.
• Monoclonal antibody – the class of biologic drug whose purification uses these columns.
• Dupilumab – the anti-interleukin 4 receptor antibody named in the dependent claims.
• Sodium hydroxide – the alkali whose concentration is scanned in the claimed method.
• Guanidinium chloride – the strong denaturant used as the benchmark strip to see what regeneration missed.

---

AU 2026202255 was published in the Australian Official Journal of Patents on 9 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.