April 25, 2016

Principles of automated multistep protein purification

By Christine Markeland, Senior Research Engineer at GE Healthcare

During the webinar Automated two-step purification of antibody drug candidates, we collected a number of interesting questions on multistep processing. In this blog post, I have summarized questions related to principles of multistep processing and the case study presented by Erwin van Puijenbroek and Matthias Hermann, Scientists at F. Hoffman-La Roche.

Questions and answers

The first five questions are related to principles of multistep processing.

First, the initial step should be rapid and concentrating. Affinity chromatography or ion exchange is a great option. Secondly, the different steps should be orthogonal, for example, it is useful to combine ion exchange (separation by charge differences) with size exclusion chromatography (SEC) (separation by size differences). Thirdly, steps should be combined so that sample preparation between steps is minimized. For example, use a concentrating technique (like affinity chromatography or ion exchange) before size exclusion chromatography (which is limited by sample volume). For more info, see the Strategies for Protein Purification handbook.
As any other purification run, multistep purification requires method optimization. Investigate elution pattern for each step before combining them into the multistep run. Collect different peaks in different loops. Sometimes, all collected peaks need to be processed in the second step.
This is possible, but requires text instruction editing. Start peak detection watch after a certain volume (breakpoint), or set a watch for conductivity level and start a new watch for peak detection when target conductivity value is reached (watch in watch).
A watch command should be included in the UNICORN method to detect start and end of the peak. When a peak is detected, the valves will switch (after programmed delay volume) and direct the detected peak into the loop.
The system configuration presented at the webinar is for an automated, two-step purification. Because of the flexibility of ÄKTA pure, there are many other automation possibilities. You need to keep in mind that the amount of programming and the amount of testing/optimization increases for each step added. Full automation of the five steps in the purification scheme that you indicate (affinity chromatography – buffer exchange – anion exchange – buffer exchange – SEC) will require a considerable method development effort. It can be more efficient to reconsider the individual steps of the purification procedure, and/or to consider automation of some, but not all, of the chromatography steps.

The following questions are related to the case study presented during the webinar.

Yes, the autosampler program has an integrated wash step.
The optimized double injection method needs 13.3 min longer.
The multiple-column valve is needed to connect two columns, the protein A and the SEC columns.
In the setup used in the case study (100 µL protein A column at 12 s residence time and Superdex 200 Increase 10/300), the biggest limitation lies in the 500 µl collected from the protein A column. When higher amounts were eluted from the protein A step, we lost protein as part of the peak was not collected. A solution would be to find a SEC column, onto which larger volumes can be injected, while keeping a good resolution. Increasing the residence time to, for example, 30 s will also increase the yield but triple the run time of the protein A step and consequently reduce the number of samples that can be processed. You also have to keep in mind that one should not load more than 1 mg on the Superdex 200 Increase 10/300 column.
It is 2 ml. However, 1 ml would be enough, as only 500 µl is collected from the protein A column.
There is no neutralization of pH of the protein A peak. Instead, the collected peak is directly loaded onto the SEC column without neutralization. However, the buffer exchange over the SEC column constitutes a neutralization step.
The 100 µl column was packed in-house at F. Hoffman-La Roche.
A 4-fold injection would increase the run time by approximately 30 min and reduce the capacity of the autosampler to only 6 samples (6 × 4 vials). We would use another system for samples of this volume.
The concentration was determined by peak integration using UNICORN software.

If you missed it, you can view the webinar on demand.

In a separate post, I have summarized questions related to methods for multistep processing and the example methods.