Figure 4. Protein A load volume and titer for the duration of the run. Black circles and triangles represent the start of the cycles for column 1 and 2, respectively.
Chromatogram overlays of the 35 cycles performed on each of the two capture columns were generated for analysis (data not shown). Constant peak height and position are observed across all cycles. Tailing was noticed on some cycles, especially towards the end of the run. Analysis of conductivity traces showed a change in transition curve profile for these cycles which indicates a change in bed structure rather than a change in impurity profile or issue with the UV meter. The root cause was likely air on the columns. This can easily be mitigated for future processes by installing a bubble trap prior to the protein A step, which was not in place during this run.

Virus Inactivation

The continuous low pH virus inactivation step was successfully controlled to meet pH and residence time specifications. The virus inactivation step was performed in 3 parts by starting and pausing this step 3 times during the 14-day process. This pausing was required to account for the change in rate of product elutions from the upstream protein A step and maintain a constant residence time across the virus inactivation column. The virus inactivation step was triggered to automatically start when the protein A product vessel of 10L filled to a high target volume. During the run, the virus inactivation step drained the vessel until it reached a low target volume, at which point the virus inactivation step automatically paused until the high target level was reached again. Each of the three virus inactivation sub-runs lasted approximately 1 day.
The virus inactivation step used an SEC column to provide the required inactivation residence time. The SEC column models a plug flow reactor and this choice benefits from the abundance of industry knowledge around column packing and qualification. Additionally, residence time distribution is independent of flow rate in a packed bed with no liquid-solid mass transfer, unlike tubular flow reactor designs which rely on dean vortices for mixing (Amarikwa, 2019). Other options, such as alternating tank VI were not considered due to the increased space, equipment, and automation requirements.
The pH of the capture product was titrated inline down to a pH of 3.5 with 1 M of acetic acid, before passing through the SEC column. After exiting the SEC column, pH was titrated inline up to a target pH of 7.2 with 1M tris base. Inline pH adjustments were performed using feedback control. The pH was well controlled for and was mostly maintained within the specified ranges of ±0.1 and ±0.2 for the acidification and neutralization, respectively. Any disturbances outside of the specified ranges were diverted to waste until pH returned within spec and remained stable for 2 minutes. pH data can be seen in Figure 5.