Ask the Expert: Clarification of Lentiviral Vectors Using Scalable Filtration

Researchers traditionally have produced lentivirus (LV) in adherent cultures by transient transfection using media containing animal serum. The method is inexpensive for process development (PD) and early clinical trials, but it requires increased operator manipulations and costs during scale-up. Development scientist Adam McLeod delivered an Ask the Expert webinar on 27 August 2019 to illustrate how GE Healthcare Life Sciences at the Centre for Advanced Therapeutic Cell Technologies (CATCT) in Toronto, Canada, approaches lentiviral vector (LVV) manufacturing. McLeod explained that the team induced a LV producer cell line and then recovered LVVs by filtration. Upfront costs increase with that strategy, but McLeod says that the changes greatly simplify scale-up.

McLeod’s Presentation
For transient transfection processes, the goal is to scale cell culture development to manufacturing levels as quickly as possible. However, to reduce cost at larger scales and to decrease batch-to-batch variability, a LV producer cell line is recommended. The CATCT team uses a producer cell line and small-molecule inducers to generate LVVs in suspension cultures with animal-derived-component–free media and reagents. That approach enabled production of 28 L in a closed, single-use Xcellerex XDR-50 bioreactor.

CATCT has found that having LV producer lines simplifies and improves the cost-effectiveness of future processes. Clarification becomes the real challenge. Especially for batch production with suspension-adapted cells, debris must be removed to prevent filter clogging and maximize target recovery. Researchers have traditionally relied on centrifugation, but it is costly, labor-intensive, difficult to scale up, and not always aseptic.

The self-contained nature of single-use filtration (SUF) offers a more direct path from PD scale to manufacturing-level LVV recovery. It decreases downstream costs and contamination risks because it enables customized clarification setups comprising inexpensive, aseptic disposables. More important, SUF scales up easily. Increasing filter surface area and flow rate can eliminate producer cells and debris at large scales with minimal damage to vector infectivity.

In-house infectious titer (IT) analytics suggest that the above strategy is promising. The team completed 16 LV clarifications, 10 with filtrate volumes between 700 mL and 1.1 L, and six between 1.8 L and 2.1 L. They used an ULTA filter train with the prefilter rated at 5 μm, the intermediate at 0.6 μm (both glass fiber), and the microfilter at 0.2 μm nominal (polyethersulfone, PES). An automated cell counting device and microscopy both showed removal of all cell producers and debris. Equally important, LVV titers were high, and the virus lost no infectivity. At 0.7–1.1 L scale, IT was recovered at 110.7% ± 11.2% (n = 10); 1.8–2.1 L samples showed IT of 101.0% ± 6.1% (n = 6). In aggregate, IT was recovered at 107.1% ± 10.7% (n = 16).

Those strong results might be attributed to how the equipment was prepared. Presoaking filters with HyCell TransFX-H media helped to achieve >60% IT recovery. One hypothesis is that the media coated the PES membrane and fibers in ways that prevented viral vector sticking.

The CATCT team soon hopes to realize 30% LVV recovery over our entire downstream process, vector concentrations of 1 x 108 TU/mL, ≥2 log removal of host cell proteins and DNA, and process times under eight hours. Those goals seem feasible because SUF simplifies downstream processing.

Questions and Answers
Have you tested SUF using adeno-associated viruses (AAVs)? We have not, but such a process should work with small modifications. E.g., you might need to adjust the ratio of the prefilter to the microfilter because AAV processes contain more cell debris and fewer intact cells. Something similar can be done in terms of membranes.

What cell culture concentrations were achieved in these runs? We harvest 4–6 million cells per mL, which is small compared with a monoclonal antibody (MAb) process. But LVVs cytotoxic, so 4–6 million is the high end of what is possible. If a client has a higher cell concentration than that, we can scale the filters to match.

Where are you losing most of the recovery? We initially optimized our unit operations in isolation, but we highly discourage that. Even small changes in upstream materials and operations manifest in the purification train. Due to a change in upstream, we needed to go back to reoptimize clarification. With that process completed, we now will focus on our subsequent steps.

Do you use polyethersulfone (PES) for the final filter, and if so, how do you prevent binding? We use PES filters for clarification and sterile filtration. We have not studied this in detail but believe that flushing or soaking filters before a process prevents viruses from interacting with membrane surfaces — and that seems to be the key to the whole filter train.