Bai-wei Gu, Juan Lagos, and Matthew Weaver (heads of cell line development, upstream process development, and downstream process development groups, respectively, at WuXi Advanced Therapies, ATU) joined forces on 29 October 2019 to feature their company’s viral-vector manufacturing capabilities for cell and gene therapies. In addition to adherent platforms for lentivirus (LV) and adenoassociated virus (AAV) vectors, ATU soon will offer suspension-cultured viral vector platforms for them as well as analytical measures that support release testing. Transitioning from adherent to suspension processes required up- and downstream adaptation. Gu, Lagos, and Weaver shared what they learned about process development (PD) during that transition and detailed what challenges remain before their platforms can launch later in 2020.
The Presentation
Current Capabilities: ATU’s work with suspension PD stems from success with its adherent viral vector platforms for AAV and LV. The former includes a conventional upstream process with 22–25 days between thaw and harvest and a “hybrid†downstream approach. Because AAVs typically are used in vivo, harvests for that platform usually undergo an initial multicolumn-chromatography step (optimized by serotype), followed by tangential-flow filtration (TFF).
The company’s LV platform offers additional opportunities to enhance project success. It can run at verification, pilot (optional), and good manufacturing practice (GMP) scales, and it generates appropriate release- testing certification. The platform also offers strong culture and recovery capabilities. Bulk and clarified harvests average 1.03 × 10ⶠand 1.02 × 10â¶, respectively, in 36-layer hyperstacks. Unlike its AAV counterpart, ATU’s LV platform does not require initial chromatography, so it offers straightforward but highly effective downstream processing. Operators aim to produce concentrations 200× larger than those of the bulk harvest. In 36-plate hyperstacks, infectious titers usually range between 2.4 × 10⸠and 2.0 × 1010. Overall yields average 50%, varying from 35 to 65% depending on gene of interest (GoI) and plasmid construct. Although ATU has optimized its proprietary cell lines, it also allows sponsors to transfer their own plasmids in during PD. Thus, the company can move sponsors from feasibility study through GMP manufacturing efficiently depending on their process needs.
Upstream Prospects and Challenges: ATU is building on current capabilities by developing four new producer cell lines for LVs and AAVs (two each). Baseline data confirm the cell lines’ efficiency. New analytical platforms also have been established to support release-testing requirements. The plan now is to ensure strong scalability at 1,000 L and above. PD toward that goal has been illuminating because it has revealed key challenges that will help ATU — and the industry — optimize suspension operations.
Cell aggregation was particularly intractable during adaptation to suspension culture of LVs for transfection into human embryonic kidney (HEK293) cells. Proof of concept for ATU’s LV platforms was conducted initially in shake flasks and a 2-L bench bioreactor, then in a 50-L single-use bioreactor. Harvests from both reactors exhibited similar cell growth and density rates until transfection (three days after inoculation). Two days after that, viable cell density remained steady at both scales, but aggregation significantly increased in the 50-L bioreactor. Samples from that vessel also exhibited half the infectious productivity of those from the flasks and 2-L device.
It should be noted that the proof-of-concept team did not add anticlumping agents to the 50-L batch. Combining those with shear-protection agents and adequate agitation could reduce cell aggregation in future processes. The same goes for additives that facilitate DNA transcription and inactivate cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP) — an approach that the proof-of-concept team found promising.
Careful study of bioreactor conditions helped to curb aggregation in later assessments. Bioreactors differ in geometry, working volume, and material construction — which all can converge to promote aggregation and/or sheer. Thus, it was critical to compare power-to-volume ratios (p/V), impeller designs, mixing speeds, and gas-exchange strategies. Computational fluid dynamics (CFD) can be helpful in this respect, presenters noted, but its technological complexity currently limits its widescale use in PD.
Given those factors, ATU teams designed a suspension process that transitions incubated material to 48 parallel minibioreactors and culminates in bench- (2-L) and manufacturing-scale (50- and 200-L) reactors. Presenters stressed that a minibioreactor strategy facilitates scalability because those units are designed to approximate mixing conditions in larger bioreactors. Using that LV process achieved cell viability rates >90% and cell densities between 4.5 and 5.9 million/mL. It also became clear that aggregation and glucose production related inversely to agitation: Strong agitation promoted density and diminished glucose.
Chromatographs from similar tests of the two AAV suspension platforms revealed similarly strong results, although viral serotype and GoI significantly dictated expression levels and viral productivity. Thus, process sponsors are advised to request preliminary testing of multiple vector designs — including chromatography for viral particle detection and analytical ultracentrifugation for viral productivity screening. ATU now is focused on achieving corresponding outcomes at and above 1,000 L.
Downstream PD: Scalability is the goal for downstream, too. To that end, downstream specialists modified their adherent platforms’ clarification steps to accommodate large-scale manufacture, considering four basic questions: Which process maximizes yield? Which minimizes turbidity? Which has the highest maximum capacity? Which is best operationally?
The downstream team performed clarification screening studies to determine whether filters from adherent-cell–based processes translated well for harvests from suspension cultures. ATU tested five filters (including an adherent-platform filter), with capacities ranging from 47 to 117.5 L/m². ATU’s preferred filter (47-L/m² capacity) generated yields of 90–95% with turbidities of 7.1 NTU. It also has proven performance in batches up to 180 L. Two filters — a 114.5-L/m² stand-alone and 117.5-L/m² two-filter train — offered better capacity than the preferred filter and equivalent yields (90–100%). Both also exhibited low turbidity (1.31 and 2.72 NTU). But they required considerable processing times (100 min). The train also needed more operator intervention than comparators. The team decided against the one-filter train because of the large-scale equipment complexity and determined to use the 47-L/m² adherent-process filter train or the 114.5-L/m² single filter depending on sponsor needs.
PD revealed that concentration strategies matter just as much as filter specifications. Conventional TFF calls for constant feed flux, adjustment of transmembrane pressure (TMP) for established critical process parameters, and concentration of drug product. That works for adherent-cell platforms and at small scales but complicates suspension purification, especially at manufacturing levels. Too much concentration can disrupt process parameters and diminish drug-substance quality. Long TTF steps also require more personnel and increase setup and breakdown times. Thus, it is important to balance maximum concentration and purity.
ATU scientists are accomplishing that by reversing the conventional strategy: adjusting feed flux while keeping TMPs and feed–retentate stream differences (∆P) steady. Initial testing of AAV concentration using 0.2-m² membranes and 1 L of material demonstrated that a TMP-focused strategy can keep process times low (89 min) while still preserving product quality. The downstream team will continue to investigate this strategy with an eye to scalability and automation.
New Platforms on the Horizon: Having gathered 50-L baseline data for its LV and AAV suspension platforms, ATU is augmenting its processes, developing master cell banks (MCBs), and validating new equipment for GMP compliance. PD teams are engineering producer cell lines with proprietary plasmids, and baculovirus platforms are under investigation. ATU hopes to unveil those developments as well as a modified platform for quality control later in 2020. These innovations can help sponsors move efficiently through all aspects of viral-vector manufacture.
Questions and Answers
What are key advantages of moving from adherent to suspension culture? Adherent cell culture requires significant manual intervention. So it is slower and more prone to operational error than are self-contained suspension cultures. Suspension cultures also use chemically defined media to eliminate process variability stemming from animal-derived components. But most important is that suspension culture enables strong control over dissolved oxygen, pH, and other parameters. That brings process developers close to the culturing environment.
What are some challenges with developing a suspension platform? Suspension platforms complicate cell-line development. HEK293 cells are prone to aggregation and require strict culture and clone-selection processes. But such hurdles are easy to negotiate.
Yield is the concern downstream. A purification process using the same vector serotype can produce vastly different results depending on the GoI. Even with well-understood vectors, sponsors should perform development screening for factors such as specificity to anticipate chemical, logistical, and financial challenges with GoI isolation.
Why should downstream operators control TMP and ∆P as opposed to feed flow? Conventional TFF limits how much drug substance can be concentrated because, at a certain point, TMP cannot be kept within acceptable ranges. Controlling TMP and ∆P across the membrane enables shorter and stronger concentrations than typical filtration. It also might allow intensification of the process and completion of the first three downstream processing steps on the same day.
Are PD activities preestablished? PD should be a “toolbox†for manufacturing success rather than a one-size-fits-all approach. For example, downstream engineers know what conditions work well for vectors of given serotypes, but they cannot be sure about process challenges without more information. Initial screening across an entire downstream platform can tailor PD to specific processing needs such that sponsors learn whether they can take calculated risks (e.g., with clarification). Thus, PD must be customized based on phase, indication, and process needs.
More Online
The full presentation of this webcast can be found on the BioProcess International website at the link below.