Adeno-associated virus perfusion enhanced expression: A commercially scalable, high titer, high quality producer cell line process

Abstract

With safety and efficacy demonstrated over hundreds of clinical trials in the last 30 years, along with at least six recent global marketing authorizations achieved since 2017, recombinant adeno-associated viruses (rAAVs) have been established as the leading therapeutic gene transfer vector for rare, monogenic diseases. Significant advances in manufacturing technology have been made in the last few decades to address challenges with GMP production of rAAV products, although yield, cost, scalability, and quality remain a challenge. With transient transfection processes established as a manufacturing platform for multiple commercial AAV products, there remains significant yield, cost, robustness, and scalability constraints that need to be resolved to enable a reliable supply of rAAV products for global patient access. The development of stable producer cell lines for rAAV products has enabled scalability and, in some cases, improvements in productivity. Herein we describe a novel AAV perfusion-enhanced expression (APEX) process, resulting in higher maximum cell densities in the production bioreactor with a 3- to 6-fold increase in volumetric productivity. This process has been successfully demonstrated across multiple serotypes in large scale cell culture with titers approaching 1 × 10¹² GC/mL. The APEX production platform marks a significant leap forward in the efficient and effective manufacturing of rAAV vector products.

Keywords: AAV; cell density effect; cost of production; high-yield; perfusion; process development; producer cell line; scalable.

Graphical abstract

Figure 1

AAV/helper virus Pinnacle PCL platform The Pinnacle PCL platform incorporates the AAV Rep/Cap and transgene, flanked by inverted terminal repeat (ITR) sequences stably integrated into the production cell genome. AAV production is performed in serum-free suspension culture through the addition of a helper virus (adenovirus) to initiate AAV replication in the bioreactor (only key genes for AAV production shown in the figure). Subsequent downstream purification works to remove impurities, cellular and helper-virus by-products, and concentrate purified AAV particles.

Figure 2

Cell density effect observed in rAAV production demonstrated at 1.3 × 10⁶ vc/mL(blue), 3.2 × 10⁶ vc/mL (red) and 5.2 × 10⁶ (green) vc/mL infection VCD (A) VCD time profile. (B) Glucose (dash) and lactate (solid) time profile. (C) Volumetric titer. (D) Cell qP. Data are presented as mean ± SD.

Figure 3

Nutrient deficiency is not the cause for cell density effect Normalized metabolic profile post infection are shown in (A). Batch process with infection VCD of 1.2 × 10⁶ vc/mL (blue), 1.8 × 10⁶ vc/mL (red), and 1.8 × 10⁶ vc/mL with glucose and glutamine daily feed (green) were compared in (B) glucose consumption rate, (C) glutamine consumption rate, and (D) Cell qP. Data are shown represent n = 2 (where relevant) and presented as mean ± SD.

Figure 4

Waste metabolites facilitate the cell density effect Impact of adding lactate (0–30 mM) and ammonium (0–3 mM) at the time of infection is shown in (A). Lower pH culture from 7.8 to 7.4 in cultures with infection VCD of 1.3 × 10⁶ vc/mL and 3.2 × 10⁶ vc/mL impacted the lactate accumulation (B), lactate cell specific production rate (C), and CSPR (D). We used 6 mM, 8 mM, and 10 mM glutamax instead of 6 mM glutamine and evaluated impact of ammonium accumulation (E), ammonium cell specific production rate(F) and cell qP (G). Data are presented as mean ± SD.

Figure 5

Perfusion process parameters were derived from historical batch process data (A) The percentage of extracellular rAAV trend with dead cell percentage, suggesting perfusion should stop at 48 hpi. A metabolic model derived from batch process data showed close prediction (dash line) to the experiment data (solid dot) for (B) VCD and (C) nutrient and metabolites prediction with 2.5 VVD perfusion rate. (D) The APEX process demonstrated consistent titer increase from the batch process with three capsid serotypes. Data are presented as mean ± SD. ∗p < 0.05.

Figure 6

Characterization and optimization of APEX process (A) Correlation between cell qP and CSPR at infection VCD of 3–5 × 10⁶ vc/mL. Cell qP of batch process is shown as the dashed line. (B). Full capsid percentage directly correlates with the CSPR. (C). Optimized APEX process parameters capable of supporting predicted rAAV titer up to 1.2 × 10¹² GC/mL. (D) rAAV titer for hu37-clone 1 in batch, APEX P1, and APEX P2 in 3L bioreactors. Data are presented as mean ± SD.

Figure 7

APEX process scalability up to 250L and comparison between batch and APEX processes (A–C) The VCD for batch process at 3L and APEX processes across AMBR250, 3L, and 250L, respectively, for clone 1 in APEX P1 (A), clone 2 in APEX P2 (B), and clone 3 in APEX P2 (C). The clone 1 and 2 are hu37 serotype PCLs and the clone 3 is a AAV8 serotype PCL. (D) Harvest titer. All titer data was measured by dPCR except that AAV8 in 3L Batch and Ambr250 was measured by qPCR. Data are presented as mean ± SD. The average titer for batch and APEX processes (across all scales) are labeled above the respective batches.