Perfusion-Based Production of rAAV via an Intensified Transient Transfection Process

Abstract

Increasing demand for recombinant adeno-associated virus (rAAV)-based gene therapies necessitates increased manufacturing production. Transient transfection of mammalian cells remains the most commonly used method to produce clinical-grade rAAVs due to its ease of implementation. However, transient transfection processes are often characterized by suboptimal yields and low fractions of full-to-total capsids, both of which contribute to the high cost of goods of many rAAV-based gene therapies. Our previously developed mechanistic model for rAAV2/5 production indicated that the inadequate capsid filling is due to a temporal misalignment between viral DNA replication and capsid synthesis within the cells and the repression of later phase capsid formation by Rep proteins. We experimentally validated this prediction and showed that performing multiple, time-separated doses of plasmid increases the production of rAAV. In this study, we use the insights generated by our mechanistic model to develop an intensified process for rAAV production that combines perfusion with high cell density re-transfection. We demonstrate that performing multiple, time-separated doses at high cell density boosts both cell-specific and volumetric productivity and improves plasmid utilization when compared to a single bolus at standard operating conditions. Our results establish a new paradigm for continuously manufacturing rAAV via transient transfection that improves productivity and reduces manufacturing costs.

Keywords: adeno‐associated virus; continuous manufacturing; gene therapy; mechanistic modeling; transfection.

Figure 1

Model‐based re‐transfection. (A) Schematic of perfusion bioreactor system. The cell culture fluid is continuously circulated through the tangential flow filtration system, where full and empty capsids are removed via a harvest stream. A cell‐ and capsid‐containing bleed stream is periodically discharged to modulate the cell culture density. A feed flow consisting of fresh cell culture media was controlled to maintain a constant cell culture volume. Plasmids encoding for rAAV were dosed on days 0, 5, 7, and 9. (B) Model predictions of the total extracellular plasmid concentration (combined pAAV‐GFP, pRC5, and pHelper). Plasmids were added to the perfusion cell culture when the predicted total extracellular plasmid concentrations approached zero. (C) Model predictions of the concentration of total plasmids within cellular nuclei (blue, left axis) and experimental measurements of the percentage of cells expressing GFP (orange, right axis). In both plots the vertical dotted lines represent re‐transfection events. Error bars represent measurement SD.

Figure 2

Mechanistic model predicts experimental dynamics. (A) Total cell density and (B) viability experimental measurements (markers) and model predictions (solid lines). Error bars represent measurement SD. (C) Total capsid and (D) vector genome experimental measurements (markers) and model predictions (solid lines). Total capsids were measured using ELISA and vector genomes (vg) were measured using ddPCR. Error bars represent measurement SD. (E) Relative viral protein (VP) synthesis rate per cell (Equation 15) and capsid filling rate per cell (Equation 18) as predicted by the kinetic model. Rates are relative to the relevant maximum rate predictions from the mechanistic model simulation. (F) Relative concentrations of Rep proteins per cell and empty capsids in the nucleus per cell as predicted by the kinetic model. Concentrations are relative to the relevant maximum concentrations predictions from the mechanistic model simulation. In all plots the vertical dotted lines represent re‐transfection events.

Figure 3

Extended production of total and full capsids. (A) The total capsid titer in the bioreactor vessel is plotted with the cumulative total capsid titer harvested in the cell bleed and harvest streams. (B) The vector genome titer in the bioreactor vessel is plotted with the cumulative vector genome titer harvested in the cell bleed and harvest streams. In both plots the black line represents the sum of all three product sources (vessel, cell bleed, harvest) and the vertical dotted lines represent re‐transfection events.

Figure 4

Total and full capsid yields for each transfection event. The yields from each transfection were calculated as described in Materials and Methods section 4.13. A transfection event starts with the addition of the first plasmid bolus (as specified in Table 1) and concludes either when the next transfection event begins or when the culture ends. (A) Net total capsid titer (left axis, black) and full capsid titer (right axis, orange) produced during each transfection event. (B) The total capsids produced per unit weight of plasmid added (left axis, black) and full capsids produced per unit weight of plasmid added (right axis, orange) during the transfection events. Cellular specific productivity of (C) total capsid and (D) full capsid. The specific productivity was calculated using the amount of full capsids produced by the cell culture and the integrated viable cell density (IVCD) as outlined in the Materials and Methods section 4.14. Error bars represent the forward propagation of the measurement SD.

Figure 5

Simulated full capsid yields for the deconvoluted transfection days. The yields of each transfection were calculated using simulations of the fit mechanistic model and the input conditions specific to each transfection event. The length of each simulation was four days. Intracellular concentrations were set to zero at the start of each simulation. (A) Net full capsid titer. The relative full capsids (left axis, grey) and the relative full capsids produced per unit weight of plasmid added during the transfection (right axis, orange). (B) Relative cellular specific productivity. The specific productivity was calculated using simulation predictions for the amount of full capsids and integrated viable cell density (IVCD). Error bars represent the 95% confidence interval from 10,000 Monte Carlo simulations. (C) Cell‐specific productivity accounting for only transfected cells as measured by green fluorescent protein (GFP). Cell‐specific productivity calculation details can be found in Materials and Methods section 4.14.