October 28, 2019 PAP-Q3-19-CL-006
With the ability to treat and the potential to cure diseases for which no treatments are currently available, cell and gene therapies continue to attract significant attention. According to the Alliance for Regenerative Medicine, there were over 917 regenerative medicines in development worldwide by the end of Q1 2019, including gene, cell and tissue-engineering therapies.1
Of the 1,060 clinical trials in regenerative medicine underway worldwide by the end of Q1 2019, 372 involved gene therapies, with 123 in phase I, 217 in phase II and 32 in phase III. In that quarter alone, $2.1 billion was raised to advance gene and gene-modified cell therapies, most of which utilize viral vectors for gene transfer.1 A few cell and gene therapy products have already been commercialized, with four approved by the U.S. Food and Drug Administration since 2017, including the two AAV products cited above. Regulatory authorities in China, Australia, New Zealand, South Korea, India, Japan, Europe and Canada have also approved gene therapy products.2 It is predicted that as many as 40 cell and gene therapies will be on the market by 2022.3
Viral vectors used in cell and gene therapies include adenovirus (Ad), lentivirus (LV), retrovirus (RV), herpesvirus (HSV) and adeno-associated virus (AAV). Recombinant AAV (rAAV) vectors are attractive for the development of gene therapies due to their efficacy, long-term gene expression and safety.4 They were first discovered in 1965 as a contaminant of adenovirus stocks and were thus given the name “adeno-associated virus.” AAVs are among the smallest viruses and are non-pathogenic members of the Dependovirus genus of parvoviruses. More than 12 AAV serotypes have been identified, along with hundreds of naturally occurring capsid variants.4 The AAV gene therapy products that are in clinical evaluation target a variety of diseases affecting multiple organ types in varying patient populations.
The first approved in vivo gene therapy products in Western markets were AAV vectors. These include Glybera (alipogene tiparvovec, (approved by the European Medicines Agency (EMA) in 2012) and the FDA-approved drugs Luxturna® (voretigene neparvovec, 2017) and Zolgensma® (onasemnogene abeparvovec-xioi, 2019). In addition, all viral vector therapeutics that received Priority Medicine status from the EMA in 2017 were rAAV vectors.4,5
Initial laboratory procedures for the production of AAV viral vectors involved the use of adherent cells in plasticware – an approach with limited scalability. Today, however, the large-scale production of AAV vectors using suspension cell culture in conventional bioreactors has been successfully demonstrated. The two predominant methods include transient transfection of suspension-adapted mammalian HEK293 cells6–8 and the infection of Spodoptera frugiperda (Sf9) insect cells using the baculovirus expression vector system (BEVS).4 rAAV vectors produced in both systems have been shown to have comparable biological activity.4 Production platforms, such as Pall’s iCellis large-scale adherent cell culture platform, are also gaining wider use.9
Each of these methods have been used to produce rAAVs that are being evaluated in late human clinical trials and are on the precipice of commercialization, with several more products in preclinical and early clinical development. The manufacturing processes are similar in that both start with thawed cells that are expanded through a seed train to a terminal bioreactor. In the differentiating step, HEK293 cells are transiently transfected with plasmid DNA, while Sf9 cells are infected with baculoviruses. After approximately three days, the rAAV virions are harvested, purified, formulated and filled into vials.
There are additional differences between the processes. For transient transfection, a master and working cell bank for HEK293 cells is required. It is also necessary to produce large amounts of plasmid DNA and to use chemical agents that mediate transfection, both of which are raw materials required in sufficient supply for each batch.
In the transfection step, multiple DNA plasmids (helper and transgene-containing) are mixed with the transfection agent to enable delivery of the DNA across the cell membrane and to the nucleus of the producer HEK293 cells. The fluid dynamics of this mixing process must be carefully controlled to prevent the formation of plasmid–reagent complexes that are too large or too small and to ensure an even distribution of the complex into the cell suspension. These requirements can create challenges in obtaining the batch scale at which these processes are required to achieve production demand.
Additionally, it is possible that some transgenes carried by the therapeutic vector may not be compatible with the mammalian manufacturing system. Expression of these genes during production, which is driven by the same promoters that are designed to function in humans during gene therapy, may negatively impact the performance of the culture system. Gene-suppression systems to regulate expression of the therapeutic gene during production are being developed to overcome these challenges.
In contrast to the transfection process, the infection of Sf9 cells does not require carefully controlled mixing times. In addition, any potentially negative impacts to the culture system, due to the expression of the therapeutic transgene, are minimized since the typical mammalian promoters used in vectors have little or no expression in insect cells. During production, a small quantity of the baculoviruses harboring the necessary genes to generate the recombinant AAV vector are infused into the terminal bioreactor to infect the Sf9 cell culture.
For the Sf9–baculovirus production system, the engineering of two recombinant baculoviruses (rBV) must be performed. One rBV contains the AAV helper genes (rep and cap), while the second rBV contains the genome of the AAV therapeutic vector. Master and working banks must be made for these viruses, as well as master and working cell banks for the Sf9 cells. Therefore, this system requires a significant upfront investment in time (several months) and resources needed to clone and bank each of these components.
The average production yield from the Sf9–baculovirus platform is approximately 10-fold higher per cell as compared with transient transfection of HEK293 cells. With the Sf9–baculovirus system, the significant upfront investment in time and resources needed to clone and bank each of the recombinant baculoviruses and Sf9 cells pays off with the ability to rapidly manufacture very large batches of rAAV. In both systems, a significant portion of the capsids do not contain the vectors (i.e., empty capsids), and the full capsids can be enriched in subsequent purification steps.
The final AAV vectors produced using HEK293 and Sf9 cells differ with respect to their impurity profiles. With the HEK293 platform, potential impurities include specific host-cell DNA/proteins and residual plasmid used to generate the vector. For the baculovirus system, the potential impurities include Sf9 host DNA/proteins and baculovirus DNA/proteins. A potential risk associated with the mammalian system is the encapsidation of host-cell genetic material, which may contain oncogenes harbored in the HEK293 cells, but there may be similar risks with packaging Sf9 host-cell DNA. As is typically the case in manufacturing of biologic drug substances well-established downstream purification methods have been shown to remove residual host-cell proteins and DNA that may be present in intermediates produced. Because the HEK293 and Sf9 feed streams are slightly different, the specific purification steps taken for the two platforms may also differ accordingly. To monitor these risks (as with any biologically produced product), tests are utilized to determine whether residual DNA/protein is present in the AAV product.
AAV vectors are typically purified using affinity chromatography and end with formulation into a desired buffer, filtration and filling into vials. The specific steps taken to optimize the downstream purification process for any given AAV vector will correlate directly with the specific impurity profile and level of empty capsids in the feed stream. These are addressed with the incorporation of additional chromatography and filtration steps. For each platform, differences in titer will determine scaling parameters and process steps needed during chromatography and filtration. As indicated earlier, common to both platforms is the presence of empty capsids in the feed stream. The degree of optimization taken to limit the quantity of empty particles in the upstream production process will impact the degree of optimization required downstream to remove the empty particles. Purification modalities (chromatography or ultracentrifugation) can be implemented to remove empty capsids irrespective of the production system.
Many of the methods used to characterize AAV vectors are utilized independent of the production process and are intended to evaluate both product quality and attributes. They include standard enzyme-linked immunosorbent assay (ELISA) and quantitative polymerase chain reaction (qPCR) techniques that are widely applied in the biopharmaceutical industry. Digital droplet PCR (ddPCR) for absolute rather than relative quantification of target DNA molecules is now becoming the industry standard for vector genome and residual DNA titration.10
Vector purity is determined by analyzing samples for process-related impurities (e.g., host-cell DNA and proteins) and for product-related impurities (e.g., the ratio of full and empty capsids). Analytical ultracentrifugation (AUC), capillary electrophoresis (CE) and HPLC-based methods are powerful techniques for determining the purity of AAV products. Thermo Fisher’s Viral Vector Services business has developed robust analytical technique which was developed in house, for the characterization and composition of empty, partial and full AAV capsids.
Identity testing is conducted on the vector in order to confirm the proper identity of the capsid proteins; this is typically performed using specific antibodies or liquid chromatography – mass spectrometry. Vector genome identity is also performed to confirm the integrity of the therapeutic transgene sequence contained within the vector; this is typically performed using next-generation sequencing (NGS) or Sanger sequencing. The formulated product is also subjected to osmolality, pH and other analytical chemical testing to confirm that the vector was formulated correctly.
Product strength e.g., genome titer (number of vector genome copies per milliliter) is universally used to determine dosing; infectivity and potency (expression of the transgene and therapeutic effect) are also evaluated. Safety tests are performed to confirm that impurities or contaminants like replication-competent AAV (rcAAV), adventitious agents, endotoxin and microbes are below specified levels.
The need to develop more sensitive, accurate, robust assays constantly pushes the industry to search for new technologies, while at the same time adopting proven analytical methods used in the biologics industry.
The choice to use the mammalian or insect production platform is contingent on several factors. Overall, the batch size and frequency of AAV production will be dictated by the drug dose, the size of the patient population and the expected market penetration for the product. Some gene therapies are given as ~100 µL doses, such as those administered via subretinal injection. Others are delivered intravenously at doses of ~100 mL. Across the dosing range, the patient population may be very small or very large. Consequently, a commercial process could be viably supported either by a 20- to 200-L bioreactor scale or may require large-scale manufacturing in a 2000-L bioreactor.
The funding paradigm for the gene therapy developer will also have an impact on the bioprocessing platform decision. Many startups supported by venture capital funding receive money from these firms in increments as the project progresses. Milestone payments are typically made when toxicology data is obtained, after phase I trials are completed and safety in humans is confirmed, and as the gene therapy meets efficacy endpoints in later clinical trials. This model encourages minimizing time to the clinic, and the fastest way to achieve proof-of-concept data in humans is typically for companies to employ transient transfection in the adherent mammalian platform. However, the adherent platform has inherent challenges in scale-up, which must be overcome in order to produce consistent, large-scale batches required for commercial production.
On the other hand, there is a new accelerated approval paradigm at the FDA through which consideration for a commercial license can be achieved in an approximate four-year time frame (rather than the typical 10–11 years) if the CMC section is sufficiently robust and a well-designed early clinical study has demonstrated efficacy.11 For companies that wish to leverage this new paradigm, it is essential to invest in the development of a commercial-ready manufacturing platform and to generate robust data for the CMC package up front and to leverage a process that will enable rapid scale-up to commercial production.
The large-scale production of AAV vectors using both mammalian HEK293 and Sf9–baculovirus suspension cell cultures has been successfully demonstrated and will likely continue to gain in favor owing to their scalability and ability to produce large, commercial batches of drug product.
Thermo Fisher Scientific offers many of the key reagents needed to support both baculovirus and mammalian suspension processes, as well as adherent mammalian processes for the manufacture of AAV vectors. These reagents are available through the catalog to researchers and in larger quantities for manufacturing.
With the acquisition of Brammer Bio, Thermo Fisher’s Pharma Services business has added a complementary capability servicing the growing demand for viral vector manufacturing. Customers today not only have access to important research reagents for their early work, but the support of a team with decades of experience in GMP viral vector manufacturing that can help them successfully move their projects from development to the clinic and on to commercialization.
Dr. Snyder was the founder of Florida Biologix, which was spun out of the University of Florida in 2015 and merged to create Brammer Bio in 2016. Dr. Snyder has been investigating virus biology, vector development, cGMP manufacturing and analytical technologies, and viral vector–mediated gene transfer for over 32 years. Dr. Snyder received his doctoral degree in microbiology from the State University of New York at Stony Brook and obtained his BA in biology from Washington University in St. Louis.