Optimizing Capsid Separation with Anion Exchange Chromatography

Gene therapy and AAVs

More than 40 years after their first precarious iteration, gene therapies have begun to realize their potential. In fact, the majority of domestically marketed gene therapies have been approved by the Food and Drug Administration (FDA) within the last calendar year. Dozens more are in late-stage clinical trials and slated for imminent review, filling pharma pipelines for years to come. Moreover, with a genomic blueprint robustly mapped and open to modification, a technology initially based on ameliorating rare, monoallelic anomalies at a very small scale and with great inefficiencies has become much more refined.

In principle, gene therapy targets diseased cells with an engineered sequence that generates a functional protein to override a defective one. To accomplish this, an investigator packages DNA within a virus that can enter a target cell type and deposit its cargo, without infecting and replicating throughout the recipient. Several such viral vectors, including retroviruses and herpes simplex virus (HSV-1), can reliably serve specific contexts. Although retroviruses are used extensively in research and autologous cell therapies such as CAR-T, their role as gene therapy vectors per se has been hampered in part by their tendency to insert unpredictably into gene regulatory regions. On the other hand, HSV-1 is a lytic DNA virus that, correspondingly, can drive lesion-targeted effects such as tumor ablation and treating skin wounds.

What is needed most broadly, however, is a vector that can be predictably and reproducibly engineered, purified, and seamlessly scaled across a rapidly growing pipeline. For this purpose, biomedical professionals have settled firmly on recombinant adeno-associated viruses (AAVs). These replication-defective DNA viruses can impart long-term therapeutic expression without insertion, to a high degree of safety and consistency and a low degree of immunotoxicity. 

The importance of capsid separation

Given this consensus, one of the greatest challenges is to streamline and universalize the upstream and downstream components of the manufacturing process: that is, first producing AAVs, and then isolating and purifying the appropriate particles. Each process has critical process parameters that necessitate rigorous development and adherence to process definitions that match FDA guidance, ideally without having to redefine the process for every new serotype or therapeutic payload. In brief, these steps include: introduction of virus, helper, and target genes into host cells; culture and harvest; and steps to remove nucleic acid, protein, and other impurities, including target vector capture, buffer exchange, and a “polishing” step to separate empty and partially filled, unwanted capsids from full ones containing viral genome and therapeutic sequence that will be delivered to the patient.

The focus of this article is to underscore the importance of this polishing step and present some actionable parameters for optimizing capsid separation. There is no current FDA-established threshold ratio of full-to-empty capsids. However, regulatory standards define empty ones as product impurities that must therefore be measured, characterized, and mitigated to the greatest extent possible. Excess empty capsids can lead to higher-volume or -frequency doses to ensure delivery of the full therapeutic payload, potentially prolonging acute disease management, increasing the risk of liver toxicity, and stimulating innate and adaptive immunities. However, rather than achieving 100 percent full capsid purification, a reasonable goal is to first create a universal process by which to obtain an attractive and reproducible percentage. 

Small-scale separation: Ultracentrifugation

In research laboratories, investigators typically optimize capsid separation with density gradient ultracentrifugation (DGUC). Factors including the amount and G-C content of DNA contained within a capsid impart a particular density to it. Capsids can therefore be isolated along a density gradient, with full, partially full, and empty ones predictably segregating out of preparations within their own respective class. The most common centrifugation matrices are cesium chloride and iodixanol. Both are high-density, low-viscosity compounds, with iodixanol having the additional benefit of being non-toxic. While DGUC provides a high degree and consistency of separation, it is necessarily a small-scale and labor-intensive method that needs to be calibrated specifically to each target and is not conducive to GMP scaling and effective real-time monitoring for separation efficiency. 

Separation at scale: anion exchange chromatography

For an industrial-scale process, the technology with the most long-term promise is anion exchange chromatography (AEX). Instead of differing densities, it leverages a small discrepancy in the isoelectric point between the presence (pI = 5.9) and absence (pI = 6.3) of the complete vector genome inside viral capsids. Both species can be retained on a chromatographic column with a porous resin or a membrane containing a strong anion exchanging group such as a quarternary amine (QA). After equilibration, the positive charge of the QA provides a matrix that is sticky for negatively charged particles, including both full and empty capsids. Elution conditions are developed to separate the two. This method is a significant improvement compared to DGUC in terms of flow rate and process yield, and hence, scalability. 

Investigators and industry professionals have worked to improve and standardize AEX workflows to reliably maximize the full/empty ratio, pushing the peaks farther apart. Early protocols resulted in a large overlap of full and empty capsid peaks, forcing a choice to narrowly collect eluted fractions from the part of the peak without an overlap, which sacrificed yield, or broadly collecting fractions including overlapping peaks, which required another purification step to remove the bulk of contaminating empty capsids.

The right buffer with appropriate pH and components is required to maximize the binding of both full and empty capsids. Salts influence the ionic strength and conductivity of the solution to further impact the interaction between the viral capsids and the AEX resin—careful selection and adjustment of the concentration enhances the separation of full and empty capsids. Additional variables like flow rate, temperature, and additives also contribute to the process efficiency.

To extend their utility for the clinic, developed processes must be assessed for their compatibility with good manufacturing practice (GMP) environments and their ability to maintain critical quality attributes of the purified rAAVs when the process is scaled up. And, ideally, processes will have broad applicability across AAV serotypes with differing cargoes. Recent improvements include innovative chromatography media, fine-tuning buffers, and optimizing elution conditions. For instance, dextran surface extenders have been shown to increase elution efficiencies by providing a stickier lattice for more elusive serotypes such as AAV9. Magnesium or sodium chloride may substitute for TMAC more effectively in some elution paradigms. Yield may improve through the use of step salt gradients instead of linear ones. 

Infrastructure and technology are evolving rapidly to accommodate what will soon be a wealth of AAV-based gene therapies. Anion exchange chromatography is the scalable platform that can best serve to maximize the purity of therapeutic products at the downstream end. With some of the enhancements described in this article, great strides have been made in separating empty and full capsids on AEX resins, membranes, and monoliths.