Rational Design of Anion-Exchange Chromatography for Lentiviral Vectors Through Mechanistic Insights into Sorption Behaviour

The expansion of cell and gene therapy clinical trials has sharply increased demand for viral vectors. Lentiviral vectors (LVs) have become the primary tools for engineering patient cells to produce CAR-T therapies, an emerging form of cancer treatment. However, efficient LV purification remains a major bottleneck to delivering these therapies affordably and at scale. Product loss during anion-exchange (AIEX) chromatography, the dominant method for primary capture, is the main contributor. Large-scale recoveries are exceptionally low (~20%), and reported yields vary widely (0-100%), reflecting limited understanding of the complex adsorption phenomena and loss mechanisms underpinning LV AIEX. Consequently, no consensus has been reached on optimal chromatography configurations. Addressing this challenge requires identifying dominant loss mechanisms and defining how LV structural attributes and adsorbent morphology influence performance. Such insights will support the rational design of next-generation AIEX strategies capable of reliably achieving high LV recovery. This research demonstrates that time spent in the adsorbed state (contact time) is a critical, and likely the major, source of LV product loss on commercial Q-membrane adsorbents. This loss was attributed to a process of increased irreversible binding, rather than vector inactivation. Characterization of loss kinetics demonstrated half of all recoverable product was lost within 13-19 minutes, highlighting the rapid rate of material loss. By using low contact times (5 min), significant improvement to AIEX recovery is achieved (~60%). Increasing adsorbent occupancy reduced this effect, while no correlation with overall LV charge was observed. These findings identify contact time as the predominant factor limiting LV recovery in commercial Q-membranes, suggesting that their current interaction strengths are incompatible with high LV recovery. Linear gradient elution consistently produced a two-peak profile, as widely reported in the literature. Elucidating the LV structural factors driving this behaviour is therefore essential for rational process design. Enzymatic digestion and peak re-injection experiments revealed two distinct LV subpopulations, each binding via different envelope components: a strongly binding population interacting through glycosaminoglycans (Peak 2), and a weaker-binding population likely interacting via the phospholipid membrane and/or envelope proteins (Peak 1). Importantly, these subpopulations displayed differences in primary T cell transduction efficiency, highlighting a potential new source of product heterogeneity. Manipulating structural features of Q-nanofiber adsorbents influenced both LV recovery and the distribution of the two-peak elution profile. Increasing adsorbent porosity improved recovery, likely by mitigating time-dependent loss, while nanofiber diameter and ligand density primarily influenced peak separation within the two-peak gradient elution profile. The best performing adsorbent configuration achieved high LV recovery (∽70%) and enhanced resolution between LV subpopulations, offering a potential route to isolate high-potency LVs. These insights were formalised in a physics-based mathematical model of LV AIEX chromatography. A modified multi-state steric mass action framework accurately captured both two-peak elution behaviour and contact time-dependent loss, providing a key first step in developing a predictive tool for in silico process development. Collectively, the findings of this thesis provide a deeper understanding of LV sorption behaviour, offering a foundation for the rational design of high-yield LV AIEX processes, in silico process models, and future vector-specific separation materials.