Scalable Liquid-Phase Oligonucleotide Synthesis Enabled by Organic Solvent Resistant Membrane Separation

Recent years have seen significant investment in therapeutic oligonucleotides as a modality to treat a plethora of diseases. There are currently over 20 approved oligonucleotide therapeutic drugs. Most approvals were limited to rare genetic diseases until the recent approval of Inclisiran for cholesterol-lowering treatment which has marked a turning point for the field. Oligonucleotides, typically 15-25 nucleotides long, are currently manufactured in an iterative addition of nucleotide monomers by solid-phase oligonucleotide synthesis (SPOS) which is limited to batches sizes of approximately 10 kg due the lack of in-line product analysis and increasing column sizes, resulting in nonlinear flow rates and reduced product purity. Despite the high coupling yields, automation and simplicity of purification SPOS generates large volumes of solvent and reagent waste which is an issue from an economical and environmental perspective. Due to the rapid growth in the field with hundreds of oligonucleotides in clinical trials it is expected that there will be a need for an alternative manufacturing platform. This thesis focuses on the development of a liquid-phase oligonucleotide synthesis (LPOS) platform assisted by nanofiltration and ultrafiltration membrane separation for improved scalability and sustainability. Growing intermediate oligonucleotides are attached to a soluble support to facilitate membrane separation by imparting a large molecular weight difference in comparison to excess reagents and by-products, in addition to improving solubility in organic solvents. Liquid-phase chemistry and membrane filtration is highly scalable due high rates of reaction and large reaction and filtration volumes. Several commercially available polymeric (Borsig oNF-3 and SolSep UF) and ceramic membranes (Inopor NF 750 Da and UF 2000 Da) were screened for use in harsh oligonucleotide synthesis conditions which use various organic solvents along with acidic and basic conditions. It was found that the ceramic membranes were most suitable for integration with LPOS due to their superior chemical stability, quantitative product retention and high permeance. An ultrafiltration pore size (~3 nm) was most suitable for the process due to greater selectivity and permeance compared to nanofiltration membranes (~1 nm). A four-3 armed PEG was chosen as the soluble support to enhance membrane separation as it is highly retained by the membrane, easily functionalised by the first nucleoside and very soluble in the standard organic solvents used during LPOS. A one-pot synthesis strategy was devised which enabled sequential coupling, sulfurization and detritylation without intermediate separation by quenching excess phosphoramidites and achieving complete detritylation. Use of a one-pot strategy enables a reduction in the number of separation steps and more efficient purification compared to previously reported LPOS two-pot strategies. Reactions could be monitored and optimised by in-line analysis by NMR and HPLC. Using the developed one-pot synthesis with ceramic membranes, 6mer and 18mer 2’-OMe phosphorothioate oligonucleotides were successfully synthesised with nearly complete product membrane retention (99.5-100%), efficient diavolume usage (5-10 diavolumes per separation), and high crude purities (81% for 6mer and 72% for 18mer) after cleavage from PEG supports by ammonolysis. The use of commercially available ceramic membranes for OSN and OSU proved highly effective, demonstrating stability across all tested reagents and solvents. The PEG-oligonucleotide products exhibited high solubility in acetonitrile, and solubility could be further enhanced by adding methanol to improve filtration performance, especially for longer chain lengths, where diafiltration performance becomes a limiting factor. This approach can be used in its current state to generate up to around 18mer oligonucleotides or shorter fragments (~3-10mer) for enzymatic ligation.