Increased recombinant adeno‐associated virus production by HEK293 cells using small molecule chemical additives

Recombinant adeno‐associated virus (rAAV) has established itself as a highly efficacious gene delivery vector with a well characterised safety profile allowing broad clinical application. Recent successes in rAAV‐mediated gene therapy clinical trials will continue to drive demand for improved rAAV production processes to reduce costs. Here, we demonstrate that small molecule bioactive chemical additives can significantly increase recombinant AAV vector production by human embryonic kidney (HEK) cells up to three‐fold. Nocodazole (an anti‐mitotic agent) and M344 (a selective histone deacetylase inhibitor) were identified as positive regulators of rAAV8 genome titre in a microplate screening assay. Addition of nocodazole to triple‐transfected HEK293 suspension cells producing rAAV arrested cells in G2/M phase, increased average cell volume and reduced viable cell density relative to untreated rAAV producing cells at harvest. Final crude genome vector titre from nocodazole treated cultures was >2‐fold higher compared to non‐treated cultures. Further investigation showed nocodazole addition to cultures to be time critical. Genome titre improvement was found to be scalable and serotype independent across two distinct rAAV serotypes, rAAV8 and rAAV9. Furthermore, a combination of M344 and nocodazole produced a positive additive effect on rAAV8 genome titre, resulting in a three‐fold increase in genome titre compared to untreated cells.


INTRODUCTION
Recent regulatory approvals for recombinant adeno-associated virus (rAAV) mediated gene therapy products (Luxturna in 2017 and Zolgensma in 2019) -together with a significant increase in the number of rAAV gene therapies in clinical trials [1,2] -has highlighted the need for improved production process technology with respect to titre, product quality and cost. Indeed, the high cost of manufacturing rAAV gene therapies to support the typically high therapeutic dosages used (e.g., 6 × 10 13 -2 × 10 14 vector genomes/kg [3] ) has limited the economic viability of AAV gene therapies despite evidence of promising clinical efficacy. Therefore, there is an urgent need for robust, intensified process technology to support production of Good Manufacturing Practice (GMP) quality material. As nearly all current rAAV production processes rely on co-transfection of three plasmids encoding the necessary AAV and helper genes into HEK293 cell hosts for transient production of rAAV within 3-5 days, [4] efficient use of plasmid DNA (itself produced using a costly GMP-approved process) is also of paramount importance.
rAAV is a complex macromolecular product requiring a diverse network of cellular processes and molecular interactions to enable coordinated cellular synthesis -involving host-cell proteins, transiently expressed viral helper, capsid and replicase genes, as well as the single stranded therapeutic viral DNA payload itself. [5][6][7] Small molecule enhancers of recombinant protein production have demonstrated efficacy in a wide variety of mammalian cell lines [3,8,9] and the addition of chemicals as diverse as sodium chloride, sodium butyrate (NaBu) and soy peptones have been shown in previous studies to improve rAAV production yields. [9][10][11] Targeting of discrete pathways involved in the replication, packaging and trafficking of viral particles by bioactive small molecule cell culture additives offers a simple and cost-effective way of increasing viral titre and reducing overall production costs.
Here, we show that microplate plate-based screening of chemical additives can be used to rapidly identify positive effectors of rAAV synthesis in human embryonic kidney (HEK) cells and that significant improvements in viral genome titre can be obtained by subsequent optimisation of dosing regimen. Furthermore, we also demonstrate that a combination of two functionally distinct small molecule enhancers may act additively, resulting in a three-fold increase in viral vector genome titre and thus a substantial improvement in the efficiency of plasmid DNA usage.

Cell culture
Proprietary suspension adapted HEK293 cells were provided by REGENXBIO (Rockville, MD). Cells were cultured in a serum free medium supplemented with L-glutamine (Thermo Scientific, Waltham, MA). Cells were maintained in an orbital shaking incubator (Infors, Bottmingen, Switzerland) at 30 ml culture volumes in 125 ml Erlen-meyer flasks (Corning) at 37 • C, 5% CO 2 and 85% humidity, with agitation at 140 rpm. Smaller scale cultures were grown in shallow-well 24-well plates (0.7 ml culture volume/well, 240 rpm shaking) (Corning, Acton, MA) using the Deutz system. [12] Routine cell density and

rAAV vector production in suspension adapted HEK293 cells
An HEK293-derived cell line adapted to suspension in Dynamis media, SKMB (REGENXBIO Inc.), was subjected to a triple-transfection protocol as previously described, [13] with the following modifications: (1) cis plasmid containing an enhanced green fluorescent protein (eGFP) expression cassette flanked by AAV2 inverted terminal repeats (ITRs); (2) trans plasmids pAAV2/8 and pAAV2/9 containing the AAV2 rep gene and capsid protein genes from AAV8 and 9, respectively and (3)

Request for materials
All requests for materials used in this article should be directed to the primary author at: j.scarrott@sheffield.ac.uk.
Each additive was used at three different concentrations (low, medium and high -values based on previous literature, see Table S1). Total cell culture was harvested at 72 HPT, cells were chemically lysed and the rAAV8 genome titre in the crude supernatant was measured by ddPCR.
Crude genome titre in cells containing chemical additives was com-  [27][28][29] Nocodazole treated cells were found to be up to 27% larger than control cells. Nevertheless, this phenotypic variation was not observed with M344, in which the increase in titre was not accompanied by an increase in cell volume. It may therefore be inferred that these molecules modulate rAAV expression via distinct mechanisms independent of each other and that increased control of rAAV production may be achieved by the application of specific combinations of small molecule effectors.

Early addition of nocodazole enhances rAAV8 production in HEK293 cells
A recent study showed that cap gene expression, translation and assembly of Cap proteins into empty AAV particles is a kinetically rapid process. [30] Approximately 80% of capsids are assembled within the first 24 HPT whereas Rep mediated replication of rAAV genome from the cis-ITR plasmid is slower, peaking after capsid protein levels have plateaued due to the inhibitory action of Rep binding to the packaging plasmid and inhibiting transcription or by translational repression of cap mRNA. As capsid secretion from the nucleus into the cytoplasm is independent of genome loading, pre-assembled empty capsids can be secreted from the site of rAAV genome loading, thus depleting the pool of empty capsids available to newly replicated rAAV genome. [32,31] Therefore, we hypothesised that the effect of a given chemical effector may be positive (or further enhanced), neutral or negative with respect to the timing of its deployment.
To further investigate the impact of nocodazole in enhancing rAAV production, we deployed high dose nocodazole (4 μM) between 0 and 48 HPT (Figure 2). Addition of 4 μM nocodazole immediately (0 HPT) and 4 HPT resulted in a considerable reduction in VCD at 72 HPT of 74% and 64% respectively, compared to rAAV8 producing cells without nocodazole (Ctrl) (Figure 2A). Also apparent at early addition timepoints was a slight reduction in cell viability (p > 0.05; Figure 2B) and a significantly higher mean cell volume compared to cultures not treated  Figure 2D). Addition at the earlier timepoint of 0 HPT was sub-optimal, with a 10% reduction in rAAV8 titre compared to the titre of cells treated at 4 HPT ( Figure 2D). We therefore conclude that early addition (∼4 HPT) of nocodazole to rAAV producing cells is an effective positive mediator of rAAV production in a small-scale, transient expression and suspension HEK cell system.

Nocodazole improves rAAV genome titre in two different serotypes
To investigate the effect of nocodazole addition in a system more representative of large scale rAAV production, culture volume was scaled up from the initial microplate culture volume of 0.7-30 ml culture volume in shake flasks. To rule out a serotype-dependent effect of nocodazole addition, plasmids for both rAAV8 and rAAV9 serotypes were separately transfected in HEK293 cells and cultured with and without the addition of 4 μM nocodazole. rAAV genome titre was measured from 24 to 72 HPT. All measured parameters of nocoda-zole induced cell cycle arrest (reduced VCD, decreased viability and increased mean cell volume) and increased rAAV genome titre were consistent with those seen at smaller culture volumes ( Figure 3A,B).
Comparable to microplate-based cultures, mean viral genome titre was increased by up to 2.5-fold in nocodazole-treated cultures compared to untreated cultures ( Figure 3C). Importantly there were no observed significant differences in the above-described measurements between cultures producing rAAV8 or rAAV9, suggesting a broad applicability for nocodazole addition in rAAV manufacturing.

rAAV producing cultures treated with nocodazole have an increased proportion of cells in G2/M phase
Nocodazole is an anti-mitotic agent, used as both a chemotherapeutic and as a common agent of cell cycle synchronisation. [17,[32][33][34] Noco- . Data shown are the mean ± SEM. n = 3 independent biological replicates. Fold change between nocodazole treated and untreated cultures was analysed for each serotype separately by Student's unpaired two-tailed t-test with respect to untreated cultures. **p < 0.01, ***p < 0.001. HPT, hours post-transfection; rAAV, recombinant adeno-associated virus; SEM, standard error of mean; VCD, viable cell density by apoptosis. [34][35][36][37] In addition to its use as a cell synchronisation agent, nocodazole has previously been shown to increase transient recombinant protein expression in a mammalian cell system. [38] Flow cytometry was carried out to determine the cell cycle status of rAAV8-producing cultures treated with nocodazole. Untreated rAAV8 producing cells were found to be almost entirely in either G1 or S phase at 72 HPT ( Figure 4A) such that later additions to cell culture are sub-optimal in terms of producing high rAAV titres, likely due to the rapid assembly process of cap proteins that largely occurred within 24 HPT. [30] The nucleolus is a dynamic compartment within the nucleus which undergoes extensive remodelling during cell cycle progression, particularly during mitosis. [39] Previous immunocytochemical examination of nucleolar localisation throughout normal cell cycle progression shows a distinctive dispersal of nucleolar protein staining throughout the cytoplasm during prometaphase and metaphase -concomitant with the breakdown of the nuclear membrane during mitosis. [40][41][42] Nucle-olar proteins in cells in interphase and prophase are typically found located within the nucleus itself. Of note, the nucleolus is considered to be a likely site for AAV capsid assembly and Rep-mediated loading of AAV genome into capsids, [43][44][45][46] as well as being linked more generally to viral replication in other human viruses. [47] Cells from rAAV8 producing cultures -both with and without nocodazole addition 4 HPT -were harvested and fixed at 24 HPT. Cells were stained with a nucleolar marker (fibrillarin -a protein component of the nucleolus associated with ribosomal RNA processing [41] ) and DAPI (a nuclear DNA stain). Widefield fluorescent microscopy was

Small molecule additives can be combined to further enhance rAAV8 production
In order to evaluate whether specific combinations of effectors could act synergistically to further increase rAAV production, we utilised  [10] Z-VAD-fmk is a pan-caspase inhibitor. [19] Each chemical was added to rAAV8 producing HEK293 cultures together with nocodazole at 4 HPT in 24-well microplates. The inclusion of z-VAD-fmk in the initial screening experiment was predicated on its anti-caspase activity, as we hypothesised caspase mediated apoptosis -linked to rAAV production and expression of AAV2 Rep proteins -would negatively affect final crude titre. [50,49] Additionally, caspase-mediated apoptosis induced by nocodazole-mediated cell cycle dysregulation would further increase cell death within the production cultures. Increased cell viability and VCD in cultures treated with nocodazole/z-VAD-fmk relative to untreated, nocodazole treated or nocodazole/M344 treated cells suggest that apoptosis was reduced ( Figure 5A,B) but with an unexpected reduction in crude genome titre ( Figure 5D). The observed reduction may be a consequence of off-target induction of autophagy via inactivation of n-glycanase 1 (NGLY1) that has been shown to occur in HEK293 cells treated with z-VAD-fmk. [50] While the role of autophagy in rAAV production is undetermined, inhibition of autophagy has been shown to increase recombinant protein expression in CHO cells [51,52] and unintentional upregulation of autophagy may result in a reduction in viral component proteins necessary for production of high viral titres. Interestingly, the highest mean cell volume, which appeared to correlate positively with genome titre, was measured in cells treated with both nocodazole and z-VAD-fmk ( Figure 5C). This may be a result of reduced apoptotic cell death allowing cell size to increase but with the caveat that any benefit gained from this from a production perspective is attenuated by the potential negative off-target effects of z-VAD-fmk.
We observed that the combination of 2.5 μM M344 and 4 μM nocodazole produced an additive effect, increasing crude rAAV genome titre 2.6-fold compared to untreated cultures, an improvement on nocodazole alone (two-fold increase compared to untreated) ( Figure 5D). This additive effect was also observed in 30 ml shake intact capsid-specific ELISA analysis of total capsid titre showed a significant increase in intact capsids after addition of nocodazole (4.3fold increase) and nocodazole/M344 (11.3-fold increase) compared to untreated controls (data not shown). While this increase in the relative ratio of empty to full capsids is suboptimal in terms of commonly assessed product quality attributes, recent improvements in downstream processing of crude AAV lysate will likely reduce the impacts of excess empty capsids during final product formulation, whilst retaining the benefits of increased genome titre. [53] Additionally, we anticipate that rAAV vector development could complement the process engineering strategy to further maximise rAAV titre (e.g., high intact capsid levels are likely to benefit from hybrid Rep with improved genome packaging efficiency) [54] while combinatorial empirical modelling will enable systematic determination of optimal chemical dosage and timing.
Taken together, these data show that nocodazole, either alone or in combination with select small molecules, can reproducibly boost both genome and total viral particle titre in a transient suspension HEK293 rAAV production system and that it is both scalable and applicable to production of rAAVs derived from two phylogenetically distinct and clinically translatable pseudotyped capsids.

DISCUSSION
Improving the yield of intact, genome-containing rAAV particles during viral vector production is a critical step to reducing overall production costs. Here, we describe a simple and robust method by which viral wildtype AAV replication within the G2/M phase of the cell cycle. [59,60] Nocodazole has been shown to significantly improve transient transfection efficiency in CHO cells, [38] which may stem from increased nuclear permissibility of transfection complexes due to the breakdown of the nuclear membrane during mitosis. A reduction in cell proliferation caused by nocodazole addition may also benefit viral production by reducing plasmid copy number dilution and maintaining mRNA transcript levels. [56] A recent study utilising a CRISPRmediated genome wide screening strategy identified two target genes (ITPRIP and SKA2) that when modulated in cells increased rAAV genome titre and improved full/empty capsid ratios, with both target genes (strongly, in the case of SKA2) associated with cell cycle modulation. [61] Further to this, a proteomic study of HEK cells during AAV5 production highlighted a number of proteins involved in cell cycle and proliferation as being strongly downregulated during production. [62] The molecular effects of cell cycle arrest within the G2/M phase on rAAV production are unknown, but there is a potential correlation between the loss of essential nucleolar functions and the nucleolar localisation of viral proteins, while the volume increase could possibly minimise crowding effects due to accumulation of viral proteins. [63] This apparent link between the cell cycle and AAV production, and the abundance of cell cycle modulating molecules, necessitates further investigation into the use of cell cycle modulators for both improving rAAV production yields and ultimately improving our understanding of the underlying biological processes governing rAAV production.
The apparent important temporal aspect of cell cycle regulation within the production process may also provide avenues for nonchemical interventions to improve vector yield as this relationship becomes better understood. Whilst we have not investigated the mechanism behind M344-mediated titre enhancement, HDAC6 (of which M344 is a selective inhibitor) has been shown to bind to, and regulate clearance of, ubiquitinated proteins via induction of the heatshock cellular response. [64] AAV capsid proteins are a known target of ubiquitination post-viral entry, [65] and the ubiquitin-proteasome pathway (UPP) has been suggested to play an active role in AAV capsid monomer degradation, [66] therefore we posit a link between M344mediated UPP dysregulation and increased crude viral titre. Due to the robustness of the results between small-scale plate-based cultures and larger scale shake flasks, we believe that the screening process shown here could be further scaled down and automated to increase throughput, due to the availability of instrumentation that can rapidly and accurately dispense very small volumes of drugs into culture. As minimal volumes are required for ddPCR analysis of genome titre, identification of novel enhancers of rAAV could be rapidly incorporated into existing rAAV production platforms with minimal changes to existing protocols.
In summary, we show that the use of readily available small molecule enhancers can significantly improve rAAV production yield. We show that small molecule enhancers of rAAV production are amenable to optimisation in an existing suspension HEK293 cell system and that positive hits from initial small-scale screening of enhancer molecules are translatable to improvements in genome titre up to at least 30 ml shake flask scale, with the potential for translation to commercially viable volumetric production (>10 L), as has been demonstrated for a similar production system. [67] We also show that increased titre resulting from nocodazole treatment is consistent across two different serotypes, suggesting broad applicability in rAAV manufacturing.

CONFLICTS OF INTEREST
Part of this work was conducted under a research agreement between the University of Sheffield and REGENXBIO Inc. Ping Liu and Ayda Mayer are employed by REGENXBIO Inc. and hold shares in the company.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the lead author upon request. Please contact j.scarrott@sheffield.ac.uk for data.