A novel hydrogen peroxide evolved CHO host can improve the expression of difficult to express bispecific antibodies

Abstract The manufacture of bispecific antibodies by Chinese hamster ovary (CHO) cells is often hindered by lower product yields compared to monoclonal antibodies. Recently, reactive oxygen species have been shown to negatively impact antibody production. By contrast, strategies to boost cellular antioxidant capacity appear to be beneficial for recombinant protein expression. With this in mind, we generated a novel hydrogen peroxide evolved host using directed host cell evolution. Here we demonstrate that this host has heritable resistance to hydrogen peroxide over many generations, displays enhanced antioxidant capacity through the upregulation of several, diverse antioxidant defense genes such as those involved in glutathione synthesis and turnover, and has improved glutathione content. Additionally, we show that this host has significantly improved transfection recovery times, improved growth and viability properties in a fed‐batch production process, and elevated expression of two industrially relevant difficult to express bispecific antibodies compared to unevolved CHO control host cells. These findings demonstrate that host cell evolution represents a powerful methodology for improving specific host cell characteristics that can positively impact the expression of difficult to express biotherapeutics.

Nonenzymatic antioxidants include several small molecules such as vitamin E but are largely centered around glutathione (GSH; Valko et al., 2007). GSH is a tripeptide that is present in all cell types at millimolar concentrations and acts as a major redox buffer affecting a broad range of intracellular systems (Forman et al., 2009). Antioxidants are thought to be important for recombinant protein expression. Indeed, the depletion of the reduced form of GSH (active GSH) has been linked to decreased specific productivity (qP) of manufacturing cell lines (Handlogten et al., 2020) and proteomic work demonstrated that high antibody-producing CHO cell lines upregulated GSH biosynthetic pathways (Orellana et al., 2015). These data are supported by observations that high producer cell lines had increased cellular GSH content (Chong et al., 2012). Consistently, the modulation of GSH synthetic enzymes, through targeted genetic overexpression, was shown to improve mAb titers (Orellana et al., 2017). In addition to GSH, other antioxidants such as thioredoxin reductase 1 and peroxiredoxin 6 were shown to be elevated by depletion of microRNA 23 and linked to improved recombinant protein expression (Kelly et al., 2015). More recently, the transcription factor Forkhead BoxA1 (Foxa1) has been linked to improved expression of difficult to express (DTE) antibodies through a mechanism involving reduced oxidative stress (Berger et al., 2020). Taken together, the upregulation of antioxidants appears to be beneficial for the expression of recombinant proteins in CHO cells.
The development of novel DTE antibodies is often hindered by manufacturing challenges resulting from low product yields (Spiess et al., 2015) which have been associated with high levels of cellular stress, including oxidative stress (Chevallier et al., 2020). Rational genetic engineering approaches involving the manipulation of specific genes through overexpression or targeted genetic ablation to alter subcellular processes have been implemented to relieve production bottlenecks and produce more predictable and robust cell lines. To date, this strategy has been employed to alter diverse subcellular processes including cell cycle (Fussenegger et al., 1998), metabolism (Fogolin et al., 2004), protein secretion (Mohan et al., 2007), and importantly, cellular redox (Banmeyer et al., 2004;Orellana et al., 2017;Warner et al., 1993). Although these strategies have been used to boost cell line characteristics to moderate success, the dawn of the "omics" era suggests that targeting specific genes in this way may not be as effective as originally hypothesized. This is due, in part, to the complex interplay of intracellular pathways that give rise to dynamic web-like interaction systems that are capable of compensating for the misexpression of an individual gene. In fact, using a method that induces global cellular changes as opposed to those that target individual genes may prove more effective for boosting productivity. An example of this is directed host cell evolution, a technique that can offer a relatively unbiased, simple but effective method of engineering CHO cells such that they are evolved to be endowed with specific characteristics that make them superior to their predecessor. With this in mind, we evolved our suspension adapted CHO host cells in the presence of hydrogen

| Production of BisAbs A and B by stable CHO pools
Stable CHO pools expressing BisAbs were generated by transfecting either CHO Control or H 2 O 2 evolved host cells with a plasmid encoding either BisAb A or B and the GS selectable marker using an

| Glutathione assays
Relative changes in intracellular total glutathione (total GSH) and oxidized glutathione (GSSG) were determined with GSH/GSSG-Glo™ Assay Kit (Promega) according to the manufacturer's instructions.
Briefly, CHO Control or H 2 O 2 evolved host cells in culture were harvested and resuspended in fresh AstraZeneca proprietary medium supplemented with 6 mM L-glutamine (untransfected hosts) or 50 µM MSX (transfected pools). Cells were seeded at 10,000 cells/ well in a white 96-well luminometer-compatible plate (medium-only wells were used for background luminescence detection). A 25 µl volume of either total glutathione lysis reagent or oxidized glutathione lysis reagent was added to cell-containing wells and incubated at room temperature on a plate shaker for 5 min. Then, 50 µl of freshly prepared luciferin generation reagent was added to all wells followed by a 30 min incubation at room temperature. Finally, 100 µl of luciferin detection reagent was added to each well and incubated for 15 min. Luminescence was measured using an EnVision Microplate Luminometer (PerkinElmer). The analysis was performed according to the manufacturer's instructions.

| Chemstress assays
Chem stress assays were performed according to the manufacturer's instructions (ChemStress®, Valitacell Ltd). In brief, CHO or H 2 O 2 evolved host cells were seeded into Valitacell ChemStress plates at 18,000 cells/well in 90 µl AstraZeneca proprietary medium supplemented with 6 mM L-glutamine. A control well was incubated with medium alone. Plates were incubated for 72 h in a static incubator at 36.5°C, 6% CO 2 . Following this, 10 µl of neat PrestoBlue dye (Thermo Fisher Scientific) was added to all wells before plates were mixed for 20 s and incubated for a further 30 min at 36.5°C, 6% CO 2 .
Plates were analyzed using a PHERAstar plate reader (BMG LAB-TECH with preconfigured protocols (excitation 560 nm, emission 590 nm). Data were analyzed using the ValitaAPP software (Valitacell Ltd).  remained unchanged (Figure 3a    Day 12 were comparably low between both hosts (Figure 7b,g).

| MSB survival assays
Second, all hosts displayed favorable lactate profiles with H 2 O 2 evolved hosts A and B having lower lactate levels throughout the majority of the fed-batch process (Figure 7c,h). Finally, the titers of BisAb A were 3.5-fold higher from H 2 O 2 evolved host A compared to the CHO control host (Figure 7d). Improvements in titer were also  All qPCR data were normalized to MMADHC mRNA expression and SD calculated on fold change relative to control. The graphs show the mean ± SD, N = 3 in all cases, statistics determined using an unpaired t-test. CHO, Chinese hamster ovary. **p < 0.005, ***p < 0.0005, **** = p < 0.00005 1.75-fold higher than the CHO control host (Figure 7i). Interestingly, H 2 O 2 evolved host A demonstrated a significant increase in specific productivity (qP) of 1.6-fold compared to CHO control A (Figure 7e).
Indeed, the increased volumetric titer observed for BisAb A is likely derived from a combination of improved qP as well as cell growth and viability. By contrast, the improved titer seen for BisAb B appears to result from improved growth and viability as qP was not significantly different between hosts (Figure 7j). In addition, intracellular expression of BisAb A and B, assessed by flow cytometry, showed comparable profiles for Hc and Lc in both CHO control and H 2 O 2 evolved host cells (data not shown).

| DISCUSSION
The production of DTE biopharmaceuticals by manufacturing cell lines is often hindered by low product yields (Spiess et al., 2015) that are, in some cases, associated with enhanced cellular stress. One such stress pertains to alterations in cellular redox state where elevated ROS generation arises due to a complex interplay between ER and mitochondrial burden (Templeton et al., 2013;Tu & Weissman, 2004;Turrens, 2003) as well as fluctuations in bioreactor conditions such as changes in dissolved oxygen concentrations (Handlogten et al., , 2020 and cell culture medium components (Halliwell, 2014;Kelts et al., 2015;Schnellbaecher et al., 2019). The subsequent accumulation of ROS can damage the cell leading to poorer cell performance and lower antibody titers. To address these challenges, we generated a novel H 2 O 2 evolved host that was evaluated for the expression of two industrially relevant DTE BisAbs.
This host demonstrated heritable resistance to H 2 O 2 (Figure 1a,b) and improved survival in the presence of several prooxidant chemicals that were selected to mimic bioreactor stressors (Figure 2a-d).
These chemicals affect a diverse subset of intracellular redox pathways such as those involved in GSH biosynthesis and turnover (BSO and MS; Dunning et al., 2013;Lee et al., 1992) Figure 4a,b). Interestingly, previous studies have drawn links between the expression of antioxidants such as peroxiredoxin 5 and mnSOD and improved CHO cell survival in response to oxidative stress (Banmeyer et al., 2004;Warner et al., 1993). Perhaps consistent with these observations, H 2 O 2 evolved hosts A and B also demonstrated a significant upregulation in a panel of diverse antioxidant defense genes such as those involved with GSH biosynthesis and turnover (GCLC, GCLM, GSS, and GPrx; Figure 5). Perhaps unsurprisingly, enhanced expression of GCLM and GSS have been linked to high producer cell lines (Orellana et al., 2015), and GCLM overexpression has been shown to improve mAb production (Orellana et al., 2017). Furthermore, we observed a significant upregulation in total GSH in H 2 O 2 evolved hosts A and B (Figure 5a-d).
This phenotype observed in the H 2 O 2 evolved host is supported by data linking increased GSH content with improved mAb titers (Chong et al., 2012). In addition, Geoghegan et al demonstrated that the mRNA expression of the cysteine transporter, xCT, was upregulated in CHO cells during increased mAb production in the stationary phase of growth. Moreover, this phenotype was sensitive to xCT inhibition by sulfasalazine and linked to oxidative stress induced by high mAb production (Geoghegan et al., 2018). These data taken together with data in this study, suggest that elevated xCT expres- activity could be enhanced using directed host cell evolution (Spitz et al., 1988) and may facilitate the elimination of excessive H 2 O 2 production, therefore preventing oxidative stress. In addition to the findings presented here, a recent transcriptomic analysis performed by our group in which non-expressing H 2 O 2 evolved host cells were compared with CHO control cells (data not shown) revealed that in addition to the upregulation of diverse antioxidant genes there was a concomitant downregulation of a number of pro-oxidant genes.
These data further exemplify the importance of modulating diverse cellular pathways, as opposed to the targeted expression of single genes, when engineering cells to improve the manufacture of biotherapeutics. It is also important to highlight that in addition to improved antioxidant capacity other mechanisms, not investigated here, may also be involved in improving BisAb titers such as elevated transcription of antibody genes or increased gene copy number.
Collectively, our data suggest that the H 2 O 2 evolved host is better equipped to combat excessive H 2 O 2 production and that H 2 O 2induced evolution can improve cell performance through alterations in the expression of diverse redox pathways. Indeed, given that these phenotypes appear to be stable in transfection pools, which for CHO cells are known to exhibit considerable phenotypic instability, we anticipate that these beneficial phenotypes will also be maintained upon the generation of more phenotypically stable clonal cell populations, although this remains to be investigated.
To conclude, we have generated a novel H 2 O 2 evolved CHO host that has been evaluated for the expression of two industrially relevant DTE BisAbs. The data presented here indicate that global changes in antioxidant pathways, such as those involved in GSH biosynthesis and turnover as well as H 2 O 2 elimination, can confer cellular resistance to a diverse subset of oxidative stressors. Moreover, boosting antioxidant capacity appears to have advantages for better cell growth, viability, and DTE BisAb titers. Improving the expression of DTE biotherapeutics is of great importance as the range of novel formats in the biopharmaceutical industry is expanding rapidly and poses great challenges for their developability and manufacture. By using a host that is better equipped to deal with these stresses alleviates one such challenge. This study highlights the beneficial effects of directed host cell evolution in augmenting global cellular redox networks to improve the manufacturing of DTE biotherapeutics. Finally, these data offer insights into the role of cellular antioxidants in the production cell lines and therefore support growing research efforts to control cellular redox and boost recombinant protein production.