DISCUSSION

A multi-omics data-driven systems biotechnology approach is presented here, characterizing the effect of pH on titer and various quality profiles of mAbs produced in CHO cells. This approach is necessary for the dissection of underlying molecular mechanisms as the changes in specific rates of by-product production alone could not explain the variations in observed quality profiles. Specifically, our work clearly demonstrated that N-glycan structures are influenced by transcript and protein expression changes in various pathways including N-glycan biosynthesis, ER and Golgi vesicular transport, fatty acid β-oxidation, and responses to unfolded proteins (UPR), ER stress and oxidative stress. Similarly, temporal variations in gene expression profiles of UPR and response to ER stress were associated with high molecular weight (HMW) species as the culture progressed. The pH-dependent changes in charge variants were also positively correlated with the expression differences of relevant post-translational modification enzymes.
The relative abundance of early glycans in low-pH cultures may be explained by the pH specificity of intracellular organelles where the resting pH of the Golgi ranges from ~pH 6.7 in the cis-Golgi to ~pH 6.0 in the trans-Golgi (Figure 4 ). The maintenance of ER and Golgi pH homeostasis is important for ER and Golgi functions such as protein glycosylation, membrane trafficking and protein sorting (Kellokumpu, 2019; Paroutis et al., 2004). We postulate that the influx of H+ into the cell in low-pH cultures decreases the pH of the cytoplasm and the luminal pH of the ER and Golgi via increased expression of v-ATPases (Supplementary Figure 7 ) and PAM (Supplementary Figure 8A ), the latter a proposed v-ATPase associated pH sensor within the secretory pathway (Rao et al., 2019) (Figure 5, left panel ). The resultant disruption of luminal pH gradient through acidification may constitute a significant perturbation in the more basic compartments of the early secretory pathway and is associated with an increased expression of the corresponding localized N-glycosylation enzymes. Coupled with higher expression of members of ER and Golgi vesicular trafficking pathways in low-pH cultures (Supplementary Figure 6 ), the implied increase in general secretory pathway trafficking and thus a shorter residence time for cargo proteins in ER and Golgi will result in higher abundance of early mAb glycoforms.
Cellular stress resulting from prolonged high production of mAb may account for the decreased relative abundance of late glycans, with culture time having a much more pronounced effect in comparison to culture pH. As the cell culture progresses, we hypothesize there is a disruption of ER homeostasis due to ER stress associated with increased levels of unfolded proteins (Gerlach et al., 2012) as well as oxidative stress associated with oxidative protein folding in the ER and increased fatty acid β-oxidation that generate reactive oxygen species (ROS) (Malhotra and Kaufman, 2007) (Figure 5, right panel ); the increase in the bioenergetic fatty acid β-oxidation is possibly a response to higher energy demand for protein productivity by the cells. ER stress and oxidative stress are also known to exacerbate each other (Nakka et al., 2016), and their resultant disruption of ER homeostasis promotes accumulation of unfolded proteins which can lead to aggregation of mAb into HMW species (Li et al., 2016; Zhou et al., 2018). The decrease in complex (late) mAb glycoforms, resulting in hypoglycosylation and reduced glycodiversity, may arise from accumulated irreversible oxidative damage to glycosylation enzymes that causes increasing loss of functionality over time (Korovila et al., 2017). Such damage will naturally have a larger cumulative effect further down in a sequential chain of enzymatic reactions.
While previous studies have investigated the effects of pH on quality attributes such as N-glycosylation (Aghamohseni et al., 2014; Ivarsson et al., 2014; Jiang et al., 2018), aggregation (Paul et al., 2018) and charge variants (Xie et al., 2016), to the best of our knowledge, no report has investigated the underlying causes of these pH-specific effects. In this regard, this study systematically characterizes the effects of pH on quality attributes at various levels of the cellular hierarchy, i.e. transcripts, proteins and metabolites. Such integrative analyses allow us to understand that some cellular processes emerge at transcriptional level whereas others emerge only at the post-transcriptional or metabolite levels. The increased expression of vesicular transport genes in low-pH cultures, and its potential contribution to higher levels of early glycans, was evidenced by transport, protein and metabolite datasets. Such concordant observations across multiple omics datasets strongly suggest the activation or suppression of certain pathways and its consequences at downstream layer, thereby allowing us to derive relevant hypotheses at the global level which may not be possible with only one of the omics datasets. However, certain observed changes in product quality were found to be linked to a specific omics dataset – for example, the variations in mAb aggregation levels were found to be associated with differential expression of several protein disulfide isomerases only at the protein, i.e. post-transcriptional, level. We also noted a decrease in glycodiversity as the cell culture progressed, irrespective of pH with concomitant increased expression of multiple ER stress and oxidative stress markers in all three omics datasets (Supplementary Figures 3, 4 ).
Overall, the integration of multi-omics data allowed us to unravel mechanisms involving multiple pathways and enzymes that are associated with variations in product quality, and also highlighted the complex and dynamic interactions between these quality profiles and pathways. Taken together, the systematic approach adopted in the current study augments our understanding of factors affecting mAb quality attributes, and we believe that the insights gained from this study can contribute towards the development of targeted approaches that result in more effective product quality control strategies.