INTRODUCTION
Chinese hamster ovary (CHO) cells are one of the main host systems used
by the biopharmaceutical industry to manufacture biologics such as
monoclonal antibodies (mAbs). While advances in process development,
cell line engineering and clonal selection have largely improved
recombinant mAb yields, ensuring consistent product quality still
remains a major challenge (Hong et al., 2018). A key product quality
attribute is N-linked glycosylation, which has been reported to
significantly impact the biological activity of mAbs (Jennewein and
Alter, 2017). For example, the absence of fucose in the terminal
N-glycan of the fragment crystallizable (Fc) region of mAbs is reported
to increase antibody-dependent cell-mediated cytotoxicity up to 50-fold
(Shields et al., 2002), while N-glycan structures with high mannose
content lead to decreased in vivo half-life (Goetze et al.,
2011). In addition to N-glycosylation, protein aggregation is another
key determinant of mAb quality (Paul et al., 2018). Particularly, high
molecular weight (HMW) aggregates have been shown to enhance the risk of
undesirable immunogenicity (Vázquez-Rey and Lang, 2011). Variations in
protein charge can also adversely affect the pharmacokinetics of the mAb
(Xie et al., 2016), especially decreases in net positive charge (leading
to formation of acidic variants) can decrease tissue retention and
increase whole body clearance of antibody (Boswell et al., 2010).
Therefore, to ensure cells are in desired metabolic state for consistent
quality profiles of mAbs, cell culture processes need to be well
controlled by maintaining process parameters within an acceptable range.
Several cell culture process parameters including media components,
feeding strategy, temperature, pH and osmolality are known to influence
the quality profiles of mAbs. Among these, pH is a critical parameter
which has been shown to greatly affect N-glycosylation profiles
(Aghamohseni et al., 2014; Ivarsson et al., 2014; Jiang et al., 2018),
aggregation levels (Paul et al., 2018) and charge variant species (Xie
et al., 2016) of the mAb product. Current understanding of how process
conditions affect quality profiles has stemmed from empirical
understanding and experience within each organization without detailed
scientific discussion in literature. With the greater emphasis placed by
regulatory agencies on adopting the quality by design (QbD) approach in
process understanding and product quality control, a more detailed
characterization of cellular mechanisms arising from changes to process
conditions, in part, facilitates the development of rational control
strategies (Sommeregger et al., 2017). In this regard, a
knowledge-driven approach based on the combined use of systems modeling
and high-throughput “-omics” data profiling offers immense potential
to link process conditions and/or clonal traits with product quality in
a systematic manner (Lakshmanan et al., 2019; Yusufi et al., 2017). Only
a handful of studies have adopted high-throughput “-omics”
methodologies to understand the biological mechanisms in cell culture
conditions that affect productivity and/or quality profiles.
Despite the enormous potential, significant resource demand has limited
the ability of organizations to take on these omics approaches. Most
efforts have utilized only a single omics or at most two different omics
approaches for their investigations since substantial resources and
knowledge are required to extend it further for multi-omics analyses.
Such examples include the application of transcriptomics and proteomics
profiling to identify the pathways that are associated with improved
recombinant protein titers in low-temperature CHO cultures grown under
batch or perfusion modes (Baik et al., 2006; Kaufmann et al., 1999;
Tossolini et al., 2018). Similarly, a combination of metabolomics and
proteomics detected higher reactive oxygen species (ROS) and glutathione
peroxidase expression levels during the CHO inoculum scale-up step,
indicating hypoxia as the cause of lower productivity (Gao et al.,
2016). More recently, proteomics was used to assess the effect of
pCO2, media hold duration, and media manganese content
on mAb N-glycosylation structures in CHO fed-batch cultures (Nguyen Dang
et al., 2019). However, to the best of our knowledge, no in-depth
investigation has combined more than two “-omics” datasets to discern
the complex underlying mechanisms between changes in process conditions
and their impact on the product titer and/or quality profiles.
In this study, we
obtained
comprehensive time-based transcriptomics, proteomics and metabolomics
data from CHO fed-batch cell cultures as a result of varying pH set
points to understand the cellular effect of pH controls on the biology
of CHO cells. Relevant statistical analyses of multi-omics datasets were
carried out to investigate the variations in both mAb titer and quality,
including N-glycosylation, protein aggregation and charge variants, as a
consequence of pH variation. Our study pinpointed the influences of pH
and cell culture time on mAb quality attributes through varying
responses of several pathways including N-glycosylation, intracellular
vesicular trafficking, and response to ER and oxidative stress.
Specifically, we noted culture pH set points impacted organellar
homeostasis, N-glycosylation, and vesicular trafficking efficiency, thus
resulting in differences in cell productivity and product quality.