Introduction
Chinese hamster ovary (CHO) cells are the most common host organisms for
production of recombinant therapeutic proteins, including monoclonal
antibodies (mAbs). Fed-batch (FB) cell culture processes using CHO cells
have been established as a platform for most mAbs in the
biopharmaceutical industry (Zhu, 2012). As cell culture platforms
continue to evolve, perfusion process stands out as an attractive next
generation alternative to fed-batch process. High demand, competitive
landscape and fast-paced biopharmaceutical markets require higher
productivity, manufacturing flexibility and reduction in development
time (Rodrigues et al., 2010). In order to accept the current market
need, continuous downstream manufacturing combined with perfusion
upstream process offers great advantage on the intensification of the
volumetric productivity with limited investment. In the perfusion
process, media is fed on a regular basis and the product containing
media is harvested continuously. The continuous product removal and
short residence time reduce the variability in product modification and
lead higher product heterogeneity. Short residence time adds another
advantage for unstable molecules such as enzymes or bispecific
antibodies (Karst et al., 2018).
Perfusion cell culture have diverged to two mode of operations, steady
state (SS) and non-steady state (NSS) (Bielser et al., 2018). SS
perfusion mode maintains constant cell density over a lengthier culture
duration. The harvest stream is used to remove excess cells to keep a
steady concentration. NSS perfusion mode focuses on a higher intensified
cell density, which varies during the culture duration. Therefore,
perfusion rate must be adjusted dynamically to meet the cellular demand
and allow cells to grow at the maximum rate (Karst et al., 2017).
Changing the upstream processes can impact the host cell protein (HCP)
profile and the relative abundance of particular proteins in harvest
cell culture fluid (HCCF), as well throughout the downstream
purification (Hogwood et al., 2013; Park et al., 2017). Despite great
research interest in HCPs, there are no published investigations to
compare perfusion and fed-batch processes for differences in HCP
production and profiles, as well as the two modes of perfusion cell
culture process.
In this study, we collected HCCF from various bioreactor cultures (FB,
SS perfusion, NSS perfusion) at different growth phases of a
mAb1-producing CHO DG44 cell line. A complete analysis of HCPs will
establish the early proof of concept to understand the differences in
HCP production based on differences in cell culture processes. We
utilized liquid chromatography-mass spectrometry (LC-MS) based
proteomics throughout the study, which have recently shown great promise
for HCP characterization (Bracewell et al., 2015; Walker et al., 2017).
Discovery-based proteomics by data-dependent acquisition (DDA) is a
powerful technique for unbiased identification of proteins in a sample.
However, this method is not ideal for the consistent detection of low
abundance proteins. HCPs co-exist with highly concentrated mAb product
(dynamic range of concentrations ≥ 5 orders) in the HCCF and
purification intermediates, which puts a great challenge for detection.
Here, to meet our needs, we developed a parallel reaction monitoring
(PRM) approach for the target proteins of interest. PRM uses targeted
tandem MS to simultaneously monitor product ions of a targeted peptide
with high mass resolution and accuracy (Kreimer et al., 2017).
In our HCP profiling studies, we focused on the comparison of one
particular group of HCPs that are known to affect product stability.
Particle formation in the polysorbate (PS)-formulated biotherapeutics
that has become a major quality concern and potential risk factor in the
industry. Recent published data and our own investigation have provided
clues for the leading cause of PS degradation by HCPs remaining in drug
product. Low levels of residual HCPs that have lipase activity, such as
lipoprotein lipase (LPL) (Chiu et al., 2017), hydrolyze polysorbates
cleaving them into fatty acids. The accumulation of free fatty acids
ultimately precipitates to form particles upon long or in some cases
short-term storage.
Particle formation in drug substances was observed in mAb1 and this
molecule was used in this study. Source materials were generated by
three different modes of cell culture processes, subsequently the
enzymes that potentially can hydrolyze polysorbates were identified,
quantified and compared. The aim of this body of work was to establish
proof of concept and methodology to compare FB and perfusion cell
culture processes. Perfusion process is shown to be a viable alternative
to utilize on the problematic molecules, such as mAb1 that have product
quality issues when cultured in traditional FB mode.