Identification and quantification of HCPs by LC-MS
Total HCPs measurements of HCCF from different modes of cell culture processes, FB, SS and NSS are listed in Table 1. Samples from both unpurified (i.e. HCCF) and Protein A purified materials were analyzed. For the perfusion systems, HCP counts across the harvest duration provides insights to the HCP profile changes from early to late growth stage.
During the early stages of cell culture growth, there is a significant lag between central metabolism and post-translational modification. A strategy was developed to exclude harvest product from the early days (1-5 days) of production to ensure stable product quality. Hence, the mAb1 quality and composition of HCPs were less varied depending on culture days but maintained significant changes with the different culture processes. The level of HCPs in HCCF were greater than ~105 ppm for FB and NSS perfusion but approximately 104 ppm for SS perfusion. Low cell viability for both FB and NSS perfusion culture processes can mean high levels of HCPs released in the feed stream due to cell lysis.
General patterns of HCP distribution was visualized in 2D-PAGE shown Figure 2. The loading of the gel is normalized by the amount of product. Different culture processes show a comparable spot pattern and intensity with mAb1 spots. For HCP comparison, SS shows less overall HCP population. Day 5 perfusion product also shows less HCPs detected compared to perfusion day 14. Interestingly, NSS perfusion demonstrates additional HCPs in the HMW area (across pI range) while the FB shows additional HCPs across the MW and PI ranges. There are several HCP species (spots) and abundance (spot volume) varied among the culture processes. These minor variations in HCP composition and abundance between different processes suggest influence of cultivation process on HCP in cell culture supernatants.
ELISA and 2D-PAGE have limited capacity to detect HCPs that are in low abundance or poorly immunogenic. Orthogonal detection and monitoring technique using MS offers the opportunity not only to arrive at an accurate quantification of total HCPs but also provides the ability to identify the individual protein species with high sensitivity (Levy et al., 2014). Samples were analyzed by high-resolution LC-MS and HCPs were identified by SEQUEST search against the CHO K1 proteome database (GCF 000223135.1 CriGri 1.0 from www.ncbi.nlm.nih.gov) through Thermo Scientific Proteome Discoverer v2.2. Table 1 summarizes the group number of HCPs in HCCF in the cultures with at least 2 identified unique peptides. A total of 1310 HCPs were identified in FB culture supernatants. Perfusion SS contained fewer detected HCPs (~300) and NSS perfusion shows comparable numbers of HCPs (~1700) as FB that were likely due to the high cell concentration. Approximately 83% of fed -batch, 70% NSS perfusion, 92% SS perfusion were commonly identified with other culture processes using. HCP composition was significantly more diverse in the FB and NSS perfusion culture than in the SS perfusion culture. These data confirm similar trends in HCP population using ELISA.
The HCCF from the three types of cultures were later purified by ProA chromatography and analyzed for HCP population. As indicated in Table 1, results from both MS and ELISA method show drastic decrease on the HCP level for both perfusions, but HCP levels remain relatively high for FB. This result indicates that it might be easy to remove host cell protein from perfusion process by ProA column. To investigate any significant differences in product quality from FB, SS and NSS and their potential impacts on methodologies used in the study, product quality indicators such as aggregation, fragmentation, N-linked glycosylation patterns and charge variants were evaluated after ProA purification (Supplementary Table 1). Both aggregation and fragmentation were slightly reduced in perfusion cultures compared to fed-batch. N-glycosylation showed similar pattern for both types of culture. Perfusion NSS has been observed a simpler glycosylation (G0F) increased as the runs progress and cell densities go up due to the decrease in nutrient availability in all culture processes. At the same time, high mannose (Man5) of NSS perfusion increased and a similar correlation has also been reported in other publications, which indicate that this could be a general phenomenon (Pacis et al., 2011).
The charged variants analysis of mAb1 shows that FB has 24.4% acidic peaks, and 8.4% basic peaks (Figure 3). With perfusion processes, the abundance acidic species is significantly reduced to 13.6% for SS and 15.8% for NSS while more basic forms are produced to 14.6% for SS and 13.4% for NSS compared to FB. Other mAbs also exhibited similar trends in charged variants with perfusion processes due to the increase C-terminal lysine and/or less light chain N-terminal glutamine cyclization (Gomez et al., 2020; Walther et al., 2019). The general consensus according to a number of scientific publications and various conventions is that C-terminal lysine and N-terminal glutamine cyclization, are likely non-CQAs (critical quality attributes), which may not have substantial effects on antigen binding affinity, efficacy and safety of the antibody product (Du et al., 2012; Sing h et al., 2016). However, charged variants can alter the composition of HCP impurity profile if HCPs are binding to mAbs.
The relatively minor charge differences have been shown to have a large impact on the total HCP association due to the single abundant charge species that bind substantially more strongly to HCPs (Levy et al., 2014). For ProA chromatography, post load wash is a critical step to reduce HCPs. In the mAb1 purification process, neutral pH (pH 7.4) was employed for the capture and wash step. Since HCPs tend to have lower isoelectric points (pI) (Lavoie et al., 2019), the majority of HCPs are anionic whereas mAb1 (pI 8.5) is cationic at the neutral pH. Thus, strong hydrophobic and electrostatic interactions can form between mAb1 and HCPs. Based on our proteomic analysis after ProA purification, FB culture contains HCPs that have theoretical pI below neutral pH, pI < 6 ≈ 50%, pI < 7 ≈ 74%, and pI < 8 ≈ 82%. This result showed that these low isoelectric points tend to bind with acidic species of mAb1 in FB culture. Our work brings up the awareness of potential differences of charged variants between fed-batch and perfusion and the needs for further characterization of the product, as well as HCPs. In general, the differences we have observed in all product qualities tested, i.e. aggregates, N-glycan, charged variants, are not expected to have a noticeable impact on the methodologies used in this work.