Optimized HCP Assay with Reduced Matrix Interference from Protein A: Improved Performance in Dilution Linearity and Spike Recovery

HTTPS://STOCK.ADOBE.COM

Chinese hamster ovary (CHO) cell lines are a highly efficient expression platform widely used in biotherapeutic manufacturing. However, CHO cell-derived impurities, including host-cell proteins (HCPs), remain significant concerns for patient safety and product stability. HCPs originating from manufacturing cells can induce adverse immune responses in patients (1). HCPs also have been shown to reduce stability and shelf life of therapeutic monoclonal antibodies (mAbs) and excipients (2, 3). Strong interactions between biotherapeutic materials and HCPs can occur, leading to copurification of HCPs alongside biologic products. Thus, HCPs are critical quality attributes (CQAs) that require constant process optimization for clearance throughout biomanufacturing (4).

Several purification steps can reduce and remove HCPs: e.g., depth filtration and chromatography based on protein A affinity, anion exchange (AEX), cation exchange (CEX), and hydrophobic interaction (HIC). HCP levels should be tested and controlled throughout the life cycle of biologics development and commercialization. The biopharmaceutical industry follows the United States Pharmacopeia General Chapter <1132> and ICH Q6B guidance for HCP testing, characterization, and release (e.g., HCP specification <100 ng/mg) (5, 6). Although HCP testing can be performed with generic commercial HCP immunoassay kits during early clinical phases, process-specific HCP kits are preferred for clinical and postapproval products because they can detect a spectrum of HCP species for specific manufacturing processes (6). Despite progress made with next-generation HCP technologies and automation such as the Ella system (Bio-Techne), the GyroLab platform (Gyros Protein Technologies), the AlphaLISA assay (Revvity Inc.), and liquid chromatography with tandem mass spectrometry (LC-MS/MS) (7–9), HCP enzyme-linked immunosorbent assays (ELISAs) remain the gold standard for release testing and process support. Therefore, improvements to HCP ELISA assays can help scientists in the biopharmaceutical industry to investigate, troubleshoot, and improve methods for HCP analysis.

Here we report three cases of assay optimizations during HCP testing of an early phase mAb. The first case describes how a commercial HCP kit did not detect HCPs present in bulk drug substance (DS), as indicated by failed dilution linearity. Optimization of sample treatment and reagents did not resolve the issue completely, and a process-specific assay was introduced for those HCP samples. The second case describes our observation of drift in spike recoveries that was observed to be independent of sample type but was associated closely with plate-well positions during method qualification of a process-specific HCP assay. Similar low-recovery issues were reported previously by other quality control (QC) laboratories performing HCP validations. In the third case, we identified an interference of protein A molecules with an HCP ELISA assay. We investigated root causes and propose optimizations herein to address such issues in HCP testing. The same classical HCP ELISA format has been used widely for two decades and remains a gold standard in the biopharmaceutical industry. We hope that our findings can help scientists to improve their assay quality and reduce invalid rates for HCP testing.

Materials and Methods

A commercial CHO HCP kit (Cygnus catalog F550-1), an Alexion process-specific CHO HCP kit, and protein A standards were used in the HCP interference studies. All mAb DSs and process intermediates came from Alexion, SpectraMax M2, and M5 plate readers (Molecular Devices). HCP standards at 2, 6, 12, 25, 50, and 100 ng/mL were obtained from HCP kits. HCP spikes at 3, 6, 10, and 50 ng/mL were prepared by mixing HCP standards with diluted samples.

We pipetted 100 uL of diluted horseradish peroxidase (HRP)–conjugated, anti-CHO detection antibody to precoated microtiter strips, followed by 50 uL each of HCP standards, controls, and samples. Solutions were incubated on a shaker at room temperature for two hours and were washed manually four times with phosphate-buffered saline (PBS) containing 0.05% Tween 20 (MilliporeSigma). Washed plates were incubated with 100 uL of tetramethylbenzidine (TMB) substrate (Thermo Fisher Scientific) for

30 minutes at room temperature without shaking and then were examined using colorimetric optical densities (ODs) of 450 nm and 650 nm at a plate reader in reference to HCP standards.

Results and Discussion

Dilution Linearity: A mAb-based DS from process optimization was tested by multiple analysts using a generic commercial HCP kit. All samples failed multiple times to achieve acceptable sample dilution linearity within a 25% coefficient of variation (CV) (Table 1). Two data sets indicated the complexity of potential root causes:

• HCP spike recoveries for DS samples consistently passed assay acceptance criteria while nonspiked DS samples always failed dilution linearity (>25% CV)

• DS sample denaturing by heating (95 °C for 10 minutes) improved dilution linearity but was not sufficient to pass dilution-linearity criteria (Table 1). In addition, heating caused protein aggregation, precipitation, and loss of HCP reactivity and thus was not considered ideal for QC use.

Table 1: A commercial host-cell protein (HCP) kit failed dilution linearity with limited improvement from sample heating. Dilution linearity and spike recovery were reported from original Softmax files prior to rounding. Final reported spike recovery is adjusted from spike control.

Application of a process-specific HCP kit for testing that DS showed improved assay performance, especially for sample-dilution linearity (Table 2). We believe that poor dilution linearity observed in previous testing was caused by a low abundance of HCP antibody species in commercial HCP kits relative to our HCP sample, as well as low-binding affinities between our HCP samples and antibodies from the commercial HCP kit. Both factors may have been overcome by our process-specific HCP kit. That supports the industry-recognized benefits of process-specific HCP assays, especially for late-stage processes and approved commercial products, as a preferred method by regulatory agencies. We measured 60% loss of HCP values from the heating step.

Table 2: Improved dilution linearity from process-specific host-cell protein (HCP) kit.

Spike-Recovery Optimization: Our process-specific HCP kit used the same assay design and setting as the commercial HCP kit (Cygnus F550-1), for which detection HCP antibodies, samples, and standards were incubated spontaneously with HCP antibodies on a coated microtiter plate. During method qualification of the process-specific HCP plate, wells located farther away from HCP standards gave lower spike recoveries (Table 3). When those samples were moved closer to HCP standards, measured spike recoveries returned to acceptable levels, suggesting that the lower spike recovery (assay shift) was independent of sample type and matrices but instead was associated with well positions and timing of loading the HCP standards and samples to assay plates. The pattern occurred both in our development facility and at commercial QC laboratories, prompting us to find root causes and a solution to resolve that common issue associated with high invalid rates in HCP testing. We suspected prolonged single-channel pipetting and sample transferring to be root causes.

Table 3: Assay shift and low spike recovery from single-channel sample loading; AEX = anion exchange, CEX = cation exchange, DS = drug substance, HCP = host-cell protein, VI = viral inactivation.

Plate Format: Samples were loaded to assay plates from left to right and top to bottom. Lower spike recoveries frequently were detected at the right and bottom sides of the assay plates.

A dilution plate was used to load all HCP standards, controls, and samples first before transfer by a multichannel pipette to a HCP-reaction plate preincubated with 100 μL of detection antibody. Multichannel transfer cut down pipetting time from 10–15 minutes to 2–3 minutes, significantly improving spike recovery within acceptable ranges and eliminating poor spike recoveries associated with plate format (Table 4 and Figure 1). HCP-testing optimization for several mAb products improved assay valid rate and accuracy.

Figure 1: Recovery trending before and after introduction of intermediate dilution plate and multichannel pipetting; the blue trends represent recovery-value percentages, whereas orange trends represent standard deviations within the plates.

Table 4: Multichannel sample loading improved spike recovery; AEX = anion exchange, DS = drug substance, HCP = host-cell protein, VI = viral inactivation.

Protein A Interference: HCPs and residual protein A are process-derived impurities both independent of and nonrelevant to each other during HCP testing. When protein A molecules were spiked into DS samples, positive HCP values were detected — a new finding that, to our knowledge, had not been reported in literature. Such reactivity or interference to HCP assays seemed to be related to DS types. Because of the extensive use of protein A resin for mAb purification, it is important to understand whether and how protein A molecules interfere with HCP assays and to provide an awareness to scientists who test the two impurities per day. Here in our troubleshooting experiments, we spiked three common recombinant protein A resins — MabSelect SuRe (Cytiva), Amsphere A3 (JSR Life Sciences), and MabSelect PrismA (Cytiva) — into one human immunoglobulin G (IgG) product and two variable heavy chain (VHH) products.

Results demonstrated significant interference of protein A impurities to HCP assays in VHH products, but no protein A interference was observed in the presence of the IgG product (Table 5). Biomolecule type showed a slight difference in its contribution to HCP testing, but product type determined if protein A molecules acted as HCP-like impurities in the assay.

Table 5: Interference of protein A molecules to host-cell protein (HCP) assay; DS = drug substance, IgG = immunoglobulin G, VHH = variable heavy chain domain.

Data trending and risk assessment demonstrated low protein A levels in our process intermediates and DSs. Percentages of HCP values for protein A impurity were theoretically below the limit of detection for HCP assays and therefore considered low risk for product quality. However, we still explored solutions to eliminate interference from protein A impurity to HCP assays.

At that point, the data confirmed our hypothesis that protein A molecules have multiple IgG-binding domains and can build bridges between HCP-capture antibodies (rabbit) and detection antibodies (rabbit), thereby generating false-positive HCP values during testing (Figure 2). Human IgG products bind strongly with protein A molecules, neutralizing and eliminating further binding with both HCP-capture and detection antibodies. VHH products, however, bind weakly with protein A impurities and fail to compete with high-affinity assay IgG molecules, thus generating false-positive HCP values.

Figure 2: The protein A (ProA) molecule bridges two host-cell protein (HCP) antibodies (Abs) in the absence of HCP molecules (left) and generates false-positive HCP values (middle). Nonspecific rabbit immunoglobulin G (IgG) neutralizes ProA molecules and eliminates interference (right); Ab = antibody, HRP = horseradish peroxidase.

Protein A spikes into assay buffers (MabSelect SuRe F610 kit and MabSelect PrismA 29707299 kit, both from Cytiva) confirmed interference in our rabbit-derived process-specific HCP assay (Table 6). Note that protein A binds poorly with goat and sheep IgGs, so interference was not detected in a goat-derived HCP assay (Cygnus F550-1 HCP kit) (data not shown). To resolve protein A interference issues, we incubated samples with a nonspecific rabbit IgG and demonstrated a complete abolishment of protein A interference and false-positive HCP values, especially from VHH products (Table 7). No significant HCP values were measured from protein A spikes in reference to unspiked DS samples.

Table 6: Protein A in buffer contributed to positive host-cell protein (HCP) measurements in a process-specific HCP assay.

Table 7: Blocking of protein A interference by nonspecific rabbit immunoglobulin G (IgG) in host-cell protein (HCP) assays; mAb = monoclonal antibody, VHH = variable heavy chain domain.

Conclusion

We report herein three case studies in which assay optimizations successfully resolved HCP-testing issues. We demonstrated improved sample-dilution linearity by introducing a process-specific HCP kit and improved HCP-assay spike recoveries with multichannel pipetting for sample loading. Protein A molecules and HCPs are considered to be two nonrelevant impurities from the biomanufacturing industry’s testing perspective. However, our discovery of residual protein A interference to HCP-impurity testing can build awareness and enable risk assessment for HCP-testing strategies. Considering that the HCP ELISA is a major analytical assay for impurity control and patient safety of biologics, cell and gene therapies, and vaccines, we hope that our work can help to support and guide daily HCP testing and troubleshooting.

Acknowledgments

We thank Rahul Godawat, Siguang Sui, and Rohan Patil for process support and Scott Umlauf and Jon Borman for HCP knowledge-sharing and support. David Courchesne provided sample management. Dan Su and Eugenie Cheng provided laboratory coordination and training.

References

1 Yasuno K, et al. Host Cell Proteins Induce Inflammation and Immunogenicity As Adjuvants in an Integrated Analysis of In Vivo and In Vitro Assay Systems. J. Pharmacol. Toxicol. Methods 103, 2020: 106694; https://doi.org/10.1016/j.vascn.2020.106694.

2 Li X, et al. Identification and Characterization of a Residual Host Cell Protein Hexosaminidase B Associated with N-Glycan Degradation During the Stability Study of a Therapeutic Recombinant Monoclonal Antibody Product. Biotechnol. Prog. 37(3) 2021: e3128; https://doi.org/10.1002/btpr.3128.

3 Jones M, et al. “High-Risk” Host Cell Proteins (HCPs): A Multi-Company Collaborative View. Biotechnol. Bioeng. 118(8) 2021: 2870–2885; https://doi.org/10.1002/bit.27808.

4 Singh SK, et al. Understanding the Mechanism of Copurification of “Difficult To Remove” Host Cell Proteins in Rituximab Biosimilar Products. Biotechnol. Prog. 36(2) 2020: e2936; https://doi.org/10.1002/btpr.2936

5 USP Chapter <1132>. Residual Host Cell Protein Measurement in Biopharmaceuticals. United States Pharmacopeial Convention: Rockville, MD, 2016.

6 Yao Z, Wang W, Zhou W. Current Trends in Host Cell Protein Detection for Biologics Manufacturing. BioPharm Int. 36(2) 2023: 30–34; https://biopharminternational.com/view/current-trends-in-host-cell-protein-detection-for-biologics-manufacturing.

7 Petrovic S, et al. Host-Cell Protein Analysis To Support Downstream Process Development: A High-Throughput Platform with Automated Sample Preparation. BioProcess Int. 17(11–12) 2019: 32–49; https://www.bioprocessintl.com/process-development/host-cell-protein-analysis-to-support-downstream-process-development-a-high-throughput-platform-with-automated-sample-preparation.

8 Voigtmann M, Foettinger-Vacha A. Using Automated Immunoassays for HCP Analysis in Early Bioprocess Development. BioProcess Int. 21(11–12) 2023: 20–23; https://www.bioprocessintl.com/separation-purification/using-automated-immunoassays-for-hcp-analysis-in-early-bioprocess-development.

9 Van Manen-Brush K, et al. Improving Chinese Hamster Ovary Host Cell Protein ELISA Using Ella®: An Automated Microfluidic Platform. Biotechniques 69(3) 2020: 186–192; https://doi.org/10.2144/btn-2020-0074.

Bulat R. Ramazanov, Shenjiang Yu, Khushboo Kapadia, Daria V. Sizova, Helen SooHoo, Emily H. Sciortino, Rebecca K. Phillips, Samantha R. Cote, and Tony Dang, all work in analytical development and quality control at Alexion Pharmaceutical Development and Clinical Supply; 100 College Street, New Haven, CT 06510. Formerly of Alexion, Bing Hu is executive bioassay director at Eli Lilly and Company (Indianapolis, IN); [email protected].