Host Cell Protein: Minimizing Impurities for Safe Biologics

Host cell proteins are unwanted molecular hitchhikers that tag along when manufacturers produce biologic drugs like monoclonal antibodies, vaccines, and gene therapies. Every biologic is made inside living cells, and those cells shed thousands of their own proteins into the production mix. Even after extensive purification, trace amounts of these host cell proteins (HCPs) can remain in the final drug product, where they risk triggering immune reactions, degrading the drug itself, or shortening its shelf life.¹ Controlling them is one of the trickiest problems in biologics manufacturing, and the strategies for doing so span everything from genetically engineering the production cells to deploying machine-learning models that predict which impurities will be hardest to remove.

Why Host Cell Proteins Are a Safety Concern

When you inject a biologic drug, your immune system encounters not just the therapeutic protein but any stowaway HCPs that made it through purification. Some of those impurities can provoke an immune response, potentially causing anything from mild injection-site reactions to antibodies that neutralize the drug itself. The risk is not uniform across all HCPs. Research using computational immunology tools has shown that HCPs whose amino acid sequences are less similar to their human counterparts tend to be more immunogenic, because the immune system recognizes them as foreign more readily.² An HCP that closely resembles a human protein might slip under the radar, while one that looks distinctly non-human is more likely to set off alarms.

Beyond immune reactions, certain HCPs carry enzymatic activity that can directly damage the drug product. A good example involves polysorbates, the detergent-like stabilizers added to many antibody formulations to prevent the drug from clumping. Hydrolytic HCPs from Chinese hamster ovary (CHO) cells, the workhorse production system for most antibody drugs, can break down polysorbates over time. That degradation leads to visible particles forming in the drug vial and a shorter usable shelf life.³ In one documented case, a residual enzyme called hexosaminidase B chewed through the sugar structures on a monoclonal antibody during stability testing, altering the drug’s molecular profile under accelerated storage conditions.

Not every HCP poses an equal threat, though. A study evaluating residual HCPs from rice-endosperm-produced human serum albumin found low immunogenicity even at doses equivalent to 25 times the clinical amount, suggesting that the choice of host organism matters and that plant-based systems can produce relatively benign HCP profiles.⁴ The takeaway is that HCP risk is not one-size-fits-all. It depends on the individual protein’s identity, its enzymatic activity, and how closely it resembles human proteins.

How HCPs Survive Purification

Most biologics go through multiple purification steps designed to strip away impurities, yet a stubborn minority of HCPs persist all the way to the final product. Understanding why they persist is essential to designing better removal strategies. There are two main culprits: direct binding to the drug itself, and indirect association through DNA and chromatin complexes.

The direct-binding route, often called “hitchhiking,” occurs when an HCP physically sticks to the therapeutic antibody. Recent quantitative work measuring these interactions found that problematic HCPs bind to antibodies with relatively high affinity but fast-on/fast-off kinetics, with binding strengths in the low nanomolar range for the strongest binders. When researchers ran a model HCP called PLBL2 through a standard Protein A purification column with two different antibodies, the amount of PLBL2 that carried through correlated directly with how strongly it interacted with each antibody.⁵ Some of these hitchhikers bind right at the antibody’s antigen-binding site, making them especially hard to detect because they are essentially shielded by the drug molecule itself.⁶

The indirect route involves DNA and chromatin fragments from ruptured cells. These negatively charged complexes can act as a bridge, pulling HCPs along through the Protein A capture step. In experiments across three different biosimilar antibodies, researchers found that reducing DNA in the material loaded onto the Protein A column consistently lowered HCP levels in the output, while reducing HCP levels before the column did not meaningfully improve the column’s own clearance performance.⁷ In practical terms, this means that tackling DNA contamination early in the process can be more effective than trying to remove HCPs directly at the capture stage.

The combined effect of these persistence mechanisms means that a handful of repeat offenders show up across many different antibody products. Large-scale profiling studies have identified groups of common, persistent HCPs in antibody manufacturing, with both high abundance and direct drug interaction contributing to which proteins survive.⁸

Engineering the Production Cells

One increasingly popular strategy is to stop problematic HCPs from being produced in the first place. Using gene-editing tools, manufacturers can knock out the genes responsible for the most troublesome impurities directly in the production cell line. Researchers have successfully eliminated genes for two well-known problem proteins, Anxa2 and Ctsd, from CHO cells, confirming the complete absence of these HCPs in the resulting cell lysates without harming cell growth or productivity.⁹

A more ambitious version of this approach targets the polysorbate-degrading enzymes that cause formulation instability. One team removed nine different hydrolase genes from CHO cells simultaneously. In the unmodified parent cell line, about 70% of the polysorbate in culture supernatant was degraded over 14 days. In the multi-knockout line, less than 25% degradation occurred over the same period, a dramatic improvement that could translate directly into longer shelf life for the finished drug.¹⁰ These engineered cell lines are still being refined, but the approach represents a fundamental shift from relying solely on downstream purification to addressing the problem at its source.

Downstream Purification Strategies

Even with cleaner cell lines, downstream purification remains the primary workhorse for HCP removal. The standard antibody purification train starts with Protein A affinity chromatography, which captures the antibody and washes away the bulk of impurities, followed by one or two polishing steps that mop up whatever remains. But optimizing those polishing steps is where much of the innovation is happening.

Before the formal chromatography train begins, harvest clarification can make a real difference. A flocculation process using a stimulus-responsive polymer called benzylated poly(allylamine), followed by depth filtration, has shown efficient reduction of HCPs, DNA, and aggregates across multiple antibody products while maintaining high yield.¹¹ By clumping impurities together early, this approach reduces the burden on the expensive chromatography steps that follow.

In the polishing stages, anion-exchange chromatography in flow-through mode is standard: the antibody passes through while negatively charged impurities stick. But the performance varies considerably depending on the medium used. A comparative study of three anion-exchange media found that membrane-based devices could reduce HCP levels from around 8,000 parts per million down to as low as 10 ppm at competitive loading densities, though the exact performance depends on the specific product and conditions.¹² Meanwhile, newer multimodal chromatography resins, which combine different binding chemistries on a single bead, offer another lever. One study found that optimizing the hydrophobic component of the resin ligand and using larger-pore base beads could reduce HCP levels by three orders of magnitude under high antibody loading.¹³

Measuring What You Cannot Easily See

You can only control what you can measure, and measuring HCPs is surprisingly difficult. The traditional workhorse assay, ELISA, uses polyclonal antibodies raised against HCPs to capture and quantify them. But this approach has a fundamental blind spot: if the polyclonal antibody reagent does not recognize a particular HCP, that impurity goes undetected. Coverage gaps are a known and persistent issue.

One approach to evaluating coverage involves using the ELISA antibodies to capture HCPs from a sample, then identifying what was captured (and what was missed) using mass spectrometry. Several groups have developed variations on this theme, and the combined ELISA-mass spectrometry workflow has become an important quality check during assay development.¹⁴ A quantitative version of this approach not only identifies which HCPs the ELISA misses but also estimates how much of each missed protein is present, enabling more informed risk assessment.¹⁵

Researchers have also asked whether using ELISA antibodies raised in multiple different animal species could improve coverage. The intuitive logic is appealing: different host animals might recognize different HCPs. But a direct comparison found that mixing antibodies from five host species did not substantially improve ELISA performance over single-species reagents, while raising both the cost and the ethical burden of antibody production.¹⁶

Where ELISA tells you “how much,” mass spectrometry tells you “what.” Tandem mass spectrometry can identify individual HCPs by name, even at very low concentrations, and this capability has become central to modern HCP risk assessment.¹⁷ Newer instruments and methods are pushing detection limits further. An analysis using the Orbitrap Astral mass spectrometer identified 236 individual HCPs across therapeutic antibody samples, with about 55% of those quantifiable at meaningful confidence levels.¹⁸ A separate “deep field scan” method, which uses iterative data acquisition to avoid repeatedly measuring the same abundant peptides, improves the detection of low-level HCPs without requiring sample enrichment or cleanup steps.¹⁹

Capillary western blots represent another orthogonal approach. Unlike traditional slab-gel westerns, capillary versions are semi-automated and quantitative. In a head-to-head comparison during COVID-19 vaccine production in Vero cells, capillary western results agreed closely with SDS-PAGE-based measurements, while ELISA readings were outliers, suggesting capillary westerns may sometimes get closer to the true HCP concentration.²⁰ Capillary zone electrophoresis coupled with mass spectrometry has also shown promise, generating roughly five times higher signal intensity for low-level spiked proteins compared to conventional liquid chromatography approaches.²¹

Predicting Which Impurities Will Be Hardest to Remove

Running experiments to optimize the removal of every possible HCP is expensive and time-consuming. Computational models are starting to fill some of that gap. Quantitative structure-property relationship models, which predict a protein’s behavior during chromatography based on its three-dimensional structure, have been trained to forecast how E. coli HCPs will behave during ion-exchange chromatography. A relatively simple model using just two features, the protein’s isoelectric point and the sum of its negative surface electrostatics, was able to predict the retention times of 288 HCPs with an error of 5% or less.²² A complementary study found that focusing on monomeric proteins improved predictive accuracy, with the model’s goodness of fit reaching 0.70.²³

Similar modeling has been applied to plant-derived biologics. Researchers reconstructed three-dimensional structures of over 100 tobacco host cell proteins and used those models to predict which would stick to different chromatography resins, then validated the predictions experimentally.²⁴ While no model is perfect, these tools can narrow the experimental search space considerably, especially when dealing with host systems where less HCP characterization data exists.

Different Hosts, Different Problems

CHO cells dominate antibody manufacturing, but E. coli, yeast, plant cells, and insect cells are all used for various biologics. The HCP challenge looks different depending on the host. A telling comparison between CHO and E. coli systems found that the HCP population changes substantially as a product moves through purification in E. coli, to the point that an ELISA kit developed from the initial harvest material significantly underestimated HCP levels in later purification intermediates. Supplementing the ELISA with antibodies raised specifically against the intermediate-stage material improved detection. By contrast, CHO-derived products showed no significant shift in HCP composition between harvest and the equivalent purification stage.²⁵ The practical lesson: analytical strategies that work for CHO may fail for bacterial systems, and each host type needs its own tailored approach to HCP monitoring.

Even within a single CHO cell culture, the HCP landscape is not static. A long-term study monitoring the extracellular HCP profile over 500 days of continuous culture found that 92 HCPs showed expression changes of up to 48-fold, and 34 of those had already been flagged as difficult to purify. This kind of drift matters especially for continuous manufacturing processes, where cells are maintained for extended periods rather than being regrown fresh for each batch.

Gene Therapy Vectors Bring Their Own Complications

The HCP challenge takes on a different character in gene therapy manufacturing, particularly for adeno-associated virus (AAV) vectors. The viral capsid proteins that make up the therapeutic product are enormously abundant relative to the trace HCPs, creating a dynamic range problem that makes detection extremely difficult. Standard mass spectrometry sample preparation tends to be overwhelmed by capsid-derived peptides, drowning out the signal from low-level impurities.

A systematic comparison of sample preparation methods for AAV products found that a “native” digestion strategy, which keeps the capsid intact while selectively digesting exposed HCPs, dramatically improved detection. The optimized workflow achieved the deepest host-cell proteome coverage, particularly in highly purified drug substance samples where the dynamic range challenge is greatest.²⁶ As gene therapies move from rare-disease niches toward broader use, refining these analytical methods is becoming urgent.

Industry Alignment and Regulatory Expectations

There is no universally mandated HCP limit for biologic drugs. Regulatory agencies like the FDA and EMA expect manufacturers to demonstrate that residual HCPs are reduced to levels that do not compromise patient safety, but the acceptable threshold varies by product, route of administration, and dosing frequency. A recent industry-wide survey captured current default limits being used for both total and individual HCP impurities across companies, along with approaches for assessing immunogenicity risk and feedback from global health authorities.²⁷ The survey highlights an ongoing push toward harmonization, but for now, companies largely set their own internal specifications based on product-specific risk assessments.

This ambiguity can feel unsatisfying, but it reflects a genuine scientific reality: the risk of a given HCP depends on what it is, not just how much of it is present. A hundred parts per million of a harmless structural protein is very different from ten parts per million of an active lipase that degrades your formulation. Regulators increasingly expect companies to identify the specific HCPs in their products and assess the risk of each one individually, rather than relying on a single total-HCP number as a blanket measure of purity. The analytical and cell-engineering advances described above are what make that kind of granular risk assessment feasible in the first place.