Multispecific Analyses of Biologics Pose Multiple Challenges

Multispecific antibodies are of interest to drug developers because they provide an avenue for achieving more effective or safer treatments for a wider range of diseases. Those attributes result from their ability to simultaneously target or engage multiple protein targets. Unfortunately, a consequence of this enhanced capability is greater structural complexity, which is often accompanied by greater numbers of charge and structural variants and potentially a wider range of post-translational modifications (PTMs). Drug developers need to obtain deep understanding of these highly complex biomolecules as early as possible in the drug development process to avoid investing time and cost into nonviable candidates. Analytical techniques that provide better information, earlier, are needed to facilitate multispecific drug development.

Greater complexity than traditional monoclonal antibodies

Multispecifics are made up of similar domain structures as monoclonal antibodies (mAbs). They are more complex, however, and that complexity is increasing as antibody developers seek to tailor binding and half-life to specific biological requirements while also addressing steric considerations for multiple target engagement, according to Donmienne Leung, head of protein engineering at Absolute Antibody.

The challenge, says Susan Darling, senior director of capillary electrophoresis and biopharma product management and marketing at SCIEX, is that there are many more variants associated with these complex structures.

“Standard mAbs have identical heavy and light chains, but there are charge and sequence variants and many different post-translational modifications that afford many different isoforms. For multispecifics, the number of possible structural variants is greater and depends on how many different components are being combined together. Each one of those structure variants has its own range of isoforms. That makes the analysis of multispecifics exponentially more complicated,” she explains.

If multispecifics are produced with stochastic expression systems, inappropriate association of the different chains may occur. That makes even size assays for identifying intact mAbs inapplicable, because the different structural combinations may have the same size, but only the correctly assembled molecule will exhibit the desired activity. All of the rest are considered impurities. In addition, the greater structural complexity of multispecifics may result in a greater tendency to misfold and aggregate, comments Leung.

Furthermore, Darling notes that even if a strategy, such as the “knob-in-hole” approach, is used to afford multispecifics with the correct structures, the bioengineering process typically introduces more attributes to these proteins, such as different modifications, that make these biologics more complex not only structurally but from an affinity perspective.

The increased number of undesired side products may potentially make purification of the requisite therapeutic by traditional chromatography techniques more difficult, adds Leung. “Multiple chromatography steps and mass spectrometry (MS) become necessary for full characterization of the multispecific. The evaluation of multiple target binding will also be less straightforward when the geometry of the multispecifics and accessibility of each binding domain can be highly dependent on assay formats,” she states.

Similar characterization needs, plus much more

That greater complexity means that for multispecifics, all of the characterization and other assays required for monospecifics are required, plus many more. Specifically, it is crucial to confirm the correct folding and intramolecular pairing of various domains. As an example, Leung highlights the need to confirm the correct pairing of heavy chain to light chain for each specific antibody arm to retain affinity to the required target of interest.

“We can no longer assume that two heavy and two light chains will come together in a symmetrical molecule,” Darling agrees. “Many different combinations with minimal mass differences can be generated depending on whether the platform involves stochastic expression or not,” she says.

In addition, while the mispairing of both the heavy and light chains is often of critical importance for the potency and safety of mAb-based bispecific formats, it may not be a factor for other formats, observes James Graham, director of protein and process analytics within the Bio R&D group at Lonza. “Even though similar methods can be used, the range of multispecific formats means that ‘new’ attributes are specific to the individual format for the protein rather than broadly relevant for all multispecifics,” he notes.

Multispecifics built using Fc and antibody fragments, Leung continues, can be purified and analyzed similarly to traditional antibodies and proteins using affinity chromatography (Protein A, nickel, etc.). “The purity and homogeneity of multispecifics can be achieved and evaluated by a combination of chromatography (e.g., hydrophobic interaction, ion exchange, and size exclusion) and MS similar to traditional biologics, while PTMs on recombinant proteins and multispecifics are typically characterized by mass spectrometry,” she says.

Assumptions made for immunoglobulin (IgG) monospecifics don’t apply to multispecifics, Darling adds. “The assumption for monospecific IgGs that there will be Fc and target binding affinities is also no longer applicable. Two or three different binding affinities can be competitive against different targets,” she contends. In addition, she notes that there are often modifications in multispecifics that are not observed for mAbs because of the way the cell lines used to produce these biomolecules are engineered. There can be additional PTMs as well, and in particular more variability in disulfide bonds.

It is important to remember, though, Graham underscores, that mAbs themselves are already multispecifics, so there already is a good baseline in terms of knowing what needs to be monitored. “In addition to established attributes we monitor during mAb development (e.g., safety, purity, potency, and the range of product variants such as glycosylation), it is the specific modifications that have been made from the established mAb framework that need special attention. These may vary substantially depending on the protein architecture,” he concludes.

All of these variations, Darling asserts, can impact binding affinities and thereby make many other assumptions about monospecific IgGs also nonrelevant for multispecifics. “There can be different cross-reactivities, different catabolite forms, and different isoforms generated during the biotransformation process that can all have modified binding capacities. Simple ligand binding assays are no longer sufficient,” she says.

Leung also stresses that in addition to all of these issues, it is important to remember that multispecifics must be monomeric, soluble, and stable in nature to enable clinical development and downstream commercialization.

Need to manage greater variability

From a characterization standpoint, the big challenge for multispecifics is the increased possibility for greater higher-order-structure variability, according to Darling. “The molecular asymmetry and potential for multiple product-related impurities based on the combinations of chains being expressed is the source of this variability,” Graham notes.

The fact that protein structures, including multispecifics, when in solution are dynamic in nature with flexible structural isomers and potentially aggregated molecules further complicates bioanalysis, according to Leung. “Temperature can also influence the overall stability of the correctly folded molecule and the respective domains. Moreover, the surface interactions between molecules and between domains may have significant impact on self-association,” she comments.

Where there are significant molecular weight differences between potential product variants, Graham adds that the assessment of aggregation can become challenging because a high-molecular-weight species could represent a protein aggregate or a mispairing variant.

This variability can also create hurdles during process optimization and scale up once a multispecific product is defined and an investigational new drug application (IND) has been filed. “With multispecifics, there simply are more chances because of the higher variability that any changes to the process will lead to the formation of the wrong structures and/or isoforms. That means the risk is greater. It also means that many more samples must be evaluated in a way that provides detailed information very quickly,” Darling says.

It is generally necessary, observes Darling, to use a combination of intact mass measurements and characterization with reductive techniques such as digestion prior to subunit analysis or peptide-level analysis to identify disulfide bonds.

But more understanding wanted sooner

A key trend in the biopharma industry affecting all drug development is the desire to compress development timelines down to just a few years. “Companies are looking to get development times down to 1000 days; they want to develop everything at pandemic speed, even when they have diverse portfolios that include highly complex next-generation products such as multispecifics, gene and cell therapies, and RNA-based treatments, all of which require new and different analytical capabilities,” Darling asserts.

The result is a unique situation in which a usually risk-averse industry, because it is rapidly moving to more complicated molecules, needs access to newer, perhaps initially slightly more experimental technologies. “The industry is really driving to be creative, not just in terms of analytical solutions, but in how they manage data and pull it together to knock the inefficiencies out of every step,” Darling concludes. “Our customers are challenging us to develop new analytical techniques that provide more information more quickly, and when we bring them new solutions, they are adopting them at a much faster rate than could ever be imagined even a decade ago,” she says.

Failing fast a major strategy

The reason for this transformation within the industry is a new focus on the developability of biologics. “Developability is increasingly critical for downstream commercialization,” Leung states. Darling agrees. She says that the strategy of failing fast and early has been increasingly adopted at an accelerating rate over the past few years, with companies looking for technologies that enable more detailed characterization and stability assessments early on.

“Simply selecting the highest-yielding cell line without considering the product quality is no longer the typical approach. Developers want to find the right cell line and clone that produces a high yield of the desired product—mAb or multispecific—with the right structure and modifications from the outset to avoid costly late-stage failures,” Darling explains.

For instance, Leung points to the use of various in silico and pre-clinical screening techniques for the evaluation of non-specific binding, self-association, and chemical stability as means for identifying problematic molecules early on. She also notes that while sequence analysis and in vitro assays have been recognized for the pre-clinical prediction of immunogenicity risks associated with traditional mAbs, these techniques have not been correlated for multispecifics with confidence.

Developers are also looking for high-throughput, highly informative, easy-to-use analytical solutions that support this strategy. For multispecifics, one of the challenges is the inability to apply platform analytical strategies because of the variety of formats. Product-specific methods become critical, and these methods are required early in development, according to Graham. “As well as ensuring that method performance is sufficient to monitor the attribute in question, throughput and turnaround times must be compatible with screening stages during cell line and process development,” he says.

Orthogonal approach to analysis essential

Comprehensive characterization of multispecifics, because these biomolecules are so complex, generally requires the use of several orthogonal methods. In addition to traditional antibody assays, many other methods are being explored.

“There are many different attributes of multispecifics that need to be determined, and because this type of biologic is relatively new, there is still work to be done to determine which methods are best,” Darling observes. There is a need, she adds, for iteration and customization of technologies to determine which will be most impactful and robust. As a result, multispecific developers are leveraging orthogonality, using multiple assays to confirm results and build more confidence in their data.

For instance, in addition to leveraging established techniques such as ultra-high performance liquid chromatography (UHPLC), capillary electrophoresis (CE), surface plasmon resonance (SPR), MS, and automated cell-based analysis platforms, techniques that can easily identify peaks within a profile are crucial, according to Graham. Native MS is one such technology that can be used to generate peak identity information in parallel to the primary analysis method, rather than having to perform subsequent follow-on testing. “Native MS is particularly interesting due to the information density of the technique and the fact that non-covalent variants or impurities may be present for some multispecific formats that can be challenging to characterize using other methods,” he remarks.

In general, adds Leung, a combination of high-resolution LC and MS techniques are necessary to fully characterize multispecifics. Thermal stability analyses such as differential scanning calorimetry (DSC) and differential scanning fluorimetry (DSF) allow developers to select for better-behaved molecules. CE and other electrophoretic techniques such as isoelectric focusing (IEF) and capillary IEF (cIEF) have improved resolution of traditional sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) methods in elucidating the correct pairing of polypeptide chains in multispecifics.

Intact mass analysis using Orbitrap-based LC-MS can resolve all theoretically expected species in a complex mixture, while using electrospray ionization time-of-flight (ESI-TOF) MS on quadrupole instruments enables quantification of various species with improved sensitivity and robustness.

While target binding can be confirmed using simple enzyme linked immunosorbent assay (ELISA) methods, biochemical characterization of multispecifics requires detailed kinetic analysis to compare the integrity of the final product. Association and dissociation phases of molecular interactions can be measured by established techniques including SPR, biolayer interferometry (BLI), and isothermal titration calorimetry (ITC).

Ultimately, though, Leung concludes that cell-based assays are necessary to confirm target engagement and the desired functional activity of the multispecific.

With so many options for analyses to perform, it can be quite challenging to determine which methods should be used for a given multispecific. Lonza has found that having a template library of different analytical methods based on the chemistry of the therapeutic protein provides a starting point for method development. “When combined with a design-of-experiment approach, development cycle for the deployment of a new method can be greatly streamlined and development times reduced from weeks to days or less!” Graham avers.

New tools for accelerating multispecific development

With an overall goal of accelerating biopharma development, SCIEX has focused on meeting the needs for easier-to-use MS solutions that provide rapid, detailed results. In 2021, the company introduced the accurate mass ZenoTOF 7600 system, which leverages both Zeno trap with electron activated dissociation (EAD) fragmentation technology.

In traditional TOF instruments, Darling explains, the fragment ions arrive at the pulser region with a spread of velocities, so only a small percentage of the ion population gets accelerated into the flight tube. This duty cycle problem is overcome because the Zeno trap enables all of the ions to arrive at the TOF pulser at the same time independent of their masses. More than 90% of the ions are injected, resulting in significantly improved MS/MS ion sampling for all m/z and greater sensitivity for MS/MS and high-resolution multi-reaction monitoring (MRM) acquisition scan modes and greater ion fragment coverage.

EAD, meanwhile, provides for generation of electrons with tunable velocities and allows for different fragmentation mechanisms, including the hot ECD mechanisms ideal for peptide mapping of glycopeptides and disulfide-bonded peptides, according to Darling.

Importantly, EAD preserves valuable side chain information, such as for labile PTMs very common in multispecifics, that enables confirmation of various isomers and affords excellent MS/MS coverage for all charge states in one experiment. “We are finding real interest in this technology because it allows for much easier and more robust characterization of even highly complex molecules. In fact, for the first time it could be possible that electron dissociation could become the go-to-choice for fragmentation,” Darling asserts.

Another advance in MS technology is in the works at SCIEX and is based on the direct coupling of cIEF with MS. Initially developed by IntaBio, which SCIEX acquired in early 2021, the system enables high-resolution separation of charge variants combined with detailed MS data for each variant. “The combination of isoelectric point directly coupled with MS data allows for confident identity assignment during intact analysis in much less time than is required for conventional peptide mapping analyses,” Darling notes. The company expects to launch the system sometime in 2022.

On the CE front, SCIEX has introduced in 2021 a new system that allows for much higher throughput analysis. “The BioPhase 8800 system leverages a new cartridge that allows parallel processing of eight different CE samples (same or different molecules) simultaneously using CE-SDS and/or cIEF with UV or LIF detection, providing all the advantages of CE while delivering consistent, accurate results so that more samples can be analyzed in less time,” Darling explains.

Importantly, methods developed for SCIEX’s widely used PA 800 system can be transferred to the BioPhase 8800, avoiding any need to do bridging studies as projects move to late-stage development and commercialization. “That can save weeks to months of time,” Darling says. “In addition,” she emphasizes, “companies can rapidly perform design-of-experiment studies to quickly optimize both assays and manufacturing processes.”

Automation on the front and back end

More complex molecules and the need to perform more orthogonal analyses means that more samples must be prepared and much larger data sets are generated that must be analyzed. Automation of both of these aspects of multispecific analysis is becoming another crucial strategy for speeding up the development process. “Biopharma companies are looking for any opportunities to remove unnecessary manual processing, whether with respect to sample preparation or data processing,” comments Darling. That includes investing in more specific software products, many of which leverage artificial intelligence and machine learning, to help with data integration.

While standardizing assays for multispecifics is not possible given the unique nature of each of these biologics, automation of sample preparation and data acquisition and analysis can help simplify these solutions, Darling insists. High-throughput data analysis platforms, agrees Graham, are critical to the deployment of newer analytical methods for multispecifics into process development, because otherwise the data analysis time quickly dominates the overall time to result.

Interesting but not a major risk

The challenges associated with multispecific analysis, while real, are not so great that the bioanalytical experts with experience in biologics development cannot manage them. “Multispecific analysis is definitely an interesting field, but it should not be considered a major risk,” Graham asserts. “Rather, analytical development for these complex biomolecules should be approached with a structured but flexible workflow combined with a good understanding of each molecule’s chemistry,” he concludes. Having multiple answers on hand for multiple multispecifics works!

About the author

Cynthia A. Challener, PhD, is a contributing editor to Pharmaceutical Technology.