Advanced bioanalytical technologies accelerate drug development in the new age of protein therapeutics

In 1986, the FDA approved a new transplant rejection drug that was derived from an intact monoclonal antibody (mAb).¹ Although it would be almost a decade before a second mAb drug was approved, this event was the first small step in what has become a foundational shift in the pharmaceuticals industry. From the mid-1990’s, FDA approvals of mAb drugs accelerated steadily, and development pipelines have become increasingly dominated by a widening array of mAb-based candidates. As of 2016, there were over 60 approved mAb-based drugs with a total market of approximately $89 billion per year.¹ It has been estimated that there are more than 550 antibodies in clinical development, approximately 50 of which are in phase III trials.¹ The market forces driving the shift towards mAbs are clear: While 90% of prescriptions in the US may be for small molecule drugs, four of the top five earning drugs are mAbs (or mAb-based protein therapeutics) including repeated top-earning drug Humira (Adalimumab) which grossed $18.43 billion in 2017, even while technically off-patent.² The therapeutic driving forces are equally clear: mAbs use the uniquely high specificity of protein-protein interactions to tightly and unerringly bind their targets, even in the exceedingly complex molecular environment of the cell. This can have the effect of greatly improving therapeutic power while drastically reducing off-target effects, which no doubt contributes to the relatively high success rate for mAbs drugs, more than 15% of which survive from first-in-human clinical trials through to approval.¹

This is not to suggest that mAb drug development is any less challenging than for small molecule drugs. In fact, there are aspects of development, particularly in the pre-clinical phases, where working with biologically manufactured macromolecules can present a broad set of unique challenges. Among these is the very real struggle to determine exactly what the biologically active form of the molecule is, or even if there is a singular form that is responsible for the majority of the activity. While molecular characterization is usually straightforward for small molecules, protein therapeutics have complex, dynamic 3D structures that dictate biological activity, clearance rate and immunogenicity. Furthermore, protein therapeutics are often covalently modified in a manner that is non-uniform and often exceedingly challenging to control during biomanufacturing. The result is invariably a distribution of drug molecules whose on- and off-target effects are determined by the properties of the distribution rather than any one species. A key factor in overcoming these challenges is the application of powerful new bioanalytical tools that can rapidly provide a detailed picture of candidate protein therapeutics, their molecular, structural and dynamic distributions and their molecular mechanisms of action.³

The ‘Technology-Enhanced Biopharmaceuticals Development and Manufacturing’ program (TBio-DM), led by the Wilson group at York University, is providing leadership in the translation of advanced bioanalytical techniques to drug development. This $1.6M initiative is funded by the NSERC Collaborative Research and Development program, together with industrial partners SCIEX, Fluidigm Canada and Sanofi Pasteur. Now early in it’s third year, TBio-DM has already delivered on it’s principal aim of accelerating drug development through the application of new technologies in ‘real-world’ projects, often with a direct and substantial impact on business trajectories. Below are three narratives from TBio-DM associated projects and their impact on drug development.

“As of 2016, there were over 60 approved mAb-based drugs with a total market of approximately $89 billion per year.¹ It has been estimated that there are more than 550 antibodies in clinical development, approximately 50 of which are in phase III trials.^1″

Rapid Characterization of an Avastin Biosimilar. The shift to protein therapeutics has created additional challenges for generic drug makers, who must now demonstrate biosimilarity for molecules whose structure and activity can be highly sensitive even to minute changes in manufacturing processes, and for which it can be a struggle to demonstrate molecular mechanisms of action. In pre-clinical development of biosimilars, it is necessary to establish biosimilarity not only in terms of conventional biochemical metrics (i.e., K_d; IC₅₀; PK/PE), but also in terms of structural properties, conformational stability and molecular mechanism of action. Apobiologix, a division of generic drug-maker Apotex, has partnered with TBio-DM to develop analytical tools for rapid, facile characterization of an Avastin biosimilar in development. The approach combines newly established advanced bioanalytical methods, like native mass spectrometry (Figure 1A), ion mobility spectrometry (Figure 1B) and technologies unique to TBio-DM, like microfluidics-enabled Time-Resolved ElectroSpray Ionization mass spectrometry with Hydrogen Deuterium exchange (TRESI-HDX, Figure 1C), to provide a detailed picture of the biosimilar and its interaction with the Avastin target, VEGF (Figure 1D, bottom).

In the current example, this approach revealed that: (i) The biosimilar exhibited the same epitope on VEGF, indicating the same molecular mechanism of action as the reference product. (ii) The biosimilar glycoforms were somewhat more varied and mature compared to the reference product, corresponding mostly to the presence or absence of one or more terminal galactose residues. Such glycosylation differences can underlie differences in activity and immunogenicity. (iii) The global structure (molecular size) of the biosimilar was essentially identical to that of the reference product, with no significantly populated alternate conformations in the ground state, which can indicate differences in conformational stability, potentially impacting potency and shelf-life. (iv) The manufacturing process delivered good lot-to-lot similarity, with only minor differences in the glycoform profile noted.

While these results are not GMP release methods (as yet), they can nonetheless contribute to regulatory filings for biologics approvals as orthogonal supportive structural characterization data. They can provide critical insights into the molecular bases of any functional differences that may be observed between the biosimilar and the reference product. The straightforwardness and rapidity of these analyses also allow them to be used in direct support of manufacturing and Quality Control (QC), providing a detailed picture of the biosimilar product at each stage of development.

 

Small Molecule Drug Development Neurodegenerative Disease. While much of the motivation to employ advanced bioanalytical tools comes from the transition to protein therapeutics, the same tools can also substantially enhance the design of small molecule drugs with challenging protein targets. Amyloidotic neurodegenerative diseases (a group that includes Alzheimer’s, Parkinson’s and a host of other neuropathologies) provide a clear example. These diseases are characterized by the intra- and/or extracellular accumulation of protein aggregates, in the case of Alzheimer’s corresponding to A-beta and Tau protein. The challenge of amyloidogenic targets is that they are invariably ‘intrinsically disordered’, meaning that they do not adopt a well-defined ‘native’ structure, but rather a weighted distribution of structures (called the ‘conformational ensemble’) that defines their biological function or pathology. Intrinsically disordered proteins (IDPs) are notoriously difficult to characterize as they are ill-suited to analysis by the classical tools of structural biology, X-ray crystallography and structural NMR. It is therefore often not possible to apply the conventional strategies of ‘structure-guided’ drug design, nor to (experimentally) uncover molecular mechanisms of action for drug candidates that directly target amyloidogenic proteins.

In 2016, Treventis corp., a Toronto company specializing in the development of novel anti-amyloid drugs, partnered with TBio-DM to apply a unique approach for acquiring structural characterization and mechanism of action data candidates targeting the Alzheimer’s implicated Tau protein. Using TRESI-HDX, the TBio-DM team was able to show precisely how Treventis candidates were impacting the conformational ensemble of Tau protein in order to exert their therapeutic effect. These data were specific enough that submolecular features of the drug-altered Tau conformational ensembles could be correlated to drug efficacy in vitro (Figure 2), including the ‘intensity’ of binding to a known amyloidogenic region, which correlated to the ‘steepness’ of IC₅₀ curves, and the extent of global conformational collapse, which correlated to the IC₅₀ value. These molecular mechanism of action data provided a basis for candidate selection, direction for iterative improvements in subsequent rounds of development and, importantly, played a key role in setting the stage for a multimillion dollar research agreement with French pharmaceutical giant Servier.

“The challenge of amyloidogenic targets is that they are invariably ‘intrinsically disordered’, meaning that they do not adopt a well-defined ‘native’ structure, but rather a weighted distribution of structures (called the ‘conformational ensemble’) that defines their biological function or pathology.”

Vaccine Potency Assay Reagent Development. Unlike their drug-maker counterparts in the pharmaceuticals industry, vaccine developers have always been faced with the challenges of biological production. The active ingredient in vaccines typically corresponds to one or more protein (or immunogenic carbohydrate) from the target pathogen, which is produced by the pathogen itself in host cells or by other organisms using a recombinant expression system. As a consequence, vaccine developers can in many cases be viewed as ‘early adopters’ for many of the bioanalytical techniques that will be needed to support mAb-based drug development. One clear example of this is the translation of mass spectrometry-based proteomics techniques (developed largely in academia in the 1990’s) to detect and quantify low-level protein contaminants from pathogen or host cells in formulation. Structural assays to reveal the conformational properties of the antigen and, if possible, the complex with mAbs raised against it, have largely relied on classical techniques like X-ray crystallography and structural NMR.

Another important aspect of pre-clinical vaccine development is the ability to predict potential antigen ‘potency’ in terms of the extent to which it presents surfaces that can be tightly bound by mAbs generated in the patient. This is typically achieved by western blot, which can be carried out in a high throughput manner, but is relatively uninformative in the sense that it provides no information about the nature of the complex (e.g., precisely where and how binding is occurring) and is only weakly quantitative. Western blots can be complemented by surface plasmon resonance (SPR) or ForteBio® assays to acquire dissociation constants for mAb/antigen complexation, and/or to identify the location of binding through scanning of antigen peptides. However, this approach is labour intensive and often fails, especially when the mAb binds to an antigen surface that corresponds to discontinuous sequence of amino acids (i.e., when there is a ‘conformational epitope’).

Sanofi Pasteur has partnered with TBio-DM to implement hydrogen-deuterium exchange (HDX)-based epitope mapping in pre-clinical vaccine development. This approach uses a heavily automated system that is commercially available from Waters Corp. to define the interaction surface for antibody/antigen interactions. The advantage of this approach is that it is relatively facile, and yet can provide a highly detailed picture of linear and conformational epitopes with moderate throughput. Consequently, it is possible to analyze panels of antibodies for those that might be expected to have ‘protective’ (i.e., inactivating) interactions with the antigen, and to simultaneously determine the mechanism of protection. As an example, Figure 3 illustrates a TBio-DM HDX analysis for two protective antibodies against diphtheria toxin (DTx), revealing distinct mechanisms of inactivation. This information is crucial to reagent (mAb) selection for vaccine potency assays, and can ultimately inform antigen selection for incorporation into vaccine products.

Outlook. The ongoing transformation of drug development pipelines to favor protein therapeutics over small molecules is driving an increasingly pressing need to translate advanced bioanalytical technologies to all stages of drug development in industry. To the companies that invest in the necessary technologies and expertise will go the competitive advantages of substantially accelerated pre-clinical development, knowledge-enhanced decision-making for advancement of early-mid-stage candidates, better control over manufacturing and production QC and, as more sophisticated assays are validated by regulatory agencies, ultimately more success in regulatory filings. Programs like TBio-DM at York university provide a critical bridge for drug developers at all stages, from emerging biotechs to multi-nationals, allowing low-risk, low-cost access to advanced – and sometimes entirely unique – bioanalytical tools that will have a direct impact on business trajectories.

 

References.

1. Carter, P. J. & Lazar, G. A. Next generation antibody drugs: pursuit of the ‘high-hanging fruit’. Nat. Rev. Drug Discov. 17, 197–223 (2018).
2. Souriau, C. & Hudson, P. J. Recombinant antibodies for cancer diagnosis and therapy. Expert Opin. Biol. Ther. 3, 305–318 (2003).
3. Deng, B., Lento, C. & Wilson, D. J. Hydrogen deuterium exchange mass spectrometry in biopharmaceutical discovery and development – A review. Anal. Chim. Acta 940, 8–20 (2016).
4. Liuni, P. & Wilson, D. J. Understanding and optimizing electrospray ionization techniques for proteomic analysis. Expert Rev. Proteomics 8, 197–209 (2011).