Bioprocessors used to say “the process is the product,” which reflects biotech’s deathly fear of regulatory risk more than the belief that one can treat diabetes with filtration. Increasingly, we are learning that if one characteristic deserves designation as “the product” it is a therapeutic protein’s three-dimensional structure.
In addition to the chemical attachment of amino acids in a particular sequence, the primary structure, proteins exist in nature as complex, semi-self-assembled, dynamic entities whose topography changes, at both the macro and micro level, through interactions with their surroundings. These higher-order structures (HOS) include secondary (e.g. local folding patterns, alpha and beta sheets), tertiary (overall folding), quaternary (subunit arrangement), and even quinary (the spatial and organizational relationship between protein and cell interior).
Biosimilars and HOS
Quantifying HOS differences or similarities between two protein samples has become standard for demonstrating biosimilarity for that specific FDA-approved designation, and is increasingly used to compare different batches of the same drug. Instrument makers, aware—hoping is perhaps a better word—that biosimilars will eventually enter the U.S. market in meaningful numbers (just three are available to patients), have been adapting physical and spectroscopic analysis methods to probe for subtle but meaningful variations within HOSs.
One technique tailor-made for HOS is circular dichroism (CD). CD quantifies differences in absorbance for left- and right-circularly polarized light occurring when a molecule contains one or more chromophores that are either in a chiral environment or are themselves chiral.
Doug Marshall, Ph.D., product manager at Applied Photophysics, explains that in proteins the dominant chromophore is the amide carbonyl bond of the peptide backbone, which absorbs in the far-UV. “The angle between adjacent carbonyl bonds differs depending on the secondary structure, e.g. alpha helix or beta sheet. These differing angles affect the environment of the bonds and give rise to a far-UV CD spectrum that informs on the secondary structure of the protein.”
Other chromophores in proteins amenable to CD analysis include the aromatic side chains on tryptophan, tyrosine, and phenylalanine, plus disulfide bonds that absorb in the near-ultraviolet and give rise to a near-UV CD spectrum, which is sensitive to any factor that affects the local environment of these groups.
“Far-UV is also used in this fashion, to detect structural differences, assess stability through thermal or chemical denaturation experiments, or to assess ligand binding interactions,” Marshall adds.
Although biosimilars are the most obvious application for HOS, CD determinations serve in this capacity throughout biotherapeutic development. “The ability to understand the effects of product optimization, scaleup, formulation, manufacturing changes, fill/finish, and storage conditions on HOS is vital,” he continues. “CD also plays an important role in batch-to-batch monitoring and in post-release pharmacovigilance.”
For small molecules, CD helps establish a molecule’s absolute configuration, for example through comparison with a molecule of known configuration or through calculations from first principles using density functional theory (DFT). For proteins, CD combined with statistics can allow objective detection of changes in HOS resulting from oxidation, deamidation, glycosylation, or other post-translational modifications. By introducing a CCD fluorescence accessory, systems from Applied Photophysics assess HOS as three quality attributes simultaneously: secondary structure by far-UV CD, tertiary structure by near-UV CD, plus an orthogonal fluorescence method.
Overcoming vendor specificity
Because it potentially draws on so many varied analytical methods, HOS analysis must often tie together results from multiple platforms or vendors. As more companies become deeply involved in HOS, this issue needs to be overcome.
Protein Metrics, for example, provides vendor-neutral software for post-acquisition data analysis of LC, LC-MS, CE, and CE-MS data that works with data files from all major mass spectrometry vendors. “We have also worked with younger, smaller detector and instrument companies to provide enhanced data analysis for high-throughput applications, and for established workflows that fall short of customers’ expectations,” says Eric Carlson, Ph.D., president. This includes mass spec data from QTOF, ion trap, and Orbitrap instruments, and from most LC and CE systems. “HOS generally, and its specific components and analytical methods, vary significantly depending on the research objective.”
Among these techniques, native MS analysis comes immediately to mind. A literature search uncovers thousands of references to the technique, which often incorporates MS and front-end separation like LC or relatively exotic modalities like rapid oxidation or combination with top-down proteomics MS.
Native MS analyzes intact proteins and their complexes without destroying noncovalent interactions critical to HOS analysis. It achieves this by ionizing macromolecules directly, from a non-denaturing solvent, to preserve many solution characteristics in the gas phase. Through native MS, charge state distributions of protein ions provide conformational information as well as gross molecular weights of intact molecules.
Protein Metrics’ main contribution to native MS is a charge deconvolution algorithm, dubbed the "parsimony algorithm," that differs from maximum entropy approaches used by most all other vendors. “Our algorithm avoids the known artefacts and harmonics that plague maximum entropy analysis, which are a real problem when deconvolving over a wide mass range,” Carlson explains, “which is particularly important in native mass spec analysis to preserve native structure.” Applications of the parsimony algorithm include complexes (dimers, trimers, hexamers, etc), and ligand binding, which is of great interest in drug development.
Some methods, like hydrogen-deuterium exchange (HDX) and oxidative footprint (aka chemical labeling), are complementary and orthogonal. HDS, the prototypical NMR technique for HOS determinations, which provides insight into a molecule’s solvent accessibility, “senses” the protein backbone, usually at the level of pepsin-digested peptides.
Oxidative footprint requires an oxidation step. “Until recently, inexpensive off-the-shelf solutions to the oxidation requirement were unavailable. We recently published with collaborators to demonstrate this using a benchtop Fenton chemistry,” Carlson explains. The Fenton reaction uses an iron catalyst to potentiate the oxidizing potential of hydrogen peroxide. “For this reason oxidative footprint has not been a mainstream technique, but we hope our software will make it more widely available.”
HDX and oxidative footprint methods are normally vendor-specific, but Carlson says his company’s software is not, and can therefore work with multiple instruments.
Another method Carlson favors is crosslinking analysis. This technique uses chemical crosslinkers to probe HOS. By using spacers of different lengths analysts can determine through-space distances among various amino acid residues, and even individual atoms. The relative atomic or amino acid positions relate directly to a protein’s shape.
The mystery and beauty of HOS analysis becomes apparent when one considers that epitope mapping, the most commercial technique in protein analysis, uncovers both linear and conformational epitopes. Everyone is familiar with a protein’s linear structure, the physical sequence of amino acids. Conformational epitopes consist of non-contiguous amino acids that assemble in space to create a kind of virtual active site. “It is generally the epitope, and not the full target protein, that is claimed as intellectual property,” notes Carlson, and highly significant conformational epitopes arise through higher-order conformational processes. “HOS studies are therefore critical for securing the commercial value of biotherapeutics.”