Methods and Purposes for Determining Higher Order Structures of Biopharmaceuticals

The determination of higher order structures of biopharmaceuticals has become an important subject for biopharmaceutical firms and regulatory agencies. This overview covers the important approaches, as described in recent books, journals, and magazine articles, as scientific meetings.

In starting this discussion, a statement from a recent issue of BioProcess International seems appropriate: "It is important to note that no analytical test or combination for higher order structures (HOS) has yet been sufficiently validated for analytical testing as a substitute for clinical studies, in the development of a biosimilar therapeutic antibody (mAb) drug substance"(1). A goal is to hasten this process so that, in the near future, such methods will be adopted and accepted by the industry and all regulatory agencies involving proteins. The analysis of HOS has become an extremely important aspect of determining the eventual biological activity of a new biopharmaceutical and related species, such as proteins, fusion proteins, bioaffinity proteins, antibodies (mAbs), and antibody–drug conjugates (ADCs). (Recall your college biochemistry: "structure equals function.") The analysis of HOS has become more important for firms seeking to demonstrate that their biosimilar product has the same or close physical, biological, and biopharmaceutical properties as the original, proprietary product. It is generally accepted that if the HOS (in solution) of the biosimilar are not identical or very close to those of the innovator, then the biosimilar will not have the same desired medicinal properties. It is also expected that innovator companies determine the HOS of their products, and demonstrate that the HOS remain similar or identical from batch-to-batch, within experimental error. It has become obvious that information about HOS is crucial in submittals of all biopharmaceuticals for all regulatory agencies, worldwide.

What then is meant by HOS? As described in several annual symposia in recent years, HOS relate to the conformation, size, and arrangement of atoms that make up a biopharmaceutical (2), specifically tertiary or quaternary structures. The term can also relate to nucleic acids, DNA, RNA, and so forth, but here the term is used as it relates to protein based biopharmaceuticals. At the 7th annual California Separations Society (CASSS) HOS Conference, speakers presented different aspects of the latest approaches to determine the HOS of proteins. The final program, and copies of some slide sets, are available online at the CASSS website. There are analogous HOS conferences, such as the Gordon Research Conferences, with similar goals.

There are no figures in this column, a first, because current methods for determining HOS do not really provide true and total information about HOS for any protein, in solution. Some methods come close, but none succeed 100%. Does it make sense to show spectroscopy spectra that do not provide total HOS?

Two recent books are devoted to HOS (3,4), with the earlier having a second edition in progress. This book is a key reference, as it also discusses how HOS evidence applies to biopharmaceutical products, both proprietary and biosimilars, and their comparisons. Two vital aspects should be kept in mind, one being production of the total HOS alone for a protein as a marker, and the other being the use of HOS for comparing biopharmaceuticals, either comparing products batch-to-batch or comparing proprietary products to biosimilars. These are very different applications.

Why, then, has the topic of HOS become so important in the biopharmaceutical industry, with regulatory agencies, in scientific publications, and now in complete books? Virtually all that follows here is covered in much greater depth and brilliance in the Berkowitz and Houde text (3). However, we do not always march in lockstep.

Background and Rationale to Determining the HOS of Proteins

Because biopharmaceuticals are such large, biologically active molecules, often 150 kDa or larger, and because they are generated through a variety of expression systems, often mammalian in nature, their structures tend to be very complex, of high molecular weights (MWs), and usually with post-translational modifications (PTMs) on various, and varying, amino acid sites (5–7). They are very complicated molecules to deconstruct, but analytical biotechnology has been able to do that, though not quite for HOS. It has become clear that we are able to decipher virtually all PTMs, amino acid variants, disulfide bridges, points of attachments of low MW drugs in ADCs, and so forth. For such analyses, numerous techniques can be used, especially high performance liquid chromatography (HPLC), almost always interfaced with some form of modern mass spectrometry (MS), and sometimes using multidimensional HPLC. These techniques have now become semi-routine, and capable of fully characterizing almost any biopharmaceutical under development, and (eventually) new molecules coming to market (5–8). The issue we now face is not the lower (1st or 2nd) dimensional characterization of the original protein and all of its PTMs, but rather the HOS (3rd and 4th dimensional) characterization. These terms of structural dimension are further discussed in references 3–6. Analytical methods to characterize these HOS abound, but the remaining question is whether they are able to provide the true and complete HOS of typical biopharmaceuticals. Our interpretation of which methods provide more HOS information than others may differ from the views of others in the field (3–4). This is our true goal: to explain why some methods provide more HOS information than others.

Unless it is possible to describe the true and complete HOS of innovator biopharmaceuticals, biosimilars may have greater difficulties being approved without human clinical trials. Recall the introductory sentence above. Most biopharmaceutical firms today are able to provide virtually complete analytical structures, even of very complex antibodies (mAbs) (5–7). The US Food & Drug Administration (FDA) has come to expect such structural analytical data on all submittals. But final decisions as to which analytical methods will provide 100% HOS, in those same submittals, appear lacking. This situation needs to be corrected.

HOS ultimately determine the biological activity and biopharmaceutical benefits of biopharma drugs in their desired, native state, with full biological import. It will be the HOS data, in comparing a biosimilar to its proprietary, that will determine if the two products are interchangeable, and whether the biosimilar can go to market without obtaining animal or human clinical data. It is the ability of a biopharmaceutical to bind to certain biological receptors or targets that will determine its success. If a molecule is not perfectly expressed or purified, leading to changes in the intended HOS, the molecule may not be approved for use. This is particularly relevant when proteins are expressed in inclusion bodies in E. coli, and then need to be purified and refolded to their native state. Science and engineering must improve, develop, or invent better instrumentation, methods, and applications to define, with 100% certainty, the true and complete HOS. Analytical methods cannot always define if biomolecules are active against their receptors or antigens until they are tested. Analytical methods for determining HOS should, at least, be able to strongly suggest biological activity. It will be an actual demonstration of recognition against targets that will demonstrate that a biopharmaceutical has the correct HOS. Usually, this relies on various approaches, such as affinity recognition in HPLC, surface plasmon resonance (SPR) spectroscopy, biolayer interferometry (BLI), enzyme linked immunosorbent analysis (ELISA) methods, and other related (already established) analytical approaches.

What, then, defines HOS? The closest thing to a full determination may be the X-ray crystallographic (XRC) picture one obtains from conventional, low MW drugs or other targets, which typically represents the most energetically stable states (structures). These XRC pictures have been realized for decades for crystallized proteins, but never in solution. Two-dimensional nuclear magnetic resonance spectroscopy (2D NMR) provides two different NMR spectra, but still do not provide total and complete information about the HOS of a protein (vide infra). HOS might be defined by the radius of gyration (Rg) or hydrodynamic radius (Rh) via multiple angle light scattering (MALS) or viscometric methods. Such a radius is a part of the HOS, but it does not, alone, define the true and complete HOS. This is also true of analytical ultracentrifugation (AUC). The observance of aggregates by the above technique is not the true HOS of the parent protein. In fact, although aggregation may be considered an aspect of HOS, here we will not consider it as preventing aggregation and maintaining native HOS, which is our focus. To consider that Fourier transform infrared spectroscopy (FT-IR) or Raman spectroscopy, define (singly or together) the total and complete HOS, is also, to our minds, not true. Similarly, probably no spectroscopic technique, which now might include nuclear magnetic resonance (NMR) or electron spin resonance (ESR), defines the true and complete HOS. However, recent 2D NMR methods, especially high-resolution ones, appear to come very close to a true HOS, protein and software dependent (vide infra). Again, the recent text of Houde and Berkowitz should be considered as required reading on HOS methods of determination (3).

When we come to something such as electron microscopy (EM) (now quite popular for defining HOS by some), that, too, does not really define all the nuances and intricacies of HOS. EM cannot provide the same degree of refinement and structure description as XRC. It does not yet reach the atomic or molecular levels. Companies and researchers have been applying 2D NMR to define HOS, and there are several publications and presentations (3,9–12) to this effect. It is our inclination, as with others, that XRC and 2D NMR may well provide the highest level of HOS yet available. Yet even these two techniques have serious limitations, vide infra. Our estimation of the inability of an analytical method to define HOS, with 100% assurance and clarity, is based on our own readings and experience. Others may differ in opinion.

What Are the Most Important and Successful Methods Available Today for Determining HOS?

What, then, are the most important analytical approaches, now existing, that can support the determination of HOS for biopharmaceuticals? If we utilize the first few references, we can derive about 10 individual analytical methods (2–4), These are (not ranked in order of importance or application): X-ray crystallography (XRC), 1D and 2D nuclear magnetic resonance (NMR), hydrogen–deuterium exchange mass spectrometry (HDX-MS), ultraviolet–fluorescence–Fourier transform infrared (UV–FL–FT-IR) spectroscopy, circular dichroism (CD), light scattering (static or dynamic) techniques (LS, SLS, DLS, X-ray scattering), analytical ultracentrifugation (AUC), differential scanning calorimetry (DSC), and perhaps, cross-linking MS (CLMS). Clearly, all of these currently used methods cannot be adequately covered in this single LCGC North America column. We have taken license to choose some of these to evaluate, hopefully those of the widest utility and applications today. We will pique the interest and ire of the reader by pointing out that no single method, not even 2D NMR, provides everything needed and wanted for determining 100% of HOS for a protein in solution, and although XRC can actually locate the precise location of each and every atom in a biopharmaceutical, the analysis is done in the solid state, which is perhaps irrelevant to in vivo solution conditions. None of the above analytical methods (and others) are ideal; each and every one has problems, some very serious and some less so. The only analytical technique that does provide total HOS is most likely X-ray crystallography (XRC). However, it has problems, at times, in working with crystallizing complex proteins, but when it works, it very clearly provides the closest thing possible today to a true HOS. It also requires somewhat expensive instrumentation and trained operators. It pinpoints exactly where every single atom exists in 3D space, and the distances and angles from one atom to the others. However, when it comes to HOS, it is a much less accurate and accessible technique than for considering the tertiary structures of small monomeric proteins. Again, it is only useful in the solid state, which may be totally irrelevant to the solution conditions of the final, biopharma formulation before injection or administration. It provides a true HOS, which analytical ultracentrifugation (AUC), FT-IR, multiple angle light scattering (MALS), 1D NMR, and all the other conventional analytical methods, as above, cannot yet realize.

As of yet, it is not really a requirement of any regulatory agency that the full and complete HOS be defined in order to compare a biosimilar to its proprietary cousin, so as to gain rapid, market access. It is simply necessary to demonstrate, perhaps by multiple methods, that the biosimilar demonstrates very comparable spectra to the proprietary, original protein already approved for marketing. One can actually use many analytical techniques, including MALS, hydrogen deuterium exchange (HDX), AUC, FT-IR, 2D NMR, and any others, so long as the spectra or data (HDX, DSC, and so on) derived are very comparable between the biosimilar and proprietary products. Some readers may think that HDX can provide true HOS for a biopharmaceutical; that is not true. It can be used (and very nicely, too) to compare the proprietary and biosimilar products under the very same HDX conditions or family of conditions, and, when those results are compared in a mirror image format, they should show no real differences if they indeed have the same HOS. That is also very suggestive, as are comparisons using any other of the various methods being touted as providing HOS determination. We are not certain that any spectroscopic (even 2D NMR), ELISA, HDX, or XLMS methods are capable of providing the true HOS of a biopharmaceutical. However, they don't have to do that; they just have to show no differences when used with both the proprietary (innovator) drug and its biosimilar cousin.

Why, then, do we believe that XRC may well be the only viable approach to thoroughly define HOS for any protein? As mentioned above, it only works with a solid crystalline protein, and will not show the actual HOS of that protein in a solution. However, it can pinpoint the exact position of every single atom in 3D space for the most energetically stable form relative to all the other atoms of the protein. It can indicate the precise distance between all the atoms, as well as the angles involved between any three atoms, regardless of their identity. That surely appears to be a true HOS, at least of the solid state, both of which should overlap 100% between the proprietary and biosimilar biopharmaceuticals. The issues are that it requires a crystalline protein (which is no small feat), the usual X-ray spectrometer, qualified operators, and suitable computer software facilities, not to mention the time and resources being unsustainable for HOS determination on a robust level for all biopharmaceuticals. But it certainly does work when the protein can be crystallized. That crystal HOS may well change in solution, depending on conditions, and the solution HOS may be different between the innovator and biosimilar products. Also, dynamics of the protein are not considered in the XRC determination.

One final word of caution comes from the recent article, mentioned in the introduction, that discusses gaps in our science that limit the development of an industry standard for regulatory approval (1). It said, in part, when discussing therapeutic monoclonal antibodies (TmAbs) and other innovator products,"It is important to note that no analytical test or combination for HOS has yet been sufficiently validated for analytical testing as a substitute for clinical studies in the development of a biosimilar TmAbs drug substance. Without such validation, creating a quantitative industry standard to establish HOS similarity seems premature. Instead, renewed investment in developing predictable and standardized HOS measurements of validated reference materials is envisioned as a path forward" (1).

Most studies using the methods indicated below do not really provide information about the total HOS. Most spectroscopic methods, such as FT-IR, Raman, CD, fluorescence, and others, do not provide information about HOS, but only provide spectra related to HOS. It should be said that if all such approaches when applied to comparison of biosimilar to proprietary biopharmaceuticals indicate similar or analogous data, such as by CD; and that they are supportive of indicating the same HOS they are not confirmatory. No analytical regime is as confirming as XRC combined with 2D NMR.

Individual Spectroscopic and Other Methods Suggested to Determine HOS

At scientific conferences dealing with HOS, certain things become more obvious regarding current efforts to derive HOS of biopharmaceutical products (10). Most analytical methods for HOS are spectroscopic in nature, such as FT-IR, CD, Raman, fluorescence, and others. Others are microscopic, such as electron microscopy (EM), which shows actual shapes and approximate sizes of mAbs or proteins, but little about HOS, per se (nice pictures, no numbers). It is not apparent that EM can ever be used to demonstrate chemical or HOS equivalency between a biosimilar and the proprietary protein. It is not equivalent to XRC in the details it can provide. None of these, and others, are able to provide a true, complete definition of the HOS of a typical protein. Most provide some type of information, such as Rg or Rh from MALS, but that is far from complete information about HOS, or MWs using AUC, which has very little to do with HOS. Yes, AUC can separate aggregates from monomer and indicate MWs of all species, but that again has little to do with HOS. Most of what is in the literature or at scientific meetings in the way of defining HOS does not really do this. Even HDX can suggest equivalent HOS between the biosimilar and proprietary proteins, especially by mirror plots of their peptides (after HDX), but that is not defining HOS. It does appear that most, if not all, analytical methods to define HOS are not hitting the true mark. Why not? As we continue, we'll discuss in more depth what appear to be the most popular HOS methods in vogue today.

Analytical Ultracentrifugation (AUC)

AUC (3, Chapter 9) is a method that provides the molecular weight (MW) of a mixture of proteins or a highly purified individual protein, using a series of standards separated together with the target biopharmaceutical in a high-density medium, at high rates of spinning in a commercial centrifugation instrument. After suitable resolution of the species present, each one can be "determined" by its approximate MW (not radius of gyration (Rg), or hydrodynamic radius (Rh), size or shape, but by molecular weight (MW). The approach allows for the accurate and precise determination of monomer, aggregates, and fragments of the starting protein or mAbs. It certainly works quite well for these determinations, and has become a strong component of any laboratory studying protein aggregates, adducts, cross-linked proteins, and so forth.

There is but one small problem in attempting to utilize AUC to derive HOS: It cannot provide such information. What it can provide is an indication of true MW, and from knowing the MW of the monomer, whether one is dealing with a dimer, trimer, hexamer, or fragments of the monomeric starting material. However, such information can be obtained more accurately these days via native mass spectrometry. It says very little, if anything at all, about whether the protein has x% random coil, y% beta-sheet, or z% alpha helix, and, more importantly, it does not define the true conformation, shape, or configuration of the protein under any biological solution conditions. However, it should be indicated that even knowing the x%, y%, or z% of a protein still does not tell us the true HOS shape and conformation of the entire protein. AUC and most of the other methods to be described and discussed below, also, do not provide x%, y%, or z% but perhaps for CD. However, even that does not define total conformation or HOS.

Isothermal (ITC) or Differential Thermal Calorimetry (DTC)

We now resume this (perhaps painful) realization and discussion, that basically very few current analytical approaches can determine the full HOS of a protein in solution under its true, biological reaction conditions. Not even isothermal (ITC) or differential calorimetry (DSC) methods of today can provide HOS information (3, Chapter 11). They indicate changes in reactivity and reaction thermodynamics, but really say nothing about absolute HOS. Again, they can be useful in comparing a biosimilar to its proprietary analog protein and in suggesting similar HOS, but these are mainly assumptions without true evidence.

X-Ray Crystallography (XRC)

XRC (3, Chapter 8) has been around for decades, and it works extremely well for those proteins that crystallize well before being measured. It has the ability to define precisely where each and every atom of the molecule resides in space, its neighboring partner, to what it is bonded, even the angles of most bonds, and it certainly does indicate HOS, probably better than any other known method, even for 2D NMR. The main problems with XRC are twofold. One is that the protein must be obtained (most often) in a crystalline form, and not all proteins crystallize suitably for good XRC measurements. The second, and perhaps even more crucial, is that XRC is usually done on a protein in the solid state (crystallized typically in the energetically most stable state), which is not how it exists or reacts in solution. The solid state conformation of a compound's HOS may well be very different from its HOS in solution, when undergoing a reaction or complexation. Thus, XRC delivers a conformation that one must then prove is the same in a solution reaction, which is often very difficult to prove or disprove. Suffice it to say, at this point, that XRC has now become the method of choice to determine HOS for a typical protein in the solid state. It remains to be demonstrated that the HOS of a typical protein in the solid state is identical to that in solution. This may not be required for regulatory purposes when introducing biosimilars. It seems very likely that XRC, combined with 2D NMR in a biological or aqueous medium for comparative purposes, may well come closest to the needs of the regulatory agencies in comparing biosimilars to proprietaries.

Nuclear Magnetic Resonance (NMR) Spectroscopy

NMR (3, Chapter 13) has, since its inception in the 50s and 60s, become the workhorse for determining the structures of low MW compounds. It has also been shown to be extremely useful for higher MW proteins and mAbs, though because of the intensity of the data produced, it often takes considerable time, effort, and computer power to derive true 3D structures. Nevertheless, especially in the 2D format, it has probably shown more promise than any other method for determining protein HOS, but is not without its own problems. 2D NMR does not refer to the use of two spatial dimensions but rather two different isotopes are used to generate each spectrum (¹³C, ¹H, ¹⁵N usually). It does not truly provide a spectroscopic picture (as in XRC) of the HOS, but a pattern of nuclei shifts that is reflective of the parent protein. And, it becomes very useful when the biosimilar and proprietary proteins have exactly the same 2D NMR patterns, suggesting the same HOSs.

However, as with XRC, its problem is that it cannot easily be performed in a biological medium, identical to when a protein is undergoing its usual reactions. The presence of the biological medium, almost always including water and salts, as well as numerous other biological compounds in the system, also appear in the NMR spectrum. Sometimes these NMR active materials overwhelm the spectrum of the protein of interest. This all makes it very difficult to determine the true conformation of the protein of interest, as it is actually reacting or in any, biological medium.

That is not to say that 2D NMR can actually define a HOS for that protein alone in D₂O. It is not actually showing the HOS but rather the positions of the ¹³C or ¹H or ¹⁵N positions in 3D space. That is not a 3D representation. If one only wants to demonstrate the equivalency of HOS between a proprietary and biosimilar product of the same protein, then surely 2D NMR can seemingly accomplish this. It takes quite some time, depending on the MW of the protein involved, but with suitable computer hardware and software, as well as a high resolution NMR, a useful pseudo-HOS determination is produced, now on a regular basis. However, what is actually being generated is a series of ¹H or ¹³C spectra, which can be represented as a two-dimensional diagram suggesting the HOS, but it is not really the same refined HOS given by XRC as above. One could then assume that, under biological reaction conditions, these two species also have the same HOS. That is, unfortunately, a huge leap of faith in most instances.

Differential Scanning Calorimetry (DSC)

DSC (3, Chapter 11) has become one of the most important approaches for studying the thermodynamics of chemical or biological reactions. It can indicate whether a reaction is exothermic, endothermic, or unreactive (no reaction occurs). These are extremely useful things in studying reaction kinetics and thermodynamics, but totally useless for determining absolute HOS, because if all we wish is to determine if the HOS of a stable protein before it undergoes any chemical reactions, DSC does not provide that ability. It can indicate the relative speed of a biological reaction, whether it gives off or absorbs energy in its reaction, rates of chemical reactions, and other thermodynamics, but it can say nothing whatsoever about HOS. Why then, do others believe it does possess this ability?

Cross-Linking MS (CLMS)

CLMS has become a very useful approach for demonstrating the closeness of one atom to another by the use of various cross-linking reagents towards certain functional groups, commonly found in most proteins (3, Chapter 12). Once two atoms in the protein backbone are cross-linked, as a function of the reagent used, it can then be digested into smaller peptides, and the peptide map will demonstrate the flat structure of the original protein. It does not provide bond angles involved before cross-linking occurred. It can easily indicate which atoms in the protein were close enough or not to undergo crosslinking with that particular reagent. It may provide useful information by top-down protein sequencing, as well as bottom-up. It really says very little about the angles of any bonds in the original protein, but only the nearness of those atoms undergoing the actual cross-linking. However, it can be quite effective at identifying higher order oligomers (dimers, trimers), but one must be cautious as to whether they are biologically relevant or not. Its ability to demonstrate HOS remains somewhat limited, though a comparison of proprietary and biosimilar products of the same biopharmaceutical can be made using CLMS. How much that demonstrates equivalency of their HOS remains questionable, if not outright dubious.

Hydrogen–Deuterium Exchange–Mass Spectrometry (HDX-MS)

HDX (3, Chapter 12) has become a somewhat, standard analytical routine in many, if not most, biopharma firms. More and more submittals for biosimilars or proprietary drugs are now containing HDX studies. There are at least two major instrumentation firms that now offer commercial grade HDX instrumentation, namely a high resolution, ultrahigh performance liquid chromatography (UHPLC) system and a research grade MS. What HDX provides is the ability to discern which specific sequences of amino acids, in any protein, native or denatured, because of its placement in the HOS structure, is able or not to undergo hydrogen-deuterium exchange as a function of time and conditions. Those sequences of amino acids not protected by other pieces of the protein's 3D conformation more readily undergo HDX than others. This is a function of the protein's HOS, temperature of the exchange conditions, speed of the exchange, and other experimental, well described parameters of running a typical HDX set of experiments.

What HDX allows, though not providing true HOS, is the ability to directly compare the biosimilar to proprietary proteins under the very same HDX conditions, side-by-side, if desired, and when these are represented in a mirror plot of the peptides studied, as well as their percent H-D exchange (percent exchange), that is very strong evidence that these two proteins have the very same or very similar, HOS. If these mirror plots indicate total equivalence of H-D exchange percentages, the regulatory agencies are very likely to agree it is strongly suggestive of both species having the same HOS. The degree of agreement between these two sets of HDX numbers per peptide, that will be taken as 100% demonstration of identical HOSs, is up to the regulatory agencies to determine. It seems that the biopharma industry is now coming to agree that HDX data can be very persuasive in convincing FDA of the equivalency of these two protein species. However, test cases are only now going thru the usual FDA review process for final approval of newer biosimilars.

Spectroscopic Methods of Determining HOS

We will close by discussing generic spectroscopic methods and, finally, multiple angle laser light scattering (MALS) methods. Generic spectroscopic methods include Fourier transform infrared (FT-IR) spectroscopy, ultraviolet-visible absorbance (UV-Vis), fluorescence, Raman, and circular dichroism (CD). These are standard traditional methods used for decades to characterize, in part, protein characteristics, but never to produce any evidence as to HOS (3, Chapters 5–6). Indeed, there is general agreement that virtually all purely spectroscopic methods cannot provide information on HOS. What is provided are spectra related to bending, stretching, or vibrational mode changes under the influence of (usually) light, linear, or circularly polarized.

However, CD has become recognized as offering much more information about protein conformations than any other purely spectroscopic method today. Why is this? The method relies on using optically active polarized light as the source, which is then rotated as it passes through chiral species, such as a protein's backbone. The light is then rotated to a certain degree, one direction or the other, depending on the specific protein's conformation(s) in solution. Each portion of the light source (wavelengths) that passes through the protein structure is thus rotated as a function of wavelength, and a plot is then generated of how much that light source has been rotated from the null. Each section of the wavelength scale used reflects a certain type of protein conformation, alpha helix, beta-sheet, or random coil. Each protein, having different conformational parts or HOS, will then have its own, unique CD, which will define percent alpha-helix, percent random coil, and percent beta-sheet. Unfortunately, that is not as useful as the actual HOS for the protein, and nobody has yet been able to go from a CD spectrum to the HOS for any protein. While each CD for a specific protein may well be different from the next, as seen by itself, it is not defining HOS for that protein. In fact, it is really only suggesting that your protein has some defined three-dimensional structure and is not an aggregate. Nevertheless, it is useful in comparing a biosimilar to its proprietary cousin, but only if the CD spectra are very similar, ideally identical. If there are differences in the two CD spectra, say above a few percentage changes, then all assumptions as to HOS equivalency may be denied by the regulatory agencies.

Electron Microscopy (EM) and Cryo-EM

EM is a form of normal light microscopy, but using an electron beam to view the 3D structure of the protein (3, Chapters 2,3,10). It, too, was not capable of providing HOS, but it does present a greatly enlarged picture of the protein or antibody. What is amazing about looking at these images is that they do indeed look just like we would expect a protein or antibody to look, but without any actual size dimensions–that is, there were shapes evident, but no absolute sizes or angles possible for the blob that represents, for example, an antibody. There was no indication of HOS, nor numbers to indicate the size of the protein or where anything actually resided, in terms of functional groups, disulfide bonds, and so forth.

However, the above assessment of modern low temperature electron microscopy (cryo-EM) has changed, just within the current year (13–16). Though the technique was not very satisfactory for conventional proteins in the past, some very recent publications indicated that cryo-EM can now be utilized for biological proteins of ordinary MWs with high resolution and the fine structure of each protein. Thus, proteins such as human adenosine A1 receptor-Gi complex, serotonin 5-HT1B receptor coupled to heterotrimeric Go, inhibitory G protein, and others, can all be structurally determined, as have been typical proteins using XRC in the solid state. However, cryo-EM is working at low temperatures, although not the solid state of the protein, but rather with its native, active state intact. These seem to be very significant advances for the future of all cryo-EM studies with typical native proteins, going forward.

Light Scattering Techniques

LST (3, Chapter 8) includes approaches such as dynamic LS (DLS), and static LS (SLS), with a variety of commercial instruments (Malvern, Wyatt and others). There are instruments for doing multiple angle, right angle, low angle, and other approaches but they are all using scattering measurements as a function of the protein size and shape. The information regarding HOS is limited to things such as radius of gyration (Rg), or hydrodynamic radius (Rh), as well as polydispersity, incremental, refractive index (dn/dx), and absolute MWs. However, there is very little information regarding, true HOS at the atomic levels, as with certain of the above approaches, especially cryo-EM, XRC, and 2D NMR.

Conclusions and Apologies

There is little question but that everyone has realized the terrible importance of being able to define HOS of a typical protein in the shortest, fastest, and least expensive ways possible today. In our opinion, and perhaps that of others (3), this may well involve combining XRC in the solid state of the protein in question, together with some form of 2D NMR, depending on what combination of isotopes one wishes to utilize. It seems possible that 3D NMR may also be feasible depending on the sensitivity of the instrument to the level of ¹⁵N present in any given protein. At the moment, 2D NMR (¹H, ¹³C) appears able to differentiate different structural isomers of the same protein quite well, and to compare a biosimilar to its proprietary cousin to show similarities or differences. These are, indeed, very useful attributes but still do not actually provide the complete HOS, as can XRC in the solid state. We fully realize that the above discussions might be frustrating to the readers but the conclusions seem inescapable so far.

Acknowledgements

We acknowledge, with appreciation, the ability to have the inputs and advice of Steve Berkowitz in the above column.

References

(1) S.J. Kaur, D. Sampey, L.W. Schultheis, L.P. Freedman, and W.E. Bentley, BioProcess Int. 14(9), 12-21 (2016).

(2) Higher Order Structure 2018, Omni Providence Hotel, Providence, RI, April 9-11, 2018, 7th International Symposium on Higher Order Structure of Protein Therapeutics. http://www.casss.org/page/HOS1801.

(3) D.J. House, and S.A. Berkowitz (Eds.), Biophysical Characterization of Proteins in Development Biopharmaceuticals (Elsevier Science Publishers, Amsterdam, 2015).

(4) G. Misra (Ed.), Introduction to Biomolecular Structure and Biophysics, Basics of Biophysics (Springer International Publishing AG, Part of Springer Nature, www.springernature.com/us, 2017).

(5) C.T. Walsh, S. Garneau-Tsodikova, and G.J. Gatto, Jr., Ang. Chem. Int. Ed. 44, 7342-7372 (2005).

(6) J.R. Lill and W. Sandoval (Eds.), Characterisation of Biotherapeutics (John Wiley & Sons, Hoboken, NJ, 2017).

(7) G. Chen (Ed.), Characterisation of Protein Therapeutics Using Mass Spectrometry (Springer Science, New York, 2013).

(8) M.W. Dong, Modern HPLC for Practicing Scientists (John Wiley & Sons, Hoboken, NJ, 1st ed., 2006).

(9) B. Japelj, G. IIc, J. Marusic, J. Sencar, D. Kuzman, and J. Plavec, Biosimilar structural comparability assessment by NMR: from small proteins to monoclonal antibodies, www.nature.com/scientific reports/August, 2016.

(10) Gordon Research Conference, Molecular Structure Elucidation, August, 2016, Mt. Stowe, VT; Gordon Research Conference, Molecular Structure Elucidation, August, 2018, Sunday River, ME.

(11) https://www.bruker.com/applications/pharma-biopharma/drug-development/structure-elucidation/protein-analysishos.html.

(12) Biophysical Characterization of Proteins in Development Biopharmaceuticals, Edited by D.J. House and S.A. Berkowitz, Elsevier Science Publishers, Amsterdam, 2015, Chapter 13, Y. Aubin, D.J. Freedberg, and D.A. Keire, One and Two-Dimensional NMR Techniques for Biopharmaceuticals.

(13) A. Koehl, H. Hu, S. Maeda, Y. Zhang et.al. Nature 558, 547–552 (2018).

(14) C.J. Draper-Joyce, M. Khoshouei, D.M. Thai, Y.L. Liang et. al., Nature 558, 559–563 (2018).

(15) J. Garcia-Narria, R. Neeme, P.C. Edwards, and C.G. Tate, Nature 558, 620–623 (2018).

(16) Y. Kang, O. Kuybeda, P.W. de Waal, S. Mukherjee et. al., Nature 558, 553–558 (2018).

ABOUT THE AUTHOR

Ira S. Krull

Ira S. Krull is a Professor Emeritus with the Department of Chemistry and Chemical Biology at Northeastern University in Boston, Massachusetts, and a member of LCGC's editorial advisory board.

Jared R. Auclair

Jared R. Auclair is currently the Director of Executive Training and Biotechnology Programs in the Department of Chemistry and Chemical Biology at Northeastern University. In addition to being Director of Biotechnology, Dr. Auclair also directs the Biopharmaceutical Analysis Training Laboratory and the Asia-Pacific Economic Cooperation Center of Regulatory Excellence in Biotherapeutics. This latter appointment allows Dr. Auclair to collaborate with both academic researchers and industry in the area of biopharmaceutical development and analysis. He has expertise in molecular biology, protein biochemistry, analytical chemistry, protein crystallography, and biological mass spectrometry. He is interested in understanding the molecular mechanisms of neurodegenerative diseases as well as advancing diagnostics for women's health.

ABOUT THE COLUMN EDITOR

Anurag S. Rathore

Anurag S. Rathore is a professor in the Department of Chemical Engineering at the Indian Institute of Technology in Delhi, India.