1.4 Association and aggregation

Understanding proteins

metal ions or disul de bridges. Domains often form functional units, such as the calcium-binding EF hand (helix-loop-helix) domain of calmodulin. Because they are independently stable, domains can be swapped by genetic engineering between one protein and another to make chimeric proteins. It is independent because domains may often be cloned, expressed, and puri ed independently of the rest of the protein, and they may even show activity if there is any known activity associated with them. Some proteins contain only a single domain, while others may contain several domains. A protein domain is assigned a certain type of fold. Domains with the same fold may or may not be related to each other functionally because nature has used and reused the same fold many times in different contexts. The currently available protein 3D structures in the Protein Data Bank (PDB; http://www.wwpdb.org/) are a repository for the 3D structural data of large biological molecules, such as proteins and nucleic acids. The domains can be divided into four main classes based on the secondary structural content of the domain: • All-α domains have a domain core exclusively built from αhelices. This class is dominated by small folds, many of which form a simple bundle of helices running up and down. • All-β domains have a core composed of antiparallel β-sheets, usually two sheets packed against each other. Various patterns can be identi ed in the arrangement of the strands, often giving rise to the identi cation of recurring motifs, for example, the Greek key motif. • The α+β domains are a mixture of all-α and all-β motifs. The classi cation of proteins into this class is dif cult because of overlaps between the other three classes and, therefore, is not used in the CATH (class, architecture, topology, homologous superfamily) domain database. • The α/β domains are made from a combination of β–α–β motifs that are predominantly from a parallel β-sheet surrounded by amphipathic α-helices. Domains have limits on the size and vary from 36 residues in E-selectin to 692 residues in lipoxygenase-1, but the majority, 90%, have less than 200 residues with an average of approximately 100 residues. Very short domains, less than 40 residues, are often stabilized by metal ions or disul de bonds. Larger domains, greater than 300 residues, likely consist of multiple hydrophobic cores.

The HOS is stabilized through a large number of weak and strong bonds including weak noncovalent bonds formed ionic, dipoles (hydrogen bonds), nonpolar (hydrophobic), and van der Waals interactions. These 11

Biosimilarity: The FDA Perspective

12 bonds involve the interaction of amino acid side chains and the polypeptide chain. Since the transition from a polypeptide chain to HOS requires a signi cant loss of entry (structuring), it must be compensated for by enthalpy released from the forming of a bond (energy is released when a bond is formed); as a result, the protein structure can remain a dynamic state of structuring that may affect its activity as well as its stability. In most instances, the changes are transitory, and the protein returns to its native structure. However, the possibility of dynamic changes to a protein structure makes it possible for a molecule to have a different activity if its physicochemical properties are altered; additionally, if there is aggregation, this may lead to a loss of activity and a likely increase in the immunogenicity of the protein. Protein aggregation is caused by two factors: colloidal and conformational stabilities. The attractions on the surface of proteins can make colloidal dispersions that can be dynamic and signi cantly affect the safety and the effectiveness of proteins under stress conditions; the conformational changes are brought about by the hydrophobic interactions of the buried functional groups. There is a likelihood of both types of aggregates and, in some instances, one leads to another. So far, the regulatory authorities have not focused on these differentiations but over time, it is likely that these would be included as part of the risk analysis of the manufacturing process.

There is also a likelihood of aggregation due to molecular crowding when the drug is exposed to a high concentration of other proteins in the plasma. Would it ever be a requirement to study the nature of a circulating protein drug? This remains to be seen. However, a recent trend in the reformulation of proteins like adalimumab and rituximab in high-concentration formulations is an alarming trend; motivated by intellectual property (IP) protection as these drugs come off patent, the originators are reformulating their products, without fully realizing that the molecular crowding at the site of administration, if not in the vial or the syringe, is likely to increase the aggregation potential. The regulatory agencies should require a demonstration of safety at this level when a formula change request is made. This aspect of safety consideration is a topic of a citizen’s petition led by the author with the FDA.

The HOS a protein takes is intrinsically dictated by its primary structure and posttranslational modi cations (PTMs). In the 1960s, Cyrus Levinthal proposed an interesting observation regarding protein folding. In a 100-residue protein, allowing 5 × 1047 conformations possible and if each con rmation takes 1 ps, it will take 1018 times as long as the age of the universe. It is truly amazing how a protein HOS is repeatedly formed approximately the same way, even if there are a few defects left in the primary structure. It is this possibility of variability that makes the development of biosimilar products challenging. Misfolded proteins often reach a stage of energy level that may be dif cult to overcome and return them to their native state—the

Understanding proteins

conditions under which proteins are manufactured can signi cantly alter this pro le.

Protein synthesis involves a complex array of cellular machinery, primarily ribosomes. Proteins are synthesized from the N-terminus to the C-terminus in a sequential manner at a rate of 50–300 amino acids/ minute; the folding begins once the chain has acquired 50–60 amino acids—cotranslational protein folding that constraints and limits the pathways a protein can take into HOS, and this may explain why Levinthal’s Calculations come short.

Chaperones are proteins that help other proteins fold correctly in addition to proteolytic apparatus available in the cells. There are some proteins that have no well-de ned HOS. These are disordered or unstructured random coils, like the synthetic polymer chains or denatured proteins. This state may be a transitory state during the binding process and may be responsible for a multitude of protein actions in the cell. This intrinsic disorder creates a challenge to the demonstrate structure–function relationship, and whereas these aspects are not yet recognized by the regulatory authorities, it is only a matter of time when the biosimilar product developers may be required to demonstrate the disordered state comparisons as well—that will signi cantly raise the bar on the development of biosimilar products. There are two types of proteins that can be labeled as unnatural construction—the fusion proteins or the conjugate (e.g., pegylated) proteins and very a large assembly of virus particles or nanoparticle delivery systems. The fusion of an Fc part of an antibody (typically an immunoglobulin G1 [IgG1] antibody) with that of another pharmaceutically relevant protein through recombinant genetic technology results in fusion proteins. The Fc portion of an antibody increases the circulation time just as does the pegylation; examples include fusion of Fc to the blood-clotting factor VIII and factor IX. In June 2014, the FDA approved Eloctate, antihemophilic factor (recombinant), Fc fusion protein, for use in adults and children who have hemophilia A. Eloctate is the rst hemophilia A treatment designed to require less frequent injections when used to prevent or reduce the frequency of bleeding. In March 2014, the FDA approved Alprolix, coagulation factor IX (recombinant), Fc fusion protein, which is a recombinant DNA–derived coagulation factor IX concentrate. It temporarily replaces the missing coagulation factor IX needed for effective hemostasis. Etanercept is a fusion protein produced by recombinant DNA technology. It fuses the TNF receptor to the constant end of the IgG1 antibody. First, the sponsors isolated the DNA sequence that codes the human gene for soluble TNF receptor 2, which is a receptor that binds to TNF-alpha. Second, they isolated the DNA sequence that codes the human gene for the Fc end of IgG1. Third, they linked the DNA for TNF receptor 2 to the DNA for IgG1 Fc. Finally, they expressed the

13