Computational methods in preformulation study for pharmaceutical solid dosage forms of therapeutic proteins

1 Introduction

Formulation development requires a balance between physical and chemical properties to achieve a robust formulation with the desired physical and bioequivalent results. Stages of formulation development include description of formulation platforms for preclinical studies, development of preclinical formulations, pre-formulation studies of active pharmaceutical ingredient (API) to support drug–product development and finally formulation development. Preformulation is a branch of pharmaceutical sciences that utilizes biopharmaceutical principles in the determination of physicochemical properties of a drug substance. Preformulation testing encompasses all studies enacted on a new drug compound in order to develop a stable, elegant and biopharmaceutically suitable dosage form and successful commercial product. Goals of preformulation studies are to choose the correct form of API, to evaluate its physical properties and generate a thorough understanding of the material’s stability under various conditions, thereby leading to the optimal, marketable drug delivery system. Preformulation study is therefore considered an exploratory tool initiated early in the development of any pharmaceutical dosage form [1].

2 Challenges to formulation development of therapeutic proteins

Significant strides in the field of biotechnology have opened up possibilities for large-scale production of therapeutic proteins which have immense therapeutic potential in the treatment of a range of diseases, starting from rheumatoid arthritis to cancers. Different therapeutic proteins available in the market include monoclonal antibodies, erythropoietins, interferons, growth factors, insulin, interleukins, tissue plasminogen activator, blood clotting factors and replacement enzymes. Monoclonal antibodies are highly preferable owing to their target specificity. Worldwide sales of monoclonal antibodies are expected to reach US$125 billion by 2020 [2]. However, complex nature of these protein macromolecules and their marginal stability even in solid state act as barriers to successful formulation development of therapeutic proteins. Other major problems associated with therapeutic proteins are their exogenous sources, variations in processing condition of the same molecule from one manufacturer to the other, inter-batch differences in the conditions and finally heterogeneity. Since the processing involves numerous steps, minor modification in a single step can be detrimental to the stability of the protein molecule [3, 4]. Preformulation studies therefore form an integral part of protein-based product development which will accelerate clinical manufacturing and investigational new drug filing process. A thorough understanding of physicochemical characteristics of the protein molecule is essential towards setting up the preformulation study protocol. Stability of protein molecules is affected by manufacturing variables, sterilization conditions and container-closure system. The two major issues that are of primary concern during development of solid dosage forms for therapeutic proteins include aggregation propensity at various stages of manufacturing and possible immunogenicity with fatal outcomes [5].

Assessment of physical stability of a protein-based therapeutic is a challenging task as it is linked to the primary sequence of amino acids in the protein. Secondary structure of the protein or its conformation results from intra- and intermolecular hydrogen bonds resulting in the formation of a typical geometric shape. The conformations that usually occur are α−helix, β−sheets, turns, bends and irregular random coil [6, 7]. Stability and biological activity of proteins are closely linked to their global flexibility, fluctuations and other dynamic processes. Flexibility which is essential for its activity poses a challenge to the manufacturers as it compromises on the stability. Although the unfolded state of the protein is unstable, the folded state is only marginally stable and any alteration in manufacturing conditions or in the neighbouring environment may cause various types of instability problems for the protein molecule such as aggregation, degradation by oxidation, hydrolysis, deamidation and lastly, inactivation. Thermodynamic stability of proteins is thus a complex balance between folded and unfolded conformations. Protein stability is greatly affected by those factors which disturb the delicate balance between the stabilizing and destabilizing forces. Protein dynamics should be studied in detail in order to develop and optimize manufacturing conditions for protein-based drug delivery systems. During manufacture and storage, proteins are vulnerable towards aggregation, leading to degradation, denaturation and resultant loss in biological activity. Rational design of protein-based biopharmaceuticals and self-assembled protein-inspired supramolecular aggregates requires a thorough understanding of the protein structure, thermodynamic events and factors responsible for switching of conformation. Information about the thermodynamics of the process of unfolding enables redesigning of proteins, enzymes and antibodies with better solubility and stability through modification of amino acid sequence [8].

3 Aggregation of therapeutic proteins

Protein aggregation is a natural phenomenon because it results in the formation of non-covalent bonds which are very similar to those present in native, stabilized structure [9]. Aggregation-induced degradation with antibodies may evoke fatal immunological reaction. Elevated temperature during storage or mechanical vibration during transportation can contribute to aggregation [10]. Various other physical stress factors have been found equally responsible to cause aggregation and include pH, ionic strength, presence of metal ions and surface adsorption [11]. Although stability of protein-based dosage forms can be enhanced by the process of lyophilization or freeze-drying, the dried powder needs to be reconstituted with suitable vehicle prior to administration. However, aggregation may also occur in the lyophilized proteins. Lyophilization process variables and composition of the solid determine the extent of aggregation which is greatly affected by the nature of the protein [12]. Stringent processing conditions during lyophilization may result in thermodynamic instability and reversible unfolding of the protein. If no irreversible damages occur to the protein during storage or reconstitution, then refolding takes place as soon as product is reconstituted, and there is no negative impact on the pharmaceutical stability of the product [13]. Aggregation in proteins can also be induced by the antimicrobial preservatives added to multiple-dose protein formulations. Therefore, suitable biophysical computational tools and instrumental analytical techniques are essential for prediction of aggregation propensity and profiling and for studying the factors affecting alteration in conformation [14].

Local interactions (e. g. hydrogen bonding, steric interactions) are responsible for the formation of α- and β-sheets, loop structure which govern packing dynamics of the molecule to a specific 3D-structure and thus dictate the specific biological function exhibited by the protein. These interactions are constituted by networks of amino acid residues. These networks are assumed to represent a subset of all potential interactions of residues. Literature surveys have revealed the availability of different structural and computational methods to provide insight into amino acid networks. The methods which can provide valuable information even in the absence of any data on protein structure include multiple sequence alignments (MSAs) and algorithms like statistical coupling analysis (SCA), mutual information (MI), McLachlan-based substitution correlation (McBASC) and observed minus expected square (OMES). These computational methods will guide the scientists in the design of engineered proteins with enhanced biological function or biobetters [15]. Physicochemical properties of amino acids, presence of aromatic side chains and charged residues are determinants for aggregation tendency of protein molecules [16]. There are various approaches for correlating solid-state protein stability with inherent protein characteristics such as glass transition temperature, moisture content and free volume measurement with the help of instrumental techniques relying on bulk properties or population averaged properties. However, meaningful correlation cannot be established always because protein instability is highly dependent on the nature of the functional groups [12].

4 Computational tools in assessment of immunogenicity of therapeutic proteins

Therapeutic antibodies, exogenous enzymes, signalling peptides of well-defined structure and function can be immunogenic, if they are of non-human origin. Non-immunogenic or de-immunized variant of the antibody, immunotolerant variant of the enzyme with unaltered binding specificity, affinity and stability can be developed by grafting of key functional residues from an exogenous therapeutic antibody onto human antibody framework. Grafting can be done on a trial-and-error basis or it can be done on a rational basis with a prior knowledge of structure–function relationship. An important obstacle to grafting is the absence of common modular structure in these therapeutic proteins and also lack of homologous human counterpart. Computational methods have been used on a small subpopulation of peptides (T-cell epitopes) for identification of mutations to minimize MHC II binding, responsible for eliciting immune response. Epitope databases like Immune Epitope Database (IEDB) and the proprietary T Cell Epitope Database™ (TCED™) have also been used for the purpose. Improved computational algorithms will target amino acid sequence for prediction of immunogenicity of the entire protein and will ultimately reveal the best structure with reduced immunogenicity potential. Such dynamic programming-based algorithms will enable mutations at the most promiscuous amino acids that are elements of multiple overlapping immunogenic peptides and resulting in the elimination of over six epitopes per mutation [31, 32]. Although several predictive models have been employed till date, the success rate in accurate prediction is not very encouraging. Low success rate is attributed to lack of sufficient knowledge relevant to the underlying mechanisms for immunogenicity of therapeutic proteins [33].

5 Conclusion

Owing to their large, complex and conformationally heterogeneous structures, therapeutic proteins are vulnerable to different instability problems induced by various physicochemical and mechanical stress factors during the various processing steps starting from the first process in multi-step manufacturing to administration. Propensity of the molecules towards aggregation needs attention as it can have deleterious effect on biological activity and safety. Therefore, preformulation studies need to be conducted prior to development of stable and efficacious solid dosage forms for the therapeutic proteins. Stability assessment is of primary concern in the preformulation study which utilizes different low-, intermediate- and high-resolution instrumental methods. Application of computational tools to low-resolution data helps in the prediction of secondary and tertiary structure of proteins with high level of accuracy. Data extracted from such analyses can be exploited in the redesigning and engineering of therapeutic proteins or biobetters with improved solubility and stability, which can be easily developed into a stable and effective formulation.