Abstract
Introduction
Modern, industrialized drug discovery usually starts with the selection of a molecular target that has strong disease association and preferably some preclinical or clinical validation. The goal of pharmaceutical and biotechnology companies is to discover and develop safe and efficacious drugs that either alleviate the symptoms of a disease or preferably halt or reverse its progression, often by modulating the function of a known validated target. After targets have been selected, an increasingly common question that must be asked during the early life of a drug discovery effort is the choice of therapeutic modality, as reviewed in this special issue of the Journal of Biomolecular Screening by Smith. 1
Choice of Modality
At one end of the spectrum of potential modalities, we have small molecules that are synthesized or manufactured at scale in medicinal chemistry facilities. Small-molecule drug discovery and development is a well-established discipline that relies on synthetic chemistry to produce, in a relatively inexpensive way, low-molecular-weight substances that are often administered to patients orally. Small-molecule drug discovery emerged more than 100 y ago with the introduction of important medicines such as morphine, aspirin, and penicillin, and today, more than 90% of all drugs on the market are formulations of small-molecule chemical entities. Because of their relatively straightforward production methods, small-molecule drugs are rapidly susceptible to significant competition from generic copies when their composition-of-matter or methods-of-use patents expire.
At the other end of the spectrum, we have large, complex, protein-based molecules, collectively called biologics. Biologic drug discovery and development is an increasingly attractive and important discipline that has advanced rapidly in recent years to yield dozens of approved products. Biologic drug discovery relies on molecular biology and genetic engineering approaches to produce protein-based drugs using genetically-modified living cells. Isolated and purified recombinant human insulin (Humulin) for the treatment of diabetes was the first genetically engineered drug to be approved by the U.S. Food and Drug Administration (1982). Other recombinant proteins that have been approved include multiple versions of human growth hormone (since 1985), multiple interferons (since 1986), and various growth factors (since 1989). Etanercept (Enbrel; 1998) is an example of a fusion protein that combines the tumor necrosis factor receptor 2 and the Fc region of an IgG1 antibody, whereas exenatide (Byetta; 2005) is an example of a synthetic peptide agonist of the GLP1 receptor, a member of the Class B family of G-protein–coupled receptors (GPCRs).
The first monoclonal antibody (anti-CD3) to be approved (1985 and subsequently withdrawn) was muromonab-CD3 (Orthoclone OKT3), for the management of transplant rejection. This drug was a murine antibody, which caused issues in patients (e.g., an immune response and poor pharmacokinetic properties). Subsequently, novel strategies were adopted in an attempt to produce less immunogenic murine antibodies, leading to the development and approval of chimeric antibodies, such as abciximab (ReoPro; 1994) and rituximab (Rituxan; 1997), and humanized antibodies, such as daclizumab (Zenapax; 1997 and subsequently withdrawn), palivizumab (Synagis; 1998), and trastuzumab (Herceptin; 1998). Thanks to scientific advances including phage display and transgenic mouse technologies, fully human antibodies can now be produced: adalimumab (Humira; 2002) was developed using a phage display approach, whereas panitumumab (Vectibix; 2006) was developed using the XenoMouse technology. More recently, novel antibody scaffolds have been approved (e.g., the bispecific T-cell engager, blinatumomab; BLINCYTO; 2014).
In a major show of force, a majority (7) of the 10 best-selling drugs worldwide in 2013 were biologics, with Humira, Enbrel, and Remicade taking the top 3 spots. As the production of biologics is complex and relies on the use of living organisms, it is difficult to produce exact copies of them. Consequently, generic competition has not been as rife as in the realm of small molecules. However, the landscape is changing with the creation of licensure pathways for copycat biologics, called biosimilars, when (preclinical and clinical) data demonstrate convincingly that the product is highly similar to an already approved biologic.
Small molecules and biologics differ in more than just their size and method of generation: they are discovered in different ways, they have different mechanisms of action and are characterized in different ways, they are formulated and administered in different ways, and they are treated differently by the human body. Small molecules readily cross cell membranes and can easily access small pockets and crevices on proteins. For instance, small molecules can be efficacious modulators of intracellular enzymes with pockets that bind nucleotides (e.g., the ATP pocket in kinases or the cGMP pocket in esterases). In contrast, because of their large size, protein-based molecules do not readily cross cell membranes, and so most intracellular targets are effectively off limits for biologics. Where biologics excel is in the disruption of protein-protein interactions and in the targeting of secreted proteins and cell-surface targets. Although receptor tyrosine kinases have been targeted successfully (e.g., cetuximab; Erbitux, an antibody inhibitor of the epidermal growth factor receptor that was first approved in 2004 for the treatment of various cancers), significant challenges remain for the successful targeting of more complex membrane proteins such as GPCRs and ion channels (see Wilkinson et al. 2 ).
Lead Discovery and Optimization
Most pharmaceutical companies own or have access to large libraries of chemical compounds. Among the millions of chemicals in existence is a subset that possesses pharmacological activity at a molecular target of interest. A key step is to identify those active chemicals, most often using automated high-throughput screening (HTS), which involves testing each chemical once at a single concentration (typically 1–25 µM) in a biologically relevant biochemical or cell-based assay. Initial hits are generally not very potent (µM levels of affinity) or selective over related targets, and so a frequent challenge during the initial optimization of small-molecule drugs is to improve their pharmacological properties by synthesizing and characterizing thousands of novel analogs over the course of many months or years. Progress here can be accelerated using knowledge about the structure of the binding pocket to support the rational design of analogs with improved activity and selectivity.
In contrast, the early discovery phase for a biologic focuses on identifying the optimal sequence and scaffold for the drug. Starting points for the development of antibody-based drugs include phage display libraries and bespoke hybridoma-derived libraries of antibodies that are generated on a target-by-target basis. Phage display libraries consist of large collections of variable region (VH and VL) antibody fragments displayed on the surface of bacteriophage. Antibody fragments with high affinity for the target can be identified and isolated through several rounds of selection against the target antigen in vitro followed by the determination of the DNA sequence encoding these fragments. The large size of phage libraries (typically >1012) permits the sampling of a large number of diverse variants, more than the largest small-molecule libraries.
In the hybridoma approach, antibodies can be often generated in vivo by immunizing animals such as mice and rabbits with an antigen that represents the target of interest. The antigen could be a soluble protein, such as a cytokine, or it could be a cell-surface target, such as a receptor. The antigen may be the full-length target or it may be a truncated version that contains functionally important domains or peptide sequences. For cell-surface targets, the antigen may be overexpressed on by engineered cells and administered to the animals either in the form of whole cells or membrane fragments prepared from the cells. The isolation of cell lines expressing high levels of the target can be challenging, especially for complex, multisubunit receptors (see Butler et al. 3 ).
Alternatively, DNA encoding the antigen may be administered to the animal so that the protein of interest is expressed in vivo. After a period of several weeks, putative antibody-producing B-cells can be isolated and fused with myeloma cells to yield thousands of hybrid cells (hybridomas) that display properties of both B-cells (i.e., they produce and secrete antibodies) and myeloma cells (i.e., they divide indefinitely). Single hybridoma cells can be selected if they secrete antibodies that bind to and modulate the functional activity of the target of interest. As with small-molecule HTS, this screening process can be automated to allow routine testing of thousands of hybridoma samples, as demonstrated by Tickle et al. 4 A variety of assay technologies are typically used to detect binding of antibodies to isolated proteins, for example, homogeneous time-resolved fluorescence, 5 or proteins expressed in cells.6,7 Unlike small-molecule starting points that emerge from HTS campaigns, active antibodies are often very potent (pM levels of affinity) and extremely selective for the target of interest; so selective in fact that they often do not cross-react with homologous targets in species such as rodents, dogs, or nonhuman primates that are most commonly used for in vivo efficacy, pharmacokinetic, and toxicologic studies.
After selection, the most interesting antibody hits will be sequenced and then developed into optimized lead candidates through protein engineering. Protein engineering involves modification or substitution of specific amino acid residues, either rationally (leveraging structural knowledge) or through directed evolution. Protein engineering goals may be to increase antibody expression, improve the lead candidate’s druglike properties, and reduce the likelihood that it may provoke an immune response in patients. Protein engineers may also seek to improve product stability by reducing the likelihood of aggregation, degradation due to chemical modifications (such as oxidation of methionine residues and deamidation of asparagine residues), or fragmentation by clipping. In certain cases, protein engineers may also seek to improve the lead candidate’s binding affinity and efficacy at the target, its selectivity over related targets, as well as its cross-reactivity with the target in other species, most notably rodents and nonhuman primates, to enhance its efficacy in preclinical disease models. It is also during the protein-engineering stage that antibodies with the desired properties can be reengineered to create novel antibody derivatives such as bispecifics, 8 antibody fragments, and non-IgG antibody formats.
Beyond the Research Phase
After a biologic lead candidate has been optimized, it needs to be produced in sufficient quantities for use in patients, both during clinical trials and after product launch. Development of a robust and reliable large-scale, high-yield production process is complicated and expensive and requires the use of engineered plant, bacteria, yeast, or mammalian cell lines. Many challenges must be overcome during process development, and rigorous quality control must be applied to both the production cell line and to the product itself. Production cell lines must be created, selected, characterized, stored, and banked for future use during the entire life cycle of the product. 9 Lot-to-lot monitoring must be conducted on unprocessed and purified bulk material as well as the final product. In contrast to many small molecules, biologics are usually not absorbed intact from the gastrointestinal tract, and so they must be administered by injection to achieve systemic exposure and efficacy. Before a biologic can be administered to patients, its structural integrity must be verified and the consistency of its biological potency and specificity must be confirmed. In addition, its thermal stability must be demonstrated 10 in an appropriate formulation 11 under well-defined and carefully controlled storage conditions.
Another key difference between small-molecule drugs and biologics can be found in the intellectual property arena. Typically, a small molecule is protected by a composition-of-matter patent that covers the chemical structures of the drug and related molecules. However, because of the uniqueness of protein-based drugs produced by living cells, the intellectual property landscape is very different for biologics and can even vary between different patent-granting agencies. 12
Summary
In this special issue of the Journal of Biomolecular Screening, we have assembled a series of articles that exemplify and discuss various aspects and challenges associated with the discovery, development, and manufacture of biologics with an emphasis on those topics that we feel will appeal to readers of this journal. We hope you enjoy them!
