Shark variable new antigen receptor biologics – a novel technology platform for therapeutic drug development

Abstract

Introduction: Biologics drugs have succeeded in achieving a commercial dominance in the global market for new therapies and large pharmaceutical companies' interest remains strong through a continued commitment to pipeline development. It is not surprising, therefore, that next-generation biologics, particularly antibody-like scaffolds that offer many of the advantages of the original biologic drugs but in simplified formats, have entered the clinic as competing substitute therapeutic products, to capture market share.

Areas covered: Specifically, this paper will position shark-derived variable new antigen receptors (VNARs) within an overview of the existing biologics landscape including the growth, diversity and success to date of alternative scaffolds. The intention is not to provide a comprehensive review of biologics as a whole but to discuss the main competing single-domain technologies and the exciting therapeutic potential of VNAR domains as clinical candidates within this context.

Expert opinion: The inherent ability to specifically bind target and intervene in disease-related biological processes, while reducing off-site toxicity, makes mAbs an effective, potent and now proven class of therapeutics. There are, however, limitations to these ‘magic bullets’. Their size and complexity can restrict their utility in certain diseases types and disease locations. In contrast, a number of so-called alternative scaffolds, derived from both immunoglobulin- and non-immunoglobulin-based sources have been developed with real potential to overcome many of the shortcomings documented for mAb treatments. Unlike competing approaches such as Darpins and Affibodies, we now know that shark VNAR domains (like camel VHH nanobody domains), are an integral part of the adaptive immune system of these animals and have evolved naturally (but from very different starting molecules) to exhibit high affinity and selectivity for target. In addition, and again influenced by the environment in which they have evolved naturally, their small size, simple architecture, high solubility and stability, deliver additional flexibility compared to classical antibodies (and many non-natural alternative scaffolds), thereby providing an attractive basis for particular clinical indications where these attributes may offer advantages.

Keywords::

• binding domains
• biologics
• drug development
• mAb
• novel scaffolds
• shark
• variable new antigen receptor

1. Overview of the market

Since the 1970s, advances in recombinant DNA, protein engineering and phage display technologies [1-10], coupled with a greater understanding of disease processes, have resulted in the development and production of the growing number of biopharmaceuticals or biologics currently used by clinicians in their assessment and treatment of patients. According to Leader et al. [11], biologics can be divided into four main categories: i) replacement protein therapies with enzymatic or regulatory activity such as insulin; ii) specific targeting protein therapeutics including antibodies and alternative binding domains; iii) protein-based vaccines; and iv) protein-based diagnostics. Recent years have seen the biologics market go from strength to strength to a position currently where they dominate the top 10, billion dollar revenue generators or so-called blockbuster drugs. This growth is predicted to continue from a market value of $150.1 billion in 2011 to $251.8 billion by 2017, representing a compound annual growth factor (CAGR) of 9% [12]. Worldwide spending on new drug development reflects this strong growth with reported drug development expenditure in 2012 to be in excess of $80 billion with spending on biopharmaceutical research accounting for around 50% of this total budget. The USA currently represents the greatest market segment for biologics, due in part to the high price for this drug class; however, this is likely to change over the next 5 years or so with developing countries such as China, Brazil, India and Russia gaining an increased importance in the global marketplace [13].

2. Existing technologies

The past 10 years have been tough for many drug discovery and development companies. There has been a paucity of new small molecule entities entering the clinic and an urgent need to expand and identify new disease targets. In this climate the attractiveness of complex, multi-domain drug candidates which show shorter drug development timelines and are challenging to copy have provided a rare opportunity to many companies and this in turn has driven the recorded increase in biologic resource and development activity. Within the realm of biologics, mAbs are currently the king. In 2011, they represented close to 32% of overall revenue (∼ $48 billion). This is expected to grow further to a staggering $90 billion by 2017, giving an estimated CAGR of 11.8%. As of February 2013, 39 mAbs had been approved for therapeutic use and over 350 were in clinical development [12,14].

Inherent advantages of mAbs over small molecules include high levels of specificity for target, reduced off-site toxicity and the efficacy benefits of induced cell death and long serum half-life, both achieved through the Fc domain. These latter attributes can be altered by molecular engineering to further tailor potency and pharmacokinetic properties, building flexibility into the design of mAbs, depending on the chosen application of the drug [15-20].

Cancer and chronic inflammatory disease are currently the primary indications for antibody therapy due in part to the systemic accessibility of the target antigens. Of the 28 USA and Europe market-approved mAbs reported by Reichert in 2012 [14], less than half recognised unique targets, with four targeting TNF-α and CD20 and two each targeting EGFR and VEGF. Whereas whole mAbs still dominate the list of clinically approved therapeutics, smaller, antibody fragments have been developed for both in vivo imaging and therapy. In particular, certolizumab pegol (Cimzia), which is a humanised anti-TNF-α Fab’ fragment conjugated to a PEG moiety, of ∼ 40 kDa, has shown good efficacy against rheumatoid arthritis and Crohn’s disease [14,21-25].

Although the clinical successes of mAbs and buoyant market growth have resulted in a significant shift of investment towards biologic pipeline development, there remain biological limitations and economic barriers to overcome for many forms of antibody therapy [26-28]. As small molecules are typically orally available, they can be readily prescribed and self-administered at home by the patient. This is simply unachievable for mAb therapy which currently has to be administered by intravenous infusion or subcutaneous injection often in a clinical setting by a medical professional. Although the frequency of dosing is greatly reduced compared to that of small molecules, the inconvenience, discomfort and cost of treatment has considerable impact on both the patient and the healthcare provider. Another limiting factor is the cost of manufacturing complex, large globular glycoproteins which requires considerable investment in good manufacturing practice level facilities, optimised eukaryotic expression systems and standardised purification protocols, to produce a consistent supply of clinical grade material. At 150 kDa, the size of mAbs also dictates their site of action within the human body, limiting this to systemically accessible targets or cell surface molecules with minimal tissue penetration. These considerations and perceived limitations has fuelled a continuous drive in the field of immunotherapeutic drug discovery to develop smaller, stable binding molecules that retain the specificity of antibodies but overcome some of the recognised cost and biological challenges. Examples of these antibody-based fragments and novel scaffolds have been the subject of many comprehensive reviews [29-34] and as such this paper will only focus on a handful that have progressed towards the clinic and have similar attributes to variable new antigen receptor (VNAR) domains.

3. How the technology works

To replace current gold standard therapies, next-generation biologics not only have to be as good as antibodies but also need to demonstrate additional features that differentiate them in this highly competitive landscape. Decreased size to enhance tissue penetration and reduce immunogenicity, increased solubility to assist formulation, simple molecular architecture that facilitates multiple re-formatting options and promotes greater yields for cost-effective manufacturing are just some examples of additional desirable properties. One candidate biologic and the subject of this technology review is the immunoglobulin (Ig)-based VNAR derived from the Ig new antigen receptor (IgNAR) found in cartilaginous fish [35,36]. IgNAR is known to play a role in the adaptive immune system of these animals and shares structural homology with cell adhesion molecules, antibody variable light (VL) chains and T-cell receptors, indicating a possible common ancestral receptor molecule that has evolved and been recruited by the immune system to function as antibody-like structures [35,37-39]. The secreted form of the IgNAR molecule is a heavy-chain protein dimer, each polypeptide comprising five constant domains and a single variable domain, with no associated light chain [35,36]. The variable domain of IgNAR or VNAR defines the specificity of the molecule and can be classified into different types based on the position and number of non-canonical cysteine resides within the hypervariable (HV) and framework (FW) regions [40,41]. To date, three main categories of IgNAR variable domains (VNARs) have been defined in the literature, as isotypes I, II and III. Interestingly, the type I VNAR has only been identified in nurse sharks and may be unique to this species [37]. This isotype possesses two cysteine residues in FWs 2 and 4, and two paired cysteines in CDR3. From analysis of the crystal structure of a type I VNAR in complex with lysozyme, it was determined that germline cysteine residues present in FWs 2 and 4 are required to form disulphide bridges with those in CDR3, thereby creating a tightly packed structure within which the CDR3 loop is constrained towards the HV2 region (Figure 1) [42].

Figure 1. Illustration of VNAR isotypes showing the similarities and differences between different isotypes of VNAR domain.

Type II VNAR domains possess the characteristic protruding CDR3, created by intra-molecular disulphide bonds formed between cysteine residues in CDR1 and CDR3 [36,37,43]. Type II sequences typically have shorter CDR3s than type I with an average of 15 and 21 residues, respectively.

The remaining class of VNAR described in the literature is type III, the expression of which is confined to young sharks (< 1 year old) [44]. Type III VNARs have highest similarity to the type II VNARs with the addition of a conserved tryptophan residue in CDR1. The D regions of shark genes, which form CDR3 and give rise to such diversity in types I and II VNARs, are fused with the V-gene in type III neonates and results in short (∼ 15 residues) limited CDR3s lacking in sequence diversity. It is hypothesised that type III VNARs are a first line of defence to protect newborn pups against a common pathogen or may play a role in the development of the adult immune system. This isotype is then quickly overtaken by the more mature classes of VNAR which provide a more expansive immune repertoire [44]. In addition to types I, II and III, there are also variants that contain only the highly conserved canonical cysteine residues at positions 35 and 107 that act to hold the Ig fold and lack any other disulphide bridges throughout the molecule. Liu et al. [45] and Streltsov et al. [37] referred to these as type IIb, a nomenclature that we have also adopted to assign these domains to different isotypes. An expansion of this designation has also been employed to describe a very similar domain to the type IIb which lacks the non-canonical cysteine residues but retains an invariant tryptophan in CDR1, like the type III isotype – this has therefore been called type IIIb. All currently named isotypes are shown in Figure 1. Unlike conventional antibodies, VNARs have a severely truncated CDR2 region [42]. To compensate for this, diversity is achieved by long variable protruding CDR3s, with additional diversity evident in CDR1 and also through two HV regions (HV2 and 4), unique to this class of molecule. The benefit of a structurally complex CDR3 combined with supporting intermolecular disulphide bridging is that VNARs form more compact structures with unusual paratopes which can access cryptic epitopes such as pockets or grooves within the target antigen [42,46-48]. In addition to this, the deleted CDR2 makes VNARs the smallest naturally occurring antigen-specific binding domains in the vertebrate kingdom with a molecular mass of ∼ 12 kDa [42]. A comparison of relative sizes of antigen-binding domains currently being developed for therapeutic applications is shown in Figure 2.

Figure 2. Relative size of Ig and Ig-like biologics is shown. Molecular models depicting the relative size of four different antibody-based scaffolds currently under clinical development are shown. VNAR and VHH are illustrated in two potential therapeutic formats. In general, the VNAR protein is ∼ 20% smaller than the VHH equivalent. The authors wish to thank Springer Science and Business Media, LLC and Landes Bioscience for their kind permission for inclusion of this modified figure from Barelle et al., 2009 [41].

4. VNARs as therapeutic candidates

Naturally designed to bind antigen with high affinity and selectivity, VNARs are inherently good candidates for therapeutic development [37,49-52]. Functioning naturally in the harsh environment of shark sera that consists of 350 mM urea and 1000 mOsmol, they exhibit tremendous stability and solubility [40,53,54]. Their simple single chain structures afford remarkable refolding properties after exposure to high temperatures (> 80°C), chemical and enzymic degradation [49,53-55]. Perhaps the greatest differentiating attributes from classical antibodies are their small size and unique topologies, opening up opportunities to penetrate tissues and bind to novel, cryptic targets intractable to existing larger more globular binding domains.

4.1 Drug discovery

Identifying a naturally occurring antibody-like scaffold with the potential for clinical development is the end point of a process that could now be considered validated and de-risked. The VNAR immune platform and drug pipelines now boast: robust means of isolating therapeutically relevant VNARs and selecting those with the desired binding affinity, the ability to engineer and adapt the biophysical properties of lead clones so they are fit for purpose and the means of high expression and purification in both eukaryotic and prokaryotic systems. Two platforms are available for the isolation of VNARs. The first is based on the role of IgNAR as part of the adaptive immune system of sharks. It has been shown now, in at least three different species of shark, that an IgNAR response can be elicited in response to antigen challenge [49,51,56]. As VNAR domains are amenable to phage display, a simple blood sample can be taken from these animals after an iterative process of immunisation followed by boosts, RNA extracted and the VNAR repertoire amplified from cDNA generated from this total message [57,58]. From this point cloning into a standard phagemid vector and selecting against target should provide positive hits. A second and complimentary means of isolating VNAR domains is the construction of a naïve or semi-synthetic phage display library based on naturally occurring VNAR domains that often include significant additional diversity through the engineering of the CDR regions [45-48,55,59-64]. Both these methods have pros and cons. The evolutionarily distant position of the shark immune system encourages a good response to many mammalian proteins. Therefore, immunisation typically provides high-affinity domains directed against antigen following an in vivo maturation processes. This process is a little slower than the related process seen in mammals as it can take ∼ 4 – 6 months to achieve the desired response. Selection from a synthetic library shortens this time frame, however the level of success (specificity and affinity) is dependent on the quality and size of the library being screened and generally will deliver domains of lower affinity that may require further refinement through in vitro maturation. Both methods, however, have been used successfully to isolate a number of VNAR domains against multiple target classes as illustrated in Table 1.

Table 1. Published VNAR domains isolated by selection against specific targets.

Target	Application	Source	Ref.
VHSV	Anti-viral	Semi-synthetic, Banded houndshark	[63]
HEL	Proof of concept	Semi-synthetic, Banded houndshark	[64]
TNF-α	Endotoxic shock	Immunised, Horn shark	[56]
HSA	Half-life extension	Immunised, Spiny dogfish	[51]
HBeAg of HBV	Anti-viral	Semi-synthetic, Wobbegong	[62]
Ebola virus	Immunodiagnostic	Immunised, Nurse	[50]
Cholera toxin	Biosensor	Naïve, Spiny dogfish	[45]
Staphylococcal enterotoxin B	Sensor	Semi-synthetic, Spiny dogfish	[55]
Ricin	Sensor	Semi-synthetic, Spiny dogfish	[55]
Botulinum toxin	Sensor	Semi-synthetic, Spiny dogfish	[55]
Leptin	Proof of concept	Semi-synthetic, Nurse	[59]
AMA1	Malaria diagnosis	Semi-synthetic, Wobbegong	[48]
Tom70	Diagnosis & therapy	Semi-synthetic, Wobbegong	[47]
HEL	Proof of concept	Immunised, Nurse	[49]
Gingipain K protease	Proof of concept	Naïve/semi-synthetic, Wobbegong	[46,61]

HBV: Hepatitis B virus; HEL: Hen egg lysozyme; HSA: Human serum albumin; VNAR: Variable new antigen receptor; VHSV: Viral haemorrhagic septicaemia virus.

Controlled mutagenesis through ribosome display has enabled the structure of a model type I VNAR to be interrogated and assessed based on retention of binding function [65]. This study highlighted some key hallmark residues within VNAR isotypes across different species and has illustrated the tolerance of these domains for alterations within the sequence and retention of function, informing possible sites for conjugation and humanisation. A recent publication by Kovalenko et al. [66], focussing on an anti-human serum albumin (HSA) VNAR domain isolated from an immunised dogfish [51] showed that humanisation is achievable without loss of function. The question of the need to humanise has yet to be answered in vivo but the humanised versions of these VNAR domains show negligible antigenicity in human dendritic cell assays (personal communication). Their small, compact structure and lack of aggregation when expressed suggest that this may not be as significant an issue as one might expect for a shark-derived protein. This position was further supported by a study in non-human primates designed to address the serum half-life of the anti-HSA VNAR mentioned above. Müller et al. [51] found no evidence of an immune response after two administrations (intravenous followed by subcutaneous) of the VNAR fusion protein. It is not always easy to predict but the key question of immunogenicity will also be affected by the target, the route of administration, serum half-life and the number of doses required to treat the patient. Ultimately, the true test of immunogenicity against a particular biologic in the patient population will only become apparent during clinical trials.

4.2 Drug development

The single chain composition of VNAR domains makes any molecular alteration relatively simple as the binding domain can be expressed as a single cassette without the requirement for two chains to associate and fold correctly. Dimer and trimer constructs can be expressed and purified and still exhibit functionality [51]. They are also amenable to re-formatting as Fc if the cell killing function and half-life extension is required [66].

Size, affinity and tissue penetration are correlated and in a cancer tissue penetration model, provided a high-enough affinity is achieved, the advantages of a smaller protein-binding domain will be realised [67]. This profile perfectly matches that of the VNAR making it a highly plausible candidate technology for solid tumour therapies. Being small, however, does usually mean that the protein will have a short serum half-life. Typically, any molecule below the renal filtration cut-off of ∼ 60 kDa will be cleared rapidly from the body [68]. This has value if the domain is designed to be an in vivo imaging agent where signal–to-noise ratios determine assay sensitivity but can be potentially limiting for the development of an efficacious drug. There are multiple ways to improve the pharmacokinetic profile of small proteins or peptides [69-80]. Fusion to an antibody Fc domain, as mentioned previously, is one way [81]; however, if being small is advantageous, then adding multiple extra domains (Fc) counteracts this size benefit. To circumvent this problem, the team at the University of Aberdeen have developed a VNAR-based half-life extension tool that increases the serum circulating time of a fusion partner from hours to weeks [51]. The domain, known as E06, was isolated from a spiny dogfish immunised with HSA and has exhibited efficacy across three different models of pharmacokinetics (PK) (mouse, rat and monkey). Tolerating both N- and C-terminal fusions, this opens up the opportunities to re-format not only VNARs but also scFvs and peptides as dimers or trimers, increasing affinity for target through avidity or increasing utility through bi- or tri-functionality. Development of this PK bio-tool increases the repertoire of re-formatting options open to VNAR biologics, correlating optimal design with clinical application as shown in Figure 3.

Figure 3. Schematic representation of VNAR formats showing different re-formatting options for VNAR domains.

Tailoring PK is a highly effective way of increasing the therapeutic window of short-lived molecules, decreasing doses and reducing the frequency of administrations. This provides a huge benefit to patients who currently have to undergo invasive transfusions of drugs resulting in both discomfort and increased risk of complications and infection. Taking the combined PK data for E06 in fusion with a partner VNAR, and applying allometric extrapolation, the predicted half-life of this molecule in humans is ∼ 19 days (Figure 4). This value correlates extremely well with that measured for the half-life of albumin itself validating the potential use of E06 as an effective bio-tool to improve the PK of partner therapeutics [51].

Figure 4. Allometric scaling of the anti-albumin VNAR domain, E06 is shown. Comparison of the published half-life of albumin in mouse, rat and cynomolgus monkey (⧫) and that of 2V-E06 (▪) from the data published in Müller et al., 2012 [51] is depicted.

5. Alternative technologies

As we have already reported, the overwhelming success of antibodies as therapeutic drugs has ignited a great deal of interest in the biopharmaceutical sector. The goal of replicating an antibody’s exquisite specificity for target while bringing added value through a differentiating quality such as small size has been central to the recent progress seen for several alternative platforms. Many promising antibody-based binding domains and alternative scaffolds have resulted from this recent intense research activity. The critique of novel scaffolds that follows is by no means exhaustive but focuses on those entering or in clinical trials or with an approved product already on the market. Table 2 specifically collates the current information available about next-generation biologics in clinical trials.

Table 2. Next-generation scaffolds currently in the clinic.

Drug	Scaffold	Target/indication	Structure/origin	Commercial developer	Clinical trial phase	Year last verified
Kalbitor (Escallantide)	Kunitz domain	Hereditary angiooedema	Human protease inhibitor LACI-D1	Dyax	FDA approved drug	–
DX-890 (Depelestat, EPI-hNE4)	Kunitz domain	Acute respiratory distress syndrome	Human protease inhibitor ITI-D2	Debiopharm S.A./Dyax	Phase II	2008
ABY-025	Affibody	Imaging Her2 expression in breast cancer	Bacterial protein A	Affibody	Phase II	2013
ALX-0081 (Caplacizumab)	Nanobody	von Willebrand’s factor/cardiovascular	Camelid domain	Ablynx	Phase II	2014
ATN-103 (Ozoralizumab)	Nanobody	TNF-α and HSA/RA	Camelid domain	Ablynx	Phase II	2013
CT-322 (Angiocept)	Adnectin	VEGFR2/NHL/glioblastoma/CRC	FN3 domain of fibronectin	Adnexus/Bristol-Myers Squibb	Phase I/II	2009/10
ALX-0061	Nanobody	IL6R/autoimmune	Camelid domain	Ablynx/Abbvie	Phase I/II	2012
MP0112	DARPin	VEGF-A/ophthalmology	Ankyrin repeats domain	Molecular Partners	Phase I/IIa	2011
Angiocal (PRS-050)	Anticalin	VEGF/cancer	Lipocalins	Pieris	Phase I	2011
ALX-0761	Nanobody	Anti-inflammatory	Camelid domain	Ablynx/Merck Serono	Phase I	2014
ALX-0141	Nanobody	RANKL/musculoskeletal	Camelid domain	Ablynx	Phase I	2011
ALX-0171	Nanobody	RSVF-protein/respiratory disease	Camelid domain	Ablynx	Phase I	2012
GSK1995057	dAb	TNFR1/lung Injury	Human single-domain antibody	GSK	Phase I	2014
GSK2374697	dAb	Anti-HSA dAb-GLP1/diabetes	Human single-domain antibody	GSK	Phase I	2013
BI 1034020	Nanobody	Not disclosed/Alzheimer’s disease	Camelid domain	Boehringer Ingelheim	Phase I	2014

Ig and non-Ig based scaffolds in the clinic at the date this manuscript was submitted. Information was collated from references [105-107].

CRC: Colorectal cancer; GSK: GlaxoSmithKline; HSA: Human serum albumin; NHL: Non-Hodgkin lymphoma; RANKL: Receptor activator of nuclear factor kappa-B ligand; RSVF-protein: Respiratory syncytial virus fusion-protein.

5.1 Antibody-based biologics

As discussed, whole mAb sales currently dominate and represent ∼ 75% of the total biologics market in the USA and Europe [14]. Of this 75%, the largest grouping is humanised mAbs (36%), with their origins firmly linked to monoclonal production via hybridomas. However, almost the same number are now fully human (32%) indicating a shift in technology preference and the use of approaches that enable the selection of human binding domains from phage display libraries or the innovative creation of genetically modified mice that express ‘human’ Igs in response to antigen challenge.

Antibody fragments such as Fabs, scFvs and Fc fusions have been developed for clinical use and have been comprehensively described in a recent book by William and Lila Strohl in addition to a number of review papers [82-86]. For the purposes of this technical review, we will concentrate on camelid nanobodies and the single human domains known as dAbs. We consider these to be the most important antibody-based competitors to the VNAR approach. Discovered in the early 1990s, the naturally occurring single-domain antibodies in Camelidae (camels, dromedaries and llamas) known as VHHs or nanobodies were shown to form part of the adaptive immune system in these mammals [87]. The ability to raise antigen-specific nanobodies through immunisation led to the isolation of domains against multiple targets and the launch of a Belgium-based company, Ablynx that successfully developed these domains into drug candidates. Showing similarities to VNARs in terms of their structure and properties, these single heavy-chain binding domains exist naturally and exhibit high affinity for target. However, in sharp contrast to VNARs, VHHs have evolved directly from an IgG lineage and at some point they have lost their partner VL domain and have undergone a truncation of the CH1 region resulting in the remaining VH variable domain being fused directly to the hinge region. Isolated VHH domains have been the subject of extensive research for almost 20 years (VNARs around 10 years) and there are currently seven nanobodies in clinical trials (Table 2) with more than 25 programs in company pipelines across six major therapeutic areas, including cardiovascular, infectious disease, musculoskeletal, oncology, respiratory and immunology.

Single-domain human antibodies or dAbs were the basis of the technology platform developed by Domantis which was subsequently acquired by GlaxoSmithKline (GSK) in 2006. The technology comprises large synthetic phage display libraries built from isolated heavy or light human variable domains into which significant diversity was introduced through randomised engineering of the two CDR regions [88]. Three reported dAbs are currently in the clinic (Table 2). One interesting programme is the fusion of an anti-HSA dAb (GSK2374697) to GLP-1 for the possible treatment of diabetes. The inclusion of an anti-HSA dAb as part of this drug formulation has validated the approach of ‘piggy-backing’ on the long-lived serum protein as a method of increasing the half-life of biologically active but short-lived peptides in the body [89,90].

Just as fully human mAb can be derived from immunisation of an engineered mouse, so also a second platform developed by Crescendo Biologics Ltd allows for the isolation of fully human single domains from a transgenic mouse strain. This approach used as a starting point a ‘triple knockout’ (TKO) mouse where all three murine Ig loci were functionally silenced. TKO mice were then crossed with mice containing a yeast artificial chromosome construct that included both human heavy chain genes and murine constant genes. Following immunisation, these animals produce human heavy chain-only antibodies fused to a mouse Fc region. Initial success with the “Crescendo Mouse” was announced in January 2013, and the company is currently developing their own early stage pipeline of therapeutic products.

5.2 Novel scaffolds

In the past 20 years, around 30 different engineered binding proteins have been proposed as potential FWs for therapeutic drug development [29,30,33,91-93]. The generation of engineered protein scaffolds generally begins with a natural protein structure that is used as a scaffold for modification. Typically, a core molecular FW provides the structure and stability onto which a diverse library of flexible loops or exposed residues can be introduced to create a synthetic library of binding proteins analogues to the variable domains found in antibodies. When compared to classical antibodies, these novel binding protein repertoires can offer a number of potential advantages. With the more obvious advantages of a small size (penetration, expression, re-formatting), the synthetic approach allows for the selection of a core scaffold that is chosen because it possess desirable characteristics such as low immunogenicity, expressability, solubility, thermal stability and/or protease resistance. Although many remain in the earlier, exploratory stages of platform technology optimisation, the exploitation of engineered protein scaffolds is proving to be an attractive route for binding molecules suitable for therapeutic development. Approaches, such as Affibodies, DARPins, Kunitz domains, Adnectins and Anticalins have progressed towards clinical candidates, either for therapy or for in vivo diagnostic or imaging applications.

Kunitz-type inhibitors are the earliest example of successful modifications to pre-existing binding molecules to deliver a library of related binding structures. They are robust scaffolds of human origin, based on a small engineered disulphide-bridged serine protease inhibitor of 60 amino acids in length with a molecular mass of ∼ 7 kDa. Diverse binding can be achieved through the engineering of an exposed peptide loop. Kalbitor (Escallantide) became the first approved domain in this class in 2009 for the treatment of acute attacks of hereditary angiooedema [94]. A second drug candidate DX-890 (Depelestat, EPI-hNE4) is an inhibitor of human neutrophil elastase and is currently being assessed for the treatment of patients with impaired lung function [95-97].

Libraries of Affibody-binding sites can be created by the mutagenesis of 13 amino acid positions exposed on the surface of α-helixes 1 and 2 in the Z-domain of protein A from Staphylococcus aureus. Although binders to a number of targets have been successfully isolated and characterised, one potential disadvantage of this class of protein scaffold is their bacterial origin which may increase the risk of immunogenicity after repeated therapeutic administration to patients. However, their small size (6.5 kDa) offers wide applicability for in vivo diagnostic imaging, where rapid distribution and elimination are desirable features of a disease-specific biologic [98]. To date, the only Affibody enrolled in a clinical study is ABY-025 for PET imaging of HER2 expression in metastatic breast cancer.

DARPins are derived from natural ankyrin proteins and consist of at least three or more repeat motifs which generate a series of binding site loops at a large cysteine-free surface [34,92]. This structural arrangement encourages the construction of diverse DARPin repertoires and the selection from these libraries of DARPins with picomolar affinities for their target proteins. They have a molecular size range of 14 – 20 kDa (four- or five-repeat DARPins units). Molecular Partners has developed a portfolio of drug candidates based on DARPins for ophthalmology, oncology and inflammation. DARPin MP0112 is an anti-angiogenic drug that has now been used clinically to treat wet age-related macular degeneration and more recently diabetic macular oedema. In both cases, MP0112 reduces VEGF-A concentrations in the aqueous humour of the eye after intraocular injection and appears to stabilise visual acuity and reduce retinal oedema.

Adnectins are based on a fibronectin type III backbone with molecular mass of 10 kDa. They are structurally very similar to antibody variable domains, but without a core disulphide bond [99]. CT-322 (Angiocept) is the first engineered adnectin developed by Adnexus/Bristol-Myers Squibb and is an antagonist of VEGFR-2. CT-322 is currently in Phase II clinical trials for treatment of glioblastoma having completed Phase I studies without any significant adverse effects and with promising anti-tumour effects seen in some patients [100,101].

Anticalins® are proteins derived from human lipocalins which naturally bind, store and transport a diverse range of molecules in the body [102]. Recombinant Anticalins are capable of binding both proteins and haptens. Structurally they are barrel-shaped domains which are formed by eight anti-parallel β-strands connected by flexible loops and an attached α-helix. Diversity is engineered into the loop regions giving a repertoire of molecules from which those with the desired binding properties can be selected. Anticalin libraries have been constructed by Pieris, and clinical Phase I testing of Angiocal (PRS-050), an anti-VEGF PEGylated protein of 40 kDa, has been successfully completed in 26 patients with advanced solid tumours. PRS-050 was safe and well tolerated and had a half-life of 6 days without immunogenicity.

6. Conclusion

Biologic drug development is an exciting and dynamic field where innovation is driving the next wave of new products. Antibodies are big players in this industry and exemplify perfectly how nature can be harnessed and exploited to benefit humans. We are proposing that another naturally occurring class of molecules, shark VNAR domains, can become the next ‘small’ big thing. VNARs are perfectly designed and adapted to fit the requirements of drug discovery development and exhibit a range of significant additional attributes to give confidence that they too will become a valuable platform for multiple medical treatments.

7. Expert opinion

Antibody-based biologics have gone from strength to strength, demonstrated by their current and growing dominance in the therapeutic drug market. Such success breeds opportunity in the development of the next generation of these molecules to enhance their clinical impact and increase their economic value. A common thread across these new biological domains is the desire to retain affinity and selectivity for target while decreasing the size and complexity. There appears to be two main categories for these new molecules: naturally occurring antibody or antibody-like domains and alternative scaffolds that have been re-engineered to adopt the features of their natural counterparts.

Taking a current snapshot of these new biologic domains shows quite clearly that the nanobodies have progressed significantly with seven candidates presently in the clinic. Perhaps one aspect of this success can be attributed to the timeframe since these single-domain antibodies were discovered; however, it is clear from the breadth of the utility and application of these domains that a key strength is that they naturally exhibit the attributes required for drug development with additional benefits such as their size providing added value for many indications. Although this review has focussed on the therapeutic arm of development, these single-domain antibodies have also shown utility as in vivo imaging agents and in vitro diagnostics and are, therefore, excellent candidates for stratified medicine in the form of companion diagnostics [103,104].

It is the opinion of the authors that this broad utility and success can be translated directly from the camel VHH to the shark VNAR platform. In fact, the nanobodies by trailblazing a path into patients have signposted a rapid route for shark VNAR into clinical studies.

Correlations can be made regarding the natural role of shark VNARs and nanobodies as components of the adaptive immune systems in their relative hosts. It is fascinating that convergent evolution has resulted in two very different species developing a single chain binding domain repertoire to protect against external challenge. This is a huge advantage when it comes to desirable attributes in a therapeutic drug platform as the domains can be raised through immunisation resulting in in vivo maturation, with high affinity, highly selective binders. An important additional benefit of immunisation in the shark system is that these animals are evolutionary distinct from mammals (∼ 450 million years) providing a robust platform for raising a response against proteins that are highly conserved between warm-blooded animals.

A greater understanding of the natural repertoire of multiple shark species coupled with in vitro characterisation has resulted in the ability to construct and screen semi-synthetic VNAR phage display libraries. In contrast to the VHH approach, the increased number of isotypes present in different shark species has allowed the construction of more structurally and clonally diverse ‘naïve’ libraries from which single domain nanomolar affinity antibodies, against varied targets, have been quickly isolated. As such, VNAR domains can be isolated from a series of complementary platforms providing a route to additional diversity and candidate generation.

Size is another key attribute of these shark single domains. As discussed previously, there is a desire to retain the affinity and specificity of whole antibodies but capturing this within the smallest binding domain possible. VNARs are the smallest naturally occurring binding domains in the vertebrate kingdom (up to 20% smaller than VHH) and, therefore, have achieved this goal without the need for re-engineering. They possess four regions of high variability per domain (VHH has three) and, therefore, have at their disposal the greatest density of binding loops by size of any competing technology. When this is combined with the natural propensity of multiple different structural families to target these loops into crevices and pockets (as well as additional more flexible structures that demonstrate induced fit) [42,43], the potential of the VNAR technology as a potent novel drug discovery ‘engine’ is clear.

Article highlights.

• Biologics are dominating the therapeutic drug market with mAbs showing the greatest commercial impact and most sustainable market growth.

• There has been a shift in the development of therapeutic antibodies towards smaller, more stable, more flexible formats to increase their efficacy and broaden their clinical utility.

• There are a growing number of next-generation antibody-like and novel scaffolds entering biotechnology and biopharmaceutical company pipelines designed to address some of the limitations of conventional antibodies and open up new opportunities for novel disease targeting and alternative modes of administration.

• Shark single domains (variable new antigen receptors) have multiple natural and downstream attributes such as small size, high affinity, selectivity for target and ease of production that makes them attractive candidates for drug development.

Declaration of interest

The authors were supported by Scottish Enterprise and the Biotechnology and Biological Sciences Research Council (BBSRC). The authors have no other relevant affiliations or financial involvement with any organisation or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Notes

This box summarizes key points contained in the article.