Elsevier

Drug Discovery Today

Volume 26, Issue 6, June 2021, Pages 1409-1419
Drug Discovery Today

Review
Keynote
PepTherDia: database and structural composition analysis of approved peptide therapeutics and diagnostics

https://doi.org/10.1016/j.drudis.2021.02.019Get rights and content

Highlights

  • PepTherDia is a database containing 105 approved peptide pharmaceuticals.

  • 86% of approved peptide therapeutics and diagnostics are natural or naturally-derived.

  • Bimodal distribution of peptide molar mass, with the large majority <2000 g/mol.

  • Balance between polar and hydrophobic residues within the peptide structures.

  • 46% of approved peptides are cyclic with 5 to 7 membered macrocycles being most common.

As of 2020, there were >100 approved peptides with therapeutic or diagnostic applications. However, a complete database providing information on marketed peptides is not freely available, making the peptide chemists’ job of designing future peptide drug candidates challenging. Unlike the rules for small-molecule drugs, there is no general set of guidelines for designing a successful peptide-based drug. In this review, together with our freely available database (PepTherDia, http://peptherdia.herokuapp.com), we provide insights into what a successful peptide therapeutic or diagnostic agent looks like and lay the foundation for establishing a set of rules to help future medicinal chemists to design peptide candidates with increased approval rates.

Introduction

Following the approval of the first peptide therapeutic agent, the 51-amino acid (AA) hormone insulin, in 1923 1, 2, drug discovery has been progressively expanding into the chemical space between small molecules and large proteins. Subsequently, a significant number of peptides (and peptidomimetics) have received regulatory approval. Recently, peptides emerged as novel modalities for various applications in the therapeutic and diagnostic markets, providing new opportunities for the modulation of difficult targets. Since the second half of the last century, the number of peptides on the therapeutics and diagnostics market has steadily increased, reaching the milestone of >100 approved peptide drugs in 2020 (Fig. 1). These drugs represent a unique class of chemical compounds that bridges the gap between small molecules (typically molar mass <500 g/mol) and large biologics (typically molar mass >5000 g/mol). Occupying an intermediate region of complexity and molar mass, they combine many of the benefits of these two categories. The main disease areas presently treated with peptide drugs are metabolic disorders, cancers, and cardiovascular diseases, with emerging therapeutic applications in the areas of infectious diseases, pain, and urinary tract, gastrointestinal, and respiratory disorders 1, 3. Given that the incidence of metabolic disorders, cancers, and cardiovascular diseases in the Western world is increasing alarmingly 4, 5, 6 and the need for new effective medicines to treat emerging health problems [e.g., severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)/coronavirus disease 2019 (COVID-19)] is growing, it is likely that the demand for peptide drugs will continue to grow.

According to the latest report from Transparency Market Research, the worldwide market for peptide pharmaceuticals has been growing at a compound annual growth rate (CAGR) of ∼8%; this is expected to increase over time with the same trend, reaching a value of USD 50 billion in 2027 from USD 25 billion in 2018 [7].

As of November 2020, according to our database analysis, there were 105 peptide pharmaceutical products (see definitions later) with regulatory approval in the main pharmaceutical markets (North America, Europe, and Japan), of which 89 were peptide drugs and 16 were diagnostic agents. Moreover, numerous clinical studies involving peptide agents are in progress: 4859 in total, 468 of which are in Phase III trials [8]. This suggests that the pharmaceutical industry is committed to exploring the role of peptide therapeutics in modulating previously ‘undruggable’ targets and addressing unmet medical needs. The main advantages of peptides over small molecules and proteins are illustrated in Table 1. Despite these important benefits, the drug development process for future therapeutic peptides from laboratory to approval has traditionally presented many obstacles. In fact, unique challenges, such as chemical and physical instability, short circulating half-life, high proteolytic degradation, rapid renal clearance, poor membrane permeability, poor oral bioavailability, and low solubility must be addressed to bring a peptide to clinical use 1, 9. Nonetheless, lately, advances in drug delivery and emerging medicinal chemistry strategies have brought peptides to a significant renaissance, by overcoming their issues and eventually improving pharmacokinetic (PK) profiles and oral bioavailability.

Over the past decade, several research groups have tried to reach a better understanding of approved peptides and their properties. In 2010, Vlieghe et al. reviewed and listed the synthetic therapeutic peptides that have reached the main pharmaceutical markets (USA, Europe, and Japan) [10]. Six years later, Raghava and collaborators filed a repository (PEPlife) to provide the scientific community with data on peptide half-lives [11], followed, 1 year later, by a database containing 852 US Food and Drug Administration (FDA)-approved biologics, among which 28 were peptides [12]. Recently, an extensive review of approved peptide therapeutics targeting G-protein-coupled receptors (GPCRs) was published by Davenport and collaborators [13], demonstrating the dominant presence of such therapeutics in the market.

Nevertheless, to the best of our knowledge, a complete database with structural analysis, production methods, PK properties (i.e., terminal half-life and protein binding), indications, and routes of administration of regulatory-approved peptides has not been freely available online (as of November 2020). In fact, the information is scattered throughout the scientific literature and on various websites, making the search for approved peptides challenging. In addition, there is a crucial lack of rules that makes the medicinal chemists’ job of designing entirely new potential peptide drugs difficult. In fact, as already pointed out by Tyagi and colleagues [14], the ‘drug-likeness’ criteria used for small molecules [principally, Lipinski’s Rule of Five (Ro5) [15], but also models such as the central nervous system multiparameter optimization (CNS MPO) [16]] are not applicable to peptides because of their entirely different intrinsic properties and applications.

In this context, we have developed and made accessible online PepTherDia (Peptide Therapeutics and Diagnostics: http://peptherdia.herokuapp.com*), a manually curated database containing a searchable list of approved peptide drugs and diagnostic agents, with information on their physicochemical and PK properties, as well as their routes of administration and indications. Its purpose is to provide assistance to medicinal chemists and scientists in the field of peptide drug discovery. For the compounds enumerated and described in PepTherDia, we performed a detailed analysis of the structural features and collected information on their terminal half-life, plasma protein binding, indication, route of administration, production methodologies, marketing authorisation (year and agency of first approval), and origin of their design. The information contained in this database will be updated on a regular basis (e.g., yearly) with new approvals as well as new properties investigated. For the first time, in this review, we highlight important trends in peptide approvals and we provide insights into the features and characteristics that are common in approved peptide agents. We envisage that this information will aid the scientific community to more successfully design or prescreen candidates at an early stage of the peptide drug discovery process to increase longitudinal approval rates.

A challenge in designing this study was the breadth and diversity in the so-called ‘peptide’ molecule because of differences in structure, size, and composition. Therefore, the first thing the authors felt necessary to clarify was: what is a peptide? The International Union of Pure and Applied Chemistry (IUPAC) defines peptides as ‘amides derived from two or more amino carboxylic acid molecules (the same or different) by formation of a covalent bond from the carbonyl carbon of one to the nitrogen atom of another with formal loss of water’ [17]. By contrast, the recommendations of the IUPAC-IUB Joint Commission on Biochemical Nomenclature (JCBN) defines a peptide as a chemical entity presenting from 2 to 50 AA residues [18], while the currently used regulatory FDA definition delineates a peptide as ‘any polymer composed of 40 or fewer amino acids’, regardless of their production method 19, 20. Finally, the European Medicines Agency (EMA), instead of describing peptides based on their size, considers them as small molecules if chemically synthesised, while treats them as biological entities if they are extracted from natural sources or produced with recombinant methodologies [21], which highlights the fact that the definition of peptide remains ambiguous.

At this stage, the reader might state that this definition can be considered as a philosophical debate that could elicit controversial answers. Indeed, the scientific community differs greatly on where to stop using the term ‘peptide’ and start using the term ‘protein’. In Box 1, the definition of peptide that led our research, together with other terminologies are explained.

A repository of 105 compounds was obtained by searching in DrugBank [22], FDA and EMA web pages 19, 23, Pharmaceutical and Medical Devices Agency website [24], and Drug Central website [25]. The key inclusion and exclusion criteria used in the curation of PepTherDia are listed in Table 2. Examples of peptides not included and the reason for their exclusion are shown in Table 3. With the aim of providing each peptide with a complete profile comprising relevant information regarding terminal half-life, protein binding, therapeutic indications, and routes of administration, specific searches were carried out in DrugBank [22], National Centre for Advancing Translational Sciences web page [26], Drugs.com [27], and pharmaceutical companies’ websites, using the generic name of the individual peptide. References specific to each approved peptide as well as SMILES codes used to calculate the peptide molar mass values can be found on our website PepTherDia.

As discussed and defined earlier, peptide medicines generally comprise natural AAs, unnatural AAs, and non-amino acidic modifications. Fig. 2 provides an example of how the peptide daptomycin can be divided into the above components. However, in some cases, such as the glycopeptide antibiotics (dalbavancin, telavancin, oritavancin, and teicoplanin), this is a complex (if not impossible) task. Similarly, the complexity of some multicyclic peptides does not allow the unambiguous identification of a defined single macrocycle and, therefore, the members of each macrocycle were not counted here.

Each constitutional member can be further classified as polar, acidic, basic, nonpolar aliphatic, or aromatic based on its structural and physicochemical characteristics. For the natural amino acidic residues, the designations polar, acidic, basic, nonpolar aliphatic, or aromatic, derived from literature precedent [28], are generally ascribed by the nature of the side chain. Given that non-natural AA members form an amide backbone in the same way as natural AAs, they can be classified following the same principles used for natural AAs. By contrast, peptide modifications are a broad and varied structural class, and their classification requires consideration of their complete structure and the way in which they are conjugated to the peptide. For example, in daptomycin (Fig. 2), decanoic acid can be classified as a nonpolar aliphatic modification because the carboxylic acid moiety becomes part of an amide and its contribution to the final polarity predominantly increases lipophilicity.

The complete list of non-natural amino acids and modifications, together with their polarity classification, can be found on the PepTherDia website.

Thanks to improvements in synthetic and manufacturing technologies, it is now possible to synthesise ever-larger peptides in a short time, yielding high purities and quantities. Nonetheless, from our study, it emerged that most approved peptides (68%) are relatively ‘small’ peptides, comprising 2–16 constitutional members, with a second minor cluster (27%) of larger size peptides with ∼28–37 members (Fig. 3a). This is mirrored in a bimodal molar mass distribution in the ranges 300–1750 g/mol (major: 71%) and 2750–4250 g/mol (minor: 22%), with a remarkable lack of mid-length approved peptides (Fig. 3b). Hence, the data suggest that there are two main groups of peptides: low molar mass and high molar mass with only a few examples in between (e.g., sinapultide and ziconotide, with molar mass of 2469.45 and 2639.14 g/mol, respectively). To evaluate the possibility of bias in the peptide design, which might have led to the deliberate development of peptides of certain sizes to mimic specific biomolecules, it is necessary to analyse the origin of the peptide design case by case (Fig. 3e). In this context, it emerges that natural peptides account for 30% of our sample. However, peptide analogues account for 54%. Finally, heterologous peptides account for 16% of the sample. This underlines that heterologous peptides are difficult to design a priori and are mostly discovered by library screening. In light of these findings, we can state that, unsurprisingly, there is a clear trend (in 84% of the cases) toward following the route of inspiration from nature as a greater promise of success and that the bimodal distribution could be attributed to the characteristics of the natural molecules that have inspired the design. Examples that demonstrate this are the natural nonapeptide oxytocin and the 32-membered peptide calcitonin. Here, in both cases, their length results from the size of the natural molecule of origin.

Among all the constitutional members, most (around 81%) are represented by natural L-AAs. The residual 19% comprises non-natural AAs and modifications. A careful analysis of the AA residues contained in each approved peptide (Fig. 3c) showed that the most common AAs in the sequences are the nonpolar aliphatic leucine (L) and glycine (G), followed by the polar serine (S). By contrast, the least common residues are methionine (M), histidine (H) and isoleucine (I). This is largely in agreement with the occurrence of natural AAs in proteins: in nature, leucine (L) accounts for 9.1%, serine (S) for 6.8%, glycine (G) for 7.2%, and alanine (A) for 7.8%, being the most common amino acids. By contrast, methionine (M) and histidine (H) account each only for 2.3%, cysteine (C) for 1.9%, and tryptophan (W) for 1.4%, being the least common AA residues found in proteins [28]. The occurrence of cysteine (C) in pharmaceutical peptides is higher than in proteins, because of the frequent use of disulfide bonds as a tool for macrocyclisation (see ‘Conformational and shape properties’). In general, the AA composition of proteins and peptides is highly variable; some AAs might occur only once or not at all in a given peptide and might be repeated several times in another peptide sequence. An example of where this is seen is the repetition of the moiety KL4 in the peptide sinapultide, designed to mimic the C-terminal domain of the surfactant protein B [29].

D-AAs with natural side chains account only for a small percentage (∼4% of the total AAs with natural side chains) and are mainly represented by phenylalanine, alanine, tryptophan, and arginine. The selective replacement of L-AAs by their enantiomers (D-AAs) can protect the molecule from protease degradation [30]. This is a common technique to obtain proteolytic stabilisation by backbone modification, even if this causes conformational changes that might affect biological activity. An example is the somatostatin-like peptide octreotide, in which natural phenylalanine and tryptophan are replaced with their mirror-image forms, leading to a 100-fold increase in the terminal half-life 31, 32.

In peptide drug discovery, the use of non-natural AAs as well as conjugation with non-amino acidic members are common techniques to overcome peptide limitations and have been widely explored in both protein and peptide design 33, 34. In Fig. 3d, the most common non-natural amino acids and structural modifications in approved peptides are reported.

Position-specific incorporation of non-natural AAs bearing a variety of bespoke side chains can provide improvements in peptide properties, activities, and functions [35]. In this study, non-natural AAs were found to be present not only in heterologous peptides (e.g., ACE inhibitors, macimorelin, and argatroban) and peptide analogues (e.g., GnRH agonists, buserelin, carbetocin, icatibant, pasireotide, and carfilzomib), but also in one-third of natural peptides. In fact, from our analysis, the percentage of proteinogenic AAs in natural peptide antibiotics (e.g., capreomycin, vancomycin, bleomycin, and oritavancin) has been estimated to be between 0 and 10% of all the constitutional members, a value that is significantly lower than the average percentage of natural AAs in the pool of 105 approved peptides (81%).

Ornithine (Orn), 2,4-diaminobutyric acid (Dab), and 2,3-diaminopropionic acid (Dap) are homologues of lysine, where the variability in structures results from the difference in the number of side chain carbons. They all contain an amino group on the side chain, which exhibits a basic/ionisable contribution and, at the same time, allows opportunities for cyclisation or conjugation. Moreover, it has been reported that the use of Dap and Dab in antimicrobial peptides (AMPs) can prevent the hemolytic activity of positively charged natural AAs (i.e., arginine and lysine), solving a characteristic issue of AMPs [36]. AMPs have also shown improved stability to trypsin while retaining their biological activity, when arginine and lysine residues are replaced by Dab, Dap, or homoarginine [37], demonstrating that lysine homologues are not a suitable substrate of this hydrolytic enzyme. Likewise, lower susceptibility to tryptic hydrolysis has been reported after lysine replacement with Orn [38]. In our pool of approved peptides, Dap and Dab are only found in natural peptides: Dap in capreomycin, viomycin, and enviomycin, but Dab in colistin and polymyxin B. By contrast, Orn is used in a variety of peptide analogues (e.g., ornipressin, atosiban, anidulafungin, caspofungin, and micafungin) as well as in natural peptides (e.g., daptomycin and bacitracin). Overall, AA side chains bearing a primary amine, such as lysine and its homologues, represent one of the most frequently used AA moieties (∼7% of the total AAs). Specifically, the Dab residue was encountered 12 times in total, but appears only in two peptide structures (colistin and polymyxin B), whereas Orn appears only four times, each in different peptides (daptomycin, ornipressin, atosiban, and bacitracin).

4-hydroxyproline (Hyp) is a proline containing a hydroxyl group on the pyrrolidine ring. In general, the relatively frequent use of Hyp derivatives (nine in total in the pool of approved peptides, e.g., caspofungin, icatibant, and voxilaprevir, to name but a few) might be ascribable to the polarity enhancement, additional hydrogen-bonding ability, and the possibility of further conjugation gained with their introduction. Moreover, the presence of Hyp stabilises the triple-helical structure of collagen [39]. Similarly, this stabilisation can occur in peptide secondary structures.

Naphthyl-alanine (Nal) is often used to mimic tryptophan and to explore potential improvements in peptide pharmacological profiles 40, 41, 42; however, it is not clear whether 1-Nal or 2-Nal adequately replicate the effects of tryptophan aromatic interactions. In fact, as with any modification, the consequence of these replacements upon the peptide potency needs careful assessment, because it has been demonstrated that substitution of tryptophan with 1-Nal or 2-Nal decreases the potency of cholecystokinin analogues [43]. Nonetheless, this strategy has been successfully used in the development of GnRH receptor (GnRH-R) antagonists (abarelix, ganirelix, degarelix, and cetrorelix) and other peptides (lanreotide, nafarelin, and pralmorelin). Specifically, in GnRH-R blockers, the histidine–tryptophan motif of the natural hormone GnRH, has been replaced with the three-AA motif 2-Nal–(4-Cl)Phe–3-Pal, suggesting an intention to improve the aromatic contribution at the peptide N-terminus, which might be important for the binding and antagonistic activity at the receptor.

An additional common strategy used to improve peptide drug-likeness is the introduction of non-AA appendages to tune the PK properties. These are usually linked to the main chain by only one functional group (e.g., −COOH for fatty acids or −OH for sugars) and are generally attached to AAs containing polar functional groups (e.g., OH, NH2, and COOH) or to the N-or C-termini. Enhancement of stability, protein binding, and membrane permeability can be obtained through peptide lipid acylation, whereas improved solubility and bioavailability can be achieved through glycosylation 44, 45. Indeed, lipid acylation and glycosylation are the most common modifications encountered in the pool of approved peptides.

Lipid acylation is a post-translational modification of proteins that has found applications in peptide design to improve PK and pharmacodynamic (PD) properties while retaining the ability to bind the target receptor 44, 46, 47. Of marketed peptides, 13% present a lipophilic carbon chain attached to their structure and, in some cases, demonstrated prolonged terminal half-life (e.g., in oritavancin and dalbavancin, which have a terminal half-life of 245 and 346 h, respectively,) and high protein binding (>90% in most approved lipidated peptides). The length of the carbon chain may influence the half-life duration but, currently, we do not have enough data around approved lipidated peptides to state so. Other examples of approved peptides presenting a lipophilic carbon chain include the popular diabetes drugs and glucagon-like peptide-1 (GLP-1) receptor agonists, liraglutide (conjugated with palmitic acid) and semaglutide (conjugated with an octadecanedioic acid). They both bind with high affinity to plasma proteins (98–99% of the peptide bound), promoting greater peptide stability, which results in significantly extended half-lives of 13 and 168 h, respectively, compared with the parent GLP-1.

Carbohydrate groups are less frequent but still significant modifications, being found in eight out of 105 peptides on the market (e.g., bleomycin and vancomycin). Given the synthetic challenges that glycochemistry presents, these are typically found in peptides of natural origin. In these peptides, glycosyl units are attached to the main structure via an N-terminal amine group or hydroxyl group on the side chain, similar to recombinant glycoprotein therapeutics, in which carbohydrates are commonly N linked to asparagine or O linked to serine and threonine [48]. It is probably not by chance that these eight examples comprise a large aromatic core or aliphatic chain, in which hydrophobicity is balanced by one or more sugars. In fact, glycosylation improves the physicochemical and PK properties of peptide drugs through enhanced solubility and an increase in bioavailability and oral absorption 21, 45.

Other modifications can include metal cation-chelating agents (DOTA in dotatate and DTPA in pentetreotide), typically found in diagnostic agents or linkers (2-amino-4,6-dimethyl-3-oxo-3H-phenoxazine-1,9-dicarbonyl in dactinomycin). Finally, pyroglutamic acid (cyclic lactam of glutamic acid) is naturally found at the N-terminus of many neuronal peptides and hormones, but its function in living cells is still unclear [49]. In drug design, N-pyroglutamyl formation is a common modification used to cap the N-terminus to modulate peptide activity and increase resistance to degradation [50]. Indeed, in some cases, pyroglutamyl is essential to achieve full biological activity [51]. This modification is found in eight approved peptides, such as leuprolide and all GnRH agonists.

Notably, voxilaprevir was the first example of fluorinated peptide on the market, bearing four fluorine atoms: a difluoromethylene adjacent to a benzopyrazine modification involved in forming a macrocyclic structure, and a difluoromethyl group on an aliphatic non-natural AA: 1-amino-2-(difluoromethyl)cyclopropane-1-carboxylic acid. We predict that the exploitation of fluorine in peptides will follow what has already happened in small molecules, by becoming a key medicinal chemistry tool in which the judicious addition of a small and highly electron-withdrawing atom, such as fluorine, has been shown to have a key role in improving PK and physicochemical properties [52]. Hence, in the current peptide drug discovery pipeline, fluorination has been found to increase thermal stability and proteolytic stability, without affecting biological activity [53]. This has been applied to glucagon-like peptide-1 (GLP-1), in which the substitution with hexafluoroleucine in different positions has been shown to improve both binding affinity and protease (DPP IV) stability [54].

Further examination of the peptide structures reveals that there is, on average, a balance between polar and hydrophobic residues, when the polar contribution is derived by the summation of the polar, basic, and acidic constitutional members and the hydrophobic contribution is the summation of the aromatic and aliphatic constitutional members. Most approved peptides contain from 35% to 75% of polar residues, indicating that these molecules do not present a high excess of either hydrophilic or lipophilic components (Fig. 4a). This perhaps should not be a surprise, given that both polar and hydrophobic components are generally required for drugs with good PK profiles. A small number of outliers that comprise 100% polar or 100% hydrophobic building blocks are present. However, these exceptions are generally represented by a few building blocks (two, three, or five) and their small size makes these peptides more similar to small molecules. Indeed, the hydrophilic dipeptide spaglumic acid as well as the hydrophobic tripeptides ACE inhibitors (enalapril, perindopril, ramipril, quinapril, and trandolapril) respect the Lipinski’s Ro5 tailored for small molecules (computed by ChemAxon and Chemicalize [55]). Other outliers are represented by the growth hormone secretagogue receptor agonist macimorelin, comprising only three building blocks, and the hydrophobic antiviral peptides telaprevir, boceprevir, and ombitasvir, comprising four or five members. Finally, cyclosporine is the only case in which a marketed peptide comprising more than five members (11 in this case) completely lacks the polar component. This is reflected in its very long half-life (19 h) [56] and the fact that, after administration, 90% is found to be bound to serum proteins, mainly lipoproteins [57]. In this respect, the classification of natural and non-natural AAs is based solely on the nature of their side chain. Assuming that the impact of the backbone on the final properties is largely consistent with the size of the peptide, its polar contribution has not been considered. This is the clearest way to distinguish between hydrophilic and hydrophobic AAs, without overshadowing the contribution of the side chain with that of the backbone. To provide the reader with a visual overview of the relative physical property balance, the colour map in Fig. 4b shows the percentage composition of each approved peptide when the constitutional members are colour-coded according to their side chain properties: polar, acidic, basic, nonpolar aliphatic, and aromatic.

Another important aspect of structural composition is the proportion of peptide drugs and diagnostics that are either linear or contain a macrocycle. In nature, cyclic peptides of various sizes (from 8 to 50 AAs) occur in all kingdoms of life. Their enhanced stability and advantageous biopharmaceutical properties make their application in drug design common [58]. Indeed, 53% of the marketed peptides are linear, whereas 47% present one or more macrocycles in their structure (Fig. 3f). Among the approved cyclic peptides, 39% are of natural origin, 55% are analogues and only 6% are heterologous; perhaps nature has once again demonstrated how to develop stable, biocompatible peptides and, consequently, is a rich source of inspiration for candidates with optimal drug-like properties. Interestingly, our analysis highlighted that, overwhelmingly, peptide macrocycles comprise five-to-seven residues, with only a few exceptions (e.g., nesiritide and carperitide comprise a 17-membered ring) (Fig. 3g). Examples of peptides with five-to-seven-membered macrocycles include oxytocin, desmopressin, lanreotide, and eptifibatide. In general, smaller macrocycles tend to have greater conformational stability because of physical restraints and fewer rotatable bonds. As such, macrocyclisation is a common medicinal chemistry technique used to enhance peptide conformational stability and restrict the usual peptide chain flexibility 59, 60. This can stabilise the peptide conformation for optimal receptor complexation and confer a protein-like secondary and tertiary structure 60, 61, 62.

Depending on the desired site of cyclisation, there are various strategies to generate cyclic peptides [63]. These can involve the peptide head (peptide C-terminal moiety), the peptide tail (peptide N-terminal moiety), or AA side chains. According to our findings (Fig. 3h), the most common technique is side chain-to-side chain cyclisation (58% of all the marketed cyclic peptides), with 25 out of 26 side chain-to-side chain macrocycles formed by a disulfide bond between cysteine thiol pairs. The exception is bremelanotide, in which a lactam is formed between the amine side chain of a lysine residue and the carboxylic acid side chain of an aspartic acid residue. However, disulfide bridges are not always metabolically stable in vivo, limiting their application [64]. Macrocycles within a peptide can be formed also by head-to-side chain cyclisation (24%), mainly via lactamisation between the C-terminal carboxylic acid and a side chain amine (e.g., lysine). The head-to-tail cyclisation between the N-and C-termini (7%) generates an all-amide end-to-end cyclic lactam, thus abrogating exopeptidase hydrolysis. Finally, another cyclisation strategy encountered in two out of 47 cyclic peptides (i.e., grazoprevir and elcatonin) is side chain-to-tail macrocyclisation. When the head and/or the tail of the peptide is involved in the macrocycle, protease access to the backbone is reduced because cyclisation removes the free N-and C-termini that are targeted by amino- and carboxy-peptidases, respectively [65]. This explanation bears out in the approved peptides, with the mean experimental terminal half-life of cyclic peptides compared with linear peptides being 27 h and 12 h, respectively.

Among marketed peptide therapeutics and diagnostics, 38% present an amidated C-terminus, whereas 10% present an acetyl group at the N-terminus (Fig. 3i, j). Modifications of the peptide N-terminus also include the addition of pyroglutamic acid (7%) or deamination of the last AA (4%). Similar to head-to-tail cyclisation, this abrogates exopeptidase hydrolysis by masking N-and C-termini [66]. Moreover, N-terminal acetylation or C-amidation precludes ionisation and hydrogen bonding of NH2 and COOH groups, respectively 67, 68, thus better mimicking natural proteins.

Section snippets

Concluding remarks and prospects

Together with the database PepTherDia, this review offers the possibility of exploring common trends in approved peptide drugs and diagnostics. We have highlighted strategies most commonly used in peptide drug design, which have successfully brought these peptides to the market. The trends underlined cannot be ascribed to luck or coincidence. Most approved peptides (84%) follow the rules ‘established’ by nature over several millennia of evolution because they are naturally derived or analogues

Conflict of interest

C.R.C is a director of Pepmotec Ltd, a peptide synthesis spin-out company from Durham University, UK.

*The PepTherDia website contains the full list and classification of non-natural AAs and modifications together with the detailed methodologies used for data collection, computational analysis, and structural analysis.

Acknowledgements

The authors gratefully acknowledge Daniele Tomasi for the development of the website PepTherDia. This work was supported by a Liverpool John Moores University Vice Chancellor’s PhD Scholarship (V.D’A.) and a University Alliance Doctoral Training Alliance COFUND/Marie Skłodowska-Curie (European Union’s Horizon 2020 research and innovation programme, grant agreement No 801604) PhD Fellowship Programme in Applied Biosciences for Health (P.D.). The funding sources had no involvement in the study

Vera D’Aloisio received her Master’s degree in Pharmacy from the University of Camerino (Italy) in 2017, with a thesis on the nanoformulation of small peptides, carried out at Liverpool John Moores University, under the supervision of Gillian A. Hutcheon. She is currently pursuing a PhD in peptide drug discovery and formulation at Liverpool John Moores University, mentored by Gillian A. Hutcheon and Christopher R. Coxon. Her PhD research is focused on the development and the formulation of

References (68)

  • M. Rodriguez

    Synthesis and biological activities of cholecystokinin analogues substituted in position 30 by 3-(1-naphthyl)-l-alanine [Nal(1)] or 3-(2-naphthyl)-l-alanine [Nal(2)]

    Eur. J. Med. Chem.

    (1991)
  • S.V. Moradi

    Glycosylation, an effective synthetic strategy to improve the bioavailability of therapeutic peptides

    Chem. Sci.

    (2016)
  • B.P. Ward

    Peptide lipidation stabilizes structure to enhance biological function

    Mol. Metabol.

    (2013)
  • L. Thorstholm et al.

    Discovery and applications of naturally occurring cyclic peptides

    Drug Discov. Today Technol.

    (2012)
  • N. Qvit

    Peptidomimetic therapeutics: scientific approaches and opportunities

    Drug Discov. Today

    (2017)
  • A. Henninot

    The current state of peptide drug discovery: back to the future?

    J. Med. Chem.

    (2018)
  • J. Ferlay

    Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012

    Int. J. Cancer

    (2015)
  • M.L. Daviglus

    Preventing cardiovascular disease in the 21st century: therapeutic and preventive implications of current evidence

    Am. J. Cardiovasc. Drugs

    (2006)
  • Transparency Market Research Peptide

    Therapeutics Market to Expand at a CAGR of 7.9% from 2019 to 2027

    (2019)
  • ClinicalTrials.Gov https://www.clinicaltrials.gov/ [Accessed February 22,...
  • D.J. Craik

    The future of peptide-based drugs

    Chem. Biol. Drug Des.

    (2013)
  • D. Mathur

    PEPlife: A Repository of the Half-life of Peptides

    Sci Rep

    (2016)
  • S.S. Usmani

    THPdb: database of FDA-approved peptide and protein therapeutics

    PLoS One

    (2017)
  • A.P. Davenport

    Advances in therapeutic peptides targeting g protein-coupled receptors

    Nat. Rev. Drug Discov.

    (2020)
  • M. Tyagi

    Drug syntheses beyond the Rule of 5

    Chemistry

    (2020)
  • T.T. Wager

    Central nervous system multiparameter optimization desirability: application in drug discovery

    ACS Chem. Neurosci.

    (2016)
  • A.D. McNaught et al.

    IUPAC. Compendium of Chemical Terminology

    (1997)
  • IUPAC-IUB Joint Commission on Biochemical Nomenclature

    Nomenclature and symbolism for amino acids and peptides: recommendations 1983

    Eur. J. Biochem.

    (1984)
  • FDA

    Impact Story: Developing the Tools to Evaluate Complex Drug Products: Peptides

    (2019)
  • L.C. Wu

    Regulatory considerations for peptide drug products

  • Go.DrugBank. https://go.drugbank.com/ [Accessed February 22,...
  • European Medicines Agency (EMA) www.ema.europa.eu/en [Accessed February 22,...
  • Pharmaceuticals and Medical Devices Agency (PMDA) www.pmda.go.jp/english/index.html [Accessed February 22,...
  • Drug Central. http://drugcentral.org [Accessed February 22,...
  • Cited by (59)

    • Acetone-precipitated zein protein hydrolysates from blue-maize selectively target hepatocellular carcinoma and fibroblasts in a dose-dependent manner

      2023, Food Hydrocolloids for Health
      Citation Excerpt :

      Bioactive peptides are a popular cancer therapeutic molecule among research projects worldwide (Trinidad-Calderón et al., 2021a). These amino acid polymers consist of 2 to 50 amino acids that are inactive within the protein they contain and are typically produced from macronutrients (D'Aloisio, Dognini, Hutcheon & Coxon, 2021; Daliri, Oh & Lee, 2017). Reports of their bioactive effects and their obtention sources are varied (Felício, Silva, Gonçalves, Santos & Franco, 2017).

    • Effect of dilution solvent and injection volume on the analysis of basic hydrophilic therapeutic polypeptide salts with pressurized carbon dioxide mobile phases

      2022, Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences
      Citation Excerpt :

      Lau and Dunn pointed the evolution of chain length in therapeutic peptides observed between the 1980s’ and 2018 [6]. Short-chain peptides containing less than 10 amino acids (molecular weight below 1000 Da) were still major in clinical development and standing for 75% of commercialized peptides in northern countries until 2021 [7]. However, following the developments in peptide synthesis and manufacturing improvement to achieve molecular weights up to 5000 Da [8,9], larger peptides appeared and are now more commonly observed in newly developed drugs.

    View all citing articles on Scopus

    Vera D’Aloisio received her Master’s degree in Pharmacy from the University of Camerino (Italy) in 2017, with a thesis on the nanoformulation of small peptides, carried out at Liverpool John Moores University, under the supervision of Gillian A. Hutcheon. She is currently pursuing a PhD in peptide drug discovery and formulation at Liverpool John Moores University, mentored by Gillian A. Hutcheon and Christopher R. Coxon. Her PhD research is focused on the development and the formulation of novel small peptide antagonists of the calcitonin gene-related peptide receptor (CGRP-R), as a novel treatment for migraine.

    Paolo Dognini graduated cum laude in medicinal chemistry and pharmaceutical technologies from the University of Pavia in 2018. In 2019, he entered the DTA3/MSCA COFUND PhD Fellowship Programme in Applied Biosciences for Health and is currently undertaking doctoral studies in medicinal chemistry at Liverpool John Moores University, mentored by Francesca Giuntini and Christopher R. Coxon. His research focusses on the synthesis and development of new peptide conjugates with applications as multifunctional drugs, diagnostic agents, and materials. A particular focus is on nucleophilic aromatic substitution to couple peptide cysteine residues with perfluoroaromatic compounds.

    Christopher R. Coxon is an associate professor in synthetic chemistry and a Bicentennial Research Leader in the School of Engineering and Physical Sciences at Heriot-Watt University, UK. His research programme focusses on peptide drug discovery and using peptides as tools for biology. In 2010 he completed his PhD in the development of novel inhibitors of protein kinases at The Northern Institute for Cancer Research, Newcastle University, before postdoctoral appointments in chemical biology and peptide chemistry at Durham University from 2010 to 2013. He previously held academic positions at Durham University and Liverpool John Moores University (between 2013 and 2019) and is now a director and cofounder of Pepmotec Ltd.

    View full text