Small-molecule drugs (<500 Da) have formed the backbone of modern pharmacotherapy for over a century; however, their inherent physicochemical and pharmacological attributes also set limitations that are increasingly difficult to breach in drug discovery today. The active-site paradigm restricts discovery in enzymes/receptors without defined binding pockets which makes 75–80 % of human proteins including most transcription factors and regulatory hubs undruggable under traditional occupancy-driven pharmacological methods. Second, features that make small molecules synthetically tractable, and orally bioavailable (small size, moderate lipophilicity, rigid scaffolds) also limit their capacity for interactions with a macromolecular target. As a result, potency is often bought at the cost of promiscuity: one small molecule may interact with dozens of off-target proteins, ion channels or transporters causing off-target toxicities that may only be detected in late stage development or post-marketing surveillance. Third, small molecules are often optimised for target affinity with limited or no co-optimization of their ADME properties. This myopic optimization approach results in candidate drugs with excellent in vitro IC50 values, but poor translational potential; about 40 % of attrition in Phase II/III is now ascribed to sub-optimal pharmacokinetics rather than lack of efficacy.
It has been demonstrated that on average small molecule drugs bind to at least 6-11 targets (not counting their intended pharmacological target). These targets are generally known as off-targets and interactions with these additional targets can be potentially adverse. Off-target interactions are generally lower affinity than those with the intended pharmacological target, but can be significant if the off-target has higher cellular expression or if there is high systemic exposure (for example in preclinical toxicity screens, where much higher doses are interrogated to determine the toxicological profile of the compound, clinical mis-dosings, accidental or intentional overdose, drug-drug interactions (may cause higher systemic exposures), or any other unexpected individual differences (also may cause higher systemic exposures). The number of off-targets for a small molecule is probably always significantly greater than what is being reported at the time of a marketing application, because pharmaceutical companies profile compounds using a predefined (restricted) non-overlapping target panel consisting of a limited set of targets (typically<100). Current in silico off-target prediction platforms mitigate but do not eliminate these risks, because training sets based on historical patent data systematically under-represent novel chemical scaffolds leading to false negatives as well as false positives. As a result "clean" selectivity profiles typically can only be achieved through iterative cycles of medicinal chemistry and extensive safety pharmacology testing, increasing time and cost.
Small-molecule candidates face severe difficulties due to inconsistent and poor oral bioavailability. Certain physicochemical characteristics determine a drug's effectiveness towards its target (i.e. high lipophilicity, number of aromatic rings, and conformational rigidity) often restrict aqueous solubility to values below those required for dissolution in the gastrointestinal tract. Compounds with an intrinsic solubility < 10 µg/mL often have unpredictable absorption and food effects, leading to imprecise dose prediction. Permeability can also be a limiting factor: molecules with >5 hydrogen-bond donors or a polar surface area > 140 Å2 are poorly absorbed across the intestinal epithelium, requiring non-oral dosing and decreasing patient compliance. Once absorbed, small molecules are subjected to extensive first-pass metabolism by cytochrome P450 enzymes in the liver and intestine. Chemical "soft spots" such as unhindered aliphatic hydroxylation sites, electron-rich aromatic rings, or labile amide bonds will often yield rapid clearance, short half-lives, and potentially toxic or reactive metabolites. Attempts to increase metabolic stability (e.g., deuteration, fluorination, or macrocyclization) can inadvertently increase lipophilicity, which in turn decreases solubility and increases plasma protein binding to produce lower exposures of free drug at the site of action. Efflux transporters such as P-glycoprotein will also lower intracellular concentrations, especially at the blood–brain barrier and tumor tissues, leading to sub-therapeutic exposure. Balancing these interdependent parameters demands complex multi-parameter optimization (MPO) workflows, in which ADME assays are integrated as early as possible in lead optimization, but the trade-offs are often finely poised and late-stage failures due to pharmacokinetic liabilities remain one of the dominant attrition modalities in drug discovery and development.
Fig. 1 Key factors influencing small-molecule drug bioavailability.1,2
Cyclic peptides are polypeptide chains that consist of canonical and non-canonical amino acids, and the N- and C-termini are linked together by covalent bonds. They adopt macrocyclic structures and a large group of cyclic peptides of various shapes, sizes and chemical compositions were isolated from natural sources which reflects the broad structural diversity of cyclic peptides as a class of molecules. A well-known example of this is the family of cyclic peptides called microcystins that are generated by cyanobacterial blooms in freshwater lakes. These cyclic peptides are animal toxins. By contrast, the pentacyclic peptide nisin is a common food additive, used for example in processed cheeses, meat products, canned foods and drinks, because it is bacteriostatic (growth-inhibiting) against a wide range of gram-positive bacteria and so prolongs shelf-life. This natural molecule is produced by some strains of the bacterium Lactococcus lactis, so many of us eat it without knowing we are ingesting a cyclic peptide.
Peptide cyclization yields a macrocyclic molecule. The conformational rigidity and the absence of termini of macrocyclic peptides make them resistant to enzymatic degradation. Linear peptides are targets for proteolytic cleavage by exopeptidases that recognize free terminal amino- or carboxyl-groups and by endopeptidases that cleave the amide bonds along the peptide backbone. The half-life of peptides in plasma is typically < 5 min. The closed backbone of cyclic peptides safeguards termini while their constrained conformation organizes them into compact structures. The concomitant reduction in conformational entropy of the peptide reduces the likelihood of a protease adopting the binding geometry required for hydrolysis, with kinetic studies on trypsin, chymotrypsin and pronase showing up to 100-fold slower cleavage rates for monocyclic analogues and >1000-fold slower rates for bicyclic variants. The backbone's stiffness promotes secondary formations including β-turns and β-hairpins which protect amide bonds from solvent exposure and active sites of proteases. The gains in stability are not an in vitro artifact: Several FDA-approved cyclic drugs including cyclosporine A, pasireotide and rezafungin demonstrate oral bioavailability and plasma half-lives suitable for once- or twice-daily administration despite their peptide structure. The incorporation of various chemistries such as lactam bridges and hydrocarbon staples allows scientists to regulate the structural properties of molecules so that they sustain biological activity after a full day in serum while their linear counterparts disappear quickly. In this manner, cyclization can overcome the Achilles' heel of peptide therapeutics (proteolytic lability) without sacrificing biological potency.
The large flat surfaces involved in protein–protein interactions as well as transcription-factor surfaces and disordered protein regions remain undruggable because traditional small molecules lack the ability to engage such targets. Cyclic peptides represent a privileged molecular class: their size (MW of 0.5–2 kDa and 10–20 Å in diameter) affords an extended recognition surface that can form multiple hydrogen bonds, salt bridges, and hydrophobic contacts over more than 500 Å2 of surface area, approaching the contact footprint of antibody complementarity-determining regions. Macrocyclization pre-organizes the contact residues into a low-entropy conformation so that the entropic penalty upon binding is minimized and the enthalpic gains are maximized, which results in the low- to sub-nanomolar affinity range of best-in-class cyclic peptides. Recent examples span the full range of these challenging targets. The stapled α-helical peptide ALRN-6924 reactivates p53 signaling by binding with 0.7 nM affinity to the previously "undruggable" PPI interface of HDM2/X. Highly selective synthetically evolved bicyclic peptides (bicyclic peptide phage display) targeting coagulation factor XIIa or complement C5 reach picomolar affinities against challenging homologous proteases. Even amyloidogenic interfaces that lack defined pockets have been successfully targeted: a rationally designed cyclic β-hairpin binds the Aβ42 oligomer interface with Kd = 3 nM and prevents fibril elongation in vitro and in mouse models of Alzheimer's disease. The modularity of cyclic peptide chemistry also allows rapid diversification (D-amino acids, N-methylation, or non-canonical side chains) to further optimize both affinity and physicochemical properties, broadening chemical space beyond rule-of-five limitations for small molecules. As a result, cyclic peptides are rapidly emerging as precision tools to drug the "undruggable" proteome.
Cyclization is not a 'cookie-cutter' approach, but rather a strategy for creating a diverse array of topologies to match the size and shape requirements of a given target class. Head-to-tail amide macrocyclization produces the familiar 'belt' of macrocycles which adopt compact β-turn or β-hairpin shapes for active-site blockade or reverse-turn epitope mimicry. Side-chain-to-side-chain lactam or disulfide crosslinks graft on α-helical elements that can invade coiled-coil interfaces or inhibit viral fusion peptides. Hydrocarbon staples enforce α-helicity and concomitantly enhance cell permeability to reach intracellular targets like BCL-2 family proteins. Longer recognition faces are spanned by bicyclic or tricyclic scaffolds produced by click chemistry or ring-closing metathesis. These nanoscale scaffolds with well-defined cavities and several recognition 'loops' are the antibody paratope analogs synthesized in one step. The synthetic modularity is further extended by the ease of introducing non-proteinogenic building blocks: β-amino acids, γ-turn mimetics, PEGylated staples, or photoreactive diazirines can be incorporated at user-defined locations to adjust half-life, tissue distribution, or introduce covalent photo-crosslinking to validate targets. Phage, mRNA or yeast display libraries of up to 1012 unique cyclic sequences provide deep sampling even for targets with novel topologies to rapidly identify binders. The resulting library of chemically and topologically diverse scaffolds can be matched to bind to extracellular receptors, intracellular PPIs, membrane channels, nucleic-acid structures, or pathogen surfaces, making cyclic peptides one of the most adaptable modalities in the drug-discovery arsenal.
Clinically utilized cyclic peptides originate primarily from naturally occurring cyclic peptides. As cyclic peptides have several features that make them attractive lead compounds for drug development as well as nice tools for biochemical research, scientists have made diverse efforts to develop biologically active cyclic peptide compounds. Peptides can be prepared by either genetic or synthetic method. The genetic method, as will be described below, is usually limited to ribosomal 20 amino acids, whereas the sequence determination of hit compounds is straightforward. On the other hand, the synthetic method can provide more versatile cyclic peptide compounds as the repertoire of amino acids and the way of forming cyclic peptides are diverse. Solid-phase peptide synthesis combined with split-and-pool synthesis can prepare fairly large libraries. However, sequence determination is challenging after screening of these libraries. Conventional Edman degradation cannot be used for cyclic peptides once the free N-terminus disappears after cyclic peptide formation by N-to-C cyclization.
Cyclization remains the primary obstacle during cyclic peptide synthesis. The four types of cyclization are side chain-to-side chain, head-to-tail, head-to-side chain and side chain-to-tail. The natural cyclic peptides are in majority head-to-tail cyclic peptide probably because the lack of amino or carboxyl end that imparts metabolic stability. Cyclization of peptides have been obtained in a variety of ways including: amide, disulfide, thioacetal, thioether, ether, C-C, C=C (i.e. alkene metathesis), C≡C triple bond (alkyne metathesis), triazole formation and multicomponent reactions (e.g. Ugi reaction). In the case of unprotected peptides, the classical amide bond formation between amino and carboxyl groups is chemoselectivity-lacking and could lead to C-terminal amino acid epimerization thus unsuitable for cyclization in solution. Therefore, aminolysis-based reactions can only be used to get cyclic peptides in solution using a partially protected linear peptide and mild activation C-terminus. Epimerization of the C-terminus could still take place. Cyclization by aminolysis can also be done on resin, e.g. using a diaminobenzyoyl linker to link the peptide to the solid support. The drawbacks of the cyclization through amide bond formation can be overcome by using chemoselective ligations. The main benefits of chemoselective ligations are the fact that unprotected peptides can be used, which eliminates the solubility issues seen with partially protected sequences and minimal or no activation of C-terminus.
Incorporation of non-canonical amino acids (ncAAs) into cyclic peptides provides chemical diversification as well as a possible introduction of proteolytic resistance, membrane permeability, or alternative cross-linking handles; however, their incorporation increases synthetic complexity multiplicatively. During solid-phase peptide synthesis (SPPS), each ncAA must be chemically pre-synthesized (or purchased) with orthogonal protection groups, and is then coupled to the growing peptide chain using conditions orthogonal to Fmoc/t-Boc chemistry. As many ncAAs are expensive (>$500/g) and only commercially available in milligram quantities, the overall material costs for library production can increase dramatically. Coupling efficiency is also typically lower than for native residues (especially when steric bulk in the side chain causes hindrance), which increases the formation of deletion sequences that are difficult to purify. Modification types such as N-methylation, sulfation, or glycosylation cannot be recapitulated in SPPS, which is driving use of cell-free or in vivo incorporation methods. Genetic code expansion using orthogonal aaRS–tRNA pairs allows for site-specific insertion of ncAAs during ribosomal synthesis; however, suppression efficiency is rarely >30–50 %, and release factor 1 (RF1) competes with charged tRNA, truncating full-length product. Methods that use quadruplet codon or sense-codon reassignment can encode multiple unique ncAAs, but require extensive host-strain engineering and may tax the metabolism of the producer strain.
Successful cyclization produces a product peptide that remains a crude blend of linear precursors and other impurities which requires purification to improve its starting purity below 5 %. Drug-grade purities (≥95 %) are therefore only reached by an effective purification workflow. Although reversed-phase HPLC is frequently used for standard purification, one ncAA's retention-time shifts or altered ring topology require re-optimized gradients for each sequence which leads to higher solvent use and longer analytical time. Diastereomeric impurities, resulting for example from epimerization events or racemization during cyclization are particularly problematic, because their physicochemical properties are so similar to the product that their peaks overlap, resulting in a need for ultra-high-resolution columns (C18, C8, or phenyl-hexyl) and high temperatures. Mass-directed purification will isolate the correct mass, but isotope patterns of halogenated ncAAs or for heavily N-methylated species may hide minor impurities. During SPPS batch-to-batch variability emerges from changing resin loadings which affects coupling efficiencies and cyclization yields because of minor variations in solvent water content or base strength. Oxidation of methionine or cysteine residues during lyophilization (if not completely removed) by TFA carried over from the HPLC purification can also lead to inactivation by sulfoxide formation or disulfide scrambling. As a consequence, programs are increasingly switching to more integrated quality-by-design (QbD) workflows with in-line analytics (UPLC-MS) in real time, DoE for cyclization conditions, and defined raw-material specifications for resins and ncAAs.
The application of these synthetic strategies is driven by our 10+ years experience in implementing even the most challenging ring-closure strategies. Whether it's classical head-to-tail lactamizations or the latest in programmable multicomponent macrocyclizations, we always start with a retrosynthetic risk assessment of sequence length, steric bulk, β-turn propensity, and inclusion of ncAAs to inform the best cyclization modality (on-resin vs. off-resin, chemoselective vs. enzymatic, or traceless ligation) before the first amino acid is coupled. For sequences that oligomerize in high concentration, we use high-dilution flow reactors to generate pseudo-dilution conditions while still processing at gram-scale and we are accustomed to routinely achieving cyclization yields >65 %, even for 14- to 18-mer scaffolds. To prevent epimerization at the C-terminus, we screen activating reagents (COMU, PyBOP, HATU) with non-nucleophilic bases (2,4,6-collidine) and use in-line chiral HPLC to monitor the stereochemical integrity. If ring strain is expected to be an issue with small or bulky sequences, we incorporate turn-inducing D-Pro, N-methylated residues, or aziridine-containing dipeptide mimetics to pre-organize the linear precursor into a cyclization-competent conformation. For targets that require non-peptidic linkages, we have chemists that are experts in copper-free click cycloadditions, thiol–ene stapling, and Staudinger ligations all performed under strictly anhydrous conditions to preclude oxidation artifacts. Finally, our in-house enzymology unit is also equipped to engineer sortase- and butelase-mediated cyclizations for aqueous-phase macrocyclization of sensitive sequences. This toolbox means that every cyclic peptide – from 6-mer disulfide-constrained epitopes to 30-mer bicyclic degraders – is manufactured with the highest yield and lowest side-product profile.
The cornerstone of our supply philosophy is the guarantee of drug-grade purity, starting with an analytical cascade embedded into every synthetic step rather than tacked on at the end. UPLC-MS systems operate in both low- and high-resolution modes, enabling the detection and quantification of linear deletion sequences, epimeric adducts, and oxidation products at 0.1 % loadings. Crude cyclization mixtures are screened by ion-mobility mass spectrometry (IMS-MS) to separate isobaric conformers that co-elute on typical C18 columns, so that only the correctly folded macrocycle is advanced to purification. Peptides containing ncAAs receive additional verification by 2D 1H-13C HSQC NMR, which can confirm site-specific incorporation while also ruling out racemization or scrambling. Finally, each batch is subjected to endotoxin testing (LAL assay), residual solvent analysis (GC-MS), and bioburden counting in our GMP-compliant QC suite, with release of certificates of analysis (CoAs) in compliance with ICH Q7. Clients therefore receive not only milligram-to-kilogram quantities of material but also a comprehensive analytical dossier to fast-track regulatory filing timelines.
Our process architecture includes speed and scalability as foundational design elements. Automated peptide synthesizers operate continuously, 24/7, in parallel with microwave-assisted modules to dramatically shorten assembly times (typical Fmoc-SPPS cycle time for a microwave-assisted module is 4 min, reducing overall assembly time of a 20-mer cyclic precursor to < 8 h). If recalcitrant building blocks are necessary for non-canonical or heavily modified sequences, a switch to semi-automated, single-vessel synthesis is available with on-the-fly intervention by our in-house chemists to alter coupling stoichiometries or solvent polarity and guarantee delivery of even the most challenging peptides in 5–7 business days. Purification following cyclization reactions is highly streamlined by a direct-connect RP-HPLC skid, which automatically loads crude reaction mixtures, elutes under optimized gradients, and lyophilizes the product, all without manual intervention, thereby shortening downstream processing by 30 %. Scale-up is enabled by a two-tier platform: Tier 1 uses 200 L solid-phase reactors to manufacture hundreds of grams of lead candidate material. Tier 2 features solution-phase fragment condensation and enzymatic cyclization, which can be scaled to multi-kilogram quantities under cGMP conditions. One recent program served to highlight this capacity: we delivered >200 unique cyclic peptides (avg. 14 residues, 2–3 ncAAs each) at ≥95 % purity within eight months to enable a client to advance from hit triage through in vivo efficacy studies without delays for resynthesis. Whether your target projects 5 mg for an initial SAR panel or 5 kg for IND-enabling tox packages, our vertically integrated workflow means you can have your quality, speed, and scalability too.
Peptide Synthesis Services at Creative Peptides
References
