Combinatorial chemistry has proven to be an indispensable tool for the discovery of cyclic peptides. This approach systematically constructs the diversity of potential binders, while imposing a required structure that is necessary for function. By coupling multiple orthogonal reactions together to create building blocks and assembling these into libraries in a one pot-no touch (OPTNT) fashion, combinatorial chemistry is distinct from one-at-a-time synthesis by providing access to thousands to millions of peptide variants from a single campaign. Consequently, rare high affinity binders are enriched for during discovery, and vast areas of sequence space are explored in a fraction of the time. The ease with which such libraries can be prepared, screened, and decoded is further expanded when integrated with parallel synthetic approaches such as solid-phase beads, solution, or DNA-encoded synthesis. For cyclic peptides, where the control of topology is important to stability and target engagement, the combinatorial approach enables multiple cyclisation chemistries (disulfide, lactam, thioether) to be tested in parallel with sequence variation to define structure–property relationships that can be used to direct lead optimization and is a valuable source of drug candidates, research tools and molecular probes.
One example of such combinatorial chemistry in peptide discovery is where large libraries of compounds are synthesized, where each member of the library is a different mixture of amino acids. Combinatorial chemistry allows very large peptide libraries to be generated in a short space of time. Parallel synthesis is important to the success of combinatorial chemistry because, as a single member of a peptide library is synthesized, many other variants can be synthesized in parallel, greatly increasing the diversity of the library.
OBOC libraries are a very potent solid-phase combinatorial technique, where each microbead is functionalized with a single unique peptide sequence, allowing the simultaneous synthesis and physical separation of millions of compounds. This strategy is ideally suited to the exploration of cyclic peptides, since the bead serves both as a solid support for stepwise assembly, and as a convenient, physical carrier for individual assay of each compound, directly linking binding phenotype with chemical identity. The OBOC strategy does not require pre-encoded DNA tags, but rather takes advantage of the direct chemical synthesis of the peptide on the bead surface, followed by on-bead screening against fluorescently labelled or enzyme-tagged targets. Positive beads can then be physically isolated under a microscope, with the attached peptide sequence identified by Edman degradation or mass-spectrometric decoding. The advantages of this approach are immediate access to hit structures, as well as the ability to generate true chemical diversity (non-canonical amino acids, D-residues, multiple cyclisation motifs etc) within a simple bead-based format, including the potential for high-throughput visual screening.
Fig. 1 Illustration of the on-chip screening for peptide-based protein binders.1,5
The split-and-mix approach is the underlying mechanism that enables OBOC diversity, allowing a straightforward linear synthesis to generate an exponential number of peptide sequences. A uniform population of polymeric microbeads each bearing a common linker and starting amino acid residue is used as the reaction support. Each reagent containing a diversity residue is added by first splitting the bead pool into separate reaction vessels, each containing a different protected amino acid. After coupling, the beads are vigorously mixed in order to randomize their spatial distribution, then washed and deprotected to unmask the free amine for the next cycle. The process of splitting the bead pool → coupling a diversity reagent → mixing to redistribute the beads is repeated for every residue in the sequence such that the path each bead takes through the array of building blocks is strictly unique. After synthesis, the library will contain one bead for every possible sequence in the diversity set. The entire library is a single, homogeneous population of beads, each bearing a unique peptide sequence, and can be handled as a bulk solid sample. For cyclic peptides, the final split, couple, and wash step is replaced with a cyclisation chemistry, which is often an intramolecular lactam or disulfide bridge formed by selectively deprotecting the appropriate side-chain functionalities, followed by activation of the C-terminus. Each bead in the library undergoes ring closure while remaining tethered to the solid support. The workflow from coupling to cyclisation requires high efficiency and orthogonal chemistry. Each coupling must be driven to high conversion on every bead in the library, as incomplete sequences will generate truncated side-products that muddy screening. Automated dispensers are used to ensure that every split fraction is exposed to the correct reagent, and thorough mixing after each coupling is necessary to ensure beads are completely and randomly redistributed, avoiding positional bias. After coupling the final residue, the bead-bound linear peptide is ready for cyclisation. For disulfide-based libraries, beads are moved to an oxidizing buffer (typically DMSO/air or dilute H2O2) to generate intramolecular disulfides between genetically encoded cysteines, while for lactam bridges, selective deprotection of Lys and Asp/Glu side chains followed by HATU-mediated activation forms the side-chain crosslink.
Screening OBOC libraries is based on on-bead binding assays that report target engagement without the need to cleave the peptide from the solid support. In its simplest form the bead library is incubated with a fluorescently labelled target protein in a buffer designed to minimize non-specific binding, then washed to remove unbound material. Beads that exhibit fluorescence when viewed under a fluorescence microscope are scored as positive and their position is recorded for physical isolation. Assay readout can be done by manual or automated bead picking. The former relies on visual inspection, while the latter is based on bead-sorting flow cytometry and interrogates up to millions of beads per hour. The fluorescent label used also has an effect on assay sensitivity and artefacts: small organic dyes such as fluorescein or rhodamine provide robust signals but can have hydrophobic artefacts, while protein fusions such as GFP or R-phycoerythrin allow the target to remain in its native conformation during the binding event. Bead throughput can be improved by arraying the beads into microwell plates, which allows screening against multiple targets or concentration series in parallel in a single experiment. Having identified positive beads, the peptide sequence must now be determined. Direct Edman degradation of the bead-bound peptide provides reliable N-terminal sequencing for shorter peptides (≤15 residues), but is not possible if the N-terminus is blocked or if the peptide contains non-standard residues. This limitation can be overcome by using an encoding strategy, in which a unique chemical or optical barcode is co-synthesized on each bead during split-and-mix synthesis, either as a PNA tag or a series of fluorophore-doped silica nanoparticles. The barcode is decoded by hybridization or fluorescence microscopy after screening, revealing the synthesis history of each bead and thus the peptide sequence. Mass spectrometric sequencing is becoming more prevalent: the peptide is cleaved from a single bead using photocleavable linkers, spotted onto a MALDI target, and then analyzed by tandem MS/MS, with de novo sequencing algorithms reconstructing the sequence from fragment ion patterns. This approach is fast, sensitive, and tolerant of non-canonical residues, and is well suited to cyclic peptide libraries.
DNA-encoded libraries of cyclic peptides (DELs) are synthetic collections of macrocyclic peptides prepared in a single reaction vessel and then used directly for screening in affinity selection or high throughput screening (HTS). The first step in the preparation of DELs is the encoding of every peptide with a unique DNA oligonucleotide barcode that permanently records the synthetic history of that peptide, directly linking genotype and phenotype. This DNA encoding is a major departure from existing platforms and overcomes the scalability problem of current one bead one compound combinatorial libraries, whose size is inherently limited by the physical separation of individual members. DNA encoding also means that each member of the library can be directly identified by massively parallel DNA sequencing following affinity selection. The synthesis of DELs involves cycles of chemical coupling and enzymatic DNA ligation: at each cycle the growing peptide–DNA conjugate is split into a number of reaction vessels, a building block is added to the peptide and a DNA tag to the encoding strand. The pools are then recombined and the cycle is repeated for three to four steps, until libraries containing billions of individual cyclic peptides can be created. This split-and-pool strategy is based on solid-phase synthesis and exceeds the number of compounds accessible to high-throughput screening. The resulting mixture is then used in affinity selection against an immobilised target, non-binding members are washed away and bound peptides recovered and their DNA barcodes amplified and identified by next-generation sequencing. This approach allows rapid hit identification, on the order of days rather than months, and the ability to explore wider chemical space, including non-standard amino acids, non-standard ring sizes, and non-standard cyclisation chemistries, that may not be available using purely biological display platforms. DELs have had a significant impact on the discovery of cyclic peptides, especially as leads against previously difficult targets such as protein–protein interfaces and undruggable enzymes. The platform continues to evolve with improvements to DNA-compatible chemistries and bioinformatic analysis.
Fig. 2 DP-DEL library architecture and synthesis.2,5
DNA tagging is based on a one-to-one correspondence between the chemistry of each library member and a nucleotide sequence used as a molecular barcode. In a typical DNA-recorded synthesis, each combinatorial cycle is recorded by ligation of a short, predefined oligonucleotide that encodes the identity of the building block added during that cycle. A cyclic peptide library prepared by four cycles of amidation, for example, would have a concatenated DNA tag made up of four segments, each encoding an amino acid incorporated at that position. The combinatorial split-and-pool logic ensures that each peptide–DNA conjugate is decorated with a unique combination of these tags, leading to a DNA-encoded library in which the chemical diversity of the peptide moiety is directly reflected by the sequence diversity of the attached DNA. This encoding allows all library members to be pooled into a single reaction vial for affinity selection, conserving reagents and screening time relative to spatially arrayed formats. The DNA tag can then be amplified by PCR, allowing exponential enrichment of bound peptides and generating enough material for high-throughput sequencing. A critical requirement is that the DNA barcode must be chemically inert under conditions used for peptide synthesis and cyclisation; strong acids, oxidizing agents, or heavy-metal catalysts that would degrade the nucleic acid must be avoided or applied under strictly controlled, mild conditions. The tag length is generally optimized to minimize any potential interference with peptide binding while still allowing robust amplification, with most libraries using 10–20 base-pair segments that are concatenated to yield a 60–80 base-pair coding strand. After selection, the amplified DNA is sequenced and bioinformatic pipelines translate each read into a corresponding peptide structure, filter out low-quality reads, and determine enrichment scores by comparing the frequency of each sequence in the selected pool to its abundance in the naive, unselected library. This quantitative enrichment analysis distinguishes true binders from stochastic noise, and structure–activity relationships can be identified and used to guide chemical optimization. The principle of DNA tagging thus fuses the synthetic tractability of combinatorial chemistry with the analytical power of sequencing, creating a self-contained discovery ecosystem in which each molecule carries its own instruction manual.
Cyclisation of DELs has been shown to be required to satisfy two conflicting criteria: the ring-closing chemistry must be efficient on the peptide moiety, but at the same time the DNA barcode must remain intact and unmodified as it is susceptible to reagents and extreme pH. As a result, mild, aqueous-compatible reactions are the mainstay of DEL cyclisation. Amide-based macrocyclization can be achieved in water using water-soluble carbodiimides such as EDC in the presence of N-hydroxy-succinimide. This reagent activates the C-terminal carboxylate under neutral conditions, and allows intramolecular capture by the N-terminal amine with no detectable DNA degradation. However, the success of this approach depends on the physical ability of the peptide sequence to sample a cyclic conformation under aqueous buffer conditions, and aggregation-prone sequences may resist cyclisation, leading to partial reactions and loss of diversity. Click chemistry, and particularly copper-catalyzed azide–alkyne cycloaddition (CuAAC) has become a popular alternative, as it is compatible with DNA and proceeds rapidly at ambient temperature to give stable, non-hydrolysable macrocycles. In this approach the linear peptide–DNA conjugate is synthesized with N-terminal azide and C-terminal alkyne, and treatment with a copper(I) catalyst links the two termini via a triazole linkage to form a macrocycle. The reaction is highly selective and tolerates unprotected side chains, making it suitable for DEL conditions where other functional groups are present. Post-synthetic cyclisation strategies provide another approach: the peptide can be assembled on solid support using orthogonal protecting groups, then cyclized using native chemical ligation or lactam bridge formation, and finally conjugated to the DNA barcode using a mild, terminal alkylation or click reaction. In this two-step approach, the cyclisation chemistry is decoupled from DNA exposure, and reagents that would otherwise damage the nucleic acid are acceptable. Thioether bridges have also been introduced in DEL peptides by introducing cysteine residues at both termini and reaction with a bis-electrophilic linker; the resulting thioether macrocycle is stable and redox-insensitive, overcoming a significant issue with disulfide-based cyclisation. The method of cyclisation is therefore chosen based on the structural needs of the target, the desired level of metabolic stability and the tolerance of the DNA tag to the reaction conditions. Optimization of cyclisation efficiency is an active area of research: recent examples focus on catalysts, DNA-compatible reagents, and microfluidic platforms that can perform cyclisation on miniaturised scales to retain the full theoretical diversity of the DEL in the final, screened population.
DELs are characterized by the unique combination of chemical flexibility and sequence-readability. They combine the advantages of screening in solution like those based on phage display or OLE with the synthetic diversity of combinatorial chemistry. In contrast to OBOC libraries where individual macrocycles are loaded on separate physical beads and require individual screening, DELs allow to generate, pool and screen millions of cyclic peptides in a single container in multiplexed fashion, which reduces discovery time, and saves materials and labor. Owing to the lack of physical separation between individual library members, the potential theoretical diversity of a DEL is no longer limited to the millions, but more than ten billion unique sequences can be created by repeated split-and-pool synthesis reactions. In addition, the encoding DNA strand is a highly stable and amplifiable identifier that is not removed or altered in aqueous buffers, in mild organic solvents, or during washing procedures of affinity-based selections, as is the case for chemical barcodes on individual beads which may be bleached, leached off, or become obscured by proteins and other contaminants. In comparison to phage display, which is also genotype-linked, DELs are much more chemically flexible, allowing direct integration of non-proteinogenic residues, N-methylated amides, D-amino acids, and backbone-modified monomers during solid-phase peptide synthesis, instead of being limited to the 20 standard amino acids. In addition, DEL selections are performed in solution or on immobilized targets under defined chemical conditions, rather than requiring a biological selection in phage hosts, which can suffer from variable clone-specific infectivity or toxicity that biases against rare library members. The analysis of DELs is also more informative compared to standard high-throughput screening assays where a single endpoint is measured per well: next-generation sequencing not only allows to monitor the relative abundance of all library members, but also reveals quantitative enrichment scores for each compound in the library, which can be used to assess subtle structure–activity relationships, and to train statistical models to predict the optimal cyclisation point.
| Method | Diversity | Scalability | Detection |
|---|---|---|---|
| OBOC | 10⁶–10⁸ | Easy | On-bead |
| DEL | 10⁹–10¹² | Excellent | Sequencing |
DELs have become a powerful technology in drug discovery, ligand optimization and materials science. The integration of DELs into the industrial peptide pipeline as a front-end lead generation engine is ongoing. DELs in drug discovery have resulted in highly active hits across challenging targets such as protein–protein interaction surfaces, allosteric kinase pockets and GPCR extracellular domains, where high-throughput screening is usually not an option. Simultaneously screening billions of macrocycles in one vessel allows for the rapid identification of cyclic peptide inhibitors of oncogenic signalling complexes, viral entry mechanisms or immune checkpoint pathways, and several have been developed into lead optimization programmes. The modular DEL synthesis is easily matured to get from hits to lead molecules: once a consensus sequence is established, off-DNA resynthesis with a series of systematic modifications such as N-methylation at key positions, substitution of key residues with non-canonical amino acids or change of ring size allows affinity, selectivity and PK tuning within weeks instead of months. DELs have also found applications in materials science, where DEL-inspired encoding strategies are being developed to create functional peptide coatings that self-assemble on the surface of nanoparticles or biosensors. In these cases, the DNA tag identifies not only the active sequence but also programs the peptide spatial arrangement on the material interface in order to control properties such as biocompatibility, catalytic activity or stimulus-responsive release. The industrial integration of DELs into peptide pipelines has matured beyond boutique academic efforts to scalable, good-manufacturing-practice-compatible processes. Automated split-and-mix synthesizers run on 96-well plates or microfluidic cartridges minimize reagent consumption and maximize reproducibility. Quality-control (QC) protocols now include DNA-based assays that rapidly diagnose cyclisation efficiency and only release libraries for screening after confirming that linear impurities will not confound hit identification. The close coupling of DEL screening and downstream pharmacological profiling, where hit peptides are synthesized off-DNA and tested in cellular assays, xenograft models or toxicity panels, has created a seamless discovery-to-development continuum that speeds the translation of cyclic peptide leads into clinical candidates.
DELs are expected to become the starting point for first-in-class medicines for targets that are otherwise refractory to small molecules. This transition will be enabled by a number of concurrent developments. DNA-compatible chemical reactions will be applied more broadly, allowing DEL libraries to incorporate complex macrocycles with a diversity formerly only available in scaffold-hopping combinatorial libraries. The impact of AI in DEL design will also begin to be felt, where AI methods are applied to the identification of chemical scaffolds and library design parameters that maximise hit rates in high-throughput screens. Automated DNA-parallel split-and-mix synthesis workstations will be the norm, with split-and-pool cycles being performed at micro-litre scale using liquid handlers. This will allow libraries of >1010 members to be constructed at a fraction of the reagent costs. Improved in-line quality control in these automated workstations will also allow more detailed tracking of the cyclisation efficiency of DEL synthesis and identification of linear impurities via DNA-hybridization assays, to ensure that cyclisation sub-populations do not contaminate DEL screening pools. Machine-learning algorithms will be developed that are able to predict, based solely on DEL sequence data, the cyclisation junctions most likely to produce stable macrocycles, as well as which target classes will be most likely to be druggable with a cyclic peptide. These data will be used in the rational design of DELs that are more likely to be successful, and thus have higher hit rates in DNA-encoded screens. New technologies, such as phage–mRNA hybrid display, or cell-free transcription–translation platforms, will begin to be used in tandem with DELs to enable true continuous evolution. Hits identified from DNA-encoded selection events can be immediately amplified and diversified in situ, in a self-optimizing discovery loop. This will be supported by a closer integration of DEL discovery and structural biology/computational docking methods, delivering next-generation cyclic peptide therapeutics that match the binding affinity of antibodies and exceed the cell permeability and oral availability of small molecules.
Combinatorial and DNA-encoded libraries (DELs) enable unparalleled diversity and screening efficiency for cyclic peptide discovery. If you're exploring new binders, optimizing target selectivity, or initiating large-scale hit campaigns, our advanced library technologies provide the flexibility and depth you need.
We offer:
Our integrated solutions help you rapidly identify high-value cyclic peptide leads while reducing time, risk, and cost.
References
