A typical human cell is estimated to contain 130,000 binary protein-protein interactions at any given time. Disruption of any of these protein-protein interactions (PPIs) has the potential to alter cellular physiology and ultimately lead to human disease. It is therefore not surprising that PPIs have become an emerging class of targets for the development of inhibitors, both as research tools to perturb the complex interaction networks that underlie cellular processes and as therapeutics against diverse human diseases and conditions.
Protein-protein interfaces are fundamentally different from the defined cavities of enzymes or GPCRs to which small molecules have been historically optimized for binding. As a result of this, the contact surface is usually large (1,500–3,000 Å2), flat and highly hydrophobic with no apparent grooves or cavities to accommodate a small ligand. As the binding energy is spread over many weak van-der-Waals and hydrogen-bond interactions, it is enthalpically challenging for a single 200–500 Da molecule to simultaneously displace dozens of amino-acid side chains to obtain a nanomolar affinity (rarely achieved in practice). Interface residues are also typically discontinuous in sequence space, which presents a structural obstacle that small molecules, limited by Lipinski's Rule of Five, simply cannot reach to mimic extended epitopes. Flat surfaces also lack concavity, which rules out shape-complementarity and means small molecules can only make limited contact with one face of the interface, resulting in at best micromolar affinities. PPI hot-spots contain limited nucleophilic residues thus eliminating covalent chemical approaches to strengthen weak non-covalent interactions. High-throughput screenings of standard libraries produce limited successful results and medicinal chemistry efforts struggle to achieve necessary therapeutic potency levels thereby pushing researchers toward alternative methods including cyclic peptides.
Cyclic peptides are a synthetic class of molecules that bridges the gap between small molecules and biologics: they offer the extended contact area needed to address flat PPI interfaces and are synthetically accessible. Cyclization preorganizes the pharmacophore by fixing the peptide backbone into a macrocycle. Binding leads to reduced entropy penalties to achieve antibody-like affinities (Kd ranges from low nanomolar to picomolar) for molecules weighing less than 2 kDa. Cyclic peptides lack free N- and C-termini, making them impervious to exoproteases, and backbone N-methylation or D-amino acid substitution make the scissile bonds unrecognizable to endoproteases, greatly increasing plasma half-life from minutes to hours. Tunable ring size and side-chain diversity also permit cyclic peptides to encircle the PPI surfaces and recapitulate critical α-helical or β-strand elements that drive the native interaction. Cell-penetrating stapled peptides and lipidated macrocycles have been shown to be active in cells against canonical intracellular protein–protein targets (HDM2/p53, BCL-2 family, β-catenin/Tcf) long thought to be undruggable by small molecules. Moreover, cyclic scaffolds can be tuned to achieve membrane permeability through carefully adjusted lipophilicity and conformational rigidity, achieving oral bioavailability in certain cases. The use of phage display, mRNA display, and more recently AI-driven de-novo design, allows cyclic peptide libraries of >1012 variants to be screened with high throughput, opening up the previously unapproachable PPI space to next-generation therapeutics.
Chemical modifications allow peptides to be tailored for success. They aim at locking the peptide's secondary structure into a stable and specific conformation to improve its biophysical properties. Notable approaches include cyclisation and backbone modification which have become prominent methods to bypass such limitations. The development of these strategies has enabled the creation of peptide-based inhibitors with superior inhibitory activity and PK properties when compared to their non-modified peptide counterparts. Peptide cyclisation has drawn significant interest as a result of the properties they confer. Cyclic peptides have demonstrated the ability to increase bioactivity and decrease toxicity, as well as show improved affinity for their targets. Peptide cyclisation aims to rigidify the structure of the peptide in its active conformation. An array of strategies (hairpins, stapling, hydrogen bond surrogates, etc.) have been developed to stabilise turns, helices and extended conformations within peptide scaffolds. The overall strategy is presented below according to the conformation of the native ligand, but is often based on the stabilisation of the peptide secondary structure.
Protein-protein interfaces are typically large (1,500–3,000 Å2), flat, and solvent exposed and thus difficult to target using conventional small molecules. In contrast, cyclic peptides are structurally pre-organized and present an extended, well-defined surface that can recapitulate the shape and charge complementarity of an interfacial epitope. The key to successful design is high-resolution structural data (X-ray, cryo-EM or NMR) to define the "hot spots" and secondary-structure motifs (α-helices, β-strands, or flexible loops) that drive the majority of the binding free energy. For α-helical epitopes, we apply hydrocarbon stapling to pre-organize the peptide into the desired helical conformation with the critical side chains aligned on the same face to match the topology of the target groove. Cyclic peptides derived from a loop are designed by transplanting the native dimerization loop (e.g. PHGDH residues 120–140) onto a macrocyclic scaffold followed by disulfide or lactam bridges to stabilize loop geometry without disrupting critical hydrogen bonds. Ring size and topology is optimized computationally (RosettaRemodel, flex ddG) to avoid entropic costs: a ring that is too small is strained, while an oversized ring is poorly defined. Finally, surface complementarity is fine-tuned by mutating peripheral residues to improve electrostatic or hydrophobic complementarity, as in the case of A7L, K10R and R16E substitutions that increased affinity to the PHGDH dimer interface.
Fig. 1 Main strategies applied to peptide-based inhibitors for protein–protein interactions (PPIs) in cancer therapy.1,2
Selectivity towards homologous proteins and metabolic stability are essential for the development of PPI-targeting cyclic peptides into drugs. Selectivity can be engineered by first performing in silico alanine scanning/mutational profiling of a binder to determine residues that either contribute disproportionately to binding (hot spots) or that can be varied to confer selectivity towards the desired target and away from off-targets. For example, substitutions like D11N or K13S create increased specificity for hydrogen bonds, whereas hydrophobic mutations such as L91F provide deeper pocket engagement without promiscuity towards other BCL-2 family members. Conformational locking of peptides with disulfide, lactam or hydrocarbon staples rigidifies the structure and also protects the scissile amide bonds from proteases; cyclic YAP-TEAD inhibitors were five-fold more potent and had significantly longer plasma half-lives than linear analogues. Similarly, non-natural amino acids like D-enantiomers, N-methylated residues, or β-amino acids, can be incorporated at solvent-exposed locations to further protect against aminopeptidases and carboxypeptidases while not disrupting the binding interface. For cell permeability, modifications such as lipidation or attachment to cell-penetrating peptides can be used, but are carefully selected to not affect selectivity. A C-terminal TAT cell-penetration tag was used to increase the intracellular delivery of anti-TEAD macrocycles without sacrificing target selectivity. Collectively, these structure-guided and chemical-biology approaches result in cyclic PPI inhibitors with nanomolar potency, exquisite selectivity, and improved metabolic stability that are suitable for in vivo testing and drug development.
It is intrinsically more challenging to synthesize the macrocyclizing peptides that we deliberately design to recapitulate large flat PPI interfaces, compared to more conventional therapeutic peptides. First, ring strain is significant when the sequence exceeds ~20 residues or the target epitope necessitates a rigid α-helix or β-hairpin spanning >12 Å of the partner protein surface. Given these limitations, the entropic cost of forcing the linear precursor into an intramolecular "closed" conformation typically relegates head-to-tail lactamization yields to under 30 %, with intermolecular oligomerization and C-terminal epimerization as strong competitors. Second, non-standard residues—D-amino acids, N-methylated amides, fluorinated aromatics, β-amino acids, or click handles (azide/alkyne)—are commonly incorporated to either stabilize bioactive conformations or to introduce membrane-permeating lipophilicity. These building blocks tend to be expensive, commercially rare, or incompatible with standard Fmoc chemistry. Bulky side chains hinder coupling efficiency leading to deletion sequences, and sensitive functional groups (e.g. azides) can decompose under TFA cleavage cocktails. We therefore maintain an in-house library of >250 Fmoc-protected ncAAs, deploy elevated-temperature coupling protocols (50 °C, HATU/HOAt), and apply orthogonal protecting-group strategies (allyl/alloc, ivDde) to enable late-stage diversification without compromising stereochemical integrity. Finally, for sequences where classical cyclization fails we turn to chemoselective stapling—ring-closing metathesis or CuAAC—performed in flow reactors that achieve pseudo-dilution conditions at gram scale, routinely delivering cyclization yields >70 % even for 25-mer macrocycles.
The crude mixture becomes a 'soup' with linear precursors together with both epimerized macrocycles and oligomers plus oxidized side products which only vary by a single stereocenter or amide bond. RP-HPLC resolution is no longer trivial because the additional macrocycle rigidity often shrinks retention-time windows; peak pairs that differ by<0.05 min must be resolved at 60 °C on sub-2-µm C18 or phenyl-hexyl columns to achieve baseline separation. High-resolution MS/MS is essential to distinguish isobaric dimers or oxidized products; ion-mobility spectrometry (IMS) separates chromatographically co-eluting conformers. NMR characterization is made more difficult by the lack of rotational freedom in the macrocycle; severe resonance overlap and ambiguous NOE assignments are the norm. We therefore acquire 2D 1H-13C HSQC and ROESY data at 600 MHz using cryogenic probes and supplement them with data collected on 15N-labeled samples produced using cell-free expression to unambiguously assign all backbone amide signals. Chiral purity is confirmed by Marfey derivatization and ultra-high-resolution LC-MS/MS; epimer content >0.1 % at this stage leads to re-purification or route redesign. Batch-to-batch consistency is also maintained by statistical process control (SPC) charts that monitor critical process parameters (temperature, coupling equivalents, cyclization time) across >30 consecutive runs.
Flat, solvent-accessible interfaces typical of PPIs often extend to >1,500 Å2 and are regarded as being intrinsically "undruggable" by small molecules. Our bespoke cyclic-peptide platform is specifically designed to address this problem. The process begins with structure-guided design: X-ray, cryo-EM or NMR data are interrogated to define the smallest "hot-spot" epitope (8–20 residues) accounting for ≥ 70 % of the binding free energy. Linear precursors are designed to mimic the native α-helix, β-hairpin or loop conformation and have flexible flanks removed to reduce entropic penalty. Linkers (disulfide, lactam, hydrocarbon, or triazole) are frequently employed to pre-organize the peptide into the bio-active topology; for instance, a disulfide-cyclized 17-mer (from YAP interface 3) was optimized to an IC50 of 25 nM against TEAD1, outperforming the corresponding full-length YAP fragment (IC50 = 40 nM). Non-natural amino acids (N-methylated residues, fluorinated aromatics, constrained dipeptides) can be readily incorporated via microwave-enhanced SPPS and verified by HR-MS/MS, to achieve 10- to 100-fold gains in affinity without loss of synthetic tractability. Each analogue is further characterized with a full SAR package (KD/Kpa, thermodynamic signatures, mutational scans), allowing iterative refinement to reach single-digit nanomolar potency.
Biological insight can sometimes exceed chemical supply rate in PPI programs, leading us to develop a 48-hour "hit-to-lead" cycle. Automated microwave synthesizers assemble linear precursors overnight, and microfluidic cyclization modules provide milligram quantities of macrocycles the following day. A cloud-based LIMS stores real-time UPLC-MS and SPR data to allow medicinal chemists to accept or reject analogues within the same week. A recent client campaign against the APC–Asef PPI evolved from a 57 nM lead to a 516 nM optimized variant in just three iterative rounds by utilizing diamino-diacid linkers that simultaneously enhanced cyclization yield and membrane permeability. Rapid orthogonal chemistries (click, RCM, enzymatic cyclization) are archived as digital protocols so that route switching involves only a barcode scan, not weeks of re-optimization.
Seamless scale-up is planned for all cyclic PPI inhibitors that make it to lead stage. We have a fully integrated GMP suite with 200 L solid-phase reactors, flow-cyclization skids and PAT-enabled purification trains, all capable of delivering multikilogram lots under ICH Q7 controls. Process parameters like temperature, base equivalents and redox potential have been fixed via Design-of-Experiments (DoE) studies to ensure batch-to-batch CV<3 %. We regularly hand over validated routes to partner CMOs with zero re-qualification and, for example, a recent YAP-TEAD program was scaled from 50 mg discovery material to 2 kg IND-grade API within 12 weeks and with a full CMC dossier (residual solvent profiles, endotoxin data, forced-degradation stability). Included in regulatory toxicology packages are cell-penetration studies (MDCK-II, Caco-2) and plasma stability panels, to ensure that the same peptide that performs in in vitro potency assays also survives in vivo exposure.
Peptide Synthesis Services at Creative Peptides
References
