Amino acids are bifunctional organic molecules that possess both a basic amino group (―NH₂) and an acidic carboxyl group (―COOH). Their chemical diversity arises from differences in the types of side chains (R groups), which are formed when the hydrogen atom on the carboxyl carbon is replaced by an amino group. The structural differences among amino acids depend on the nature of their side chains, and the 20 amino acids are classified according to the chemical structure or properties of their side chains^[1-2].They can be divided into polar and nonpolar amino acids. As the simplest biological building blocks of living organisms, amino acids not only participate in metabolic activities and signal transduction but also form polypeptide chains through peptide bond condensation, serving as the cornerstone of biomolecular self-assembly. A peptide is a biologically active molecule formed by the covalent linkage of two or more amino acids via a peptide bond, and polypeptides—sequences of amino acids linked by peptide bonds—are among the most important classes of biomolecules^[3-4]. Their self-assembly is widespread in nature. Polypeptides, with their sequence design, dynamic folding, and abundant noncovalent interaction sites (hydrogen bonds, π–π stacking)^[5], exhibit unique advantages in nanotechnology and biomedicine. However, linear polypeptides are susceptible to enzymatic degradation, and their conformational flexibility makes it difficult to precisely control assembly pathways. This limitation has driven the development of cyclic peptide molecular platforms, which offer greater stability and functional plasticity.

Cyclic dipeptides are composed of two amino acids that condense and dehydrate, with the terminal —COOH of one amino acid and the terminal —NH₂ of the other amino acid linked via a peptide bond, releasing a molecule of water to form a six-membered heterocyclic compound. They represent the smallest cyclic peptides containing a core heterocyclic lactam ring with 6 members, featuring two amide groups at opposite positions in the ring. The simplest example is cycloglycine, also known as 2,5-diketopiperazine (DKP)^[3].Cyclic dipeptides form molecular chains or layers through N―H…O hydrogen bonds between adjacent molecules^[6-8].Compared to linear peptides, cyclic dipeptides are more attractive due to their superior properties, including higher structural rigidity, resistance to proteolysis, metabolic stability, and enzymatic stability. They can also form multiple hydrogen-bonding sites (including two hydrogen-bond donors and two hydrogen-bond acceptors), enabling intermolecular hydrogen bonding between adjacent cyclic peptide molecules. These hydrogen-bonding sites may play a potential role in the formation of self-assembled structures or complexes, further facilitating the construction of higher-order supramolecular architectures. As a result, cyclic peptides and their derivatives have attracted attention in areas such as photoluminescence, drug research, biomedicine, and as potential building blocks for advanced nanodevices^[9-11].However, the precise molecular design of cyclic dipeptide-based materials and the flexible control of their self-assembly processes remain significant challenges. By systematically summarizing the structure and properties of cyclic peptide self-assembly, we not only gain a deeper understanding of the intrinsic relationship between structure and properties but also enable targeted manipulation of their self-assembly structures, thereby inspiring and guiding the design of novel functional bio-inspired materials.

This article presents a systematic study and discussion centered on three research levels: structural analysis of cyclic dipeptide self-assembly, assembly mechanisms, and functional applications. At the molecular design level, single-crystal X-ray diffraction and molecular dynamics simulations are combined to reveal the cooperative driving mechanisms of molecular packing modes and non-covalent interactions. In terms of assembly mechanisms, the article elucidates the entropy-driven regulation of assembly pathways in real time. Regarding functional applications, it focuses on breakthrough advances in piezoelectric sensing, optical waveguiding, and bioactivity, with illustrative examples provided. By analyzing representative systems, the article uncovers the structure–property relationships among molecular design, microstructure, and macroscopic performance. Addressing current challenges—such as the unclear regulatory mechanisms of molecular chirality on assembly pathways, the difficulty in reconciling long-range order in nanostructures with large-scale fabrication, and biocompatibility trade-offs—the article proposes interdisciplinary solutions that integrate machine learning prediction, dynamic responsive design, and biomimetic hybridization. This work not only provides a network framework for the rational design of non-covalent interaction networks but also outlines technological pathways for developing intelligent drug carriers and biodegradable electronic devices, thereby facilitating the translation of cyclic dipeptide-based materials from fundamental research into applications in biomedicine and flexible optoelectronics industries.

In summary, existing scoring systems have limited predictive capabilities for bleeding events, and their results are inconsistent^[25,30,33].

The functional properties of cyclic dipeptides and their derivatives can be confirmed through crystalline self-assembly. By selecting an appropriate benign solvent, the sample is dissolved at high temperature and then cooled according to a programmed protocol. When the supersaturation of the system reaches a critical threshold, crystals precipitate. Crystallographic analysis reveals that cyclic dipeptides based on tryptophan form monoclinic or orthorhombic structures with a high aspect ratio and a needle-like morphology^[14-15]. Basavalingappa et al.^[16]designed and crystallized cyclo-Dip-Dip (Dip: β,β-diphenyl-Ala-OH), and used single-crystal X-ray analysis to gain deeper insights into the molecular structure and non-covalent interactions of the cyclic peptide. Cyclo-Dip-Dip forms an orthorhombic space group P2₁2₁2₁(Fig. 1b). Along the crystallographic aaxis, parallel stacking generates a structure of parallel β-sheet arrays, whose stability is synergistically driven by two hydrogen bonds between aromatic rings and two face-to-face π–π interactions, as shown in Fig. 1c. On the bcplane, hydrophobic interactions near the crystal lead to the assembly into a monolayer. In higher-order structures, cyclo-Dip-Dip forms layer-by-layer arrangements, where adjacent layers achieve a stable higher-order structure through a “zipper-like” interlocking of more loosely packed aromatic rings (Fig. 1d). This multi-level assembly pattern validates the core principle in entropy-driven mechanisms, where local order guides long-range order.

Molecular self-assembly refers to the spontaneous association of individual molecules under thermodynamic conditions via non-covalent interactions to form a well-defined, orderly arranged, and relatively stable supramolecular structure; it is a ubiquitous process in nature^[17]. At its core, molecular self-assembly reduces the system's free energy, transforming the system from a disordered state into a highly ordered, functionally superior structure, with intermolecular interactions driving molecules into a stable, low-energy state^[18]. Self-assembly enables materials to organize and scale up in a "bottom-up" manner^[19-20], skillfully assembling biological building blocks into precisely sized, ordered suprastructure, thereby providing an effective pathway for the design and development of functional biomaterials.

From a chemical structural perspective, cyclic dipeptides consist of a DKP scaffold and various amino acid side chains. DKPs provide a broad range of hydrogen bonds that facilitate the self-assembly of cyclic dipeptides, while different amino acid sequences or side chains modified by functional groups offer multiple non-covalent interactions—such as hydrophobic interactions, van der Waals forces, π–π stacking, aromatic interactions, and electrostatic interactions—that play distinct roles in the self-assembly process. In addition, external conditions such as temperature, pH, ion concentration, solvent, and enzymes also influence the kinetic processes of cyclic dipeptide self-assembly. Ultimately, driven by the coordinated interplay of multiple interactions, self-assembly depends on the cumulative effect of numerous low-energy-state interactions^[21-23],which can maintain the integrity and stability of the self-assembled architectural framework and further assemble into various ordered supramolecular nanostructures (including nanofibers, nanobelts, and nanotubes), as shown in Figure 2 ^[17,24]. Different assembly structures arise from the distinct, orderly intramolecular packing controlled by various non-covalent forces within the cyclic dipeptide building blocks. Non-covalent interactions are widespread in nature; therefore, a thorough understanding of the properties of different non-covalent interactions and the effective utilization of their interactions is essential^[25]. Under the driving force of multiple non-covalent interactions, the manner in which cyclic dipeptides and solvents pack at the molecular level varies, leading to differences in the final self-assembled nanostructures formed. By precisely designing the intermolecular interactions of cyclic dipeptides and the self-assembly conditions, and by further exploring the thermodynamic and kinetic processes of cyclic peptide self-assembly, it is crucial to skillfully balance the intermolecular non-covalent interactions. Peptides dissolved in a solvent can adopt a specific conformation that determines whether self-assembly occurs, thereby favoring the formation of secondary structures such as α-helices, β-sheets, and β-hairpins^[17],and giving rise to functional materials with outstanding chemical and physical properties. On the other hand, due to their cyclic configuration, cyclic dipeptides eliminate free amino and carboxyl groups, thereby exhibiting a stronger propensity for self-assembly and the formation of organized supramolecular structures^[14,26]. From the perspective of molecular engineering, the rigidity of the cyclic structure and the broad, superior hydrogen-bonding capacity of cyclic dipeptides endow them with more complex architectures. The unique, hydrogen-bond-driven strong intermolecular interactions make cyclic dipeptides highly prone to molecular self-assembly, and depending on the nature of the α-substituent (or R group), cyclic dipeptides can participate in intermolecular interactions to form one-dimensional and two-dimensional hydrogen bonds^[27-28].

In essence, the self-assembly behavior of cyclic dipeptides is a dynamic equilibrium process governed by their intrinsic network of noncovalent interactions. By introducing specific amino acid side chains, the assembly pathways and functional outputs can be precisely tuned. Amino acids containing aromatic groups, such as phenylalanine, tryptophan, and tyrosine, can provide hydrogen bonding, π–π stacking, or polar interactions^[29-31]. For example, Jeziorna et al.^[32]studied two diastereomers of cyclo-tyrosine-alanine, cyclo-L-Tyr-L-Ala and cyclo-L-Tyr-D-Ala. The structure contains phenyl and hydroxyl groups, which are bifunctional residues that facilitate intermolecular contacts, including π–π interactions and hydrogen bonding, jointly driving self-assembly. Similarly, Govindaraju et al.^[19]designed and investigated the self-assembly of differently chiral cyclo-phenylalanine-phenylalanine, cyclo-L-Phe-L-Phe and cyclo-D-Phe-L-Phe, where hydrogen bonding and aromatic π–π stacking synergistically drive the formation of highly stable two-dimensional sheet structures with large lateral surface areas. In addition, the introduction of charged amino acids such as arginine and lysine generates electrostatic forces^[33]; their long side chains can both regulate assembly kinetics through electrostatic repulsion and serve as structural water molecules to form water-mediated hydrogen-bonding networks. Moreover, the different geometries of DKPs and the arrangement of water molecules also influence self-assembly. Subtle differences in these cooperative mechanisms and variations in chirality lead to significant differences in self-assembly and the formation of higher-order structures: cyclo-L-Tyr-L-Ala forms nanotubes and nanowires, whereas cyclo-L-Tyr-D-Ala forms only nanotubes^[32]. With the exception of glycine, each DKP has four possible stereoisomers. Because the chiral centers are located in the rigid core scaffold, each stereoisomer arranges its side chains in a specific spatial orientation, thereby directly influencing its interactions with biological targets and potentially eliciting markedly different biological effects. For example, all four stereoisomers (LL/LD/DL/DD) of cyclo-Leu-Pro, cyclo-Val-Pro, and cyclo-Phe-Pro^[11]can be distinguished by electronic circular dichroism, with each enantiomer exhibiting a symmetric yet mirror-image spectral response, further highlighting the critical role of chirality in regulating function. These advances demonstrate that rational design of noncovalent interaction networks can endow cyclic dipeptide materials with precise structures, laying the foundation for the development of next-generation intelligent biomaterials.

The core objective of cyclic peptide self-assembly research is to achieve precise customization of assembly morphology and function through rational amino acid design and the regulation of non-covalent interaction networks. The development of any class of self-assembling materials requires an understanding of the fundamental mechanisms that drive their behavior, which is of great significance for deepening our understanding of interactions between biomolecules and for advancing innovative materials. Many research groups have designed and synthesized cyclic peptides composed of different amino acids, crystallized these peptides, and investigated their microstructures, as well as the roles and synergistic driving forces of various non-covalent interactions—including hydrogen bonding, electrostatic interactions, and aromatic interactions—in the self-assembly process. By elucidating the self-assembly mechanisms and patterns, researchers aim to gain a deeper understanding of the intrinsic relationships between various non-covalent forces and polypeptide supramolecular nanostructures, thereby enabling targeted manipulation of assembly structures. In addition, the fundamental thermodynamics and kinetics underlying the crystallization mechanisms of peptide self-assembly are analyzed and studied. Current research indicates that the synergistic interplay of hydrogen bonding and π–π stacking is the core mechanism driving the formation of β-sheet-like crystalline structures (cyclo-Dip-Dip), while chiral differences (L/D-type amino acids) can also selectively give rise to nanowires or nanotubes.

To enhance the performance and practical applications of cyclic peptide-based biomaterials, several issues and challenges remain in this field: First, there is a bottleneck in mechanistic understanding; in situ characterization data on the kinetic parameters of chiral cyclic peptide assembly pathways are lacking, which hinders the controlled growth of long-range ordered structures. Second, large-scale preparation poses significant challenges: single-crystal growth at the laboratory scale takes a long time, and polycrystalline, dendritic, and twinned materials lead to performance variability. Third, application adaptability is limited, as the biodegradation rate and device durability are difficult to reconcile. The existence of these challenges severely restricts the design and development of novel functional materials with ideal performance for future applications. To overcome these challenges, a multidisciplinary, integrated development strategy is required. First, computational-assisted design should be employed, using AI-driven computational simulations to model assembly pathways for specific sequences, enabling the design of larger cyclic oligopeptides (such as cyclic tripeptides and cyclic tetrapeptides), amino acid substitutions (including side-chain modifications, enantiomers, peptide sequences, or derivatives), and flexible assembly methods (co-assembly, covalent conjugation). More complex reactions and self-assembly under different conditions (e.g., various solvents, physical vapor deposition) can be introduced to further tune the structure and properties, while machine learning is used to optimize solvent systems and predict morphological structures and performance orientations. Second, dynamic regulation technologies should be developed, such as light/heat-responsive cyclic peptides that enable in situ reversible adjustment of nanotube diameters. Finally, biomimetic hybrid engineering should be pursued by combining cyclic peptide crystals with conductive polymers to enhance the cycling stability of flexible devices. Although cyclic peptide-based materials have demonstrated potential in cancer treatment, antibacterial applications, targeted drug delivery, and biosensing, their clinical translation still requires further experiments to comprehensively and systematically evaluate long-term in vivo toxicity, side effects, and metabolic pathways. Future research should focus on innovation across the entire value chain—molecular design, process development, and application validation—to advance these green, intelligent materials from the laboratory to industrial-scale production^[55-56]..