Programmable Steric Control of Collagen Peptide-to-Fiber Assembly Authors: Pengfei Jin, David M. Chenoweth Published: 2026 ABSTRACT Peptide-based collagen-mimetic materials have drawn growing interest as fibrillar collagen surrogates due to their accessibility and versatility. However, an efficient strategy for assembling collagen peptides into well-defined collagen filaments remains elusive, owing to the nonspecific intermolecular interactions in collagen triple-helix formation. Here, we develop a new strategy for designing peptides that form triple-helical collagen-mimicking filaments using a single-residue side-chain modification. By organizing the intermolecular steric interactions in a “bump-gap” design, the peptides self-assemble into interlocked, endlessly growing triple-helical filaments with exceptional specificity. The peptide filaments exhibit high aspect ratios with micrometer lengths and are capable of forming networked structures that build hydrogels. These collagen-mimicking hydrogels, triggered by pH-dependent self-assembly, demonstrate superior stiffness over natural collagen with remarkable shear-thinning properties. Collagen is the primary structural component of the extracellular matrix (ECM), providing mechanical support to tissues and mediating cellular behaviors. , Materials derived from fibrillar collagen are widely used in biomedical applications, including tissue engineering, drug delivery, and regenerative medicine, because of their biocompatibility and biodegradability. − However, natural collagen carries risks of immunogenicity and pathogen contamination, and its chemical modification is challenging, limiting broader applications. − In contrast, collagen-mimetic peptides (CMPs) offer a safer and versatile alternative, as they can be efficiently synthesized and functionalized via solid-phase peptide synthesis (SPPS). Collagen fibers form through a characteristic triple-helix structure arising from repeating (Gly-Xaa-Yaa) triplets, often containing proline (Pro, P) at Xaa positions and (2S,4R)-4-hydroxyproline (Hyp, O) at Yaa positions. This modular architecture inspires the use of short CMPs to self-assemble into structures that mimic fibrillar collagen, although the design of a strategy that connects short peptides into continuous filaments with lengths comparable to natural collagen remains a significant challenge. In this work, we focus on constructing continuous single collagen triple helices by periodic assembly of CMPs. Although collagen peptides can be cross-linked as building blocks through disulfide bond formation, native chemical ligation, metal complex formation, , peptide amphiphile aggregation, π–π stacking, − cation−π interactions, and electrostatic interactions, the resulting materials comprise short triple-helical fragments limited by peptide length. In comparison, the structural unit of natural collagen, tropocollagen, spans over 1000 residues (∼300 nm) in each strand, which is critical for its structural strength. Using CMPs to build continuous collagen triple helices can overcome the limitations of natural collagen while resembling its structural features and mechanical properties. Collagen triple-helical self-assembly is relatively nonspecific in chain registers, resulting in peptide trimers with different configurations (Figure A). Building an elongated peptide filament requires one specific chain register; undesired registers could interrupt filament growth. Specific chain registers can be achieved by covalent linkages such as disulfide bonds, creating “sticky ends” and enabling filament growth (Figure B1). This strategy achieved ∼400 nm filaments by Raines and co-workers and was used to produce a collagen-mimicking gel by Koide and co-workers. Another strategy uses electrostatic interactions to enhance the specificity of peptide chain registers, first reported by Hartgerink and co-workers , and then developed by Raines and co-workers. , Side-chain salt-bridge-forming residue pairs, including the Lys-Asp pair, have been extensively investigated for promoting collagen heterotrimer formation. − By incorporating multiple Lys and Asp residues, the pairwise electrostatic interaction can provide enough specificity to encode the staggered peptide arrangement (Figure B2), resulting in peptide filaments, fibrillar bundles, and peptide gels. (A) Variation in chain registers of collagen self-assembly; (B) three strategies for constructing triple-helical collagen filaments. Herein, we report a new strategy that employs steric interactions to lock filament-forming peptide chain register. Periodic assembly of uniform short peptides along an endless filament naturally forms gaps between the peptide termini along the triple helix. We designed a bump substitution on adjacent peptide strands that fits into the gap without disturbing the triple-helical structure (Figure B3). By optimizing the gap width, bump size, and bump position, we can ensure that the bump residues can only be satisfied in the gap when self-assembled into triple helices, forming a specifically interlocked peptide filament. This design introduces three key innovations: (1) chain-register specificity is achieved by destabilizing all undesired registers rather than stabilizing the preferred one; (2) the single bump residue design provides minimal modification to the peptides, avoiding the synthetic complexity of branched peptides and permitting potential functionalization outside the bump-gap motif; and (3) incorporation of a charged bump residue enables pH-dependent self-assembly. The steric bump design relies on substitution at the glycine position. The tightly packed collagen triple helix is highly sensitive to steric perturbation; substitution of an interior buried glycine by any side-chain-bearing amino acid is extremely destabilizing, reducing the melting temperature (T m, the temperature at which 50% of peptides unfold) of a collagen triple helix by up to 60 °C. − The gap structure is achieved by deleting a Gly-Pro segment in a poly­(Gly-Pro-Hyp) peptide strand, exposing the amidated Hyp C-terminus on one side of the gap and the positively charged Hyp N-terminus on the other side. We initially chose aspartic acid as the bump residue due to its proper size and negative charge, which was expected to form a salt bridge with the N-terminal Hyp of the gap, resulting in additional stability and pH-dependence (Figure B). (A) Schematic self-assembly of a filament-forming CMP. (B) Structure model of collagen triple-helix with a bump-gap motif (modified from PDB: 3B0S). The Asp bump side chain is highlighted in the space-filling model. (C, D) Two types of filament-forming CMP sequences derived from arranging the repeating unit (dashed orange boxes) in different screw directions. The helical symmetry of the collagen triple helix imposes rules for designing the filament-forming peptide sequence in which the positions of bump residues are the critical parameters. By defining a certain helical periodicity, the chain register of a triple-helical filament can be determined by placing the bump-gap motif at regular intervals along the triple helix. The bump-gap motif can distribute in left-handed or right-handed helical symmetry, yielding two types of sequences with the Asp bump near the C- or N-terminus, respectively (Figure C,D). Consequently, a filament-forming CMP sequence is determined by two parameters: (1) the relative position of the bump residues (close to an N- or C-terminus); (2) the length of the repeating units. To investigate the minimal CMP length necessary for self-assembly into filaments, following the formulation in Figure , we started with short peptides that are either 22 residues (peptide 1) or 28 residues (peptides 2–4) in length (Figure A). All peptides were prepared with C-terminal hydroxyproline amidation and a free amino group on the N-terminal hydroxyproline. Aza-glycine (azGly, azG) incorporation was essential for promoting triple-helix formation. Gly to azGly replacement was originally developed to stabilize the triple helix through additional hydrogen bonding and conformational preorganization. − A single azGly substitution can increase the triple-helix T m by 10 °C at the center of a CMP and around 5 °C near the termini; the incorporation of multiple azGly residues can hyperstabilize short CMPs. Filament formation was expected when the peptides were buffered to allow the proposed salt bridge to form; triple-helix formation was characterized using circular dichroism (CD) by monitoring the 225 nm characteristic band. Peptide 1 did not form a triple-helix structure, and variable-temperature CD measurements indicate a random coil conformation. Peptide 2 precipitated upon buffering, indicating unusual structures since CMPs with similar lengths and charges are typically highly water-soluble. Considering that the six triplets near the N-terminus tend to form homotrimers and compete with filament formation (Figure S2), we designed peptides 3 and 4 with fewer azGly residues to destabilize the homotrimer and enhance filament-forming specificity. Peptide 3 precipitated similarly to peptide 2, whereas peptide 4 gelled at neutral pH and 2 mM concentration (0.5% w/v) (Figure B). This gelation behavior at low concentrations strongly supports the formation of long, one-dimensional structures with physically cross-linking properties. (A) Sequences and self-assembly results of filament-forming CMPs. *2 mM peptide in 20 mM tris buffer (pH = 7.4). (B) Hydrogel of peptide 4. (C) CD melting curve of peptide 4 at varied pH. (D–F) TEM (D), cryo-EM (E), and AFM (F) images of self-assembly structures of peptide 4. (G) Cryo-EM image of a single filament. (H) Filament diameter determined by stacking the radial intensity profile from panel G. (I and J) TEM images of peptide 4 self-assembled for 24 h (I) and 7 days (J). The gelation of peptide 4 is pH-dependent because the salt bridge between the Asp bump and the N-terminal Hyp forms only when the Asp side chain is deprotonated. Under acidic conditions, CD melting data indicate that the triple helices unfold at a lower temperature (T m = 25 °C, Figure C), implying peptide 4 is partially folded at room temperature. When buffered to neutral pH, the salt bridge stabilized triple helix of peptide 4 exhibits a T m of 40 °C, leading to a higher degree of supramolecular polymerization and a longer filament length that facilitates hydrogel formation. Multimeric peptide aggregates were further confirmed by size-exclusion chromatography with multiangle light scattering (SEC-MALS) analysis of peptide 4 at low temperature (Figure S4). Evidence for both peptide filaments and filament bundles was obtained by combining SEC-MALS with concentration-dependent CD measurements (Figure S5). The filament and gel network structures of peptide 4 were confirmed by TEM and AFM imaging. Peptide 4 self-assembles into filaments with high specificity, resulting in micrometer-length rigid filaments (Figures D, S6). The peptide filaments exhibit affinity for each other and afford networks for gelation, as characterized by AFM and cryo-EM (Figures E,F, S12–S20). Quantitative analysis of a single peptide filament in the cryo-EM image yields a diameter of 1.7 nm, which is consistent with the 1.5 nm diameter of a typical collagen triple helix (Figure G,H). To further investigate the aggregation of peptide 4, we varied the filament growth time at 4 °C. Compared to the 24 h incubation, filaments equilibrated for 7 days exhibited thicker bundles, with width increasing from 4 to 8 nm (Figures I,J, S9) and persistence length from 0.8 to 1.9 μm (Figure S10, fitted by FiberApp), indicating higher stiffness. Small-angle X-ray scattering (SAXS) analysis was further used to characterize peptide 4, revealing a broad peak at 1.2 nm consistent with the distance between triple helices within nanofibers (Figure S7). In addition, peptides 2 and 3 also form filamentous structures, according to TEM imaging. Higher numbers of azGly residues appear to hinder filament-forming specificity but promote filament aggregation (Figure S8). Additional peptides were prepared to validate our filament assembly model (Tables S1, S2). The efficiency of the Asp bump was supported by alternate bump residues, including Asn, Glu, His, Cys, Phe, and Tyr, in variants of peptide 4. While the Glu bump resulted in weaker gelation, all other bump variants completely disrupted gelation, supporting the salt bridge stabilization effect in the bump-gap motif. Other variants of peptide 4 lost gelation properties when gap size, bump position, peptide length, or terminal capping was altered, suggesting that peptide 4 forms triple-helical filaments as designed rather than other types of aggregation. Additional details are provided in the Supporting Information. The gelation of peptide 4 was monitored by using rheology at various concentrations (Figure A). All samples of peptide 4 gelled within 1 min after buffering, indicated by G′/G″ > 1. The peptide hydrogel exhibits high stiffness with a storage modulus (G′) of 1.6 × 103 Pa, 1.0 × 104 Pa, and 3.6 × 104 Pa at 0.5%, 1.0%, and 2.0% w/v concentrations, respectively (Figure B). Unusual shear sensitivity was observed, with the gel–sol transition (G″ > G′) occurring at 5% strain and 1 Hz oscillation for a 0.5% w/v concentration (Figure B). This pronounced shear-thinning property likely results from shear-induced fragmentation of the supramolecular filaments, as evidenced by the greater shear sensitivity at higher peptide concentrations, where elevated cross-link density restricts filament mobility and promotes breakage. To compare the peptide hydrogel with natural collagen, we characterized bovine collagen and Matrigel under the same conditions (Figure D). The hydrogel of peptide 4 exhibits a storage modulus G′ 3-fold higher than bovine collagen and 160-fold higher than Matrigel. The higher G′ of the peptide 4 hydrogel under these specific conditions likely arises from its longer filament length, higher rigidity (due to its proline-rich sequence), and filament aggregation behavior. Unlike the peptide 4 hydrogel, which melts under 5% shear strain, bovine collagen remains gelled under 100% strain, and Matrigel is not shear-thinning. In variable-temperature rheology experiments, peptide 4 remained gelled at 37 °C (Figure S11). The unique rheological characteristics of the filament-forming CMP hydrogel could allow it to function as a high-stiffness, injectable gel, enriching the array of biocompatible hydrogels. (A) Monitored gelation of peptide 4 via oscillatory shear rheology (strain = 0.5%, frequency = 1 Hz). Samples were buffered at t = 0 at 25 °C. (B and C) Amplitude and frequency sweeps of gelled peptide 4 at varied concentrations. (D) Amplitude sweeps of bovine collagen, Matrigel, and peptide 4. All samples were prepared at 0.5% (w/v) in 20 mM tris buffer and allowed to gel for 30 min. Manipulating steric interactions provides an efficient strategy for controlling the self-assembly of CMPs and constructing collagen-mimetic filaments. The bump-gap design introduces a novel class of peptides capable of self-assembling into triple-helical filaments that mimic fibrillar collagen. Beyond the long aspect ratio filaments and unique gelation properties, the introduction of a single side-chain bump residue may be compatible with further functionalization, enabling the design of peptide hydrogels with tunable physical and protein-binding properties, for a broad range of bioengineering applications. Supplementary Material