Rigid helical-like assemblies from a self-aggregating tripeptide

Abstract

The structural versatility, biocompatibility and dynamic range of the mechanical properties of protein materials have been explored in functional biomaterials for a wide array of biotechnology applications. Typically, such materials are made from self-assembled peptides with a predominant β-sheet structure, a common structural motif in silk and amyloid fibrils. However, collagen, the most abundant protein in mammals, is based on a helical arrangement. Here we show that Pro-Phe-Phe, the most aggregation-prone tripeptide of natural amino acids, assembles into a helical-like sheet that is stabilized by the dry hydrophobic interfaces of Phe residues. This architecture resembles that of the functional PSMα3 amyloid, highlighting the role of dry helical interfaces as a core structural motif in amyloids. Proline replacement by hydroxyproline, a major constituent of collagen, generates minimal helical-like assemblies with enhanced mechanical rigidity. These results establish a framework for designing functional biomaterials based on ultrashort helical protein elements.

The self-assembly of peptides into diverse nanostructures is of key interest because of their wide-ranging applications in bionanotechnology, regenerative medicine and bioengineering¹. Successful design of the short peptide building blocks that form functional supramolecular nanostructures often relies on a trial-and-error approach or on mimicking biological elements^2–4. Among the minimal recognition modules, the Phe-Phe dipeptide, which forms β-sheet assemblies, is most studied due to its mechanical, optical and electrical properties^3,5–7. In a pioneering work, Ulijn, Tuttle and co-workers performed thorough in silico analyses aimed to predict the tendency of de novo designed short peptide sequences to form nanostructures in aqueous solution at neutral pH. Initially, all 400 possible dipeptide sequences were analysed by molecular dynamic simulations, confirming the unique aggregation propensity of sequences containing phenylalanine or tryptophan residues⁸. Subsequently, all the 8,000 tripeptides that could be constructed by combinatorial arrangement of the 20 natural amino acids were computationally analysed and ranked according to their aggregation propensity⁹. The results suggested Pro-Phe-Phe to be the most aggregation-prone tripeptide. Thus, atomic-level determination of the supramolecular arrangement of the Pro-Phe-Phe building block is highly important to understand the molecular basis of its great aggregation potential.

Peptides and proteins adopt two distinct atomic-level architectures of self-assembled hierarchical organization: β-sheet and helical conformations. The β-sheet structure is characteristic of amyloid fibres, which are well established to be associated with various neurodegenerative disorders¹⁰ and have also been suggested to be the basic functional motif in prebiotic Earth¹¹. The self-assembled nanostructures formed by short peptide sequences are predominantly based on amyloidal β-sheet organization^5,12. In this structural pattern, the most notable repetitive feature of the fibril is a set of β-sheets that are parallel to the long axis of the fibril, whereas the corresponding strands are perpendicular to this axis¹³. However, Landau and co-workers have recently provided atomic-level insight into the structural organization of functional amyloids formed by the 22-residue phenol-soluble modulin α3 (PSMα3) peptide secreted by Staphylococcus aureus¹⁴. X-ray analysis of the PSMα3 single crystal revealed a distinct architecture in which α-helices run perpendicular to the fibre axis. This original study implied that, in contrast to the prevalent dogma, a super-helical architecture in which the helical peptide backbone runs perpendicular to the fibre axis actually represents a generic structural motif in self-assembled peptide nanostructures.

Here, aiming to understand the higher-order organization of Pro-Phe-Phe, we have studied the self-assembly mechanism of the tripeptide by examining its circular dichroism (CD) and infrared spectral signatures, which are confirmed by single-crystal X-ray analysis. Surprisingly, our findings demonstrate that, markedly different from Phe-Phe, the Pro-Phe-Phe peptide forms unique helical-like sheets that mate via aromatic dry interfaces, an organization that has not been shown previously for any short peptide composed of coded amino acids. Furthermore, reengineering of Pro-Phe-Phe by substituting proline with hydroxyproline (Hyp), an essential component of collagen, affords super-helical assembly with mechanical rigidity comparable to that of collagen fibres.

Assembly and structural characterization of Pro-Phe-Phe

The self-assembly of Pro-Phe-Phe was explored using high-resolution scanning electron microscopy (HRSEM) and atomic force microscopy (AFM) (see Methods). The peptide adopted an elongated and unbranched helical fibre network morphology, with a high aspect ratio in phosphate buffer at pH 7.4 above the critical aggregation concentration (Fig. 1a-d and Supplementary Figs. 1-3). The amyloidogenic nature of the fibres was examined using a thioflavin-T (ThT) binding assay, an amyloid-specific fluorescent dye (see Methods). Staining the fibres with ThT resulted in high fluorescence levels, as well as a typical amyloid-binding emission signal, thus establishing their amyloidogenic nature (Fig. 1e and Supplementary Fig. 4). Kinetic studies using preformed matured fibres aimed at understanding the structural stability also exhibited a high fluorescence emission spectrum of the final plateau regime following ThT binding, resembling a characteristic kinetic profile of amyloid assemblies (Supplementary Fig. 4). To study the secondary structure of the peptide in solution, we carried out CD spectroscopy (see Methods). Surprisingly, unlike self-assembling nanostructures formed by Phe-Phe and other building blocks from this structural family, the tripeptide assemblies exhibited a CD signal comprising double-negative maxima at 210 nm and 224 nm, with a positive maximum at around 198 nm, characteristic of a helical conformation (Fig. 1f and Supplementary Fig. 5)¹⁵. These peaks are slightly redshifted compared to the canonical helical signal, which shows negative maxima at 208 and 222 nm, with a lower intensity. The lower intensity of the signals is consistent with an earlier observation that the depth of ellipticity rapidly decreases as the length of the helix is shortened¹⁵. Thus, the CD spectra are in accordance with a helical-like structural arrangement of the minimal peptide. This was also supported by the Fourier transform infrared (FTIR) spectrum (see Methods), which presented a sharp amide I band at 1,647 cm⁻¹ with a shoulder at 1,682 cm⁻¹, indicating the presence of a predominant helical conformation (Fig. 1g)¹⁶. To gain further information about the molecular spacing of the Pro-Phe-Phe assemblies, X-ray diffraction of the dried fibres was performed (see Methods). The powder pattern suggested monoclinic unit cells with a=5.330 Å, b=11.946 Å, c=34.360 Å and β=92.17° (Supplementary Fig. 6), considerably different from the typical diffraction of β-sheet amyloids. Thus, these results signify the stacking of distinctive units, rather than β-strands, as also suggested by the secondary structure analysis.

Fig. 1. Amyloid-like fibrillar assembly of Pro-Phe-Phe.

a,b, HRSEM micrographs of the tripeptide fibres prepared in phosphate buffer at pH 7.4. c,d, AFM images of the tripeptide fibres. e, Confocal fluorescence microscopy images of Pro-Phe-Phe fibres stained with ThT in phosphate buffer at pH 7.4 using 20 mM peptide and 20 μM ThT. f, CD spectrum of the tripeptide in solution, showing the presence of helical assemblies. g, FTIR analysis of the tripeptide with a characteristic helical peak. Scale bars, 10 μm (a,b,e).

Single-crystal X-ray structure of the tripeptide

To examine the molecular basis of fibre formation with non-canonical β-sheet structure, we crystalized Pro-Phe-Phe (see Methods). The space group and unit cell parameters of the formed crystals highly resembled those of the fibres (Supplementary Fig. 6), indicating a similar molecular organization. The tripeptide crystallized in the P2₁ space group, with one tripeptide molecule in the asymmetric unit, as shown in 50% probability displacement ellipsoids (Fig. 2a). The two aromatic side chains arranged in the same face relative to the peptide backbone, creating a hydrophobic region, with the charge termini displayed on the opposite site of the peptide backbone. Thus, Pro-Phe-Phe adopted an overall amphiphilic conformation with net segregation of the hydrophobic and hydrophilic components on opposite sides of the backbone¹⁷. The torsion angles around the Phe₂ residue appeared to play a pivotal role in dictating the overall structural features. The allowed torsion angles of the Phe₂ residue were found to be localized within the right-handed helical region of the Ramachandran plot, with Ψ₂ and Ψ₂ values of −78.5° and −38.9°, respectively (Supplementary Table 1). Two types of hydrogen bond facilitated the crystal packing of Pro-Phe-Phe: head-to-tail interactions involving terminal amine and carboxylic acid groups (Fig. 2b) and side-by-side stacking involving the amide groups (Fig. 2c). In the crystallographic b direction, Pro-Phe-Phe propagated through head-to-tail intermolecular hydrogen bonds, generating a helical-like arrangement of the peptide backbone, as shown in Fig. 2b,d,e. The void space inside the helical framework is demonstrated in top view presentation in Fig. 2d.

Fig. 2. Single-crystal structure of Pro-Phe-Phe in P2, space group.

a, ORTEP diagram of the asymmetric unit in 50% probability displacement ellipsoids. b, Head-to-tail hydrogen bonds viewed approximately along the b axis. c, Side-by-side stacking through amide hydrogen bonds along the a axis. d,e, Top view (d) and side view (e) of the helix. f, Aromatic zipper-like molecular packing of the adjacent helical-like assemblies along the c direction. g, Formation of an elongated structure by stacking of helices through intermolecular hydrogen bonds along the a direction. For clarity, the peptide helix is superimposed over an ideal helical model in e, f and g. Nitrogen and oxygen heteroatoms are shown in blue and red, respectively.

Tripeptides constituted from canonical amino acids and organized in helical assemblies have been reported very rarely¹⁸. Moreover, the three residues in these peptides are not sufficient to complete a helical turn, and at least one water molecule is required as an additional residue to link the translationally related peptide segments along the helical axis. In contrast, the Pro-Phe-Phe tripeptide forms a helical assembly without any additional solvent molecules in the crystal, making it a highly unique organization. The fourth residue required to complete a single helical turn is fulfilled by the terminal hydrogen-bonded addition of the next peptide molecule. As a result, the directionality of the amide groups essential for the i+4 → i backbone hydrogen bond parallel to the α-helix axis is changed. Instead, the backbone hydrogen bonds are oriented in the perpendicular direction of the helix. In the canonical helical conformations, these backbone hydrogen bonds are organized parallel to the helical axis. Thus, although the torsion angles of Phe₂ were found to be localized within the right-handed helical region of the Ramachandran plot, the hydrogen-bonding network produced a unique structural arrangement not previously observed in any biological system. Based on the torsion angle preference, the Pro-Phe-Phe supramolecular organization can be described as a helical-like structure differing from canonical helical conformations in its hydrogen-bonding direction. To the best of our knowledge, no atomic structure of a supramolecular helical tripeptide comprised only of natural amino acids has been reported. In the resulting structure, the centre of the helix is composed of a hydrophilic segment while its surface comprises hydrophobic Phe residues, as observed in the b-c plane (Supplementary Fig. 7). Adjacent helical-like structures run laterally with respect to each other in the c direction (Fig. 2f), with their interface stabilized by intermolecular interactions between the aromatic moieties of Phe residues arranged in an approximate T-shape, as commonly observed in helical peptide and protein crystals. Thus, the amphipathic conformations of Pro-Phe-Phe form a dry ‘aromatic zipper’ arrangement, which holds the superstructures together and stabilizes the helical-like assemblies¹⁷. In the perpendicular direction (a direction), nearby helices interact in a parallel pattern by alternately incorporating amide NH and CO groups in hydrogen bonding, stacking into a closely packed helicallike sheet¹⁹. As the helical strands run perpendicular to the stacking axes, addition of hydrogen-bonded Pro-Phe-Phe molecules in the growing sheet account for the formation of elongated fibre structures (Fig. 2g and Supplementary Fig. 7).

Several previous studies have explored the self-assembly of longer helical peptides into coiled coil nanofibres in which the helices are orientated parallel to the fibre axis^20,21. Longitudinal propagation of the coiled coil bundles leads to the formation of the fibres. Distinct from coiled-coil structures, a few longer helical peptides have been shown to arrange into layers of parallel or antiparallel packed amphiphilic helices perpendicular to the long axis of helix propagation with net segregation of polar and apolar surfaces, thus producing a bilayer structure with alternating hydrophilic and hydrophobic interfaces^22–25. A recent study has demonstrated the self-assembly of a range of tripeptides, Phe-Xaa-Phe (with alternating stereochemistry), all displaying β-sheet structures and organized in a parallel-to-fibre arrangement¹⁷. However, excluding the direction of backbone hydrogen bonds, fibre formation by the minimal tripeptide Pro-Phe-Phe (comprising a helical-like arrangement perpendicular to the fibre axis) closely resembles fibre formation by the 22-amino-acid PSMα3 peptide (Fig. 3a)¹⁴. The allowed torsion angle around the Phe₂ of the central core Lys-Phe-Phe residues in PSMα3 is also in the right-handed helical region of the Ramachandran plot, thus confirming the Pro-Phe-Phe tripeptide as a minimalistic model for the newly defined structural motif. For the tripeptide, helicallike sheets run parallel to the elongated axes (Figs. 2g and 3b). The molecular arrangement of Pro-Phe-Phe helical-like sheets is unique, with a structure that buries the hydrophilic component in a solvent-inaccessible environment and extends the hydrophobic parts outwards. Adjacent sheets are mated through hydrophobic dry steric zipper interactions between aromatic rings and stack laterally to generate a continuous row of combined sheets composed of helices along the fibre axis, thus forming an ordered helical-like supramolecular architecture. As these dry steric zipper assemblies of sheets expose bulky phenyl rings on both sides, the inter-sheet distance is 11.5 Å. The inter-strand distance of 5.3 Å resembles the characteristic meridional diffraction of an aggregated β-sheet structure (4.9 Å)¹³, resulting in an overall super-helical packing arrangement (Fig. 3c). In a typical cross-β and in the recently reported cross-α structure, the mated sheets are also stacked laterally through a similar hydrophobic dry interface, yet the rows of pairs of mated sheets are fully separated by wet interfaces lined with water molecules, forming alternating dry and wet interfaces. The pair of mated sheets featuring a dry steric zipper interface is considered the basic unit of amyloid-like fibrils in these arrangements. In contrast, the absence of wet interfaces in the Pro-Phe-Phe structure indicates continuous stacking of helical-like sheets through a dry interface along the c axis. This indeed explains the growth of the fibre in two directions (along the a axis and the c axis), both perpendicular to the helical axis. However, the shorter packing distance (5.3 Å) between helices implies dense packing and predominant elongation of the fibre along the a direction. Consequently, the two-dimensional growth of Pro-Phe-Phe produces wider fibres, as compared to typical amyloids, which elongate predominately in one direction as the solvent surrounds the pairs of mated sheets. An important aspect of this super-helical structure is the perpendicular orientation of the backbone hydrogen bonds relative to the helix axis, a counterintuitive packing compared to conventional helices and the newly described PSMα3. This configuration may have some resemblance to the folding of the nascent helices, which is dominated by short stretches of the amino acid (preferably three to four residues long), displaying a very strong preference for Ψ, Ψ angles corresponding to the helical region of the Ramachandran plot²⁶.

Fig. 3. Supramolecular helical architecture.

a, Crystal structure of PSMα3 and the central core Lys-Phe-Phe residues (PDB code 5i55)¹⁴. b, Formation of helical-like sheet and interaction between adjacent sheets in the crystal packing of Pro-Phe-Phe. The helices of two nearby sheets are shown in grey and red. Inter-sheet and inter-strand distances are shown. c, Arrangement of Pro-Phe-Phe in the crystallographic a-b plane. The aromatic ring of the Phe residue is represented as a green sphere.

Side-chain modification of Pro-Phe-Phe

To understand whether this helical-like conformation is unique to the Pro-Phe-Phe tripeptide or a robust structure that represents a more generic conformation, we modified the tripeptide sequence by replacing proline with hydroxyproline (Hyp). The hydroxylation of proline plays an important role in stabilization of the collagen triple helix, a central structural element in connective tissues, underlying its well-adapted physical properties²⁷. Structural modifications of short peptide sequences often abrogate the backbone conformation, as well as the higher-order packing of the relevant peptide building blocks²⁸. However, in spite of the additional hydrogen-bonding site, X-ray single-crystal analysis of Hyp-Phe-Phe revealed unprecedented similarity to Pro-Phe-Phe in both its backbone conformation and its supramolecular organization (Supplementary Fig. 8 and Supplementary Table 1). Supplementary Fig. 8a represents the asymmetric unit of Hyp-Phe-Phe in 50% probability displacement ellipsoids. The formation of a super-helical architecture was also found to be similar to Pro-Phe-Phe (Supplementary Figs. 8 and 9). The additional hydrogen-bonding site was utilized though hydrogen bond formation between adjacent strands, suggesting that helical-like sheets of Hyp-Phe-Phe are more tightly packed than those of Pro-Phe-Phe (Fig. 4a,b). This finding clearly indicates that the helical-like conformation and supramolecular helical packing of the Pro-Phe-Phe tripeptide are indeed characteristics of the peptide backbone, which are not affected by side-chain modifications.

Fig. 4. Strength modulation of helical-like architecture via side-chain modification.

a,b, Comparison of the inter-strand hydrogen bonding in the crystal packing and Young’s modulus of Pro-Phe-Phe (a) and Hyp-Phe-Phe (b). Carbon atoms of two adjacent helices are shown in green and magenta. Inter-strand hydrogen bonds are shown in orange. c, Comparison of Young’s moduli of different biological and non-biological materials, with Phe-Phe, Pro-Phe-Phe and Hyp-Phe-Phe marked in brown.

The unique super-helical organization of both Pro-Phe-Phe and Hyp-Phe-Phe, and specifically the further reinforcement of the inter-sheet interaction by the additional hydrogen bond in the latter tripeptide, inspired us to study the macroscopic manifestation of such a packing module. As mechanical properties are directly correlated with higher-order packing of biomolecules²⁹, we analysed both tripeptides using an indentation-type AFM experiment (see Methods). The Young’s modulus of the Pro-Phe-Phe assemblies was found to be about 44 GPa, indicating a remarkable stiffness of thesuper-helical conformation (Fig. 4a). The self-assembled fibres also exhibited a similar rigidity, displaying a comparable Young’s modulus value (Supplementary Fig. 10). In comparison, the nanotubes of the Phe-Phe dipeptide, the minimal core recognition motif of amyloid-β and one of the shortest and stiffest β-sheet-based protein-aceous structures, showed a Young’s modulus of 19 GPa (Fig. 4c)³⁰. Moreover, incorporation of the hydroxyl group in the side chain to generate Hyp-Phe-Phe significantly increased the stiffness due to stabilization via additional hydrogen bonds along the long axis, resulting in a Young’s modulus of 102 GPa (Fig. 4b), comparable to the mechanically rigid collagen matrix (Fig. 4c)²⁹. Taken together, these experiments demonstrate that the super-helical organization of the tripeptides affords one of the stiffest biological materials. Importantly, even in such a short sequence, it was possible to modulate the robustness of the higher-order assemblies in a predictable manner through side-chain modification.

Sequence modification of Pro-Phe-Phe

The generic nature of the helical-like conformation of the Pro-Phe-Phe tripeptide backbone was further investigated by sequentially mutating the terminal residues with an intrinsic helix-stabilizing amino acid, Ala, while conserving the central Phe, which adopted dihedral angles confined in a helical conformation. The conformations of the modified sequences, Ala-Phe-Phe and Ala-Phe-Ala, were analysed in atomic detail by single-crystal X-ray crystallography, which revealed that both peptides organized into a β-sheet-rich structure (Supplementary Note 1). In addition, sequence shuffling of the native Pro-Phe-Phe into the other two possible combinations, namely Phe-Pro-Phe and Phe-Phe-Pro, also completely altered the secondary conformation to a β-sheet-rich structure (Supplementary Note 2). This comprehensive analysis of the tripeptide library further confirmed that the helical-like conformation of Pro-Phe-Phe is truly a nature of the peptide backbone (Supplementary Table 2). Moreover, all four modified peptides, which are not among the top 400 high-aggregation-prone tripeptides⁹, show β-sheet-rich organization, clearly demonstrating that the helical arrangement is a fundamental secondary structure required for high aggregation of short peptides.

Outlook

An atomic-level understanding of supramolecular peptide assembly is critical, not only for the design of functional biomaterials, but also for uncovering the role of the core recognition motifs in physiological self-organizing systems. The short interdigitated dry steric zipper represents a fundamental structural motif conferring exceptional stability to β-sheet-rich amyloids¹³. The super-helical domain, stabilized by the dry steric zipper, is common to the packing of both the minimal Pro-Phe-Phe tripeptide and the much longer PSMα3 pep-tide. Thus, the organization into helical structural elements emerges as an additional major paradigm for peptide aggregation, analogous to the β-sheet-rich architectural design. Furthermore, unlike many β-sheet-based modules, the minimal helical-like motif described here allows the rationally designed modulation of the structural features at the molecular level, resulting in higher-order organization with predictable sequence-structure relationship. Thus, this minimal motif can be manipulated to display unique biophysical characteristics in the bulk state. Finally, the single-crystal X-ray analysis presented here could pave the way for the future design of modular super-helical self-assembling nanostructures by incorporating Pro-Phe-Phe or its variants into peptide sequences.

Online content

Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41563-019-0343-2.

Methods

Preparation of peptide assemblies. The peptides were synthesized by DGpeptidesCo. by conventional Fmoc solid-phase synthesis then purified to more than 95%, as verified by HPLC followed by mass spectrometry confirmation of their identity (Supplementary Figs. 15-20). All peptides were stored at −20 °C. For assembly, peptides in the required concentration were dissolved in phosphate buffer at pH 7.4 by vigorous vortexing for 2 min. The peptide solutions were then incubated at 18 °C for two weeks with frequent shaking before examination.

SEM. The peptides were dissolved in phosphate buffer at pH 7.4 at a concentration of 5 mg ml⁻¹. The samples were incubated at 18 °C for two weeks with frequent shaking. A 5 μl aliquot was allowed to dry on a microscope glass coverslip under ambient conditions overnight and coated with Au. SEM images were recorded using a JSM-6700F FE-SEM (JEOL) operating at 10 kV.

AFM. The peptides were dissolved in phosphate buffer at pH 7.4 at a concentration of 5 mg ml⁻¹. The samples were incubated at 18 °C for two weeks with frequent shaking. AFM images were taken by depositing 5 μl solutions onto freshly cleaved V1 grade mica (Ted Pella). The samples were allowed to dry under ambient conditions overnight. Images were obtained with an AIST-NT Smart AFM system in non-contact (tapping) mode using 100-mm-long silicon nitride cantilevers (OMCL-RC800PSA-W, Olympus) with resonance frequency of 70 kHz. The images were analysed and visualized using WSxM imaging software³¹ (Nanotec Electronica S.L).

ThT binding assay. For ThT binding analysis, 5 μl of 20 mM peptide solution was dropcast over a microscope glass slide. Then, 5 μl of 20 μm ThT was immediately added and the sample was covered with a coverslip. The stained samples were visualized using an inverted LSM 510 META confocal laser scanning microscope (Carl Zeiss Jena) at excitation and emission wavelengths of 458 nm and 486-593 nm, respectively. The fluorescence images were edited using Carl Zeiss AIM software. The ThT-bound peptide fibres were represented using an FITC fluorescence filter cube in green. Peptide fibres without ThT under similar experimental conditions did not show any significant fluorescence. Peptide fibres after completion of ThT fluorescence kinetics also showed similar bright-field emission.

ThT fluorescence kinetic assay. The peptides were dissolved in phosphate buffer at pH 7.4 at a final concentration of 20 mM. The samples were incubated at 18 °C for two weeks with frequent shaking. An aged peptide sample was then added to 40 μM ThT in water to a final concentration of 20 μM ThT. Data were collected in a Greiner bio-one black 96-well flat-bottomed plate and immediately covered with a silicone sealing film (ThermalSeal RTS). The plate was incubated in a plate reader (CLARIOstar, BAAG LABTECH) at 37 °C, with 5 s shaking before each cycle, and data were collected for 250 cycles with 5 min intervals between each cycle. On excitation at 438 nm, ThT fluorescence emission spectra at 485 nm were recorded over time. Measurements were performed in triplicate. All triplicate values were averaged and plotted against time, and the standard error of means was represented as error bars. Phosphate buffer without any peptide was used as a control.

FTIR spectroscopy. The peptides were dissolved in phosphate buffer at pH 7.4 to a final concentration of 5 mg ml⁻¹. The samples were incubated at 18 °C for two weeks with frequent shaking. A 30 μl aliquot of the peptide solution was deposited on disposable KBr infrared sample cards (Sigma-Aldrich), then allowed to dry under vacuum. The samples were saturated twice with 30 μl of D₂O and vacuum dried. FTIR spectra were collected using a nitrogen-purged Nicolet Nexus 470 FTIR spectrometer (Nicolet) equipped with a deuterated triglycine sulfate detector. Measurements were performed using a 4 cm⁻¹ resolution and by averaging 64 scans. The absorbance maxima values were determined using an OMNIC analysis program (Nicolet). The background was subtracted using a control spectrum.

CD spectroscopy. The peptides were dissolved in phosphate buffer at pH 7.4 at a concentration of 5 mg ml⁻¹. The samples were incubated at 18 °C for two weeks with frequent shaking, and the experiments were performed without further dilution. CD spectra were collected using a Chirascan spectrometer (Applied Photophysics) fitted with a Peltier temperature controller set to 25 °C, using quartz cuvettes with an optical path length of 0.1 mm (Hellma Analytics). Absorbance of the sample was kept within the linear range of the instrument during measurements. Data acquisition was performed in steps of 1 nm in a wavelength range of 190-240 nm with a spectral bandwidth of 1.0 nm and an averaging time of 3 s. The spectrum of each sample was collected three times and averaged. The baseline was similarly recorded for phosphate buffer and subtracted from the sample spectra. Data processing was performed using Pro-Data Viewer software (Applied Photophysics).

Powder X-ray diffraction. The lyophilized powder of the peptide was dissolved in double distilled water and allowed to self-assemble by incubating at 18 °C for four weeks. The sample was then centrifuged for 10 min at 2,900g and the solution was decanted to remove non-assembled peptide molecules. The assembled fibres were lyophilized and poured inside a glass capillary (0.5 mm diameter). X-ray diffraction data were collected using a Bruker D8 Discover theta/theta diffractometer with a liquid-nitrogen-cooled intrinsic Ge solid-state linear position detector. The cell parameters were determined using the GSAS-II software³². Due to lower peak intensities, the Rietveld refinement of the powder diffraction patterns did not result in an adequate fit. The diffraction patterns were therefore analysed using the whole profile fitting (Pawley) method with a final Rwp=1.1% (ref.³³).

Crystal preparation and data collection. Crystals used for data collection were grown using the vapour diffusion method. The dry peptide was first dissolved in water at a concentration of 5 mg ml⁻¹. Then, 50 μl was deposited into a series of 8 × 40 mm vessels. Each tube was sealed with Parafilm, into which a single small hole was pricked using a needle. The samples were placed inside a larger vessel filled with 2 ml of acetonitrile. The systems were ultimately capped and incubated at 18 °C for several days. Crystals grew within 7-8 days. For data collection, crystals were coated in paratone oil (Hampton Research), mounted on a MiTeGen cryo-loop and flash-frozen in liquid nitrogen. Diffraction data were collected at 100 K on a Rigaku XtaLabPro with a Dectris 200 K detector using CuKα radiation at λ=1.54184 Å.

Processing and structural refinement of crystal data. The diffraction data were processed using CrysAlisPro 1.171.39.22a. Structures were solved by direct methods in SHELXT-2014/5³⁴. The refinements were performed with SHELXL-2016/4 and weighted full-matrix least-squares against |F²| using all data. Atoms were refined independently and anisotropically, with the exception of hydrogen atoms, which were placed in calculated positions and refined in a riding mode. Crystal data collection and refinement parameters are shown in Supplementary Tables 3 and 4 and the complete data can be found in the cif files appended as Supplementary Data 1-4. The crystallographic data have been deposited in the Cambridge Crystallographic Data Centre (CCDC) under nos. 1565666, 1823367, 1862583 and 1834550 for Pro-Phe-Phe, Hyp-Phe-Phe, Ala-Phe-Phe and Ala-Phe-Ala, respectively.

Young’s modulus. AFM experiments were carried out using a commercial AFM (JPK, Nanowizard II). The force curves were obtained using the commercial software from JPK and analysed by a custom-written procedure based on Igor pro 6.12 (Wavemetrics). Silica cantilevers (SSS-SeIHR-50, Nanosensor Company: half-open angle of the pyramidal face, θ<10°; tip radius, 2-10 nm; frequency in air, ∼96-175 kHz) were used in all experiments. The spring constant of the cantilevers was in the range of ∼7-35 N m⁻¹. The maximum loading force was set to 368 nN for the Pro-Phe-Phe and Hyp-Phe-Phe crystals. All AFM nano-indentation experiments were carried out at room temperature. In a typical experiment, the peptide crystals were spread over the surface of a freshly cleaved mica substrate. Then, the cantilever was moved over a crystal with the help of an optical microscope at a constant speed of 15 μm s⁻¹ and held on the crystal surface at a constant force of 368 nN. The cantilever was retracted and moved to another spot for the next cycle. The indentation fit was performed using a custom-written Igor program and manually checked after fitting was complete. Each approaching force-deformation curve in the range of 20 nm was fitted, or from the contact point to the maximum indentation depth if the maximum indentation depth was less than 20 nm. By fitting the approaching curve to the Hertz model (1), the Young’s modulus of the crystals was obtained. Typically, four to five such regions (3 μm × 3 μm, 600 pixels) were randomly selected for each sample to construct the elasticity histogram:

$$F\left(h\right)=\frac{2}{\pi }\tan \alpha \frac{E}{1-{\nu }^2}{h}^2$$ (1)

where F is the force acting on the cantilever, h is the indentation depth of the crystal by the cantilever tip, α is the half angle of the tip, E is the Young’s modulus of the sample and v is the Poisson ratio. v=0.3 was used in the calculation.

Supplementary Material

Supplementary information is available for this paper at https://doi.org/10.1038/s41563-019-0343-2.

Reporting Summary.

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

Crystal data for Pro-Phe-Phe, Hyp-Phe-Phe, Ala-Phe-Phe and Ala-Phe-Ala are available from the Cambridge Crystallographic Data Centre (CCDC) under reference nos. 1565666, 1823367, 1862583 and 1834550, respectively (https://www.ccdc.cam.ac.uk/structures/). The remaining data supporting the findings of this study are within the Article and its Supplementary Information files and are available from the corresponding author upon reasonable request.