Protein nanofibril design via manipulation of hydrogen bonds

The paper addresses how molecular interactions, particularly hydrogen bonds and aromatic side-chain interactions, shape the assembly pathway and material properties of amyloid protein fibrils. Amyloid self-assembly underpins both pathological aggregates linked to neurodegenerative diseases and functional materials like silk. Despite diverse sequences, amyloids share β-strands perpendicular to the fibril axis connected by dense hydrogen-bonded networks, with hydrophobic and aromatic residues influencing fibrillation kinetics and organization. The research question is whether controlled intervention in these interactions—especially the H-bonded network—can modify amyloidogenic assembly and its physical properties without abolishing fibril formation. The study focuses on introducing aromatic residues (F, Y, W) into a glycine-rich, amyloidogenic peptide motif to impose steric constraints and alternative H-bond donors, thereby probing effects on nucleation, growth, morphology, structure, and nanomechanics.

Prior work shows amyloid fibrils, whether pathological or functional, feature universal backbone-driven H-bonded β-sheets, with hydrophobic and π–π interactions often accelerating assembly. Sequences rich in alanine/glycine motifs promote amyloid formation and are implicated in diseases (e.g., Huntington’s, prion-related, Aβ, α-synuclein) as well as functional amyloids (silk fibroin/spidroin, Pmel17). Aromatic interactions can nucleate fibrillation, as in Aβ (KLVFF core) where diphenylalanine interactions initiate aggregation; conversely, FF dipeptide assembly can start via H-bonds followed by π–π stacking. Mechanical properties vary substantially across amyloids, e.g., Aβ fibrils exhibit 2–5 GPa modulus, whereas FF assemblies reach 20–30 GPa. Viral capsid assembly illustrates robust, directed noncovalent assembly under varied conditions. Solvent effects, including the use of DMSO for hydrophobic peptides, have been widely discussed: DMSO can stabilize monomeric or misfolded states yet is reported not to change end-point amyloid fibril structures; its viscosity may better model crowded cellular microenvironments for hydrophobic biomolecules. These foundations motivate testing how backbone H-bond networks and aromatic side-chain placement modulate assembly pathways and mechanics.

Design: A glycine/alanine-rich amyloidogenic peptide core (GAGAGSGAGAGSGAGAGSGAG, P1) was modified by substituting glycine residues with aromatic amino acids phenylalanine (F), tyrosine (Y), or tryptophan (W) at low (~10%, two residues, typically termini) or higher (~30%, six residues) fractions to generate P1–P7: P1 GAGAGSGAGAGSGAGAGSGAG; P2 FAGAGSGAGAGSGAGAGSGAF (10% F); P3 FAGAFSGAGAFSGAFAFSGAF (30% F); P4 YAGAGSGAGAGSGAGAGSGAY (10% Y); P5 YAGAYSGAGAYSGAYAYSGAY (30% Y); P6 WAGAGSGAGAGSGAGAGSGAW (10% W); P7 WAGAWSGAGAWSGAWAWSGAW (30% W). Structural propensity of monomers was predicted using PEP-FOLD3. Sample preparation: Lyophilized peptides dissolved in DMSO, aggregated at 65 °C for 11 days. Characterization: - Morphology: AFM (JPK NanoWizard 4; tapping mode; AC240/AC160 cantilevers) and TEM (not detailed in Methods but used). AFM sample prep on mica; images processed with JPK software. - Nanomechanics: AFM PeakForce QNM on Bruker Multimode; AC160/RTESP probes (40–50 N/m); HOPG (18 GPa) for calibration; deformation ~1 nm; DMT model used to compute elastic modulus. - Secondary structure: FT-IR (Nicolet 6700), 400–4000 cm−1, 32 scans/sample, Happ-Genzel apodization; DMSO spectra subtracted; analysis via second derivative and peak areas (Origin). - H-bond network: XPS (Kratos AXIS-Ultra DLD, Al Kα, 15–75 W, pass energy 20–80 eV; chamber <1×10−9 torr). Charging corrected by referencing C 1s to 285.0 eV; repeated scans and flood gun adjustments minimized differential charging; low-power work-function measurements aided tracking surface potential; beam-damage assessed and minimized. Deconvolution focused on C 1s components (CC/CH, Cα, Cam, CH) and N 1s (NH not H-bonded vs H-bonded) to quantify H-bond signatures and binding-energy shifts ΔEB(N). - Assembly kinetics: Thioflavin T (ThT) assay. Peptides dissolved in DMSO, sonicated at 37 °C for 10 min; centrifuged at 14,000 rpm for 10 min. Kinetics at 65 °C in 96-well plate: 0.3 μM peptide + 20 μM ThT; monitored on Clariostar reader (excitation 440 nm, emission 490 nm) until saturation. - Electron diffraction: Performed on peptide assemblies to assess cross-β spacings. Stability in water: Titration of DDW into DMSO solutions and storage of powders at ~50–60% RH; peptides precipitated with <1% water; powders eventually fibrillated after 4–12 months under humidity.

- Amyloidogenic propensity retained: Tailored changes in the H-bonded network via aromatic substitutions did not abolish amyloid fibrillation and had minimal effect on nucleation at low substitution levels (10%). - Morphology: P1 formed elongated fibrils (lengths 100 nm to 1.5 μm). P2 (10% F, termini) produced relatively longer micron-scale fibrils; P4 (10% Y) formed shorter fibrils (100–500 nm). P6 (10% W) produced coexisting spherical and short fibrillar morphologies (50–150 nm). Increasing aromatic content to 30% altered morphologies: P3 (30% F) yielded bead-like fibril structures and abundant spheres (20–80 nm diameter; spheres 40–60 nm accounted for >50% population). In contrast, P5 (30% Y) and P7 (30% W) formed continuous fibrils; P5 mostly micron-length fibrils; P7 showed an approximately even distribution among 100–500 nm, 500–900 nm, and >900 nm fibrils. - Kinetics: 10% substitution peptides (P2, P4, P6) nucleated within ~36–38 h, similar to P1. Higher aromatic content increased nucleation times: P3 ~41 h, P5 longer (not specified numerically), and P7 the longest at 181 h. Elongation rates varied: P5 (30% Y) showed the highest elongation rate; P4 (10% Y) slower than P5; F-containing P2 and P3 also exhibited increased elongation rates versus P1, P6, P7. Interpretation: Aromatic side chains introduce steric constraints that impede primary nucleation (longer lag at higher content) but, once nuclei form, promote directional elongation via facilitated H-bond alignment; solvation, sterics, and H-bonding capacity together govern kinetics. - Secondary structure (FT-IR): All assemblies showed antiparallel β-sheet features (amide I peaks at ~1623 and 1695 cm−1). P1 also displayed bands at 1730 cm−1 (non-H-bonded C=O), 1546 cm−1 (amide II, non-H-bonded NH), and amide A <3300 cm−1 with ~3450 cm−1 shoulder (non-bonded OH), indicating incomplete participation of groups in H-bonds. F substitution (P2, P3) induced a shoulder at 1644–1682 cm−1 (increased random coil/α-helix), disappearance of 1546 and 1730 cm−1 bands, emergence of 1518 cm−1 (amide II, β-turn), and a small 3055 cm−1 band (π–π stacking). Y substitution (P4, P5) promoted β-sheet formation via OH H-donor while preserving random coil/α-helix; 1730 cm−1 band reappeared and ~3450 cm−1 shoulder disappeared, correlating with long persistent fibrils. W substitution (P6, P7) showed aggregative β-sheet with disordered content (shoulder 1640–1685 cm−1). - H-bond network (XPS): Deconvolution identified CH (amide carbonyl involved in H-bond) and NH contributions. 10% substitutions decreased NH/Ntot but did not necessarily reduce CH/(CH+Cam), implying steric hindrance and/or alternative donors (Y: OH; W: NH indole). P2 (10% F) decreased NH/Ntot and slightly lowered ΔEB(N), suggesting weaker backbone H-bonds. P4 (10% Y) showed up to twofold reduction in NH/Ntot, increased CH/Cam, and increased ΔEB(N), indicating fewer but stronger backbone H-bonds while OH competes as an H-donor. P6 (10% W) decreased NH/Ntot and increased CH/Cam, consistent with additional H-bonding via indole NH. At 30% Y/W (P5, P7), fewer yet stronger backbone H-bonds formed (increased ΔEB(N)). - Electron diffraction: All examined assemblies (P1, P2, P3, P6, P7) exhibited cross-β characteristics. Interstrand distances increased for F and Y substitutions (P1 ~4.7 Å to P2/P3 ~4.9 Å; P4/P5 ~4.8 Å) and decreased for W substitutions (P6/P7 ~4.6 Å). Intersheet distances remained ~10.8–10.9 Å across samples, indicating localization of hydrophobic residues at fibril interfaces. - Nanomechanics: Fibrillar assemblies of P3 and P7 exhibited elevated elastic moduli (26 GPa and 29 GPa, respectively), while corresponding spherical assemblies showed <5 GPa. Continuous H-bonded networks and aromatic content contributed to higher stiffness; morphology strongly influenced modulus. - Solvent effects: F and especially W side-groups display attractive interactions with DMSO, stabilizing spherical intermediates and increasing incubation times (e.g., P3, P6), whereas Y showed no noted affinity to DMSO. Peptides were unstable in aqueous environments: <1% water induced precipitation; powders under 50–60% humidity eventually formed fibrils within 4–12 months.

The study demonstrates that deliberate manipulation of backbone hydrogen-bond networks via strategic aromatic substitutions does not prevent amyloid formation but tunes higher-order assembly and material properties. Low-level substitutions minimally affect nucleation times, while higher aromatic content introduces steric hindrance that lengthens nucleation but can enhance elongation directionality once nuclei form. The side groups also modulate secondary structure distributions and provide alternative H-bond donors (Y OH, W indole NH), yielding fewer yet stronger backbone H-bonds at higher Y/W content. Consistent cross-β spacings with altered interstrand distances indicate that hydrophobic aromatics localize at fibril interfaces, influencing 2D packing and 3D assembly pathways. Nanomechanical measurements link continuous β-sheet H-bond networks and aromatic content to increased stiffness in fibrils, while spherical morphologies are softer. Overall, three competing factors determine outcomes: steric constraints from side groups (affecting nucleation and packing), H-bond formation pathways mediated by side groups, and side-group affinity for solvent (competing with fibrillar growth). These insights address the central question by showing how controlled molecular-level interventions sculpt fibril morphology, structure, and mechanics, informing both disease-related aggregation understanding and biomaterial design.

Controlled intervention in molecular interactions—implemented by substituting glycine residues with aromatic amino acids in an amyloidogenic peptide—modulates hydrogen-bond networks and reshapes fibrillation pathways without abolishing amyloid formation. The approach yields distinct morphologies, secondary structures, interstrand spacings, and notably altered nanomechanical properties, with certain designs (e.g., 30% F or W) producing stiff fibrils (up to ~29 GPa). The results highlight the interplay of steric constraints, H-bonding capabilities of side groups, and solvent interactions in determining assembly kinetics and outcomes. These findings provide a framework for rationally tuning protein supramolecular constructs and pave the way for designing amyloid-based materials with bespoke mechanical properties, while offering deeper understanding of factors governing pathological versus functional amyloid formation.

- Solvent environment: Experiments were conducted in DMSO to solubilize highly hydrophobic peptides; while argued not to alter end-point fibril structures, DMSO can affect intermediate secondary structures and assembly kinetics, limiting direct extrapolation to physiological aqueous conditions. - Aqueous instability: Peptides precipitated with <1% water and only formed fibrils after long-term humidity exposure as powders, indicating behavior may differ in water-rich environments. - XPS artifacts: Quantitative H-bond assessment via XPS is sensitive to differential charging and potential beam-induced damage; although minimized via methodology, residual uncertainties remain. - Scope: Study focuses on a specific glycine/alanine-rich motif and selected substitution patterns (10% and 30% F/Y/W); generalizability to other sequences and substitution distributions requires further validation. - TEM details and in vivo relevance were not addressed; mechanical measurements rely on AFM DMT fitting with assumptions of contact mechanics and calibration standards.