Nanoscale Assembly of Functional Peptides with Divergent Programming Elements

Self-assembling peptides are being applied both in the biomedical area and as building blocks in nanotechnology. Their applications are closely linked to their modes of self-assembly, which determine the functional nanostructures that they form. This work brings together two structural elements that direct nanoscale self-association in divergent directions: proline as a β-breaker and the β-structure-associated diphenylalanine motif, into a single tripeptide sequence. Amino acid chirality was found to resolve the tension inherent to these conflicting self-assembly instructions. Stereoconfiguration determined the ability of each of the eight possible Pro-Phe-Phe stereoisomers to self-associate into diverse nanostructures, including nanoparticles, nanotapes, or fibrils, which yielded hydrogels with gel-to-sol transition at a physiologically relevant temperature. Three single-crystal structures and all-atom molecular dynamics simulations elucidated the ability of each peptide to establish key interactions to form long-range assemblies (i,e., stacks leading to gelling fibrils), medium-range assemblies (i.e., stacks yielding nanotapes), or short-range assemblies (i.e., dimers or trimers that further associated into nanoparticles). Importantly, diphenylalanine is known to serve as a binding site for pathological amyloids, potentially allowing these heterochiral systems to influence the fibrillization of other biologically relevant peptides. To probe this hypothesis, all eight Pro-Phe-Phe stereoisomers were tested in vitro on the Alzheimer’s disease-associated Aβ(1–42) peptide. Indeed, one nonfibril-forming stereoisomer effectively inhibited Aβ fibrillization through multivalent binding between diphenylalanine motifs. This work thus defined heterochirality as a useful feature to strategically develop future therapeutics to interfere with pathological processes, with the additional value of resistance to protease-mediated degradation and biocompatibility.


S1. Spectroscopic data (NMR and ESI-MS
Chart S1. Fibril size distribution for 2a (counts = 150) with mean diameter of individual (n=1) fibril corresponding to 1.6 ± 0.2 nm, two fibrils (n=2) corresponding to 3.2 ± 0.3 nm, and three fibrils (n=3) corresponding to 4.6 ± 0.4 nm.  Chart S2. Fibril size distribution for 3a (counts = 25, due to rare instances).   Generation of tripeptide and Aβ structural models. Models of zwitterionic tripeptides were built upon the experimental structure (2a) or generated using the AmberTools19 package 1 and the VMD1.9.3 software 2 through in-house tcl scripts (3a). The initial coordinates of the protein were taken from the NMR structure with PDB ID 2NAO, 3 namely from the chain A of the corresponding pdb file.

MD simulations.
Multi-copy MD simulations of the self-assembly process for 216 2a and 3a tripeptides were performed as described previously. 4 To further investigate the molecular determinants of 2a fibrillization, an additional set of simulations was performed with 1000 such peptides. The initial structures were generated by placing the center of mass of the peptide repeatedly on the 15 Å-spaced points of a 10x10x10 (2a) or  6x6x6 (2a, 3a) grid. Initial orientations of peptides were randomized, and the systems were solvated with water molecules.
To investigate the molecular interactions occurring in solvent between 3a and Aβ, we performed a MD simulation of Aβ and 20 tripeptides in water solution. In this case, due to the net electrostatic charge carried by the protein, the system was neutralized by adding 13 K + and 10 Clions.
All the simulations were performed as follows. First, three consecutive restrained structural optimizations (up to 25,000 steps) were performed in the presence of harmonic restraints (k = 1 kcal mol -1 Å -1 ) applied to: a) all non-hydrogenous atoms of the system; b) backbone atoms; c) Cα atoms. Reference structures at steps b) and c) were the final ones from the previous step. Next, up to 50,000 cycles of unrestrained optimization were performed. Each system was then heated to 310 K in 1 ns via constant-pressuretemperature (NTP) MD simulations, followed by an equilibration phase of 10 ns. Starting from the equilibrated structure, multiple MD simulations were performed for each system (see Table S1). Pressure and temperature were set to 1 atm and 310 K (after the equilibration phase) using the isotropic Berendsen barostat 5 and the Langevin thermostat, 6 respectively. A time step of 2 fs was used for all the simulation steps but the production runs, where it was set to 4 fs after hydrogen mass repartitioning. 7 Periodic boundary conditions were employed, and electrostatic interactions were estimated using the Particle Mesh Ewald scheme with a cutoff of 9.0 Å for the short-range evaluation in direct space and for Lennard-Jones interactions (with a continuum model correction for energy and pressure).
The AMBER force fields parm14SB 8 was employed to model the Aβ peptide and the tripeptides, while the TIP3P 9 model was used describe water molecules, and the parameters for the ions were taken from ref. 10. The TIP3P (transferable intermolecular potential with 3 points) model describes a water molecule as three interaction sites corresponding to the oxygen and the two hydrogen atoms, linked by rigid bonds. Oxygen and hydrogens bear fraction charges amounting to -0.834 and 0.417, respectively. Coulomb interactions are present between all intermolecular pairs of charges, while Van der Waals interactions are described by a single Lennard-Jones term between oxygens.
Despite its simplicity, this model is still largely employed in the simulation of biomolecules, in view of its computational efficiency and of the reasonable thermodynamic and structural description provided for liquid water without the need for three-body corrections.  total of 90 images were collected. Reflections were indexed and integrated using the software Mosflm, 1 space group P21 was determined using POINTLESS. 2 The resulting data set was scaled using AIMLESS. 3 Phase information were obtained by direct methods using the software SHELX-T. 4 Refinements cycles were conducted with SHELXL-18, 5 operating through the WinGX GUI, 6 by full-matrix least-squares methods on F 2 .

S32
Congruence of structure model and calculated electron density was inspected using the software Coot. 7 The asymmetric unit contains a single molecule of the peptide. Hydrogen atoms were added at geometrically calculated positions and refined isotropically. All nonhydrogen atoms within the asymmetric unit have been refined with anisotropic thermal parameters. Unit cell parameters, scaling statistics, and refinement statistics are reported in Table S2.

Compound 3a (CCDC2026055)
A stick-shaped single crystal of the peptide was collected with a loop, cryoprotected by dipping the crystal in polyethylene glycol with average molecular weight 200 g/mol, PEG 200, and stored frozen in liquid nitrogen. The crystal was mounted on the diffractometer at the synchrotron Elettra, Trieste (Italy), beamline XRD1, using the robot present at the facility. Temperature was kept at 100 K by a stream of nitrogen on the crystal. Diffraction data were collected by the rotating crystal method using synchrotron radiation, wavelength 0.70 Å, rotation interval 1°/image, crystal-to-detector distance of 85 mm. A total of 180 images were collected. Reflections were indexed and integrated using the software Mosflm, 1 space group P21 was determined using POINTLESS. 2 The resulting data set was scaled using AIMLESS. 3 Phase information were obtained by direct methods using the software SIR 2014. 8 Refinements cycles were conducted with SHELXL-18, 5 operating through the WinGX GUI, 6 by full-matrix least-squares methods on F2.
Congruence of structure model and calculated electron density was inspected using the software Coot. 7 The asymmetric unit contains 10 crystallographically independent molecules of the peptide. In the residual electron density of the asymmetric unit, 12 water molecules were located, 7 of which in positions at full occupancy, 5 in positions statistically occupied in 50% of the unit cells. Restraints were applied to bond lengths and angles of the proline moiety of two crystallographically independent peptides, using the cards DFIX and DANG of the SHELXL-18 software, 5 in particular when a statistical occupancy of two close positions was observed. In addition, restrains were applied to keeps similar values for anisotropic displacement parameters for adjacent atoms in the direction of bonds. Hydrogen atoms of the peptide molecules were added at geometrically calculated positions and refined isotropically. When a clear electron density could be observed, hydrogen atoms of the water molecules at full occupancy were added considering the hydrogen bonding pattern, and refined with restrains on bond lengths and angles, using the cards DFIX and DANG of the SHELXL-18 software. 5 All the atoms, S33 except the hydrogen atoms, within the asymmetric unit have been refined with anisotropic thermal parameters. Unit cell parameters, scaling statistics, and refinement statistics are reported in Table S2.

Compound 4b (CCDC2021318)
A stick-shaped single crystal of the peptide was collected with a loop, cryoprotected by dipping the crystal in glycerol and stored frozen in liquid nitrogen. The crystal was mounted on the diffractometer at the synchrotron Elettra, Trieste (Italy), beamline XRD1, using the robot present at the facility. Temperature was kept at 100 K by a stream of nitrogen on the crystal. Diffraction data were collected by the rotating crystal method using synchrotron radiation, wavelength 0.70 Å, rotation interval 1°/image, crystal-todetector distance of 85 mm. A total of 180 images were collected. Reflections were indexed and integrated using the XDS package, 9 space group C2 was determined using POINTLESS, 2 and the resulting data set was scaled using AIMLESS. 3 Phase information were obtained by direct methods using the software SHELXS. 10 Refinement cycles were conducted with SHELXL-14, 5 operating through the WinGX GUI, 6 by full-matrix leastsquares methods on F2. Congruence of structure model and calculated electron density was inspected using the software Coot. 7 The asymmetric unit contains a single molecule of the peptide and a molecule of water in a special position, with the oxygen atom located along the 2-fold symmetry axis, resulting in an occupancy of 50% in the asymmetric unit.
During refinement, no restraints were applied on distances, angles or thermal parameters of the peptide or the water molecule. Hydrogen atoms of the peptide were added at geometrically calculated positions and refined isotropically, with thermal parameters dependent on those of the attached atom. All non-hydrogen atoms were refined with anisotropic thermal parameters. Unit cell parameters, scaling statistics, and refinement statistics are reported in Table S2.     Thioflavin T fluorescence assay