Non-natural 3-Arylmorpholino-β-amino Acid as a PPII Helix Inducer

A new non-natural β-amino acid, named 3-Ar-β-Morph, was designed and synthesized via a regio- and diastereoselective Pd-catalyzed C(sp3)H-arylation of the corresponding 2S,6S-(6-methoxymorpholin-2-yl)carboxylic acid, readily available from glucose. According to the computational prevision and confirmed by IR and NMR data, the insertion of 3-Ar-β-Morph in a model foldamer represents a way to stabilize a PPII-like helix through the presence of two γ-turns, secondary structure motifs induced by the morpholine ring, and the trans-tertiary amide bond.


Simulation of (3S)-4 in a biological complex
We investigated on the possibility for peptide (3S)-4 to mimic a PPII helix within a biological complex. As a reference, we chosen the structure of the complex between human platelet profilin (HPP) and a poly-L-proline decamer (L-Pro10), 2 for which a crystal structure is available (1AWI.pdb).
The complex is formed by two molecules of HPP bound to L-Pro10, where this latter adopts a PPIIhelix. We performed MD simulations (100 ns) of both the HPP:L-Pro10 and HPP:(3S)-4 complexes, the latter obtained by a protein-protein docking approach. The binding energy of both L-Pro10 and (3S)-4 was then computed using the Nwat-MMGBSA method. 3 Results confirmed that peptide (3S)-4 can actually behave as a PPII mimic in a biological complex. Figure S3 shows the geometry of the complex between HPP and peptide (3S)-4, as obtained by a cluster analysis of the last 50 ns of MD trajectory. Binding energies for both L-Pro10 and (3S)-4 are reported in Table TS2.  -64.5 ± 4.9 -91.2 ± 9.6 -48.9 ± 6.5 -80.3 ± 10.5 a Nwat-MMGBSA binding energies were computed considering no explicit waters (Nwat=0) or including 30 explicit waters (Nwat=30). In this case, selected waters are the closest to the ligand in each frame of the MD trajectory and are considered part of the receptor, according to the Nwat-MMGBSA method. In both cases, entropy was neglected and the computed values should be considered as a "score" rather than an absolute binding free energy.

Computational methods
Parameterization of 3-Ar-β-Morph. Charge parameterization for β-Morph was performed using the R.E.D.IV tools. 4 The amino acid structure was capped by acetyl and a NHMe group at the N and C termini, respectively, and subjected to a conformational search using the low mode method, the AMBER10EHT force field and the Born solvation model implemented in MOE. 5 The two conformations corresponding to the E and Z configurations at the peptide bond linking the acetyl cap to the residue were used for charge parameterization. For each conformation, two orientations were used to derive conformation and orientation independent RESP charges. Gaussian09 6 was used to perform quantum mechanical calculations at the HF/6-31G* level, accordingly to the force field specifications. All the molecular dynamics simulations were conducted with the Amber18 and AmberTools18 packages, 7 using the ff14SB forcefield. 8 Parameters for the peptide bond rotation were modified as suggested by Doshi and Hemelberg. 9 Hamiltonian replica exchange molecular dynamics. H-REMD simulations were conducted starting from the extended configuration of (3R)-and (3S)-4. Twelve different Hamiltonians were generated by progressively lowering the torsional potential of the φ, ψ and ω dihedrals, starting from the default values. All the 12 replicas were subjected to a geometry minimization (1000 cycles of steepest descent and 1000 cycles of conjugated gradient, up to a gradient of 0.1 kcal/mol·Å), followed by constant volume (NVT) equilibration (5 ns, 300 K, Langevin thermostat with a collision frequency = 2.0 ps -1 , electrostatic cutoff = 8.0 Å, PME, SHAKE to constrain bonds involving hydrogens). A production run of 1.5 µs was then conducted under the same conditions. Simulations were conducted on a cluster of GPU-equipped nodes using the pmemd.cuda.MPI executable 10,11 of the Amber18 package. 7 Trajectory analyses were conducted on the final 500 ns of the unmodified replica, using cpptraj and cpptraj.cuda. Cluster analyses were done requesting 10 clusters, using the average-linkage algorithm and the pairwise mass-weighted root mean squared deviation (RMSD) on the Cα as the metric.
Convergence was evaluated by doing a cluster analysis every 500 ns and comparing results in terms of population of the most populated clusters and RMSD between the main cluster representative conformations. All the simulations resulted converged between 1000 and 1500 ns of simulation time.
Simulation of HPP complexes. The HPP models containing L-Pro10 and (3S)-4 were prepared starting from the 1AWI.pdb file. HPP:L-Pro10 model: the complex was processed with the Structure Preparation module of the software MOE and protonated at pH=7 using the Protonate3D tool. Waters were removed and the system was minimized up to a gradient of 0.1 kcal/mol·Å, using the Amber10:EHT force field and the Born solvation model for water, and keeping the backbone atoms restrained. The model was used for MD simulations as described below. HPP:(3S)-4 model: starting from the HPP:L-Pro10 model, the L-Pro10 chain was removed and the system was subjected to a backbone-restrained minimization, as described above. The system was used as the receptor in a docking experiment using the Protein-Protein docking algorithm implemented in MOE. The representative geometry of the most populated cluster for (3S)-4, obtained by H-REMD simulations, was used as the ligand to be docked. The top-scored pose showed a decent superposition between (3S)-4 and the L-Pro10 chain as found in the crystal structure. Thus, the HPP:(3S)-4 complex was used for the next steps.
The HPP:L-Pro10 and HPP:(3S)-4 models described above were used as the starting point for MD simulations. The systems were neutralized by adding 4 Cl-atoms and solvated by an octahedral box of TIP3P water, extending up to 10 Å from the solute. The system was equilibrated as described in previous works, 12 and subjected to 100 ns of MD simulations using pmemd.cuda of Amber18. The resulting trajectory was analyzed by computing time dependent RMSDs to verify that systems were stable during the dynamics. Then, the last 50 ns were subjected to a cluster analysis, using the same

Synthesis of compounds 2-4,7-9,11,12
General information. Chemicals were purchased from Sigma Aldrich and were used without further purification. Mass spectra were recorded on an LCQESI MS and on a LCQ Advantage spectrometer from Thermo Finningan and a LCQ Fleet spectrometer from Thermo Scientific. The NMR spectroscopic experiments were carried out either on Varian MERCURY 300 MHz (300 and 75 MHz for 1 H and 13 C, respectively), or Bruker Avance I 500 MHz spectrometers (500 and 125 MHz for 1 H and 13 C, respectively). Optical rotations were measured on a Perkin-Elmer 343 polarimeter at 20 °C (concentration in g/100 mL). Chemical shifts (δ) are given in ppm relative to the CHCl3 internal standard, and the coupling constants J are reported in Hertz (Hz). The synthesis of dipeptide 10 1 and compound 1 1 are reported in the literature.