Navigating complex peptide structures using macrocycle conformational maps

Identification of turn motifs that are stabilized by intramolecular hydrogen bonds can be useful in describing the conformation of peptide systems. However, this approach is somewhat insufficient for cyclic peptides because peptide regions that are not positioned within a hydrogen bond can be left with no description. Furthermore, non-regular secondary structures and other rarely-observed conformations can be left without detailed evaluation. Herein, we describe “higher-order” ϕ/ψ plots termed macrocycle conformational maps (MCMs) as a tool for evaluating and comparing the conformations of a series of structurally related macrocyclic peptides.


S3
1 H, 13 C-gHSQC, and 1 H, 13 C-HMBC or 1 H, 13 C-CIGARAD NMR experiments were recorded and used to confirm the cyclic topology of each molecule. 1 H NMR spectra were referenced to DMSO-d6 (δ 2.50 ppm). 13 C NMR spectra were referenced to DMSO-d6 (δ 39.52 ppm). Mixing times for zTOCSY spectra were 80 ms, for ROESY spectra 150 ms. gHSQC spectra were recorded with a direct proton carbon coupling constant of 140 Hz, and gHMBC spectra with a long-range formic acid in HPLC-grade acetonitrile). Method A: A linear gradient starting from 5% of B to 95% over 4 min at a flow rate of 1.0 mL/min. Stays constant at 95% for 1 min and then returns to 5% over 0.5 min. Method B: Stays constant at 5% of B for 0 min at a flow rate of 1.5 mL/min, followed by a linear gradient to 95% over 4.0 min. Stays constant at 95% of B for 3.0 min and then returns to 5% B over 0.5 min. Method C: A linear gradient starting from 5% of B to 95% over S4 15 min at a flow rate of 1.0 mL/min. Stays constant at 95% for 1 min and then returns to 5% over 0.5 min.

Synthesis of Dominant Rotors and Macrocycles
The synthesis of Fmoc-protected dominant rotor building blocks was accomplished through condensation of a Fmoc-amino acid hydrazide and a 1,3-benzoxazine.
Fmoc-amino acid hydrazide synthesis: Under a nitrogen atmosphere, the desired Fmocamino acid (1.0 eq) was stirred in dry THF (0.1 M) and the mixture was cooled to -15 °C in an ice/NaCl bath. N-methyl morpholine (1.5 eq) and isobutyl chloroformate (1.0 eq) were added to solution, and the mixture was stirred for 5 minutes. Boc-hydrazide (1.1 eq) was added dropwise as a solution in dry THF. The mixture was allowed to stir for 2 hours and warm to rt. The precipitate was then filtered off and the solvent was removed by rotary evaporation. The resultant Bocprotected Fmoc-amino acid hydrazide was then purified by flash chromatography (1:1 EtOAc:hexanes). The resultant material was treated with 1:1 v/v TFA:DCM and stirred for 1 hour.
The mixture was concentrated in vacuo and the residue was extracted from NaHCO 3 /CHCl 3 three times. The Fmoc-amino acid hydrazide is generally sparingly soluble in CHCl3. The organic layers were mixed, dried over magnesium sulfate and the solvent was removed in vacuo.

1,3-benzoxazine synthesis:
Anthranilic acid or a structural analogue was dissolved in excess acetic anhydride and the mixture was refluxed for 8 hours.

Previously reported compounds
The following compounds were prepared according to literature procedures: The following compounds were previously reported by our lab: a, b, Ra-1, Sa-1, Ra-2, and Sa-2. 2

cyclo-[(Ra)-Trz(methyl/(S)-methyl)-Ala-Gly-Phe] (Ra-4)
Synthesized according to the procedure in 1. Hz. The output structures were ranked by energy. The structures were then checked for violations of the experimental distance and dihedral restraints. The lowest energy structure which satisfied these tests was passed for molecular dynamics study as the representative study. Solvent explicit molecular dynamics simulations were carried out with the Desmond Molecular Dynamics software module (D.E. Shaw, v4.4) running inside Maestro (Schrodinger LLC, v2015-2). The OPLS4 force S19 field was used for parameterization of the peptidomimetic macrocycle. The macrocycle representative structure was placed in an octahedral box solvent box (DMSO) with a minimum distance of 12 Å between solute atoms and the box boundary. The solvated box was minimized then brought to 300 K from 10 K using a restrained dynamics regime. Coulombic interactions were grouped into near-and far interactions with a near-interaction cutoff of 9 Å. Bonds were constrained with the SHAKE algorithm and an integration time step of 2 fs was used. The final MD production run was 100 ns in length with energy value recording every 1.2 ps and trajectory recording every 4.8 ps. The run trajectory was clustered using the Trajectory Clustering script within Maestro with a 0.4 Å RMSD cutoff for variation between backbone heavy atoms and a sampling frequency of 10%. The most populated cluster was taken as the "preferred" structure i.e. that conformation which the molecule spent most time in the dynamics run. This process was repeated for each potential diastereomer to assign the stereochemistry of the new sp 3 center in the linker. The distances from the most populated cluster for each diastereomer were then measured and compared with the initial NOE derived distances. The simulation distances for both diastereomers is included in the individual entry for each compound. The cluster that displayed the least violations was selected as the probable species isolated.

Van't Hoff analysis of dominant rotor macrocycles
The Van't Hoff analysis for Ra-1/Sa-1 and Ra-2/Sa-2 has been reported previously but the data and methods are also presented here for clarity. 2 This analysis was not carried out for single-welled system Sa-3 or Ra-4, which was isolated as a single atropdiastereomer amongst three other minor species.
General procedure for measuring equilibrium constants: A 1-dram oven-dried vial containing ~3.5 milligrams of macrocycle was dissolved in 0.25 mL of DMSO-d6 using a 1 mL syringe and transferred into a three-millimeter NMR tube. The samples were heated for 24 h to each specified using an oil bath controlled by an IKA programmable temperature probe. After 24 h, the sample was removed and immediately submerged in a dry ice/isopropanol bath for 1 minute. The sample was left to thaw to room temperature and a quantitative 1 H NMR experiment at 25 ºC was conducted on a Varian 500 MHz spectrometer (Xsen5mm probe with 1 H S/N of 1245.3) with the following parameters: number of scans (nt = 1), receiver gain = 30, and pulse width pw = 90. The equilibrium constants for each compound were measured using variable temperature NMR on a Bruker 600 MHz spectrometer. The sample was heated to each specified temperature for approximately 30 min followed by quantitative 1 H NMR analysis. The equilibrium constants were calculated from the relative integration of the Sa-to Ra-atropdiastereomer (see Fig below   exemplified for Ra-1/Sa-1).
Summary of the equilibrium constants (Keq) and standard Gibbs free energy differences for macrocycles determined at specific temperatures. Accuracy of temperature (± 3%) and 1 H NMR integration (± 4%) measurements were estimated and propagated through calculations. Uncertainty values in kcal/mol reported as ± 2*standard error.
Van't Hoff plot analysis: A graph of the natural logarithm of the equilibrium constants with respect to the inverse temperature (K -1 ) was plotted and fitted to the linear form of the van't Hoff equation (eq. 1). The change in enthalpy and entropy for the atropisomerization process was estimated from the slope and intercept of the van't Hoff plot.