Polarity effects in 4-fluoro- and 4-(trifluoromethyl)prolines

Fluorine-containing analogues of proline are valuable tools in engineering and NMR spectroscopic studies of peptides and proteins. Their use relies on the fundamental understanding of the interplay between the substituents and the main chain groups of the amino acid residue. This study aims to showcase the polarity-related effects that arise from the interaction between the functional groups in molecular models. Properties such as conformation, acid–base transition, and amide-bond isomerism were examined for diastereomeric 4-fluoroprolines, 4-(trifluoromethyl)prolines, and 1,1-difluoro-5-azaspiro[2.4]heptane-6-carboxylates. The preferred conformation on the proline ring originated from a preferential axial positioning for a single fluorine atom, and an equatorial positioning for a trifluoromethyl- or a difluoromethylene group. This orientation of the substituents explains the observed trends in the pKa values, lipophilicity, and the kinetics of the amide bond rotation. The study also provides a set of evidences that the transition state of the amide-bond rotation in peptidyl-prolyl favors C4-exo conformation of the pyrrolidine ring.

The synthesis of the model compounds 3-7 was performed on 0.2-0.1 g scales using the following general schemes: All manipulations were performed at room temperature under air atmosphere. An N-Boc amino acid was treated with an excess of 4 M hydrogen chloride in dioxane for 90-120 min. The solvent was removed under reduced pressure, the residue dissolved in water and freeze-dried. To the residue anhydrous dichloromethane was added, followed by acetic anhydride (2-5 equiv), and trimethylamine (2-5 equiv). The resulting mixture was stirred for about 1 hour. The solvent was removed under reduced pressure, the residue dissolved in water and freeze-dried. The resulting mixture was dissolved in water and filtered through a short cation exchange column. The acidic fractions were collected and freeze-dried. The N-acetylated compound obtained in this way was dissolved in methanol (≈ 1-2 mL) and trimethylsilane (≈ 0.1-0.2 mL) was added to acidify the mixture. The mixture was stirred for several hours, the solvent was removed under reduced S3 pressure, and the residue purified on a short silica gel column using ethyl acetate/methanol 40:1 to 20:1 gradient elution. The compounds 3-7 were obtained as colorless oils.

NMR experiments
NMR experiments were performed with spectrometers operating at 500 and 700 MHz proton frequency. The variable temperature unit was calibrated according to conventional methanol standard. The pKa values were obtained from analytical samples according to the detailed protocols described earlier. S1,S4 The samples were titrated at 295-298 K, and the measurements were performed at 298 K. The lipophilicity measurements were performed as described. S1 A compound (4-6 mg) was shaken with deionized water and octan-1-ol (1.00 mL each) for about 24 hours at 295-298 K. Then, 0.30 mL of each phase were taken in NMR tubes and diluted with 0.30 mL of acetonitrile-d3. The measurements were performed with either 1 H or 19 F detection using datasets with identical settings. The equilibrium ratio (P) was determined as the ratio between absolute integral values from the octanol and water samples. The measurements were performed in triplicate.
For the amide isomerism measurements 4-6 mg of an analyte (1-7) was dissolved in a solvent (0.53 mL) in a standard 5 mm NMR tube. The 1 H and 19 F NMR measurements were performed manually in one scan in order to enable complete prerelaxation of the nuclei. The ratios between equivalent integrals originating from rotameric species were taken as the equilibrium constants (K). The value was averaged from few different resonances, few different spectra, 1 H and 19 F measurements.
The rotation velocities were measured in 2D cross-relaxation pulse sequences with z gradients: 1 H EXSY (for all samples), 19 F EXSY (for 3 and 4) and 19 F{ 1 H} EXSY with proton decoupling during evolution, mixing and detection (for 1 and 2). The measurements in aqueous samples (D2O) were performed at 310 K, the measurements in benzene (C6D6), and dichloromethane (CD2Cl2) were performed at 298 K. The frequency domains were zoomed to the regions of interest, and the experiments were performed with standard settings and prolonged recycling delays (≥3•t1 of the analyzed resonances). The frequency domain spectra were baseline corrected and integrated. The exchange rate matrices were calculated using EXSYCalc freeware (Mestrec). A detailed description is provided below.

EXSY procedure description
For the compound/solvent combination 4/C6D6, first the Ktrans/cis was identified from the 1D 1 H and 19 F NMR spectra as 4.69 ± 0.14: S5 Then, 1 H EXSY were acquired with mixing times 0.5 and 1 s, with the parameters as outlined below: Note prolonged recycling delay (d1). According to our experience, a prolonged recycling delay does not change the final exchange rate, but it substantially improves the agreement between the rate constants derived from analysis of individual resonances. Effectively, this helps to reduce the standard deviation in the final value.
There is also a relatively high number of data points in the indirect dimension (TD1 512). This enables suppression of truncation artifacts in the indirect dimension, and effectively the sharp CH3-group resonances could be integrated and analyzed. S6 19 F EXSY spectra were acquired with mixing times 0.3, 0.6, and 0.9 s with the parameters as shown below: Here, prolonged recycling was also applied. The t1 value was estimated as ≈ 1.4 s according to an inversion recovery experiment. Hence, the spin recycling aq+d1 = 5.7 s ≥ 4•t1. Interestingly, the inversion recovery showed that the t1 for the s-cis resonance is slightly longer.

S7
The frequency domain spectra were phased in direct and baseline corrected in both dimensions. In the 1 H EXSY, the CH3O and acetyl CH3 resonances were integrated. In the 19 F EXSY, the fluorine resonance was integrated. Relative integrals were input into EXSYcalc as shown: The exchange rate matrix is defined as: The main diagonal elements contain the relaxation terms (R1 and R2) that are incorrect in the current treatment. To obtain a fully correct exchange rate matrix, a reference EXSY spectrum with zero mixing time should be acquired, and absolute integral values should be inserted into the EXSYCalc. With the existed method, only the secondary diagonal elements that represent the exchange rates are correct.
In the current example (4/C6D6), the following exchange rates were obtained: The rates were averaged to deliver the final value that is presented in Table S2. The standard deviation values were also calculated from the measured values.

Molecular modeling
The molecular modeling was performed with compounds 1-4 using Scigress Modelling Suite (Fujitsu, Poland). Dipole calculations were performed after geometry optimization using the PM6 Hamiltonian included in the MOPAC package.
The C 4 -exo/endo conformations were extracted from dynamic simulations of potential energy map generated in MM3 force field. The dihedral angles H-C 2 -C 3 -H were estimated from resulting conformations. The experimental 2 JHCCH were then converted to dihedral angles using MestReJ freeware (Mestrec) with the following substitutes: S1=NRC(=O)R(s-cis), S2=COOR, S3,S4=H/CHClMe as the closest ones from the existing list. The conformations were readjusted to fit both dihedral angles within ± 10° accuracy. The ranges for the dihedral angles in the C 4endo and C 4 -exo conformations were estimated from the ones obtained in the simulation run before and after the adjustments. The estimated dihedral angle ranges were then converted to the J value ranges. These are shown in Figure 4.

Figure S1
Dipole size and orientation in methyl acetate and N-methyl-and N,Ndimethylacetamides. The simulations were done using PM6 Hamiltonian from MOPAC package. The pKa values are obtained according to the equation:

Extended physicochemical data
The pKa* values are obtained according to the equation S4 : Thus, the first value is obtained from the titration experiments, while the second is obtained from equilibrium measurements at extreme pH. In principle, both values (columns highlighted in yellow) should be equal within the accuracy of the experimental determination.
The values for fluoroproline derivatives are from reference. S4 S10  Note that the chemical shifts were read out from the 1 H NMR spectra acquired at 298 K. The spectra were referenced according to the conventional deuterium lock referencing (Bruker Avance III console, for solvent details see command 'edlock' in Topspin). No additional referencing was applied. S13