A Unique and Stable Polyproline I Helix Sorted out from Conformational Equilibrium by Solvent Polarity

Polyproline I helical structures are often considered as the hidden face of their most famous geminal sibling, Polyproline II, as PPI is generally spotted only within a conformational equilibrium. We designed and synthesized a stable Polyproline I structure exploiting the striking tendency of (S)-indoline-2-carboxylic acid to drive the peptide bond conformation toward the cis amide isomer, when dissolved in polar solvents. The cooperative effect of only four amino acidic units is sufficient to form a preferential structure in solution. We shed light on this rare secondary structure with a thorough analysis of the spectroscopic and chiroptical properties of the tetramer, supported by X-ray crystallography and computational studies.


■ INTRODUCTION
Since the very first synthesis of poly-L-proline in 1954, by Ephraim Katchalski-Katzir and collaborators, 1−3 the conformational behavior of proline-containing oligomers intrigued many scientists. Two recurring helical secondary structures are generally observed with proline and proline-mimetic oligomers: the right-handed polyproline I (PPI) and the lefthanded polyproline II (PPII). 4−6 The characteristic backbone dihedral angles and geometric properties of PPI and PPII are summarized in Table 1.
The dualistic and interdependent nature of these two structures lies mainly in the different orientation of the amide bonds, with all amide bonds in the cis conformation (ω = 0°) for PPI and all in the trans (ω = 180°) for PPII. The PPII helix has been deeply studied, 7−10 as it is the dominating form in aqueous solution for proline-enriched oligomers, it is involved in many biological processes, and it is the helical conformation adopted by the single strands of collagen. 11−13 It is worth mentioning the recent advances on the comprehension of the factors, which stabilize PPII over PPI structures: the influence of concentration and of the polarity of the solvent, 14,15 the effect of terminal functional groups or substituents on the proline moieties, 16−20 and the temperature and pH. 21−23 However, the PPI helix is a rare secondary structure, mostly observed in organic solvents in its dichotomic relation with PPII, through the mutarotation phenomenon of oligoprolines. More insight into the synthesis and characterization of PPI helices came from the study of peptoids, N-substituted glycine oligomers. 24−26 In particular, the group of Taillefumier described PPI-like structures, stabilized by sterically hindered aliphatic side chains, which lock the amidic bond exclusively in the cis conformation. 27−29 Another example is the stabilization of a PPI-like peptoid helix through metal coordination by Maayan et al. 30 Recently, the group of Tedesco crystallized enantiomorphic right-and left-handed polyproline type I helices, obtained by conformationally restricted cyclic dodecapeptoids. 31 We have recently described the preference of (S)-indoline-2carboxylic acid (2-(S)-Ind) derivatives, to drive the peptide bond conformation toward the cis amide isomer, when dissolved in polar solvents (Scheme 1). 32 A thorough experimental and computational study on the conformational properties of the simple acetamide derivative of 2-(S)-Ind demonstrated the influence of the molecular dipole moment on the conformational equilibrium, and that steric and electrostatic interactions contribute to stabilize the amide cis geometry. 2-(S)-Ind is an excellent model molecule because the NMR signals of the different conformers are well separated and defined, allowing a careful measurement of the thermodynamic and kinetic constants for the cis/trans isomerism. Moreover, the behavior of 2-(S)-Ind is opposite to the general preference of proline for the trans isomer, in polar solvents. 33 −35 Therefore, we speculated that longer oligomers of 2-(S)-Ind might show a cooperative effect strong enough to stabilize a PPI helix structure, even with a few amino acidic units. Here, we report the synthesis and structural characterization of oligomers containing one to four units of 2-(S)-Ind and prove that the tetramer indeed adopts a PPI conformation both in the solid state and in solution.
Scheme 1 for angle definition). The amide angle ω is expected to assume values around 0°and 180°leading to cis and trans isomers, respectively, distinguished by NMR, while the rotational barrier for angle ψ is too low to be recognized by NMR. Therefore, it is reasonable to expect 2 n possible sets of NMR signals corresponding to the number n of amide bonds of the oligomers. Particularly, the regions of the spectra between 4.1 and 5.2 ppm, corresponding to the H α signals, are explicative of the conformational equilibrium. Figure 1 (left) shows, as expected, an increase of complexity for the H α signals of the chemically pure oligomers dissolved in CDCl 3 , upon increasing the oligomer size, indicative of the concurring presence of multiple helices. Conversely, Figure 1 (right) shows a major population of cis isomers in oligomers 1−3, when dissolved in DMSO-d 6 . Moreover, the 1 H NMR spectrum at 25°C of tetramer 4 shows the presence of one predominant fold, suggesting that a cooperative effect stabilizes the formation of a well-defined conformation, possibly a PPI helix, already at the level of the tetramer. To demonstrate the effective prevalence of a single helical structure in solution, we performed more refined NMR experiments (described below) comparing the results with the crystal structure of the tetramer.
X-ray Crystal Structure of Tetramer 4. Crystals of tetramer 4 suitable for X-ray diffraction analysis were obtained by slow evaporation of a 0.1 M solution of 4 dissolved in ethyl acetate, a polar solvent. The diffraction pattern showed a 2/m symmetry. The unit cell metrics and systematic absences along with the chiroptical properties of the molecule indicate the P2 1 space group. The more relevant crystal data and structure Scheme 1. Representation of the Cis−Trans Isomerization Equilibrium of the Amidic Bond in Ac-(2S)-Ind-OMe Depending on the Polarity of the Solvent 32 refinement parameters are listed in Table S1, while a representation of the crystal structure of 4 is shown in Figure  2.
The dihedral angle values measured on the crystal are reported in Table 2, showing the typical values of a polyproline I helix. As a crystal structure could represent only one of the possible helices in solution, we performed afterward a series of NMR and CD analyses in solution, to confirm the predominant presence of a PPI helix in polar solvents.
Two-Dimensional NMR Analysis. In DMSO-d 6 solution, a largely prevalent species of Boc-((2S)-Ind) 4 -OMe (4) (>86%) was detected, which was completely characterized by comparing scalar and dipolar interactions detected in 2D homo-and heteronuclear maps. An exhaustive description of the methodology used for the attribution of each signal is reported in the Experimental Section.
For the definition of the conformation of the prevailing species of the tetramer in DMSO-d 6 Figure 3) supported an all-cis conformation of the tetramer, which is in fact the sole conformation keeping the CH α methine protons in spatial proximity to each other. Such a conformation was also supported by the following spatial proximity constraints imposed by ROE data and represented in Figure 3: (i) intraunit ROE H α −H β(cis) and interunit ROE H α − H β′(trans)* (the star denotes the adjacent previous unit) showed comparable intensities ( Figure S1); (ii) the methyl moieties of the t-butyl group gave ROE on peri aromatic protons H B1 and H C1 of units B and C; and (iii) a ROE between the protons of   Residues depicted in Figure 3. b The ψ angle referred to the ester bond NCCO(Me) of the last residue D.  Figure S1. the two terminal units H α A and H D1 was also detected. The conformation of the tetramer in DMSO-d 6 solution defined by NMR data is in very good agreement with that obtained in the solid state by X-ray, as shown in Figure 3. Thus, the PPI helix is maintained in DMSO solution.
Thermal Stability of Tetramer 4 in DMSO. We measured 1 H NMR spectra at variable temperatures of tetramer 4 dissolved in DMSO-d 6 , to check the stability in solution of the PPI conformation. The temperature was increased by 10°C steps for each measurement, starting from 25 until 105°C. Figure 4 highlights an exceptionally stable structure up to the highest temperature. No other signals, corresponding to other isomers, are visible at higher temperatures in the H α protons region (5.0−5.6 ppm), while the ratio between their integrals and the integrals of peri aromatic H 1 protons (7.75−8.20 ppm), characteristic of cis isomeric units, remains basically unchanged.
ECD and VCD Spectra of 1−4. Electronic and vibrational circular dichroism (ECD and VCD) spectra were recorded for all compounds 1−4 in several solvents, CH 3 OH, CH 3 CN, and CHCl 3 for ECD (DMSO could not be used due to its absorption bands up to 265 nm), and CD 3 CN, CDCl 3 , and DMSO-d 6 for VCD. The ECD spectrum of monomer 1 is faintly visible. Upon increasing the oligomer length n, a consistent structured spectrum appears in all solvents, reaching maximum intensity for tetramer 4 ( Figure 5). The most intense negative band at 250−255 nm is absent for 1 and approaches Δε max values of −24, −60, and −113 M −1 cm −1 for 2, 3, and 4, respectively. The nonlinear increase with n witnesses a progressive stabilization of a well-defined conformation, which is similar in CH 3 OH and CH 3 CN, and possibly also in CHCl 3 . In comparison, an almost linear trend is observed for the UV spectra ( Figure S2). Variabletemperature ECD spectra measured in CH 3 CN between 20 and 50°C showed only a modest decrease in intensity upon heating ( Figure S3), witnessing again the stability of the major conformation responsible for the ECD profile. Because of the presence of the aniline-like chromophore, a comparison with ECD spectra commonly associated with the PPI structure, due to the amide chromophore transitions, is not immediate. A distinctive ECD signature of PPI is a positive band around 210 nm in water or alcohol solvents, allied with the amide π−π* transition. 6,36 Although our spectra do contain a positive ECD band in that region ( Figure 5), transition and molecular orbital (MO) analysis after TD-DFT calculations (see below) highlights a major contribution of MOs localized on the aromatic rings in this region too, in addition to amide-centered transitions (Supporting Information).
VCD spectra of compounds 2−4 parallel the behavior observed by ECD. The spectra become progressively stronger and better defined upon increasing the oligomer length n, and for the tetramer they acquire a consistent shape in the fingerprint region (<1450 cm −1 ) in all solvents (spectra for 4 in Figure 6; all other spectra shown in Figures S5 and S6). IR/ VCD characterization of PPI helix has been sporadic, also because of its coexistence with PPII. 37,38 According to Dukor and Keiderling, the VCD signature of PPI helix adopted by poly-L-proline consists of a negative couplet in the amide I region around 1650 cm −1 in D 2 O, similar to PPII but shifted to higher frequencies. 36 Our spectra for the tetramer and the trimer ( Figure S6) consistently display a negative VCD couplet between 1650 and 1700 cm −1 , with solvent-dependent intensity, which by DFT calculations (vide infra) is assigned to amide I stretching vibrations. The IR spectra reported for PPI in D 2 O consist of three bands in the amide I region, 36 also consistent with our results ( Figure 6). We notice some discrepancy in this region between CDCl 3 and the other two solvents, which we attribute to a different conformational manifold also suggested by NMR results. Thus, we may confirm the occurrence of the diagnostic IR and VCD signature established by Dukor and Keiderling on poly-Lproline also for other peptides when they assume a similar PPI helix conformation.

Molecular Dynamics of Tetramer 4.
To better characterize the possible conformations of tetramer 4, we employed a  The Journal of Organic Chemistry pubs.acs.org/joc Article molecular dynamics (MD) enhanced sampling strategy in explicit chloroform and DMSO solvents. Briefly, parallel-bias metadynamics (PBMETAD) 39,40 was used to enhance the exploration of the tetramer free-energy landscape along with the four ω angles and the three amide ψ angles. A clustering procedure was employed to distinguish different conformers, and representative conformer structures were refined at the DFT level (see the Computational Details). Surprisingly, we found very small energy differences between the various conformers, many of them being within 1 kcal/mol from each other (see Table S2). As the intrinsic error associated with our calculations is likely larger than 1 kcal/mol, it is not possible to draw definitive conclusions from the calculations, but it is still possible to note some trends. First, of all of the conformers obtained from the MD simulations, the vast majority have at least two amide cis bonds, with only one having a single cis bond being selected among the stable ones per solvent. This suggests that conformations with more cis amide bonds were generally more stable. Interestingly, the all-cis conformer (see Figure 7a) is observed in both solvents. In DMSO, it is the second most stable conformer, with an energy difference of just 0.9 kcal/mol with respect to the absolute minimum (see Figure  7b). As was already mentioned, such an energy difference is smaller than the intrinsic error associated with the method, and therefore we can only conclude qualitatively that the all-cis conformer is thermodynamically favored in DMSO. This is not the case in chloroform, where such a conformer is predicted to be about 3 kcal/mol less stable than the minimum. The different behavior can be understood by looking at the large dipole moment of such a conformer, and in general of all of the predominantly cis conformers, which explains their higher stability in a polar solvent such as DMSO. Furthermore, the conformer with four cis amide bonds is very close to the X-ray structure (RMSD ≈ 0.6 A; see Figure 7a), which further supports the identification of such a conformer as the most thermodynamically stable one.
To obtain a more direct link with NMR experiments in DMSO, we analyzed the distances involved in interunit ROE interactions observed in NMR (Figures 3 and S1) along the PBMetaD. As the conformers interconvert slowly, we calculated ROE-like distances separately for each cluster (Table S4). While the all-cis conformer respects all of the experimental ROE constraints, all of the other conformers present at least two violations. This suggests that, of all of the conformations explored in our simulation, only one is compatible with all of the observed ROEs.
Calculations of the VCD and ECD Spectra of 4. X-ray crystallography, NMR evidence, and simulations all converge on a strongly preferred all-cis PPI-like structure for the tetramer. To reconcile these findings with chiroptical measurements, we run VCD calculations on two different structures, one obtained after relaxing the X-ray structure through DFT geometry optimizations at the B3LYP-D3/6-311+G(d,p) level and the other one being the all-cis structure obtained after the computational procedure at the same level of theory. As was stated above, the two structures are very similar to each other ( Figure 7a). In particular, the two central units B and C are almost coincident, while the largest discrepancy occurs for the aromatic ring of unit D.
In both cases, calculations were run both in vacuo and using the implicit PCM solvent model for DMSO. In fact, we aimed at reproducing VCD spectra recorded in this solvent where the major conformation found by NMR recalls the X-ray one. Figure 8 shows the comparison of VCD spectra calculated from the two aforementioned structures with PCM for DMSO with the experimental spectrum recorded in d 6 -DMSO. The  The Journal of Organic Chemistry pubs.acs.org/joc Article agreement is remarkable in the fingerprint region, while it is poorer in the amide I region where, however, a negative couplet is correctly predicted. The normal modes responsible for the calculated couplet, shown in Figure S7, correspond to amide carbonyl stretching vibrations. Thus, VCD calculations further demonstrate that the X-ray geometry is largely retained in DMSO solution and confirm the VCD signature for the PPI helix established by Dukor and Keiderling, as a negative couplet in the amide I region.
A consistent approach was adopted for ECD calculations of 4. DFT-optimized geometries obtained starting from the X-ray and MD all-cis structures were employed in this case too. TD-DFT calculations were then run in vacuo or with the PCM solvent model for CH 3 CN or CH 3 OH, with B3LYP and CAM-B3LYP functionals, employing either the def2-SVP or the def2-TZVP basis sets. The results (Figures 9 and S8−S10) were all quite consistent and nicely reproduced the experimental ECD spectra. As an example, Figure 9 shows the comparison between the ECD spectrum of 4 in CH 3 CN and those calculated at the CAM-B3LYP/def2-SVP level using the two geometries (Figure 7a). The consistency between the experimental ECD spectra recorded in various solvents ( Figure  5) and the calculation results demonstrates again that the all-cis PPI-type structure found in the solid state, and obtained from MD/DFT simulations, is preserved in different conditions and represents a strongly preferred fold for (2S)-Ind-based oligomers.

■ CONCLUSION
We demonstrated the formation of a stable PPI helix structure, obtained from a short oligomer of (S)-indoline-2-carboxylic acid, in polar solvents. A thorough characterization, by NMR and X-ray diffraction, strongly supports the presence of a stable and basically unique structure in solution and in the solid phase. Measurements of the chiroptical properties, of the wellfolded small oligomer, through ECD and VCD techniques, further confirmed the prevalence of a stable PPI fold and allowed the identification of diagnostic signals. The wealth of experimental evidence, combined with the conformational analysis and DFT simulation of the chiroptical properties, supports the assignment of a single, stable PPI conformation in polar solvents. In conclusion, the rare Polyproline I has been uniquely isolated, in solution and in the solid phase, allowing for a fine characterization, which will be useful for future identification of naturally occurring PPI or in the design of structured peptide sequences. These findings could aid in the design of β-turns or β-hairpins or in the synthesis of macrocyclic peptides. ■ EXPERIMENTAL SECTION General Information and Materials. All nonaqueous reactions were run in oven-dried glassware under a positive pressure of argon, with exclusion of moisture from reagents and glassware, transferring solvents and liquid reagents with hypodermic syringes. The glassware has been dried with a heating gun under vacuum and allowed to cool under argon. Anhydrous solvents and liquid reagents were obtained  The Journal of Organic Chemistry pubs.acs.org/joc Article using standard drying techniques. Solid reagents were of commercially available grade, used without further purification, and, when necessary, stored in a controlled atmosphere and/or at −20°C.
(S)-Indoline-2-carboxylic acid was purchased from abcr GmbH. Reactions were monitored by thin-layer chromatography using Merck silica gel 60 F254 plates. Visualization of the developed chromatogram was performed by UV absorbance, aqueous potassium permanganate, or iodine. Flash chromatography was performed using Sigma-Aldrich silica gel 60, particle size 40−63 μm, with the indicated solvent system. Melting points were measured using a "Buchi Melting Point B-545" instrument.
NMR spectra of synthetic intermediates were recorded on a Bruker Avance DRx 400. Compounds 1, 2, and 3 were fully characterized with 1 H NMR and 13 C NMR on a Jeol instrument JNM-ECZ500R. Structural assignments were made with additional information from gCOSY, gHSQC, and gHMBC experiments. For compound 4, 1 H NMR, 13 C NMR, gCOSY, gHSQC, ROESY, TOCSY, and temperature titration were recorded on a VARIAN INOVA 600 MHz instrument. Chemical shifts are reported in ppm with the deuterated solvent signal as the internal standard. Data are reported as follows: chemical shift, integration, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, qn = quintet, m = multiplet, and br = broad), with the coupling constant in hertz.
The crystal structure measurements have been done on a Bruker SMART BREEZE CCD diffractometer equipped with a graphite monochromate Mo Kα radiation.
HPLC-ESI-Q/ToF flow injection analyses (FIA) were carried out with a 1200 Infinity HPLC (Agilent Technologies, U.S.), coupled with a quadrupole-time-of-flight tandem mass spectrometer (6530 Infinity Q-TOF; Agilent Technologies) through a Jet Stream ESI interface (Agilent). Mass Hunter Workstation Software (B.04.00) was used to control the HPLC and the mass spectrometer, for data acquisition and for data analysis.
Boc-(2S)-Ind-OMe (1). (S)-Indoline-2-carboxylic acid (H-(2S)Ind-OH) (2.8 g, 17.2 mmol) was dissolved in 280 mL of methanol and cooled to 0°C with an ice bath. To this suspension was added dropwise thionyl chloride (1.87 mL, 25.7 mmol, 1.5 equiv). The reaction mixture was stirred for 1 h and allowed to reach room temperature. Afterward, the resulting solution was heated at 70°C, in an oil bath, and allowed to stir at reflux for 16 h. After being cooled at room temperature, the reaction mixture was concentrated under reduced pressure. The residue was dissolved in ethyl acetate (EtOAc) and washed with an aqueous saturated solution of sodium bicarbonate (NaHCO 3 ) three times. The combined organic layers were washed with brine, dried over Na 2 SO 4 , and concentrated under reduced pressure. The resulting crude material was purified by flash chromatography on silica gel using 10−50% EtOAc in hexane to give the desired product (H-(2S)-Ind-OMe) as a white solid in 82% yield (2.5 g, 14.1 mmol). 1  3 mmol) of (Boc) 2 O was dissolved in 10 mL of dioxane and dropped into the reaction mixture, which was left to stir for 16 h at rt. 250 mL of EtOAc was added to the solution, and the resulting mixture was washed with HCl 1 M for 3 × 250 mL. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated under reduced pressure. The resulting solid was purified by flash chromatography on silica gel using 10−50% EtOAc in hexane. The desired product was obtained as a white solid in an isolated yield of 75% (9.3 mmol, 2.58 g). TLC R f = 0.3 (Hex:EtOAc = 8:2), mp = 54−56°C. Boc-((2S)-Ind) 2 -OMe (2). (S)-Indoline-2-carboxylic acid (H-(2S)-Ind-OH) (1.2 g, 7.2 mmol) was dissolved in 20 mL of a dioxane:water (1:1) mixture, and 1.7 g (7.8 mmol) of (Boc) 2 O was added. The reaction mixture was left to stir at rt for 16 h. The resulting solution was then concentrated under reduced pressure, and the residue acidified with HCl 1 M was washed three times with EtOAc. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated under reduced pressure. The crude product Boc-(2S)-Ind-OH was obtained as a white solid with 87% yield (6.3 mmol, 1.66 g) and used without further purification. When needed, the reaction was performed several times on the same scale. 1  To a 45 mL dry dichloromethane stirred solution of Boc-(2S)-Ind-OH (1.75 g, 6.63 mmol) and Mukaiyama reagent (2.37 g, 9.3 mmol) was added 1.17 g (6.63 mmol) of H-(2S)-Ind-OMe. 2.6 mL of TEA (18.6 mmol) was dropped into the solution, and the reaction mixture was heated at reflux, in an oil bath, for 16 h before being allowed to cool at room temperature. The reaction mixture was then diluted with DCM and washed with HCl 1 M, a saturated solution of Na 2 CO 3 , and brine. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography on silica gel using 10−50% EtOAc in hexane, obtaining the desired product as a white solid with 70% yield (4.64 mmol, 1.96 g). TLC R f = 0.43 (Hex:EtOAc = 7.5:2.5), mp = 87−89°C. 1  Boc-((2S)-Ind) 3  To a 25 mL dry DCM stirred solution of Boc-((2S)-Ind) 2 -OH (997 mg, 2.4 mmol) and Mukaiyama reagent (845 mg, 3.3 mmol) was added 434 mg (2.45 mmol) of H-(2S)-Ind-OMe. 920 μL of TEA (6.6 mmol) was dropped into the solution, and the reaction mixture was heated at reflux, in an oil bath, for 16 h before being allowed to cool at room temperature. The reaction mixture was then diluted with DCM and washed with HCl 1 M, a saturated solution of Na 2 CO 3 , and brine. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography on silica gel using 10−30% EtOAc in toluene, to obtain the desired product as a white solid with 60% yield (1.44 mmol, 818 mg) To a 10 mL dry DCM stirred solution of Boc-((2S)-Ind) 3 -OH (900 mg, 1.6 mmol) and Mukaiyama reagent (580 mg, 2.27 mmol) was added 290 mg (1.63 mmol) of H-(2S)-Ind-OMe. 630 μL of TEA (4.5 mmol) was dropped into the solution, and the reaction mixture was heated at reflux, in an oil bath, for 16 h before being allowed to cool at room temperature. The reaction mixture was then diluted with DCM and washed with HCl 1 M, a saturated solution of Na 2 CO 3 , and brine. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography on silica gel using 10−40% EtOAc in toluene, to obtain the desired product as a white solid with 32% yield (0.  13  Crystal Structure Determination. The structure solution was obtained by automated direct methods contained in the SHELX suite. 41 The absolute configuration of the molecule was inferred starting from the known configuration of the stereogenic carbons of proline. After the hydrogen atoms were placed in calculated positions, the non-hydrogen atoms were refined with anisotropic thermal parameters. At the end of refinement, the thermal ellipsoids of some atoms of the t-butyl carbamate protecting group were excessively elongated, suggesting the presence of some conformational disorder, which, however, could not be modeled. NMR Characterization. 1 H NMR characterization of the PPI helix for compound 4 in DMSO-d 6 was performed on a 600 MHz instrument. On the basis of the 1 H− 13 C scalar correlations detected in the HSQC map (Supporting Information), resonances between 4.9 and 5.5 ppm and between 2.8 and 4.1 ppm were attributed to the methine (H α ) and diastereotopic methylene protons of the pentatomic rings, respectively. Aromatic protons originated a complex pattern of signals in the high-frequency spectral region (6.6−8.2 ppm), and sharp singlets at 1.52 and 3.86 ppm were unambiguously assigned to the tbutyl and methoxy groups, respectively. In the attribution of NMR signals of Boc-((2S)-Ind) 4 -OMe, the four monomeric units were named sequentially as A−D (Figure 3), with end A bearing the t-butyl moiety. Through-space dipolar interactions detected in the 2D ROESY map (Supporting Information) and scalar correlations detected in the 2D COSY and 2D TOCSY allowed one to distinguish the resonances bearing to every unit (Table 3). Among methine protons CH α , only the one centered at 5.05 ppm originated a dipolar interaction at the frequency of the t-butyl group and was then attributed to CH α A . Intraunit ROEs produced by CH α A allowed one to distinguish H β A (3.45 ppm) and H β′ A (2.92 ppm), cisoid and transoid to it, respectively. Importantly, CH α A gave a dipole−dipole interaction with the methine proton resonating at 5.18 ppm, which was then attributed to CH α B of the adjacent unit B. The above said dipolar interaction also led one to assess that the amide junction between the two adjacent A and B units of the prevailing species showed a strong preference for the cis conformation, as was already demonstrated for the dimer Ac-((2S)-Ind) 2 -OMe in the same solvent. Dipolar interactions generated by H α B at the frequencies of 3.67, 3.32, and 5.35 ppm led to the assignment of adjacent cissoid H β B (3.67 ppm) and transoid H β′ B (3.32 ppm) and to the identification of the methine at the chiral center of the third unit, H α C . Likewise, ROE effects produced by this last proton allowed one to attribute the methylene protons of unit C (H β C at 3.98 ppm and H β′ C and 3.47 ppm, cisoid and transoid to H α C , respectively) and the methine proton H α D (5.40 ppm) of the last unit. Intraunit ROEs given by this last proton led to the assignment of its cissoid H β D (3.84 ppm) and transoid H β′ D (3.48 ppm). Regarding the aromatic proton resonances, peri protons H ar4 of every unit were assigned on the basis of the ROE effects produced by their adjacent methylene protons, and the TOCSY and COSY correlations detected starting from peri protons led to the complete assignment of aromatic protons (Supporting Information). Computational Details. Molecular Dynamics Simulations Run with the Amber Software. The tetramer was described by adapting the protein ff14SB parameters for the amide backbone, 42 while specific parameters for the aromatic side chain were taken from the General Amber Force Field (GAFF). 43 Torsional parameters for rotation around the amide bond were refit from calculations at the B3LYP/6-311G(d,p) level and substituted to the force field parameters. The tetramer was placed in a DMSO or chloroform solvent box, and the system was equilibrated (the equilibration protocol is described in detail in the Supporting Information). PBMETAD simulations were run for 1.5 μs in both solvents, applying a parallel metadynamics bias on each ω angle and the three ψ angles between the indoline residues. We used PBMETAD as it allows biasing several collective variables (CVs) at once in a parallel fashion. As the barriers separating the conformers are almost exclusively along these torsions, this strategy allows an exploration as much as possible unbiased of the tetramer's conformational space. PBMETAD simulations were run with Amber interfaced with the PLUMED enhanced sampling library. 44 The sampled structures were reweighted using the approach by Tiwary and Parrinello and clustered using the K-means algorithm based on the ω and ψ angles. 45 One structure from each cluster, representative of one conformation, was extracted for the QM analysis.
QM Calculations. All QM calculations were run with Gaussian 16 software. 46 The structures extracted from DMSO and chloroform simulations were optimized in PCM solvent 47 at the B3LYP-D3(BJ)/ 6-311+G(d,p) level of theory. Frequency calculations were used to ensure the achievement of a minimum, to compute the free energy differences, and to calculate the VCD and IR spectra. Excited-state calculations were run with the TD-DFT method employing the B3LYP and CAM-B3LYP functionals, the def2-SVP or def2-TZVP basis sets, either in vacuo or in PCM solvent. The number of roots was 100 for CAM-B3LYP and 150 for B3LYP. IR/VCD and UV/ ECD spectra were generated using SpecDis v. 1.71; 48 plotting parameters are reported in the figure legends.
Equilibration Protocol. The tetramer was placed in a truncated octahedron box with 1080 DMSO molecules (or 952 CHCl 3 molecules) and minimized with 2000 conjugate gradient steps. Each box was heated to the desired 300 K temperature in a 100 ps simulation run. Additional 100 ps NPT simulations were performed to equilibrate the density of the solvent. Finally, a production run of 40 ns in the NPT ensemble was run without any bias. In all simulations, we used a 2.0 fs integration time step combined with the SHAKE algorithm. The Langevin thermostat was used to control the temperature, whereas the pressure was controlled by the Monte Carlo barostat implemented in Amber. In all unbiased runs, the tetramer remained in its initial conformation as was monitored by the ω and ψ angles.
Details on the Parallel-Bias Metadynamics. To explore the conformations of the tetramer in both solvents, we employed welltempered Parallel-bias Metadynamics (PBMETAD) simulations. In PBMETAD, multiple one-dimensional bias potentials are applied on several collective variables (CV) of the system. As in standard Metadynamics, the bias potential is expressed as a sum of Gaussian functions. However, here the bias on one CV depends on how much the other CVs have been biased, ensuring the convergence of the monodimensional bias to the correct limit. Here, we considered each peptide ω and ψ angle as a CV, for a total of seven CVs in the PBMETAD simulations. We found it necessary to bias also the ψ angles, as the indoline side chains give rise to significant barriers also along ψ rotation. Well-tempered metadynamics was applied, with a bias factor of 20. This value was estimated from the potential energy barrier along ω rotation (∼16 kcal/mol).
Full characterization of the synthesized compounds, NMR experiments, calculation details, HPLC−MS spectra, and crystallographic data for compound 4 (PDF) FAIR data, including the primary NMR FID files, for compounds 1, 2, 3, and 4 (ZIP)