Synthesis and Conformational Properties of 3,4-Difluoro-l-prolines.

Fluorinated proline derivatives have found diverse applications in areas ranging from medicinal chemistry over structural biochemistry to organocatalysis. Depending on the stereochemistry of monofluorination at the proline 3- or 4-position, different effects on the conformational properties of proline (ring pucker, cis/ trans isomerization) are introduced. With fluorination at both 3- and 4-positions, matching or mismatching effects can occur depending on the relative stereochemistry. Here we report, in full, the syntheses and conformational properties of three out of the four possible 3,4-difluoro-l-proline diastereoisomers. The yet unreported conformational properties are described for (3 S,4 S)- and (3 R,4 R)-difluoro-l-proline, which are shown to bias ring pucker and cis/ trans ratios on the same order of magnitude as their respective monofluorinated progenitors, although with significantly faster amide cis/ trans isomerization rates. The reported analogues thus expand the scope of available fluorinated proline analogues as tools to tailor proline's distinct conformational and dynamical properties, allowing for the interrogation of its role in, for instance, protein stability or folding.


■ INTRODUCTION
Fluorination of organic molecules has proven to be a highly useful tool for the manipulation of their conformational and electronic properties with minimal steric effects. 1−7 Fluorination of the L-proline ring has been heavily exploited for conformational control of its ring pucker. 8 For example, the five-membered proline ring conformation can be biased to either a C γ exo or a C γ endo pucker by introducing a (4R)fluoro group (1, Figure 1) or a (4S)-fluoro group (2), respectively, an effect attributed to σ CH → σ* CF hyperconjugation interactions. 9 Besides a ring pucker, fluorination also strongly influences the cis/trans ratio of the Xaa-Pro peptide bond relative to proline in a solvent-dependent way. 10 The inductive effect of fluorine reduces the capacity for the nitrogen lone pair to conjugate with the amide carbonyl group and thus to contribute to the double bond character of the amide bond. As a consequence, the rotational energy barrier is decreased and accelerated cis/trans isomerization is observed. 11−13 The same effect renders fluorinated prolines less basic 11,13,14 and the carboxylic acid group more acidic. 15 The combination of both conformational and dynamical effects make fluoroprolines valuable tools for determining the significance of proline's unique structural properties within proteins or peptides. 8,14 Nevertheless, the first syntheses of (4R)-FPro 1 and (4S)-FPro 2 date back to 1965, 16 although it took until the late 1990s for this potential to be fully recognized. In a landmark study investigating the mechanism behind collagen stability, 9,17,18 Raines and co-workers applied fluoroprolines to revise the origins behind the extraordinary thermostability of this protein, which forms triple helices out of Pro-Hyp-Gly repeats. Replacing (4R)-4-hydroxyproline (Hyp) with (4R)-FPro 1 led to a more thermostable collagen mimic, which, since fluorine is a weak hydrogen bond acceptor, disproved that a hydrogen bond network involving the hydroxyl moiety of Hyp induces collagen stability. In contrast, replacing Hyp with (4S)-FPro 2 led to less stable collagen mimics. Since fluorine is more electronegative than a hydroxyl group, (4R)-FPro favors the C γ exo pucker more strongly than Hyp, and because (4S)-FPro favors the C γ endo pucker, this revealed that it is the strong preference for the C γ exo pucker of Hyp that plays a key role for collagen stability. This ring pucker preorganizes the dihedral angles in such a way that a favorable n → π* interaction is promoted between the carbonyl groups of two adjacent peptide bonds, favoring the trans amide bond rotamer. 19 Interestingly, the ring pucker of the Pro residue preceding Hyp is also relevant for collagen stability, 20 which has equally been investigated using both 4-and 3monofluorinated proline variants. 21 The case of collagen initiated many other demonstrations of the potential of proline fluorination to investigate the distinct structural and dynamical properties of proline residues within peptides and proteins, exploiting both the modulations of proline structure and cis/trans isomerization kinetics. 8 Indeed, modulating these properties by fluorination, rather than just fully eliminating them by mutating proline to nonproline residues, 22 can provide a more elegant approach toward uncovering the functional significance of proline's unique properties. Moreover, the introduction of fluorine allows the use of 19 F NMR as a powerful means to monitor residuespecific information. The exceptionally high responsivity of the 19 F nucleus to changes in its (local) environment, in addition to the sparsity of the 19 F spectrum, make 19 F NMR a very attractive means to monitor protein structural and dynamical changes, enzyme catalysis, and ligand binding. 23−28 Despite these clear advantages and earlier suggestions, 29−31 to the best of our knowledge, there are only a very limited number of reports involving the full potential of 19 F NMR in a fluoroproline peptide context. 32,33 However, if the FPro residue is to be used purely as a 19 F NMR probe, the conformationally perturbing effects of fluorine must be carefully considered. We recently introduced (3S,4R)-3,4difluoroproline ((3S,4R)-FPro) 4 ( Figure 2) where the two fluorines have opposing preorganizing effects, thus resulting in a proline analogue with minimal conformational bias and minimal homonuclear coupling complications for 19 F NMR purposes. 34 Given the well-demonstrated importance of having fluoroprolines available with matching conformational, kinetic, and NMR properties for the application at hand, the synthesis of novel fluorinated variants in an optically pure form continues to be of interest. 35 In addition, regardless of whether such applications require conformationally neutral, C γ exo or C γ endo pucker promoting fluoroprolines, the availability of more than one variant with similar conformational properties, but well-separated 19 F NMR chemical shifts, is of interest for site-specific multiresidue-labeling strategies of proteins, especially in the case of low-complexity sequences found in prolinerich proteins such as collagen, but also many transcriptional activators. Hence, we envisaged a convenient synthesis of the 3,4-difluoro-L-prolines 4−7 (Figure 2), in order to expand the toolbox of proline analogues.
In this work, we describe in detail the synthesis of the yet unreported (3S,4S)-3,4-difluoroproline 5 and a novel, more concise route for (3R,4R)-3,4-difluoroproline 6, both as their N-Boc derivatives, and as their N-acetylated methyl esters 21 and 22 (Scheme 2). Following our earlier communication, the development of the synthesis of N-Boc-4, including further optimization efforts of the bis-deoxyfluorination step as well as a direct synthesis of (N-Fmoc)-4, is described. The ring pucker analyses, prolyl bond cis/trans ratios, and isomerization kinetics of 21 and 22 are described and compared to those of unmodified proline and the four known monofluorinated proline derivatives. Since 5/21 can be regarded as a combination of (4R)-FPro and (3R)-FPro, both known to be biased to the C γ exo pucker and trans peptide bond configuration relative to proline, 14 it was anticipated that 5/ 21 will display a conformational bias in the same direction. Similarly, 6/22 was expected to have a larger proportion of the C γ endo pucker and of the cis peptide bond configuration relative to proline, as it is a combination of (4S)-FPro and (3S)-FPro. approach for 3,4-difluorination, with an excellent precedence available from the Marson group, who obtained trans-3,4difluoropyrrolidine from trans-3,4-dihydroxypyrrolidine via the corresponding triflates. 38 While the epoxides and diols would be accessed from 3,4-dehydroderivatives 25a−c, direct functionalization of 25a−c such as vicinal difluorination or a halofluorination/fluoride halide displacement could also lead to the desired 3,4-difluoroprolines. 3,4-Dehydroproline is a commercially available (expensive) building block but can also be obtained by a well-described elimination process involving 26a−c starting from cheap (4R)-4-hydroxyproline. Finally, an electrophilic fluorination approach as recently described by Ciulli et al. 39 leading to 27a/28a was also envisaged. With facile deprotection and versatility in mind, a benzyl ester in combination with various amine protecting groups were used throughout our investigations.
3,4-Dehydroproline Synthesis. Initial efforts focused on achieving a large-scale synthesis of 3,4-dehydroproline 25. Following a literature protocol, conversion of protected (4R)-4-hydroxyproline 26a to the corresponding iodide, via a Mitsunobu reaction, 40 followed by DBU-promoted HI elimination, gave a ±5:1 mixture of alkene regioisomers, from which the desired alkene 25a could be isolated in an excellent combined 76% yield (not shown), with 16% of the undesired 4,5-alkene 31a. While this elimination reaction gave 25a as a pure enantiomer (>97% ee, see Supporting Information), the separation of the alkene isomers was cumbersome. Moreover, it was found that conversion of 26b to the corresponding 4-OMs derivative 30b (Scheme 3), followed by elimination using the same base, led to a mixture (±2:1 ratio) of racemic alkene 25b and partially racemized 31b. A 89:11 ratio of amide rotamers of 31b was observed in the NMR spectra, with NOESY analysis showing the trans isomer being the major rotamer (see Supporting Information). Pleasingly, a one-pot Grieco elimination sequence 41 directly starting from 26a gave enantiopure 25a as the major regioisomer with an increased regioselectivity (>10:1 ratio), and with a negligible degree of racemization. The smaller Scheme 1. Precedence for the Synthesis of 3,4-Difluoroprolines

Scheme 2. Retrosynthetic Analysis
The Journal of Organic Chemistry Article amount of 31a facilitated purification considerably. Furthermore, in contrast to base-mediated elimination reactions, it was found that direct Grieco elimination of (4R)-4-hydroxyproline could also be performed with an N-Fmoc-protecting group (26c) in a very good yield. This procedure is an improvement over the previously reported two-step elimination via 4-SePh intermediates, which are typically prepared from the corresponding 4-OMs or 4-OTs derivatives. 42−46 Vicinal Difluorination and Halofluorination. Direct vicinal difluorination of alkene 25a was attempted using recent methods developed by Gilmour and Jacobsen, both based on the in situ generation of a hypervalent iodoarene difluoride. 47,48 Unfortunately, both methods were unsuccessful and only led to recovered starting material (not shown). Subsequent attempts to effect halofluorination on 25a using different combinations of NBS, NCS, or NIS with either HF· pyridine or Et 3 N·HF were unsuccessful as well, and this line of research was abandoned.
Epoxide-Based Strategy. Epoxidation of 3,4-dehydroproline derivatives is known, but not with the Boc/Bn-or Ac/Bnprotecting group combinations. Following protocol, treatment of 25a/b with mCPBA led to a mixture of epoxides 9a/b and 10a/b in good yields with the trans isomer 10a/b isolated as the major isomer after chromatography (Scheme 4). While determination of the epoxide stereochemistry was achieved by 1 H NMR analysis as reported by Robinson et al. on N-Cbz-3,4epoxyproline benzyl esters (Supporting Information), 49 unambiguous conformation of the stereochemistry was obtained by X-ray crystallographic analysis of 9b (Supporting Information).
First, epoxide 10a was investigated as a substrate for direct fluoride opening with HF reagents (Table 1). Reaction with Et 3 N·3HF in dichloroethane (DCE) at 80°C for 3 days resulted in a complex mixture of chlorinated and fluorinated products (±15%), alongside 68% of the recovered starting material (not shown), but conducting the reaction neat with increasing the reaction temperature to 130°C (entry 1) induced deprotection and aromatization, leading to pyrrole 35 in a quantitative yield. Due to its low reactivity, the use of Et 3 N·3HF is often characterized by long reaction times and high reaction temperatures, which can be alleviated by microwave irradiation. 50 However, with a short reaction time, no product was observed and increasing the reaction time and temperature led to pyrrole 35 (entries 2−4). With the more reactive DMPU·HF, 51 reaction of 10b did lead to fluorohydrin 33b in a 15% yield, together with 30% of the recovered starting material (entry 5). Unfortunately, raising the reaction time and temperature did not improve the yield (entry 6). These reactions suffered from gel formation, which impeded the isolation of the products. The use of hexafluoroisopropanol (HFIP) as an additive successfully disrupted gel formation, but no fluorination was observed (not shown). Next, epoxide opening was attempted with Bu 4 NH 2 F 3 . Unexpectedly, the reaction at reflux in DCE yielded chlorohydrin 32a (entry 7). Presumably, decomposition of the solvent under these conditions must have released chloride ions, which subsequently opened the epoxide. In toluene, Bu 4 NH 2 F 3 was found to be too basic, with fluoride causing H α deprotonation, leading to the formation of allylic alcohol 34a (entry 8). This was also the major pathway upon reaction with TBAF in t-BuOH (entry 9). Interestingly, in contrast to the 4,5-dehydro isomer 31b, the 13 C and 1 H NMR spectra of 34b only showed a single set of resonances, which could indicate the presence of a single rotamer. The NOESY NMR spectrum of 34b is consistent with the trans rotamer (Supporting Information). With KHF 2 in ethylene glycol at 150°C (entry 10), aromatization and transesterification was observed, yielding 36.
Starting from 10b, regioselective opening with HCl, HOTs, and HBr (or MgBr 2 ) led to the corresponding 4-substituted 3hydroxyprolines 32b, 37b, and 38b in excellent yields. However, subsequent DAST-mediated deoxyfluorination reactions mostly led to aromatization: for the chlorohydrin 32b, pyrrole 39b was the only product isolated, while, with the βhydroxy tosylate 37b, a low yield of the desired 3-fluorinated product 40b was obtained, alongside 62% of pyrrole 39b. Tentative assignment of the expected stereochemistry of 40b at C β was based on the observed coupling constant of 5 Hz between H α and H β . Attempts to achieve fluorination at the 4position in the presence of the 3-OH group by bromide or tosylate displacement with TBAF-t-BuOH were also unsuccessful. Starting from 37b, a mixture of allylic alcohol 34b and epoxide 10b was obtained. Despite the reduced basicity due to hydrogen bonding with t-BuOH, fluoride must have deprotonated the alcohol group of 37b causing epoxide formation, followed by H α deprotonation, resulting in epoxide opening to give 34b. Using bromohydrin 38b, the same allylic alcohol 34b was the only product isolated. Interestingly, treating 38b with AgF in nitromethane only led to epoxide formation in a quantitative yield.
At this point, the epoxide-based strategy was abandoned, and attention shifted to fluorine introduction via a vicinal diol group.
Marson et al. previously demonstrated that, starting from a trans-3,4-ditriflate substituted pyrrolidine ring 43 (Scheme 6), vicinal difluorination with TBAF can yield the corresponding trans-3,4-difluoropyrrolidine 44 in a good yield, 38,56 and this transformation has also been successful on the corresponding Cbz derivative. 57 However, treatment of 3,4-dihydroxyproline 19a with triflic anhydride already resulted in the formation of pyrrole 39a in a 64% yield. Hence, reaction with nonafluorobutanesulfonyl fluoride (NfF) 58 in combination with tetrabutylammonium difluorotriphenylsilicate (TBAT) 59 was attempted, as this process generates sulfonates in the presence of fluoride. Pleasingly, this led to 20a as the only observed 3,4difluoroproline diastereoisomer ( 19 F NMR analysis), with an enol sulfonate 46a as major byproduct along with its hydrolysis product, 3-oxoproline, as a minor, but persistent, impurity (not shown). Interestingly, no pyrrole side product was observed. While separation of all products was possible by HPLC, purification was considerably facilitated by subjecting the reaction mixture to NaBH 4 in order to reduce the 3-oxoproline byproduct to the corresponding alcohol (not shown). The regiochemistry of enol sulfonate 46a was established by means of a 2D HOESY NMR experiment. As a Fmoc-protecting group does not tolerate basic conditions, TBAT could not be used as a fluoride source for the NfF fluorination. Even when (diluted) Et 3 N·3HF/Et 3 N was employed as a fluoride source, 60 no difluorination was observed in the crude 19 F NMR and pyrrole 35 was the only product obtained from the reaction.
The reaction of the 3,4-diols 19a and 19c was also investigated with DAST (Scheme 7). With 19a, this led to a complex reaction mixture, in which the desired difluorinated 20a was clearly visible by 19 F NMR analysis, next to two minor byproducts, which presumably were monofluorinated hydroxyfluoroprolines 47a. As the desired 20a coeluted with another byproduct, identified as the corresponding cyclic sulfite, the crude reaction mixture was subjected to typical oxidation conditions, leading to the formation of the cyclic sulfate 48a. Isolation was now possible, leading to 20a in a 26% yield. According to MS analysis, the sulfite oxidation was not accompanied by possible 61 proline C5-oxidation to the corresponding lactam. Similarly, when this sequence was applied to the Fmoc-protected 19c, the desired 3,4difluoroproline 20c was also isolated, albeit in a reduced 14% yield.
Despite the low yield of this double deoxyfluorination process, the very short synthesis (only three steps from protected (4R)-hydroxyproline) was deemed an acceptable The Journal of Organic Chemistry Article and practical synthesis, as gram-scale quantities of 20a could readily be obtained.
Electrophilic Fluorination Strategy. With no straightforward access to other 3,4-dihydroxyproline diastereoisomers as substrates for bis-deoxyfluorination, investigations turned toward an electrophilic fluorination approach. Barraclough et al. had demonstrated the regioselective conversion of a 4ketoproline derivative to the corresponding silyl enolether, 62,63 which was used to stereoselectively introduce deuterium at C3. Hence, formation of the silyl enol ether 49a was achieved upon treatment of 29a, 64−66 synthesized by Dess−Martin periodinane oxidation of 26a in 94% yield (not shown), with LDA and TMSCl, and subsequently fluorinated with SelectFluor (Scheme 8). In our hands, this transformation proved to be low-yielding and was found difficult to optimize, leading to a mixture of isomers 27a/28a in a maximum 31% yield. Reduction of the 4-keto group led to a mixture of two separable fluorohydrin isomers, 50a and 51a, in a moderate yield. In the course of the optimization process, Ciulli and coworkers reported the synthesis of 27a/28a in 50% yield using this procedure, and of 50a/51a in 58% and 30% yields, respectively. 39 Interestingly, they also isolated a third diastereomer. Preliminary assignment of the stereochemistry at C β was based on the observed coupling constant between H α and H β , which was ∼6 Hz for 50a and ∼2 Hz for 51a. This value for 50a is in line with the coupling constant observed between H α and H β in 20a. In addition, for 50a, clear NOESY cross peaks were observed between H α and H β and between H β and H γ , suggesting all protons are on the same α-face of the pyrrolidine ring. This assignment was in agreement with the Ciulli work. 39 Deoxyfluorination of both 50a and 51a was achieved in a very good yield by treatment with the NfF and TBAT reagent combination. The stereochemistry of 24a was unambiguously assigned by means of X-ray analysis ( Figure 3).
With the new 3,4-difluoroproline derivatives 23a and 24a in hand, conversion to the required N-acetyl methyl ester derivatives 21 and 22 was carried out to allow conformational studies, including comparison with other, known, N-acetylated fluoroproline methyl esters. 10,13,17,67 Hence (Scheme 9), the benzyl-protecting group was removed by hydrogenolysis, and the N-Boc group by treatment with methanolic HCl. These conditions also simultaneously effected methyl ester formation. Finally, the amine groups were converted to their corresponding N-acetyl derivatives 21 and 22.
It was possible to obtain single crystals of 21, and crystallographic analysis ( Figure 4) provided unambiguous proof of its relative stereochemistry.
Conformational and Kinetic Analyses. The experimental cis/trans ratios in chloroform and water, the experimental cis/trans isomerization rate constants in water, and the DFTcalculated pucker preferences for the N-Ac-X-OMe model compounds of proline, the (3S,4R)-, (3R,4R)-, and (3S,4S)-3,4-difluorinated prolines and their monofluorinated progenitors are reported in Table 3. The entries are organized according to pucker preference. The data for the (3S,4R)variant 56 has been reported and discussed previously, 34 but are included in Table 3 for the sake of completeness. In the following discussion, the term "bias" assumes the conformational preference of the nonfluorinated N-acetyl proline methyl ester as a reference.
The amide cis/trans ratios in both chloroform and water of the 3,4-difluorinated proline 21 are very similar to those of each of their monofluorinated progenitors 52 and 53. For 22, the ratios are closer to those of (4S)-fluoroproline 54 than the (3S)-derivative 55.
The cis/trans isomerization rates (represented here by k ex = k cis/trans + k trans/cis ) typically increase with an increasing number of fluorine substitutions, mostly due to the electron-withdrawing effect of the fluorine atoms decreasing the double bond character of the amide bond. 11 As expected, both the (3S,4S)-and (3R,4R)-difluorinated variants, 21 and 22, indeed show higher isomerization rates than their monofluorinated progenitors. Interestingly, the (3R)-variant 53 has a markedly higher isomerization rate than all other monofluorinated prolines, 29 and even exchanging faster than the (3R,4R)difluorinated variant 22. This remarkable acceleration by fluorination at the 3-position with this stereochemistry is retained when combined with fluorination at the 4-position, resulting in even higher isomerization rates for the (3S,4S)variant 21. The isomerization rate for 21 is also much higher than that of the previously described (3S,4R)-and (4,4)difluorinated variants. 34 Figure 3. X-ray structure of (3R,4R)-3,4-difluoroproline 24a. Thermal ellipsoids drawn at the 50% probability level.

The Journal of Organic Chemistry
Article Finally, the calculated ratios between C γ endo and C γ exo puckers using DFT with chloroform or water as an implicit solvent are provided (Table 3). Unmodified proline has a higher preference for the C γ endo than the C γ exo pucker. 9 Both (4S)-and (3S)-fluoroprolines, 54 and 55, strongly bias these pucker ratios to the C γ endo form, with negligible C γ exo pucker populations, both in chloroform and water. 21 As expected, the (3R,4R)-difluoroproline variant 22 is heavily biased to the C γ endo pucker as well, with essentially the same C γ endo/C γ exo ratio as that of its (3S)-and (4S)-progenitors. The (4R)-and (3R)-fluoroprolines, 52 and 53, are biased to the C γ exo pucker relative to Pro, albeit to different degrees. Where the (4R)variant 52 shows a similar C γ exo bias in both solvents and for both trans and cis forms, the cis rotamer of the (3R)-variant 53 shows a high C γ exo bias in chloroform, but a low bias in water. The (3S,4S)-difluorinated proline 21 shows a bias to the C γ exo pucker in the same order of magnitude as its progenitors. Interestingly, especially in the cis rotamer, the C γ exo pucker is highly populated in both solvents, even higher than in its trans rotamer and than in its progenitors.
Experimental verification of these computational results can in principle occur via analysis of vicinal scalar couplings. Unfortunately, 3 J FF couplings are known not to be practically exploitable to assess the dihedral angle, 69 while quantitatively calculating the ring pucker from experimental 3 J HF and 3 J HH couplings was in our hands found not to be reliable due to the limited accuracy of Karplus relations for difluorinated fivemembered pyrrolidine rings. Instead, these couplings can  Ac-Pro-OMe n/a n/a n/a n/a 8.9 2.6 8.8 4.7 (3S,4R)-56 n/a 7.4 n/a 13.7 7.9 n/a 7.3 n/a (3R)-53 exo n/a 25.9 n/a 28.2 5.1 n/a 4.8 n/a (4R)-52 exo n/a n/a n/a n/a 8.7 8.2 7.8 10.1 (3S,4S)-21 exo n/a 27.3 n/a 29.8 5.2 n/a 5.0 n/a (3S)-55 endo 19.8 n/a 13.7 n/a n/a <1.0 d n/a 1.0 (4S)-54 endo n/a n/a n/a n/a 9.7 <1.0 d e e (3R,4R)-22 endo 21.2 n/a 24.4 n/a n/a <1.0 d n/a <1.0 d a1 H− 1 H couplings measured using PSYCHEDELIC. 70 1 H− 19 F couplings were read from the 1D 1 H spectrum on the H α proton. b Bias refers to the conformational preference of the nonfluorinated N-acetyl proline methyl ester as a reference. c "Cisoid" indicates that the coupled atoms are on the same side of the proline ring, whereas "transoid" indicates that the coupled atoms are on different sides of the ring. d Value smaller than signal line width. e Degenerate H β chemical shifts. Individual couplings could not be extracted.

The Journal of Organic Chemistry
Article qualitatively be compared to those of the monofluorinated progenitors (Table 4), bearing in mind that the different fluorine substitution patterns may significantly influence the Karplus relation. The (4R)-and (4S)-monofluoroprolines, which are established as strongly biased to, respectively, C γ exo and C γ endo, clearly display distinct transoid 3 J H α H β coupling constants of 8.2 cis /10.1 trans Hz and <1.0 Hz, respectively, implying this coupling provides a sensitive measure for the endo/exo ratio. Both the similar small magnitude of this coupling in (3S)-monofluoroproline, known to have a pronounced C γ endo pucker, 21 and the larger values found for proline (2.6 cis /4.7 trans Hz), consistent with intermediate endo/exo ratios and a higher endo population in the cis-form, confirm the relevance of 3 J H α H β coupling constants for a qualitative analysis of a fluorinated proline ring pucker. Hence, given the (3R,4R)-difluorinated variant 22 also shows a small 3 J H α H β coupling value of <0. 5 Hz, its calculated preference for a C γ endo pucker is consistent with these experimental data.
In contrast, the cisoid 3 J H α H β coupling constants of the (4R)and (4S)-fluoroprolines and proline show similar values of 8.7 cis /7.8 trans Hz, 9.7 Hz, and 8.9 cis /8.8 trans Hz, respectively, implying this coupling is not very sensitive to the endo/exo ratio. Indeed, both the (3R)-fluoroproline, known to prefer an exo pucker, 21 and the (3S,4S)-difluoroproline show lower cisoid 3 J H α H β couplings of 5.1/4.8 Hz and 5.2/5.0 Hz, respectively, which suggests the fluorine substitution pattern is in this case the most significant factor determining the value. Nevertheless, the similarity of both the 3 J H α H β and 3 J H α F β couplings observed for the (3R)-and (3S,4S)-variants suggests both fluoroprolines have mostly similar endo/exo ratios. In addition, these couplings differ significantly with those of the (3S,4R)-variant, which is expected given the latter displays virtually no pucker preference.
The clear C γ exo pucker bias observed for (3S,4S)difluorinated proline 21 in solution by NMR is also observed in its crystal structure ( Figure 4). A single crystal of 22 was not obtained, but the C γ endo pucker bias of the (3R,4R)difluoroproline ring could be observed in the crystal structure of its N-Boc-protected precursor 24a ( Figure 3). It should be noted that the packing of molecules in the solid state, and their resulting conformations, is determined from the sum of a multitude of inter-and intramolecular interactions, and often deviates from the conformation in solution, which in turn is typically solvent-dependent. With this caveat in mind, the observed conformations in the crystal structures strongly suggest that the 3,4-difluorination instills the expected conformational bias.

■ DISCUSSION
The potential of fluorinated prolines as tools for protein research has a long track record. Next to the well-known example of collagen, stabilized forms of proteins such as barstar, 13 ubiquitin, 71 Trp cage mini protein, 72 and GFP 73 incorporating 4-fluoroprolines were obtained with the C 4stereochemistry selected to reinforce the pucker observed in the native protein. Both 3-and 4-monofluorinated prolines have been used to probe the effect of β-turn stability on the self-assembly of elastin peptide mimics. 68 Accelerated peptide folding, as a consequence of the accelerated cis/trans kinetics, was observed when fluoroprolines were integrated in thioredoxin (Trx), 74 β2-microglobulin (β2m), 75 and ribonu-clease (RNase) A. 76 Fluorinated prolines have also been used to reveal the relevance of a proline ring pucker in ribosomal peptide synthesis. 77,78 The extended range of cis/trans isomerization kinetics offered by the 3,4-difluoroprolines, in conjunction with either a bias to trans and the C γ exo pucker, to cis and the C γ endo pucker, or a similar structural preference to proline, clearly will be of interest within such studies, allowing us to deconvolute the roles of ring pucker and cis/ trans preferences from isomerization kinetics.
Recently, Bernardes and Corzana and co-workers used a rational Pro-to-FPro substitution to stabilize an antigen− antibody complex. 79 As a result of its proximity to a highly electronegative fluorine, the polarization of a nearby CH bond was increased. This led to an enhanced CH−π interaction, which stabilized the antigen−antibody complex. A similar improved CH−π interaction has been observed between a fluoroproline-modified phosphopeptide and the WW domain of Pin1. 80 Clearly, 3,4-difluorinated proline analogues, especially with a 3,4-cis stereochemistry, will be of great interest in that regard, as enhanced C−H polarization and thus enhanced CH−π interactions can be expected. 81 Regarding the use of fluoroprolines as 19 F NMR reporters, the simultaneous fluorination at the 3-and 4-positions provides for very distinct chemical shifts compared to the monofluorinated progenitors. The experimental 19 F chemical shifts and n J FF coupling constants for the N-Ac-X-OMe model compounds the (4,4)-difluoroproline, (3S,4R), (3R,4R)-, and (3S,4S)-3,4-difluorinated prolines, and their monofluorinated progenitors are shown in Table 5. For all 3,4-difluoroprolines, the homonuclear coupling constant between the vicinal 19 F nuclei is small, as opposed to that of the geminal difluorinated (4,4)-variant. This property is very useful for advanced 19 F NMR experiments, as it minimizes any potential complications from J modulation during spin−echo pulse sequences, or from second-order effects, which is an issue in geminal difluorinated prolines. 30 In addition, the 3,4-difluorinated derivatives have very distinct 19 F chemical shift values compared to their monofluorinated progenitors, even though they possess similar structural properties. The 3,4-difluoroprolines can thus be used complementary to the monofluoroprolines for 19 F NMR purposes, allowing for the design of combinatorial incorporation schemes aimed at studying poly proline-and prolinerich sequences, due to maximum chemical shift dispersion between these residues, but with minimal complications from homonuclear couplings.
The Journal of Organic Chemistry As part of a program to expand the scope of available fluorinated prolines, we report here in full the effective syntheses of three 3,4-difluorinated proline analogues (as summarized in Scheme 10). In addition, we report the first conformational characterization (trans/cis ratios and isomerization kinetics, and ring pucker preferences) of the (3R,4R)and (3S,4S)-3,4-difluoroproline analogues. The (3S,4R)-difluorinated proline derivative could not be synthesized directly from 3,4-dehydroproline, or from the 3,4epoxyproline derivative, with the former being unreactive under conditions of alkene difluorination or halofluorination and the latter typically suffering from aromatization, leading to pyrrole derivatives. However, a direct bis-deoxyfluorination strategy with the easily accessible 3,4-dihydroxyproline as a substrate led to the desired target using both NfF and DAST, with the former giving the highest yield when N-Boc was used as a protecting group and the latter suitable with an N-Fmocprotecting group. Yields were low (26% and 14%, respectively), but as only two transformations were required from the protected 3,4-dehydroproline, gram quantities are easily available. In this context, we report that the direct synthesis of Fmoc-protected 3,4-dehydroproline from the corresponding 4-hydroxyproline is possible using the one-pot Grieco elimination procedure, in contrast to the usually employed basic conditions. The (3R,4R)-and the novel (3S,4S)-difluorinated proline derivatives were synthesized using a two-step fluorination strategy: the first being an electrophilic fluorination starting from protected 4-ketoproline and the second by a DASTmediated deoxyfluorination after 4-ketoreduction. Hence, starting from 26a, the (3S,4S)-and (3R,4R)-N-Boc 3,4difluoroproline benzyl esters were obtained in an overall yield of 9% for 23a and 3% for 24a, in a combined 5 steps (with three common steps before diastereomer separation). It may be pointed out that the linear 10-step Fleet synthesis 37 of (3R,4R)-N-Bn difluoroproline methyl ester 17 (cf. Scheme 1) has a higher overall yield (24%), although this delivers a single diastereomer only. In addition, Ciulli reported higher yields for the electrophilic fluorination step. 39 X-ray crystallographic analysis allowed unambiguous determination of the relative configuration of the obtained 3,4-difluoroprolines.
Due to the opposing conformational effects of each individual fluorine in the (3S,4R)-difluorinated proline derivative, this analogue has previously been described as having a minimal conformational bias to proline. 34 In contrast, it is shown here that a combination of 3-and 4-fluorine substitutions with similar preorganizing effects results in 3,4difluorinated proline derivatives with similar conformational preferences as monofluorinated prolines. While the (3R,4R)difluorinated proline derivative resembles most closely the (4S)-fluoroproline, the (3S,4S)-difluorinated proline derivative resembles the (4R)-fluoroproline, though with a somewhat higher preference for a C γ exo pucker in its cis rotamer. Given the distinct 19 F chemical shifts of both 3,4-difluorinated derivatives to their monofluorinated progenitors, they will be of interest for multiresidue fluorine-labeling strategies, for instance, in the study of repetitive or low-complexity protein sequences, where similar conformational preorganizing effects are desired, but distinct residue-specific 19 F NMR chemical shifts are needed.
A clearer difference between the 3,4-difluoprolines and their monofluorinated progenitors is the faster amide rotamer isomerization rates. This is expected given the larger electron-withdrawing effect of two fluorines compared to that of one. Especially the (3S,4S) variant shows a remarkably high isomerization rate, higher than any previously described difluorinated variant. These new variants will thus be very useful toward studying the role of Xaa-Pro cis/trans isomerization kinetics for biological function, 82 protein folding, 83 or amyloid assembly. 75 Applications of the 3,4-difluoroprolines are in progress and will be reported in due course, as are deeper investigations on revealing the structural origins of their conformational properties and cis/trans isomerization kinetics.  The Journal of Organic Chemistry Article mass spectrometer via a Dionex Ultimate 3000 autosampler and uHPLC pump and eluted in 5 min at 0.6 mL min using a gradient of 20% acetonitrile (0.2% formic acid) to 100% acetonitrile (0.2% formic acid) through an Acquity UPLC BEH C18 (Waters) 1.7 μm 50 mm × 2.1 mm column. High-resolution mass spectra were recorded using positive ion electrospray ionization. 1 H, 19 F, and 13 C NMR spectra were recorded at room temperature on a Bruker Ultrashield 400 or 500 MHz spectrometer. 1 H and 13 C chemical shifts (δ) are quoted in ppm relative to residual solvent peaks as appropriate. 19 F spectra were externally referenced to CFCl 3 . The coupling constants (J) are given in hertz (Hz). The NMR signals were designated as follows: s (singlet), d (doublet), t (triplet), q (quartet), quin (quintet), sxt (sextet), spt (septet), m (multiplet), or a combination of the above. For all compounds, a detailed peak assignment was performed through the combined use of HSQC, HMBC, NOESY, and COSY NMR experiments.
Reaction of 38b with AgF (Scheme 5). To a solution of 38b (99.0 mg, 0.289 mmol) in nitromethane (5.0 mL) was added AgF (183.5 mg, 1.447 mmol), and the resulting mixture was stirred at room temperature. After 2 h, no conversion of the starting material was observed on TLC and subsequently the mixture was heated at 45°C. After 15 h, the mixture was filtered through Celite and the solvent evaporated in vacuo to yield 10b (77.0 mg, quant) as a clear oil.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.8b02920. 1 H, 13 C, and 19 F NMR spectra of all novel compounds, chiral HPLC chromatograms (for 3,4-dehydroproline synthesis), J value analysis for epoxides 9a,b,d and 10a,b,d, X-ray crystallographic data for 9b, 21, and 24a, computational data of the proline conformers of 21, 22, 52−55, and Ac-Pro-OMe in CHCl 3 and water calculated by DFT including Gibbs free energies, electronic energy values, and Cartesian coordinates, and general NMR conditions for conformational and kinetic analysis (PDF) Crystallographic data for compound 21 (CIF) Crystallographic data for compound 9b (CIF) Crystallographic data for compound 24a (CIF) Notes