Regioselective Reduction of 1H-1,2,3-Triazole Diesters

Regioselective reactions can play pivotal roles in synthetic organic chemistry. The reduction of several 1-substituted 1,2,3-triazole 4,5-diesters by sodium borohydride has been found to be regioselective, with the C(5) ester groups being more reactive towards reduction than the C(4) ester groups. The amount of sodium borohydride and reaction time required for reduction varied greatly depending on the N(1)-substituent. The presence of a β-hydroxyl group on the N(1)-substituent was seen to have a rate enhancing effect on the reduction of the C(5) ester group. The regioselective reduction was attributed to the lower electron densities of the C(5) and the C(5) ester carbonyl carbon of the 1,2,3-triazole, which were further lowered in cases involving intramolecular hydrogen bonding.


Introduction
Sodium borohydride (NaBH 4 ) is a mild reducing agent used widely for converting aldehydes and ketones to alcohols [1]. It has been reported that some esters can also be reduced by NaBH 4 in alcohols, although higher proportions of the reagent and longer reaction times are required than for reduction of aldehydes and ketones [2,3]. Based on these reports, we developed an interest in the opportunity to prepare and selectively reduce heterocyclic diesters, which became our goal. Triazole and other five-membered ring heterocycles having vicinal diesters became available following the pioneering reports by Huisgen of 1,3-dipolar cycloaddition of a 1,3-dipolar species with an activated disubstituted alkyne [4,5]. The 1,4-disubstituted triazoles made from terminal alkynes are available by the regioselective method known as CuAAC (copper(I) azide-alkyne cycloaddition) [6,7], whereas regioselective synthesis of 1,5-disubstituted triazoles employs other catalysts [8][9][10].
Reduction of both ester groups of oxazole 2,4-diesters has been reported using NaBH 4 in ethanol at room temperature [11]. Dimethyl pyrrole-2,4-dicarboxylate was selectively reduced to form ethyl 4-(hydroxymethyl)-1H-pyrrole-2-carboxylate (A of Scheme 1) using diisobutylaluminum hydride (DIBAH). The selectivity of the reduction was not as effective for dimethyl pyrrole-3,4-dicarboxylates and did not occur for dimethyl furan-2,4-dicarboxylate, suggesting a role of the pyrrole nitrogen in complexing with the reagent and a neighboring carbonyl oxygen [12]. Dimethyl pyridine-2,5-dicarboxylate has been selectively reduced at C(2) using NaBH 4 /CaCl 2 in ethanol/THF to form methyl 6-(hydroxymethyl)-3pyridinecarboxylate (B of Scheme 1) [13]. The distinction between the relative reactivity of unsymmetrical diesters is based upon differences in electronic factors [14], steric factors [15], or a combination of these factors [16]. Two research groups have shown there to be a difference in reactivity of bulky tert-butyl esters and methyl esters [17,18]. One of these employed sodium triacetoxyborohydride [NaBH(OAc) 3 ] [18]. Selective reduction of diesters to form hydroxy esters using lithium borohydride (LiBH 4 ), relying upon differences in steric effects, has been reported [19]. Another example using DIBAH was for selective reduction of diesters by formation of an aldehyde monoester [12,20]. Esters were relying upon differences in steric effects, has been reported [19]. Another example usi DIBAH was for selective reduction of diesters by formation of an aldehyde monoes [12,20]. Esters were reduced using NaBH4-CaCl2 [13,21] or by NaBH4 with added sodiu methoxide (NaOMe) via sodium monomethoxyborohydride (NaBH3OCH3), wherein t authors propose that the active reagent of NaBH4 in methanol is indeed (NaBH3OCH [22]. The reduction of nicotinate esters with a large excess of NaBH4 at reflux [23] a nicotinate esters having other electron-withdrawing groups by an excess of reductant room temperature [24] have been observed. Prior work on the NaBH4 reduction of dieth 2,4-pyrroledicarboxylate took place regioselectively at the C(4) ester group. Diethyl 2 pyrroledicarboxylate gave predominate conversion to the diol [12]. A: Previous work [12] B: Previous work [13] C: This work

Results
We elected to pursue work directed towards possible selective reduction of fivemembered ring heteroaromatic diesters. As representatives of these, neither of the ester groups of dimethyl imidazole-4,5-dicarboxylate were reduced by NaBH 4 in MeOH/THF, but chemoselective reduction [25] was achieved for the N(1) ketone of dimethyl 1-(2-oxo-2-phenylethyl)-1H-1,2,3-triazole-4,5-dicarboxylate (1a) (C of Scheme 1) to form dimethyl 1-(2-hydroxy-2-phenylethyl)-1H-1,2,3-triazole-4,5-dicarboxylate (2a), followed later by regioselective reduction to form methyl 5-(hydroxymethyl)-1-(2-hydroxy-2-phenylethyl)-1H-1,2,3-triazole-4-carboxylate (3a). In our earliest experiment for reaction of 1a, we anticipated the reduction with NaBH 4 to reduce the ketone, but not the ester. The ketone was reduced, but some ester reduction was also observed. Adding more NaBH 4 and waiting for the reaction to finish gave 3a in good yield, with the ester group at C(4) being unreactive, although initially we did not know which of the ester substituents had been reduced. This turned out to be an excellent example of a regioselective process. We chose to investigate the scope of the selectivity of diester reduction of 1a, along with related compounds. Ester reduction was understandably more likely for 1a than for the more electron-rich imidazole 4,5-diesters. An important issue was how a hydroxy group as part of the N(1) substituent might influence the outcome of the ester reduction or the reduction rate of 1a. We posited that a hydroxy group might be required in order to observe selective reduction of 1a, which turned out not to be the case.
A detailed study of a series of the 1H-1,2,3-triazole diesters revealed that a carbonyl function at N(1) was not necessary for reduction of the C(5) ester, but reduction might be enhanced via six-membered lactone formation between an OH group on an N(1) substituent and the C(5) ester prior to reduction. Five-membered lactones are known to be reduced by NaBH 4 [26]. We have not investigated the possibility of lactone formation. However, evidence was not seen for formation of a reactive borate complex [27]. The C(5) ester group was found to be much more reactive with NaBH 4 than a C(4) ester in the presence or absence of a carbonyl group (or a hydroxy group) at N(1). The scope of the selective ester reduction of triazole diesters at C(5) and the factors responsible for the selectivity were examined by reduction of a number of triazole esters and diesters.
unreactive, although initially we did not know which of the ester substituents had been reduced. This turned out to be an excellent example of a regioselective process. We chose to investigate the scope of the selectivity of diester reduction of 1a, along with related compounds. Ester reduction was understandably more likely for 1a than for the more electron-rich imidazole 4,5-diesters. An important issue was how a hydroxy group as part of the N(1) substituent might influence the outcome of the ester reduction or the reduction rate of 1a. We posited that a hydroxy group might be required in order to observe selective reduction of 1a, which turned out not to be the case.
A detailed study of a series of the 1H-1,2,3-triazole diesters revealed that a carbonyl function at N(1) was not necessary for reduction of the C(5) ester, but reduction might be enhanced via six-membered lactone formation between an OH group on an N(1) substituent and the C(5) ester prior to reduction. Five-membered lactones are known to be reduced by NaBH4 [26]. We have not investigated the possibility of lactone formation. However, evidence was not seen for formation of a reactive borate complex [27]. The C(5) ester group was found to be much more reactive with NaBH4 than a C(4) ester in the presence or absence of a carbonyl group (or a hydroxy group) at N(1). The scope of the selective ester reduction of triazole diesters at C(5) and the factors responsible for the selectivity were examined by reduction of a number of triazole esters and diesters.
Reduction of 1a-f with NaBH4 in methanol afforded 2a and 2b in a chemoselective process after just 3 min ( Table 1). The N(1) ester group of dimethyl 1-(2-ethoxy-2-oxoethyl)-1H-1,2,3-triazole-4,5-dicarboxylate (1c) was reduced faster than the C(5) ester, but not fast enough to be synthetically useful. Additional reaction time allowed all C(5) esters to undergo reduction. The products 3a-f were formed in the times indicated, along with the number of equivalents of NaBH 4 needed for the starting diesters to be fully converted to the C(4) esters 3a-f [28,29]. Notably, diesters 1d-f, containing only alkyl or aralkyl functionality at N(1), required longer times for reduction. The structures of the products were elucidated using 1D and 2D-NMR spectroscopy. As it was not clear from 1D spectroscopy alone which ester group had been reduced, 2D-NMR HMBC spectroscopy was necessary to help elucidate the product structures of 3a-f [29]. The structure of the reduction product 3f was further confirmed by X-ray crystallography, as shown in Figure 1.
Additional reaction time allowed all C(5) esters to undergo reduction. The products 3a-f were formed in the times indicated, along with the number of equivalents of NaBH4 needed for the starting diesters to be fully converted to the C(4) esters 3a-f [28,29]. Notably, diesters 1d-f, containing only alkyl or aralkyl functionality at N(1), required longer times for reduction. The structures of the products were elucidated using 1D and 2D-NMR spectroscopy. As it was not clear from 1D spectroscopy alone which ester group had been reduced, 2D-NMR HMBC spectroscopy was necessary to help elucidate the product structures of 3a-f [29]. The structure of the reduction product 3f was further confirmed by Xray crystallography, as shown in Figure 1. Triazoles with substituents at the C(5) position were prepared using ruthenium catalysis [10] by dichloro(pentamethylcyclopentadienyl)ruthenium polymer (RuCl2CP*)x producing 1-substituted 1H-1,2,3-triazole-5-esters 4a-d, as shown in Table 2. The reduction of 4a-d with NaBH4 in methanol afforded 5a-d. The reduction of 4d required more NaBH4 and more time than 4a-c, similar to the conditions needed for the reduction of 1f. The absence of an electron-withdrawing ester group at C4 for 4a-d also slowed the reduction rate of the C5 ester group compared to the rates of 1a-f. Triazoles with substituents at the C(4) position were prepared using copper catalysis [6,7] producing methyl 1-substituted 1H-1,2,3-triazole-4-carboxylates 6a-d, as shown in Table 3. The reduction of 6a-d with NaBH4 in methanol afforded 7a-b and 8c-d. It was observed that the reduction of the N(1) carbonyl for 6a-b occurred at an acceptable rate, Triazoles with substituents at the C(5) position were prepared using ruthenium catalysis [10] by dichloro(pentamethylcyclopentadienyl)ruthenium polymer (RuCl 2 CP*) x producing 1-substituted 1H-1,2,3-triazole-5-esters 4a-d, as shown in Table 2. The reduction of 4a-d with NaBH 4 in methanol afforded 5a-d. The reduction of 4d required more NaBH 4 and more time than 4a-c, similar to the conditions needed for the reduction of 1f. The absence of an electron-withdrawing ester group at C4 for 4a-d also slowed the reduction rate of the C5 ester group compared to the rates of 1a-f. Table 2. Preparation of methyl 1-substituted-1H 1,2,3-triazole-5-carboxylates using dichloro(pentamethylcyclopentadienyl) ruthenium polymer (RuCl 2 CP*) x catalyst and their reduction by NaBH 4 .
Additional reaction time allowed all C(5) esters to undergo reduction. The products 3a-f were formed in the times indicated, along with the number of equivalents of NaBH4 needed for the starting diesters to be fully converted to the C(4) esters 3a-f [28,29]. Notably, diesters 1d-f, containing only alkyl or aralkyl functionality at N(1), required longer times for reduction. The structures of the products were elucidated using 1D and 2D-NMR spectroscopy. As it was not clear from 1D spectroscopy alone which ester group had been reduced, 2D-NMR HMBC spectroscopy was necessary to help elucidate the product structures of 3a-f [29]. The structure of the reduction product 3f was further confirmed by Xray crystallography, as shown in Figure 1. Triazoles with substituents at the C(5) position were prepared using ruthenium catalysis [10] by dichloro(pentamethylcyclopentadienyl)ruthenium polymer (RuCl2CP*)x producing 1-substituted 1H-1,2,3-triazole-5-esters 4a-d, as shown in Table 2. The reduction of 4a-d with NaBH4 in methanol afforded 5a-d. The reduction of 4d required more NaBH4 and more time than 4a-c, similar to the conditions needed for the reduction of 1f. The absence of an electron-withdrawing ester group at C4 for 4a-d also slowed the reduction rate of the C5 ester group compared to the rates of 1a-f. Triazoles with substituents at the C(4) position were prepared using copper catalysis [6,7] producing methyl 1-substituted 1H-1,2,3-triazole-4-carboxylates 6a-d, as shown in Table 3. The reduction of 6a-d with NaBH4 in methanol afforded 7a-b and 8c-d. It was observed that the reduction of the N(1) carbonyl for 6a-b occurred at an acceptable rate, Ph(CH) 2 CH 2 13 5d Ph(CH) 2 CH 2 80 10. 5 21 Triazoles with substituents at the C(4) position were prepared using copper catalysis [6,7] producing methyl 1-substituted 1H-1,2,3-triazole-4-carboxylates 6a-d, as shown in Table 3. The reduction of 6a-d with NaBH 4 in methanol afforded 7a-b and 8c-d. It was observed that the reduction of the N(1) carbonyl for 6a-b occurred at an acceptable rate, but the reduction of the C(4) position ester group of 8c-d was slow and required multiple days and a large excess of NaBH 4 , as shown in Table 3. Table 3. Preparation and reduction of methyl 1-substituted 1H-1,2,3-triazole carboxylates.
Molecules 2021, 26, x FOR PEER REVIEW 5 of 14 but the reduction of the C(4) position ester group of 8c-d was slow and required multiple days and a large excess of NaBH4, as shown in Table 3. Observations based on the data in Tables 1-3 are: 1. β-Ketone substituents at N(1) undergo rapid reduction as expected. 2. β-Ester substituents at N(1) undergo reduction more rapidly than C(5) ester substituents. 3. Reduction of C(5) ester substituents is facilitated by a hydroxy group at the β carbon of N(1) substituents, whereas C(5) esters having only aralkyl substituents at N(1) react slower with NaBH4. 4. A ketone substituent at N(1) facilitates reduction of a C(5) ester more than an ester at N(1) because the ketone is reduced more rapidly, exposing a facilitating hydroxy group for reduction of a C(5) ester. 5. Regardless of the substituents at N(1), C(5) ester substituents are reduced more rapidly than C(4) ester substituents. 6. The presence of a C(4) ester substituent enhances the rate of reduction of a C(5) ester substituent. Stronger reducing agents such as LiAlH4 should be chosen when complete reduction of triazole and other heteroaromatic diesters is desired.
Compounds 1a-f and 4a-d each gave reduction of the C(5) ester. The relative rates of reduction varied considerably with the nature of the N(1) substituents. An N(1) substituent containing a ketone was reduced by NaBH4 to the corresponding alcohol within a few minutes, as observed for products 2a-b. The resulting hydroxy substituent situated at the δ-position relative to the ester carbonyl undergoing reduction showed a rate-enhancing effect on the ester at C(5), such as in the reduction of 1a-b to 3a-b as compared to the reduction rates of 1d-f to 3d-f. When N(1) contained a keto or ester group, as for 1a-c and 4a-c, reduction of that keto or ester group to an alcohol was observed. The rate-enhancement was lower for the reduction of 1c because the N(1) ester group was reduced more slowly than the N(1) ketones of 1a and 1b, also observed for the reduction rate of 6b compared to that of 6a. The reduction rate of the C(5) ester group was significantly slower when the N(1) substituent contained only alkyl, alkenyl, or aralkyl groups as in 1d-f and 4d. The times necessary to complete the reactions in this study provide a clear distinction between triazole C(5) esters bearing an N(1) substituent containing a β-carbonyl group (1a-c and 4a-c) and those having N(1) alkyl or alkenyl substituents (1d-f and 4d). All ketone and ester groups of 1ac and 4a-c were reduced to alcohols. The resulting alcohols 2a-c and 5a-c gave reduction of the C(5) ester groups more readily than C(5) esters 1d-f and 4d.
Our hypothesis to explain the difference in reactivity of the C(4) and C(5) diesters is that there are consistent differences in electron density at the ester carbonyl carbons, leading to preferred reaction of the C(5) carbonyl group of the C(5) esters. Molecular modeling studies [30] furnished insight into the preferential reduction of the C(5) esters, which indicated that the calculated electron density at the C(5) carbonyl carbon was lower than at the C(4) carbonyl carbon for 1a-f and 4a-d, supporting the observation that NaBH4 gave reduction of the C(5) ester group in preference to the C(4) ester group. Plots of electron Observations based on the data in Tables 1-3 are: 1. β-Ketone substituents at N(1) undergo rapid reduction as expected. 2. β-Ester substituents at N(1) undergo reduction more rapidly than C(5) ester substituents. 3. Reduction of C(5) ester substituents is facilitated by a hydroxy group at the β carbon of N(1) substituents, whereas C(5) esters having only aralkyl substituents at N(1) react slower with NaBH 4 . 4. A ketone substituent at N(1) facilitates reduction of a C(5) ester more than an ester at N(1) because the ketone is reduced more rapidly, exposing a facilitating hydroxy group for reduction of a C(5) ester. 5. Regardless of the substituents at N(1), C(5) ester substituents are reduced more rapidly than C(4) ester substituents. 6. The presence of a C(4) ester substituent enhances the rate of reduction of a C(5) ester substituent. Stronger reducing agents such as LiAlH 4 should be chosen when complete reduction of triazole and other heteroaromatic diesters is desired.
Compounds 1a-f and 4a-d each gave reduction of the C(5) ester. The relative rates of reduction varied considerably with the nature of the N(1) substituents. An N(1) substituent containing a ketone was reduced by NaBH 4 to the corresponding alcohol within a few minutes, as observed for products 2a-b. The resulting hydroxy substituent situated at the δposition relative to the ester carbonyl undergoing reduction showed a rate-enhancing effect on the ester at C(5), such as in the reduction of 1a-b to 3a-b as compared to the reduction rates of 1d-f to 3d-f. When N(1) contained a keto or ester group, as for 1a-c and 4a-c, reduction of that keto or ester group to an alcohol was observed. The rate-enhancement was lower for the reduction of 1c because the N(1) ester group was reduced more slowly than the N(1) ketones of 1a and 1b, also observed for the reduction rate of 6b compared to that of 6a. The reduction rate of the C(5) ester group was significantly slower when the N(1) substituent contained only alkyl, alkenyl, or aralkyl groups as in 1d-f and 4d. The times necessary to complete the reactions in this study provide a clear distinction between triazole C(5) esters bearing an N(1) substituent containing a β-carbonyl group (1a-c and 4a-c) and those having N(1) alkyl or alkenyl substituents (1d-f and 4d). All ketone and ester groups of 1a-c and 4a-c were reduced to alcohols. The resulting alcohols 2a-c and 5a-c gave reduction of the C(5) ester groups more readily than C(5) esters 1d-f and 4d.
Our hypothesis to explain the difference in reactivity of the C(4) and C(5) diesters is that there are consistent differences in electron density at the ester carbonyl carbons, leading to preferred reaction of the C(5) carbonyl group of the C(5) esters. Molecular modeling studies [30] furnished insight into the preferential reduction of the C(5) esters, which indicated that the calculated electron density at the C(5) carbonyl carbon was lower than at the C(4) carbonyl carbon for 1a-f and 4a-d, supporting the observation that NaBH 4 gave reduction of the C(5) ester group in preference to the C(4) ester group. Plots of electron densities of the C(5) carbonyl and C(5) carbons of 1a-f and separately 4a-d showed linear correlations with R-squared values of >0.9. When intramolecular hydrogen bonding was included in the calculations and compared to the data for no hydrogen bonding, a lower electron density was observed at C(5) and a much lower electron density at the C(5) Electron density differences and, where relevant, intramolecular hydrogen bonding, are postulated to explain the faster rates of reduction of the C(5) esters. Other possible explanations of a higher rate of reactivity at C(5) such as for 2a and 2b are the formation of a lactone intermediate [26] or a borate complex intermediate [27]. These explanations are not possible for 1d-f, which lack reducible carbonyl groups in substituents at N(1) such as in 1a-c. Similar observations were made for the modeling studies of 4a-d. The overall rates of reduction of 4a-d showed a decrease compared to those for 1a-f, indicating a favorable rate enhancement for reduction of a C(5) ester when an ester group is also present at C(4).

Conclusions
The regioselective NaBH 4 reduction of the C(5) 1H-1,2,3-triazole esters was found to be a dependable reaction that occurred with all 1H-1,2,3-triazole 4,5-diesters examined in this study. The 1H-1,2,3-triazole 5-esters were also reduced faster than the 1H-1,2,3triazole 4-esters. The results are supported by 2D-NMR and X-ray data. When a hydroxy substituent was available at the N(1) position by the initial reduction of a ketone or ester at N(1) of the triazole esters and diesters studied, reduction times were shorter for the C(5) esters of both triazole-4,5-diesters and triazole-5-esters. The lower electron densities found at the C(5) position for all of the 1H-1,2,3-triazole esters and diesters studied are posited as the enabling factor for reduction even in the cases when no hydroxy substituent is available at the N(1) position to assist in the regioselective reduction of the C(5) position ester. In cases where a hydroxy substituent becomes available by reduction of a group at N(1), the rate of reduction at C(5) is further increased. Intramolecular hydrogen bonding could be the reason for the high reactivity of C(5) esters where hydrogen bonding is relevant.

Materials and Methods
All starting materials were purchased from commercially available sources and used as obtained. All synthesized organic azides were stored at 0 • C until needed. All azides should be considered as hazardous and potentially explosive. Plastic or ceramic spoons were used when weighing solid azides. All synthesized organic azides were stored at 2-8 • C to ensure safety of potentially explosive material and to reduce degradation of the azide moiety. All reactions were performed in a ventilated hood. Thin layer chromatography (TLC) was performed on Agela Technologies aluminum-backed silica dioxide plates and products observed under 254 nm UV light. Flash column and radial chromatography (T-Squared Technology, Inc., 206 Lassen Dr., San Bruno CA 94066 USA) were performed with SiliCycle silica gel 60, 0.040−0.063 mm (230−400 mesh) using distilled ethyl acetate and distilled hexanes. Microwave-assisted synthesis was performed using a CEM Discover SP Microwave Synthesizer. NMR spectra (400 or 500 MHz for 1 H and 100 MHz for 13 C) were measured in CDCl 3 or DMSO-d 6 . Chemical shifts (δ) are given in ppm relative to the resonance of their respective residual solvent peak, CHCl 3 (7.27 ppm, 1 H; 77.16 ppm, the middle peak, 13 C). Multiplicities were described using the following abbreviations: s = singlet, d = doublet, t = triplet, q = quartet, dd = doublet of doublets, dt = doublet of triplets, m = multiplet. Regioselectivity of the reduction was determined by HSQC and HMBC correlation 2D-NMR analysis. FTIR experiments were performed on a Perkin Elmer

General Procedures for the Synthesis of Organic Azides
Metal azides are shock sensitive and should be handled with smooth edged, nonmetallic spoons or spatulas. Purification of initially isolated products is required to remove traces of azides.