Preparation of Acidic 5-Hydroxy-1,2,3-triazoles via the Cycloaddition of Aryl Azides with β-Ketoesters

Herein, a high-yielding cycloaddition reaction of β-ketoesters and azides to provide 1,2,3-triazoles is described. The reactions employing 2-unsubstituted β-ketoesters were found to provide 5-methyl-1,2,3-triazoles, whereas 2-alkyl-substituted β-ketoesters provided 5-hydroxy-1,2,3-triazoles (shown to be relatively acidic) in high yields and as single regioisomers. Several novel compounds were reported and characterized including long-chain 5-hydroxy-1,2,3-triazoles potentially bioisosteric to hydroxamic acids.


■ INTRODUCTION
1,2,3Ttriazoles are important scaffolds employed in medicinal chemistry, 1 catalysis, 2 materialscience, 3 and biology. 4 The electronic and physicochemical properties of triazoles bear a close similarity to those of the amide functionality and therefore can be classified as amide bioisosteres. 5,6 Coppercatalyzed alkyne−azide cycloaddition is the most accredited method to synthesize substituted 1,2,3-triazoles, 7−11 while the less well-known reactions of malonates or β-ketoesters with aromatic azides have become attractive alternatives as they do not require metal catalysts to proceed ( Figure 1). 12 β-Ketoesters react quickly with mild bases, providing highly reactive enolates that have a myriad of reported applications in organic chemistry. 13,14 Dimroth reported a cycloaddition of βketoesters and azides in 1902 15 where ethyl acetoacetate reacted with azidobenzene in the presence of sodium ethoxide to provide 5-hydroxytriazoles in poor yields. This material was properly identified; however, it was not characterized beyond the physical constants and, notably, their remarkable acidity was overlooked. More recently, Wang and co-workers reported the cycloaddition of β-ketoesters and azides, via organocatalysis, to yield 1,4-disubstituted 1,2,3-triazoles. 16 Intrigued by these reports 15, 16 where the same reagents gave rise to different products under similar basic conditions, and in a continuation of our studies on the reactivity of azides and enolates (Figure 1), 17 we decided to re-examine the cycloaddition of simple ketoesters and aromatic azides in the presence of organic bases.

■ RESULTS AND DISCUSSION
This work involved reacting a selection of pyridyl-or arylazides and β-ketoesters and showed that distinct triazole products could be obtained in high yields via a divergent reaction pathway that led to either disubstituted 1,2,3-triazoles or 5-hydroxy-1,2,3-triazoles depending upon the nature of the β-ketoester. These findings were subsequently exploited to prepare a family of 5-hydroxytriazoles, whose acidity has been shown to be even stronger (pK a = 4.20 for 5a) than those for analogous compounds (pK a = 5.14−6.21) reported by Sainas and co-workers. 18 Our observation of the relatively strong acidity of 5-hydroxytriazoles has the potential to be applied toward hydroxamic acid bioisosteres ( Figure 1).
Our investigation began with the cycloaddition reaction of pyridyl azides 2a−c with ethyl acetoacetate 1a in the presence of organic soluble bases, from which 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) gave the best yield. The reactions of pyridyl azides 2a and 2b with 1a furnished triazoles 3a and 3b in 80% and 72% isolated yields, respectively, whereas the reaction of pyridyl azide 2c, bearing an ortho-nitrogen, showed no conversion to 3c (Table 1).
Compound 3c, however, was obtained in average yields (50%) when a copper trifluoromethanesulfonate toluene complex (Cu(OTf) 2 ·C 6 H 5 CH 3 ) in dimethyl sulfoxide (DMSO) was added to the reaction mixture. The reaction of 1a and phenyl azide 4a ( Table 2, entry 1) gave the disubstituted triazole ester 3d 16 in an excellent isolated yield. In summary, the reaction of unsubstituted 1a with azides 2a−c (Table 1) or 4a (Table 2) under basic conditions provided 4,5disubstituted 1,2,3-triazoles of a similar type to those reported by Wang. 16 However, when 2-substituted β-ketoester 1b was reacted with azide 4a ( Table 2, entry 2), a sharp diversion of the reaction course was observed, and 5-hydroxy-triazole 5a was isolated as the sole product ( Table 2 entry 2). We then investigated the scope of the reaction, applying the same reaction protocol to β-ketoesters 1b and 1c and aryl azides (4a−j) ( Table 2). Reactions of ketoester 1b with aryl azides 4h−j did not lead to the formation of any product, and only starting materials were recovered after 18 h. Reactions of ketoesters 1b and 1c with azides 4a−g and 4a under identical conditions provided 5-hydroxytriazoles 5a−g and 5k, which with the exception of 5a are reported here for the first time. The chemical structure of 5a was confirmed by single-crystal X-ray diffractometry (see Figure 2 and the Supporting Information).
The reaction of ketoesters 1b and 1c with aryl azides 4a−j in the presence of 1.2 equiv of DBU in acetonitrile at 50°C for 18 h provided 5-hydroxytriazoles 5a−k in moderate to good yields. The reaction was, however, sensitive to the electronic effects of substituents of the azide, where electron-neutral azides 4a, 4b, 4d, and 4e gave highest yields. The electron-rich azide 4c reacted well, albeit the yields were lower. Aromatic azides 4f−g bearing strong electron-withdrawing groups provided the corresponding 5-hydroxytriazoles 5f−g in only moderate yields. Steric hindrance from aryl substituents of aryl azides 4 was not found to affect reaction yield. The potential for the organo-catalytic activity of DBU in this reaction was also investigated; however, no evidence to support this was found. We then proceeded to optimize the reaction conditions between 1b and 4a by varying the base, solvent, temperature, and reaction times.
The high yield obtained when reacting 1b with 4a in the presence of solid KOH (potassium hydroxide) as a base and TBAB (tetrabutylammonium bromide) as a PTC (phasetransfer catalyst) encouraged us to study the scope of this transformation (Table 3). Phase-transfer conditions were found to be particularly effective with electron-poor and electron-neutral azides, such as 4a, 4g, and 4h, with yields up to 95%. However, when bulky azide substituents where present, such as 4e, only the hydrolyzed product 6 was favored over cyclization (Table 3). Since compound 5a    The Journal of Organic Chemistry pubs.acs.org/joc Article behaved as a relatively strong Brønsted acid, we decided to carry out a titration experiment to better characterize its properties. The pK a of 5a was found to be 4.2, which is comparable to that of a carboxylic acid (see the Supporting Information). 19 This result was very unexpected compared to the pK a values of the related 4-hydroxy-1,2,3-triazoles, which were found to be significantly less acidic than those reported by Pippione and co-workers. 20 Moreover, we found 5hydroxytriazoles to be highly soluble in water when accompanied by both organic and inorganic bases, thus justifying their attractiveness as candidates for drug discovery. We further expanded the scope of the cyclization reaction to include cyclic β-ketoesters 7a−c to provide novel long-chain 5hydroxytriaxoles 8. Interestingly, poor conversion was observed when the reaction was performed in solvent, while we saw a notable improvement when the cyclization was performed solvent-free with yields up to 79% (Table 4).
A literature search highlighted the structural similarity of 5hydroxytriazoles and hydroxamic acids. 21 Hydroxamic acids and 5-hydroxytriazoles share similar pK a valuesand amide-like bioisosterism, which makes them excellent ligands for enzymebound metals. Compounds 5a−g and 8 reacted with Fe 2+ and Cu 2+ salts to provide blue-violet and red-colored solutions, indicating a ligand-like behavior similar to that of hydroxamic acids. Suberanilohydroxamic acid 9 (SAHA) is a hydroxamic acid that is active as a histone deacetylase (HDAC) inhibitor. 22,23 To exemplify the potential of the 5-hydroxytriazole nucleus in medicinal chemistry, we set out to convert long-chain 5-hydroxytriazoles 8a−c into terminal N-phenylamide-substituted triazoles 10a−c, which bear a close structural relationship to hydroxamic acid 9 (Figure 3).
The preparation of 10a−c (n = 1−3, respectively) is reported in the Experimental section. The structural and functional similarities between hydroxamic acid 9 and 5hydroxytriazoles 10 are highlighted in Figure 3 and include the following: (i) nominally similar scaffolds and chemical functionalities; (ii) analogous lone pairs (circled in red) of the carbonyl oxygen of the hydroxamic acid and the pyridyllike N of the triazole system; 24 (iii) hydrophobic backbones (circled in green), which are essential for interaction with active sites of HDAC isoforms, e.g., zinc-binding groups; and (iv) aromatic rings (in pink), which are essential for the correct positioning in the enzyme active site via π−π stacking. 25 Based on the reactivity observed, two reaction mechanisms have been proposed (Scheme 1), which lead to distinct products via analogous intermediates 12a and 12b. Intermediates 12a and 12b arise from reaction of enolate 11 with aryl azides (pathway a or pathway b), respectively. In pathway a, species 12a is formed and will evolve toward cyclic amide 13a, which contains no enolizable proton, and the concomitant elimination of ethoxide. A subsequent attack of ethoxide to the acetoxy group in 13a will lead to the elimination of ethyl acetate and the formation of 5. Conversely, in the presence of an enolizable proton such as in 12b (pathway b), the following cyclization to 13b and its subsequent protonation will generate 14b, which will provide compounds 3 after dehydration. A rationale similar to ours used to explain the mechanism in Scheme 1 has been reported by Pedersen and Begtrup for the reaction between phenyl azides and amides of malonic acids. 26

■ CONCLUSION
In conclusion, we have demonstrated that aryl azides undergo two distinct cycloaddition reactions with enolizable βketoesters depending on the substitution pattern on the βketoester, leading to different products. The cycloaddition of 2unsubstituted β-ketoester with pyridyl azides and phenyl azide was found to lead to 5-methyl-triazoles, whereas the cycloaddition of 2-alkyl-substituted β-ketoesters with phenyl azide and substituted aryl azides was found to lead to 5hydroxytriazoles, where the structure of one of the products has been confirmed via X-ray diffractometry (see Figure 2 and the Supporting Information). 27 The reaction of phenyl azide and substituted aryl azides with 2-alkyl-substituted β-ketoesters has been shown to be a fast, mild, and high-yielding method for the synthesis of 5-hydroxy-1,2,3-triazoles. The relatively acidic nature we observed for the 5-hydroxytriazoles has led us propose the study of 5-hydroxytriazole analogues as a new class of bioisosteres of hydroxamic acids. Future work will involve an investigation of this novel class of compounds as potential biological targets and their potential as a bioisosteric relative of biologically active hydroxamic acids.  (Hz). Melting points were measured using a Stuart scientific melting point apparatus and were uncorrected. Infrared spectra (IR) were recorded with KBr discs using a Bruker Tensor27 FT-IR instrument. Highresolution mass spectra were obtained on a Waters Micromass GCT PremierMS spectrometer or on a Bruker microTOF-Q III LC-MS spectrometer (APCI method). Optical rotations were measured on a PerkinElmer 343 polarimeter. HPLC chromatograms was recorded on a YMC-Triart Phenyl 150 × 4.6 mm column using a 5 μL injection volume (60:40 MeCN/H 2 O) at two different wavelengths of 190 and 254 nm, respectively. The purity of the final products was verified by HPLC analysis and 1 H and 13 C{ 1 H} NMR spectroscopy. Analyticalgrade solvents and commercially available reagents were used as received. Dry DCM was purchased from Sigma-Aldrich. Reactions were monitored by TLC (Merck, silica gel 60 F254). Flash column chromatography was performed using silica gel 60 (0.040−0.063 mm, 230−400 mesh). Substituted arylazides 4a−j 28 and pyridyl azides 2a− c were prepared according to reported procedures. 29,30 β-Ketoesters 1a−c and modified ketoesters 7a−c were purchased from Sigma-Aldrich and used without further purification. 5-Hydroxy-1,2,3triazoles 5a, 5e, 5f, and 5g were synthesized via phase-transfer catalysis according to GP3 and via DBU-promoted synthesis according to GP2, while 5-hydroxy-1,2,3-triazoles 5b, 5c, 5d, and 5k were synthesized according to GP2 via a DBU-promoted synthesis. 5-Methyl-1,2,3-triazoles 3a−b were synthesized according to the GP1 procedure. 5-Methyl-1,2,3-triazole 3c was synthesized according to a modified version of GP1. Long-chained 5-hydroxy-1,2,3-triazoles based SAHA analogs 10b−c were synthesized according to GP6 via long-chained 5-hydroxy-1,2,3-triazole precursors 8a−c, which were synthesized according to GP4. The long-chained 5-hydroxy-1,2,3triazole-based SAHA analog 10a was synthesized according to a modified version of GP4. General Procedure for the DBU-Promoted Synthesis of 5-Methyl-1,2,3-triazoles 3a and 3b (GP1). To a solution of pyridyl azides 2a or 2b (0.5 mmol, 1 equiv) and β-ketoester 1a (0.6 mmol, 1.2 equiv) in MeCN (2.5 mL, 0.2M) was added DBU (0.6 mmol, 1.2 equiv), and the reaction mixture was stirred at 50°C in an oil bath overnight. The crude mixture was evaporated under vacuum and purified via flash column chromatography (MeOH/DCM/AcOH 90:10:0.1) to afford title compounds 3a and 3b as solids.
Ethyl 5-Methyl-1-(pyridin-4-yl)-1H-1,2,3-triazole-4-carboxylate 3a. Yellow solid (93 mg, 80% yield). 1 3, 14.4, 9.9. All analytical data are consistent with those reported in the literature. 31 Synthesis of Ethyl 5-Methyl-1-(pyridin-2-yl)-1H-1,2,3-triazole-4-carboxylate 3c. To a solution of 2c (100 mg, 0.83 mmol, 1 equiv) in DMSO (1.4 mL, 0.6M) were added 1a (141 ul, 1 mmol, 1.2 equiv), DBU (150 ul, 1 mmol, 1.2 equiv), and Cu(OTf) 2 ·C 6 H 5 CH 3 (43 mg, 0.083 mmol, 0.1 equiv). The reaction mixture turned from light brown to black upon the addition of the catalyst and was stirred for 8 h at reflux in an oil bath. After 8 h, TLC showed the complete consumption of 2c, and the mix was cooled to room temperature and extracted with DCM/H 2 O three times. The collected organic phases were filtered through a Celite pad and concentrated in vacuo. The crude product was purified by flash column chromatography (DCM/ AcOEt 90:10) to afford the product 3c in a modest yield (58 mg, 50% yield) as a yellow oil. TLC showed the product to be visible as a brilliant purple spot under an UV lamp at a short wavelength. All analytical data are consistent with those reported in the literature. 1  General Procedure for the DBU-Promoted Synthesis of 5-Hydroxy-1,2,3-triazoles 5a−k (GP2). To a solution of aryl azides 4 (0.5 mmol, 1 equiv) and β-ketoester 1 (0.6 mmol, 1.2 equiv) in MeCN (0.2M) was added DBU (0.6 mmol, 1.2 equiv), and the reaction mixture was stirred at 50°C in an oil bath overnight. The crude mixture was evaporated under vacuum and purified via flash column chromatography (MeOH/DCM/AcOH 90:10:0.1) to afford title compounds 5a−k as solids. In some cases after chromatography, products 5a−k still contained traces of the DBU salt, which were easily removed by the trituration of the solid with a minimum quantity of water.