Kinetic Resolution of Cyclic Secondary Azides, Using an Enantioselective Copper-Catalyzed Azide–Alkyne Cycloaddition

Abstract

An enantioselective copper-catalyzed azide− alkyne cycloaddition (E-CuAAC) is reported by kinetic resolution. Chiral triazoles were isolated in high yield with limiting alkyne (up to 97:3 enantiomeric ratio (er)). A range of substrates were tolerated (>30 examples), and the reaction was scaled to >1 g. The er of a triazole product could be enhanced by recrystallization and the recovered scalemic azide could be racemized and recycled. Recycling the azide allows efficient use of the undesired azide enantiomer.

Graphical abstract

The copper(I) catalyzed azide−alkyne cycloaddition reaction (CuAAC)^1,2 is an important transformation. Applications for CuAAC penetrate numerous subdisciplines, including chemical biology, medicinal chemistry, and polymer chemistry.^3–5 This click reaction⁶ is operationally simple, functional-group-tolerant, high yielding, and broad in scope. Furthermore, the triazole ring is gaining recognition as a protease-resistant peptidomimetic.⁴ Biologically active α-chiral triazoles have been reported (Figure 1).^7–12 Because of the prevalence of α-amino acid-derived amides, an enantioselective CuAAC (E-CuAAC) reaction would be useful for accessing α-N-chiral triazoles. However, E-CuAAC has proven challenging in part because (i) both cycloaddition components have linear geometries, (ii) CuAAC does not require a ligand, and (iii) the product triazole is a competent CuAAC ligand.¹³ Therefore, ECuAAC must kinetically outcompete the background CuAAC reaction.

Figure 1.

Representative bioactive α-N-chiral triazoles.

A few reports describe kinetic resolution,^14–16 dynamic kinetic resolution,¹⁷ or desymmetrization^18–21 for E-CuAAC.²² The majority of successful E-CuAAC reactions generate α-C-chiral triazoles, where the new stereocenter is alkyne-derived.^{15,16,18,20,21} The original E-CuAAC reported by Fokin and Finn is the state-of-the-art E-CuAAC kinetic resolution for α-N-chiral triazole formation, where the new stereocenter is derived from the azide component (Scheme 1a).¹⁴ The authors provided only two examples of kinetic resolution with a selectivity factor (s) of 3.2 and 8, respectively (ca. 70:30 er and 84:16 er, assuming 40% conversion of the azide). Using the same conditions, the desymmetrization of two bis-azides were reported to result predominantly in the formation of bistriazole (not shown).

Scheme 1.

Azide Kinetic Resolution by E-CuAAC

Our laboratory became interested in using the unique properties of allylic azides to establish α-N-chiral centers though dynamic kinetic resolution.^23–25 Recently, we reported the first E-CuAAC reaction that proceeded by dynamic kinetic resolution.¹⁷ This work prompted an exploration of E-CuAAC by azide kinetic resolution for α-N-chiral triazole synthesis (Scheme 1b). Reported herein is a successful expansion of the kinetic resolution by E-CuAAC.

Our investigation began with azide 1a, alkyne 2a, and commercially available PYBOX ligands (Table 1, entries 1−3). Minimal enantioselectivity was observed with these ligands, confirming the report from Fokin and Finn.¹⁴ A wide variety of chiral ligands were screened consisting of numerous ligand classes (see Table 1, entries 4 and 5, as well as the Supporting Information). Only the aryl-PYBOX ligands provided reasonable enantioselectivity, with ligand L5 providing the best selectivity among the ligands screened. The data obtained are reported here with er instead of the selectivity parameter s because (i) er is more synthetically meaningful, (ii) er is directly obtained by chiral HPLC, and (iii) several kinetic assumptions are made when deriving s, which are likely faulty for E-CuAAC, based on reported CuAAC kinetics.^26,27 One could convert the reported er to a presumed s by assuming 40% conversion for reactions approaching completion. Lowering the reaction temperature resulted in an increase in enantioselectivity, albeit at the cost of conversion (Table 1, entries 6−8). The reaction was faster in DME or PhCF₃ as the solvent, relative to CH₂Cl₂, and the enantioselectivity was maintained. When using PhCF₃ as the solvent, the reaction was complete within 48 h (entry 11; see the Supporting Information for additional optimization).

Table 1.

E-CuAAC Optimization by Kinetic Resolution^a

entry	ligand	solvent	Temperature (°C)	yieldb (%)	enantiomeric ratio, erc
1	(R)-L1	CH2Cl2	rtd	88 (35)	55:45
2	(R)-L2	CH2Cl2	rtd	55 (22)	50:50
3	(R)-L3	CH2Cl2	rtd	92 (37)	22:78
4	(R)-L4	CH2Cl2	rtd	95 (38)	21:79
5	(S)-L5	CH2Cl2	rtd	96 (38)	80:20
6e	(S)-L5	CH2Cl2	0	84 (34)	84:16
7e,f	(S)-L5	CH2Cl2	−15	64 (26)	86:14
8e,g	(S)-L5	CH2Cl2	−20	34 (14)	89:11
9e,f	(S)-L5	DME	−15	87 (35)	86:14
10e,f	(S)-L5	PhCF3	−15	92 (37)	88:12
11e,h	(S)-L5	PhCF3	−15	92 (37)	88:12

^aReactions conducted with azide 1a (0.125 mmol) and alkyne 2a(0.05 mmol) at 0.1 M in varying solvent with (CuOTf)₂·PhMe (0.62 μmol) and varying ligand (2.5 μmol). All yield and er values reflect the average of duplicate trials.

^bYield determined using calibrated GC with triphenylmethane as an internal standard. The yield is calculated based on either the alkyne (limiting reagent) or azide (kinetic resolution component).

^cChiral HPLC was used to determine er.

^drt = room temperature.

^eConcentration was 0.125 M.

^fTime = 72 h.

^gTime = 96 h.

^hTime = 48 h.

With the optimized conditions in hand, the scope of the alkyne coupling partner was investigated (Scheme 2). The model substrate was isolated in comparable yield and er as expected (3a). Other aryl alkynes were tolerated with electron-rich (3b and 3c), electron-neutral (3d and 3e), and electron-deficient substituents (3f−3h) on the arene. Substituents positioned meta and ortho on the aryl alkyne provided good enantioselectivity (3i−3l). Ethyl propiolate (3m), cycloalkyl alkynes (3n and 3o), and a heterocyclic alkyne (entry 3p) were tolerated, although several substrates required higher catalyst loadings or longer reaction times to reach higher conversion. Note that long reaction times are not uncommon for other ECuAAC reactions.^15,18,19

Scheme 2. Substrate Scope of Alkyne Coupling Partner^a.

^aIsolated yields are reported. The yield is calculated based on either the alkyne (limiting reagent) or azide (kinetic resolution component). Enantiomeric ratio (er) was determined by chiral HPLC. Yield and er values are the average of duplicate trials. ^b2.5 mol % (CuOTf)₂·PhMe, 10 mol % (S,S)-4-F-Ph-PYBOX, and 0.1 M in PhCF₃.

The scope of the azide component that could be kinetically resolved was explored (Scheme 3). Substituents on the chromane arene core were tolerated, including methyl (5a) and halogens (5b−5d). Groups could also be added α to the oxygen atom (5e). Azido-tetrahydroquinolines with various N-protecting groups (5f−5h) and the free NH (5i) provided triazoles in high yield and acceptable enantioselectivity. Azidothiochromanes (5j and 5k), azido-dihydrobenzofuran (5l), and azido-indane (5m and 5n) could also be resolved. Substrates 5m and 5n are noteworthy because the original E-CuAAC reported by Fokin and Finn described this as a problematic substrate (s < 1.3).¹⁴ Acyclic substrate 5o provided slower conversion and slightly reduced selectivity.

Scheme 3. Substrate Scope of Azide Coupling Partner^a.

^aIsolated yields are reported. The yield is calculated based on either the alkyne (limiting reagent) or azide (kinetic resolution component). Enantiomeric ratio (er) was determined by chiral HPLC. Yield and er values are the average of duplicate trials.

Chiral alkyne 6 was used to test for matched/mismatched behavior (Scheme 4). Treating azide rac-1a with chiral alkyne 6 and either enantiomer of ligand L4 resulted in moderate yield with opposite diastereoselectivity (14:1 and 1:13). This reversal of diastereoselectivity indicates this reaction is under catalyst control.

Scheme 4.

Matched/Mismatched Experiment

The kinetic resolution could be successfully scaled to provide more than 1 g of triazole product (Scheme 5). The initial enantioselectivity was 89:11 er, which corresponds to an s = 13.5. The enantiopurity could be enhanced upon recrystallization (99:1 er). The excess azide was recovered (76:24 er) and racemized upon exposure to catalytic AgPF₆.²⁸ The azide could then be recycled and used in a subsequent reaction. The ability to recycle the recovered azide improves the overall efficiency of the reaction. The (R)-1a enantiomer preferentially reacted with the catalyst derived from ligand (S,S)-L5. This was determined by analyzing scalemic azide recovered from the reaction. The scalemic azide was compared to a sample of (S)-1a which was accessed via diazo transfer from commercially available (S)-3,4-dihydro-2H-chromen-4-amine. The same analysis was conducted with azide recovered in the synthesis of triazoles 3j and 3p, which confirmed the absolute configuration of the product. The configuration of the other triazole products were assigned based on analogy.

Scheme 5.

Gram-Scale Reaction and Racemization of Recovered Azide

This report describes an expanded scope for both the azide and alkyne coupling partners in an E-CuAAC. The products of this kinetic resolution are α-N-chiral triazoles and can be obtained in up to 97% yield and up to 97:3 er. The reaction can be conducted to isolate more than 1 g of product and the excess azide can be recovered, racemized, and recycled. The er of the product triazoles can be readily improved to 99:1 with a single recrystallization.