Copper-Catalyzed Cross Coupling of Benzylic C–H Bonds and Azoles with Controlled N-Site Selectivity

Abstract

Azoles are important motifs in medicinal chemistry, and elaboration of their structures via direct N–H/C–H coupling could have broad utility in drug discovery. The ambident reactivity of many azoles, however, presents significant selectivity challenges. Here, we report a copper-catalyzed method that achieves site-selective cross coupling pyrazoles and other N–H heterocycles with substrates bearing (hetero)benzylic C–H bonds. Excellent N-site selectivity is achieved, with the preferred site controlled by the identity of co-catalytic additives. This cross-coupling strategy features broad scope for both the N–H heterocycle and benzylic C–H coupling partners, enabling application of this method to complex molecule synthesis and medicinal chemistry.

Graphical Abstract

Pd-¹ and Cu-catalyzed² C(sp²)–N cross coupling reactions are some of the most widely used methods for pharmaceutical synthesis.³ Recent efforts have begun prioritizing complementary methods for C(sp³)–N coupling⁴ as a means to expand the topological diversity and physiochemical properties of the resulting molecules.^{3, 5} C–N coupling methods that directly functionalize C(sp³)–H bonds bypass the need for pre-functionalized alkyl electrophiles and represent important targets for medicinal chemistry.⁶ Significant progress has been made in C(sp³)–H amination reactions that install ammonia surrogates via nitrene transfer⁷ or azidation,⁸ while C–H/N–H cross-coupling reactions, for example, with secondary amines/amides or N–H heterocycles, are more limited⁹ and often require excess C–H substrate.¹⁰ Here, we report a copper-catalyzed method for selective cross-coupling of azoles with (hetero)benzylic C–H substrates as the limiting reagent, affording N-benzylic heterocycles featured in pharmaceutical and agrochemical compounds (Figure 1A).^11–14 In addition to the challenge of C–H site selectivity, these reactions feature a second selectivity challenge arising from azoles that incorporate two (or more) nucleophilic nitrogen atoms.¹⁵ This issue has important implications for medicinal chemistry because regioisomeric N-substituted azoles can have very different properties and/or bioactivity (Figure 1B).¹⁶ The method described herein achieves excellent C–H site selectivity via hydrogen-atom transfer (HAT) from (hetero)benzylic C–H bonds (Figure 1C). Reaction with the azole coupling partners via Cu^II-mediated radical-polar crossover exhibits excellent N¹/N² site-selectivity (e.g., with pyrazoles), and variation of the reaction conditions lead to selective formation of either regioisomeric N-benzyl product with many coupling partners (Figure 1D).

Figure 1.

(A) Importance of benzylic N-azoles in drug discovery. (B) Impact of regioselectivity of heterocyclic compounds in medicinal chemistry. (C) Copper-catalyzed regioselective cross couplings of benzylic C–H bonds and N–H heterocycles enabled by various additives.

Investigation of oxidative cross coupling of benzylic C–H substrates and N–H azoles started with ethylbenzene (1a) and ethyl 3-(trifluoromethyl)-1H-pyrazole-4-carboxylate (2a), a pyrazole featured in previous drug discovery efforts.¹⁷ Potential reaction conditions were inspired by recent studies involving the use of copper catalysts in combination with N-fluorobenzenesulfonimide (NFSI) as the oxidant. Diverse nucleophilic coupling partners have been used in these reactions, including pseudohalides (cyanide,¹⁸ azide,^8d isocyanate¹⁹), alcohols,²⁰ carbamates,²¹ and carbon-based nucleophiles (Zn(CF₃)₂,²² ArB(OH)₂,²³ alkynes²⁴)^25,26 Cu^I-mediated activation of NFSI generates an N-centered radical that promotes selective HAT from benzylic C–H bonds, and subsequent coupling with heteroatom nucleophiles appears to favor a radical-polar crossover pathway involving formation of a benzylic cation intermediate.^8d,19,20 Complementary studies have shown that mild reductants, such as dialkylphosphites, can promote these reactions by buffering the redox state of the Cu catalyst and ensuring that both Cu^I and Cu^II are present in the reaction.^19,20,27

The above mechanistic considerations guided a survey of reaction conditions. Initial screening data showed that Cu/NFSI-catalyzed oxidative coupling of 1a and 2a favors the N² isomeric product 3aa when the reaction is conducted in chlorinated solvents PhCl and dichloromethane (DCM) (Figure 2A, entries 1–2; see Supporting Information for full screening details). Use of more polar solvents [hexafluoroisopropanol (HFIP) and MeNO₂] led to a switch in selectivity, favoring the N¹ isomer 3aa’ (entries 3–4). Tetrabutylammonium (TBA) bromide helped to solubilize the Cu catalyst and improved the reactivity and selectivity. Conventional ancillary ligands, such as phenanthroline or bioxazolines, led to lower yields (see Table S1 and S2 for detailed information). A mixture of DCM:HFIP (7:3) enhanced the conversion of 1a and improved the product yield to 71% (3aa + 3aa’, entry 5). Increasing the loading of TBABr to 0.3 equiv further improved the reaction yield and increased the N²:N¹ selectivity to 36:1 (entry 6). Optimal results results were obtained upon replacing TBABr with TBACl, affording 3aa in 75% yield (entry 7). Further screening experiments revealed that the N²/N¹ regioselectivity could be switched to favor the N¹ product 3aa’ with certain additives (Figure 2B, see Table S2 for additional data). The initial entries in Figure 2B show that additives with halide anions favor formation of the N² regioisomer 3aa. In contrast, additives with triflate anion or Lewis acids, such as silyl triflates or BF₃•OEt₂, favor formation of the N¹ regioisomer 3aa’. Optimal results for the formation of 3aa’ were achieved with trimethylsilyl triflate (TMSOTf) as the additive at 60 °C (67% yield; N²:N¹ = 1:66).

Figure 2.

Evaluation of effects of various solvents and additives on the regioselectivity. (A) Effects of solvents on benzylic C–N cross coupling reaction. (B) Regioselectivity switch observed in cross coupling of 1a and 2a with different additives. Conditions identical to those shown in part A, using 10 mol% additive instead of TBABr, 10 mol % CuBr₂ instead of CuCl, and DCM:HFIP (7:3) solvent. ^a Monitored by ¹H NMR spectroscopy, yield determined using 0.2 mmol mesitylene as internal standard. TBA, tetrabutylammonium; DCM, dichloromethane; HFIP, hexafluoroisopropanol. ^b Reaction run with TBA⁺Cl⁻ instead of TBABr. ^c Reaction run at 60 °C.

The pyrazole reagent 2a has an N–H bond at the N¹ position, as revealed by X-ray crystallography and depicted in Figure 2A (see Section 7 in the Supporting Information for details). This structure is consistent with previous reports for other electron-deficient, 3-substituted pyrazoles.²⁸ We postulated that the switch in pyrazole regioselectivity could arise from kinetic versus thermodynamic control over the C–N bond-forming step. Experimental data supported this hypothesis: addition of TMSOTf to 3aa induced isomerization of this compound to 3aa’, whereas no isomerization of 3aa’ was observed in the presence of TBABr (Figure 3A). These results suggest 3aa is the kinetic product, and they are rationalized by the non-basic reaction conditions, which result in alkylation of the pyrazole nitrogen atom lacking the proton. Reactivity with the pyrazole N² lone pair (Figure 3B, top) contrasts previously reported reactivity with deprotonated pyrazolide reagents which react preferentially at the N¹ site.^28b,29 The isomerization data in Figure 3A suggests that strong Lewis acids, such as trimethylsilyl (TMS) cation, can promote isomerization via the benzylic cation and N^2-TMS species, to the thermodynamically favored N¹ product.^28b,30 The data in Figure 2A, entries 1–4 suggest that solvents capable of stabilizing charged intermediates also favor the thermodynamic product.

Figure 3.

Mechanistic origin of pyrazole regioselectivity. (A) N¹/N²-isomerization test, implicating the the N¹. (B) Proposed mechanism rationalizing the influence of TMSOTf (and other Lewis acids) on the N²/N¹ regioselectivity.

Access to both isomeric pyrazole coupling products is noteworthy because most coupling reactions with pyrazoles employ a base and access only the thermodynamic products.^29,31 For example, reaction of (1-bromoethyl)benzene with pyrazole 2a affords exclusively the N¹ isomer 3aa’ (see Section 4 in the Supporting Information for details). To explore the scope of this reactivity, we evaluated numerous other C–H/N–H cross-coupling reactions with (hetero)benzylic and azole coupling partners (Figures 4 and 5). Reactions with (hetero)benzylic C–H substrates were initially tested using the TBACl conditions, which tend to be more robust and promote formation of the unique N²-pyrazolyl product (Figure 4A). Ethylbenzene derivatives with different p-substituents are well tolerated (70–86%, 3aa-3da). While the N² product is typically observed under these conditions, only the N¹ product 3ba’ is observed with p-MeO-ethylbenzene (1b). This result is rationalized by the stability of the benzylic carbocation, which facilitates isomerization to the thermodynamic product (cf. Figure 3). The lack of C–N coupling at tertiary C–H sites in reactions with isobutyl- and isopentylbenzene (3ea, 3fa) highlights the exquisite site-selectivity of the HAT steps with the NFSI-derived imidyl radical.^8d,20 The strongly electron-withdrawing nitro group in the indane substrate 1g reduces the product yield (33%, 3ga). Benzhydryl N-azoles, a class of compounds that exhibit aromatase inhibitory activity,³² exhibit good reactivity (86%, 3ha; 95%, 3ia) and form only the N¹ regioisomers, again rationalized by the benzylic cation stability.

Figure 4.

Assessment of various benzylic C–H substrates in cross coupling reactions with N–H heterocycles with (A) TBACl as the additive for N² regioselectivity and (B) with TMSOTf as the additive for N¹ regioselectivity. Regioisomers >5% were isolated and reported. ^aConducted in 0.5 mL DCM:HFIP (7:3). ^bConducted with 10 mol % CuBr₂ and 30 mol % TBABr. ^c Conducted with 10 mol % TBACl. ^dConducted at 40 °C. ^eConducted at 60 °C. ^fConducted in DCM. ^gConducted at 30 °C. ^hConducted with 10 mol % BF₃•OEt₂.

Figure 5.

Assessment of various N–H heterocycles in cross coupling reactions. Substrate scope with diverse N–H heterocycles and indane under kinetically controlled TBACl conditions (A) and in the presence of BF₃•OEt₂ as a Lewis acid cocatalyst (B). Exploration of cross-coupling reactions of diverse N–H heterocycles and (hetero)benzylic C–H scaffolds (C) under the TBACl conditions of Figure 5A, unless noted otherwise. Regioisomers formed in >5% yield were isolated. ^aConducted in DCM. ^bConducted at 50 °C. ^cConducted at 30 °C. ^dTrace amount of the other benzylic regioisomer was observed. ^eConducted at 40 °C. ^fConducted with 10 mol % BF₃•OEt₂.

Benzylic pyrazoles of chromans,³³ thiophene,^14,34 and other substrates bearing oxygen-, sulfur- and nitrogen-containing heterocycles react successfully (1j–1u). These reactions demonstrate a tolerance for a formyl group (3ka), heteroarylbromides (3la), and pyrimidines (3ma, 3nb).³⁵ The pyridine substrate 1o reacts in lower yield (26%, 3oa). Substrates with two possible (hetero)benzylic sites (1g, 1m, 1n, and 1o) generate a mixture of products, with moderate to good regioisomeric ratios (r.r.).

Pyrazoles 2a and 3-bromopyrazole (2b) undergo successful late-stage reactivity with bioactive molecules. Examples include an antidiabetic intermediate 1p, which affords only the N¹ coupling product (90%, 3pa’); benzbromarone methyl ether (77%, 3qa), a derivative of a xanthine oxidase inhibitor;³⁶ celestolide (70%, 3rb); ibuprofen methyl ester (65%, 3sa); an anti-inflammatory anti-allergy agent³⁷ (49%, 3tb, r.r. = 7.2:1) and a precursor to a GnRH antagonist³⁸ (42%, 3ua). These results were complemented with a focused assessment of the Lewis acid co-catalytic conditions (Figure 4B). These reactions lead to exclusive or high N¹ pyrazole site-selectivity in the C–N coupling reactions.

Different pyrazoles and other azole coupling partners were then tested in reactions with indane, motivated by the relevance of N-indanyl azoles in medicinal chemistry (Figure 5A).³⁹ Symmetrical 4-substituted pyrazoles, bearing fluoro, chloro, iodo, formyl, trifluoromethyl and nitro substituents, undergo coupling in good-to-excellent yields (60–82%, 3vc-3vh). Other di- and trisubstituted pyrazoles with di- and trifluoromethyl (2i, 2j)^17a and sulfonyl chloride (2k) substituents react effectively (52–55% yields, 3vi-3vk). Successful reactivity, but lower yields are observed with indazole coupling partners 2l and 2m (32%, 3vl; 30%, 3vm).^40,41 The dichloropurine derivative 2n generates both N⁹ and N⁷ regioisomeric products (58% yield, N⁷:N⁹ = 1:1.5, 3vn). Triazoles, including 1,2,4- and 1,2,3-isomers 2o and 2p and a tetrazole 2q undergo successful reactivity (77%, 3vo; 45%, 3vp; 68%, 3vq). The reaction with 2p favors formation of the less common N¹ regioisomer (N¹:N² = 4.6:1),⁴² consistent with the mechanistic rationale above.

Inclusion of BF₃•OEt₂ in these reactions enabled modulation of the azole regioselectivity (results with BF₃•OEt₂ were slightly better than with TMSOTf). In each of the four cases shown in Figure 5B, the favored regioisomer is different from that observed in the TBACl conditions. For example, the reaction with 2j completely switches from N² to N¹ selectivity (46%, 3vj’). In other cases, use of BF₃•OEt₂ ensures only a single isomer is formed (63%, 3vn’; 66%, 3vp’).

In a final assessment of the method, we explored C–N cross coupling reaction with different (hetero)benzylic and N-heterocyclic partners (Figure 5C). These reactions proceed in moderate-to-excellent yields. The reaction conditions show promise for heterocyclic coupling partners beyond azoles, including the beta-lactam 2s (80%, 3ps) and sultams 2r and 2t (47%, 3vr, Figure 5A; 52%, 3tt, Figure 5C). Reactions of the azoles 2a, 2i, and 2p react with the regioselectivity expected from the observation elaborated above. For example, 1p affords the thermodynamically favored coupling product upon reaction with 2p, reflecting the stability of benzylic cation. On the other hand, the reaction of 1w and 2a afforded the kinetically favored the N² regioisomer even with BF₃•OEt₂ as a cocatalyst (51%, 3wa). The combined effect of BF₃•OEt₂ cocatalyst and HFIP as a cosolvent, however, supported a switch in the observed regioselectivity (46%, 3wa’). The reaction of 1t and 2i affords only one of the four possible C–N coupling products (55%, 3ti), demonstrating both C–H and N-nucleophile selectivity.

In summary, the results outlined herein introduce a unique synthetic method for direct C(sp³)–H/N–H coupling of (hetero)benzylic substrates and azoles. Multiple features contribute to the potential impact of these methods. Perhaps the most notable feature is the ability to control the azole N-site selectivity, often enabling access to either regioisomeric product. Other highlights include the high-to-exclusive (hetero)benzylic C–H site selectivity, use of the C–H substrate as the limiting reagent, excellent scope for both coupling partners, and broad functional group compatibility. Overall, these C(sp³)–N coupling methods provide efficient access to complex, pharmaceutically relevant structures, and they should find widespread utility for library synthesis and exploration of chemical space in medicinal chemistry and related disciplines.⁴³