Solvent-Mediated Tunable Regiodivergent C6- and N1-Alkylations of 2,3-Disubstituted Indoles with p-Quinone Methides

Indium-catalyzed, solvent-enabled regioselective C6- or N1-alkylations of 2,3-disubstituted indoles with para-quinone methides are developed under mild conditions. Notably, highly selective and switchable alkylations were selectively achieved by adjusting the reaction conditions. Moreover, scalability and further transformations of the alkylation products are demonstrated, and this operationally simple methodology is amenable to the late-stage C6-functionalization of the indomethacin drug. The reaction pathways were explained with the support of experimental and density functional theory studies.


■ INTRODUCTION
Indole is one of the most significant heteroaromatic structural units found in numerous bioactive natural products, medicinal compounds, and synthetic compounds and plays an important role in their bioactivity. 1 Particularly, the substructures of 2,3disubstituted indole skeletons (as fused or substitutions) are medicinally and synthetically valuable scaffolds and present as a core structural motif in numerous pharmaceutically relevant compounds. 2−5 Velbanamine and a small subgroup of tetracyclic natural bases belonging to the Iboga family are representative fused indole substructure examples (Figure 1). 6 The 2,3-disubstituted indole scaffold is also a basic structural motif for the nonsteroidal anti-inflammatory drug (NSAID) indomethacin. 7 Therefore, there is interest in the development of efficient methods to access 2,3-disubstituted indoles. Most of these methods for functionalized 2,3-disubstituted/fused indoles include the architecture of the parent heterocycle by cyclization via different transition metals and metal-free catalysis conditions using phenylhydrazine, aniline, and nitrobenzene as main substrates. 8 Other approaches to these functionalized indoles involve derivatization of the indole nucleus via crosscoupling methodology or direct C−H/N−H activation. 9 Meanwhile, N1, C3, and C6 positions of 2,3-disubstituted indoles may exhibit mainly three types of nucleophilic behavior toward electrophiles due to their electronic nature (Scheme 1a). The C3-and N1-positions of 2,3-disubstituted indoles have been well documented as the most reactive sites among the three positions (N1, C3, C6), and the majority of their reactions were focused on the C3-and N1-positions. 10−12 Furthermore, 2,3disubstituted/fused indoles can couple with electrophiles to permit Friedel−Crafts-type remote C6-functionalization, which is relatively more difficult (Scheme 1a). In this context, significant progress has been recently achieved via remote C6−H functionalization of indoles. Limited elegant approaches, such as template-assisted C6-olefination, 13 ligand-controlled C6-borylation, 14 C6-arylation via a transient mediator, 15 and C6-alkylation with σ-activation, 16 are available for the remote C(sp 2 )−H activation (Scheme 1a, middle side). In addition, Lewis and Bronsted acid promoted Friedel−Crafts-type reactions were also reported for the C6−H alkylation (Scheme 1a, right side). 17−23 Within this scope, Zheng, You, and coworkers, pioneers in the field, first described scandium triflatecatalyzed direct C6 alkylation of 2,3-disubstituted indoles with aziridines via Friedel−Crafts-type functionalization. 17 Subsequently, Shi and co-workers reported another pioneering study via chemo-and regiospecific C6-alkylation of 2,3-disubstituted indoles using 3-indolylmethanols with organocatalysts. 18 Also, Zhang and co-workers described chiral phosphoric acid (CPA)catalyzed remote C6-enantioselective C−H functionalization of 2,3-disubstituted indoles with isatin-derived N-Boc ketimine through the dual H-bonds and π−π interaction strategy. 19 However, the enantioselective C6-functionalization of 2,3disubstituted indoles with benzofuran-derived azadienes using chiral phosphoric acid led to a range of enantioenriched heterotriarylmethanes. 20 Recently, electrophiles such as ohydroxybenzyl alcohol, 21 trifluoromethylated 3-indolylmethanol, 22 and carbene 23 have been utilized in the Friedel−Craftstype C6-alkylations of 2,3-disubstituted indoles.
p-Quinone methides (p-QMs) have attracted considerable attention as powerful acceptors in 1,6-nucleophilic conjugate addition reactions with a wide range of carbon and heteroatom nucleophiles. 24 To date, indoles have been also tested as donor molecules on this 1,6-conjugate addition platform to access triarylmethanes. 25, 26 Despite those elegant advancements using this concept, the synthesis of triarylmethanes on the benzenoid ring of the indole motif is still challenging and there are a limited number of examples. 27,28 In 2016, Yao, Lin, and co-workers disclosed a single example (in a yield of 66%, its structure was drawn incorrectly in the article) of C6-alkylation of 2,3dimethylindole with p-QM under metal-free conditions (Scheme 1b). 29 Later, Zhang and co-workers reported another single example (in a yield of 83%, 62% ee) of CPA-catalyzed remote C6-enantioselective C−H alkylation of a 2,3-fused indole with a p-QM. 19 In 2020, Lin, Sun, and co-workers reported the synthesis of an unsymmetrical diarylmethane (1 example) and triarylmethanes (5 examples) through p-TsOHcatalyzed 1,6-conjugate addition of 2,3-dimethyl/fused indoles to in situ generated p-QM intermediates (Scheme 1b). 26 Consequently, the preceding examples of the C6-functionalization of indoles with p-QMs suffered from limited substrate scope (in terms of both indoles and p-QMs). However, the development of efficient protocols for the chemoselective functionalization through both C6−H and N1−H bonds of unprotected indoles is still a difficult problem and challenge. Herein, we describe In(OTf) 3

■ RESULTS AND DISCUSSION
In continuation of our research interest in C−H and N−H alkylation reactions, 27 2,3,4,9-tetrahydro-1H-carbazole (1a) and p-QM 2d were selected as the model substrates to optimize the reaction conditions (Table 1). The initial experiments were conducted in 1,2-dichloroethane (DCE) at room temperature for 12 h by using 10 different metal triflates as Lewis acid promoters (Table 1, entries 1−10). Zn(OTf) 2 (10 mol %) and Gd(OTf) 3 (10 mol %) were found to promote the reaction quite well, giving the N1-alkylation product 4ad in 82 and 77% yields, respectively (entries 1 and 2), whereas Bi(OTf) 3 (10 mol %) furnished C6-alkylation product 3ad in an 84% yield (entry 3). Subsequently, when other Lewis acids such as Cu(OTf) 2 and LiOTf were tested for the reaction, no alkylation products were detected, and only starting materials were recovered (entries 4 and 5). The subsequent screening of other triflates, such as scandium(III), tin(II), silver(I), ytterbium(III), and indium-(III) triflates, revealed that the desired C6-alkylation product 3ad was generated with excellent regioselectivity and yields (entries 6−10). We also checked the reaction with Bronsted acids in DCE (Table 1, entries 11 and 12). When TfOH was sufficient to promote the reaction, 3ad was obtained in 82% yield (entry 11). In contrast, NaHSO 4 as the solid heterogeneous acid catalyst afforded N1-alkylation product 4ad (entry 12). Furthermore, when the reaction was performed with HFIP as a catalyst, any alkylation products were not formed (Table 1,  entry 13). Notably, entries 1, 6, and 10 indicated that the catalysts such as Zn(OTf) 2 , Sc(OTf) 3 , and In(OTf) 3 played an important role in regioselective switching. Next, to test the effect of the solvent with these catalysts, further additional optimization was conducted ( Interestingly, the reaction in ethyl acetate afforded 3ad as a major product in 56% yield and with almost 2:1 regioselectivity (entry 18), while the reaction in dioxane gave 4ad as a major product in 84% yield and with almost 10:1 regioselectivity (entry 22). When dichloromethane, acetonitrile, and nitromethane were applied to the reaction instead of DCE, similar results were found (entries 19−21). Significantly, the use of THF in the In(III)-catalyzed reaction system gave the best result for the N1alkylation product 4ad with a 90% yield (entry 23). We found the yield of C6-alkylation product 3ad was improved to 92% with toluene as solvent (entry 24). Based on the above results, 10 mol % In(OTf) 3 is the preferred catalyst for these regioselective reactions. Furthermore, the use of THF for N1-alkylation and toluene for C6-alkylation as solvent were determined to be the other optimal reaction parameters at room temperature for 12 h (entries 23 and 24). Under the established optimized reaction conditions, we attempted to survey the substrate scope of various p-QM substrates 2a−x for C6-alkylation of 1a (Schemes 2a). By employing 1a as the coupling partner, we first investigated the generality of p-QMs 2a−x. A series of C6-alkylated products 3aa−ax were obtained in good to excellent yields (75−92% yield) under the optimized conditions (entry 24, Table 1). To our delight, this indium(III)-catalyzed 1,6-hydroarylation reaction in toluene demonstrated a broad scope for the p-QM reaction partner. In the case of p-QM 2a without any substituent on the phenyl ring, the reaction proceeded well and delivered the corresponding product 3aa in 88% yield. p-QMs bearing a halogen group at the para-position of aryl ring afforded C6alkylated indoles 3ab−ae in good to excellent yields (78−92%). A series of p-QMs including both electron-donating substituents (2f−h) such as methyl, tert-butyl, and methoxy and electronwithdrawing substituents (2i−k) such as carboxylic acid, nitro, and trifluoromethyl at the para-position of the phenyl ring were well tolerated to yield the desired products (3af−ak) in the range of 83−89% yields under the optimal conditions. The p-QM 2l bearing a diphenylamine group at the para-position of the aryl ring was also a suitable reaction partner for the 1,6-addition 3al in 80% yield. Even the reactions of substrates 2m−r with sterically hindered ortho-monosubstituted (Br, OAc, and OH), meta-monosubstituted (OH), meta/meta-disubstituted (Br), and ortho-/meta-disubstituted (OMe) aromatic rings proceeded efficiently and led to the corresponding products 3am−ar in excellent yields between 82 and 90%. Further, p-QMs 2s and 2u bearing naphth-2-yl and pyren-1-yl groups were also compatible and delivered the addition products 3as and 3au in 75% and 82% yields, respectively, whereas p-QMs 2t bearing an anthracen-9-yl group failed to yield 3at due to the possible steric effect of the anthracene ring. On the other hand, the corresponding p-QM 2v of terephthalaldehyde gave tandem double 1,6-conjugate addition product 3av in 83% yield. Finally, p-QMs 2w−x bearing the heterocyclic ring were used as a substrate instead of the carbocyclic aryl ring under standard conditions. First, N-tosylindole-substituted p-QM 2x gave the desired product 3ax in an excellent yield of 85%. Second, Nmethylindole-substituted p-QM 2w was employed as the substrate; however, no reaction occurred under the standard conditions. This result indicated that the reactivity of Nmethylindole-substituted p-QM 2w is lower than that of Ntosylindole-substituted p-QM 2x in the conjugate addition with 1a. We hypothesize that the electron-rich N-methylindole ring provides electrons by resonance, reducing the reactivity of the corresponding p-QM. With the optimized conditions in hand, we next evaluated the substrate scope of 2,3-disubstituted indole derivatives. As depicted in Scheme 2b, a wide range of 2,3disubstituted indoles were suitable reaction partners for p-QM 2d and provided the desired products in good to excellent yields. Pleasantly, 5-, 7-, and 8-membered fused-ring substrate triad 1b−d provided C6-alkylation products 3bd−dd with high yields and regioselectivity. Also, indoles 1e−g bearing dimethyl, ethyl/ methyl, and phenyl/ethyl on the 2,3-positions tolerated C6alkylation with high efficiencies (3ed−gd, 84−88%). With 6membered fused-ring indoles 1h−j bearing electron-donating groups (−Me and −OMe) and an electron-withdrawing group (−Cl) on the 5-position, the expected products 3hd−jd were obtained in high yields (73−87%). Notably, the N-protected 6membered fused-ring indoles 1k−l with N-methyl and N-benzyl groups also exhibited pleasing results, and the corresponding products 3kd−ld were obtained in high yields (86% and 74%). The N-acetyl-protected indole 1m failed to react with p-QM 2a under the standard reaction conditions. This result can be attributed to the π-electron-withdrawing effect of the acetyl substituent. The structures of the C(6)−H alkylation products were elucidated utilizing NMR spectroscopy and highresolution mass spectrometry (HRMS). As a representative example, the alkylation at the C6-position of 3ad was also determined by a nuclear Overhauser effect (NOE) study. NOE correlations between the C(7)−H and N(1)-H, hydrogens (see green arrow) agree with the alkylation depicted (Scheme 2a). To demonstrate the practicality of the current protocol, a gramscale synthesis of 3ad was also investigated under the optimized standard reaction conditions, and the yield of the desired product was only slightly decreased with no effect on regioselectivity (Scheme 2a). This result reveals that the current transformation can be used to synthesize C6-alkylated indoles with practical usefulness.
As depicted in Scheme 3, the N(1)−H alkylation reactions of various 2,3-disubstituted/fused indoles 1b−j mediated by In(OTf) 3 in THF were also explored under the optimized conditions (Table 1, entry 23). In this context, the N-alkylation product 4aa was generated and isolated in 85% yield (Scheme 3a). On the other hand, the p-QMs (2b,c, 2f−h, and 2j−l) containing different electron-donating and -withdrawing groups such as fluoro, chloro, methyl, tert-butyl, methoxy, nitro, trifluoromethyl, and diphenylamine at the para-position of the phenyl ring were suitable substrates to react with 1a in this transformation, affording the corresponding alkylation products 4ab−ac, 4af−ah, and 4aj−al in 63−90% yields.
Additionally, when the electron-donating and -withdrawing ortho-substituted p-QMs 2m−n were used as the substrates for the reaction, the alkylation process took place at the nitrogen atom in indole derivative 1a to afford the corresponding products 4am−an in 75% and 90% yields, respectively (Scheme 3a). Furthermore, the introduction of mono-and disubstitution at the meta-position did not significantly affect the yield of products 4ap (84% yield) and 4aq (82% yield) (Scheme 3a). p-QMs with other aromatic rings, such as naphthalene and pyrene, were also tested, giving satisfactory results. The corresponding products 4as and 4au were obtained in good yields (74% and 80%) (Scheme 3a). Unfortunately, in the case of anthracenylsubstituted p-QM 2t, no product (4at) formation was observed (Scheme 3a). It is assumed that steric hindrance due to the size of the anthracene ring blocks the approach of the nucleophile. When the terephthalaldehyde-p-QM 2v was used, both p-QM groups reacted and the desired product 4av was obtained in 82% yield (Scheme 3a). Both the 5-, 7-, and 8-membered fused-ringindoles 1b−d and indoles 1e−f including substituents such as dimethyl and ethyl/methyl on 2,3-positions can react efficiently with 4-(4-bromobenzylidene)-2,6-di-tert-butylcyclohexa-2,5dien-1-one (2d) to give the N(1)−H alkylation products 4bd−fd in 70−80% yields (Scheme 3b). Nevertheless, a complex mixture was obtained when 1g was reacted with 2d under standard conditions, and no desired product 4gd was isolated (Scheme 3b). Notably, substrates containing methyl, methoxy, and chloride groups at the C5-position of 6-membered fused-ring-indole 2a also provided the target products (4hd− 4jd) in moderate yields (40−64%) (Scheme 3b). To examine the synthetic potential of the protocol, postfunctionalizations of the C6-alkylation products were also performed (Scheme 4).
To elucidate the reaction mechanism of the regioselective alkylations of 2,3-disubstituted/fused indoles with p-QMs, a series of control experiments were then performed (Scheme 5). For this, we tested the interconversion between these two different alkylation products 3ad and 4ad. When the pure 4ad was subjected to the conditions for C6-alkylation, 3ad was isolated in 90% yield as a sole product (Scheme 5a). On the contrary, when the pure 3ad was subjected to the conditions for N-alkylation, the reaction did not proceed (Scheme 5a). Moreover, no conversion from 3ad to 4ad was observed, even when the N-alkylation reaction time was prolonged to 24 h. Under the thermal conditions in THF (Scheme 5b), the progress of the reaction was further investigated. The reaction between 1a and 2d with the 10 mol % In(OTf) 3 catalyst system in THF for 4 h in a sealed tube at 90°C surprisingly provided 3ad in 89% yield instead of 4ad (Scheme 5b). In addition, when the stirring of 4ad occurred under the same catalyst system in THF for 4 h in a sealed tube at 90°C, the conversion of 4ad to 3ad was observed in 87% yield (Scheme 5b). When the reaction between 2d with N-methyl-protected indole 1k for 12 h was conducted only in THF as a solvent, 3kd was isolated in 85% yield (Scheme 5c). These control experiments disclose that the N-alkylated product is probably the kinetically controlled product and the C-alkylated product is the thermodynamically controlled product. Furthermore, we also investigated whether our catalyst system would work on ortho-quinone methides (o-QMs), a reactive intermediate, to access the corresponding Cand N-alkylated products 10 and 11 (Scheme 5d). o-Hydroxybenzyl alcohol 9 was selected as a representative o-QM precursor to test In(OTf) 3 -catalyzed target reactions. Gratifyingly, the use of the reaction conditions for the Calkylation provided product 10 with 89% yield. However, an unstable or not isolable product was observed when the Nalkylation conditions were employed. These reactions indicate that the C-alkylation conditions are more suitable for the generation and trapping of the o-quinone methide intermediate 12.
From the above experimental results, it can be seen that the solvent, Lewis acid, and temperature strongly affect the course of the reaction. Based on the results of the optimization and control experiments, together with the related reports, 26,10a a plausible reaction mechanism for the N-and C-alkylation was proposed, as shown in Scheme 6a. Three pathways are possible via hydrogen bonding-and metal coordination-assisted catalyst (paths 1−3). With 1a and 2d as representative examples, this might involve initial activation of the substrates by In(OTf) 3 to form 12-membered transition state A (TS1). We assume that less membered transition state A facilitates the N-nucleophilic attack of 1a to generate the zwitterionic intermediate B and form the C−N bond. Finally, intramolecular proton transfer from N− H and the regeneration of the In(OTf) 3 catalyst affords the desired product 4ad (at room temperature in THF, path 1). In contrast, pathway 2 for the C−H functionalization involves a 15membered transition state C (TS2) between catalyst, donor, and acceptor compounds leading to an increase in the nucleophilicity at the C6-position of the indole ring. Next, the Friedel− Crafts-type nucleophilic addition of 1a to 2d generates intermediate D. Eventually, proton transfer, catalyst regeneration, and isomerization occur to afford the desired product 3ad (at room temperature in toluene or in heating conditions, path 2). However, the steric crowding 4ad would be selectively activated by In(OTf) 3 to yield the intermediate E, followed by C−N bond cleavage to form the intermediates F and G, which produces 3ad (under heating conditions or at room temperature in toluene, path c).
To gain detailed insights into the above reaction mechanism, we performed a series of DFT computations using the Gaussian 16 program 31 to explain the experimental results, particularly the interconversion between kinetic and thermodynamic products. The DFT method, hybrid functional B3LYP, was performed using model substrates bearing the bromine group (1a, 2d, 3ad, 4ad, intermediates, and transition states). 32−35 The energy diagram is depicted in Scheme 6b. We further computed their vibrational frequencies to characterize each stationary structure. In all computations, Ahlrichs' def2-TZVP basis set was utilized. 36 The solvation model based on density (SMD) was used to evaluate the effects of toluene (ε = 2.3741) and THF (ε = 7.4257) on the computed Gibbs energies. 37 As depicted in Scheme 6b, first, the interaction of 2d and In(OTf) 3 leads to complex H with a relative energy of −10.8 and −13.0 kcal mol −1 in toluene and THF, respectively. Subsequently, the nucleophilic attack by the indole 1a bifurcates into two paths which are N1-and C6-attacks. Our computations showed that N1alkylation proceeds apparently barrierless. For the N1alkylation, H → B, the computed reaction-free energies are 23.2 and 22.7 kcal mol −1 with toluene and THF, respectively. For C6-alkylation, H → D, while the computed reaction-free energies are 12.7 and 12.1 kcal mol −1 , the activation-free energies are 19.3 and 20.0 kcal mol −1 with toluene and THF, respectively. The Gibbs free energies indicated that N1alkylation in THF has lower energy by 0.5 kcal mol −1 compared to N1-alkylation in toluene. Additionally, we determined that the reaction barrier for C6-alkylation in toluene is lower in energy by 0.7 kcal mol −1 than C6-alkylation in THF. However, the N1-functionalized product 4ad is higher in energy compared to the C6-functionalized product 3ad by 7.6 and 8.2 kcal mol −1 in toluene and THF, respectively. The DFT results are rationally explained and supported our experimental observations that the C6-alkylation is more thermodynamically favorable, while the N1-alkylation is more kinetically favorable. Remote hydrogen bonding interactions between indole NH and catalyst lower the energy of the steps and promote the formation of the intermediates or transition states.

■ CONCLUSION
In summary, we have developed highly regiodivergent alkylation reactions of 2,3-disubstituted indoles with p-QMs via indium-(III)-catalyst in different solvents. The reactions feature kinetic and thermodynamic control, mild conditions, a broad substrate scope, and good functional group tolerance, providing an efficient method for the synthesis of C6-and N1-alkylated indole derivatives. Essentially, the reaction parameters (such as catalyst, solvent, and temperature) are found to be crucial for achieving in receiving high regioselectivity. Experimental and DFT studies supported that the reactions proceed through kinetic and thermodynamic control. Furthermore, the synthetic utility of the current approach and the alkylation products were demonstrated by scaling up to gram-scale, removing the tert-butyl group, and aryl−aryl coupling. Moreover, the protocol was successful for the late-stage modification of indomethacin as a biorelevant motif.   3 (10 mol %), and the mixture was stirred at room temperature for 12 h. After the reaction was complete (monitored by TLC), the solvent was removed under reduced pressure and the residue was purified by silica gel column chromatography using hexane/ EtOAc solvent mixture as the eluent.
Large-Scale Synthesis of 3ad. To a solution of indole 2a (514 mg, 3.0 mmol) and p-QM (2a−x) (1.23 g, 3.3 mmol) in toluene (0.1 M) was added In(OTf) 3 (169 mg, 0.3 mmol, 10 mol %), and the mixture was stirred at room temperature for 12 h. After the reaction was complete (monitored by TLC), the solvent was removed under reduced pressure and the residue was purified by silica gel column chromatography using hexane/EtOAc (95:5) to give the compound 3ad (1.39 g, 85% yield).