Nickel-Catalyzed Dearomative Arylboration of Indoles: Regioselective Synthesis of C2- and C3-Borylated Indolines

Abstract

Indole dearomatization is an important strategy to access indolines: a motif present in a variety of natural products and biologically active molecules. Herein, a method for transition-metal catalyzed regioselective dearomative arylboration of indoles to generate diverse indolines is presented. The method accomplishes intermolecular dearomatization of simple indoles through a migratory insertion pathway on substrates that lack activating or directing groups on the C2- or C3-positions. Synthetically useful C2- and C3-borylated indolines can be accessed through a simple change in N-protecting group in high regio- and diastereoselectivities (up to >40:1 rr and >40:1 dr) from readily available starting materials. Additionally, the origin of regioselectivity was explored experimentally and computationally to uncover the remarkable interplay between carbonyl orientation of the N-protecting group on indole, electronics of the C2–C3 π-bond, and sterics. The method enabled the first enantioselective synthesis of (−)-azamedicarpin.

Graphical Abstract

INTRODUCTION

Indolines are prevalent scaffolds that appear in many natural products and biologically significant molecules. For example, the monoterpenoid indole alkaloids include many examples of highly functionalized indolines.¹ Additionally, indolines appear in FDA-approved pharmaceuticals, including naturally derived vinblastine and pentopril (Scheme 1A).² Due to the ubiquitous nature of this motif, various strategies to access indolines have been developed over the last 70 years.³ Of these, indole dearomatization has emerged as an important strategy for preparing indolines;³ this is, in part, due to the wide availability and synthetic accessibility of indoles. Significant effort has led to the development of a multitude of dearomatization strategies, which include cycloadditions, hydrogenation, radical-mediated processes, or reactions of electrophilic indoles.³ Perhaps the most well-established indole dearomatization strategy exploits the inherent nucleophilicity of indole through electrophilic attack at the C3-position. The resulting iminium species is trapped by a nucleophile to generate the indoline (Scheme 1B).³

Scheme 1.

Synthesis of Indolines by Dearomatization

More recent efforts have focused on the dearomatization of indoles through the aid of transition-metal catalysis.^4–6 Many of these processes take advantage of the inherent nucleophilicity of indole to direct the regioselectivity of the dearomatization.⁴ However, an alternative approach that is not reliant on adding electrophiles to C3 and nucleophiles to C2 involves a migratory insertion of the C2–C3 π-bond of indole into a [M]-R bond to access diverse indoline products (Scheme 1C).⁵ The intramolecular version of this strategy has been advanced in a series of elegant studies from Jia, Lautens, and others.⁶ While the selectivity of these reactions is not reliant on inherent reactivity, the tether dictates the regiochemical outcome of the reaction.

Within the field of transition-metal catalyzed indole dearomatization, carboboration variants have also been investigated (Scheme 1D).^{4g–j,k–m,5,6k,p,7} These processes are particularly valuable, as in addition to generating a new C–C bond, the resulting C–B bond can be transformed to various functional groups through stereospecific transformations.⁸ One approach to carboboration involves 1,2-metalate rearrangements of boronates induced by various electrophiles (often activated by transition metals).^4h–j,k–m While these reactions have been developed with several distinct classes of electrophiles, they do rely on the inherent nucleophilicity of indoles as outlined in Scheme 1B. The migratory insertion approach illustrated in Scheme 1C has also been advanced with carboboration. Ma and Xu^6p as well as Jia and Lautens^6k have developed intramolecular carboboration reactions. Despite the value of intramolecular indole dearomatization, intermolecular variants that would allow for greater synthetic versatility remain quite limited.⁹ Ito has shown that substrates with extended conjugation, such as indole-carboxylate esters, undergo Cu-catalyzed carboboration.¹⁰ Additionally, Engle has demonstrated that a tethered 8-aminoquinoline directing group can facilitate a Pd-catalyzed dearomative intermolecular arylboration of indole.¹¹ While these examples are important advances in indole dearomatization, the requirement of an activating group (e.g., carboxylate) or a tethered directing group to achieve satisfactory reactivity limits the broad utility of these methods and only allows for one regioisomer to be generated.

Recently, our group has developed a Ni-catalyzed 1,2-arylboration reaction of alkenes.^12,13 On the basis of the principles learned, we envisioned advancing this area to the challenging problem of dearomatization. In particular, dearomative arylboration of indoles without the aid of activating or directing groups at C2/3 was undertaken, as it would allow for exploration of unprecedented and challenging reactivity.¹⁴ Herein, we present the realization of this goal. Most importantly, since the method is not reliant on substituents at C2/3, both regioisomers of product can be accessed to synthesize valuable C2- and C3-borylated indolines (Scheme 1E).

Mechanistically, we hypothesized that [Ni(I)]-Bpin complex II would form through base-assisted transmetalation with B₂pin₂ (Scheme 2), followed by syn-addition across the C2–C3 π-bond of indole to generate two possible regioisomers (III or IV).¹² Reaction with the arylbromide electrophile would form arylboration product V or VI. A significant challenge in the development of this intermolecular dearomatization could be overcoming the energy of aromaticity, ~18.7 kcal/mol,¹⁵ at a rate faster than Miyaura borylation. Additionally, controlling the regioselectivity of the migratory insertion event was identified as an opportunity to access both regioisomers of the product (V or VI).

Scheme 2.

Proposed Catalytic Cycle

RESULTS AND DISCUSSION

Preliminary investigation of a variety of N-protected indoles revealed that N-Boc derivative 6 reacted to produce arylboration product in moderate yield and high regio- and diastereoselectivity under Ni-catalyzed arylboration conditions (Scheme 3). The regio- and diastereoisomer of the product derived from 6 was established by X-ray crystallography as the syn-C2-borylated product A. Remarkably, a reversal in regioselectivity was observed when N-Piv-indole 7 was subjected to the reaction conditions, forming C3-borylation product B as a single regioisomer, albeit in low yield. This regiodivergence arising from a simple swap between two similar N-protecting groups was both mechanistically intriguing and synthetically desirable, as both C2- and C3-borylated indolines could be accessed from readily available indoles. It should be noted that while change of the N-protecting group to alter the regiochemical course of indole functionalization reactions is known (e.g., N-Ts vs N-H^16a,d, N-CO₂Me vs N-Bn,^16b N-H vs N-TIPS,^16c N-H vs N-Boc^16e the surprisingly subtle change (i.e., N-Boc vs N-Piv) demonstrated here has, to our knowledge, no precedent.

Scheme 3. Initial Investigations^a.

^aReactions run on a 0.2 mmol scale. All yields, rr’s, and dr’s were determined by ¹H NMR analysis of the crude reaction mixture relative to an internal standard.

Optimization of the reaction of N-Boc-indole revealed that increased quantities of NaOt-Bu and B₂pin₂ led to higher yields (Table 1, entry 1). While lower equivalents of PhBr was tolerated, the use of 3.0 eq was chosen for scope investigation when less reactive ArBr were evaluated (Table 1, entry 4). The cosolvent mixture of THF and DMA was crucial to obtain the optimal yields (Table 1, entries 2 and 3). In addition, lower equivalents of NaOt-Bu and B₂pin₂let to lower yields (Table 1, entry 5). Finally, the catalyst loading could be reduced with minimal impact on yield (Table 1, entries 6 and 7).

Table 1.

Optimization of Reaction Conditions^c

Entry	N-R	Change from above conditions	% yield	A:B
1	N-Boc	none	84	16:01
2	N-Boc	THF only	28	8:01
3	N-Boc	DMA only	69	16:01
4	N-Boc	1.5 equiv PhBr	85	20:01
5	N-Boc	1.5 equiv NaOtBu / 2 equiv B2pin2	71	13:01
6	N-Boc	2 mol% Ni(DME)CI2	82	20:01
7	N-Boc	10 mol% Ni(DME)CI2	79	19:01
8	N-Piv	none	31	1:>40
9	N-Cy-acyl	none	72	1:4
10	N-Boc-piperidine-acyl	none	10	1:9
11	N-Boc-piperidine-acyl	30 °C	31	1:9
12	N-Boc-L-proline	none	76a	1:>40
13	N-Boc-L-proline	3 equiv. B 2 pin 2 , pre-stir with NaO t Bu	77 b	1:>40

^cReactions run on 0.2 mmol scale. All yields, rr’s, and dr’s were determined by ¹H NMR analysis of the crude reaction mixture relative to an internal standard.

^aProduct formed in 9:1 dr, 89:11 er, 90% es (major diastereomer).

^bProduct formed in 9:1 dr, 98:2 er, 99% es (major diastereomer)

With respect to the C3-borylation process, despite extensive investigations, optimization of the reaction of the N-Piv indole was unsuccessful (Table 1, entry 8). It was found that the yield of the reaction could be increased if N-Cy-acyl-indole was used; however, the regioselectivity was diminished (Table 1, entry 9). While, the increased yield is likely the result of a less sterically demanding N-protecting group, sterics clearly play an important role in regiocontrol. At this stage, the hypothesis was advanced that the more electron-withdrawing N-Piv (vs N-Boc) favors C3 borylation, and that perhaps use of a more electron-withdrawing N-protecting group of similar size to N-Cy-acyl may be optimal. Studies along these lines eventually led to the finding that reaction of N-Boc-piperdine-acyl-indole led to superior regiochemical outcome compared to N-Cy-acyl (Table 1, entries 10–11). Further design of the N-protecting group by “moving” the N-Boc substituent closer to the indole, led to use of N-Boc-L-proline, which resulted in good yield and regioselectivity (Table 1, entry 12). Furthermore, this substrate was particularly attractive because it allowed for control of stereochemistry with respect to the existing stereogenic center (9:1 dr). The major diastereomer was confirmed by X-ray crystal structure analysis (see the SI for details). One final aspect of optimization was needed, as it was found that the N-Boc-L-proline was epimerizing under the reaction conditions. Since epimerization likely occurs due to the presence of NaO^tBu prior to complexation with B₂pin₂, the conditions were modified to prestir B₂pin₂ with NaO^tBu before adding the other reagents. With this modified procedure, epimerization was minimized, forming product in 98:2 er (Table 1, entry 13).

After optimization of the reaction conditions for both the C2- and C3-borylation reactions, the scope was evaluated. With respect to the C2 borylation reaction, various substituents were tolerated in the C4–C7 positions of the N-Boc-indoles, producing diverse indoline products as the syn-diastereomers (Scheme 4). 4-Fluoroindole reacted to obtain product 9 in moderate yield, likely due to unfavorable steric interaction with the nearby fluorine in the migratory insertion step. However, 5-fluoroindoline 10 and chloro-substituted indolines 11–13 were formed in good yields and regioselectivities. Additionally, electron-rich 5- and 6-methoxyindoles underwent dearomatization in excellent yields (product 15–16), though 6-methoxyindoline 16 formed in diminished regioselectivity (7:1 rr). This observation can perhaps be attributed to destabilization of the forming benzylic alkyl-[Ni] complex in the transition state by electron donation from the methoxy substituent, resulting in higher rates of formation of the other regioisomer. The reverse effect was observed with an amide-substituent in the C6 position leading to formation of a single observable regioisomer of product 17, likely due to a stabilizing effect on the forming electron-rich alkyl-[Ni] bond. Finally, substitution at the C7-position was tolerated with Cl, Me and amide substituents (product 13, 14, 18), or in the form of a cyclized amide (product 19).

Scheme 4. Reaction Scope^d.

^dYields are reported as the average of two isolated experiments, and products were isolated as single diastereomers. Crude yields, dr’s, and rr’s are reported as the average of two experiments in parentheses and were determined through ¹H NMR analysis of the crude reaction mixture relative to an internal standard. ^a4.5 eq B₂pin₂, 4 eq NaOt-Bu used. ^b1.5 eq ArBr used. ^cReaction run at 30 °C.

Excitingly, 2-substituted N-Boc-indoles underwent efficient dearomatization to produce the opposite regioisomer containing a N-substituted quaternary carbon with a C3-benzylic boronic ester (Scheme 4). 2-Arylindoles, including substrates with electron-neutral (20), electron-deficient (21), and electron-rich (22) aryl substituents were effective in the dearomative arylboration. Additionally, a Bdan substituent was tolerated in the C2-position, resulting in a highly functionalized indoline product (23). The products were formed as single diastereomers in high regioselectivity. It is proposed that the reversal of regioselectivity is the result of addition of the [Ni(I)]-Bpin to the C2–C3 π-bond to minimize steric pressure between the Bpin unit and the C2 substituent. Formation of a tertiary alkyl-[Ni] complex is rare^12,17 and this species is likely stabilized by both the α-aryl and α-carbamate groups.^12c,18 With respect to the limitations, 3-phenyl indole was unreactive and alkyl substituents were not tolerated at the C2 and C3 positions.

Investigation of the aryl bromide scope of the dearomative arylboration reaction revealed that methoxy- (26) and fluoro- (28) substitutions in the para-position as well as para-aniline derivatives (29 and 32) reacted efficiently (Scheme 4). Additionally, acetal and tertiary amine functionalities were well tolerated (33 and 34). The aryl bromides could also be substituted in the meta-position with oxygen and nitrogen functionalities (24-25, 30-31). Finally, heterocycles such as carbazole (35) and N-Me-indole (36) participated in the arylboration reaction, with no observable dearomatization of the N-Me-indole ring.

However, reactions with electron-deficient and sterically hindered aryl bromides resulted in significant Miyaura borylation and minimal dearomatization. We hypothesized that an alternative N-protecting group could be used to tune the electronics of indole in order to increase the rate of the dearomatization reaction relative to Miyaura borylation. N-Dimethylamide-protected indole was identified to enable reaction with challenging electrophiles, including with electron-deficient aryl bromides (products 37 and 38), sterically hindered 2-bromotoluene (product 39), and a heterocyclic aryl bromide (product 40) (Scheme 4). Notably, all yields and regioselectivities were superior to those observed with N-Boc-indole (see the SI for details). For example, reaction 2-fluorobromobenzene did not lead to desired product (<2% by NMR) with N-Boc-indole. A significant increase in regioselectivity was also observed with 6-methoxy N-dimethylamide indole (product 41, 28:1 rr), in contrast to the lower selectivity observed with N-Boc-6-methoxy indole (product 16, 7:1 rr).

With respect to the C3-borylation process, in addition to simple indole 42, electron-donating methoxy-groups were tolerated in the C5 and C6 positions of the indole ring (products 44 and 47), and halides such as fluorine and chlorine were tolerated in the C5 (46) and C6 (45) positions, respectively (Scheme 4). Additionally, aryl bromide electrophiles with diverse functionality reacted smoothly, including electron-donating methoxy-groups (48 and 50), an electron-withdrawing chlorine (49), and N-containing heterocycles (51 and 52). In all cases the diastereoselectivity with respect to the existing proline stereogenic center was 8:1 or greater. In addition, single diastereomers of product could be easily isolated by silica gel column chromatography. Finally, a model that rationalizes the observed stereochemical outcome is shown in Scheme 4.¹⁹ It is proposed that minimization of A(1,3) strain for both amides is the primary driver for the illustrated conformation. In this conformation, the bottom face of the indole is partially blocked by the N-Boc unit, while the top face remains accessible.

To demonstrate the utility of the method, the dearomative arylboration reaction of N-Boc-indoles was amenable to glovebox-free, gram-scale reaction setup using standard Schlenk techniques. The indoline products 8 and 53 were obtained with no appreciable difference in yield from that achieved with glovebox setup on 10 × smaller scale (Scheme 5A). The arylboration products could be further elaborated through functionalization of the Bpin moiety to afford diverse indoline products. Matteson homologation produced products 54 and 59. Metal-free cross-coupling with 2-bromoquinoline delivered diarylated indoline 56, and with benzofuran produced highly substituted indoline 58. Boc-deprotection delivered the N-H indoline 55. Additionally, the benzylic Bpin moiety could be oxidized with sodium perborate to deliver benzylic alcohol 57.

Scheme 5. Synthetic Utility of the Arylboration Reaction.

^aReaction conditions for Scheme 5A: arylboration: indole (5 mmol, 1 equiv), Ni(DME)Cl₂ (0.25 mmol, 5 mol %), ArBr (15 mmol, 3 equiv), B₂pin₂ (12.5 mmol, 2.5 equiv), NaOt-Bu (10 mmol, 2 equiv), THF/DMA (4:1, 0.1 M),10 °C, 24 h. a) n-BuLi, CH₂Br₂, THF, −78 °C to −20 °C.b) TFA, DCM, 0 to 25 °C. c) 2-bromoquinoline, n-BuLi, Et₂O, −78 °C; Troc-Cl, −78 to 25 °C; NaOH, H₂O₂, THF, 25 °C. d) NaBO₃.H₂O, THF/H₂O (1:1), 25 °C. e) n-BuLi, benzofuran, THF, −78 °C; NBS. f) n-BuLi, CH₂Br₂, THF, −78 to 25 °C. ^bReaction conditions for Scheme 5B: arylboration: indole (2 mmol, 1 equiv), Ni(DME)Cl₂ (0.1 mmol, 5 mol %), ArBr (6 mmol, 3 equiv), B₂pin₂ (6 mmol, 3 equiv), NaOt-Bu (4 mmol, 2 equiv), THF/DMA (9:1, 0.1 M), 30 °C, 18 h. g) n-BuLi, CH₂Br₂, THF, −78 to 25 °C. h) NaBO₃.H₂O, THF/H₂O (1:1), 25 °C. (i) NaH, DMF, 25 °C. j) NaOMe, MeOH/MeCN, 80 °C. k) Pd(OH)₂, H₂, EtOAc, 25 °C

We envisioned that the C3-borylative dearomatization reaction would be a useful tool to enable the first enantioselective synthesis of (−)-azamedicarpin (63), a pterocarpan natural product that has previously shown modest activity against leukemia cell lines and inhibition of bacterial and fungal biofilm formation (Scheme 5B).²⁰ Arylboration product 61 derived from 6-methoxy indole 60 and a complex arylbromide was isolated as a single diastereomer. Subsequent Matteson homologation, oxidation, and S_NAr led to formation the tetracyclic core. Deprotection of the N-Boc-proline and benzyl ether formed (−)-azamedicarpin (63) in 10% overall yield over six steps.

The origin of regiodivergence in the dearomative arylboration reaction was explored through experimental and computational investigations at the ωB97X-D/SDD–6–311++G(d,p)/SMD(THF)//B3LYP-D3(BJ)/SDD–6–31+G(d,p) level of theory. The computed Gibbs free energy profiles of the reaction with N-Boc- and N-Piv-indoles indicate that the migratory insertion into a Ni(I) boryl species is irreversible and thus determines the regioselectivity (see SI Figures S9–S12).^12b The migratory insertion in the C2- and C3-borylation pathways with N-Boc- and N-Piv-indoles involve different rotamers of the π-indole nickel complex (C1–C4) (Scheme 6A, see SI for other higher-energy transition state conformers). In agreement with experiment, the reaction with N-Boc-indole prefers C2-borylation via TSA from conformer C2 in which the carbonyl is syn-periplanar with the C2-position of the indole (Scheme 6A). The C2-borylation transition state TSA is stabilized because the nucleophilic boryl ligand²¹ prefers addition to the more electrophilic C2 position, and this process generates a stable benzylic alkyl-[Ni] species.^12c The lowest energy pathway for C3-borylation of N-Boc-indole proceeds through TSA1 from a different carbonyl conformer C1 with the carbonyl anti-periplanar with the C2-position of the indole, indicating that carbonyl conformation influences regioselectivity. In the migratory insertion with N-Piv-indole, the computed regioselectivity reverses to favor C3-borylation via TSB1 (ΔG^‡ = 16.7 kcal mol⁻¹), in agreement with experimental observation. The lowest energy pathway for C2-borylation proceeds through the syn-periplanar conformer C4, which is significantly higher in energy due to steric strain between the fused benzene ring and the N-Piv tert-butyl group. The C3-borylation proceeds via the anti-periplanar conformer C3 in which the steric repulsion with the fused benzene ring is diminished.

Scheme 6. Mechanism Studies^a.

^aArylboration conditions for Scheme 6B and C: indole (0.5 mmol, 1 equiv), Ni(DME)Cl₂ (0.025 mmol, 5 mol %), PhBr (1.5 mmol, 3 equiv), B₂pin₂ (1.25 mmol, 2.5 equiv), NaOt-Bu (1.0 mmol, 2 equiv), THF/DMA (4:1, 0.1 M), 10 °C, 24 h. Values given in red and parentheses are relative Gibbs energies in kcal/mol.

To probe the origins of regiodivergence more deeply, two hypotheses were advanced. The first is that the t-butyl group of N-Piv-indole blocks addition of Bpin to C2, therefore resulting in C3-borylation (Scheme 6B). To explore this hypothesis, N-neopentyl acyl-indole 64 which is iso-steric with N-Boc-indole, was subjected to the reaction. C3-borylation product 65 formed preferentially, albeit in diminished regioselectivity, indicating that sterics are important but not the primary driver for regiodivergence. The second hypothesis is that carbonyl orientation could influence the electronics of the C2–C3 π-bond of indole to change regioselectivity, which is supported by the previously discussed transition state geometries. Conformationally locked N-Piv-indole derivative 66 was designed to constrain the carbonyl in the same syn-periplanar orientation as the reactive conformer of N-Boc-indole (Scheme 6C). Upon subjection to the reaction, the regioselectivity was flipped to C2-borylation (product 67), indicating that carbonyl orientation is a major factor in determining regioselectivity.

Intrigued by the remarkable influence that carbonyl orientation has on the regioselectivity, we calculated CHelpG²² atomic charges for C2 and C3 of both conformers of N-Boc- and N-Piv-indoles (Scheme 6D). For N-Boc-indole, both conformations have comparable energies but different polarities of the C2–C3 π-bond. The syn-periplanar conformer I2 has a more polarized C2–C3 π-bond and prefers C2-borylation due to polarity matching with the nucleophilic boryl ligand at the more electrophilic C2 position. Similarly, the syn-periplanar conformer of the N-Piv-indole (I4) has a more polarized C2–C3 π-bond but is much higher in energy due to steric repulsions. Thus, lower energy conformer I3 is the reactive conformer and undergoes C3-borylation in the lowest energy pathway. The less polarized C2–C3 π-bond in I3, and thus less electrophilic C2 position, combined with a steric interaction between the approaching Bpin moiety and the pivalate tert-butyl group in I3 favors C3-borylation. CHelpG atomic charges of the various N-protected indoles investigated in this study were also calculated (Scheme 6D). Indoles undergoing C2-borylation all have a more polar C2–C3 π-bond (Δe > 0.34), indicating electronic control of regioselectivity. Indoles undergoing C3-borylation have a less polar C2–C3 π-bond (Δe ≤ 0.3) and both electronics of the C2–C3 π-bond as well as sterics appear to play a role in the observed regioselectivity. Calculated C2–C3 π-bond polarizations of various C6-substituted N-Boc-indoles are also linearly correlated with experimental regioselectivities, further supporting the importance of C2–C3 π-bond polarization on regioselectivity.²³

CONCLUSION

This work presents an intermolecular transition-metal catalyzed dearomatization of simple indoles proceeding through a migratory insertion pathway that does not require extended conjugation or a tethered directing group on the C2- or C3-positions of indole. The process is regiodivergent due to the remarkable interplay between carbonyl orientation of the N-protecting group, C2–C3 π-bond polarization, and sterics. The C2- and C3-borylated indoline products are synthetically useful intermediates that are easily elaborated to a variety of indolines as demonstrated in the first enantioselective synthesis of (−)-azamedicarpin.