Reductive dearomative arylcarboxylation of indoles with CO₂ via visible-light photoredox catalysis

Abstract

Introduction

Catalytic reductive coupling of two different electrophiles (or cross-electrophiles coupling) has emerged as a powerful strategy to form C–C bonds¹. Compared with the conventional cross-coupling, transition-metal (TM)-catalyzed reductive coupling streamlines the preparation and handling of air- and moisture-sensitive organometallic reagents, thus showing significant advantages, including easily available substrates, operational simplicity, and high-step economy (Fig. 1a, path i)^2,3. Recently, great achievement has been realized in the three-component reductive coupling, that is, difunctionalization of one unsaturated bond with two electrophiles, which could rapidly generate highly functionalized skeletons by forming two new bonds (Fig. 1a, path ii)^4–9. Notably, if one electrophile is tethered with the unsaturated bond, a ring could be constructed via intramolecular cyclization^7–9. Although the transition-metal catalysis with two-electron transfer process (Fig. 1a) is powerful in tuning the reactivity and selectivity, there are still some challenges in the tandem reductive cyclization/cross-coupling. For example, the ipso-functionalization (Fig. 1a, path i), that is, direct two-component coupling, would be favored if the unsaturated bond is not reactive and migratory insertion step is not fast enough. Moreover, the generated organometallic intermediate could undergo other side reactions, including protonation, β-hydride elimination, or isomerization (Fig. 1a, path iii)^10,11. Therefore, it is highly desirable to develop other strategies, such as successive single-electron transfer (SSET) strategy (Fig. 1b), to prevent side reactions and realize such transformations with high selectivity.

Open in a separate window

Fig. 1

Strategies for tandem reductive couplings.

a Widely investigated transition-metal-catalyzed reductive coupling via two electron transfer process. (i) ipso reductive coupling, (ii) tandem reductive cyclization/coupling, (iii) β-hydride elimination or isomerization as other side reactions. b Rarely investigated radical-type reductive coupling via successive single-electron transfer process. O.A. oxidative addition, SET single-electron transfer. The gray square represents common organic structure. The pink circle represents organic electrophiles.

Dearomatization of (hetero)arenes is an efficient and powerful method to provide 3D cyclic skeletons, which exist widely in pharmaceuticals and bioactive natural products^12–16. Dearomative functionalization of indoles are particularly interesting to generate important indolines^15,16. Although many methods have been developed in this field^17–25, the dearomative reductive coupling of indoles with two electrophiles is still challenging due to the stability of carbon–carbon double bond within aromaticity and thus slow migratory insertion rate. In 2017, Qin et al. realized an elegant Ni-catalyzed asymmetric reductive hydroarylation of carbon–carbon double bond in indoles with water as proton source (Fig. 2a)²¹. When alkyl bromides were used as quenching electrophiles, however, the desired alkylarylation was not efficient and suffered from competitive side reaction, ipso cross-electrophiles coupling. With our continuous interest in green and sustainable synthesis with CO₂, which is abundant, recyclable and nontoxic C-1 source, we wondered whether we could realize the dearomative arylcarboxylation of indoles with aryl halides and CO₂ as electrophiles via dearomative process^26,27, generating the valuable and difficultly accessible indoline-3-carboxylic acids^28,29. However, due to the thermodynamic stability of both CO₂ and aromaticity, it is highly challenging to prevent the ipso-carboxylation of aryl halides and realize selective dearomative carboxylation via transition-metal catalysis^30–34. Inspired by the merit of visible-light photoredox catalysis^35–45, especially the visible-light-driven redox-neutral difunctionalization of alkenes with CO₂^46–49, we envisioned that the reductive dearomative difunctionalization might be achieved via a radical relay SSET strategy (Fig. 1b). We speculated that the visible-light-induced single-electron reduction of carbon–halide bond could generate a highly active aryl radical, which would prefer to undergo radical addition to the C2–C3 double bond in indoles instead of CO₂^50,51 (for visible-light-induced decarboxylation to access aryl radical, in which aryl radical neither reduced to aryl anion nor trapped by carbon dioxide, see ref. ⁵⁰; as long as radical reaction is concerned, carbon radical is hardly coupling with CO₂ unless CO₂ is converted to its corresponding radical anion, see ref. ⁵¹). The resulting benzylic radical might be further reduced via a second SET process to generate carbanion intermediate, which could react with weakly electrophilic CO₂ to deliver the desired dearomative arylcarboxylation product (Fig. 2b).

Open in a separate window

Fig. 2

Reductive dearomative difunctionalization of indoles.

a Tandem Ni-catalyzed asymmetric reductive dearomative functionalization of indoles by using proton or alkyl bromides as another electrophile. b Visible-light-promoted reductive dearomative arylcarboxylation of indoles with CO₂ via SSET process, in which carbon centered benzylic radical and benzylic anion as key intermediate. PC photocatalyst.

Herein, we report the success of such SSET process via visible-light photoredox catalysis, realizing the dearomative arylcarboxylation of indoles with CO₂ to give indoline-3-carboxylic acids with high selectivities (Fig. 2b). This reaction shows low loading of photocatalyst, generally good yields, mild reaction conditions (room temperature, 1 atm), good functional group tolerance, and broad substrate scope. Mechanistic studies indicate that the benzylic radicals and anions might be generated as the key intermediates. This strategy not only provides a direction for dearomative difunctionalization but also represents a way to realize selective tandem reductive cyclization/cross couplings, preventing the undesired two-component couplings and reductive Heck-type reactions via β-hydride elimination^10,11.

Results and discussion

Optimization study

With this hypothesis in mind, we investigated the reaction of ethyl 1-(2-bromobenzoyl)-1H-indole-2-carboxylate 1 with CO₂ under visible-light irradiation conditions (Table 1). When we used 1,2,3,5-tetrakis(carbazol-9-yl)-4,6-dicyanobenzene (4CzIPN) as a photocatalyst, Cs₂CO₃ as a base, N,N-diisopropylethylamine (DIPEA) as an electron donor, and DMSO as solvent (Table 1, entry 1), the desired carboxylic acid 2 was obtained in 88% isolated yield with high diastereoselectivity (d.r. >19:1). Control experiments revealed that CO₂, visible light, photocatalyst, and reductant were all essential for this transformation (Table 1, entries 2–5). Other photocatalysts, such as Ir- and Ru-complex, gave lower yields (Table 1, entries 7 and 8). When Et₃N was used as a reductant or 2 equiv of DIPEA was used, the yield was dropped to 75% (Table 1, entries 9 and 10), which might arise from lower efficiency for reductive quenching of the excited photocatalyst. Besides Cs₂CO₃, other bases, such as K₂CO₃ and KOPiv, were also efficient for the reaction, while the reaction with Cs₂CO₃ is the most efficient (Table 1, entries 1, 5, 11, and 12), which might quench the in-situ generated acidic components and stabilize the carboxylates. Moreover, the reaction could also be conducted in DMF to give the desired product in 75% yield (Table 1, entry 13). Compared with aryl bromide 1, aryl iodide 1’ and chloride 1” were also tested to give lower yields (Table 1, entries 14 and 15). In line with our expectations, not even trace of ipso-carboxylation of Ar–X (X = Cl, Br, I) bond was found in this reaction. The structure of 2 was further confirmed by X-ray crystal analysis, which indicated the cis-relationship between the newly generated two C–C bonds (please see ^{Supplementary Information} (^SI) for more information).

Table 1

Optimization of the reaction conditions^a.

Entry	Variation from standard conditions	Yieldb
1	None	88%
2	Without CO2	n.d.
3	Without light	n.d.
4	Without 4CzIPN	n.d.
5	Without DIPEA	Trace
6	Without Cs2CO3	55%
7	Ir[(dF(Me)ppy)2(dtbbpy)](PF6) instead of 4CzIPN	78%
8	Ru(bpy)3(PF6)2 instead of 4CzIPN	52%
9	Et3N instead of DIPEA	75%
10	DIPEA (0.4 mmol, 2.0 equiv)	75%
11	K2CO3 instead of Cs2CO3	71%
12	KOPiv instead of Cs2CO3	74%
13	DMF instead of DMSO	75%
14	ArI 1’ instead of ArBr 1	77%
15	ArCl 1” instead of ArBr 1	20%

Open in a separate window

LED light-emitting diode, DMF N,N-dimethylformamide, n.d. not detected.

^aReaction conditions: 1 (0.2 mmol), 4CzIPN (0.002 mmol), Cs₂CO₃ (0.6 mmol), DIPEA (0.6 mmol), DMSO (2 mL), 1 atm of CO₂, 30 W blue LEDs, room temperature (RT), 24 h.

^bIsolated yield.

Evaluation of substrate scope

With the optimized reaction conditions in hand, we first tested the substituent effect of the indole ring (Fig. 3). Notably, high diastereoselectivity (d.r. >19:1) was observed in most cases. We found that several esters (2–4) with different alkyl groups showed high reactivity. Moreover, a variety of amides also underwent this reaction well to give corresponding products 5–8 in moderate to high yields. In contrast, the substrates bearing phenyl (9) and electron-donating methyl group (10) at the C2 position, which might hamper the addition of aryl radical to indole, showed low reactivity. When no substitution was located at C2 position, the desired product 11 was obtained in 65% with poor diastereoselectivity of 2:1, which indicates that the substituention located at C2 position is necessary for high diastereoselectivity. Although all-carbon quaternary centers are difficult to synthesize due to the steric hindrance, we further challenged us using the substrate with C2- and C3 disubstitution. Different from previous transition-metal catalysis to give reductive Heck-type products^10,11, we could obtain the desired product 12 in high chemo- and diastereoselectivity, albeit with moderate conversion, indicating potential of this strategy in organic synthesis. Encouraged by these results, we further investigated other indole derivatives. The substrates with electron-donating groups or electron-withdrawing groups at the C5-position were all suitable for such a reaction to deliver the desired products in moderate to good yields. Many kinds of functional groups, such as methoxyl (14), fluoro (15), chloro (16), bromo (17), and ester group (18), were all tolerated well. The substrates bearing fluoro (19) and methoxyl (20) group at C6-position also worked well. To our delight, the reaction was not sensitive to steric hindrance, as the substrate bearing methyl group at C4-position (21) also worked well. Notably, the substrates bearing C5- and C6-disubstitution, such as dimethoxyl groups (22), were also tolerated. Besides various substitutions on the indole rings, we further tested the substitution effects to the aryl bromides (R⁴). Several functional groups were tolerated at para- and meta-position. Different from previous Pd-catalysis²⁰, in which the oxidative addition is sensitive to the steric hindrance, the methyl (25) and fluoro (27) groups at ortho-position of C–Br bond did not significantly affect the reaction, delivering the desired products in satisfactory yields. Moreover, difluoro (28) and dimethoxyl (29) substituents were also compatible. Notably, heterocycles, such as thiophene (30), which might inhibit transition-metal catalysis due to the strong coordination, could also survive in this reaction.

Open in a separate window

Fig. 3

Scope of substrates with substituents on indoles.

The standard reaction conditions were used, as shown in Table 1, entry 1. Isolated yields are presented.

In light of these results, we turned our attention to unactivated aryl halides, which are more electron rich and thus more challenging to undergo single-electron reduction by photocatalyst. Under slightly modified conditions (Fig. 4), a variety of aryl bromides and iodides underwent the reaction to generate desired products 31–38 in good yields, albeit with moderate diastereoselectivity. The single-crystal analysis of the major isomer of product 31 indicated the trans-relationship between the newly generated C–C bonds (please see Supplementary Information for more information), which is different from 2. As shown in Fig. 4, both unactivated aryl bromides and iodides could react smoothly with good functional group tolerance. Notably, carbon−halo bonds, especially the C–Br bond (36), which are reactive in transition-metal catalysis, survived in our reaction conditions.

Open in a separate window

Fig. 4

Scope of substrates bearing unactivated aryl bromides and iodides.

Reaction conditions: Indole derivatives (0.2 mmol), Ir-catalyst (0.002 mmol), Cs₂CO₃ (0.6 mmol), DIPEA (1.3 mmol), DMSO (2 mL), 1 atm of CO₂, 30 W blue LEDs, RT, 24 h, isolated yield.

Furthermore, 1-(2-bromophenethyl)-1H-indole-2-carboxylate and ethyl 1-(3-(2-bromophenyl)propyl)-1H-indole-2-carboxylate were tested in the standard reaction conditions (Fig. 5a) to give the corresponding products bearing 6-membered 39 and 7-membered ring 40 in 42% and 14% isolated yields, respectively, demonstrating that this strategy is not restricted to 5-membered ring formation. To further demonstrate the utility of this dearomative arylcarboxylation, we conducted a gram-scale reaction of 1 under standard conditions, which is also smooth to generate 2 in 80% yield (Fig. 5b). Moreover, facile and selective derivatizations, such as bromination and reduction, of carboxylic acid 2 provide easy access to valuable motifs (Fig. 5c).

Open in a separate window

Fig. 5

Synthetic applications.

a Synthesis of 6-membered and 7-membered ring-bearing products under the standard conditions. b Gram-scale synthesis of 2. c Transformations of the product 2. (i) bromination with Br₂ and AcOH. (ii) selective reduction of the C3-ester group by borane. (iii) reduction of amide and ester groups by LiAlH₄.

Proposed mechanism and mechanistic studies

To gain more insight into the reaction mechanism, we further did a variety of control experiments (please see Supplementary Figs. ¹–⁹ for more information). First, we found that the desired reaction of 1 was significantly inhibited by 2,2,6,6-tetramethyl-1-piperdinyloxy (TEMPO). Meanwhile, a TEMPO-radical adduct 43 was isolated, indicating that benzylic radical was involved in the reaction (Fig. 6a). As transient aryl radical can be trapped by TEMPO⁵², the fact that no trapping of the aryl radical with TEMPO indicated that the radical cyclization step might be fast. Second, we did isotope-labeling studies (Fig. 6b). When DMSO-d₆ was used as solvent in the reaction of 44, no deuterium incorporation in the hydroarylation product 45 was observed, thus ruling out the possibility of hydrogen-atom transfer with DMSO⁵³. We found a 82% deuterium incorporation at C3-position of 45 when 20 equivalent of D₂O was added under N₂ atmosphere, indicating the formation of benzylic anion intermediate. Furthermore, with slightly modified the reaction conditions, 4-fluorobenzaldehyde 46 was applied as electrophile instead of CO₂ to deliver the desired product 47 in 81% yield (Fig. 6c), further confirming that the benzylic anion might be involved in the reaction.

Open in a separate window

Fig. 6

Preliminary mechanistic studies.

a Trapping experiment by radical scavenger 2,2,6,6-tetramethyl-piperidinyloxyl (TEMPO), supporting that radical process might be involved. b Isotopic labeling experiments in DMSO-d₆ or in the presence of different amounts of deuterated water, suggesting that benzylic anion was involved. c 4-fluorobenzaldehyde was used as an electrophile instead of CO₂, further indirectly confirming the formation of benzylic anion intermediate.

Moreover, Stern–Volmer fluorescence quenching experiments were performed (Fig. 7, please see Supplementary Fig. ¹⁰ for more information). The luminescence of 4CzIPN at λ_max = 536 nm was readily quenched by DIPEA with a slope of 512.5, which was much more significant than indole 1 (1.9) or the combination of indole 1 and Cs₂CO₃ (17.4). These results suggested that DIPEA had the priority to undergo SET with excited 4CzIPN.

Open in a separate window

Fig. 7

Optical experiment with fluorescence spectrum.

a Steady-state Stern–Volmer experiment of 4CzIPN and DIPEA, the luminescence of 4CzIPN was readily quenched by DIPEA. b Stern–Volmer fluorescence quenching experiments using 4CzIPN with DIPEA, 1 as well as 1 and Cs₂CO₃.

Based on above-mentioned results and previous work, we proposed the possible mechanism for the reaction of 1 (Fig. 8). The reductive quenching of photo-activated 4CzIPN (E_1/2 [4CzIPN/4CzIPN^•−] = −1.21 V vs. SCE in MeCN)⁵⁴ by DIPEA (E_1/2^Ox = +0.63 V vs. SCE in DMF)⁵⁵ leads to the corresponding radical anion 4CzIPN^•− and the radical cation DIPEA^•+. The reduced 4CzIPN^•− undergoes a single-electron reduction of aryl bromide (ArBr) 1, producing the [ArBr]^•− radical anion I and regenerating the neutral 4CzIPN to complete one catalytic cycle. Then, the generated [ArBr]^•− radical anion undergoes fragmentation to release a bromide anion and an aryl radical II^56,57, which undergoes facile intramolecular radical addition to the C2–C3 double bond of indole to afford the benzylic radical III. The following SET with 4CzIPN^•− delivers the carbon anion intermediate IV^58,59, which could undergo nucleophilic addition to CO₂^60–63. Following protonation provides the dearomative arylcarboxylation product 2. However, at this stage we could not exclude another possibility that I undergoes radical cyclization via C–C bond formation and following re-aromatization via release of bromide anion to give III. Further experimental and computational investigations are necessary to explore more details for this process.

Open in a separate window

Fig. 8

Mechanistic proposal for the arylcarboxylation of 1.

PC = 4CzIPN.

In conclusion, we report a strategy of successive single-electron transfer (SSET) process for the tandem reductive cyclization/cross couplings, realizing the first dearomative arylcarboxylation of indoles with CO₂ via visible-light photoredox catalysis⁶⁴ (during the revision of this paper, Li reported an elegant and very similar visible-light photoredox-catalyzed reductive arylcarboxylation of styrenes, see ref. ⁶⁴). Notably, this reaction is highly chemoselective, as common side reactions, such as ipso coupling of aryl halides and β-hydride elimination, are avoided. This reaction shows high selectivity, low loading of photocatalyst, generally good yields, mild reaction conditions (room temperature, 1 atm), good functional group tolerance and broad substrate scope, providing great potential for the synthesis of valuable but difficultly accessible indoline-3-carboxylic acids. Mechanistic studies indicate that the benzylic radicals and anions might be generated as the key intermediates. Further application of this strategy in other reductive couplings with diverse electrophiles is underway in our laboratory.

Methods

Supplementary information

Acknowledgements

Author contributions

D.G.Y. and W.J.Z. conceived and designed the study, and wrote the paper. W.J.Z., Z.H.W., L.L.L., Y.X.J., K.G.C., T.J., Y.L., and G.M.C. performed the experiments and mechanistic studies. Z.H.W. performed the crystallographic studies. All authors contributed to the analysis and interpretation of the data.

Data availability

The authors declare that the data supporting the findings of this study are available within the article and its ^{Supplementary Information files}. Extra data are available from the author upon request. The crystallography data have been deposited at the Cambridge Crystallographic Data Center (CCDC) under accession number CCDC: 1950734 (compound 2) and CCDC: 1972327 (compound 31) can be obtained free of charge from www.ccdc.cam.ac.uk/getstructures.

Competing interests

The authors declare no competing interests.

Footnotes

References