Introduction

The venerable Diels-Alder (D-A) reaction featuring [4 + 2]-cycloaddition of a conjugated diene with a dienophile is highly regarded in organic chemistry1,2. On the other hand, an efficient multicomponent reaction of readily accessible reactants would give a large library of products in rich structural diversity3. In this context, a combined strategy on aza-D-A reaction involving in situ formation of imines serving as aza-diene has long been invented4,5. By constrast, the corresponding reaction through in situ generation of conjugated dienes by coupling of two readily available components remains unknown.

ortho-Quinodimethanes (o-QDMs) are useful, short-lived transient synthetic intermediates. They often serve as highly reactive diene species to participate in D-A reactions6,7,8,9,10. Given the perspective on synthesis of indole alkaloids or construction of frameworks containing tetrahydrocarbazole motifs, the development of efficient approaches to generate indolo-2,3-quinodimethane under mild conditions has received considerable attention in the past decade11,12,13,14,15,16,17,18,19,20,21,22. Among these, the identification of catalytic methods was of particular interests, as the reactivity of the in situ formed o-QDMs and the stereochemistry of the products could be modulated by a specific catalyst15,16,17,18,19. For example, in 2011, Melchiorre and co-workers found that acidity of the proton at the 2-methyl N-Boc protected indole moiety could be enhanced in presence of a proper aminocatalyst. Thus, active indole-2,3-quinodimethane could be generated facilely15. Shortly after this elegant work, Chen and co-workers demonstrated a Brønsted acid mediated reaction of 2- methyl-3-indolylmethanols, which could also generate active indole-2,3-quinodimethane intermediates in situ by releasing H2O as byproduct16. By using N-heterocyclic carbene as catalyst, Chi et al. reported an innovated method on functionalization of benzylic C(sp3)-H bonds of heteroaryl aldehydes through the formation of active o-QDM species17. By utilization of intramolecular Heck reaction, Fuwa and Sasaki employed α-phosphono enecarbamate as indole-2,3-quinodimethane precursor20. Later on, Mukai and co-workers developed a palladium-catalyzed reaction of allenylanilines to generate similar o-QDMs21. In 2014, Zhang and co-workers described an elegant strategy for in situ generation of furan-based o-QDMs from acyclic 2-(1-alkynyl)-2-alken-1-ones through a gold-catalyzed dehydrogenative cyclization of 2-(1-alkynyl)-2-alken-1-ones18. In spite of these remarkable advances, all o-QDM species described were derived from a single reactant (Fig. 1a, left). Some of the o-QDM precursors required multi-step’s synthetic manipulation. The structural diversity of the desired o-QDM intermediates was largely depend on the availability of the corresponding single reactant in hand. In view of the significance of diversity-oriented synthesis3, the development of intermolecular reactions of easily available starting materials to generate this highly reactive species is in high demand (Fig. 1a, right).

Fig. 1: Catalytic reactions for heterocyclic ortho-quinodimethane formation.
figure 1

a Catalytic approaches to generate heterocyclic o-QDMs. b This work: pallladium-catalyzed carbene bridging o-QDM formation. c Bioactive molecules and functional materials containing carbazole and indole nucleus.

Full size image

Because of the high economy and overall efficiency, transition-metal-catalyzed C-H bond activation has evolved as efficient tool to transform readily available materials into value added products23,24,25,26. In line with our recent research interests in carbene bridging C-H activation27,28,29,30,31,32,33,34,35, we sought to develop a palladium carbene migratory insertion36,37,38,39,40,41 enabled C(sp3)-H bond activation to generate indole-2,3-quinodimethane intermediate by the reaction of aryl halides 1 with N-tosylhydrazones 2 derived from 2-methylindole-3-carboxaldehydes (Fig. 1b). We assumed that the dearomatization of intermediate III could be relatively facile due to the weak aromaticity of the indole moiety42,43,44,45,46,47 and the following rearomatization via D-A reaction of in situ formed o-QDM intermediate with proper dienophiles. Herein, we describe a carbene migratory insertion enabled dearomative C(sp3)-H bond activation through palladium-catalyzed coupling of N-tosylhydrazones with aryl halides. The synthetic utility of this strategy is demonstrated by trapping of o-QDMs with 3-nitroindoles48,49,50,51,52,53,54,55,56,57 and N-sulfonylimines58,59,60, to give indolocarbazole derivatives and tetrahydrocarboline type of indole alkaloids, both of which are core structural motifs of bio-active molecules and functional materials (Fig. 1c)7,8,61,62,63,64,65,66,67,68,69. Compared with other regio-isomers, indolo[2,3-b]carbazoles are a relatively less explored class of molecules. SR1366866 was identified as a potent oral anticancer agent, and TPEICE68 was discovered as a potential sensor for detection of explosives (Fig. 1c).

Results

Reaction development

On the basis of our mechanistic assumption, we believe that heterocyclic o-QDMs might be trapped by 3-nitroindoles. To our delight, this carbene bridging o-QDM formation from bromobenzene 1a (0.2 mmol) and N-tosylhydrazone 2a (0.3 mmol) was indeed feasible. Indolo[2,3-b]carbazole 4a could be obtained through a two-step, one-pot procedure by the reaction of phenyl bromide 1a with N-tosylhydrazone 2a and 3-nitroindole 3a. A set of conditions with a catalyst assembled from Pd2(dba)3•CHCl3 and SPhos, in presence of K3PO4 in toluene at 80 °C was found to be optimal (for details on condition investigation, see Table S1). Treatment of the mixture with DDQ in ethyl acetate could afford the target product 4a in 73% yield upon isolation. The reaction could be scaled up to 1 mmol, while maintaining similar reaction efficiency. Replacement of phenyl bromide 1a with phenyl iodide 1a’ led to a slightly decreased yield of 4a.

The current one-pot two-step procedure for indolo[2,3-b]carbazole synthesis through carbene bridging o-QDM formation is applicable for a broad range of aryl and heteroaryl bromides (Fig. 2). As depicted, aryl bromides containing different functional groups variating with electron-donating and electron-withdrawing properties were tolerated (4b-4n). In specific, reactants bearing strong electron withdrawing groups, such as CN (4 d), CO2Me (4e, 4 g and 4j), NO2 (4i) and Ac (4n), could participate in current three-component reactions. The reactions could proceed well even when steric bulky groups were introduced. For example, arylbromides, decorating with OMe (4n), NO2 (4i) and CO2Me (4j) groups at the ortho position of the phenyl ring, and 1-bromonaphthalene (4p) were competent reactants. Heteroaryl bromide bearing like 3-bromobenzo[b]thiophene was viable reactant as well. The reaction gave 4q in 62% isolated yield under current optimal conditions.

Fig. 2: Reaction scope of bromides 1 for the synthesis of indolo[2,3-b]carbazoles 4.
figure 2

a Phenyl iodide 1a’ was used instead of phenyl bromide 1a. Me: methyl, Boc: t-butyloxycarbonyl, DDQ: 2,3-dichloro-5,6-dicyanobenzoquinone, RT: room temperature.

Full size image

Subsequently, the substrate scope with respect to N-tosylhydrazones 2 and nitro-heteroaromatics 3 was investigated. This two-step one-pot procedure could be applied for various hydrazones with different substituents on the indole ring system (Figs. 3, 4a–w). N-tosylhydrazones containing other heteroaromatic system, like benzothiophene (4x) and benzofuran (4 y, 4ae and 4af), were viable carbene precursors for this bridging o-QDM formation, giving the corresponding products in up to 61% isolated yields. For some reactions of the 3-nitroindoles, isolation of the cycloaddition adducts (4z’-4ad’) was necessary, which were beneficial for the purification of the final indolocarbazoles after oxidation (step 2). The reactions of 3-nitroindoles bearing tosyl and Me protecting groups showed lower efficiencies (4ah, 4ai vs. 4a)70. To our delight, 2-nitrobenzofuran was a competent substrate as well51,71,72,73, and the reaction gave the corresponding product 4ai in 46% isolated yield.

Fig. 3: Reaction scope of tosylhydrazones 2 and nitro-heteroaromatics 3 for the synthesis of indolo[2,3-b]carbazoles 4.
figure 3

a Scope of N-tosylhydrazones. b Scope of nitro-heteroaromatics. Me: methyl, Boc: t-butyloxycarbonyl, DDQ: 2,3-dichloro-5,6-dicyanobenzoquinone, RT: room temperature.

Full size image
Fig. 4: Reaction scope of bromides 1 for the synthesis of tetrahydrocarbolines 6.
figure 4

Me: methyl, Ts: 4-methylbenzenesulfonyl.

Full size image

After establishing modular approach to prepare indolo[2,3-b]carbazole analogs, the potential of current three-component D-A reaction enabled by dearomative C(sp3)-H bond activation was tested by cycloaddition with dienophiles N-sulfonylimines 5. This time, a catalytic system assembled from Pd(OAc)2/CyDPEPhos in toluene (0.1 M) at 80 °C turned out to be optimal (for details on reaction condition studies, see Table S2). The transient active diene species could be trapped by N-sulfonylimine 5a to give 6a in 68% isolated yield. The structure and geometry of the product were determined by single-crystal X-ray analysis (Fig. 4). A wide range of aryl bromide with electron-donating or electron-withdrawing groups (6a-6n, 6q, 6r, 6 v and 6w) on the phenyl rings, were viable reaction partners. Meta- or para- substituents regardless of the electronic nature were all tolerated. The reactions proceeded smoothly to give the target products in 44-68% isolated yields (6a-6k). The reaction could be scaled up, and 6a was isolated in 58% yield. It is worthwhile to mention that the reaction seems to be not sensitive the steric hindrance factors, as arylbromides with ortho substituents could react under the standard conditions to give 6 l to 6n in 37 to 46% yields. Furthermore, bromides bearing heteroaromatic systems, such as furan, thiophene, thianaphthene, dibenzofuran rings, could all react to generate the corresponding o-QDMs in situ and to participate in the subsequent [4 + 2] cycloaddition reactions, to give the corresponding indolocarbazoles in up to 67% yields (6o-6p, 6s-6u). 2-Bromonaphthalene and 2-bromophenanthrene could react as well, affording 6 v and 6w in 56% isolated yields. Besides arylbromide, viny bromide like β-bromostyrene was competent reactant. Product 6x was obtained in 35% yield.

Subsequently, the substrate scope with respect to N-tosylhydrazones 2 deriving from 2-methylindole-3-carboxaldehydes was investigated. As depicted in Fig. 5a, variation the group on the nitrogen atom from methyl to benzyl group resulted in a lowering the yield indolocarbazole 6 y. N-tosylhydrazones 2 with tert-butyl carbamate on the nitrogen atom (R = Boc) or deriving from free indole (R = H) were not good carbene precursors to react, and no desired products were detected from the reaction mixture under the standard conditions. By contrast, a styrene moiety was tolerated, and the corresponding tetrahydrocarboline 6ab was obtained in 40% isolated yield. Similarly, hydrazones bearing substituents on the phenyl ring in the indole nucleus, ranging from electron-donating groups to electron-withdrawing groups, all reacted well to give the desired products (6ac-6ai). The structure and geometry of the product was further confirmed by single-crystal X-ray analysis of 6af.

Fig. 5: Reaction scope of N-tosylhydrazones 2 and imines 5 for the synthesis of tetrahydrocarbolines 6.
figure 5

a Scope of N-tosylhydrazones. b Scope of imines. Me: methyl, Ts: 4-methylbenzenesulfonyl.

Full size image

Encouraged by the results obtained, the generality and limitation of N-sulfonylimines 5 were also surveyed briefly. As shown by the results compiled in Fig. 5b, N-tosylimines with different substituents variated by electronic and steric hindrance effects could react to give final three-component adducts in up to 69% isolated yields (6aj-6aq). The reaction of N-sulfonylimines deriving from 2-naphthaldehyde and thiophene-2-carbaldehyde could also proceed to give the corresponding tetrahydrocarbolines 6ar and 6as in 33% and 48% yields, respectively. Variation of the protecting group from tosyl to Boc, tert-butylsulfinyl and phenyl groups, resulted in no observation of desired products (6at-6av), which probably attributed to the lower electrophilicity of these imines. The reactions of other dienophiles like benzaldehyde, ethyl acrylate which proved to be less electrophilic than N-sulfonylimine 5a failed to provide desired cycloaddition adducts under current conditions.

Mechanistic investigation

It is worthwhile to mention that a two-component one-pot two step reaction of 2a with 3a in presence of K3PO4 in toluene, could produce indolocarbazole 7a in 67% yield after oxidation (Fig. 6a, right). By contrast, the reaction of 2a and 5a in presence of K2CO3 in toluene at 80 °C failed to give the corresponding tetrahydrocarboline 8a (Fig. 6a, left). For comparison, N-tosylhydrazone 9a deriving from indol-3-ylphenylmethanone was prepared, and subjected to the reactions with 3a and 5a. Interestingly, 4a and 6a were obtained in 90% and 40% isolated yields (Fig. 6b), respectively. These somewhat surprising observations may indicate that: (1) 3-nitroindole 3a is more reactive than N-sulfonylimine 5a towards the in situ generated o-QDM; (2) o-QDM bearing additional phenyl group react much faster than the one generated from N-tosylhydrazone 2a in presence of base (See Supplementary Information for comparison on the reactivity through DFT calculation); (3) The cycloaddition of the in situ generated o-QDM with dienophiles 3a or 5a may not require the participation of a palladium catalyst. Hence, a catalytic pathway involving π-Lewis base catalysis74 reported by Chen and co-workers was unlikely. Treatment of tosylhydrazone 9a with suitable base offers an alternative method to generate o-QDM intermediate, which may complement current approach through palladium-catalyzed reaction of hydrazone 2a with phenylbromide 1a.

Fig. 6: Two-component reactions of N-tosylhydrazones with 3-nitroindole 3a or N-tosylimine 5a.
figure 6

a Reaction of 2a with 3-nitroindole 3a or imine 5a in presence of base. b Reaction of 9a with 3-nitroindole 3a or imine 5a in presence of base. Reaction condition A: K2CO3 (3.0 equiv), toluene (0.1 M), 80 °C, Ar, 12 h. Reaction condition B: i K3PO4 (4.0 equiv), toluene (0.1 M), 80 °C, Ar, 12 h; ii DDQ (1.2 equiv), ethyl acetate (0.2 M), RT, 6 h. Me: methyl, Ts: 4-methylbenzenesulfonyl, Boc: t-butyloxycarbonyl, DDQ: 2,3-dichloro-5,6-dicyanobenzoquinone, RT: room temperature.

Full size image

To delve deeper into the reaction mechanism, several deuterium labeling experiments were performed. Under otherwise identical conditions, three-component reaction of 1a, 2a and 3a in presence of 10 equiv. of D2O gave 4a’ with significant amount of deuterium incorporated on the methylene carbon atom (Fig. 7a, up). Similar reaction phenomenon was observed for the reaction with imine 5a (Fig. 7a, bottom). The relatively lower value of kH/kD indicated C–H bond cleavage in the methyl moiety in 2a was not involved in the rate-determining step (Fig. 7b).

Fig. 7: Mechanistic experiments.
figure 7

a Three-component D-A reactions in presence of D2O. b KIE experiments. Condition A: see Fig. 3, step 1; Condition B: see Fig. 4.

Full size image

DFT calculations

DFT calculations were performed to further investigate the mechanism of o-QDM formation in Pd-catalyzed coupling of PhBr 1a with N-tosylhydrazone 2a. The computed energy profiles are shown in Fig. 8a. Under the experimental conditions, the oxidative addition of 1a with the active Pd(0) catalyst supported by SPhos requires a barrier of 19.6 kcal/mol, leading to the Pd(II) intermediate INT1 (Fig. 8a, see details in Fig. S9). Treatment of N-tosylhydrazone 2a with base can give diazo compound 2a’, which readily reacts with INT1 via TS1 (ΔG‡ = 12.9 kcal/mol) to produce the palladium carbene intermediate INT3. Subsequently, intramolecular migratory insertion of the phenyl group can take place smoothly (TS2, ΔG‡ = 3.0 kcal/mol)75, affording the ƞ3-allyl-Pd(II) species INT4. These steps involving palladium carbene have feasible kinetics and favorable thermodynamics, verifying the proposed carbene bridging process (I → II → III in Fig. 1b).

Fig. 8: Computational studies.
figure 8

a Computed energy profiles for the formation of o-QDM Species. b Computed transition states of [4 + 2] cycloadditions. Energies are with respect to the separated PhBr and SPhos-Pd(0) catalyst. Key bond distances are shown in Å.

Full size image

Based on INT4, the key o-QDM species could be delivered either by β-H elimination or C-H bond activation. The computations show that the β-H elimination requires a high barrier of 41.5 kcal/mol (TS3). In contrast, the computed barrier for C-H activation is much lower (TS4, ΔG‡ = 3.5 kcal/mol). The C-H activation directly leads to the complex of SPhos-Pd(0) with o-QDM (INT5) rather than the five-membered palladacycle (see details in Fig. S10). Although the o-QDM formation is a dearomatization process, it can be compensated by the stability of the extended π conjugation. The release of o-QDM from INT5 is endothermic by 9.0 kcal/mol, which offers partial reversibility for the methyl C-H activation, in agreement with the experimentally observed deuterium incorporation on the methylene carbon (Fig. 7a). The barrier of C-H activation is much smaller than those of PhBr oxidative addition and palladium carbene formation, which is consistent with the relatively low KIE values (Fig. 7b). Finally, the in situ formed o-QDM undergoes D-A cycloadditions with dienophiles 3-nitroindole and N-sulfonylimine.

Interestingly, the D-A cycloadditions with N-sulfonylimine affording tetrahydrocarbolines exhibit excellent diastereoselectivity (Fig. 4 and Fig. 5). The computed [4 + 2] transition states indicate that TS5, leading to the observed diastereomer 6a, is 2.9 kcal/mol lower than TS6 (Fig. 8b), which is consistent with the experimental observations. The favored TS5 is mostly stabilized by the secondary orbital interactions76 between the phenyl group in imine and the π moieties in o-QDM (Fig. 8b), which is absent in the disfavored TS6. These asynchronous concerted transition states TS5 and TS6 are superior to their corresponding stepwise [4 + 2] transition states (see details in Fig. S11).

Synthetic application

Preliminary manipulations on 4a and 6a were performed (Fig. 9). Monobromination of 4a took place selectively on the phenyl ring arise from N-tosylhydrazone 2a, to give 10 in nearly quantitative yield, which could further undergo Suzuki-Miyaura coupling to produce 11 and 13 in 71% and 70% isolated yields, respectively. The Boc group was removed facilely by treating 11 with TFA, giving 12 in 79% yield upon isolation. Treatment of 6a with DBU and DDQ in DCM could give pyridoindole 14. The tosyl group could be removed in presence of Li/naphthalene in THF to produce 15. To furthur demonstrated the synthetic potential of current dearomative C(sp3)-H bond activation to generate o-QDM intermedate, a palladium-catalyzed three-component reaction of phenylbromide 1a, dizao precursor 2a with trifluoroacetophone 16 was also examined. To our delight, tricyclic adduct 17 containing a trifluoromethyl group could be obtained in 92% isolated yield under slightly modified procedure (Fig. 9c).

Fig. 9: Synthetic application.
figure 9

a Manipulation from 4a. b Manipulation from 6a. c Trapping the in situ generated o-QDM with trifluoroacetophone 16. Conditions and reagents: (i) 4a, NBS, DCM, 0 °C, 6 h. (ii) PhB(OH)2, Pd(PPh3)4, K2CO3, 1,4-dioxane:H2O = 4:1, 80 °C, 12 h. (iii) TFA, DCM, 50 °C, 2 h. (iv) ArB(OH)2, Pd(PPh3)4, K2CO3, 1,4-dioxane:H2O = 4:1, 80 °C, 12 h. (v) 6a, DDQ, DBU, DCM, RT, 6 h. (vi) 6a, Li/naphthalene, THF, 0 °C, 2 h. (vii) a 1a, 2a, 16, Pd(OAc)2 10 mol%, (4-MeC6H4)3P 15 mol%, K3PO4, THF, 80 °C, 12 h; b TsOH·H2O, DCM, RT, 6 h. Me: methyl, Ts: 4-methylbenzenesulfonyl, DDQ: 2,3-dichloro-5,6-dicyanobenzoquinone, RT: room temperature.

Full size image

Discussion

In summary, we have described a strategy on o-QDM formation through palladium carbene migratory insertion enabled dearomative C(sp3)-H activation. The in situ generated heterocyclic o-QDMs could be trapped facilely by 3-nitroindoles and sulfonylimines, to give a large number of indolo[2,3-b]carbazole and tetrahydrocarboline derivatives with rich structural diversity. Control experiments revealed that the heterocyclic o-QDM intermediate could react with dienophiles 3-nitroindoles and N-sulfonylimines readily without the aid of a palladium catalyst. Compared with existing methods11,12,13,14,15,16,17,18,19,20,21,22, our reaction represents a rare example on catalytic intermolecular reaction to generate o-QDM intermediates. We anticipate such a reaction mode may find application on diversity-oriented indoline containing alkaloid synthesis. According to our mechanistic studies, the achievement of asymmetric synthesis through chiral ligand control via palladium catalysis is quite challenge. Further studies on mechanistic details, trapping of o-QDMs with other less activated dienophiles are currently underway in our laboratory.

Methods

General procedure for the synthesis of 4

To an oven-dried reaction tube containing a stirring bar was added Pd2(dba)3•CHCl3 (5 mol%), SPhos (15 mol%), K3PO4 (4.0 equiv), starting materials 1 (1.5 equiv), 2 (1.5 equiv) and 3 (1.0 equiv). This mixture was dissolved in toluene (0.1 M) and stirred at 80 °C for 12 h. Then the reaction mixture was was cooled to RT and filtered through a pad of Celite and eluated with DCM (20 mL). The filtrate was evaporated under reduced pressure to give the crude mixture. Treatment of the mixture with DDQ (1.2 equiv) in ethyl acetate (0.2 M) at RT for 6 h, the filtrate was evaporated under reduced pressure to give the crude mixture, which was purified by flash column chromatography on silica gel to give the desired product 4.

General procedure for the synthesis of 6

An oven-dried 25 mL test-tube equipped with a septum and a magnetic stir bar was charged with Pd(OAc)2 (10 mol%) and CyDPEPhos (40 mol%). The tube was then evacuated and back filled with argon. This cycle was repeated three times, and toluene (1.0 mL) was added via syringe. The resulting solution was stirred at RT for 30 min. Then to the reaction mixture was added 2 (1.5 equiv), 5 (1.5 equiv), K2CO3 (3.0 equiv), 1 (0.2 mmol, 1.0 equiv) and toluene (1.0 mL) under an atmosphere of argon. This mixture was stirred at 80 °C for 16 h. Then the reaction mixture was cooled to RT and filtered through a pad of Celite and eluted with DCM (20 mL). The filtrate was evaporated under reduced pressure to give the crude mixture, which was purified by flash column chromatography on silica gel to give the desired product 6.