Volume 65 (2017) Issue 5 Pages 409-425
C–H functionalization reactions involve the activation of otherwise unreactive C–H bonds, and represent atom economical methods for the direct transformation of simple substrates to complex molecules. While transition metal-catalyzed C(sp2)–H functionalization reactions are regularly used in synthesis, C(sp3)–H functionalization is rarely applied to the synthesis of complex natural products because of the difficulties associated with controlling selectivity. With this in mind, we focused on the development of new palladium (Pd)(0)-catalyzed C(sp3)–H functionalization reactions for the synthesis of complex molecules, resulting in several new methods capable of solving these problems. We initially developed a concise synthetic method for the facile construction of oxindoles and spirooxindoles via a Pd-catalyzed benzylic C(sp3)–H functionalization reaction. This method was subsequently extended to the synthesis of various heterocycles, including 2-arylindoles, benzocarbazole, indolocarbazole, indoloquinazolinone, and indoloquinazolinedione, as well as the total synthesis of several pyrrolophenanthridine alkaloids without the need for any protecting groups. This method was also successfully applied to the synthesis of the right-hand fragment of benzohopane from tetrahydro-2H-fluorene, which was constructed by a Pd-catalyzed benzylic C(sp3)–H functionalization. In this review, we provide a detailed discussion of our most recent investigations pertaining to Pd(0)-catalyzed benzylic C(sp3)–H functionalization.
During the synthesis of a target compound, reactive functional groups such as halogen and triflate substituents are traditionally employed to access intermediates bearing the desired functional groups. A wide variety of reliable functional group transformations has been developed to date, and these reactions have paved the way for the total synthesis of numerous complex natural products.1–3) However, the introduction of a desired functional group invariably requires the application of a stepwise synthetic route starting from a reactive intermediate, which can lead to higher and higher costs depending on the number of synthetic steps required. To solve this problem, C–H functionalization,4) which involves the activation of an otherwise unreactive C–H bond, has attracted considerable attention as an atom economical method for the direct transformation of a simple substrate to a complex molecule.
C–H functionalization reactions can be roughly classified into C(sp2)–H and C(sp3)–H functionalizations.4) C(sp2)–H functionalization dates to the work of Murahashi’s group5) in 1955, when they reported the co-catalyzed carbonylation of N-benzylideneaniline via C(sp2)–H bond activation. In 1993, Murai et al. reported an Ru-catalyzed C(sp2)–H functionalization6) as a practical and precise method for the introduction of alkyl groups at the ortho positon. Following on from these early works, many other chemists have developed an academic interest in the efficiency of C(sp2)–H functionalization, and developed various new reactions. In contrast, the development of practical C(sp3)–H functionalization reactions has lagged because of the low reactivity of C(sp3)–H bonds compared with C(sp2)–H bond. In 2000, Hartwig and colleagues reported the selective borylation of the C–H bonds of methyl groups.7) In the same year, Chatani et al. reported an acylation reaction involving C(sp3)–H functionalization.8) Based on these reports, C(sp3)–H functionalization attracted considerable attention from many synthetic chemists.
Palladium (Pd) is one of the most commonly used catalysts in transition-metal-catalyzed C–H functionalization reactions.9–20) Under Pd catalysis, there are two main approaches for activating unreactive C(sp2)–H and C(sp3)–H bonds, including: i) an intramolecular approach; and ii) a directing group approach. A variety of different directing groups and reaction conditions involving Pd catalysts have been developed for extending the substrate scope of C–H functionalization reactions.21–29) However, very little progress has been made towards controlling the selectivity of C(sp3)–H functionalization in the presence of C(sp2)–H bonds. This issue is further confounded by the lack of suitable methods for distinguishing between the different C(sp3)–H bonds found in complex molecules. The application of C(sp3)–H functionalization to the total synthesis of complex natural products therefore remains a big challenge in synthetic chemistry, whereas C(sp2)–H functionalization is employed routinely.30–32) We recently became interested in the development of Pd(0)-catalyzed C(sp3)–H functionalization reactions for the synthesis of complex molecules, and developed several new methods that addressed some of the problems described above. In this review, we provide a summary of our recent research on the Pd(0)-catalyzed benzylic C(sp3)–H functionalization reaction and its application.33–43)
Spirooxindole natural products, which are characterized by the fused spiro-structure of their oxindole core bearing various rings at its C3 position, have attracted considerable attention from numerous synthetic chemistry groups because of their complex structures and their pronounced biological activities44–53) (Fig. 1). The spirooxindole skeleton could be used as a valuable synthetic intermediate for the construction of complex indole alkaloids. The development of efficient methods for the construction of this structure is therefore particularly important. A variety of different methods have been reported for the synthesis of spirooxindoles, including an intramolecular Heck,54) oxidative rearrangement,55) intramolecular Mannich,56) and ring expansion reactions.57) We previously developed a novel method for the construction of oxindole skeletons from carbamoyl chlorides, as well as a spirooxindole synthesis that proceeded via the bismetallation of 1,3-dienes.58–61) Furthermore, this method was subsequently applied to a formal synthesis of elacomine.60) Although these synthetic methods represent efficient strategies for the construction of spirooxindole moieties, their application has been limited by their requirement for the preparation of the carbamoyl chloride precursors. For example, the treatment of N-[2-(1,3-butenyl)aryl]carbamoyl chloride 2 with hexamethyldisilane and a catalytic amount of [Pd(η3-allyl)Cl]2 resulted in an intramolecular carbosilylation reaction to give oxindoles 3 in good yield bearing an allylsilane functional group61) (Chart 1). The subsequent Sakurai-type cyclization of 3 provided the tricyclic spirooxindoles 4 and 5 by controlling the stereochemistry of the three contiguous stereogenic centers. Although this strategy provided facile access to several spirooxindoles, N-[2-(1,3-butenyl)aryl]carbamoyl chloride 2 (R=CH2CH2OTBS) had to be synthesized from the corresponding tert-butoxycarbonyl (Boc)-protected iodoaniline 1 via the construction of the diene moiety (5 steps), followed by the introduction of the carbamoyl chloride group (1 step). To establish a much simpler synthetic strategy for the construction of these spirooxindole systems, we switched our focus to the use of C–H functionalization chemistry because C–H bonds can be used directly, thereby avoiding the need for the introduction of an olefin moiety. We therefore planned the intramolecular C(sp3)–H functionalization of carbamoyl chloride 6 according to a benzylic C(sp3)–H bond activation strategy (Chart 2).
When we started this project, there were very few reports on intramolecular Pd-catalyzed C(sp3)–H bond activation followed by a C–C bond forming reaction. For example, Fagnou and colleagues described the synthesis of dihydrobenzofuran 8 using a Pd-catalyzed C(sp3)–H functionalization strategy63) (Chart 3). The authors of this report found that the addition of pivalic acid (PivOH) promoted the C–H bond activation process. Although several reports were available at the time describing the activation of the C(sp3)–H bonds of methyl groups for the synthesis of cyclobutene62) and five membered rings,63–68) there were no reports pertaining to Pd-catalyzed C(sp3)–H activation followed by a C–C bond forming reaction to give a methylene or methine moiety. For this reason, several groups became interested in the development of C(sp3)–H functionalization chemistries.
We initially investigated the development of a new method for the synthesis of simple oxindoles via the C(sp3)–H functionalization of a methyl group attached to a benzene ring. Given that the oxindole ring is a fundamental heterocyclic skeleton that can be found in a wide range of biologically active natural products and medicines,69,70) the development of a concise and versatile method for the construction of oxindoles and spirooxindoles is highly desirable. We envisioned that the oxidative addition of carbamoyl chloride 9 to Pd(0) would give intermediate 10, which would be followed by the C(sp3)–H activation of the methyl group to give palladacycle 11 (Chart 4). Finally, intermediate 11 would undergo a reductive elimination step to give oxindole 12. Given that the starting carbamoyl chloride 9 could be readily prepared from a simple ortho-methyl aniline by introduction of a substituent (R) and a carbamoyl chloride moiety, this C(sp3)–H activation method for the construction of oxindoles would provide a concise method for the construction of various oxindoles. At the time, there were no other examples of C–H functionalization reactions involving carbamoyl chlorides. Furthermore, if successful, this idea could be extended to the synthesis of various other heterocycles based on the methyl, methylene, or methine C(sp3)–H activation, followed by an intramolecular C–C bond forming reaction.
To assess the feasibility of our initial idea, we synthesized carbamoyl chloride 9a bearing two ortho methyl groups by the formylation of 2,6-dimethylaniline (13), followed by the reduction of the resulting formamide and the formation of a carbamoyl chloride moiety (Chart 5).
Based on earlier reports, carbamoyl chloride 9a was treated with Pd(OAc)2 (3 mol%), PCy3·HBF4 (6 mol%), and Cs2CO3 (1.1 equiv.) in mesitylene at 135°C under argon.63) Fortunately, the desired oxindole 16a was obtained in 10% yield along with a large amount of N-2,6-trimethylaniline (15) (Table 1, entry 1). As reported by Fagnou and colleagues,63) the addition of PivOH led to an improvement in the yield, with aniline 15 being obtained in a slightly higher yield of 20% (Table 1, entry 2). To avoid the elimination of CO, the reaction was performed under a CO atmosphere both with and without PivOH. Both reactions proceeded smoothly to give 16a in 67 and 35% yields, respectively (Table 1, entries 3 and 4). These results indicate that the presence of CO was important to the success of this reaction, and that its inclusion was compatible with the addition of PivOH. Based on these results, we screened a variety of different reaction temperatures, ligands, and additives against this reaction (Table 1, entries 5–14). The optimum temperature for this reaction was determined 120°C, although the reaction also proceeded at 100°C, albeit in a lower yield (Table 1, entries 5 and 6). It was also found that N-hydroxypivalamide (PivNHOH) is a good additive, leading to improvements in the yield up to 84% (Table 1, entries 7–10). In terms of the ligands, Ad2PnBu,71,72) which is a bulky trialkyl phosphine, gave the best results (Table 1, entries 11–14). The optimum conditions were therefore determined as follows: Pd(OAc)2 (3 mol%), Ad2PnBu (6 mol%), Cs2CO3 (1.1 equiv.), and PivNHOH (0.3 equiv.) under CO in mesitylene at 120°C (Table 1, entry 14).
The optimum reaction conditions were subsequently applied to the synthesis of various oxindoles (16b–f, h–p) from the corresponding carbamoyl chlorides (9b–f, h–p) bearing a wide range of different substituents on their aromatic ring and nitrogen atom (Table 2). These investigations revealed the following features of this reaction.
Having established the optimal reaction conditions, we proceeded to investigate: (i) the extension of this method to the synthesis of spirooxindoles; and (ii) improving the selectivity of this reaction.
When we began our work towards the development of a Pd-catalyzed benzylic C(sp3)–H functionalization for the synthesis of spirooxindoles, there was very little precedent in the literature for the activation of methine C(sp3)–H bonds.73,74) Furthermore, previous research indicated that there could be potential regioselectivity issues.74) For example, the intramolecular cyclization of an aryl bromide containing a cyclopropyl group gave a 3 : 1 mixture of dihydrobenzofuran and chromane products via the C(sp3)–H functionalization of the cyclopropyl methine and methylene groups, respectively.74)
We initially applied our newly developed conditions to carbamoyl chlorides bearing cycloalkyl groups instead of a methyl group.33,35) In the case of carbamoyl chloride 17a bearing a cyclopropyl group the desired spirooxindole 18a was selectively obtained in 60% yield (Chart 6). In contrast, carbamoyl chlorides 19 and 21 containing cyclobutyl and cyclopentyl groups failed to afford the desired spirooxindole products 20 and 22 under the same conditions, and gave the corresponding anilines. Given that methine C(sp3)–H activation was rare at the time73,74) and a spirooxindole fused with a cyclopropane moiety could be readily converted to various other fused systems,57) we proceeded to investigate the cyclization of carbamoyl chlorides with cyclopropyl groups.
Extensive investigations of the catalyst, phosphine ligand, and base revealed that the conditions developed for oxindole synthesis33) were applicable to the spirooxindole synthesis. With this in mind, we proceeded to examine the scope and limitations of this reaction (Table 3). As well as being suitable for the synthesis of oxindoles, we confirmed that the optimized conditions could be applied to synthesis of various spirooxindoles fused with cyclopropanes. It is noteworthy that the methine C(sp3)–H bond of the cyclopropyl group was selectively functionalized in the presence of the methylene C(sp3)–H of the ethyl group and the methine C(sp3)–H bond of the isopropyl group. Although the substituent at the other ortho position was found non-essential, it had a positive effect on the cyclization (e.g., 18g vs. h). The methine C(sp3)–H functionalization proceeded without any erosion in the stereochemistry (e.g., 18i vs. j). In short summary, we established a new method for the synthesis of spirooxindoles based on the Pd(0)-catalyzed methine C(sp3)–H functionalization of a cyclopropyl group.35)
Prior to conducting our own research towards the synthesis of oxindoles, it was well known that C(sp2)–H activation proceeds selectively in the presence of competitive C(sp3)–H bonds. For example, Glorius and colleagues75) reported an intramolecular competition experiment between C(sp2)–H and C(sp3)–H bonds in an amination reaction. The activation of the C(sp3)–H bond was only observed under their reaction conditions. Fagnou and colleagues76) also reported an exclusive selectivity for the activation of 2-bromo-N-methyl-N-phenylbenzamide (26), which only reacted at the arene C(sp2)–H bond. It was therefore clear that we would have to address the selectivity of the C(sp3)–H activation step to allow for the application of this C–H activation chemistry to complex molecules (Chart 7).
Based on the selective C(sp3)–H activation that we observed for the synthesis of oxindole 16p, we proceeded to investigate whether the selectivity could be controlled by changing the electron density of the aromatic ring. In the case of substrate 9s bearing a methoxy group, the competitive C(sp2)–H activation was almost completely suppressed, whereas the reactions of substrates 9q and 9r bearing methyl and trifluoromethyl groups gave the desired oxindoles 16q and 16r by C(sp3)–H activation, along with small amounts of the corresponding isooxindoles 29q and 29r (11–15%), which were derived from C(sp2)–H activation (Chart 8). When the previously reported reaction conditions of another group (i.e., Pd(OAc)2 (3 mol%), Cy3P·HBF4 (6 mol%), and Cs2CO3 (1.1 equiv.) in mesitylene at 135°C)63) were applied to 9p, no selectivity was observed, resulting in the formation of 16p (24%), 29p (23%), and the corresponding aniline (36%). The reaction of carbamoyl chloride 9t with a naphthalene moiety gave only oxindole 16t under the developed conditions. These results therefore indicate that C(sp3)–H activation was favored over C(sp2)–H activation under our conditions. To the best of our knowledge, this work represents the first reported example of a Pd-catalyzed chemoselective C(sp3)–H activation.33)
Next, we investigated the regioselectivity of this reaction for two functionalizable C–H bonds using carbamoyl chlorides containing cyclopropyl, alkyl, phenyl, and allyl groups. Compounds 17k–m were treated under the optimized conditions as shown in Chart 9. These reactions revealed that the functionalization process occurred preferentially in the order Heck reaction>cyclopropyl methine C(sp3)–H activation>methyl C(sp3)–H activation>arene C(sp2)–H activation under the optimized conditions.35)
The proposed mechanism is shown in Chart 10. According to this mechanism, the reaction would begin with the oxidative addition of the carbamoyl chloride to Pd(0) to give intermediate 34.77) Under the optimum conditions, the elimination of CO from 34 would be suppressed by performing the reaction under a CO atmosphere. A ligand exchange process would lead to the formation of intermediate 36, which would undergo C(sp3)–H activation through a concerted metalation-deprotonation78,79) to give palladacycle 37. A subsequent reductive elimination would give oxindole 16 and spirooxindole 18 along with the regeneration of Pd(0). The effect of PivNHOH currently remains unclear, but we assume that it assists in the activation of the benzylic C(sp3)–H in 36, as well as stabilizing intermediate 35.80)
Having established a synthetic method for the construction of oxindoles from carbamoyl chloride,33,35) we investigated the extension of this method to the synthesis of several other heterocycles. We initially focused on a cascade reaction involving a combination of benzylic C(sp3)–H functionalization with Pd-catalyzed isocyanide insertion81–86) for the synthesis of indole derivatives, which are an important class of nitrogen-containing heterocycles in the pharmaceutical sciences. Several other groups had previously reported concise methods for the synthesis of carbo- and heterocyclic systems through isocyanide insertion and C(sp2)–H activation steps.87–89) While there have been several reports describing the formation of indoles via the Ru-catalyzed C(sp3)–H activation of 2,6-disubstituted isocyanides,90,91) there have been no reports pertaining to the synthesis of heterocycles through Pd-catalyzed isocyanide insertion and C(sp3)–H functionalization steps. We envisioned that the reaction of o-methylphenyl isocyanide 38 with aryl halide 39 in the presence of a palladium catalyst would give 2-arylindole 40 through a cascade process consisting of the oxidative addition of Pd(0) to the aryl halide, isocyanide insertion, benzylic C(sp3)–H functionalization, and reductive elimination (Chart 11). Notably, this strategy would be divergent and applicable to simple starting materials; however, the coordination of the isocyanide substrate to palladium would hamper the Pd(0)-catalyzed cascade process, especially the C(sp3)–H functionalization step.
To realize this strategy, we investigated a variety of different reaction conditions using 2,6-dimethylphenyl isocyanide 38a (R1=H, R2=Me) and iodobenzene 39a (R3=H) as model substrates. The results revealed that Pd(OAc)2 (5 mol%), Ad2PnBu (10 mol%) and Cs2CO3 (1.2 equiv.) in toluene at 100°C gave the best results for the desired product 40a. The key to the success of this reaction was determined the slow addition of isocyanide 38a to the reaction mixture, because the presence of an excess of this reagent would deactivate the catalyst via the formation of palladium clusters.92) Notably, no additives were required for this reaction, although a bulky carboxylic acid was found essential for the synthesis of oxindoles via benzylic C(sp3)–H functionalization.
With the optimal conditions in hand, we evaluated the scope of this reaction using various aryl halides 39 bearing a range of electron-withdrawing and electron-donating groups (Table 4, 40b–g). Bromobenzene was also successfully converted to indole 40a in 73% yield, whereas chlorobenzene did not react under the optimized conditions (not shown in Table 4). Heteroaromatic rings such as thiophenes and pyrroles were also successfully introduced at the 2-position of the indole ring, as exemplified by compounds 40h and i. Isocyanides bearing nitro and methoxy groups also reacted smoothly to give the corresponding 2-substituted indoles 40j and k, respectively. The replacement of one of the ortho methyl groups on the isocyanide substrate with a chloro or ethyl group was also well tolerated, with the C–H bond of the remaining methyl group reacting selectively to give the desired corresponding indoles 40l and m as single products. The use of an isocyanide substrate bearing only one ortho substituent (i.e., 2,4-dimethylphenyl isocyanide) failed to afford any of the desired product 40n, most likely because of the poor stability of the less hindered isocyanide93,94) and the flexible conformation of the reaction intermediate.
This reaction was subsequently extended to the synthesis of several benzocarbazole 43 and indolocarbazole 46 systems34) (Chart 12). It was envisioned that o-alkynylphenyl isocyanide 41 would react with iodo-2,6-dimethylbenzene (42a) in the presence of a Pd catalyst to give a tetracyclic carbazole via: i) oxidative addition of the iodobenzene substrate to Pd(0); ii) sequential insertion steps involving the isocyanide and alkyne moieties; iii) benzylic C(sp3)–H functionalization; and iv) reductive elimination. An initial attempt under the optimal conditions (i.e., Pd(OAc)2, Ad2PnBu, and Cs2CO3 in toluene at 100°C) resulted in a low conversion. This decrease was attributed to the reaction proceeding via a seven-membered palladacycle, which would be much less favorable than a five- or six-membered palladacycle. In this case, the addition of PivOH led to a considerable increase in the yield of the tetracyclic carbazole 43a (66% yield). Notably, the electronic state of the alkyne had no discernible impact on the yields of carbazoles 43b, c (R=NO2C6H4, MeOC6H4). Furthermore, this strategy can be readily applied to the synthesis of several indolo[2,3-a]carbazole derivatives, which have interesting biological effects.95) The treatment of 1-Boc-2-bromo-3-methyl-1H-indole 45 with PivOH under the optimal conditions gave the desired carbazole 46 in 36% yield. These results represent the first example of a Pd(0)-catalyzed domino reaction containing a C(sp3)–H functionalization step.
We also developed a new benzylic C(sp3)–H functionalization method for the synthesis of indoloquinazolinones, which represent an important structural class found in several biologically active alkaloids, including tryptanthrin,96,97) phaitanthirins,98) and ophiuroidine38,99) (Fig. 2). The tetracyclic skeleton of an indoloquinazolinone, which consists of a quinazolinone ring fused to an indole, is generally prepared by the coupling of an isatin to an isatoic anhydride under basic conditions.97) However, considerable opportunities still exist to improve the scope and limitations of these methods, especially for the direct synthesis of indoloquinazolinones bearing a methylene moiety at their C6 position.100–102)
The reaction conditions were investigated using chloroquinazolinone 47a as a model substrate, which was prepared from 2,6-dimethylaniline in four steps, including the formation of the isocyanate, its subsequent condensation with anthranilic acid, and the chlorination of the 2-position with POCl3. We initially applied the optimum conditions for the oxindole synthesis33) without CO (i.e., Pd(OAc)2 (5 mol%), Ad2PnBu (10 mol%), Cs2CO3 (2 equiv.), and PivNHOH (30 mol%) in mesitylene at 140°C; Table 5, entry 1). Fortunately, the reaction gave indoloquinazolinone 48a in 22% yield. To improve the yield, we evaluated a wide range of different Pd sources, bases, ligands, and solvents. The results revealed that the use of Pd(PPh3)4 (5 mol%), Na2CO3 (2 equiv.), and PivNHOH (30 mol%) in PhCl at 120°C gave the best results for the synthesis of indoloquinazolinone 48a. It is noteworthy that the addition of an additive was critical to the success of this reaction, with PivNHOH giving a much higher yield than PivOH (Table 5, entries 4–6).
We subsequently investigated the scope and limitations of this reaction (Table 6). Substrates bearing an alkyl, methoxy, or chloro group at the second ortho position of their benzene ring (i.e., the C6 position) reacted smoothly to give the corresponding indoloquinazolinones 48 (Table 6, entries 1–4) in good to excellent yields. However, substrates bearing no substituent at the C6 position gave much lower yields of the desired products, along with significant amounts of unreacted starting material (Table 6, entries 5–8). In contrast, substrates bearing a fluoro, chloro, methoxy, phenyl, vinyl, or ester group on the phenyl ring of their 2-chloroquinazolin-4(3H)-one moiety were tolerated under the optimal conditions, and gave the desired indoloquinazolinones 48 in moderate to excellent yields (Table 6, entries 9, 10, 12–17). However, the introduction of a bromo group at the C5′ position of the substrate resulted in the complete failure of the reaction (Table 6, entry 11). As described previously,103) most indoloquinazolinones are unstable and readily oxidized to the corresponding indoloquinazolinediones on exposure to the atmosphere.
We subsequently investigated the possibility of conducting the benzylic C(sp3)–H functionalization and oxidation steps in one-pot for the synthesis of indoloquinazolinediones (Table 7). Chloroquinazolinone 47a was initially subjected to the reaction conditions described above (Table 5, entry 9), and the resulting mixture was then agitated under an atmosphere of oxygen at room temperature for several hours to give indoloquinazolinedione 50a in 57% yield (Table 7, entry 1). Several indoloquinazolinediones bearing fluoro, chloro, and alkyl groups were successfully synthesized using this procedure, albeit in low yields compared with the corresponding indoloquinazolinone 48 (Table 7, entries 2–5).
In summary, we established an effective procedure for the construction of heterocycles based on a 2-arylindole skeleton with multi-bond formation via Pd-catalyzed isocyanide insertion and benzylic C(sp3)–H functionalization reactions.36) We also developed a new method for the synthesis of indoloquinazolinones and indoloquinazolinediones based on benzylic C(sp3)–H functionalization.38) In both cases, the presence of an additional ortho substituent was important to the success of the reaction. These results indicate that there is an important steric factor at play resulting from the second ortho substituent, which positively affects the distance between the C(sp3)–H bond and the Pd center during the C(sp3)–H activation step.
Pyrrolophenanthridine alkaloids, which can be biogenetically produced by the dehydration and aromatization of lycorine, belong to the Amaryllidaceae alkaloids104) (Fig. 3). Assoanine (51),105) pratosine (52),106) hippadine (53),107) and dihydroanhydrolycorin (54)108) have been isolated from plants belonging to the Amaryllidaceae family. These alkaloids exhibit various biological activities, including acetylcholinesterase inhibitory activity, anticancer activity, and antitrypanosomal activity.104) Consequently, they have received considerable attention from chemists and biological scientists alike. Among the many total syntheses of these alkaloids,109,110) much attention has been focused on construction of the biaryl moiety.111) There have, however, been no reports in the literature concerning the total synthesis of these alkaloids using a C(sp3)–H functionalization. With this in mind, we investigated the application of our newly developed method for the construction of oxindoles to the total synthesis of Amaryllidaceae alkaloids to evaluate its scope and generality.33)
Retrosynthetically, it was envisioned that tetracyclic compound 55 could be used as a common intermediate for the synthesis of pyrrolophenanthridine alkaloids. The oxindole and phenanthridine rings could be constructed by C(sp3)–H and C(sp2)–H functionalization reactions involving the methyl and phenyl groups, respectively (Chart 13). In other words, compound 55 could be accessed from iodotoluene (59) and benzylaniline 60 through stepwise C(sp3)–H and C(sp2)–H functionalization reactions to give the lactam ring and the dihydrophenanthrine skeleton (path a). In contrast, compound 55 could be synthesized from compound 56 or 57 according to a one-pot process involving C(sp3)–H and C(sp2)–H functionalization reactions (path b). Initial attempts indicated that the latter of these two aims would be difficult to achieve because of undesired side reactions37); therefore the stepwise strategy was employed for the total synthesis.
The synthesis started with the coupling of compound 60a, which was prepared from 6-bromoveratraldehyde,112) to 2-iodotoluene 59 using a Catellani reaction113) via sequential C(sp2)–H functionalization and oxidation reactions to give phenanthridine 61a in 66% yield. The subsequent reduction of compound 61a with NaBH3CN, followed by the introduction of a carbamoyl moiety with triphosgene, gave the cyclization precursor 58a in 79% yield over two steps. We then applied our newly developed conditions to 58a to allow for the formation of an oxindole through C(sp3)–H functionalization. The treatment of 58a with Pd(OAc)2 (5 mol%), Ad2PnBu (10 mol%), Cs2CO3 (1.1 equiv.), and PivNHOH (0.3 equiv.) in mesitylene (0.2 M) under an atmosphere of CO at 100°C gave the desired tetracyclic product 55a (30%) together with a significant amount of phenanthridine 61a (60%), which was presumably derived from the elimination of CO and aerobic oxidation. Unfortunately, we were unable to suppress this side-reaction despite screening an extensive range of different Pd sources, ligands, bases, and additives. Compound 61a was subsequently separated from the product by column chromatography over silica gel and recycled by conversion back to 58a. Because the carbamoyl moiety was fixed by the rigid tricyclic skeleton, the C–H bond of the methyl group was positioned further away from the Pd center compared with the simple carbamoyl chloride 9a derived from 2,6-dimethylaniline. The positioning of the carbamoyl group in this way made the elimination of CO much more competitive. The resultant tetracyclic compound 55a was converted to dehydroassoanine 62a in 57% yield by the reduction of the lactam moiety with diisobutylaluminum hydride (DIBAL-H). The reduction of this common intermediate with NaBH3CN114) afforded assoanine (51). In contrast, the oxidation of 62a with BaMnO4 gave pratosine (52).115) The established synthetic route was also applied to the synthesis of hippadine (53) and dehydroanhydrolycorine (54) using amine 60b as a starting material, which was derived from commercially available 6-bromopiperonal in two steps. In a similar manner, the tetracyclic compound 55b, which was obtained from carbamoyl chloride 58b by an oxindole formation, was converted to dehydroanhydrolycorine (54) using DIBAL-H. Hippadine (53) was also synthesized by the oxidation of dehydroanhydrolycorine (54). Spectroscopic data for the synthetic materials 51–54, including their high-resolution mass spectra, were consistent with those previously reported for these compounds. The total synthesis of several pyrrolophenanthridine alkaloids was achieved using 7 and 8 step sequences.37) This new route is short and concise because it does not require the use of protecting groups, and could therefore be used to provide facile access to various analogs from simple starting materials (Chart 14).
Benzohopanes (63) have been isolated from rock extractions and petroleum samples collected from Guatemala by Hussler’s group116,117) (Fig. 4). Although these natural products have not been detected in living organisms, they can be generated by the aromatization of C35 hopanoids (ex. bacteriohopanetetraol (64)) via an interesting mechanism. Structurally, these compounds consist of a tetracyclic skeleton with a cis-fused bicyclo[4.3.0]nonane core. Before our report, there had been no synthetic studies about these compounds in the literature.
Retrosynthetically, it was envisioned that benzohopane (63) could be synthesized by the coupling of tricyclic compound 68 to farnesyl halide 67, followed by sequential cationic or radical polyene-cyclization118) and deoxygenation reactions (Chart 15). Tetrahydro-2H-fluorene 69 was identified as a key intermediate for the stereoselective construction of the right-hand cis-fused hexahydrofluorene core 68. Although tetrahydro-2H-fluorenes are considered useful intermediates for the construction of complex systems, there were no general synthetic methods for preparation of these compounds in the literature when we initiated this work.119,120) Considering our experience of C(sp3)–H functionalization reactions, we envisaged that these structures could be synthesized from enol triflate 70. With this in mind, we developed a new method for the synthesis of tetrahydro-2H-fluorene based on a Pd(0)-catalyzed benzylic C(sp3)–H functionalization reaction for the synthesis of the right-hand fragment of benzohopane.42)
The synthesis of tetrahydro-2H-fluorene 69 would begin with the oxidative addition of enol triflate 70 to Pd(0) (Chart 16). Because there are two methyl groups at its ortho positions (R=Me), C(sp3)–H activation would only occur to give a six-membered palladacycle 71. Finally, reductive elimination would give tetrahydro-2H-fluorene (69). Although enol triflates are useful synthetic intermediates, there have been very few examples of their use in Pd(0)-catalyzed C(sp3)–H activation reactions121) compared with C(sp2)–H activation processes.122,123)
Enol triflate 70, which was prepared from cyclohexanone through sequential Pd-catalyzed α-arylation124) and triflation reactions, was treated with Pd(PPh3)4 (10 mol%), Cs2CO3 (1.1 equiv.), and PivOH (30 mol%) in N,N-dimethylformamide (DMF) at 140°C, resulting in the formation of tetrahydro-2H-fluorene 69 (Table 8, entry 1). We then screened a variety of different Pd sources, ligands, reaction temperatures, and additives, as shown in Table 8, and the optimal conditions were determined as follows: Pd(OCOCF3)2 (10 mol%), PPh3 (20 mol%), Cs2CO3 (1.2 equiv.), and PivOH (30 mol%) in DMF at 80–100°C. For this transformation, PPh3 was found better than bulky trialkyl phosphine ligands, including Cy3P, Ad2PnBu, and tBu3P, which were otherwise effective for our oxindole synthesis33,35) through C(sp3)–H activation (Table 8, entries 4–10). Additionally, this reaction proceeded at temperatures in the range 80–100°C, which are lower than those required of our oxindole synthesis. This result can be explained in terms of the greater accessibility of the Pd center to the benzylic C(sp3)–H bond because of the flexible sp3 carbon linkage. The presence of a carboxylic acid additive (i.e., 1-AdCOOH) was found essential to the success of this reaction, which proceeded through a concerted metalation deprotonation (CMD) pathway, as previously reported.78,79)
We subsequently investigated the scope and limitations of the optimal conditions. Various tetrahydro-2H-fluorenes were synthesized from the corresponding enol triflate substrates, and the substituent at the opposite ortho position was found essential for avoiding the undesired formation of benzocyclobutene 73 (Chart 16). Notably, all these products were unstable and readily decomposed at room temperature after a few days.
With tetrahydro-2H-fluorene 69 in hand, we proceeded to synthesize the right-hand fragment of benzohopane 68 (Chart 17). Freshly prepared 69 was converted to the cis-fused tricyclic ketone 74 through sequential hydroboration and tetra-n-propylammonium perruthenate (TPAP) oxidation reactions.125) The site-selective bis-methylation was difficult; therefore an ester group was introduced to ketone 74 using Mander’s procedure.126) The stepwise methylation reaction was followed by the hydrolysis of the ester group under basic conditions to give methyl ketone 76 as a 3.3 : 1 mixture, along with a small amount of the corresponding trans-fused isomers. After triflation with Comins’ reagent,127) Pd-catalyzed carbonylation of 77 gave the α,β-unsaturated ester 78, which was converted to the right-hand fragment of benzohopane 68 using a DIBAL-H reduction and Dess–Martin oxidation. The cis-fused stereochemistry of the product was confirmed by nuclear Overhauser effect (NOE) experiments involving 79. This synthetic route was found robust for the construction of the cis-fused tricyclic skeleton of benzohopane. Our newly developed Pd(0)-catalyzed benzylic C(sp3)–H activation method for the synthesis of tetrahydro-2H-fluorene was extended to the concise synthesis of the substructure of benzohopane.42)
We initially developed a new synthetic method for the construction of oxindoles and spirooxindoles via a Pd(0)-catalyzed benzylic C(sp3)–H functionalization reaction. This method was subsequently extended to the synthesis of various other heterocycles, including 2-arylindoles, benzocarbazole, indolocarbazole, indoloquinazolinone, and indoloquinazolinedione. It is noteworthy that our synthetic method for the preparation of oxindoles was successfully applied to the total synthesis of several pyrrolophenanthridine alkaloids. In this case, a concise synthetic route was achieved without protecting groups using a Pd-catalyzed C(sp3)–H functionalization reaction. Additionally, a new method was developed for the construction of tetrahydro-2H-fluorenes based on a Pd-catalyzed benzylic C(sp3)–H functionalization strategy. These structures are useful building blocks for the introduction of other functional groups and the construction of increasingly complex structures. This method was also successfully applied to synthesis of the right-hand fragment of benzohopane. In terms of our future work, we hope to apply C(sp3)–H functionalization chemistry to the total synthesis of complex natural products by developing more concise and direct methods.
This review of the author’s work was written by the author upon receivingthe 2016 Pharmaceutical Society of Japan Award for Young Scientists.
First, I would like to express my gratitude to Prof. Y. Takemoto of Kyoto University for his kind and valuable suggestions, and his encouragement during these early studies. The experimental results reviewed in this paper were realized by the dedication and passion of my colleagues, including Dr. T. Nanjo, Mr. M. Okuno, Dr. L. Zhao, Dr. S. Suetsugu, Mr. M. Nakajima, Mr. E. Iderbat, Mr. S. Yamamoto, Mr. N. Muto, and Ms. M. Horinouchi. These works were financially supported from the following organizations: a Grant-in-Aid for Young Scientists (Start-up) from the Japan Society for Promotion of Science and Scientific Research on Innovation Area “Molecular Activation Directed toward Straightforward Synthesis” from The Ministry of Education, Culture, Sports, Science and Technology of Japan.
The author declares no conflict of interest.