2020 Volume 68 Issue 6 Pages 491-511
In spite of only a few naturally occurring products having one or more fluorine atoms, organofluorine compounds have been widely utilized in pharmaceutical, agrochemical, and functional material science fields due to the characteristic properties of the fluorine atom. Therefore, the development of new methods for the introduction of fluorine-containing functional groups has been a long-standing research topic. This article discusses our contributions to this area. The first topic is on the trifluoromethylations of C–C multiple bonds using Togni reagent based on our working hypothesis that hypervalent iodine could be activated by coordination of the carbonyl moiety to the Lewis acid catalyst. The second topic relates to asymmetric fluorofunctionalization of alkenes. A newly designed phase-transfer catalyst consisting of a carboxylate anion functioning as a phase-transfer agent and a primary hydroxyl group as a site that captures the anionic substrate was revealed to be an effective catalyst for asymmetric fluorolactonization. Inspired by the mechanistic studies of fluorolactonization, we produced a linked binaphthyl dicarboxylate catalyst, which catalyzes the 6-endo-fluorocyclization and the deprotonative fluorination of allylic amides in a highly enantioselective manner. The third topic is on C–H fluorofunctionalizations using either catalysis or photoactivation. Benzylic trifluoromethylation, which is still a rare reaction, using Togni reagent and aromatic C–H trifluoromethylation using Umemoto reagent under simple photoirradiation conditions were achieved. In addition, the Csp3–H fluorination of alkyl phthalimide derivatives is demonstrated.
Although very few natural products contain fluorine atoms, artificial organofluorine compounds have found widespread applications in pharmaceutical, agrochemical, and material science fields.1–6) The introduction of a fluorine atom(s) into organic molecules results in changes in the properties of compounds. Fluorine is the second smallest atom after hydrogen, and the C–F bond is only 1.28-fold longer than the C–H bond.7) Nevertheless, its bond energy is higher than that of the C–H bond (C–F bond = 105.4 kcal/mol, C–H bond = 98.8 kcal/mol), and the C–F bond is found to be the strongest bond in organic chemistry.8) Therefore, the replacement of hydrogen with fluorine in organic materials does not significantly change their size but often improves their stability.
Another feature of the fluorine atom is its large electronegativity (Pauling scale: F: 4.0, O: 3.5, H: 2.1).8) Fluorine affects the property of proximal functional groups. For example, the pKa value of an acidic proton is reduced if fluorine exists on a neighboring carbon. In terms of electronic properties, fluorine is more similar to oxygen than to hydrogen, and it has been used as a functional quasi-isostere of oxygen.1–6) Additionally, fluorine has low polarizability, suggesting that the lipophilicity of organic compounds is affected by incorporating fluorine atoms.1,2)
Considering the above features, the incorporation of fluorine-containing functional groups at appropriate positions is an important topic, and tremendous efforts have been made to fluorofunctionalize organic materials. Since useful fluorinating and fluorofunctionalizing reagents have been produced, this area of chemistry has made remarkable progress in recent years.9–42) Various fluorine-containing functional groups have been attached to molecules, although available structures are still limited. Thus, a new class of fluorofunctionalizations is desired to expand the chemical space of organofluorine compounds.
Based on this background, we have focused on the development of methods for the incorporation of a fluorine and a trifluoromethyl group into organic molecules since 2010. In this review article, our recent work on 1) the trifluoromethylation of C–C multiple bonds, 2) the asymmetric fluorofunctionalization of alkenes, and 3) C–H fluorination and trifluoromethylation is described. Many related investigations on these topics have been reported, but not all publications could be introduced in this article, and it is recommended that readers refer to recent reviews for further details.9–42)
The trifluoromethyl group is one of the major fluorine-containing functional groups and has been introduced into various bioactive organic molecules in place of a methyl group to improve their properties, such as metabolic stability and lipophilicity, although its size is closer to that of an isopropyl group than to a methyl group.43) As the Csp2–CF3 bond is found in many bioactive compounds, the introduction of the trifluoromethyl group into aryl and vinyl moieties has received widespread attention. Special attention has been paid to transition metal-catalyzed and -mediated cross-coupling-type reactions, which are initiated by cleavage of the C–B, C–X, and C–H bonds, and to the addition of the in situ-generated trifluoromethyl radical to aromatic rings.27,44–47)
At the start of our research project, we focused on the trifluoromethylation of indole derivatives, which are pivotal core structures found in many natural and synthetic bioactive compounds. Although its trifluoromethylated derivatives seem to be an intriguing target in pharmaceutical science, the direct trifluoromethylation of indole derivatives was scarce and a challenging subject at that time.48–50) As indoles have a nucleophilic property, we envisaged that the trifluoromethylation of indoles could be achieved by the combination of electrophilic trifluoromethylating reagent 1, so-called Togni reagent.22,51,52) However, the reaction did not proceed in the absence of additives. We subsequently planned to use a Lewis acid catalyst to enhance the electrophilicity of Togni reagent by coordination to the Lewis acid, stimulated by Togni’s work53) on O–CF3 bond formation using 1 and a stoichiometric amount of zinc salt (Chart 1).
Optimization experiments revealed that the copper salt worked as an efficient catalyst for trifluoromethylation using 1, and the desired reaction proceeded smoothly in MeOH at room temperature. Under the optimized conditions, some indole derivatives were trifluoromethylated54) (Table 1). The trifluoromethyl group was selectively attached at the C2-position. The electronic properties of the five-membered ring of indoles were important for the reaction to proceed (3a, b), although the electronic properties of substituents at the C5-position had less effect (3c, d). Bioactive compounds, such as melatonin 2g and tryptophan derivative 2h, could be converted to the corresponding trifluoromethylated products in good yields. N-Alkylated-3-methyl indoles were suitable substrates for the present trifluoromethylation reaction, and 3i and 3j were obtained in 90 and 67% yields, respectively. Interestingly, trifluoromethylation proceeded selectively at the C2-position even when the C3-position was not substituted (3l).
 |
In the course of this study, curious results were encountered (Chart 2). The desilylated di-trifluoromethylated product 4a was obtained in 10% yield in the reaction of N-trimethylsilyl (TMS)-3-methylindole 2m, while the reaction of N-tert-butyldimethylsilyl (TBDMS)-3-methylindole 2n provided only 3n. In the latter case, the TBDMS group remained intact on the nitrogen of the indole ring. Therefore, we hypothesized that the silyl cation would be an effective activator for trifluoromethylation using 1.
Based on this hypothesis, trifluoromethylation was examined under the presence of trimethylsilyl trifluoromethanesulfonate (TMSOTf), which is a representative silyl cation source, as a catalyst55) (Table 2). To our surprise, the reaction rate was remarkably fast, and the reactions were normally completed within less than 5 min. The use of 1.3 equivalent (equiv.) of 1 mainly afforded the C2-trifluoromethylated products in good yields. An electron-donating group at the C5-position accelerated the di-trifluoromethylation, and 4c was obtained in 45% yield together with 32% of 3c when an excess amount of 1 was used. As is the case with copper-catalyzed reactions, various functional groups were tolerant under these reaction conditions.
 |
Various reaction systems for the regioselective trifluoromethylation of indoles were reported,56–65) and the reactions enable access to indole derivatives bearing a trifluoromethyl group at the appropriate positions.
2.2. Trifluoromethylation of AllylsilanesThe Csp3–CF3 bond formation reaction was mainly studied using carbonyl chemistry until 2010, that is, the trifluoromethyl group was introduced by nucleophilic attack of a trifluoromethyl anion equivalent on a carbonyl group or by the reaction of enolates and enamines with an electrophilic trifluoromethylating reagent.66–68) On the other hand, double functionalization of the C=C bond is recognized as a powerful method for the construction of multifunctional aliphatic compounds.69–79) However, this type of trifluoromethylation was unexplored before the start of our project. Hence, our research group became interested in the development of new methods for the trifluoromethylation of C–C multiple bonds.
The allylic trifluoromethylation of simple alkenes was independently achieved by Parsons and Buchwald,80) Wang et al.,81) and Xu et al.82) under copper-catalysis conditions with either Togni reagent or Umemoto reagent in 2011. During our independent study on the trifluoromethylation of alkenes, which was based on our indole trifluoromethylation, we noticed that the substrate scope was limited to mono-substituted terminal alkenes. To overcome this problem, allylsilane, which is a more nucleophilic substrate than unactivated alkenes, was attractive. Various 2-substituted allylsilanes were investigated under the conditions described in Table 3 (left).83) It was found that the reaction with substrates bearing an aryl or a styryl group at the C2-position provided the desilylated products in good yields (6a–c). An alkyne moiety somewhat impeded the reaction, probably due to the coordination of either the substrate or the product to the copper ion (6d). However, 70% yield of 6d was obtained when 0.5 equiv. of CuI was used. Allylsilane with an alkyl group was successfully transformed to the trifluoromethylated compound (6e), and the reaction system was applicable to 2- and 3-substituted allylsilane (6f). It was noteworthy that simple allylsilanes having no substituents on the alkene moiety predominantly afforded the vinyl silane derivatives (Table 3, right). Although a bulky substituent on the silicon atom had a negative effect on the reaction (7d), the reactions with other allylsilanes proceeded smoothly (7a–c). These vinyl silanes were successfully utilized for the Pd-catalyzed Hiyama cross-coupling reaction84) and Rh-catalyzed 1,4-addition to cyclohexanone,85) suggesting that the 1,1,1-trifluoro-3-butene unit could be introduced in various organic molecules.
 |
The stability of the carbocation at the β-position of the silicon atom was found to be a key factor to determine the product outcome. Our proposed mechanism is described in Chart 3. After an electrophilic active species generated from CuI and 1 reacts with an alkene, the reaction with a 2-substituted allylsilane proceeds via a carbocationic intermediate. Subsequently, a silyl cation is released to form a C=C bond, and the silyl cation is trapped by MeOH or 2-iodobenzoate. If a substrate has no substituent at the β-position, deprotonation at the allylic position occurs preferentially after coordination of the alkene unit to an electrophilic active species. As a consequence, the product is released with the formation of a C–CF3 bond. Gouverneur’s group independently reported a similar reaction system86) and the trifluoromethylation of allylsilanes under photoredox catalysis.87)
Unique transformations of epoxide 8, easily obtained by treatment of 6a with meta-chloroperoxybenzoic acid (mCPBA), were discovered83) (Chart 4a). Deoxygenative alkylation at the α-position of the trifluoromethyl group occurred when 8 reacted with alkyl lithium reagents. As the same product was obtained from allylic alcohol 9 produced from 8 with a non-nucleophilic base, sodium bis(trimethylsilyl)amide (NHMDS), an allylic alcoholate seems to be the reaction intermediate (Chart 4b). Mechanistic studies including density functional theory (DFT) calculation indicated that the reaction proceeds via a rate-determining carbolithiation and Li2O elimination and is accelerated and controlled by chelation between lithium alkoxide and a fluorine of the trifluoromethyl group.88)
In the course of our study on the trifluoromethylation of allylsilanes,83) we eventually found that di-trifluoromethylation product 10 having a methoxy group at the benzylic position was produced when an electron-rich substrate was used (Chart 5). Coupled with our proposed mechanism of the trifluoromethylation of allylsilanes (Chart 3), we considered that MeOH would be trapped by the cationic intermediate derived from 6i during the second trifluoromethylation. This hypothesis prompted us to investigate the oxy-trifluoromethylation of alkenes. As expected, the introduction of alcohols at the benzylic position was observed when the trifluoromethylation of 11a, which is an electron-rich styrene derivative, was carried out in alcoholic solvents89) (Chart 6). The 2-iodobenzoate adducts were simultaneously isolated in these reactions, suggesting that 2-iodobenzoate derived from 1 could function as a nucleophile for oxy-trifluoromethylation.
Since the 2-iodobenzoate unit is easily convertible, we considered that acyloxy-trifluoromethylation could be an attractive reaction. Accordingly, oxy-trifluoromethylation was examined in an aprotic solvent, CH2Cl2 (Table 4). If a styrene substrate has no substituent at the α-position (R = H), [(MeCN)4Cu]PF6 is the catalyst of choice, and the desired products 13 were afforded in good to high chemical yields.89) Although an electron-rich aromatic ring was essential for this reaction, various functional groups as well as heteroaromatic rings were compatible under these reaction conditions. In contrast, the use of CuI as a catalyst provided much better results in the reaction of α-substituted styrene derivatives, because the deprotonative products were mainly afforded when [(MeCN)4Cu]PF6 was used.90) As a carboxylic acid can be generated from an electron-rich aromatic ring under oxidative conditions,91) 13b was additionally transformed to the 3-trifluoromethyl lactic acid derivative.
 |
The substrate scope of this reaction could also be expanded to dienes.90) 1,2-Oxy-trifluoromethylation product 15a was obtained selectively in the reaction of 1,1-disubstituted butadiene, while 1,4-addition products were observed selectively in the reaction of 1-substituted and 1,2-substituted dienes (Chart 7). To our delight, cyclohexylbutadiene 14d was found to be an effective substrate for this reaction. The system was also applicable to the reaction with alkynes89) (Chart 8). Trifluoromethylated vinyl esters were obtained in these reactions, and the stereochemistry of the products was determined to be trans.
Szabó’s group independently reported a similar oxy-trifluoromethylation of both alkenes and alkynes,92) and Zhu and Buchwald achieved intramolecular oxy-trifluoromethylation using the copper/1 system and its asymmetric version.93,94) In addition, many research groups have studied the oxy-trifluoromethylation of a variety of olefinic substrates under various reaction conditions including the copper/1 system as well as photo-redox catalysis.14,16,22,24,26,78)
Trifluoromethyl-substituted styrenes are key structural components in both the pharmaceutical and material chemistry fields. Thus, the oxy-trifluoromethylation products were transformed to the corresponding β-trifluoromethylstyrene derivatives90) (Chart 9). Both basic and acidic conditions could essentially be used for this elimination, although an electron-rich aromatic ring would be required for the reaction to operate under acidic conditions. In addition, the direct synthesis of β-trifluoromethylstyrenes was demonstrated under oxy-trifluoromethylation conditions in the presence of para-toluenesulfonic acid (p-TsOH). It was noteworthy that the reaction provided only the trans isomer.
As a part of our studies on the oxy-trifluoromethylation of alkynes, we focused on propargylic alcohols as a substrate. This trifluoromethylation was expected to proceed via a Meyer–Schuster rearrangement to afford α-trifluoromethyl enones, which would be useful building blocks for further synthesis. It was found that the dual catalytic system of CuI and Re2O7 accelerated the desired reaction, providing Z-trifluoromethyl enones or β-unsubstituted α-trifluoromethyl enones95) (Table 5). A time–course experiment in the reaction of 17a revealed that the Z-isomer was a kinetic product and that the ratio of Z/E gradually decreased over the course of the reaction (Chart 10a). As the E-isomer was provided selectively when the silyl allenol ether was used as a substrate (Chart 10b), this reaction appears to proceed via endo-dig cyclization with the oxygen of the alcohol or rhenium oxide to generate a cyclic intermediate, such as oxetene I or six-membered ring intermediate II including rhenium (Chart 10c). The group of Liu and Tan also reported a similar reaction with CuI alone, affording the E-isomer, and thus the reaction is complementary to our system.96)
 |
A double functionalization-type trifluoromethylation of alkenes together with the formation of a C–C bond is attractive as a useful method for the construction of new carbon frameworks. In 2012, Liu’s group reported the palladium/ytterbium-catalyzed oxidative aryl-trifluoromethylation of α,β-unsaturated amide derivatives to provide trifluoromethylated oxindole derivatives.97) On the other hand, we found that 5-phenyl-1-penten-3-ol provided not only the oxy-trifluoromethylation product (epoxide) but also the carbo-trifluoromethylation product (tetralin), suggesting that an intramolecular aryl group could work as a nucleophile under Cu–Togni reagent conditions. Subsequently, our focus moved to the development of carbo-trifluoromethylation. Carbocycles and heterocycles, such as tetralin and indoline, are found in many bioactive compounds, and their trifluoromethylated derivatives would be potentially useful. Therefore, we decided to investigate the trifluoromethylation of unactivated alkenes having an aromatic ring as the nucleophilic part98) (Table 6), although the treatment of an alkene bearing allylic protons under trifluoromethylation conditions has the potential to provide a deprotonative trifluoromethylation product (section 2.2.).80–83)
 |
The choice of the copper catalyst and the reaction solvent was important for determining product selectivity. The cationic copper source, [(MeCN)4Cu]PF6, had a tendency to afford a deprotonative trifluoromethylation product, whereas the neutral copper salt, CuI, preferentially provided a carbo-trifluoromethylation product. In addition, the use of a less polar and coordinating solvent, such as 1,4-dioxane, favored carbo-trifluoromethylation, while the reaction rate became slower. As the construction of a six-membered ring was faster than that of a five-membered ring, chlorinated solvents could be used for the six-membered ring forming trifluoromethylation, selectively providing tetrahydronaphthalenes bearing a trifluoromethyl group (20d–i). This indicated that orbital interaction between an aryl nucleophile and alkene is crucial for this trifluoromethylation. The reaction system was applicable to substrates having no quaternary carbon center (20k), and heteroaromatic compounds could be prepared without difficulty (20n–s). Unlike five- and six-membered rings, seven-membered ring formation was not observed. Instead, 1,6-oxy-trifluoromethylation products 21 were obtained via a 1,5-hydride shift, suggesting the possibility of trifluoromethylation-initiated remote functionalization (Chart 11). Some research groups subsequently reported remote difunctionalizations using similar reaction systems.99–104)
As an acryloanilide has no allylic proton and a nucleophilic aromatic ring, our reaction system could be applied for the synthesis of oxindole derivatives having a trifluoroethyl group at the C3-position105) (Table 7). High chemical yields of various oxindole derivatives 23 were generally observed, and functional groups were compatible under these reaction conditions. As for the carbo-trifluoromethylation of a simple alkene (vide supra), an electron-withdrawing group impeded the trifluoromethylation. After our initial work, various reaction conditions have been applied to the same reaction.14,22,106–112)
 |
Encouraged by the above trifluoromethylations generating carbocycles and heterocycles via the addition of an aromatic ring, we became interested in a semi-pinacol rearrangement-type difunctionalization to produce β-trifluoromethyl-α-aryl ketones, which would be a useful building block of bioactive compounds. Thus, we examined the reaction of 1,1-diaryl-2-propanols with 1113) (Table 8). Interestingly, Fe(OAc)2 was found to work as a more efficient catalyst compared to copper salts in this reaction. The use of K2CO3 accelerated the present reaction. Symmetrical substrates (Ar1 = Ar2) generally provided the corresponding trifluoromethylated products in high yields, irrespective of the electronic property of aromatic rings (entries 1–5). In contrast to other carbo-trifluoromethylations, the reaction of substrates bearing an electron-withdrawing group was faster than that of substrates bearing an electron-donating group (entries 2–6). The ortho-substituent did not hinder the reaction (entries 3 and 7). Unsymmetrical substrates (Ar1 ≠ Ar2) were then examined (entries 8–15). The reaction of 24h, which possesses 4-chloro and 2-chlorophenyl groups, proceeded with the migration of the 4-chlorophenyl group in a highly selective manner (entry 8). In contrast, a fluorine substituent did not induce selective migration, and a 1 : 1 mixture of 25i and 26i was obtained (entry 9). The phenyl group migrated selectively in the reaction of substrates having a phenyl and a 2-substituted aryl ring (entries 10–12). These results indicate that steric hindrance on the aryl ring reduces the migration rate. Interestingly, the electronic property on the aryl ring affects the migration rate, and the electron-withdrawing group accelerates aryl migration (entry 13). This phenomenon is inconsistent with a semi-pinacol rearrangement via the carbocation intermediate, suggesting that the reaction proceeds via a radical intermediate, as with the neophyl rearrangement.114) The pentafluorophenyl group and alkyl group did not migrate, and 25n and 25o were predominantly afforded (entries 14 and 15). Similar reaction systems were independently reported by Li and Wu’s group115) as well as Tu’s group116) in the same year.117)
 | ||||||
|---|---|---|---|---|---|---|
| Entry | Ar1 | Ar2 | 25 : 26a) | 25 [%] | 26 [%] | |
| 1b) | Ph | Ph | a | — | 82 | — |
| 2b) | 4-Me-C6H4 | 4-Me-C6H4 | b | — | 72 | — |
| 3c,d,e) | 4-MeO-C6H4 | 4-MeO-C6H4 | c | — | 78 | — |
| 4f) | 4-F-C6H4 | 4-F-C6H4 | d | — | 80 | — |
| 5f) | 4-Cl-C6H4 | 4-Cl-C6H4 | e | — | 92 | — |
| 6g) | 3-CF3-C6H4 | 3-CF3-C6H4 | f | — | 91 | — |
| 7c,d) | 2-MeO-C6H4 | 2-MeO-C6H4 | g | — | 92 | — |
| 8f) | 4-Cl-C6H4 | 2-Cl-C6H4 | h | 25 : 1 | 80 | 4 |
| 9b,h) | 4-F-C6H4 | 2-F-C6H4 | i | 1 : 1 | 42 | 42 |
| 10d) | Ph | 2-Cl-C6H4 | j | 9 : 1 | 79 | 9 |
| 11b,c) | Ph | 2-Br-C6H4 | k | 12 : 1 | 81 | 7 |
| 12b) | Ph | 2-MeO-C6H4 | l | 3 : 1 | 71 | 21 |
| 13b,h) | 4-Cl-C6H4 | 4-MeO-C6H4 | m | 10 : 1 | 80 | 7 |
| 14f) | Ph | C6F5 | n | — | 79 | NDi) |
| 15c,e,f) | 4-Cl-C6H4 | Me | o | — | 64 | NDi) |
a) Determined by 19F-NMR analysis of the crude mixture. b) Run for 12 h. c) Run at 80°C. d) Run for 9 h. e) Run in the absence of K2CO3. f) Run for 3 h. g) Run for 6 h. h) Run at 23°C. i) Not detected.
During the screening of substrates for carbo-trifluoromethylation to construct heterocycles (Table 6), N-migratory oxy-trifluoromethylation product 28 was serendipitously obtained when the unprotected aryl-allyl-amine was subjected to trifluoromethylation (Chart 12). Because 1,2-aminoalcohol is a versatile building block for the synthesis of pharmaceutical, agrochemical, and functional material compounds, we rapidly optimized the reaction conditions. Consequently, the reaction proceeded smoothly with CuI and 1 in t-BuOH, and 90% yield of 28a was obtained118) (Chart 12). The 2-iodobenzoate ester underwent hydrolysis under basic conditions, and the aryl group could be removed by treatment with (NH4)2Ce(NO3)6. This unique trifluoromethylation was successfully demonstrated using various aryl-allyl-amines and benzyl-allyl-amines.
N-Migratory oxy-trifluoromethylation was considered to proceed through an aziridine-forming trifluoromethylation and a ring-opening reaction of the aziridine. From the perspective of synthetic utility, the aziridine group is a useful functional group owing to its reactivity. Encouraged by the above findings, we also investigated amino-trifluoromethylation to provide aziridines having a trifluoromethyl group. Both low catalyst loading and a short reaction time were important factors to suppress the formation of 28 and to obtain 29 selectively (Table 9). A greater amount of CuI (5 mol%) was required in the reaction of substrates having an electron-withdrawing group due to its lower reaction rate, while the chemical yields were generally good to high.
 |
Having established suitable conditions for selective aziridine formation, the direct transformation of the in situ-generated aziridine was examined in one pot118) (Chart 13). Thiophenol, alkyl thiol, aniline, and alcohol were successfully introduced into the products when AgBF4 was added as an additional Lewis acid catalyst in the second step. Tryptamine bearing a trifluoroethyl group could be synthesized using indole in the presence of BF3.
The mechanistic studies were investigated by means of kinetic studies, NMR, and MS analyses.119) The dependence of the rate on the concentration of both CuI and Togni reagent 1 was found to be of first order. The reaction rate was also in proportion to the substrate concentration, suggesting that the rate-determining step involves not only the reactive copper species and 1 but also the substrate. The reaction showed inverse first order dependence on the product concentration, indicating that aziridine inhibits the reaction by coordination to the copper species. Based on MS analysis and control experiments using copper salts, the Cu(II) complex generated from CuI and 1 was considered to play a major role. Taking the results of the experiments into account, the proposed catalytic cycle is illustrated in Chart 14. CuI is oxidized by 1 to provide Cu(II) species. Togni reagent 1 is activated by coordination to Cu(II) ion, and the electrophilicity of hypervalent iodine is enhanced. As allyl amine can coordinate to the Cu(II) ion, intermediate I would be in equilibrium with II and III. When both 1 and allyl amine are coordinated to the Cu(II) species, the C=C bond can interact with hypervalent iodine to allow aziridine formation to proceed. This is followed by C–CF3 bond-forming elimination.
This reaction system could be applied to the synthesis of trifluoromethylated pyrrolidines118,119) (Chart 15). The group of Tan and Liu reported similar amino-trifluoromethylation to provide trifluoromethylated pyrrolidines and indolines including the asymmetric version.120,121) Other amino-trifluoromethylations have been summarized in a recent review.25)
During the process of the optimization of the Ar-migratory trifluoromethylation (section 2.4.), we noticed that hydro-trifluoromethylation product 38a was observed when DMF was used as a solvent (Chart 16). Quick screening of the reaction conditions to obtain 38a revealed that Fe(OAc)2 was not required for hydro-trifluoromethylation. This reaction pathway was found to be different from our previous studies on Lewis acid-catalyzed reactions, and the reaction would proceed via the addition of a trifluoromethyl radical to alkene. In 2013, Li and Studer reported pioneering work on trifluoromethylations using Togni reagent and an electron-donor molecule system, which produces a trifluoromethyl radical in situ.122) Consequently, we became interested in trifluoromethylation using Togni reagent and electron donors.123) It was found that the hydro-trifluoromethylation of terminal alkenes could be conducted with 1 and K2CO3 in DMF124) (Table 10). As usual, the trifluoromethyl radical predominantly attacked the alkene moiety even in the presence of an electron-rich aromatic ring. The compatibility of various functional groups was high, and aryl iodide was intact under the reaction conditions (38b). gem-Disubstituted alkene could be transformed to 38g, albeit with moderate yield.
 |
We also found that vinylic trifluoromethylation and iodo-trifluoromethylation proceeded when iodide salts were used as an electron donor. Fortunately, the products could be controlled.124) Tetrabutylammonium iodide (TBAI) selectively afforded vinylic trifluoromethylation product 40, whereas potassium iodide (KI) provided iodo-trifluoromethylation product 41 (Table 11). The solubility of 2-iodobenzoate salt derived from 1 was verified to be crucial for product selectivity. That is, tetrabutylammonium 2-iodobenzoate is soluble in 1,4-dioxane, and thereby the salt could work as a base to eliminate HI from 41. Meanwhile, the insoluble potassium salt could not promote the elimination, providing the iodo-trifluoromethylation products selectively. Many reports on hydro-, vinylic, and iodo-trifluoromethylations have been published and summarized in various reviews.14,16–18,28)
 |
Difunctionalization of alkenes is recognized as one of the most useful transformations for increasing molecular complexity, and thus tremendous efforts have been directed toward the development of the fluorofunctionalization of alkenes.30,33–35,37,39) Since compounds present in nature are in a chiral environment, biological activities between stereoisomers generally differs. Therefore, the development of an efficient method to synthesize chiral fluorinated compounds is highly desirable. In this context, asymmetric fluorofunctionalization of alkenes has attracted much attention, because the reaction is a promising approach to preparing structurally diverse chiral fluorinated compounds.125) However, such reactions remain largely unexplored in comparison with the asymmetric fluorination of carbonyl compounds.30,37,68,126)
Pioneering work on asymmetric fluorination of silyl enol ethers using a stoichiometric amount of a chiral fluorinating reagent derived from a cinchona alkaloid was independently achieved by Shibata et al.127) and Cahard et al.128) Subsequently, this reaction system was applied to fluorinations with allylsilanes and indole derivatives, and its catalytic version has been actively studied for a decade.129–135) Another representative approach to asymmetric fluorofunctionalization was the use of a chiral hypervalent iodine.136) Shibata’s group reported fluoroaminocyclization using a chiral diiodo-binaphthalene catalyst.137) Ishihara-type chiral iodoarene138) was also utilized for asymmetric fluorofunctionalizations by Nevado’s group,139) Jacobsen’s group,140–142) and Gilmour’s group.143,144)
On the other hand, the innovative concept of chiral phase-transfer catalysis was applied to asymmetric fluorination of alkenes by Toste’s group in 2011.145) A chiral phosphate anion acts as a phase-transfer catalyst to transport an electrophilic fluorinating reagent, Selectfluor, which is insoluble in non-polar solvents, thereby preventing undesired background reactions. Consequently, fluorination occurs in a chiral pocket constructed by a chiral phase-transfer catalyst. The availability of Selectfluor, which is a cationic and reactive reagent, without significant loss of reactivity is advantageous in this system. Subsequently, phosphate anion catalysis has been utilized for other fluorinations of the C=C bond including enamides and phenols.146–155)
3.1. Asymmetric FluorolactonizationInspired by the pioneering work of Toste’s group,145) we became interested in anionic phase-transfer catalysis for the enantioselective fluorofunctionalization of alkenes. As isobenzofuranone derivatives are found in many bioactive compounds, we initially focused on asymmetric fluorolactonization, which had not been reported before we started this project. Although the basic concept of anionic phase-transfer catalysis seemed promising, we were aware that simple application of a chiral phosphate anion catalyst would be difficult for the following two reasons: 1) the substrate carboxylate anion generated under basic phase-transfer conditions could work as an achiral phase-transfer catalyst to give the undesired racemic product; and 2) either flexible ion pairing of the anionic substrate with Selectfluor or loose salt bridging between the catalyst and the anionic substrate would result in low asymmetric induction.
To overcome the abovementioned inherent issues, we designed a new anionic phase-transfer catalyst consisting of a carboxylate anion functioning as a phase-transfer agent, a primary hydroxyl group as a site that captures the anionic substrate, and bulky substituents at the 3,3′-positions for improving solubility and making a chiral pocket156) (Chart 17a). The precatalysts were synthesized from chiral BINOL in 8 steps (Chart 17b). Dicarboxylic acid 42 was prepared by the sequential reaction of the introduction of a phosphate ester and reductive CO2 fixation.157) After protection of carboxylic acid moieties, bromination was carried out with Mg(TMP)2 (TMP = 2,2,6,6-tetramethylpiperidine). Ar groups at the 3,3′-positions were introduced through the Suzuki–Miyaura cross-coupling reaction, and subsequently the trimethylsilylethyl group was removed by treatment with tetrabutylammonium fluoride (TBAF). One of the carboxylic acids was selectively converted to the methyl ester under Kawabata’s conditions,158) and the ester moiety was reduced to afford precatalysts 45.
The synthesized precatalysts were evaluated for their efficiency in the fluorolactonization of 46a (Table 12). As the size of the substituents at the 3,3′-positions increased, not only the enantioselectivity but also the chemical yield improved remarkably (entries 1–3). When solvents less polar than toluene were used, the ee value of 47a increased (entries 4 and 5). The use of Na2SO4 as a dehydrating agent in cyclohexane provided the best result (entry 6). Interestingly, fluorolactonization itself proceeded even without a phase-transfer catalyst (entry 7).
 | ||||
|---|---|---|---|---|
| Entry | Precatalyst | Solvent | Yield [%]a) | Enantiomeric excess (ee) [%] |
| 1 | 45a | Toluene | 30 | 33 |
| 2 | 45b | Toluene | 37 | 68 |
| 3 | 45c | Toluene | 84 | 83 |
| 4 | 45c | n-Hexane | 77 | 87 |
| 5 | 45c | c-Hexane | 85 | 88 |
| 6b) | 45c | c-Hexane | Quant (99)c) | 88 |
| 7 | — | c-Hexane | 51 | — |
a) Determined by 1H-NMR analysis. b) Run with Na2SO4. c) Isolated yield.
Under the optimized reaction conditions, the substrate scope was investigated (Table 13). The substituent on the benzoic acid side did not have significant impact on the asymmetric induction, and the ee values were generally high irrespective of its electronic property (47b–g). Meanwhile, the substituent on the aryl ring of the styryl unit (R′) somewhat affected the enantioselectivity (47h–k). The reaction proceeded smoothly, even if the substrate had a heteroaromatic ring (47l). To our delight, alkyl-substituted vinyl benzoic acids 46m and 46n could be utilized, and the corresponding lactones were obtained in a reasonably enantioselective manner. The absolute stereochemistry was determined to be R by X-ray analysis of optically pure 47d.
 |
Based on analysis of NMR experiments, the proposed mechanism of the fluorolactonization is illustrated in Chart 18. The precatalyst tends to aggregate after treatment with a base. The substrate anion facilitates dissociation of the aggregate to generate binary complex I, for which the efficiency of transportation of Selectfluor is better than that of the catalyst aggregate. Selectfluor is brought into the liquid phase from the solid phase by complex I to make complex II, and then fluorination occurs in the chiral pocket.
During the mechanistic studies of fluorolactonization, we noticed that a binary complex between the catalyst and an anionic substrate is responsible for bringing Selectfluor into the liquid phase. This suggests that the cooperative action of two carboxylate anions located at an appropriate distance would enable high reaction efficiency. Taking this into account, we designed a linked-binaphthyl dicarboxylic acid precatalyst that could form a dicarboxylate catalyst under basic conditions159) (Chart 19a). The synthetic procedure of 52a, which gave the best result (vide infra), is described in Chart 19b. After the preparation of linked-BINOL derivative 51 through mono-protection of 49 and etherification with a linker unit, the phosphate group was introduced. Four bromines were replaced with the phenyl group, and then the following reductive CO2 fixation with lithium naphthalenide afforded 52a.157)
As the newly synthesized catalysts have no hydrogen bond donor unit, we choose an allylic amide as a substrate. To select the optimal catalyst, the fluorination of cinnamyl amine derivative 53a was carried out under the conditions described (Table 14). The length of the methylene linker affected the reaction efficiency, and the C3 linker provided the best result (entries 1–3). Phenyl groups at the 3,3′-positions were found to be essential for catalysis. Both the chemical yield and the ee value were reduced significantly when 52d and 52e were used, probably because of the decreased solubility and increased conformational freedom with the change in the steric environment around the carboxylate (entries 4 and 5). Hydroxy carboxylic acid precatalyst 45c, which is an efficient precatalyst for asymmetric fluorolactonization, and binaphthyl dicarboxylic acid 55 provided only 6 and 5% yields of 54a with low enantioselectivity, respectively (entries 6 and 7). Phosphoric acid 56 was totally ineffective under the reaction conditions (entry 8). As is the case with our fluorolactonization, the addition of Na2SO4 was effective in accelerating the fluorination of 53a (entry 9). In order to determine the stereochemistry of 54a, Mosher’s ester method160) was applied after hydrolysis of the dihydrooxazine ring under acidic conditions. This revealed that the diastereomers are derived from the stereoisomer at the benzylic position, while the facial selectivity of the fluorination step is well regulated.
 | ||||
|---|---|---|---|---|
| Entry | Precatalyst | Yield [%]a) | dra) | ee [%] |
| 1 | 52a | 58 | 1 : 1 | 89/96 |
| 2 | 52b | 24 | 1 : 1 | 75/87 |
| 3 | 52c | 32 | 1 : 1 | 75/90 |
| 4 | 52d | 10 | 7 : 3 | 40/— |
| 5 | 52e | 6 | 2 : 1 | 11/— |
| 6 | 45c | 6 | 2 : 1 | −20/−40 |
| 7 | 55 | 5 | 2 : 1 | −9/— |
| 8 | 56 | trace | — | — |
| 9b) | 52a | 76c) | 1 : 1 | 93/98 |
a) Determined by 1H-NMR analysis. b) Run with Na2SO4. c) Isolated yield.
Under the optimized reaction conditions, the fluorination of some cinnamyl amine derivatives was examined (Table 15). Irrespective of the electronic properties on the aromatic rings, high enantioselectivities were observed. However, the diastereoselectivities were almost 1 : 1. Coupled with the findings in the literature,161–164) fluorocyclization proceeds via the formation of a benzylic carbocation intermediate, which undergoes less stereoselective intramolecular C–O bond formation. Thus, we anticipated that the stereoselectivity would be improved if a substituent was attached at the β-position of an allylic amide to restrict the conformation of the carbocation intermediate.165,166) According to this hypothesis, fluorination with acyclic tri-substituted allylic amides was investigated under the described conditions (Table 16). As expected, the reactions proceeded with high diastereoselectivity, and the desired oxazine derivatives, which have a fluorine on a quaternary carbon center, were obtained in up to 99% ee. Although the heteroaromatic rings could be applicable to this reaction system, reduced diastereoselectivity was observed in the reaction of electron-rich substrates (54i and 54j).
 |
 |
In contrast to an acyclic substrate, a cyclic allylic amide was expected to provide the corresponding oxazine as a single diastereomer. Thus, we also carried out the fluorocyclization of cyclic allylic amides (Table 17). Aryl halide units were compatible under these reaction conditions, though the reaction rate decreased (58c and 58d). Substituents on the chromene framework did not have a significant impact on enantioselectivity, and the dimethyl unit of the chromene core was not essential for asymmetric induction. The electronic property on the aryl group of the amide unit slightly affected the reaction, but the ee values were generally good to high (58k–n).
 |
Encouraged by the promising results of the fluorocyclization of di- and tri-substituted allylic amides, we decided to examine the fluorination of tetra-substituted substrates under our phase-transfer catalysis, because vicinal tetra-substituted carbon centers would be obtained if the 6-endo-cyclization proceeds via a carbocation intermediate. However, it was found that the reaction of 59 preferentially afforded deprotonative fluorination product 60 in a highly enantioselective manner167) (Table 18). The absolute stereochemistry of the product was determined to be S by HPLC analysis in comparison with Toste’s report.149) Cyclized products were observed as a by-product in these reactions, though the range of the ee values was from 30 to 71%. If our working hypothesis is correct, the deprotonative fluorination product and the fluorocyclization product are formed from the same carbocation intermediate and a similar level of enantioselectivities should be observed for both products. However, this was not the case, as mentioned above. Although the reason is not clear at the moment, we speculate that there are two possibilities: 1) the reaction mechanism of deprotonative fluorination is distinct from that of fluorocyclization; or 2) kinetic resolution occurs during the carbocation intermediate. Further investigations regarding the reaction mechanism are ongoing.
 |
Among various methods for the introduction of trifluoromethyl groups, direct C–H trifluoromethylation is the most atom- and step-economical strategy. Compared with aromatic C–H trifluoromethylation reactions through either metal-catalyzed C–H activation or the addition of a trifluoromethyl radical,44–47) Csp3–H trifluoromethylation has been less well investigated, even up until today. Additionally, most researchers employed tetrahydroisoquinoline, which is known to be a highly reactive and well-studied substrate for Csp3–H functionalization, before we started this project.168–174)
In the course of our studies on trifluoromethylation chemistry, we found the solvent-dependent C–H trifluoromethylations of phenols175) (Table 19). Although phenol is a key component of many bioactive compounds, few attempts to introduce a trifluoromethyl group into phenols have been made. Togni’s group reported the trifluoromethylation of sodium phenolate derivatives in 2008, though control of the reaction site was difficult.176) Benzylic Csp3–H trifluoromethylation products 62 were obtained selectively in aprotic solvents, such as DMF (Table 19, left). In these reactions, only the para-benzylic position predominantly reacted, and substitution at the ortho-methyl group and at the aromatic C–H bond was negligible. Since Cu(II) species could also work as a catalyst, we considered that the Cu(II)–phenoxide complex was a key intermediate to generate the benzylic radical via hydrogen atom abstraction with a phenoxy radical. Meanwhile, the use of an alcoholic solvent, such as t-BuOH, allowed the aromatic C–H trifluoromethylation to proceed to provide 63 (Table 19, right). This solvent-dependent switching may be due to competitive decomplexation of the putative Cu–phenoxide complex with an excess amount of alcoholic solvent.
 |
To demonstrate the utility of this trifluoromethylation, potent FabI inhibitor 66, which is effective against resistant bacteria,177) was synthesized from 62b in 3 steps (Chart 20). The tert-butyl group was easily removed under acidic conditions, and the pyridine ring was connected using pyridine difluoride. Subsequently, the methyl ether was transformed to a hydroxy group to provide target molecule 66 with good efficiency.
Encouraged by the above findings on the C–H trifluoromethylation of phenols, we became interested in the development of other types of Csp3–H trifluoromethylations. ortho-Alkylbenzophenone derivatives are known to generate a photoenol via the generation of a ketyl radical intermediate by n–π* excitation and hydrogen atom abstraction under photoirradiation conditions.178) Photoenols have been utilized for direct functionalizations of benzylic Csp3–H bonds since the mid-20th century, and several types of electrophiles successfully undergo the C–C bond-forming reaction.179–183) Coupled with our trifluoromethylation chemistry, we envisaged that a photoenol would react with an electrophilic trifluoromethylating reagent. Thus, the C–H trifluoromethylation of ortho-alkylbenzophenone derivatives was attempted for investigation under photoirradiation conditions184) (Table 20). Optimization experiments revealed that the wavelength was important for this reaction, and the best result was observed when 365 nm of LED was used as a light source. Interestingly, only mono-trifluoromethylated product 68h was obtained even when the substrate had two methyl groups at both ortho-positions. In addition, the ortho-methyl group was selectively trifluoromethylated even in the presence of an additional methyl group at the meta- or para-position (68i and 68j). Acetophenone derivatives (R = alkyl) could also be applicable to this reaction, albeit with modest yields (68l and 68m).
 |
The proposed reaction mechanism is illustrated in Chart 21.185,186) Initially, the substrate ketone is excited by photoirradiation, and the resulting triplet state oxygen radical abstracts the hydrogen atom at the ortho-benzylic position. This is followed by rapid intersystem crossing (ISC) to produce a photoenol. Subsequently, the photoenol attacks the hypervalent iodine moiety of 1, followed by reductive elimination to provide 68. As the reaction proceeded smoothly even in the presence of radical trapping reagents, such as TEMPO and BHT, a radical coupling reaction between the biradical intermediate and a trifluoromethyl radical is unlikely.
During trifluoromethylation via a photoenol, we noticed that Togni reagent 1 slowly decomposed and a trifluoromethyl radical was generated by simple UV irradiation. Although there are many reports on the trifluoromethylation of aromatic rings by the addition of a trifluoromethyl radical,44–47) we considered that this simple protocol could be an alternative for trifluoromethylation using a trifluoromethyl radical. Hence, the trifluoromethylation of aromatic and heteroaromatic rings was investigated with an electrophilic trifluoromethylating reagent under simple photoirradiation conditions187) (Table 21). As a result of screening of electrophilic trifluoromethylating reagents, it was found that Umemoto reagent II 71188) produced the best result under 375-nm LED irradiation (Table 21). Electron-rich aromatic rings, such as phenols and anilines, could be successfully trifluoromethylated selectively (70a–d). Although pyridine alone was not transformed, pyridines having electron-donating groups underwent the reaction, providing 70e and 70f. The reaction of 8-aminoquinoline proceeded to afford 5-trifluoromethylated and 7-trifluoromethylated products (70g and 70g′), together with a small amount of the di-trifluoromethylated compound. As expected, electron-rich heteroaromatic compounds were a suitable substrate for this reaction (70h–l), and 70m was obtained from caffeine in 58% yield.
 |
Csp3–H fluorination has been studied for a long time, and the classical method is to use molecular fluorine or CF3OF under photoirradiation.189–191) Since these fluorinating reagents have extremely high reactivity, selective mono-fluorination is difficult in such a classical reaction. Reflecting the development of useful, safer electrophilic fluorinating reagents bearing the N–F bond, however, recent progress in this area has enabled more selective C–H fluorination of benzylic and aliphatic compounds in the presence of metal catalysts or organocatalysts.31,34,38,42,192)
On the other hand, an amino group exists in various molecules, including naturally occurring products, pharmaceutical compounds, and functional materials. A substitution with phthalimide, which is exemplified by Gabriel reaction193) and Mitsunobu reaction,194) is one of the promising protocols for the introduction of amino equivalents. Phthalimide is normally used as a protective group and is removed at a later stage to generate the corresponding amino group.
Therefore, the phthalimide unit could be found in various synthetic intermediates. Nevertheless, the reactivity of phthalimide under photoirradiation has been studied in less detail. In 1972, a photocyclization reaction via C–H abstraction by a photoexcited phthalimide was reported by Kanaoka.195) The photoexcitation of phthalimide was utilized to generate azomethine ylides from phthalimide derivatives having a silicon-containing electron donor by the group of Mariano and Yoon.196) Later on, some transformations initiated by photoexcitation of the phthalimide unit were studied.
Inspired by this background, we tried to investigate the C–H fluorination of aliphatic compounds with a phthalimide unit.197) Under the optimal conditions, various phthalimide derivatives were investigated (Table 22). While the reaction of N-butyl phthalimide proceeded selectively at the δ-position and 73a was obtained in 72% yield, N-propyl phthalimide provided β-fluorinated product 73b. When using a pentyl substrate, fluorination occurred at both the γ- and δ-positions, and the C–H bond at the δ-position was more reactive (73c). These results suggested that phthalimide works as an electron-withdrawing group and deactivates proximal C–H bonds.198) A tertiary carbon center underwent C–H fluorination, although the reaction rate was slower, probably due to its steric hindrance (73d). Benzylic C–H bonds were also reactive under these reaction conditions (73e–g). In addition, phthalimide-protected amino acid esters could be applied to the present fluorination reaction conditions, and 73h and 73i were obtained in 45 and 70% yield, respectively. Compound 73j was afforded as a diastereomeric mixture. Interestingly, the C–H bond at the α-position of the oxygen atom was selectively fluorinated to provide 73k and 73l in the reaction of amino acid esters 72k and 72l. These results indicate that the hydrogen atom abstraction of this reaction does not proceed by the intramolecularly excited phthalimide group because the reactive C–H bond is far from the phthalimide group.
 |
To shed light on the importance of the phthalimide unit, other protective groups including benzyl amide, o-nitrobenzene sulfone amide, succinimide, and 4,5,6,7-tetrachlorophthalimide were introduced. As a result, no reaction was observed with these protective groups. The triplet-state energy of N-methyl phthalimide was reported to be 3.1 eV,199) and the energy gap between the singlet and triplet states of Selectfluor was reported to be 2.6 eV.200) Based on the experimental results and from literature, we consider that our C–H fluorination proceeds via a triplet–triplet energy transfer between phthalimide and Selectfluor and hydrogen atom abstraction with the resulting nitrogen radical, which may occur either in an intramolecular or intermolecular fashion.
This article summarizes our recent contributions to fluorine chemistry. Based on the working hypothesis that Togni reagent could be activated by coordination to a Lewis acid catalyst, we developed the trifluoromethylations of indole and allylsilane, oxy-trifluoromethylations, carbo-trifluoromethylations, and amino-trifluoromethylations. In particular, mechanistic studies on amino-trifluoromethylation revealed that the rate-limiting step included the Cu(II) ion, allylamine substrate, and Togni reagent, indicating that our working hypothesis is reasonable. Furthermore, hydro-, iodo-, and vinylic trifluoromethylations were accomplished using the Togni reagent–electron donor molecule system.
In asymmetric fluorofunctionalizations, we designed and synthesized two new phase-transfer catalysts. The hydroxy carboxylic acid precatalyst 45c was found to be effective for asymmetric fluorolactonization. Taking the results of NMR experiments into account, the hydroxyl group interacted with the anionic substrate, and the temporary binary complex between 45c and the substrate was better for the transportation of Selectfluor. According to this mechanistic consideration, a linked binaphthyl dicarboxylate catalyst was designed. Notably, 52a showed high efficiency in the asymmetric fluorocyclization and deprotonative fluorination of allylic amides.
We also investigated the introduction of a fluorine-containing functional group to C–H bonds. The solvent-dependent C–H trifluoromethylation of phenol derivatives and trifluoromethylation of ortho-alkylbenzophenone derivatives were achieved. We noted that a trifluoromethyl radical could be efficiently generated from Umemoto reagent II under simple photoirradiation conditions, and the trifluoromethylation of aromatic compounds was demonstrated using this method. Meanwhile, the C–H fluorination of N-alkyl phthalimide derivatives was accomplished with Selectfluor under photoirradiation.
In terms of trifluoromethylation chemistry, our reaction systems have contributed to developing new trifluoromethylations and preparing various trifluoromethylated compounds. The construction of new trifluoromethylation systems that can be utilized for the selective C–H transformation at later stages will be topics for future study. On the other hand, we believe that the high reactivity of our phase-transfer catalyst will open the way for new asymmetric fluorofunctionalizations using Selectfluor and that their integration with other reaction systems, for example, C–H fluorination, will lead to the development of novel stereoselective fluorinations.
My deep appreciation goes to all of the author’s co-workers. This work was carried out under the direction of Prof. Mikiko Sodeoka at RIKEN and Prof. Yoshitaka Hamashima at the University of Shizuoka. I am grateful to Prof. Sodeoka and Prof. Hamashima for helpful discussions and support. I would like to thank Dr. Daisuke Hashizume of RIKEN for the X-ray analysis measurements of 47d. This work was supported by Grant-in-Aid for Young Scientists (B) and for Scientific Research (C) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan; Naito Foundation (Japan); Uehara Memorial Foundation; Research Foundation for Pharmaceutical Sciences; and Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS) from Japan Agency for Medical Research and Development (AMED). I am also thankful towards Tosoh F-Tech, Inc. for the generously providing the Ruppert–Prakash reagent.
The author declares no conflict of interest.
This review of the author’s work was written by the author upon receiving the 2019 Pharmaceutical Society of Japan Award for Young Scientists.
