2020 年 68 巻 10 号 p. 907-945
Oxygen atoms have a lone pair of electrons, so they have high chelation ability, high nucleophilic ability, stabilizing ability of adjacent cations, and take a chelate or oxocarbenium ion structure with Lewis acids and metals. I took advantage of these properties to develop three new reactions, 1) asymmetric synthesis of chiral quaternary carbon centers, 2) asymmetric synthesis using acetal functions, and 3) organic chemistry using acetal-type reactive salt chemical species, and applied them to biologically active natural products synthesis. New reactions described here are all innovative and useful for natural products synthesis. In particular, the first asymmetric synthesis of fredericamycin A, and concise asymmetric synthesis of anthracycline antibiotics, scyphostatin, (+)-Sch 642305, (−)-stenine, clavolonine, (+)-rubrenolide, (+)-rubrynolide, (+)-centrolobine, and decytospolide A and B, etc., are noteworthy. Furthermore, since reactions using acetal-type reactive salt chemical species allow the coexistence of functional groups that normally cannot coexist, the reactions using reactive salts have potential to change the retrosynthesis planned based on conventional reactions.
Oxygen is the most popular heteroatom. Oxygen atoms have a lone pair of electrons, so they have high chelation ability, high nucleophilic ability, stabilizing ability of adjacent cations, and take a chelate or oxocarbenium ion structure with Lewis acids and metals. Although the characteristic reactivity of the oxygen atom has been well known for a long time, I took advantage of these properties to develop new and innovative synthetic organic reactions and applied them to natural product synthesis. This review describes the development of new reactions, 1) asymmetric synthesis of chiral quaternary carbon centers, 2) asymmetric synthesis using acetal functions, and 3) organic chemistry using acetal-type reactive salt chemical species. Research on their applications to the synthesis of biologically active natural products is described. In the following section, for each response, background of the research and overview are first described, followed by a detailed study.
Background of the research and overview: My predecessor, Professor Yasuyuki Kita, was engaged in asymmetric total synthesis of fredericamycin A (1.1)1–3) and the determination of its absolute configuration, and I cooperated in developing an absolute configuration determination method in it. 1.1 shows potent anticancer activity against various in vivo tumor models, such as P388 leukemia, B16 melanoma, and CD8F mammary. In addition, 1.1 does not show mutagenicity in the Ames test.4–7) Therefore 1.1 attracted much attention from many synthetic chemists. As shown in Fig. 1, 1.1 has a unique structure, consisting of two sets of peri-hydroxy tricyclic aromatic moieties connected via a spiro quaternary carbon, which has only one asymmetric center by the presence of a single methoxy group at the farthest position on the A-ring. When we started our project, five racemic8–12) and one optically active13) syntheses were already reported. However, one such optically active synthesis was achieved by optical resolution of the racemic precursor using a chiral column by HPLC. Therefore the absolute configuration of fredericamycin A (1.1) was still unknown at the time.
(Color figure can be accessed in the online version.)
To determine the absolute configuration of fredericamycin A (1.1), we developed an asymmetric quaternary carbon construction method by the rearrangement reactions of α,β-epoxy acylates with clear absolute configuration, and succeeded in the first asymmetric synthesis and determination of the absolute configuration of 1.1.
Furthermore, as a development of reactions using epoxy alcohol derivatives, we developed two new reactions, a rearrangement reaction of α,β-epoxy sulfonates and a reaction of epoxy alcohols with phenyliodine(III) bis(trifluoroacetate) (PIFA), and applied them to concise asymmetric synthesis of natural products.
1.1. Asymmetric Synthesis by Stereospecific Rearrangement of α,β-Epoxy Alcohol Derivatives14)Rearrangement of α,β-epoxy acylates: It is well known that determining absolute configuration of chiral quaternary carbon centers is generally difficult. Especially, the determination is quite difficult for those with similar substituents. Fredericamycin A (1.1) is such a compound; two substituents forming the quaternary spiro center are very similar and distinguished only by a single methoxy group at the farthest position. X-ray analysis has powerful potential to determine the absolute configurations. However, crystallization of compound was necessary for X-ray analysis at that time, and fredericamycin A (1.1) did not crystallize. Therefore I surmised that a new method was necessary to determine the absolute configurations of such quaternary carbon centers whose absolute configurations were difficult to determine by spectroscopic methods. In other words, I considered that the synthesis of asymmetric quaternary carbon center of fredericamycin A by stereospecific conversion of the compound, whose absolute configuration was obvious, would give the answer of the absolute configuration of the quaternary carbon center of fredericamycin A (1.1). As such a new method, I assumed the stereospecific rearrangement of tetra-substituted α,β-epoxy alcohol derivatives, whose optically active form could be easily synthesized. As for the rearrangement of α,β-epoxy alcohol derivatives, the reactions of their silyl ethers had been reported at the time. However, in these cases, the R2 substituent next to the silyloxy group is rearranged due to the electron-donating nature of silyl functional group14) (Chart 1, eq. 1). On the other hand, the acyl group is well recognized to be a representative functional group that participates in adjacent carbocations. Indeed, although only two reports on Lewis acid treatment of α,β-epoxy acylate were previously reported, the reaction gave many products including the rearrangement product, and was not suitable for rearrangement reaction.15,16) However, I envisioned, if the rearrangement reaction of α,β-epoxy acylates would proceed selectively via α-cleavage of the epoxy ring, the electron-withdrawing property of the acyl group suppress the rearrangement of the R2 group and the R5 group is rearranged (Chart 1, eq. 2). It was thought that an obtained product would be valuable for asymmetric synthesis of fredericamycin A (1.1) and give a good tool for determination of its absolute configuration. α,β-Epoxy acylates with unambiguously determined absolute configurations are synthesized easily, as described later.
(Color figure can be accessed in the online version.)
As model studies, we first examined the reactions of bicyclic and tricyclic compounds. Acetyl and benzoyl groups were used as acyl group (Chart 2). Eqs 1 and 2 show the results in bicyclic system.17,18) The reactions of cis-α,β-epoxy acylates with BF3·Et2O proceeded via β-cleavage of epoxy ring to give the rearranged products in high yields (eq. 1). The reason that β-cleavage of epoxy rings occurred must be due to the stability of the cation intermediates. Thus the β-cation intermediates are more preferable than the α-cation intermediates because of electron-withdrawing nature of acyl groups. However, the reactions of trans-α,β-epoxy acylates proceeded through neighboring-group participation mainly to give the mixture of the diol mono-acylate derivatives with small amount of the rearranged products (eq. 2). On the other hand, the reactions of both tricyclic epoxy acrylates having a benzene ring closer to fredericamycin A proceeded via α-cleavage of oxirane ring because of stability of benzylic carbocation irrespective of their stereochemistry.19) Although cis-isomers proceeded to afford ortho ester and enone, respectively (eq. 3), trans-isomers gave the desired rearranged products in high yields (eq. 4).
(Color figure can be accessed in the online version.)
The difference of the reactivity of trans-isomers in bicyclic and tricyclic systems was attributed to differences in the stabilities of the intermediates. The tetracyclic intermediate 1-i, in which the acyl group participated in the adjacent cation, is supposed in tricyclic system (Fig. 2). But the presence of benzene ring caused its instability compared with the tricyclic intermediate in eq. 2 and inhibited the formation of 1-i. As a result, rearrangement reaction proceeded.
(Color figure can be accessed in the online version.)
Asymmetric synthesis and determination of absolute configuration of fredericamycin A (1.1)20): Using this rearrangement reaction as key reaction, we synthesized natural and unnatural fredericamycin A21,22) (Chart 3). Asymmetric reduction of tricyclic enone 1.2 with CBS′ reagent23) to give (R)-allyl alcohol 1.3 in 74% enantiomeric excess (ee). Oxidation of 1.3 with vanadyl acetylacetonate [Va(acac)2] followed by Mitsunobu reaction24) using optically pure (−)-camphanic acid afforded trans-epoxy camphanate (−)-1.4 in 74% diastereomeric excess (de), which was purified by column chromatography to give optically pure (−)-1.4 (≥99% de). The absolute stereochemistry of (−)-1.4 was clearly determined by X-ray analysis. The rearrangement reaction of (−)-1.4 (≥99% de) using BF3·Et2O proceeded with complete stereoselectivity, and the spiro compound (+)-1.5 (≥99% de) was obtained in 94% yield. The absolute stereochemistry of (+)-1.5 was also confirmed by X-ray analysis. Compound (+)-1.5 was converted to natural fredericamycin A without affecting the absolute configuration of spiro center. Acetalization of (+)-1.5 was performed prior to alkaline hydrolysis to prevent easy racemization of the spiro center by retro-aldol and aldol reaction. Dess–Martin oxidation of the hydroxyl group obtained by hydrolysis followed by sulfinylation with lithium bis(trimethylsilyl)amide and PhSSO2Ph gave keto acetal 1.6. Deacetalization of 1.6 followed by continuous removal of PhSH and oxidation by m-chloroperoxybenzoic acid (mCPBA) resulted in an optically active CDEF-ring unit 1.7 as a diastereomeric mixture at sulfur centers (approximately 1 : 1). The following several-step sequences, (1) treatment of 1.8 with sodium hydride (NaH), (2) [4 + 2]-cyclocondensation with 1.7, (3) methylation of hydroxyl group, (4) selective demethylation of the methyl ether on the F-ring by treatment with NaBr and p-toluenesulfonic acid (p-TsOH),25) and (5) oxidation with SeO2, afforded the hexacyclic product (S)-1.9 (54, 97% ee). Wittig reaction of (S)-1.9 with trans-2-butenyl triphenylphosphonium bromide yielded a 5 to 1 mixture of (E,E)- and (E,Z)-side chain isomers in 46% yield. The mixture was deprotected with BBr3 followed by autoxidation to give a 5 to 1 mixture of 1.1 and its (E,Z)-side chain isomer in a total yield of 74%. The pure natural fredericamycin A (1.1) was obtained in 40% yield after purification by HPLC column. Unnatural fredericamycin A, ent-1.1, was also synthesized in the same manner via (R)-1.9, obtained by condensation of 1.7 with 1.10 in place of 1.8. Since the natural fredericamycin A (1.1) was synthesized from the coupling reaction of 1.7 and 1.8, the absolute configuration of its spiro center was unambiguously determined to be S.
(Color figure can be accessed in the online version.)
The rearrangement reaction of α,β-epoxy acylates has been also applied to acyclic and monocyclic systems. Figure 3 shows α,β-epoxy acylates examined. They gave the corresponding rearranged products in high yields.18,26–29)
Rearrangement reaction in a monocyclic system was applied to the concise synthesis30) of (S)-(+)-sporochnol A,31) which has a significant feeding deterrence toward herbivorous fish activity. Asymmetric reduction of the enone 1.11 with Corey’s reagent,23) stereoselective epoxidation, and Mitsunobu reaction24) with p-NO2-C6H4COOH gave the trans-epoxy p-nitrobenzoate 1.12 (99% ee). Treatment of 1.12 with Lewis acid, (C6F5O)3Al, afforded the rearranged product 1.13 in 96% yield without loss of chirality. LiAlH4 reduction of 1.13 followed by treatment with PhI(OAc)2 gave the dialdehyde 1.14 in 74% yield. Olefin aldehyde 1.15 was obtained by selective monoacetalization of 1.14 with meso-hydrobenzoin, methylenation by the Wittig reaction, and deacetalization by acetic acid. Wittig reaction of the aldehyde 1.15 with isopropenyl triphenyl phosphine and n-BuLi followed by MeMgI treatment32) yielded optically pure (S)-(+)-sporochnol A (Chart 4).
As mentioned above, the acyl groups, the electron-withdrawing groups, have caused a new rearrangement. Incidentally, the sulfonyl groups are also strong electron-withdrawing groups and do not participate in adjacent cations. We then assumed that the rearrangement reactions of α,β-epoxy sulfonates would proceed in the same way as α,β-epoxy acylates. That is, the reaction of the epoxy sulfonates in which two substituents do not have any electron-stabilizing ability proceeds via β-cation intermediates due to the electron-withdrawing ability of the sulfonyl group. On the other hand, the reaction of the epoxy sulfonates having electron-stabilizing substituent such as aromatic group at the α- or β-position is thought to proceed via α- or β-cation intermediate next to aromatic group.
In fact, the rearrangement reaction of the epoxy tosylates with two alkyl substituents proceeded via a β-cation intermediate to give an α-keto tosylate containing β-quaternary carbon (Chart 5, eq. 1). On the other hand, as expected, the reactions of the substrates having phenyl group at their α- or β-position yielded the corresponding products via the cation intermediates next to phenyl group (Eqs. 2 and 3).33)
(Color figure can be accessed in the online version.)
Chart 6 shows asymmetric formal synthesis of (−)-aphanorphine by application of the rearrangement reaction of epoxy sulfonate.33) The reaction of optically active epoxy tosylate 1.16 having meta-methoxyphenyl group at the β-position with EtAlCl2 afforded the rearranged product 1.17 in 93% yield with 98% ee. Reductive removal of its tosyloxy group resulted in 77% yield of optically active 1.18 with 98% ee, which had been already converted to (−)-aphanorphine by Ogasawara and colleagues.34)
(−)-α-Herbertenol and (−)-herbertenediol are known as chemical markers for liverworts belonging to the genus Herbertus.35) These compounds were also synthesized using the rearrangement reaction of epoxy tosylates 1.19a [for (−)-α-herbertenol] and 1.19b [for (−)-herbertenediol]33,36) (Chart 7). Rearrangements of α,β-epoxy tosylates 1.19a and 1.19b with EtAlCl2 proceeded without loss of ee, and 1.20a (93%) and 1.20b (99%) were obtained, respectively. Next reduction of tosyloxy group also proceeded in good yields to give the corresponding products 1.21a and 1.21b. MeLi attacked their ketones followed by Burgess reagent37) treatment, and subsequent cyclopropanation in the usual way38) afforded cyclopropyl compounds 1.22a and 1.22b, respectively. Reductive opening of the cyclopropane ring and acidic cleavage of the methyl ether bond yielded (−)-α-herbertenol and (−)-herbertenediol.
While studying the rearrangement reactions of α,β-epoxy alcohol derivatives, we found an interesting domino reaction. That is, the reaction of epoxy alcohols with PIFA in the presence of water resulted in different types of products depending on the substitution pattern of the epoxide ring (Chart 8). Tetrasubstituted or 2-alkyl trisubstituted epoxy alcohols reacted with PIFA in four-steps domino reaction: 1) substitution of the trifluoroacetoxy group of PIFA with an alcohol of the substrate, 2) intramolecular nucleophilic substitution of the oxygen atom of the epoxy ring followed by nucleophilic addition of water at 3-position, 3) oxidative cleavage of the carbon bond between the vicinal diol, and 4) lactol formation by intramolecular attack of the hydroxyl group to the aldehyde function, and monocyclic lactols were obtained, respectively. In the case of 2-unsubstituted epoxy alcohols, the bicyclic lactols were formed by one more intramolecular lactol formation. This reaction had generality, and bicyclic and monocyclic α,β-epoxy alcohols in Fig. 4 could be used.
(Color figure can be accessed in the online version.)
This new domino reaction was applied to the concise asymmetric synthesis of (+)-tanikolide (Chart 9). Reaction of the optically active epoxy alcohol 1.23 (98% ee) with PIFA in the presence of water provided the bicyclic lactol 1.24 in one-pot. Diisobutylaluminium hydride (DIBAL-H) reduction of 1.24 followed by oxidation of the resulting lactol yielded (+)-tanikolide (98% ee).
Background of the research and overview: Acetals have been originally used as protecting groups for carbonyl compounds.41–43) After the discovery of the acetal Mukaiyama aldol reaction,44) they have been also recognized as synthetic equivalents of carbonyl groups in addition reactions. Thus the first asymmetric syntheses using chiral acetals were reactions using chiral acetals as chiral synthetic equivalents of carbonyl groups. From the mid-1970s, asymmetric synthesis of optically active secondary alcohols using optically active acetals has been studied. In particular, C2-symmetric acetals from optically active C2-symmetric diols have been actively used because of the single formation of acetals even from non-symmetric carbonyl compounds (Chart 10, route b). This reaction was actively studied until the mid-1980s and was used for asymmetric synthesis of natural products. However, advances of enantioselective synthesis of secondary alcohols by reaction of aldehydes and nucleophiles in the presence of asymmetric catalyst abolished this type of asymmetric synthesis. Another category of asymmetric synthesis using C2-symmetric acetals is asymmetric synthesis of chiral centers from prochiral centers using chiral acetals as chiral protecting groups for carbonyl groups (route a). However, when we began researching asymmetric synthesis, in 1984, only one such type of asymmetric synthesis had been reported. Since then, various types of asymmetric syntheses in this category have been developed.45,46)
(Color figure can be accessed in the online version.)
All asymmetric reactions using chiral C2-symmetric acetals had been categorized to routes a and b before we developed the reactions classified to route c. The reactions in route c can form at least two chiral centers in remote positions. As far as I know, all reactions developed are new asymmetric reactions using C2-symmetric acetals.
We first developed asymmetric synthesis belonging to route a and its application to natural products synthesis. We next developed one asymmetric synthesis belonging to route b. Thereafter, asymmetric reactions belonging to route c and their applications to many natural products synthesis were developed. Furthermore, as an extension of route c asymmetric desymmetrization of meso-diols has been achieved.47–49)
2.1. Asymmetric Synthesis Using C2-Symmetric Acetals2.1.1. Asymmetric Synthesis Using α-Keto Acetal Derivatives (Categorized to Route a)Optically active tert-alcohol moieties are found in many natural products. The most direct way to construct them is asymmetric nucleophilic addition to ketones. I envisioned nucleophilic addition to chiral α-keto acetals, in which the ketone is located next to chiral C2-symmetric acetal.
We found the reactions of chiral C2-symmetric acetal 2.1, prepared from optically pure (2S,3S)-1,4-dimethoxy-2,3-butanediol, with Grignard reagents gave the best results in cyclic and acyclic systems.50–54) The presence of oxygen atom in the side chain of acetal was essential to obtain high de (92–100% de). Thus the use of the acetals prepared from optically pure (2R,3R)-2,3-butane diol instead of (2S,3S)-1,4-dimethoxy-2,3-butanediol gave poor results. The reaction was then supposed to proceed via chelation transition state 2-i to give the optically active tert-alcohol derivatives 2.2 in high diastereoselectivity (Chart 11).
(Color figure can be accessed in the online version.)
Diastereoselective nucleophilic addition to α-keto acetal giving tert-alcohol moiety was applied to asymmetric synthesis of several natural products such as (R)- and (S)-mevalolactone,52,53) and anthracycline antibiotics.55–60) Chart 12 shows asymmetric synthesis of γ-rhodomycinone.57,58) Nucleophilic addition of EtMgCl to 5,8-dimethoxy-1-oxo-β-tetralone 1-acetal 2.4, prepared from 5,8-dimethoxy-1-tetralone 2.3 in three steps, afforded a single diastereomer 2.5 in good yield. Acid hydrolysis of 2.5 followed by KBH4 reduction gave diol 2.6. Ceric ammonium nitrate (CAN) oxidation afforded quinone 2.7, which was coupled with 4-acetoxy-5-methoxyhomophthalic anhydride 2.8.61) The anhydride 2.8 was then treated with NaH and coupled with 2.7 to afford the adduct 2.9. Treatment of 2.9 with aqueous CF3CO2H resulted in deacetylation and transfer of the quinone moiety, yielding 2.10. Demethylation of 2.10 gave (−)-γ-rhodomycinone.
The same chelation of chiral acetal and organometallics was also effective in the reaction of α-aldoxime-ether acetal. Nucleophilic addition of an organocerium reagent, in situ prepared from Grignard reagent and CeCl3, to α-aldoxime-ether acetal gave optically active amines in highly diastereoselective manner62) (Chart 13).
(Color figure can be accessed in the online version.)
Asymmetric reduction of chiral α-keto-β,γ-unsaturated acetals with LiAlH4 was also attained in a highly diastereoselective manner63) (Chart 14). It is worthy to note that the stereochemistry of the obtained allyl alcohols depended on the added inorganic salt. Addition of LiBr caused si-face attack to give (R)-allyl alcohols as major products. On the other hand, addition of MgBr2 instead of LiBr gave (S)-allyl alcohols as major products by re-face attack.
(Color figure can be accessed in the online version.)
The diastereoselective reduction was applied to the synthesis of both enantiomers of 3′-methoxy-4′-O-methyljoubertiamine64) (Chart 15). Reduction of 2.11 with LiAlH4 in the presence of LiBr afforded (R)-allyl alcohol 2.12. Claisen-Eschenmoser [3,3]-sigmatropic rearrangement65) of 2.12 gave the rearranged product 2.13 without loss of ee. 2.13 was reduced by LiAlH4 followed by deprotection of acetal to give (R)-(−)-3′-methoxy-4′-O-methyljoubertiamine. On the other hand, (S)-allyl alcohol 2.14 obtained by LiAlH4 reduction of 2.11 in the presence of MgBr2 was converted to (S)-(+)-3′-isomer in the same way as (R)-(−)-3′-isomer.
As a development of organic synthesis using Beckmann cleavage reaction,66–68) a new asymmetric carbon–carbon bond formation classified as route b in Chart 10 was achieved by combining Beckmann cleavage reaction and C2-symmetric acetal chemistry69) (Chart 16). When racemic α-methoxycycloalkanone oxime acetate 2.16 was treated with a catalytic amount of trimethylsilyl trifluoromethanesulfonate (TMSOTf) in the presence of (2R,4R)-2,4-bistrimethylsilyloxypentane, the chiral acetal 2.17 was obtained quantitatively via oxocarbenium ion intermediates 2-ii. By asymmetric nucleophilic substitution of 2.17, optically active ω-cyano compounds 2.18 were obtained in high yields with high de. The secondary hydroxyl group of the allylic 2.18 (Nu = allyl, n = 1) was oxidized by pyridinium chlorochromate (PCC) followed by aqueous alkaline treatment to remove auxiliary unit and alkaline hydrolysis of nitrile. Methylation of the resulting carboxylic acid afforded the hydroxy methylester 2.19. Ozonolysis of the double bond of 2.19 followed by NaBH4 reduction gave the dihydroxy methylester 2.20, a key intermediate for the synthesis of α-(R)-lipoic acid.
As shown in Chart 10, the reactions belonging to routes a and b are basically reactions that form one chiral center with the exceptions of Diels–Alder reactions. Incidentally, C2-symmetric acetal has two chiral centers. Therefore I next planned an intramolecular haloetherification of ene acetals, envisioning efficient use of introducing at least two chiral centers in the product (Chart 17). That is, the intramolecular haloetherification of ene acetals in the presence of an alcohol first forms a halonium ion on the olefin, which is attacked by the oxygen atom of the acetal, followed by alcohol attack on acetal carbon to produce the products with two chiral centers. The use of chiral acetal in this reaction allows for asymmetric induction. Since the products are another type of acetals, nucleophilic substitution followed by removal of the acetal accessary yields diol derivatives with two remote optically active centers. Therefore as an optically active diol, I used chiral hydrobenzoin, which is abundantly available70) and various methods can be used to remove its accessary since it should be considered two sets of benzyl alcohol.
(Color figure can be accessed in the online version.)
We first examined the reaction of ene acetals.71,72) The ene acetal 2.21a was prepared from (R,R)-hydrobenzoin and 4-pentenal in quantitative yield. Treatment of 2.21a with iodonium di-sym-collidine perchlorate [I(coll)2ClO4] in the presence of H2O gave two hydroxyl aldehyde ethers 2.22a and 2.23a in a 6 : 1 ratio. On the other hand, when N-bromosuccinimide (NBS) was used instead of I(coll)2ClO4, both the yield and the selectivity decreased. The reaction must proceed via the 8-membered hemiacetal intermediate by initial formation of a halonium ion at the olefin unit, an intramolecular nucleophilic attack of the acetal oxygen atom, followed by introduction of water (Chart 18).
(Color figure can be accessed in the online version.)
Attempts to use carbon nucleophiles instead of H2O have failed. However, using various alcohols (ROH), the 8-membered acetals 2.24a (70–80% de) were obtained diastereoselectively. Compound 2.24a (R = CH3OCH2CH2–, 100% de) excluding the minor diastereomer was converted to optically active 1,4-diol 2.25a by a three-step sequence: 1) LiAlH4 reduction, 2) stereospecific substitution by Grignard reagent, and 3) removal of the accessary of the acetal unit by Birch reduction (Chart 19).
(Color figure can be accessed in the online version.)
In the case of hexenal, the ene-acetal prepared from chiral hydrobenzoin could not form a 9-membered acetal and use of (1S,3S)-1,3-diphenyl-1,3-propanediol instead of chiral hydrobenzoin was needed. Chart 20 summarizes the syntheses of optically active 1,4- or 1,5-diols from the ene-acetals 2.21a,b. The reaction process is the same as Chart 19. The nucleophilic substitution reactions using NaCN or NaCH(CO2Me)2 in addition to LiAlH4, proceeded without any problem. The deprotection process from 2.26a,b obtained by Grignard reaction to 2.25a,b could be achieved by Birch reduction or hydrogenolysis.
(Color figure can be accessed in the online version.)
The above method was applied to concise synthesis of solenopsin A and a civet constituent72) (Chart 21). Solenopsin A is a component of the venom of the fire ant S. invicta. Birch reduction of 2.27, synthesized from hexenal and (1S,3S)-1,3-diphenyl-1,3-propanediol as shown in Chart 20, yielded optically pure 1,5-diol 2.28, which was already converted to Solenopsin A by Solladie´ and Huser.73)
(+)-(S,S)-(cis-6-Methyltetrahydropyran-2-yl)acetic acid is a constituent isolated from the perfume material civet,74) a glandular secretion of the civet cat Viverra civetta. OsO4 oxidation of the olefin of 2.29 followed by NaBH4 reduction then catalytic hydrogenolysis yielded the triol 2.30. Cyclic ether 2.31 was obtained by the reaction of 2.30 with PhC(OMe)3.75,76) Alkaline hydrolysis of 2.31 followed by Jones oxidation yielded the cyclic ether 2.32, a civet constituent.
Table 1 shows the results of the reactions of ene ketals 2.33 having hydroxyl or carboxylic acid group in the same molecule.77) On intramolecular iodoetherification with bis(2,4,6-collidine)iodonium hexafluorophosphate [I(coll)2PF6] in the absence of external alcohol, the ene ketals 2.33 gave the corresponding spirocyclic acetals 2.34, respectively. The reaction proceeded in a highly diastereoselective manner via fused bicyclic oxocarbenium ions and various ene ketals were available.
 |
(Color figure can be accessed in the online version.)
The spirocyclic acetal 2.34a in Table 1 was converted to the optically active pheromone 2.36 of the wasp Paravespula vulgaris78) (Chart 22). Radical reduction79–81) of 2.34a gave the corresponding methyl substituted spirocyclic ketal 2.35. Treatment of 2.35 with excess CAN in CH3CN–H2O (1 : 1)82–84) afforded the target pheromone 2.36 through hydroxy hemiacetal and successive simultaneous intramolecular cyclization.
Next, a desymmetric haloetherification of σ-symmetric diene acetals was assumed. If the intramolecular haloetherification of diene acetals would proceed in a diastereoselective manner, products with olefin, halogene, and acetal functions would be obtained, which are expected to be a useful synthon. Incidentally, the synthesis of natural products requires some selective reactions regarding regiochemistry and stereochemistry during the transformation. As shown in the previous section, the chiral auxiliary, the chiral acetal unit, served as the source of asymmetric induction, protecting group, and hydroxyl groups. In addition to these advantages, we planned actively to use the chiral acetal moiety as template to achieve regioselectivity and stereoselectivity during transformation of natural products synthesis. In other words, we envisioned a new concept: chiral auxiliary multiple-use method (Chart 23).
(Color figure can be accessed in the online version.)
Haloetherification of σ-symmetric cyclohexa-1,4-diene acetals: Cyclohexa-1,4-diene derivatives 2.37a–c having chiral C2-symmetric acetal from (R,R)-hydrobenzoin were treated with NBS in the presence of methoxyethanol. The reactions proceeded via oxocarbenium ion intermediates to give the expected haloetherification products 2.38a–c in fairly high level of diastereoselectivity. However, the use of I(coll)2ClO4) instead of NBS did not give fruitful results85) (Chart 24).
(Color figure can be accessed in the online version.)
We then examined the intramolecular bromoetherification of 2.37a in the presence of MeOH in place of methoxyethanol for considering the asymmetric synthesis of natural products. Radical reduction of the optically pure 2.39 (100% ee) purified by column separation afforded 2.40 in high yield (Chart 25).
Compound 2.40 was applied to asymmetric synthesis of scyphostatin.86,87) Scyphostatin displays the most potent activity (IC50 = 1.0 µM) to a neutral sphingomyelinase (N-SMase).88) Chart 26 shows the synthesis route. SeO2 oxidation of 2.40 produced the tertiary-allylic alcohol 2.41 as a single stereoisomer. Methanolysis of 2.41 followed by reduction using dissolving metal yielded the dihydroxydimethyl acetal 2.42. Protecting the secondary alcohol as a tert-butyldimethylsilyl (TBS)-ether followed by trimethylsilyl (TMS)-protection of the tertiary alcohol, followed by deprotection of acetal by successive treatment with TMSOTf and water (For deprotection of acetal group, see the next Chapter 3.1.1) produced aldehyde 2.43. Nucleophilic addition of the lithium reagent, formed by the reaction of nBuLi and tributyl(2,4-dimethoxyphenylmethyloxymethyl)stannane (2,4-DMPMOCH2SnBu3) afforded an (R)-alcohol (56%), which was converted to the amino alcohol 2.44 by reaction with diphenylphosphoryl azide (DPPA), PPh3, diethyl azodicarboxylate (DEAD) and LiAlH4 reduction. Amide formation by condensation of 2.44 with unsaturated fatty acid 2.4589) followed by desilylation gave dihydroxyamide 2.46. Stereoselective and chemoselective epoxidation of 2.46 followed by oxidation afforded the epoxy ketone 2.48 via 2.47. Cyclohexanone 2.48 was converted to the conjugated cyclohexenone 2.49. Removal of the 2,4DMPM-ether group in 2.49 resulted in the desired natural product, scyphostatin.
Furthermore, we succeeded in synthesizing more useful synthon for natural product synthesis. That is, optically pure 2.39 (100% ee) was hydroborated with thexylborane regioselectively, and the following oxidation produced the corresponding enone acetal 2.50 through β-bromoketone intermediate and sequential β-elimination of HBr (Chart 27).
The cyclohexene acetal 2.50 was so useful to concisely synthesize many natural products in optically active forms.
Chart 28 shows asymmetric synthesis of (+)-cryptocaryone.90) Diphenylethylene unit of 2.50 was removed by treatment with CAN in CH3CN–H2O.81–83) Cleavage of the acetal group occurred at the same time and successive intramolecular cyclization gave the lactol 2.51. MeOH treatment of 2.51 afforded methoxy compound 2.52. Acylation of 2.52 followed by acidic treatment gave enol keto lactol 2.53. Oxidation of 2.53 finally produced (+)-cryptocaryone.
Enone acetal 2.50 was also applied to asymmetric synthesis of (+)-Sch 64230591) (Chart 29). (+)-Sch 642305 is a potent inhibitor of the bacterial DNA primase, DnaG92) and a potent inhibitor of human immunodeficiency virus type 1 (HIV-1) Tat transactivation.93) Chiral auxiliary served most efficiently in this synthesis. The aldol reaction of 2.50 with the protected-hydroxy-hexanal 2.5494) proceeded stereoselectively to produce a single aldol adduct 2.55. Without acetal ring, this aldol reaction gives a mixture of diastereomers at C6-position. Conversion of 2.55 to xanthate 2.56 and treatment with PhSH gave the β-sulfinyl ketone 2.57, which was reduced to α-alkyl ketone 2.58 by radical reduction. Sequential acid hydrolysis, pyridinium dichromate (PDC) oxidation, and tert-butyldiphenylsilyl (TBDPS)-ether deprotection gave the diketo-carboxylic acid 2.59. Macrolactonization of 2.59 along with β-elimination of PhSH produced the 10-membered lactone 2.60. Finally, mCPBA oxidation and hydrolysis gave the desired natural (+)-Sch 642305. During this synthesis the acetal and the moiety from chiral auxiliary presented from the starting material to just the precursor 2.60 of the final natural product.
Chart 30 shows application to the formal asymmetric synthesis of (−)-stenine.95) Allylation at C6-position of 2.50 took place readily to afford 2.61 in 92% yield. Michael addition of 4-pentenyl magnesium bromide to the enone 2.61 afforded 2.62 in 90% yield. 2,3-Dichloro-5,6-dicyano-p-benzoquinone (DDQ)-mediated hydrolysis96) of the acetal group in 2.62, followed by reductive amination of the resulting aldehyde 2.63 and protection of the amine as a 2-nitrobenzenesulfonyl (Ns) group led to formation of cyclohexanone derivative 2.64 in a total yield of 61% in three steps. Treatment of 2.64 with CAN then furnished the target alcohol 2.65 in 62% yield. Intramolecular Mitsunobu cyclization of 2.65 with diisopropyl azodicarboxylate (DIAD) in 1,4-dioxane gave 2.66 in 55% yield. α-Ethylation of 2.66 gave 80% yield of 2.67, taking into account the recovered starting material. Conversion of 2.67 to ester 2.68 was achieved in a total yield of 57% by a series of reactions involving ozonolysis, PDC oxidation, and esterification with trimethylsilyldiazomethane. Removal of the Ns group in 2.68 followed by formation of the lactam ring gave the desired tricyclic lactam 2.69 in 95% yield. NaBH4 reduction of the ketone moiety in 2.69 afforded lactone 2.70, which has been previously employed by Aubé and colleagues97) and Zhang and colleagues98) as an intermediate in their total syntheses of stenine.
ent-2.50, obtained by converting the intramolecular bromoetherification product of ent-2.37a with (S,S)-hydrobenzion in the same as 2.50, was applied to the asymmetric synthesis of clavolonine99) (Chart 31). Clavolonine is a member of the lycopodine class of alkaloids, one of the subfamilies of the Lycopodium alkaloids. When we reported our synthesis studies in 2011, three total syntheses had already been reported.100) The coupling reaction between ent-2.50 with 3-chloropropyl triflate afforded the chloroalkyl-enone 2.71 stereoselectively. Compound 2.71 was then converted to the azide 2.72 via the Michael adduct. The acetal moiety in 2.72 was hydrolyzed to give aldehyde 2.73. Chemoselective Grignard reaction to the aldehyde and Dess–Martin periodinane oxidation of the resulting diol 2.74 yielded diketo azide 2.75. Octahydroquinoline derivative 2.76 was obtained by chemoselective Staudinger/aza-Wittig cyclization. Direct exposure of 2.76 to methanolic HCl provided the tricyclic amine 2.78 via the intermediacy of compound 2.77. Finally, treatment of 2.78 with HBr in AcOH converted the methoxy group to bromide resulting in intramolecular nucleophilic substitution and saponification to afford clavolonine.
Double haloetherification of acyclic diene acetals101,102): To our surprise, a new reaction of acetal was found. That is, the acyclic diene acetal 2.79a was reacted with 2.5 equivalents (equiv) of N-iodosuccinimide (NIS) in the presence of H2O to give the bicyclic compound 2.80a. Initially, it was thought that hydroxyl aldehyde 2.81 was formed by opening the lactol intermediate 2-iv. However, 2.80a lacked hydroxyl and aldehyde groups, and its NMR spectrum showed a bicyclic structure. The structure of 2.80a, including its stereochemistry, could be clearly determined by X-ray analysis of the LiAlH4-reduced compound of 2.80a. The reaction mechanism is considered as follows. Lactol intermediate 2-iv is formed by the first intramolecular iodoetherification, and the iodonium ion attacks the second olefin because the introduced hydroxyl group and the second olefin are in place, resulting in the second intramolecular iodoetherification to produce the fused bicyclic acetal 2.80a (Chart 32).
(Color figure can be accessed in the online version.)
This new double iodoetherification reaction is general, and a variety of acyclic diene acetals 2.79a–f yielded fused bicyclic acetals 2.80a–f in a diastereoselective manner (Chart 33).
Double haloetherification of acyclic diene acetal has been applied to asymmetric synthesis of rubrenolide and rubrynolide (Chart 34). The fused bicyclic acetal 2.80d obtained in 62% yield by intramolecular iodoetherification reaction of diene-aldehyde acetal 2.79d was converted to the epoxy lactone 2.82 by sequential DDQ-mediated hydrolysis96) and oxidation. Reaction of 2.82 with CAN in CH3CN–H2O (1 : 1)83,84) followed by epoxide formation gave epoxy lactone 2.83. Reaction of 2.83 with 8-nonenyl magnesium bromide, followed by unwinding of the ring of the resulting hydroxyl lactone 2.84a, afforded epoxy lactone 2.85a. Ring opening of the epoxide with water using bismuth(III) trifluoromethanesulfonate [Bi(OTf)3] catalyst gave (+)-rubrenolide. A similar reaction of 2.83 using 8-trimethylsilyl-8-nonynylmagnesium bromide and tetra-n-butylammonium fluoride (TBAF) treatment at the final step afforded the structurally related (+)-rubrynolide.
(Color figure can be accessed in the online version.)
Haloetherification of acyclic diene ketals: Double intramolecular iodoetherification reactions were also observed in acyclic diene ketal system. Treatment of the symmetric and unsymmetric diene-ketals 2.86a,b with I(coll)2PF6 in the presence of H2O afforded the spirocyclic ketals 2.87a,b, respectively. Conversion of 2.87b to the spiroketal 2.88 was achieved by reaction with CAN77,82–84) (Chart 35).
Iodoetherification of this diene ketal was applied to asymmetric synthesis of exogonic acid103) (Chart 36). Nucleophilic substitution of less hindered iodine by cyanide was achieved by treatment of ent-2.87a, prepared from diene ketal derived from (S,S)-hydrobenzoin, with NaCN to give cyanide 2.89 in 65% yield. Radical reduction of the remaining iodine afforded methyl cyanide 2.90, which was converted to carboxylic acid 2.91 by DIBAL-H reduction and subsequent oxidation. Reaction of 2.91 with CAN82–84) in the presence of H2O led to acetal cleavage, removal of diphenylethylene unit, and continuous cyclization to produce exogonic acid.
Although many enzymatic and chemical methods have been developed for enantioselective desymmetrization of meso-diols, no procedure for highly enantioselective desymmetrization of acyclic meso-1,2-diols had been reported at the time.
Incidentally, in the previous chapter, new chiral centers were formed in the olefins of the ene aldehydes by intramolecular haloetherification of chiral ene acetals prepared from racemic ene aldehydes and chiral non-racemic C2-symmetric diol, chiral hydrobenzoin. The reactions proceeded via oxocarbenium ion intermediates, and the realization of high degree of diastereoselectivity was dependent on the large energy difference between the possible transition states that gave the oxocarbenium ion intermediates. This finding suggests that if a large energy difference existed between the possible transition states formed in the intramolecular haloetherification of the ene acetal synthesized from the appropriate chiral non-racemic ene aldehyde and symmetric meso-diol, the reaction should proceed through the least crowded transition state. As a result, two oxygen atoms of meso-diol were thought to de distinguished. (1R,2R,3S,4S)-3-Methyl-5-norbornene-2-carboxaldehyde (2.92) was selected as a non-racemic ene aldehyde for the following four reasons: (1) 2.92 can be easily synthesized by asymmetric Diels–Alder reaction104) followed by LiAlH4 reduction and Swern oxidation; (2) Acetalization would proceed stereoselectively to give cis isomer105); (3) The double bond is fixed in the ring, and the acetal is located inside the skeleton, so that a new chiral center (the *carbon of the product) formed by haloetherification is stereospecifically formed. Most importantly (4) the sterically rigid forms of the oxocarbenium ion intermediates are expected to cause a large energy difference. In other words, comparing Ts-1 leading to exo-isomer and Ts-2 leading to endo-isomer of the two transition states, the repulsion between substituent R and the norbornene skeleton in the endo-isomer is greater than in the exo-isomer. Therefore the reaction is expected to proceed through Ts-1 (Chart 37).
(Color figure can be accessed in the online version.)
We first studied desymmetrization of meso-1,2-diols.106–108) Meso-cyclohexane-1,2-diol 2.93A was used as a model substrate (Chart 38). Acetalization of norbornene aldehyde 2.92a with 2.93A yielded cis-ene-acetal 2.94aA as a single isomer. The intramolecular bromoetherification reaction of 2.94aA in the presence of MeOH gave the mixed acetals 2.95aA with high diastereoselectivity. As expected, the reaction proceeded through more preferable exo-transition state Ts-1 in Chart 37. Debromoetherification reaction of 2.95aA produced the corresponding hydroxyl acetal 2.96aA. It was stable and did not acetalize to form dioxorane 2.94aA. Protection of the hydroxyl group of 2.96aA as a benzyl ether followed by transacetalization of the resulting benzyl ether 2.97aA with one equivalent of meso-diol 2.93A afforded the mono-protected diol 2.98aA in good yield. At the same time, the ene-acetal 2.94aA was regenerated quantitatively.
Chart 39 shows the asymmetric desymmetrization cycle of meso-1,2-diols 2.93. After meso-1,2-diols 2.93 are once converted to ene acetals 2.94a, they are led to acetals 2.97a, and then treated with 2.93 again, to give optically active diol derivatives 2.98a, and at the same time ene acetals 2.94a are regenerated.
(Color figure can be accessed in the online version.)
Many types of cyclic and acyclic meso-1,2-diols are available in this desymmetrization cycle. In each case, intramolecular bromoetherification proceeded with high diastereoselectivity to give almost single isomer. The high chemoselectivity of these reactions is also noteworthy because the alkene or alkadiene in the substrates remained intact. A significant feature of our desymmetrization method allowed the use of many types of protecting groups of hydroxyl group such as benzyl (Bn), TBDPS, benzoyl- (PhCO), or p-methoxybenzyl (PMB). Figure 5 shows the structures and % ee values of the meso-1,2-diol derivatives we have prepared using this approach.
This desymmetrization method was next applied to meso-1,3-diols. However, in the reaction using (1R,2R,3S,4S)-3-methyl-5-norbornene-2-carboxaldehyde 2.92a, good diastereoselectivity could not obtained by the intramolecular bromoetherification reaction. Therefore we thought that other norbornene aldehyde derivatives with larger substituent were needed to desymmetrize 1,3-diols. To that end, we designed a kinetic optical resolution of norbornene aldehyde derivatives using chiral non-racemic hydrobenzoin as shown in Chart 40. Acetalization of racemic norbornene aldehyde derivative (±)-2.92 with chiral non-racemic hydrobenzoin, (S,S)-hydrobenzoin, gave the inseparable diastereoisomeric ene acetals 2.99 and 2.100. Treatment of these materials with 0.5 equiv of NBS in the presence of H2O (5 equiv) provided the hydroxyl-aldehyde 2.101 from 2.99 and unreacted 2.100, which were easily separated by SiO2 column chromatography. The debromoetherification of 2.101 and acid catalyzed hydrolysis of 2.100 resulted in optically pure norbornene-aldehydes 2.92b–d, respectively.109,110)
As noted above, chiral non-racemic norbornene aldehyde derivatives with bulky substituent were available, and we again tried the desymmetrization of meso-1,3- and meso-1,4-diols. After several studies, 3-endo-phenyl-4-exo-methyl-norbornene aldehyde 2.92c proved to be the best reagent for the desymmetrization of meso-1,3- and 1,4-diols111) (Chart 41). Acetalization of (2R)-2.92c with the meso-diol 2.93 afforded the single ene-acetal isomer 2.94c. Intramolecular bromoetherification of 2.94c in the presence of MeOH proceeded to give the mixed acetal in a highly stereoselective manner. For meso-1,2-diol, the next reaction was debromoetherification. However, in this case, the hydroxyl acetal such 2.96a in Chart 39 obtained by debromoetherification of the mixed acetal was unstable because there was no repulsion of the two R groups, and 2.94c was regenerated by acetal cyclization. Therefore hydroxyl aldehyde 2.102 was obtained by intramolecular bromoetherification and continuous acid hydrolysis. Protection of the hydroxyl group in 2.102 gave 2.103, which was debromoetherified to afford optically active meso-diol derivative 2.98c along with norbornene aldehyde 2.92c. Figure 6 shows the structures and % ee of meso-1,3- and 1,4-diol derivatives desymmetrized using this method.
(Color figure can be accessed in the online version.)
Pyridinium salts are excellent electrophiles and commonly used as substituted pyridine derivatives after nucleophilic addition reactions. On the other hand, phosphonium salts are well known very useful reagents in Wittig reaction and so on.
Meanwhile, we have discovered completely new synthetic aspects of pyridinium and phosphonium salts prepared from acetals. Both salts acted as leaving groups and nucleophilic substitutions proceeded in high yields. Phosphonium salts also acted as in situ protecting groups, allowing the reversal and control of the reactivity of two carbonyl groups.
3.1. Organic Reactions Using Pyridinium Salt IntermediatesBackground of the research and overview112): Acetals, including ketals, are recognized as good protecting groups of carbonyl groups. Deprotection of acetals to form the parent carbonyls usually occurs under acidic conditions. The rate of deprotection of acetals depends on the stabilities of the oxocarbenium ion intermediates. Thus the acetals, which give more stable intermediates, are deprotected much faster. In this regard, deprotection of ketals from ketones, which produce more stable intermediates, proceeds much faster than deprotection of acetals from aldehydes (Chart 42). This concept is general knowledge in organic chemistry. Indeed, although many methods are available for deprotecting ketals in the presence of acetals, prior to our method, no method has been reported that allows the deprotection of acetals in the presence of ketals.41–43,113,114)
(Color figure can be accessed in the online version.)
We found that treatment of acetals with trialkylsilyl trifluoromethanesulfonate (R3SiOTf)-pyridine base such as collidine gave the parent carbonyl groups in high yields via pyridinium-type salt intermediates such as collidinium salt. The characteristic points of our method are as follows: 1) The reaction is unprecedented in that acetals from aldehydes can be deprotected in the presence of ketal, and 2) The reaction proceeds under weakly basic conditions (Chart 43).
(Color figure can be accessed in the online version.)
Our method has the unique features described above that are not available in conventional methods. Therefore many groups have used our method in their synthetic chemistry. For example, Iwabuchi and colleagues115) succeeded in selective deprotection of acetal in the presence of ketal using triethylsilyl trifluoromethanesulfonate (TESOTf) and 2,6-lutidine combination to achieve total synthesis of (−)-CP55940. Inoue and colleagues116) often used our method to synthesize acid-labile leukotriene derivatives, having conjugated dienes, 1,4-dienes, and silyl ethers of allyl alcohols in addition to acetal. Other groups have also used our method to deprotect acetal in the presence of acid-labile functional groups.117–121)
Pyridinium-type salt intermediates are cationic electrophilic species. Because the acetal substitution reactions proceeded under weakly basic conditions, it was possible to use substrates and reagents with functional groups that were unstable to acids and bases, greatly expanding acetal chemistry.
In addition, deprotection of acetal-type protecting groups of hydroxyl groups such as tetrahydropyranyl (THP)-, trimethylsilylethoxymethyl (SEM)-, benzyloxymethyl (BOM)-, methoxyethoxymethyl (MEM)-, methoxymethyl (MOM)-ether, and methylene acetal was also successful. In these cases, 2,4,6-collidine was effective as a base for deprotection of THP-ether, whereas other ethers needed to use 2,2′-bipyridyl instead of 2,4,6-collidine. The most notable aspect of our method is that the order of deprotection is reversed from the conventional methods, as shown in Fig. 7.
(Color figure can be accessed in the online version.)
Deprotection of acetals122,123): In the schyphostatin synthesis in Chart 26, we first looked at gradual conversion. That is, hydroxy dimethylacetal 3.1 prepared by tert-butylsilylation of 2.42 in Chart 26 was treated with TESOTf and 2,6-lutidine to try to obtain triethylsilylated dimethylacetal 3.3. Surprisingly, however, the reaction at 0 °C for 30 min yielded triethylsilylated aldehyde 3.2 in 82% yield (Chart 44).
To confirm the repeatability of this deprotection reaction, we next examined the reaction of model compound, dimethyl acetal 3.4 having tert-hydroxyl group under the same reaction conditions just changing TESOTf (Table 2). The reaction using 1.2 equiv of TMSOTf gives silylated acetal 3.6 (R = TMS) as a major product with a small amount of silylated aldehyde 3.5 (R = TMS) (entry 1). The use of 2.0 equiv of TMSOTf or TESOTf afforded the corresponding silylated aldehydes 3.5 (R = TMS, TES) in excellent yields (entries 2, 3). However, tert-butyldimethylsilyl trifluoromethanesulfonate (TBSOTf), trimethylsilyl chloride (TMSCl), and triethylsilyl chloride (TESCl) did not give the aldehyde at all even though 4.0 equiv of reagents was used (entries 4–6). It is noteworthy that trifluoromethanesulfonic acid (TfOH) caused no reaction (entry 7), although TfOH is formed by the reaction of TMSOTf or TESOTf and the substrate with hydroxyl group, 3.4 (entries 1–4).
 |
We then treated acetal 3.7a, which has only an acetal function, with 2.0 equiv of TMSOTf (or TESOTf) and 3.0 equiv of 2,6-lutidine. As a result, the deprotected parent aldehyde 3.8a was obtained in good yield in each reaction (Chart 45).
To clarify the reaction mechanism, 1H-NMR experiments of the reaction using TESOTf were performed (Fig. 8). The upper chart shows 1H-NMR spectrum of 2,6-collidine, the second graph is the spectrum of acetal 3.7a. The third is the spectrum of the reaction mixture (A). The fourth is the spectrum of the crude product after H2O work-up (B). It is noteworthy that the acetal proton (4.2 ppm) of 3.7a moved to around 6.0 ppm, strongly suggesting N,O-acetal structure. Then, aldehyde proton (9.75 ppm) was generated by post-treatment of H2O.
(Color figure can be accessed in the online version.)
The reaction behavior was also tracked by TLC. As a result, when the acetal 3.7a was treated with TESOTf and 2,6-lutidine, 3.7a was first converted to a highly polar compound rather than aldehyde 3.8a, and post-treatment with neutral H2O gave the product, aldehyde 3.8a.
From 1H-NMR study of the reaction and TLC check, the reaction mechanism can be considered as follows (Chart 46). The 2,6-lutidinium salt 3-i was first formed by reaction of 3.7a and 2,6-lutidinium triethylsilylate, in situ prepared from TESOTf and 2,6-lutidine.124) Indeed, mass spectroscopy of the reaction mixture showed the peak for salt 3-i (m/z 278.25). Next, addition of neutral water resulted in nucleophilic substitution, yielding hemiacetal 3-ii, which was spontaneously converted to aldehyde 3.8a.
(Color figure can be accessed in the online version.)
Chart 47 shows the reactions of various acetals with TMSOTf (or TESOTf)–2,6-lutidine followed by H2O work-up. Deprotection of acetal 3.7a and dioxolane 3.7b proceeded without any problem, and the corresponding aldehyde was obtained regardless of silyltriflate. However, the reaction of dioxane 3.7c showed different results. TMSOTf gave the aldehyde whereas TESOTf gave the enol ether 3.9. Reactions of dioxolane 3.7d and dimethylacetals 3.7e–i with various functional groups, including those that are acid labile, also gave good results. The reaction of the ketal 3.7j gave a very interesting finding. The dioxolane 3.7b from the aldehyde gave the aldehyde irrespective of silyltriflate, whereas the dioxolane 3.7j from ketone behaved differently depending on the silyltriflate. That is, reaction with TMSOTf resulted in deprotection, and use of TESOTf did not.
Deprotection of acetals in the presence of ketals: Ketals are deprotected faster than acetals.113) It is common sense in organic chemistry, and many textbooks on organic chemistry explain so.114) In fact, there were no reports of deprotection of acetals in the presence of ketals before we first reported this finding.122,123) The results of acetal 3.7b and ketal 3.7j in Chart 47 indicated that the TESOTf–2,6-lutidine combination could reverse the reactivity, ie, deprotect the acetal in the presence of ketal. To address this issue, we developed an unprecedented deprotection reaction in which acetal from aldehyde was deprotected faster than ketal (Chart 48). Thus treatment of a 1 : 1 mixture of dioxolanes 3.10 and 3.7j from aldehyde and ketone with TESOTf and 2,6-luitidine gave the intact ketal 3.7j and aldehyde 3.11 in 82%.
Characteristics of our method were demonstrated by comparison with conventional methods (Chart 49). Deprotection of compound 3.12, which has an acetal and a ketal in the same molecule, may yield three carbonyl compounds, aldehyde acetal 3.13, keto acetal 3.14, and keto aldehyde 3.15. Our method yielded only 3.13, in which the acetal was deprotected in the presence of a ketal. On the other hand, in the usual methods, p-TsOH treatment in wet acetone or trimerthylsilyl iodide (TMSI) treatment in CH2Cl2, 3.14 was obtained as a main product by deprotection of ketal in the presence of an acetal, and 3.13 was not formed. These results show that our method is completely different from the usual methods.
Incidentally, it transpired that the choice of base was very important for the success of this deprotection. Chart 50 shows the effect of various pyridine bases on the deprotection of dimethyl acetal 3.7a. All pyridine bases such as pyridine, 2-picoline, 2,4-lutidine, 2,6-lutidine, and 2,4,6-collidine gave their corresponding highly polar salts, but with neutral water the subsequent hydrolysis proceeded only with salts from 2,6-lutidine and 2,4,6-collidine, yielding aldehyde 3.8a in high amounts. On the other hand, other salts from pyridine, 2-picoline, and 2,4-lutidine required aq. 0.1 N HCl solution to obtain aldehyde 3.8a. These observations implied that adequate repulsion between the nitrogen and carbon atoms in the salt was essential for deprotection of acetals, and pyridine bases with bulky substituent at 2- and 6-positions gave good results. The use of base was essential in this reaction, because it did not result in a complex mixture. It is noteworthy that 4-dimethylaminopyridine did not cause a reaction. 2,4,6-Collidine was better than 2,6-lutidine between 2,6-lutidine and 2,4,6-collidine, probably because of the higher electron-donating ability of 2,4,6-collidine.
Chart 51 shows the structures of the starting substrates with an acetal and a ketal group in the same molecule and the yields of the products obtained by chemoselective deprotection using TESOTf–2,6-lutidine or 2,4,6-collidine combination. In each substrate an acetal was selectively deprotected in the presence of a ketal. In cases of substrates with hydroxyl group triethylsilylated aldehydes were obtained in high yields.
A discussion of the reaction mechanism in Chart 46 indicated that during deprotection process, the pyridinium salt reacted with nucleophile, H2O, first to generate the substituted product, hemiacetal. In other words, the pyridinium salts were considered electrophilic species. Next, we investigated the nucleophilic substitution of the pyridinium salt intermediates by various types of nucleophiles.
Reaction with hetero nucleophiles125): Traditional nucleophilic substitution reactions of acetals are carried out under acidic conditions. Therefore substrates and reagents with acid-labile functional groups cannot be used. On the other hand, our method was performed under weakly basic conditions that could leave acid-labile functional groups such as epoxides, silyl ethers, trityl (Tr) ethers, and tert-butyl esters intact. Our mild reaction conditions should also allow the use of base-labile compounds. Next, we examined the use of hetero nucleophiles containing functional groups that were unstable to acids and bases.
Chart 52 shows the alcohols used and the product yields obtained for the nucleophilic substitution reaction of various alcohols on the acetal 3.7a. All reactions proceeded without problems and the corresponding O,O-mixed acetals 3.16 were obtained in high yields. Alcohols such as allyl alcohol, propargyl alcohol, and benzyl alcohol tend to generate stable cation species by SN1-type elimination of the hydroxyl group under acidic conditions, so reaction conditions are important, but our reaction system gave the products in good yields without any problem. Bulky alcohols, such as secondary and tertiary alcohols, which are usually difficult to react with acetals, also produced the mixed acetals in high yields. Since the reaction conditions are very mild, alcohols with acid-labile functional groups such as an epoxide, dimethyl acetal, and a sugar unit, etc. also gave the desired products in sufficient yields.
This mixed acetal formation reaction could be applied to various symmetric acetals. Chart 53 shows the product structures and the yields. Nucleophiles such as allyl alcohol, lithium phenylsulfide and sodium azide provided the corresponding O,O-mixed acetals, O,S- and N,O-acetals in high yields.
Incidentally, homo acetals are often obtained as by-products due to excessive reaction in a mixed acetals synthesis method by transacetalization of alcohol under acidic conditions. On the other hand, in our method, the starting homo acetals are first converted to the salt intermediates then the addition of alcohol causes nucleophilic substitution, resulting in mixed acetals. Therefore there is no overacetalization in our method. Furthermore, our method proceeds under weakly basic conditions and many acid-labile functional groups are unaffected. Figure 9 shows the mixed acetals and their yields obtained by our and conventional methods. The upper values are the yields from our method. The lower ones are the yields by conventional methods: the mixed acetal 3.17 from dimethyl acetal and geraniol (H3PO4, HCO2H, xylene, 140 °C)126); the mixed acetal 3.18 from diethyl acetal and 2,2,2-trichloroethanol (p-TsOH·H2O, benzene, 81 °C)127); the mixed acetal 3.19 from dimethyl acetal and ethanol (concd HCl aq, CHCl3, room temperature (r.t.)).128) In each case, our results are better than the reported results.
As described above, the pyridinium-type salt intermediates are cationic electrophilic species and the reactions proceed under weakly basic conditions. Furthermore, nucleophiles can be added after the formation of salt intermediates. Therefore many compounds with acid- and base-labile functional groups are available. On the other hand, the usual methods of acetal substitution reactions proceed via oxocarbenium ion intermediates under acidic conditions. The oxocarbenium ions are so unstable that it is necessary to add a nucleophile from the beginning of the reactions. Therefore compounds with acid-labile functional groups cannot be used. Based on these observations, our method promised a new development of nucleophilic substitution reactions of acetals. So we next looked at the use of various types of carbon nucleophiles.
Reaction with carbon nucleophiles: Chart 54 shows the results using Gilman reagents as nucleophiles.129) The reactions proceeded in good-to-high yields and the corresponding products were obtained via collidinium salt intermediates. As expected, acid-labile functional groups such as trityl ether and silyl ether and base-labile groups such as acetate were not affected.
The method is characterized by a high chemoselectivity that could not be achieved by previously reported methods. Thus the reaction of compound 3.20 with acetal and ketal by our method afforded only ketal 3.21 (R = TES), not acetal 3.22 (R = TES) and diether 3.23 (R = TES). On the other hand, in other methods, reaction with Gilman reagent130) or with a manganese ate-type reagent131) in the presence of BF3·Et2O did not yield 3.21 (R = H) at all, but gave the mixture of 3.22 (R = H) and 3.23 (R = H) (Chart 55).
Table 3 shows the reactions of various carbon nucleophiles with collidinium salt intermediates.132) Many aldol-type reactions between acetals with silyl enol ethers or silyl ketene acetals have been reported to proceed under acidic conditions, but they are limited to simple substrates that can withstand acidic conditions.133) On the other hand, our method was very mild and the substrates with acid-labile functional groups were also available. Enamines also gave the corresponding products in high yields. Isocyanoacetoamide having an amide moiety in the same molecule reacted with the salt intermediates directly to provide the oxazoles in high yields by substitution and spontaneous cyclization. This result is the first example of the reaction of acetals and isocyanoacetoamide. No significant effects on the co-existing functional groups were observed, even the acid-sensitive ones. When 1-isocyano-2-methylbenzene was used as a simple isocyanide, corresponding amines were obtained in moderate yields by in situ LiBH4 reduction of isocyanide adducts. The substitution reaction of the collidinium salt intermediates with trimethylsilyl cyanide (TMSCN) proceeded in moderate yields, and the 2,2′-bipyridinium salt intermediates prepared from 2,2′-bipyridyl instead of 2,4,6-collidine gave the corresponding products in high yields.
 |
Chemoselective substitutions was also achieved (Table 4). The reaction of the substrate 3.24 with acetal and ketal groups with carbon nucleophiles proceeded acetal selectively to give the products 3.25.
 |
The order of deprotection of acetals from aldehydes by conventional methods depends on the stability of oxocarbenium ion intermediates, whose formations are the rate-determining steps (Chart 56). Therefore deprotection of acetals having electron-donating alkyl groups at the α-position proceeds faster than unsubstituted ones.113,135) Moreover, in the case of different acetals, deprotection of the bulkier alkoxy acetals proceed faster than smaller alkoxy or cyclic ones.136–138)
On the other hand, as shown in the former chapter our method relies on the bulk of acetals and ketals, and less bulky acetal can be deprotected faster. We next looked at 1) deprotection of the same acetals with different substitution patterns, 2) deprotection of different types of acetals, and 3) synthesis of single ethers from mixed acetals.
Deprotection of same acetals with different substitution patterns: We used dimethyl acetal 3.26 as the unsubstituted, dimethyl acetal 3.27 as the monosubstituted, and dimethyl acetal 3.28 as the disubstituted group as model compounds for investigation of the effect of α-substituent of an acetal group. A 1 : 1 mixture of 3.26 and 3.27 or 3.26 and 3.28 was treated with TMSOTf (or TESOTf)–2,4,6-collidine followed by BnOH addition. As a result, the combination of 3.26 and 3.28 using TESOTf showed the highest chemoselectivity producing a mixed acetal 3.29 from 3.26 and recovering 3.28. 3.30 was not obtained. In addition, the combination of 3.26 and 3.27 gave the mixture of 3.29 and 3.31 even with TESOTf (Chart 57).
Next, chemoselective deprotection of α-unsubstituted acetals in the presence of α-substituted or aromatic acetals was examined (Chart 58). Our method was successfully applied to both cyclic acetals, dioxolane 3.32 and dioxane 3.33. When the α-substituent was an isopropyl group, one substituent was sufficient. Furthermore, the present method was useful for the selective deprotection of the substrate 3.35 having an aliphatic acetal and an aromatic one, and the product in which the aliphatic acetal was deprotected was obtained in good yield.
Deprotection of the different types of acetals: Remarkable reversal of the reactivity of acetal deprotection by our method was more pronounced in the reaction of different types of acetals. After treating a 1 : 1 mixture of dimethyl acetal 3.26 and diethyl acetal 3.36a with TESOTf and 2,4,6-collidine followed by BnOH addition, mixed acetal 3.29 (92%) derived from dimethyl acetal 3.26, mixed acetal 3.37a (33%) from diethyl acetal 3.36a, and the recovered 3.36a (6%) were obtained. On the other hand, when the diisopropyl acetal 3.36b was used instead of 3.36a, a highly selective reaction to dimethyl acetal 3.26 was observed, producing mixed acetal 3.29 in quantitative yield and quantitative recovery of diisopropyl acetal 3.36b (Chart 59).
Next, deprotection of the substrate 3.38 containing both dimethyl acetal and a diisopropyl one and the substrate 3.40 having both cyclic acetal and diisopropyl acetal was examined (Chart 60). Diisopropyl acetal unit is very unstable under acidic conditions.139) However, our method did not affect it and ω-alkanoyl diisopropyl acetal 3.39 was selectively obtained with 77% yield. Further treatment of 3.40 with TESOTf and 2,4,6-collidine gave 3.39 in 76% yield with full selectivity, leaving the diisopropyl acetal unit intact.
Formation of single ethers from mixed acetals: Since our method was able to identify the bulkiness of the oxygen atoms as described above, we then applied it to mixed acetals consisting of two oxygen atoms with different pendant groups. As a result, selectively reductive alkyl ether synthesis from mixed acetals was achieved.
Applying the usual method using TMSOTf and triethylsilane (Et3SiH),140) which is used for the conversion of acetals to ethers, on mixed acetal 3.41a yields two ethers 3.42a and 3.42a′ (Table 5, entry 1). On the other hand, after generating a collidinium salt by our method, TESOTf–2,4,6-collidine, NaBH4 was added to reduce the salt giving the single ether 3.42a in 74% (entry 2). In this case, THF was used as reaction solvent for NaBH4 reduction. Even when TMSOTf was used instead of TESOTf, single 3.42a was obtained with a better 80% yield (entry 3). Reducing the reaction temperature increased the yield of 3.42a (entry 4).
 |
The above results show the complex from R3SiOTf–2,4,6-collidine can discriminate between two oxygen atoms and the less bulky oxygen atom is replaced by 2,4,6-collidine (Chart 61). This result indicates that the complex from R3SiOTf–2,4,6-collidine has a very high ability to discern subtle differences in the circumstances around the oxygen atom. TMSOTf, on the other hand, is not sufficient to discriminate between two oxygen atoms in the mixed acetal, as shown in Table 5, entry 1.
(Color figure can be accessed in the online version.)
Chart 62 illustrates the generality of this selective reduction. Good-to-excellent yields were obtained in the presence of many functional groups including both base and acid labile ones, such as acetoxy (OAc), benzoyloxy (OBz), tert-butyldimethylsilyloxy (OTBS), trityloxy (OTr), halogens, and methyl ester (3.41b–3.41h). In addition, this method could be applied to sterically dense substrates 3.41i–3.41k, using 2-methoxypyridine instead of 2,4,6-collidine. These results indicate that our method is useful for the synthesis of various types of alkyl ethers.
Protecting groups of hydroxyl functional groups such as THP-, MOM-, SEM-, and BOM-ethers, and methylene acetal, which is diol protecting group, also have an acetal structure. Although these two oxygen atoms are different from each other, the complex of TESOTf and 2,4,6-collidine distinguishes two oxygen atoms with high discrimination ability as described above. We next examined the deprotection of acetal-type protecting groups for hydroxyl groups.
Deprotection of THP-ethers141,142): The two oxygen atoms of THP-ether are different from each other. That is, one is the oxygen atom in the ring and the other is the oxygen atom of the hydroxyl group. However, as expected, the reaction proceeded highly chemoselectively to give a single collidinium salt intermediate and H2O work-up provided the deprotected product in good yield. Since the reaction conditions were weakly basic, the reaction could proceed without affecting many functional groups including acid-labile ones such as TBS- and Tr-ethers. The utility of this selective deprotection of THP-ether in the presence of other functional groups could be verified by the reaction of substrate 3.43 having 4-methoxyphenylmethyl- (MPM-), TBS-, and allyl TES-ethers in addition to THP-ether. That is, treatment of 3.43 with TESOTf–2,4,6,-collidine in CH2Cl2 at 0 °C for 30 min then H2O work-up gave THP-deprotected alcohol 3.44 in 71% without affecting other functional groups (Chart 63).
Deprotection of MOM-, MEM-, BOM-, and SEM-ethers143,144): The two oxygen atoms of MOM-ether are very similar to each other. The only difference between the two oxygen atoms is methyl and methylene groups. However, under our conditions two atoms could be distinguished and a single salt intermediate was obtained. However, it was stable and could not be converted to aldehyde by neutral H2O work-up (Chart 64, upper equation). The difference between the salt from THP-ether, which was converted to alcohol by H2O work-up (see Chart 63), and the salt from MOM-ether is the presence or absence of alkyl substitution on the acetal carbon. This means that proper repulsion is required to convert salt to an alcohol via hemi acetal. But the acetal carbon of the salt from MOM-ether has no alkyl substituent. The next study of various pyridine bases that could be more easily removed from the salt intermediate showed that 2,2′-bipyridyl gave the best result (Chart 64, reaction formula below).
The generality of this deprotection reaction is shown in Chart 65, in which the structures of starting MOM-ethers and the product yields are described. Both TMSOTf and TESOTf worked well as silyl triflates, and the corresponding alcohols were obtained in high yields from primary, secondary, and tertiary MOM-ethers. Other protecting groups for hydroxyl groups such as acetate, benzoate, benzyl ether, acid labile TBS- and Tr-ethers, and other functional groups such as olefin, tert-butyl ester, methyl ester, bromine, and amide were not affected during the reaction. It is noteworthy that aromatic MOM-ether, which is easily deprotected under normal acidic conditions, produced small amounts of the deprotected product in the case of TMSOTf and no deprotected product in the case of TESOTf.
The difference between our method and conventional methods was evident from the reaction of substrate 3.45, which contains aliphatic and aromatic MOM-ethers in the same molecule (Chart 66). When 3.45 was treated with TESOTf–2,2′-bipyridyl followed by H2O work-up, aromatic MOM alcohol 3.46 was obtained in 85% yield, but aliphatic MOM phenol 3.47 was obtained in 76% yield in the conventional method.145)
The combination of TMSOTf–2,2′-bipyridyl was also useful for the deprotection of other acetal-type protecting groups such as MEM-, BOM-, and SEM-ethers, and the corresponding alcohols were obtained in high yields. Chart 67 shows the structures of starting ethers and the product yields. In these cases, acid-labile functional groups such as Tr-ether also survived during reaction.
Acetal-type protecting groups such as MOM-, MEM-, BOM-, and SEM-ethers have two oxygen atoms in different environments. Chart 68 shows the reaction mechanism estimated from several studies. TMSOTf–2,2′-bipyridyl complex (Bipy-TMS+OTf−) first activated one of acetal oxygen atoms, then 2,2′-bipyridyl attacked the acetal carbon to give the pyridinium intermediate A, which was reacted with H2O to form the deprotected alcohol via hemiacetal (path a), or TMS-ether B, which was hydrolized by work-up (H2O) to give the deprotected alcohol (path b). In the case of the small protecting groups such as MOM- or MEM-ethers, the reaction proceeded mainly through path a. Although BOM-ether was bulky protecting group, its reaction also proceeded preferentially through path a. However, in the case of SEM-ethers the reactions were found to proceed through approximately the same rate in the two reaction paths.
(Color figure can be accessed in the online version.)
Incidentally, aromatic MOM-ethers could not be deprotected under the above reaction conditions as shown in Chart 65. However, quite recently we found a remarkable effect of CH3CN as reaction solvent. In other words, reaction in CH3CN instead of CH2Cl2 enabled smooth conversion of the aromatic MOM-ethers to the corresponding aromatic alcohols.146) Chart 69 shows the structures of starting MOM-ethers and the product yields. With aromatic MOM-ethers having electron donating group such as methyl and methoxy group, the reaction proceeded without any problem, and the corresponding products were obtained in high yields. However, substrate with electron withdrawing nitro group required longer reaction time and heating, albeit with higher product yield. This reaction proceeded satisfactorily even with a substrate having a sterically hindered MOM group or a substrate having an ester. MOM-ether having acid-labile Tr-group was also deprotected without affecting it. This result indicates that the reaction was very mild and non-acidic. In addition, the method was applicable to naphthol MOM-ether. The intermediates leading to aromatic alcohols were TMS-ethers, not 2,2′-bipyridinium salts like aliphatic MOM-ethers, via path b (R = Ar) in Chart 68.
Representative protecting groups for 1,2- and 1,3-diols are acetonide, benzylidene acetal, and methylene acetal.41–43) Among them acetonide and benzylidene acetal are stable from strong basic to neutral conditions, whereas methylene acetal is stable from strong basic to medium acidic conditions. Methylene acetal is therefore considered the best protecting group for diols, but its deprotection requires strict acid conditions. On the other hand, acetonide and benzylidene acetal are deprotected under mild acidic conditions. Benzylidene acetal is also deprotected by hydrogenolysis. Therefore acetonide and benzylidene acetal are commonly used as protecting groups for diols. However, if methylene acetal can be deprotected under mild conditions, methylene acetal is the best protecting group of diols.
Deprotection of methylene acetals 3.48 with trialkylsilyltriflate–2,2′-bipyridyl was investigated. After salt intermediates were easily formed, different types of products were obtained depending on the subsequent work-up. That is, when the salts derived from TMSOTf–2,2′-bipyridyl were post-treated with 1N HCl aq., the diols 3.49 were obtained in high yields (Chart 70). 1N HCl aq. was a weak acid, and acid-labile functional groups such as TBS- and Tr-ethers were not affected. On the other hand, use of TESOTf instead of TMSOTf and K2CO3 aq. work-up afforded hydroxyl TES-ethers 3.50.
The above reactions proceeded via trialkylsilylated 2,2′-bipyridinium salt intermediates 3-iii (Chart 71). Silylation occurred at the less sterically hindered oxygen atom, and substitution reaction with H2O gave the hemiacetals 3-iv, which, depending on the trialkylsilyl triflate used, were spontaneously converted to the corresponding products, 3.49 or 3.50.
We next envisioned that MOM-ethers would be obtained by reaction of the salts with MeOH. Reaction of 3.48 with TMSOTf–2,2′-bipyridyl followed by MeOH addition gave the hydroxyl MOM-ethers 3.51 in high yields, as expected. On the other hand, when the reaction using TESOTf–2,2′-bipyridyl was followed by treatment with MeOH–Et3N, triethylsilylated MOM-ethers 3.52 were obtained in high yields. In the case of methylene acetals 3.48a–g from asymmetrical diols, silylation occurred with the less sterically hindered alcohol, and the MOM-ether was formed with the more sterically hindered alcohol (Chart 72).
As mentioned above, the rate of deprotection of acetal-type protecting groups by conventional methods depends on the stability of the oxocarbenium ion intermediates. Ketal-type protecting groups (acetonide, cycloalkylidene ketals) are deprotected much faster than acetal-type protecting groups (MOM-ether, methylene acetal). That is, the order of deprotection is cycloalkylidene ketals, acetonides > MOM-ethers > methylene acetals.41–43) Indeed, when compounds 3.53a with acetonide and methylene acetal or 3.55a with cyclohexylidene ketal and a MOM-ether were treated with aq. CF3COOH, ketal-type protecting groups were deprotected, and methylene diol 3.54 and MOM-ether diol 3.56 were obtained in good yields (Chart 73).
On the other hand, in our method the complex from R3SiOTf and pyridine base can identify the steric bulkiness. Since ketal-type protecting groups are bulkier than acetal-type protecting groups, our method was thought to be able to deprotect acetal-type protecting groups selectively in the presence of ketal-type protecting groups.
The reaction of 3.53b, having cyclohexylidene ketal and methylene acetal in the same molecule, with TESOTf and 2,2′-bipyridyl was first investigated (Chart 74). As a result, a small amount of the desired product 3.57a, obtained by deprotection of methylene acetal unit, and major unexpected enol ether 3.57a′ were obtained. A plausible reaction mechanism for 3.57a′ is shown in parenthesis. Therefore the bulk of the complex (Bipy-TES+OTf−) from TESOTf was insufficient, and non-selective coordination to the oxygen atom of the cyclohexylidene ketal followed by β-elimination of hydrogen atom by 2,2′-bipyridyl afforded 3.57a′.
(Color figure can be accessed in the online version.)
We next used bulky TBSOTf instead of TESOTf. As a result, deprotection reactions of methylene acetal unit proceeded preferentially in compounds containing methylene acetal and ketal-type protecting groups such as acetonide (3.53a), cyclohexylidene acetal (3.53b), and cyclopentylidene acetal (3.53c) to give the products 3.58a–c in high yields (Chart 75). It is noteworthy that 3.58a was obtained in high yield even from 3.53a with acetonide, the smallest ketal-type protecting group of 1,2-diol. This indicates our method has good ability to distinguish the three-dimensional environments of protecting groups.
The method using TBSOTf could be employed in a diversity array. That is, the intermediate, 2,2′-bipyridinium salt, prepared by the reaction of 3.53a with the combination of TBSOTf–2,2′-bipyridyl combination could be converted to various compounds, mono-TBS protected diol 3.58a, MOM and TBS protected diol 3.58d, diol 3.58e, and mono-MOM protected diol 3.58f in high yields by changing the conditions of subsequent processing (Chart 76).
The TBSOTf–2,2′-bipyridyl deprotection method described above was also applied to selective deprotection of MOM-ethers in the presence ketals (Chart 77). Deprotection of MOM-ethers occurred in high yields without adversely affecting the cyclohexylidene (3.55a), cyclopentylidene (3.55b), and acetonide (3.55c) ketal groups to give the ketal alcohols 3.59a–c. Selective deprotection also proceeded with the bulky sec- and tert-MOM-ethers (3.55d,e), giving the corresponding alcohols 3.59d and 3.59e in 93 and 91% yields, respectively.
Background of the research and overview of in situ protection methodology: Carbonyl groups such as aldehydes and ketones are the most important functional groups in organic chemistry and their reactivity is well known. The order of the reactivity of carbonyl groups toward nucleophiles is generally aldehyde > ketone > ester > amide and nitrile150) (Fig. 10). Therefore various chemoselective reactions of highly reactive carbonyl groups are possible in the presence of less reactive ones. On the other hand, generally a three-step sequence of protection/reaction/deprotection is needed to transform less reactive carbonyl groups in the presence of more reactive ones. Figure 11 shows some ketone-selective reactions of keto-aldehydes using conventional methods. If it is possible to reverse the reactivity of the carbonyl groups, one-step chemoselective conversion of the less reactive carbonyl groups can be performed in the presence of more reactive groups. In situ protection method, in which the more reactive carbonyl groups are protected during the transformation of the less reactive carbonyl groups then reformed during workup, is a powerful tool for these transformations.151) However, they have several drawbacks such as narrow substrate coverage, the need for strict stoichiometric control, and the use of expensive reagents. Therefore more practical and facile methods are strongly desirable.
As an extension of pyridinium salt chemistry, we discovered the interesting reactivity that α-alkoxyphosphonium salts formed by reaction of acetals with trialkylsilyl triflate and trisubsituted phosphines depends on the trisubsituted phosphine used. That is, the α-alkoxyphosphonium salts from (o-tol)3P reacted with nucleophiles to afford the substituted products in high yields. In contrast, the salts from PPh3 were stable and the next substitution reaction was very slow (Chart 78).
Therefore we next examined an in situ protection method for the carbonyl groups using phosphonium salts, which were stable against nucleophilic substitution. As a result, we developed two very practical in situ protection methods, 1) Reverse of reactivity and 2) Control of reactivity151,152) (Fig. 12).
(Color figure can be accessed in the online version.)
The reactivity of α-alkoxyphosphonium salts formed by reaction of acetal 3.26 with TESOTf and phosphine was first examined (Table 6). The reactivity of these salts was determined by the balance between steric and electronic factors, respectively. That is, the phosphonium salt from (o-tol)3P occurred the next nucleophilic substitution with H2O smoothly to give aldehyde 3.60 in high yield via hemi acetal. On the other hand, the phosphonium salt from PPh3 was stable, the next hydrolysis was very slow, and the yield of aldehyde 3.60 was only 47% even after 96 h. In addition, the phosphonium salts from (p-tol)3P, (m-tol)3P, and (o-MeOPh)3P were very stable and no hydrolysis occurred at all after 24 h.
 |
Chart 79 summarizes the reactions of α-alkoxyphosphonium salts derived from (o-tol)3P with nucleophiles. The reactions with H2O, RMgBr, TMSCN, and PhLi all proceeded in high yields, and the corresponding substituted products were obtained.
(Color figure can be accessed in the online version.)
Selective transformations of less reactive carbonyl groups in the presence of aldehydes: Table 6 in the previous section shows that the reactivity of α-alkoxyphosphonium salts with water is strongly dependent on the phosphine used. Therefore PPh3, which was easily available, was selected from the phosphines that provided stable α-alkoxyphosphonium salts. Next, we compared the reactivity of the PPh3 salt with the (o-tol)3P salt using Grignard reagent as nucleophile (Chart 80). The phosphoniun salt formed by the reaction of the acetal and TMSOTf–(o-tol)3P reacted smoothly with PhMgBr and gave the desired product in high yield within 1 h. On the other hand, the salt of PPh3 was less reactive, and only a small amount of the desired product was obtained over 1 h. Therefore we attempted to develop an in situ protection method to reverse the reactivity of keto-aldehydes using PPh3.
First, the reaction was studied using a substrate that had a ketone and an aldehyde in the same molecule (Chart 81). Compound 3.61a was first treated with PPh3 and trialkylsilyltriflate at 0 °C in CH2Cl2 for 1 h, then BH3·THF was added at 0 °C. After confirming completion of the reaction by TLC check, treatment of the reaction mixture with sat. NaHCO3 aq. (or TBAF) gave a high yield of hydroxyl aldehyde 3.62a, in which ketone was selectively reduced. Even using excess amounts of PPh3, R3SiOTf, and BH3·THF, 3.62a was still obtained in high yields (Chart 81, eq. 1). The results reveal that phosphonium salts were formed only from aldehyde, not ketone. According to further studies, the optimized conditions for this protocol were: 1) treatment of 1.2 equiv of PPh3 and 1.2 equiv of TMSOTf at r.t., 2) reduction with 1.5 equiv of BH3·THF at −40 °C, followed by 3) stirring at 40 °C with sat. NaHCO3 aq., and 3.62a was obtained in 96% yield (Chart 81, eq. 2).
The highly chemoselective formation of phosphonium salt from aldehydes only, not ketones, was also confirmed by NMR studies of the reaction mixture prepared from the reaction of keto-aldehyde 3.61b, PPh3, and TMSOTf in CD2Cl2 (Fig. 13). The 1H-NMR chart of the reaction mixture showed disappearance of the aldehyde signal observed in the starting 3.61b, new generation of O,P-acetal signals, and intact acetyl function signals. The 13C-NMR chart also confirmed that the aldehyde carbon had disappeared and ketone carbon remained intact.
(Color figure can be accessed in the online version.)
A variety of substrates containing two different carbonyl groups, i.e., ketones or esters and aldehydes, in the same molecule were available in this way. Chart 82 shows the product structures and yields. The in situ protection method, which proceeded via O,P-acetal phosphonium salt intermediates, could be used for alkylation with Grignard reagents and reduction with borane or DIBAL-H to give the corresponding products in high yields. The lactols, the right compounds in upper stand, were obtained by intramolecular attack of the hydroxyl group, generated by selective ketone reduction and alkylation of 5-oxoaldehyde, to the aldehyde. The method was very gentle and the enolizable ketone was selectively reduced in high yield. Bulky ketone and benzophenone ketone were reduced to afford the desired products in moderate yields (Second to fourth compounds in middle stand). Ester groups in substrates that also contain aldehyde moieties could also be selectively converted to the corresponding alcohols by DIBAL-H reduction or Grignard addition. DIBAL-H reduction of an ester-aldehyde at −78 °C produced the semi-reduced dialdehyde in good yield (Compounds in lower stand).
The in situ protection method enabled selective conversions of amide and nitrile groups in substrates containing aldehyde moieties. Charts 83 and 84 show the starting substrate structures and product yields. The reduction and Grignard addition of Weinreb amides proceeded smoothly, and the corresponding dialdehydes and keto aldehydes were obtained in moderate-to-good yields (Chart 83). The nitrile groups in nitrile aldehydes were selectively transformed by addition DIBAL-H or Grignard reagent, yielding the corresponding dialdehydes and keto aldehydes in moderate yields, too (Chart 84).
Selective transformations of less reactive carbonyl groups in the presence of ketones: Ketones do not react with PPh3 even when used in excess, due to the steric bulk and low nucleophilicity of PPh3. However, the use of highly reactive and commercially available PEt3 instead of PPh3 allowed selective in situ protection of ketones. As shown in Chart 85, PEt3–TESOTf in situ ketone protection of keto esters followed by DIBAL-H reduction and Grignard reaction furnished the corresponding primary and tertiary alcohols in good yields, respectively.
This in situ protection method with the combination of PEt3–TESOTf also helped in the selective conversion of Weinreb amides in the presence of ketone moieties (Chart 86).
Selective transformations of similar reactive carbonyl groups in the presence of α,β-unsaturated ketones (enones): PPh3 did not convert ketones to their corresponding phosphonium salts even with excess PPh3, as shown in Chart 81. On the other hand, it was reported that treatment of enones with PPh3 and TBSOTf yielded the phosphonium silyl enol ethers by 1,4-addition of PPh3.157–159) Therefore if the phosphonium silyl enol ethers from enones could survive during conversion of ketones, we thought that an in situ protection method for discriminating two functional groups having similar reactivity could be established.
Control of the reactivity between α,β-unsaturated ketones (enones) and saturated carbonyl groups was first studied.160,161) Ketones and enones displayed nearly identical reactivity toward nucleophiles. Indeed, when reduction of a 1 : 1 mixture of enone 3.63a and ketone 3.64a using 1.0 equiv of DIBAL-H was performed, the recovered 3.63a (41%), an allylic alcohol, an alcohol 3.65a (45%), and the recovered 3.64a without chemoselectivity were obtained. However, pretreatment of the mixture with TMSOTf and PPh3 (1.5 equiv each) for enone protection followed by DIBAL-H reduction and workup under weakly basic (K2CO3/MeOH) gave alcohol 3.65b (82%) from ketone 3.64a and the recovered enone 3.63a (82%) (Chart 87).
Chart 88 shows the generality of the method. After pretreatment of the mixture of aromatic or aliphatic enones 3.63a–c and other carbonyls such as various ketones 3.64 including an easily enolizable 3.64e or esters 3.66 including an α,β-unsaturated ester 3.66a was treated with PPh3 and TMSOTf followed by DIBAL-H reduction, and the recovered enone 3.63 and the corresponding alcohol 3.65 and 3.67 were obtained in high yields. Pyridyl ketone 3.64c and unprotected indolyl ester 3.66d were also selectively reduced. The acid-labile MOM and TBS groups in 3.64f and 3.66c survived under the reaction conditions. The substrates 3.68a–e, which contained both enone and other carbonyl groups in the same molecule, also gave the corresponding products 3.69a–e in high yields. It is worthy to note that compound 3.68e containing Weinreb amide provided the aldehyde 3.69e in 76%.
The in situ protection method was also applicable to ketone-selective alkylation of keto-enones with ketone and enone in the same molecule, as shown in Chart 89. After pretreatment of keto-enones, the nucleophile was added to the reaction mixture. As a result, ketone-selective alkylated products were obtained in high yields by Reformatsky-type and Grignard reactions. Barbier-type allylation using In(0) also reacted ketone-selectively. These were the first reported examples of ketone-selective alkylation of keto-enones.
This in situ protection method using 1,4-addition of PPh3 to the enone was very effective in distinguishing enones from other carbonyl compounds with quite similar reactivity.162) The reactivity of 1,2-addition to β-disubstituted enones and β-monosubstituted ones is almost the same. However, using our method, we were able to distinguish these two carbonyl compounds. Furthermore, our method was able to discriminate between β-monosubstituted enones and ynones in high selectivity. Chart 90 shows the results of the DIBAL-H reduction of a 1 : 1 mixture of β-monosubstituted enone 3.63a and β-disubstituted one 3.70a (eq. 1 and eq. 3) or a 1 : 1 mixture of β-monosubstituted enone 3.63a and ynone 3.71a (eq. 2 and eq. 4). Without in situ protection, the reactions did not give good selectivity (eq. 1 and eq. 2). Approximately equal amounts of the alcohols 3.63aR and 3.72a in the reaction in addition to the recovered 3.63a and 3.70a (eq. 1) or similar amount of 3.63aR and 3.73a along with the recovered 3.63a and 3.71a (eq. 2) were obtained. On the other hand, the in situ protection method achieved complete discrimination of two carbonyl compounds. That is, only an alcohol 3.72a from 3.70a and the recovered enone 3.63a were obtained from the mixture of 3.63a and 3.70a (eq. 3). Ynone 3.71a was also reduced selectively in the presence of enone 3.63a (eq. 4).
Chart 91 shows the generality of this in situ protection method. A variety of β-monosubstituted enones 3.63, β-disubstituted enones 3.70, and ynones 3.71 were available, and reduced products, 3.72 from 3.70 or 3.73 from 3.71, were obtained in good-to-high yields along with the recovered 3.63. The substrates with two carbonyl groups in the same molecule, 3.74 with β-mono- and β-di-substituted enone moieties, or 3.76 with β-monosubstituted enone and ynone units, also gave the products, 3.75 and 3.77, both of which have intact β-monosubstituted enone units, in high yields.
Discriminative two-way transformations of α,β-unsaturated esters in the presence of enones; Concise synthesis of oxacyclic compounds163): As described in the previous chapter, our in situ protection method was applied to enones to give the phosphonium silyl enol ethers in CH2Cl2, and other carbonyl derivatives such as ketones, esters, Weinreb amides, β-disubstituted enones, and ynones were converted selectively in the presence of β-monosubstituted enones. We next considered applying this method to combinations of α,β-unsaturated esters and enones. During that study, other phosphonium salts were formed from enones in a protic solvent, and double bond-selective conversion of α,β-unsaturated esters was successfully achieved in the presence of enones. Furthermore, as an extension of the work, one-pot synthesis of oxacyclic compounds in six steps was achieved.
That is, a 1 : 1 mixture of enone and α,β-unsaturated ester or the substrate having enone and α,β-unsaturated ester in the same molecule was treated with PPh3 and TMSOTf in CH2Cl2 followed by DIBAL-H reduction or Grignard reaction then TBAF work-up afforded the allyl alcohol from α,β-unsaturated ester and the recovered enone or the substrate with allyl alcohol and enone unit (Chart 92, eq. 1). On the other hand, when PPh3 and TMSOTf were reacted in protic solvent such as wet acetone instead of CH2Cl2, 3-oxoalkyltriphenylphosphonium salts were formed, and the olefin units of α,β-unsaturated esters in the presence of enones were successfully converted selectively (Chart 92, eq. 2).
(Color figure can be accessed in the online version.)
Figure 14 shows 1H-NMR charts of the enone 3.63a, the reaction mixture obtained by treatment of 3.63a with PPh3 and TMSOTf in CD2Cl2 (Chart A), and the reaction mixture obtained by treatment of 3.63a with PPh3 and TMSOTf in CD3OD (Chart B). Chart A shows disappearances of acetyl methyl and two olefinic proton signals, and appearances of trimethylsilyl, vinyl methyl, one olefin, and one benzyl proton signals. The stereochemistry of the olefin was determined cis by nOe experiment. On the other hand, in Chart B acetyl methyl, methylene, and benzyl proton signals were observed. In addition, the structure of the salt in Chart B was unambiguously determined by X-ray analysis.
(Color figure can be accessed in the online version.)
The bidirectional discriminative transformations shown in Chart 92 were examined using β-phenyl and β-alkyl enones 3.63a,f, β-phenyl and β-alkyl α,β-unsaturated esters 3.78a,b, and the substrates 3.79a,b containing both functional groups in the same molecule (Fig. 15). The products and their yields are shown in Fig. 15. Pretreatments of a 1 : 1 mixture of 3.63a and 3.78a, 3.63a and 3.78b, 3.63f and 3.78a, or 3.63f and 3.78b with PPh3 (1.5 equiv) and TMSOTf (1.5 equiv) in CH2Cl2 followed by DIBAL-H reduction or Grignard reaction afforded the corresponding products 3.80a,b from 3.78a or 3.81a,b from 3.78b in high yields with the recovered 3.63a or 3.63f. In the same way, the selective conversion of ester sites of 3.79a,b gave the corresponding products 3.82a,b and 3.83a,b in high yields. On the other hand, a 1 : 1 mixture of 3.63a and 3.78a, 3.63a and 3.78b, 3.63f and 3.78a, or 3.63f and 3.78b was treated with PPh3 (1.0 equiv) and TMSOTf (1.0 equiv) in MeOH/AcOEt (7/1) followed by hydrogenation (H2 balloon, 5% palladium carbon (Pd-C)) or in acetone/water (50/1) followed by dihydroxylation (10 mol% K2OsO4, 5 equiv NMO) to give the recovered 3.63a or 3.63f and the corresponding products 3.84a,b from 3.78a or 3.85a,b from 3.78b in high yields in all the reactions. Similarly, the olefin moieties of 3.79a,b were selectively converted to give the corresponding products 3.86a,b and 3.87a,b in good-to-high yields.
This in situ protection method using these two-types phosphonium salts was applied to pot-economic synthesis of oxacyclic compounds using the substrates 3.88a–d having enone and α,β-unsaturated ester in the same molecule (Chart 93). The enone unit of 3.88a–d was first converted to the phosphonium silyl enol ether intermediate 3-v by reaction in CH2Cl2, then reduced with DIBAL-H to convert the ester group to an alcohol intermediate 3-vi. After concentration, the reaction solvent was changed to acetone/H2O (50/1) and the mixture treated with conc. HCl to convert the phosphonium silyl enol ether 3-vi to the 3-oxoalkyltriphenylphosphonium salt 3-vii. Dihydroxylation of the olefin unit of 3-vii gave dihydroxylphosphonium salt 3-viii. Treatment of 3-viii under alkaline conditions regenerated the enone 3-ix, and successive oxa-Michael reaction produced the oxacyclic compounds 3.89a–d in moderate yields. These six-step reactions were performed in one-pot. Asymmetric dihydroxylation using dihydroquinidine 4-chlorobenzoate catalyst was performed, and asymmetric synthesis of optically active oxacyclic compound was also successful.164,165)
(Color figure can be accessed in the online version.)
The above in situ protection methods were applied to concise asymmetric synthesis of natural products. In other words, (+)-centrolobin was synthesized using the reversal of the reactivity, and decytospolides A and B were synthesized using the control of the reactivity.
(+)-Centrolobin155): (+)-Centrolobine isolated from the heartwood of Centrolobium robustum has anti-inflammatory, antibacterial, and anti-leishmanial activities.166–168) Asymmetric syntheses of centrolobine have been reported by several groups, but they involve multiple steps and/or give low total yields.169) Our synthetic route is summarized in Chart 94. One-pot conversion to optically active lactol 3.92 was achieved by applying our in situ protection method to reverse the reactivity of keto-aldehyde 3.91 obtained by ozonolysis of commercially available cyclopentene 3.90. Asymmetric reduction of ketone was performed using CBS reagent.23) The Horner–Wadsworth–Emmons (HWE) reaction of 3.92 with 3.93 furnished the desired 2,6-cis-THP 3.94. Decarbonylation of 3.94 and the following Ullmann coupling with CuI–NaOMe yielded (+)-centrolobine in five steps from a commercially available 3.90 in 75% overall yield.
(Color figure can be accessed in the online version.)
Decytospolides A and B: Decytospolides A and B were isolated from Cytospora sp., an endophytic fungus from Ilex canariensis.170) Decytospolide B shows in vitro cytotoxic activity against tumor cell lines A549 and QGY. There were two reports on the total synthesis of these natural products, which required multiple steps.171,172) We could synthesize decytospolides A and B from 3.95 in six and seven steps, respectively, with a total yield of 47% and 46% (Chart 95). That is, asymmetric reduction of commercially available cyclopentanone 3.95 with the CBS reagent,23) protection of the hydroxyl group, and one-pot ozonolysis–Wittig reaction using 3.96 were performed, and optically active keto-enone 3.97 was obtained in three steps. The oxacyclic compound 3.98 was obtained by chemoselective reduction of the ketone of 3.97 applying our in situ protection method to control the reactivity, followed by an intramolecular oxa-Michael reaction and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) treatment. The BOM-ether of 3.98 was deprotected by hydrogenolysis to produce decytospolides A, whose hydroxyl group was acetylated to give decytospolide B.
In this review, I have described our challenges on synthetic organic chemistry. All the reactions introduced here were developed using the characteristic reactivity of oxygen atoms with lone pairs.
Although the rearrangement reactions of epoxy derivatives were already known an excellent tool for skeleton transformations, there was no report of epoxy acylate rearrangement reaction at a practical level before our work. To obtain a new tool for skeletal transformation, we developed epoxy acylate rearrangement reactions at a practical level. In particular, determination of the absolute configuration of fredericamycin A using the rearrangement reaction of epoxy acylate is a good example showing the usefulness of this method, which can be applied to the determination of absolute configurations, which are difficult to achieve by spectroscopic methods. Furthermore, the rearrangement of epoxy tosylates having an electron-withdrawing functional group and the application of this to the asymmetric synthesis of several natural products show the usefulness of these rearrangement reactions.
Asymmetric synthesis using C2-symmetric acetal derivatives also offers a new approach. It is still difficult to obtain optically active tert-alcohols by asymmetric addition to ketone. Nucleophilic addition to α-keto acetals provides a reliable method of chiral tert-alcohol synthesis. Asymmetric synthesis of intramolecular haloetherification using chiral ene or diene acetals can provide two or four chiral centers in a single step. Many asymmetric syntheses using chiral C2-symmetric acetals have been reported so far, but our reaction is the only one that does not fall into those categories. A new concept using this reaction, chiral auxiliary multiple-use method, was proposed and applied to the concise asymmetric synthesis of many natural products to demonstrate its usefulness. As an extension of this method, we developed an asymmetric synthesis of optically active meso-diol derivatives using intramolecular bromoetherification of acetals synthesized from chiral ene aldehyde and meso-diols. The most common way to desymmetrize meso-diols is to use enzyme or asymmetric catalyst. However, in these cases, the protecting groups of hydroxyl function are only acyl groups. On the other hand, our method can use almost all protecting groups of hydroxyl function.
Organic reactions using acetal-type reactive salt chemical species are also revolutionary. Our reactions use the pyridinium-type salts as good leaving groups. As a result, the order of hydrolysis of the acetal was reversed. It is common knowledge in organic chemistry that ketals are hydrolyzed faster than acetals, as described in organic chemistry textbooks. On the other hand, our method is the only method that can deprotect acetal in the presence of ketal. This was reported in 2004, and remains the only method that can reverse the order of hydrolysis of acetal and ketal. Another characteristic of our method is that the reaction proceeds under weakly basic conditions. Therefore this method can be applied to acid- and base-labile substrates and reagents. Its usefulness is evident from the use of many other groups. We also succeeded in deprotecting the acetal-type protecting groups of the hydroxyl groups in the reverse order of conventional methods. For reactions using phosphonium salt species, the in situ protection methods using acetal-type phosphonium salts as protecting group for the carbonyl functions are noteworthy. That is, the usual order of reactivity of carbonyl groups to nucleophiles is aldehyde, ketone, ester, amide, and nitrile, but the reactivity can be reversed using this method. Several in situ protection methods that cause reactivity reversal have been reported by other groups, but they require the use of special reagents and strict stoichiometry control. Conversely, our method does not require such strict conditions, and the reagents used are commercially available. Therefore this in situ protection method is the most practical, to the best of our knowledge. Furthermore, our in situ protection method using phosphonium salt species could also control reactivity between similarly reactive carbonyl functionalities. This is the first method successfully to do this. Reactions using pyridinium salts and phosphonium salts allow the coexistence of functional groups that normally cannot coexist because the individual reactivity of the functional groups can be reversed or controlled. Therefore I believe that reactions using reactive salts have potential to change the retrosynthesis planned based on conventional reactions.
I am very grateful to the emeritus professors of Osaka University, the late Professor Yasumitsu Tamura and Professor Yasuyuki Kita for their suggestions and supports. My collaborators, Associate Professor Tomohiro Maegawa, Associate Professor Mitsuhiro Arisawa, Assistant Professor Kenichi Murai, and especially all students who contributed to the works described here, are gratefully acknowledged with deepest appreciation. I am deeply grateful for financial support from KAKENHI by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and from the Japan Society for the Promotion of Science.
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.
