Enantioselective Total Syntheses of Preussomerins: Control of Spiroacetal Stereogenicity by Photochemical Reaction of a Naphthoquinone through 1,6‐Hydrogen Atom Transfer

Abstract We report the enantioselective total syntheses of preussomerins EG1, EG2, and EG3. The key transformation is a stereospecific photochemical reaction involving 1,6‐hydrogen atom transfer to achieve retentive replacement of a C−H with a C−O bond, enabling otherwise‐difficult control of the spiroacetal stereogenic center.


Introduction
Spiroxins 1 and 2 [1] and preussomerins 3-5 [2,3] are the most elaborate members of the spirodioxynaphthalenes, a class of fungi-derived, highly oxygenated naphthoquinone dimers with potent bioactivities (Scheme 1). [4] Formidable synthetic challenges are provided by their molecular complexity due to the triply linked bis-naphthoquinone scaffolds. Note that 1 and 2 are linked by a CÀ C bond and two CÀ O bonds, while 3-5 are linked by three CÀ O bonds.
In our study toward the enantioselective syntheses of these compounds, we previously reported the first enantioselective total syntheses of 1 and 2 [5] by exploiting a photochemical reaction of naphthoquinone A [6] that enabled us to control the spiroether stereogenic center (pink). The benzylic CÀ H bond in A is nicely replaced by an internal CÀ O bond in D with retention of the configuration by the following mechanism: Upon n-π* photoexcitation of A, the subsequent processes are 1) 1,5-hydrogen atom transfer (1,5-HAT) to give biradical B, 2) single electron transfer (SET) to form zwitterion C, and 3) cyclization to cyclic ether D, which proceed rapidly enough relative to the internal CÀ C bond rotations to enable a stereospecific transformation. This process can be regarded as an intramolecular redox reaction, in which the benzylic oxidation level is increased and the quinone is reduced.
We turned our attention to preussomerins 3-5, which are even more challenging targets in view of the obvious difficulty in controlling a spiroacetal stereogenic center (yellow). [7,8] Faced with this problem, we focused again on the above-mentioned photochemical reaction, but using naphthoquinone substrate E designed by inserting an oxygen atom into A en route to chiral, non-racemic spiroacetal H.
Two concerns were foreseen: 1) Viability: Process E!F involves a 1,6-hydrogen atom transfer (1,6-HAT), known to be a less facile process than its 1,5-counterpart. [9][10][11] 2) Stereospecificity: The anchoring oxygen atom in E adds conformational mobility (with respect to A) that may increase the chances for racemization during the course of the reaction. [5,12] Herein we describe the design of viable naphthoquinone substrates that allow the desired stereospecific photochemical reaction involving a 1,6-HAT process, and the first enantioselective total syntheses of preussomerins EG 1 (3), EG 2 (4), and EG 3 (5).

Results and Discussion
The initial feasibility study is summarized in Table 1. We selected O-benzyl juglone 6 as a substrate, which was irradiated with fluorescent light (CH 3 CN, RT, entry 1).
[6a] As the resulting product 7 was unstable to air, it was protected as a methyl ether. The fact that acetal 8 was obtained established the feasibility of the projected photochemical reaction, although the yield was low. Interestingly, this is the first report on the photochemical reaction of 6, which had been long known, and must have previously been exposed to light. [13] In addition, the dimeric product 12 was obtained in 17 % yield, which came from an oxidative dimerization of phenol 7 followed by methylation. [14] Unfortunately, the reaction took a very long time (5.5 days), which stands in contrast to the cases with a 1,5-HAT (Scheme 1), [6a] suggesting the expected difficulty of the 1,6-HAT. At this juncture, we considered that the low efficiency of the photochemical reaction could be attributed to the scarcity of the reactive conformer I (R = H), in which the benzylic CÀ H bond is disposed near the quinone carbonyl. Note that J is an unreactive conformer: Even if photoactivated, the excited state derived from J would not be productive due to the spatially unfavorable arrangement, and would relax back to the ground state.
With the above consideration in mind, we came up with the idea of installing a bromine atom as in 9 a, inspired by a report by Baldwin and Brown in 1969. [15] Our hope was that the population of the reactive conformer I (R = Br) would be increased, providing an entropic advantage for the 1,6-HAT to occur. Indeed, the irradiation of 9 a with fluorescent light afforded acetal 11 a in high yield (80 %), and the reaction time was shortened (2 days, entry 2). [16] Moreover, the reaction time was further shortened (20 min) by using LED irradiation (448 nm, 680 mW), affording 11 a in 84 % yield (entry 3). The LED light also shortened the reaction time of non-bromo substrate 6 (2 h), although the yield was lower (42 %, entry 4).
Having obtained this promising result, we examined the substrate scope of the reaction by using bromonaphthoquinones 9 b-l with various ether substituents ( Table 2). [14] Two experimental precautions were implemented: 1) Considering the potential instability of bromonaphthoquinones 9 b-l toward room light, they were prepared from the corresponding naphthalenes 13 b-l immediately before the photoreactions, and 2) as the resulting products were unstable to air, they were isolated as the corresponding methyl ethers 11 b-l by treatment with dimethyl sulfate (for the general procedure, see the Supporting Information).
Quinone 9 b with an allylic CÀ H bond (R = CH=CH 2 , R' = H) was also a good substrate, giving the corresponding acetal 11 b in 71 % yield, although Norrish-Yang product 14 was obtained as a minor product (entry 1). On the other hand, quinone 9 c with a methyl group (R = R' = H) was a poor substrate, requiring a longer reaction time and giving only poor yield (entry 2). Better results were obtained with larger alkyl substituents. Quinone 9 d with an ethyl group (R = Me, R' = H) underwent the reaction in a shorter time, giving an improved yield (entry 3). The improvement was even more prominent with quinone 9 e with an isopropyl group (R = R' = Me, entry 4), showing that the dialkyl substitution leads to good substrates. Indeed, quinones 9 f and 9 g with cycloalkyl groups gave the corresponding products 11 f and 11 g with spiroacetal structures (entries 5 and 6), which were promising results for the synthesis of the preussomerins.
Given the reaction mechanism (Scheme 1), this trend could be rationalized by considering the substitution effect in terms of: 1) the bond dissociation energies of the reacting CÀ H bond 2) the electron-donating ability to facilitate the SET step [6] Scheme 1. Photochemical approach to spiroxins and preussomerins: Stereospecific 1,5-HAT (previous work) and 1,6-HAT (this work). Along these lines, we expected quinone 9 h (R = OMe, R' = H) would be a good substrate, as the methoxy group weakens the CÀ H bond and exerts an electron-donating ability (entry 7). However, orthoester 11 h was obtained in only 30 % yield, owing to the chemical lability of the methoxymethyl group, which was partially cleaved during the reaction. Our related expectation for the electron-withdrawing group R was that it would discourage the SET process. Indeed, quinone 9 i (R = COOEt, R' = H) gave acetal 11 i in low yield (entry 8), and 9 j (R = CN, R' = H) gave an intractable mixture of unidentified products (entry 9). [6a] The reaction of quinone 9 k with a 2-phenylethyl group (R = CH 2 Ph, R' = H) gave 11 k in 67 % yield as the sole product arising from the expected 1,6-HAT (entry 10). Our interest in this particular substrate 9 k was that a 1,7hydrogen atom transfer (1,7-HAT) may compete by participation of the distal CÀ H bond (a weak benzylic bond). As it turned out, none of 11 k' expected from the 1,7-HAT was detected (see the bottom of Table 2).
Quinone 9 l with a cyclopropylmethyl group (R = cyclopropyl, R' = H) gave acetal 11 l in 69 % yield (entry 11), suggesting that the SET process was much faster than the ring opening of the cyclopropylmethyl radical and/or the corresponding cationic species, [17,18] as evidenced by NMR analysis of the crude products: the lack of vinyl protons suggested the absence of the ring-opened products.
Having studied the substrate scope, we proceeded to plan the synthesis of preussomerin EG 1 (3), paying particular attention to the stereospecificity. Scheme 2 shows our retrosynthetic analysis. The bis(spiroacetal) structure in 3 would be derived from quinone 15 through an intramolecular quinone/hydroquinone redox reaction. [5b, 7a] Benzoquinone 15 could be derived from phenol (R)-16, presuming the installation of an oxygen atom para to the C9 phenol.
We were pleased to realize that the reaction proceeded in a stereospecific manner (> 99 % ee, Scheme 3b), as verified by HPLC analysis of 22 on a chiral stationary phase (CHIRALPAK ® IB, eluent: hexane/EtOAc = 80/20, flow rate: 1 mL min À 1 , 25°C, retention time: 13.5 min for the S isomer, 14.9 min for the R isomer). The absolute configuration of the spiroacetal center in 22 was assigned as S by X-ray diffraction analysis (Figure 1), [23] verifying the stereoretentive nature of the key photochemical reaction.
We are pleased to document that even a photochemical process including 1,6-HAT can nicely transmit stereochemical information to the product without any loss of the enantiomeric excess. [24] Scheme 4 shows the rationale for the stereospecific conversion of (S)-17 into (S)-21 based on our working hypothesis. [5] Among possible conformations of (S)-17, the key conformer K has the hydrogen atom near the quinone carbonyl group. Upon photoexcitation, a 1,6-HAT process occurs to generate biradical L, whose molecular shape reflects that of K, in which the stereochemical information of (S)-17 is retained as axial chirality.
Given the following processes L!M!N were fast enough relative to the conformational flipping of two intermediates L (to give O) and M (to give P), the final cyclization regenerates the central chirality in the product N [= (S)-21] with retention of configuration, which is consistent with the present result. [25] In some cases in our previous study partial stereomutation was observed, suggesting the intervention of such conformational changes of reactive intermediates, while the present case was not disturbed by such stereochemical deterioration. [26] With enantiomerically pure (S)-22 in hand, the next task was its conversion into quinone 24, the key intermediate toward the preussomerins (Scheme 5). Hydrogenolysis of spiroacetal (S)-22 removed the benzyl group and the bromine atom, giving phenol 16. Addition of triethylamine was crucial for the debromination. [27] Attempted oxidation of phenol 16 to quinone 24 did not proceed at all. We attributed this failure to the poor πelectron density of phenol 16 by the presence of a carbonyl group. After considerable experimentation, we decided to transform the ketone into the corresponding alcohol. Ketone 16 was reduced with NaBH 4 , and the resulting alcohol was treated with PhI(OCOCF 3 ) 2 in wet THF to produce the desired quinone 23 in 41 % yield.   many other possibilities were examined, this was the optimum yield. A dilemma was that the application of stronger oxidants seemed to cause competitive oxidation of the naphthalene moiety.
Oxidation of alcohol 23 with Dess-Martin periodinane gave labile ketone 24, which was used without silica gel chromatography for the next step in the synthesis of preussomerin EG 1 (3). Acetate 24 was treated with K 2 CO 3 in methanol (Scheme 5). Although the projected reaction (24!3) indeed proceeded, it was accompanied by the facile 1,4-addition of methanol to give preussomerin EG 3 (5) in 85 % yield. All the spectroscopic data of the synthetic material 5 matched with the reported data for the natural product. [2] The structure was further verified by X-ray diffraction analysis. [23] The facile conversion of 24 into 5 could be explained as follows: Removal of the acetyl group in 24 gives phenolate 25, [28] which undergoes an internal redox reaction to give the bis(spiroacetal) structure of 3. [5b, 7a] This process could be expressed as a formal intramolecular 1,6-addition (see the electron arrows in 25, Scheme 5). The α,β-unsaturated system in 3 underwent facile attack of a methanol to afford 5 as a single stereoisomer. [7a] To explain this rigorous stereoselectivity, we could conceive three relevant factors, albeit of unknown relative significance. 1) convex/concave (outside from the cage) 2) Felkin-Anh (anti to the α-oxygen) 3) axial attack (minimizing the torsional strain) Seeking direct access to 3, we screened several reaction parameters to suppress the extra 1,4-addition to produce 5, which unfortunately turned out to be unfruitful (Table 3). Entry 1 shows the result already presented in Scheme 5. Lowering the temperature (À 20 and À 40°C) afforded 3 in low yield (entries 2 and 3). Changing the solvent (EtOH and t-BuOH) resulted in a complex mixture and gave only a small amount of 3 without the corresponding adduct 27 and 28, respectively (entries 4 and 5). When hydroxide was employed as a nucleophile, compound 3 was not obtained at all (entry 6). The major product was angular alcohol 29 formed in 43 % yield, which was generated by a 1,4-addition of hydroxide to the highly electrophilic β-position of acylquinone 24. [29] The essential issue was that the acetyl protecting group on the phenol was much more resistant to removal than expected. Namely, these conditions generated hydroquinone 26 bearing an acetyl group, which was derived from the reduction of 24. The reductant would be phenolate 25, functioning as an intermolecular two-electron donor to the reactive electron acceptor 24. [30] With these results, we judged that the direct synthesis of 3 from 24 was not practical, and alternatively focused on the transformation of 5 into 3 (Scheme 6). Treatment of 5 with TMSOTf and Et 3 N caused the β-elimination of methanol, [7a] while the ketone and phenol in 3 were simultaneously transformed into the corresponding silyl ether to furnish 30. [31] Hydrolysis under weakly acidic conditions gave 3 in  83 % yield (2 steps). [32] All the spectroscopic data of the synthetic material 3 were identical with those reported. [2] For the synthesis of (À )-preussomerin EG 2 (4), the direct 1,4-addition of H 2 O to 3 was difficult. [33] On the other hand, a two-step protocol-stereoselective epoxidation and reductive opening of the oxirane ring-nicely enabled the total synthesis of 4 in 88 % yield (2 steps). [7d] All spectroscopic data of the synthetic material 4 were identical to those reported. [2] The structure of 4 was further confirmed by Xray diffraction analysis. [23] The [α] D values of the synthetic materials 3-5 were the same in sign to those of the natural samples, while substantially different in magnitude. We are confident in the chemical purity and the [α] D values of our synthetic materials, so the discrepancy may come from the low enantiomeric excess and/or insufficient chemical purity of the natural samples. [34]

Conclusion
In conclusion, photochemical reactions of naphthoquinones involving 1,6-HAT have been developed. The installation of a bromine atom ortho to the substituent on the naphthoquinones rendered the 1,6-HAT facile by an entropic effect. The substrate scope and limitations have been described with respect to the substitution pattern. The photochemical reaction proceeded in a stereospecific manner (retention), which enabled the first enantioselective total syntheses of preussomerins EG 1 , EG 2 , and EG 3 .