Asymmetric Syntheses of the Flavonoid Diels–Alder Natural Products Sanggenons C and O

Abstract

Metal-catalyzed, double Claisen rearrangement of a bis-allyloxyflavone has been utilized to enable a concise synthesis of the hydrobenzofuro[3,2-b]chromenone core structure of the natural products sanggenon A and sanggenol F. In addition, catalytic, enantioselective [4+2] cycloadditions of 2′-hydroxychalcones have been accomplished using B(OPh)₃/BINOL complexes. Asymmetric syntheses of the flavonoid Diels–Alder natural products sanggenons C and O have been achieved employing a stereodivergent reaction of a racemic mixture (stereodivergent RRM) involving [4+2] cycloaddition.

Sanggenon-type natural products (Figure 1) are intriguing synthetic targets due to their complex chemical structures and potent biological activities.¹ In particular, the congener sanggenon C² has antitumor, antiviral, and anti-inflammatory properties^1c,d,3 which makes it an attractive synthetic target. Although chemical syntheses of related Diels–Alder (DA)-type natural products, including sorocenol B⁴ and brosimones A and B,⁵ have been achieved, there are no reported catalytic, enantioselective [4+2] cycloadditions of 2′-hydroxychalcones, nor are there published reports of DA cycloadditions of complex flavonoid dienes. Here we report the first total syntheses of sanggenon A (1) and sanggenol F (2), both featuring complex benzofuro[3,2-b]chromenone structures. Further structural complexity was obtained employing enantioselective [4+2] cycloaddition to construct sanggenon C (3) and sanggenon O (4)⁶ using a unique stereodivergent RRM.⁷

Figure 1.

Structures of sanggenon-type natural products.

Sanggenons C (3), D (5),⁸ and O (4) are apparent DA cycloadducts between a flavonoid diene (7) and a 2′-hydroxychalcone (6) (Scheme 1). Among these compounds, sanggenons C and O were shown to be endo cycloadducts and epimers at C-2 and C-3. We considered stereodivergent RRM of chiral, racemic diene 7 as a promising strategy, assuming that issues of endo/exo selectivity, face selectivity, and enantioselectivity could be addressed. Compound 7 may be derived from dehydrogenation of the prenyl group of sanggenol F or its protected derivative (9).^5a,9 Alternatively, 7 may be derived from isomerization of the chromene ring in sanggenon A or protected substrate (8).¹⁰ Sanggenol F core structure 9 may arise from olefin cross-metathesis of precursor 10.We envisioned that 10 may be derived from metal-catalyzed double Claisen rearrangement¹¹ of bisallyloxyflavone ether 11, followed by hemiketalization.

Scheme 1.

Retrosynthetic Analyses for Sanggenons C and O

Synthesis of 11 commenced with tetra-MOM group protection of the commercially available compound morin (12) and subsequent 5-allylation to afford the protected intermediate 13 (Scheme 2). Selective 3-MOM deprotection was accomplished using NaI and a catalytic amount of aqueous HCl.¹² In this transformation, protonation of the 4-carbonyl and O-3 of 13 appears to activate the 3-MOM group for chemoselective deprotection. Finally, 3-allylation and global MOM deprotection afforded the desired substrate 11 in 33% overall yield (5 steps).

Scheme 2.

Synthesis of Morin Di-allyl Ether^a

^aReagents and conditions: (a) MOMCl (5 equiv), DIPEA (5 equiv), CH₂Cl₂, 0 °C to rt, 55%; (b) allyl-Br (1.2 equiv), Cs₂CO₃ (1.2 equiv), DMF, 65 °C, 12h, 93%; (c) NaI (0.8 equiv), 1 N HCl (0.1 equiv), 40 °C, 12h, 73%; (d) allyl-Br (2.0 equiv), Cs₂CO₃ (2.0 equiv), DMF, 65 °C, 12h, 96%; (e) conc. HCl, MeOH, 40 °C, 93%.

A number of rare earth (RE) metal triflates were evaluated for double rearrangement of 11 (Table 1).¹³ Among the metal triflates, Yb(OTf)₃ was found to produce the double rearrangement product 15 in 72% yield. In this reaction, hexafluoro-2-propanol (HFIP) was used as a polar, non-coordinating¹⁴ cosolvent to solubilize the polar rearrangement substrate. Other RE triflates were found to produce 15 along with the single rearrangement product 16 in varying ratios. Based on examination of a DFT molecular model¹³ for 11, there are two possible chelation sites between O-3, O-4 (2.85 Å) and O-5, O-4 (2.72 Å) through either five- or six-membered arrangements. Selective binding to either or both chelation pockets may occur depending on the ionic radius¹⁵ of the Lewis acid to enable 3-allyl rearrangement and/or 5-allyl rearrangement.¹⁶ Different modes of chelation by RE metal triflates may explain the product distributions observed in Table 1. For example, La(OTf)₃ with a larger ionic radius (1.03 Å) may preferentially chelate between O-3 and O-4 with a larger binding pocket to afford the 3-allyl rearrangement product 16 as the major product.

Table 1.

Lewis Acid-mediated Double Rearrangement^a

entryb	M(OTf)x	ionic radii (pm)c	yields of 15, 16 (%)
1	Yb(OTf)3	86.8	72, 0
2	Y(OTf)3	90.0	60, 36
3	Gd(OTf)3	93.8	48, 40
4	Nd(OTf)3	98.3	38, 62
5	La(OTf)3	103.2	0, 70

^aSee SI for experimental details.

^bAll reactions were carried out with substrate 11 (0.13 mmol) and M(OTf)₃ (15 mol%) at 0.1 M in CH₂Cl₂/HFIP (4:1) at 50 °C.

^cFor effective ionic radius, see ref 15.

After obtaining the desired hydrobenzofuro[3,2-b]chromenone core structure (±)-15 via double rearrangement, we investigated syntheses of the derived natural products 1 and 2. Although the prenyl group could be installed by cross-metathesis¹⁷ of hemiacetal (±)-15 to directly access (±)-sanggenol F (2), we found that silylation of (±)-15 followed by cross-metathesis produced the tri-silyl-protected (±)-sanggenol F (±)-17 in excellent yield (Scheme 3). This sequence enabled purification of the final product from residual ruthenium byproducts. Desilylation of (±)-17 afforded (±)-2, which was dehydrogenated with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ)^10,18 to afford (±)-sanggenon A (1) in 78% yield.

Scheme 3.

Syntheses of Sanggenol F and Sanggenon A^a

^aReagents and conditions: (a) TBSOTf (3.2 equiv), NEt₃ (3.0 equiv), CH₂Cl₂, rt, 6h; (b) Grubbs second generation catalyst (10 mol%), isobutene, 40 °C, 24h, 92%, 2 steps; (c) NEt₃·3HF (8.0 equiv), CH₃CN, 0 °C, 3h, 91%; (d) DDQ (1.2 equiv), THF, 60 °C, 12h, 78%.

We next evaluated 1 as a diene precursor for [4+2] cycloaddition. Treatment of 1 under Brønsted acidic, thermal, or photochemical conditions in the presence of silica-supported silver nanoparticles (AgNPs)¹⁹ and 2′-hydroxychalcone (cf. 20) as dienophile did not lead to the desired [4+2] cycloaddition; only decomposition of (±)-sanggenon A was observed. Accordingly, we elected to prepare a protected variant of sanggenon A to serve as a diene equivalent. Specifically, transformation of the prenyl flavonoid (±)-17 to chromene (±)-18 was investigated. Treatment of (±)-17 with DDQ in tetrahydrofuran (THF) cleanly afforded (±)-18 (73%) (Scheme 4). However, severe decomposition of 17 was observed when halogenated solvents (e.g. CH₂Cl₂, CHCl₃, or PhCl) were employed. As DDQ is a strong electron-acceptor,²⁰ a charge-transfer complex may be produced when an electron-donating solvent such as THF is employed.²¹ Formation of such a complex may modulate the reactivity of DDQ,²² which appears to largely suppress decomposition observed using oxidative conditions with halogenated solvents. We anticipated that the TBS-protected chromene 18 could generate the desired diene 19 in situ through retro 6π-electrocyclization followed by deprotonation/protonation (formal [1,7] hydrogen shift).²³ Accordingly, (±)-18 was employed in cycloadditions with acetylated 2′-hydroxychalcone 20 in the presence of AgNPs to yield a mixture of two endo cycloadducts and minimal production of exo diastereomers. The mixture of endo cycloadducts was sequentially treated with aqueous NaHCO₃ and NEt₃·3HF²⁴ to yield a mixture of (±)-sanggenons C (3) and O (4) in 36% yield (3 steps, 4:1 d.r.).

Scheme 4.

Syntheses of (±)-Sanggenons C and O^a

^aReagents and conditions: (a) DDQ (1.5 equiv), THF, 60 °C, 73%; (b) 20 (1.2 equiv), AgNP (0.25 mol%), AcOH (2 equiv), DCE, 65 °C, 3d; (c) sat. NaHCO₃, MeOH, rt, 12h; (d) NEt₃·3HF (8.0 equiv), CH₃CN, rt, 12h, 36% (3 steps), sanggenon C (3):sanggenon O (4) = 4:1.

Unfortunately, our previously developed conditions for asymmetric Claisen rearrangement¹¹ failed to give useful levels of enantioselectivity employing bis-allyloxyflavone substrate 11. We considered that sanggenons C and O are both endo cycloadducts with the same absolute configurations for the chiral cyclohexene moiety and epimers at both C-2 and C-3. Accordingly, we proposed that enantioselective [4+2] cycloaddition of chiral, racemic substrate 18 and dienophile 20 should efficiently deliver two natural products simultaneously utilizing a stereodivergent process. In our initial studies,²⁵ we found that borate complexes derived from chiral 1,1′-bi-2-naphthol (BINOL)^5b,26 could be used in catalytic, enantioselective [4+2] cycloadditions of 2′-hydroxychalcones. A two-dimensional screen was conducted using a number of borates (B(OPh)₃, tris(p-chlorophenyl), tris(pentafluorophenyl), and tris(hexafluoroisopropyl) borate) and BINOL ligands. We found that asymmetric [4+2] cycloaddition of the model dienophile 2′-hydroxychalcone 21 and diene 22²⁷ using a catalytic amount of (S)-3,3′-dibromoBINOL 23 and triphenylborate afforded the [4+2] cycloadducts 24 and 25 in 91% combined yield, 10:1 endo/exo ratio, with >99% and 41% ee, respectively (Scheme 5a). By mixing 21, 23, and B(OPh)₃ in CH₂Cl₂, we obtained borate complex 26, the structure of which was confirmed by single-crystal X-ray structure analysis.¹³ The predicted absolute stereochemistry of the corresponding endo cycloadduct 24 based on the expected face selectivity of 26 was also confirmed by X-ray crystallography.¹³ We found that 1 equiv of phenol was also bound to the borate complex as shown in the X-ray crystal structure (Scheme 5b). The bound phenol could serve to further activate the chalcone for cycloaddition and may also protonate the cycloadduct to turnover the boron/BINOL catalyst for additional catalytic cycles. Wulff and coworkers have reported a hydrogen-bonded complex of a boroxinate catalyst and a protonated iminium substrate.^26b We also found that treatment of crystalline 26 with diene 22 led to similar results, producing 24 in high enantioselectivity.¹³ The latter result reinforces the involvement of borate complex 26 in the catalytic asymmetric DA cycloaddition, which is also supported by literature precedent.²⁸

Scheme 5.

Model Reaction and Proposed Active Complex for [4+2] Cycloaddition

We next applied this catalytic system to a stereodivergent RRM strategy to synthesize enantioenriched sanggenons C and O. Based on the model reaction (cf. Scheme 5), a catalytic amount of B(OPh)₃ and (R)-BINOL 23 were used to mediate asymmetric [4+2] cycloaddition of diene precursor (±)-18 and dienophile 20 (Scheme 6). After sequential deprotection of both acetate and silyl protecting groups, promising enantioselectivities were observed. A number of BINOL ligands were evaluated; best results were obtained when B(OPh)₃ and 23 were employed. A mixture of cycloadducts was observed, which was followed by consecutive deprotections (NaHCO₃; NEt₃·3HF) to afford sanggenon C (3) and sanggenon O (4) in 2:1 ratio with 98% and 93% ee, respectively. The high endo/exo selectivity (cf. Scheme 5) suggested that minimal amounts of exo diasteromers were generated which we were not able to isolate and characterize. In this transformation, we expected four different stereoisomers: sanggenon C (3) and ent-sanggenon O (4) from (2R,3R)-18, and ent-sanggenon C (3) and sanggenon O (4) from (2S,3S)-18 (Scheme 6). As shown in the (R)-BINOL/B(OPh)₃/chalcone complex 27, the re face of the chalcone dienophile is blocked by the bulky bromo substituent. As a result, the derived dienes (2R,3R)-19 and (2S,3S)-19 approach from the si face of the chalcone 20 to afford 3 and 4. Using (S)-BINOL 23/B(OPh)₃ as catalyst, ent-sanggenon C (3) and ent-sanggenon O (4) were isolated in 2:1 ratio with 99% and 93% ee, respectively.¹³We also noticed that, in the stereodivergent RRM, sanggenon C is produced in both high enantiomeric excess and diastereomeric ratio relative to sanggenon O. This result suggests that matched/mismatched cycloadditions may also be operative based on differential interactions of the chiral borate catalyst with both enantiomers of diene 19.

Scheme 6.

Asymmetric Syntheses of Sanggenons C and O^a

^aReagents and conditions: (a) B(OPh)₃ (20 mol%), 23 (22 mol%), 20 (1 equiv), PhCF₃, 80 °C, 48h; (b) sat. NaHCO₃, MeOH, rt, 12h; (c) NEt₃·3HF (8.0 equiv), CH₃CN, rt, 12h.

To understand the greater preference for formation of sanggenon C vs O (cf. Scheme 4), and the higher enantioselectivity observed for sanggenon C, we conducted computational studies of dienophile 27 using both enantiomers of a simplified variant of the diene (19′, TMS instead of TBS) to analyze the interacting complex of reagents engaging in DA cycloaddition.^13,29 Cycloaddition models A and B are shown in Scheme 6, with the lowest energy conformer of diene 19. It appears that cycloaddition through model A is favored in comparison to the corresponding model B using (R)-BINOL, which results in greater amounts of sanggenon C derivatives and therefore favored production of (2R,3R) stereoisomers in the product mixture. Our preliminary reaction assembly calculations¹³ show that there are significant steric interactions between the prenyl and phenyl groups on the chalcone dienophile which are likely responsible for the significantly increased energy in assemblies related to model B. Based on the X-ray structure of the chiral borate complex (cf. Scheme 5), we predicted that the use of (R)-BINOL as catalyst should yield (3″S,4″R,5″S) for both sanggenons C and O. This prediction is in agreement with absolute stereochemistry determinations reported by Nomura and coworkers.^2b

In summary, we have achieved the first asymmetric syntheses of the flavonoid Diels–Alder natural products sanggenons C and O. The syntheses employ a Lewis acid-promoted double Claisen rearrangement to construct the hydrobenzofuro[3,2-b]chromenone core of sanggenon A and sanggenol F. The first catalytic enantioselective [4+2] cycloadditions of 2′-hydroxychalcones have been developed using BINOL-boron catalysis. The high enantio- and diastereoselectivity of this catalytic system enabled a stereodivergent reaction of a chiral, racemic flavonoid diene precursor to afford enantioenriched sanggenons C and O. Preliminary calculations of the interactions of diene–dienophile complexes support the stereochemical outcomes observed in stereodivergent cycloadditions. Further studies on the chemistry and biology of DA natural products are ongoing and will be reported in due course.