Enantioselective Photocycloaddition of 3-Hydroxyflavones: Total Syntheses and Absolute Configuration Assignment of (+)-Ponapensin and (+)-Elliptifoline

Abstract

We have previously reported development of biomimetic, asymmetric [3+2] photocycloadditions between 3-hydroxyflavones and cinnamate dipolarophiles to access (−)-rocaglamide and related natural products. Herein, we describe enantioselective syntheses of aglain cycloadducts leading to the first total syntheses and absolute configuration assignments of the aglain natural products (+)-ponapensin and (+)-elliptifoline.

Introduction

The plant genus Aglaia produces a series of secondary metabolites, including methyl rocaglate (1),¹ the densely functionalized cyclopenta[b,c]benzofuran (rocaglate) anticancer agent silvestrol (2),² and the cyclopenta[b,c]benzopyrans (aglains) (−)-ponapensin (3)³ and (−)-elliptifoline (4)⁴ (Figure 1). Relative stereochemical configurations of 3 and 4 were proposed in their respective isolation reports. Although aglains do not typically possess the strong biological activity of their rocaglate counterparts, (−)-ponapensin (3) was found to exhibit potent NF-κB inhibitory activity in an enzyme-based ELISA assay (IC₅₀ = 60 nM) while rocaglamide⁵ and methyl rocaglate (1) exhibited IC₅₀ values of 2.0 and 2.3 µM, respectively.³ Other aglain natural products including 4-epi-aglain A,³ aglain B,⁶ 10-O-acetylaglain B,⁷ and aglain C⁶ were not active in the NF-κB assay (IC₅₀ > 5 µM).³ Unlike silvestrol and related rocaglates,^8,9 thus far no total syntheses of aglain natural products including 3 and 4 have been reported to date.

Figure 1.

Representative natural products from the genus Aglaia.

We have developed a biomimetic approach to the rocaglates using [3+2] photocycloaddition between 3-hydroxyflavones (3-HF’s) 5 and cinnamate derivatives 6 as dipolarophiles followed by a ketol rearrangement/reduction sequence.^8a,9b,10 Herein, we present the first enantioselective total syntheses and absolute configuration assignments of the aglain natural products (+)-ponapensin (3) and (+)-elliptifoline (4), the optical antipodes of metabolites found in nature.

Results and Discussion

Cyclopenta[b,c]benzopyran Core Synthesis

Based on previous results,^10c [3+2] photocycloaddition between oxidopyrylium 7 derived from Excited State Intramolecular Proton Transfer (ESIPT) of 3-HF 5 and methyl cinnamate 6a afforded the hydrated aglain (±)-8a in 61% yield which was recrystallized from acetonitrile (Scheme 1).¹¹ Various conditions were evaluated on substrate 8a in order to achieve stereoselective reduction of the bridgehead ketone including DIBAL-H, L-Selectride, and aluminum isopropoxide (Al(OiPr)₃), all of which did not lead to reduced products. When 8a was treated with NaBH₄ in MeOH, no reaction was observed (Table 1, Entry 1). NaBH₄ reduction proceeded when aglain 8a was better solubilized in THF (Entry 2) and afforded reduced products 9a and 10a in a 1:5 ratio. Improved diastereoselectivity was not observed using Bu₄NBH₄¹² which afforded reduced aglain 10a as the major product (dr = 5:1) in 98% yield (Table 1, Entry 3). Fortunately, treatment of 8a with Me₄NBH(OAc)₃¹³ in benzene (entry 4) afforded reduction product 9a which has the 10S configuration required for synthesis of both ponapensin (3) and elliptifoline (4). Relative configuration assignment of C10 was determined via NOE correlation observed between H10/H4 of aglain 10a and an absence of this correlation in its epimer 9a.¹¹

Scheme 1.

Cycloaddition of 3-HF 5 and Cinnamic Esters 6

Table 1.

Diastereoselective Reduction of Aglain (±)-8a

entry	reagenta	solvent	additive	9a:10ab	yield (%)c
1	NaBH4	MeOH	-	-	0
2	NaBH4	THF	-	1:5	93
3	Bu4NBH4	THF	-	1:5	98
4	Me4NBH(OAc)3	benzene	AcOH	3:1	95
5	Me4NBH(OAc)3	PhCF3	AcOH	5:1	95
6	NaBH(OAc)3	PhCF3	AcOH	10:1	95

^aSee Supporting Information for detailed experimental procedures.

^bDetermined by ¹H NMR analysis of the crude reaction mixture.

^cIsolated after purification by silica gel chromatography.

After evaluation of solvents, we found that 2,2,2-trifluoromethylbenzene (PhCF₃) was an optimal solvent for hydroxyl-directed reduction of aglain 8a to afford 9a in 10:1 dr (Entry 5). The counterion size of the reducing agent appears to be significant in enhancing diastereoselectivity of the reduction as NaBH(OAc)₃¹⁴ provided the 10S diastereomer 9a in 10:1 dr (Table 1, Entry 6). Treatment of hydrated ketone 8a with 4Å MS in CH₂Cl₂ led to the formation of the derived bridgehead ketone 11 (Figure 2).¹¹ Reduction of 11 with NaBH(OAc)₃ led to full conversion to 9a in a shorter time (2 h vs. 16 h for 8a) without loss of diastereoselectivity.

Figure 2.

Proposed assemblies for diastereoselective reduction.

A rationale for reagent-controlled, diastereoselective reduction of bridgehead ketone 11 is shown in Figure 2. When using a borohydride reducing agent (e.g. NaBH₄, Bu₄NBH₄), the bridgehead ketone 11 may be preferentially attacked on the si face due to larger dihedral angles for O-C10-C5-C4 and O-C10-C2-C3 (134° and 138°, respectively) versus the smaller dihedral angles of O-C10-C5-C5a and O-C10-C2-O (113° and 102°, respectively), thereby allowing more optimal orbital alignment for hydride addition to the bridgehead carbonyl.¹⁵ Furthermore, use of a borohydride with a large counterion (e.g. Bu₄N⁺) does not affect the diastereoselectivity. When a hydroxyl-directed reducing reagent¹³ was used (e.g. NaBH(OAc)₃), diastereoselectivity did not appear to be derived from the 5-hydroxyl group in the usual fashion as this moiety is coplanar with the C10-carbonyl. We propose that borate complexation with the 5-hydroxyl takes place (cf. Figure 2, assembly 12) leading to selective hydride delivery to the re face of the activated carbonyl due to preferred orientation of the borate near the less sterically encumbered phloroglucinol ring system. Furthermore, we considered that the oxygen lone pair of the 6-methoxy group may also exchange with the borate and further direct the reduction to the re face as a secondary effect (cf. Figure 2, assembly 13). The latter issue was examined using the model aglain 14¹⁶ which lacks a methoxy at C6 (Scheme 2). Sodium triacetoxyborohydride reduction of 14 furnished reduced aglain 15 in 8:1 dr which was not significantly different from the corresponding reduction of aglain 8a (10:1 dr). Therefore, the secondary effect illustrated in assembly 13 (Figure 2) is not likely a significant factor to the diastereoselectivity for the reduction. Furthermore, reduction of ketone 14 with Bu₄NBH₄ yielded the reduced aglain 16 in 7:1 dr, which is comparable to the diastereoselectivity observed in the Bu₄NBH₄ reduction of aglain 8a (5:1 dr). Relative configurations of 15 and 16 were determined by NOESY NMR spectroscopy.¹¹

Scheme 2.

Reductions of Demethoxy Aglain 14

Enantioselective Photocycloaddition to the Aglain Core

Having established a method to access aglain core structure 8a, we synthesized enantioenriched aglain 9a using the tetrakis-9-phenanthrenyl TADDOL 17 (Scheme 3).^10a Host-guest complexation of 17 and 3-HF 5 promoted enantioselective [3+2] photocycloaddition leading to isolation of the hydrated-aglain cycloadduct (+)-8a in 69% yield and 85.5:14.5 er. We attempted recrystallization of (+)-8a to obtain enantiopure crystals. However, in this case we observed crystals of centrosymmetric racemate (±)-8a^10a,11 with enantiomeric enrichment of the mother liquor. Fortunately, enantiopure crystals of the reduction product (+)-9a were obtained after two recrystallizations (>96:04 er after recrystallization from isopropanol and >98.5:1.5 er after recrystallization from acetonitrile).¹¹ Relative and absolute stereochemistries of enantiopure aglain (+)-9a were confirmed by X-ray analysis using Cu-K_α radiation (Figure 3)¹⁷ and the absolute configuration further confirmed by conversion of (+)-8a to (−)-methyl rocaglate (1) (Scheme 4).^{10a, 11}

Scheme 3.

Enantioselective Photocycloaddition and Subsequent Reduction

Figure 3.

Determination of the absolute stereochemistry of enantiopure aglain (+)-9a using Cu-K_α radiation X-ray analysis.

Scheme 4.

Confirmation of the Absolute Stereochemistry of Aglain (+)-8a by Synthesis of (−)-Methyl Rocaglate (1)

Further Elaboration of the Aglain Core

Before using enantiopure material, we first optimized our synthetic route to ponapensin (3) and elliptifoline (4) using chiral, racemic materials. In initial experiments, several attempts to hydrolyze methyl ester (±)-9a were made. However, in general decomposition resulted under basic or nucleophilic (e.g. LiBr) deprotection conditions (Scheme 5). We next turned our attention to hydrolysis and further advancement of the reduced aglain core structure 9b. Photocycloaddition of 3-HF 5 and benzyl cinnamate 6b led to the production of benzyl ester aglain (±)-8b in 60% yield (Scheme 1). The corresponding endo cycloadduct was then recrystallized from acetonitrile¹¹ and was subsequently reduced to afford the 10S-aglain (±) 9b (10:1 dr). Benzyl ester 9b was hydrogenolyzed and the derived carboxylic acid subsequently coupled with 4,4-dimethoxy-1-butanamine¹⁸ 18 to provide amidoacetal (±)-19 in 95% yield (two steps, Scheme 6). In an alternative pathway, transesterification of methyl ester (±)-9a was achieved using Otera’s catalyst¹⁹ in benzyl alcohol as solvent to afford benzyl ester (±)-9b in 63% yield along with recovered starting material (28%). Several reagents were evaluated in order to achieve direct ester-amide exchange of 9a including Otera’s catalyst, Sc(OTf)₃, Zn₄(OCOCF₃)₆O, Mg(OTf)₂, and Zr(O^tBu)₄ either as catalysts or stoichiometric additives.²⁰ Modest conversions of aglain (±)-9a to (±)-19 were observed using Otera’s catalyst and Mg(OTf)₂ in 25% and 38% yields, respectively. Interestingly, when using trimethylaluminum(III)²¹ for ester amide exchange, we observed clean formation of amido acetal (+)-19 in 64% yield without epimerization of (+)-19 (Scheme 6).

Scheme 5.

Attempted Hydrolysis of 9a

Scheme 6.

Transesterification/ Ester-Amide Exchange toward Ponapensin and Elliptifoline

a) From 9a: Me₃Al (20 equiv), amine 18 (20 equiv) toluene (0.3 M) 70 °C for 12 h (64%); from 9b: i) H₂-Pd/C then ii) EDCI (1.5 equiv), HOBt (1.2 equiv), amine 18 (2.5 equiv) (95%, 2 steps); b) Otera’s catalyst (20 mol%), BnOH, 110 °C, house vacuum, 24 h (64%, recovery 28% 9b).

Final Cyclizations to Ponapensin and Elliptifoline

With enantiopure aglain amide (+)-19 in hand, we next examined cyclization to the hemiaminal of ponapensin (Scheme 7).²² Treatment of amido acetal (+)-19 with CSA (50 mol%) in EtOAc led to a mixture of products that were isolated after two purifications. The major product observed was (+)-ponapensin in 30% yield as a single hemiaminal diastereomer ([α]_D²⁵ = +170° synthetic (c 0.2, CHCl₃), [α]_D²² = −167° natural (c 0.4, CHCl₃)). The other major product isolated was the ethyl hemiaminal derivative of ponapensin, aglain 20, isolated in 12% yield. The production of this compound was due to trace ethanol present in the ethyl acetate. Cyclic intermediate 21 was also isolated in 2% yield and its relative configuration was assigned via NOE correlation observed between H13 and H4 (Figure 4). Cyclic intermediate 21 provides important evidence for the intermediacy of acyliminium²³ 22 and accounts for formation of a single diastereomer of 3. However, the tertiary hydroxyl may not be the only contributing factor to the diastereoselectivity. Figure 5 shows a Spartan ’10 model of the ground state conformation (B3LYP/6-31G**) of putative acyliminium intermediate 22. Trans²⁴ acyliminium 22 may be formed to avoid unfavorable A^1,3 strain interactions²⁵ leading to the conformer depicted in Figure 5. In this conformer, the oxygen of the 6-methoxy group is approximately 2.7 Å from C13 which bears a π* orbital.²⁶ A n → π* interaction appears to stabilize this conformation and allows nucleophilic attack specifically from the opposite face of iminium 22. Furthermore, we also isolated trace amounts of the dihydropyrrole derived from 22 and possible dimeric products from the reaction mixture. Cyclizations were attempted in other solvents such as THF and 1,4-dioxane to avoid formation of 13-OEt derivative 20, in which case ponapensin was still not produced cleanly. After considerable optimization, we found that treatment of (+)-19 with CSA in EtOAc containing methanol (15 equiv.) led to the production of ponapensin (3) in 68% yield (Scheme 8) as a single diastereomer. When methanol was used as solvent, the reaction stopped at approximately 50% conversion to 3 suggesting that an equilibrium had been reached. In a similar manner, treatment of aglain (+)-9a with 5 equivalents of tigloyl amide and 50 mol% CSA in EtOAc led to the production of (+)-elliptifoline (4) as a single diastereomer. Elliptifoline was isolated in 81% yield ([α]_D²⁵ = +172° synthetic (c 0.2, CHCl₃), [α]_D²² = −88.9° natural (c 0.6, CHCl₃)).

Scheme 7.

Acid-Catalyzed Cyclization without Additional Methanol

Figure 4.

B3LYP/6-31G* ground state conformer of cyclic aminal 21 and assigned relative configuration based on the observed 2D NOESY

Figure 5.

B3LYP/6-31G** ground state conformer of acyliminium 22 and stereoselective approach of the nucleophile.

Scheme 8.

Optimized Syntheses of Ponapensin (3) and Elliptifoline (4)

The relative configurations at C13 of both 3 and 4 were confirmed via X-ray crystallography using chiral, racemic natural products to be the same for ponapensin (3) and elliptifoline (4) (Figure 6). The X-ray structures show that the proline amide moiety is folded as a trans-rotamer over the neighboring aromatic ring. A C–H–π interaction may also stabilize this conformation where the phenyl ring donates electron density to the electron deficient C–H bonds of the 13–OMe in ponapensin (3).^24a Kinghorn and coworkers correctly predicted this conformation for ponapensin (3) using molecular modeling and observed the ¹H NMR shielding of the C13 methoxy protons.³ It should be noted that synthetic elliptifoline (4) was determined to have the opposite configuration at C13 than reported by Duh and coworkers (cf. Scheme 8).⁴ The crucial NOE between H4 and H13 was observed in our hands as reported by Duh and coworkers.⁴ The major rotamer in solution is likely the trans conformation displayed in the crystal structure (Figure 6). Although there is a large discrepancy in magnitude between the synthetic and natural samples’ specific rotations (172° vs. 88.9°), the ¹H and ¹³C NMR data support that we have made elliptifoline (4)¹¹ and the configuration at C13 is therefore opposite of what Duh and coworkers reported.⁴

Figure 6.

X-ray crystal structures of the racemic natural products. The structures shown have the absolute configuration of (+)-ponapensin (3) and (+)-elliptifoline (4), enantiomers of the natural products.

As previously described, the specific optical rotations of synthetic (3) and (4) were determined to be dextrorotatory. Accordingly, the enantiomers of both natural products 3 and 4 were synthesized. Figure 7 shows the assigned absolute configurations of the natural products which were further confirmed by the synthesis of enantioenriched (−)-methyl rocaglate (1) from enantioenriched (71% ee) (+)-9a (c.f. Scheme 4).¹¹ We propose that a kinetic resolution may occur in nature wherein (−)-aglain ketone core structures undergo reduction and (+)-aglains do not (Figure 8).^8a,27 Racemic bridgehead ketones have been utilized as substrates for kinetic resolution using microbes such as C. lunata and R. rubra and also in the presence of ketoreductase/NADP.²⁸ (+)-Aglains may then be converted to their (−)-rocaglate counterparts via a reductoisomerase/NADPH pathway (e.g. DXR, ketol-acid reductoisomerase). Keto-acid reductoisomerase is known to catalyze the conversion of 2-acetolactate to 2,3-dihydroxyisovalerate and DXR is involved in terpenoid biosynthesis in E. coli.²⁹ Furthermore, it is also possible that a parallel kinetic resolution (PKR)³⁰ may occur where (+)-aglains selectively undergo ketol rearrangement/reduction and (−)-aglains are reduced via reductoisomerase/NADPH leading to enantiopure (−)-rocaglates and (−)-aglain structures, respectively.

Figure 7.

Relative and absolute stereochemical assignments of (−)-ponapensin (3) and (−)-elliptifoline (4). (−)-Elliptifoline (4) is drawn with its reassigned configuration at C13.

Figure 8.

Proposed kinetic resolution of racemic aglains.

Conclusion

In summary, we have employed enantioselective [3+2] photocycloaddition methodology to synthesize the aglain natural products (+)-ponapensin and (+)-elliptifoline. Diastereoselective, reagent-controlled reduction of the hydrated bridgehead ketone led to an intermediate with the correct S configuration at C10. Subsequent ester-amide exchange and acid-catalyzed cyclization provided the natural products. These expedient syntheses demonstrated that late stage installation of the C13 hemiaminal and aminal stereocenters may be substrate controlled by stereoelectronic stabilization of an acyliminium species. This may explain why the relative stereochemical outcome at C13 is similar for both (−)-ponapensin and (−)-elliptifoline. The unexpected finding that (−)-ponapensin and (−)-elliptifoline are enantiomeric to (−)-methyl rocaglate has led to the hypothesis for biosynthesis involving kinetic resolution of aglain ketone core structures. Further work toward development of enantioselective photocycloadditions as well as syntheses and biological evaluations of additional aglain natural products are currently in progress and will be reported in due course.