Asymmetric Dearomatization/Cyclization Enables Access to Polycyclic Chemotypes

Abstract

Enantioenriched, polycyclic compounds were obtained from a simple acylphloroglucinol scaffold. Highly enantioselective dearomatization was accomplished using a Trost ligand-palladium(0) complex. A computational DFT model was developed to rationalize observed enantioselectivities and revealed a key reactant-ligand hydrogen bonding interaction. Dearomatized products were used in visible light-mediated photocycloadditions and oxidative free radical cyclizations to obtain novel polycyclic chemotypes including tricyclo[4.3.1.0^1,4]decan-10-ones, bicyclo[3.2.1]octan-8-ones and highly-substituted cycloheptanones.

Enantioenriched, polycyclic compounds were obtained from a simple acylphloroglucinol scaffold. Highly enantioselective dearomatization was accomplished using a Trost ligand- palladium(0) complex. Dearomatized products were used in visible light-mediated photocycloadditions and oxidative free radical cyclizations to obtain novel polycyclic chemotypes.

Introduction

Polyprenylated polycyclic acylphloroglucinols (PPAPs) are a natural product class well-known for their diverse biological activities,^[1] and their complex structures have attracted the attention of many chemists in the field of natural product total synthesis.^[2] We have previously shown that dearomative alkylation at the 6-position of 4,6-diallyl-2-benzoyl-5-O-methyl-phloroglucinol (1) enables the synthesis of PPAPs such as 7-epi-nemorosone (2)^[3] and plukenetione A (3)^[4] (Figure 1).

Figure 1.

Natural products derived from acylphloroglucinol scaffold 1

Diversity-oriented synthesis (DOS) enables the assembly of small molecule screening libraries with high sp³ carbon and chiral center content for use in drug discovery.^[5][6] In order to access complex compound libraries in minimal steps from simple starting materials, the development of efficient, complexity-generating synthetic methodology is critical. Herein, we report a new strategy for generation of enantioenriched, polycyclic compounds via two-step dearomatization/cyclization sequences. Asymmetric, dearomative cinnamylation of simple benzoyl phloroglucinol 1 produces chemotype 4 bearing multiple olefinic functional handles that can be exploited for subsequent intramolecular cyclizations to produce novel polycyclic chemotypes (Figure 2).

Figure 2.

Asymmetric cinnamylation of 1 and further cyclizations of 4 gives rise to novel chemotypes

Results and Discussion

We first evaluated dearomative, asymmetric cinnamylation^[7] of substrate 1 using Trost ligand-Pd⁰ complexes (Table 1). Treatment of 1 with cinnamyl carbonate 5a, (S,S)-DACH-phenyl Trost ligand (L1), Cs₂CO₃, and Pd₂(dba)₃^[8] afforded the cinnamylated product 4a in 72% yield and 99 % ee (Entry 1). The absolute stereochemistry of 4a was determined by X-ray crystallography of the derived p-bromobenzoate 6a (Figure 3).^[9] Use of (R,R)-DACH-phenyl Trost ligand (L2) afforded the opposite enantiomer (ent-4a, not shown) in 81% yield (Entry 2). The DPEDA-type Trost ligand L3 gave a slightly lower yield and % ee than was observed with L1 (Entry 3). (S,S)-DACH-naphthyl ligand (L4) was not effective for the reaction (Entry 4). Use of the (S,S)-ANDEN-phenyl type ligand (L5) provided the opposite enantiomer in 44% yield and 24 % ee (Entry 5).

Table 1.

Asymmetric Cinnamylation: Ligand Screening

Entry	Ligand	Yield[a] [%]	ee[b] [%]
1	(S,S)-DACH-phenyl Trost (L1)	72	99
2	(R,R)-DACH-phenyl Trost (L2)	81	-97
3	(S,S)-DPEDA-phenyl Trost (L3)	63	86
4	(S,S)-DACH-naphthyl Trost (L4)	0	--
5	(S,S)-ANDEN-phenyl Trost (L5)	44	-24

^[a]Yields of isolated products.

^[b]Determined by HPLC analysis.

Figure 3.

Determination of absolute stereochemistry of p-bromobenzoate 6a

We next investigated substrate scope (Table 2). Both naphthyl- (5b) and 4-bromophenyl-substituted (5c) substrates were well tolerated. Reactions of carbonates with electron-donating groups such as 4-methoxyphenyl (5d) or 2-furanyl (5e) were more sluggish, requiring 48 h for full consumption of 1. Enantiopurities of the products (4b-e) were high (91-99 % ee). Conversely, 4-nitrophenyl substrate (5f) showed only moderate enantioselectivity. Altering the substitution at the allylic double bond completely prevented the reaction (g and h).

Table 2.

Asymmetric Cinnamylation: Substrate Scope

Entry	Substrate	R’	Product	Yield[a] [%]	ee[b] [%]
1	5b		4b	96	97
2	5c		4c	87	98
3	5d		4d	69[c]	91
4	5e		4e	79[d]	99
5	5f		4f	77	41
6	5g		4g	0	-
7	5h		4h	0	-

^[a]Yields of isolated products.

^[b]Determined by chiral HPLC analysis.

^[c]48 h reaction time.

Our provisional model to rationalize enantioselectivity is depicted in Figure 4. For Tsuji-Trost allylic alkylations involving prochiral π-allyl complexes, a wall/flap cartoon mnemonic aids in predicting stereochemical outcomes.^[10] This model has also been used to predict the facial discrimination of simple prochiral enolates.^[10b-d] Lloyd-Jones, Norrby, and coworkers have reported computational and NMR studies indicating that, in contrast to the C₂-symmetric geometry assumed in the wall/flap model, experimental evidence suggests that the allyl group is situated remotely from the DACH ligand in a complex that is not symmetric on the NMR timescale.^[11] Given the complexity of 1 and the asymmetry of the cinnamyl cation, we relied on computational modeling with DFT-optimized L1-Pd-η3-cinnamyl complexes using the Lloyd-Jones/Norrby geometry. DFT optimizations of pro-R and pro-S outer-sphere approaches of the naked anion of 1 (a hydrogen-bond stabilized exocyclic enolate)^[12] in toluene indicated that for each input complex, the pro-R (observed) approach was energetically favored.^[13] In the lowest-energy complex, L1 participates in a hydrogen bond with 1, as well as a π-stacking interaction with the cinnamyl cation. A similar hydrogen bond was observed in Lloyd-Jones’ transition state calculations of cesium malonate with an L2-Pd-η3-allyl complex, giving rise to a quasi-inner sphere transition state wherein enantioselectivity is governed by pre-coordination of the nucleophile to the ligand rather than the Pd atom.^[14]

Figure 4.

DFT-optimized (B3LYP-LACV3P**++) structure (Schrödinger’s Jaguar) of the lowest-energy L1-Pd-η3-cinnamyl-enolate approach indicating a hydrogen bond between the incoming enolate and ligand N-H

With enantioenriched dearomatized cores bearing multiple olefinic moieties in hand, we next pursued oxidative cyclization reactions as a strategy to rapidly generate structural complexity. We first explored visible light photoredox catalysis.^[15] Knowles and coworkers have shown that visible light catalysts can generate heteroatom radicals by proton-coupled electron transfer (PCET), and that the radicals may be trapped by alkenes to access heterocycles.^[16] Inspired by this approach, we postulated that the enol 4 may be an excellent substrate for radical generation at the 2-position, with subsequent trapping by the pendant cinnamyl group.

In the event, compound 4a was converted to three different chemotypes on treatment with Ir(dFCF₃ppy)₂(dtbbpy)PF₆ ("Ir catalyst") under blue LED irradiation (Table 3, Entry 1). After detailed NMR analysis, it was determined that the expected photoredox product (9a) was produced in 11% yield, as well as tricycles 7a and 8a in 22% and 13% yields, respectively. These compounds presumably arise via [2+2] photocyclization (De Mayo reaction)^[17] between the styrenyl alkene and exo-enol, with the Ir catalyst serving as photosensitizer.^[18] In contrast, 2-naphthyl enol (4b) did not afford 8b and 9b, presumably due to steric hindrance (Entry 2). The 4-bromophenyl (4c) and 4-nitrophenyl (4e) substrates were slow to react (Entries 2 and 5). 4-Methoxyphenyl-substituted 4d reacted smoothly with a product distribution similar to that observed with 4a.

Table 3.

Photoreactions of cinnamylated substrates 4^[a]

Entry	Substrate	Time [h]	Yield [%] of[a]
			7	8	9
1	4a	24	22 (7a)	13 (8a)	11 (9a)
2	4b	24	20 (7b)	-	-
3	4c	48	17 (7c)	13 (8c)	5 (9c)
4	4d	6	27 (7d)	13 (8d)	11 (9d)
5	4e	48	23 (7e)	-	-

^[a]Yields of isolated products.

Subsequent in situ NMR experiments indicated that prior to cyclization to 7a/8a, clean photoisomerization of the styrene of 4a occurs, suggestive of triplet energy transfer. Identical E/Z ratios were obtained irrespective of the starting isomer (Scheme 1). Based on these observations, a mechanistic proposal for the formation of [2+2]-cycloadducts and the cyclization product 9a is shown in Scheme 2. First, the styrenyl moiety (ca. 60 kcal/mol triplet energy)^[17a] is sensitized by the visible light-excited Ir catalyst (61 kcal/mol triplet energy).^[11a] The exo-enol is then excited through intramolecular triplet energy transfer from the excited styrene and reacts via [2+2] cycloaddition with the trans-cinnamyl group to afford compounds 7a and 8a. The conformer enabling [2+2] cycloaddition with the cis-cinnamyl substrate is likely disfavored due to allylic strain. Instead, cis-cinnamyl isomer 4a′ is converted to [3.2.1]-bicycle 9a via a redox process.

Scheme 1.

Photoisomerization of the styrene moieties of 4a and isomer 4a′

Scheme 2.

Mechanistic proposal for formation of cyclobutanols 7a/8a and [3,2,1]-tricycle 9a

In order to further probe the intramolecular reactivity of 4, we also investigated oxidative free radical cyclization using Mn(OAc)₃/Cu(OAc)₂.^[19] While treatment with Mn^III and Cu^II gave intractable mixtures, treatment of 4a with Mn(OAc)₃ in acetic acid produced acetate 10a in 19% yield (Table 4, Entry 1). Scheme 3 depicts a proposed mechanism. The radical generated at the 2-position cyclizes with the cinnamyl group positioned to minimize torsional strain with the benzoyl group in the 5-centered transition state. The resultant benzyl radical may undergo one electron oxidation by Mn^III to afford a benzylic cation, stabilized as the oxonium by the adjacent benzoyl carbonyl, which may be trapped by acetic acid with stereochemical inversion to afford 10a. 4-Bromophenyl derivative 4c similarly gave 10c (Entry 2). A competing pathway, in which the cinnamyl alkene undergoes 5-exo-trig cyclization with the 1-O-radical, was observed with 4-methoxy-substituted substrate 4d. We propose that the electron-donating group enables further Mn^III-mediated oxidation of the resultant benzylic radical to a stabilized benzyl cation which may then trapped by acetate to form 11 (Entry 3).^[13]

Table 4.

Oxidative free radical cyclization of cinnamylated substrates 4^[a]

Entry	4	R	Yield of[a]
			10 [%]	11 [%]
1	4a	Ph-	19 (10a)	-
2	4c	4-BrPh-	20 (10c)	-
3	4d	4-MeOPh-	24 (10d)	51 (11d, dr = 1:1)

^[a]Yields of isolated products.

Scheme 3.

Proposed mechanism for formation of [3.2.1]-bicycle 10

We next explored further reactivity of the dearomatization/cyclization-derived polycycles. First, [3.2.1]-bicycle 10a was treated with LiOH in methanol to give the highly substituted 7-membered ester 12a via a retro-Dieckmann process (Scheme 4A). Compound 7a was subjected to similar conditions, producing decarbonylated cycloheptanone 13a as a single diastereomer in 53% yield (Scheme 4B). Next, compound 7a was treated with the non-nucleophilic base NaH. Under these conditions, bicyclo-[3.2.1] derivative 14a was obtained in 20% yield along with 22% of recovered 7a. This product may presumably be obtained via protonation of an anti-Bredt enolate carbanion^[20] after retro-aldol cleavage of the cyclobutanol. [3.2.1]-Bicycle 14a was converted to 13a in 38% yield upon LiOH treatment (Scheme 5). We propose that the one-pot conversion of 7a to 13a may also arise via a retro-aldol–retro-Dieckmann sequence. Alternatively, a mechanism in which retro-Dieckmann fragmentation precedes retro-aldol reaction may be operative. In either case, the retro-Dieckmann fragmentation for 14a occurs orthogonally to that observed for 10a, presumably due to the absence of a benzoyl group stabilizing the resulting anion.

Scheme 4.

Fragmentation of compounds 10a (A) and 7a (B)

Scheme 5.

Further reactions of scaffold 7a

Conclusions

In summary, we have established a new strategy for generating unique enantioenriched mono- and polycyclic chemotypes. Highly enantioselective cinnamylation of a phloroglucinol scaffold provided dearomatized substrates which were converted by divergent cyclizations to structures including tricyclo[4.3.1.0^1,4]decan-10-ones and bicyclo[3.2.1]heptanones. The polycyclic chemotypes were ring-opened to afford highly substituted 7-membered products. Preliminary computational studies indicated the potential for interesting pre-organizing intramolecular interactions between the ligand/Pd⁰/substrate complex and the incoming enolate nucleophile. Studies exploring further intramolecular transformations of the dearomatized scaffolds, as well as rigorous computational examination of the asymmetric allylic alkylation using transition state calculations, are currently in progress and will be reported in due course.

Experimental Section

Full experimental details for the preparation of the compounds described herein, as well as details related to computational and mechanistic experiments, are provided in the Supporting Information.