An Enantioselective Synthesis of the ABD Tricycle for (−)-Phomactin A Featuring Rawal’s Asymmetric Diels-Alder Cycloaddition

Abstract

An enantioselective synthesis of the ABD-ring of (−)-phomactin A is described here. The sequence features Rawal’s asymmetric Diels-Alder cycloaddition. The overall length is significantly reduced from our previous attempt.

We have been pursuing the total synthesis of phomactin A, a biologically interesting and structurally unique natural product from the culture filtrate of a parasitic fungus Phoma sp. [SANK 11486] found on the shell of Chinoecetes opilio off the coast of Fukui prefecture in Japan.^1–3 In the last 15 years, phomactin A and its other family members have attracted a number of impressive synthetic efforts^4,5 including Yamada’s synthesis of (+)-phomactin D,⁶ total syntheses of (±)- and (+)-phomactin A by Pattenden⁷ and Halcomb,⁸ respectively, and most recently, Wulff’s total synthesis of (±)-phomactin B2.⁹ We¹⁰ disclosed our synthetic approach to (±)-phomactin A potentially through ABD tricycle 1,¹¹ which can be accessed from an intramolecular oxa-[3 + 3] annulation of vinyl iminium salt 2.^12–14

However, our total synthesis efforts have been seriously compromised by the fact that the annulation precursor 2 required 22 steps from 2-methyl-cyclohexanone, adopting a route not amenable for asymmetric total synthesis. Therefore, we have been exploring a conceptually different strategy. Specifically, by constructing the A-ring from a Diels-Alder cycloaddition instead of buying from 2-methyl-cyclohexanone, we hope to establish a much shorter and asymmetric synthesis of 1. We report herein an enantioselective synthesis ABD-tricycle for (−)-phomactin A featuring Rawal’s asymmetric Diels-Alder cycloaddition.

Our initial efforts quickly identified that Diels-Alder cycloaddition of tiglic aldehyde with Danishefsky’s diene proceeded only at high temperatures [Scheme 2]. After acidic workup, this cycloaddition afforded, with variable yields, a complex mixture of products 3–5 including para-hydroxy-acetophenone 5,¹⁵ likely derived from dimerization of the diene. On the other hand, cycloaddition with Rawal’s diene^16–20 did occur with remarkable ease. Although the reaction was slow at −10 °C [ice/acetone], at 0.22 mol scale, it was complete within 20 h at 50 °C in THF to afford the desired cycloadduct (±)-6 in high yields as a single endo isomer.

Scheme 2.

Danishefsky’s diene versus Rawal’s diene.

It is noteworthy that the better reactivity towards tiglic aldehyde is likely a consequence of the dimethylamino group possessing stronger donating ability into the diene [significantly raising its HOMO level] relative to Danishefsky’s diene.¹⁸ This key endo-selective cycloaddition sets up a majority of the stereochemistry in the A-ring, and provides extensive oxygenation for subsequent transformations. However, attempts to append a tethering fragment through nucleophilic addition to the aldehydic motif in (±)-6 were not fruitful, and (±)-6 was rather unstable to chromatography, and underwent clean elimination to give the enone under acidic conditions. With the knowledge of Halcomb’s success in adopting Suzuki-Miyaura coupling,⁸ we proceeded to olefinate the aldehyde in (±)-6, and after hydrolysis, enone (±)-7 was attained in 72% yield [Scheme 2]. A nucleophilic epoxidation and ketone protection gave vinyl epoxy acetal (±)-8, which was used for investigating the Suzuki-Miyaura coupling [vide infra].

We then turned our attention to pursuing an asymmetric Diels-Alder cycloadditions of tiglic aldehyde with the Rawal’s carbamate diene 9^18–22 [Scheme 3]. Screening Corey’s oxazaborolidine catalyst 12²¹ only led to 1,4-addition and 1,2-addition products 16a and 16b.²³ Rawal¹⁹ had reported a related cycloaddition of tiglic aldehyde that gave 62% yield with 93% ee when using Jacobsen’s Cr(III)-salen catalyst²⁴ [see 13 with SbF₆ as counter anion]. However, our attempts using diene 9 again resulted 16a and 16b with the best yield for the desired cycloadduct 10 being 36%. The use of Co(III)-salen catalyst 14 was also not useful and gave products related to 1,4-addition and 1,2-addition.²³ Ultimately, we were able to obtain 10 in 78% yield with ee up to 90% by using the (R,R)-Cr(III)-salen catalyst 15 with BF₄ as counter anion. The ee was determined by making the (S)-naproxen ester 11 [not crystalline and so no X-ray structure could be attained].

Scheme 3.

Rawal’s asymmetric Diels-Alder cycloaddition.

With optically enriched cycloadduct 10 in hand, we had trouble transforming it into enone (±)-7. As shown in Scheme 4, unlike that of 6, Wittig olefination of 10 was <10% yield. While a related DA-cycloadduct 17 [R = allyl, only 74% ee] gave improved yield for the olefination, and while hydrolysis to alkene 19 was fast, the elimination of resulting ketone 20 to (±)-7 was neither efficient nor clean. Subsequently, we examined the possibility of first pursuing hydrolysis/elimination sequence. However, while treating 10 with HF or HF/TFA quantitatively gave 21, the elimination was again slow and sluggish, leading to enone (+)-3 in low yields accompanied with significant decompositions.

Scheme 4.

Struggles in accessing enone (+)-7.

Intriguingly, only after we reduced the aldehyde using LAH, the hydrolysis/elimination sequence could be achieved to cleanly give alcohol (−)-22. However, subsequent Dess-Martin periodinane [DMP] oxidation and Wittig olefination were again disastrous in an attempt to synthesize (+)-7. On the other hand, alcohol (−)-22 proved to be useful in accessing vinyl epoxy acetal (+)-8 through epoxy acetal (−)-23 in an overall of 4 steps [Scheme 5]. LAH-reduction of the epoxide in 8 followed by TES-protection led to vinyl acetal (+)-24 in 87% overall yield.

Scheme 5.

A suitable path to vinyl epoxy acetal (+)-8.

At this point, although we have already examined Suzuki-Miyaura coupling of (±)-8 with vinyl bromide 25,¹⁵ it was not very desirable. Thus, we focused on the cross-coupling of vinyl acetal (+)-24 with 25 [Scheme 6]. Upon closer examination of the reaction mixture, we were disturbed to find significant quantities of the unexpected olefin 27 [E/Z unassigned] in addition to the desired Suzuki-Miyaura cross-product (−)-26. It was extremely difficult to separate 27 from (−)-26.²⁵

Scheme 6.

A strange Suzuki-Miyaura coupling.

We were certain that this compound was not coming from contamination during the preparation of vinyl bromide 25.¹⁵ The use of exotic bases such as Cs₂CO₃, Tl₂CO₃, Ag₂CO₃, Ag₂O, or Tl(OEt)₂,²⁴ often critical to the success of these reactions, did not seem to help matters. After trying the PdCl₂dppf and finding it produced a larger amount of by-product 27, We stuck to our traditional mode of employing PdCl₂(PPh₃)₂ for the reaction. While we remain uncertain of the mechanism for this unusual crossed-product, it appears that the addition order mattered. After the 9-BBN hydroboration had occurred, a respective base must be added first to the borane solution before the addition of the palladium/vinyl bromide solution. The most effective catalyst was Pd(PPh₃)₄, and when it was used at 4.5 mol% along with 2.0 M aq KOH, the formation of 27 was just barely observable by ¹H NMR.

Resolving this quagmire in the Suzuki-Miyaura coupling allowed us to complete an asymmetric synthesis of the ABD-tricycle. As shown in Scheme 7, desilylation followed by DMP oxidation led to acetal enal (+)-28, and a careful hydrolysis of acetal (+)-28 with 1.0 N H₂SO₄ gave the annulation precursor (−)-29. Under the oxa-[3 + 3] annulation conditions,^10,13 ABD tricycle (+)-1 was attained in 30% yield²⁷ in an overall 13 steps from diene 9. The absolute configuration at this point is solely based on Rawal’s assignment of related asymmetric Diels-Alder cycloadducts.

Scheme 7.

Synthesis of ABD-ring (+)-1 for (−)-phomactin A.

We have described here an enantioselective synthesis of the ABD-ring for constructing (−)-phomactin A. The synthetic sequence features Rawal’s asymmetric Diels-Alder cycloaddition. The overall length of this new approach is significantly reduced from our previous attempt. Completion of a total synthesis of (−)-phomactin A is in progress.

EXPERIMENTAL SECTION

Asymmetric Diels-Alder

To a solution of tiglic aldehyde (3.95 mL, 41.0 mmol, 2.0 equiv) in CH₂Cl₂ (20 mL) were added 4 Å molecular sieves (16.5 g), (R,R)-Cr(III)-Salen-BF₄ (697.0 mg, 1.02 mmol, 5 mol%) and then diene 9 (7.12 g, 20.5 mmol, 1.0 equiv) and the reaction was stirred at rt for 2 d at rt. The reaction mixture was filtered through a plug of silica gel. The filtrate was concentrated to give a light brown oil. This crude oil was purified via silica gel flash column chromatography (gradient eluent: 5–10% EtOAc in hexanes) to afford the cycloadduct Diels-Alder product 10 (6.90 g) in 78% yield as a yellow oil. 10: R_f = 0.56 (25% EtOAc in hexanes); [α]²³_D = + 48.0 [c 2.50, CHCl₃];¹H NMR (400 MHz CDCl₃) δ −0.15 (s, 3H,), −0.06 (s, 3H), 0.67 – 0.93 (m, 12H), 1.16 (s, 3H), 1.69 – 1.77 (m, 1H), 2.11 – 2.17 (m, 1H), 2.23 – 2.46 (m, 1H), 3.57 (s, 3H), 4.45 (s, 1H), 4.58 – 4.61 (m, 2H), 4.84 (s, 1H), 7.08 – 7.23 (m, 5H), 9.58 (s, 1H); ¹³C NMR (125 MHz, CDCl₃) δ −4.6, 14.1, 16.6, 18.1, 25.8 28.4, 35.0, 48.5, 53.0, 53.6, 59.7, 99.5, 126.2, 126.7, 128.3, 139.8, 154.8, 158.0, 205.9; IR (neat) cm⁻¹ 2956w, 2931w, 2859w, 1721w, 1692w, 1670w, 1451w, 1393m, 1373m, 1219w, 1201w, 1169m, 835w, 780m; mass spectrum (APCI): m/e (%relative intensity) 267 (13) (M+1)⁺, 432 (100).

LAH Reduction and HF-Hydrolysis

To a solution of the Diels-Alder cycloadduct 10 (6.89 g, 16.0 mmol) in 160 mL of Et₂O at 0 °C was added LAH (1.21 g, 32.0 mmol) portion wise. The reaction was then warmed to rt and stirred overnight. The reaction was cooled to 0 °C, and H₂O (1.21 mL), 10% aq NaOH (1.21 mL) and H₂O (3.63 mL) were added slowly in that order. The mixture was stirred for 1–2 h, and then the white precipitate was formed. The white solid was removed by filtration. The filtrate was concentrated under reduced pressure to give a pale yellow oil.

The above crude oil was taken up in CH₃CN (32 mL) and to this solution was added a HF solution (1.10 mL, 48%, 29.0 M, 32.0 mmol) and the reaction was stirred at rt overnight. The reaction solution was neutralized by addition of aq Na₂CO₃ until the CO₂ bubbles stopped forming. The solid (mostly NaF) was removed by filtration and the filtrate was concentrated in vacuo. The crude oil was purified via silica gel flash column chromatography eluting with 10% EtOAc in hexanes to give alcohol (−)-22 (2.26 g), as a pale yellow oil, in 91% for two steps from the Diels-Alder cycloadduct 10. Upon sitting, colorless crystals can be formed from crystallization. (−)-22: R_f = 0.19 (40% EtOAc in hexanes); [α]²³_D = −18.0 [c 2.50, CHCl₃]; ¹H NMR (500 MHz, CDCl₃) δ 0.97 (s, 3H), 0.98 (d, 3H, J = 7.0 Hz), 1.73 (brs, 1H), 2.29 (dd, 1H, J = 13.5, 17.0 Hz), 2.36–2.44 (m, 2H), 3.59 (d, 1H, J = 11.0 Hz), 3.62 (d, 1H, J = 11.0 Hz), 6.01 (d, 1H, J = 10.0 Hz), 6.82 (d, 1H, J = 10.0 Hz); ¹³C NMR (500 MHz, CDCl₃) δ 15.2, 15.5, 31.8, 41.6, 41.8, 68.1, 128.7, 158.1, 200.1; IR (neat) cm⁻¹ 3448s, 2965m, 1662s; mass spectrum (APCI): m/e (%relative intensity) 155 (100) (M+1)⁺.

Naproxen Ester Formation for Analyzing Isomeric Ratios

To a solution of give alcohol (−)-22 (10.0 mg, 0.067 mmol) in CH₂Cl₂ (2 mL) were added (S)-naproxen (15.0 mg, 0.067 mmol), DMAP (0.080 mg, 0.0067 mmol), and DCC (14.0 mg, 0.067 mmol). The resulting mixture was stirred for at rt for 18 h. The reaction mixture was simply filtered through a Celite^™ plug and the clear filtrate was concentrated in vacuo to give the crude ester 11 as a solid that was subjected to crude NMR analysis for the assessment of diastereomeric ratios. 11: R_f = 0.62 (40% EtOAc in hexanes); ¹H NMR (400 MHz, CDCl₃) δ 0.80 (d, 3H, J = 6.6 Hz), 0.91 (s, 3H), 1.54–1.61 (m, 1H), 1.57 (d, 3H, J = 6.8 Hz), 1.66–1.71 (m, 1H), 1.88–1.96 (m, 1H), 1.99–2.17 (m, 1H), 2.17 (d, 2H, J = 9.2 Hz), 3.91 (s, 3H), 4.03 (d, 2H, J = 3.6 Hz), 5.85 (d, 1H, J = 10.4 Hz), 6.59 (d, 1H, J = 10.4 Hz), 7.10–7.15 (m, 2H), 7.35 (dd, 1H, J = 1.6, 8.0 Hz), 7.63–7.68 (m, 3H).

The Oxa-[3 + 3] Annulation: Asymmetric Synthesis of ABD-Ring (+)-1

To a solution of enal (−)-29 (107.0 mg, 0.35 mmol) in anhyd EtOAc (15 mL - distilled from CaH₂) was added anhyd piperidine (42.0 μL, 0.422 mmol, 1.2 equiv – distilled from and stored with CaH₂) at −10 °C. The solution was stirred to −10°C for 5 min, and then, acetic anhydride (200.0 μL, 2.11 mmol, 6.0 equiv) was added carefully dropwise. The resulting mixture was stirred at rt for 3–4 h before it was washed with equal volume of 1.0 N aq NaOH. The organic layer was dried over Na₂SO₄, filtered, and concentrated in vacuo. The crude product was purified using silica gel flash column chromatography (gradient eluent: 0–25% EtOAc in hexanes) to give the desired ABD-tricycle (+)-1 (32.0 mg) in 30% yield along with byproduct 30a/b (27.0 mg) in 25% yield. (+)-1: R_f = 0.41 (10% EtOAc/hexanes); [α]²³_D = + 95 [c 0.64, CHCl₃]; ¹H NMR (500 MHz, CDCl₃) δ 6.45 (d, 1 H, J = 10.5 Hz), 5.15 (brd, 1 H, J = 10.5 Hz), 4.99 (d, 1 H, J = 10.5 Hz), 2.87 (dd, 1 H, J = 6.5, 20.0 Hz), 2.29–2.40 (m, 2 H), 2.12 (m, 1 H), 2.10 (d, 1 H, J = 19.5 Hz), 2.06 (ddd, 1 H, J = 1.5, 7.5, 14.5 Hz), 1.94 (dt, 1 H, J = 3.5, 14.0 Hz), 1.94 (ddd, 1 H, J = 2.0, 4.0, 14.5 Hz), 1.89 (ddd, 1 H, J = 3.5, 12.5, 15.0 Hz), 1.80 (td, 1 H, J = 3.5, 14.5 Hz), 1.70 (ddd, 1 H, J = 3.0, 6.0, 15.0 Hz), 1.47 (s, 3 H), 1.38 (brm, 3 H), 1.07 (s, 3 H), 1.02 (d, 3 H, J = 7.5 Hz); ¹³C NMR (75 MHz, CDCl₃) δ 198.6, 169.1, 134.6, 125.3, 119.8, 118.0, 116.0, 82.9, 47.1, 46.1, 39.5, 37.0, 35.7, 34.5, 31.3, 25.2, 24.6, 20.0, 16.1; IR (neat) cm⁻¹ 2976m, 2929m, 2901m, 2835w, 1640vs, 1607s, 1449m, 1416s; mass spectrum (ESI) m/e (% relative intensity) 309.2 (M + Na)⁺ (100), 197.0 (13); m/e calcd for C₁₉H₂₆O₂Na (M + Na)⁺ 309.1830, found 309.1818 [δ = 3.89 ppm].

Scheme 1.

A synthetic approach to phomactin A.