Synthetic efforts toward the bicyclo[3.2.1]octane fragment of rhodojaponin III

Abstract

Rhodojaponin III is a grayanane-type diterpenoid natural product with a novel chemical scaffold. It shows potent antinociceptive activity and may represent a new class of natural non-opioid analgesics with a novel mode of action. We explored the Au(I)-catalyzed Conia-ene cyclization and the Mn(III)-mediated radical cyclization of alkynyl ketones for the synthesis of the bicyclo[3.2.1]octane fragment of rhodojaponin III. These strategies will be applicable in the synthesis of rhodojaponin III and analogs for future biological studies.

1. Introduction

Pain is “an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage”.¹ Pain reduces the quality of life and imparts high health costs and economic loss to society. Current pain management heavily relies on analgesic medications. Analgesics mainly target the enzymatic cascade related to inflammatory processes (e.g., nonsteroidal anti-inflammatory drugs) or the endogenous opioid system (e.g., opioids). However, they exhibit limited efficacy, unwanted side effects, and drug abuse problems.² In particular, the United States is experiencing a nationwide public opioid crisis that continues to escalate. According to the US Department of Health & Human Services, more than 130 people died every day from opioid-related drug overdoses in 2016 and 2017.³ Therefore, to overcome this serious socioeconomic issue, the search for new antinociceptive compounds that are effective for both acute and chronic pain and do not produce tolerance or dependence is important.

Among the plants clinically used to relieve pain in China, Rhododendron molle G. Don (Ericaceae) is one of the most potent medicines for pain management.^4,5 It has been traditionally used as an anodyne and anesthetic. Of the natural products isolated from Rhododendron molle, a grayanane-type diterpenoid, rhodojaponin III (1, Fig. 1), showed significant antinociceptive activity in an acetic acid-induced writhing test (74% inhibition of the writhing events at 0.08 mg/kg).⁶ Rhodojaponin III (1) was also more potent than morphine in both acute and inflammatory pain models and 100-fold more potent than gabapentin in a diabetic neuropathic pain model.⁶ More importantly, naloxone showed no significant effect on analgesia induced by rhodojaponin III (1), suggesting that the endogenous opioid peptidergic system is not involved in the antinociceptive activity of 1.⁶ Taken together, rhodojaponin III (1) may represent a new class of natural non-opioid analgesics. Due to the important biological activities of rhodojaponin III (1) and structurally related grayanane-type natural products,⁷ a considerable amount of effort has been made to establish an efficient synthetic approach to this class of natural products,^8–10 culminating in the first total synthesis of grayanotoxin III by Masuda, Shirahama and co-workers in 1994.¹⁰

Fig. 1.

Structure and retrosynthesis of rhodojaponin III (1).

We were intrigued by the great potential of rhodojaponin III (1) for a novel non-opioid analgesic and a chemical probe for identification of novel drug targets for pain management. Herein, we describe a stereoselective synthesis of the bicyclo[3.2.1]octane moiety of rhodojaponin III (1), enlisting the intramolecular radical-mediated cyclization of alkynyl ketones.

2. Results and Discussion

As illustrated in Fig. 1, a convergent synthesis of rhodojaponin III (1) could be realized by coupling bicyclo[3.2.1]octane 3 and cyclopentane 2. Towards this goal, we embarked on a stereoselective construction of 3. We envisioned that the stereoselective synthesis of 3 could be achieved via the Conia-ene-type reaction of alkynyl ketone 4 which could be prepared from α,β-unsaturated cyclic enone 5.

There have been reports on stereoselective syntheses of structurally similar bicyclo[3.2.1]octanes. For example, the first total synthesis of principinol D by Newhouse and co-workers features a Ni-catalyzed α-vinylation reaction for the bicyclo[3.2.1]octane fragment of principinol D.¹¹ Ding and co-workers explored a Ti(III)-mediated reductive epoxide-opening/Beckwith–Dowd rearrangement process to efficiently assemble the bicyclo[3.2.1]octane framework of rhodomolleins XX and XXII.¹² More recently, Jia and co-workers reported the investigation of a radical-mediated cyclization of alkynyl ketones for the synthesis of the bicyclo[3.2.1]octane framework of (−)-glaucocalyxin.¹³

In order to explore the feasibility of the proposed approach to bicyclo[3.2.1]octane 3, our synthesis began with the preparation of alkynyl ketone 4 for intramolecular cyclization. The conjugate addition of vinylmagnesium bromide in the presence of CuBr•SMe₂ and TMSCl to α,β-unsaturated cyclic enone 5, which was prepared by PMB protection of the known 2-(2-hydroxyethyl)cyclohex-2-en-1-one,¹⁴ afforded TMS enol ether 6 in 76% yield (Scheme 1).¹⁵ Propargylation of 6 via the resulting lithium enolate generated by MeLi provided the desired alkynyl ketone 4 (37%) as well as the C2-epi-diastereomer 7a (35%).¹⁶ In addition to C-propargylation products, we also observed the formation of O-propargylation product 7b (15%). X-ray crystallography was used to establish the structure and stereochemistry of the propargylation products.¹⁷

Scheme 1.

Preparation and initial attempts for the Conia-ene-type cyclization of alkynyl ketone 4. Reagents and conditions: (a) vinylmagnesium bromide, CuBr•SMe₂, TMSCl, HMPA, THF, −78 °C, 3 h, 76%; (b) MeLi•LiBr, THF, 0 °C, 1 h, then, CHCCH₂Br, HMPA, −78 to 25 °C, 16 h, 37% (rac-4), 35% (rac-7a), 15% (rac-7b); (c) KOt-Bu, DMSO, 25 °C, 1.5 h, 23%; (d) Et₃N, TBSCl, NaI, MeCN, reflux, 16 h, 84%; (e) [(CyJohnPhos)Au(MeCN)]SbF₆ (0.5 equiv), acetone, 45 °C, 4 h, 58%.

Having prepared alkynyl ketone 4, we turned our attention to the crucial Conia-ene-type cyclization for the construction of the bicyclo[3.2.1]octane fragment of rhodojaponin III (1). The intramolecular addition of enols to alkynes or alkenes represents one of the most powerful and widely employed methods for the formation of carbon-carbon bonds, known as the Conia-ene reaction.¹⁸ Originally, Conia and co-workers studied the thermal intramolecular cyclization of ketones onto alkynes leading to valuable five- or six-membered carbocycles. However, high temperature is required for this reaction to occur, and many functional groups are not compatible with these restrictive reaction conditions. To overcome this shortcoming, Trauner and co-workers reported a Conia-ene cyclization of unactivated alkynes under basic conditions.¹⁹ Unfortunately, alkynyl ketone 4 did not undergo the desired cyclization under basic conditions. Instead, 4 was isomerized to the corresponding internal alkyne 8 (Scheme 1).²⁰

The Conia-ene cyclization catalyzed by transition metals has been widely employed for the formation of carbon–carbon bonds. Toste and co-workers reported a catalytic Conia-ene reaction that proceeds at ambient temperatures and under neutral conditions using gold(I) complexes.²¹ Carreira and co-workers adopted this catalytic version of the Conia-ene reaction towards the total synthesis of (±)-gomerone C.²² Treatment of 4 with Et₃N, TBSCl, and NaI provided 9, setting the stage for the Au(I) catalyzed Conia-ene reaction. However, the Au(I) catalyzed cyclization of 9 failed to provide the desired 5-exo-dig cyclization product. Instead, it resulted in the formation of the undesired 6-endo-dig cyclization product 10.²³

Since the Conia-ene reaction failed to provide the desired 5-exo-dig cyclization product, we searched for other types of cyclization reactions. After an extensive investigation of conditions for the cyclization of alkynyl ketones, we adopted the Mn(III)-initiated cyclization method, which was originally reported by Snider and co-workers.²⁴ They found that the free radical cyclization of (trimethylsilyl)alkynyl ketones mediated by Mn(OAc)₃ provided 5-exo-dig alkenes as the major products. Gratifyingly, upon treatment of 11 with Mn(OAc)₃, the desired 5-exo-dig cyclization product 12 was obtained in 50% yield as an (E)/(Z) mixture (Scheme 2).²⁵ Selective removal of the TMS group without the accompanying PMB deprotection was achieved by treatment of 12 with p-TsOH in MeCN (73%).

Scheme 2.

Mn(OAc)₃-mediated radical cyclization of alkynyl ketone 11. Reagents and conditions: (a) MeLi•LiBr, THF, 0 °C, 1 h, then, TMSCCCH₂Br, HMPA, −78 to 25 °C, 16 h, 34%, d.r. = 1:1; (b) Mn(OAc)₃ (20 equiv), EtOH/HOAc, 100 °C, 72 h, 50%; (c) p-TsOH, MeCN, 0 °C, 2 h, 73%; (d) NaBH₄, MeOH, 0 °C, 2 h, 65%; (e) DIAD or DEAD, 4-nitrobenzoic acid, PPh₃, THF, −20 to 40 °C.

Having achieved the stereoselective construction of the bicyclo[3.2.1]octane by employing the Mn(III)-mediated radical cyclization reaction, we turned our attention to the stereoselective reduction of ketone 13 to the desired equatorial alcohol 3. When 13 was treated with NaBH₄, the reduction reaction provided a single stereoisomer. Careful analysis of the ¹H NMR spectral data indicated that the NaBH₄ reduction afforded the undesired axial alcohol 14 (see the ESI for details). Reduction of 13 under various reduction conditions such as DIBAL-H and L-Selectride did not afford the desired alcohol 3. These results were consistent with observations made by Newhouse and co-workers during the synthesis of principinol D.¹¹ To obtain the desired alcohol 3, we also attempted the Mitsunobu inversion (PPh₃, DIAD or DEAD, 4-nitrobenzoic acid) of 14, but the reaction resulted in either recovery of 14 or decomposition.

Since no attempts with the PMB protected γ-hydroxyl ketone 13 afforded the desired equatorial alcohol, we explored β-keto esters or β-hydroxy ketones for the stereoselective reduction to give the desired alcohol. Starting from known cyclic enone 15,²⁶ the Cu(I)-mediated addition of vinylmagnesium bromide (54%) followed by TMS-propargylation provided TMS-alkynyl ketone 17 (73%) (Scheme 3). As expected, subjecting 17 to the Mn(III)-mediated radical cyclization reaction afforded the desired bicyclo[3.2.1]octane 18 in 43% yield as an (E)/(Z) mixture. TMS deprotection of 18 was accomplished by treatment with p-TsOH to give 19 (69%). We explored a wide range of reducing agents (e.g., NaBH₄, MnCl₂•NaBH₄, Zn(BH₄)₂, Me₄NBH(OAc)₃, NH₃• BH₃) to stereoselectively reduce the β-keto ester to the desired β-hydroxy ester, but none of these conditions gave the desired equatorial alcohol 20a. Instead, the undesired axial alcohol 20b was obtained, as was the case with 14 (see Scheme 2 for details). We also attempted the S_N2 inversion of 20b to 20a, but the S_N2 inversion reaction resulted in a limited success. When the chloromethanesulfonate 21, prepared from 20b and ClCH₂SO₂Cl, was treated with KO₂ and 18-crown-6, the S_N2 inversion reaction afforded the desired alcohol 20a (5%) and the recovered undesired alcohol 20b (14%).

Scheme 3.

Mn(OAc)₃-mediated free radical cyclization of alkynyl β-keto ester 17 and reduction of bicyclo[3.2.1]octane 19. Reagents and conditions: (a) vinylmagnesium bromide, CuBr•SMe₂, THF, −78 °C, 1 h, 54%; (b) KOt-Bu, t-BuOH, reflux, 30 min, then, TMSCCCH₂Br, reflux, 1 h, 73%, d.r.>10:1; (c) Mn(OAc)₃ (12.5 equiv), EtOH/HOAc, 90 °C, 48 h, 43%; (d) p-TsOH, MeCN, 0 to 25 °C, 2 h, 69%; (e) SmI₂, PhSH, HMPA, THF, 0 °C, 5 h, rac-20a:rac-20b = 1:3; (f) ClCH₂SO₂Cl, 2,6-lutidine, CH₂Cl₂, 0 °C, 1 h, quantitative; (g) KO₂, 18-crown-6, DMSO, 25 °C, 2 h, 5% (rac-20a), 14% (rac-20b).

Then, we decided to adopt Newhouse’s β-hydroxy ketone substrate for the formation of the desired alcohol. LiAlH₄ reduction of keto ester 19 (81%) followed by mono TBS protection of the resulting diol gave the corresponding TBS protected alcohol 22 in 84% (Scheme 4). PCC oxidation and TBS deprotection set the stage for the SmI₂ reduction. Following Newhouse’s conditions,¹¹ when 23 was subjected to the SmI₂ reduction (SmI₂, PhSH, and HMPA), the reaction proceeded smoothly to give the desired equatorial alcohol 24 in 78% yield, as a single stereoisomer.¹¹

Scheme 4.

SmI₂-mediated reduction of β-hydroxy ketone 23. Reagents and conditions: (a) LiAlH₄, THF, 0 °C, 1 h, 81%; (b) TBSCl, imidazole, CH₂Cl₂, 25 °C, 24 h, 84%; (c) PCC, CH₂Cl₂, 25 °C, 1 h, 79%; (d) 6 N HCl, THF, 0 to 25 °C, 4 h, 67%; (e) PhSH, HMPA, SmI₂, THF, 0 °C, 5 h, 78%.

Having established the stereoselective route to the bicyclo[3.2.1]octane fragment of rhodojaponin III (1), we prepared the enantiopure β-vinyl cyclic ketone 16 by exploiting Helmchen’s auxiliary (see the ESI for details).^27,28 Enantiopure 16 can be converted to the enantiopure bicyclo[3.2.1]octane fragment 24 of rhodojaponin III (1) by following the procedure established for racemic β-vinyl cyclic ketone 16.

3. Conclusion

Rhodojaponin III (1) shows great potential as a non-opioid analgesic agent for pain management. Our synthetic efforts were focused on the stereoselective synthesis of the bicyclo[3.2.1]octane fragment of rhodojaponin III (1). The Cu-catalyzed conjugate addition of a vinyl Grignard reagent to α,β-unsaturated cyclic enone 17 followed by subsequent TMS-propargylation set the stage for the key cyclization reaction of alkynyl ketones. After an extensive search for cyclization conditions, Mn(III)-initiated radical cyclization of alkynyl ketones (11 and 17) provided the desired bicyclo[3.2.1]octanes (13 and 19, respectively). We expect that our stereoselective approach to the bicyclo[3.2.1]octane fragment of rhodojaponin III (1) in conjunction with Helmchen’s chiral auxiliary will be applicable in an enantioselective total synthesis of rhodojaponin III (1) and related natural products. The total synthesis of rhodojaponin III (1) will not only allow opportunities for new, innovative, and efficacious therapies to treat pain, but also offer a new drug target to overcome the current opioid crisis.

Highlights.

• Bicyclo[3.2.1]octanes were prepared by radical cyclizations of alkynyl ketones

• Conia-ene-type reactions were attempted for the synthesis of bicyclo[3.2.1]octanes

• A chiral auxiliary was used for the synthesis of enantiopure β-vinyl cyclic ketone