A Metathesis-Based Approach toward the Convergent Synthesis of Populusene A

Abstract

Our group was interested in developing a synthesis of populusene A, a diterpene isolated from the tree resin of Populus euphratica. The natural product displays submicromolar activity against the expression of pro-inflammatory proteins and transcription factors responsible for the regulation of inflammatory cytokines. The focus of this manuscript is on an effective synthesis of cycloheptenone 5, a promising building block for populusene A. An efficient synthesis of diene precursors for a proposed metathesis cyclization was developed. Unfortunately, initial attempts at metathesis were almost entirely unsuccessful. Two key structural alterations to the metathesis precursor made the cyclization practical and high yielding: replacing the terminal olefin with a phenyl-equipped disubstituted alkene and protecting the β-keto phosphonate as an enol carbonate. These changes limited homodimerization, and the O-carbonate enforced a conformation more favorable to cyclization. This approach may be helpful in addressing other challenging metathesis cyclizations.

Introduction

In 2021, Liu et al. investigated plant resins to elucidate medicinally relevant natural products. , Populus euphratica, a tree grown in arid environments in China, became an attractive target as the resin produced under saline-alkali stress has historically been used in traditional Chinese medicine to treat swelling and pain. Through their detailed investigation into Populus euphratica, cembrenoid populusene A (1, Figure ) was discovered and characterized by spectroscopic and computational methods. Populusene A contains an unprecedented bridgehead alkene system containing a bicyclo[8.4.1]­pentadecane core. Due to the unique structure of populusene A, it defines its own skeletal class known as populusane. Although multiple solvent systems were attempted to generate an X-ray-quality crystal of populusene A, none were successful, and its relative configuration was ultimately determined through electronic circular dichroism calculations.

1.

3D computational model of populusene A (1).

Populusene A displayed biologically significant activity as an anti-inflammatory agent. The natural product displayed a dose-dependent decrease in the expression of pro-inflammatory proteinsTNFα, IL-6, iNOS, and COX-2in bacterial LPS-stimulated RAW264.7 cells. Additionally, at a low concentration of 125 nM, populusene A displays inhibitory activity against the transcription factor responsible for the regulation of inflammatory cytokines, nuclear factor-κB (NF-κB), in LPS-induced RAW264.7 cells. There is potential for selective COX-2 inhibition through upstream regulation of cytokines that induce an inflammatory response. Evaluation of off-target inhibition of COX-1 was unfortunately not evaluated but would be an interesting future direction.

Populusene A attracted our attention as a synthetic target, and our retrosynthetic plan is outlined in Scheme . We envisioned two major building blocks: unsaturated aldehyde 4 and cycloheptenone phosphonate 5. Our discussion will focus on the development of a practical and scalable synthesis of cycloheptenone 5 and on how we overcame a problematic metathesis cyclization.

1. Populusene AProposed Retrosynthesis.

Results and Discussion

Cycloheptenone 5 was envisioned to be assembled in five steps from commercially available 2,3-dibromopropene (7) (Scheme ). A nucleophilic substitution reaction between ethyl acetate (6) and 2,3-dibromopropene (7) under basic conditions would afford vinyl bromide intermediate 8. Subsequent deprotonation of excess diethyl methylphosphonate and acylation with ester 8 was very effective; however, due to the volatility of 8, further route scouting was investigated as detailed below.

2. Initial Forward Synthesis toward the Eastern Fragment (5).

Several cross-coupling reactions were attempted, including a Pd­(PPh₃)₄-catalyzed Suzuki reaction and NiCl₂(dppp)-catalyzed cross coupling; however, the most promising results were from an iron-mediated cross-coupling reaction. Using Fe­(acac)₃ and the Grignard reagent, i-PrMgBr, we were able to achieve this coupling reaction to produce alkene 10 in good yields.

Optimization to improve the intermediate’s physical properties and reduce the overall step count began with diethyl­(2-oxopropyl)­phosphonate (20; Scheme ). Alkylation with 2,3-dibromopropene (7) using NaH and n-BuLi to generate the dianion gave the vinyl bromide intermediate 9 in 86% yield on a multigram scale. This improved sequence is presented in the optimized final route in Scheme . To set up the metathesis cyclization, a second alkene was introduced by alkylation of 10 with allyl bromide, leading to the metathesis substrate 11.

3. Summarized Eastern Fragment Synthetic Route.

Initial cyclization studies are summarized in Table . Standard RCM conditions were attempted on intermediate 11 (entry 1) utilizing the Hoveyda–Grubbs catalyst in CH₂Cl₂ at reflux (HGII, 13, Figure ). However, after 3 days, the only observable product was homodimerization 14, a common byproduct for this type of transformation.

1. Select RCM Optimization Reactions for Intermediate 5 .

entry	catalyst	solvent	temp.(°C)	molarity (M)	time	% component
1	HGII	CH2Cl2	40	0.06	3 days	0% 11
						100% 14
						0% 5
2	HGII	CH2Cl2	40	0.01	3 days	0% 11
						100% 14
						0% 5
3	GII	CH2Cl2	40	0.01	3 days	20% 11
						67% 14
						13% 5
4	HGII	C6F6	120 (μW)	0.03	2 × 15 min	0% 11
						76% 14
						24% 5

^aEntries used 10 mol % catalyst.

^bEntry used 30 mol % catalyst.

2.

Structures for the Grubbs 2nd generation catalyst (GII, 12) and Hoveyda–Grubbs 2nd generation catalyst (HGII, 13).

Kirkland and Grubbs had previously published on the importance of concentration with respect to the RCM of cyclic-trisubstituted alkenes through examination of their kinetics. They found that the more dilute the reaction is, the more likely it would be to favor the desired product over homodimerization. We attempted changing the concentration to 0.01 M from 0.06 M; however, this procedural change did not result in reduction of the undesired homodimerization product 14 (Table , entry 2).

Another procedural modification that we attempted was to change the catalyst within the reaction mixture. Molybdenum-based catalysts were considered but not attempted due to their high sensitivity to oxygen and moisture. Instead, Grubbs second generation (GII, 12) and Hoveyda–Grubbs second generation (HGII, 13) catalysts were utilized (Figure ). Changing the catalyst to HGII from GII (Table , entry 3) resulted in slightly more promising results with a mixture of starting material (20%), homodimerized product (67%, E/Z mixture), and the desired ring-closing product (13%) based on NMR analysis. These conditions gave our first sample of cycloheptenone 5, but the yield was very low. Additionally, due to GII being more prone to decomposition, 10 mol % catalyst was used with two additional charges of 10 mol % added per day. Consequently, we decided to continue moving forward with HGII for further optimization.

One of the last procedural changes that we attempted was to change the reaction solvent. Samojłowicz et al. had reported a doping effect of fluorinated aromatic hydrocarbon (FAH) solvents on the rate of ruthenium-catalyzed RCM reactions. They found that FAHs could improve the initiation step of Ru-catalyzed RCM reactions by π–π stacking with the N-aromatic substituent, which was supported experimentally and computationally. We used hexafluorobenzene (Table , entry 4) as an FAH solvent, as well as more forcing conditions in a microwave reactor. Slightly more promising results were observed by NMR with a 24% yield of the desired RCM product 5. Unfortunately, we did not find a practical way to cyclize diene 11 from our screening.

Grubbs and co-workers discussed a general model for alkene cross-metathesis selectivity. This insightful paper categorized olefins (types I–IV) based on their propensity toward homodimerization (Figure ). Mapping this analysis onto substrate 11, the 1,1-disubstituted olefin was categorized as type III (it displays no homodimerization), and the terminal olefin was type I (it displays rapid homodimerization, which was experimentally observed). Adding steric encumbrance to the terminal olefin might disfavor homodimerization and favor intramolecular cyclization. This change would be analogous to using a type II olefin in the cross-metathesis reactions discussed by Grubbs.

3.

Grubbs’ olefin categorization for cross-metathesis reactions applied to cyclization substrate 11.

Cyclization of more hindered olefin substrates with a HGII catalyst is presented in Table . The substrates were prepared by the alkylation of phosphonate 10 with different allylic bromides. Installing a methyl group in place of a hydrogen atom (entry 2) led to improved cyclization under microwave conditions and resulted in a 20% isolated yield of 5 with the majority of the mass still going to the homodimerized product. A phenyl group dramatically shifted the composition of the mixture, with no homodimerized product, and 74% of the mixture component was the desired metathesis product 5 by NMR (entry 3). Unfortunately, the isolated yield for cycloheptenone 5 was only 36% after purification and isolation, suggesting unidentified side reactions.

2. Evaluation of Steric Influence of Formation of Desired RCM Product 5 .

entry	R1	solvent	temp.(°C)	molarity (M)	time	% component	yield
1	H (11)	C6F6	120 (μW)	0.03	2 × 15 min	0% 15	N/A
						76% 14
						24% 5
2	Me (16)	PhMe	160 (μW)	0.01	15 min	7% 15	20%
						79% 14
						16% 5
3	Ph (17)	PhMe	150 (μW)	0.05	2 × 1 h	26% 15	36%
						0% 14
						74% 5

^aEntries used 10 mol % HGII catalyst.

In a further effort to improve the metathesis cyclization, we introduced a conformational restriction by preparing the enolate and trapping with methyl chloroformate to give the Z-enol carbonate (based on precedent). Cyclized product 19 was generated initially in modest yield from enol carbonate 18. Microwave irradiation for 1 h at 200 °C led to a mixture of 68% cyclized product and 32% starting material based on NMR analysis of the crude reaction mixture (Table , entry 1). Purification via chromatography resulted in 49% yield of cyclization product, which was the highest yield of any metathesis cyclization to date. The enol phosphate appears to play the largest role in the increased yield of the metathesis cyclization in comparison to the introduction of steric bulk. , This promising result encouraged us to evaluate the conformations of the different cyclization substrates.

3. Final Optimization of Eastern Fragment Metathesis.

entry	HGII loading	temp. (°C)	time	yield
1	10 mol %	200 (μW)	1 h	49%
2	10 mol %	120	2 days	74%
3	20 mol %	120	1 week	92%
4	10 mol %	160 (μW)	1 h	83%
5	10 mol %	170	1 h	75%

We analyzed phenyl substrate 18 by using molecular modeling with Avogadro and identified the extended orientation between the two alkenes (Figure ). Unprotected intermediate 17 had its two alkenes splayed apart, which would reduce the reactivity during the metathesis. The enol carbonate substrate 18 showed a minimized conformation, with the two alkenes held in close proximity. This conformation appears to be much more favorable for the metathesis cyclization.

4.

3D modeling of intermediates 17 and 18 was performed using Avogadro software. Energy minimization was performed using force field MMFF94s.

With the promising new substrate 18, we optimized the reaction conditions to improve the yield (Table ). The first reaction attempted was to use the same conditions that were initially employed but at 120 °C for 2 days, resulting in an improved 74% yield from 49% (Table , entry 2). To push the reaction, we doubled the amount of catalyst and allowed it to react over a week span to yield 92% yield (entry 3). Due to the prolonged reaction time, we decided to revisit utilizing the microwave reactor. At 160 °C in a microwave with the HGII catalyst, the desired metathesis product was isolated in 83% yield on a gram scale (entry 4). Alternatively, the reaction could be performed in a pressure reactor with similar yields after 1 h (entry 5). When the reaction was run in traditional glassware with heating, the yield was slightly lower; however, the reaction was easily scalable to multigram batches (entries 2 and 3). The O-carbonate-protecting group was best removed by treatment with methyllithium at low temperatures. These conditions gave cycloheptenone 5 in 82% yield on a gram scale. The optimized route is presented in Scheme and includes the preferred use of commercial keto phosphonate 20, which was discussed above.

Conclusions

Synthesis of cycloheptenone 5, a building block for our proposed synthesis of populusene A, was developed. Initial efforts to cyclize diene 11 in a metathesis reaction were almost completely unsuccessful. Two key changes to the substrate made the cyclization practical and high yielding: introduction of steric bulk at the type-I terminal alkene and protection of the β-keto phosphonate as an enol carbonate. These modifications improved the metathesis dynamics and strongly favored cyclization over dimerization. Cycloheptenone 5 is now a practical intermediate for its synthesis. Further investigation of the populusene A synthetic approach has been described elsewhere.

The data underlying this study are available in the published article and its Supporting Information.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c00302.

• Experimental procedures and full characterization, including ¹H NMR, ¹³C NMR, R _f, and HRMS data for all products (PDF)

†.

Novartis Institutes for BioMedical Research, 10675 John Jay Hopkins Drive, San Diego, California 92121, United States

‡.

Department of Chemistry, University of California at Davis, 141 Physical Sciences Mall, Davis, California 95616, United States.

§.

Department of Chemistry and Biochemistry, University of California at Los Angeles, 607 Charles E Young Dr E, Los Angeles, California 90095, United States.

All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.