Total synthesis of rupestine G and its epimers

Rupestine G is a guaipyridine sesquiterpene alkaloid isolated from Artemisia rupestris L. The total synthesis of rupestine G and its epimers was accomplished employing a Suzuki reaction to build a terminal diene moiety. The diene was further elaborated into the desired guaipyridine structure by a ring-closing metathesis reaction. Over all, rupestine G and its three epimers were obtained as a mixture in a sequence of nine linear steps with 18.9% yield. Rupestine G and its optically pure isomers were isolated by chiral preparative HPLC and fully characterized by 1H ,13C NMR, HRMS, optical rotation value, and experimental and calculated electronic circular dichroism spectroscopy.


Introduction
Guaipyridine sesquiterpene alkaloids are a family of natural compounds that share a unique structure consisting of a fused pyridine ring and seven-membered carbocycle [1]. For example, patchoulipyridine (1, figure 1) and epiguaipyridine (2, figure 1) were first isolated from the essential oil of Pogostemon patchouli Pellet by Büchi et al. in 1966 [2]. Another representative guaipyridine alkaloid, cananodine (3, figure 1), was isolated from the fruits of Cananga odorata. Cananodine shows potent activity against Hep G2 cell lines with a sub-micromolar IC 50 value [3]. Recently  used for detoxification, antitumour, antibacterial and antiviral activity, and for protecting the liver [4][5][6][7]. Owing to their structural similarities when compared with cananodine, it is suggested that rupestines might also possess promising cytotoxic activity. Unfortunately, biological evaluations of these alkaloids were limited by their scarce availability from natural sources. Hence, their scarcities and their unique structural features render them worthy targets for their total synthesis.
The first synthesis of guaipyridine sesquiterpene was accomplished by Büchi et al. [2]. Exposure of β-patchoulene to hydrazoic acid in the presence of H 2 SO 4 , followed by dehydrogenation in hot 1-methylnaphthalene over Pd/C produced patchoulipyridine (1) as the major product. Van der Gen et al. [8] isomerized the 1,5-double bond of guaiol to obtain the desired isomer with a 4,5-double bond, which was further oxidized with ozone and treated with hydroxylamine. By this way, the 5-epimer of epiguaipyridine (2) was synthesized. It should be noted that the absolute configuration of van Der Gen's synthetic product is different from that of the 'natural' one proposed by Büchi et al. [2]. Since neither β-patchoulene nor guaiol are commercially available, it is inevitably necessary to isolate them before the initiation of the synthesis. Decades later, Craig & Henry [9] applied a microwave-assisted decarboxylative Claisen rearrangement to synthesize (+)-cananodine in 2006 (scheme 1).
Another strategy was to build the seven-membered ring of guaipyridine compounds using derivatives of pyridine as the starting material. Applying this strategy, the Vyvyan group explored a base-promoted epoxide-opening and an intramolecular Heck cyclization to build the guaipyridine core (scheme 2) [10][11][12]. This approach subtly uses cheap and commercially available chemicals to launch the synthesis and deserves to be further developed.

Results and discussion
Natural rupestines are usually isolated as isomeric compounds, with different configurations at 5-and 8-positions. For example, rupestine B and C, rupestine H and I as well as rupestine L and M are natural isomeric compounds (electronic supplementary material, figure S0) [7]. Rupestine E was once erroneously assigned as its (5R,8R)-isomer, i.e. rupestine, a compound that has actually not been isolated from the natural plant [4][5][6][7]. In view of the confusion regarding the structural elucidation of isomers rsos.royalsocietypublishing.org R. Soc. open
of rupestine, preparation of all isomers would be beneficial for the confirmation of their individual structural characterizations and biological evaluations. Thus, we chose a nonstereoselective route to provide the four isomers in a single reaction.
Compound 12 was envisaged to be constructed by a ring-closing metathesis (RCM) reaction from the substituted diene 11. By application of a Suzuki cross-coupling reaction and alkylation, compound 11 could be obtained smoothly starting from compound 9. Furthermore, compound 9 could be accessible by decarboxylative Blaise reaction of picolinonitrile 8, which could be rapidly prepared from commercially available 5-bromo-2-picoline (6) (scheme 3).
The final synthesis strategy of rupestine G is shown in scheme 4. The 2-cyanopyridine 8 was readily prepared by m-CPBA oxidation and modified Reissert-Henze reaction from 5-bromo-2-methylpyridine (6) following the method developed by Fife [13][14][15]. The methyl nicotinoylacetate 9 was obtained from decarboxylative Blaise reaction of 8 with potassium methyl malonate in 82% yield [16][17][18][19]. Treatment of 9 with allyl bromide in the presence of sodium ethoxide provided 10 in 97% yield. After screening several Suzuki cross-coupling conditions, it was found that using isopropenylboronic acid pinacol ester, instead of unstable prop-1-en-2-ylboronic acid, gave compound 11 in 92% yield [20][21][22][23][24]. The pivotal RCM reaction catalysed by the Grubbs II catalyst was carried out to build the seven-membered ring in 53% yield [25][26][27][28]. According to the NMR data, the ring-closed product favoured the enol form 13 rather than the keto form 12, although both of the two tautomers were detectable on thin layer chromatography. The moderate but still acceptable yield of RCM reaction probably was the result of an undesired intermolecular reaction. The reaction in low concentration provided less intermolecular by-product, but also low conversion of the starting material 11. To sum up, the six-step reaction successfully constructed the frame of guaipyridine. Compound 13 was then reduced by NaBH 4 in MeOH. Theoretically, reduction of compound 13 would present one additional chiral carbon in the product, hence we did not purify the compound 14 but directly dehydrated it with MsCl in pyridine at 60°C and obtained diene 15 in a total yield of 67.1% in two steps. Hydrogenation of compound 15 catalysed by Pd/C in MeOH gave rupestine G and its epimers as a mixture in an overall 91.4% yield. Thus, from 5-bromo-2-picoline (6), the desired target was obtained in an overall 18.9% yield.
The mixture (46.4 mg) was first isolated on a preparative TLC to give two pairs of diastereoisomers (31.0 and 10.6 mg, i.e. 16a and 16b, respectively). These two pairs of compounds were further separated by chiral separation with a Shimadzu LC-20A preparative HPLC, to give four optically pure isomers.
The structures of these four isomers were intensively elucidated by extensive analysis with 1 H NMR, 13

Conclusion
In summary, we have achieved the first total synthesis of rupestine G and its epimers in a sequence of nine linear steps starting from commercially available 5-bromo-2-picoline. Notable transformations include a decarboxylative Blaise reaction between potassium methyl malonate and picolinonitrile and a Suzuki reaction to induce an isopropenyl group. The construction of the seven-membered ring was accomplished by a RCM reaction. Hydrogenation of the diene moiety finalized the synthesis of rupestine G and its epimers. Preparative HPLC obtained four optically pure isomers and their structures were fully characterized by 1 H, 13 C NMR, HRMS, optical rotation value, and experimental and calculated ECD. The synthetic approach demonstrated herein would be equally effective for the synthetic preparation of other guaipyridine sesquiterpene alkaloids. Biological evaluations of rupestine G and its epimers are ongoing and will be published in due course.

Material and methods
All reactions were performed in oven-dried flasks. Reagents and solvents were purchased from commercial vendors and used as received. Reaction progress and purity of the compounds were monitored by TLC. 1 [29][30][31][32]. Absolute configuration was assigned by using optical rotation spectra, circular dichroism spectroscopy and time-dependent density functional theory calculations at BP/TZVPP level. The ground-state geometries were optimized with density functional theory calculations. All atoms were estimated with the basis set def-TZVP and the functional BP. Electronic circular dichroism corresponding to the optimized structures was calculated using the TDDFT method at BP/def-TZVP level. The results were subsequently optimized by the Gaussian method. 4. 1