Total Syntheses of Marrubiin and Related Labdane Diterpene Lactones

Total syntheses of the labdane diterpene lactones marrubiin, marrulibacetal, desertine, marrulibacetal A, marrubasch F, cyllenine C, marrulanic acid, and marrulactone are described. The trans-decalin moiety of these molecules was constructed in a stereoselective manner by a Pauson-Khand reaction, and the resultant cyclopentenone was oxidatively cleaved for formation of the lactone ring. Elongation of the side chain at C9 was achieved by an epoxide-opening reaction with a variety of nucleophiles, and the functional group manipulations completed the syntheses of these natural products. Stereochemistries of desertine could be established by the transformations.


Retrosynthetic Analysis
Our retrosynthetic analysis of marrubiin-related natural products is depicted in Scheme 1. Since the difference between the target molecules lies in the variation in the side chain at C9, we planned to install the full C9 side chain late in the synthesis by a nucleophilic ring-opening reaction of advanced epoxide intermediate 22, which could be derived from ester 23 through a chemoselective reduction of the ester functionality in the presence of the lactone moiety. While the methyl group at C4 would be introduced from the less-hindered diastereoface by an enolate alkylation, C5−C6 bond formation, carboxylation at C4, and oxidative cleavage of the C−C bond at C6 were required for the conversion of enyne 20 to lactone 23. In light of these requirements, we chose to use the intramolecular Pauson−Khand reaction (PKR) [24,25] to fashion the six-membered ring and introduce the carbonyl group at C4. Although some concern arose over the formation of cis-fused isomer cis-24 from embodies the C8−C10 stereotriad of C9-oxygenated labdane diterpenoids, compound 20 can serve as a useful chiral building block for the synthesis of pharmacologically interesting, marrubiin-related natural products. In this article, we describe the total syntheses of members of the marrubiin family including marrubiin (1), marrulanic acid (10), cyllenine C (12), marrulibacetal (13), marrulactone (14), marrulibacetal A (17), desertine (18), and marrubasch F (19) [23].

Retrosynthetic Analysis
Our retrosynthetic analysis of marrubiin-related natural products is depicted in Scheme 1. Since the difference between the target molecules lies in the variation in the side chain at C9, we planned to install the full C9 side chain late in the synthesis by a nucleophilic ring-opening reaction of advanced epoxide intermediate 22, which could be derived from ester 23 through a chemoselective reduction of the ester functionality in the presence of the lactone moiety. While the methyl group at C4 would be introduced from the less-hindered diastereoface by an enolate alkylation, C5−C6 bond formation, carboxylation at C4, and oxidative cleavage of the C−C bond at C6 were required for the conversion of enyne 20 to lactone 23. In light of these requirements, we chose to use the intramolecular Pauson−Khand reaction (PKR) [24,25] to fashion the six-membered ring and introduce the carbonyl group at C4. Although some concern arose over the formation of cis-fused isomer cis-24 from

Retrosynthetic Analysis
Our retrosynthetic analysis of marrubiin-related natural products is depicted in Scheme 1. Since the difference between the target molecules lies in the variation in the side chain at C9, we planned to install the full C9 side chain late in the synthesis by a nucleophilic ring-opening reaction of advanced epoxide intermediate 22, which could be derived from ester 23 through a chemoselective reduction of the ester functionality in the presence of the lactone moiety. While the methyl group at C4 would be introduced from the less-hindered diastereoface by an enolate alkylation, C5−C6 bond formation, carboxylation at C4, and oxidative cleavage of the C−C bond at C6 were required for the conversion of enyne 20 to lactone 23. In light of these requirements, we chose to use the intramolecular Pauson−Khand reaction (PKR) [24,25] to fashion the six-membered ring and introduce the carbonyl group at C4. Although some concern arose over the formation of cis-fused isomer cis-24 from enyne 20 considering that [RhCl(CO) 2 ] 2 -catalyzed PKRs of substituted 5-(pent-4-ynyl)cyclohexa-1,3-dienes are precedented to afford cis-decalin derivatives stereoselectively [26], we expected that the stereocenter at C5 would be epimerized after oxidative cleavage of the double bond at C6.

Synthesis of Advanced Intermediate 22
The synthesis of advanced intermediate 22 commenced with the key PKR of enyne 20 (Scheme 2). Initial attempts to cyclize dicobalt complex 25, prepared by treatment of enyne 20 with Co2(CO)8 in CH2Cl2 (97% yield), upon heating in refluxing acetonitrile failed to produce any of the PKR products, leading to decomplexation. After considerable experimentation, we found that the desired tricyclic compound 24 could be produced with the aid of a promoter, with CyNH2 (Cy = cyclohexyl) [27] being optimal for this purpose. Gratifyingly, the product, obtained in (CH2Cl)2 at a substrate concentration of 10 mM under optimized conditions in 97% yield, proved to be the desired stereoisomer trans-24 as confirmed by the absence of a cross-peak between C10-CH3 and C5-H in the nuclear Overhauser effect spectroscopy (NOESY) spectrum. It has been suggested on the basis of quantum mechanical studies that the stereochemistry of PKR is determined by the irreversible olefin insertion step [28]. Since the chair−chairlike transition state (TS) A suffers from steric repulsion between C8-CH3 and C1-Hax, the reaction proceeded through chair−boatlike TS B, leading to the exclusive formation of trans-decalin trans-24. To enhance the synthetic utility of PKR, a number of catalytic methods have been reported, and a variety of metal complexes have been employed for this purpose [29]. However, our attempts to Scheme 1. Retrosynthetic analysis.

Synthesis of Advanced Intermediate 22
The synthesis of advanced intermediate 22 commenced with the key PKR of enyne 20 (Scheme 2). Initial attempts to cyclize dicobalt complex 25, prepared by treatment of enyne 20 with Co 2 (CO) 8 in CH 2 Cl 2 (97% yield), upon heating in refluxing acetonitrile failed to produce any of the PKR products, leading to decomplexation. After considerable experimentation, we found that the desired tricyclic compound 24 could be produced with the aid of a promoter, with CyNH 2 (Cy = cyclohexyl) [27] being optimal for this purpose. Gratifyingly, the product, obtained in (CH 2 Cl) 2 at a substrate concentration of 10 mM under optimized conditions in 97% yield, proved to be the desired stereoisomer trans-24 as confirmed by the absence of a cross-peak between C10-CH 3 and C5-H in the nuclear Overhauser effect spectroscopy (NOESY) spectrum. It has been suggested on the basis of quantum mechanical studies that the stereochemistry of PKR is determined by the irreversible olefin insertion step [28]. Since the chair−chairlike transition state (TS) A suffers from steric repulsion between C8-CH 3 and C1-H ax , the reaction proceeded through chair−boatlike TS B, leading to the exclusive formation of trans-decalin trans-24.

Synthesis of Advanced Intermediate 22
The synthesis of advanced intermediate 22 commenced with the key PKR of enyne 20 (Scheme 2). Initial attempts to cyclize dicobalt complex 25, prepared by treatment of enyne 20 with Co2(CO)8 in CH2Cl2 (97% yield), upon heating in refluxing acetonitrile failed to produce any of the PKR products, leading to decomplexation. After considerable experimentation, we found that the desired tricyclic compound 24 could be produced with the aid of a promoter, with CyNH2 (Cy = cyclohexyl) [27] being optimal for this purpose. Gratifyingly, the product, obtained in (CH2Cl)2 at a substrate concentration of 10 mM under optimized conditions in 97% yield, proved to be the desired stereoisomer trans-24 as confirmed by the absence of a cross-peak between C10-CH3 and C5-H in the nuclear Overhauser effect spectroscopy (NOESY) spectrum. It has been suggested on the basis of quantum mechanical studies that the stereochemistry of PKR is determined by the irreversible olefin insertion step [28]. Since the chair−chairlike transition state (TS) A suffers from steric repulsion between C8-CH3 and C1-Hax, the reaction proceeded through chair−boatlike TS B, leading to the exclusive formation of trans-decalin trans-24. To enhance the synthetic utility of PKR, a number of catalytic methods have been reported, and a variety of metal complexes have been employed for this purpose [29]. However, our attempts to  To enhance the synthetic utility of PKR, a number of catalytic methods have been reported, and a variety of metal complexes have been employed for this purpose [29]. However, our attempts to carry out the catalytic reaction with enyne 20 met with failure: the use of Krafft conditions [30] resulted in recovery of enyne 20 in 49% yield, whereas a complicated mixture was obtained when using 2-naphthaldehyde as a CO donor in the [RhCl(cod)] 2 -catalyzed reaction [31]. In contrast to these unsuccessful results, the transformation of enyne 20 to enone trans-24 could be refined to a one-pot procedure, wherein the cobalt complex was formed in (CH 2 Cl) 2 and heated under reflux after addition of CyNH 2 and 10-fold dilution with (CH 2 Cl) 2 . As anticipated, alkylation of the potassium enolate generated from trans-24 with MeI in THF took place exclusively from the less-hindered α-face to provide enone 26 in 89% yield.
With tricyclic compound 26 in hand, efforts were next focused on the transformation of the cyclopentenone moiety to the corresponding γ-butyrolactone. With regard to the oxidative cleavage of the cyclopentenone ring, it was found that desired γ-ketocarboxylic acid 27 could be obtained upon exposure of enone 26 to ozone in CH 2 Cl 2 at −78 • C followed by either reductive (Me 2 S) or oxidative (H 2 O 2 , NaOH) workup [32], but the reaction suffered from low yield (33%) and reproducibility issues. While α-ketocarboxylic acid 28, detected as a byproduct, could be converged to γ-ketocarboxylic acid 27 upon treatment with H 2 O 2 in aqueous NaOH/THF, no improvement in overall yield was observed. We then investigated a stepwise approach via α-ketocarboxylic acid 28. After an extensive screening of oxidants, the KMnO 4 /NaIO 4 system proved to be effective for the conversion of enone 26 to 28. To our surprise, submission of crude α-ketocarboxylic acid 28 to activated carbon for removal of the residual Mn salt to avoid decomposition of H 2 O 2 effected desired decarbonylation. As a consequence, γ-ketocarboxylic acid 27 could be obtained in 53% yield over two steps. The reason for the decarbonylation is unclear at present, but the possibility of involvement of contaminated Mn salt was excluded due to the fact that the reaction occurred from α-ketocarboxylic acid 28 produced by ozonolysis [33]. To the best of our knowledge, this is the first example of activated carbon-mediated decarbonylation of α-ketocarboxylic acid. Lactone formation from γ-ketocarboxylic acid 27 was achieved following the precedents of Wheeler [34] and Mangoni [19]: selective reduction of the carbonyl group at C6 with NaBH 4 in EtOH at 0 • C was followed by lactonization through a mixed anhydride upon treatment with ClCO 2 Et in the presence of Et 3 N in CH 2 Cl 2 at 0 • C to give lactone 23 in 88% yield over two steps [35].
The remaining operation necessary for the synthesis of advanced intermediate 22 involved chemoselective reduction of the ester functionality in the presence of the lactone moiety (Scheme 3). As a prelude to the conversion, TMS ether 23 was converted to α-hydroxyester 29 by exposure to Bu 4 NF in THF at 0 • C (98% yield). With regard to the reduction, the use of NaBH 4 or LiBH 4 resulted in no reaction even at the reflux temperature, whereas the lactone moiety in 29 was selectively reduced with diisobutylaluminum hydride (DIBALH) in CH 2 Cl 2 at −78 • C. We were gratified to find that the desired chemoselective reduction could be achieved by the use of sodium bis(2-methoxyethoxy)aluminum hydride (Red-Al ® ) as a reducing agent. It should be noted that the hydroxyl-directed reduction was accompanied by some lactone reduction when performed using Et 2 O or THF as a solvent or when performed at temperatures above −20 • C. The use of CH 2 Cl 2 proved optimal in terms of chemoselectivity and solubility of substrate 29, but the collapse of the five-membered aluminate intermediate was retarded under the optimal conditions, resulting in the formation of a mixture of aldehyde 30 and 1,2-diol 31 after aqueous workup. Thus, the mixture needed to be subjected again to Red-Al ® in CH 2 Cl 2 at −23 • C for the full reduction to 1,2-diol 31. The synthesis of advanced intermediate 22 was completed upon treatment of 1,2-diol 31 with p-toluenesulfonylimidazole and NaH in THF at 0 • C (84% yield over three steps).

Total Syntheses of Marrubiin and Related Labdane Diterpene Lactones
Having established a route to epoxide 22, the stage was now set for elongation of the side chain for total syntheses. In this regard, Welch and co-workers reported the synthesis of isomarrubiin (C9epi-marrubiin) by a CuI-catalyzed epoxide ring-opening reaction of the C9-epimer of epoxide 22 in Et2O at room temperature (40% yield) [36]. With this reaction serving as a reference, we first examined Cu(I)-catalyzed epoxide-opening reaction with Grignard reagent 32 [37] (Scheme 4). After some experimentation, we found that the use of a stoichiometric amount of CuBr·SMe2 was more effective, providing marrubiin (1), [α]D 21 +34.4 (c 1.04, CHCl3) [lit. [38], [α]D 20 +35.8 (c 3.1, CHCl3)] in 65% yield. However, Grignard reagent 32 is prone to isomerization due to the high acidity of the furan 2position, resulting in the formation of isomer 33 in 9% yield [39]. This problem was circumvented by the use of Grignard reagent 34 [40], the TMS group of which was uneventfully removed with Bu4NF in THF [41] after the epoxide-opening reaction.

Total Syntheses of Marrubiin and Related Labdane Diterpene Lactones
Having established a route to epoxide 22, the stage was now set for elongation of the side chain for total syntheses. In this regard, Welch and co-workers reported the synthesis of isomarrubiin (C9-epi-marrubiin) by a CuI-catalyzed epoxide ring-opening reaction of the C9-epimer of epoxide 22 in Et 2 O at room temperature (40% yield) [36]. With this reaction serving as a reference, we first examined Cu(I)-catalyzed epoxide-opening reaction with Grignard reagent 32 [37] (Scheme 4). After some experimentation, we found that the use of a stoichiometric amount of CuBr·SMe 2 was more effective, providing marrubiin (1) ] in 65% yield. However, Grignard reagent 32 is prone to isomerization due to the high acidity of the furan 2-position, resulting in the formation of isomer 33 in 9% yield [39]. This problem was circumvented by the use of Grignard reagent 34 [40], the TMS group of which was uneventfully removed with Bu 4 NF in THF [41] after the epoxide-opening reaction.

Total Syntheses of Marrubiin and Related Labdane Diterpene Lactones
Having established a route to epoxide 22, the stage was now set for elongation of the side chain for total syntheses. In this regard, Welch and co-workers reported the synthesis of isomarrubiin (C9epi-marrubiin) by a CuI-catalyzed epoxide ring-opening reaction of the C9-epimer of epoxide 22 in Et2O at room temperature (40% yield) [36]. With this reaction serving as a reference, we first examined Cu(I)-catalyzed epoxide-opening reaction with Grignard reagent 32 [37] (Scheme 4). After some experimentation, we found that the use of a stoichiometric amount of CuBr·SMe2 was more effective, providing marrubiin (1), [α]D 21 +34.4 (c 1.04, CHCl3) [lit. [38], [α]D 20 +35.8 (c 3.1, CHCl3)] in 65% yield. However, Grignard reagent 32 is prone to isomerization due to the high acidity of the furan 2position, resulting in the formation of isomer 33 in 9% yield [39]. This problem was circumvented by the use of Grignard reagent 34 [40], the TMS group of which was uneventfully removed with Bu4NF in THF [41] after the epoxide-opening reaction. An inspection of the structures of marrulibacetal (13) and marrulibacetal A (17) revealed that the highly oxidized tetrahydrofuran ring of these molecules could be formed by oxidation of the furan moiety in marrubiin (1) followed by internal acetalization. While unprecedented in the transformation of the marrubiin class labdane diterpenoids, successive oxidations of a furan ring have been documented in semi-synthesis of the neoclerodane diterpene natural products salvinicins A and B by Prisinzano and co-workers [42]. Furthermore, Frontana-Uribe and co-workers reported construction of the [6,6,5,5]-tetracyclic framework, in which 4,5,6,7a-tetrahydro-2H-furo [2,3-b]pyran is spirolinked to a trans-decalin ring system, by an electrochemical oxidation of hispanolone [43]. When marrubiin (1) was exposed to pyridine tribromide in EtOH/CH 2 Cl 2 at 0 • C, oxidative acetalization occurred to give bisacetal 36 in 79% yield with a cis/trans ratio of 2:1, albeit with no sign of internal acetalization in contrast to Frontana-Uribe's work. This result is attributed to the conformational constraint imposed by the lactone ring. The chemical yield was improved to 84% in the presence of K 2 CO 3 as an acid scavenger. As expected from the precedent [44], the reaction rate of bisacetal 36 with OsO 4 /4-methylmorpholine N-oxide (NMO) in aqueous THF was influenced by the substrate structure: the reaction of a 1:1 mixture of diastereomers with the cis relative configuration at room temperature proceeded to completion within 1 h to give diols 38 and 39 in a stereoselective manner, whereas the trans-isomers were almost recovered unchanged under these conditions. Although the remaining trans-isomers could be consumed at an elevated temperature after prolonged reaction times, the lower π-facial selectivity (desired:undesired = 1:1.2), together with the fact that recovered trans-isomers could be isomerized to cis-isomers upon exposure to pyridinium p-toluenesulfonate (PPTS) in EtOH for 1 h, prompted us to perform the reaction at room temperature. Submission of an inseparable mixture of diols 38 and 39 to TsOH in benzene effected internal transacetalization, affording marrulibacetal (13) (1). It should be mentioned that one of two diastereomers, obtained as an inseparable mixture in a ratio of 1:1 by dihydroxylation of bisacetal 37, matched by 1 H-NMR with desertine (18), although the chemical correlation to establish the stereochemistry will be presented later (vide infra). The low chemical yield (16%) of 17 was due to the formation of two isomers with the trans H15/H16 stereochemistry in 38% yield.
Having synthesized marrubiin (1) and natural products possessing the same carbon framework in a higher oxidation state, we next addressed the conversion of epoxide 22 to natural products 12 and 14, which required two-and three-carbon nucleophiles, respectively (Scheme 5). Fortunately, epoxide 22 underwent nucleophilic ring-opening reaction with commercially available lithium acetylide ethylenediamine complex, affording alcohol 43 in 95% yield. Oxidative lactonization of homopropargyl alcohol 43 could be attained by the gold(I)-catalyzed cycloisomerization/oxidation sequence under Ye conditions [47], completing the synthesis of cyllenine C (12). On the other hand, the use of Grignard reagent 44 [48] as a three-carbon nucleophile under conditions identical to those with 32 gave alcohol 45 (48% yield), which was desilylated with Bu 4 NF in THF to furnish 1,5-diol 46 in 82% yield.
Of the eight natural products synthesized, desertine (18) could not be purified due to the difficulty of separation from its diastereomer 40. With regard to this natural product, stereochemistries were determined by an NOESY experiment by Dijoux-Franca and co-workers, who indicated two structures differing in the configuration at C15 [14]. Furthermore, evidence that supports the stereochemical relationship between the decalin and tetrahydrofuran moieties, separated by the C11−C12 two-carbon bridge, was not provided in their report. Therefore, we felt compelled to perform experiments to determine the stereochemistries of 18, and we found that exposure of marrulibacetal A (17) and its diastereomer 42 to TsOH in refluxing MeOH to effect transacetalization resulted in the formation of triol 40 and desertine (18), respectively, albeit in low yields (Scheme 6). These results suggest that the stereocenters at C13 and C14 of 18 have configurations opposite to those of 17. Together with our previous observation that desertine was produced from cis-37, the stereochemistries of this natural product can be established as shown for 18. Scheme 5. Total syntheses of cyllenine C (12), marrulactone (14), and marrulanic acid (10) from epoxide 22. DMSO = dimethyl sulfoxide, Tf = trifluoromethanesulfonyl, Ms = methanesulfonyl, TBS = tert-butyldimethylsilyl. Scheme 5. Total syntheses of cyllenine C (12), marrulactone (14), and marrulanic acid (10) from epoxide 22. DMSO = dimethyl sulfoxide, Tf = trifluoromethanesulfonyl, Ms = methanesulfonyl, TBS = tert-butyldimethylsilyl.
Of the eight natural products synthesized, desertine (18) could not be purified due to the difficulty of separation from its diastereomer 40. With regard to this natural product, stereochemistries were determined by an NOESY experiment by Dijoux-Franca and co-workers, who indicated two structures differing in the configuration at C15 [14]. Furthermore, evidence that supports the stereochemical relationship between the decalin and tetrahydrofuran moieties, separated by the C11−C12 two-carbon bridge, was not provided in their report. Therefore, we felt compelled to perform experiments to determine the stereochemistries of 18, and we found that exposure of marrulibacetal A (17) and its diastereomer 42 to TsOH in refluxing MeOH to effect transacetalization resulted in the formation of triol 40 and desertine (18), respectively, albeit in low yields (Scheme 6). These results suggest that the stereocenters at C13 and C14 of 18 have configurations opposite to those of 17. Together with our previous observation that desertine was produced from cis-37, the stereochemistries of this natural product can be established as shown for 18. Molecules 2020, 25
This sequence was repeated employing bis(2-methoxyethoxy)aluminum hydride in toluene (3.3 M, 0.16 mL, 0.53 mmol) and CH 2 Cl 2 (4 mL) with the reaction time of 16 h at −23 • C. The crude product (216 mg) was used without further purification.