|Home | About | Journals | Submit | Contact Us | Français|
We report our full account of the enantioselective total synthesis of (−)-acylfulvene (1) and (−)-irofulven (2), which features metathesis reactions for the rapid assembly of the molecular framework of these antitumor agents. We discuss (1) the application of an Evans’ Cu-catalyzed aldol addition reaction using a strained cyclopropyl ketenethioacetal, (2) an efficient enyne ring-closing metathesis (EYRCM) cascade reaction in a challenging setting, (3) the reagent IPNBSH for a late stage reductive allylic transposition reaction, and (4) the final RCM/dehydrogenation sequence for the formation of (−)-acylfulvene (1) and (−)-irofulven (2).
The illudins are a family of highly cytotoxic sesquiterpenes isolated from the bioluminescent mushroom Omphalotus illudens (Jack O’Lantern mushroom) and other related fungi.1 Illudin M (3) and illudin S (4) (Figure 1) are among the most cytotoxic members of this family, and have been studied extensively for their promising antitumor activity.2 Despite their high cytotoxicity, these illudins exhibit low therapeutic indices in solid-tumor systems.3 Consequently, several analogs of the natural illudins have been prepared and evaluated for the treatment of various cancers.4 One such semi-synthetic derivative, irofulven (2), was prepared from illudin S through treatment with excess acid and formaldehyde, and has demonstrated greatly enhanced therapeutic potential against several solid tumor systems.5 The superior pharmacological properties of irofulven (2) are accompanied by a markedly lower cytotoxicity than that of illudin S (4).6 Several studies have been directed toward elucidating the mechanism of biological activity of the illudins, acylfulvene (1), and irofulven (2) in order to understand the nature of this selective toxicity.7 The mechanism is believed to involve an initial activation step by conjugate addition of a hydride (NADPH) or thiol (glutathione, or cysteine) nucleophile into the enone moiety followed by nucleophilic addition of DNA to the strained cyclopropane ring to generate a stable aromatic DNA adduct 18 (Scheme 1). The observed onset of apoptosis is believed to be a result of DNA alkylation followed by strand cleavage through this general mechanism. Irofulven (2) is currently undergoing clinical trials for the treatment of various cancers as both a monotherapy and in combination with other chemotherapeutics.8
The promising antitumor properties and the highly reactive molecular framework of (−)-irofulven (2) and other illudins have rendered them interesting synthetic targets.9 Our laboratory has disclosed concise enantioselective syntheses of (−)-acylfulvene (1) and (−)-irofulven (2).10 Key features of our approach include a stereoselective aldol addition of a strained ketenehemithioacetal 26, which secures the C2 stereocenter and enables ready access to aldehyde (+)-22 (Scheme 2). A key enyne ring-closing metathesis (EYRCM)11 cascade reaction of trienyne 21 generates the AB-ring system 20. A reductive allylic transposition then sets the stage for the final ring-closing olefin metathesis (RCM) to build the C-ring and complete the syntheses of (−)-acylfulvene (1) and (−)-irofulven (2). Herein we describe the development of our general synthetic strategy to these fascinating molecules.
Since aldehyde 22 contains the reactive cyclopropane and tertiary alcohol substructure common to acylfulvene (1), irofulven (2), and most members of the illudin family, its efficient synthesis was of critical importance. Initially, we developed a synthetic route that enabled us to rapidly generate large quantities of the racemic aldehyde 22 for evaluation of our synthetic strategy (Scheme 3).12 This route involved treatment of pentane-2,4-dione (27) with 1,2-dibromoethane and potassium carbonate in dimethylsulfoxide (DMSO) to afford cyclopropyl diketone 28 in 61% yield (Scheme 3). Mono olefination using the Wittig reaction afforded intermediate 29 in 56% yield. Silylcyanation with stoichiometric TMSCN in the presence of catalytic InBr3 then afforded cyanohydrin 30 in 81% yield, and DIBAL-H reduction afforded the racemic aldehyde 22 in multi-gram quantities.
The enantioselective total synthesis of the target compounds required an enantioselective synthesis of aldehyde 22. Initially, we considered an asymmetric silylcyanation strategy to generate the tertiary alcohol stereocenter (Scheme 4), based on the route to the racemic aldehyde 22. Examination of Jacobsen’s thiourea catalyst 3113 provided the desired optically enriched cyanohydrin 30; however, the conversion and level of stereoselection with ketone 29 was non-ideal (50 h, 13%, 53% ee). Furthermore, the selectivity was detrimentally affected by the long reaction times that were required for full conversion of the starting material (8 d, 71%, 34% ee). The use of ketone 29 as substrate with Hoveyda’s catalyst 3214 in the presence of Al(OiPr)3 and Ph3PO afforded the desired compound in good yields (79%), but unfortunately without enantioselection. Likewise, the use of Deng’s silylcyanation reaction15 conditions employing a cinchona alkaloid based catalyst ((DHQD)2AQN) also proved problematic, highlighting the challenge in developing a solution strictly based on the proven route to racemic 30.16
We investigated several asymmetric oxidation reactions as a means of accessing the tertiary alcohol stereocenter including a Sharpless dihydroxylation, a Sharpless epoxidation, and a substrate directed epoxidation relying on a stereocenter set by a Carreira alkynylation reaction (Scheme 5). Double olefination of diketone 28 afforded the volatile diene 33, which was subjected to Sharpless’ dihydroxylation conditions.17 While the desired diol 34 was generated in 50% yield, the diene 33 proved to be a poor substrate for enantioselective dihydroxylation. We proceeded to explore the Sharpless asymmetric epoxidation18 reaction with alcohol 35, which was prepared from ketone 29 through a Shapiro reaction with dimethylformamide (DMF) followed by a Luche reduction. Unfortunately, the Sharpless epoxidation of diene 35 provided a complex mixture of products likely resulting from the oxidation of the undesired olefin. Also, alternative synthesis of racemic 36 highlighted its undesired propensity to undergo a Lewis acid catalyzed rearrangement to aldehyde 37. An approach based on asymmetric alkynylation of aldehyde 37 followed by substrate directed epoxidation also did not provide the desired C2-stereocenter.19 While Carreira’s alkynylation reaction provided the desired product 38 with excellent stereoselectivity (99% ee) using superstoichiometric Zn(OTf)2 and N-methylephedrine (NME), the subsequent epoxidation of the allylic alcohol 38 using meta-chloroperbenzoic acid (mCPBA) resulted in the formation of a complex mixture of products. Since oxidation reactions20 aimed at forming the stereocenter adjacent to the cyclopropane proved to be problematic, we pursued an alternative route.
We sought to use Evans’ copper catalyzed aldol reaction for the formation of the desired tertiary alcohol stereocenter,21 in which we needed to generate a highly strained cyclopropyl silylketenehemithioacetal nucleophile 26 (Scheme 2). Initial studies by Ainsworth and coworkers aimed at generating the O-silylated cyclopropyl keteneacetal 41, revealed that formation of this strained exocyclic double bond was problematic. They reported that the product 41 was generated in at most 10% yield (R = Me, Equation 1).22 Instead, the C-silylated product 42 was formed as the major product (40%, R = Me, Equation 1). Following this report, Pinnick and coworkers observed the formation of the trimer 43 in addition to the C- and O-silylated products 41 and 42 (R = Et, Equation 1).23 These cyclopropyl ester enolate anions are generally regarded as pyramidalized carbanion centers rather than the O-lithiated planar methylene cyclopropane species.24
Our studies revealed that enolization of 1-cyclopropylethanone (44) at the cyclopropyl carbon is problematic if competing enolization pathways are accessible. Both hard and soft enolization conditions afforded the undesired silyl enol ether 45 exclusively (Scheme 6).25
Interestingly, Seebach and coworkers were able to generate a lithium cyclopropanecarbothioate anion from the corresponding thiol ester and characterize it through X-Ray crystallographic analysis.26 This structure exhibited features characteristic of a normal planar O-lithiated enolate, as opposed to a pyramidal C-lithiated center. Guided by this observation, we reasoned that the enolate of cyclopropylthiol esters might prefer the formation of the O-silylated ketenehemithioacetal rather than the C-silylated product. To our delight, the O-silylated ketenehemithioacetals 26a and 26b were generated as the major products through treatment of the cyclopropylthiol esters 46a and 46b with lithium diisopropylamide (LDA) and trimethylsilyl chloride (TMSCl) in THF at −78 °C (Scheme 7). This reaction afforded an inseparable mixture of the O- and C-silylated products 26 and 47. The highest selectivity was achieved with the ethylthiol ester 26b to generate a 9:1 mixture of 26b and 47b in 70% yield; whereas, the tert-butylthiol ester 46a led to a 3:2 mixture of 26b and 47b in 67% yield. Fortunately, the undesired C-silylated products 47a and 47b did not interfere with the planned aldol reaction. The mixture of compounds 26b/47b (9:1) could be generated on multi-gram scale and could be stored under an argon atmosphere at −10 °C for greater than a month without any decomposition or O- to C-silyl transfer. To the best of our knowledge, this is the first example of the formation of a cyclopropyl silylketenehemithioacetal that can be applied in a Mukaiyama aldol reaction.27
Due to the strain associated with the exocyclic double bond, the cyclopropyl ketenehemithioacetals 26a and 26b are highly reactive and are excellent substrates for Evans’ copper catalyzed aldol reaction21 (Table 1). Under optimal conditions, treatment of silylketenehemithioacetal 26b (1.1 equiv, mixture of 26b:47b = 9:1) with methylpyruvate (25) in the presence of 10 mol% of (R,R)-CuBox provided the enantiomerically enriched thiol ester (+)-48b (R2 = TMS) in 95% yield and 92% ee (entry 10, Table 1).28 This reaction was performed on large scale to generate a 20-gram batch of the desired product (+)-48b, and the (R,R)-Box ligand was recovered in approximately 85 % yield from the reaction mixture. As a part of these studies, we also evaluated the (R,R)-CuPybox catalyst, but it proved to be inferior to the CuBox system for this transformation (entries 2 and 3, Table 1). While the t-butylketenehemithioacetal substrate, 26a, was competent for this transformation under the optimized conditions (entry 6, Table 1), attempts to derivatize the resulting t-butylthiol ester 48a proved to be ineffective (vide infra, Scheme 8).
With the bisesters 48a and (+)-48b in hand, we proceeded to derivatize the thiol ester selectively. Initially, we investigated methylcuprate addition into the C4 thiolester.29 Attempts to functionalize the tert-butylthiol ester 48a proved to be inefficient (Scheme 8). Surprisingly, using a large excess of methylcuprate (10 equiv), methyl addition occurred exclusively at the C1 methyl ester to afford the lactone 49 in 45% yield. In contrast, addition of 1 equivalent methylcuprate to the more reactive ethylthiol ester (+)-48b afforded the desired product (+)-50 in 25% yield. However, this reaction was complicated by significant decomposition of the sensitive cyclopropylketone (+)-50 under the reaction conditions.
We found that the ethanethiol ester (+)-48b could be selectively derivatized through a modified Fukuyama cross-coupling protocol.30 Using the reported reaction conditions,30a we obtained the desired product (+)-50 in 42% yield (entry 1, Table 2). Under these conditions, the reaction suffered from incomplete conversion of the starting material (27% recovered (+)-48b) and the instability of the catalyst, which was evident from the precipitation of palladium black over the course of the reaction. We developed the optimal conditions for the substrates of interest by evaluating various ligands, reaction temperatures, and solvents (Table 2). Using the optimal conditions, multi-gram quantities of the methyl ketone (+)-50 were efficiently prepared in 83% yield via the cross-coupling of thiol ester (+)-48b with iodomethylzinc using 2-dicyclohexylphosphino-2′,6′-dimethoxy-1,1′-biphenyl (SPhos)30b as a supporting ligand in a 1:1.5 THF:NMP30c solvent mixture (entry 7, Table 2). SPhos proved to be the ideal ligand for this difficult transformation providing improved stability for the palladium metal center and increased reaction rates.
Methylenation of the sensitive and sterically hindered ketone (+)-50 was achieved through a Takai olefination (Scheme 9).31 Treatment of ketone (+)-50 with CH2I2, Zn dust, TiCl4, and catalytic PbCl2 afforded olefin (+)-51 in 89% yield.32 The ester (+)-51 was then treated with DIBAL-H to afford a mixture of the desired aldehyde (+)-22 and the corresponding fully reduced primary alcohol (1:2.5 respectively). Without purification, this mixture was immediately oxidized with Dess-Martin periodinane (DMP) to give aldehyde (+)-22 exclusively in 91% yield over the two steps.
The configuration of C2 in aldehyde (+)-22 was verified through X-ray crystallographic analysis of a corresponding derivative with (−)-brucine (Scheme 10).10,33 This efficient aldol-based approach for securing the C2 stereochemistry enabled us to generate multi-gram quantities of the key aldehyde (+)-22. Notably, aldehyde (+)-22 possesses a substructure that can be mapped on to most of the illudin sesquiterpenes.
With the routes to the racemic and optically enriched aldehyde 22 established, we developed a two-step sequence to generate several substrates for the evaluation of the EYRCM reaction.12 Addition of a series of alkynes 53a–m to aldehyde 22 followed by desilylation provided the diols 54a–m as a mixture of C3 diastereomers (3S:3R, 4–9:1) favoring the Felkin-Ahn mode of carbonyl addition (Scheme 11).34 We then added the allylsilane tether12 for the planned enyne metathesis cascade. Thus, monosilylation of diols 54a–j with allyldimethylsilyl chloride afforded the enynes 55a–j (Scheme 11).
The EYRCM sequence described in Scheme 12 represented our planned approach toward the synthesis of the functional AB-ring system common to the illudins. The enyne metathesis between the tethered olefin and the alkyne of 56 could generate a ruthenium alkylidene 58, that would undergo a ring-closing olefin metathesis to afford a tetrasubstituted alkene on a highly substituted B-ring 59.35 We envisioned that elaboration of the functionalized side chain of 59 would potentially allow rapid access to various members of the illudin family.
The initial studies of the key EYRCM step were carried out on the enynes 55a–d containing a functional side chain potentially en route to our targets. These trienynes 55a–d were treated with the first- or second-generation Grubbs’ ruthenium catalyst (G136 and G237 respectively), and the reactions were monitored by 1H NMR spectroscopy (Scheme 13).12 However, none of the trienynes 55a–d afforded the desired EYRCM products 60a–d. The lack of reactivity of these substrates indicated that the efficiency of the EYRCM is highly sensitive to steric congestion around the alkyne. A similar lack of reactivity was observed with the trienyne 55n,38 which suggested that unhindered terminal olefins competitively reacted with and reduced the activity of the metathesis catalyst toward the desired EYRCM cascade.
Accordingly, we selected trienyne 55f bearing a side chain with a less reactive trisubstituted olefin, and monitored the EYRCM reaction of this substrate using 1H NMR (Scheme 14). We were delighted to find that treatment of trienyne 55f with G2 for 1 h at 65 °C generated the desired cyclic silane 60f with good conversion (90%, 1H NMR). Interestingly, when the G1 catalyst was used, the enyne 55f was converted to the cyclopentenyl product, 61. Extensive 2D-NMR analysis and X-ray crystallographic analysis of a related product39 allowed for the assignment of the structure of the cyclopentene 61.40
A plausible mechanism for the formation of the two metathesis products 60f and 61 is described in Scheme 15. In the presence of the G2 catalyst, the initial metathesis occurs at the terminal olefin 62 to give, after the EYRCM, the desired cyclic silane 60f. Conversely, it is plausible that the less reactive G1 allows reversible formation of ruthenium alkylidene 64, which undergoes a more facile enyne metathesis reaction to produce cyclopentene 61.41
The encouraging result obtained with enyne 55f using G2 prompted us to evaluate the efficiency of the key metathesis reaction on other substrates. Thus, enynes 55f–p were subjected to G2 (10 mol%) in PhH at 65 °C for 1 h to afford the desired tricyclic dienes 60f–p in modest to good yields (Scheme 16).10 In situ 1H NMR monitoring of these reactions revealed clean conversion in all cases. Thus, the moderate yields are attributed to the sensitivity of these silanes towards silica gel chromatography. Notably, the enyne metathesis conditions proved to tolerate sensitive functional groups such as the aldehyde of 55o42 and the primary iodide of 55p.43
Encouraged by these results, we focused our efforts on building the C-ring of the illudins. Our initial strategy was inspired by the work of Hoye and coworkers on the relay ring-closing metathesis reaction (Scheme 17).44 Initial metathesis of the allyl group of a tetraene 68, obtained from the cyclic silyl ether 59, would generate the ruthenium alkylidene 69. This would set the stage for an intramolecular olefin metathesis providing compound 70 with the ruthenium at the site required for the final cyclization to generate 71.45
Thus, we prepared the substrates 73 and 74 for the relay ring-closing metathesis (Scheme 18). Wittig olefination of 60o afforded 60n in a low 30% yield, complicated due to the sensitivity of the allylic silane. The triol, 72, was then prepared in 30% yield through a Tamao oxidation.46 The allyl silane tether was then selectively appended to the terminal allylic alcohol to afford 73 in 47% yield. Alternatively, the cyclic ether 60n could be treated with allylmagnesium bromide to afford the allylsilane 74 directly in 45% yield. Unfortunately, when we evaluated the relay RCM with the tetraenes 73 and 74 using G1 or G2 catalysts, we only observed dimerization or decomposition of the substrates.47 These findings prompted us to consider a different approach for assembling the C-ring of the target illudins that would involve a more reactive olefin.
Accordingly, we revised our synthesis to incorporate a reductive allylic transposition reaction (Scheme 19). Through this strategy, alcohol 76 could be elaborated to the terminal olefin 77, which could then be converted to tricycle 78 via a RCM reaction. Oxidative dehydrogenation would then provide the fulvene 79.
Due to the difficulty of forming the triol 76 from the allylsilane through oxidative methods (60n→72, Scheme 18), we investigated alternative olefin tethers for the EYRCM cascade.12 In the midst of these studies, we made a tactical change to use allyloxydialkylsilyl tethers in the EYRCM (Scheme 20). These tethers obviated the problematic oxidation step and allowed direct access to the stable triol product from the EYRCM reaction via in situ removal of the tether. During the preliminary screening of several tethers using a model substrate,12 allyloxydiethylsilyl tether 80 demonstrated an optimal combination of stability and reactivity. Selective monosilylation of diol 54k–m with allyloxydiethylsilyl chloride 8048 gave the enyne metathesis substrates 81k–m in good yields (83–95%).
Our first generation synthesis of the tricyclic system began with the OPMB substrate 81m and featured a Stille cross coupling reaction to append the appropriate isopropenyl side chain for the final RCM step (Scheme 21). EYRCM of the p-methoxybenzyl ether substrate 81m followed by in situ TBAF cleavage of the oxysilane tether furnished the desired cyclohexenyl product 82m directly in 64% yield.12 In contrast to the allyldimethylsilyl tether (Scheme 16), the allyloxysilane tethered substrate 81m required a higher temperature (110 °C) to achieve complete conversion in the EYRCM reaction. Selective TBS protection of the triol 82m at the primary allylic alcohol followed by protection of the diol as a carbonate with triphosgene afforded compound 83. Removal of the PMB group by the action of 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) followed by bromination of the pendant alcohol then afforded intermediate 84, poised for a Stille cross-coupling. Isopropenyltributylstannane was coupled to the allylic bromide to generate substrate 85 in 35% yield to set up the olefinic side chain for the final RCM reaction. Desilylation of ether 85 gave the allylic alcohol 86, which was primed for a reductive allylic transposition reaction.
Exposure of allylic alcohol 86 to 2-nitrobenezenesulfonyl hydrazide (NBSH)49 under Mitsunobu conditions furnished the terminal olefin 87 (30%, 6S:6R, 3:1)34 via Myers’ reductive allylic transposition chemistry (Scheme 22). Gratifyingly, the planned RCM reaction employing G2 in benzene at 65 °C generated the C-ring to afford cyclopentene 88 in 45% yield (6S:6R, 3:1). Dehydrogenation with DDQ furnished the fulvene 89 in 93% yield, and hydrolytic cleavage of the carbonate afforded the diol 90 in 99% yield. o-Iodoxybenzoic acid (IBX) oxidation9g then provided acylfulvene (1) in 83% yield.50 In the course of these studies, we found a more efficient route that would circumvent derivatization of the side chain (Scheme 20). Thus, we used the optimal acetylides 53k and 53l (Scheme 11) that could be directly applied in the C-ring RCM step for the synthesis of acylfulvene (1).
We first evaluated the tandem EYRCM-desilylation sequence with the phenethyl derivative 81k (Table 3). As with intermediate 81m (Scheme 21), the EYRCM of 81k required high temperature (110 °C) and high catalyst loading of G2 (30 mol%) to achieve complete conversion. We reasoned that at this high temperature, the lifetime of the catalyst might be reduced. Using the optimal concentration (0.01M) and catalyst loading of G2 (30 mol%), the desired triol 82k was isolated in 52% yield, after removal of the silyl moiety with TBAF (entry 4, Table 3). Decreasing the concentration and raising the catalyst loading did not improve the yield of the final triol 82k (entries 5 and 6). Moreover, the use of milder desilylation condition or use of ruthenium scavengers51 during isolation afforded similar yields of the triol 82k. We speculated that at high temperature, partial loss of the allyloxydiethylsilyl tether, promoted by the vicinal hydroxyl group, was responsible for the low efficiency of the reaction.
In order to increase the stability of the enyne metathesis substrate and improve the yield of desired triol, the C2 tertiary hydroxyl group was converted to the corresponding trimethylsilyl ether. The reactivity of the silyl ether substrates 91k and 91l were significantly enhanced under the EYRCM conditions and required only 15 mol% catalyst loading of G2 at 90 °C (Scheme 23).10 After in situ desilylation of the EYRCM product, a mixture of the desired triol 82k and byproduct 92 were isolated in 52% and 20% yield, respectively. Conversely, the styrenyl derivative 91l10 containing a C7–C8 trisubstitued styrenyl alkene underwent the EYRCM cascade and desilylation reaction smoothly to afford the desired triol 82l exclusively in 79% yield (Scheme 23). The undesired product 92 was not observed for substrate 91l, which is consistent with the lower reactivity of styrenyl olefins under the EYRCM conditions.
The formation of the unexpected triol 92 was investigated in detail. In situ 1H NMR studies revealed that some trienyne 91k diverges from the desired EYRCM pathway (91k→82k, Scheme 24) to undergo a competing olefin metathesis with the C7–C8 alkene affording a ten-membered ring intermediate 95 (Scheme 24). Subsequent enyne metathesis and olefin isomerization52 of cyclic alkyne 95 produced the tricyclic disiloxane 96. In situ nOe analysis of intermediate 96 confirmed the E geometry for the C7–C8 olefin, which was opposite to the triol derived from desilylation of alkyne 95. Interestingly, lower reaction temperatures (80 °C) led to an increase in the yield of the olefin metathesis product 95, which is attributed to the higher energy barrier generally required for an EYRCM as compared to a RCM. The sensitive cyclic alkyne 95 was isolated and resubmitted to the optimal enyne metathesis conditions at higher temperature (90 °C) to give the triol 92 after silyl cleavage.
With the key triols 82k and 82l in hand, we evaluated the reductive allylic transposition reaction and RCM reaction for the completion of the synthesis of (−)-acylfulvene (1) and (−)-irofulven (2). We found it necessary to mask the tertiary and secondary alcohols, and developed a tandem process to generate the carbonates 97k and 97l (Scheme 25). Monosilylation of the allylic alcohols 82k and 82l with tbutyldimethylsilyl trifluoromethanesulfonate (TBSOTf, 1 equiv) selectively protected the primary alcohol. Sequential treatment with triphosgene and treatment with TBAF afforded the desired carbonates 97k and 97l in good overall yields in a single flask.
Substrates 97k and 97l were subjected to Myers’ reductive allylic transposition reaction to give desired trienes (Table 4).49 Low temperature Mitsunobu displacement with NBSH generates the allylic hydrazide derivatives, which upon warming, spontaneously lose 2-nitrobenzene sulfinic acid followed by dinitrogen to afford the desired terminal olefins 99k–l.
When treated with diethylazodicarboxylate (DEAD), triphenylphosphine (PPh3), and NBSH at 0.02M concentration in N-methyl morpholine (NMM), the allylic alcohol 97k provided the desired product 99k (6S:6R, 3:1) along with a significant amount of unreacted starting material (entry 1, Table 4). By increasing the concentration of the reaction mixture49a we observed full consumption of the alcohol 97k; however, the yield was still unsatisfactory (43%, entry 2, Table 4). Careful examination of this reaction revealed that thermal decomposition of the unreacted NBSH generated diimide in the reaction mixture, which reduced a significant amount (19%) of the product 99k at the C7–C8 terminal olefin. Gratifyingly, addition of allylbenzene as a scavenger for the diimide and further increasing the reaction concentration afforded the desired product 99k in 75% yield (entry 3). Unfortunately, when we tried to apply these conditions to the reductive allylic transposition of substrate 97l, the yield of the isolated product 99l was modest (54%, entry 4) as a result of the poor solubility of this substrate.10 To address the lack of reactivity of alcohol 97l, we added neopentyl alcohol to improve the efficiency of the Mitsunobu displacement.53 Unfortunately, neopentyl alcohol further decreased the solubility of the substrate resulting in poor yield of olefin 99l (35%, entry 5). The use of THF in place of NMM improved the homogeneity of the reaction mixture, but also increased the formation of undesired byproducts (entry 6). Furthermore, a mixture of THF and NMM as solvent did not improve the efficiency of allylic transposition (entries 7–8). Due to the insolubility of the substrate in the reaction media at low temperature and at high concentration, variable yields of the desired product were obtained.
In order to address the complications associated with substrate 97l, we considered the use of a more stable derivative of NBSH that would allow us to carry out the challenging Mitsunobu displacement at higher temperatures and lower solvent concentrations. Thus, the acetone hydrazone derivative, N-isopropylidene-N’-2-nitrobenzenesulfonyl hydrazine (IPNBSH),54 was prepared and used for the reductive allylic transposition of alcohol 97l (Table 5). We were pleased to find that the Mitsunobu displacement of alcohol 97l with IPNBSH proceeded smoothly at temperatures between 5–23 °C and at lower concentrations to give the stable hydrazone intermediate 100l. Exposure of intermediate 100l to hydrolytic conditions then afforded transposition product 99l. Water alone was insufficient for the hydrolysis of the hydrazone (entry 1, Table 5); however, the addition of alcoholic co-solvent greatly enhanced the yield and rate of formation of the product 99l (entries 2–5). Interestingly, the solvolysis of hydrazone 100l using 2,2,2-trifluoroethanol (TFE) at 0 °C occurred with the greatest efficiency to afford the desired olefin 99l in 71% yield (entry 5).
Preliminary studies on the final steps of the synthesis were carried out using triene 99k. Treatment of triene 99k with 15 mol% of G2 at 65 °C resulted in clean conversion to the desired diene 88 (82%, 6S:6R, 7.6:1, Scheme 26), which was accessed in our first generation synthesis (Scheme 22). Isolation of the carbonate from this reaction mixture was found to be problematic. The (6R)-diastereomer of carbonate 88 was particularly sensitive to silica gel chromatography. Furthermore, oxidation of the minor isomer (6R)-88 proved to be very slow. Therefore, we carried forward only the major diastereomer 6S-88 through the remaining steps of the sequence shown in Scheme 26.
We were pleased to find that oxidation of the cyclopentene (6S)-88 with DDQ afforded the desired fulvene carbonate 89 in 93% yield (Scheme 26) in a manner similar to the first generation route described above (Scheme 22). Subsequent hydrolysis of the carbonate 89 gave the diol fulvene 90 as reported by Brummond and coworkers.9f The synthesis of acylfulvene (1)55 was then completed by oxidation of the secondary alcohol with IBX.9g
With the final steps of the synthesis of acylfulvene (1) in place, we focused on streamlining the final stages of syntheses of (−)-acylfulvene (1) and (−)-irofulven (2). These final optimizations were performed on enantiomerically enriched samples of triene 99l prepared from the key aldehyde (+)-22. In order to efficiently convert both diastereomers of triene 99l to the final diol fulvene 90 we bypassed the isolation of the sensitive carbonate 88 (Scheme 27) via an in situ hydrolysis of the carbonate. Thus, after the RCM of triene 99l, the mixture was sequentially diluted with dimethylformamide (DMF) and treated with aqueous lithium hydroxide. The resulting diol 101 was quickly subjected to aqueous work up, filtered through silica gel, and immediately oxidized to the desired diol fulvene 90 using chloranil in 70% yield over the three steps.
We then established a tandem process to include the RCM, hydrolysis, and dehydrogenation in a single flask. Thus, the triene 99l was subjected to a three step sequence involving the RCM, carbonate hydrolysis, and sequential chloranil oxidation to afford the desired diol fulvene 90 directly in 70% yield (Scheme 28). Interestingly, by replacing chloranil with DDQ, a more potent oxidant, the triene 99l could be converted directly to the target (−)-acylfulvene (1) in 30% yield without isolation of any intermediates (Scheme 28). Finally, (−)-acylfulvene (1) was converted to (−)-irofulven (2) in 63% yield using the protocol described by McMorris and coworkers.5,10 All spectroscopic data for (−)-acylfulvene (1) and (−)-irofulven (2) matched those reported in the literature.
We have described the development of our synthesis of two potent antitumor agents (−)-acylfulvene (1) and (−)-irofulven (2). The optimal sequence is summarized in Scheme 29. The asymmetric copper catalyzed Evans aldol addition reaction with the strained ketene acetal 26 secured the C2 stereocenter of the target compounds. The powerful EYRCM cascade reaction with the allyloxysilane tether was successfully employed for the B-ring construction. The successful implementation of this strategy required the identification of optimal derivatives for rapid post-EYRCM derivatization. The reagent IPNBSH efficiently provided the necessary reductive transposition of an advanced allylic alcohol. Finally, a tandem RCM/dehydrogenation process was employed for the C-ring construction to complete the synthesis of (−)-acylfulvene (1) and (−)-irofulven (2).
A flame-dried flask was charged with (R,R)-2,2′-isopropylidene-bis(4-tert-butyl-2-oxazoline) (2.03 g, 6.90 mmol, 0.10 equiv)56 and copper (II) trifluoromethanesulfonate (2.50 g, 6.90 mmol, 0.10 equiv) in a glove-box under a dinitrogen atmosphere. The flask was sealed with a rubber septum and removed from the glove-box. The flask containing the solids was charged with THF (304 mL) at 23 °C and was flushed with argon. After 1h, the resulting bright green solution was cooled to −78 °C, and methyl pyruvate (25, 7.80 g, 76.0 mmol, 1.10 equiv) was added via syringe followed by (cyclopropylidene-ethylsulfanyl-methoxy)-trimethyl-silane (26b [mixture of 26b:47b = 9:1], 15.5 g, 69.0 mmol, 1 equiv 26b) via syringe. After 19 h, the reaction mixture was diluted with diethyl ether (300 mL), and filtered through a plug of silica gel (6 × 6 cm, eluent: 1% triethyamine in diethyl ether). The filtrate was concentrated under reduced pressure and the residue was purified by flash column chromatrography (silica gel: diam. 9 cm, ht. 15 cm; eluent: 1% triethylamine in [2% ethyl acetate in hexanes] to 1% triethylamine in [20% ethyl acetate in hexanes]) to afford the desired (2R)-2-(1-ethylsulfanylcarbonyl-cyclopropyl)-2-(trimethyl-silanyloxy)-propionic acid methyl ester (48b, 19.8 g, 95%, [α]20D = +30.2 (c 2.22, CHCl3)) as a colorless liquid. Protodesilylation of the C2-trimethylsilyloxy group of 48b afforded samples of the corresponding C2-alcohol that were found to be of 92% ee by chiral HPLC analysis [Chirapak AD-H; 1.5 mL/min; 10% iPrOH in hexanes; tR(minor) = 4.65 min, tR(major) = 5.17 min]. The (R,R)-2,2′-isopropylidene-bis(4-tert-butyl-2-oxazoline) ligand was recovered from the reaction mixture (~85%) and purified by flash column chromatography (silica gel: diam. 2.5 cm, ht. 10 cm; eluent: 20% ethyl acetate in dichloromethane). TLC (10% ethyl acetate in hexanes), Rf: 0.4 (UV, CAM). 1H NMR (500 MHz, CDCl3): δ 3.72 (s, 3H), 2.79 (q, J = 7.3 Hz, 2H), 1.58–1.54 (m, 1H), 1.53 (s, 3H), 1.27–1.19 (m, 2H), 1.19 (t, J = 7.3 Hz, 3H), 1.12–1.08 (m, 1H), 0.07 (s, 9H). 13C NMR (125.8 MHz, CDCl3): δ 200.9, 173.4, 75.4, 52.1, 41.8, 24.2, 23.0, 15.3, 14.8, 11.6, 1.5. FTIR (neat) cm−1: 2954, 1747, 1666, 1456, 1413, 1372, 1289, 1263. HRMS (ESI): calc’d for C13H24NaO4SSi [M+Na]+: 327.1057, found: 327.1066.
n-Butyllithium (2.50 M in hexanes, 100 μL, 250 μmol, 1.30 equiv) was added dropwise via syringe to a solution of diisopropylamine (37.0 μL, 270 μmol, 1.40 equiv) in THF (300 μL) at 0 °C. After 30 min, the mixture was cooled to −78 °C and a solution of alkyne 53c (55.0 mg, 230 μmol, 1.20 equiv) in THF (0.9 mL) was added dropwise via cannula. After 35 min, a solution of the aldehyde 22 (43.0 mg, 190 μmol, 1 equiv) in THF (0.6 mL) was added dropwise via cannula. After 2 h, saturated aqueous ammonium chloride solution (0.5 mL) was added. The resulting mixture was allowed to warm to 23 °C, was diluted with diethyl ether (40 mL) and was washed with water (10 mL). The aqueous layer was extracted with diethyl ether (2 × 40 mL), and the combined organic layers were dried over anhydrous magnesium sulfate and were concentrated under reduced pressure. The crude silyl ether residue was dissolved in THF (3 mL) and to this solution was added hydrogen fluoride-triethylamine complex (20.0 μL, 190 μmol, 1.00 equiv) at 0 °C. After 2 h, saturated aqueous sodium bicarbonate solution (3 mL) was added, and the resulting mixture was warmed to 23 °C and was diluted with diethyl ether (100 mL) and water (10 mL). The aqueous layer was extracted with diethyl ether (2 × 40 mL), and the combined organic layers were dried over anhydrous sodium sulfate and were concentrated under reduced pressure. The resulting crude oil was purified by flash column chromatography (silica gel: diam. 2 cm, ht. 16 cm; eluent: 75% diethyl ether in n-pentane) to afford the desired diol 54c (46 mg, 63%, (3S:3R, 4:1), 2:1 mixture of C6 diastereomers). TLC (15% diethyl ether in n-pentane) Rf: 0.15 (Anis). 1H NMR (400 MHz, C6D6, 4:1 mixture of (3S)- and (3R)-diastereomers; major (3S)-diastereomer reported): δ 6.06 (ddd, J = 17.2, 10.8, 6.3 Hz 1H), 5.11–5.03 (m, 3H), 4.90 (br-s, 1H), 4.43 (d, J = 5.6 Hz, 1H), 4.09 (br-s, 1H), 1.89 (br-s, 1H), 1.76 (br-s, 3H), 1.64 (br-s, 1H), 1.27 (s, 3H), 1.27–1.17 (m, 1H), 1.15 (s, 3H), 1.14 (s, 3H), 0.99 (s, 9H), 0.95–0.84 (m, 1H), 0.60–0.52 (m, 1H), 0.46–0.40 (m, 1H), 0.25 (s, 3H), 0.10 (s, 3H). 13C NMR (100.6 MHz, C6D6): δ 147.9, 145.2, 118.1, 113.0, 87.6, 85.3, 75.0, 71.3, 69.7, 43.1, 33.2, 26.2, 23.6, 23.2, 23.0, 22.9, 18.6, 10.9, 9.5, −3.9, −4.9. FTIR (neat) cm−1: 3457, 3082, 2958, 1637, 1472, 1252, 1082. HRMS (ESI): calc’d for C23H40NaO3Si [M+Na]+: 415.2639, found: 415.2631.
To a solution of the diol 54j (100 mg, 420 μmol, 1 equiv, (3S:3R, 6.7:1)) in dichloromethane (2 mL) at 23 °C was added triethylamine (175 μL, 1.26 mmol, 3.00 equiv) followed by allylchlorodimethylsilane (79.0 μL, 510 μmol, 1.20 equiv) via syringe. After 40 min, the resulting mixture was purified directly by flash chromatography (silica gel: diam. 3.0 cm, ht. 15 cm; eluent: 9% diethyl ether in n-pentane) to afford dienyne 55j (127 mg, 83%, (3S:3R, 8:1)) as a clear colorless oil. TLC (25% diethyl ether in hexanes) Rf: 0.50 (UV, Anis). 1H NMR (400 MHz, C6D6, 8:1 mixture of (3S)- and (3R)-diastereomers; major (3S)-diastereomer reported): δ 5.93–5.78 (m, 1H), 5.28 (d, J = 2.5 Hz), 5.02–4.91 (m, 3H), 4.64 (t, 1H, J = 1.7 Hz), 2.14 (br-s, 1H), 1.91–1.89 (m, 5H), 1.78–1.60 (m, 3H), 1.54–1.46 (m, 1H), 1.36 (br-s, 3H), 1.17–1.10 (m, 1H), 0.86 (d, J = 6.7 Hz, 6H), 0.69–0.63 (m, 1H), 0.56–0.50 (m, 1H), 0.22 (s, 3H), 0.20 (s, 3H). 13C NMR (100.6 MHz, C6D6, 8:1 mixture of (3S)- and (3R)-diastereomers; major (3S)-diastereomer reported): δ 147.7, 134.2, 117.7, 114.0, 86.7, 81.0, 74.6, 70.3, 32.6, 28.1, 28.0, 25.2, 23.8, 23.5, 22.0, 11.0, 9.2, −1.5, −1.9. FTIR (neat) cm−1: 3570, 3078, 2959, 2925, 2230, 1632, 1374, 1253, 1062, 859. HRMS (ESI): calc’d for C20H34NaO2Si [M+Na]+: 357.2220, found: 357.2235.
Silyl ether 55o (176 mg, 530 μmol, 1 equiv) was dissolved in benzene (35.0 mL) in a Schlenk vessel. The resulting solution was degassed thoroughly by passage of a stream of argon and G2 (44 mg, 53 μmol, 0.10 equiv) was added as a solid. After 5 min, the light pink reaction mixture was heated to 65 °C by placement in a pre-heated oil bath. After 1 h, the catalyst was quenched by addition of ethylvinyl ether (0.5 mL). After 5 min, the reaction mixture was cooled to 23 °C and the solvent volume was reduced to ~50% under reduced pressure. The resulting mixture was immediately purified by flash chromatography (silica gel: diam. 4 cm, ht 15 cm; eluent: 10% ethyl acetate in hexanes) to afford the desired diene 60o (96 mg, 59%) as a clear colorless oil. TLC (10% ethyl acetate in hexanes), Rf: 0.3 (UV, Anis). 1H NMR (500 MHz, C6D6): δ 9.32 (t, J = 1.5 Hz, 1H), 5.80–5.75 (m, 1H), 4.19 (s, 1H), 2.60 (br-s, 1H), 2.43–2.38 (m, 2H), 2.08–1.96 (m, 2H), 1.30 (dd, J = 13.0, 7.5 Hz, 1H), 1.20 (dd, 1H, J = 13.0, 7.5 Hz), 1.14 (s, 3H), 1.12 (s, 3H), 1.00–0.94 (m, 1H), 0.78–0.72 (m, 1H), 0.56–0.50 (m, 1H), 0.49–0.44 (m, 1H), 0.05 (s, 3H), 0.02 (s, 3H). 13C NMR (125.8 MHz, C6D6): δ 200.3, 137.9, 132.2, 128.9, 120.8, 76.4, 71.8, 43.0, 28.9, 22.3, 21.2, 14.6, 13.9, 8.1, 7.6, −0.6, −1.0. FTIR (neat) cm−1: 3535, 2927, 1720, 1377, 1253, 1102. HRMS (ESI) calc’d for C17H26NaO3Si [M+Na]+: 329.1543, found: 329.1548.
M.M. is an Alfred P. Sloan Research Fellow, a Beckman Young Investigator, and a Camille Dreyfus Teacher-Scholar. We thank Dr. Michael A. Schmidt, Justin Kim, and Dr. Peter Müller for assistance with X-ray crystallographic analysis. We are grateful to Professor A. H. Hoveyda and Professor E. N. Jacobsen for generous samples of their silylcyanation catalysts. We acknowledge financial support in part by the Damon Runyon Cancer Research Foundation (DRS-39-04), NIH-NIGMS (GM074825), Amgen, and Astrazeneca.
Supporting Information Available: Experimental procedures and spectroscopic data for new products. This material is available free of charge via the Internet at http://pubs.acs.org.