|Home | About | Journals | Submit | Contact Us | Français|
This report describes a number of new synthetic approaches towards methyl-substituted mono- and diepoxy alcohols that serve as substrates for endo-selective epoxide-opening cascades. The key transformations involve the manipulation of alkynes. Highlighted are the directed methylmetalation of bishomopropargylic alcohols, the bromoallylation of alkynes, and Pd-catalyzed cross-coupling between an alkenyl boronate ester and allylic bromides.
Skipped dienes, also known as 1,4- or homoconjugated dienes, are a key structural motif found in a number of important natural products, including several leukotrienes, the omega-3 fatty acids, insect pheremones,i and the macrolide amphidinolide A, among many others. Skipped polyene structures are also notable for their proposed intermediacy in the biosynthesis of ladder polyether natural products (such as brevetoxin B, 3, Scheme 1).
These natural products, notorious for their acute toxicity and for the environmental destruction they wreak, have long fascinated organic chemists due to their intriguing mixture of structural complexity and stereochemical uniformity.ii It was Nakanishi who, nearly 25 years ago, advanced a concise proposal for the biosynthesis of brevetoxin B and other ladder polyethers, a synthesis that culminates in a cascade of regio- and stereospecific epoxide openings.iii The appeal of this hypothesis lies in its economy. In vivo, a single stereoselective epoxidase, chemoselective for E alkenes, acting on polyene 1 could generate all (R,R)-“skipped polyepoxide” 2, which could in turn undergo an all-endo cascade of epoxide-opening cyclizations to afford brevetoxin B, 3.iv
Abbreviated skipped polyenes similar to 1 can be useful substrates for epoxidation and subsequent studies of biomimetic in vitro epoxide-opening cascades. Construction of these polyepoxides requires a protocol that is reliably highly (R,R)- or (S,S)-selective for a wide variety of trans-disubstituted and trisubstituted alkenes. The powerful Shi asymmetric epoxidation,v in which the active catalyst is a chiral dioxirane generated in situ from a simple, fructose-derived ketone, has emerged as just such a method. If asymmetric epoxidations of this type are justifiably taken to be something of a “solved problem,” the stereocontrolled synthesis of the necessary trans-disubstituted and E-trisubstituted alkenes (and polyenes) then becomes the primary challenge.
The difficulty of synthesizing highly substituted alkenes with careful stereo- and regiocontrol is well documented. vi Furthermore, the construction of skipped dienes can be additionally troublesome due to the rather delicate doubly allylic protons found in these compounds. The pKa of 1,4-pentadiene (divinylmethane) in DMSO is approximately 35, vii making doubly allylic protons sensitive to strong base, and the same C-H bonds are also susceptible to hydrogen atom abstraction, viii which can lead to subsequent di-Π-methane rearrangement.ix Thus the synthesis of skipped dienes and polyenes must generally avoid very strong bases (such as alkyllithiums) and reagents that promote single electron processes.
Strategies for skipped polyenes previously explored in our group targeted the synthesis of trimethylsilyl (TMS)-substituted skipped polyenesx and all-trans-disubstituted polyenes xi from bishomopropargylic alcohol 4 (Scheme 2). These polyenes were then transformed into the polyepoxy alcohols 5 and 6, respectively, with high diastereoselectivity via Shi epoxidation. Polyepoxy alcohols 5 and 6 then underwent endo-selective epoxide-opening cascades (accompanied by concomitant protiodesilylation in reactions of 5) to afford THP tetrad 7, a characteristic structural subunit of the ladder polyether natural products.10,11
Versatile alkyne 4 was transformed into the skipped triene precursor to 5 via the copper-mediated cross-coupling of a propargyl nucleophile and a TMS-substituted alkenyl iodide.10c The all-trans-disubstituted triene precursor to 6 was obtained from dissolving metal reduction of a skipped triyne. Thus the two quite differently substituted triepoxides 5 and 6 could both be straightforwardly derived from common intermediate 4, and both 5 and 6 cyclized to afford tetrad 7.
It is notable that the ring junctions of 7 are all hydrogen-substituted. A cursory glance at the ladder polyether natural products reveals that methyl (Me) substituents are also frequently encountered at ring junctions (as in brevetoxin B (3, Scheme 1), which bears Me groups at five of its ten ring junctions). Methyl substituents also appear in 1 and 2, Nakanishi’s hypothesized polyepoxide and polyene precursors to brevetoxin B. The Me groups that decorate the trisubstituted epoxides of 2 are encountered both distal and proximal to the internal nucleophile.xii This point is critical, as the Me group can exert a rather powerful directing effect on the regioselectivity of epoxide opening, particularly under acidic activation.xiii
We recently reported that water overcomes methyl group directing effects in endo-selective epoxide-opening cyclizations to enable the rapid construction of fused tetrahydropyran (THP) bicycles and tricycles with the incorporation of angular Me groups.xiv In that report we described the effects of Me substitution in epoxide-opening reactions of diversely methyl-substituted mono- and diepoxy alcohols 8–12 (Scheme 3). The key propargylation and dissolving metal reactions used for diene and triene syntheses en route to 5 and 6 were unfortunately not readily amenable to the controlled incorporation of Me substituents into alkenes and skipped polyenes. We were therefore obligated to develop a number of new synthetic strategies for the synthesis of 8–12. We herein describe these strategies for the stereo- and regiocontrolled synthesis of trisubstituted epoxides (and their trisubstituted alkene precursors). Two of the key transformations highlighted in this work, Thompson’s directed methylmetalation of alkynesxv and Kaneda’s haloallylation of alkynes,xvi have to our knowledge been rather infrequently used in total synthesis and related applications, but both methods have proved indispensable to our work. Furthermore, all of the syntheses detailed here begin from common intermediate 4. We believe that this work further emphasizes the utility and versatility of alkyne 4 as a synthetic platform.
Bishomopropargylic alcohol 4 is prepared easily on a gram scale in five steps from 2,3-dihydropyran.xvii We began our exploration into new transformations of 4 by tackling the synthesis of monoepoxy alcohol 8. Compound 8 bears a proximal methyl substituent and is therefore Me-substituted at both sides of the epoxide. Consequently, we planned to approach 8 via syn-selective 1,2-difunctionalization of alkyne 4 (or a protected derivative of 4).
Repeated attempts at 1,2-difunctionalization of 4 and its TBS ether 13 via Negishi carbometalationxviii resulted either in very low or undetectable conversion under standard conditions (2 equiv AlMe3 and 1 equiv Cp2ZrCl2 in CH2Cl2 or 1,2-dichloroethane at room temperature) (Scheme 4). Heating above 30° C led only to the decomposition of starting material. Attempted methylalumination of 4 and 13 with the stoichiometric addition of water according to the modified protocol of Wipf and Limxix likewise led to poor conversion and significant decomposition. Similarly, methylcuprationxx of 13 was too slow to be practical. Thus the highly beguiling rapid, syn-stereoselective methylmetalation for the eventual 1,2-dimethylation of alkynes was unfortunately but a siren’s song, leading only to failure in our early studies.
Other attempts at 1,2-difunctionalization of 13 via silylcupration and stannylcuprationxxi were also unsuccessful, with low conversion observed in all cases and only stannylcuprationxxii yielding even a trace of the desired adduct.
It appeared that 4 and 13 were relatively inert under standard undirected carbometalation and metallometalation conditions. In the case of 13, the minimal alkyne reactivity may perhaps be due to the steric demand of the neighboring bulky silyl group in the lowest energy conformer of the molecule (alkyne and silyl ether both equatorial).
This inferred proximity could be turned to our advantage, however, in carbometalation directed by the free hydroxyl group in 4 (Scheme 4). Thompson’s method for the directed methylmetalation of propargylic, homopropargylic, and bishomopropargylic alcohols with TiCl4 and AlMe3,15 recently used in the total synthesis of (−)-borrelidin by the Omura group,xxiii proved highly effective. Methylmetalation of 4, I2 treatment, and subsequent silyl protection afforded alkenyl iodide 16 in a modest 41% yield, but with >20:1 regio- and stereoselectivity.
Alkenyl iodide 16 could be transformed into epoxide-opening cyclization substrate 8 in three simple steps via Negishi coupling with dimethylzinc, Shi asymmetric epoxidation (Shi AE), and TBAF-induced deprotection of the TES ether (Scheme 5). The Shi asymmetric epoxidation step proceeded with considerably lower than is typically observed in Shi epoxidations of trisubstituted alkenes. We conjecture that this poor diastereoselectivity arises from a mismatched relationship between the chiral dioxirane epoxidation agent and the stereocenters of the neighboring THP ring. However, as we have not attempted epoxidation with the antipodal ketone catalyst (derived from expensive L-fructose), we cannot confirm this hypothesis.
With epoxy alcohol 8 in hand, we turned to the synthesis of its isomer 9, an epoxide-opening cyclization substrate that bears a distal Me substituent (located on the far side of the epoxide with respect to the alcohol). This Me group was appended via alkylation of a derivative of alkyne 4 with iodomethane to afford bishomopropargylic alcohol 19 (Scheme 6). As the synthetic elaboration of internal alkyne 19 requires the incorporation of only one additional Me group, we initially attempted to perform a hydrometalation/methylation sequence on 19 and its silyl-protected derivatives with Schwartz’s reagent, DIBAL, or 9-BBN, but observed in all cases low yield, poor regioselectivity, or both.
Instead of hydrometalation, standard Thompson carbometalation conditions (2.2 equiv AlMe3 and 1.1 equiv TiCl4 in CH2Cl2 at −78°) proved best. Upon quenching with aqueous acid, carbometalation smoothly afforded trisubstituted alkene 20 in 70% yield (79% based on recovered starting material) and with no observable trace of undesired regio- or stereoisomers. Silyl protection, asymmetric epoxidation, and deprotection then provided distally Me-substituted epoxy alcohol 9.
In general, the alkenylmetal products of carbometalation are highly versatile and thus highly valuable, as they can be functionalized directly in a quenching operation with a variety of simple electrophiles (e.g. H+, I2, a few highly electrophilic carbon electrophiles) or transmetalated to Pd or Cu and subsequently used as nucleophiles in conjugate addition and cross-coupling processes.xxiv Unfortunately, however, Thompson carbometalation involves delivery of the metal to the proximal end of alkynes 4 and 19 (and Me to the distal end). This regiochemistry precludes using transmetalation of the alkenylmetal and subsequent coupling for the construction of linear skipped dienes or polyenes.
For the assembly of the linear skipped dienes necessary for the synthesis of diepoxy alcohols 10, 11, and 12, the development of new strategies became necessary. While terminal alkynes 4 and 13 were nearly inert to undirected carbometalation with Cu and Al/Zr systems, we elected to try these and closely related substrates in a catalytic halopalladation process. To our delight, halopalladation proved straightforward and efficient.
The palladium-catalyzed haloallylation of alkynes was discovered by Kaneda, Teranishi, and coworkers.16 It has seen some use in recent years, with notable extensions of the method reported by the Rawal groupxxv and a synthetic application described by Hoye and coworkers.xxvi It has proved invaluable to our own work. The method effects chloro- and bromoallylation of alkynes via the proposed mechanism shown in Scheme 7.16,xxvii The first step is regio- and syn-stereoselective alkyne hallopalladation to afford alkenylpalladium halide 23. After binding and 1,2 insertion of an allylic halide to give 24, β-halide elimination gives diene 25.
Trans-selective halocrotylation of 26, the TBDPS ether of 4, would provide direct access to the synthetically useful skipped diene 27 (Scheme 8), which bears the appropriate substitution at both the disubstituted and trisubstituted alkenes. This transformation proved problematic in practice. Bromocrotylation is not straightforward, as the requisite 3-bromo-1-butene is not commercially available and isomerizes to crotyl bromide at ambient temperature.xxviii Chlorocrotylation with 3-chloro-1-butene was successful, but the E:Z selectivity was only 1.7:1. Moreover, all attempts to effect the cross-coupling of the isomeric alkenyl chloride products with dimethylzinc via a Pd/P(tBu)3 catalyst systemxxix were unsuccessful.
We found more success with simple bromoallylation and subsequent homologation. Bromoallylation of 26 under conditions slightly modified from the original Kaneda conditions (dropwise addition of the alkyne to a room temperature solution of 5 mol% PdCl2(PhCN)2 in allyl bromide, with the addition of 5 equiv of NaHCO3xxx) afforded diene 28 in 88% yield, with no trace of undesired regio- and stereoisomers (Scheme 9). Methylation of alkenyl bromide 28 via Negishi coupling with dimethylzinc and a subsequent Shi asymmetric epoxidation chemoselective for the more electron-rich trisubstituted alkene provided epoxy alkene 29. This monosubstituted alkene was readily and nearly quantitatively converted to the methylated disubstituted alkene via cross metathesisxxxi with cis-2-butenexxxii using the Hoveyda-Grubbs 2nd generation metathesis catalyst 31 (5 mol%),xxxiii in respectable 4.1:1 E:Z stereochemical purity. While a simple Me group is added in this reaction, cross metathesis is of course a versatile strategy that also makes possible the appendage of more complex epoxy alkenes for the eventual synthesis of longer epoxide chains.xxxiv Shi epoxidation of the resultant disubstituted alkene afforded the diepoxide in modest (2.7:1) overall dr, but we were able to improve this to better than 20:1 dr via preparative HPLC. The choice of TBDPS as silyl protecting group was in large part due its UV activity, which made detection on HPLC possible. Finally, cleavage of the silyl ether afforded diepoxy alcohol and epoxide-opening cascade substrate 10.
These results in hand, we were initially hopeful that Kaneda haloallylation could be extended to the synthesis of diepoxy alcohol cascade substrates 11 and 12. Both of these substrates bear a Me substituent on the distal side (the endo site of attack) of the epoxide closer to the THP ring. Unfortunately, bromoallylation of internal alkyne 32 unexpectedly proceeded with a reversal of regioselectivity as compared to that observed with terminal alkyne 26 such that undesired regioisomer 33 predominated (Scheme 10).
Ultimately, for the synthesis of diepoxide 11, we took recourse to the hydrometalation/cross-coupling strategy first investigated for the preparation of distally trisubstituted monoepoxide 9. Hydrozirconation of 32 with Cp2Zr(H)Cl (Schwartz’s reagent) proceeded in good yield, but the regioselectivity of addition across the alkyne was nearly 1:1, and separation of the resultant regioisomers after either iodinolysis of the C-Zr bond or transmetalation to Zn and quench with carbon electrophiles proved troublesome.
Uncatalyzed hydroboration of 32 with pinacolborane was so slow as to be entirely impractical, but the addition of catalytic quantities (0.1 equiv) of Cp2Zr(H)Cl and triethylamine according to the modification of Wang, et al.xxxv to the method developed by Pereira and Srebnikxxxvi improved the reaction rate (Scheme 11). However, even with Cp2Zr(H)Cl catalysis and upon heating the hydroboration reaction mixture to 60° for 28 h, conversion of starting alkyne 32 to pinacolates 35 and 36 was typically only 60–80%. While less than ideal, we found that these conditions represented a reasonable compromise, as longer reaction times and higher temperatures led to lower mass recovery. While poor conversion is a drawback of this transformation, unreacted 32 could be recovered and recycled. Furthermore, the stereochemistry of addition was cleanly syn, the regioselectivity was 2.0:1 in favor of the desired 35 over the undesired 36, and, most important, the two regioisomeric pinacolate esters were readily separable by column chromatography.
After purification, alkenyl boronate ester 35 was cross-coupled with allyl bromide via Pd-catalyzed conditions adapted from those developed by Miyaura and coworkers.xxxvii The addition of 2 equivalents of water was found to improve the reaction rate substantially, presumably by effecting in situ hydrolysis of the boronate ester to the more reactive boronic acid. Bidentate bisphosphine ligands like dppe and the best ligand identified, dppf, were found to improve reaction rate as compared to those with monodentate ligands such as PPh3, PCy3, or Buchwald ligands.xxxviii
The crude diene was carried into Shi epoxidation (again chemoselective for the more electron-rich trisubstituted olefin over the monosubstituted) to afford epoxy alkene 37. Cross metathesis of 37 with cis-2-butene again proceeded in excellent yield and good stereoselectivity (4.2:1 E:Z), and the resulting disubstituted alkene was subjected to another Shi epoxidation to afford the diepoxide in disappointing 1.5:1 overall dr. The 1.5:1 mixture of diastereomers was enhanced to greater 9:1 diastereopurity via preparative HPLC, and final TBAF deprotection gave diepoxy alcohol and cascade substrate 11 in excellent yield.
The synthesis of the third and final diepoxy alcohol cascade substrate 12, which bears methyl substituents on both epoxides, intercepted the synthesis of 11 at alkenyl boronate ester 35 (Scheme 12). Cross coupling of 35 with prenyl bromide (rather than allyl) proceeded in similarly good 75% overall yield, but the coupling, which goes through an unsymmetrical π-allyl intermediate, resulted in a 2.5:1 mixture of the desired SN2 product 39 along with undesired SN2’ product 40.
These diene isomers were inseparable by column chromatography, but Shi epoxidation proved chemoselective yet again, epoxidizing both trisubstituted alkenes in 39, but leaving untouched the monosubstituted olefin in 40 to give a readily separable mixture of monoepoxide 41 and diepoxide 42 (65% yield based on 39, 3.5:1 dr). The diastereopurity of 42 was improved via preparative HPLC, and samples of ≥15:1 dr were subjected to TBAF deprotection to afford diepoxy alcohol 12.
We have developed synthetic routes to trisubstituted mono- and diepoxides 8–12, which serve as substrates in endo-selective epoxide-opening cyclizations and cascades. The longest of these routes (the synthesis of 11) is eight steps from known homopropargylic alcohol 4. We believe that these transformations highlight the remarkable versatility of common intermediate 4. Further exploration into new methods for the stereocontrolled construction of skipped polyenes and epoxide chains is ongoing in our laboratory.
Full experimental details for compounds 4, 8–12, 16–17, 19–21, 26, 28–30, 32, 35, 37–38, and 42–43 are available in the Supporting Information to reference 14.
All air-sensitive reactions were performed under an oxygen-free atmosphere of argon with rigid exclusion of moisture from reagents and glassware. Dichloromethane was either distilled from calcium hydride or purified via an SG Water USA solvent column system. Tetrahydrofuran (THF) and Et2O were either distilled from a blue solution of benzophenone ketyl or purified via an SG Water USA solvent column system. Triethylamine was purified via an SG Water USA solvent column system. Methyl iodide and allyl bromide were purified by filtration through basic alumina before use. K3PO4 was oven-dried overnight before use. Chiral ketone 18, used in Shi asymmetric epoxidation, was prepared from D-fructose according to the procedure of Vidal-Ferran and coworkers.xxxix All other reagents and solvents were used as obtained, without further purification.
To a solution of bishomopropargylic alcohol 4 (265 mg, 1.89 mmol) in CH2Cl2 (20 mL) at −78° was added slowly a 2 M solution of AlMe3 in hexanes (2.08 mL, 4.16 mmol). This solution was warmed for 3 min. by removing the flask from its cold bath to ensure complete deprotonation. After recooling to −78°, a 1 M solution of TiCl4 in CH2Cl2 (2.08 mL, 2.08 mmol) was added dropwise. The solution was stirred 2 h. at −78° and then quenched with a solution of I2 (2.4 g, 9.45 mmol) in Et2O (20 mL). The reaction flask was then wrapped in foil and allowed to warm to room temperature for 10 h., at which point H2O (2 mL) was added. The quenched reaction solution was diluted with Et2O (100 mL) and washed with aqueous 3 M NaHSO3 (40 mL) until the organic layer was colorless. The aqueous layer was extracted with Et2O (3 × 40 mL), and the combined organics were washed with sat. NaCl, dried over MgSO4, and concentrated in vacuo. The crude alkenyl iodide was carried forward without purification; Rf = 0.72 (50% EtOAc in hexanes), UV active.
Upon dissolution of this crude in DMF (1 mL), imidazole (322 mg, 4.73 mmol) and TESC1 (380 µL, 342 mg, 2.27 mmol) were added, and the solution was stirred 2 h. at room temperature. The reaction was applied directly to column of SiO2 and chromatographed (2% EtOAc in hexanes) to afford 16 (310 mg, 0.78 mmol, 41% over 2 steps), which was isolated along with some silylated bishomopropargylic alcohol (43 mg, 0.17 mmol). Rf of 16 = 0.60 (10% EtOAc in hexanes), UV active; [α]22D = −19.2 (c = 4.0, CHCl3).
IR (thin film, NaCl) 2955, 2876, 1461, 1415, 1274, 1239, 1127, 1102, 1004 cm−1
1H NMR (500 MHz, CDCl3) δ 5.64 (app qt, J = 6.3, 0.7 Hz, 1H), 3.90-3.85 (m, 1H), 3.39-3.28 (m, 3H), 3.04 (app d, J = 14.8 Hz, 1H), 2.41 (dd, J = 14.8, 9.1 Hz, 1H), 2.02 (m, 1H), 1.75 (d, J = 6.3 Hz, 3H), 1.68-1.62 (m, 2H), 1.54-1.44 (m, 1H), 0.96 (t, J = 7.9 Hz, 9H), 0.59 (q, J = 7.8 Hz, 6H); 13C NMR (125 MHz, CDCl3) δ 131.5, 107.3, 81.3, 70.8, 68.1, 47.8, 33.8, 25.8, 22.5, 7.1, 5.3.
HR-MS (ESI) m/z calcd for C15H29IO2Si (M+Na)+: 419.0874, found 419.0893.
To a of bishomopropargylic alcohol 19 in CH2Cl2 (10 mL) at 0° was added slowly a 2 M solution of AlMe3 in hexanes (0.96 mL, 1.93 mmol).S10 This solution was stirred for 3 min. and then recooled to −78°. A 1 M solution of TiCl4, in CH2Cl2 (0.96 mL, 0.96 mmol) was added dropwise. The solution was stirred 2 h. at −78° and then quenched with cold MeOH (1 mL), upon which the solution turned pale yellow. The solution was diluted with Et2O (10 mL) and washed with a saturated solution of Rochelle’s salt (10 mL). The aqueous layer was extracted with Et2O (3 × 20 ml), dried over MgSO4, and concentrated in vacuo, and this crude product was purified by column chromatography (25% EtOAc in hexanes) to afford 20 as a colorless oil (105 mg, 0.62 mmol, 70%): Rf = 0.38 (30% EtOAc in hexanes); [α]22D = −19.4 (c = 1.2, CDCl3). Some unreacted 19 (16 mg, 0.10 mmol, 12%) was also recovered.
IR (thin film, NaCl) 3412, 2928, 2855, 1451, 1376, 1340, 1277, 1095, 1036, 944 cm−1
1H NMR (500 MHz, CDCl3) δ 5.29 (app t, J = 7.0 Hz, 1H), 3.94-3.89 (m, 1H), 3.42-3.30 (m, 2H), 3.08 (ddd, J = 8.7, 7.2, 4.5 Hz, 1H), 2.52 (app dt, J = 15.0, 5.8 Hz, 1H), 2.24 (app dt, J = 15.0, 7.0 Hz, 1H), 2.10 (m, 1H), 1.73 (s, 3H), 1.72-1.64 (m, 5H), 1.40 (dddd, J = 17.4, 11.3, 6.3, 5.4 Hz, 1H); 13C NMR (100 MHz, CDCl3) δ 134.1, 120.5, 82.4, 70.9, 67.9, 32.8, 31.6, 26.1, 25.7, 18.2.
HR-MS (ESI) m/z calcd for C10H18O2 (M+Na)+: 193.1199, found 193.1194.
To PdCl2(PhCN)2 (40 mg, 0.11 mmol) and NaHCO3 (880 mg, 10.5 mmol) was added allyl bromide (9 mL, 105 mmol). The resulting solution was stirred 15 minutes at room temperature. Alkyne 26 (795 mg, 2.1 mmol) as a solution in THF (2 mL) was then added dropwise at ambient temperature via syringe pump over 90 min. After addition the reaction was stirred a further 30 min., directly concentrated in vacuo, and chromatographed (gradient 2% to 3% to 5% EtOAc in hexanes) to afford alkenyl bromide 28 (920 mg, 1.84 mmol, 88%): Rf = 0.54 (10% EtOAc in hexanes); [α]22D = −11.5 (c = 27.0, CDCl3).
IR (thin film, NaCl) 3072, 3011, 2933, 2858, 1639, 1472, 1428, 1362, 1218, 1127, 1103, 1048 cm−1.
1H NMR (500 MHz, CDCl3) δ 7.87 (app t, J = 8.4 Hz, 4H), 7.56-7.46 (m, 6H), 5.92 (dddd, J = 16.5, 10.1, 6.2, 6.2 Hz, 1H), 5.85 (app t, J = 6.8 Hz, 1H), 5.22 (dd, J = 17.1, 1.7 Hz, 1H), 5.13 (dd, J = 10.1, 1.5 Hz, 1H), 3.92-3.87 (m, 1H), 3.71 (app td, J = 9.2, 1.5 Hz, 1H), 3.55 (ddd, J = 10.2, 9.1, 4.6 Hz, 1H), 3.39 (app td, J = 11.3, 2.7 Hz, 1H), 3.32 (app d, J = 14.7 Hz, 1H), 3.13-3.02 (m, 2H), 2.43 (dd, J = 14.8, 10.0 Hz, 1H), 2.05-2.00 (m, 1H), 1.70-1.62 (m, 1H), 1.60-1.48 (m, 2H), 1.23 (s, 9H); 13C NMR (125 MHz, CDCl3) δ 135.9, 135.9, 134.9, 134.3, 133.5, 129.9, 129.7, 127.8, 127.6, 126.0, 115.5, 80.0, 71.9, 67.6, 44.5, 35.8, 33.5, 27.1, 25.5, 19.3.
HR-MS (ESI) m/z calcd for C27H35 BrO2Si (M+Na)+: 521.1482, found 521.1489.
Alkyne 32 (616 mg, 1.56 mmol) was added to a dry, cooled sealed tube. The tube was pumped under high vacuum and then backfilled with argon three times. Pinacolborane (287 µL, 253 mg, 1.98 mmol) was added, followed by Et3N (22 µL, 16 mg, 0.16 mmol) and Schwartz’s reagent (41 mg, 0.16 mmol). The resulting slurry was heated to 60° and stirred vigorously for 28 h. while protected from light. After cooling, the crude reaction mixture was filtered through a short pad of SiO2 (100% Et2O) and concentrated in vacuo to a heavy oil containing a 2:1 mixture of pinacolate 35 (Rf = 0.36, 10% EtOAc in hexanes) and its regioisomer 36 (Rf = 0.38, 10% EtOAc in hexanes), along with unreacted 32, traces of a proton quench product, and borate and other impurities. These were separated via careful column chromatography (gradient 3% to 5% EtOAc in hexanes) to afford 35 (235 mg, 0.45 mmol, 29% (39% based on recovered 32) in >20:1 regioisomeric purity along with unreacted 32 (163 mg, 26%): [α]22D = −2.1 (c = 1.9, CDCl3).
IR (thin film, NaCl) 3072, 2932, 2246, 1632, 1472, 1428, 1371, 1301, 1103 cm−1
1H NMR (500 MHz, CDCl3) δ 7.71-7.67 (m, 4H), 7.45-7.36 (m, 6H), 6.43 (app tq, J = 6.4, 1.5 Hz, 1H), 3.81-3.76 (m, 1H), 3.45-3.39 (m, 1H), 3.30-3.24 (m, 2H), 2.81-2.75 (m, 1H), 2.08-2.00 (m, 1H), 1.82-1.77 (m, 1H), 1.63 (d, J = 1.5 Hz, 3H), 1.49-1.38 (m, 3H), 1.27 (app s, 12H), 1.04 (s, 9H); 13C NMR (125 MHz, CDCl3) δ 143.43, 136.1, 136.1, 134.9, 133.9, 129.9, 129.7, 127.9, 127.6, 83.3, 82.4, 72.6, 67.9, 33.6, 32.1, 27.2, 25.8, 25.1, 25.0, 19.5, 14.5 (no signal was observed for the boron-bound carbon).
HR-MS (ESI) m/z calcd for C31H46BO4Si (M+Na)+: 543.3089, found 543.3089.
[1, 1'-Bis(diphenylphosphino)ferrocene] dichloropalladium(II) (PdCl2(dppf)) (73 mg, 0.10 mmol) was added to a flame-dried, cooled sealed tube. K3PO4 (1.83 g, 8.63 mmol) was added, and the tube was pumped under high vacuum and backfilled with argon three times. Alkenyl boronate 35 (600 mg, 1.15 mmol) was then added as a solution in dry THF (2 mL). The mixture was allowed to stir under Ar for 5 min. Degassed water (42 mg, 42 µL, 2.3 mmol, degassed via sparging) was then added, followed immediately by prenyl bromide (859 mg, 666 µL, 5.76 mmol). The sealed tub was capped, and the slurry was heated to 80° and stirred vigorously for 42 h. After cooling and dilution with Et2O (5 mL), the crude product was filtered through SiO2 (washed with Et2O) and concentrated in vacuo to yield a 2.5:1 mixture of 39 and 40. These inseparable diene isomers were purified away from phosphine and other impurities via column chromatography (gradient 2% to 3% EtOAc in hexanes) to give a 2.5:1 mixture of 39:40 (400 mg, 0.86 mmol, 75% combined yield, Rf = 0.63, 10% EtOAc in hexanes). This mixture was carried forward into Shi epoxidation without further purification.
To this mixture (400 mg, 0.86 mmol) in 2:1 v/v DMM:MeCN (23.2 mL) was added a 0.05 M solution of Na2B4O7•10H2O in 4 × 10−4 Na2EDTA (15.5 mL), nBu4HSO4 (75 mg, 0.22 mmol), and chiral ketone 18 (222 mg, 0.86 mmol). This biphasic mixture was stirred vigorously at 0°. To this mixture was added, simultaneously over 30 min. via syringe pump, a solution of Oxone (1.06 g, 1.73 mmol) in 4 × 10−4 Na2EDTA (7.75 mL) and a 0.89 M solution of K2CO3 (7.75 mL, 6.9 mmol). After the K2CO3 and Oxone solutions had been added, the resulting mixture was stirred an additional 20 min., at which point it was diluted with EtOAc (25 mL). The aqueous layer was extracted with EtOAc (3 × 25 mL), and the combined organics were washed with sat. NaCl, dried over MgSO4, and concentrated in vacuo to provide desired diepoxide 42 and diastereomers. 40 was partially oxidized under these conditions to monoepoxide 41 (Rf = 0.77, 20% EtOAc in hexanes. Column chromatography (15% EtOAc in hexanes) afforded diepoxide 42 in 3.5:1 overall dr as a colorless oil (276 mg, 0.56 mmol, 65% (49% yield over 2 steps), Rf = 0.54 (20% EtOAc in hexanes)) contaminated with a small quantity of ketone 18. Diepoxide 42 was further purified via preparative HPLC (Supelco SUPELCOSIL LC-SI 20 mm achiral SiO2 column, 5 µm particle size; 99.5:0.5 hexanes:iPrOH, 20 mL/min.; tR of desired diastereomer = 11.9 min.) to afford 42 free of 20 and in 15:1 to 20:1 overall dr (depending on batch): [α]22D for a sample in 20:1 dr = −7.5 (c = 3.3, CDCl3).
IR (thin film, NaCl) 3072, 2958, 2930, 2857, 1590, 1472, 1462, 1428, 1379, 1102 cm−1
1H NMR (500 MHz, CDCl3) δ 7.71-7.66 (m, 4H), 7.46-7.41 (m, 2H), 7.40-7.36 (m, 4H), 3.85-3.80 (m, 1H), 3.43 (ddd, J = 9.3, 4.8, 4.5 Hz, 1H), 3.29 (app td, J = 9.3, 2.5 Hz, 1H), 2.93 (app t, J = 6.1 Hz, 1H), 2.89 (app t, J = 6.0 Hz, 1H), 2.11 (ddd, J = 14.4, 6.4, 2.7 Hz, 1H), 1.85-1.80 (m, 1H), 1.77-1.73 (m, 2H), 1.60 (ddd, J = 14.9, 9.5, 5.8 Hz, 1H), 1.51-1.39 (m, 3H), 1.33 (app s, 6H), 1.27 (s, 3H), 1.04 (s, 9H); 13C NMR (100 MHz, CDCl3) δ 136.1, 136.1, 134.7, 133.6, 130.0, 129.8, 127.9, 127.7, 81.4, 72.5, 67.9, 61.3, 61.2, 58.8, 58.0, 38.2, 33.5, 31.8, 27.2, 25.6, 24.9, 19.5, 19.0, 17.3.
HR-MS (ESI) m/z calcd for C30H42O4Si (M+Na)+: 517.2745, found 517.2751.
This work was supported by the NIGMS (GM72566). C.J.M. thanks the George Büchi Summer Graduate Fellowship for fellowship support. Li Li (MIT) acquired HR-MS data. We are grateful also to Dr. Jeffery A. Byers, Aaron van Dyke, Ivan Vilotijevic, and Brian S. Underwood (all of MIT) for many helpful discussions. Without their insights this work would not have been possible.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.