|Home | About | Journals | Submit | Contact Us | Français|
We report the application of diphenyloxazinone glycinate chiral templates to asymmetric syntheses of cylindrospermospin, 7-epi-cylindrospermopsin, 7-deoxycylindrospermopsin, and spirotryprostatins A and B. Synthetic studies toward quinine, nakadomarin A, and palau’amine using these templates are also described.
We have previously described (1) the use of commercially available diphenyloxazinones 1 and 2 as versatile templates for the asymmetric synthesis of amino acids and natural products containing amino acids. Herein we will detail the use of these templates for the synthesis of complex natural products, many with no apparent amino acid functionality.
Cylindrospermopsin (3) was isolated from the marine cyanobacterium Cylindrospermopsis raciborskii in 1992 (2). Other members of the family include 7-epi-cylindrospermopsin (4), isolated from A. ovalisporum (3), and 7-deoxycylindrospermopsin (5), isolated from C. raciborskii (4). While 3 and 4 were shown to be equipotent hepatotoxic glutathione biosynthesis inhibitors, the deoxy analogue 5 showed no activity. During the course of our efforts, Snider (5) and Weinreb (6) had reported racemic syntheses of 3, while White (7) had synthesized 4 in optically active form.
We began by alkylation of the enolate of (+)-2 with homoallyl iodide to yield alkene 6 (Scheme 1), planning for the new chiral center to set by relay the remaining stereocenters of the molecule. Removal of the chiral template via Birch reduction furnished protected amino acid 7; treatment with acetyl chloride cleaved the Boc group and converted the acid to its mixed anhydride, whose LAH reduction furnished the free amino alcohol 8. Treatment with phenyl bromoacetate under basic conditions gave oxazinone 9, which was unstable to dimerization; immediate oxidation with m-CPBA yielded the stable nitrone 10.
At this point we attempted our key intramolecular nitrone 1,3-dipolar cycloaddition and were gratified to obtain the desired adduct 11 in 78% yield as a single diastereomer, albeit as a 10:1 mixture with the regioisomer resulting from endo cyclization. The stereochemical outcome, confirmed by X-ray analysis, results from suprafacial addition to the alkene by the nitrone in an exo fashion, as shown in the transition state A. This cycloadduct contained the necessary stereochemistry for the A ring of the cylindrospermopsin family. (8)
The lactone of 11 was reduced to the corresponding lactol, and reductive amination under hydrogen gave the free amine while also reducing the labile NO bond; treatment with p-nitrophenylcarbonate yielded the urea 12 (Scheme 2). After much experimentation, we found that subjection of the diol to TEMPO and PhI(OAc)2, along with a catalytic amount of MsOH to aid disproportionation of the nitroxyl radical, gave the desired aldehyde 13 resulting from selective oxidation of the primary alcohol. Treatment of 13 with lithiated nitromethane provided the corresponding β-hydroxynitro compound as a mixture of diastereomers; treatment with acetic anhydride resulted in elimination to the nitroalkene as well as protection of the secondary alcohol, and in situ conjugate reduction with sodium borohydride gave the nitroalkane 14. Refluxing in neat TFA removed the PMB protection, and conversion to the O-ethyl isourea 15 was effected with Meerwein’s salt.
TBAF-mediated nitroaldol reaction of 15 with the dimethoxypyrimidine aldehyde 16 gave the adduct 17; immediate quenching was required to prevent retro-nitroaldol reaction and equilibration to an equimolar mixture of all four possible diastereomers. The quenched adduct 17 was immediately hydrogenated to effect reductive guanidinylation of the nitro group onto the isourea, yielding 18 as an inseparable 1:0.8 mixture of diastereomers. Acidic hydrolysis of the acetyl group and pyrimidines produced a separable mixture. Finally, treatment of 18 with sulfur trioxide-pyridine and molecular sieves in DMF gave a moderate yield of 7-epi-cylindrospermopsin (4) as a 2:1 mixture with the bissulfate (9). Cylindrospermopsin (3) was synthesized in a similar fashion (10).
Nitroaldol reaction between 15 and benzylpyrimidine 19 with cesium fluoride and acetic anhydride produced directly the E-nitroalkene 20 (Scheme 3). Treatment with sodium borohydride effected conjugate reduction, and hydrogenation gave the pyrimidine 21 as an equimolar mixture of diastereomers at C8. Cleavage of the acetate and sulfonation yielded 7-deoxycylindrospermopsin (5) and its diastereomer 22. With this material in hand, we showed that the proposed structure for 5 was incorrect, and that contrary to previous reports, 5 showed potent protein synthesis inhibitory activity (11).
Quinine (23, Figure 3) is among the most famous alkaloids due to its antimalarial properties and has long been the target of synthetic research. Stork closed the quinuclidine ring at the N1-C6 bond in his stereoselective synthesis (12), while Jacobsen (13) and Kobayashi (14) used the classical Rabe N1-C8 disconnection. Our plans called for a novel intramolecular SN2′ reaction to install the C3-C4 bond.
Our synthesis began with formation of the silyl enol ether 24 of lactone (+)-2 in near-quantitative yield (Scheme 4). Treatment of 24 with TBAF in the presence of aldehyde 25 led to a diastereoselective aldol reaction to give the thermodynamic adduct, containing the correct C8 and C9 stereochemistry as confirmed by X-ray crystallographic analysis of the TES-protected alcohol 26.
Removal of the N-Cbz group followed by a three-step protocol to remove the chiral auxiliary yielded the fully protected amino acid 27. Direct reduction of the ester to the aldehyde 28, followed by a Grignard reaction, produced a single diastereomeric alcohol. It was subsequently found that the stereochemistry of the secondary alcohol was crucial for a successful piperidine-forming reaction, and the product directly obtained from this sequence did not lead to the desired piperidine system. Thus, the resultant alcohol was oxizided and diastereoselectively reduced to secure 29 with the opposite carbinol stereochemistry, which led to a productive piperidine-forming reaction. Acetate protection, conversion of the benzyl ether to the corresponding mesylate 30 in two steps, and treatment with sodium hydride effected cyclization to the piperidine ring of 31 in excellent yield.
At this point we converted the acetate of 31 to the corresponding O-TMS ether substrate 32 (Scheme 5). Thus, treatment of 31 with lithium aluminum hydride, followed by treatment with trimethylsilyl triflate and N-alkylation, provided 32 in high yield for the three steps. Acidic removal of the O-TMS residue and oxidation to the ketone 33 followed by formation of the silyl enol ether furnished the key substrate 34. Treatment of 34 with tributyltin fluoride in the presence of Pd(II) and a phosphine ligand effected the desired C3-C4 closure to the quinuclidine and installed the vinyl group with the correct relative stereochemistry. The incipient keto-quinuclidine proved rather labile to purification and handling and was immediately reduced to furnish 7-hydroxyquinine 35 (15). We are currently examining methods for deoxygenation of the quinuclidine 35 to yield quinine 24.
We have developed concise asymmetric syntheses of spirotryprostatins A and B, structurally interesting antimitotic arrest agents isolated by Osada (16) that have attracted considerable attention from the synthetic community (17). Our approach deploys a novel three-component azomethine ylide cycloaddition strategy to assemble the prenylated spiro-oxindole pyrrolidine in a single stereocontrolled step. Our synthesis of spirotryprostatin B commenced with condensation of oxazinone 36 with aldehyde 37 to generate the incipient iminium species, whose deprotonation yields the azomethine ylide 39 (Scheme 6). Dipolar cycloaddition reaction with the readily available dipolarophile 38 provided the tetracyclic cycloadduct 40 in excellent yield. This reaction sets four consecutive stereogenic centers, including the quaternary spiro-ring junction, whose stereochemistry was confirmed by X-ray crystallographic analysis.
Hydrogenation of 40 in the presence of palladium chloride quantitatively cleaved the chiral auxiliary from 40 (Scheme 7) to furnish the free amino acid 41. Fortuitously, we found that 41 could be selectively coupled with D-proline benzyl ester without the need to protect the proline amine of 41. This is presumably a manifestation of the severe steric hindrance of the proline amine on both faces, which obviates self-condensation. Removal of the benzyl ester and cyclization gave dioxopiperazine 42 in excellent yield. Treatment with two equivalents of tosic acid in refluxing toluene effected regioselective elimination of methanol to furnish alkene 43. The final oxidative decarboxylation of 43 to spirotryprostatin B proved extremely difficult and required the extensive evaluation of methods to effect this seemingly trivial transformation. Standard Kochi-type oxidative decarboxylation methods led to significant decomposition and over-oxidation side products. Eventually, we found that Barton-modified Hunsdiecker conditions effected the oxidative decarboxylation and installed the desired alkene in the central pyrrolidine ring. In the event, lithium iodide cleaved the ethyl ester of 43 to yield the corresponding acid. Subjection of the resultant acid 44 to a Barton-modified Hunsdiecker protocol installed the desired alkene in modest, but reproducible, yield. Finally, the proline residue was epimerized under thermodynamic conditions (2:1 favoring the natural stereochemistry) to complete the synthesis of spirotryptostatin B (44) (18).
We next pursued a similar strategy toward spirotryprostatin A, but unfortunately observed that the Hunsdiecker reaction on a substrate corresponding to 43 completely failed in this case. In light of this, we decided to do away with the troublesome carboalkoxy residue entirely.
Thus, dipolar cycloaddition of the labile exocyclic alkene 45, prepared in situ from the corresponding β-TMS alcohol, with the ylide derived from condensation of 36 and 37 gave the desired adduct 46 along with the methanol elimination product (Scheme 8). Hydrogenation removed the bibenzyl auxiliary, and thermodynamic epimerization at C9 yielded an inseparable mixture of diastereomers 47 that was carried forward to dioxopiperazine 48, which could be separated from its C9 diastereomer. Treatment of 48 with tosic acid effected methanol elimination to complete the synthesis of spirotryprostatin A (49) (19).
We recently applied the dipolar cycloaddition methodology to synthesis of the ADE ring system of the manzamine alkaloid nakadomarin A (20). The requisite aldehyde 50 could be generated easily from protected mannitol, but the exocyclic enone dipolarophile 51 proved highly reactive and unstable to polymerization. We then deployed a two-step procedure to generate 51 in situ, which permitted the dipolar cycloaddition with diphenyloxazinone 36 to yield spirocycle 52 containing the AD ring system (Scheme 9).
The chiral auxiliary was easily cleaved with Pearlman’s catalyst to give amino acid 53 (Scheme 10). Acylation of the amine yielded alkene 54, and installation of the methyl ester and acetonide deprotection gave diol 55, which was converted to diene 56. Ring-closing metathesis with Grubbs’ second-generation catalyst completed the ADE ring system, giving alkene 57.
The ketone of 57 was converted in a two-step process to alkene 58 (Scheme 11); reduction of the ester to the alcohol and reoxidation to the aldehyde 59 allowed installation of the desired alkyne 60. Epoxide alkylation of the alkyne gave β-hydroxyalkyne 61 (21), but the yields of this reaction have proven variable and the reaction capricious to scale-up.
Currently, alternatives to this homologation are being evaluated such that the planned palladium-mediated conversion to the furan system can be interrogated. If successful, this will allow subsequent closure of the B ring, positioning the molecule for the acylation-metathesis sequence planned to complete the synthesis of nakadomarin A (62).
Palau’amine (63, Figure 4) is a highly complex, densely functionalized alkaloid, with a carbon-to-nitrogen ratio approaching 2:1. The presence of six fused rings, including a spiro junction, a guanidine hemiaminal, and a guanidine hemi-amino aminal, and potent biological activity (22) makes the molecule a highly challenging yet inviting synthetic target, currently being pursued by many groups (23). We have recently undertaken studies utilitzing an intramolecular dipolar cycloaddtion strategy to assemble the difficult cyclopentane core.
Our approach begins with α-acylation of oxazinone (−)-2 to yield alkene 64 (Scheme 12). N-Boc removal followed by addition of paraformaldehyde furnishes the incipient iminium species, whose deprotonation generates the azomethine ylide; spontaneous intramolecular dipolar cycloaddition afforded the desired tricyclic species 65 as a single diastereomer. Treatment of 65 with lithium aluminum hydride furnished diol 66, which was subsequently subjected to hydrogenation to remove the bibenzyl residue. Reaction of the resultant amino alcohol with Boc anhydride effected acylation of both the nitrogen and the primary alcohol to give 67. Treatment of 67 with ruthenium trichloride not only oxidized the secondary alcohol, but also effects the oxidation α to the ring nitrogen to provide the ketoamide 68 (24). Conversion of the ketone to enone 69 under standard conditions followed by a Bayliss-Hillman reaction furnished alcohol 70, which was subjected to saponification to provide the ester 71. Conjugate addition of nitromethane to enone 71 in the presence of tetramethyl guanidine proceeded in a diastereoselective manner to give 72, which was followed by reduction of the ketone to alcohol 73. Compound 73 embodies all of the requisite carbon atoms and suitably poised functionality to serve as a reasonable substrate from which palau’amine might be constructed. Studies are currently underway to effect the end-game conversion of 73 to palau’amine and congeners.
In parallel with this approach, several alternative routes were being concomitantly evaluated for installation of the two aminomethyl arms. As shown in Scheme 13, enone 69 was subjected to dipolar cycloaddition to give isoxazolidine 74; treatment with DBU restored the enone while liberating the N-hydroxyl moiety, which was converted to its acetate 75. Conjugate addition of nitromethane allowed closure of the resultant α-anion onto the N-acetoxy species to give protected spiro-aziridine 76. Studies are similarly underway to futher elaborate this tricyclic system to palau’amine.
In conclusion, the diphenyl oxazinone templates (1, 2) have proven to be versatile and useful systems from which numerous complex natural products and their analogs can be constructed in a stereocontrolled, asymmetric manner. We continue to explore and exploit the chemistry of this versatile glycine template to the asymmetric total synthesis of many different families of biomedically significant nitrogenous substances.
The authors are grateful to the National Institutes of Health (GM068011), the National Science Foundation, Ajinomoto Co., and the Japanese Society for the Promotion of Science for financial support. This manuscript is dedicated to the memory of Professor Albert I. Meyers.