|Home | About | Journals | Submit | Contact Us | Français|
A practical route to the labile tetracyclic isoxazolo[4,3,2-de]phenanthridinone moiety of the antifungal parnafungins has been developed. Zinc reduction of a methyl 2'-hydroxymethyl-2-nitro-3-biphenylcarboxylate, which was prepared by a Suzuki coupling, afforded a benzisoxazolone that was treated with MsCl and base to generate the labile tetracyclic ring system in 37–47% yield. This compound decomposes to the phenanthridine in CDCl3 and the phenanthridine N-oxide in aqueous base.
A Merck group recently reported the isolation of parnafungins A1 (1a), A2 (1b), B1 (2a) and B2 (2b) that interconvert readily by an elimination that opens the xanthone ring to form an enone and a phenol and then a conjugate addition from either phenol OH group to either face of the enone to reform the xanthone (see Scheme 1).1 The parnafungins demonstrate broad spectrum antifungal activity with no antibacterial activity by inhibiting fungal polyadenosine polymerase (PAP) and show in vivo efficacy against Candida albicans in a mouse model.1b Affinity selection/mass spectrometry demonstrated that the linear parnafungin A binds preferentially to PAP.1c The O-methylated parnafungins C and D were recently isolated.1d
The isoxazolone ring of 1 and 2 is very labile. Treatment of 1 and 2 at neutral or basic pH generated phenanthridines 3 and 4, respectively, in less than 1 hour. The same compounds were generated over 10–20 hours at pH 3. Unfortunately, phenanthridines 3 and 4 are biologically inactive.1a
The parnafungins might be biosynthesized by oxidative coupling of 5 with anthranilic acid to give 6 (see Scheme 2). Further oxidation of the benzylic methyl group and aniline could give parnafungins A1 (1a) and A2 (1b). Blenolide C (5b) was recently isolated2 and numerous natural products are known that are derived from blenolide C by oxidative dimerization or coupling at the carbon marked by an asterisk. Bräse3 and Nicolaou4 have recently reported syntheses of blenolide C (5b).
We therefore decided to start our synthetic studies by developing a route to the tetracyclic isoxazolo[4,3,2-de]phenanthridinone moiety 10 that can then be extended to the synthesis of the parnafungins by starting with blenolide C or an analog. We were guided by the studies of Wierenga5 that were based on earlier work of Bamberger6 and Cohen.7 Wierenga reported that reduction of methyl o-nitrobenzoate (7) with zinc, NH4Cl, and Na2CO3 in MeOH containing formaldehyde afforded hydroxymethylbenzisoxazolone (9) in 44% yield (see Scheme 3). A similar reduction without formaldehyde gave 8 in low yield, which could be converted to 9 by reaction with basic formaldehyde.
Our retrosynthesis of 10 is shown in Scheme 4. An intramolecular Friedel-Crafts alkylation of 11 could give 10. Hydroxymethylbenzisoxazolone 11 should be accessible by Wierenga's procedure from biphenylcarboxylate 12, which has been prepared by Liu from methyl 3-chloro-2-nitrobenzoate (16) by Suzuki coupling with phenylboronic acid (15, R = H).8 However, the Friedel-Crafts reaction may not work well with an unactivated aromatic ring and this route differs from the proposed biosynthesis in which the ring is closed by C-N rather than C-C bond formation. We therefore considered an alternate approach in which the C-N bond of 10 is formed from 13 by an intramolecular SN2 reaction. Suzuki coupling of 16 with the appropriate boronic acid 15, R = CH2OH, should provide 14, which should form 13 by zinc reduction without formaldehyde.
Suzuki coupling of 16 and phenylboronic acid (17a) as described by Liu afforded 12 in 65% yield (see Scheme 5). The yield was improved to 99% using SPhos, Pd(OAc)2 and K3PO4 in wet THF.9 Reduction of 12 with Zn and NH4Cl in THF/MeOH/H2O with sonication for 30 minutes afforded a 3:1 mixture of the desired benzisoxazolone 19a and amino ester 20a. Benzisoxazolone 19a decomposed on chromatography, but it could be obtained in 60% yield by washing the mixture with 9:1 hexanes/Et2O to remove 20a. Reaction of 19a with formaldehyde and Na2CO3 in aqueous THF afforded the desired hydroxymethylbenzisoxazolone 11 in 61% yield. This two-step sequence gave pure 11; impure 11 was obtained in one step by reduction of 12 with Zn and formaldehyde. Unfortunately, all attempts to form 10 by intramolecular Friedel-Crafts alkylation on the phenyl ring of 11 were unsuccessful.
We then turned to a more electron rich aromatic ring to see if the Friedel-Crafts alkylation would work with an optimized substrate. The analogous Suzuki coupling of 16 with 3,5-dimethoxyphenylboronic acid (17b) gave 18b in 99% yield. Zinc reduction gave 19b in 66% yield, which was treated with formaldehyde to give 21b in 76% yield. We were delighted to find that treatment of 21b with 10% TFA in CH2Cl2 generated an iminium cation that added to the aromatic ring to give 22b in 27% yield. Although this route to 22b provided the first synthesis of an isoxazolo[4,3,2-de]phenanthridinone, it is not applicable to the synthesis of the parnafungins because the Friedel-Crafts reaction would have to occur meta to the oxygen substituents and para to the carbonyl group. We therefore turned our attention to the alternate route using C-N bond formation.
The analogous Suzuki coupling of 16 with 2-hydroxymethylphenylboronic acid (23a) afforded biphenyl 24a in 95% yield (see Scheme 6). Zinc reduction afforded a mixture of benzisoxazolone 25a and the amino ester analogous to 20. Treatment of this mixture with MsCl and Et3N in CH2Cl2 for 15 min at 0 °C afforded the mesylate of 25a, which was treated with Na2CO3 in 1:1 THF/H2O for 40 minutes to give the desired tetracyclic isoxazolo[4,3,2-de]phenanthridinone moiety 10 in 37% yield from 24a.
Application of this route to a coupling product of blenolide C (5b) would require functionalization of a benzylic methyl group. We therefore prepared 24b in 95% yield by Suzuki coupling of 16 with o-tolylboronic acid (23b). Bromination of 24b with NBS and (BzO)2 in CCl4 at reflux gave 24d, R = H, X = Br. Reaction of 24d with KOAc in DMF afforded 24e, R = H, X = OAc, which was hydrolyzed with K2CO3 in MeOH to give 24a in 79% overall yield from 24b. Other approaches were unsuccessful. Reduction of 24d with Zn gave 25b resulting from reduction of both the nitro and bromomethyl groups. Functionalization of the methyl group must be carried out before formation of the benzisoxazolone ring. Zn reduction of 24b gave 25b in 62% yield, which decomposed on treatment with NBS.
Suzuki coupling of 16 with boronic acid 23c10 provided biphenyl 24c (83%) with the oxygen substituents on the same carbons as in the parnafungins. Zinc reduction gave benzisoxazolone 25c, which was treated with MsCl and Et3N to give the mesylate. Treatment of the mesylate with Na2CO3 in 1:1 THF/H2O provided 26c in 47% overall yield from 24c. This sequence provides a very short and high yield route to the labile tetracyclic isoxazolo[4,3,2-de]phenanthridinone moiety of the parnafungins that should be applicable to the synthesis of the natural product.
The Merck group suggested that the rearrangement of parnafungins 1 and 2 to 3 and 4, respectively, could occur by either an E2 reaction or by hydrolysis of the isoxazolone followed by loss of water. We explored the stability of 10, 22b and 26c in CDCl3. All three compounds rearranged cleanly to the respective phenanthridine 27a,11 27b and 27c without any evidence for the formation of an intermediate (see Scheme 7). This indicates either that the reaction proceeds by an E2 elimination or that loss of water is much faster than hydrolysis of the isoxazolone. Tetracycle 26c with the oxygen substituents in the same position as the parnafungins rearranges fastest with a half life of 6 days. Tetracycle 10 rearranges with a half life of 10 days and tetracycle 22b rearranges slowest with a half life of 27 days. The C-methyl group of m-methoxytoluene is deprotonated twice as fast as that of toluene by lithium amide bases, whereas the C-methyl group of o-methoxytoluene is deprotonated ten times slower than that of toluene.12 Deprotonation occurs in the rate determining step of an E2 elimination even if protonation of the carbonyl group by adventitious HCl is the initial step. Therefore, the two methoxy groups meta to the methylene group of 26c should accelerate deprotonation and E2 elimination, whereas the methoxy groups ortho and para to the methylene group of 22b should retard deprotonation and elimination. The parnafungins have oxygen substituents meta to the methylene group as in 26c and a carbonyl group para to the methylene group, which should further increase its acidity and accelerate the E2 elimination.
We also explored the decomposition of 10 under basic condition with NaOD in CD3CN/D2O. Under these conditions, we detected an intermediate with the CH2 group shifted upfield to δ 4.32 from δ 4.72 in 10. We tentatively assigned structure 28 to this intermediate (see Scheme 8). To our surprise, the final product is not 27a, but a 1:10 mixture of 27a and the N-oxide 29 resulting from oxidation of 28, presumably by air. The autoxidation of hydroxylamine in aqueous base has been extensively studied13 and the oxidation of alkylhydroxylamines to oximes in MeOH has been reported.14 The formation of 29 by this decomposition pathway is noteworthy because we were unable to prepare 29 directly by oxidation of 27a using procedures that work well on phenanthridine iteself. Both electron withdrawing groups and peri substituents are known to retard the N-oxidation of heterocycles.15
In conclusion, a practical route to the labile tetracyclic isoxazolo[4,3,2-de]phenanthridinone moiety (10 and 26c) of the antifungal parnafungins has been developed. Zinc reduction of methyl 2'-hydroxymethyl-2-nitro-3-biphenylcarboxylates 24a and 24c, which were prepared by Suzuki couplings, afforded benzisoxazolones 25a and 25c that were treated with MsCl and then base to generate the labile tetracyclic ring systems 10 (37%) and 26c (47%). These compounds rearrange to phenanthridines 27a and 27c in CDCl3 and 10 decomposes to phenanthridine N-oxide 29 in aqueous base.
We are grateful to the National Institutes of Health (GM-50151) for support of this work. We thank Professor Stephen Buchwald, Massachusetts Institute of Technology for advice on Suzuki couplings.