Solid Phase Total Synthesis and Structure Proof of Callipeltin B

Abstract

The cytotoxic, cyclic heptadepsipeptide natural product callipeltin B was synthesized on a solid-phase support in 15% overall yield. Comparison of the ¹H NMR spectra of three synthetic isomers with those of callipeltin B confirmed the configurational reassignment of its threonine residues as D-allothreonine and the assignment of the configuration of its β-methoxytyrosine residue as (2R, 3R).

The cyclic depsipeptide callipeltin B (1) was isolated from the Lithistid sponge callipelta sp. and structurally characterized by Minale and coworkers in 1996.¹ Along with the related cyclic depsipeptide callipeltin A,² 1 possesses a 22-membered macrolactone composed of the unique, non-proteinogenic amino acids β-methoxytyrosine (β-MeOTyr) and (3S,4R)-3,4-dimethyl-L-pyroglutamic acid (DiMePyroGlu) as well as several D- and N-methylated amino acids. Both callipeltins A and B showed broad-spectrum cytotoxicity against a number of tumor cell lines, including several drug-resistant cell lines,¹ and callipeltin A shows potent anti-HIV activity.²

Although the initially published structure of callipeltin B (1a) has never been revised, revisions to the structure of the closely related cyclic depsipeptide callipeltin A have strongly implied a need for a structural revision of 1. Specifically, D'Auria and coworkers, after isolating and characterizing two smaller, linear peptidic fragments of callipeltin A that they termed callipeltins D and E, suggested that one of the residues initially identified as L-threonine should be reassigned as D-allothreonine.³ Such a change thus implied that the structure of callipeltin B should be reassigned as 1b, in which the configuration of C-21 is R.

More recently, Bifulco and D'Auria have published computational studies that suggest that both of the residues in callipeltin A initially assigned as L-threonine by Minale and coworkers are actually D-allothreonine.⁴ At the same time, the configuration of the β-MeOTyr residue of callipeltin A, which could not be determined in Minale's original structural study, was determined to be (2R, 3R) by D'Auria and co-workers, using a combination of synthesis and degradation studies.⁵ Both of these structural reassignments were confirmed in our laboratory by the synthesis and spectral correlation of two diastereomers of callipeltin E.⁶ Taken together, these studies implied that the structure of callipeltin B should be further revised to 1c, in which the configuration of C-18 has been changed and those of C-1′ and C-6 are assigned as R. Herein we wish to report the synthesis of the three candidate structures 1a-c and the confirmation of 1c as the structure of callipeltin B by ¹H NMR correlation.

A solid phase synthetic strategy was chosen to expedite the synthesis of 1, its analogues and related depsipeptides. To permit macrocyclization on a solid support, it was decided to anchor the sidechain of the N-methylglutamine residue to a peptide amide synthesis resin, thereby leaving available both N- and C-termini for macrocyclization by amide bond formation. The Tentagel-based TG Sieber amide resin⁷ (2) was chosen for the relatively low amounts of TFA (1-5% in CH₂Cl₂) needed for efficient cleavage of the peptide, thereby minimizing the possibility of acid-catalyzed β-MeOTyr decomposition during cleavage from the resin. A standard Fmoc-based, C-to-N peptide synthesis approach thus required the use of allyl-based protecting groups for N- and C- termini immediately prior to macrocyclization. All sidechain protecting groups were selected for their ability to be removed either during the mildly acidic cleavage from the Sieber resin or a subsequent catalytic hydrogenation.

The initial plan to close the macrocycle at the MeGln/β-MeOTyr peptide bond, permitting a C-to-N synthesis from the MeGln residue, had to be abandoned because of spontaneous diketopiperazine formation upon deprotection of the Leu-MeGln dipeptide; instead, it was decided to close the macrocycle at the β-MeOTyr/N-methylalanine peptide bond, despite the known problem of slower acylation of secondary amines. Additionally, an initial plan to install the N-terminal (3S,4R)-3,4-dimethyl-L-pyroglutamic acid after macrocyclization was revised because of rapid O to N transacylation upon deprotection of the N-terminal Fmoc group. As a result, it was decided to install the (3S,4R)-3,4-dimethyl-L-pyroglutamic acid residue prior to forming the ester bond to prevent transacylation.

All of the constituent residues needed to make 1a-c were either commercially available or had been previously synthesized. In addition to D'Auria's synthesis of β-MeOTyr,⁵ several others had been published;⁸ several different syntheses of DiMePyroGlu had also been reported.⁹ The syntheses of 1a-c began with the deprotection and acylation of 2 (Scheme 1) with the cyclic anhydride of Fmoc-N-methylglutamic acid (3),¹⁰ followed by activation of the resin-bound acid with HATU and coupling with (2R, 3R)-O-benzyl-β-methoxytyrosine allyl ester (4)^7c to afford a resin-bound dipeptide. All deprotection and coupling reactions were monitored by removal of an aliquot of 1-2 mg of resin, cleavage of the peptide from the resin and analysis of the crude cleavage mixture by reverse phase HPLC and MALDI-MS. It was found that use of HATU with no added HOAt afforded complete consumption of starting material with no epimerization evident by HPLC analysis. Sequential installation of Fmoc-Leu, Fmoc-D-Arg(NO₂), Fmoc-D-aThr(THP) and Fmoc-D-aThr using coupling conditions optimized for each step proceeded without incident to afford the resin-bound hexapeptide 5 in >95% purity as determined by HPLC analysis. For the syntheses of 1a and 1b, protected L-threonine was substituted in this sequence for the second, or both, D-allothreonine residues. Additionally, in the syntheses of 1a and 1b, Fmoc-D-Arg(Z,Z) was substituted for Fmoc-D-Arg(NO₂).

Scheme 1.

Solid Phase Synthesis of 1c

Deprotection of the N-terminal Fmoc afforded a β-amino alcohol that could be selectively N-acylated with the N-hydroxysuccinimide ester of (3S,4R)-3,4-dimethyl-L-pyroglutamic acid (6). Esterification of the hydroxyl with alloc-N-methylalanine using the sulfonyl nitrotriazole reagent MSNT¹¹ in conjunction with N-methylimidazole afforded a resin-bound, protected heptadepsipeptide in high purity. Palladium-catalyzed deprotection of N- and C-termini followed by macrolactamization using PyAOP to minimize epimerization of the activated β-methoxytyrosine afforded the resin-bound macrocycle in roughly 80% purity as judged by HPLC analysis of the crude deprotection mixture. Removal of the cyclized product from the resin with 2% TFA-CH₂Cl₂ and complete deprotection by catalytic transfer hydrogenation afforded 1c, after purification by reverse phase HPLC, in 15% overall yield. The isomeric depsipeptides 1a and 1b were obtained in 15% and 14% overall yields, respectively.

Because of the unavailability of a sample of natural callipeltin B, the correlation of synthetic 1a-c with callipeltin B involved the comparison of spectroscopic data. Comparison of the ¹H NMR of 1c with that of callipeltin B showed no significant differences between the two apart from the absence of several exchangeable amide protons in CD₃OD.¹² On the other hand, the ¹H NMR spectra of 1a and 1b showed considerable differences from that of callipeltin B; in fact, the spectra of 1a and 1b showed evidence of conformational heterogeneity at ambient temperature. On the basis of these comparisons, we now conclude that the correct structure of callipeltin B is 1c. Current studies are underway to synthesize analogues of 1c and several related cyclic depsipeptides using the methodology described herein.

In summary, the synthesis of the cyclic depsipeptide natural product callipeltin B has been accomplished, thereby providing confirmation of the recent configurational reassignment of the previously assigned L-threonine residues as D-allothreonine and the recent configurational assignment of the β-methoxytyrosine residue as (2R, 3R). This synthesis illustrates the facility with which even complex natural products can be synthesized on solid supports. In addition, the expeditious solid-phase synthesis of 1 opens the door for the rapid synthesis of analogues to explore structure-function relationships and for the synthesis of other, more complex cyclic depsipeptides.

Supplementary Material

1si20060913_07. Supporting Information Available.

Complete experimental details and spectroscopic details for 1 and the novel protected amino acids used in its synthesis. A tabulated comparison of ¹H NMR spectra of 1 and callipeltin B and reproduced spectra of both are also provided. This information is available free of charge via the Internet at http :// pubs.acs.org.