The Total Syntheses of Phorboxazoles—New Classics in Natural Product Synthesis

Abstract

The marine sponge metabolites phorboxazoles A (see structural formula) and B were the targets of several clever total syntheses, whose special features are described in this Highlight. Now that the synthetic entry to these compounds has been cleared, their cytostatic activity can be studied thoroughly in vitro and in vivo. In September 1995 at the Eighth Symposium on Marine Natural Products in Santa Cruz de Tenerife, Spain, the phorboxazoles first caught the interest of natural products chemists. The phorboxazoles had been isolated by Molinski and co-workers from the marine sponge Phorbas sp. found in the Indian Ocean.1 The structures as well as the relative and absolute configurations were elucidated from 2D NMR and NOESY experiments, Mosher ester analysis, correlation with synthetic compounds, and CD spectroscopy.2 The phorboxazole skeleton consists of six rings—including two 2,4-disubstituted oxazoles—and 15 stereogenic centers organized into a macrolide (C1–C26) and a side-chain substructure (C27–C46). Phorboxazoles A and B differ only in the absolute configuration at C13. Both compounds show exceptional cytostatic activity and antifungal activity against Candida albicans but no antibacterial activity against E. coli, Pseudomonas aeruginosa, and Staphylococcus aureus. Phorboxazoles A and B were tested against the 60 tumor cell lines of the U.S. National Cancer Institute (NCI), and they effected in vitro inhibition of cell growth with mean GI50 values of <1.6×10−9 m. Phorboxazole A induced cell cycle arrest in the S phase in Burkitt lymphoma cells and showed neither inhibition of tubulin polymerization nor interference with the microtubules. Since selective modulation of cell cycles by natural and synthetic compounds is of both scientific and pharmaceutical importance,3 the phorboxazoles and (semi)synthetic analogues are attractive compounds for detailed in vitro and in vivo investigations to determine their biological target and mode of action. These investigations largely depend on a synthetic entry to the phorboxazole skeleton. The first total synthesis of phorboxazole A was completed by Forsyth and co-workers in early 1998.4 They relied on a highly convergent strategy for the assembly of the phorboxazole skeleton, employing three fragments of comparable complexity (C3–C17, C18–C30, and C31–C46) (Figure 1). The Forsyth synthesis of phorboxazole A.4 a) 1) EDCI⋅MeI, HOBt, 2) DMP; b) 1) (BrCCl2)2, PPh3, 2,6-tBu2-4-Me-C5NH2, 2) DBU (67 % overall). EDCI=N′-3′-dimethylaminopropyl-N-ethylcarbodiimide, DBU=1,8-diazabicyclo[5.4.0.]undec-7-ene, DMP=Dess–Martin periodinane, HOBt=1-hydroxybenzotriazole, K-Selectride=potassium tris(sec-butyl)borohydride, TPS=tert-butyldiphenylsilyl. A most elegant and unprecedented feature of the Forsyth synthesis is the biomimetic assembly of the oxazole units. This assembly commenced with the formation of the C16–C18 oxazole from a C3–C17 amino alcohol and a C18–C30 carboxylic acid by application of the Wipf protocol (yellow box in Figure 1).5 Amide formation was followed by oxidation to yield the amide aldehyde. Aromatization to give the 2,4-disubstituted oxazole involved bromooxazoline formation and subsequent dehydrobromination. The C1–C23 macrocycle was closed by an intramolecular Still–Gennari olefination of a C3 aldehyde with C1–C2 bis-(2,2,2-trifluoroethyl)phosphonoacetate, thus forming the intermediate with a C2–C3 double bond in 77 % yield as a 4:1 mixture of Z and E isomers. Finally, the second biomimetic formation of an oxazole unit (C29–C31) completed the skeleton assembly although in significantly lower yield (22 %) than that for the C16–C18 oxazole synthesis (67 %). For the construction of the remote stereogenic center at C43 Forsyth et al. relied on Corey's oxazaborolidine-catalyzed reduction protocol. In contrast to this reagent-controlled method (highlighted in red), all other stereocenters, which are clustered around C5–C15, C22–C26, and C33–C38, were either taken from abundant chiral starting materials (highlighted in green) or generated in a substrate-directed manner (highlighted in blue). This strategy is exemplified by Forsyth's synthesis of the C3–C17 fragment (Scheme 1). Starting from (S)-glyceraldehyde acetonide 1 an endo-selective BF3-mediated hetero-Diels–Alder reaction formed the syn-C11/C15 tetrahydropyran (THP) fragment 2 as the major product in a 16:4:1 mixture with exo cycloadducts. The C13 axial hydroxy group was generated diastereoselectively by substrate-controlled reduction with bulky K-Selectride as the hydride source. The formation of the C11, C13, and C15 stereocenters was therefore guided by the intermediate stereocenter at C16, which is removed en route to the natural product. Finally, Nozaki–Kishi coupling of 3 with allylic bromide 4, which was derived from (S)-malic acid, generated homoallylic alcohol 5 as a 3:2 mixture of C9 epimers. The unwanted diastereomer was removed and recycled by Mitsunobu inversion. This sequence developed during the race for the first total synthesis illustrates a drawback of Forsyth's substrate-controlled strategy: In several instances absence of a substrate-intrinsic stereochemical bias led to a mixture of diastereomers, which necessitated additional recycling steps. a) 1) BF3⋅OEt2, Et2O, −78 °C, 2) TBAF, TsOH, THF, 60 % (2 steps), 16:4:1 d.r.; b) 1) K-Selectride, THF, 2) TPSCl, Et3N, 3) DDQ, CH2Cl2, 4) DMP, 70 % (4 steps); c) CrCl2/NiCl2, THF, 80 %, 3:2 d.r. TBAF=tetrabutylammonium fluoride, DDQ=2,3-dichloro-5,6-dicyano-p-benzoquinone, PMB=p-methoxybenzyl, TES=triethylsilyl. The first total synthesis of phorboxazole B was completed by the Evans group in 2000.6 Their retrosynthetic analysis differs significantly from Forsyth's strategy in that only two fragments of equal complexity (C1–C19 and C20–C38) are employed and the C39–C46 side chain is added late in the synthesis (Figure 2). Fragment assembly started with an E-selective Wittig olefination (E/Z>95:5) to generate the C19–C20 double bond followed by a Yamaguchi macrolactonization. In order to avoid isomers at the C2–C3 double bond, the (Z)-alkene was masked as a triple bond and regenerated after macrolactonization by Lindlar hydrogenation (Z/E>95:5). Addition of the C39–C46 side chain had been achieved initially by a chemoselective halogen–metal exchange of a vinyl iodide (in the presence of the terminal vinyl bromide) with tert-butyllithium and subsequent addition to a C1–C38 aldehyde to yield a 1:2 mixture of C38 epimers. When the initially formed vinyllithium species was transmetalated with magnesium bromide, the desired C38 diastereomer was formed almost exclusively (>20:1) by chelation-controlled nucleophilic addition. The synthesis of the C20–C38 fragment included the regioselective α-metalation of the C29–C32 methyloxazole with lithium diethylamide (yellow box in Figure 2) and subsequent diastereoselective addition to the C33–C37 lactone to generate the C33 hemiacetal. The Evans synthesis of phorboxazole B.6 a) LDA, THF, −78 °C; b) LiNEt2, THF, −78 °C. LDA=lithium diisopropylamide, TIPS=triisopropylsilyl. This phorboxazole synthesis features an array of asymmetric catalytic processes developed in the Evans laboratory over the years. Only the forlorn stereocenter at C43 is derived from a chiral starting material, while the other stereocenters are either formed by ligand- and auxiliary-controlled asymmetric reactions or are generated with high substrate control. The elegance of the Evans synthesis and their brilliant retrosynthetic analysis is exemplified by the synthesis of the C4–C19 fragment (Scheme 2). The C5 stereocenter was formed with perfect stereoselection by addition of bis(trimethylsilyl)dienol ether 7, which functions as an efficient polyacetate aldol equivalent, to (benzyloxy)acetaldehyde in the presence of the copper(II)-Ph-pybox catalyst A (Figure 2).7 Axial nucleophilic attack at a cyclic oxonium ion8 formed in situ from the anomeric acetate 9 resulted in the diastereoselective formation of the C9 stereocenter (89:11 d.r.). Addition of a silylketene acetal to α-oxazole aldehyde 11 catalyzed by [Sn((S,S)-Ph-box)](OTf)2 (Figure 2) allowed the stereocontrolled generation of the C15 hydroxy group.9 Addition of dibutylboryl enolate derived from ketone 10 to β-silyloxyaldehyde 13 afforded the aldol product 14 under 1,5-anti induction.10 Finally, the C11 stereocenter was set by axial nucleophilic hydride delivery to a cyclic oxonium ion. a) 2 mol % A, CH2Cl2, 85 %, >99 % ee; b) TMSOTf, 2-(trimethylsiloxy)propene, cat. pyr, CH2Cl2, 89 %, 89:11 d.r.; c) 10 mol % B, CH2Cl2, 91 %, 94 % ee; d) nBu2BOTf, iPr2NEt, CH2Cl2, 82 %, >95:5 d.r.; e) 1) TIPSOTf, CH2Cl2, 99 %, 2) HF⋅pyr, pyr, THF, H2O, 99 %; f) BF3⋅OEt2, Et3SiH, CH2Cl2, 96 %, >95:5 d.r. TMS=trimethylsilyl, OTf=trifluoromethanesulfonate, pyr=pyridine. The Smith total synthesis of phorboxazole A, which was completed in 2001,11 features the application of the Petasis–Ferrier rearrangement for the construction of two of the four tetrahydropyran units (yellow box in Figure 3). In 1996 Petasis described the rearrangement of enol acetals to give tetrahydropyranones. This process is related to earlier investigations by Ferrier et al. on enol ether rearrangements induced by mercuric ions. However, never before was the enol acetal rearrangement used on such challenging targets. It has proved to be very valuable for the assembly of highly functionalized cis-substituted tetrahydropyran ring systems. The Smith synthesis of phorboxazole A. 11 a) AgOCN, CH2N2; b) Et3N, Tf2O. The synthesis of the central and most complex cis-substituted C22–C26 THP system started with the condensation of β-hydroxy acid 17, which was prepared under Evans aldol conditions, with propargyl aldehyde 18 (Scheme 3). A 3:1 mixture of cis and trans acetals was obtained, from which the trans acetal could be recycled. After conversion of dioxanone 19 into sulfone 20, a Julia-type reaction with an α-halo Grignard reagent generated the enol ether 21 without stereocontrol of the exo double-bond geometry. However, in the presence of Me2AlCl the Petasis–Ferrier rearrangement took place, and pyranone 22 was formed exclusively in high yield. This observation was explained by the boatlike transition state of the E isomer, caused by 1,3-diaxial interactions in the chair transition state, in contrast to the chairlike transition state of the Z isomer. Diastereoselective reduction of the C24 ketone in 22 was achieved with sodium borohydride. a) 1) HMDS, 2) TMSOTf, 66 % cis + 19 % trans (2 steps); b) nBuLi, 1-chloro-1-iodoethane, iPrMgCl, −78 °C, 95 %, E/Z 1:1; c) Me2AlCl, CH2Cl2, 91 %. DMB=3,4-dimethoxybenzyl, HMDS=hexamethyldisilazane. The C5–C9 trans-substituted THP system was built up from Danishefsky's diene and a protected β-hydroxyaldehyde in a hetero-Diels–Alder reaction (Scheme 4). 1,4-Vinyl cuprate addition, hydroboration, generation of the exo-methylene functionality by a Wittig reaction and oxidation of the C11 primary alcohol delivered aldehyde 25, which was ready for a Nagao aldol reaction using a tin enolate (92 %, 10:1 d.r.). The resulting C3–C13 β-hydroxycarboxylic acid 27 was condensed with an oxazole aldehyde to generate a dioxanone. By use of the Petasis modification of the Tebbe reagent, the enol acetal required for the Petasis–Ferrier rearrangement was established. In the presence of Me2AlCl even this highly functionalized C3–C19 segment underwent the rearrangement cleanly. Reduction of the C13 ketone using K-selectride selectively formed the axial hydroxy group. In prior investigations Smith had shown that the vinyl acetal 31 was resistent to the rearrangement conditions due to a dead-end chelation with the neighboring oxazole nitrogen. 1,3-Transposition of the enol ether oxygen allowed the successful rearrangement of 28 to the desired tetrahydropyranone system. a) 1) Sn(OTf)2, (−)-26, N-ethylpiperidine, 92 %, 10:1 d.r., 2) LiOH, H2O2, 98 %; b) Me2AlCl, CH2Cl2, −78 °C, 89 %. After conversion of the C19 hydroxy group into the phosphonium chloride, the C19–C20 double bond was established in high yield (94 %) and E selectivity (E/Z 12:1) by condensation with aldehyde 23. One further key feature of the Smith synthesis is the application of the 2,4-disubstituted difunctional oxazole 33 (Figure 3) as a linchpin for the combination of the acyclic precursor of the macrolide core and the functionalized side chain. Revisiting a protocol developed in 1949 by Sheehan, Smith et al. used bromoacetyl bromide, silver isocyanate, and diazomethane to generate a 2-bromomethyl-4-trifluoromethanesulfonyloxyoxazole. Oxazole 33 is perfectly suited at one terminus for a Stille cross-coupling reaction and at the other for a Grignard addition to a C33–C46 lactone side chain. Finally, a Still–Gennari olefination was applied to form the macrolide (95 %, Z/E 4:1). In their total synthesis of phorboxazole A Pattenden et al. utilized a fragment-coupling strategy combining features of the Evans and Forsyth syntheses.12 In contrast to the Evans synthesis, this fragment assembly started with the diastereoselective addition of a metalated C20–C32 oxazole to a C33–C46 lactone. Next, the C19–C20 double bond was set by Wittig olefination, and finally macrolide ring closure was achieved by an intramolecular Still–Gennari olefination. Thus, this highly efficient fragment assembly employing three segments of roughly equal complexity (C3–C19, C20–C32, C33–C46) might well be suitable for providing compounds for studies of the structure–activity relationship (SAR). For the generation of the stereocenters Pattenden relied both on reagent control (e.g. Brown allylation for C13, C15, and C35) and substrate control (e.g. an intramolecular oxyanion Michael addition for C11) as well as on chiral building blocks (e.g. malic acid for C43 and D-xylose for C37–C38). The fourth synthesis of phorboxazole A was completed by the group of D. R. Williams and is characterized by a far-reaching asymmetric allylation strategy (Figure 4).13 Originally established in Williams's hennoxazole synthesis, the in situ transmetalation protocol, which employs an allylstannane, boron tribromide, and (R,R)- or (S,S)-1,2-diamino-1,2-diphenylethane as the chiral controller, was adapted in this phorboxazole synthesis (yellow box in Figure 4). The chiral ligand had been introduced by Corey in enantioselective Diels–Alder reactions and asymmetric allylations. The boron bromide ligand system built up from the C2-symmetric ligand and BBr3 is used as a chiral electrophilic boron atom ready for the SE2′ transmetalation step with the allylic stannane, which is derived from malic acid. The Williams synthesis of phorboxazole A. 13 LS-Selectride= lithium tris(1,3-dimethylbutyl)borohydride, HWE=Horner–Wittig–Emmons olefination. The diastereofacial selectivity of the addition is determined solely by the chiral auxiliary. An iterative asymmetric allylation process allowed construction of all stereocenters (except C13) in the C5–C15 bistetrahydropyran region. Ring closure towards the THP systems was achieved by mesylation and subsequent SN2 displacement. The C13 hydroxy functionality was generated under substrate control by using LS-Selectride. The allylation strategy was utilized a third time for the generation of the C37 stereocenter. As a result, three of the four tetrahydropyran systems were established by using the asymmetric in situ transmetalation–allylation strategy. As seen before in the Smith synthesis, a 2,4-disubstituted oxazole was used as a bifunctional linchpin for the convergent assembly of fragments. In contrast to the Smith group, Williams et al. used a masked iodomethyloxazole aldehyde for this purpose. First, a SmI2-mediated Barbier coupling of the C28–C32 iodomethyloxazole with the C33–C46 aldehyde yielded a 1:1 diastereomeric mixture of C33 alcohols. Subsequently, the oxazole aldehyde was set free by a deprotection–oxidation sequence and combined with a C20–C27 β-ketophosphonate in a Horner–Wadsworth–Emmons reaction, which delivered the trisubstituted C27–C28 olefin with excellent E selectivity. The C22–C26 stereopentade had been created by two successive Evans aldol processes, in which four contiguous stereocenters had been generated (Scheme 5). Under Luche conditions the C26 ketone could be reduced regio- and diastereoselectively. In a very elegant manner, the C22–C26 tetrahydropyran was established by internal capture of a transoid allylic cation at C26 by the C22 oxygen atom with concomitant cleavage of the MOM ether (Scheme 5). The configuration at C26 in the neutral precursor is irrelevant in this cyclization step. Finally, the macrolide was formed by Wittig olefination (C19–C20, >95:5 E/Z) and intramolecular Horner–Wadsworth–Emmons olefination (C2–C3, 4:1 Z/E). a) Tf2O, pyr, CH2Cl2, −20 °C, 12 h, 55 %. MOM=methoxymethyl. Preliminary results from SAR studies were reported by Uckun and Forsyth.14 Seven synthetic analogues were tested against three different human cancer cell lines. These studies revealed that the macrolide, the central C29–C31 oxazole, and the C39–C46 side chain are necessary for anticancer activity, whereas the C45–C46 vinyl bromide and the free C33 hemiketal can be replaced by an alkyne and a mixed ketal, respectively. It was speculated from these results that the biological activity might depend on multiple point contacts on a cell receptor. In addition, the rigid C27–C31 vinyloxazole substructure might place the macrolide and side-chain substructures in the necessary relative orientation for receptor binding. In this context, further SAR studies are necessary to identify a minimal pharmacophore. As the configuration of C38 is thought to play a crucial role in the overall molecular shape of phorboxazole, a variation of this stereocenter is of special interest as well as deletions within the macrolide core structure. The work on the phorboxazoles demonstrates how the judicious application and development of reagent and substrate control as well as the chiral pool have successfully enabled chemists to tackle a most challenging stereochemical problem. Besides improving known protocols and inventing new ones in the race for the phorboxazoles, researchers have shown how important creativity and nonlinear thinking can be. Total synthesis remains a yardstick for measuring the progress of organic chemistry and natural product chemistry.