Recent progress on the total synthesis of acetogenins from Annonaceae

Introduction

Annonaceous acetogenins (ACGs) constitute a series of natural products isolated exclusively from Annonaceae species [1-5] that are widely distributed in tropical and sub-tropical regions. Since uvaricin [6], the first acetogenin identified from the roots of Uvaria acuminata, more than 400 members of this family of compounds have been isolated from 51 different species [7].

The common skeleton is most often characterised by an unbranched C₃₂ or C₃₄ fatty acid ending in a γ-lactone. Several oxygenated functions, such as hydroxyls, ketones, epoxides, tetrahydrofurans (THF) and tetrahydropyrans (THP), may be present, as well as double and triple bonds. Thus several types of ACGs have been characterised, based on the nature of the functional groups which are present. These including mono-THF, adjacent bis-THF, nonadjacent bis-THF, tri-THF, adjacent tetrahydrofuran-tetrahydropyran (THF-THP), nonadjacent THF-THP, mono-THP, and acetogenins containing only γ-lactones.

These compounds are known to exhibit a broad range of biological activities, the precedent for which came from early South American populations, who used extracts of Annonaceae plants as pesticidal and antiparasitic agents [8]. The proven activities of the acetogenins now include (but are not limited to): pesticidal, antifeedant, antiprotozoal, immunosuppressive and probably most importantly, antitumor [3]. In this respect they are known to be very potent cytotoxic compounds, targeting the reduced nicotinamide adenine dinucleotide (NADH): ubiquinone oxidoreductase (also known as complex I) which is a membrane bound protein of the mitochondrial electron transport system, and the ubiquinone linked NADH oxidase in the plasma membrane of cancerous cells [9,10]. Inhibition by these mechanisms results in adenosine triphosphate (ATP) deprivation, which leads to apoptosis of the highly energy demanding tumor cells [11]. The acetogenins are now considered as the most potent (effective in nanomolar concentrations) known inhibitors of the mitochondrial complex I [9,12]. More recently the annonaceous acetogenins have also been shown to overcome resistance in multidrug resistant (MDR) tumors [13].

Because of their structural diversity and the numerous biological properties, many authors are working on the total synthesis of ACGs. Previous reviews on the total synthesis of ACGs have already been published [14-18]; since 1996 over 100 total syntheses of all types of ACGs have appeared in the literature, illustrating the creativity of chemists. Convergent, linear, and biomimetic approaches have been used, relying on the use of cheap chiral starting materials (e.g. amino acids, sugars, tartaric acid, etc.) or on asymmetric reactions {e.g. Sharpless asymmetric epoxidation (AE), Sharpless asymmetric dihydroxylation (AD), diastereoselective Williamson etherification, etc.}. Semi-synthesis of natural ACGs as well as derivatised ACGs (e.g. amines, esters, and glycosylated ACGs) and preparation of structural analogues (e.g. simplified mimics, chimeras) have also been reported. This overview covers only examples of the first total synthesis and the syntheses that helped to correct structures achieved during the past 10 years (largely up to 2007). Indeed, total synthesis is a key tool for the complete structure determination of ACGs, since several absolute configurations of stereogenic centres are rather difficult to determine without comparison of the spectroscopic data, and/or the chromatographic properties of several stereoisomers. This review of total synthesis of every type of ACGs is presented in order of publication.

Review

Total synthesis of ACGs

The significant biological activity of the acetogenins, as well as their interesting and diverse structures, has stimulated substantial interest in their chemical synthesis.

1 Mono-THF ACGs

Total synthesis of longifolicin

In 1998, Marshall’s group [19] reported the total synthesis of longifolicin (1) (Scheme 1). Treatment of the advanced intermediate 2 with tetrabutylammonium fluoride (TBAF) in THF gave the THF product 3. Attachment of the butenolide moiety to the THF intermediate 4 was achieved through introduction of a chiral propargylic alcohol segment to yield 6, to which an allenyl Pd hydrocarbonylation methodology was applied to form the butenolide 7. The MOM protecting groups in 7 were cleaved to give longifolicin (1) in high yield. The ¹H and ¹³C NMR spectra were identical to those of an authentic sample. Furthermore, the rotation ([α]_D +13.5, c 0.37, CH₂Cl₂) and mp (79–80 °C) were in close agreement with the reported values ([α]_D +13.0, c 0.001, CH₂Cl₂; mp 83 °C) [20]; so this total synthesis confirmed the structural assignment of longifolicin.

Total synthesis of corossoline

Corossoline (8) was originally isolated [21] from the seeds of Annona muricata in 1991. Its absolute stereochemistry except for the C-8′ position was deduced by applying Mosher’s new methodology to the monotetrahydrofuranyl ACGs such as reticulatacin [22] and by a total synthesis of (8′RS)-corossoline by Wu’s group [23,24]. In 1996, Tanaka’s group [25] reported the total synthesis of two possible diastereomers (8′R)- and (8′S)-8a to confirm the stereochemistry of the C-8′ hydroxyl group (Scheme 2). Their synthetic approach started from 1-iodododecane (9) and 5-(tetrahydropyran-2-yloxy)pent-1-yne (10). Asymmetric dihydroxylation with AD-mix-β on 11 and subsequent acid-catalyzed cyclization with camphorsulfonic acid (CSA) resulted in THF ring-containing building block 12, which was converted into alkyne 13. The alkylation of iodide 14 with the sodium enolate of 15 afforded 16. Transformation of 16 following two different procedures afforded isomeric epoxides 17 and 18, respectively. Coupling between the lithium salt of 13 and 17 (or 18) followed by hydrogenation, oxidation and deprotection afforded (8′S)- or (8′R)-8a. Comparison of the mp, [α]_D, IR and NMR data of both synthetic materials with those reported for natural corossoline did not allow for the strict determination of the configuration at the C-8′ hydroxyl group of 8.

To establish the previously undefined configuration at C-8 of the natural 8, in 1999, Wu’s group [26] reported the synthesis of (8R)- and (8S)-corossoline and assigned the absolute configuration of natural-8 at C-8 to R (Scheme 3). In their synthesis, asymmetric epoxidation of 19 directed by L-(+)-diisopropyl tartrate gave epoxy alcohol 20; the THF ring of 21 was then constructed under CSA catalyzed one-pot transformation. After deprotection of the isopropylidene acetal of 21 and oxidative cleavage of the resulting diol, the resultant aldehyde was treated with chiral allenylboronic ester to afford 22. The terminal epoxides 24 and 26 were prepared from the same intermediate 23. Coupling of 22 with 24 (or 26) in the same manner followed by hydrogenation and removal of the MOM protecting group gave (8S)-8b (or (8R)-8b) respectively. Comparison of the physical data of both synthetic corossolines with those reported for the natural one showed that the configuration at C-8 of the natural corossoline was highly likely to be R.

Synthesis of pseudoannonacin A

Annonacin A was isolated from seeds of Annona squamosa and obtained as an amorphous solid, [α]_D +23.8 (c 0.4, CH₂Cl₂) [27]. It was characterized spectroscopically, and the relative configuration of the central THF ring was assigned as being threo-trans-erythro (C_16R,C_19R). But the configurations at C-4 and C-10 remained unknown. In 1998, Hanessian’s group [28] reported the stereocontrolled first total synthesis of a diastereomer of the presumed mono-THF type acetogenin annonacin A utilizing the “Chiron approach” (Scheme 4). The well-known lactone 30 obtained from L-glutamic acid was elaborated to a lactone intermediate 31; further manipulation and chain elongation at both ends of 31 produced an advanced intermediate 32. The anion of the phenyl sulfone 33, readily prepared from D-glutamic acid, was condensed directly with the ester function of 32 to give the α-keto sulfone 34, which was futher elaborated to ester 35. Condensation of the enolate derived from 35 with O-THP (S)-lactaldehyde followed by mesylation, elimination and deprotection afforded 29. The synthetic product (a mixture of epimers at C-10) had spectroscopic data identical to those of the natural product, but a different optical rotation. They designated the (C_15R,C_16S,C_19S,C_20S) erythro-trans-threo isomer 29 as pseudo-annonacin A. The actual configuration of the natural product remains unknown.

Total synthesis of tonkinecin

Tonkinecin (36) is a mono-THF acetogenin with a hydroxyl group at C-5, and was recently isolated from roots of Uvaria tonkinesis by Yu’s group [29]. This compound has demonstrated potent cytotoxicity against hepatoma (Bel 7402) (IC₅₀ = 1.5 µM), gastrocarcinoma (BGC) (IC₅₀ = 5.1 µM), colon adenocarcinoma (HCT-8) (IC₅₀ = 0.38 µM), and leukemia (HL-60) (IC₅₀ = 0.52 µM) human tumor cell lines [29]. Tonkinecin (36) was firstly synthesized by Wu’s group in 1999 [30] which used a palladium-catalyzed cross-coupling reaction between the butenolide 39 and the THF unit 22 as the key step for constructing the backbone of 36 (Scheme 5). The synthesis of aldehyde 37 began with D-xylose and involved construction of a γ-lactone moiety utilizing Wu’s own methodology. Wittig reaction of the aldehyde 37 and the Wittig reagent 38 furnished the butenolide unit 39. The tetrahydrofuran part of 22 was constructed from D-glucose via epoxide 40. The entire carbon skeleton of 41 was constructed by Pd(0)-catalyzed cross-coupling reaction between 39 and 22. Selective hydrogenation and removal of the MOM protecting group gave tonkinecin (36). The physical data of their synthetic sample were identical to those of the natural one. In 2001, the full details of this total synthesis were reported [31].

Total synthesis of gigantetrocin A

Gigantetrocin A (42) was isolated by McLaughlin’s group from Goniothalamus giganteus [32] and showed significant and selective cytotoxicity to human tumor cells in culture [32,33]. In 2000, Shi’s group reported the first simple total synthesis of gigantetrocin A (Scheme 6) [34]. They obtained the trans-THF ring building block 45 by means of the Sharpless AD reaction and cyclization catalyzed by Co(modp)₂ (the Mukaiyama epoxidation method) under mild reaction conditions.

The connection of the THF unit 46 with the γ-lactone segment 47 was carried out by means of a Wittig reaction. The target compound 42 had specific rotation and spectral data matching those reported in the literature [32].

Total synthesis of annonacin

Annonacin (48), the first mono-THF acetogenin discovered, was isolated by Cassady’s group from the stembark of Annona densicoma in 1987 [35]. This compound demonstrated astrocytoma reversal (9ASK) activity (15−30% reversal at 100 µg/mL) and high cytotoxicity against human nasopharyngeal carcinoma (KB) and mouse leukemia (P388) [35,36]. In 2000, Wu’s group reported the first total synthesis of annonacin (Scheme 7) [37]. In their synthetic route, the R-hydroxy ester 49 obtained from L-ascorbic acid was elongated by two carbons to give ester 50 using a four-step sequence, the protected diol of which was then treated with H₅IO₆ to give the chiral aldehyde 51. On the other hand, 49 was converted to phosphonium salt 52, which was treated with aldehyde 51, and the resulting alkene was further elaborated to afford the epoxide 53. Next, the lithiated derivative of THF alkyne 22, which was prepared from the D-glucono-δ-lactone-derived α-hydroxyl ester 54 through Sharpless AD reaction as a key step, was treated with epoxide 53 in the presence of BF₃·Et₂O to afford alkynol 56. Catalytic hydrogenation of 56 and subsequent construction of the butenolide segment finished the total synthesis of annonacin (48), whose R_f value and spectroscopic data were identical to those reported for the natural product. In 2001, the full details of this total synthesis were reported [31].

Total synthesis of solamin

Solamin (57), a cytotoxic mono-tetrahydrofuranic γ-lactone acetogenin isolated from Annona muricata seeds in 1991 [38], was synthesized by Kitahara’s group in 1999 [39] through a direct coupling reaction (Scheme 8). The mono-THF unit 58, which was prepared from D-glutamic acid, was treated with the sodium enolate of 59 to afford the main structure 60. Oxidation of the sulfide 60 followed by thermal elimination and deprotection completed the total synthesis of solamin (57). The data of the synthetic 57 were identical to those of an authentic sample [38].

In their total synthesis of solamin (57), Mioskowski’s group [40] reported the first application of the RCM reaction using a ruthenium catalyst thus demonstrating the efficiency of this reaction for the construction of ACGs (Scheme 9). Both allyl alcohol 63 and vinyl-substituted epoxide 66 were synthesized via alkyne reduction yielding (E)- or (Z)-allylic alcohol followed by Sharpless AE using (+)- or (−)-DET. This synthesis was quite flexible and all stereoisomers of the central THF core of solamin were easily obtained. The allyl alcohol 63 and the vinyl epoxide 66 were then coupled to construct the solamin skeleton 67. Subjecting of 67 to 1,3-dimesitylimidazol-2-ylidene ruthenium benzylidene catalyst (RuCl₂(=C(H)-Ph)(PCy₃)(IMes)) after protection of the free hydroxyl group afforded the RCM product 68. Hydrogenation of the double bond of dihydrofuran 68 and iodination of the primary alcohol gave 69, which was then utilized to alkylate the sodium enolate of lactone 15 to afford 70. Oxidation of the sulfide 70 followed by thermal elimination and finally removal of the two silyl protective groups gave solamin (57). But after careful comparison the configuration of the OH-group in C13 between the literature for isolation [38] and this total synthesis by Mioskowski’s group, we found that Mioskowski’s group may have made a mistake in the configuration of the OH-group in C13, and the configuration of the OH-group in C13 should be R as in Scheme 8, not S in Scheme 9.

Total synthesis of cis-solamin

cis-Solamin is a mono-THF acetogenin isolated from Annona muricata in 1998 [41]. To establish the absolute configuration of cis-solamin, two candidates 71a and 71b were synthesized by Makabe’s group in 2002 employing a TBHP-VO(acac)₂ diastereoselective epoxidation followed by a cyclization strategy (Scheme 10) [42]. In their synthesis, diastereoselective epoxidation of 72 and spontaneous cyclization afforded the diastereomers 73a and 73b. Protection of the hydroxyl group of 73a as a MOM ether afforded 74, which was coupled with the γ-lactone precursor 75 by a Sonogashira cross-coupling reaction to give compound 76. Catalytic hydrogenation of 76 using Wilkinson’s catalyst and oxidation of the sulfur with mCPBA followed by thermal elimination afforded the candidate 71a. The other candidate 71b was synthesized using the same reaction sequence as that employed for 71a. By comparison of the optical rotation of the two possible diastereomers, it was suggested that the absolute configuration of natural cis-solamin was 71a.

In 2002, Brown’s group [43] also reported the synthesis of cis-solamin (71a) and its diastereomer 71b using the diastereoselective permanganate-promoted oxidative cyclization of 1,5-dienes to create the THF diol core (Scheme 11). Notably, no protecting groups were required during the stages of fragment assembly. The synthesis of precursor 78 for the oxidative cyclization reaction was completed by hydrolysis of 77 and activation of the resulting unsaturated acid as the acid chloride followed by reaction with lithiated (2S)-10,2-camphorsultam. The key oxidative cyclization reaction, conducted under phase-transfer conditions, introduced the C15, C16, C19, and C20 stereocenters present in cis-solamin in one step, then the auxiliary was best removed from 79 by reduction using NaBH₄. The resulting diol was taken forward by conversion to the epoxide 80. Addition of the C3-C13 fragment in a copper-catalyzed Grignard reaction afforded 81. The butenolide ring in 4,5-dehydro-cis-solamin (83) was put in place using a ruthenium catalyzed Alder-ene reaction of 81 with 82. Final selective reduction of the 4,5-double bond completed the synthesis of 71a. The diastereomeric structure 71b was also synthesized following the same route but using the (2R)-10,2-camphorsultam. In 2004, the full details of this total synthesis were reported [44].

In 2005, Donohoe’s group [45] reported the formal synthesis of (+)-cis-solamin using oxidative cyclization of diol 84 with catalytic osmium tetroxide which gave the THF product 85 in high yield (Scheme 12).

In 2006, Stark’s group [46] accomplished an enantioselective total synthesis of cis-solamin using a highly diastereoselective ruthenium tetroxide catalyzed oxidative cyclization as a crucial transformation (Scheme 13). In their synthetic route, commercially available (E,E,E)-1,5,9-cyclododecatriene 86 was readily converted into diene 87. Standard silyl protection of diol 87 afforded the cyclization precursor 88. Treatment of diene 88 under the ruthenium tetroxide catalyzed oxidative conditions resulted in a smooth conversion of the starting material into THF 89. Triol (+)-90 was obtained with lipase Amano AK desymmetrization. For the appropriate side chain attachment, the termini were differentiated to give lactone (−)-91. Reduction of (−)-91 followed by a Wittig reaction yielded fully deprotected product (−)-93. Finally, the introduction of the butenolide segment using a ruthenium(II)-catalyzed Alder-ene reaction followed by selectively reduction furnished cis-solamin (71a). Spectroscopic data for this compound were identical to those reported for cis-solamin isolated from natural sources [41].

Total synthesis of mosin B

Mosin B (94) is a mono-THF acetogenin isolated by McLaughlin’s group [47] from the bark of Annona squamosa and shows selective cytotoxic activity against the human pancreatic tumor cell line, pancreatic cancer cells (PACA-2) (ED₅₀ = 2.5 × 10⁻⁴ µg/mL), with a potency 100 times that of adriamycin (ED₅₀ = 1.8 × 10⁻² µg/mL) [47]. In 2001, a total synthesis of the threo/trans/erythro-type acetogenin mosin B (94a) and one of its diastereomers 94b had been achieved by Tanaka’s group (Scheme 14) [48]. The THF core segment 97 was stereoselectively constructed by iodoetherification of E-allylic alcohol 96, which was prepared from chiral alcohol 95a. The γ-lactone segment 99 was synthesized by α-alkylation of α-sulfenyl γ-lactone 15 with 98b. The carbon skeleton 100 was assembled in a convergent fashion from 97 and 99 through a Nozaki-Hiyama-Kishi reaction. Oxidation of 100 followed by deprotection afforded the candidate 94a. The other candidate 94b was synthesized from 95b using the same procedure. Comparison of the specific rotation of synthetic 94a and 94b with the naturally occurring mosin B suggested that the absolute configuration of natural mosin B was 94a.

Total synthesis of longicin

In 1995, McLaughlin’group reported the isolation of longicin (101) from Asimina longifolia (Annonaceae). Longicin was reported to exhibit over 1 million-fold selective antitumor activity against PACA-2 (IC₅₀ = 1.25 × 10⁻⁹ µg/mL) compared to adriamycin (IC₅₀ = 1.95 × 10⁻³ µg/mL)[49]. In 2005, Hanessian’s group [50] reported the first total synthesis, stereochemical assignment, and structural confirmation of longicin (101) (Scheme 15). The strategy involved the use of Grubbs’ RCM reaction as a “chain elongation” strategy for the synthesis of acetogenin-type structures and a new protocol for butenolide incorporation. Prepared from D-glutamic acid, lactone 102 was converted to the desired threo-trans-erythro THF isomer 103, which could be converted to the different two diolefins 104 and 105. The 14- (106) and 11-membered (107) ring lactones were obtained by an ester-tethered RCM macrocyclization. Hydrogenation of the olefins followed by saponification of macrolactones 106 and 107 with NaOMe and subsequent MOM protection gave the common intermediate 108 with identical physical data independent of the route used. The lithium enolate formed from 108 with LDA was treated with 109 and the resulting aldol product was desilylated and in situ lactonization gave 110. Removal of the MOM groups afforded longicin (101), which was identical to the natural product on the basis of the reported physical constants [49].

Total synthesis of murisolin

Murisolin (111) is a mono-THF acetogenin, isolated from the seed of Annona muricata by Cortes’s group [51], which shows selective cytotoxic activity against human lung carcinoma (A-549) (ED₅₀ = 5.90 × 10⁻⁸ µg/mL), human colon adenocarcinoma (HT-29) (ED₅₀ = 6.58 × 10⁻⁸ µg/mL), and human kidney carcinoma (A-498) (ED₅₀ = 1.09 × 10⁻⁹ µg/mL) with potency from 10⁵ to 10⁶ times that of adriamycin (ED₅₀ = 3.99 × 10⁻³ µg/mL for A-549, ED₅₀ = 2.43 × 10⁻² µg/mL for HT-29 and ED₅₀ = 2.26 × 10⁻³ µg/mL for A-498) [52]. In 2004, Tanaka’s group [53] reported the first total synthesis of murisolin (111) (Scheme 16). The threo/trans/threo-type THF ring moiety 116 was constructed with excellent stereoselectivity by asymmetric alkynylation of α-tetrahydrofuranic aldehyde 114 with 1,6-heptadiyne (115). Then Sonogashira coupling of 116 and the iodide 117 followed by hydrogenation and deprotection provided murisolin (111). The spectroscopic data of synthetic 111 (¹H NMR, ¹³C NMR, IR, MS, mp) were in good agreement with those reported. In 2005, the full details of this total synthesis [54] were reported, along with the total synthesis of natural 16,19-cis-murisolin 112 and unnatural 16,19-cis-murisolin 113 from 118 and ent-118 respectively using a similar procedure.

In 2004, Curran’s group [55] reported a 4-mix/4-split strategy for the synthesis of a stereoisomer library of (+)-murisolin and 15 of its isomers, which relied on solution phase technique of fluorous mixture synthesis (Scheme 17). In their synthetic route, a single mixture of M-119, which was tagged with different fluorous PMB tags, was transformed into alkene M-120 and was then followed by two splits. First, each of the two mixtures was subjected to a Shi epoxidation with enantiomeric ketone catalysts. Later, these two mixtures were split again, half being subjected to a Mitsunobu reaction and the other half not. Ultimately, they obtained four mixtures M-111a–d, each containing four isomers, which were demixed and detagged to provide all 16 target isomers. In 2006, the full details of this work were described [56].

In 2006, Makabe’s group [57] reported the total synthesis of murisolin (111), (15R,16R,19S,20S)-cis-murisolin (112), and (15R,16R,19R,20S)-murisolin A (121) (Scheme 18). The mono-THF moieties were synthesized from epoxy alcohol 122 by using Sharpless AD-mix-β for the threo-trans-threo THF moiety 123, a AD-mix-β followed by the Mitsunobu reaction for the erythro-cis-threo THF moiety 125, and AD-mix-α for threo-trans-threo THF moiety 124. The α,β-unsaturated γ-lactone segment 127 was synthesized through alkylation of lactone 15 and iodide 126. The THF moiety 123 and γ-lactone 127 were coupled by a Sonogashira cross-coupling reaction to afford 111. The syntheses of 112 and 121 were carried out as described for 111.

Total synthesis of reticulatain-1

Reticulatain-1 (128) is a mono-THF acetogenin, isolated from Annona reticulata in 1995 [58]. To determine the absolute configuration of natural 128, Makabe’s group [59] reported the total synthesis of 128a and 128b in 2004 (Scheme 19). 130 was obtained by using the Sharpless epoxidation and dihydroxylation of 129. Compound 130 was then subjected to the Mitsunobu inversion to afford 131, which was transformed into 125. Then the THF moiety 125 and γ-lactone moiety 132 were coupled by a Sonogashira cross-coupling, and diimide reduction followed by deprotection allowed completion of the total synthesis of candidate 128a. The other candidate 128b was synthesized from 133 using the same procedure as that employed for 128a. However, both of the specific rotation of synthetic 128a ([α]_D³⁰ +9.68, c 1.00, CHCl₃) and 128b ([α]_D²⁷ +2.34, c 1.00, CHCl₃) are lower than the reported value of natural occurring reticulatain-1 ([α]_D +22, c 1, CHCl₃) [58], so comparison of the specific optical rotations of 128a and 128b did not allow for the strict determination of the absolute configuration. So the structure of natural reticulatain-1 has not been confirmed.

Total synthesis of muricatetrocin

In 1996 McLaughlin’s group reported the isolation of muricatetrocin C (134) from the leaves of Rollinia mucosa [60]. The molecule exhibited potent inhibitory action against prostatic adenocarcinoma (PC-3) (ED₅₀ = 1.35 × 10⁻⁷ µg/mL), PACA-2 (ED₅₀ = 5.69 × 10⁻⁷ µg/mL), and A-549 (ED₅₀ = 5.55 × 10⁻⁶ µg/mL) [60]. In 2000, Ley’s group [61] reported the first total synthesis of muricatetrocin C (134) (Scheme 20). The anti-1,2-diol component 136 was obtained through selective chemical differentiation of the hydroxyl termini in diol 135. The 2,5-trans-disubstituted THF unit 140 was then constructed by ozonolysis of the alkenol 137 to give the lactol 138. Treatment of 138 with an excess of propargyl alcohol afforded 139, which was followed by anomeric O-C rearrangement to give 140. The hetero-Diels-Alder (HDA) reaction between diene 141 and nitrosobenzene followed by N-O bond cleavage and elimination of the aryl amine to reintroduce the butenolide unsaturation afforded 143. Then coupling reaction between 136 and 140 provided 144, which was readily transformed into 145. Sonogashira cross-coupling of 145 with the vinyl iodide 143 followed by selective hydrogenation, desilylation and removal of the butane diacetal group finished the total synthesis of 134. The spectroscopic data for synthetic 134 (¹H NMR, ¹³C NMR, IR, MS, mp and specific rotation) were in excellent agreement with those reported for naturally occurring muricatetrocin C (134). In 2002, the full details of this work were described [62].

In 1993 McLaughlin’s group reported the isolation of two new mono-THF acetogenins from Annona muricata [33]. They were named muricatetrocin A and B. In 1994 Yang’s group published analytical data for howiicin E isolated from Goniothalamus howii, which indicated a constitutional identity and a stereochemical match for muricatetrocin A and howiicin E [63]. In 2000, Koert’s group [64] reported the total synthesis of (4R,12S,15S,16S,19R,20R,34S)-muricatetrocin (146) and (4R,12R,15S,16S,19R,20R,34S)-muricatetrocin (147) based on a modular synthetic strategy which had been used in the total synthesis of mucocin (Scheme 21) [65]. The ylide 149, which was synthesized from the cis-THF alcohol 148a, was coupled with the butenolide aldehyde 150 via a Wittig reaction to afford the THF aldehyde 151 after further 3 steps. Then addition of the magnesium derivative of iodide 152 to aldehyde 151 followed by global deprotection provided compound 146. The compound 147 was synthesized from the trans-THF alcohol 148b following a similar procedure. Compound 146 showed analytical data in agreement with howiicin E and a fit with the data for muricatetrocin A if one reassigns the reported ¹³C signals for C(13) and C(14). Compound 147 matched muricatetrocin B in respect to all NMR data. However, a lower optical rotation was found for 147 ([α]_D²⁸ = +6.7) than was reported for the natural product ([α]_D²⁵ = +15.0).

2 Adjacent bis-THF ACGs

The core unit of the adjacent bis-THF acetogenins contains six oxygenated stereocenters, and much of the synthetic work on the family has been focused in that direction. The first successful approach was recorded in 1991 by Hoye’s group who employed a two-directional inside-out epoxide cascade sequence to prepare a core enantiomer of uvaricin [66]. This synthesis was important in establishing the absolute stereostructure of the natural product. Subsequently, numerous synthetic approaches to related core THF arrays have been reported.

Total synthesis of parviflorin

Parviflorin (153), a relatively rare C₃₅ adjacent bis-THF acetogenin, was isolated by McLaughlin’s group both from Asimina parviflora [67] and from Annona bullata [68]. Parviflorin showed remarkable selectivity in its cytotoxicity against certain human solid tumor cell lines [67,68]. In 1996, Hoye’s group [69] achieved the first synthesis of parviflorin (153) by using a highly efficient construction of the adjacent bis-THF backbone (Scheme 22). 1,5,9-Cyclododecatriene (86) was converted to the bis-allylic alcohol 154 through selective oxidation and Wittig extension followed by DIBAL-H reduction. The stereogenic centres in the bis-THF backbone 156 were then installed by sequential double Sharpless AE/Sharpless AD. The building block 157 was then constructed through bidirectional chain synthesis strategy. The propargylic alcohol 159, obtained from 1,4-bis(alkenyloxy)benzene 158, was converted into butenolide 160 through Red-Al reduction, iodine treatment, and carbonylation. Oxidative release of 160 followed by Swern oxidation and Takai reaction provided the terminal vinyl iodide 161. Final Pd⁰-catalyzed coupling of alkyne 157 with vinyl iodide 161 gave the enediyne 162, which underwent selective hydrogenation and desilylation to give (+)-parviflorin (153).

In 1997, Trost’s group [70] reported the total synthesis of (+)-parviflorin (153) through a flexible approach (Scheme 23). The bis-acetonide 165, which was constructed from the aldehyde 163 and 164, was hydrolyzed and then treated with base to give the bis-THF 166. Then Ru-catalyzed Alder–ene type reaction of 166 with 82 yielded the butenolide 167. Hydrogenation of the double bond afforded (+)-squamocin K, while diastereoselective dihydroxylation of the same double bond yielded (5S)-hydroxyparviflorin 168. Chemoselective deoxygenation of the C-5 hydroxyl group afforded parviflorin, spectroscopically identical to the natural product.

Total synthesis of trilobacin

Trilobacin (169), the first known member of the adjacent bis-THF ACGs with a threo, trans, erythro, cis, threo backbone, was isolated from the the bark of Asimina triloba by McLaughlin’s group. Studies with human solid-tumor cell lines showed that trilobacin (169) (ED₅₀ <10⁻¹⁵ µg/mL) was over 1 billion times more potent against HT-29 than adriamycin (ED₅₀ = 6.69 × 10⁻³ µg/mL) [71,72]. In 1996, Sinha’s group [73] reported the first total synthesis of 169 (Scheme 24). The phosphonium salt 172 and aldehyde 175 were prepared using the AD reaction from alkenes 170 and 173, respectively. Coupling of the Wittig reagent 172 with the aldehyde 175 produced the alkene 176. Oxidative cyclization with Re₂O₇/lutidine afforded the corresponding trans-substituted tetrahydrofuran 177. Alternatively, Mitsunobu inversion of the free alcohol within 177, prior to its activation and ring-closure, gave lactone 178, which was converted to the primary Wittig salt 179. Treatment of 179 with BuLi and then with aldehyde 180 followed by hydrogenation and deprotection afforded 169, which was found to be identical (¹H NMR, [α]_D, IR, MS) to the naturally occurring trilobacin.

Total synthesis of annonin I and asiminacin

Among the 19 ACGs that were isolated from Annona squamosa Born’s group found annonin I (181a) [74] to be the most potent compound concerning cytotoxic and insecticidal activity. With the aim to prepare interesting substances for biological assays, Scharf’s group reported the first total synthesis of 15-epi-annonin I (181b) in 1996 (Scheme 25) [75]. Ring opening – ring closing epoxide cascade [66] was performed on 182 in the presence of hexafluorosilicic acid. This produced the bis-THF moiety 183. Opening of epoxide 183 with alkyne 184 followed by epoxide formation at the other end of the molecule afforded 185, which was attached to lactone 186 by another epoxide opening, which produced the main structure 187. Finally, formation of the butenolide followed by removal of the silyl protecting groups finished the total synthesis of 15-epi-annonin I (181b).

Squamocin A (181a) [76] (also called annonin I [74]) and squamocin D (188) [77] (also called asiminacin [78]) belong to a subclass of ACGs with an adjacent bis-THF subunit and an extra hydroxy group in the left side chain (C-28). Both natural products show remarkable cytotoxic activity and are interesting antitumor candidates [76,77]. In 1999, Koert’s group reported the total synthesis of these two natural products (Scheme 26) [79]. The bis-THF core of 190 with the relative threo-trans-threo configuration was constructed by an established multiple Williamson reaction on 189. Monoprotection of 190 followed by oxidation gave the aldehyde 191, which could readily be converted into the aldehyde 192. The bromide 193 was transformed into the corresponding Grignard reagent, which was allowed to react with the aldehyde 192 to afford the two epimers 194 and 195. The dianion of 196 was allowed to react with (S)-propene oxide, and subsequent deprotection of the three silyl ethers gave the target compound squamocin A (181a). Along the same route squamocin D (188) was obtained from 197. The spectrocopic data for squamocin A and squamocin D matched the literature data. In 2000, the full details of this synthetic work were reported [80].

Total synthesis of asiminocin

McLaughlin’s group published a report on the isolation and structure elucidation of asiminocin (198), a C₃₇ ACG with nearly one billion times the cytotoxic potency of a standard reference, adriamycin, as measured against breast carcinoma (MCF-7) (ED₅₀ <10⁻¹² µg/mL compared to adriamycin, ED₅₀ = 1.76 × 10⁻² µg/mL) [78,81]. In 1997, Marshall’s group reported the total synthesis of asiminocin (198) through a bidirectional approach starting from the (S,S)-tartrate derived dialdehyde 200 and the (R)-α-OSEM stannane 199 (Scheme 27) [82]. Addition of 199 to 200 in the presence of InCl₃ afforded the bis-adduct, anti-diol 201. The derived tosylate 202 was converted to the bis-THF core unit 203 upon treatment with TBAF. Oxidation to aldehyde 204 followed by InCl₃-promoted addition of the (S)-allylic stannane 205 gave the anti adduct 206. Removal of the OH group by reduction of the tosylate 207 with LiBEt₃H yielded the SEM ether 208. Conversion to the vinyl iodide 209 followed by Pd⁰-catalyzed coupling with the (S)-alkynyl butenolide 210 gave the asiminocin precursor 211. Selective hydrogenation of the enyne moiety with diimide and cleavage of the SEM protecting groups completed the synthesis of triol 198, which exhibited ¹H and ¹³C NMR spectra indentical to those of asiminocin [81].

Total synthesis of asiminecin

In 1997, Marshall’s group reported the total synthesis of asiminecin (212) starting from aldehyde 204 and the OTBS allylic stannane 205 (Scheme 28) [82]. Addition of the latter to the former in the presence of InCl₃ afforded the anti adduct 213 which was protected as the SEM ether 214. Hydrogenation followed by OTBS cleavage with TBAF and selective silylation of the primary alcohol with TBSCl and Et₃N-DMAP led to the secondary alcohol 215. Tosylation and hydrogenolysis with LiEt₃BH removed the C-30 OTs group affording the SEM ether 216. The remaining steps for the completion of total synthesis of asiminecin were carried out along the lines described for asiminocin.

Total synthesis of (+)-(30S)-bullanin

(+)-Bullanin was isolated from the stem bark of Asimina triloba as an inseparable mixture of 30S and 30R diastereomers [83]. In 1998, Marshall’s group reported the total synthesis of (+)-(30S)-bullanin (217) through S_E2′ additions of oxygenated non-racemic allyl stannane (Scheme 29) [84]. Transmetallation of stannane 219 with InCl₃ in the presence of aldehyde 218a afforded the expected anti-adduct 220. Addition of stannane 222 to aldehyde 221 in the presence of BF₃·OEt₂ afforded the syn-adduct 223 as the only detectable stereoisomer. Tosylation of the alcohol followed by exposure to TBAF promoted bis-THF cyclization. Introduction of the butenolide moiety finished the total synthesis of (+)-(30S)-bullanin (217). The identity of this material with that of the 30S natural isomer was established through ¹H NMR comparison of the tri-(S)-MTPA (Mosher) ester with that of the (S)-Mosher ester of the mixture derived from natural sources. The optical rotation of their synthetic material, [α]_D +24, was in close agreement with the reported value for the mixture, [α]_D +28.

Total synthesis of uvaricin

Uvaricin (225), an adjacent bis-THF acetogenin which was isolated in 1982 from Uvaria acuminata, was of special historical value because it was the first ACG discovered [6]. In 1998, the group of Sinha and Keinan reported the first total synthesis of the naturally occurring isomer 225 using three consecutive Sharpless AD reactions to place the necessary oxygen functions on a “naked” carbon skeleton 226 in a regio- and enantioselectively controlled manner (Scheme 30) [85]. The appropriate bis-THF ring system 227 was constructed using a Williamson type etherification reaction on a functionalized bis-mesylate intermediate. A Sonogashira cross-coupling reaction of the terminal acetylene 228 with vinyl iodide 229 followed by hydrogenation and thermal elimination finished the total synthesis of 225. ¹H NMR and ¹³C NMR data were found to be identical to the reported spectral data.

In 2001, Burke’s group reported the synthesis of a known intermediate 232 in the synthesis of uvaricin (225) (Scheme 31) [86]. A chiral DPPBA ligand controlled double cyclization of 230 allowed the selective formation of a single diastereomer 231 in one step, thus providing general access to annonaceous acetogenins containing trans/threo/trans or cis/threo/cis bis-THF core structures. Desymmetrization of diene 231 with AD-mix-β provided the known triol 232, which has served as an intermediate in a total synthesis of uvaricin (225).

Total synthesis of trilobin

Trilobin (233), which was isolated from the the bark of Asimina triloba by McLaughlin’s group [71] in 1992, has high potency against human lung cancer, breast cancer, and colon cancer cell lines (10⁶ to nearly 10¹⁰ times the cytotoxic potency of the reference compound, adriamycin) [71]. In 1999, Marshall’s group reported the total synthesis of trilobin (Scheme 32) [87]. Addition of the γ-alkoxy allylic indium reagent derived from the (R)-α-OMOM allylic stannane 234 and InCl₃ to aldehyde 218b afforded the anti-adduct 235 as a single diastereomer, which could be cyclized to the cis-threo-THF 236. Completion of the bis-THF core was effected by addition of the (S)-γ-OMOM allylic stannane 237 to aldehyde 236, affording the syn adduct 238. Treatment of this adduct with Bu₄NOH led to the bis-THF 239. Introduction of the butenolide moiety was achieved through condensation of ester 240 with a protected lactic aldehyde, which afforded product 233, identified as (+)-trilobin through comparison of the ¹H and ¹³C NMR spectra with those of the natural product.

At the same time, the group of Sinha and Keinan also reported the first total synthesis of 233 using the different “naked” carbon skeleton strategy [88], with all of the asymmetric centers in the bis-THF fragment 243 being produced by the Sharpless AD and AE reactions, starting with alcohol 242 (Scheme 33). Then the reaction of epoxide 244 with trimethylsilylethynyllithium and subsequently with the butenolide precursor 246 finished the total synthesis of trilobin. Synthetic trilobin (233) and its R- and S-Mosher’s esters showed ¹H NMR data identical to those of the naturally occurring trilobin and its R- and S-Mosher ester derivatives.

Total synthesis of asimilobin

Asimilobin (247) was isolated by McLaughlin’s group, both from the seeds of Asimina triloba [89] and from the bark of Goniothalamus giganteus (Annonaceae) [90], and showed cytotoxicity values comparable with adriamycin against six human solid-tumor cell lines [89,90]. In 1999, the group of Wang and Shi reported the first total synthesis of asimilobin (Scheme 34) [91]. Compound 248 was smoothly oxidized and cyclized to form a C₂-symmetrical bis-THF compound 249 using Co(modp)₂ as a catalyst under an oxygen atmosphere. Mono-protection of the diol 249 followed by Swern oxidation, and then reaction of the resulting aldehyde with CH₃(CH₂)₁₃MgCl gave the bis-THF segment 250. The coupling reaction between the aldehyde prepared from 250 and the ylide prepared from 47 gave the enyne 251, which was hydrogenated. Global deprotection allowed completion of the synthesis of 247a. The spectral data (¹H and ¹³C NMR, HRMS) of the synthetic compound 247a were consistent with those reported for the title compound in literature. However, the specific rotation was opposite to that reported. {Synthetic compound 247a: [α]_D²⁰ −11.4 (c 0.70, CHCl₃), [α]_D²⁶ −11.9 (c 0.43, CH₂Cl₂); Lit. [90] [α]_D +6.0 (c 0.05, CHCl₃); Lit. [91] [α]_D +11.3 (c 1.00, CH₂Cl₂)}. In order to clarify this problem, they immediately synthesized diastereomer 247b using the enantiomer of segment 250 made via the same procedures. They found that 247b had the same spectral data and close specific rotation as that reported in literature. {[α]_D²⁴ +6.4 (c 0.36, CHCl₃); [α]_D²⁵ +7.0 (c 0.10, CH₂Cl₂)}. Thus, this work strongly suggested that the natural product had the opposite absolute configuration on the bis-THF unit to that reported in the literature. In 2000, the full details of this total synthesis were reported [92].

Total synthesis of squamotacin

Squamotacin (252), which was isolated from the bark of Annona squamosa, showed cytotoxic selectivity for PC-3 (ED₅₀ = 1.72 × 10⁻⁹ µg/mL), with a potency of over 10⁸ times that of adriamycin (ED₅₀ = 3.42 × 10⁻¹ µg/mL) [93]. Its structure had been proposed on the basis of ¹H and ¹³C NMR, MS, and IR spectral data [93]. In 1999, the group of Sinha and Keinan reported the first total synthesis of (+)-squamotacin (252) [94] through a “naked” carbon skeleton strategy where all asymmetric centers in the bis-THF fragment of the molecules 256 were produced by the Sharpless AD and AE reactions (Scheme 35). Elongation of the carbon skeleton of 256 was achieved by a ring-opening reaction using 257 to afford alkyne 258. Then Wittig reaction of the corresponding Wittig reagent prepared from 258 with aldehyde 259 followed by catalytic hydrogenation and deprotection afforded 252. The synthetic compound 252 was found to be identical, by ¹H and ¹³C NMR, and MS, with the naturally occurring squamotacin.

Total synthesis of asimicin

Asimicin (260), which was isolated from the pawpaw tree, Asimina triloba [95], was synthesized by Marshall’s group in 1997 [96]. The approach employed the (R)-α-OSEM allylic stannane 261 reaction with the dialdehyde 262 obtained from (S,S)-diethyl tartrate to afford the bis-adduct 263 (Scheme 36). Treatment of 263 with TBAF led to the core bis-THF intermediate, diol 264. Mono tosylation and subsequent hydrogenolysis with LiBEt₃H gave alcohol 265. The iodide 266 was coupled with the higher-order vinylcyanocuprate to afford olefin 267, which could be converted to the epoxide 268. Addition of (R)-lithio-2-(OTBS)-3-butyne afforded the trifluoroacetate 269, then 269 was converted into the butenolide 270. Cleavage of the SEM protecting group afforded (+)-asimicin (260). The ¹H and ¹³C NMR spectra and optical rotation of the synthetic 260 were identical to those reported for (+)-asimicin, [α]_D 15.0 (c 0.2, CHCl₃), reported [α]_D 14.7 (c 0.3 CHCl₃) [95].

In 2000, the group of Sinha and Keinan reported the total synthesis of asimicin (260) [97] and demonstrated the advantages of three different strategies for the synthesis of the tricyclic intermediate 274 (Scheme 37), which represented the key fragment of the bis-THF ACGs. The naked carbon skeleton strategy was based on the production of all asymmetric centers by selective placement of the oxygen functions onto an unsaturated, non-functionalized carbon skeleton 271. Diversity in this approach arose from the relative timing of highly stereoselective reactions, such as the Sharpless AD reaction and the Kennedy oxidative cyclization (OC) with rhenium(VII) oxide. The convergent strategy, which was based on the combinatorial coupling of two series of diastereomeric fragments 275 and 276, to produce intermediate 277, enjoyed the advantages of both efficiency and versatility. The third approach, which was based on partially functionalized intermediates, such as 278, combined the advantages of both the linear and the convergent strategies, synthetic efficiency and diversity. The phosphonium salt 279, which was synthesized from 274, was reacted with aldehyde 280 in a Wittig reaction, which, after global deprotection, allowed completion of the total synthesis of asimicin (260). The spectral data (¹H and ¹³C NMR) of synthetic asimicin was identical to those of naturally occurring compounds.

In 2005, Roush’s group synthesized the bis-THF core of asimicin [98] from two sequential chelate-controlled [3+2] annulation reactions of allylsilanes and appropriately substituted aldehydes (Scheme 38). Subjecting the protected allylsilane 281 to the [3+2] annulation reaction with α-benzyloxyacetaldehyde (282) afforded the 2,5-trans-THF 283. Conversion of 283 to aldehyde 284 was achieved by reductive removal of the benzyl group and subsequent oxidation of the alcohol. Treatment of aldehyde 284 with allylsilane 285 mediated by SnCl₄ afforded the bis-THF 286. Finally, the butenolide ring was installed using a procedure developed by Marshall’s group [96] to provide synthetic (+)-asimicin.

In 2006, Marshall’s group reported the total synthesis of asimicin by a highly convergent route in which Grubbs cross-metathesis played a key role (Scheme 39) [99]. The bis-THF core unit 289 was constructed through a bidirectional outside-in hydroxy mesylate cascade cyclization route from 288. The bisbutenolide analogue 292 was prepared from diene 290 and the butenolide segment 291 through Grubbs cross-metathesis. Reaction of the bisbutenolide 292 with 1-decene catalyzed by Grubbs II catalyst led to the asimicin precursor 293, which was selectively hydrogenated, and subsequent global deprotection afforded asimicin (260). Analogues that differed in the length of the alkyl chain were also obtained in this way.

Total synthesis of 10-hydroxyasimicin

In 2005, Ley’s group reported the total synthesis of 10-hydroxyasimicin (294) (Scheme 40) [100]. Williamson cyclization of 295 led to the formation of the bis-THF core 296, which could be transformed into the fragment 297 in 9 steps. Sonogashira cross-coupling of vinyl iodide 298 with the propargylic alcohol 297 proceeded smoothly to produce the skeleton 299. The enyne functional group was reduced selectively and final global deprotection with BF₃·Et₂O in dimethyl sulfide afforded 294 as a colorless wax. The spectroscopic data for synthetic 294 (¹H NMR, ¹³C NMR, IR, MS, and specific rotation) were in excellent agreement with those reported for naturally occurring 10-hydroxyasimicin (294) [101].

Total synthesis of asimin

In 1994, McLaughlin’s group reported the isolation of asimin (300) from the stem bark of the North American paw-paw tree, Asimina triloba [78], depicted asimin as the C-10(S) isomer. However, in a subsequent paper the stereochemistry at C-10 was shown as R based upon chemical shift differences [81,102]. In view of the rather subtle basis for this assignment, Marshall’s group undertook a total synthesis of both C-10 epimers of asimin, reported in 1999 (Scheme 41) [103]. Their synthesis started with alcohol 301, which was converted into bistosylate 302 in 10 stpes, then the threo, trans, threo, trans, threo-bis-THF core unit 303 could be obtained from 302 upon stirring with TBAF in THF. The side chain of asimin in 307 was introduced through stereoselective addition of the organozinc reagent 305 to aldehyde 304. The ester 308 was condensed with the TBS ether of (S)-lactic aldehyde 309 to afford the γ-lactone adduct 310. Exposure of the alcohol 310 to trifluoroacetic anhydride and triethylamine led to the triol 300a. 10(S)-Asimin (300b) was prepared from aldehyde 304 by an identical sequence, using the enantiomer of 306 in the addition of the organozinc reagent. By comparing the MTPA ester of diastereomeric alcohols 300a and 300b with the authentic MTPA ester, the stereochemistry at C-10 of asimin (300) assigned as R.

Total synthesis of bullatacin

In 2000, the group of Sinha and Keinan reported the total synthesis of bullatacin (311) [97] and demonstrated the advantages of three different strategies for the synthesis of the tricyclic intermediates 274 using the same procedure as in the total synthesis of asimicin (Scheme 42).

In 2005, Roush’s group reported the total synthesis of (+)-bullatacin (311) via a diastereoselective [3+2] annulation reaction (Scheme 43) [104]. Racemic aldehyde 314, which was prepared from allylsilane (±)-312 and α-benzyloxy acetaldehyde (313), was treated with the highly enantiomerically enriched allylsilane 315 in the kinetic resolution manifold, providing the key bis-THF fragment 316 as a single diastereomer. Protodesilylation of the bis-THF 316 was accomplished by treatment with TBAF to give tetraol 317. The butenolide ring was then installed completing the total synthesis of (+)-bullatacin (311).

In 2006, Pagenkopf’s group reported the total synthesis of bullatacin (311) in an efficient route from commercial starting materials (Scheme 44) [105]. The bis(THF) core 319 was constructed from bis-epoxide 318 through double allylation and oxidative cyclization. Then titanium acetylide 320 was reacted with bis(THF) 319 to afford 321. Introduction of the unprotected butenolide (as 323) by epoxide opening with lithiated 322 followed by selective reduction and deprotection afforded bullatacin.

Total synthesis of rollidecins C and D

Rollidecin C (324) and rollidecin D (325) were discovered in the bioactive leaf extracts of Rollinia mucosa [106]. Both compounds 324 and 325 have exhibited cytotoxicity against six human tumor cell lines. Compound 324 was found to be uniformly more potent than 325 and showed selectivity toward HT-29 (ED₅₀ = 6.26 × 10⁻² µg/mL), exhibiting potency that approaches that of adriamycin (ED₅₀ = 2.81 × 10⁻² µg/mL for HT-29) [106]. In 2001, the group of Sinha and Keinan reported the total synthesis of rollidecins C and D (Scheme 45) [107]. Wittig reactions between the ylide derived from 326 and either of the two homologous butenolide aldehydes, 327 and 328, produced respectively 329 and 330, both in the form of a mixture of E and Z isomers. The MOM ether in 329 and 330 was selectively cleaved using TMSBr at low temperature, affording 331 and 332, respectively. The oxidative biscyclization reaction was carried out with either 331 or 332 using CF₃CO₂ReO₃ with trifluoroacetic anhydride (TFAA), affording the desired bis-THF products 333 or 334 respectively. Then 334 was converted to 325 by hydrogenation, and hydrogenation of compound 333 followed by desilylation afforded 324. Both synthetic compounds 324 and 325 were found to be identical by ¹H and ¹³C NMR with the naturally occurring rollidecins C and D, respectively.

Total synthesis of 30(S)-hydroxybullatacin

In 2003, Marshall’s group disclosed a modular synthetic approach to the adjacent bis-THF rings (Scheme 46) [108]. This approach featured highly selective additions of chiral R-oxygenated allylic stannane and indium reagents such as B¹ and D¹ (M = SnBu₃ or InBr₂) to an acylic core aldehyde precursor (A¹ then C¹) followed by core ring closure (E¹ → F¹) and ensuing Sonogashira coupling (F¹ + G¹ → H¹) to append the butenolide segment. This straightforward strategy permitted the efficient assembly of the acetogenin structure from four basic subunits. By interchanging these subunits a variety of natural acetogenins and their isomers should be accessible in relatively few steps. They extended the scope of their modular four-component synthesis of annonaceous acetogenins to 30(S)-hydroxybullatacin (335). The ¹H and ¹³C NMR spectra of the tetraol product 335 were in complete agreement with the reported spectra [109].

Total synthesis of uvarigrandin A

In 2003, Marshall’s group extended the scope of their modular four-component synthesis of ACGs to uvarigrandin A (338), and 5(R)-uvarigrandin A (339) (Scheme 47) [108].

Total synthesis of membranacin

Membranacin (343), a cytotoxic anti-tumor acetogenin isolated from the seeds of the fruit tree Rollinia mucosa [110,111] was synthesized by Brown’s group in 2004 (Scheme 48) [112]. The bis-THF precursor 346 was constructed from the lactone 344 using metal-oxo and metal-peroxy-mediated oxidative cyclisations as the key steps. The butenolide portion of membranacin (343) was introduced using Trost’s ruthenium-catalysed Alder-ene reaction, and afforded compound 343 whose spectroscopic data were consistent with those of membranacin [110].

In 2005, Lee’s group also reported the total synthesis of membranacin (343) (Scheme 49) [113]. Radical cyclization of 347 proceeded stereoselectively to give cis-2,5-disubstituted oxolane product 348, which was converted into (E)-β-alkoxyvinyl (S)-sulfoxide 349. Radical cyclization proceeded uneventfully to yield bis-oxolane 350 in high yield. Homoallylic alcohol 351 prepared from 350 could serve as a pivotal intermediate for the natural products via cross metathesis reaction of its terminal olefin. A cross olefin metathesis reaction of 351 and 352 followed by the established three-step sequence finished the total synthesis of membranacin (343).

Total synthesis of rolliniastatin 1, rollimembrin

Rolliniastatin 1 (353) and rollimembrin (354) are ACGs isolated from the seeds of Rollinia mucosa and Rollinia membranacea [9,110,114]. Lee’s group reported the first total synthesis of rollimembrin along with the total synthesis of rolliniastatin 1 in 2005 (Scheme 50) [113]. Homoallylic alcohol 351 could serve as a pivotal intermediate for the two natural products via cross metathesis reaction with their terminal olefin. A cross olefin metathesis reaction of 351 and 355 provided intermediate 357, which was converted into rolliniastatin 1 (353) via the three-step sequence. Rollimembrin (354) was synthesized in the same manner using 356 as the metathesis partner.

Total synthesis of longimicin D

Longimicin D (359), which is a structural isomer of asimicin, isolated by McLaughlin’s group from leaves and twigs of Asimina longifolia in 1996 [115], exhibits selective cytotoxic activities against A-549 (ED₅₀ = 4.93 × 10⁻⁴ µg/mL), PC-3 (ED₅₀ = 2.42 × 10⁻⁴ µg/mL), and PACA-2 (ED₅₀ = 1.69 × 10⁻⁷ µg/mL), with potency from 10³ to 10⁵ times that of adriamycin [115]. In 2006, the group of Maezaki and Tanaka reported the first total synthesis of longimicin D (Scheme 51) [116]. The bis-THF alcohol 360 was oxidized to give aldehyde 361. Introduction of the alkyne 362 into the bis-THF core 361 proceeded successfully giving the desired propargyl alcohol 363, which was converted into iodide 364. Alkylation of the γ-lactone 15 with the iodo-bis-THF core 364 afforded 365. The total synthesis of longimicin D (359) was accomplished from 365 by subsequent reactions – (1) oxidation of the sulfide, (2) thermolytic elimination of the sulfoxide, and (3) global deprotection with acidic MeOH – to give 359 in excellent yield. The spectroscopic and physical data of synthetic 359 (¹H NMR, ¹³C NMR, IR, MS) were in good agreement with those reported, while the specific rotation of synthetic 359 {[α]_D²⁵ = +23.2 (c 0.48 in EtOH)} was higher than that reported in the literature {[α]_D = +14 (c 0.1 in EtOH)}, so the structure of longimicin D needs to be further investigated.

Efforts toward the synthesis of mucoxin

Mucoxin (366), an ACG isolated from bioactive leaf extracts of Rollinia mucosa, was the first acetogenin containing a hydroxylated trisubstituted THF ring [117]. This natural product is a highly potent and specific antitumor agent against MCF-7 cell lines (ED₅₀ = 3.7 × 10⁻³ µg/mL compared to adriamycin, ED₅₀ = 1.0 × 10⁻² µg/mL) [117]. In 2006, Borhan’s group reported the total synthesis of the proposed structure of mucoxin via regio- and stereoselective THF ring-forming strategies (Scheme 52) [118]. The 2,3,5-trisubstituted THF portion (C13-C17) 368 was accessed using a highly regioselective cyclization of epoxydiol 367, and the 2,5-disubstituted THF ring (C8-C12) in 370 was conveniently assembled from 369 via a 1,2-n-triol cyclization strategy. The spectral data of the synthetic material did not match the reported data for the natural product. On the basis of detailed spectroscopic analysis of the synthesized molecule, they reasoned that the spectral discrepancies were due to stereochemical misassignment of the natural product. The structure of natural mucoxin need to be further revised through a different total synthesis.

Modular synthesis of adjacent bis-THF annonaceous acetogenins

In 2003, Marshall’s group reported a synthesis of four adjacent bis-THF ACGs, asiminocin, asimicin, asimin, and bullanin, by a modular approach from seven fundamental subunits, A–G (Scheme 53) [119]. The approach employed a central core aldehyde segment, C, to which were appended an aliphatic terminus, A or B, a spacer subunit, D or E, and a butenolide terminus, F or G. Coupling of the A, B, D, and E segments to the core aldehyde unit was effected by highly diastereoselective additions of enantiopure allylic indium or tin reagents. The butenolide termini were attached to the ACD, BCE, or BCD intermediates by means of a Sonogashira coupling. The design of the core, spacer, and termini subunits was such that any of the C30, C10, or C4 natural acetogenins or their stereoisomers could be prepared.

3 Nonadjacent bis-THF

Total synthesis of 4-deoxygigantecin

4-Deoxygigantecin (371) was isolated from the bark of Goniothalamus giganteus by McLaughlin’s group [120]. The absolute stereochemistry of natural 4-deoxygigantecin had not yet been determined, however, it was assumed that 371 possessed, except for the C-4 carbinol center, the same absolute configuration as that of gigantecin (372), whose absolute stereostructure had been established by an X-ray crystallographic analysis [121]. In 1997, Tanaka’s group reported the total synthesis of natural (+)-4-deoxygigantecin (371) (Scheme 54) [122], which was the first example of the synthesis of a non-adjacent bis-THF type ACG. Starting with (−)-muricatacin (373), benzoate 374 was obtained, which was transformed into 375 in 11 steps. Mesylate formation from 375 followed by the Sharpless AD using AD-mix-α, and subsequent cyclization with Triton B furnished the key bis-THF ring-containing synthon 376. A Pd⁰-catalyzed cross coupling reaction of compound 376 with vinyl iodide 377 gave 378. Finally, catalytic hydrogenation of 378 and subsequent deprotection of the MOM group finished the total synthesis of (+)-4-deoxygigantecin (371). Its ¹H-NMR data were in good agreement with those recorded for natural 371 and the optical rotation value {[α]_D²³ +16.0 (c 0.05, MeOH)} of the synthetic sample was also consistent with that of natural 371 {[α]_D +15.5 (c 0.2, MeOH)}. In 1998, they reported the full details of this total synthesis [123].

Total synthesis of squamostatin D

In 1994, Fujimoto’s group [124] described the isolation and structure elucidation of five nonadjacent bis-THF ACGs, squamostatins A–E. In 1998, Marshall’s group reported the total synthesis of squamostatin D (379) (Scheme 55) [125]. The tosylate 381 was converted to the eventual threo,trans,threo C16-C34 segment 382 of squamostatin-D upon treatment with TBAF in THF. Introduction of the C12 stereocenter along with the C1-C11 chain 384 of squamostatin D was conveniently achieved through addition of the zinc reagent 305 to the aldehyde 383. The derived tosylate 385 cyclized upon treatment with TBAF to afford the bis-THF ester 386. At last, the butenolide segment of squamostatin D was introduced by a modification of the method of Wu’s group [23], affording squamostatin D (379), [α]_D +8.4 (lit. [124] +7.9), mp 112–113 °C (lit. [124] mp 112–113.5 °C), The ¹H and ¹³C NMR spectra were identical to the spectra of the natural product as was the ¹H spectrum of the (R)-Mosher ester derivative 380.

Total synthesis of gigantecin

Gigantecin (388), a representative nonadjacent bis-THF acetogenin, was isolated from the bark of Goniothalamus giganteus in south east Asia [126] and the seed of the Brazilian plant Annona coriacea [121]. The relative and absolute configurations of gigantecin were assigned after extensive spectroscopic and Mosher ester analysis, and the assignment was confirmed by single crystal X-ray analysis. Gigantecin displayed potent cytotoxicity against A-549, HT-29, MCF-7, and glioblastoma multiforme (U251MG) human tumor cell lines at ED₅₀s of 0.4, 0.001, 4.3, and 0.003 µg/mL, respectively [121,126]. In 2004, Crimmins’s group reported first total synthesis of (+)-gigantecin exploiting a modified asymmetric aldol protocol using chlorotitanium enolates of oxazolidinone glycolates (Scheme 56) [127]. The diene 389 was subjected to the Grubbs catalyst resulting in formation of the dihydrofuran 390, which was then converted to the aldehyde 391. Addition of acetylene 392 to aldehyde 391 produced the propargylic alcohol 393. The final C-C bond was fashioned by palladium-mediated coupling of the acetylene 394 with vinyl iodide 395 to provide enyne 396. Selective hydrogenation followed by removal of the protecting groups led to the completion of the synthesis of (+)-gigantecin (388). Synthetic gigantecin was identical (¹H, ¹³C NMR, [α]_D²⁴) to the natural material.

In 2006, Hoye’s group described an efficient, highly convergent chemical synthesis of (+)-gigantecin (388) utilizing a one-pot, three component olefin metathesis coupling strategy (Scheme 57) [128]. Mixed silaketal 399 was prepared by sequential loading of 397 and then 398 onto Ph₂SiCl₂, then triene 399 and alkene 400 were combined and exposed to Grubbs II catalyst [Ru=CHPh(Cl)₂ (PCy₃)(DHIMes)] to induce a ring-closing/cross-olefin methathesis sequence which afforded the major product 401. Diimide reduction and global deprotection gave 388.

Total synthesis of cis-sylvaticin

cis-Sylvaticin (402), isolated in 1995 from the leaf extracts of Rollinia mucosa, was an interesting natural product with nonadjacent THF rings [129]. cis-Sylvaticin displays potent activity as an antitumor agent and exhibits nanomolar cytotoxicity toward human solid tumor cell lines [129]. In 2006, Donohoe’s group first reported the total synthesis of cis-sylvaticin (402) (Scheme 58) [130]. Oxidation of commercial tetradecatetraene 403 under AD conditions gave a tetraol which was immediately converted into bisacetonide 404. The key double oxidative cyclization was then applied on bisacetonide 405 to afford the bis-THF 406. Subsequently, the bis-THF 407 was elaborated from 406. Then construction of 409 was accomplished by a cross metathesis reaction between 407 and 408. The synthesis of cis-sylvaticin (402) was completed by a selective diimide reduction and acid promoted deprotection of the three OTBS groups. The synthetic material had spectroscopic data (¹H and ¹³C NMR, [α]_D, HRMS) identical to that reported.

4 Three adjacent THF rings

Total synthesis of goniocin

Goniocin (410), which was isolated from Goniothalamus giganteus [131], possess three adjacent THF rings and, therefore, represented the first example of a new subclass of ACGs. Structure 410 was proposed for goniocin on the basis of its MS and ¹H and ¹³C NMR data. In 1997, Sinha’s group reported that all trans-4,8,12-trienol substrates indeed underwent a highly stereospecific triple oxidative cyclization reaction in the presence of a rhenium(VII) reagent to produce a single stereoisomer of a tris-THF product (Scheme 59) [132]. Surprisingly, however, the product’s stereochemistry was not trans-threo-trans-threo-trans-threo as expected, but trans-threo-cis-threo-cis-threo. Consequently, they synthesized 17(S),18(S)-goniocin (411) rather than 410. The key intermediate in their synthesis was the “naked” carbon skeleton 413 which was easily prepared from 412. When trienol 413 was treated with a mixture of CF₃CO₂ReO₃ and trifluoroacetic anhydride, a stereochemically pure tris-THF product 414 was obtained, which was then converted to the phosphonium salt 415. Wittig reaction of 415 with aldehyde 416 afforded alkene 417. Finally, hydrogenation and removal of the protecting groups afforded 411.

In 1998, the group of Sinha and Keinan reported the first asymmetric total syntheses of goniocin (410), and cyclogoniodenin T (418) (Scheme 60) [133]. Oxidative cyclization of 419 with CF₃CO₂ReO₃ and lutidine produced the trans-THF derivative 420. Asymmetric dihydroxylation of 420 using AD-mix-α followed by double mesylation produced 421. Acidic cleavage of the acetonide and the silyl ethers followed by heating of the resultant tetraol in pyridine produced the desired all-trans tris-THF diol 422. The Wittig reagent 423 was reacted with aldehyde 424 to produce alkene 425. Finally, catalytic hydrogenation and deprotection of both MOM and BPS groups afforded goniocin (410). Cyclogoniodenin T (418) was prepared from the ent-419 using the same procedure. The absolute stereochemistry of their synthetic 410 and 418 was proved by comparison of the ¹H NMR spectra of their (R) and (S) bis Mosher esters with the original spectra of the esters of the naturally occurring 410 and the semisynthetic 418.

5 Adjacent THF-THP rings

Total synthesis of jimenezin

In 1998, Mata’s group isolated a new ACG from the seeds of Rollinia mucosa (Annonaceae) and named it jimenezin (426a) [134]. This natural product was quite active in the BST assay (IC₅₀ 5.7×10⁻³ µg/mL) [135], and exhibited potent cytotoxic activity against six human solid tumor cell lines. In 1999, Takahashi’s group reported the first total synthesis of jimenezin that dictated revision of the formula to 426b (Scheme 61) [136]. The coupling reaction between 427 and 428 afforded a 92:8 mixture of the desired carbinol 429a and its diastereomer 429b. Hydrogenation of the mixture using PtO₂ gave the desired 19β-alcohol 430a along with its 19-epimer 430b. Dess-Martin oxidation of the mixture (430a and 430b), and subsequent reduction with L-Selectride could transform the 430a into 430b. The 19β-alcohol 430a was then transformed to the central core 431a in 10 steps. Finally, the complete carbon skeleton of 426a was assembled by joining 431a and 432 under Hoye’s conditions. The spectroscopic and physical properties of the synthetic material 426a were found to differ from those of natural jimenezin, so the 19α-alcohol 430b was transformed into the terminal acetylene derivative 431b, which was coupled with 432 affording 426b, whose physical and spectral data ([α]_D²⁰, ¹H and ¹³C NMR) were identical with those of the natural jimenezin.

In 2005, Lee’s group reported the total synthesis of jimenezin (426b) via radical cyclization of β-alkoxyacrylate and β-alkoxyvinyl sulfoxide intermediates and intramolecular olefin metathesis reaction (Scheme 62) [137]. Hydroxy oxane 434 was prepared from a β-alkoxyacrylate aldehyde precursor 433 via samarium(II) iodide-mediated cyclization. Oxolane derivative 436 was obtained via radical cyclization of β-alkoxyvinyl sulfoxide 435. The homoallylic alcohol prepared from aldehyde 437 was converted into carboxylate ester 438, which could serve as a precursor for macrolactone 439 via ring-closing olefin metathesis. Incorporation of an (S)-propylene oxide unit into 439 and further manipulations generated jimenezin (426b).

In 2006, Hoffmann’s group reported the total synthesis of jimenezin (426b) [138] through a highly stereoselective intramolecular allylboration to establish the tetrahydropyran ring and an intramolecular Williamson reaction to close the THF ring (Scheme 63). Treatment of the E-allyl boronate 440 with LiBF₄ or Yb(OTf)₃ in acetonitrile with 2% water led to the desired allylboration product 442, which was transformed into compound 443. An iodine-lithium exchange reaction of 443 gave the corresponding organolithium compound, which added to the aldehyde 444 afforded the desired stereoisomer 445. Tosylation of the secondary hydroxy group in 445 followed by hydrogenation and refluxing in pyridine produced compound 446. The alcohol 446 was converted into the 1-phenyl-1H-tetrazol-5-yl (PT) sulfone 447. A Julia-Kocienski olefination of the sulfone 447 with the aldehyde 448 gave the alkene 449. Chemoselective hydrogenation of the double bond in 449 followed by cleavage of all three silyl ethers provided (−)-jimenezin (426b), which was found to be identical to the natural product with respect to the spectroscopic data [134].

Total synthesis of muconin

Muconin (450), which was isolated from the leaves of Rollinia mucosa by McLaughlin’s group in 1996 [117], has exhibited potent and selective in vitro cytotoxicity against PACA-2 (ED₅₀ = 5.4 × 10⁻⁴ µg/mL) and MCF-7 (ED₅₀ = 2.4 × 10⁻⁴ µg/mL) in a panel of six human solid tumor cell lines [117]. In 1998, Jacobsen’s group reported the total synthesis of muconin through a chiral building block approach (Scheme 64) [139]. The hydrolytic kinetic resolution (HKR) of (±)-tetradeceneoxide using complex (S,S)-454 afforded (R)-tetradecane-1,2-diol 451, which could be converted to acid 452. Pyranol 453 was constructed by the hetero-Diels-Alder condensation of 1-methoxy-3-[(trimethylsilyl)oxy]-1,3-butadiene with p-bromobenzyloxyacetaldehyde catalyzed by (S,S)-455. Esterification of 453 with acid 452 followed by ring-closing metathesis and further elaboration afforded 456. Coupling of 457 with aldehyde 456 followed by elimination and deprotection finished the total synthesis of 450 which exhibited spectral properties identical to those of the natural product.

In 1999, Kitahara’s group also reported the total synthesis of (+)-mucocin (450) (Scheme 65) [140]. The epoxide 458, which was constructed from the D-glutamic acid, could be transformed into 459. Palladium(0)-mediated coupling of the alkyne 459 with the iodoalkyne 460 followed by hydrogenation afforded 461 and 462, and the undesired β-alcohol 462 was inverted to α-alcohol 461 by means of a Dess-Martin oxidation/LiAl(Ot-Bu)₃H reduction sequence. Finally, global deprotection provided (+)-muconin (450) with spectral properties identical to those of the natural product [117]. In 2000, they reported the full details of this total synthesis [141].

In 2002, Takahashi’s group reported the total synthesis of muconin (450) through a coupling reaction of a THF–THP segment and a terminal butenolide (Scheme 66) [142]. The cyclic ether 464, which was obtained by heating 463 with sodium methoxide in methanol, could be transformed into a terminal acetylene 465. Then the complete carbon skeleton of 450 was assembled by joining 465 and 466 under Hoye’s conditions to give enyne 467, which underwent regioselective reduction and deprotection to give muconin (450). The spectroscopic and physical properties of 450 were identical those of natural 450 [117].

In 2004, the group of Yoshimitsu and Nagaoka reported the total synthesis of (+)-muconin (450) starting from (−)-muricatacin (373) (Scheme 67) [143]. (−)-Muricatacin (373) was converted to δ-lactone 468. Reduction of 468 with diisobutylaluminum hydride provided lactol 469, the sodium alkoxide derivative of which subsequently underwent Wittig olefination with phosphonium compound 470 to give olefin 471. 471 was oxidized with mCPBA to provide an epoxide whose opening with CSA gave tetrahydropyran 472. The triflate 473 was reacted with the lithium enolate generated from the known α-thiophenyl γ-lactone 15 to provide lactone 474; subsequent elimination and deprotection finished the total synthesis of (+)-muconin (450), whose spectroscopic and analytical data were consistent with those of the natural product.

6 Nonadjacent THF-THP rings

Total synthesis of mucocin

Mucocin (475), which was isolated from the leaves of Rollinia mucosa, was the first ACG reported that bears a hydroxylated THP ring along with a THF ring [144]. Mucocin was found to be quite active in the BST assay (IC₅₀ 1.3 µg/mL) and showed selective inhibitory effect against A-549 (ED₅₀ = 1.0 × 10⁻⁶ µg/mL) and PACA-2 (ED₅₀ = 4.7 × 10⁻⁷ µg/mL) in a panel of six human solid tumor cell lines [144]. Its selective potency was up to 10,000 times that of adriamycin. Interestingly, mucocin was found to be as active as bullatacin in inhibition of oxygen uptake by rat liver mitochondria (LC₅₀ 18 and 9 nM/mg protein, respectively). In 1998, the group of Sinha and Keinan reported the first total synthesis of mucocin via the “naked” carbon skeleton strategy (Scheme 68) [145]. All eight asymmetric centers in the key fragment 479 of the molecule were introduced by double AE reaction of 476 followed by double AD reaction of 478. Treatment of 479 with a catalytic amount of TsOH induced the double ring closure to afford the nonadjacent THP-THF ring system 480, which was then transformed into the alkyne 481. Cross-coupling with Pd(PPh₃)₂Cl₂ catalyst of 481 and 482 afforded enyne 483. Homogeneous catalytic hydrogenation and acid-catalyzed deprotection of all four protecting groups in 483 afforded 475, which was found to be identical (MS, ¹H and ¹³C NMR, [α]_D) with the naturally occurring mucocin.

In 1999, Takahashi’s group also reported the total synthesis of mucocin (475) (Scheme 69) [146]. Reaction of aldehyde 484 with the lithiated alkyne 485 produced the alcohol 486 [147]. Then the complete carbon skeleton of 475 was assembled by joining 487 and 432 under Hoye’s conditions to give the labile enyne 488, which underwent regioselective reduction followed by deprotection thus completing the total synthesis of 475, whose spectral properties were indistinguishable from those of the natural product [144]. In 2002, the full procedure for this total synthesis was reported [148].

In 1999, Koert’s group reported the total synthesis of (−)-mucocin (475) by using a stereocontrolled coupling reaction (Scheme 70) [65]. An acid-catalyzed intramolecular 6-endo attack on the alkenyl epoxide of the acetonide in 489 afforded the THP ring 490, which was transformed into the iodide 491 in 4 steps. Finally, addition of an organometallic reagent derived from 491 to the THF aldehyde 492 followed by deprotection provided (−)-mucocin (475) ([α]_D = –12.7, c 0.27 in CH₂Cl₂), which was found to be identical to the naturally occurring product in respect to the spectroscopical data. In 2000, the full details of this total synthesis were reported [149].

In 2002, the group of Takahashi and Nakata reported the total synthesis of 475 based on the SmI₂-induced reductive cyclization as a key step (Scheme 71) [150]. The THP ring in the central core 494 was constructed from 493 by the SmI₂-induced reductive cyclization, whereas the trans-THF ring was synthesized by oxidative cyclization of a homoallylic alcohol. The γ-lactone 466 was synthesized by aldol condensation of chiral ester 495 and aldehyde 496a. Finally, a Pd-catalyzed cross-coupling reaction of the THP/THF segment 494 and vinyl iodide 466 followed by hydrogenation and global deprotection finished the total synthesis of 475.

In 2003, Evans’s group reported the total synthesis of (−)-mucocin (475) by using a temporary silicon-tethered (TST) RCM homo-coupling reaction (Scheme 72) [151]. The enantioselective addition of the alkynyl zinc reagent derived from 497 to the aldehyde 498 furnished the propargylic alcohol. Protection of the alcohol as the triisopropylsilyl ether followed by deprotection of the p-methoxyphenyl ether afforded the allylic alcohol 499. The TST-RCM cross-coupling reaction between 499 and 500 furnished 501 and completed the construction of the carbon skeleton of mucocin (475).

In 2005, Mootoo’s group reported the total synthesis of mucocin (475) in a three component modular approach based on olefinic coupling reactions (Scheme 73) [152]. They used a cross-metathesis on tetrahydropyran 502 and THF 503 to assemble a stereochemically complex bicyclic ether 504, which was further elaborated to sulfone 505. Then 505 was reacted with butenolide aldehyde component 416 in a Julia–Kocienski olefination to provide the mucocin framework 506, which was converted to the natural product 475 after alcohol deprotection.

In 2006, Crimmins’s group reported the enantioselective total synthesis of (−)-mucocin (475) (Scheme 74) [153]. Both fragments 508 and 510 were prepared via an asymmetric glycolate aldol-RCM sequence. Then 508 and 510 were coupled through a cross-metathesis reaction to afford bicyclic ether 511. The coupling of advanced acetylene 511 and known butenolide 512 finished the total synthesis of (−)-mucocin (475).

7 mono-THP ACGs

Total synthesis of Pyranicin

Pyranicin (513), which was isolated from the stem bark of Goniothalamus giganteus, was the first mono-THP acetogenin isolated [154]. The acetogenin 513 was quite active in the BST assay (LC₅₀ = 0.3 µg/mL) [155] and showed selective inhibitory effects against PACA-2 cell lines (ED₅₀ = 1.3 × 10⁻³ µg/mL) with potency 10 times that of adriamycin (ED₅₀ = 1.6 × 10⁻² µg/mL) [154]. In 2003, the group of Takahashi and Nakata reported the first total synthesis of 513 in a stereocontrolled manner (Scheme 75) [156]. SmI₂-induced reductive cyclization of 516 afforded a 16,20-cis-19,20-anti-THP derivative 517. Through utilization of Mitsunobu lactonization, stereoinversion at the C-19 position was achieved affording 518, which was transformed into the phosphonium salt 519 through DIBAL reduction and Wittig reaction. Construction of the complete carbon skeleton of 513 was achieved through a Wittig reaction, then global deprotection of 521 produced pyranicin (513). The synthetic 513 showed [α]_D²⁴ +19.5 (c 0.55, CHCl₃), while the [α]_D²³ value of natural 513 was reported to be –9.7 (c 0.008, CHCl₃). However the NMR data of the corresponding MTPA esters (514 and 515) were revealed to be well matched with those reported. As the optical rotation of the natural product was measured at very low concentration, the difference might be due to experimental error or the presence of impurities.

In 2005, Rein’s group reported a convergent total synthesis of 513 employing asymmetric Horner-Wadsworth-Emmons (HWE) reactions (Scheme 76) [157]. Their synthesis began with the desymmetrization of meso-dialdehyde 522 through an asymmetric HWE olefination which gave the secondary alcohol 524. A Mitsunobu reaction followed by basic hydrolysis of the resulting chloroacetate then gave the inverted product 525. In the subsequent hetero-Michael cyclization, the desired cis-cis-THP 526 was formed, which was then transformed into the desired vinyl iodide 527. Lactonization of 528 under acidic conditions gave the desired lactone 529 as a diastereomeric mixture, which was then treated with base to afford the propargylic alcohol 530. The complete pyranicin framework was assembled through a Sonogashira coupling of 527 and 530, giving ene-yne 531. Finally, a selective diimide reduction followed by global deprotection using HF in MeCN afforded pyranicin (513). In 2006, they reported the full details of this total synthesis [158].

Total synthesis of Pyragonicin

Pyragonicin (532), which was isolated from the stem bark of Goniothalamus giganteus [154], was active in the BST assay (LC₅₀ = 0.9 µg/mL) [155] and showed a selective inhibitory effect against PACA-2 (ED₅₀ = 5.8 × 10⁻² µg/mL) [154]. In 2005, the group of Takahashi and Nakata reported the first total synthesis of the proposed structure of pyragonicin 532 (Scheme 77) [159]. SmI₂-induced reductive cyclization of 533 gave THP ester 534. Stereoinversion at the C-17 position was achieved using a Mitsunobu lactonization of 534, subsequent DIBAL reduction and Wittig reaction afforded olefin 536, which was transformed into the phosphonium salt 537. Then Wittig reaction of aldehyde 416 completed the construction of pyragonicin (532). Compound 532 had spectroscopic data consistent with that of natural pyragonicin, but a different optical rotation.

In 2005, Rein’s group also reported the total synthesis of pyragonicin (532) using the asymmetric Horner-Wadsworth-Emmons (HWE) methodology (Scheme 78) [160]. The THP-fragment 540, which in turn was accessed from meso-dialdehyde 538 via an asymmetric HWE desymmetrization, coupled with 542, which was also constructed from rac-541 by a parallel kinetic HWE resolution, completed the total synthesis of pyragonicin (532). The spectroscopic data of 532 (IR, ¹H and ¹³C NMR) were, within the normal error limits for such data, identical to those reported by McLaughlin. However, there was a strong discrepancy in the optical rotation. In 2006, the full details of this total synthesis were reported [158].

In 2006, Takahashi’s group described a second-generation synthesis of pyragonicin (532) (Scheme 79) [161]. The key step involved an olefin cross-metathesis between the THP segment 546 and the terminal γ-lactone residue 548 in the presence of Grubbs’ first-generation catalyst 549 affording the desired coupling product 550 exclusively. The olefin 550 underwent hydrogenation followed by syn-elimination of the sulfoxide and global deprotection to finish the total synthesis of pyragonicin (532).

8 Only γ-lactone ACGs

Total synthesis of squamostanal A

Squamostanal A (551) was isolated from the seeds of Annona squamosa and characterized by usual spectroscopic methods (NMR, mass spectrometry, and circular dichroism) as (5S)-3-(12-formyldodecyl)-5-methyl-2,5-dihydrofuran-2-one [162]. In 1996, Figadère’s group reported the total synthesis of squamostanal A (551) in only 4 steps (Scheme 80) [163]. 552 was enolized and added to (S)-propylene oxide to afford the macrolactone 553 and butyrolactone 554. After separation, 554 was first oxidized with H₂O₂ to afford the corresponding butenolide, and then treated with PDC to afford the desired product 551, whose spectroscopic data were in accord with those of natural squamostanal A.

Total synthesis of diepomuricanin

Diepomuricanin (555), which was isolated from Annona muricata by Cavé’s group [164], was assumed to be a biosynthetic intermediate for tetrahydrofuranic annonaceous acetogenins. In 1996, Tanaka’s group reported the total synthesis of (15S,16R,19S,20R,34S)-diepomuricanin (555) (Scheme 81) [165]. A Pd-mediated cross-coupling reaction between 556 and 229 yielded enyne 557, then catalytic hydrogenation followed by treatment with MsCl/Et₃N, dil. HCl/MeOH and KOH/THF afforded 558. Oxidation to the sulfoxide with m-CPBA/NaHCO₃ and subsequent thermal elimination by refluxing in toluene led to (15S,16R,19S,20R,34S)-diepomuricanin (555). By comparing the IR, ¹H and ¹³C NMR data and the optical rotation values (synthetic [α]_D +17.0; natural [α]_D +13.5), the absolute configuration of diepomuricanin was likely to be 15S,16R,19S,20R,34S.

Total synthesis of muricatacin

Muricatacin, an acetogenin derivative that showed cytotoxic activity against certain human tumor cell lines, had been isolated from the seeds of Annona muricata [166], and remarkably, the natural compound comprises both (−)-muricatacin [(R,R)-373a] and its enantiomer (+)-muricatacin [(S,S)-373b]. In 1997, the group of Rassu and Casiraghi reported the synthesis of both enantiomers of muricatacin, (R,R)-373a and (S,S)-373b (Scheme 82) [167]. (+)-(R)-glyceraldehyde acetonide (R-560) coupled with (tert-butyldimethylsilyl)-2-hydroxyfuran (TBSOF) afforded the 4,5-syn-configured adduct 560. Compound 561 was subjected to sequential catalytic hydrogenation and protection of the OH function afforded the seven-carbon intermediate 562. The oxidative removal of the C-7 carbon atom generated the six-carbon aldehyde 563. Wittig olefination of aldehyde 563 with the appropriate C₁₁ ylide followed by catalytic hydrogenation and BF₃ etherate-promoted desilylation afforded (−)-muricatacin [(R,R)-373a]. Its enantiomer (+)-muricatacin [(S,S)-373b] was synthesized from (+)-(S)-glyceraldehyde acetonide (S-560) using the same procedure.

In 1997, Scharf’s group reported the total synthesis of both enantiomers of epi-muricatacin (+)-(S,R)-373c and (−)-(R,S)-373d by means of a change in the sequence of side-chain introduction from the same chiral precursor 564 (Scheme 83) [168].

In 1998, Uang’s group reported the synthesis of (−)-muricatacin [(R,R)-373a] and 5-epi-(−)-muricatacin [(R,S)-373d] from thioglycolic acid employing (1R)-(+)-camphor as the chiral auxiliary (Scheme 84) [169]. Oxidation of 569 with OsO₄ afforded 373a, while m-CPBA promoted oxidation of 569 afforded 373d.

In 1998, Yoon’s group reported the synthesis of four stereoisomers of muricatacin 373a–d through the reaction of corresponding aldehydes 570a–d [170], prepared from D-glucose, with the anion of triethylphosphonoacetate followed by reduction and cyclization under acidic conditions (Scheme 85).

In 1998, Figadère’s group reported the synthesis of muricatacin (373b) through addition of TBSOF to an achiral aldehyde promoted by Ti(OiPr)₄ in the presence of (R)-1,1′-bi-2-naphthol (BINOL) followed by hydrogenation (Scheme 86) [171]. It is worth noting that the major threo product was obtained with 90% ee through this titanium-mediated addition of TBSOF to tridecanal in (R)-BINOL at −20 °C in Et₂O.

In 1999, Couladouros’s group reported the total synthesis of (−)-muricatacin (373a) and (+)-epi-muricatacin (373c) from the same γ-lactone 572 (Scheme 87) [172].

In 1999, Trost’s group reported the total synthesis of muricatacin (373a) through ruthenium-catalyzed cycloisomerization-oxidation on 577, which was synthesized from enyne 576 via asymmetric dihydroxylation (Scheme 88) [173].

In 2000, the group of Heck and Mioskowski reported the total synthesis of (−)-(4R,5R)-muricatacin (373a) using as a key step a regio- and stereospecific ring-opening of a substituted vinyl epoxide 578 under Lewis acid catalysis (Scheme 89) [174].

In 2002, the group of Carda and Marco reported the stereoselective synthesis of muricatacin (−)-373a through a ring-closing metathesis (Scheme 90) [175]. The acrylate 581, which was synthesized from (R)-2-benzyloxytetradecanal 580, underwent the RCM reaction, thus furnishing lactone 582. Hydrogenation of 582 finished the total synthesis of muricatacin (−)-373a.

In 2003, Popsavin’s group reported a novel general approach using an enantiodivergent synthesis of 373a and 373b from D-xylose (Scheme 91) [176]. The lactol 585, which was obtained from 583, was transformed into the formate 587 through oxidation. 587 was treated with aqueous trifluoroacetic acid to yield the α-lactone 589. 5-O-Benzoyl-1,2-O-cyclohexylidene-α-D-xylofuranose 584 was transformed into the corresponding saturated ester 586 through a Wittig olefination followed by catalytic hydrogenation, and 586 was treated with sodium methoxide in methanol to furnish the hydroxylactone 588, whose oxidation afforded aldehydo-lactone ent-589. The chiral synthons 589 and ent-589 were converted into the targets 373a and 373b through a known two-step sequence [177].

In 2003, the group of Bernard and Piras reported the total synthesis of (−)-(4R,5R)-muricatacin (373a) from cyclobutanone 591, which was obtained by lithium salt catalyzed ring expansion of the optically pure oxaspiropentane 590. (R,S)-591 was transformed into the corresponding γ-lactone (R,R)-592 through a Baeyer-Villiger oxidation. Synthesis of the γ-lactone (R,R)-593 constituted a formal synthesis of (−)-muricatacin (373a) (Scheme 92) [178].

In 2003, the group of Yoshimitsu and Nagaoka reported the total synthesis of (−)-muricatacin (373a) (Scheme 93) [179] through α-C-H hydroxyalkylation of THF with tridecanal using triethylborane-TBHP, which provided alcohols 594. Then α′-C-H oxidation of THF (+)-595 with ruthenium tetroxide under modified Sharpless conditions followed by deprotection finished the total synthesis of (−)-muricatacin (373a). This study presented a novel method for C-H bond functionalization as a means for preparing γ-(hydroxyalkyl)-γ-butyrolactones.

In 2004, Quinn’s group reported the total synthesis of (−)-muricatacin (373a) (Scheme 94) [180] by using the highly regioselective and stereoselective tandem ring-closing/cross metathesis reaction of 596 to construct the lactone and the alkyl chain in 597. Then (−)-muricatacin (373a) was obtained by catalytic hydrogenation/hydrogenolysis of 597.

Total synthesis of montecristin

(+)-Montecristin (598) isolated in 1997 from the roots of Annona muricata [181] might be an intermediate between the less and the more oxygenated acetogenins. In 2001, Brückner’s group reported the total synthesis of both 598a and 598b (Scheme 95) [182], and by comparing their specific rotations with those of montecristin, demonstrated that 598a was ent-5-epi-montecristin while 598b was the enantiomer of (+)-montecristin. Alkylating the dilithiated hydroxylactone S,S-599 with iodide 600 delivered trans-alkylated hydroxylactone 601a. The ensuing β-elimination of 601a followed by acetonide cleavage finished the total synthesis of 598a, while 598b was prepared from R,R-599 using the same procedure.

Total synthesis of acaterin

Acaterin (602a), which was isolated from a culture broth of Pseudomonas sp. A92 by Endo’s group in 1992 [183], is an inhibitor of acyl-CoA:cholesterol acyltransferase (ACAT) [183]. In 2002, the group of Franck and Figadère reported the synthesis of (−)-acaterin (602a) through the first application of the Baylis–Hillman reaction to α,β-unsaturated lactone (S)-603 (Scheme 96) [184].

In 2002, Singh’s group reported a short and efficient synthesis of acaterin from 604 (Scheme 97) [185], which was constructed from caprylic aldehyde and methyl acrylate through a Baylis–Hillman reaction. Ring closing metathesis reaction on 605 using Grubbs’ catalyst followed by deprotection afforded natural (−)-acaterin (602a) and its diastereomer (602b).

In 2003, Kumar’s group reported the total synthesis of (−)-acaterin (602a) (Scheme 98) [186]. Starting from octan-1-ol, the phosphonium salt 608 was obtained by employing the Sharpless AD procedure and a Wittig olefination. Then the coupling of phosphonium salt 608 with aldehyde 496b and subsequent cyclization afforded 602a.

Total synthesis of rollicosin

Rollicosin (610a), isolated in low yield from Rollinia mucosa in 2003, was a new subclass of acetogenins containing two terminal γ-lactones [187]. Quinn’s group reported the first total synthesis of rollicosin in 2005 using a tandem RCM/CM strategy for allyl butenolide preparation (Scheme 99) [188]. Butenolide 613 was produced by tandem RCM/CM with initial RCM of acrylate 612 preceding CM with the benzyl ether of 10-undecen-1-ol (611). 613 was exposed to H₂ in the presence of Pd/C to effect removal of the benzyl ether and concomitant alkene reduction to provide alcohol 614, TPAP oxidation to the corresponding aldehyde and one-carbon Wittig homologation then gave terminal alkene 615. Treatment of 615 with AD-mix-β provided diol 616, which after suitable protection was coupled with the enolate of 15 to produce 617. Oxidation of sulfide 617 and thermal elimination followed by TBS deprotection provided rollicosin (610a), which displayed spectral data (IR, ¹H and ¹³C NMR) and optical rotation consistent with that of naturally occurring rollicosin.

In 2005, Makabe’s group reported the total synthesis of (4R,15R,16R,21S)-rollicosin (610a) and (4R,15S,16S,21S)-rollicosin (610b) (Scheme 100) [189]. Sharpless AD using AD-mix-β on 618 furnished lactone 619. The hydroxy lactone 620a and the α,β-unsaturated lactone 621 were coupled by the Sonogashira cross-coupling reaction. Subsequent diimide reduction and deprotection afforded 610a. (4R,15S,16S,21S)-Rollicosin (610b) was also obtained starting from 620b using the same procedure as that employed for 610a. In 2006, the full details of this total synthesis were reported [190].

Total synthesis of squamostolide

Squamostolide (622), which was isolated from Annona squamosa by Wei’s group [191], showed a remarkably weak inhibitory activity compared to ordinary acetogenins such as bullatacin [191]. In 2006, Makabe’s group reported the total synthesis of squamostolide (Scheme 101) [190]. The lactone 622 was obtained by alkylation of the enolate prepared from 15 using NaHMDS with diiodide 623. The α,β-unsaturated lactone 625 was obtained after oxidation of 624 with mCPBA followed by thermal elimination of the resulting sulfoxide. Then segments 626 and 625 were coupled by a Sonogashira reaction to furnish product 627. Diimide reduction with p-TsNHNH₂ and NaOAc in ethylene glycol diethyl ether under reflux afforded squamostolide (622). The optical rotation, melting point, ¹H NMR, and ¹³C NMR spectra of the synthetic 622 were in good agreement with those of the reported values.

Total synthesis of tonkinelin

Tonkinelin (628a), which has a simple structure in the acetogenins (compared with other types of ACGs posessing THF ring or THP ring), was isolated from Uvaria tonkinesis in 1996 by Chen’s group [192]. This compound has two hydroxyl groups at C-17 and C-18 position, and possesses an α,β-unsaturated γ-lactone which can be seen in ordinary ACGs. In 2007, Makabe’s group reported the total synthesis of tonkinelin 628a (Scheme 102) [193]. Asymmetric dihydroxylation of 629 by the Sharpless procedure using AD-mix-α and spontaneous epoxide formation afforded epoxy alcohol 630a. Then the hydroxyl group of 630a was protected as a methoxymethyl ether (MOM ether) to give compound 631a. Alkynylation of 631a afforded 632a, and Sonogashira cross-coupling reaction of 632a with 633 gave enyne 634a. Diimide reduction of 634a followed by deprotection of the MOM ether with BF₃·Et₂O afforded 628a. The other candidate 628b was synthesized from 630b using the same procedure. By comparison of the optical rotation of the synthetic candidates and the natural compound, they suggested that the absolute configuration of natural tonkinelin was likely to be (17S,18S).

Conclusion

Annonaceous acetogenins are a relatively new class of bioactive naturally occurring products. The difficulty of isolating these compounds and elucidating their structures makes them a challenging target for total synthesis. Their wide spectrum of biological properties is probably the most intriguing and exciting domain, and the future will show whether it is possible to disclose the structure-activity relationship, probably on the basis of synthetic derivatives. Furthermore it will be useful to look for simplifications of the structure without loss of activity. As a result of these investigations, it will not be surprising if annonaceous acetogenins or related compounds with structural modifications might, in the near future, play a significant role in cancer therapy via an original mechanism of action. Hence we believe it is worthwhile to observe further developments in the field of annonaceous acetogenins.