1 Isolation 1.1 Isolation of combretastatins D series

Combretastatins comprise a large family of structurally diverse natural products divided into the “A” (cis-stilbenes), “B” (dehydrostilbenes), “C” (phenanthrenes), and “D” (macrocyclic diaryl ethers) series found in plants present on the African and Asian continent [15].

The first report of this class of compounds was made by Pettit and co-workers when they isolated combretastatin D-1 (1) from a CH2Cl2/MeOH extract of Combretum caffrum, a South African tree [16]. From 77 kg of stem wood, a fraction was obtained using a Sephadex LH-20 column by partition chromatography to afford two active fractions. One of the fractions was chromatographed on a silica gel column to give compound 1 (180 mg). The other fraction was again chromatographed on a Sephadex LH-20 column and the resulting active fraction was chromatographed on a silica gel column to afford a new fraction. Re-chromatography in a silica gel column using gradient elution afforded combretastatin D-2 (2, 5.8 mg) [17].

The general structures of combretastatins D-1 (1) and D-2 (2) were established by Pettit and Singh [16,17] by analysis of NMR and mass spectra and confirmed by X-ray crystallography in an initial report. However, attempts to determine the absolute configuration of the epoxide present in compound 1 based on crystallographic data were unsuccessful. By matching the sign of the Cotton effect curves obtained in the combretastatin D-1 spectrum with the appropriate chiral epoxides, the authors assigned the absolute stereochemistry of the epoxide ring as 3R,4S. This attribution was controversial and was only definitively established years later, as will be shown in this review.

In 2005, Vongvanich and co-workers isolated combretastatins D-3 (3) and D-4 (4) from Getonia floribunda, a woody climber commonly found in many areas of Thailand [18]. From 3 kg of dried stems of Getonia floribunda macerated in CH2Cl2 it was obtained a crude extract which was chromatographed using Sephadex LH-20. The obtained fractions were re-chromatographed on Sephadex LH-20 and the obtained fractions were further purified by silica gel column chromatography. One of the obtained fractions contained pure compound 3 (10.6 mg), while another fraction was further purified by silica gel column chromatography to give compound 4 (6.8 mg).

1.2 Isolation of corniculatolides and isocorniculatolides

Corniculatolides and isocorniculatolides, isomeric macrolides of combretastatins D, were isolated by Ponnapalli’s group from two distinct species of trees. From the bark of Aegiceras corniculatum, the authors isolated the known compound 4 and the isomeric corniculatolides. The isolation of the compounds was achieved from 5 kg of air-dried bark of the aforementioned tree, which was grounded and then extracted with CHCl3 using a Soxhlet apparatus, furnishing 32.0 g of the crude extract. This extract was subjected to a vacuum liquid chromatography on silica gel to yield several fractions. Some of them were tested and the ones that exhibited antimicrobial activity were further fractionated on silica gel chromatographic column eluted with hexane and acetone (8:2) to yield four different fractions. The fractions comprised 11-O-methylcorniculatolide A (5, 10 mg), 12-hydroxy-11-O-methylcorniculatolide A (6, 1 mg) and 11-O-methylisocorniculatolide A (8, 2 mg). The authors had some difficulties to isolate isocorniculatolide A (5) using chromatography, however, it was obtained by crystallization from hexane and acetone (4 mg). The same strategy was used to isolate the known compound corniculatolide A (4, 10 mg) as colorless crystals [14].

Later the same group reported the isolation of corniculatolide B (9), isocorniculatolide B (10), and corniculatolide C (11) from Xylocarpus granatum, a tree commonly found in Southeast Asia and along the Indian Ocean coastline. The air-dried stems of X. granatum (5 kg) were powdered and extracted successively within hexane, CHCl3, and acetone in a Soxhlet apparatus. The crude CHCl3 extract (21 g) was chromatographed on silica gel (230–400 mesh) vacuum liquid chromatography (VLC) in different gradients of hexane/acetone/MeOH to 20% MeOH. From the six main fractions, three of them showed the new structures, corniculatolide B (9, 3 mg), isocorniculatolide B (10, 2 mg), and corniculatolide C (11, 5 mg), among some other known constituents. Different 1H and 13C NMR and HRESIMS techniques were employed to elucidate the chemical structures of the isolated compounds [19].

2 Synthesis 2.1 Biosynthetic pathway

In the literature, there are two possible biosynthetic pathways for the formation of these compounds. The first one was proposed by Pettit and co-workers [16,17] based on tyrosine as the starting material. An o-phenolic coupling between two units of tyrosine furnishes the intermediate Int-1, which by deamination, selective reduction of one of the carboxylate groups, macrolactonization, and subsequent structural modifications would lead to the aforementioned combretastatins D (Scheme 1).

Scheme 1: Biosynthetic pathway proposed by Pettit and co-workers.

The second pathway was proposed by Ponnapalli and co-workers [14] and was initially based on the conversion of phenylalanine into tyrosine by phenylalanine hydroxylase and m-tyrosine via radical hydroxylation (Scheme 2). Subsequent deamination of tyrosine, with concomitant hydroxylation/deamination of m-tyrosine would give compounds 12 and 13. A coupling reaction would give the corresponding diaryl ether Int-2, in a similar way to that suggested by Pettit, which could be selectively reduced to afford the corresponding seco-acid (intermediates Int-3 and Int-4). Subsequent macrolactonization would give the corniculatolides or isocorniculatolides.

Scheme 2: Biosynthetic pathway towards corniculatolides or isocorniculatolides proposed by Ponnapalli and co-workers.

2.2 Retrosynthetic analysis

Due to the aforementioned biological activities and the low availability from natural sources to provide sufficient material for additional investigations, the combretastatin D series and their isomeric macrolides have become an attractive target for synthesis. In general, the synthesis of these macrocyclic compounds can be accomplished by using two distinct disconnections (Scheme 3): one concerns the formation of the macrocycle through macrolactonization reaction from the former seco-acid formed from the Ar–O–Ar coupling from the aryl donor/acceptor (route A), while the other corresponds to the intramolecular Ar–O–Ar coupling from the former ester (route B).

Scheme 3: Retrosynthetic approaches.

Both synthetic routes have their advantages and disadvantages. The formation of the Ar–O–Ar bond can be accomplished using different methodologies [20] such as SNAr [21], Ullmann [22], or Cham–Lam reactions [23-26]. However, it has been described that a Mitsunobu reaction of the seco-acid was particularly prone to an SN1 reaction, once the activated allylic alcohol yields an oxyphosphonium ion intermediate due to the conjugation to electron-rich aromatic ring, requiring some alternative experimental strategies to achieve the target molecule, as will be discussed further in this review.

2.3 Synthesis of combretastatins D series

Boger and co-workers were the first to report the total synthesis of compound 2 using both routes A and B to obtain the desired macrolide [27]. Initially, the authors employed an Ullmann-type condensation [28] between ester 14 and 4-iodobenzaldehyde (15) to give the corresponding diaryl ether 16 in 78% yield. The subsequent demethylation reaction using boron triiodide also promoted the hydrolysis of the ester, and thus a re-esterification step was necessary to give compound 17 in 85% yield after the two steps. Subsequent reaction of the aldehyde 17 following a modified Still–Gennari protocol [29] employing the phosphonate 18 gave the alkene 19 in 90% yield and high selectivity (cis/trans = 25:1). Removal of the silane group with TBAF furnished the carboxylic acid 20, which underwent protection with Troc-Cl and selective reduction in the presence of sodium borohydride to form the alcohol 21. After ester hydrolysis the desired seco-acid 22 was obtained in 82% yield. However, several attempts to achieve the macrolactonization of 22 using PPh3 and DEAD under different conditions [30] did not lead to the desired macrolide 2, but only the formation of the diolide was observed (Scheme 4).

Scheme 4: Attempt of total synthesis of 2 by Boger and co-workers employing the Mitsunobu approach [27].

Once the first synthetic pathway did not furnish the desired compound, the authors carried out the formation of the macrocycle using an intramolecular Ullmann-type coupling reaction. Thus, the olefination reaction of aldehyde 15 with phosphonate 23, followed by the reduction of the obtained ester 24 using DIBAL led to the alcohol 25. The latter was submitted to the reaction with carboxylic acid 26 under Mitsunobu conditions [30], giving the corresponding ester 27 in 97% yield. Subsequent intramolecular Ullmann-type reaction using CuMe under high dilution conditions [31] gave macrolide 28 in low yield. Finally, demethylation using boron triiodide [32] led to the formation of combretastatin D-2 (2, Scheme 5).

Scheme 5: Total synthesis of combretastatin D-2 (2) reported by Boger and co-workers employing an intramolecular Ullmann reaction [27].

Using this strategy, Boger succeeded to synthesize compound 2 in an overall yield of 25% after 5 steps, bypassing the macrolactonization problem evidenced in the previously envisaged route.

Intrigued by the problem encountered by Boger, Deshpande decided to investigate different reaction conditions for the formation of the macrocycle using the Mitsunobu reaction [33]. Thus, Knoevenagel condensation using the diaryl ether 29 and malonic acid gave the corresponding α,β-unsaturated compound 30, which was submitted to a concomitant hydrogenation of the double bond and the nitro group to give compound 31. Sequential diazotization/halogenation and esterification reactions gave the ester 33 which was submitted to a Sonogashira coupling reaction with propargyl alcohol to give the advanced intermediate 34 [34]. Partial hydrogenation of the triple bond in 34 using Lindlar’s catalyst led to the cis-allylic alcohol 35 and subsequent ester hydrolysis led to the formation of seco-acid 36. Macrolactonization attempts conducted under high dilution, including the Mitsunobu conditions [35] gave only the cyclic diolide. However, the use of higher dilution conditions and the dropwise addition of seco-acid 36 to a solution of DEAD (7.7 equiv) and triphenylphosphine (7.5 equiv) in PhMe led to the corresponding macrolide 28 in 20% yield. Since the demethylation step of 28 was known [27], the authors described a formal synthesis of combretastatin D-2 (2, Scheme 6).

Scheme 6: Formal synthesis of combretastatin D-2 (2) by Deshpande and co-workers using the Mitsunobu conditions in high dilution conditions [33].

Despite the low yield, the authors managed to bypass the dimerization reaction previously reported by Boger, and the formal synthesis of 2 was achieved after 8 steps with an overall yield of 4.5%.

Rychnovsky and Hwang hypothesized that the low yields from the Mitsunobu reaction in the previous synthesis of compound 2 were linked to the instability of the allylic oxyphosphonium ion formed with intermediates 22 and 36 (Scheme 4 and Scheme 6) and possibly an alkyl oxyphosphonium ion should be more stable [36]. Therefore, the authors proposed a synthetic sequence where the double bond was introduced only after the macrolactonization step.

The synthetic route was initiated by an Ullmann-type coupling [37] between halide 37 and phenol 38 leading to the formation of diaryl ether 39, which was subjected to a regioselective iodination reaction to give compound 40. Conversion of the nitrile in compound 40 into the corresponding aldehyde 41 followed by Z-selective Still–Gennari olefination gave the cis α,β-unsaturated ester 42. Conversion of the installed alkene to the corresponding thioether followed by the reduction of the ester moiety using DIBAL gave the compound 43, which was subjected to a Stille coupling reaction [38] to yield compound 45. Hydrogenation reaction in the presence of metallic Mg [39] followed by an ester hydrolysis led to the formation of seco-acid 46. By using Mitsunobu conditions at high dilution and slow addition of reagents, the authors were able to synthesize the macrolide 47 in excellent yield, thus confirming their initial hypothesis regarding the stability of oxyphosphonium ions. Further demethylation [40] and oxidation of the thioether followed by thermal elimination of the intermediate sulfoxide gave 2 in 98% yield after two steps (Scheme 7).

Scheme 7: Total synthesis of combretastatin D-2 (2) by Rychnovsky and Hwang [36].

The authors also achieved the synthesis of (±)-1 from combretastatin D-2 (2). Protection of the hydroxy group in compound 2 using acetic anhydride followed by epoxidation using m-CPBA gave protected epoxide 50. Subsequent removal of the acetate group using ammonia led to racemic compound 1 (Scheme 8).

Scheme 8: Divergent synthesis of (±)-1 form combretastatin D-2 (2) by Rychnovsky and Hwang [36].

Rychnovsky and Hwang succeeded in the total syntheses of combretastatin D-2 (2) in a 36% overall yield after 13 steps and combretastatin D-1 (1) in 23% overall yield after 16 steps.

Later, the same authors performed the enantioselective synthesis of 1 in an attempt to review its absolute configuration [41]. Thus, acetylation of compound 2 followed by the use of Jacobsen’s catalyst [42] to perform the epoxidation of the double bond gave the corresponding epoxide 51 in low enantioselectivity and only 44% yield. The subsequent deprotection reaction led to compound 1 in 86% yield (Scheme 9).

Scheme 9: Enantioselective synthesis of 1 by Rychnovsky and Hwang employing Jacobsen catalyst [41].

Besides the low enantioselectivity, the authors observed that the optical rotation value of the synthesized compound ([α]D +36.9, c 0.55, CHCl3) was different from the value reported for the natural product ([α]D −100, c 0.015, CHCl3) [16] and hypothesized that the configuration of the natural compound would be 3S,4R, different from that reported in the Pettit previous work.

Couladouros and co-workers based the synthetic design of combretastatin D on the use of computational calculations in order to find the intermediates with the lowest torsional energy for the cyclization step [43-45]. The authors came to the conclusion that both the formation of the double bond in compound 2 and the formation of the epoxide in compound 1 would only be favorable after the macrolactonization step.

The authors used a convergent route for the formal synthesis of 2. Reaction of 4-bromobenzaldehyde (52) with a commercially available stabilized Wittig reagent led to the formation of the corresponding ester 53. Further reduction of the carbonyl group followed by protection of the obtained alcohol with benzyl bromide provided compound 55, which was subjected to an epoxidation using m-CPBA followed by ring opening using DIBAL [46]. The obtained alcohol was then protected with TBSCl to give fragment 57 (Scheme 10).

Scheme 10: Synthesis of fragment 57 by Couladouros and co-workers [43,45].

Using similar conditions to Boger´s protocol, compound 58 was then subjected to an Ullmann coupling reaction in the presence of ester 59 to yield the corresponding diaryl ether 60. The hydrolysis of the ester followed by the removal of the benzyl group led to the corresponding seco-acid 62. The obtained compound showed high stability when subjected to Mitsunobu conditions with the slow addition of the seco-acid into the DEAD/PPh3 reaction mixture over a 7 h period. The desired macrolide 63 was obtained in 91% yield, without formation of the diolide being observed. Subsequent removal of the TBS group gave the corresponding alcohol 64. The formation of the double bond from alcohol 64 proved to be problematic, thus, replacement of the hydroxy group by iodine [47] followed by dehydrohalogenation using an excess of KF afforded methyl combretastatin D-2 (28) in 87% yield after two steps (Scheme 11).

Scheme 11: Formal synthesis of compound 2 by Couladouros and co-workers [43,45].

The authors also described a convergent route by which it was possible to reach both, compound 2 and 1. Initially, the authors prepared the protected alcohol 66 from aldehyde 52 in a linear sequence (Scheme 12).

Scheme 12: Synthesis of fragment 66 by Couladouros and co-workers [44,45].

In parallel, the monobenzylation of 3,4-dihydroxybenzaldehyde (67), followed by chain elongation using the Wittig reaction furnished the α,β-unsaturated ester 69. The subsequent catalytic hydrogenation led to the desired phenol 70 (Scheme 13) [44,45].

Scheme 13: Synthesis of fragment 70 by Couladouros and co-workers [44,45].

An Ullmann coupling reaction using compounds 66 and 70 gave the corresponding diaryl ether 71, which was submitted to an asymmetric dihydroxylation reaction using (DHQD)2PHAL to yield diol 72. The 3R,4S configuration of compound 72 was expected based on Pettit’s work [16,17] and the optical purity of the obtained product was more than 95% by 1H NMR using [Eu(hfc)3] as a chiral shift reagent. Subsequent silylation followed by ester hydrolysis and removal of the pivaloyl group provided the seco-acid 75. Employing the same Mitsunobu conditions previously described [35], the authors were able to obtain the macrolide 76 in 81% yield which was then subjected to deprotection to give compound 77 (Scheme 14).

Scheme 14: Synthesis of fragment 77 by Couladouros and co-workers [44,45].

The synthesis of 2 from compound 77 was achieved after hydrogenolysis of the benzyl ether. Further double bond formation in compound 78 employing triiodoimidazole and PPh3 led to 2 (route A, 32% overall yield from 52). The synthesis of combretastatin D-1 (1) was achieved from the cyclodehydration of compound 77, followed by the hydrogenolysis of the benzyl ether 79 (route B, overall yield of 29% from 52) (Scheme 15).

Scheme 15: Synthesis of combretastatins 1 and 2 by Couladouros and co-workers [44,45].

Using this strategy, combretastatin D-1 (1) was obtained in an enantiomeric purity of 96%, making it possible to establish the absolute configuration of the epoxide. Further, X-ray crystallographic analysis corroborated with the results obtained by Rychnovsky and Hwang [41] that the correct configuration was 3S,4R, evidencing that the original attribution described by Pettit and co-workers [16,17] was inaccurate.

Gangakhedkar elaborated a synthetic route where it was possible to formally synthesize compound 2 in 9 steps [48]. Conversion of the hydroxybenzaldehyde 80 into the corresponding acetal followed by Ullmann-type coupling with 52, led to the formation of diaryl ether 83. Subsequent Corey–Fuchs reaction [49] and in situ alkylation led to formation of the propargylic alcohol 85. Deprotection of the aldehyde followed by chain elongation through the Wittig reaction led to the α,β-unsaturated ester 87, which was subjected to a hydrogenation reaction in the presence of metallic magnesium, leading to the formation of alkyne 88. The cis-alkene was selectively obtained using the Lindlar catalyst. Finally, hydrolysis of the ester led to the formation of seco-acid 36. Using this approach, the authors were able to achieve the formal synthesis of 2 reaching a key intermediate in 34% overall yield after 9 steps (Scheme 16).

Scheme 16: Formal synthesis of compound 2 by Gangakhedkar and co-workers [48].

Cousin and co-workers [50] innovated by using the Chan–Lam coupling [23-25] for the diaryl ether formation and applying an intramolecular Wittig reaction to promote the macrocyclization in the formal synthesis of compound 2. Initially, the authors synthesized the fragment 14 from the starting aldehyde 80 by using a Wittig reaction followed by hydrogenation using ammonium formate (Scheme 17).

Scheme 17: Synthesis of fragment 14 by Cousin and co-workers [50].

Concomitantly, 4-formylphenylboronic acid (91) was prepared from the borylation reaction of 4-bromobenzaldehyde (52, Scheme 18) [50].

Scheme 18: Synthesis of fragment 91 by Cousin and co-workers [50].

The coupling reaction [23-25] between 14 and 91 gave the corresponding diaryl ether 16 in 68% yield. Subsequent transesterification reaction [51] using dibutyltin oxide and allylic alcohol led to the formation of compound 92, which was reduced to the corresponding alcohol and then converted into the bromide 94. Ozonolysis followed by reaction with triphenylphosphine gave the corresponding phosphonium salt 96, which was subjected to different conditions for the intramolecular Wittig reaction. The best conditions found by the authors gave the desired macrolide in only 30% yield together with the trans isomer, which was further isomerized to the cis-alkene during purification by column chromatography and light, being the first time that the trans isomer was reported (Scheme 19).

Scheme 19: Formal synthesis of compound 2 by Cousin and co-workers [50].

Another strategy employed by the authors consisted in the ring closure through a metathesis reaction using the Grubbs catalyst [52,53]. The required compound 99 was prepared by converting compound 16 into the styrene 98 via a Wittig reaction followed by a transesterification to yield the desired allylic ester. Several reaction conditions for the metathesis using the 1st generation Grubbs catalyst were attempted without success, but when 2nd generation catalyst was used, the dimerization product 100 was observed (Scheme 20).

Scheme 20: Synthesis of 2 diolide by Cousin and co-workers [50].

Despite the yield for macrolide formation, Cousin proposed alternatives for the formal synthesis of 2, employing a 10-step synthetic route with an overall yield of 9%.

Nishiyama employed electrochemical techniques as a starting point to achieve the total synthesis of combretastatin D-4 (4) [54]. Different anodic oxidation conditions and phenolic substrates were tested aiming at the formation of a diaryl ether moiety. The best result was obtained when phenol 101 was subjected to anodic oxidation, leading to the formation of spiro-dimer 102 in 61% yield. Protection of the alcohol using TBSOTf followed by cyclic ether cleavage and re-aromatization gave compound 104. Subsequent dehalogenation followed by protection with BnBr and oxidation led to the carboxylic acid 107. Esterification of the carboxylic acid followed by the cleavage of the silyl ether using TBAF and hydrolysis led to the seco-acid 108. Macrolactonization using the Mitsunobu conditions gave combretastatin D-4 (4) after cleavage of the benzyl ether using Pd/C and ammonium formate (Scheme 21).

Scheme 21: Synthesis of combretastatin D-4 (4) by Nishiyama and co-workers [54].

With this synthetic route, the authors achieved the total synthesis of combretastatin D-4, after 12 reaction steps with an overall yield of 11%, highlighting the efficient formation of the diaryl ether 102 in 61% yield without the use of metallic catalysts, through dimerization of a single molecule.

Pettit and co-workers investigated the influence of structural modifications on the biological activity of combretastatins D-2 (2) and D-4 (4). The authors also investigated the influence of solvents and functional groups in the total synthesis of the targeted compounds aiming to reach higher overall yields and fewer steps [55]. The monobenzylation [56] of aldehyde 67 followed by chain elongation using a Wittig reaction gave compound 68, which was submitted to the hydrogenation of the double bond. The use of benzene as solvent in the hydrogenation step proved to be important for the selectivity of the reaction, where significant cleavage of the benzyl group resulted when ethanol was the solvent of choice. Subsequent ester hydrolysis gave compound