The Halichondrins and E7389
Katrina L. Jackson, James A. Henderson, and Andrew J. Phillips*
Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0215
Received January 15, 2009
⦁ Concluding Comments 3077
⦁ References 3077

⦁ Introduction, Isolation, and Structure Elucidation of 3044 the Halichondrins
⦁ Biological Activity 3046
⦁ Introduction, Isolation, and Structure Elucidation of the Halichondrins

⦁ Kishi’s Total Syntheses of Halichondrin B and
Norhalichondrin B
Since the initial report of norhalichondrin A in 1985, there

⦁ C1-C13 Subunit Synthesis 3048
⦁ C14-C26 Subunit Synthesis 3048
⦁ C27-C38 Subunit Synthesis 3049
⦁ Halichondrin B C39-C54 Subunit Synthesis 3049
has been substantial interest from chemists, biologists, and most recently the medical community in the halichondrin family of compounds. This interest is without doubt a combination of the chemical synthesis challenges posed by

⦁ Norhalichondrin B C39-C53 Subunit Synthesis
⦁ Subunit Couplings and Completion of the Syntheses

their structures, their remarkable in Vitro and in ViVo antitumor activities, and the potential for a therapeutic based on a halichondrin. While the area has been reviewed from several standpoints1-4 over the past two decades, in this

⦁ Subsequent Improvements by the Kishi Group 3052
review, we endeavor to provide comprehensive coverage

⦁ Total Synthesis of Norhalichondrin B by Phillips and Co-workers
through the end of February, 2009.
In 1981, independent studies by Schueur and Schmitz

⦁ C1-C13 Subunit Synthesis 3056
⦁ C14-C26 Subunit Synthesis 3057
⦁ C27-C38 and C40-C53 Subunit Syntheses 3057
culminated in the isolation of okadaic acid, 1 (Figure 1), the
structure of which was elucidated in collaboration with Clardy.5,6 This complicated polyether was isolated from two

⦁ Subunit Couplings and Completion of the Synthesis
⦁ Synthetic Work toward Halichondrin B by Horita and Yonemitsu
sponges, Halichondria okadai Kadota, a black sponge commonly located along the Pacific coast of Japan, and Halichondria melanodocia, a Caribbean sponge collected from the Florida Keys.

Okadaic acid is remarkably toxic with intraperotineal
5.4. C37-C54 Subunit Synthesis 3062 ous studies that were guided by the observation of potent in

injections in mice showing the LC50 to be 192 µg/kg; however, it displays no in ViVo antitumor activity at subtoxic doses against P388 lymphocytic leukemia. In contemporane-

ViVo activity of crude extracts from Halichondria okadai

⦁ Synthetic Work toward Halichondrin B by Salomon
Kadota (see Figure 4), Uemura isolated and identified norhalichondrin A (2) and demonstrated that this compound

⦁ C1-C15 Subunit Synthesis 3066
⦁ C1-C21 Subunit Synthesis 3066
⦁ C27-C35 Subunit Synthesis 3066
⦁ C37-C51 Subunit Synthesis 3067
⦁ Synthetic Work toward Halichondrin B by Burke 3068
⦁ C1-C15 Subunit Synthesis 3068
⦁ C14-C22 Subunit Synthesis 3069
⦁ Synthesis of the C22-C34(36) Subunit 3070
⦁ Synthesis of the C38-C54 Subunit 3072
⦁ Subunit Couplings 3073
⦁ The Discovery and Development of E7389 3073
⦁ Preliminary SAR Studies 3074
was the likely source of the cytotoxicity (IC50 ) 5 ng/mL vs B16 melanoma).7
Although a number of key features of the structure were obtained from methods such as mass spectrometry, IR, and NMR, the structure of norhalichondrin A was ultimately secured by X-ray analysis on the p-bromophenacyl ester (Figures 2 and 3, compound 3). The absolute configuration determined as part of the crystallographic process was also consistent with the dibenzoate exciton chirality data for compound 4, which showed a positive split CD. The determination of the structures of other halichondrins has subsequently been performed by careful comparison of data (especially NMR) to that for 2.

⦁ Synthesis and SAR of ER-076349, E7389,
and Analogues
Uemura and co-workers subsequently collected 600 kg of
H. okadai, and from this material, seven further halichondrins

⦁ Preclinical Development 3076
were identified.8 All of the members of the halichondrin

⦁ Progress toward the Clinic: Current Status of E7389
* E-mail: [email protected].
family possess an unusual 2,6,9-trioxatricyclo[]decane ring system, as well as a 22-membered macrolactone ring, two exocyclic olefins, and an array of polyoxygenated pyran and furan rings that define three major classes of halichon-

10.1021/cr900016w CCC: $71.50  2009 American Chemical Society Published on Web 06/17/2009

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3045

Katrina L. Jackson obtained her B.S. degree in biochemistry from Boston College in 2002, where she performed undergraduate research with Professor Shana O. Kelley. In 2007, she obtained her Ph.D. from the University of ColoradosBoulder under the direction of Professor Andrew
J. Phillips for the total synthesis of norhalichondrin B. Currently, she is a postdoctoral researcher at Harvard University with Professor Yoshito Kishi.

James A. Henderson received a B.Sc. in Biochemistry from Tennessee Technological University. After completing a M.S. degree in Chemistry at the same institution with Professor Jeffrey O. Boles, he moved to the University of Colorado at Boulder where he received his Ph.D. degree under the direction of Professor Andrew Phillips for the total synthesis of aburatubolactam A. In 2008, he moved to Harvard University where he is currently a postdoctoral researcher in the lab of Professor Yoshito Kishi.

Andrew Phillips obtained his B.Sc. (Hons, 1st class) and Ph.D. degrees from the University of Canterbury in Christchurch, New Zealand. After a postdoctoral appointment with Professor Peter Wipf at the University of Pittsburgh, he joined the faculty at the University of Colorado in 2001, where he is currently an Associate Professor of Chemistry.

drins (Figure 5). The A, B, and C families differ in their degree of oxidation at C12 and C13, and within each family,
Figure 1. (+)-Okadaic Acid, 1, and Norhalichondrin A, 2.

Figure 2. Norhalichondrin A p-bromophenacyl ester, 3, and the C12-C13 bis-p-bromobenzoate, 4, used for absolute stereochemical determination.

there are differences beyond the C45 position that are differentiated by the prefix, or lack thereof, before “hali- chondrin”.

In 1987, Blunt and Munro identified halichondrins in two unrelated New Zealand species of sponge: Raspalia agminata a black shallow-water sponge and Lissodendoryx sp. a deep water sponge.9 Only trace amounts of halichondrins were detected in Raspalia agminata, but the Lissodendoryx sp. sponge, which was collected by benthic dredging at 100 m, produced significant amounts of halichondrins. The yield from this sponge is typically 10-fold higher than any of the other halichondrin-producing sponges ( 1 mg total hali- chondrins per kilogram of wet weight of sponge vs 0.1 mg/ kg or less for other species). Bioassay-directed fractionation of the extract of Lissodendoryx sp. produced isohomohali- chondrin B, 10, as the major component, along with lesser amounts of halichondrin B and homohalichondrin B, 6.10 A large scale purification campaign from 200 kg of sponge yielded three minor new compounds: neonorhalichondrin B (8), neohomohalichondrin B (7), and 55-methoxyisohomo- halichondrin B (11).11 In contemporaneous studies that were described in 1991, Pettit and co-workers reported the isolation of halichondrin B and homohalichondrin B from an Axinella sp. sponge that was collected in Palau.12 Three new halichondrin compounds, halistatins 1, 2, and 3 (13, 14, and 15, respectively), were subsequently reported by Pettit.13 These compounds are oxidized at C10 relative to other

3046 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Figure 3. Ball and stick (hydrogens omitted for clarity) and space-filling depictions of norhalichondrin A p-bromophenacyl ester, 3.
recommended by the National Cancer Institute (NCI) for preclinical trials in March 1992. However, the naturally low abundance of the molecule from natural sources rendered supply a seemingly difficult problem. From consideration of the potency of halichondrin B, 5, in ViVo and projection of the dose regimens for human use, the likely amount needed for preclinical trials and eventual clinical investigation would be on the order of 10 g.15 Assuming success in clinical trials, the necessary commercial supply was estimated to be quite small, at around 1-5 kg/yr. Obtaining even this relatively small amount through collection from natural sources is unlikely. For example, the halichondrin-producing Lissodendoryx sp. sponge is a rare, deep water species found

Figure 4. Left image, Halichondria okadai (black sponge); right image, Lissodendoryx sp. sponges. Photos courtesy of Professor Daisuke Uemura and Professors J. W. Blunt and M. H. G. Munro.

halichondrins and were isolated from an east Indian Ocean sponge (Phakellia carteri), a western Indian Ocean sponge (Axinella sp.), and a Pacific Ocean Phakellia sp. sponge. The isolation of halichondrins from a variety of sponges lends credence to the possibility that these compounds are bio- synthesized by sponge symbionts rather than the sponges themselves.

⦁ Biological Activity
Beyond their exquisitely complex molecular architectures, the most exciting feature of the halichondrins is the impres- sive biological activity that the class presents. With an IC50 of 0.093 ng/mL against B-16 melanoma cells, halichondrin B, 5, was singled out as the most potent congener of the halichondrin family members isolated in Uemura studies (Table 1).
Initial biological evaluation of halichondrin B against B-16 melanoma cells, P-388 leukemia cells, and L-1210 leukemia cells resulted in dramatically increased survival times in mice. In the cases of B-16 melanoma and P-388 leukemia xeno- grafts in mice, daily treatment with 5 µg/kg of halichondrin B for 9 days resulted in a 2-fold improvement in the mean survival time (Tables 2 and 3). Several other dosing regimens provided comparable results, and similar results were ob- served with L-1210 leukemia (Table 4).
Later studies determined that halichondrin B behaves as a tubulin destabilizing agent with subtle differences in mechanism of action from those of other antimitotics such as the vinca alkaloids.14 During in Vitro studies, halichondrin blocked cells at G2/M phase of the cell cycle and disrupted normal mitotic spindle architecture through microtubule destabilization with few, if any, effects on other phases of the cell cycle. Halichondrin B, 5, was also found to possess extraordinary in ViVo activity against various chemoresistant human solid tumor xenographs, including LOX melanoma, KM20L colon, FEMX melanoma, OVCAR-3 ovarian, NCI H522 lung, and MDA-MB-435 breast cancers. On account of its impressive biological activity, halichondrin B, 5, was
off the Kaikoura Peninsula of New Zealand. A survey of the region by remotely operated underwater vehicle (ROV) and a benthic camera gave an estimated total biomass of the Lissodendoryx sponge of just 200-380 t. A collection of 1 t was undertaken and from this material Blunt and Munro were able to isolate 310 mg of halichondrin B, 5, and a comparable amount of isohomohalichondrin B, 10. Although this campaign provided material that was sufficient for preclinical work at the NCI, it also underscored the fact that large-scale harvesting of deep sea sponges was unlikely to be a viable option for supply.
One potential source of halichondrin B, 5, is aquaculture of the appropriate sponge genera. The New Zealand National Institute of Water and Atmospheric Research (NIWA) in collaboration with the University of Canterbury and the NCI have carried out aquaculture feasibility trials and found that the rate of production of halichondrin B, 5, from cultured Lissodendoryx sp. is lower than that found in the wild sponge by approximately 30-60%. Additionally, it is estimated that approximately 1000-5000 tons of sponge/year would need to be harvested to fulfill the need should halichondrin become a marketed drug. Unfortunately, initial research in this area has produced only limited results on small-scale trials, making aquaculture a somewhat unlikely option for produc- ing the necessary quantity of halichondrin B, 5, at least in the near future.
At the current time, perhaps the most tractable option for a reliable source of the halichondrins is total synthesis. Nonetheless, the molecular architecture presents significant challenges to synthetic chemistry. Although three total syntheses of halichondrins have been reported, for any of them to become viable and cost-effective options, significant refinement is required. An alternative would be the identi- fication of simplified analogues, an avenue that has received significant attention, and is discussed in section 8.

⦁ Kishi’s Total Syntheses of Halichondrin B and Norhalichondrin B
In a ground-breaking piece of work, Kishi and co-workers were the first to describe total syntheses of halichondrin B, 5, and norhalichondrin B, 9, in 1992.16 As was the case with the total synthesis of palytoxin, the Kishi syntheses of

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3047

Figure 5. The halichondrin-halistatin family of natural products.
halichondrin B, 5, and norhalichondrin B, 9, underscored the utility of the Ni/Cr Nozaki-Hiyama-Kishi (NHK) reaction17,18 to construct highly complex molecules. By the time the sequence was completed, this transformation had been used to forge five carbon-carbon bonds, several of which were the key bonds formed as part of subunit couplings, with good to excellent yields. Another overarching

feature of the Kishi synthesis was the use of readily accessible carbohydrate-based starting materials. The key overall strategy for halichondrin B, 5, is shown in Figure 6. The endgame was based on the formation of the key C38-C39 bond by NHK reaction, followed by formation of the two final polyether-domain spiroketals. This analysis provided C39-C54 fragment 19, which ultimately could be

3048 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Table 1. Initial in Vitro Testing of Halichondrins Isolated by Uemura

IC50 (ng/mL) amount (mg)a
halichondrin B, 5 0.0093 12.5
halichondrin C, 18 0.35 7.2
norhalichondrin A, 2 5.2 35.0
norhalichondrin B, 9 4.2
We note at the outset that since the total synthesis was published in 1992, Kishi and co-workers have continued their research and have published improved routes to several of these fragments. The first generation synthesis is presented below, and a discussion of the improved routes to various subunits is provided at the end of this section.

norhalichondrin C, 16
homohalichondrin A, 12
⦁ C1-C13 Subunit Synthesis19

homohalichondrin B, 6 0.10 3.1
homohalichondrin C, 17 2.1
a Amount isolated from 600 kg of wet sponge.

Table 2. Halichondrin B Antitumor Activity against B-16 Melanoma in ViWo

dose (µg/kg) regimena MST (days)b T/C (%)c
The synthesis of the C1-C13 segment commenced with D-glucose diacetonide (24, Scheme 1), which was oxidized to the ketone using the Omura-Swern protocol20 and treated with sodium borohydride in order to provide 31, which has the correct alcohol stereochemistry at C8. After benzyl protec- tion, the C5-C6 acetonide was selectively removed with HCl in aqueous MeOH to give 32. Mesylation of the diol and double displacement with KOAc, followed by acetate cleav- age with sodium methoxide, provided diol 33. Acid-catalyzed rearrangement to the pyranose and treatment with acetyl chloride and methanol provided the methyl talopyranoside
qd; 1-9; ip 16
5.0 39 244
0 19
5 qod; 1, 3, 5, 7, 9; ip 37.5 197
10 39.5 208
1, 5, 9; ip 18
20 39.5 219

34. After perbenzylation, exposure of 34 to allyltrimethyl- silane and TMSOTf provided allyl glycoside 36, presumably via the intermediacy of oxacarbenium ion 35. A series of standard protecting group manipulations, oxidation reac-

0 1, 4, 7, 10; iv
27.5 157
tions,21 and a Wittig reaction produced R,§-unsaturated ester
37. Removal of the acetates, followed by treatment with

a Abbreviations: qd, every day; qod, every other day; ip, intraperi- toneal; iv, intravenous. b Median survival time. c Test group/control group.

Table 3. Halichondrin B Antitumor Activity against P-388 Leukemia in ViWo
dose (µg/kg) regimena MST (days)b T/C (%)c 0 11
1.25 15 136
10 qod; 1, 3, 5, 7, 9; ip 35.5 323
a Abbreviations: qd, every day; qod, every other day; ip, intraperi- toneal. b Median survival time. c Test group/control group.

Table 4. Halichondrin B Antitumor Activity against L-1210 Leukemia in ViWo

dose (µg/kg) regimena MST (days)b T/C (%)c

Triton-B methoxide, and finally benzoylation of the C11 alcohol, provided the heteroconjugate addition product 38 as a 2.5:1 mixture of C3 diastereomers. This mixture of diastereoisomers could be consolidated to a single diastereo- isomer, 39, by a two-step process consisting of formation of the p-methoxyphenyl acetal and treatment with Triton B methoxide. After a four-step sequence of straightforward manipulations, aldehyde 40 was obtained, setting the stage for the first Ni(II)/Cr(II)-mediated reaction. When exposed

2.5 qd; 1-9; ip 16.5 150 to CrCl2 and 0.01% NiCl2 in THF, iodotrimethylsilylacety-
5.0 26 236 lene 41 added to the aldehyde to form 42 in 80% yield and

with impressive diastereoselectivity (dr ) 10:1). Kishi and Stamos have provided a rationale for this reaction that invokes a Cornforth-type model, for example, 44 f 42 compared with 45 f C11-epi-42.22 Four additional steps were then employed to convert the TMS alkyne to vinyl iodide 23 via the intermediary of vinyl stannane 43.

⦁ C14-C26 Subunit Synthesis

50 14.5 207
70 14 200

30 qd; 1-5, 7-12; ip

50 qod; 1, 3, 5, 7, 9, 11; ip
10 143

11 138
The synthesis of the C14-C26 segment 25 was a convergent approach based on the key union of 55 and 62 via a Horner-Wadsworth-Emmons reaction (Scheme 2). Thioacetal 46, readily accessible in two steps from L-arabinose (26), was converted to 47 by removal of the acetonide and protection of

100 30 375
a Abbreviations: qd, every day; qod, every other day; ip, intraperi- toneal. b Median survival time. c Test group/control group.

traced back to L-ascorbic acid, and macrolactone 21. This macrolactone 21 is also common to the synthesis of nor- halichondrin B, 9. Analysis of 21 quickly leads to 22 as a key intermediate that should be readily assembled by the union of C1-C13 fragment 23, C14-C26 fragment 25, and C27-C38 fragment 27. The starting materials for these three fragments are D-glucose acetonide (24), L-arabinose (26), and D-galactose glycal (28), respectively. A similar analysis for norhalichondrin B, 9 (Figure 7), reveals D-galactose glycal, 28, as the starting material for C39-C53 subunit 29.
the primary alcohol as the tert-butyldiphenylsilyl (TBDPS) ether. Treatment with I2 in buffered wet acetone resulted in cleavage of the dithiane to give an intermediate tetrahydrofuran hemiacetal that was acetylated to yield 48. A highly distereo- selective oxacarbenium ion allylation with allyltrimethylsilane mediated by BF3 · OEt2 installed the three carbons needed for C12-C14 (48 f 49). Hydroboration-oxidation, protection of the resulting alcohol as the methoxytrityl ether, and removal of the acetate gave 50. Swern oxidation, followed by removal of the monomethoxytrityl (MMTr) ether, produced hydroxy- ketone 51, which was readily olefinated using the Tebbe reagent 52 to give 53. Protecting group exchange and Dess-Martin oxidation23 advanced this compound to the key aldehyde 55 via 54.

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3049

Figure 6. An overview of the retrosynthesis for halichondrin B, 5, by Kishi and co-workers.
ester hydrolysis allowed material of the correct stereochem- istry to be pooled, and then mesylation completed the sequence.

⦁ C27-C38 Subunit Synthesis25
The synthesis of the C27-C38 domain began with D-galactose gycal, 28, which was converted in a four steps to the 4-O-benzyl-3,6-O-dipropionate, 66 (Scheme 3). An Ireland-Claisen rearrangement via the intermediacy of the silylketene acetal produced 67 as an approximately 8:1 mixture of diastereoisomers. This mixture was subjected to iodolactonization followed by reductive removal of the iodide to produce γ-lactone 68. At this stage, the minor isomer from the Ireland-Claisen rearrangement was removable by re- crystallization. By a 10-step sequence of routine transforma- tions, including a one-carbon homologation (69 f 70), γ-lactone 68 was then converted into aldehyde 72.
At this juncture, a Nozaki-Hiyama-Kishi reaction be- tween aldehyde 72 and methyl-(E)-3-iodoacrylate, 73, with CrCl2 containing 1% NiCl2 in THF was employed to form trans-γ-hydroxy-acrylate, 75, in an excellent 84% yield and

Figure 7. An overview of the retrosynthesis for norhalichondrin B, 9, by Kishi and co-workers.

The synthesis of §-ketophosphonate 62 commenced with pyroglutamic acid, 56, which was reduced to the primary alcohol and protected to yield 57. Enolate alkylation with with lithium diisopropylamide (LDA)/MeI provided lactone 58, which conveniently crystallizes from hexanes. Lactone opening with methyllithium and protection (58 f 59) was followed by a Shapiro reaction with Bu3SnCl as the elec- trophile. Subsequent iododestannylation gave vinyl iodide
60. A five-step sequence produced 62.
Fragments 55 and 62 were coupled by Horner-Wads- worth-Emmons reaction in the presence of sodium hydride, and subsequent Stryker reduction24 provided saturated ketone
63. Ketone reduction gave a 2:1 mixture of secondary alcohols 64 and 65 favoring the undesired diastereoisomer. Inversion of 64 to 65 under Mitsunobu conditions and then
with a d.r. of 2:1 (favoring the desired diastereoisomer) (Scheme 4). The diastereoisomers could be separated by repeated chromatography, and the undesired diastereoisomer 74 was easily inverted to the desired alcohol 75 by a two- step process consisting of Mitsunobu reaction with p- nitrobenzoic acid and methanolysis of the nitrobenzoate. Protecting group manipulations produced 76, which, upon treatment with tetra-n-butylammonium fluoride (TBAF), underwent heteroconjugate addition to provide the desired pyran 77 with excellent diastereoselectivity. Four further steps were required to convert 77 into the C27-C38 aldehyde-containing 27.

⦁ Halichondrin B C39-C54 Subunit Synthesis26
The C39-C54 segment of halichondrin B was synthesized as shown in Scheme 5. At the time that this work was being done, there was a significant amount of ambiguity regarding

3050 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 1. The Kishi Synthesis of the C1-C13 Subunit
Scheme 2. The Kishi Synthesis of the C14-C26 Subunit

the stereochemistry of the C50, C51, and C53 centers, all of which had been assigned based on 1H-1H vicinal couplings as well as biogenetic considerations. Therefore, the Kishi synthesis was designed specifically to address whether the assigned stereochemistry was correct and to be able to easily synthesize all possible permutations of the triol should the data not match the natural product.
Conjugate addition of methylcuprate to the R,§-unsaturated γ-lactone 79 (readily available from L-ascorbic acid, 20, in five steps27) provided 80 as a single stereoisomer in 95% yield. A standard sequence of seven steps advanced 80 to the epoxide 83. This epoxide was coupled under Yamagu- chi’s conditions28 with the alkynylborane derived from acetylide 84 and reduced under Lindlar conditions to provide

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3051

Scheme 3. The Initial Stages of the Kishi Synthesis of the C27-C38 Subunit

Scheme 4. Completion of the Kishi Synthesis of the C27-C38 Subunit

cis-olefin 85. Subsequent vanadium-mediated directed ep- oxidation using Sharpless’ protocol,29 followed by acidifica- tion, furnished tetrahydrofuran 86. Importantly, all of the other possible stereochemical combinations of C50, C51, and C53 could be accessed in similar fashion by using epoxi- dations of the corresponding cis- and trans-olefins prepared from acetylide 84 and its antipode. To complete the sequence,
86 was converted to aldehyde 87 in three steps, and organolithium 88 was added to the aldehyde. Removal of the silicon with AgNO3, hexamethyldisilazane (HMDS), and NaI was followed by conversion to the vinyl stannane under radical conditions. Iododesilylation and oxidation gave 19.

⦁ Norhalichondrin B C39-C53 Subunit Synthesis30
The synthesis of the C39-C53 fragment of norhalichon- drin B, 29, began in a similar vein to the synthesis of the C27-C38 segment, with D-galactose glycal being converted

Scheme 5. The Kishi Halichondrin B C39-C54 Subunit Synthesis

to 4-O-benzyl-3,6-O-dipropionate, 66 (Scheme 6). Ireland- Claisen rearrangement of the silylketene acetal generated in the absence of hexamethylphosphoramide (HMPA) gave 89 as the major product with 5:1 diastereoselectivity. Compound 89 was advanced to 90 by the same sequence as was used earlier (see Scheme 3). From 90, reductions and protecting group changes gave 92. Mesylation, homologation with cyanide, and reduction gave 93. The same five-step sequence employed to complete the C39-C54 segment was then employed to form 94. Lastly, the TBS-protected C53 alcohol was converted to the methyl ester in a three-step process (94 f 29) to complete the C39-C53 segment of norhali- chondrin B.

⦁ Subunit Couplings and Completion of the Syntheses
The assembly of the segments of halichondrin B and norhalichondrin B to complete the total syntheses is shown in Schemes 7 and 8. In the case of halichondrin B, the first fragment coupling between 26 and 27 relied upon the use of the Ni(II)/Cr(II)-mediated coupling reaction to give 95 (d.r. ) 6:1 in favor of the desired stereochemistry). Base- induced cyclization with inversion furnished tetrahydropyran 96 in 50-60% overall yield for the two steps, along with a small amount of the undesired diastereoisomer. Subsequent treatment with LiAlH4 and then Dess-Martin periodinane converted the pivaloyl ester into aldehyde 97. To effect the union of 97 and 23, another Ni(II)/Cr(II)-mediated coupling was employed, this time resulting in an 80% yield of 98.

3052 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 6. Kishi’s Norhalichondrin B C39-C53 Subunit Synthesis

Oxidation to the enone and treatment with 2,3-dichloro-5,6- dicyanobenzoquinone (DDQ),31 followed by lithium hydrox- ide, cleaved the p-methoxybenzyl (PMB) ether and hydro- lyzed the methyl ester. Subjection of the resultant seco-acid 99 to Yamaguchi conditions32 allowed lactonization to form

21 in 81% yield. Formation of the trioxatricyclo- [,7]decane ring system was initiated by treatment of 21 with TBAF. Under these conditions, the TBS groups were cleaved, and the liberated C9 alcohol added the C12-C13 enone in Michael fashion to form the five-membered ring with approximately 5-6:1 diastereoselectivity in favor of the desired stereochemistry. Treatment with pyridinium p-toluenesulfonate (PPTS) resulted in formation of the desired polycyclic ketal 100 in an impressive 64% yield after p-nitrobenzoyl protection of the C38 primary alcohol. It is worth noting that the undesired Michael adduct, which did not undergo ketalization, could be separated and recycled by treatment with TBAF (equilibration to a 5-6:1 ratio of C12 diastereoisomers occurs) and PPTS to produce further
100. Silylation of the secondary alcohol, followed by removal of the p-nitrobenzoate, gave 101.
Coupling of the common right half aldehyde with either the left portion of halichondrin B or the left portion of norhalichondrin B was accomplished with the Ni(II)/Cr(II)- mediated coupling reaction (Scheme 8). After Dess-Martin oxidation, a 50-60% yield was obtained for the coupled products 102 and 103. In the case of halichondrin B, the synthesis was completed by treatment of 102 with TBAF to remove the TBS groups and form the hemiketal between the C48 alcohol and C44 ketone. At this point, exposure to DDQ resulted in cleavage of the PMB ether, and finally treatment with (()-camphorsulfonic acid (CSA) resulted in spiroketal formation at C38 to give halichondrin B in 50-60% yield. An overview of the presumed intermediates in the sequence of 102 f 5 is also provided in Scheme 8. In summary,

Kishi’s ground-breaking synthesis of halichondrin B was completed by a 47-step longest linear sequence (from D-galactose gycal via 27).
The norhalichondrin B synthesis paralleled the final steps of the halichondrin synthesis and required only one further step to complete the synthesis, the hydrolysis of the C53 methyl ester, which was readily achieved with LiOH.

⦁ Subsequent Improvements by the Kishi Group
Although the chemistry by Kishi and co-workers delin- eated in the previous sections led to total syntheses of halichondrin B and norhalichondrin B, they have continued to refine the approach to a number of the fragments in the period since. These improvements are described below.
While the route to the C1-C13 subunit 23 described in
Scheme 1 was relatively efficient (31 steps, 4% overall yield), a more succinct approach is shown in Scheme 9a.33 This sequence commenced with L-mannoic-γ-lactone, which pos- sessed the appropriate stereochemistry for the C8-C11 stereocenters of halichondrin. Acetonide formation, reduction of the lactone to the C7 aldehyde with diisobutylaluminium hydride (DIBAL-H), and Wittig olefination led to methyl vinyl ether 111. Upon treatment with osmium tetraoxide followed by acetic anhydride, enol ether 111 was converted to 112 with a 16:1 preference for the desired stereoisomer. C-Allylation with allyltrimethylsilane provided exclusively the expected axial allyl glycoside 113 with a C6 configuration matching that of the natural product. Rh-catalyzed hydrobo- ration of 113, followed by a PCC oxidation34 led to hemiacetal 114. Subsequent Wittig olefination furnished the unsaturated ester, which underwent in situ heteroconjugate addition reaction to generate a 1:1 mixture of C3 epimers, which was readily equilibrated upon exposure to Triton B methoxide to yield 115. Selective cleavage of the less hindered acetonide liberated diol 116. Oxidative cleavage with NaIO4 was followed by a Ni(II)/Cr(II)-mediated addition reaction with vinyl iodide 117 and removal of the TBS groups with FeCl3/SiO2 to give 118. Silylation, ozonolysis, and Takai reaction completed the C1-C13 domain 23.
Due to difficulties performing the last five steps in Scheme
9a in an acceptable efficiency on large scale, an alternate route was developed. After an extensive study of the Ni(II)/ Cr(II)-mediated coupling as well as some refinement through- out the sequence, a 12-step synthesis of the C1-C13 segment was achieved (Scheme 9b).35 L-Mannonic-γ-lactone was again used as the starting material for the sequence, and protection of the two vicinal diols as cyclohexylidene ketals gave 119. Reduction of 119 to the lactol followed by Wittig homologation yielded methyl vinyl ether 120. Catalytic osmylation with the chiral ligand dihydroquinidine (DHDQ) as described by Sharpless,36 followed by acetylation, provided 121 as a 4-5:1 ratio at C7. In effort to improve material throughput, functionalized allylic silane 122 was used as an allylating reagent for the C6 position and provided 123 as a single diastereoisomer. When 123 was treated with Triton B methoxide, a four-step sequence ensued that consisted of
⦁ cleavage of the acetate, (ii) isomerization of the olefin into conjugation, (iii) cyclization by heteroconjugate addition, and (iv) equilibration at C3 diastereomers. This sequence provided the desired product, 124, in an impressive 87% overall yield. Selective hydrolysis of the cyclohexylidene ketal protecting the alcohols at C11 and C12, followed by oxidative cleavage, generated the C11 aldehyde. Ni(II)/Cr(II)-

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3053

Scheme 7. Initial Subunit Couplings from the Kishi Synthesis of Halichondrin B and Norhalichondrin B
Scheme 8. Completion of the Kishi Syntheses of Halichondrin B and Norhalichondrin B

mediated coupling using trans-ICHdCHTMS then produced
126 in good overall yield and with better than 15:1 diastereoselectivity at C11. It is also worth mentioning that
an improved workup for Ni(II)/Cr(II)-mediated couplings was also developed by Stamos and Kishi while optimizing this reaction.37 Rather than using the commonly employed

3054 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 9. Kishi’s Second Generation (a) and Third Generation (b) Approaches to the C1-C13 Subunit

aqueous NH4Cl, the more potent metal sequestering agent ethylenediamine was utilized, improving the yield of recov- ered material from 45% to 75%. Related studies also revealed the utility of 4-tert-butyl pyridine as a beneficial cosolvent that reduces homocoupling of the halide and also that the Na or K salts of serine can be used to facilitate workup of Ni/Cr reactions. Finally, in a one-pot reaction, the cyclo- hexylidene ketal of 126 was cleaved, and the resultant triol was TBS-protected. Iododesilylation with N-iodosuccinimide provided 23 with an 11% overall yield for the 12-step sequence.

An alternative approach to the coupling of the C14-C26 and C27-C38 domains that involves a heteroconjugate addition to establish stereochemistry at C23 has been reported (Scheme 10).38 Tetrahydrofuran 127 could be activated to triflate 128 and in the same pot reacted with the anion of 129 to produce 130. Horner-Wadsworth-Emmons reaction with 131 using the Roush-Masamune conditions39 yielded enoate 132 as an inconsequential mixture of diastereoisomers. NHK reaction between 132 and 133 gave an 4:1 mixture of diastereoisomers in 85-90% yield, and this material was converted to a 17:1 mixture of diastereoisomers by oxidation and reduction using the Corey-Itsuno system40 to finally produce 134. Conjugate addition under basic conditions gave 135, and removal of the ester by radical decarboxylation in two further steps produced the final compound, 96. Although slightly longer than the original approach, this route has some significant advantages with regard to stereocontrol around the C23-C27 pyran ring.
The latest synthesis of the C14-C26 segment is a concise and impressive display of the catalytic asymmetric Ni(II)/ Cr(II)-mediated reaction and Co(I)/Cr(II)-mediated reaction methodology developed recently in the Kishi lab (Scheme 11). The sulfonamide ligands 139 and 142 were designed to effectively control the stereochemical course of the two reactions. Additionally, using a strategy first described by Fu¨rstner, the NHK reaction was rendered catalytic with the
use of TMSCl (to facilitate turnover from the chromium alkoxide product) and Mn(0) (as a reducing agent for chromium).41 The first bond formation of the scheme was achieved via the catalytic asymmetric Ni(II)/Cr(II)-mediated coupling of 137 and 138 in the presence of 139. Treatment of the crude coupled product with PPTS/pyridine/isopropanol removed the resultant TMS ether formed as part of the reaction and also resulted in cyclization to form the tetrahy- drofuran ring. Following debenzoylation, 140 was isolated in 80% yield. The primary alcohol was then oxidized to the aldehyde with Dess-Martin periodinane, and subsequently an asymmetric Co(II)/Cr(II)-mediated coupling was used to forge the C23-C24 bond with 141 in the presence of ligand
142. It is noteworthy that the cobalt reaction selects for the alkyl iodide over the vinyl iodide.

One of the key steps of the halichondrin synthesis and also the synthesis of E7389 is formation of the trioxatricyclo[,7]decane (Scheme 12). As noted earlier, treatment of 21 with TBAF results in removal of the TBS groups and heteroconjugate addition to give 143 and C12- epi-143. Subsequent acid-catalyzed ketalization provided the desired compound 144, along with the undesired heterocon- jugate addition product C12-epi-143 in ratios of 5-6:1. Separation is possible, and C12-epi-143 can be recycled by treatment with TBAF and then PPTS to produce more material. In order to facilitate this process, an automated system was fabricated in which two columns were placed in sequence and connected to a pump.42 The first was an Amberlite IRA 400 methoxide column, which was expected to facilitate the retro-heteroconjugate addition process, fol- lowed by a Rexyn 101 acid column, which was expected to facilitate ketal formation (Figure 8). When this system was loaded with the crude material obtained by desilylation of 21 with TBAF and circulated for a period of 12 h, an excellent 95% yield of the desired compound 146 was obtained (Scheme 13). It is also worth noting the practical aspects of the system: for a 40 mg scale experiment,

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3055

Scheme 10. Kishi’s Heteroconjugate Addition Approach to the C14-C38 Domain

Scheme 11. Kishi’s Catalytic Asymmetric NHK and Alkyl Co Addition Reaction Approach to the C14-C26 Subunit

approximately 0.4 cm3 of Amberlite IRA 400, 0.4 cm3 of Rexyn 101, and 0.1 cm3 of alumina were placed in each column. The total volume of solvent was ca. 4 mL (c ) ca.
0.01 M) and the flow-rate was ca. 2 mL/min.

Scheme 12. Intermediates and Side Products in the Formation of the Trioxatricyclo[,7]decane

Kishi has also described an opertaionally simple nonaque- ous workup for TBAF-mediated desilylation in the context of the formation of the polycyclic acetal.43 Although framed

Figure 8. The Kishi ion-exchange system for formation of the polycyclic acetal: (a) Amberlite IRA 400 (OMe) column; (b) Rexyn 101 (H+) column; (c) basic Al2O3 (Baker) filter with glass wool dividers; (d) septum; (e) Teflon connector tube; (f) Teflon tubing. Pump ) FMI QG50. Figure taken from ref 42. Copyright 2004 American Chemical Society.
Scheme 13. Application of the Kishi Ion-Exchange System for Formation of the Polycyclic Acetal

3056 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Figure 9. The Phillips strategy for norhalichondrin B with key disconnections.

in the context of halichondrin synthesis, there is no doubt that this protocol should be of much broader application. Many of the Kishi group’s developments in the context of catalytic asymmetric Nozaki-Hiyama-Kishi reactions have also been framed in the context of halichondrin synthesis, and although not exhaustively covered here beyond the examples given, there is no doubt that they will have broader applications.44
Although the improvements described in this section would not affect the synthesis by the metric of longest linear sequence, they have made significant inroads into the total step count and also with regard to efficiency. Perhaps most importantly, these developments are of direct relevance to the important question of brevity in the context of E7389 (see section 8).

⦁ Total Synthesis of Norhalichondrin B by Phillips and Co-workers
A second total synthesis of norhalichondrin B was reported in 2009 by Phillips and co-workers.45 The overall plans for the synthesis are summarized in Figure 9. The completion of the synthesis was planned around a late-stage Horner- Wadsworth-Emmons coupling of C1-C39-containing phos- phonate 149 with the C40-C53 domain aldehyde 147, followed by spiroketal formation. The C40-C53 domain could be traced back to furfural derivative 148. Further deconstruction of 149 via 150 yielded C1-C13 domain 151, C14-C26 domain 153, and C27-C38 domain 155. The C1-C13 domain could be traced back to furan (152), the C14-C26 domain to §-keto ester 154, and the C27-C38 domain to furfural derivative 148.

⦁ C1-C13 Subunit Synthesis
The synthesis of the C1-C13 domain commenced with the Davies Rh-catalyzed addition of diazo ester 156 to furan 152 to give oxabicyclo[3.2.1]octene 157 (Scheme 14).46 This ester was advanced by a four-step sequence to 158, and when 158 was exposed to 3 mol % of Grubbs’ second generation catalyst47 conversion of the bridged bicyclic structure to pyranopyran 159 readily occurred in 71% yield. Selective hydroboration of the terminal olefin with Sia2BH gave 160 and was followed by simultaneous oxidation of the acetal
Scheme 14. Phillips’ C1-C13 Subunit Synthesis

to the lactone and the alcohol to the acid, which was methylated with TMSCHN2, producing 161. Introduction of the C8 and C9 alcohols was achieved by dihydroxylation, and subsequent protection produced 162. The remaining three carbons were introduced by a four-step sequence that began with selective olefination of the lactone with the Petasis reagent. Hydroboration-oxidation of the enol ether provided 163, which was oxidized to the aldehyde and subjected to a Nozaki-Hiyama-Kishi reaction with vinyl iodide to give 151.
An alternative sequence to this subunit that bisects the route above at compound 161 has also been described.48 Furfuryl alcohol 164 (readily obtained in two steps from furfural by addition of butenylmagnesium bromide followed by Sato’s kinetic resolution49) was subjected to Achmatowicz oxidation to give pyranone 165 (Scheme 15). Benzoylation, reduction, and cross metathesis with methyl acrylate in the presence of the Hoveyda-Grubbs catalyst 167 yielded 168. The second pyran ring was formed by TBAF-induced heteroconjugate addition, and the sequence was completed by a Grieco type oxidation50 to yield lactone 161.

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3057

Scheme 15. Phillips’ Alternative Approach to Lactone 161

Scheme 16. The Phillips Synthesis of the C14-C26 Domain

⦁ C14-C26 Subunit Synthesis
The C14-C26 subunit was prepared by a sequence that began with Noyori hydrogenation of §-keto ester 154 and subsequent Pd-mediated allylation to give O-allyl ester 170 (Scheme 16). This ester was readily converted to diazoketone 171 and when exposed to Cu(acac)2 in THF under reflux, the expected [2,3]-rearrangement occurred to yield 2,5-anti- tetrahydrofuran 172.51 Wittig olefination led to diene 173, which was selectively hydroborated with Sia2BH at the terminal olefin. Oxidation gave aldehyde 174 and the balance of the carbons required for this subunit were introduced by the Kishi asymmetric Co/Cr addition process with 175 in the presence of 176, which after desilylation of the product produced diol 177.52 Selective acylation of the primary alcohol with pivaloyl chloride and formation of the mesylate completed the synthesis of the C14-C26 fragment.
⦁ C27-C38 and C40-C53 Subunit Syntheses
The syntheses of both the C27-C38 domain and the C40-C53 domain were based around the earlier reported process for the conversion of furans to 2,6-syn-pyranones.53 In the case of the C40-C53 domain 147, the synthesis began with furfural 148, which was subjected to Brown crotylation using (-)-Ipc2-(E)-crotylborane to produce 178 (Scheme 17).
Scheme 17. The Phillips Synthesis of the C14-C26 Domain

Achmatowicz oxidation54 produced an intermediate pyranone hemiacetal 179, which was immediately subjected to trif- luoroacetic acid-mediated ionic hydrogenation using Et3SiH to give pyranone 180 in 86% yield and as a single diastereomer.55 Removal of the TBS protecting groups under acidic conditions was followed by a Jones oxidation and in situ heteroconjugate addition of the acid to the enone to produce pyranolactone 181. The final pyran stereocenter was introduced by reduction of the ketone with NaBH4 to arrive at 182 and a straightforward four-step sequence advanced material to aldehyde 184. Addition of the lithium anion derived from iodide 185, followed by Dess-Martin oxida- tion, and ozonolysis of the olefin gave the fully functionalized C40-C53 domain, 147.
Furfural 148 also served as the starting material for the C27-C38 domain 155 (Scheme 18). In this case, Brown crotylation of 148 with (-)-(Ipc)2-(Z)-crotylborane gave 186 and was followed by the same general processs of Achma- towicz oxidation and ionic hydogenation to give pyranone 188 as a single diastereomer. This material was taken forward by an analogous three-step process to that employed above to produce alcohol 190. Reduction to the triol, selective formation of the seven-membered ketal, and protection of the secondary alcohol as the TES ether gave 191. Ozonolysis of the olefin gave aldehyde 192, which was subjected to an asymmetric Nozaki-Hiyama-Kishi reaction with methyl- trans-3-iodoacrylate in the presence of Kishi’s oxazoline- sulfonamide ligand 193 to give 194 in 75% yield (d.r. ) 12:1). The impressive diastereoselectivity of this process meant that the tedious separation of the alcohol diastereoi- somers was minimized. Protection of the alcohol as the PMB ether was followed by a one-pot, two-step process that involved removal of the TES group with TBAF and heteroconjugate addition of the liberated alcohol to produce pyranopyran 195. Four standard transformations produced the fully functionalized C27-C38 domain, 155.

3058 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 18. The Phillips Synthesis of the C27-C38 Domain Scheme 19. Initial Subunit Couplings of the Phillips
Synthesis of Norhalichondrin B

⦁ Subunit Couplings and Completion of the Synthesis

The first subunit coupling was the connection of pyran- opyran 155 and tetrahydrofuran 153, which were unified by the well-established combination of Nozaki-Hiyama-Kishi reaction and pyran ring formation by SN2 reaction (see Scheme 19)56 to give 197 in 59% yield. The diastereoselec- tivity of this process was 3.7:1, and although the dia- stereisomers could not be separated at this point, separation was possible at the point of 202 f 203 (see Scheme 20). A four-step sequence of (i) LiAlH4-mediated pivalate removal,
⦁ Dess-Martin periodinane oxidation, (iii) addition of vinylmagnesium bromide, and (iv) Dess-Martin periodinane oxidation provided enone 198. At this juncture, it was possible to introduce the C1-C13 domain 151 by cross metathesis using 20 mol % of catalyst 19957 to give 150 in 62% yield.
Steps toward the completion of the total synthesis com- menced with treatment of cross metathesis product 150 with AcOH buffered TBAF (Scheme 20). Removal of the silyl protecting groups and heteroconjugate addition under these conditions produced intermediate tetrahydrofuran 200. When the reaction mixture was subjected to nonaqueous workup conditions (CaCO3, DOWEX 50WX8-400, MeOH)58 ketal formation also occurred to form the desired 2,6,9-trioxatricyclo- [,7]decane ring system directly, providing 201 in 64% yield, along with 26% of the intermediate tetrahydrofuran 200 in which the C12 stereocenter is epimeric to the desired stereochemistry. A three-step sequence of protecting group chemistry provided seco-acid 204, which readily lactonized under standard Yamaguchi conditions5 to give macrolactone
205. After removal of the primary TBS group (205 f 206) and oxidation of the alcohol to the aldehyde, §-ketophos- phonate 149 was formed by Roskamp reaction with di- methyl(diazomethyl)phosphonate 207 in the presence of SnCl2.59 The final subunit was then introduced by Horner- Wadsworth-Emmons coupling between 147 and 149 using K2CO3 and 18-crown-6 in warm toluene to yield enone 208 in 83% yield. Treatment of enone 208 with TBAF resulted in removal of the silyl protecting groups to give an intermediate that contained the C44 spiroketal. In contrast to Kishi’s two-step DDQ then CSA approach, the removal of the PMB ether with DDQ in CH2Cl2-MeOH (10:1) also resulted in direct formation of the spiroketal, yielding norhalichondrin B methyl ester, 209. The synthesis was then completed by hydrolysis of the methyl ester to yield norhalichondrin B.
In summary, the Phillips synthesis of norhalichondrin B proceeds in 37 steps longest linear sequence from com- mercially available §-furylethanol (converted in two steps to 148). Noteworthy features of the synthesis include the furan-pyranone and tandem metathesis strategies for the synthesis of pyranopyrans and the application of cross metathesis and Roskamp reactions on highly complex substrates.

⦁ Synthetic Work toward Halichondrin B by Horita and Yonemitsu
Along with the Kishi and Salomon groups, Horita and Yonemitsu were among the early investigators with respect to halichondrin synthesis. To date they have completed syntheses of the two advanced models that are directed

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3059

Scheme 20. Completion of the Phillips Synthesis of Norhalichondrin B

toward halichondrin B (the macrolactone domain 210, Figure 10, and the polyether domain 217, Figure 11). Although a total synthesis has not been completed, the ordering of subunit assembly based on intermediates available from these studies has been proposed.

Figure 10. An overview of the Horita and Yonemitsu plans for the macrolactone domain of the halichondrins.
⦁ C1-C13 and C1-C15 Subunit Synthesis
The initial Horita-Yonemitsu synthesis of the C1-C13 segment was centered around the construction of pyran rings via heteroconjugate addition reactions.60 The synthesis com- menced with D-glucose diacetonide (213, Scheme 21), which was readily advanced by standard transformations to epoxide
223. When 223 was exposed to acetic acid, 5-exo cyclization with loss of the PMP acetal occurred to provide tetrahydro- furan 224. A series of manipulations was then used to differentially protect the C13 and C11 hydroxyl groups,

Figure 11. Horita and Yonemitsu’s plans for the polyether domain of halichondrin B.

3060 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 21. The Horita-Yonemitsu Synthesis of the C1-C13 Domain

unmask the C8 aldehyde via dithiane hydrolysis, and generate the R,§-unsaturated enoate 226 by Horner-Wadsworth- Emmons reaction. Subsequent DIBAL reduction and Sharp- less asymmetric epoxidation provided epoxide 227, which could be ring-opened (via the carbamate using Roush’s method61) to provide 228. A 14-step sequence of largely protecting group manipulations produced 233, the precursor for the first pyran-forming heteroconjugate addition reaction. Treatment of 233 with TBAF resulted in TES deprotection and subsequent cyclization to form 234. Six further steps produced 236, which cyclized upon treatment with TBAF. After protection of the primary alcohol as the trityl ether, 237 and 3-epi-237 were obtained in a 2:1 ratio. Exposure to TBAF for extended periods allowed for equilibration and produced a 19:1 ratio of the desired 2,3-cis- to the undesired 2,3-trans-disubstituted tetrahydropyrans.
An alternative synthesis of this domain, plus a further two
carbons has also been reported (Scheme 22).62 It commences with tetrahydrofuran 228, an intermediate prepared by the previous route. Protecting group manipulations provide 238 in four steps, and this allows for homologation of the primary alcohol, which is achieved by tosylation/cyanide displace- ment of the tosylate, and then a two-step reduction to yield
239. A sequence of nine steps, which includes the inversion of the C7 stereochemistry by a standard oxidation-reduction protocol, led to enoate 243, the precursor for formation of the pyran ring by conjugate addition. Desilylation with TBAF was accompanied by the desired addition, yielding 244 in close to quantitative yield. Reduction of the ester and homologation of the primary alcohol (again by tosylate formation and displacement with cyanide) provided nitrile
245. Reduction and Wittig olefination advanced material to
246, which was readily cyclized to the pyranopyran by removal of the acetonides with CSA and then treatment with TBAF. Reprotection of the C14 primary alcohol as the trityl ether, and equilibration of C3 stereochemistry by treatment with TBAF gave 247. After reduction of the ester to the alcohol and protection as the TBS ether, the introduction of the §-ketophosphonate was achieved by standard methods, providing 212 in six further steps.

⦁ C16-C26 Subunit Synthesis
Because their plans for subunit coupling in the C26-C28 region centered around an aldol reaction and then pyran formation, Yonemitsu and Horita chose to prepare two C16-C26 segments, 260 and 265, epimeric at C23 (Scheme 23).63,64 This approach allowed flexibility of planning for formation of the pyran ring in either direction by either SN2- type inversion at C23 or reduction of an acetal at C27, which would be expected to provide 2,6-syn stereoinduction. The syntheses of both 260 and 265 began with allylic alcohol 251, obtained in seven steps from the Roche ester, 216. Use of Sharpless epoxidation conditions followed by Red-Al reductive epoxide opening allowed access to both C23 epimers, 252 and 261. In the case of final target 260, 252 was advanced by PMP-acetal formation, regioselective reductive cleavage to the C21 alcohol, and Swern oxidation to provide aldehyde 253. Horner-Wadsworth-Emmons reaction between 253 and phosphonate 254 (derived from (R)-malic acid in four steps65) gave 255. Reduction of 255 with lithium aluminum hydride in the presence of lithium iodide at -100 C under chelation-controlled conditions66 proceeded stereoselectively to give the allyl alcohol in near

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3061

Scheme 22. Horita and Yonemitsu’s Second Generation Approach to the C1-C13 Domain and Homologation to the C1-C15 Domain
Scheme 23. Horita and Yonemitsu’s Approach to the C16-C26 Subunit

quantitative yield, and removal of the acetonide provided triol 256. The tetrahydrofuran ring was introduced by an iodoetherification process, whereby the treatment of 256 with iodine effected a completely stereoselective iodoetherification to form 257. Removal of the iodide by elimination to the olefin and treatment with Raney Ni/H2(g), followed by benzoyl protection of the primary alcohol, provided 258. Swern oxidation and methylenation under Wittig conditions provided 259, which was converted in seven further steps to the C16-C26 subunit 260. In a similar vein, the Horner-Wadsworth-Emmons coupling between 254 and the C23 epimeric aldehyde 262 gave 263, which was processed on via iodoetherification to the C23-epimeric C16-C26 subunit 265.
⦁ C27-C36 Subunit Synthesis
The synthetic route to the C27-C36 fragment is presented in Scheme 24 and features a stereoselective C-glycosidation at C29.67,68 The sequence begins with the conversion of dimethyl tartrate 215 in five standard steps to alcohol 266.69 Swern oxidation of 266, Z-selective Horner-Wadsworth-Emmons reaction following the pro- cedure of Still and Gennari,70 removal of the ketal, and protection of the primary alcohol as the TBDPS ether furnished 268. Exposure of 268 to p-TsOH provided an intermediate lactone, which was reduced with DIBAL and immediately methylated to give the methyl glycoside. Oxidative removal of the PMB ether afforded allylic alcohol

3062 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 24. The Horita-Yonemitsu Synthesis of the C27-C36 Subunit

269. Epoxidation of 269 with m-CPBA directed by the C32 alcohol gave epoxide 271. No epoxidation occurred in the absence of the radical scavenger 270. Selective trans-diaxial opening of epoxide was achieved with a supernatant mixture of methylmagnesium chloride and salt-free methyllithium in Et2O-THF (presumed to be “Me2Mg”71) to produce 272. Protecting group manipulations and formation of the C34 tosylate led to 273. C-Glycosidation with allyltrimethylsilane in the presence of boron trifluoride etherate was attempted with 273 but gave a mixture of R and § diastereomers, necessitating postponement of the C-glycosidation for a number of steps. Conversion of 273 to allylic alcohol 274 was achieved by cyanide displacment of the tosylate and then a reduction-Wittig reaction-reduction sequence. Sharpless epoxidation, followed by opening of the epoxide by §-elim- ination and acylation introduced the C37 stereocenter (274 f 275 f 276), and four further steps were employed to arrive at 277. In contrast to 273, allylation72 of 277 was completely R-stereoselective and high-yielding, although it was accompanied by loss of the TBS groups, which were readily reintroduced to provide 279. In four further steps, the protecting groups at C35 and C36 were exchanged, and the olefin was oxidatively cleaved to provide aldehyde 280. Interestingly, Horita and Yonemitsu have reported that 280 has in Vitro activity against KB and L1210 cells, with IC50 values of 3.9 and 3.1 nM respectively.73

⦁ C37-C54 Subunit Synthesis
Two separate syntheses for the C37-C54 subunit have been reported by Horita and Yonemitsu.74,75 The first synthesis began with L-(+)-dimethyltartrate, which was converted to 266 in the same fashion as was employed for the C27-C36 domain (Scheme 25). One carbon homologation produced 281, and a sequence of five steps provided diol 282. When Y was subjected to Sharpless asymmetric epoxidation with (-)-DET, Ti(i-PrO)4, and TBHP, epoxidation with in situ opening of the epoxide occurred to yield tetrahydrofuran 283. Manipulation of the diol by protection of the primary alcohol and activation
of the secondary alcohol as the mesylate allowed for conversion of 283 to epoxide 284. Opening of the epoxide with vinylmagnesium bromide in the presence of CuI provided homoallylic alcohol 285, which was readily converted to the C51-C54 triol domain 287 by a sequence of five steps that concluded with a Prasad reduction (286 f 287). Four steps advanced 287 to the germinal dibromide 288, which was readily converted to the acetylene by treatment with LDA. Exchange of the C54 pivalate ester for a TBDPS group and formation of the alkynoic ester produced 289. A further four steps, which included introduction of the C46 methyl group by heterogeneous hydrogenation from the exo face of 291, led to key lactone 292. This lactone was reacted with the organolithium derived from 299 (synthesized in 20 steps from (+)-diethyl tartrate as shown) in the presence of CeCl3 to yield hemiacetal 293. Simultaneous removal of the acetonide and formation of the C44 ketal was possible upon treatment with p-TsOH and gave alcohol 294. Four standard transformations completed the synthesis of a C37-C54 domain that is appropriately functionalized for coupling with a C1-C36 domain.
The second synthesis of the C37-C54 domain is shown in Scheme 26 and leverages the observation that there is local symmetry in the C38-C50 region of the molecule. The starting material was again 266, which was converted to the γ-alkoxy-R,§-unsaturated ester 300 by oxidation and olefination. Following Hanessian’s protocol,76 lithium dimethylcuprate was used to stereoselectively afford 301 with 3,4-anti-stereochemistry (d.r. ) 14:1). After conversion to the §-ketophosphonate 302, a second enone was then installed by Horner-Emmons coupling to form 303. The second conjugate addition of lithium dimethylcuprate without TMSCl cleanly provided the desired ketone 304 in excellent yield and with better than 25:1 diastereoselectivity. When 304 was exposed to 6 N H2SO4 in THF, removal of the pentylidene acetals and subsequent spiroketalization pro- ceeded gradually to yield the C2-symmetrical spiroketal 305 as a single diastereoisomer in 75% yield. Three further steps

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3063

Scheme 25. The Synthesis of a Fully Functionalized C37-C54 Domain by Horita and Yonemitsu

were required to effect the one-carbon homologation of the primary alcohols to yield 306. At this juncture, desymme- trization was necessary; however all attempts to monoprotect 306 with TBSCl following known literature procedures met with failure. However, when the dipotassium salt of 306 (generated with 2.0 equiv of t-BuOK in THF at -78 C) was trapped with 1.1 equiv of TBSCl, the desired mono- protected silyl ether 307 was obtained in 45% yield with only a trace of the disilyl ether. Furthermore, the unreacted diol could be recovered chromatographically and resubjected to the reaction conditions to give 90% yield of 307 through three recycles. Three further steps converted alcohol 307 to allylic alcohol 308, and when treated with iodine in the presence of NaHCO3, 308 underwent iodoetherification to provide the tricyclic iodohydrin 309 as a single diastereo- isomer. Subsequent exposure to potassium tert-butoxide afforded the epoxide, which was then opened with vinyl- magnesium bromide in the presence of copper(I) iodide to give 310. In five conventional steps, 310 was converted to
§-hydroxy ketone 311, which was reduced under Prasad
conditions77 and protected to form 312. Five more steps were required to form phosphonate 218 from 312 (see Scheme 25).
⦁ Subunit Couplings
Although a total synthesis has not been reported, Horita and Yonemitsu have undertaken studies on key subunit couplings, and the products of these studies are completed syntheses of both the right-hand macrolactone 210 and the left-hand polyether 217. In the context of the synthesis of the macrolactone, the first subunit coupling examined was the connection of the C27-C36 and C16-C26 subunits.78 Deprotonation of 260 with LDA and aldol reaction with 280 proceeded smoothly and stereoselectively to afford adduct 313 (Scheme 27). Dess-Martin oxidation of the resultant secondary alcohol and protecting group exchange at the C35/C36 1,2-diol (acetonide f acetates) provided 314. Treatment of 314 with Et3SiH in the presence of BF3 · OEt2 resulted in a sequence of events that included removal of the PMB ether, oxacarbenium ion formation and reduction, and also removal of the SEM ether. After protection of the primary alcohol as the TBDPS ether, 315 was obtained in an excellent 87% yield. After the C35 and C36 acetates were replaced by an acetonide, the ester at C26 was converted to the olefin in three routine steps (315 f 316). In five additional steps, the TBDPS protected alcohol 316 was converted to aldehyde 214. Horner-Wadsworth-

3064 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.
Scheme 26. Horita and Yonemitsu’s Second Generation Synthesis of the ∼C37-C54 Domain
Scheme 27. Synthesis of the Macrolactone Domain by Horita and Yonemitsu

Emmons reaction between 212 and 214, followed by Stryker reduction of the resultant enone, provided 211. Careful treatment of 211 with 1 N HCl deprotected selectively the
C8 and C11 TES groups and the C1 primary TBS group. Under these conditions, the C28 and C30 secondary alcohols remained TBS protected. Exposure of the resultant triol to

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3065

Scheme 28. Horita and Yonemitsu’s Polyether Domain Synthesis

pyridinium p-toluenesulfonate prompted intramolecular ketal ring formation between the C8 and C11 hydroxyl groups and the C14 carbonyl to provide 317 in 80% yield. In a two- step procedure, the primary alcohol was converted to the acid by treatment with Dess-Martin periodinane and expo- sure to Lindgren-Pinnick conditions.79 Treatment with TBAF then deprotected the two secondary TBS groups to reveal seco-acid 318, which could be cyclized under standard Yamaguchi conditions to give macrolactone 210, albeit in somewhat modest 35% yield. It is, however, noteworthy that the macrolactonization can occur in the presence of a free alcohol at C32.
The construction of the left-hand polyether domain began with a Horner-Wadsworth-Emmons reaction between aldehyde 280 and the lithium anion of phospho- nate 218 to provide 319 (Scheme 28).80 On treatment with TBAF, 319 underwent an intramolecular cyclization to give a 3:1 mixture of diastereoisomers that unfortunately favored the undesired product, 36-epi-320. Thankfully, this mixture could be converted to 320 by treatment with potassium carbonate in methanol at 80 C. Removal of the PMB ether was achieved with DDQ, and the resultant triol underwent intramolecular spiroketalization when treated with camphor- sulfonic acid. The spiroketal formation was not diastereo- selective (d.r. ) 1:1) and after acetylation 321 was obtained. In order to elongate the C29 position by two carbons, the benzyl ether was deprotected by hydrogenation, and the resultant hemiacetal was homologated using a Wittig reac- tion. At this juncture, the C38 stereocenter was still a 1:1 mixture, but it was discovered that treatment with CSA at this point converted the mixture to a single diastereoisomer,
322. In two final steps, 322 was treated with TBAF to promote the heteroconjugate addition, and the resultant
polyether ring system was deprotected with potassium carbonate in methanol to yield 217.

⦁ Synthetic Work toward Halichondrin B by Salomon
Salomon and co-workers have synthesized several seg- ments of halichondrin B from a variety of inexpensive and commercially available starting materials: a C1-C15 seg- ment from D-ribose, a C1-C21 segment, 325, from D-ribose and D-glucose, a C27-C35 segment, 324, from D-glucose, and a C37-C51 segment, 326, from D-mannitol (Figure 12).

Figure 12. Key subunits and starting materials of the Salomon synthetic studies.

3066 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 29. The Initial Stages of Salomon’s Synthesis of the C1-C15 Domain

⦁ C1-C15 Subunit Synthesis81
The starting material for Salomon’s synthesis of the C1-C15 segment was D-ribose, which possesses the correct stereochemistry for the C7, C8, and C9 positions of hali- chondrin, as well as an appropriately positioned aldehyde for construction of one of the pyrans. Oxidation of ribofura- noside 327 at the C10 position and Wittig olefination provided access to the unsaturated ester. An early variant of the Sharpless asymmetric dihydroxylation was then used to provide 328 with 2.5:1 diastereoselectivity (Scheme 29).82 At this point, it was necessary to invert the stereochemistry at the C11 position, and this was accomplished by conversion to the cyclic sulfate followed by a regioselective nucleophilic substitution with tetrabutylammonium benzoate.83 Deben- zoylation, acetonide formation, and partial reduction of the ester with DIBAL then provided the C12 aldehyde 330. Subsequent Wittig olefination with phosphorane 331 afforded the R,§-unsaturated enone 332.
Trifluoroacetic acid-mediated acetonide hydrolysis and treatment of the resulting tetraol with Dowex 50W (a stronger acid) generated the highly functionalized acyclic 333 via opening of the furanoside (Scheme 30). Subsequent forma- tion of the furopyran ring system ensued, and after acety- lation, 334 was obtained, albeit with the incorrect stereo- chemistry at C12. All attempts to epimerize the C12 stereochemistry at this point met with failure, and it was reasoned that epimerization might be possible at a later stage in the synthesis. To continue the sequence, tetraacetate 334 was allylated selectively by a trityl perchlorate-catalyzed axial addition reaction with allyltrimethylsilane84 to give 335. Three further steps then converted 335 to the R,§-unsaturated ester 336. Treatment of 336 with sodium methoxide resulted in transesterification and partial epimerization of the C12 stereocenter to a 2:1 mixture favoring the desired compound,
337. Upon treatment of the mixture with PPTS, ketalization
of 337 gave the polycyclic acetal 338. Fortunately, 338 was readily separable from C12-epi-337,which upon reexposure to sodium methoxide provided more 337 by epimerization. Finally, when 338 was subjected to Triton-B methoxide, intramolecular heteroconjugate addition occurred to close the final pyran ring and complete the synthesis of the C1-C15 domain, 339.

Scheme 30. Completion of Salomon’s C1-C15 Subunit Synthesis

⦁ C1-C21 Subunit Synthesis85
The C1-C21 fragment of halichondrin B was prepared as shown in Scheme 31. The sequence began with 340, readily accessed from D-glucose in three high-yielding steps on molar scale. Selective deprotection of the C16-C17 acetonide and tosylation of the resulting diol gave 341. Acid- catalyzed transketalization and tetrahydrofuran formation then provided 342. Deoxygenation was then carried out under Barton conditions,86 and four further steps provided inter- mediate 343. Reaction of 343 with enolate 344 and subse- quent Wittig reaction with 346 gave C16-C21 segment 347. Selective removal of the PMB group was accomplished with ceric ammonium nitrate, and subsequent exposure to mild acid resulted in acetonide and silyl group cleavage. Treatment with Triton B followed by acylation of the crude product gave 348. Conversion of 348 to 325 was carried out in similar fashion to the analogous conversion of 334 to 339 in the C1-C15 synthesis (Scheme 30), involving axial allylation and reaction of 349 with sodium methoxide to form a tetraol. However, the ordering of steps had to be modified to perform the heteroconjugate addition in advance of the C14 ketal- ization. When done in the opposite order, a significant amount of the known furan derivative was formed. After silylation of the remaining C21 hydroxy group, exposure to Triton B methoxide resulted in C3 equilibration to yield the desired C1-C21 segment 325.
⦁ C27-C35 Subunit Synthesis87
Scheme 32 outlines the synthesis of the C27-C35 segment 324, for which D-glucose was used to as provide much of the framework of the key pyran ring. Selective masking of

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3067

Scheme 31. The Salomon Synthesis of the C1-C21 Domain

Scheme 32. The Salomon C27-C35 Subunit Synthesis

all but the trans alcohols at C30 and C31 followed by tosylation and sodium hydride-mediated epoxide formation provided 350. Regioselective opening of the epoxide by axial attack of MeMgCl provided the undesired methyl group stereochemistry. However, oxidation of the C30 hydroxyl
Scheme 33. The Salomon C37-C51 Subunit Synthesis

group and subsequent C31 epimerization via the enol provided the thermodynamically more favored (99:1) equatorial methyl group (350 f 351). Axial reduction at C30 and hydrolysis of the protecting groups provided 352, which was reprotected as the furanose bis-acetonide to give 353. Deprotection of the C33-C34 acetonide and oxidative cleavage of the diol destroyed the incorrect C33 stereocenter, providing aldehyde 354. TiCl4-mediated and chelation- controlled reaction of aldehyde 354 and a silylketene thioacetal proceeded stereoselectively (99:1) to afford thioester 355. Base-induced hydrolysis of the thioester and acid-mediated acetonide cleavage was accompanied by conversion of furanose back to pyranose as well as in situ lactonization, yielding 356. Finally, Wittig olefination pro- duced 357, and heteroconjugate addition reaction of the intermediate R,§-unsaturated ester followed by protection of the secondary alcohol with p-methoxybenzyl trichloroace- timidate under acidic conditions delivered the target lactone 324.

⦁ C37-C51 Subunit Synthesis88
The Salomon synthesis of the C37-C51 segment is shown in Schemes 33 and 34. In designing a synthesis of this domain, Salomon and co-workers noted the C2 symmetry of the C37-C51 fragment and predicted that a thermody- namically controlled stereoselective spiroketalization could

3068 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 34. The Salomon C37-C51 Subunit Synthesis, With Terminus Differentiation

be used as the final synthetic step. Both the presence of two anomeric effects and the preference of the C42 and C46 methyl groups for equatorial disposition were invoked as rationale for the expected course of the reaction. Beginning with D-mannitol, acid-catalyzed cyclization gave tetraol 358 (Scheme 33). In six steps, 358 was converted to 359. Reductive cleavage of 359 with DIBAL generated a 6:4 regioisomeric mixture of 360 and 361, and either could be quantitatively recycled back to the starting acetal 359 by treatment with DDQ under anhydrous conditions. Compound 361 was then oxidized under Swern conditions to provide aldehyde 362, and Horner-Wadsworth-Emmons reaction furnished the R,§-unsaturated acyl silane 364, which was diastereoselectively methylated in 1,4 fashion on treatment with Me2Cu(CN)Li2 in the presence of TMSCl89 to provide
365. In three further steps, acyl silane 365 was converted into the corresponding phosphorane 366, which was then condensed with aldehyde 362 to form enone 367. A second diastereoselective methyl addition was subsequently used to form the key C2 symmetric ketone. It is of particular note that (i) the R,§-unsaturated acyl silane 364 was required, because no reaction was observed with the corresponding R,§-unsaturated ester, (ii) although the corresponding R,§- unsaturated methyl ketone underwent conjugate addition, the aldol addition of the product ketone with aldehyde 362 did not generate enone 367, and (iii) the stepwise approach was necessary, because dienones in this series tended to be unreactive toward cuprates. To conclude the sequence, ceric ammonium nitrate was used to oxidatively cleave the PMB ethers with concomitant spiroacetalization to afford the targeted compound 369.
Despite the development of a route to the C2-symmetric subunit of C37-C51, it was of pivotal importance to break the symmetry of the system in order to incorporate the fragment into a synthesis of halichondrin B (Scheme 34). For the terminus differentiation of the C37-C51 segment, Salomon and co-workers returned to both regioisomers 360 and 361 obtained from the reductive cleavage of 359. Thus 360 was converted into differentially protected aldehyde 370 through a five-step sequence involving selective hydro- genolysis of a benzyl ether in the presence of a p- methoxybenzyl ether with Raney nickel. Wittig coupling of the silyl protected aldehyde 370 with previously prepared phosphorus ylide 366 provided the enone, and subsequent conjugate methylation afforded ketone 371, now terminus differentiated. Deprotection of the p-methoxybenzyl ether with DDQ and subsequent TBAF-mediated TBS deprotection

Scheme 35. The Burke Pinacol Rearrangement Approach to the C1-C15 Domain

delivered a triol that afforded diastereoisomerically pure spiroketal 326 upon treatment with 1% HCl in THF.
Despite the significant progress made in the period 1989-1993, no further work has been published by the Salomon group.

⦁ Synthetic Work toward Halichondrin B by Burke
Burke and co-workers have synthesized a variety of halichondrin B fragments and have also examined some subunit couplings. A key feature of their studies has been to exploit elements of local symmetry within the molecule.

⦁ C1-C15 Subunit Synthesis
The synthetic route to the C1-C15 segment, which began with commercially available carbohydrate D-glycero-D-gluco- heptono-γ-lactone, 372, is shown in Scheme 35.90 With the correct stereochemistry for the C8, C9, and C10 positions, lactone 372 would require only inversion at C11. Regiose- lective bis(ketalization)91 of 372 with 3-pentanone provided 373; this result is in contrast with bis(acetonide) formation, which is known to proceed with different regioselectivity.92

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3069

Scheme 36. Burke’s Ozonolytic Desymmetrization Approach to the C15-C15 Domain

Oxidation to the C11 ketone and chelation-controlled reduc- tion93 with Zn(BH4)2 gave the epimeric C11 alcohol, which now had the correct stereochemistry for the halichondrins and was subsequently protected as the TBS ether to provide
374. The R-alkoxyorganolithium reagent derived from tin-lithium exchange on stannane 375 was then added to lactone 374 to generate diol 376. The secondary alcohol of 376 was converted to the mesylate 377, which underwent a pinacol-type rearrangement94 on exposure to ethylmagnesium bromide under heating conditions. Stereoselective reduction of the resultant pyranone with DIBAL led to alcohol 378. Ozonolysis followed by reductive workup and in situ Wittig olefination provided the (E)-enoate 379 and exposure to Triton B methoxide converted 379 exclusively to 380 via a heteroconjugate addition/equilibration process. Removal of the terminal acetal and oxidative cleavage gave aldehyde 381, which could be olefinated with phosphorane 331 to furnish enone 382. Spiroketalization with aqueous HF then provided 383.
A second synthesis of a model of the C1-C15 segment of halichondrin B has also been reported by the Burke group and features a novel ozonolytic desymmetrization of a C2- symmetric diol and the use of a C14-C15 enol ether as a precursor to the trioxatricyclodecane.95 The starting material for the second synthesis was (+)-conduritol E, 384, which was silylated to produce 385 (Scheme 36). Exposure of 385 to ozone resulted in the generation of dioxabicyclo[3.3.0]octane bearing a peroxyacetal at C12, which readily underwent elimination [with concomitant oxidation of the adjacent carbon] upon treatment with acetic anhydride and triethy- lamine to furnish hemiacetal 387 as a 2:1 mixture of anomers. Of particular mention in this desymmetrization protocol is the combination of Criegee’s tactic of trapping carbonyl oxides with internal nucleophiles and Schreiber’s ozonolytic desymmetrization of simple cycloalkenes.96 Protection of 387 with TESCl yielded the lactone 388 as a single diastereo- isomer, and subsequent treatment of 388 with the Petasis reagent resulted in olefination to provide 389. Hydroboration of the resultant exo-olefin gave primary alcohol 390 in 70% yield.
Alcohol 390 was protected with a pivalate group, and the TES ethers were cleaved in a one-pot procedure to form hemiacetal 391 (Scheme 37). At this point, the Kishi strategy
Scheme 37. Burke’s C1-C15 Halichondrin Model System

of olefination and then dihydroxylation was employed for the introduction of the C8 stereocenter. Acetylation and allylation with silane 122 gave 395. At this point, treatment with Triton B methoxide was attempted to effect olefin isomerization, acetate cleavage, and heteroconjugate addition as has been done previously by the Kishi group. However, due to the TBDPS ethers, this series of transformations was prohibitively sluggish, and an alternative strategy had to be devised. To this end, treatment with TBAF/AcOH cleaved the TBDPS ethers, and subsequent exposure to basic resin (Amberlyst IRA-400) in methanol resulted in acetate cleav- age to provide the triol. Upon treatment with 10 equiv of DBU in refluxing toluene, heteroconjugate addition occurred to provide a single diastereomer. As a side note, when the same transformation was conducted in benzene, a 2.9:1 mixture of C3 stereoisomers was obtained in 94% yield, leading to the speculation that high reaction temperatures are necessary for equilibration via the retro-Michael/Michael addition process. Exposure of the resultant pyranopyran to Amberlyst IRA-400 then cleaved the pivalate ester to provide triol 396. Global protection of the alcohols as TES ethers was achieved with TESCl and Et3N, and subsequent chemose- lective oxidation of the primary TES ether under Swern conditions then provided aldehyde 397.97 Phosphonium tetrafluoroborate salt 398 was deprotonated with n-BuLi, and the resultant phosphorane was allowed to react with the aldehyde to yield the enol ether 399 as an inconsequential E/Z mixture, which when treated with p-TsOH formed the trioxatricyclodecane 400.
⦁ C14-C22 Subunit Synthesis98
The first synthesis of the C14-C22 segment from the Burke group features a palladium-mediated, ligand-con- trolled, desymmetrization of a C2-symmetric diol synthesized via a two-directional chain elongation strategy (Scheme 38). The sequence began with C2-symmetric diol 401, which was

3070 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 38. Burke’s First Generation Synthesis of the C14-C22 Domain

converted via two-directional synthesis in six steps to diacetate 403. Treatment of 403 to desymmetrization condi- tions using Trost’s (R,R)-diphenylphosphino benzoic acid (DPPBA) ligand99 and Pd(0) at 0 C smoothly provided 404 (d.r. ) 5:1). When the desymmetrization reaction was attempted at room temperature, the products also included the double cyclization product as well as the desired compound. Protection of the secondary alcohol as the PMB ether furnished 405, which could be separated at this point from the minor diastereoisomer formed in the preceding cyclization. Selective hydroboration of the terminal olefin provided an alcohol that could be used for subunit coupling, and after protection of this alcohol, treatment with DDQ then cleaved the PMB ether to provide 406. Oxidation of the secondary alcohol was effected with Dess-Martin periodi- nane, and the installation of the exo-methylene and con- comitant deacylation was achieved with methylenetriphen- ylphosphorane to provide allylic alcohol 407. Lastly, the allylic alcohol was isomerized with a cationic iridium catalyst to generate aldehyde 408.100
In addition to the route shown above, a second synthesis of the C14-C22 segment 408 was devised by the Burke group.101 The principle highlight of the second synthesis is an ozonolytic desymmetrization of a C2-symmetric dihy- droxycyclohexene, as shown in Scheme 39. The sequence began with (S,S)-dihydroxycyclohexene 410 (synthesized from known 409 via ring-closing metathesis with Grubbs first generation catalyst). Ozonolysis under Schreiber’s conditions (modified by addition of excess acetic anhydride and DMAP) provided 411 as a separable 4:1 mixture in 75% yield. Subjection of the mixture to Lewis acid promoted allylation resulted in a 2:1 mixture of products 412 and C17- epi-412, with 412 being the desired isomer. Complete reduction to the diol with lithium aluminum hydride (LAH), and subsequent selective protection of the primary alcohol provided 413. Subsequent hydroboration (413 f 414) and selective monoacylation led to 415. In two further steps, the secondary alcohol was converted to the exo-olefin 416, which was treated with NaOAc-buffered PCC to provide 408.

Scheme 39. Burke’s Second Generation Synthesis of the C14-C22 Domain

Scheme 40. Burke’s [3,3]-Sigmatropic Rearrangement Approach to the C20-C36 Domain

⦁ Synthesis of the C22-C34(36) Subunit102
In their analysis of the challenges posed by the C22-C34 domain of the halichondrins, Burke and co-workers noted that this section was only a single epimerization away from being an achiral meso compound. In light of this, a strategy was adopted involving the asymmetric desymmetrization of the meso bis(allylic alcohol) 419 (Scheme 40). The meso diol 417 was converted to the requisite bis(allylic alcohol) 419 via a four-step sequence of (i) bis(O-alkylation) with t-butyl bromoacetate under phase-transfer conditions, (ii) ozonolytic ring cleavage, (iii) Wittig homologation, and (iv) sodium borohydride reduction. Desymmetrization of 419 was achieved with the Sharpless catalytic asymmetric epoxida- tion,103 resulting in the formation of bis(epoxy alcohol) 420. Conversion of 420 to the dimesylate was followed by iodide displacement with concomitant iodide-mediated reductive opening of the epoxide to provide 421. Treatment of 421 with trifluoroacetic acid effected lactonization to form

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3071

Scheme 41. Functionalization of the Diol 426 Scheme 42. Burke’s Second Generation Approach to the
C20-C36 Halichondrin Subunit

bis(dioxanone) 422. Upon exposure to LHMDS and TMSCl, 422 underwent kinetic enolization to form the bis(silylketene acetal) 424, and two Ireland-Claisen [3,3]-sigmatropic rearrangements subsequently occurred upon heating in toluene, thus forming bis(dihydropyran) 423.
To elaborate 423 on to a C20-C36 halichondrin fragment,
differential functionalization of the nearly symmetrical bis(dihydropyran) was required.104 Toward this end, both esters were reduced, and the alcohols formed were protected to form the bis(TBS ether) 425 (Scheme 41). Subsequent regio- and stereoselective hydroboration gave diol 426, also setting the C25 stereocenter. At this point, it was observed that the C30 and C26 possessed differential reactivity toward oxidation, acylation, and benzylation. For example, PCC oxidation proceeded with complete regioselectivity to give the C30 ketone 427. Similarly, acylation under Steglich conditions105 also occurred solely at the C30 alcohol to give the diastereomeric Mosher esters 428 and 429 in high yield. Although the C30 and C26 alcohols can be distinguished by the respective equatorial, axial, and diequatorial flanking substituents, it remains unclear why such a large reactivity preference exists for the C30 alcohol. Nonetheless, this preference was crucial for the differential functionalization of bis(dihydropyran) 426 and completion of the synthesis of the C22-C36 fragment. Though oxidation and acylation gave almost exclusively the C30 functionalized product, benzylation was found to occur with slightly less selective regiochemistry, as a 4:1 mixture was obtained favoring the desired 430.
Progressing forward from 430, oxidation and methylena- tion provided olefin 431 (Scheme 42). Next, simultaneous two-carbon chain extensions were performed at C22 and C34 via displacement of the bis(triflate) formed from 431 with lithio ethoxyacetylide to provide bis(alkyne) 432. Alkyne hydration was then carried out under mild Lewis acidic conditions106 with ZnCl2 and was followed by cleavage of the benzyl ester, which proved to be somewhat difficult, although satisfactory results were seen with boron trifluoride etherate in the presence of dimethylsulfide, as described by Fuji,107 to provide bis(ester) 433. To complete the sequence, the unmasked secondary alcohol was oxidized with PCC and subjected to a Saegusa oxidation,108 yielding the R,§- unsaturated ketone 435 designed to serve as a precursor to the fully elaborated C20-C36 halichondrin B subunit.
In a later publication, Burke disclosed a more efficient synthesis of a bis(dihydropyran) intermediate 425 (Scheme 43).109 The new strategy exploited the power of ring-opening

Scheme 43. Burke’s Improved Route to Bis-pyran 425

and ring-closing metathesis as a method for the rapid and facile assembly of functionalized rings. The sequence began with the same starting material, 417, as was used for the double Ireland-Claisen route (see Scheme 40). Although conversion of 417 to the desymmetrized bis(imide) 436 could be achieved in one step by bis(O-alkylation) with (4S)-3- (bromoacetyl)-4-isopropyl-2-oxazolidinone and silver oxide, the best conversion was observed with a two-step protocol entailing (i) formation of the meso diacid with NaH, tetrabutylammonium iodide (TBAI), and sodium bromoac- etate and (ii) formation of the bis(pivalic anhydride) followed by reaction with N-lithio-(4S)-4-isoproyl-2-oxazolidinone. Subsequent stereoselective double methallylation was achieved with the Z-enolate of 436 (formed using NaHMDS) and methallyl bromide to give 437 by virtue of the valine-derived Evans chiral auxiliaries.110 Reductive cleavage of the aux- iliaries and TBS protection provided 438. At this point, the key ring-opening/ring-closing metathesis was investigated. Treatment of 438 with Grubbs first generation catalyst furnished an inseparable mixture of diastereomeric dihydro- pyrans 439 and 440 in which only one ring-closing metath- esis occurred. Fortunately, however, exposure of 438 to the

3072 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 44. Burke’s Third Generation Approach to the C22-C35 Domain

more reactive Schrock molybdenum-based catalyst smoothly produced the desired 425 in 79% yield.
The most recent work from the Burke group toward the C22-C36 segment is shown in Scheme 44.111 The key goals of this synthesis were (i) to introduce the C19 and C26 exo- methylene units simultaneously and at a late stage, (ii) to synthesize the structure bidirectionally utilizing both sub- strate- and reagent-controlled reactions, and (iii) to use the Pd(0)-mediated asymmetric double cycloetherification previ- ously developed. The sequence began with bis-silyl-protected cyclopentenediol 441, which was oxidatively cleaved and treated with the Still-Gennari reagent 442 to give 443. Reduction and hydroboration following Kishi’s empirical rule112 allowed the generation of four stereocenters via substrate-controlled asymmetric induction to produce 444. Selective protection of the C30 and C26 secondary alcohols in 444 was achieved by formation of a PMP-acetal and DIBAL reduction to provide 445. Using a two-directional strategy, 445 was elongated via oxidation, Horner- Wadsworth-Emmons olefination, and subsequent reduction to the meso-bis(allylic alcohol). At this point, the symmetry of the system was interrupted with Sharpless asymmetric epoxidation at the allylic alcohols to provide 446. In five further routine steps, 446 was converted to the Pd(0)- mediated asymmetric double cycloetherification precursor
448. Treatment of the tetraol 448 with Pd2dba3 · CHCl3 and
the Trost (R,R)-DPPBA ligand established the bis(tetrahy-

Scheme 45. Burke’s Symmetry-Based Approach to the C38-C54 Domain

dropyran) 449 with absolute stereochemical control in high yield. Five additional steps were then required to convert 449 to 450.

⦁ Synthesis of the C38-C54 Subunit113,114

In a similar vein to both Hirata/Yonemitsu and Salomon, Burke has designed an efficient construction of the C38-C54 domain by exploiting the local C2-symmetry about the C44 spiroketal carbon. The sequence commenced with the known (S)-epoxide 451 (derived from (S)-malic acid), which was opened with the tetramethylethylenediamine (TMEDA) complex of lithium acetylide (Scheme 45). The crude alkyne was isomerized with potassium tert-butoxide to give 452. Subsequent reduction with LAH provided the E-allylic alcohol as a single diastereomer, and exposure of the allylic alcohol to Johnson orthoester Claisen conditions smoothly furnished 453. Claisen self-condensation was then carried out by treating 453 with LHMDS, and the resulting §-ke- toester was decarboxylated under Krapcho conditions115 to give the C2 symmetric ketone 454. Asymmetric dihydroxy- lation with AD-mix-R to effect a stereoselective dihydroxy- lation and acid-catalyzed spiroketalization of the resultant tetraol yielded a 1.1:1.0 mixture of the C2 symmetric 1,7- dioxaspiro[5.5]undecane 455 and the isomeric, undesired 1,6- dioxaspiro[4.5]decane 456. The mixture was debenzylated and equilibrated under acidic conditions to a more favorable mixture of the desired 457, along with 458 and 459 (5:2:1). Fortunately, after separation the undesired products 458 and 459 were easily recycled to produce more 457 by re-exposure to the equilibrating conditions.
Selective oxidation of the primary alcohol groups of tetraol
457 (presumably via oxidation of the intermediate five- membered hemiacetals) was achieved with tetrapropylam- monium perruthenate (TPAP)/N-methylmorpholine-N-oxide (NMO), forming the C2-symmetric bis(lactone) 460 (Scheme 46). To avoid forming a statistical mixture of starting material and mono- and difunctionalized products, 460 was partially olefinated with 0.7 equiv of the Tebbe reagent. Subsequent hydroboration/oxidation gave 461 as an approximately 1:1 mixture with the starting material 460. Alcohol 461 was then oxidized to the aldehyde and allylated under chelation- controlled conditions with titanium(IV) chloride and allyl-

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3073

Scheme 46. Completion of the Burke C38-C54 Domain Synthesis

Scheme 47. The r-Alkoxyphosphorane Approach to the C1-C22 Domain

tributylstannane to give the homoallylic alcohol 462 as a single diastereoisomer. Conversion to the corresponding BOC carbonate, and subsequent treatment with iodine monobro- mide afforded the iodocarbonate 463 with excellent diaste- reoselectivity (18:1). In four further steps, 463 was con- verted into the fully functionalized C38-C54 segment 465.
⦁ Subunit Couplings
The Burke group has also reported studies into the coupling of the C1-C14 and C14-C22 fragments.116 The first method used for joining the C14-C22 and C1-C13 fragments was a Wittig coupling/ketalization strategy, which relied on the preparation of a C14-C22 fragment with a phosphonium salt at the C14 terminus. Toward this end, aldehyde 408 was first treated with PPTS to provide the dimethyl acetal in high yield. Subsequent exposure to triphenylphosphine in the presence of boron trifluoride etherate provided the triphenylphosphonium tetrafluoroborate salt 466 (Scheme 47). Given the moisture sensitivity and the isolation difficulties 466 presented, it was treated without delay (in base-washed glassware) with n-butyllithium at -78
⦁ C to generate the (R-methoxyalkyl)triphenylphosphorane
467. Addition of 397 to the reaction mixture followed by slow warming provided 468, which ketalized on treatment with p-toluenesulfonic acid to provide the complete C1-C22 domain, 469, in a two-step yield of 29%.
The second method used to construct the C1-C22 segment involved a Horner-Wadsworth-Emmons coupling between 397 and 471 (Scheme 48). To prepare §-ketophosphonate 471, alcohol 416 was oxidized with freshly prepared Jones
Scheme 48. Burke’s Horner-Wadsworth-Emmons Strategy for the Synthesis of the C1-C22 Domain

reagent and esterified with trimethylsilyldiazomethane to provide 470. Addition of lithio(dimethyl)methylphosphonate to 470 then yielded the §-ketophosphonate 471. Horner- Wadsworth-Emmons reaction of 397 and 471 with KHMDS as base proceeded to form enone 472 along with an unknown product that could not be separated. Thus, the enone was carried on crude into the Michael addition/ketalization sequence to provide 469 in 10% yield over these steps.

⦁ The Discovery and Development of E7389
In 1992 samples of synthetic halichondrin B and several intermediates were provided by the Kishi group to the Eisai Research Institute with a goal of evaluating in Vitro and in ViVo activity. In a remarkable discovery, the macrocyclic macrolactone diol shown in Figure 13 (146) was found to be within an order of magnitude as potent as the parent halichondrin B against DLD-1 human colon cancer cells (IC50 for 146 ) 4.6 nM).117 Further study showed that halichondrin B and 146 both blocked cell cycle progression at the G2/M phase, both caused microtubule destabilization, and both had similar profiles in the 60-cell-lines screen at the National Cancer Institute. Armed with this information, the stage was set to develop a potential halichondrin-derived therapeutic.

3074 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Figure 13. The right-hand macrolactone diol (146) of halichondrin B.

Figure 15. SAR in the C19-C26 domain.

Figure 14. Selected SAR in the ∼C30-C38 domain.
⦁ Preliminary SAR Studies
One of the initial hurdles to overcome was the observation that 146 was not active in ViVo in a LOX human melanoma xenograft model. Flow cytometry revealed that the problem was related to reversibility of action of 146, a feature that was not found with the natural product. For example, halichondrin B was able to maintain a complete mitotic block 10 h after drug washout at levels of 10 nM, whereas 146 was ineffective even at levels as high as 1 µM. To facilitate the evaluation of compounds under conditions that would be relevant to in ViVo efficacy, ERI developed an in-house proprietary cell-based assay to evaluate compounds. A number of analogues of 146 that were modified in the terminal C30-C38 region were subsequently evaluated, and although the ability to inhibit cell growth was relatively uniform, the ability to cause an irreversible complete mitotic block was very sensitive to structure changes.118 For example, among the compounds shown in Figure 14, diol 474 was the first compound able to maintain a reasonable degree of complete mitotic block. Although many compounds prepared as part of SAR around this region of the molecule displayed good potency, there were some exceptions, as can be seen with compound 475, in which the C31 methyl group has been removed, resulting in an approximately 130-fold decrease in potency relative to 474.
Other studies evaluated the effect of deletion of function- ality in the C19-C26 domain. As can be seen in Figure 15, changes in this region were permissible, although at some penalty to either potency or reversibility.
A significant step forward came with the discovery that the C29-C36 pyranopyran domain could be replaced by

Figure 16. Tetrahydropyran and tetrahydrofuran analogues.

monocyclic pyran and furan derivatives.119 A large number of compounds were evaluated in this series, and some examples are shown in Figure 16.
Compounds 481 and 488 were tested in the LOX mela- noma xenograft model, and neither showed any efficacy. A number of possible explanations for this were possible, but ultimately the question of the stability of the macrolactone

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3075

Scheme 49. Synthesis of the Aldehyde 493

Figure 17. Eisai’s lead compounds: ER-076439, 489, and E7389,

Figure 18. An overview of the synthesis strategy for ER-076349.

in these tetrahydropyran and tetrahydrofuran analogues toward nonspecific esterases present in mouse serum became of concern. Attention turned to the preparation of nonhy- drolyzable isosteres for the ester such as ketone, ether, and amide functionalities. Among these the most promising in terms of in Vitro activity was the ketone, and two compounds rose to prominence: the diol 489 (ER-076349) and the amino alcohol 490 (E7389, previously ER-086526) (Figure 17).
ER-076349 exhibited activity in ViVo in xenograft models against a variety of cancers including MDA-MB435 breast carcinoma, COLO-205 colon carcinoma, LOX melanoma, and NIH:OVCAR-3 ovarian carcinoma. There was no evidence of toxicity at the doses tested.120 Subsequent SAR from ER-076349 lead to a variety of compounds [see the next section for details], and the lead compound to emerge from these studies was E7389. Although slightly less potent than ER-076349, E7389 has the important characteristic of having a U937 reversibility ratio of 1.

⦁ Synthesis and SAR of ER-076349, E7389, and Analogues
The Eisai synthesis of both ER-076349 and E7389 employs much of the technology laid down in Kishi’s studies. Against this backdrop, an analysis of ER-076349, from which E7389 can be readily obtained, led to three key fragments 491, 493, and the ubiquitous 25 (Figure 18).
The sequence to construct the C27-C35 aldehyde 493, began with a regioselective opening of epoxide 495 with the acetylide derived from acetylene 494 (Scheme 49). Partial reduction of the alkyne under Lindlar conditions and acylation provided cis-olefin 497. Dihydroxylation then afforded an 8:1 mixture of diastereomeric alcohols, which were converted to the corresponding mesylates, whereupon separation was possible. Treatment of mesylate 498 with Triton B resulted in deacetylation and formation of the tetrahydrofuran ring. Subsequent desulfonylation with methylmagnesium bromide, and methylation of the alcohol gave 499. In four further steps, the C27-C35 aldehyde 493 was obtained.
Nozaki-Hiyama-Kishi coupling of aldehyde 493 with vinyl iodide 25 and subsequent base-induced cyclization provided a 3:1 mixture of C27 diastereomers favoring the desired product (Scheme 50). The PMB ether was then removed to yield 500, at which point the diastereomers were separable. Alcohol 500 was converted to sulfone 501 in a four-step protocol, and subsequent reaction of lithiated 501 with aldehyde 491 and oxidation gave 502. Samarium(II) iodide was then used to mediate desulfonylation, and a Nozaki-Hiyama-Kishi macrocyclization and subsequent allylic alcohol oxidation gave enone 503. Lastly, exposure of 503 to TBAF buffered with imidazole hydrochloride and then PPTS provided the ER-076349. The synthesis is approximately 35 steps longest linear sequence from com- mercial materials, and its use to prepare material for preclinical, and possibly clinical, use stands as testament to the power of organic synthesis and the talents of scientists at the Eisai Research Institute.
With ER-076349 in hand, the C34/C35 vicinal diol was used as a handle for subsequent derivatization. As shown in Scheme 51, an array of analogues were synthesized using standard transformations.
Each of the compounds shown in Scheme 51 was evaluated for (i) cell growth inhibitory activity against DLD-1 human colon cancer cells under continuous exposure condi- tions, (ii) the ability to maintain a complete mitotic block
10 h after drug washout, and (iii) susceptibility to P- glycoprotein mediated drug efflux using murine P388/ VMDRC.04 cells. As can be seen, the compounds in this

3076 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

Scheme 50. Completion of the Synthesis of ER-076349
Scheme 51. Synthesis and in Vitro Activity Profile for Macrocyclic Ketone Analogues of Halichondrin B

series displayed almost uniformly excellent activity and, in many cases, displayed single-digit reversibility ratios. All substrates were subject to P-glycoprotein mediated drug efflux, although in some cases the fold resistance was only on the order of 2-fold. On the basis of a combination of favorable features, E7389 was selected for development as a potential anticancer chemotherapeutic agent, with the name eribulin (or eribulin mesylate for the corresponding meth- ansulfonate salt).121

⦁ Preclinical Development
As was noted in the previous section, eribulin displayed good potency, and when compared with ER-076349, vinc- ristine, and paclitaxel for antiproliferative activity against eight cell lines, an average IC50 value of 1.8 ( 1.1 nM was
obtained. This compared very favorably with ER-076349 (IC50 ) 0.45 ( 0.09 nM) and was more potent than vincristine and paclitaxel (IC50 ) 3.2 ( 0.7 and 7.3 ( 1.9 nM, respectively). A number of mouse xenograft models and dosing were used to evaluate eribulin’s efficacy, and highly encouraging results were obtained. In MDA-MB-435 xeno- graft studies, treatment led to tumor regression with no evidence of toxicity based on body weight loss or water consumption. Dosing regimens at 0.25, 0.5, and 1 mg/kg were all either equally efficacious or superior to paclitaxel (dosed at the maximal tolerated dose of 25 mg/kg). Similar results were obtained with COLO 205 colon carcinoma, LOX melanoma, and NIH:OVCAR-3 ovarian cancer xenografts.119 Toxicity studies in Vitro showed no cytotoxicity at 1 µM eribulin in quiescent IMR-90 human fibroblasts.

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3077

The mode of action of eribulin has also been studied. Treatment of U937 human histiocytic lymphoma cells with eribulin resulted in arrest of cells in the G2/M phase. There were no effects on cells in the G1 and S phases, and cells in these phases became progressively depleted upon prolonged exposure to eribulin. There was marked mitotic spindle disruption, and a biotinylated compound related to eribulin was used to demonstrate that tubulin binding occurs.119 In more recent studies, the exact nature of the interaction with tubulin has been further investigated.122 Eribulin neither destabilizes nor shortens microtubules. It acts by blocking microtubule growth, which results in G2/M arrest because the microtubules are not long enough to reach the kineto- cores. Molecular modeling suggests that eribulin binds in a cleft between the R-subunit and §-subunit of tubulin, and it is thought that this blocks or slows nucleotide growth.

⦁ Progress toward the Clinic: Current Status of E7389

The first trial of eribulin in humans was with 40 patients with refractory or advanced solid tumors. Eribulin was administered intravenously at 0.25-2/mg2 on days 1, 7, and 15 of a 28-day cycle. Among these patients, two showed partial responses, three had mixed response, and 12 had stable disease for a median 4 month period. Similar data was obtained in several other phase I trials. Initial phase II clinical trials for breast cancer and non-small-cell lung cancer (NSCLC) involved a large subject group ( 100 each), and the overall response rate for breast cancer was 15%. The NSCLC group had a slightly lowered response rates ( 10%). A number of patients suffered neutropenia and fatigue, and 40% complained of peripheral neuropathy, all common side effects with tubulin-active agents. Recently, eribulin has demonstrated activity in a heavily pretreated population of women with locally advanced or metastatic breast cancer. Subjects had received an anthracycline, a taxane, and capecitabine as prior therapy and were refractory to their last chemotherapy regimen. The overall response rate was 10-15%, and there were manageable peripheral neuropathy symptoms. Both the preclinical and clinical features of E7389
have been recently reviewed in detail.123
In light of these encouraging trials, further phase II and phase III clinical trials are ongoing in Europe, Japan, and the USA. Eisai has indicated that it plans to file an NDA with the FDA for eribulin as third-line treatment for breast cancer in 2009-2010.124
⦁ Concluding Comments
The efforts invested early on in the halichondrin story by the isolation and structure elucidation groups of Uemura, Pettit and Blunt, and Munro provided the impetus for synthetic studies. In arguably the most high profile example of the power of contemporary organic synthesis to provide a remarkably complex natural product for further study, the Kishi group’s commitment to the problem of the total synthesis of the halichondrins then provided material that has allowed the E7389 story to unfold to the point where there is realistic potential for clinical application. As a final concluding comment, we would note that despite the discovery and synthesis of a number of simplified analogues, most notably E7389, synthetic studies aimed at the synthesis of the halichondrins and halichondrin subunits remain an important goal for contemporary synthesis. Aside from
opportunities to develop new methods and strategies, syn- thesis remains the only viable method for the supply of these compounds, and developments in the synthesis arena will undoubtedly impact the future biology and medicine of these important molecules.

⦁ References
⦁ Norcross, R. D.; Paterson, I. Chem. ReV. 1995, 95, 2041.
⦁ Hart, J. B.; Lill, R. E.; Hickford, S. J. H.; Blunt, J. W.; Munro,
M. H. G. Drugs Sea 2000, 134.
⦁ Yu, M. J.; Kishi, Y.; Littlefield, B. A. In Anticancer Agents from Natural Products; Cragg, G. M., Kingston, D. G. I., Newman, D. J., Eds.; CRC Press: Boca Raton, FL, 2005.
⦁ Choi, H.-W.; Demeke, D.; Kang, F.-A.; Kishi, Y.; Nakajima, K.; Nowak, P.; Wan, Z.-K.; Xie, C. Pure Appl. Chem. 2003, 75, 1.
⦁ Tachibana, K.; Scheuer, P. J.; Tsukitani, Y.; Kikuchi, H.; Van Engen, D.; Clardy, J.; Gopichand, Y.; Schmitz, F. J. J. Am. Chem. Soc. 1981, 103, 2469.
⦁ Murakami, Y.; Oshima, Y.; Yasumoto, T. Nippon Suisan Gakkaishi
1982, 48, 69.
⦁ Uemura, D.; Takahashi, K.; Yamamoto, T.; Katayama, C.; Tanaka, J.; Okumura, Y.; Hirata, Y. J. Am. Chem. Soc. 1985, 107, 4796.
⦁ Hirata, Y.; Uemura, D. Pure Appl. Chem. 1986, 58, 701.
⦁ (a) Munro, M. H. G.; Blunt, J. W. Personal communication. (b) Lake,
R. J. Internal Report, University of Canterbury, February 26, 1988.
⦁ Litaudon, M.; Hart, J. B.; Blunt, J. W.; Lake, R. J.; Munro, M. H. G.
Tetrahedron Lett. 1994, 35, 9435.
⦁ (a) Litaudon, M.; Hickford, S. J. H.; Lill, R. E.; Lake, R. J.; Blunt,
J. W.; Munro, M. H. G. J. Org. Chem. 1997, 62, 1868. (b) Blunt and Munro have also described 53-methoxyneoisohomohalichondrin B, which has been observed to convert to isohomohalichondrin B in solution in an NMR tube, presumably via acetal hydrolysis. They have suggested that it is an artifact of isolation. See ref 2.
⦁ Pettit, G. R.; Herald, C. L.; Boyd, M. R.; Leet, J. E.; Dufresne, C.; Doubek, D. L.; Schmidt, J. M.; Cerny, R. L.; Hooper, J. N. A.; Rutzler, K. C. J. Med. Chem. 1991, 34, 3339.
⦁ (a) Pettit, G. R.; Tan, R.; Gao, F.; Williams, M. D.; Doubek, D. L.; Boyd, M. R.; Schmidt, J. M.; Chapuis, J. C.; Hamel, E.; Bai, R.; Hooper, J. N. A.; Tackett, L. P. J. Org. Chem. 1993, 58, 2538. (b) Pettit, G. R.; Gao, F.; Doubek, D. L.; Boyd, M. R.; Hamel, E.; Bai, R.; Schmidt, J. M.; Tackett, L. P.; Rutzler, K. Gazz. Chim. Ital. 1993, 123, 371. (c) Pettit, G. R.; Ichihara, Y.; Wurzel, G.; Williams, M. D.; Schmidt, J. M.; Chapuis, J.-C. J. Chem. Soc., Chem. Commun. 1995, 3, 383.
⦁ Bai, R.; Paull, K. D.; Herald, C. L.; Malspeis, L.; Pettit, G. R.; Hamel, E. J. Biol. Chem. 1991, 266, 15882.
⦁ Munro, M. H. G.; Blunt, J. W.; Dumdei, E. J.; Hickford, S. J. H.; Lill, R. E.; Li, S.; Battershill, C. N.; Duckworth, A. R. Prog. Ind. Microbiol. 1999, 35, 155.
⦁ Aicher, T. D.; Buszek, K. R.; Fang, F. G.; Forsyth, C. J.; Jung, S. H.; Kishi, Y.; Matelich, M. C.; Scola, P. M.; Spero, D. M.; Yoon, S. K. J. Am. Chem. Soc. 1992, 114, 3162.
⦁ Takai, K.; Kimura, K.; Kuroda, T.; Hiyama, T.; Nozaki, H.
Tetrahedron Lett. 1983, 24, 5281.
⦁ (a) Jin, H.; Uenishi, J.; Christ, W. J.; Kishi, Y. J. Am. Chem. Soc.
1986, 108, 5644. (b) Kishi, Y. Pure Appl. Chem. 1992, 64, 343.
⦁ Aicher, T. D.; Kishi, Y. Tetrahedron Lett. 1987, 28, 3463.
⦁ (a) Omura, K.; Sharma, A. K.; Swern, D. J. Org. Chem. 1976, 41, 957. (b) Omura, K.; Swern, D. Tetrahedron 1978, 34, 1651.
⦁ Mancuso, A. J.; Huang, S.-L.; Swern, D. J. Org. Chem. 1978, 43, 2480.
⦁ Stamos, D. P. Studies towards a practical synthesis of the c.1-c.38 segment of halichondrin b (stereodifferentiation), Ph.D. Thesis, Harvard University, Cambridge, MA, 1997.
(23) (a) Dess, D. B.; Martin, J. C. J. Org. Chem. 1983, 48, 4155. (b) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277.
⦁ Mahoney, W. S.; Brestensky, D. M.; Stryker, J. M. J. Am. Chem. Soc. 1988, 110, 291.
⦁ Aicher, T. D.; Buszek, K. R.; Fang, F. G.; Forsyth, C. J.; Jung, S. H.; Kishi, Y.; Scola, P. M. Tetrahedron Lett. 1992, 33, 1549.
⦁ Buszek, K. R.; Fang, F. G.; Forsyth, C. J.; Jung, S. H.; Kishi, Y.; Scola, P. M.; Yoon, S. K. Tetrahedron Lett. 1992, 33, 1553.
⦁ Vekemans, J. A. J. M.; Boerekamp, J.; Godefroi, E. F.; Chittenden,
G. J. F. Recl. TraV. Chim. Pays-Bas 1986, 104, 266.
⦁ Yamaguchi, M.; Hirao, I. Tetrahedron Lett. 1983, 24, 391.
⦁ (a) Sharpless, K. B.; Michaelson, R. C. J. Am. Chem. Soc. 1973, 95, 6136. (b) Tanaka, S.; Yamamoto, H.; Nozaki, H.; Sharpless, K. B.; Michaelson, R. C.; Cutting, J. D. J. Am. Chem. Soc. 1974, 96, 5254.
⦁ Fang, F. G.; Kishi, Y.; Matelich, M. C.; Scola, P. M. Tetrahedron Lett. 1992, 33, 1557.

3078 Chemical Reviews, 2009, Vol. 109, No. 7 Jackson et al.

⦁ Oikawa, Y.; Yoshioka, T.; Yonemitsu, O. Tetrahedron Lett. 1982,
23, 885.
⦁ Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M. Bull. Chem. Soc. Jpn. 1979, 52, 1989.
⦁ Duan, J. J.; Kishi, Y. Tetrahedron Lett. 1993, 34, 7541.
⦁ Corey, E. J.; Suggs, J. W. Tetrahedron Lett. 1975, 16, 2647.
⦁ Stamos, D. P.; Kishi, Y. Tetrahedron Lett. 1996, 37, 8643.
⦁ Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. ReV.
1994, 94, 2483.
⦁ Stamos, D. P.; Sheng, C.; Chen, S. S.; Kishi, Y. Tetrahedron Lett.
1997, 38, 6355.
⦁ Stamos, D. P.; Chen, S. S.; Kishi, Y. J. Org. Chem. 1997, 62, 7552.
⦁ Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.; Masamune, S.; Roush, W. R.; Sakai, T. Tetrahedron Lett. 1984, 25, 2183.
⦁ (a) Itsuno, S.; Nakano, M.; Miyazaki, K.; Masuda, H.; Ito, K.; Hirao, A.; Nakahama, S. J. Chem. Soc., Perkin Trans. 1 1985, 10, 2039.
(b) Corey, E. J.; Bakshi, R. K.; Shibata, S.; Chen, C. P.; Singh, V. K.
J. Am. Chem. Soc. 1987, 109, 7925. (c) Corey, E. J.; Bakshi, R. K.; Shibata, S. J. Am. Chem. Soc. 1987, 109, 5551.
(41) (a) Fuerstner, A.; Shi, N. J. Am. Chem. Soc. 1996, 118, 12349. (b) Fuerstner, A.; Shi, N. J. Am. Chem. Soc. 1996, 118, 2533.
(42) Namba, K.; Jun, H.-S.; Kishi, Y. J. Am. Chem. Soc. 2004, 126, 7770.
(43) Kaburagi, Y.; Kishi, Y. Org. Lett. 2007, 9, 723.
(44) (a) Namba, K.; Kishi, Y. J. Am. Chem. Soc. 2005, 127, 15382. (b)
Choi, H.-w.; Nakajima, K.; Demeke, D.; Kang, F.-A.; Jun, H.-S.;
Wan, Z.-K.; Kishi, Y. Org. Lett. 2002, 4, 4435. (c) Wan, Z.-K.; Choi,
H.-w.; Kang, F.-A.; Nakajima, K.; Demeke, D.; Kishi, Y. Org. Lett.
2002, 4, 4431. (d) Zhang, Z.; Huang, J.; Ma, B.; Kishi, Y. Org. Lett.
2008, 10, 3073. (e) Namba, K.; Cui, S.; Wang, J.; Kishi, Y. Org. Lett. 2005, 7, 5417. (f) Namba, K.; Kishi, Y. Org. Lett. 2004, 6,
5031. (g) Kurosu, M.; Lin, M.-H.; Kishi, Y. J. Am. Chem. Soc. 2004,
126, 12248. (h) Namba, K.; Wang, J.; Cui, S.; Kishi, Y. Org. Lett.
2005, 7, 5421.
⦁ Jackson, K. L.; Henderson, J. A.; Motoyoshi, H.; Phillips, A. J.
Angew. Chem., Int. Ed. 2009, 48, 2346.
⦁ Davies, H. M. L.; Ahmed, G.; Churchill, M. R. J. Am. Chem. Soc.
1996, 118, 10774.
⦁ Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999, 1, 953.
⦁ Jackson, K. L.; Henderson, J. A.; Morris, J. C.; Motoyoshi, H.; Phillips, A. J. Tetrahedron Lett. 2008, 49, 2939.
⦁ Kobayashi, Y.; Kusakabe, M.; Kitano, Y.; Sato, F. J. Org. Chem.
1988, 53, 1586.
⦁ Grieco, P. A.; Oguri, T.; Yokoyama, Y. Tetrahedron Lett. 1978, 19, 419.
⦁ (a) Roskamp, E. J.; Johnson, C. R. J. Am. Chem. Soc. 1986, 108, 6062. (b) Pirrung, M. C.; Werner, J. A. J. Am. Chem. Soc. 1986, 108, 6060.
⦁ Namba, K.; Kishi, Y. Org. Lett. 2004, 6, 5031.
⦁ Henderson, J. A.; Jackson, K. L.; Phillips, A. J. Org. Lett. 2007, 9, 5299.
(54) Ho, T. -L.; Sapp, S. G. Synth. Commun. 1983, 13, 267.
⦁ Um, J. M.; Houk, K. N.; Phillips, A. J. Org. Lett. 2008, 10, 3769.
⦁ Aicher, T. D.; Buszek, K. R.; Fang, F. G.; Forsyth, C. J.; Jung, S. H.; Kishi, Y.; Matelich, M. C.; Scola, P. M.; Spero, D. M.; Yoon, S. K. J. Am. Chem. Soc. 1992, 114, 3162.
⦁ Stewart, I. C.; Douglas, C. J.; Grubbs, R. H. Org. Lett. 2008, 10, 441.
⦁ Kaburagi, Y.; Kishi, Y. Org. Lett. 2007, 9, 723.
⦁ Holmquist, C. R.; Roskamp, E. J. Tetrahedron Lett. 1992, 33, 1131.
⦁ Horita, K.; Hachiya, S.; Nagasawa, M.; Hikota, M.; Yonemitsu, O.
Synlett 1994, 1, 38.
(61) Roush, W. R.; Brown, R. J. J. Org. Chem. 1982, 47, 1371.
⦁ Horita, K.; Hachiya, S.-I.; Yamazaki, T.; Naitou, T.; Uenishi, J. i.; Yonemitsu, O. Chem. Pharm. Bull. 1997, 45, 1265.
⦁ Horita, K.; Nagasawa, M.; Hachiya, S.; Yonemitsu, O. Synlett 1994,
1, 40.
⦁ Horita, K.; Nagasawa, M.; Sakurai, Y.; Yonemitsu, O. Chem. Pharm. Bull. 1998, 46, 1199.
⦁ (a) Saito, S.; Ishikawa, T.; Kuroda, A.; Koga, K.; Moriwake, T. Tetrahedron 1992, 48, 4067. (b) Saito, S.; Hasegawa, T.; Inaba, M.; Nishida, R.; Fujii, T.; Nomizu, S.; Moriwake, T. Chem. Lett. 1984, 8, 1389.
⦁ Mori, Y.; Kuhara, M.; Takeuchi, A.; Suzuki, M. Tetrahedron Lett.
1988, 29, 5419.
⦁ Horita, K.; Sakurai, Y.; Nagasawa, M.; Hachiya, S.; Yonemitsu, O.
Synlett 1994, 1, 43.
⦁ Horita, K.; Sakurai, Y.; Nagasawa, M.; Yonemitsu, O. Chem. Pharm. Bull. 1997, 45, 1558.
⦁ Horita, K.; Sakurai, Y.; Hachiya, S.-i.; Nagasawa, M.; Yonemitsu,
O. Chem. Pharm. Bull. 1994, 42, 683.
⦁ Still, W. C.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405.

⦁ Christensen, B. G.; Strachan, R. G.; Trenner, N. R.; Arison, B. H.; Hirschmann, R. F.; Chemerda, J. M. J. Am. Chem. Soc. 1960, 82, 3995.
⦁ (a) Lewis, M. D.; Cha, J. K.; Kishi, Y. J. Am. Chem. Soc. 1982,
104, 4976. (b) Kozikowski, A. P.; Sorgi, K. L.; Wang, B. C.; Xu,
Z. B. Tetrahedron Lett. 1983, 24, 1563. (c) Hosomi, A.; Sakata, Y.; Sakurai, H. Tetrahedron Lett. 1984, 25, 2383. (d) Giannis, A.; Sandhoff, K. Tetrahedron Lett. 1985, 26, 1479.
⦁ Horita, K.; Nishibe, S.; Yonemitsu, O. Phytochem. Phytopharm. 2000, 386.
⦁ Horita, K.; Sakurai, Y.; Nagasawa, M.; Maeno, K.; Hachiya, S.; Yonemitsu, O. Synlett 1994, 1, 46.
⦁ Horita, K.; Hachiya, S.-i.; Ogihara, K.; Yoshida, Y.; Nagasawa, M.; Yonemitsu, O. Heterocycles 1996, 42, 99.
⦁ Hanessian, S.; Sumi, K. Synthesis 1991, 12, 1083.
⦁ Chen, K. M.; Hardtmann, G. E.; Prasad, K.; Repic, O.; Shapiro, M. J.
Tetrahedron Lett. 1987, 28, 155.
⦁ Horita, K.; Nagasawa, M.; Hachiya, S.-I.; Sakurai, Y.; Yamazaki, T.; Uenishi, J. i.; Yonemitsu, O. Tetrahedron Lett. 1997, 38, 8965.
⦁ (a) Lindgren, B. O.; Nilsson, T. Acta Chem. Scand. 1973, 27, 888.
(b) Bal, B. S.; Childers, W. E., Jr.; Pinnick, H. W. Tetrahedron 1981,
37, 2091.
⦁ Horita, K.; Hachiya, S.-i.; Ogihara, K.; Yoshida, Y.; Nagasawa, M.; Yonemitsu, O. Heterocycles 1996, 42, 99.
⦁ Cooper, A. J.; Salomon, R. G. Tetrahedron Lett. 1990, 31, 3813.
⦁ (a) Jacobsen, E. N.; Marko, I.; Mungall, W. S.; Schroeder, G.; Sharpless, K. B. J. Am. Chem. Soc. 1988, 110, 1968. (b) Wai, J. S. M.; Marko, I.; Svendsen, J. S.; Finn, M. G.; Jacobsen, E. N.; Sharpless, K. B. J. Am. Chem. Soc. 1989, 111, 1123.
⦁ (a) Gao, Y.; Sharpless, K. B. J. Am. Chem. Soc. 1988, 110, 7538.
(b) Kim, B. M.; Sharpless, K. B. Tetrahedron Lett. 1989, 30, 655.
⦁ Mukaiyama, T.; Kobayashi, S.; Shoda, S. Chem. Lett. 1984, 9, 1529.
⦁ Cooper, A. J.; Pan, W.; Salomon, R. G. Tetrahedron Lett. 1993, 34, 8193.
⦁ Barton, D. H. R.; McCombie, S. W. J. Chem. Soc., Perkin Trans 1
1975, 16, 1574.
⦁ Kim, S.; Salomon, R. G. Tetrahedron Lett. 1989, 30, 6279.
⦁ DiFranco, E.; Ravikumar, V. T.; Salomon, R. G. Tetrahedron Lett.
1993, 34, 3247.
⦁ Matsuzawa, S.; Horiguchi, Y.; Nakamura, E.; Kuwajima, I. Tetra- hedron 1989, 45, 349.
⦁ (a) Burke, S. D.; Jung, K. W.; Phillips, J. R.; Perri, R. E. Tetrahedron Lett. 1994, 35, 703. (b) Burke, S. D.; Jung, K. W.; Lambert, W. T.; Phillips, J. R.; Klovning, J. J. J. Org. Chem. 2000, 65, 4070.
⦁ Masamune, S.; Ma, P.; Okumoto, H.; Ellingboe, J. W.; Ito, Y. J. Org. Chem. 1984, 49, 2834.
⦁ (a) Shing, T. K. M.; Tsui, H. C.; Zhou, Z. H.; Mak, T. C. W. J. Chem. Soc., Perkin Trans. 1 1992, 7, 887. (b) Shing, T. K. M.; Tsui, H. C.; Zhou, Z. H. J. Chem. Soc., Chem. Commun. 1992, 11, 810.
⦁ Iida, H.; Yamazaki, N.; Kibayashi, C. J. Org. Chem. 1986, 51, 3769.
⦁ For related examples see: (a) Gilchrist, T. L.; Stanford, J. E. J. Chem. Soc., Perkin Trans. 1 1987, 1, 225. (b) Schoenen, F. J.; Porco, J. A., Jr.; Schreiber, S. L.; VanDuyne, G. D.; Clardy, J. Tetrahedron Lett. 1989, 30, 3765. (c) Sisti, A. J.; Vitale, A. C. J. Org. Chem. 1972, 37, 4090.
⦁ Lambert, W. T.; Burke, S. D. Org. Lett. 2003, 5, 515.
⦁ (a) Criegee, R.; Banciu, A.; Keul, H. Chem. Ber. 1975, 108, 1642.
(b) Schreiber, S. L.; Claus, R. E.; Reagan, J. Tetrahedron Lett. 1982,
23, 3867.
⦁ Rodriguez, A.; Nomen, M.; Spur, B. W.; Godfroid, J. J. Tetrahedron Lett. 1999, 40, 5161.
⦁ Jiang, L.; Burke, S. D. Org. Lett. 2002, 4, 3411.
⦁ Trost, B. M.; Van Vranken, D. L. Angew. Chem., Int. Ed. Engl. 1992,
31, 228.
⦁ Baudry, D.; Ephritikhine, M.; Felkin, H. NouV. J. Chim. 1978, 2, 355.
⦁ Jiang, L.; Martinelli, J. R.; Burke, S. D. J. Org. Chem. 2003, 68, 1150.
⦁ Burke, S. D.; Buchanan, J. L.; Rovin, J. D. Tetrahedron Lett. 1991,
32, 3961.
⦁ Gao, Y.; Klunder, J. M.; Hanson, R. M.; Masamune, H.; Ko, S. Y.; Sharpless, K. B. J. Am. Chem. Soc. 1987, 109, 5765.
⦁ Burke, S. D.; Zhang, G.; Buchanan, J. L. Tetrahedron Lett. 1995,
36, 7023.
⦁ Steglich, W.; Hoefle, G. Angew. Chem., Int. Ed. Engl. 1969, 8, 981.
⦁ Broekema, R. J Recl. TraV. Chim. Pay-Bas. 1975, 94, 209.
⦁ Fuji, K.; Kawabata, T.; Kujita, E. Chem. Pharm. Bull. 1980, 28, 3662.
⦁ Ito, Y.; Hirao, T.; Saegusa, T. J. Org. Chem. 1978, 43, 1011.
⦁ Burke, S. D.; Quinn, K. J.; Chen, V. J. J. Org. Chem. 1998, 63, 8626.
⦁ Evans, D. A.; Ennis, M. D.; Mathre, D. J. J. Am. Chem. Soc. 1982,
104, 1737.
⦁ Keller, V. A.; Kim, I.; Burke, S. D. Org. Lett. 2005, 7, 737.

The Halichondrins and E7389 Chemical Reviews, 2009, Vol. 109, No. 7 3079

⦁ Cha, J. K.; Christ, W. J.; Kishi, Y Tetrahedron 1984, 40, 2247.
⦁ Burke, S. D.; Austad, B. C.; Hart, A. C. J. Org. Chem. 1998, 63, 6770.
⦁ Austad, B. C.; Hart, A. C.; Burke, S. D. Tetrahedron 2002, 58, 2011.
⦁ Krapcho, A. P.; Weimaster, J. F. J. Org. Chem. 1980, 45, 4105.
⦁ Lambert, W. T.; Hanson, G. H.; Benayoud, F.; Burke, S. D. J. Org. Chem. 2005, 70, 9382.
⦁ Kishi, Y.; Fang, F. G.; Forsyth, C. J.; Scola, P. M.; Yoon, S. K. Halichondrins and related compounds, US Patent, 5,338,865, 1995.
⦁ Wang, Y.; Habgood, G. J.; Christ, W. J.; Kishi, Y.; Littlefield, B. A.; Yu, M. J. Bioorg. Med. Chem. Lett. 2000, 10, 1029.
⦁ Seletsky, B. M.; Wang, Y.; Hawkins, L. D.; Palme, M. H.; Habgood,
G. J.; DiPietro, L. V.; Towle, M. J.; Salvato, K. A.; Wels, B. F.; Aalfs, K. K.; Kishi, Y.; Littlefield, B. A.; Yu, M. J. Bioorg. Med. Chem. Lett. 2004, 14, 5547.

⦁ Towle, M. J.; Salvato, K. A.; Budrow, J.; Wels, B. F.; Kuznetsov, G.; Aalfs, K. K.; Welsh, S.; Zheng, W.; Seletsky, B. M.; Palme,
M. H.; Habgood, G. J.; Singer, L. A.; DiPietro, L. V.; Wang, Y.; Chen, J. J.; Quincy, D. A.; Davis, A.; Yoshimatsu, K.; Kishi, Y.; Yu, M. J.; Littlefield, B. A. Cancer Res. 2001, 61, 1013.
⦁ Kishi has recently descibed a study aimed towards selective ami- nolysis of oxiranes in the context of E7389 synthesis: Kaburagi, Y.; Kishi, Y. Tetrahedron Lett. 2007, 48, 8967.
⦁ Dabydeen, D. A.; Burnett, J. C.; Bai, R.; Verdier-Pinard, P.; Hickford,
S. J. H.; Pettit, G. R.; Blunt, J. W.; Munro, M. H. G.; Gussio, R.; Hamel, E. Mol. Pharmacol. 2006, 70, 1866.
⦁ Newman, S. Curr. Opin. InVest. Drugs 2007, 8, 1057.
http://www.reuters.com/article/pressRelease/idUS55316⦁ +⦁ 01-Feb- 2008+PRN20080201. Retrieved May 12, 2009.