Synthetic efforts for stereo structure determination of cytotoxic marine natural product pericosines as metabolites of Periconia sp. from sea hare.

Pericosines are unique C7 cyclohexenoid metabolites of Periconia byssoides OUPS-N133 fungus that was originally isolated from the sea hare, Aplysia kurodai. Pericosines show significant in vitro cytotoxicity against P388 lymphocytic leukemia cells. Pericosine A, in particular, shows the most potent activity and significant in vivo antitumor activity against P388 cells. Thus, pericosines are promising candidates for seed compounds of anticancer drugs. However, before the total syntheses of pericosines were accomplished, their stereo structures could not be determined by spectral analyses because they have multi-functionalized cyclohexenoid structures with torsional strain. In this review, synthetic efforts for pericosines in this decade are surveyed.

Relenza and Tamiflu are the most popular anti-influenza drugs today and are therefore attractive target molecules for synthetic studies. Relenza is a potent anti-influenza drug designed from the transition structure of N-acetylneuramic acid-neuramidase complex based on X-ray analysis [32]. The oral anti-influenza drug Tamiflu has received immense attention for its effect on avian flu. To this day, however, the synthesis of Tamiflu remains an enigma for synthetic organic chemists. The independent asymmetric total synthesis of Tamiflu by Corey [35] and Shibasaki [36,37] from non-natural starting material, which was published in JACS, was a hot topic in 2006. In 2007, Fukuyama and co-workers reported the practical synthesis of Tamiflu [38].  In the past several decades, scientists working on new drug discovery have extended their research field to include the ocean because of its vast biological/chemical diversity and the fact that it composes almost 70% of Earth's surface. Marine natural products chemistry has produced a number of promising candidates for anticancer drugs and several of them have undergone pre-or phases I-III clinical trials [49,50]. Recently, however, a new tide has emerged, which involves the study of metabolites of microorganisms from marine sources [51][52][53]. It must be emphasized that researchers involved in this endeavor have been attempting preservation of environmental habitat in the ocean. In 1997, Numata and co-workers isolated unique C 7 cyclohexenoid-type metabolites from Periconia byssoides OUPS-N133 fungus that was originally isolated from the sea hare, Aplysia kurodai, and designated them as pericosines A 1 and B 2 (Figure 2), together with some macrospherides [54]. Full details of the isolation of pericosines A-E 1-5 were very recently reported by Numata and co-workers [55]. Pericosines showed significant in vitro cytotoxicity against P388 lymphocytic leukemia cells. Particularly, 1 was reported to have inhibitory activity against protein kinase and topoisomerase II in addition to significant in vivo antitumor activity against P388 cells. Thus, pericosines are thought to be promising candidates for seed compounds of anticancer drugs. However, total syntheses were required to confirm the stereo structures of most pericosines because the multifunctionalized cyclohexenoid structures with severe torsional strain have made structure determination by spectral analyses difficult. In this review, synthetic efforts to produce pericosines and analogues 1, 2, 4-7, and 9 from (-)-shikimic acid 10 or (-)-quinic acid 11 are surveyed comprehensively, together with the discovery of natural compounds.

Isolation of Pericosines from Periconia byssoides [54,55]
Numata and co-workers reported the isolation of C 7 cyclohexenoids designated as pericosines A-E 1-5 in 1997 and 2007, as summarized below. The fungal strain P. byssoides OUPS-N133 was isolated from a culture of A. kurodai by bioactivity-guided cytotoxicity assay using P388 cells. The fungal strain was cultured in artificial seawater medium containing 1% malt extract, 1% glucose, and 0.05% peptone adjusted to pH 7.5 at 27 o C. After 4 weeks, mycelia obtained by filtration of the broth were extracted with AcOEt. The AcOEt extract was separated by gel filtration through Sephadex LH-20 using MeOH-CH 2 Cl 2 (1:1), followed by silica gel column chromatography with CH 2 Cl 2 -MeOH gradient solvent system. This separation was performed along with in vitro P388 cytotoxicity assay. The active fraction that was eluted with 10% MeOH-CH 2 Cl 2 was further purified by reverse phase preparative HPLC using the MeOH-H 2 O solvent system to afford 1-5. Spectroscopic analyses including NMR, MS, and IR led to the elucidation of their structures. Early studies, however, indicated the structure of pericosine A as 6, that of pericosine C as 8, and that of pericosine D as 9 [56,57]. It is difficult to determine the relative configuration of such compounds as pericosines because they possess a multi-functionalized cyclohexenoid core with torsional strain. In their most recent report, the authors stated that they elucidated the structures of pericosines, except for pericosine A, by spectroscopic analyses. They determined the relative chemistry of pericosine B 2 on the basis of NOE between H-3 and H-5 observed in its acetonide 2a and small coupling constants J 3-4 = 5.7 Hz, J 4-5 = 3.1 Hz, and J 5-6 = 4.1 Hz. Long-range coupling between H-4/H-6, H-2/H-6, and H6/H-2 observed in 2a also supported their proposed configuration. Similar NOE between H-3 and H-5 and long-range coupling between H-2/H-6 and H-4/H-6 were observed in the spectra of 2, suggesting similar configuration. Pericosine C 3, which has the same planar structure as pericosine B 2, was transformed into acetonides 3a and 3b. Spectral analysis indicated that the 3,4,5-trihydroxyl groups on the cyclohexene ring had cis,cis configuration, and the stereochemistry at C-6 was determined to be different from that of 2. Finally, its identity was confirmed by comparing with synthesized compound 7. Interestingly, pericosine C exists as an enantiomeric mixture of 3 and 7, based on comparison of specific rotation values between natural pericosine C ([α] D 25 -4.8) and synthetic 7 ([α] D 25 +35.1). The relative chemistry of pericosine D 4 was determined on the basis of NOE between one acetonide-methyl/H-3, H-4 and another acetonide-methyl/H-5 observed in the NOESY spectrum of its 3,4-acetonide. The relative chemistry of pericosine E 5 was confirmed by X-ray analysis. Pericosine E 5 also exists as an enantiomeric mixture. Pericosines are responsible for the cytotoxicity of the extract of P. byssoides. The ED 50 values of pericosines are presented in Table 1. Compounds 1, 2, and 4 exhibited significant in vitro cytotoxicity against P388 cells. Most potent 1 also showed in vivo antitumor activity against P388 cells. Furthermore, 1 inhibited protein kinase EGFR and human topoisomerase II and showed selective growth inhibition against human cancer cell lines HBC-5 and SNB-75. Thus, pericosines were proved to be promising candidates for seed compounds of cancer preventive drugs.

Synthetic Efforts for Pericosines
As described above, the syntheses of pericosines are required not only to determine absolute configuration but also to confirm relative stereochemistry. The following are synthetic efforts by several groups carried out so far.

Total synthesis of pericosine B [58]
3.1.1. Donohoe's approach Donohoe and co-workers were the first to report the total synthesis of pericosine B (+)-2 in 1998. This is the only successful synthesis to date and has been achieved by using their original hydrogen-bond-directed dihydroxylation with osmium tetroxide to give cis,cis-triol. The synthesis is summarized in Scheme 1.
Commercially available (+)-dihydroxybromocyclohexadiene 12 was selectively protected to give silyl ether 13, which was then O-methylated with NaH and MeI and deprotected with TBAF to give alcohol 14. The subsequent hydrogen-bond-directed dihydroxylation of 14, which was the key step of this total synthesis, was carried out with stoichiometric osmium tetroxide and one equivalent of quinuclidine to give syn,syn-and syn,anti-triols, 15 and 16, in 2.2: 1 ratio. Separated 15 was protected with triethylsilyl (TES) groups to give 17, and 17 was methoxycarbonylated to yield 18. The synthesis was terminated by the addition of TFA to afford 2, which had the same spectroscopic data including specific rotation as those of the natural product, leading to the conclusion that the absolute stereochemistry is methyl (3S,4S,5S,6R)-3,4,5-trihydroxy-6-methoxycyclohexene-1-carboxylate. The problems of this synthesis included poor stereoselectivity in the crucial step and use of a stoichiometric amount of highly toxic osmium tetroxide.
3.1.2. Okamura's approach [59] Okamura and co-workers attempted the total synthesis of pericosine B also in 1998. The key reaction of their synthesis was the asymmetric Diels-Alder reaction of 3-hydroxy-2-pyrone 19 with a basic catalyst. Their synthetic approach is illustrated in Scheme 2.
The enantioselective Diels-Alder reaction between 3-hydroxy-2-pyrone 19 and acrylate with chiral auxiliary 20 was carried out in the presence of cinchonidine as basic catalyst to give endo-adduct 21 in 93% yield with 95% de. Adduct 21 was then treated with NaOMe to remove chiral auxiliary and the resultant methyl ester was dihydroxylated with catalytic osmium tetorxide and NMO to afford exo-diol, which was protected as benzylidene acetal 22. Reduction with LiAlH 4 followed by treatment with NaIO 4 gave diol 23, which was successively protected with a primary hydroxyl group such as TBS-ether followed by a secondary hydroxyl group such as pivaloyl ether to give ketone 24. Stereoselective reduction with NaBH 4 afforded β-hydroxyalcohol, which was then O-methylated and deprotected with TBAF to give hydroxymethylether 25. Unfortunately, the completion of this synthesis from the intermediate has not been reported so far.

Scheme 2.
Okamura's approach. [60] An approach taken by our group toward pericosine B is shown in Scheme 3. Racemic Diels-Alder adduct 27, which was formed by reacting methyl acrylate 26 with furan in the presence of ZnCl 2 as Lewis acid catalyst, was dihydroxylated with H 2 O 2 and catalytic osmium tetroxide with excellent exo-face selectivity. Acetonides 28 derived from the mixture of diastereoisomers 27 were treated with LHMDS in THF at -78 o C to give methyl 5-epishikimate derivative 29 as a single product, whose hydroxyl group at C-5 was protected as silyl ether 30. Compound 30 was dihydroxylated with osumium tetroxide to give sole product 31 with the desired stereochemistry. After many examinations of regioselective O-methylation and O-acylation of 31, it was protected at C-2 by acetylation to give 32, then deprotected with 5-TBS ether to give alcohol 33, which in turn was oxidized with Dess-Martin periodinane to give β-hydroxyketone 34. α,β-Unsaturated ketone 35 obtained by adding TFAA to 34 was reduced with NaBH 4 to give alcohol 36 with the desired stereochemistry of 37, followed by protection and deprotection. Alcohol 38 was oxidized with Dess-Martin periodinane to give unstable α,β-unsaturated enone 39, which could not be purified by silica gel chromatography. Crude 39 was reduced to generate hydroxyl group with the desired stereochemistry. However, not our objective reaction but an unexpected reaction occurred to yield 40. This work was published in 2004.

Garcia Ruano's approach [61]
Garcia Ruano and co-workers reported their attempt to synthesize pericosine B in 2005. As shown in Scheme 4, asymmetric Diels-Alder cycloaddition reaction between chiral 3-sulfinylacrylonitrile 41 and furan with Me 2 AlCl as Lewis acid catalyst gave endo adduct 42 in 53% yield together with exo adduct in 10% yield [62]. The double bond in 42 was dihydroxylated exo-face selectively. The newly generated two hydroxyl groups were protected as acetonide 43, and this was converted into α,β-unsaturated nitrile 44 by basic treatment with NaNH 2 . Treatment with MeOH, followed by triethylamine and TMSOTf, afforded ring-opened nitrile 45, which seemed to be the last intermediate in the synthesis of 2. However, the subsequent transformation of cyano group into methoxycarbonyl group by methanolysis has not been reported so far. Scheme 4. Garcia Ruano's approach involving asymmetric Diels-Alder cycloaddition.

Total synthesis of epimer of pericosine B: Synthesis of pericosine C [63]
As described above, our synthesis of pericosine B could not be completed. Thus, we attempted to synthesize the epimer of pericosine B 7, as summarized in Scheme 5. Known methyl shikimate derivative 46 from (-)-quinic acid 11 was converted into TBS ether 47, which was reacted with catalytic osmium tetroxide and one equivalent of trimethylamine-N-oxide under reflux with tBuOH-pyridine-H 2 O (20:5:1) to give a mixture of diols 48 and 49 in 3 : 1 ratio in 40% yield with recovery of 47 (33%). Major diol 48 was selectively O-methylated at C-6 to give 50 in 66% yield as a single product, whereas minor diol 49 was singly 6-O-methylated to give 51 in 16% yield with 20% of 1,6-bis-O-methylated product. 6-O-Methyl ether 51 was further examined for its potential use in the synthesis of pericosine B 2, but all attempts were unsuccessful. Deprotection of methyl ether 50 with TBAF produced alcohol 51, and oxidation of 51 with Dess-Martin periodinane gave ketone 52, which in turn was dehydrated with TFAA to yield unsaturated ketone 53. Enone 53 was reduced with NaBH 4 in excellent stereoselectivity to give alcohol 54, and this was then converted into target molecule (+)-7. When tested for in vitro for cytotoxicity against P388 cells, obtained (+)-7 gave an ED 50 value of 17.8 μg/mL, which indicated lower activity than natural pericosine B (ED 50 : 4 μg/mL).
Thus, we concluded that the stereochemistry of C-6 had a significant influence on cytotoxicity. When this work was published in 2004, synthesized (+)-7 was a non-natural product. However, it was elucidated later that (+)-7 had the same relative chemistry as natural pericosine A 1 and was a component of natural pericosine C, which is an enantiomeric mixture of 3 and 7 [55].

Scheme 5.
Synthesis of epimer of pericosine B from (-)-quinic acid. [64,65] After the publication of the total synthesis of pericosine B 2 by Donohoe, there had been no reports of pericosine A in spite of its antitumor activity. Then, the total synthesis of initially assigned pericosines A 6 and D 9 was attempted by our group because those two target compounds seemed to be synthesized via common intermediate enone 55 derived from (-)-quinic acid 11. Our retro-synthetic strategy is shown in Scheme 6. Key reactions of this synthesis included α-face selective chlorination of silylenolether derived from known ketone 59 [66,67] and reagent-dependent stereoselective reduction of 55. Scheme 6. Retro-synthetic strategy of 6 and 9.

Total synthesis of initially assigned pericosines A and D
The synthesis of 6 was carried out as illustrated in Scheme 7. Lactone 59 derived from (-)-quinic acid 11 according to literature [66,67] was chlorinated with NCS with excellent face selectivity to give α-chloroketone 58 via silylenolether in 45% yield in 2 steps. The following reduction of 58 with NaBH 4 gave alcohol 57 as the sole product. The chemical yield of 57 was improved by sequential reactions from 45% in 2 steps to an overall yield of 57% in 3 steps from ketone 59. Treatment of 57 with TFA in MeOH under reflux afforded tetraol 60 in 38% yield. Then, 60 was converted into acetonide 61, which was oxidized with Dess-Martin periodinane to give hydroxyketone 56. The isopropylidene moiety of 56 was removed with TFA because 56 could not be dehydrated by any dehydrating agents. One plausible explanation was that 56, conformationally fixed by an isopropylidene bridge, could not be transformed into a suitable transition state for dehydration. Resultant triol 62 was converted into reactive hydroxyketone 63 that was in turn dehydrated to enone 64 with Martin's sulfrane dehydrating agent (bis[α,α-bis(trifluoromethyl) benzyloxy] diphenyl sulfur) [68] in 65% yield.

Scheme 7. Total synthesis of initially assigned pericosine A 6.
Treatment of enone 64 with TFA in MeOH gave 55, which was reduced stereoselectively with tetrabutylammonium triacetoxyborohydride [69] to afford 6 with the desired configuration as a single product. However, disagreement of the NMR data of 6 and acetonide 65 with those of natural pericosine A and those of its acetonide described in the literature [54] led us to conclude that the proposed stereochemistry of pericosine A was incorrect. Furthermore, the fact that "the isopropylidene bridge in 65 derived from 6 was located between C-4 and C-5, whereas that of the acetonide of pericosine A was between C-3 and C-4" [54] supported our conclusion.
Common intermediate 64 was reduced with NaBH 4 to afford allyl alcohol 66 as a single product with opposite stereoselectivity. Then, 66 was deprotected to afford 9 quantitatively in 2 steps, as shown in Scheme 8. However, 9 was different from pericosine D. Then, product 9 was transformed into acetonide 67 to confirm the stereochemistry of the synthesized molecule.
The stereochemistry of all intermediates in this study was carefully determined by analyzing various kinds of 2D NMR spectra. Scheme 8. Synthesis of initially assigned pericosine D 9. [70,71] From our conclusion in the previous study described in Section 3.3, determination of the true structure of pericosine A by total synthesis became our next task. In a comprehensive review of data related to pericosines [54,63,64], close similarity between the 1 H-NMR coupling constants of natural pericosine A [54] and those of 7 [63] led us to deduce that the structure of natural pericosine A was 1, as illustrated in Figure 3. Since 1 had the same relative chemistry as 7, our basic synthetic strategy was almost the same as that for 7. After a number of trials, the total synthesis of (-)-1 was achieved, as shown in Scheme 9. Known methyl 5-epishikimate derivative 68 derived from (-)-shikimic acid 9 was subjected to Dess-Martin oxidation to afford β,γ-unsaturated ketone 69. Without purification, 69 was reduced with NaBH 4 to give alcohol (-)-70, which in turn was protected with TBSCl to give silyl ether 71. After dihydroxylation of 71, resultant diol 72 was acetylated to yield 73, and this was deprotected to give 74. Subsequent Dess-Martin oxidation of 74 gave β-hydroxyketone 75, which was dehydrated with TFAA to afford α,β-unsaturated ketone 76. Subsequent reduction of 76 was carried out carefully with stoichiometric NaBH 4 at -78 o C in dry THF to afford alcohol 77 possessing the desired stereochemistry in 95% yield.
Subsequent protection of 77 was carried out with careful addition of TBSCl and a stoichiometric amount of imidazole to give silyl ether 78 in 53% yield, followed by deacetylation with K 2 CO 3 to afford enol 79 in 74% yield. The key reaction of this total synthesis, which was aimed at introducing a Cl atom, was achieved by the addition of excess SOCl 2 to 79 in dry CH 2 Cl 2 to afford chlorinated product 80 in 42% yield [71], whereas the yield of the reaction with stoichiometric SOCl 2 was 10% [70]. To our surprise, 80 was formed with rearrangement of the double bond. The structure of key intermediate 80 was confirmed by detailed 1D and 2D NMR studies. In the NOESY spectra, cross peaks H-5/t-Bu, H-6/t-Bu, H-5/SiMe, H-6/SiMe, and H-3, H4/one of cyclohexyl methylenes, were observed. HMBC cross peak H-3, H-4/singlet carbon of 80, which was observed at 110.8 ppm, confirmed that the Cl atom was introduced not via an S N i mechanism as we had aimed early on, but via an S N 2' mechanism with syn selectivity. Another plausible mechanism is the [3,3]-sigmatropic rearrangement of chlorosulfonate derived from 79. In spite of detailed analysis of NMR spectra, the stereochemistry at C-6 in 80 could not be determined at this step. This total synthesis was completed with TFA to give final product (-)-1 in 66% yield, which was not 9 as synthesized previously by us [64,65] but pericosine A. Thus, this result proved the stereochemistry at C-6 in 80. Except for the sign of the specific rotation, (-)-1 showed the same spectroscopic data, including HPLC retention time, as the natural product, and the absolute configuration of natural pericosine A was assigned as methyl (3S,4S,5S,6S)-6-chloro-3,4,5-trihydroxy-1-cyclohexene-1-carboxylate. The first total synthesis of the antipode of natural pericosine A was completed in this manner.
Next, the total synthesis of natural pericosine A (+)-1 was examined. As the preparation of (+)-70, which is an antipode in the preceding synthesis, from 11 has been reported, a similar strategy toward (-)-1 could be applicable. The synthesis of (+)-1 is summarized in Scheme 10 [71]. In our synthesis, the preparation of (+)-70 was modified from the original method appearing in the literature. Treatment of hydroxylactone 78 derived from 11 with NaOMe in MeOH followed by neutralization with DOWEX ® 50W-X8 gave crude diol 79, which was obtained by only filtration and used in the next reaction without further extraction or purification. Diol 79 was oxidized to β-hydroxyketone 80, which was then dehydrated with TFAA to produce enone 81. Reduction with NaBH 4 afforded enol 82 in 54% overall yield from 78. Εnol 82 was converted into (+)-70 according to literature [72].
The following transformation of (+)-70 into (+)-1 was accomplished as above. Since all spectral data of synthesized (+)-1, including specific rotation and HPLC retention time, agreed with the data of natural pericosine A, the first synthesis of natural pericosine A was completed.

Further Discussion
As described above, pericosines are interesting compounds because of their potent biological activities and unique structures. In particular, the role of pericosine A 1 in anti-tumor, protein kinase EGFR inhibition or topoisomerase II inhibition was demonstrated. In terms of chemical structure, the relative chemistry of pericosines A 1, C 3, and B 2 is so unique that it is difficult to find the same configurational carbasugar natural products, whereas pericosine D 4 has the same relative configuration as (+)-MK7607. We are extremely interested in the fact that pericosines C 3 and E 5 exist as an enantiomeric mixture. How are they biologically formed? Pericosine E 5 is thought to be a conjugate of pericosines A 1 and B 2 but with different chiral sense. This suggests the possibility of the presence of antipodes of 1 and 2 or unknown analogues with other combinations. We are currently directing our efforts toward designing a more effective synthetic route for pericosine A 1 because of the low total yield in our completed first synthesis. Nevertheless, the SN2' type reaction, which is the key step for Cl introduction in our previous work [70,71], may help unravel the mystery surrounding the chirality of percosines. A reaction similar to this or the enantioselective dehydration from pro-chiral 3-epiquinate may occur in fungal metabolic systems to generate chirality. Indeed, it seems unbelievable that enantiomeric mixtures of 3 and 5, particularly 5 that bears 8 chiral centers, could be formed biologically. In that sense, pericosines are truly exciting compounds. Scheme 10. Total synthesis of (+)-pericosine A from (-)-quinic acid.
On the other hand, we would like to point out the ambiguity of the conformation of pericosines. It is absolutely difficult to determine the configuration or conformation of pericosines only from spectral analysis. 1 H-NMR coupling constants of pericosines and acetonides are shown in Figure 4. The correct structure of 1 could not be concluded only from the synthesis of 1. The stereochemistry of C-6 in intermediate 80 could be determined not from detailed spectral analysis but from the difference between deprotected 1 and 9, the latter of which was synthesized in our previous study [64,65]. We hope to solve this problem by integrating our vast synthetic efforts with data from other researchers. The synthetic study of pericosine D is ongoing in an effort to determine its absolute configuration [73]. After elucidating all the structures of pericosines, we will examine the biological activity of synthesized pericosine analogues, develop more effective synthetic routes, and design more active molecules based on these seed compounds.