Pergamon

0040-4039(94)02423 -5

Tetrahedron Letters, Vol. 36, No, 7, pp. 991-994, 1995 Elsevier Science Lid

Printed in Great Britain 0040-4039/95 $9.50+0.00

The Isolation and Structures of Unusual 1, 4-Polyketldes from the Dinoflagellste, Amphidinium sp.

Inee Bauer, Luck) Maranda, Kurt A. Young and Yuzuru Shimizu* Pharmacognosy and Marina Biotechnology Laboratory, College of Pharmacy,

University of Rhode Island, Kingston, RI 02881

Stella Huang Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, CT 06492

Abstract: Unusual 1,4-polyketides with cytotoxic activity against human colon tumor cells were isolated from the dinoflagallate, Amphidinium sp. end their significance with respect to the mode of dinoflagellate polyketide biosynthesis was discussed.

Dinoflagellates have been recognized as a valuable source of compounds with novel chemical structures and highly

specific bioactivity. They include compounds such as highly selective sodium channel blockers or activators, calcium

channel activators, protein phosphatase inhibitors and antineoplastic compounds. 1, 2 In our search for antitumor agents

in dinoflagellates, we found cytotoxic open-chain 7-tetrakstones which belong to a new type of 1, 4-polyketides.

The extract of the organism, Amphidinium sp. $1-36-5 isolated from the water at Brewers Bay, St. Thomas, U.S.

Virgin Island, 3 showed strong cytotoxicity against various cell lines including the human colon tumor HCT 116. We have

already reported the isolation of four active compounds from the fractions with medium polarity. 4,5 In the nonpolar

fraction preceding fatty acids in the normal phase silica gel chromatography, we discovered two isomeric compounds,

amphidinoketide I,1 and amphidinoketide II, 2, with cytotoxicity against HCT 116.

Amphidinoketide I, 1 was obtained as a colorless oil, [0qo25=+25.3 ° (CH2CI2) in a yield of 0.0049 % of dried cell

weight. It has a UV maximum, ;m~x=235 nm (~=11,000, MeOH). High resolution electron impact mass spectrometry

(HREIMS) gave a molecular formula, C24H380 4 (caicd 390.2771; found 390.2771), which is in good agreement with the

proton and carbon numbers obtained from the NMR spectra. Amphidinoketide II, 2, which was also obtained as an oil,

[~]D25=+33.9 (CH2CI2), in a yield of 0.014% of dried cell weight, has the same molecular formula, C24H380 4 (calcd

390.2771 ; found 390.2768) as 1 and is the deconjugated isomer of I (no UV maximum above 200 nm). The compound

2 was observed often accompanied with a trace amount of I during the isolation procedure. Thus, it is possible that I is

an artifact formed during the isolation process, although 1 was clearly present in the fresh extract of the organism. The

proton and carbon NMR spectre of I differed from those of 2 only in the region close to the conjugated ketone. The rest

of the spectra had almost identical chemical shifts and coupling pattems, thus most of structural elucidation was done

with 2, which was obtained in a larger proportion.

The compound 2 has four free ketonas (6 207.2, 212.0, 208.2 and 213.1 ppm), which account for all oxygen atoms

in the molecule. There are four methyl and two terminal methylene groups, of which one methyl and one methylene

belong to the terminal isobutenyl moiety. The analysis of the NMR data (1 H, 1 H-COSY, DEPT and H ETCOR, Table 1)

led to the five spin systems, A~E, which are disrupted by four keto groups interspaced two carbons apart. There are

twelve possible combinations for the assembly of the partial structures, and the presence of the repetitive units and

closeness of chemical shifts made the assignment rather difficult. Nonetheless, after extensive COLOC and HMBC NMR

experiments, we were able to reach a single structure.

In comparison with other open chain alkyl ketones, it was supposed that the two keto groups with larger chemical

shifts: 8 212.0 and 213.1 ppm, carry an co-methyl group each, but not two c¢-methyl groups, which would bring the

chemical shift further down field, s In fact, all of the keto-carbons showed, at least, one correlation with (~-mathylene

991

992

Figure 1. 1.71 dd (t.4, t.0)

A , B i

4.91 ~ . t 11.

1 t4 .9 .~- . . ~ I

t.06 d 3H (6.2) 16.8 2.¢4m

C D

2.91 ~ 113.3. 9.3) ' O

0.~ a 3H IS.S)

i 20.0 1.97 m ~ ~ ~O I 1.~ 1.30m2H 5.81 ddt(18.9,10.2,8.7 )

E : 21~ ,~.0 = ~ 1 % = = , : ) f 1.~m z o o m 2 . (lo.1, 2~.zt) i 2..48 old (1EU), 6.3) 1.t2m 4.9e ddt i z.32 dd (t8.9, 7.3) (16.9, 2.3, t.6)

, v . v . oyO °V.oy. .v.v=

0 1.04 d (8.4)

: : = ' ~ * 44,616J) = 2.~ rn

3.04 ~ (17.S, g3) " (~)

, , . . . . , 5 / =

2 (Arrows show the presence of H-C long-range couplings)

Scheme 1, Proposed mass fragmentation of amphidinoketide I and II, 1 & 2

L

ot +'O'--. H nYz 373 CI (t00 % in 2) nYZ 276 (30 % El ol t; 20% El & CI of 2)

* + O nYz 205 (100 % El of t. 40 % CI of 2) m/z 167 (100 % El of 2; 12.5 % El of 11

groups in the COLOC and HMBC NMR spectra. The ketone at 8 207.2 ppm was correlated with the C3 methylene group

and one of the four isolated methylene protons between 2.52 and 2.78 ppm. Furthermore, the ketone at 208.2 ppm

was correlated to one of these protons as well as to the methylene protons at ~ 2.41 and 2.91 ppm in an isopropylene

link. Similarly, a correlation was observed between the carbonyl carbon at ~ 212.0 ppm and the methylene protons at 8

2.44 and 3.04 ppm of the other insetting isopropylene moiety. Finally, a correlation was observed between the

carbonyl carbon at S 213.1 ppm and the C14 methylene which is at the end of the terminal section (E) of the molecule.

These results can be explained only by the structure, 2 with methyl groups at g and 12 positions as shown in Figure 1.

The chemical shifts of all the carbonyl carbons are in good agreement with values expected from the literature. 6

993

Table 1. NMR Data of Amphidinoketide II, 2 from Amphidinium sp.

pos 13 c 1H J [Hz] COSY

1 114.9 t a: 4.91 br.t 1.7 lb, 21 b: 4.79 m 0.9 la, 3, 21

2 140.1 s 3 52.3 t 3.12 d (2H) 0.9 1 b 4 207.2 s 5 36.4* t 2.78- 2.52 6 35.9" t m(4H) 7 208.2 s 8 46.0 t a: 2.91 dd 13.3, 9.3 8b, 9

b: 2.41 dd 13.3, 3.6 8a, 9 9 41.4 d 2.94 m 8a, 8b, 22

10 212.0 s 11 44.6 t a: 3.04 dd 17.5, 9.3 11b, 12

b: 2.44 dd 17.5, 3.6 11a, 12 12 41.7 d 2.94m 11a, 11b, 23 13 213.1 s 14 49.0 t a: 2.48 dd 16.9, 6.3 14b, 15

b: 2.32 dd 16.9, 7.3 14a, 15 15 29.0 d 1.97 m 14a, 14b, 24, 16a,b* 16 36.8 t a: 1.32 m 16b, 15", 17a*, 17b*

b: 1.12 m 16a, 15", 17a*, 17b* 17 26.8 t 1.30 m (2H) 16a*, 16b*, 18" 18 34.4 t 2.00 m (2H) 17*, 19, 20a, 20b 19 139.6 d 5.81 ddt 16.9, 10.2, 6.7 18, 20a, 20b 20 114.4 t a: 4,98 ddt 16.9, 2.3, 1.6 18, 19, 20b

b: 4,92 ddt 10.2, 2.3, 2.1 18, 19, 20a 21 22.8 q 1,71 dd (3H) 1.4, 1.0 la, lb 22 16.8 q 1,04 d (3H) 6.2 9 23 16,8 q 1,06 d (3H) 6.4 12 24 20.0 q 0.86 d {3H) 6.6 15

* Due to the overlapping of groups of signals assignments and correlations are tentative.

Table 2. NMR Data of amphidinoketide I, 1 from Amphidinium sp.

pos 13 C 1H

1 27.7 q 1.87 d (0.8) 2 155.4 s 3 123.8 d 6,08 br. s 4 198.9 s 5 38.0 t 2.78- m, 6 36.6 t 2.52 (4H) 7 208.5 s 8 44.7 t a: 3.04 dd (17,3,9.3)

b: 2.45 (dd) (17.3) 9 41.7 d 2.95 m

10 212.1 s 11 46.1 t a: 2.92 (dd) (15.1,9.5)

b: 2.45 ? 12 41.4 d 2.95 m 13 213.2 s

pos 13 C

14 49.0

15 29.0 16 36.8

17 26.8 18 34.4 19 139.6 20 114,4

21 20,8 22 16.8 23 16.8 24 20.0

COLOC & HMBC {300 & 500 MHz}

H3,H5

H6,H8 H22

H22 Hl l H23

H23 H14

1 H

t a: 2.47 dd (16.8, 6.1) b: 2.32 dd (16.8, 7.3)

d 1.99 m t a: 1.37 m

b: 1.12 m t 1.30 m(2H) t 2.00 m (2H) d 5.81 dot (17,10,5.7) t a: 4.98 m(17)

b:4.92 m(10) q 2.09 d(0.8) q 1.06 d(6.1) q 1.04 d(7.0) q 0.86 d {6.6}

The fragmentation patterns in the mass spectra also corroborate the structure. Both I and 2 characteristically gave a

prominent dehydration peaks, m/z 372 or 373 (M-18 or MH-18), although there is no hydroxyl group in the molecule. In

both El and CI spectra, m/z 205 and m/z 167 were the base peaks for 1 and 2 respectively. There is also a strong peak at

m/z 276. These peaks can be explained well by formation of a furan ring via a cyclic hemiketal intermediate and

subsequent allylic cleavage or MacLaffarly rearrangement (Scheme 1).

The structures of amphidinoketides are intriguing with respect to the biosynthesis of the pelyketide-like metabolites

found in dinoflagellates. Since Birch proposed the polyketide hypothesis, 7 a vast number of metabolites have been

found to be biosynthesized by the pathway. In most compounds, the oxygen substitutions follow the 1,3-pattern

994

expected from the chain formation by the Claisen type condensation on o¢-posltion of carboxylic acids such as acetate or

propionate, although some may be altered by dehydration or hydroxylation. In dinoflagellates, however, many

metabolites have structures which do not fit typical polyketides. In fact, in the biosynthetic studies of brevetoxins, the

feeding of 13C-acetate did not necessarily result in the expected incorporation pattern.S,9,10 As a result, the possibility

of an alternative pathway involving the condensation of dicarboxylic acids was implicated (Scheme 2a). It may be

coincidental, but the discovered structure, which has four ketones at 1,4-positions, nicely fits the hypothesis. An

alternative explanation would be the mere collapse of appropriately located epoxide rings which are formed from a

polyene chain (,Scheme 2b). Similar polyepoxides are also proposed as intermediates in polycyclic ether formation.

Scheme 2. Two possible biosynthetic pathways for 1, 4-polyketides.

a. R OH ,_,..~.~ R OH .....

O O O O

b, R ~ f ~ = ~ = R ~ ~ j '~ '~/ H ~ H O O

In a preliminary screening, amphidinoketide I, 1 showed significant cytotoxicity against the human colon tumor cell

line HCT-116 and its drug-resistance strain (ICso 4.9814g/mL). The deconjugated isomer, 2 showed much weaker

activity (ICs0 7314g/mL). Bioassay results will be published elsewhere.

Acknowledgment: This work was supported by NIH grants CA 49992 and CA 50750, which is greatly appreciated.

Thanks are also due to Dr. Craig Fairchild, Ms. Laurie Come, and Ms. Judy MacBeth, Bristol-Myers Squibb for cytotoxicity

tests and Dr. Michael McGregor of URI for NMR technical assistance.

Re fe rences

1. Shimizu, Y. Chem. Rev, 1993, 93, 1685-1698; Shimizu, Y. In Marine Biotechnoiogy, Vol. 1 ; Attaway, D. H.;

Zaborsky, O. R. Ed.; Plenum Press: New York 1993, 391-410, and references therein.

2. Kobayashi, J.; Ishibeshi, M. Chem. Rev., 1993, 93, 1753-1769, and references therein.

3. This strain of Amphidinium sp. is unusually large and probably represents a new species.

4. Bauer, I.; Maranda, L.; Shimizu, Y.; Peterson, R. W.; Comell, L.; Steiner, J. R.; Clardy, J. J. Am. Chem. Soc., 1994,

116, 2657-2658.

5. Bauer, I.; Maranda, L.; Young, K. A.; Shimizu, Y.; Fairchild, C.; Comell, L.; MacBsth, J. ; Huang, S. J. Org, Chem., in

press.

6. Braifmaier, E.; Voelter, W. Carbon-13 NMR Spectroscopy; High-Resolution Methods and Applications in Organic

Chemistry and Biochemistry, 3rd Ed.; VCH; Weinheim, 1989; pp. 216-218.

7. Birch, A. J.; Donovan, F. W. Austral. J. Chem., 1953, 6, 360.

8. Lee, M. S.; Repeta, D. S.; Nakanishi, K.; Zagoraki, M. G. J. Am. Chem. Soc. 1986, 108, 7855.

9. Chou, H. N.; Shimizu, Y. J. Am. Chem. Soc. 1987, 109, 2184.

10. Lee, M. S.; Quin, G-W.; Nakanishi, K.; Zagorski, M. G. J. Am. Chem. Soc. 1989, 111, 6234.

(Received in USA 9 November 1994; revised 6 December 1994; accepted 9 December 1994)