Oyster is an important bivalve species in the shellfish indus-try in Korea due to its high nutritional value and good taste. However, many reports have indicated their deterioration in quality including taste. Also the quality varies depending on the cultivation conditions, such as the quantity or species of feed planktons or water quality. Color changes in oysters, such as greening (Kimura, 1969) and red (Hata et al., 1982) and yellowish (Hata et al., 1987) coloration, are presumed to originate from feed plankton. Also, Hatano et al. (1990) re-ported an unacceptable taste and coloration of oysters in Ja-pan. Moreover, Parry et al. (1989) reported the bitter taste of cultured oysters during a bloom of the diatom, Rhizosolenia chunii in southern Australia. However, the compound causing the bitter taste was not identified. Despite the lack of official reports on quality problems, shellfish farmers relate stories of multiple annual quality deteriolations of cultured shellfish including oysters. Therefore, it is necessary to elucidate the reasons underlying this phenomenon to ensure quality control and food hygiene.

In Korea, Lee (1995) first reported bitter-tasting oysters cul-tured in Gamak Bay on the southern coast between November 1989 and January 1990. Smoked and boiled-canned oysters produced by using the oysters collected at the same location also tasted bitter. Although no report of illness followed the consumption of these bitter oysters, the taste seriously dam-aged the local oyster industry. The incident prompted us to elucidate the structure of the bitter compound. From oysters collected during that event, five bitter tasting compounds with peptidic natures tentatively named gamakamides A-E (derived from the name of Gamak Bay), were isolated (Lee, 1995). We aimed to report the structure of gamakamide-E .

Bitter-tasting oyster Crassostrea gigas, were collected dur-ing December, 1989, at a farm in Gamak Bay, Yousu Province, Korea, and the smoked and the boiled-canned oysters were provided by a local canned food company in Tongyeong.

The nuclear magnetic resonance (NMR) spectra were mea-sured using a JEOL GSX-400 (Jeol, Tokyo, Japan), and Var-ian Unity INOVA 600 (Palo Anto, CA, USA) in D₂O, CD₃CN and CD₃OD. High-resolution (HR) and low-resolution fast atom bombardment mass spectrometry (FAB-MS) spectra were measured with a JEOL JMS 303HF spectrometer (Jeol, Tokyo, Japan) and JEOL JMS 700 spectrometer (Jeol). Electrospray ionization mass spectrometry (ESI-MS) spectra were measured with a Finnigan Mat TSQ-700 spectrometer (San Jose, CA, USA). Circular dichroism (CD) spectra were re-corded on a JASCO J-720 spectropolarimeter (Jasco, Tokyo, Japan). Optical rotations were measured with a DIP-370 spec-trometer (Jasco).

Frozen raw (1 kg), smoked-canned (10 kg) and boiled-canned (17 kg) oysters were extracted with acetone three times. After evaporating off the acetone, the extract was parti-tioned between MeOH-H₂O (8:2) and hexane. A bitter residue obtained in the MeOH-H₂O (8:2) layer was next partitioned between CHCl₃ and H₂O. The organic fraction was evaporat-ed, dissolved in CHCl₃-MeOH (1:1), and passed through an alumina column (ICN Biomedicals, Seven Hills, NSE, Austra-lia). The column was washed with CHCl₃-MeOH (1:1) and the bitter compound was eluted with 1% NH₄OH-MeOH (1:1). The residue was chromatographed on a silica gel column (Merck, Darmstadt, Germany) with CHCl₃, CHCl₃-MeOH (9:1) and CHCl₃-MeOH (1:1) and the bitter compound was detected in the second eluate. The bitter substances were fur-ther purified on a Toyopearl HW-40 column (Toso, Tokyo, Ja-pan) with MeOH-H₂O (7:3), The constituents were dissolved in MeCN-H₂O (45:55) and passed through a Fuji Gel ODS column (Fuji Chemical, Tokyo, Japan) with the same solvent. Further high performance liquid chromatography (HPLC) pu-rification was performed on a Develosil ODS-7 column (10 × 250 mm; Nomura Chemicals, Tokyo, Japan) with MeCN-H₂O (45:55) and on an Asahipak ODP-50 column (0.8 × 250 mm; Showa Denko, Tokyo, Japan) with MeCN-H₂O (45:55, yield: 0.0002% against to the raw oyster).

Purified gamakamide-E (0.1 mg) was hydrolyzed with 6 N HCl in an evacuated tube by heating at 110℃ for 26 h. After cooling to room temperature, the reaction mixture was evaporated at 50℃ in vacuo. The resulting acid hydrolysate was mixed with 20 μL of 1% 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide (FDAA) solution in acetone and 10 μL of 0.1 M NaHCO₃, and heated at 37℃ for 1 h. After cooling to room temperature the reaction mixture was neutralized with 10 μL of 0.1 N HCl. After removing of the solvent, the re-action mixture was dissolved in MeOH. HPLC analysis was accomplished by using a Cosmosil 5C18-AR (4.6 × 250 mm; Nacalai Tesque, Kyoto, Japan) column with gradient elution by altering ratios from 1% AcOH-MeCN to 79:21 to 73:27 (85 min), to 64:36 (105 min) and to 27:73 (150 min). The flow rate maintained at 1.0 mL/min and UV detection was done at 340 nm. In Marfey analysis, gamakamide-E gave L-Met (55 min; D-Met, 88 min), L-Leu (96 min; D-Leu, 112 min), L-N-Me-phe (100 min; D-N-Mephe, 102 min) and racemic Phe-D-Lys dipeptide (108 and 111 min; Phe-L-Lys, 109 and 110 min).

From frozen and canned oysters, gamakamide-E was iso-lated as a colorless amorphous solid ([α]_D²⁰ -77.8° [c 0.036, CH₃OH]) after several chromatography steps (Lee, 1995). Amino acid analysis revealed two units of leucine (Leu) and one methionine (Met). The positive ion FAB MS/MS of ga-makamide-E (Fig. 1) gave a molecular ion peak [M + H]⁺ at m/z 836.5 and many fragment ion peaks such as [M - Leu]⁺ at m/z 723.5, [M - Leu - NMePhe]⁺ at m/z 562.4, [Phe + Lys + Leu]⁺ at m/z 415.4 and [Phe + Lys]⁺ at m/z 302.3, respectively. An intensive peak appeared at m/z 772.6 corresponding to [M - SOCH₃]⁺, suggesting the existence of methionine sulfoxide [Met(O)] in the molecule.

Conversion of L-Met(O) (Sigma, St. Louis, MO, USA) to L-Met by acid hydrolysis was confirmed by ESI MS and HPLC. Therefore, the existence of Met(O) in the molecule was confirmed. The molecular formula of C₄₃H₆₁N₇O₈S was deduced from HR FAB-MS ([M + H]⁺ m/z 836.4356 △ = -2.4 mmu).

Analysis of 2D NMR spectra, COSY, TOCSY, HSQC and heteronuclear multiple bond correlation (HMBC), allowed the complete assignment of these three amino acids as well as the assignment of signals for N-methylphenylalanine (N-Mephe), phenylalanine (Phe) and lysine (Lys). The ¹H NMR spectrum (Fig. 2) in CD₃CN displayed two sets of signals around me-thionine sulfoxide at a 1:1 ratio.

This ratio was unaffected by temperature alteration, indi-cating that the two sets of signals did not arise from confor-mational changes, but from stereoisomers of sulfoxide in Met (O). In the ¹³C NMR spectrum (Fig. 3), a ureido carbon was detected at 156.7 ppm. The remaining of 42 carbons were as-signed to the amino acid residues.

All these data about the ¹H NMR and ¹³C NMR summarized on the Table 1.

Sequencing of amino acids was mainly accomplished by

HMBC experiments (Fig. 4). Correlations between NH pro-tons and the N-methyl group with neighboring carbonyl car-bons were observed between 6-NH/C7, 8-NH/C13, Me22/C23, 24-NH/C28, and 29-NH/Cl (Fig. 5).

In the ROESY spectrum, NOE correlations among 8-NH/6-NH, 8-NH/H14, 24-NH/H29, and H14/H24 supported the amino acids sequence (Fig. 5). A hydantoin (Hy) structure was also determined by HMBC experiment. Correlations from the α-proton of Lys to both the carbonyl carbon of Phe and the ureido carbon, and from both α- and NH protons of Phe to the ureido carbon were observed. Based on these data, the planar structure of gamakamide-E was deduced (Fig. 6).

Acid hydrolysis of gamakamide-E, derivatization of the resultant amino acids with Marfey's reagent (Marfey, 1984), and subsequent HPLC analysis demonstrated L-stereochem-istry of Met(O), N-Mephe and two Leus. No D- and L-Lys or D- and L-Phe peaks were observed, suggesting that the hy-dantoin was not hydrolyzed. Instead of peaks corresponding to Phe and Lys, two unidentified peaks supposedly generated by a phenylalanine-lysine hydantoin dipeptide (Phe-Lys-Hy) were observed at 108 and 111 min. To determine the stereo-chemistries of Lys and Phe, four pairs of Phe-Lys-Hy were synthesized from the D- and L-lysine and D- and L-phenyl alanine and These amino acids were analysed by HPLC af-ter hydrolysed (6 N HCl, 110℃), resulting in two peaks on chromatograms by Marfey's method (Horigome, 2000). Di-peptides (L-Phe-D-Lys and D-Phe-D-Lys) gave two peaks with retention times identical to those of gamakamide-E (108 and 111 min), while D-Phe-L-Lys and L-Phe-L-Lys dipeptides gave two peaks at 109 and 110 min. Thus, the retention times of the two peaks depended on the stereochemistry of Lys, and the α proton of phenylalanine in hydantoin was racemized un-der hydrolysis conditions. Peaks from dipeptides containing D-Lys matched the unidentified peaks from gamakamide-E. Therefore, the lysine in gamakamide-E was determined to be in the D-form. The chemical shift of the α proton of Phe in gamakamide-E in CD₃OD agreed well with that of a methyl ester of L-Phe-D-Lys but not with that of D-Phe-D-Lys. Moreover, in the CD spectra of gamakamide-E, L-Phe-D-Lys, and D-Phe-D-Lys, both gamakamide-E and L-Phe-D-Lys but gave a negative Cotton effect at 220 nm in MeCN, while D-Phe-D-Lys gave a positive Cotton effect at 220 nm. These Cotton effects are thought to be generated from interactions between the aromatic Phe ring and the hydantoin. Therefore, the stereo-chemistry of Phe in gamakamide-E was determined to be the L-form. These results allowed us to determine the structure including the stereochemistry of gamakamide-E (Fig. 6).

Gamakamide-E was neither lethal to mice (100 μg/20 g mice ip) nor inhibited the growth of Aspergillus niger or Bacillus subtilis (Lee, 1995). Many cyclic peptides having strong and diverse biological activities have been isolated from sponges (Fusetani and Matsunaga, 1993) and ascidians (Davidson, 1993) but rarely from shellfish (Reese et al., 1996). The structure of gamakamide-E resembles the keramamides, anabaenopeptins and oscillamides isolated from sponges and cyanobacteria in having a ureido carbon and D-Lys in the mol-ecule (Kobayashi et al., 1991; Harada et al, 1995; Marsh et al., 1997). Another bitter-tasting shelIfish incident occurred in Australia and was associated with a bloom of the diatom R. chunii (Parry et al., 1989). It is thus possible that gamaka-mide-E was produced by plankton, probably either a diatom or cyanobacterium, and accumulated through the food chain. However, no bloom was observed in the water at the time of the occurrence of the bitter-tasting oysters used in this study. Our future tasks will be to elucidate the biological activities and biogenetic origin of gamakamide-E.