Structure and Stereochemistry of Amphidinolide N Congeners from Marine Dinoflagellate Amphidinium Species

Abstract

Two highly potent cytotoxic 26-membered macrolides, isocaribenolide-I (1) and a chlorohydrin 2, together with known amphidinolide N (3), have been isolated from a free-swimming dinoflagellate Amphidinium species (KCA09053 and KCA09056 strains) collected off Iriomote Island, Japan. The structures of 1 and 2 were determined to be a congener of 3 with an isobutyl terminus and the chlorohydrin form of 3, respectively, by detailed analyses of spectroscopic data. The relative stereochemistries of 1 and 2 were elucidated by the conformational analyses based on NMR data.

Introduction

Marine dinoflagellates of the genus Amphidinium have been recognized as a valuable source of polyketides with novel chemical structures and interesting bioactivities.¹⁾ The macrolide group represented by amphidinolides are unique natural products with cytotoxic activity.^2,3) In 1994, Kobayashi and colleagues had reported a 26-membered macrolide, amphidinolide N,⁴⁾ isolated from marine symbiotic dinoflagellate Amphidinium species. Amphidinolide N exhibits the most potent cytotoxic activity among the amphidinolide macrolides. Soon after, Shimizu and colleagues isolated an antitumor 26-membered macrolide, named caribenolide-I,⁵⁾ from the marine free-swimming dinoflagellate Amphidinium gibbosum.^6,7) Thereafter, Kobayashi and colleagues reported a structural revision of the dihydroxy form at C-21 and C-24 into a tetrahydrofuran structure, and the relative stereochemistry of amphidinolide N was proposed.⁸⁾ Although the planar structure of amphidinolide N is the same as that of caribenolide-I, it is unclear whether amphidinolide N is identical to caribenolide-I or not, because the spectral data of both compounds have not been compared.

As synthetic approaches to determine the structure of amphidinolide N and/or calibenolide-I, Nicolaou et al. accomplished the synthesis of several analogs of amphidinolide N and caribenolide-I in 2006.^9,10) The total synthesis of the 7,10-epimer of amphidinolide N was reported by Hayashi and colleagues,^11,12) and several synthetic studies for this amphidinolide N/caribenolide-I have been carried out.^13–17) Recently, Trost et al. accomplished the synthesis of des-epoxy amphidinolide N,¹⁸⁾ and proposed that amphidinolide N and caribenolide-I are identical on the basis of the chemical shift differences between natural specimens and their synthetic product. However, the stereostructures of amphidinolide N and calibenolide-I have not yet been elucidated.

During our continuing search for new cytotoxic metabolites from extracts of marine dinoflagellate Amphidinium species,^19,20) we investigated the cytotoxic components contained in extracts of Amphidinium KCA09053 and KCA09056 strains, and found two new highly potent cytotoxic macrolides, isocaribenolide-I^21,22) (1) and a chlorohydrin 2,^22,23) together with a known amphidinolide N (3) (Fig. 1). The structures of 1 and 2 were confirmed by spectral data analyses, indicating that compounds 1 and 2 correspond to a congener with the isobutyl terminus of 3 and chlorohydrin form of 3, respectively. The relative stereochemistries of these compounds were elucidated by conformational analyses based on NMR data. Herein, we describe the isolation and provide a structural elucidation of 1 and 2. We also report that the physicochemical properties of compound 3 were identical to those of both amphidinolide N and caribenolide-I.

Fig. 1. Structures of Isocaribenolide-I (1), Chlorohydrin 2, and Amphidinolide N (Caribenolide-I) (3)

Results and Discussion

The dinoflagellate Amphidinium species (strain number KCA09053) was separated monoclonally from benthic sea sands collected off Iriomote Island, Japan.²⁴⁾ The dinoflagellates were cultivated in Muroto deep seawater containing 1% Provasoli’s enriched supplement for 14 d with mechanical stirring and illumination. The algal cells of the KCA09053 strain (dry weight 10.3 g from 150 L culture) were extracted using a methanol (MeOH)–toluene solvent system. The toluene-soluble materials of the algal extract were subjected to a silica gel column chromatography eluted by chloroform (CHCl₃)–MeOH. The crude fraction, which exhibited >90% cell-growth inhibition of human cervix adenocarcinoma HeLa cells at 1 µg/mL, was chromatographed by octadecylsilyl (ODS) and amino silica gel column chromatography and then ODS HPLC to afford isocaribenolide-I (1, 0.006% from dry weight) together with iriomoteolides-10a, 12a,²⁵⁾ and 13a²⁴⁾ and amphirionin-5.^26,27) The Amphidinium KCA09056 strain was separated from the surface of the seagrass collected off Iriomote Island.²⁸⁾ The toluene-soluble materials of the microalgal extract were subjected to silica gel column chromatography, and then the crude fraction showing potent cytotoxic activity was chromatographed using an ODS column and then ODS HPLC to afford two cytotoxic components, compounds 2 (0.0054%) and 3 (0.0024%).

Compound 3 was found to have the same molecular formula (C₃₃H₅₂O₁₁) as amphidinolide N^4,8) and carbenolide-I⁵⁾ by high-resolution (HR) electrospray ionization (ESI)-MS data [m/z 647.3400 (M + Na)⁺, Δ −0.23 mmu]. The ¹³C and ¹H chemical shifts for 3 in benzene-d₆ (C₆D₆) (Table S1) were well consistent with those of amphidinolide N.⁴⁾ The absolute values of the ¹³C chemical shift differences between this sample and the reported data fell in the range of 0.03 ppm, and the ¹H-NMR data of 3 and amphidinolide N were also close to each other. The specific rotation {[α]_D¹⁹ +30 (c = 0.49, MeOH)} of this sample showed the same positive sign as the literature value {[α]_D³⁰ +20° (c = 0.5, MeOH)}, thus suggesting that compound 3 corresponds to amphidinolide N. Furthermore, the ¹H-NMR spectrum of 3 in chloroform-d (CDCl₃) was also identical with that of caribenolide-I reported by Shimizu and colleagues.⁵⁾ The optical rotation value {[α]_D¹⁹ +27 (c = 0.49)} in dichloromethane (CH₂Cl₂) of our sample showed the same sign as that reported for the literature value {[α]_D²⁵ +91 (c = 0.13, CH₂Cl₂)} for caribeonolide-I, thus indicating that compound 3 and caribonolide-I are identical. Therefore, we concluded that amphidinolide N and caribenolide-I were the same substance.

Structures of Isocaribenolide-I (1) and Chlorohydrin 2

Compound 1 was obtained as a colorless amorphous solid and was optically active. The molecular formula of 1 was established as C₃₃H₅₂O₁₁ based on HR-ESI-MS data [m/z 647.3398 (M + Na)⁺, Δ −0.40 mmu]. The ¹³C-NMR (Table 1) spectrum in C₆D₆ showed the presence of 33 carbon signals comprising one ketone carbonyl, one ester carbonyl, two sp² quaternary carbons, one sp² methine, one sp² methylene, one hemiketal, 13 sp³ methines (10 of which were oxygenated), eight sp³ methylenes, and five methyl carbons. Four out of eight degrees of unsaturation were accounted for by two olefins and two carbonyl groups, thus implying that 1 possessed four rings.

Table 1. ¹H- and ¹³C-NMR Data of Isocaribenolide-I (1) in C₆D₆

Position	13C		1H
1	173.6	C
2	44.8	CH	2.74	qin, 7.0
3	72.9	CH	3.87	m
4	62.4	CH	3.13	dd, 2.0, 4.8
5	54.5	CH	3.59	d, 2.0
6	147.0	C
7	69.6	CH	4.86	m
8	45.1	CH2	2.98	dd, 10.4, 16.7
			2.68	dd, 2.2, 16.7
9	210.5	C
10	47.9	CH	3.26	dq, 10.6, 6.7
11	127.2	CH	5.16	brd, 10.6
12	136.3	C
13	40.2	CH2	2.48	dd, 13.5, 3.6
			2.40	dd, 13.5, 9.6
14	70.8	CH	4.32	dd, 9.6, 3.6
15	98.1	C
16	65.8	CH	3.61	t, 2.5
17	26.8	CH2	2.21	m
			1.54	m
18	25.6	CH2	1.39	m
			1.01	m
19	65.8	CH	4.22	m
20	41.5	CH2	1.44a)	m
21	74.4	CH	4.16	m
22	32.6	CH2	1.67	m
			1.16	m
23	28.0	CH2	1.59	m
			1.22	m
24	81.0	CH	3.74	dt, 5.8, 8.2
25	73.1	CH	4.96	ddd, 10.3, 8.2, 2.0
26	40.0	CH2	1.42	m
			0.95	m
27	24.7	CH	1.73	m
28	23.6	CH3	0.88b)	d, 7.0
29	21.7	CH3	0.93b)	d, 7.0
30	14.1	CH3	1.22b)	d, 7.0
31	111.5	CH2	5.40	br s
			5.31	br s
32	15.7	CH3	1.13b)	d, 7.0
33	15.8	CH3	1.79b)	s

1: 3-OH: 4.91 (br d, 5.8 Hz), 7-OH: 3.95 (br d, 5.3 Hz), 14-OH: 4.02 (br s), 15-OH: 4.35 (br s), 16-OH: 1.45 (br). a) 2H. b) 3H.

The planar structure of 1 was elucidated on the basis of detailed NMR studies, including ¹H–¹H correlated spectroscopy (COSY), total correlation spectroscopy (TOCSY), heteronuclear multiple-quantum coherence (HMQC), methine- and methylene-selected editing heteronuclear single quantum coherence (CH- and CH₂-sel E-HSQC), heteronuclear multiple bond correlation (HMBC), and nuclear Overhauser effect spectroscopy (NOESY) spectra recorded in C₆D₆. Analyses of ¹H–¹H COSY and TOCSY spectra revealed five proton–proton networks from H₂-2 to H-5 and H₃-30, from H-7 to H₂-8, from H-10 to H-11 and H₃-32, from H₂-13 to H-14, and from H-16 to H₃-28 and H₃-29 (Fig. 2). The presence of the trans epoxide at C-4–C-5 was deduced from the ¹³C chemical shifts of C-4 and C-5 (δ_C 62.4 and 54.5, respectively) and the ¹H–¹H coupling constant values (H-4/H-5: 2.0 Hz). The E-geometry of the C-11–C-12 olefin was assigned from NOESY correlations for H-10/H₃-33 and H-11/H-13b as well as the relatively high-field chemical shift for C-33 (δ_C 15.8). Connections of these five networks and C-31 and C-33 via four quaternary carbons (C-6, C-9, C-12, and C-15) were suggested by HMBC correlations as follows: C-6/H-5, C-6/H₂-31, C-7/H₂-31, C-9/H₂-8, C-9/H-10, C-11/H₃-33, C-12/H₃-33, C-13/H₃-33, C-15/H-14, and C-16/H-14. HMBC correlations for C-1/H-2 and C-1/H-25 revealed that an ester linkage existed between C-1 and C-25, suggesting the presence of a 26-membered macrolactone ring. One of two residual rings was determined to be a tetrahydrofuran ring at C-21–C-24 from the HMBC correlations for C-21/H-24. The relatively low-field carbon resonance (δ_C 98.1) for C-15 suggested that the hemiketal carbon was involved in a 6-membered ether ring with C-19. These revealed that 1 possessed the same macrocyclic portion at C-1–C-25 with four C₁ branches (C-30, C-31, C-32, and C-33) as that of amphidinolide N (3) and an isobutyl terminus for C-26–C-29. Therefore, the gloss structure of 1 was concluded to be an analog of amphidinolide N (3) with an isobutyl side-chain as shown in Fig. 1.

Fig. 2. Selected 2D-NMR Correlations for Isocaribenolide-I (1)

The ¹H- and ¹³C-NMR data (Table 2) of 2 in CDCl₃ and C₆D₆ were similar to those of 1 and 3. The ¹³C-NMR spectra of 2 disclosed the presence of a total of 33 carbons (one ketone carbonyl, one ester carbonyl, three quaternary carbons, one sp² methine, one sp² methylene, 12 sp³ methines, 10 sp³ methylenes, and four methyls). ESI-MS of 2 in MeOH showed pseudo-molecular ion peaks [(M + Na)⁺] at m/z 683 and 685 (approx. 3 : 1), indicating that 1 had a chlorine atom in the molecule, and the molecular formula of 1 was established as C₃₃H₅₃ClO₁₁ with seven degrees of unsaturation based on HR-ESI-MS data [m/z 683.3179 (M + Na)⁺, Δ +1.06 mmu]. Because four double bonds in the molecule accounted for four out of seven degrees of unsaturation, 2 was inferred to possess three rings, indicating that 2 has one ring less than 1 or 3. The deuterio-substituted (M + Na)⁺ peaks at m/z 689 and 691 observed for ESI-MS data in MeOH-d₄ suggested the presence of six hydroxyl groups in the molecule, one more hydroxyl group than 3. Detailed analysis of NMR data in CDCl₃ revealed that the C-1–C-3 and C-8–C-29 portions in 2 were the same as those in amphidinolide N (3) (Fig. 3). The presence of the resonances of C-4 (δ_H 3.77, δ_C 74.0, CDCl₃) and C-5 (δ_H 4.68, δ_C 56.7, CDCl₃) of 2 indicated that 2 lacked the epoxy ring at C-4–C-5 in 3. In the deuterium secondary isotope shifts of the ¹³C-NMR data measured in C₆D₆ (Table S2), the relatively large shift for C-4 (Δδ_C +0.07) was observed compared to that for C-5 (Δδ_C +0.02), suggesting that C-4 and C-5 were adjacent to a hydroxyl group and chlorine atom, respectively. Therefore, the gross structure of 2 was concluded to be the chlorohydrin form of the epoxide at C-4–C-5 of 3, as shown in Fig. 1.

Table 2. ¹H- and ¹³C-NMR Data of Chlorohydrin 2 in CDCl₃ and C₆D₆

Position	CDCl3				C6D6
	13C		1H		13C	1H
1	174.2	C			174.2
2	45.1	CH	2.84	dq, 8.5, 7.0	45.1	3.13	m
3	73.5	CH	4.28	br d, 8.5	72.7	4.60	d, 8.5
4	74.0	CH	3.77	br d 9.7	74.4	4.00	br d, 9.8
5	57.6	CH	4.68	d, 9.7	57.6	5.10	d, 9.8
6	149.5	C			149.7
7	72.2	CH	4.71	dd, 3.2, 9.5	72.2	4.87	m
8	45.4	CH2	3.09	dd, 17.3, 9.5	45.4	3.19	dd, 17.6, 9.7
			2.77	dd, 17.3, 3.2		2.81	dd, 17.6, 3.7
9	212.3	C			211.3
10	47.3	CH	3.36	dq, 9.4, 7.0	47.5	3.20	m
11	126.7	CH	5.03	br d, 9.4	127.3	5.11	br d, 9.4
12	136.5	C			136.5
13	39.6	CH2	2.34	dd, 13.6, 3.1	40.0	2.42	dd, 13.2, 3.0
			2.25	dd, 13.6, 9.5		2.37	dd, 13.2, 9.0
14	71.3	CH	4.03	dd, 9.5, 3.1	71.4	4.25	m
15	97.8	C			98.2
16	65.6	CH	3.64	t, 2.5	65.9	3.67	br s
17	27.1	CH2	2.17	m	26.6	2.21	m
			1.71	m		1.66	m
18	25.1	CH2	1.56	m	25.2	1.45	m
			1.36	m		1.04	m
19	66.1	CH	4.14	m	66.4	4.25	m
20	41.5	CH2	1.61a)	m	41.7	1.44a)	m
21	74.8	CH	4.23	m	74.6	4.22	m
22	32.1	CH2	2.05	m	32.6	1.76	m
			1.53	m		1.22	m
23	27.6	CH2	2.01	m	27.7	1.57	m
			1.65	m		1.32	m
24	80.2	CH	4.02	m	80.4	3.75	m
25	75.3	CH	4.81	m	75.0	4.87	m
26	30.6	CH2	1.55a)	m	30.8	1.40a)	m
27	27.3	CH2	1.30a)	m	27.5	1.37a)	m
28	22.5	CH2	1.31a)	m	22.8	1.35a)	m
29	13.9	CH3	0.88b)	t, 7.0	14.1	0.89b)	t, 7.0
30	14.0	CH3	1.24b)	d, 7.0	14.0	1.22b)	d, 7.0
31	117.7	CH2	5.49	s	118.2	5.45	s
			5.47	s		5.44	s
32	15.5	CH3	1.13b)	d, 7.0	15.8	1.14b)	d, 7.0
33	16.7	CH3	1.78b)	br s	16.8	1.80b)	br s

a) 2H. b) 3H.

Fig. 3. Selected 2D-NMR Correlations for Chlorohydrin 2

Relative Stereochemistry of Chlorohydrin 2

To elucidate the relative stereochemistry of 13 chiral centers in these amphidinolide N type macrolides, we first demonstrated the conformation analysis of 2, without an allyl epoxide portion at C-4–C-6, based on NOESY data and scalar coupling constants in CDCl₃. The ¹H–¹H coupling constants were estimated by analysis of the resolution-enhanced ¹H-NMR spectrum, while the ^2,3J(C/H) values were obtained by employing the hetero half-filtered TOCSY (HETLOC) spectrum. The intensities of NOESY correlations were normalized with the integration value of the H-2/H₃-30 cross-peak as 100 (Table S3), and the correlations were categorized into dominant nuclear Overhauser effects (NOEs) or not. Bond rotation analyses assisted by J-based configuration analysis²⁹⁾ indicated that the relative configurations for C-2–C-3, C-3–C-4, C-4–C-5, and C-24–C-25 bonds were erythro, threo, erythro, and threo relations, respectively (Figs. 4a–4d). A couple of methine–methylene portions in the C-7–C-8 and C-13–C-14 bonds were implied to possess bond rotations with anti-relations for H-7–H-8a, 7-OH–H-8b, H-13a–14-OH, and H-13b–H-14, as shown in Figs. 4e and 4f.

Fig. 4. Rotation Analyses for (a) C-2–C-3, (b) C-3–C-4, (c) C-4–C-5, (d) C-24–C-25, (e) C-7–C-8, and (f) C-13–C-14 Bonds in Chlorohydrin 2

“a” and “b” for germinal proton pairs denoted low- and high-field resonances, respectively. All of the J-values and the most of the NOESY correlations were measured in CDCl₃. The intensities of NOESY correlations were normalized with the integration value of H-2/H₃-30 cross-peak as +100. Black arrows show correlations with intensities of 40 or more, while gray ones show correlations with intensities less than 40.

Figure 5a illustrates the conformation of the C-1–C-13 portion, and the sequential four chiral centers from C-2 to C-5 of 2 were revealed to have 2S*, 3R*, 4S*, and 5R* configurations from the bond rotation analyses described above. Considering the bond rotation for C-7–C-8 and NOESY correlations for H-5/H₂-8, the relative configuration at C-7 was deduced to be R* through the exomethylene unit at C-6. The NOESY correlation was observed for H-8b/H-10 through the ketone carbonyl at C-9, implying a β-orientation and 10S* configuration for the 32-methyl group. The anitperiplanar relationship for H-10–C-33 and H-11–C-13b was deduced from NOESY correlations for H-10/H₃-33 and H-11/H-13b, suggesting axial orientations for H-13b and the 33-methyl group.

Fig. 5. Stereostructures for (a) C-1–C-13 and (b) C-12–C-26 Portions in Chlorohydrin 2

All of the J-values and NOESY correlations were recorded in CDCl₃. The intensities of NOESY correlations were normalized with the integration value of H-2/H₃-30 cross-peak as +100. Black arrows show correlations with intensities of 40 or more, while gray ones show correlations with intensities less than 40. Dashed arrows show ROESY correlations. ¹H–¹H coupling constants (Hz) (H/H): 8.5 (2/3), <1 (3/4), 9.7 (4/5), 9.5 (7/8a), 3.2 (7/8b), 9.4 (10/11), 3.1 (13a/14), 9.5 (13b/14), 2.5 (16/17a), 2.5 (16/17b), >8 (17a/18a), <3 (17b/18a), >8 (18a/19), <3 (18b/19), and 8.4 (24/25).

For the C-12–C-26 portion of 2 (Fig. 5b), NOESY correlations for H-13a/H-16 and H-14/H₃-33 indicated that the hydroxyl group at C-14 was oriented toward the inside of the macrocyclic ring. The rather small ³J(H-16/H-17a) and ³J(H-16/H-17b) and relatively large ³J(H-17a/H-18a) and ³J(H-18a/H-19) values and the NOESY correlation for H-17a/H-19 disclosed the chair conformation of the tetrahydropyran ring at C-15–C-19 with axial orientations for 16-OH and H-19. An α-axial orientation for 15-OH on the tetrahydropyran ring at C-15–C-19 was presumed, because it also seems to be favorable from the point of view of the anomeric effect. Molecular calculation studies of simple C-13–C-19 models with α- and β-orientations for 15-OH revealed that the minimized energy for the model with an α-hydroxyl group was lower than for that with a β-one.³⁰⁾ Thus, the stereochemistries of C-14, C-15, C-16, and C-19 was elucidated to be S*, S*, R*, and R*, respectively.

The relative stereochemistry between C-19 and C-21 was considered from an equivalent signal pattern of H₂-20 [δ_H 1.61 (2H) in CDCl₃ and 1.44 (2H) in C₆D₆], which might be due to the chemical equivalence of these protons; H-20β was gauche-oriented against H-19 and 21-O and anti-oriented against H-21 and 19-O, while H-20α was anti-oriented against H-19 and 21-O and gauche-oriented against H-21 and 19-O. The rotating frame nuclear Overhauser effect spectroscopy (ROESY) correlations for H-18b/H₍₂₎-20 and H₍₂₎-20/H-22a observed in CDCl₃ also supported this stereostructure. The NOESY correlation for H-21/H-25 implied an anti-relationship for H-21–H-24 of the tetrahydropyran ring at C-21–C-24. Considering the bond rotation for C-24–C-25, the stereochemistries of C-21, C-24, and C-25 were all concluded to be R*. Therefore, the total relative configuration of 2 was concluded to be 2S*, 3R*, 4S*, 5R*, 7R*, 10S*, 14S*, 15S*, 16R*, 19R*, 21R*, 24R*, and 25R*, as shown in Fig. 1.

Relative Stereochemistry of Isocaribenolide-I (1) and Amphidinolide N (3)

Elucidating the stereostructure of isocaribenolide-I (1) and amphidinolide N (3) with an ally epoxide may have a higher degree of difficulty than elucidating that of 2, owing to the conformation change described below. We performed the conformation analysis of 1 by applying the similar data analysis method described above. Magnitudes or values for J(H/H)s needed for stereochemical assignments were obtained from resolution-enhanced ¹H-NMR spectra and selective population transfer (SPT)³¹⁾ experiments. The intensities of NOESY correlations were normalized with the integration value of H-2/H₃-30 cross-peak as +100, and then correlations with intensities of 20 or more and less than 20 were judged as strong or not, respectively. The ¹H–¹H coupling constant values and the profiles of proton pairs having NOESY correlations for the C-7–C-25 portion of 1 in C₆D₆ were close to those of 2 in CDCl₃, indicating that the conformation of the C-7–C-25 portion and relative stereochemistries of nine chiral centers contained in this portion for 1 were similar to those for 2 (Fig. 6). Moreover, NOESY correlations for H-8a/H-10 and H-8b/H-11 clearly showed the spatial relationship of C-8/C-10 through the C-9 carbonyl group. The relative stereochemistry between C-19 and C-21 was deduced from equivalent signal pattern of H₂-20 [δ_H 1.44 (2H)] to be the same as that of 2, as described above.

Fig. 6. Stereostructures for C-6–C-26 Portion in Isocaribenolide-I (1)

All of the J-values and NOESY correlations were recorded in C₆D₆. The intensities of NOESY correlations were normalized with the integration value of H-2/H₃-30 cross-peak as +100. Black arrows show correlations with intensities of 20 or more, while gray ones show those with intensities of less than 20. ¹H–¹H coupling constants (Hz) (H/H): 10.4 (7/8a), 2.2 (7/8b), 10.6 (10/11), 3.6 (13a/14), 9.6 (13b/14), 2.5 (16/17a), 2.5 (16/17b), >8 (17a/18a), <3 (17b/18a), >8 (18a/19), <3 (18b/19), and 8.2 (24/25).

Furthermore, elucidation of the NOESY data and ¹H–¹H coupling constants revealed that the conformation of the C-1–C-9 portion containing the allyl epoxide for isocaribenolide-I (1) was deduced to be different from that for 2. Both intermediate values for ³J(H-2/H-3) (7.0 Hz) and ³J(H-3/H-4) (4.8 Hz) suggest that the C-2–C-3 and C-3–C-4 bonds are rotating. The former corresponds to an average value of the coupling constant when anti (approx. 9 Hz) and gauche (approx. 3 Hz) relations for H-2–H-3 are 2 : 1, while the latter value is suitable for the average at a 1 : 2 ratio of anti and gauche-relations for H-3–H-4. The normalized integration values of dominant NOESY correlations (Table S4) showed incredible pairs of NOESY correlations, such as H-4/H-2 (integration value: 18.40) and H-5/H-2 (13.38), H-4/H-7 (21.64) and H-5/H-7 (18.07), and H-4/H₃-30 (14.12) and H-5/H₃-30 (10.43), around the trans epoxide conjugated with an exomethylene. Considering the bond rotations associated with NOESY correlations, three plausible conformations could be proposed for the C-1–C-9 portion of 1 (Fig. 7). In conformation A with H-2–H-3 anti- and H-3–H-4 gauche-relations, H-4 and H₃-30, via five bonds, came close enough to give rise to the corresponding NOESY correlation, while H-5 and H₃-30, via six bonds, drew together in conformation B with double anti-interactions for H-2–H-3 and H-3–H-4. The NOESY correlation for H-2/H-4 may occur in conformation B, and the NOESY correlation for H-2/H-5 may be seen for conformation C with double gauche interactions between H-2/H-3 and H-3/H-4. In conformations A/C and B, H-7 closes with H-4 and H-5, respectively. This coexistence of NOESY correlations for H-4/H-7 and H-5/H-7 suggests the presence of a major conformation change, plausibly derived from a flip of the trans epoxide. Therefore, the relative stereochemistry of 1 is suggested as 2S*, 3R*, 4S*, 5S*, 7R*, 10S*, 14S*, 15S*, 16R*, 19R*, 21R*, 24R*, and 25R*, which is not inconsistent with the relative configurations predicted by the structural correlation between isocaribenolide-I (1) and 2.

Fig. 7. Three Plausible Conformations for C-1–C-9 Portion in Isocaribenolide-I (1); Conformations A: Anti- and Gauche-, B: Both Anti-, and C: Both Gauche-Relations for H-2–H-3 and H-3–H-4

All of the J-values and NOESY correlations were recorded in C₆D₆. The means of black and gray arrows are the same as described in Fig. 7. ¹H–¹H coupling constants (Hz) (H/H): 7.0 (2/3), 4.8 (3/4), and 2.0 (4/5).

Because amphidinolide N (3) was isolated at the same time as chlorohydrin 2, we assumed that 3 may be structurally and biosynthetically related to 2. The structural relationship between 3 and the corresponding chlorohydrin 2 indicates that 2 may be generated from 3 through the nucleophilic addition of a chloride anion from behind the C-5 epoxy carbon and cleavage of the O–C-5 bond. Considering the stereochemical inversion at C-5, the relative stereochemistry of 3 is presumed to be the same as that of 1. To clarify the stereochemistry of 3, the ¹³C chemical shifts of isocaribenolide-I (1) in C₆D₆ were compared with those of 3, because the difference in planar structures between 3 and 1 is only a side-chain. Figure 8a shows the values of ¹³C chemical shift difference between 1 and 3 [Δδ (in ppm) = δ (1 in C₆D₆) − δ (3 in C₆D₆)] for the macrocyclic portion of C-1–C-25 with four C₁ branches (C-30, C-31, C-32, and C-33). Relatively large Δδ values were observed for C-2 (–0.5 ppm), C-24 (+0.4 ppm), and C-25 (–1.8 ppm), probably due to the neighboring side-chain structural differences. The absolute Δδ values for the residual carbons were extremely small (≤0.3 ppm), strongly indicating that the relative stereochemistry of 11 chiral carbons for C-2, C-3, C-4, C-5, C-7, C-10, C-14, C-15, C-16, C-19, and C-21 in 3 were common to those in 1 at least. In contrast, Fig. 8b describes the ¹³C chemical shift differences between 2 and 3 [Δδ (in ppm) = δ (2 in C₆D₆) – δ (3 in C₆D₆)] for the C-1–C-29 portion with four C₁ branches (C-30, C-31, C-32, and C-33). The absolute Δδ values for C-21–C-29 were so small (≤0.2 ppm), suggesting that the relative stereochemistries of C-21, C-24, and C-25 were the same as those of 2. Interestingly, the relatively large Δδ values were observed for C-6 (+2.7 ppm), C-7 (+2.3 ppm), C-9 (+1.0 ppm), C-31 (+6.4 ppm), and C-33 (+1.2 ppm), which might be derived from a deshielding effect, probably due to the neighboring electron-rich chlorine atom at C-5. Therefore, the relative stereochemistry of 3 was concluded to be 2S*, 3R*, 4S*, 5S*, 7R*, 10S*, 14S*, 15S*, 16R*, 19R*, 21R*, 24R*, and 25R*. Nevertheless, the absolute stereochemistry of 1–3 remains undetermined because the α-methoxy-α-(trifluoromethyl)-phenylacetyl (MTPA) esters and degradation product were not prepared.

Fig. 8. Differences between ¹³C Chemical Shifts for (a) Macrocyclic Portions of Isocaribenolide-I (1) and Amphidinolide N (3) [Δδ (in ppm) = δ_C of 1 − δ_C of 3] and (b) Chlorohydrin (2) and Amphidinolide N (3) [Δδ (in ppm) = δ_C of 2 − δ_C of 3] in C₆D₆

The Δδ values for C-4, C-5, and C-31 in Fig. 8b were +11.9, +3.2, and +6.4, respectively.

In this study, we isolated three 26-membered macrolides from marine dinoflagellate Amphidinium species and elucidated the structures of two new compounds, isocaribenolide-I (1) and compound 2 together with known amphidinolide N (3).^2,6) Isocaribenolide-I (1) is an analog of 3 with an isobutyl side-chain. Compound 2 is a chlorohydrin form of 3, and it may be an artifact generated in its extraction and/or separation. Moreover, we concluded that amphidinolide N (3) and calibenolide-I³⁾ are the same substance, and proposed the relative stereochemistry of 3 on the basis of the NMR data for the chlorohydrin form (2) of 3. Compounds 1, 2, and 3 exhibited extremely potent cytotoxic activity against human cervix adenocarcinoma HeLa cells (IC₅₀: 0.02, 0.06, and 0.01 nM, respectively). The cytotoxicity of compound 2 was less potent than that of 1 and 3, suggesting that the presence of the epoxide ring in 2 may be essential for the powerful cytotoxicity of this macrolide.

Experimental

General Experimental Procedures

Optical rotation and IR data were measured on a JASCO DIP-370 polarimeter and a JASCO FT/IR-5300 spectrophotometer, respectively. NMR data were recorded using 2.5 mm microcells (Shigemi Co., Ltd., Japan). NMR spectra were measured on a Bruker AMX-500 spectrometer or Varian-NMR500 spectrometer equipped with a triple resonance PFG cold probe or CH “Xsens” PFG cold probe. Chemical shifts were reported in ppm with reference to the residual proton and carbon signals of C₆D₆ (δ_H 7.20 and δ_C 128.0, respectively) and CDCl₃ (δ_H 7.26 and δ_C 77.0, respectively). NMR data reconstruction and analysis were performed with Mnova NMR 7.0.3 (Mestrelab Research, Spain). The ESI-MS spectra of 1 and 3 were measured on a ThermoFisher Exactive spectrometer, and that of 2 was recorded on a JEOL JMS-T100LC spectrometer.

Extraction and Isolation

One dinoflagellate Amphidinium species (strain number KCA09053) was separated monoclonally from benthic sea sands collected off Iriomote Island, Japan.²⁴⁾ Another dinoflagellate Amphidinium species (strain KCA09056) was monoclonally separated from the surface of a seagrass collected off Iriomote Island, Japan.²⁸⁾ Information on the biomaterials and cultivation are described in previous papers. Dried algal cells (at a total weight of 10.3 g) of the KCA09053 strain obtained from 150 L of the medium were extracted with MeOH/toluene (3 : 1) and partitioned between toluene and water. The toluene-soluble materials (1.15 g) of the extract were subjected to SiO₂ gel CC using a stepwise elution with CHCl₃ (200 mL) and CHCl₃/MeOH (98 : 2). The fraction eluted with (CHCl₃/MeOH, 98 : 2) was chromatographed successively by using an ODS [Cosmosil 140C₁₈-PREP, Nacalai Tesque Inc., Japan; eluent: acetonitrile (CH₃CN)/H₂O, 7 : 3] and amino silica gel [Wakogel^® 50NH₂, FUJIFILM Wako Pure Chemical Corporation, Japan; eluent: hexane–ethyl acetate (EtOAc), 1 : 1] columns to give a cytotoxic fraction with >90% inhibition of cell proliferation against HeLa cells at 1 µg/mL. The cytotoxic fraction was separated by ODS HPLC [YMC-Pack Pro C₁₈, 5 µm, YMC Co., Ltd., Japan; 10 ×250 mm; eluent, CH₃CN/H₂O (70 : 30); flow rate, 2 mL/min; UV detection at 210 nm] to afford isocaribenolide-I [1, 0.006%, retention time (t_R) 8.6 min].

The dried algal cells (40.95 g) of the KCA09056 strain obtained from 250 L of the culture were extracted with MeOH/toluene (3 : 1). The toluene-soluble materials (15.28 g) of the extract were subjected to silica gel CC using a gradient elution of 0–2% MeOH in CHCl₃. The fraction eluted with CHCl₃/MeOH (98 : 2) was chromatographed using an ODS column (Cosmosil 140C₁₈-PREP; CH₃CN/H₂O, 7 : 3) to give a cytotoxic fraction. The fraction was separated by reversed-phase HPLC [YMC-Pack Pro C₁₈, 5 µm; 20 mm ×250 mm; eluent, CH₃CN/H₂O (76 : 24); flow rate, 8 mL/min; UV detection at 210 nm] to afford compounds 2 (0.0054%, t_R 13 min) and 3 (0.0024%, t_R 9.7 min).

Isocaribenolide-I (1)

Colorless amorphous solid; [α]_D²³ +33 ± 4 (c = 0.27, CHCl₃); IR (KBr) cm⁻¹: 3421 (broad), 2924, 1705; ¹H- and ¹³C-NMR (500 MHz, C₆D₆): Table 1; HR-ESI-MS (MeOH) m/z 647.3399 [M + Na]⁺, (Calcd for C₃₃H₅₂O₁₁Na, 647.3402).

Compound 2

Colorless amorphous solid; [α]_D²³ +36 ± 2 (c = 1.1, MeOH); [α]_D²² +26 ± 3 (c = 1.1, CH₂Cl₂); IR (KBr) cm⁻¹: 3443 (broad), 2920, 1714; ¹H- and ¹³C-NMR (500 MHz, C₆D₆ and CDCl₃): Table 2; ESI-MS (MeOH) m/z 683 and 685 (3 : 1) [M + Na]⁺; ESI-MS (MeOH-d₄) m/z 689 and 691 (3 : 1) [M + Na]⁺; HR-ESI-MS (MeOH) m/z: 683.3179 [M + Na]⁺ (Calcd for C₃₃H₅₃³⁵ClO₁₁Na, 683.3169).

Compound 3

Colorless amorphous solid; [α]_D²⁴ +41 ± 5 (c = 0.49, CHCl₃); [α]_D¹⁹ +30 ± 4 (c = 0.49, MeOH); [α]_D¹⁹ +27 ± 4 (c = 0.49, CH₂Cl₂); IR (KBr) cm⁻¹: 3443 (broad), 2920, 1714; ¹H- and ¹³C-NMR (500 MHz, C₆D₆ and CDCl₃): Table S1; HR-ESI-MS (MeOH) m/z: 647.3400 [M + Na]⁺ (Calcd for C₃₃H₅₂O₁₁Na, 647.3402).

Cytotoxic Assay

A cytotoxicity assay involving HeLa cells was performed at 37 °C in 5% carbon dioxide, at a density of 5000 cells per well in 96-well plates, using Dulbecco’s modified eagle medium containing 10% fetal calf serum. After 72 h, the number of viable cells was counted using Cell Counting Kit 8 (Dojindo Co., Kumamoto, Japan) according to the manufacturer’s instructions. The assay reagent was a tetrazolium compound (WST-8) that is reduced by live cells into a colored formazan product; the absorbance was measured at 450 nm using a microplate reader (Bio-Rad, U.S.A.). The experiments were repeated in triplicate wells. The viability of the treated groups was estimated as a percentage of that of the control groups. The cytotoxicity is expressed as IC₅₀. Doxorubicin was used as an authentic sample, and the IC₅₀ value against HeLa cells was 520 nM.

Acknowledgments

We thank Ai Tokumitsu and Seiko Oka of the Equipment Management Center, Hokkaido University, for the measurement of MS spectra, and Satoru Ibuki, Keita Ikebe, and Takahiro Tsushima of the Kochi Prefectural Deep Seawater Laboratory for the supply of deep seawater and assistance with dinoflagellate cultivation. This study was partially supported by a Grant-in-Aid for Scientific Research (No 26670045 to M.T.) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.

Conflict of Interest

The authors declare no conflicts of interest.

Supplementary Materials

The online version of this article contains supplementary materials. The spectral data of compounds 1–3 and Tables S1–S4 are available as supplementary materials.

References and Notes

• 1) Carroll A. R., Copp B. R., Davis R. A., Keyzers R. A., Prinsep M. R., Nat. Prod. Rep., 37, 175–223 (2020).
• 2) Kobayashi J., Tsuda M., Nat. Prod. Rep., 21, 77–93 (2004).
• 3) Kobayashi J., Kubota T., J. Nat. Prod., 70, 451–460 (2007).
• 4) Ishibashi M., Yamaguchi N., Sasaki T., Kobayashi J., J. Chem. Soc. Chem. Commun., 1455–1456 (1994).
• 5) Bauer I., Maranda L., Young K. A., Shimizu Y., Fairchild C., Cornell L., MacBeth J., Huang S., J. Org. Chem., 60, 1084–1086 (1995).
• 6) Maranda L., Shimizu Y., J. Phycol., 32, 873–879 (1996).
• 7) Murray S., Jørgensen M. F., Daugbjerg N., Rhodes L., J. Phycol., 40, 366–382 (2004).
• 8) Takahashi Y., Kubota T., Imachi M., Wälchli M. R., Kobayashi J., J. Antibiot., 66, 277–279 (2013).
• 9) Nicolaou K. C., Brenzovich W. E., Bulger P. G., Francis T. M., Org. Biomol. Chem., 4, 2119–2157 (2006).
• 10) Nicolaou K. C., Bulger P. G., Brenzovich W. E., Org. Biomol. Chem., 4, 2158–2183 (2006).
• 11) Ochiai K., Kuppusamy S., Yasui Y., Okano T., Matsumoto Y., Gupta N. R., Takahashi Y., Kubota T., Kobayashi J., Hayashi Y., Chem. Eur. J., 22, 3282–3286 (2016).
• 12) Ochiai K., Kuppusamy S., Yasui Y., Harada K., Gupta N. R., Takahashi Y., Kubota T., Kobayashi J., Hayashi Y., Chem. Eur. J., 22, 3287–3291 (2016).
• 13) Seck M., Franck X., Seon-Meniel B., Hocqueniller R., Figadère B., Tetrahedron Lett., 47, 4175–4180 (2006).
• 14) Jalce G., Franck X., Seon-Meniel B., Hocqueniller R., Figadère B., Tetrahedron Lett., 47, 5905–5908 (2006).
• 15) Trost B. M., Rey J., Org. Lett., 14, 5632–5635 (2012).
• 16) Kawashima Y., Toyoshima A., Fuwa H., Sasaki M., Org. Lett., 18, 2232–2235 (2016).
• 17) Toyoshima A., Sasaki M., Tetrahedron Lett., 57, 3532–3534 (2016).
• 18) Trost B. M., Bai W.-J., Stivala C. E., Hohn C., Poock C., Heinrich M., Xu S., Rey J., J. Am. Chem. Soc., 140, 17316–17326 (2018).
• 19) Kumagai K., Tsuda M., Fukushi E., Kawabata J., Masuda A., Tsuda M., J. Nat. Med., 71, 506–512 (2017).
• 20) Tsuda M., Makihara R., Tsuda M., Suzuki T., Chem. Pharm. Bull., 68, 864–867 (2020).
• 21) Tsuda M., Kumagai K., Akakabe M., Tominaga A., Konishi Y., Tsuda M., Japan Patent, 1, 2011–184437 (2011).
• 22) Compound 2 is identical to the compound previously named neocaribenolide-I: Kumagai K., Akakabe M., Minamida M., Niahisaka T., Tsuda M., Konishi Y., Tsuda M., Tominaga A., Symposium papers for 53rd Symposium on the Chemistry of Natural Products, Osaka, 53, 359–354 (2011).
• 23) Compounds 2 and 3 were assigned as stereoisomers of C-7, C-10, C-14, C-15, C-16, C-19, C-21, C-24, and C-25 in Doctor’s thesis of M. Minamida Kochi University (2015).
• 24) Akakabe M., Kumagai K., Tsuda M., Konishi Y., Tominaga A., Tsuda M., Fukushi E., Kawabata J., Tetrahedron, 70, 2962–2965 (2014).
• 25) Akakabe M., Kumagai K., Tsuda M., Konishi Y., Tominaga A., Kaneno D., Fukushi E., Kawabata J., Masuda A., Tsuda M., Chem. Pharm. Bull., 64, 1019–1023 (2016).
• 26) Akakabe M., Kumagai K., Tsuda M., Konishi Y., Tominaga A., Tsuda M., Fukushi E., Kawabata J., Tetrahedron Lett., 55, 3491–3494 (2014).
• 27) Kanto M., Sato S., Tsuda M., Sasaki M., J. Org. Chem., 81, 9105–9121 (2016).
• 28) Tsuda M., Makihara R., Minamida M., Tsuda M., Akakabe M., Kumagai K., Fukushi E., Kawabata J., Suzuki T., Hetereocycles, 1678–1685 (2020).
• 29) Matsumori N., Kaneno D., Murata M., Nakamura H., Tachibana K., J. Org. Chem., 64, 866–876 (1999).
• 30) Minimum energies of the C-13–C-19 models with α- and β-orientations for 15-OH were calculated as 59.02 and 70.05 kJ/mol, respectively, by applying the MM2 minimization in the program Chem 3D (Chembridge Soft).
• 31) Hoffman R. A., Forsén S., Prog. Nucl. Magn. Reson. Spectrosc., 1, 15–204 (1966).