Corallolides A and B: Bioactive Diterpenes Featuring a Novel Carbon Skeleton

Abstract

Corallolides A (1) and B (2) are naturally occurring diterpenes isolated from the Caribbean gorgonian octocoral Pseudopterogorgia bipinnata collected near Providencia Island, Colombia. Their tricyclic structures are based on a uniquely substituted bicyclo[9.2.1]tetradecane ring system that was established through detailed spectroscopic analysis. Compounds 1 and 2 were shown to exhibit anti-parasitic and anti-tuberculosis activity, respectively.

For nearly four decades, investigations into the natural products chemistry of Caribbean gorgonian octocorals of the genus Pseudopterogorgia have yielded a plethora of diterpenoids of diverse molecular architectures that endow them with remarkable biological activities.¹ The gorgonian species P. bipinnata, P. kallos, P. acerosa, P. americana, and P. elisabethae are particularly noteworthy as they alone appear to account for the production of the vast majority of these fascinating natural products.² Recently, the groups of Mulzer, Trauner, and Kerr have highlighted the ever increasing number of diterpenoid skeletal classes produced by these chemically prolific Pseudopterogorgia species. Furthermore, many of the metabolites engendered by these animals are highly likely to experience subsequent rearrangement to yield even more perplexing structural variants.³ The chemical investigation of these five octocoral species has led to important pharmacological advances which in turn have spurred novel synthetic methodology and inspired many chemists and biochemists to speculate on the plausible biogenetic relationships among the diverse skeletal families of diterpenes that are routinely co-isolated from them.

As part of our continuing interest in the natural products chemistry of Caribbean gorgonian octocorals and their potential utilization in the development of novel anti-infective and anti-cancer drugs, we have previously reported on several pseudopterane lactones from a Southwestern Caribbean specimen of P. bipinnata.⁴ From the same specimen, we now report the isolation and structure elucidation of two isomeric diterpenes which feature a uniquely substituted bicyclo[9.2.1]tetradecane ring system. These compounds, which have been trivially named corallolides A (1) and B (2), feature a twelve-membered macrocycle that appears to be linked biosynthetically to the pseudopterane class of diterpenes through additional C–C bond formation leading to an intriguing bridged ring system. Their molecular structures were established by analysis of 1D and 2D NMR, IR, UV, and high-resolution mass spectral data. Biological screening of metabolites 1 and 2 revealed that they are bestowed with anti-tubercular and anti-parasitic activity.

A detailed scheme for the extraction of the dry gorgonian specimen (0.11 kg) has been provided as Supporting Information. The crude CHCl₃ extract (4.1 g) was fractionated over silica gel (150 g) with a 99:1 mixture of CHCl₃/MeOH to give 22 fractions, denoted I–XXII. Fractions IX and X were combined (126 mg) and chromatographed over silica gel (7 g) using 1% acetone in CHCl₃ leading to 7 secondary fractions (A–G). Further purification of sub-fraction D (37 mg) by normal-phase HPLC afforded pure corallolide A (1) (4.5 mg, 4.1 × 10⁻³% yield) along with corallolide B (2) (5.0 mg, 4.5 × 10⁻³% yield) (dry gorgonian weight basis).

The molecular formula of corallolide A (1)⁵ was assigned as C₂₀H₂₄O₆ on the basis of high-resolution mass measurement of the [M + Na]⁺ ion at m/z 383.1483 and overall NMR information indicating the presence of nine degrees of unsaturation in the molecule. The IR data for 1 indicated the presence of hydroxyl (3467 cm⁻¹), carbonyl (1756 and 1706 cm⁻¹) and olefin (1628 cm⁻¹) bands. The carbonyl bands at 1756 and 1706 cm⁻¹ showed comparable intensity suggesting the presence in 1 of a conjugated γ-lactone and a 2-cyclopentenone ring, respectively. The latter contention was supported by strong UV absorptions at λ_max 214 nm (ε 14 200) and 235 nm (ε 12 300). Although 1 was soluble in CDCl₃, the ¹H and ¹³C NMR signals in this solvent were broad and poorly dispersed. Consequently, 1D and 2D NMR data were gathered in CD₃OD. The ¹³C NMR spectrum (75 MHz, CD₃OD) of 1 showed 20 resolved resonances (Table 1). Six olefinic [δ 174.0 (C), 153.8 (CH), 146.1 (C), 137.0 (C), 132.6 (C), 114.7 (CH₂)] and three carbonyl [δ 212.9 (C), 208.0 (C), 174.8 (C)] resonances in the ¹³C NMR spectrum accounted for six sites of unsaturation. Therefore, the remaining sites of unsaturation required by the molecular formula had to be explained by the presence of three rings. HMQC and DEPT-135 data showed that 22 of the 24 hydrogen atoms were attached to carbons (3 × CH₃, 4 × CH₂, 5 × CH); therefore, compound 1 had to have 2 OH groups.

Table 1.

¹H NMR (300 MHz), ¹³C NMR (75 MHz), ¹H—¹H COSY, NOESY, and HMBC spectral data for corallolide A (1)^a

position	δH, mult, intgt (J, Hz)	δC (mult)b	1H–1H COSY	NOESY	HMBCc
1	2.04, br m, 1H	47.4 (CH)	H2, H12αβ	H4, H14αβ	H2, H14αβ, H3-15
2	3.95, d, 1H (4.8)	78.4 (CH)	H1	H9, H12α
3		212.9 (C)			H2, H4, H3-16
4	2.68, q, 1H (6.8)	48.6 (CH)	H3-16	H1, H14αβ, H3-16, H18β	H3-16, H18β
5		83.6 (C)			H4, H3-16, H18β
6		208.0 (C)			H18β
7		137.0 (C)			H18β, H3-19
8	5.93, br s, 1H	77.1 (CH)	H9, H18α, H3-19	H9, H3-19	H9
9	8.20, br s, 1H	153.8 (CH)	H8	H2, H8, H11α, H12α
10		132.6 (C)			H9
11αβ	2.53, br m, 2H	22.5 (CH2)	H12αβ	H9
12α	1.71, br m, 1H	27.4 (CH2)	H1, H11αβ, H12β	H2, H9	H1
12β	2.00, br m, 1H		H1, H11αβ, H12α
13		146.1 (C)			H2, H3-15
14α	4.73, br s, 1H	114.7 (CH2)	H14β, H3-15	H4, H14β, H3-15	H3-15
14β	4.75, br s, 1H		H14α, H3-15	H1, H4, H14α
15	1.60, br s, 3H	21.1 (CH3)	H14αβ	H14αβ	H14αβ
16	1.16, d, 3H (6.8)	13.2 (CH3)	H4	H4, H18β	H4
17		174.0 (C)			H18αβ, H3-19
18α	2.60, br d, 1H (17.4)	46.7 (CH2)	H8, H18β, H3-19	H18β	H3-19
18β	2.81, br d, 1H (17.4)		H18α	H4, H3-16, H18α
19	2.29, br s, 3H	19.1 (CH3)	H8, H18α	H8
20		174.8 (C)			H9, H11αβ

^aSpectra were recorded in CD₃OD at 25 °C. Chemical shift values are in parts per million relative to the residual CH₃OH (3.30 ppm) or CD₃OD (49.0 ppm) signals. Assignments were aided by 2D NMR experiments, spin-splitting patterns, number of attached protons, and chemical shift values.

^b13C NMR multiplicities were obtained from a DEPT-135 experiment.

^cProtons correlated to carbon resonances in the ¹³C column. Parameters were optimized for ^2,3J_CH = 6 and 8 Hz.

Four partial structures (A–D) were deduced from extensive analyses of the 2D NMR data of 1 including COSY, NOESY, HMQC, and HMBC spectra in CD₃OD (Figure 1). The COSY, NOESY, and HMBC correlations for corallolide A (1) are presented in Table 1.

Figure 1.

Partial structures for corallolide A (1) generated from ¹H–¹H COSY, HMQC, and HMBC spectral data.

Connectivities from C11 to C2 in substructure A, including additional connectivities to an isopropylene group attached to C1, were inferred from the ¹H–¹H COSY cross-peaks (Table 1). HMBC correlations of C1 (δ_C 47.4) with H2 (δ_H 3.95), H₂-14 (δ_H 4.73 and 4.75), and H₃-15 (δ_H 1.60), and long-range couplings of C13 (δ_C 146.1) with H2 and H₃-15 allowed assignment of the structure for unit A. Partial structure B, an isolated –CH–CH₃ spin system (C4 and C16), was clearly revealed by the ¹H NMR, COSY, and HMQC spectra. The chemical shifts of H4 (δ_H 2.68) and C4 (δ_C 48.6) (as well as the multiplicities and coupling constants of all the protons) allowed us to further link units A and B through C3 (δ_C 212.9). This was supported by complementary HMBC correlations between C3 and protons H2, H4, and H₃-16.

The presence of a butenolide moiety in the molecule (unit C) was deduced from the COSY and HMBC spectra, the proton and carbon chemical shifts at positions C8–C10 and C20, together with IR and UV data (vide supra). Long-range couplings between H9 (δ_H 8.20) and C8, C10, and C20 were consistent with partial structure C. Furthermore, substructures A and C were confidently linked by the ³J_CH correlations between H₂-11 (δ_H 2.53) and C20 (δ_C 174.8). Applying these combined NMR methods resulted in the unambiguous assignment of all the protons and carbons in units A–C.

For partial structure D (C5–C7 and C17–C19), connectivities from C17 through C19 were clearly revealed by the ¹H–¹H COSY and HMQC spectra. It was, however, difficult to obtain unambiguous evidence for connecting H₂-18 (an AB system at δ_H 2.60 and 2.81) to other spin systems. Notwithstanding, the connectivity of unit D was completed by careful analysis of the proton and carbon chemical shifts at positions C5, C6, C7, C17, C18 and C19, and by HMBC correlations for H18β/C5/C6/C7, H18αβ/C17, and H₃-19/C7/C17/C18. Thus, unit D was formulated as a 2,3,5,5-tetrasubstituted 2-cyclopentenone moiety, a contention that was in full agreement with the IR and UV spectra recorded for 1. The pivotal ¹³C NMR resonances at δ_C 208.0 (C), 174.0 (C), and 137.0 (C) were thus ascribed, respectively, to C6, C17, and C7 of the 2-cyclopentenone ring system. Confirmation of the substitution pattern of the latter system as well as its connectivity to the other ring systems found in 1 came from additional HMBC and TOCSY correlation data. Critically, HMBC cross-peaks between the hydroxyl-bearing quaternary carbon C5 (δ_C 83.6) and protons H4, H₃-16, and H18β allowed the attachment of units B and D through C4 and C5, respectively. At this point, the link between units D and C through C7 and C8, allowed the complete planar structure for 1 to be assigned. Weak but diagnostic homoallylic couplings (5J) between the protons of C18 and C19 with lactone methine H8 confirmed the latter connectivity.⁶ Interestingly, such linkage, which renders H8 bisallylic (δ_H 5.93), imposes upon the geometry of 1 in such a manner that H9, which lies close to the C=O bond of unit D, cuts into the deshielding cone of the induced magnetic field causing H9 to be unusually deshielded (δ_H 8.20). These analyses confidently established the gross structure shown for 1.

The relative configurations (1S*,2S*,4S*,5R*,8R*) for the five stereocenters about the tricyclic framework of corallolide A were assigned primarily on the basis of ¹H NMR coupling constants and NOESY data supported by distance calculations (McSpartan’04 Program). As ³J_H1-H2 was only 4.8 Hz, a refutable trans-relationship between these protons was proposed.⁴ This assignment, however, was firmly established by the strong NOESY correlations of H1 (β-orientation in planar conformation) and H4, plus the conspicuous absence of a NOE between H2 and H4 (calcd inter-proton distance = 3.2 Å), thus placing the isopropenyl side chain at C1 and H₃-16 in the same α-plane. Synchronous NOESY cross-peaks between H2 and pivotal protons H9 and H12α further supported these assignments. After calculating the conformer distribution of each possible stereomeric structure of 1 at C5 and C8, the coincident NOESY correlations of H2/H9, H4/H18β, H₃-16/H18β, and H9/H12α were consistent only with structure 1 (Figure 2cf Supporting Information). Only in this stereoisomer, which also possessed the lowest calculated energy (312 kJmol⁻¹), the intramolecular distances for the latter protons were calculated to be within 2.6–2.0 Å.

The structural characterization of corallolide B (2) was carried out in an analogous manner.⁷ The HRESIMS and ¹³C NMR data for 2 were consistent with a molecular formula of C₂₀H₂₄O₆. A side-by-side comparison between the NMR spectra of isomers 1 and 2 rapidly pinpointed their structural similarities. Nevertheless, some major differences between these compounds were observed in the ¹³C NMR spectra: the signals ascribed to C1–C4 and C16 shifted from δ_C 47.4 (CH), 78.4 (CH), 212.9 (C), 48.6 (CH), and 13.2 (CH₃) in corallolide A (1) to 50.5 (CH), 209.8 (C), 73.1 (CH), 41.6 (CH), and 7.1 (CH₃) in 2, respectively. Placement in corallolide B of a carbonyl at C2 and a hydroxyl (in the α-orientation) at C3 would account for these spectral differences. Consideration of ¹H and ¹³C NMR data as well as COSY, NOESY, and HMBC data allowed the complete structure to be assigned as 2 and led to the unambiguous assignment of all the protons and carbons as listed in Table 2. The NOEs and the small coupling constant between H3 and H4 (J_H3/H4 < 1 Hz), which require a dihedral angle close to 90° (calcd θ = −88.2°), correlated with the lowest energy conformer of a model representing the relative stereochemistry shown in structure 2.

Table 2.

¹H NMR (300 MHz), ¹³C NMR (75 MHz), ¹H—¹H COSY, NOESY, and HMBC spectral data for corallolide B (2)^a

position	δH, mult, intgt (J, Hz)	δC (mult)b	1H–1H COSY	NOESY	HMBCc
1	2.74, br d, 1H (10.4)	50.5 (CH)	H12β	H3, H4, H14αβ	H14αβ, H3-15
2		209.8 (C)			H1, H3
3	4.44, s, 1H	73.1 (CH)		H1, H4, H9, H12α	H3-16
4	1.78, br q (6.9)	41.6 (CH)	H3-16	H1, H3, H3-16	H3, H3-16
5		84.3 (C)			H3, H3-16
6		208.1 (C)			H18α
7		135.7 (C)			H18α, H3-19
8	5.89, br s, 1H	76.1 (CH)	H9, H18αβ, H3-19	H9	H9
9	8.09, br s, 1H	150.5 (CH)	H8	H3, H8, H12αβ
10		133.6 (C)			H9, H11αβ
11αβ	2.42, br m, 2H	20.8 (CH2)	H12αβ		H1, H12α
12αβ	1.54, br m, 1H; 2.17, br m, 1H	31.6 (CH2)	H1, H11αβ	H3, H9	H1
13		138.2 (C)			H1, H3-15
14αβ	5.08, br s, 1H; 4.80, br s, 1H	115.3 (CH2)	H3-15	H1, H3-15	H1, H3-15
15	1.70, br s, 3H	21.8 (CH3)	H14αβ	H14α	H1, H14αβ
16	0.88, d, 3H (6.9)	7.1 (CH3)	H4	H4, H18α	H3, H4
17		169.1 (C)			H18α, H3-19
18αβ	2.64, br d, 1H (17.3); 2.77, br d, 1H (17.3)	44.3 (CH2)	H8	H3-16	H3-19
19	2.33, br s, 3H	19.4 (CH3)	H8
20		174.0 (C)d			H9

^aSpectra were recorded in CDCl₃ at 25 °C. Chemical shift values are in parts per million relative to the residual CHCl₃ (7.26 ppm) or CDCl₃ (77.0 ppm) signals. Assignments were aided by 2D NMR experiments, spin-splitting patterns, number of attached protons, and chemical shift values.

^b13C NMR multiplicities were obtained from a DEPT-135 experiment.

^cProtons correlated to carbon resonances in the ¹³C column. Parameters were optimized for ^2,3J_CH = 6 and 8 Hz.

^dDue to its low intensity the chemical shift value of this peak was carefully estimated from HMBC experiments.

The co-ocurrence of compounds 1 and 2 with various pseudopterane lactones⁴ within P. bipinnata raises the possibility that the corallolides represent a further modification of an existing metabolite, thus suggesting the biogenetic pathway shown in Scheme 1. We suggest the semisystematic name “corallolane” to define this new carbon skeleton and a numbering system that preserves the C1–C20 numbering of the pseudopterane skeleton.⁸

In our primary in vitro anti-tuberculosis assay (at 128 μg/mL), compound 2 inhibited growth of Mycobacterium tuberculosis by 95%, but was inactive against the malaria parasite Plasmodium falciparum. Compound 1, however, displayed strong antimalarial activity (IC₅₀ = 10 μg/mL).