Six Trikentrin-like Cyclopentanoindoles from Trikentrion flabelliforme. Absolute Structural Assignment by NMR and ECD

Abstract

Six new cyclopenta[g]indoles were isolated from a West Australian sponge, Trikentrion flabelliforme Hentschel, 1912 and their structures elucidated by integrated spectroscopic analysis. The compounds are analogs of previously described trikentrins, herbindoles and trikentramides from related Axinellid sponges. The assignment of absolute configuration of the new compounds was carried out largely by comparative analysis of specific rotation, calculated and measured ECD and exploiting van’t Hoff’s principle of optical superposition. Five of the new compounds were chemically interconverted to establish their stereochemical relationships, leading to a simple chiroptical mnemonic for assignment of the this family of chiral indoles. The first biosynthetic hypothesis is advanced to explain the origin of the trikentrin-herbinole family, and proposes a pyrrole-carboxylic thioester initiated polyketide synthase mechanism.

Graphic Abstract

Introduction

The majority of marine indole alkaloids are derived from secondary metabolism of tryptophan. The indigo derivative, ‘Tyrian purple’ (6,6’-dibromoindigotin), from the hypobranchial gland of various marine mollusks, has been known since ancient times:¹ contemporary studies revealed its structure through X-ray crystallography.² Still more rare are simple indoles that lack carbon substitution at C-3; their structures imply biosyntheses that diverge substantially from tryptophan metabolism. In a study by Brennan and Erickson, one of the first investigations of red algae, 10 achiral polyhalogenated indoles were reported from Rhodophyllis membranacea Harvey, and more recent studies of R. membranacea have delivered another 18 new analogs.³

Among the more curious natural indoles are unique, optically active cyclopenta[g]indoles from two Axinellid sponges. Capon and MacLeod described antibacterial (+)-cis- and (+)-trans-trikentrins A (1, 2), (–)-trans-trikentrin B (4) and an inseparable mixture of cis- and iso-trans-trikentrin B (3 and 5, respectively) from Trikentrion flabelliforme Hentschel, 1912, collected in Darwin, Australia.⁴ Scheuer and coworkers characterized the trikentrin homologs, (–)-herbindoles A-C (6–8), from a specimen of Axinella sp. collected in Exmouth Gulf, Western Australia.⁵ The latter compounds, which exhibit modest cytotoxicity and fish antifeedant activity, are antipodal in configuration to the cis-trikentrins as established through independent total syntheses of ent-(–)-1, (+)-2,⁶ and ent-(+)-6–8 by Natsume and coworkers,⁷ and of (+)-4 by Kanematsu and coworkers.⁸ (+)-Trikentramides A-D (9–12), isolated and characterized by Quinn and coworkers from a specimen of T. flabelliforme collected at Port Hedland (500 km from Exmouth), are isatin and oxindole analogs of (+)-cis- and (+)-trans-trikentrins A.⁹

Aside from the above-referenced syntheses, the total syntheses of (+)-1,^10,11 (+)-2,¹² (±)-1 and (±)-2,^{11,13,14,15,16,17,18} (+)-3,¹¹ (±)-3,^13,19 and (±)-6–8 have been achieved.^13,19,20 A formal synthesis of (±)-3 has also been reported,²¹ and all total syntheses of trikentrin-herbindole family, up to 2009, have been reviewed.²²

In all likelihood, the biosynthesis of polyhalogenated indoles in red alga is linked to the enduring natural history of indole, indoxyl, indigo dye, and other indigoid compounds from plants of the genus Indigofera and, as shown through genetic and ecological studies, bacterial biosynthesis.²³ On the other hand, the origin of the sponge-derived 1,3-dimethylcyclopenta[g]indoles must differ considerably from the latter. To date, no hypotheses for the origin of 1–12 have been advanced.

Here, we report the structures of new trikentrin-like natural products, (+)-trans-herbindole A (13) and trikentramides E-I (14–18), from T. flabelliforme collected in Exmouth Gulf. The absolute configurations of 13–18 were assigned from chiroptical comparisons with trikentrins and herbindoles, and exploitation of van’t Hoff’s principle of optical superposition,²⁴ and chemical correlation to derive a new mnemonic for absolute stereostructure of this family of compounds. Surprisingly, the structures of 13–18 uniformly exhibit a trans configuration for the fused 1,3-dimethylcyclopentane ring, in contrast with the (–)-cis-herbindoles (6–8) obtained from a specimen of T. flabelliforme collected in the same vicinity,⁴ and are enantiomorphic with the latter. A unifying hypothesis is proposed that rationalizes the biogenesis of trikentrins, herbindoles, trikentramides and two related sponge-derived indoles from West African Trikentrion loeve.

Results and Discussion.

A MeOH extract of lyophilized Trikentrion flabelliforme, collected in Exmouth Gulf, Western Australia in 1993, was redissolved in MeOH-H₂O (9:1), and the solution was partitioned progressively against hexanes, CH₂Cl₂, and n-BuOH. The CH₂Cl₂-soluble ‘fraction B’, enriched in indoles (TLC spots stained bright fuschia with vanillin-H₂SO₄), was separated by gel filtration chromatography (Sephadex LH-20) and further purified by reversed-phase HPLC (C₁₈) to obtain pure 13 –18.

The simplest of the new compounds, trans-herbindole A (13), C₁₅H₁₉N (HRESITOFMS m/z 214.1592 [M+H]⁺), when compared with cis-herbindole A (6), confirmed that 13 and 6 are constitutional isomers, but diastereomeric in the 1,3-dimethylcyclopentane ring. Compound 13 displayed a relatively featureless¹H NMR spectrum (CDCl₃) consisting of signals for a trans-1,3-dimethylcyclopentane ring: two methyl doublets (δ_H 1.21, 3H, d, J = 6.6 Hz and δ_H 1.49, 3H, d, J = 7.2 Hz), two methines (δ_H 3.46, 1H, quintet, J = 7.2 Hz; δ_H 3.71, 1H, qdd, J = 7.2, 7.8, 9.6 Hz) and H₂-7, a diastereotopic methylene group (H-7α δ_H 2.10, 1H, ddd, J = 1.2, 7.5, 12.1 Hz and H-7β, δ_H 1.97, 1H, ddd, J = 7.2, 9.6, 12.1 Hz).²⁵ The relatively small separation in chemical shifts of the H₂-7 proton signals (Δδ = 0.13 ppm) is characteristic of pseudo-enantiotopic methylene protons in trans-1,3-dimethylcyclopenta[g]indole rings (e.g. enantiotopic CH₂ in trans-1,3-dimethylindane, Figure 1a), as opposed to the larger signal separation (Δδ >1.3 ppm) observed in the corresponding diastereotopic methylene of the cis-isomers.5^,9 A broad exchangeable singlet (δ 8.02, 1H, bs) was assigned to the indole NH. The remaining signals were pairs of aryl methyls (δ_H 2.35, 3H, s and δ_H 2.50, 3H, s), and pyrrole ring proton signals (δ_H 6.55, 1H, dd, J = 3.0, 2.4 Hz, and δ_H 7.14, 1H, t, J = 2.4 Hz). Comparisons of the ¹³C NMR chemical shifts of 13 with those of 9-12⁹ and cis-herbindole A (6)⁵further supported a 4,5-dimethylindole structure.

Figure 1.

Stereotopicity of the CH₂ group in trans- and cis-1,3-dimethylcyclopentanobenzene (1,3-dimethylindane). (a) enantiotopic (b) diastereotopic.

Trikentramide E, C₁₅H₁₉NO, (14) is an oxindole; the higher homolog of 12. The ¹³C chemical shifts of 14 at C-2 (δ_C 177.7 Cq) and C-3 (δ_C 35.7, CH₂) compare well with those of indolin-2-one,²⁶ but not indolin-3-one.²⁷ Again, the 1,3-dimethylindane ring is trans-substituted.

The formulas of trikentramides F (15, C₁₄H₁₅NO₂) and G (16, C₁₅H₁₇NO₂), with two oxygens, suggested homologous isatins, differing by substitution by Me at C-5 in 16. The red-shifted long-wavelength band in the UV-vis spectrum of each compound [λ_max 326 (log₁₀ 3.54)] was consistent with a substituted isatin (indolin-2,3-dione). The ¹³C NMR spectra (CDCl₃) showed pairs of downfield signals attributed to the aryl ketone and lactam carbonyl groups, respectively [e.g. 16, δ 183.2 (C-3), 160.2 (C-2)], matched those reported for the parent heterocycle, isatin (δ 184.5, 159.5, DMSO-d₆).²⁸ The ¹H NMR spectra of the two compounds appeared similar to those of 14 and supported the same trans-1,3-dimethylcyclopenta[g]indole ring. Full ¹H and ¹³C NMR assignments (Tables 1 and 2) were secured through analysis of COSY, NOESY, HSQC and HMBC spectra (Figures 2 and 3).

Table 1.

¹H NMR Data (δ, mult)^a for 13–18.

#	13	14	15	16	17	18
NH	8.03, bs	7.97, bs	8.80, bs	8.96, bs	7.99, bs	8.21, bs
2	7.14, t, 2.4
3	6.55, dd (2.4, 3.0)	3.40, s			4.98, s	4.99, s
4
5			6.68, s
6	3.46, quinb (7.2)	3.31, quinb (7.2)	3.27, mc	3.29, quinb (7.2, 0.9)	3.27, quinb (7.2)	3.30, mc
7α	2.10, ddd, (1.2, 7.5, 12.1)	2.01, ddd, (1.2, 7.8, 12.3)	2.02, ddd, (3.0, 7.8, 12.6)	2.04, ddd, (1.5, 8.2, 12.6)	1.97, dd, (7.8, 12.6)	1.96, bddd, (7.8, 12.6)
7β	1.97, ddd, (7.2, 9.6, 12.1)	1.83, ddd, (7.2, 9.9, 12.3)	1.91, dt, (8.4, 12.6)	1.85, ddd, (7.2, 9.6, 12.6)	1.79, dt, (9.0, 12.6)	1.79, m
8	3.71, qdd, (7.2, 7.5, 9.6)	3.43, qdd, (6.6, 7.8, 9.9)	3.31, mc	3.44, qdd, (7.2, 8.2, 9.6)	3.37, m	3.40, mc
4-Me	2.50, s	2.18, s	2.55, s	2.49, s	2.28, s	2.28, s
5-Me	2.35, s	2.14, s		2.14, s	2.15, s	2.15, s
6-Me	1.21, d, (6.6)	1.13, d, (6.6)	1.26, d, (7.2)	1.14, d, (7.2)	1.12, d, (7.2)	1.07, d, (7.2)
8-Me	1.49, d, (7.2)	1.32, d, (6.6)	1.23, d, (6.6)	1.36, d, (7.2)	1.27, d, (6.0)	1.25, d, (6.6)

^a600 MHz, CDCl₃.

^bquin = quintet.

^coverlap.

^dbdd = broad doublet of doublets.

Table 2.

¹³C NMR Data (δ, mult)^a for 13–18 (CDCl₃, 125 MHz).

#	13	14	15	16	17	18
2	122.9, CH	177.7, C	160.3, C	160.2, C	178.5, C	178.9, C
3	101.6, CH	35.7, CH2	b	183.2, C	69.7, CH	69.7, CH
3a	126.3, C	123.2, C	115.2, C	115.7, C	124.1, C	124.1, C
4	128.1, C	135.2, C	140.9, C	139.9, C	134.6, C	134.4, C
5	123.1, C	125.9, C	121.0, C	127.0, C	127.1, C	127.3, C
5a	125.7, C	148.7, C	161.4, C	160.6, C	150.9, C	150.8, C
6	38.4, CH	37.9, CH	38.5, CH	38.8, CH	37.9, CH	38.0, CH
7	43.9, CH2	43.8, CH2	43.0, CH2	43.2, CH2	43.8, CH2	43.7, CH2
8	36.5, CH	35.4, CH	34.6, CH	35.2, CH	35.6, CH	35.5, CH
8a	143.0, C	125.3, C	128.9, C	128.4, C	125.7, C	126.0, C
8b	130.8, C	131.5, C	144.5, C	142.8, C	134.3, C	134.1, C
4-Me	15.7, CH3	16.1, CH3	18.5, CH3	14.5, CH3	15.4, CH3	15.4, CH3
5-Me	15.2, CH3	16.1, CH3	–	13.4, CH3	14.9, CH3	14.9, CH3
6-Me	20.4, CH3	19.7, CH3	19.3, CH3	19.5, CH3	19.8, CH3	19.6, CH3
8-Me	20.7, CH3	20.1, CH3	19.8, CH3	19.6, CH3	20.0, CH3	19.9, CH3

^aMultiplicities determined from edited HSQC, or comparisons with 9–12. MHz (Ref. 9).

^bSignal not observed.

Figure 2.

COSY and HMBC correlations of (–)-trikentramide G (16) and (+)-trikentramide I (18) (500 MHz, CDCl₃).

Figure 3.

NOESY correlations of (–)-trikentramide G (16) (600 MHz, CDCl₃).

The formula of isomeric dioxindoles (+)-trikentramides H (17) and (+)-I (18), C₁₅H₁₉NO₂Na⁺ (HRESITOFMS m/z 268.1307 and m/z 268.1305, respectively [M+Na]⁺) confirmed the two are formal reduction products of 16. The ¹³C chemical shifts of 18 at δ_C 178.9 (C-2) and 69.7 (C-3), which are consistent with those of 3-hydroxy-5-methyl-indolin-2-one,³⁰ verified the dioxindole arrangement of keto and hydroxymethine groups at C-2 and C-3, respectively.

Absolute configurations of 13–18 were assigned from analysis of the CD and [α]_D data, and comparisons with published chiroptical data of known compounds and DFT calculations of ECD. The simplest assignment, by comparisons of [α]_Ds (Table 3), is that of (+)-trans-herbindole A (13, [α]_D +29.0) which follows from its structural similarity to (+)-trans-trikentrin A (2, [α]_D +23.3) and the relative invariance of [α]_D in trikentrins and herbindoles A and B, whether substituted at C-4 or C-5 by H, Me or Et (Table 3); therefore, both compounds bear the same (6S,8S) configuration (see below for more discussion).This is not necessarily true of 3–5, and 8; the conjugated E-butenyl chain, which constitutes a substituted styrenyl chromophore, adds complexity to the [α]_D of these indoles and abrogates simple chiroptical comparisons.

Table 3.

[α]_D for Natural and Synthetic Trikentrins, Herbindoles and Trikentramides.

Config.	Cmpd.	Sourcea	[α]Db	Ref.
	(+)-1	N	+48c	4
	(+)-1	S	+49	11
ent-	(−)-1	S	−68.6	6a,b
	(+)-2	N	+23.3d	4
	(+)-2	S	+24	6c
	(+)-2	S	+24	10b
ent-	(−)-2	S	−26.8	6a,b
	(+)-3	S	+102	6c
	(+)-3	S	+100.2	8
	(+)-3	S	+101	11
	(−)-4	N	−13e	4
ent-	(+)-4	S	+24	6c
	5	S	~0	6c
	(−)-6	N	−62	7
ent-	(+)-6	S	+56.9	7
ent-	(+)-7	S	+51.2	6a
ent-	(+)-8	S	+19.9	7
	(+)-9	N	+40.7	9
	(+)-11	N	+42.2	9
	(+)-12	N	+50.0	9
	(+)-13	N	+29.0	f
	(+)-14	N	+19.0	f
	15	N	~0	f
	(−)-16	N	−43.4	f
	(+)-17	N	+22.6	f
	(+)-18	N	+231.5	f

^aN = natural; S = synthetic.

^bAll measurements were made on solutions in CHCl₃ and, aside from the noted exceptions, within the concentration range c. ~0.027 – 0.50 g/100 cm³.

^cc = 2.47.

^dc = 1.0.

^ec = 1.97.

^fThis work.

The specific rotations of oxidized indoline analogs, trikentramides A-I, do not follow a simple pattern; the magnitude and sign vary with substitution and electronic properties in a complex manner. For example, the two isatins 15 and 16 differ only in a single Me group at C-5, but show [α]_Ds ~0 and –43.4, respectively. These are confoundingly different from 13 and 14 (+29 and +19.0, respectively). Moreover, the [α]_D of highly dextrorotatory 18 (+232) appears to be largely influenced by the stereocenter at C-3. Consequently, in order to make the stereoassignments of 15 and 16, we first chose to assign 17 and 18 by ECD and chemically correlate them to other family members.

Assignment of absolute configuration of trikentramides H (17) and I (18), which are epimeric dioxindoles, rested upon interpretation of their corresponding ECD spectra (Figure 4). The ECD spectra of oxidized indoles 14–16 exhibited relatively weak Cotton effects (CEs), not unlike Natsume’s observations of indoles 1-3.⁷ In contrast, the ECDs of 17 [λ 243 (Δε –7.4), 221 (+8.6)] and 18 [λ 243 (Δε +5.0), 224 (–5.6)] are virtual mirror images dominated by relatively strong biphasic CEs that reveal a predominant influence of the C-3 stereocenter.

Figure 4.

ECD spectra (CH₃CN, 23 ˚C) of (a) trikentramides H [(+)-17] and I [(+)-18], dash). (b) trikentramides F [15] and G [(–)-16, dash]. (c) trikentramides F [(–)-16] and E [(+)-14, dash]. (d) trans-herbindole A [(+)-13].

Asymmetric reductions of isatins to 3-hydroxoxindoles have been reported that provide comparison compounds for chiroptical assignments of 17 and 18.^29,30,31

For example, isatin undergoes enzyme-catalyzed reduction in the presence of a carbonyl reductase derived from Candida parapsilosis to give (+)-(R)-dioxindole in 21%ee.³¹ (R)-5-Methyldioxindole (19, 38.5%ee), prepared by Sonderegger and coworkers using partial asymmetric hydrogenation of 20 (H₂, Pt supported on Al₂O₃-(–)-cinchonidine, Scheme 1), exhibited an ECD spectrum dominated by a negative Cotton effect (CE) (λ 237 nm, –41, arbitrary units of ellipticity) and a second CE of opposite sign at longer wavelength (λ 264, +12.5).³⁰ The authors calculated the ECD spectrum of 19 using time dependent DFT (b3pw91 and 6–31G(d, p) basis set³²) and assigned their reduction product (R)-19. We repeated the reduction of 20 under non-asymmetric conditions (EtOH, NaBH₄) and resolved the product, (±)-19, by chiral phase HPLC (Phenomenex Lux 5U, Amylose 2) into (S)-19 and (R)-19 which, as expected, exhibited mirror image ECD spectra (see Supporting Information). The near identity of the ECD spectra of (R)-19 and (+)-18, and the expected weaker dichroic contributions from the 1,3-dimethylcyclopentane in other trikentramides (Figure 4b,c), leads to the conclusion that the configuration of the natural products are (3S,6S,8S)-17 and (3R,6S,8S)-18.

Scheme 1.

Partial asymmetric hydrogenation of 5-methylisatin. See Ref. 30 and Supporting Information for ECD.

Reduction of 16 (Scheme 2, NaBH₄, EtOH, rt) gave the C-3 epimeric dioxindoles 17 and 18, identical with the natural products by HPLC retention times, ¹H NMR, [α]_D and ECD. In addition, a minor product, from over-reduction of intermediate 17/18 at the benzylic carbon, was recovered and shown to match (+)-trikentramide E (14) by HPLC retention time, ¹H NMR and ECD. Finally, oxidation of the indole (+)-13 under the selective conditions recently reported by Wang and coworkers (I₂, t-BuOOH, DMSO, 80 ˚C, Scheme 2),³³ gave a product identical with natural (–)-16 by MS, ¹H NMR, ECD and HPLC retention time. Thus, (+)-trans-herbindole A (13) is chemically correlated to (+)-14, (–)-16, (+)-17 and (+)-18 demonstrating that the dimethylcyclopentane ring is stereochemically uniform among these five natural products. Although we have no independent correlation of (+)-trikentramide E (15), it seems highly probable the configuration is the same as to its congeneric family members.³⁴

Scheme 2.

Chemical interconversions of (+)-13 with (+)-14, (–)-16, (+)-17 and (+)-18.

The observed ECD spectra of 17 and 18 are the molar sum contributions of independent asymmetric perturbations to the benzenoid chromophore by the 1,3-dimethylcyclopentane group and the benzylic secondary OH group within their respective ‘first spheres of asymmetry’. The latter is a formal corollary of van’t Hoff’s principle of optical superposition,²⁴ which has been applied both to configurational assignments through deconvolution of both molar rotation [ϕ] and ECD (Δε),³⁵ in combination with DFT calculations (e.g. CADPAC calculated [α]_D of (–)-1,3,5,7-tetramethyl-1,3-dihydroindol-2-one^35c).

The largest contribution to both the [α]_D and CD in trans- and cis-fused trikentrin-type indoles appears to be the C-8 stereocenter. For example, both (+)-ent-cis-6 and (+)-trans-13 have the same C-8 configuration and are dextrorotatory, although the magnitude of [α]_D of the former is almost twice that of the latter (Table 3). The presence of a C-5 Me group has little effect on [α]_D; trans-(+)-2 and (+)-13 have almost the same specific rotation ([α]_D = +23.3 and +29.0, respectively). A brief conformational analysis of cis and trans herbindoles was carried out (MMFF94 and DFT) to derive their lowest energy conformations and frontier molecular orbitals. The conformation of the cyclopenta[g]indole ring in 13 (Figure 5) is distorted from an idealized C₂ conformation by torsional strain and steric repulsion between the N-H bond and the methyl group, and between the Me groups at C-5 and C-6. For example, the C-5–C-5a–C-6–C-6(Me) dihedral angle is θ = 52.1˚, but the C-8b–C-8a–C-8–C-8(Me) dihedral angle is considerably larger, ψ = 72.3˚, but the ‘unstrained’ trans-1,3-dimethylindane i – Figure 1 – which freely inverts between the two degenerate C₂ conformers – displays θ = 40.1˚; ψ = 76.1˚. In contrast, the cyclopentane ring of (+)-cis-herbindole A (ent-6) has an almost symmetric envelope conformation (θ = –53.2˚, ψ = 48.3˚).

Figure 5.

Energy minimized geometries (MMFF) and DFT calculated (ωB97X-D 6–31G*) LUMO of (a) (+)-trans-herbindole A (13) and (b) (+)-cis-herbindole A (ent-6). Conformational searching returned only one conformation for 6 or 13.

The major difference in frontier molecular orbitals of the delocalized π-system in 1,3-cyclopenta[g]indoles is the distortion of the LUMO by non-bonded interactions, particularly by the C-8 methyl group, which influences the rotational strength, and therefore ECD and [α]_D, by altering molecular orbital symmetry properties. The influence of the C-8 stereocenter upon chiroptical properties is extrapolated into a simple stereochemical mnemonic (Figure 6): for trikentrin-herbindole like indoles, the absolute configuration at C-8 determines the sign of [α]_D, whether the 1,3-dimethylcyclopentane ring is cis- or trans-fused. The exceptions are 1-butenyl substituted compounds (e.g. 3) where contributions from the additional rotational strength of the styrenyl chromophore perturbs this simple rule. As the presence of C-4 or C-5 alkyl substituents (e.g. Me, Et) appear not to significantly change the relationship between sign of specific rotation and absolute configuration, we predict that hydrogenation of 1-butenyl to an n-butyl group, prior to measuring the [α]_D, will restore the relationship; the products should conform to the mnemonic and allow stereoassignment.

Figure 6.

Chiroptical mnemonic for assignment of absolute stereostructures of trikentrins and herbindoles (a) trans- and cis-8S (b) trans- and cis-8R (R₁, R₂ = H, alkyl, but ≠ 1-alkenyl)

The biosynthesis of most natural product indoles follows familiar routes. Free indole is oxidatively liberated from Trp by the action of a tryptophanase (L-Trp indole lyase). Certain bacteria are known to harbor the enzyme and, in combination with oxidases, could give rise to polyhalogenated indoles in Rhodophycae. In considering possible biosynthetic pathways that generate 1–18, degradation of Trp seems unlikely; not only does C-3 of the indole ring lack a carbon substituent, but the introduction of the fused carbocyclic ring at C-5a,b and additional carbon substituents at C-4 and C-5 necessitates unwieldy complexity in this putative biogenesis. Consequently, alternative pathways must be considered. Alignment of the structures of two C-2 substitued indenyl-pyrrole metabolites, trikentramine (21)³⁶ and trikendiol (22)³⁷ (Figure 7, both isolated from West African Trikentrion loeve) with cis-trikentrin A (1) suggests a hypothesis: the trikentrin-herbindole family are of polyketide origin. Proline is oxidatively modified to the hypothetical pyrrole-2-carboxylic as a starter unit for ketide extensions by malonate (acetate) or methylmalonate (propionate).³⁸ In this scheme, the origin of the benzenoid ring differs slightly between two pathways, a and b. Because the C-C bond formation aligns carbons formally derived from the C=O group of the ketide extender units, it is unlikely that cyclization proceeds by aldol/Claisen condensations familiar from the biosynthesis of aromatic Type II polyketides. The indane ring of trikentramine (pathway a) is mostly likely assembled after partial reductions and polyene electrocyclization reactions – possibly involving oxidative single electron transfer (SET)³⁹ – followed by aromatization to the benzene ring. For trikentrin-like secondary metabolites, (pathway b), similar reactions proceed, but for the fused indole ring we invoke a Friedel-Craft type intramolecular alkylation at C-3 (pyrrole numbering) of the electron-rich pyrrole ring for construction of the C-3b–C-4 bond. Indeed, Natsume exploited a similar reaction for the indole ring-closing steps in the total syntheses of 2, 4, and 5.⁶

Figure 7.

Hypothesis for the biogenesis of trikentrin-like natural products, including cis-trikentrin A (1), trikentramine (21) and trikendiol (22).

The hypothesis succinctly accounts for the location of substituents at C-4 and C-5, namely H, Me, Et or 1-butenyl, through chain extensions or termination of the polyketide chain by homologated malonate units. Subsequently, a manifold of oxidative reactions of the product indoles would generate 20 or, conversely, the oxindole, isatin, and dioxindole scaffolds found in trikentramides A–D (9-12) and E–I (14–18).

Conclusion.

Six new cyclopenta[g]indole natural products – trans-herbindole A (13) and trikentramides E-I (14–18) – were isolated from the sponge Trikentrion flabelliforme and their absolute stereostructures solved by integrated analysis of MS, NMR and ECD studies, in combination with calculated DFT, and chemical correlation. A simple mnemonic emerged from analysis of the chiroptical data that allows easy identification of absolute configuration of indoles and dioxindoles within the series.

EXPERIMENTAL SECTION

General Experimental Procedures.

Optical rotations were measured on a JASCO P-2000 at the D-double emission line of Na. UV-vis spectra were measured on a JASCO V-630, spectrometer. ECD spectra were measured on a JASCO J-810 spectropolarimeter in quartz cells (1 or 5 mm pathlength) at 23 ˚C. FTIR spectra were collected on thin film samples using a JASCO FTIR-4100 fitted with an ATR accessory (ZnSe plate). Inverse-detected 2D NMR spectra were measured on a JEOL ECA (500 MHz) spectrometer, equipped with a 5 mm ¹H{¹³C} room temperature probe, or a Bruker Avance II (600 MHz) NMR spectrometer with a 1.7 mm ¹H{¹³C/¹⁵N} microcryoprobe. ¹³C NMR spectra were measured on a Varian NMR spectrometer (125 MHz) equipped with a 5 mm Xsens ¹³C{¹H} cryoprobe. NMR spectra were referenced to residual solvent signals, (CDCl₃, δ_H 7.26, δ_C 77.00 ppm). High-resolution ESITOF analyses were carried out on an Agilent 1200 HPLC coupled to an Agilent 6350 TOFMS at the Small Molecule MS Facility (UCSD). Low-resolution MS measurements were made using a Thermoelectron Surveyor UHPLC coupled to an MSD single-quadrupole detector. Preparative and analytical HPLC was performed on an Agilent 1100 HPLC, or a Jasco system comprising dual-pumps (PU-2086) and mixer, UV-vis detector (UV-2075) in tandem with an ELSD detector (Softa-A model 300).

Animal Material.

The sponge T. flabelliforme (93-07-076) was collected in Exmouth Gulf (800 m south of Bundegi Beach), Western Australia, in 1993 at a depth of –9 m using scuba. The sponge was an orange, pliable, ear-shaped specimen with a smooth surface that, with pressure, emitted a red exudate. Microscopic spicule analysis revealed the predominance of styles consistent with the species T. flabelliforme. A voucher sample of the sponge is archived at UC San Diego.

Extraction of Sponge and Isolation of (+)-trans-Herbindole A (13) and Trikentramides E-I (14–18).

Lyophilized Trikentrion flabelliforme (63.88 g dry wt) was extracted with CH₂Cl₂-MeOH (2 × 400 mL, 12 h), and the combined organic extracts were concentrated and dissolved in H₂O-MeOH (1:9, 400 mL) prior to repeated extraction with hexane (400 mL x 2). Concentration of the hexane-soluble layer gave Fraction A (1.4604 g). The aqueous-MeOH layer was adjusted to 2:3 H₂O-MeOH followed by extraction with CH₂Cl₂ (500 mL x 2) to yield Fraction B (0.6845 g). The aqueous layer was concentrated and extracted with n-BuOH (400 mL x 2) to yield the C layer (0.8460 g). The remaining aqueous layer was dried to yield Fraction D (1.7995 g). Fraction B, the TLC of which stained bright fuschia⁴⁰ with vanillin-H₂SO₄, was separated further by size exclusion chromatography (Sephadex LH-20, elution with MeOH) to give 10 fractions, which were pooled according to TLC (10% MeOH-CH₂Cl₂), UV-activity and staining with vanillin-H₂SO₄ in EtOH. Fractions 4 (0.0698 g) and 6 (0.0357 g) were further purified by reversed-phase HPLC (Phenomenex, Kinetex C₁₈ column, 150 × 21.2 mm, linear gradient, initial conditions 75:25 H₂O-0.1% TFA-CH₃CN for three minutes, 30:70 for 17 minutes to 100% CH₃CN for the last 6 minutes, flow rate = 12 mL·min⁻¹) to give trans-herbindole A (13, 4.7 mg, 0.0074 % dry mass, t_R = 25.26 min), trikentramides F (15, 2.4 mg, 0.0038 % dry mass, t_R = 18.21 min), G (16, 5.5 mg, 0.0086 % dry mass, t_R = 19.62 min), H (17, 1.6 mg, 0.0025 % dry mass, t_R = 15.12 mim) and I (18, 2.8 mg, 0.004 % dry mass, t_R = 15.42 min). Pure trikentramide E (14, 1.8 mg, 0.0028 % dry mass, t_R = 21.32 min) was obtained after slightly different reversed-phase HPLC conditions (Phenomenex, Kinetex C₁₈ column, 150 × 21.2 mm, linear gradient, initial conditions 75:25 H₂O-0.1% TFA-CH₃CN for three minutes, 30:70 for 21 minutes to 100% CH₃CN for the last 2 minutes, flow rate = 12 mL·min⁻¹).

(+)-trans-Herbindole A (13): Gray oil; UV (CH₃CN) λ_max 221 nm (logε 4.53), 271 (3.86); [α]_D²³ +29 (c 0.24, CHCl₃); ECD (CH₃CN) λ 214 (Δε +5.3), 230 –7.3), 269 (–3.0), 294 (+0.66); FTIR (ATR, ZnSe plate) ν 3430, 2953, 1393, 1375, and 727 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 214.1592 [M+H]+ (calcd for C₁₅H₂₀N⁺ 214.1590).

(+)-Trikentramide E (14)⁴¹: pale orange solid; UV (CH₃CN) λ_max 204 nm (logε 4.10), 213 (4.08), 253 (3.63); [α]_D²³ +19 (c 0.18, CHCl₃); ECD (CH₃CN) λ 215 (Δε –1.45); FTIR (ATR, ZnSe plate) ν 3206, 2948, 2871, 1670 (s), 1626 (s), and 711 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 230.1534 [M+H]⁺ (calcd for C₁₅H₂₀NO⁺ 230.1539).

Trikentramide F (15): Dark yellow solid; UV (CH₃CN) λ_max 199 nm (logε 4.23), 218 (4.03), 245 (3.98), 326 (3.54); [α]_D²³ ~0 (c 0.12, CHCl₃); ECD (CH₃CN) λ 210 (Δε +2.4), 214 (2.2), 252 (–0.7); FTIR (ATR, ZnSe plate) ν 3177, 2960, 2932, 1737 (s), 1726, 1638, 1390 and 1377 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 230.1173 [M+H]⁺ (calcd for C₁₄H₁₆NO₂⁺ 230.1176).

(–)-Trikentramide G (16): Orange solid; UV (CH₃CN) λ_max 199 nm (logε 4.36), 218 (4.14), 249 (4.14), 328 (3.70); [α]_D²³ –43 (c 0.55, CHCl₃); ECD (CH₃CN) λ 192 (Δε –2.2), 203 (–2.6), 221 (+4.9), 255 (+1.1); FTIR (ATR, ZnSe plate) ν 3311 (br), 2946, 2872, 1738 (s), 1725 (s), 1626, 1592, 1382 and 1344 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 244.1331 [M+H]⁺ (calcd for C₁₅H₁₈NO₂⁺ 244.1332).

(+)-Trikentramide H (17): Pale yellow solid; UV (CH₃CN) λ_max 269 nm (logε 3.97), 221 (4.17); [α]_D²³ +22.6 (c 0.053, CHCl₃); ECD (CH₃CN) λ 221 (Δε +8.6), 243 (–7.4), 272 (+0.6); FTIR (ATR, ZnSe plate) ν 3311 (br), 2957, 2928, 1715 (s), 1704 (s), 1633 and , 1203 and 727 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 268.1309 [M+Na]⁺ (calcd for C₁₅H₁₉NO₂Na⁺ 268.1308).

(+)-Trikentramide I (18): Tan solid; UV (CH₃CN) λ_max 221 nm (logε 3.96), 252 (3.39); [α]_D²³ +232 (c 0.093, CHCl₃); ECD (CH₃CN) λ 224 (Δε –5.6), 243 (+5.0); FTIR (ATR, ZnSe plate) ν 3397 (br), , 2958, 2925, 1681 (s), 1633 (s), 1447, 1204 1189, and 1138 cm⁻¹; ¹H NMR and ¹³C NMR (CDCl₃), Tables 1 and 2; HRESITOFMS m/z 268.1305 [M+Na]⁺ (calcd for C₁₅H₁₉NO₂Na⁺ 268.1308).

Reduction of 5-Methylisatin – Dioxindoles (S)-19 and (R)-19.

Reduction of 5-methylisatin (20, 200 mg, 1.2 mmol) with NaBH₄ (was carried out according to the protocol described by Hara and coworkers.⁴² The crude product was purified by silica flash chromatography (1:1 EtOAc/CHCl₃) to provide (±)-5-methyldioxindole [(±)-19] as a colorless solid, 101 mg (50%). UV-vis (CH₃CN) λ_max 209 nm (logε 4.11), 248 (3.65), 296 (2.88). LRMS m/z 164.20 [M+H]⁺. The ¹H and ¹³C NMR spectra of the product matched the literature values for (R)-19.³⁰

A sample of (±)-19 (~100 μg) was resolved by chiral reverse-phased HPLC (Lux 5U Amylose-2 column, 250 × 4.60 mm, isocratic conditions 2.5% i-PrOH-CH₃OH, flow rate = 1 mL·min⁻¹) to give the enantiomeric dioxindoles (R)-19 (t_R = 4.68 min) and (S)-19 (t_R = 5.52 min). (S)-19: ECD (2.5% i-PrOH-CH₃OH) λ 210 (Δε +54), 240 (–24.5), 267 (+7.4). (R)-19: λ 210 (Δε –54), 240 (+24.9), 267 (–7.0), [lit.³⁰ 237 (–41), 264 (+12.5), arbitrary units of ellipticity]. See Supporting Information for ECD spectra.

Reduction of (–)-Trikentramide G (16).

NaBH₄ (1.40 mg, 35.8 μmol) was added in portions to a solution of 16 (3.0 mg, 8.2 μmol) in anhydrous EtOH (1.0 mL) at room temperature. After 1 hr., the solution was diluted with H₂O (2.0 mL), acidified to pH ~ 6 with HCl (1N), and extracted with CH₂Cl₂ (2 × 2 mL). The organic layer was dried and the residue (5.2 mg) was purified by reversed-phase HPLC (Phenomenex, Kinetex C₁₈ column, 150 × 21.2 mm, linear gradient, initial conditions 75:25 H₂O-0.1% TFA-CH₃CN for three minutes, 30:70 for 17 minutes to 100% CH₃CN for the last 6 minutes, flow rate = 12 mL·min⁻¹) to give 14 (0.5 mg, 18%, t_R = 19.30 min), 17 (1.0 mg, 33%, t_R = 15.13 min) and 18 (1.2 mg, 40%, t_R = 15.43 min) with ECD and HPLC retention times identical with the natural products.

Oxidation of (+)-trans-Herbindole A (13) to (–)-Trikentramide G (16).

(+)-trans-Herbindole A (13) was oxidized according to the procedure of Wang and coworkers.³³ A solution of (+)-13 (1.0 mg, 4.7 μmol) in DMSO (0.05 mL) was treated with a solution of iodine (1.4 mg, 5.6 μmol) and t-butyl hydroperoxide (70% w/v solution in H₂O, 3.0 μL, 23.4 μmol) in DMSO (0.50 mL) and heated to 80 ˚C with stirring for 18 h. The solution was cooled, quenched with aqueous Na₂S₂O₃ (5% w/v, 2.0 mL), extracted with EtOAc (3 × 1.0 mL) and purified by reversed-phase HPLC (Luna C₁₈ column, 250 × 4.60 mm, linear gradient, initial conditions 65:35 H₂O-0.1% TFA-CH₃CN for three minutes, 30:70 for 20 minutes to 100% CH₃CN for the last 6 minutes, flow rate = 1.0 mL·min⁻¹) to give a product, as a yellow glass (0.5 mg, 44%) that was identical to natural (–)-trikentramide G (16) by MS, ¹H NMR, ECD and HPLC retention time.