Highly Polar Spiro-Isoxazolines from the Sponge Aplysina fulva

Abstract

Two new highly polar brominated spiroisoxazolines, araplysillin N⁹-sulfamate (1) and an N-[5S,10R)-7,9-dibromo-10-hydroxy-8-methoxy-1-oxa-2-azaspiro[4.5]deca-2,6,8-triene-3-carboxy]-4-aminobutanoic acid (2), were isolated from a sample of Aplysina fulva collected in the Florida Keys. The absolute stereostructures of the new compounds were determined from analysis of MS, ¹H and ¹³C NMR and CD spectroscopy. Compound 2 provides a structural clue that may unify the biosynthesis of brominated spiroisoxazolines.

Secondary metabolites from sponges of the Order Verongidae¹ are typically highly oxidized compounds derived from bromotyrosine. Modified alkaloids derived from 3′,5′-dibromotyrosine (Figure 1, i, R = H) have been found from various Verongid genera² – mainly Aplysina, Psammaplysilla, Pseudoceratina and Verongia³ – that are widely distributed throughout Mediterranean, Pacific and Atlantic waters.⁴ Heterocycles based on brominated spiro-isoxazolines (Figure 1, iii, R=H, (5S,10R)-7,9-dibromo-10-hydroxy-8-methoxy-1-oxa-2-azaspiro[4.5]deca-2,6,8-triene-3-carboxamide) are common natural products arising from bromotyrosine secondary metabolism.⁵ Since the first example of a spiroisoxazoline was reported by the Minale group from two Mediterranean species, Aplysina aerophoba and Verongia thiona,⁶ over 25 spiroisoxazoline analogs have been described.⁴

Figure 1.

Putative biosynthesis of spiroisoxazoline carboxamide iii (R = H, (5S,10R)-7,9-dibromo-10-hydroxy-8-methoxy-1-oxa-2-azaspiro[4.5]deca-2,6,8-triene-3-carboxamide) from a 3′5′-dibromotyrosine residue.

Recently, we reported geographic variability of the diastereomeric compositions of fistularin-3 and 11-epi-fistularin-3 in Aplysina species collected from Brazil and the USA (Florida Keys) and Agelas from Australia (the Great Barrier Reef).⁷ We have extended these investigations and now report two new polar water-soluble spiroisoxazolines – the sulfamate 1 and carboxylic acid 2. Both compounds occur in very low concentrations in the most polar fractions derived from column chromatography of methanol-soluble components of the sponge extract. Compound 2 is a lower homolog of purpuroceratic acid (5),⁸ reported by Kijoa and coworkers, and lacks the unusual aryl C-methyl group of the latter. Compound 1 is the N-sulfato derivative of the known compound araplysillin-1 (3) and is formally the decarboxylation product of the N-sulfato α-aminoacid, ianthesine D (5), reported by Okamoto et al. from Ianthella sp.⁹ Carboxylic acids 2 and 5 may provide a “missing link” that unifies the biosynthesis of several spiroisoxazoline alkaloids from Verongid sponges.

Moderately polar fractions obtained from silica chromatography of the CHCl₃-MeOH extract of Aplysina fulva contained 11-hydroxyfistularin-3 and 11-epi-fistularin-3 in variable proportions as we have reported previously. The most polar fraction obtained from elution of the silica column (100% MeOH) was further separated by C₁₈ reversed phase HPLC to give a 3:2 mixture of compounds 1 and 2, respectively. Final purification of this mixture by preparative silica TLC gave pure samples of each compound 1 and 2 as colorless solids.

Compound 1 is an optically active solid, [α]_D +100 (c 0.06, MeOH) with UV activity [λ_max206 nm (log ε 4.70), 278 (3.84)]. The compound did not dissolve in CHCl₃, but was partially soluble in H₂O, MeOH, and soluble in DMSO and CHCl₃-MeOH. Analysis of the formula of 1 by mass spectrometry was complicated by multiple pseudomolecular ions and neutral losses. The negative ion ESIMS (m/z 841 [M−H+Na₂]⁺, 794 [M–Na]⁻ and 714 [M–SO₃–Na]⁻) suggested a Na⁺ salt of sulfate-half acid. The formula of C₂₁H₂₃Br₄N₃O₈S for the neutral species, requiring ten double bond equivalents, was assigned from positive-ion MALDI HRMS (m/z 841.7643, [M–H+Na₂]⁺ Δm = 3.3 mmu). The presence of only 19 distinct signals in the ¹³C NMR spectrum suggested an element of symmetry in an aryl ring. The ¹H NMR spectrum of 1 (Table 1) indicated the presence of three vinyl and aromatic protons, comprising H-5 in the spiro-ring system (δ 6.31, s, 1H) and the two-proton signal in the second ring belonging to a symmetrical 3,5-dibromotyrosine (δ 7.41, s, 2H). The ¹H (CDCl₃-CD₃OD) and ¹³C NMR (DMSO-d₆) signals – in particular, the C=N signal (δ 154.6, s) and the characteristic AB quartet for H₂-7 (δ 3.01, d, J = 18.3 Hz; 3.79, d, J = 18.3 Hz) – were consistent with a 1-oxa-2-azaspirodecatriene-ring system as seen in the structures of aeroplysinin-1¹⁰ fistularin-3,¹² and related compounds. NMR signals were observed (Table 1) for other units including a 1,3-disubstituted propane chain terminated with an oxygen atom and an NH(CO) group, and an 1-aryl-2-ethylamine side chain.

Table 1.

NMR Data for Compound 1 (600 MHz)

	CD3OD-CDCl3 (2:1)a				DMSO-d6
no.	δCb	δH [mult., J (Hz)]	HMBC (H→C)	DQF-COSY	δCb	δH [mult., J (Hz)]	HMBC (H→C)	DQFCOSY
1	74.6 (CH)	4.16 s	3,4,5,6		73.6	3.91 d (7.5)	3,4,5,6	OH
2	122.2 (C)d				120.8
3	148.5 (C)				147.3c
4	113.9 (C)d				113.1
5	131.4 (CH)	6.31 s	1,2,3,4,6,7		131.2	6.58 s	1,2,3
6	92.1 (C)				90.2
7	39.7 (CH2)	3.02 d(18.3) 3.82 d(18.3)	1,5,6,8		39.3c	3.21 d (18.6) 3.62 d (18.6)	1,5,6,8
8	154.6 (C)				154.3c
9	160.6 (C)				158.9
10	37.7 (CH2)	3.61 t (6.4)	9,11,12	11	36.2	3.39 td (6.6, 5.7)	9,11,12	11, CONH
11	29.9 (CH2)	2.09 pent (6.4)	10, 12	10, 12	29.4	1.98 p (6.6)	10,12	10,12
12	71.6 (CH2)	4.05 t (6.4)	10,11,13	11	71.2	3.95 t (6.6)	10,11	11
13	151.8 (C)				150.8c
14, 18	118.5 (C)				117.0
15, 17	133.7 (CH)	7.41 s	13,14,18,19		133.0	7.49 s	13,14,18,19
16	139.6 (C)				140.6
19	35.3 (CH2)	2.79 t (7.3)	15,16,17,20	20	33.7	2.67 t (6.7)	15,16,17,20	20
20	45.4 (CH2)	3.20 t (7.3)	16,19	19	44.7	2.92 q (6.7)	16,19	19, SO3NH
OMe	60.3 (CH3)	3.71 s	3		59.6	3.64 s	3
OH						6.37 d (7.5)		1
CONH						8.55 t (5.7)	9	10
SO3NH						4.14 t (6.7)		20

^aRun as 3:2 mixture with compound 2.

^b500 MHz

^cAssigned by HSQC and HMBC (J = 8 Hz) at 600 MHz

^dAssignment based on calculated ¹³C chemical shifts (ChemDraw Ultra).

A database survey (MarinLit, University of Canterbury) of known Verongid sponge compounds, matched the expected formula (with the assumption of one SO₃ group) to a mono-sulfated derivative of araplysillin-1 (3).¹¹ Verification of this assignment came from ¹³C NMR chemical shifts of 1 (Table 1) which matched closely those of 3.⁹ (Table 1) The assembled structure 1 and complete chemical shifts assignments were obtained by linking substructures from interpretation of 2D NMR data (gCOSY, gDQFCOSY, gHMBC, and gHSQC).

The SO₃ group was located in 1 as follows. Three exchangeable signals were detected (ESIMS measured in CD₃OD); these appeared in the ¹H NMR spectrum (DMSO-d₆) and were attributable to an OH group (δ 6.37 d, J = 7.5 Hz), one amide NH (δ 8.55, t, J = 5.7 Hz) and an unidentified upfield NH signal (δ 4.14, t, J = 6.7 Hz). Since all other heteroatoms in the formula of 1 were accounted for, the SO₃ group was placed on a nitrogen atom that allowed us to ascribe the latter NH signal to a sulfamate group (-NHSO₃⁻). The NH chemical shift is consistent with those observed for other sulfamates (e.g., δ 4.91 for ianthenisin C¹⁰).

Compound 2, [α]_D +140 (c 0.04, MeOH), also showed an isotope pattern for the MS parent ions consistent with the presence of two Br atoms. The molecular mass of 2 was inferred as 466 amu from low-resolution positive- and negative-ion ESIMS (m/z 489 [M+Na]⁺, 465 [M−H]⁻), but the formula C₁₄H₁₆Br₂N₂O₆ could only be confirmed by MALDI HRMS (m/z 488.9297 [M+Na]⁺, Δm = 2.4 mmu). and required seven double bond equivalents. Unlike 1, the appearance of 14 distinct signals in the ¹³C NMR spectrum of 2 (Table 2) showed lack of symmetry. The ¹H NMR spectrum of compound 2 (Table 2) was much simplified compared to that of 1, but retained signals due to the spiroisoxazoline unit and a 1,3-disubstituted propane side chain. The ¹³C NMR spectrum of 2 showed two C=O ¹³C NMR signals; one corresponding to a free carboxylic acid (δ 181.5 s) and a second due to the α-oximimo amide group (δ 158.9 s). Treatment of 2 with diazomethane gave the corresponding methyl ester 2a with a new ¹H NMR signal (CD₃OD) due to a second OMe group (δ 3.66, s, 3H). Carboxylic acid 2 most closely resembles purpuroceratic acid B (5) with the only difference being the absence of a methyl group at C-5.⁵

Table 2.

NMR Data for Compound 2 (600 MHz)

	CD3OD-CDCl3 (2:1)a				DMSO-d6
no.	δCb	δH [mult., J (Hz)]	HMBC (H→C)	DQF-COSY	δCb	δH [mult., J (Hz)]	HMBC (H→C)	DQFCOSY
1	74.6 (CH)	4.15 s	3,4,5,6		73.4	3.93 s	3,4,5,6
2	122.2 (C) d				120.8
3	148.5 (C)				147.0
4	113.9 (C) d				113.3
5	131.4 (CH)	6.30 s	1,2,3, 4,6,7		131.2	6.56 s	1,2,3,4,7
6	91.9 (C)				90.1
7	39.8 (CH2)	3.01 d (18.0) 3.79 d (18.0)	1,5,6,8		39.7c	3.22 d (17.7) 3.63 d (17.7)	1,5,6,8
8	154.7 (C)				154.8
9	160.6 (C)				158.9
10	40.2 (CH2)	3.30 t (6.9)		11	39.4c	3.14 t (6.6)	9,11	11
11	26.1 (CH2)	1.81 p (6.9)	10,12,13	10, 12	24.5	1.65 p (6.6)	10,12,13	10,12
12	35.9 (CH2)	2.21 t (6.9)	10,11,13	11	34.5c	2.09 bt (6.6)	10,11,13	11
13	181.5 (C)				174.7c
OMe	60.3 (CH3)	3.71 s	3		59.6	3.63 s	3
OH
CONH
COOH

^aRun as 2:3 mixture with compound 1.

^b500 MHz.

^cAssigned by HSQC and HMBC (J = 8 Hz) at 600 MHz

^dAssignment based on calculated ¹³C chemical shifts (ChemDraw Ultra)

The relative configurations of 1 and 2 were addressed using ¹³C NMR, ROESY, and comparisons of their circular dichroism (CD) spectra with those of known spiroisoxazolines. Since the ¹³C NMR chemical shifts of the 1-oxa-2-azaspirodecatriene-ring systems in 1 and 2 were essentially identical, we could assume the relative stereochemistry was the same in both and assign them by analysis of 1. The ROESY spectrum (500 mS) of 1 (DMSO-d₆) showed dipolar coupling between the C-1 OH and the proximal diastereotopic proton of the C-7 methylene group (δ 3.62, d, J = 18.0 Hz)¹² that places both groups syn to each other. The absolute configuration of 1 and 2 were assigned by circular dichroism (CD). The CD spectra of 1 and 2 were very similar [CD MeOH, 1: λ_max 244 (Δε +8.3), 285 (+8.0). 2: λ_max 243 (Δε +7.0), 289 (+5.2)] and showed two prominent positive Cotton effects (CE) that were of the same sign and similar magnitudes to those of (1R,6S)-spiroisoxazolines, particularly aerothionin (6a).^6,13 Thus, the absolute configurations are as depicted in structures 1 and 2.

Compounds 2 and 5 contain a structural motif – a spiroisoxazoline carboxyl group N-acylated to 4-aminobutanoic acid – that shows a trend that may help explain their biosynthesis. We propose each dibromospiroisoxazoline derives from a dipeptide comprised of an N-terminal 3,′5′-dibromotyrosine and another common α-amino acid (Figure 1, i R = amino acid) that has undergone decarboxylation to the corresponding amine either before or after peptide bond formation (Figure 2). For example, if the 1,4-diaminobutane (putrescine) linker in 6a derives from ornithine (or lysine, in the case of homoaerothionin, 6b¹⁴), it would appear the 4-aminobutanoic acid unit in 2 derives from glutamic acid. In contrast, the linker in 1, 3, 4, fistularin-3¹² and most other spiroisoxazolines is 3-amino-1-propanol, which does not have an obvious biogenesis unless it is considered as the decarboxylation product of homoserine, an uncommon amino acid that is an intermediate of one pathway to S-adenosyl methionine (SAM). Since the opposite end of the 3-amino-1-propanol linker in fistularin-3 is O-alkylated to the phenolic oxygen of a 3′,5′-dibromotyramine (DBT) unit, it is conceivable that the immediate precursor to the fistularin-3 linker is SAM which participates in an aberrant S_N2 type alkylation of DBT phenoxide at the more substituted S-CH₂ carbon (C-3) of the sulfonium ion instead of the S-Me carbon. After decarboxylation, the resultant amine 7 participates in amide bond formation with a spiroisoxazoline unit ii that is derived separately by ‘normal’ O-methylation of DBT with SAM and arene oxidation (Figure 1). A putative ‘C³-alanyl-methyltransferase’ might catalyze the former transformation and rationalize the unusual N,O-substituted C₃ unit that links two of three DBT-derived groups in high-molecular mass bromotyrosine natural products from Verongid sponges.

Figure 2.

Proposed unified biosynthesis of 1–6 and aberrant S_N2 substitution of S-adenosylmethionine (SAM) by 3′5′-dibromotyrosine phenoxide (path b) by hypothetical ‘C₃-alanyl-methyltransferase’.

Insufficient quantities of 1 and 2 were available to assess their antifungal activity against Cryptococcus neoformans or Candida albicans. Further investigations are pending to identify antifungal active principles in Aplysina fulva and other sponges from the Bahamas.

Experimental Section

General Experimental Procedures

Optical rotations were measured on a JASCO DIP370 or P1020 polarimeter. UV spectra were recorded on a Hewlett Packard 8452A single-beam spectrometer and CD spectra were measured using a JASCO 810 spectropolarimeter. IR spectra were recorded on Mattson Galaxy 3000 FTIR instrument ESIMS was measured using a Finnigan LCQ Deca mass spectrometer and high-resolution MALDI FTMS spectra were provided by the University of California, Riverside, MS laboratory. ¹H NMR and 2D NMR spectra were recorded on a Bruker 600 MHz DRX-600 equipped with a 5 mm cryoprobe and ¹³C NMR spectra were recorded on a Bruker Avance 500 MHz NMR spectrometer. Chemical shifts were referenced to CD₃OD (δ_H = 3.31 ppm; δ_C = 49.0 ppm), and DMSO (δ_H = 2.50 ppm; δ_C = 39.5 ppm). Solvents used were of HPLC grade.

Animal Material

Samples of Aplysina fulva (95-03-025, total ~1.4 kg) were collected by hand using scuba from Dry Rocks near Key Largo, Florida in 1995 and immediately stored at −20 °C until needed. Voucher specimens are archived in the Department of Chemistry and Biochemistry, UC San Diego.

Extraction and Isolation

A CHCl₃-MeOH-soluble fraction (3.14 g) of the MeOH extract of the sponge was separated by silica flash chromatography and eluted using a solvent gradient (1.5–100% MeOH-CHCl₃) to give an early-eluting fraction (10% MeOH-CHCl₃) containing fistularin-3,¹² 11-epi-fistularin-3¹⁵ and 11-oxoaerothionin¹⁶ as previously described.⁷ The crude fraction eluting with 100% MeOH was submitted to further purification by reversed-phase HPLC (Dynamax C₁₈, 5 μm, 10 × 250 mm, 3:1 H₂O-CH₃CN, 4.0 mL/min) to a single peak comprising a 3:2 mixture of 1 and 2 (4.8 mg, 0.0017% wet wt) respectively. Further purification of 3 mg of this mixture by silica TLC (4:21 MeOH-CHCl₃) provided 1 (1.2 mg) and 2 (0.7 mg) as white solids.

1. [α]_D²³ +100 (c 0.06, MeOH); UV (MeOH) λ_max 206 nm (log ε 4.70), 278 (3.84); CD (MeOH) λ 244 nm (Δε +8.3), 285 (+8.0); IR (neat) ν_max 3288, 2939, 2864, 1664, 1593, 1544, 1458, 1310, 1218, 1044, 990, 933, 866, 739 cm⁻¹; ¹H NMR (600 MHz) and ¹³C NMR (125 MHz), see Table 1; ESIMS m/z 841 [M−H+Na₂]⁺, 794 [M–Na]⁻, 714 [M–SO₃–Na]⁻; MALDI HRMS m/z 841.7643 [M−H+Na₂]⁺, calcd for C₂₁H₂₂N₃O₈Na₂S⁸¹Br₄ 841.7616.

2. [α]_D²³ +140 (c 0.04, MeOH); UV (MeOH) λ _max 223 nm (log ε 3.93), 281 (3.63); CD (MeOH) λ 243 nm (Δε +7.0), 289 (+5.2); IR (neat) ν_max 3321, 2959, 2930, 2850, 1666, 1581, 1407, 1310, 1271, 1218, 1048, 990, 921, 767, 739, 703 cm⁻¹; ¹H NMR (600 MHz) and ¹³C NMR (125 MHz), see Table 2; ESIMS m/z 489 [M+Na]⁺, 465 [M−H]⁻; MALDI HRMS m/z 488.9297 [M+Na]⁺, calcd for C₁₄H₁₆N₂O₆Na⁷⁹Br₂ 488.9273.

Methyl Ester of Carboxylic Acid 2. 2a

A solution of 2 (600 μg) in MeOH (~0.5 mL) at 0 °C was treated with an excess of ethereal solution CH₂N₂ (~0.2 M) and allowed to warm to rt over 20 min. The mixture was concentrated and separated on a pencil column (silica, 15:85 MeOH-CH₂Cl₂) to give 2a as a colorless solid (310 μg). ¹H NMR (CD₃OD) δ 3.66 (s, 3H, OMe), 3.72 (s, 3H, OMe); HREIMS m/z 479.9542 [M]⁺ calcd for C₁₅H₁₈O₆N₂⁷⁹Br₂ 479.9526.