A Chemical Study of Three Novel Cyclic Depsipeptides Produced by a Sponge-Derived Fungus

Abstract

Two novel cyclic depsipeptides, guangomides A (1) and B (2), together with a new destruxin derivative (3) were isolated from the cytotoxic extract obtained from the saltwater culture of an unidentified sponge-derived fungus. The new structures were elucidated based on analysis of extensive 1D and 2D NMR data sets, and the absolute configurations of 2S, 9S, 13S, 19S, 24R, 28R of 1 was determined based on the combined X-ray and Marfey's method structure analysis. Identical absolute configurations were assumed for 2. The cytotoxicity of the extract was found to be due to brefeldin A, while 1 and 2 showed weak antibacterial activity against Staphylococcus epidermidis and Enterococcus durans.

A few marine invertebrates are known to be a source of structurally fascinating and biologically active peptides and depsipeptides. Some most noteworthy examples include the peptide anticancer drug candidates dolastatin 10,ⁱ from the sea hare Dolabella auricularia (but also the cyanophyte Symploca sp VP642ⁱⁱ) and kahalalide Fⁱⁱⁱ from the herbivorous marine mollusk Elysia rufescens (and also its dietary alga Bryopsis sp.). Significantly, both of these chemotypes are continuing subjects of clinical evaluation.^iv Two closely related PKS/NRPS cyclidepspeptides — jasplakinolide from sponges^v and chondramide C from myxobacteria^vi — are of much interest because they are both potent in causing the hperassembly of G-actin into F-actin.^vii Our attempts to date to discover marine-derived fungi as a source of unusual peptides have been somewhat successful. These include the isolation of the cytotoxic bicyclic peptide malformin C from Aspergillus niger,^viii and highly N-methylated linear peptides of the RHM family from an unidentified fungal strain.^ix Recently we began a project stimulated by the observation that extracts of an unidentified fungal strain separated from an Inathella sponge possessed potent cytotoxicity and selectivity in our disk diffusion assay system.^x Most importantly its ¹³C NMR spectra displayed clusters of peaks centered at δ175. The initial bioassay-guided dereplication efforts showed that brefeldin A^xi was a non-peptide major component responsible for the cytotoxicity. Subsequently, deeper evaluation of crude extract fractions employing bioassay and LCMS data led to discovery of three novel cyclic depsipeptides named guangomides A (1), B (2) and homodestcardin (3). We now describe the structure elucidation and the biological activity of these depsipeptides.

The molecular formula of C₃₁H₄₆N₄O₉ (m/z = 641.3152 [M+Na]⁺) was established for guangomide A (1) by the HRESIMS data. The low field ¹³C NMR resonances noted above were confirmed to be associated with peptidic functionality that was further validated by ¹H NMR data showing the presence of two amide protons, six α-protons, and two N-methyl groups. There were a total of six carbonyl carbons in the ¹³C NMR spectrum (Table 1). Four of those carbonyl carbons were assigned as amide carbons and consistent with the four nitrogen atoms in the molecular formula. Therefore, the assumption that two ester carbons were present was verified by the HMBC correlations from two α-oxy methines (δ_H 5.00, δ_C 70.5 and δ_H 5.22, δ_C 76.5) to these carbonyl carbons (δ_C 172.6, 169.0).

Table 1. ¹H and ¹³C NMR data for guangomide A (1) and B (2) in CDCl₃^a.

Structural unit		1						2
		13C (δ, mult)		1H (δ, mult., J, Hz)	gCOSY	gHMBC	NOESY	13C (δ, mult)		1H (δ, mult., J, Hz)	gHMBC
S -Hic	1	172.4 (s)						172.4 (s)
	2	70.5 (d)		5.00 (dd, 8.7, 6.2)	3a, 3b	1, 3, 4, 8	3a, 3b, 4, 5, 6, 37	70.3 (d)		5.00 (dd,8.9, 6.4)	1, 3, 4, 8
	3	38.8 (t)	a	0.98 (ddd, 14.2, 6.2, 5.3)	2, 3b	1, 4, 5, 6	2, 3b, 4, 5, 6, 37	38.8 (t)	a	0.96 (ddd, 14.2, 6.4, 5.3)	2, 4, 5, 6
			b	1.55 (ddd, 14.2, 8.7, 6.7)	2, 3a, 4	1, 2, 4, 5, 6	2, 3a, 4, 5, 6		b	1.54 m	4, 5, 6
	4	23.8 (d)		1.15 m	3a, 5, 6	3, 5, 6	2, 3a, 3b, 5, 6, 31, 32, 34, 35, 37	23.8 (d)		1.19 (hept, 6.7)
	5	22.8 (q)		0.78 (d, 6.6)	4	3, 4, 6	2, 3a, 3b, 4, 31, 32, 34, 35, 37	22.9 (q)		0.78 (d, 6.6)	3, 4, 6
	6	22.3 (q)		0.72 (d, 6.6)	4	4, 5, 6	2, 3a, 3b, 4, 31, 32, 34, 35, 37	22.2 (q)		0.72 (d, 6.6)	4, 5, 6
l-Ala	8	172.6 (s)						172.9 (s)
	9	47.1 (d)		4.97 (dq, 8.7, 7.1)	11	8, 10, 12	10, 11	47.0 (d)		4.98 (dq, 8.9, 7.4)	8, 10
	10	19.3 (q)		1.53 (d, 7.1)		8, 9	9, 11	19.3 (q)		1.52 (d, 7.4)	8, 9
	11			8.22 (d, 8.7)	9	12	9, 10, 13, 15			7.89 (d, 8.5)
S -Dhiv	12	169.8 (s)						168.5 (s)
(S -Hiv)	13	76.5 (d)		5.22 s		12, 14, 15, 16, 18	11, 15, 16	78.0 (d)		5.23 (d, 2.5)	12, 14, 15, 16
	14	71.8 (s)						30.0 (d)		2.61 (hept.d, 6.7, 2.5)
	15	26.7 (q)		1.17 s		12, 13, 14, 16	11, 13, 16, OH	19.2 (q)		0.93 (d, 6.7)	13, 14, 16
	16	24.1 (q)		1.26 s		13, 14, 15	13, 15, OH	15.8 (q)		0.92 (d, 6.7)	13, 14, 15
	OH			5.27 br.s			15, 16
l-N-MeAla	18	169.0 (s)						169.7 (s)
	19	60.5 (d)		3.69 (q, 6.8)	20	18, 20, 22, 23	20, 22	60.7 (d)		3.69 (q, 6.8)	18, 20, 22, 23
	20	13.5 (q)		1.52 (d, 6.8)	19	18, 19	19, 22	13.6 (q)		1.54 (d, 6.9)	18, 19
	22	36.9 (q)		3.20 s		19, 23	19, 20, 24, 25	36.9 (q)		3.19 s	19, 23
d-Ala	23	171.3 (s)						171.1 (s)
	24	46.2 (d)		4.85 (quint, 7.1)	25, 26	23, 25, 27	22, 25, 26	46.3 (d)		4.84 (quint, 7.1)	23
	25	18.1 (q)		1.41 (d, 6.8)	24	23, 24	22, 26, 37	18.1 (q)		1.42 (d, 6.7)	23, 24
	26			7.09 (d, 7.3)	24	27	24, 25, 28, 29a			7.12 (d, 7.4)
d-N-Me-Phe	27	168.5 (s)						168.5 (s)
	28	56.6 (d)		5.76 (dd, 11.9, 5.3)	29a, 29b	1, 27, 29, 37	26, 29a, 29b, 31, 35	56.5 (d)		5.76 (dd, 11.8, 5.3)	1, 27, 29, 37
	29	33.2 (t)	a	2.92 (dd, 15.3, 11.8)	28, 29b	27, 28, 30, 31, 35	26, 29b, 28, 31, 35	33.2 (t)	a	2.92 (dd, 15.1, 11.8)	27, 30, 31, 35
			b	3.49 (dd, 15.3, 5.3)	28, 29a	30, 30, 31, 35	28, 29a, 31, 35		b	3.47 (dd, 15.1, 5.3)	27, 30, 31
	30	137.1 (s)						137.1 (s)
	31,35	128.6 (d)		7.16 m	32, 34	29, 30, 32, 33, 34	5, 6, 28, 29a, 29b	128.6 (d)		7.16 m	29, 30, 32, 33, 34
	32,34	128.4 (d)		7.24 m	31, 33, 35	30, 31, 33, 35	5, 6	128.4 (d)		7.24 m	30, 31, 33, 35
	33	126.6 (d)		7.19 m	32, 34	31, 32, 34, 35		126.6 (d)		7.19 m	31, 32, 34, 35
	37	30.2 (q)		2.92 s		1, 28	2, 3a, 4, 5, 6, 25	30.1 (q)		2.92 s	1, 28

^aMeasured at 500 MHz (¹H) and 125 MHz (¹³C).

There were six carbonyl containing substructures envisioned based on the 2D NMR data and they are shown in Figure 1. Further, as expected, the gCOSY spectrum of 1 revealed six distinct spin systems. The first included the subset associated with the 2-hydroxyisocaproic acid (A). Another cluster included the three sets of overlapping resonances for the alanine residues (B, D and E). The protons associated with the phenylalanine residue (F) were identified by HMBC correlations (H28/C30 and H29/C31, C35) shown in Figure 1. Finally, the remaining subunit, 2,3-dihydroxyisovaleric acid (C), was established by the HMBC correlations (H13/C14, C15, C16, H15, H16/C13) also shown in Figure 1. Next, the locations of the N-methyl groups (δ_H 3.21, δ_C 37.1 and δ_H 2.94, δ_C 30.3) were affirmed as connected to the alanine (D) and the phenylalanine (F) based on the HMBC correlations (H22/C19, H37/C28). The final task of sequencing the six subunits was accomplished based on the HMBC correlations from the α-protons, the amide protons and the N-methyl groups to the carbonyl carbons (H2/C8, H9, NH11/C12, H13/C18, H19, H22/C23, H24, NH26/C27, H28, H37/C1) as detailed in Figure 2.

Figure 1. Structural units for 1.

Figure 2. Selected gHMBC correlations for 1.

Assignment of the absolute configuration for the amino acid residues was accomplished using complimentary approaches. A d-N-methyl-phenylalanine was assigned based on results derived from HPLC-analysis of the products obtained from the Marfey's acid hydrolysate.^xii Though our attempts to determine the relative configurations of chiral centers using NOESY data were unsuccessful, positive results were obtained through X-ray analysis structure of 1 shown in Figure 3. ^xiii Combining the relative configurations deduced from the X-ray data and using the d-N-methyl-phenylalanine as an anchor point supported the final absolute stereostructure as 2S, 9S, 13S, 19S, 24R, 28R.

Figure 3. Absolute structure of 1 including X-ray crystal structure.

The next compound to be analyzed was guangomide B (2), whose molecular formula of C₃₁H₄₆N₄O₈ differed from that of 1 by just a single oxygen atom. Not surprisingly, its ¹³C and ¹H NMR spectra (Table 1) were almost identical with those of 1 The differences included shifted ¹³C resonances for the C13/C14/C15/C16 of 2 versus 1 consistent with the proposal that an H group replaced the OH group at C14 of substructure C (Figure 1). Likewise the ¹H NMR spectrum of 1 showed diastereotopic H₃15 and H₃16 as doublets. Thus, these data were in firm support of substructure D as 2-hydroxyisovaleric acid, which was further confirmed by the COSY and the HMBC measurements. The stereochemistry assignments shown for 2 are based on a biogenetic analogy to 1 plus the observations that both compounds have the same sign of the optical rotation and both have parallel NMR shifts at each of the chiral centers.

The last new compound, homodestcardin (3) had the molecular formula C₃₂H₅₅N₅O₇ established by the HRESIMS data. This compound was concluded to be a member of the destruxin family based on the similar profile of ¹H and ¹³C (Table 2) to that of homodestruxin B (4)^xiv and roseocardin (5).^xv Detail analysis of the 2D NMR data of 3 pinpointed the β-methyl proline and N-methyl leucine residues versus the proline residue on 4 and N-methyl valine residue on 5, respectively. Our proposed planar structure derived by analogy to 4 and 5 was confirmed from the 2D NMR data (Figure 4). Extending the comparisons of the NMR data between this trio further revealed that 3 and 4 possessed identical shifts at the western chiral centers (C6, C11, C12, C19 and C20). Alternatively 3 and 5 possessed parallel shifts at the eastern chiral centers (C26, C27 and C33) were almost identical with those of 5. The significant NOE observed between H26 and H28 was also consistent with these stereochemical conjectures. The final relative stereoassignments proposed for 3 are shown in Figure 4.

Table 2. ¹H and ¹³C NMR data for homodestcardin (3) in CDCl₃^a.

Structural unit		3
		13C (δ, mult)		1H (δ, mult., J, Hz)	gCOSY	gHMBC	NOESY
β-Ala	1	173.8 (s)
	2	34.5 (t)	a	2.56 (ddd, 18.5, 5.0, 1.8)	2b, 3b	1, 3	2b, 3a
			b	2.68 (ddd, 18.5, 11.5, 2.1)	2a, 3a		2a, 3a, 4
	3	33.2 (t)		3.08 (m)	2b, 3b, 4		2a, 2b, 3b, 4
				4.05 (m)	2a, 3a, 4		3a
	4			8.27 (d, 8.5)	3a, 3b		2b, 3a, 6, 20, 22b, 24
N-Me-Ala	5	169.7 (s)
	6	55.5 (d)		5.16 (q, 6.8)	7	5, 8	4, 7, 11
	7	15.3 (q)		1.30 (d, 6.9)	6	5, 6	6, 9
	9	28.1 (q)		2.73 (s)		6, 10	7
N-Me-Ile	10	171.1 (s)b
	11	56.8 (d)		5.03 (d, 10.9)	12	9, 10, 12, 14, 18	6, 12, 14a, 14b, 15
	12	33.5 (d)		2.06 (m)	11, 13		11, 13, 14b, 15, 17
	13	16.2 (q)		0.85 (m)c	11	11, 12, 14	12
	14	25.8 (t)	a	0.98 (m)	14b, 15		11
			b	1.43 (m)	14a, 15		11, 12, 14b, 17
	15	11.1 (q)		0.91 (t, 7.2)	14a, 14b	12, 14	11, 12, 14a
	17	31.0 (q)		3.20 (s)		11, 18	12, 14b, 19, 20
Ile	18	173.6 (s)
	19	53.5 (d)		4.83 (dd, 9.2, 6.8)	20, 24	18, 20	17, 20, 21, 22a, 22b, 24
	20	37.5 (d)		1.92 (m)	19, 21		4, 17, 19, 21, 22b, 24
	21	15.3 (q)		0.84 (m)c	20	19, 20	19, 20
	22	24.6 (t)	a	1.29 (m)	22b, 23		19, 23, 24
			b	1.44 (m)	22a, 23		4, 19, 20, 22b, 23, 24
	23	11.4 (q)		0.85 (m)c	22a, 22b	19, 20	22a, 22b
	24			7.07 (d, 8.8)	19		4, 19, 20, 22a, 22b, 26, 30b
3-Me-Pro	25	171.0 (s)b
	26	67.1 (d)		4.27 (d, 0.3)	27	25, 27, 28, 29	24, 27, 28
	27	36.2 (d)		2.78 (m)	26, 28, 29b		26, 28, 29a, 29b
	28	19.0 (q)		1.11 (d, 7.1)	27	26, 27, 29	26, 27, 29a, 30a
	29	30.9 (t)	a	1.70 (m)	29b, 30a, 30b		28, 29b, 30a, 27
			b	2.09 (m)	27, 29a, 30a, 30b		27, 29a, 30b
	30	44.9 (t)	a	3.56 (dt, 9.5, 7.4)	29a, 29b, 30b		28, 29a, 30b, 33, 34a
			b	3.86 (td, 9.5, 3.0)	29a, 29b, 30a		24, 29b, 30a, 33
Hica	32	169.9 (s)
	33	71.9 (d)		4.90 (dd, 10.4, 3.3)	34a, 34b		30a, 30b, 34a, 34b, 37
	34	39.3 (t)	a	1.37 (ddd, 14.5, 9.0, 3.4)	34b, 35	32, 35, 36	30a, 33, 34b, 35
			b	1.97 (m)	34a	32, 35, 36	33, 34a, 35, 36
	35	24.4 (d)		1.85 (m)	34a, 36, 37		34a, 34b, 36, 37
	36	23.4 (q)		1.00 (d, 6.6)	35	35, 37	34b, 35
	37	21.7 (q)		0.96 (d, 6.6)	35	35, 36	33, 35

^aMeasured at 500 MHz (¹H) and 125 MHz (¹³C).

^bAssignment may be interchangeable.

^cCoupling constant could not be measured due to signal overlap.

Figure 4. Selected 2D NMR correlations and the relative structure of 3.

The literature, aside from that noted above of cyclic depsipetides, including those with significant bioactivity properties isolated from marine-derived fungi is rather sparse. One set of compounds includes 15G256γ, 156G256δ and 15G256ε which are lipodepsipeptides isolated from a Hypoxylon oceanium separated from mangrove wood.^xvi Exumolides A and B isolated from a marine plant-derived Scytaladium sp. are cytostatic cyclic hexadepsipetides.^xvii Sansalvamide^xviii and N-methylsansalvamide^xix isolated from two different marine alga-derived Fusarium spp. are cytotoxic cyclic pentadepsipeptides. Those previous examples contain one ester group in the molecule. By contrast, the guangomides (1 and 2) isolated in this study are different from known cyclic depsipetides from marine-derived fungi because of the presence of two ester groups in the molecule. In addition, homodestcardin (3) is the first example of destruxin analog from marine-derived fungi. The biological properties of 1 and 2 were investigated and deserve brief comment. Guangomide A (1) was inactive against murine and human tumor cell lines in a disk diffusion assay. However, 1 and 2 showed weak antibacterial activity against Staphylococcus epidermidis (MIC=100μg/mL, each) and Enterococcus durans (MIC=100μg/mL, each).

Experimental Section

General Experimental Procedures

UV/vis measurements were recorded on HP 8453 diode array spectrometer. Optical rotations were obtained on a JASCO DIP-370 digital polarimeter. The NMR spectra were recorded on a Varian UNITY INOVA-500 spectrometer, operating at 500 and 125.7 MHz for ¹H and ¹³C, respectively. High resolution mass measurements were obtained on a bench-top Mariner ESI-TOF mass spectrometer. HPLC was performed with a column of 4μm ODS.

Biological Materials

The fungus (strain no. 001314c) was isolated from an Inathella sp. (coll. no. 00314) collected using SCUBA off the coast of Guango, Papua New Guinea on December 2000 by the same techniques described previously.^xx Attempts to identify this fungal strain by the alignment of the D2 region of the 25S ribosomal DNA sequence and its fruiting body were unsuccessful (the closest fungus: Fusarium graminearum with a genetic distance of 9.75%). This fungus is maintained in a cryopreserved state at UCSC.

Culture conditions

The fungal strain was grown in a liquid medium (20 L) containing 1.5 % malt extract broth in filtered Monterey Bay seawater adjusted to pH 7.4 at 180 rpm for 28 days at room temperature (25°C).

Disk Diffusion Soft Agar Colony Formation Assay

An in vitro cell-based assay was employed to identify solid tumor selectivity for original extracts, extract partition fractions, and pure compounds. The differential cytotoxicity is expressed by observing a zone differential between any solid tumor cell (Colon 38, Colon H116, Lung H125) and either leukemia cells (L1210 or CEM) or normal cells (CFU-GM). The sample is designated as “solid tumor selective” if the zone units of solid tumor – normal cell or leukemia cells is greater than 250 units.

Extraction and Isolation

The culture was filtered under suction and the broth was extracted with equal volumes of EtOAc three times. The EtOAc extract was partitioned by Kupchan type extraction reported previously.^xxi The CH₂Cl₂ extract (EFD; 2.6 g) contained crystals, which were purified by washing away with MeOH and were identified brefeldin A (560 mg) by comparison of its spectral data to the published data.⁸ The CH₂Cl₂ soluble portion was separated with a flash silica gel chromatography using CH₂Cl₂-MeOH stepwise gradient as the eluent. B3 (170 mg) obtained from MeOH-CH₂Cl₂ (1:99) was purified by a reverse phase HPLC with MeOH-H₂O (3:7 up to 1:0) to furnish 2 (1.5 mg). B4 (190 mg) obtained from MeOH-CH₂Cl₂ (1:49) was purified by a reverse phase HPLC with MeOH-H₂O (2:3 up to 9:1) to furnish 1 (11.0 mg) and 3 (1.2 mg).

Guangomide A (1)

Colorless crystals from Hexane/EtOAc/MeOH (1/1/1). Mp. 255-257°C. [α]²⁸_D −44.6 (c 0.8, CHCl₃); λ_max (MeOH)/nm 203 (log ε 4.38); ¹H and ¹³C NMR data are listed in Table 1; HRESIMS: m/z 641.3152 [M+Na]⁺ (calcd for C₃₁H₄₆N₄O₉Na, 641.3157).

Guangomide B (2)

White amorphous powder. [α]²⁸_D −18.1 (c 0.9, CHCl₃); λ_max (MeOH)/nm 204 (log ε 4.38); ¹H and ¹³C NMR data are listed in Table 1; HRESIMS: m/z 625.3208 [M+Na]⁺ (calcd for C₃₁H₄₆N₄O₈Na, 625.3208).

X-ray crystallography of 1

The single crystal X-ray analysis was conducted as follows. Suitable crystals were obtained from Hexane/EtOAc/MeOH (1/1/1) by the vapor diffusion method. This crystal (0.60 × 0.40 × 0.30 mm³) was mounted on a Bruker SMART diffractometer (MoKα; -100°C). A hemisphere of data was taken using a narrow-scan routine (1406 frames, 0.3° steps ω-scan, exposure time was 30 s/frame, 2θ_max = 63.62°). Raw data was integrated with the Bruker SAINT+ program^xxii to yield a total 33908 reflections, of which 10117 were independent (R_int = 2.58 %, completeness 94.9 %) and 8852 with I > 2σ(I). Data were collected for absorption using SADABS program (min. and max. transmission are 0.9479 and 0.9735, respectively).^xxiii The structure was solved by direct methods and refined by full matrix least square on F² techniques using anisotropic displacement parameters for all non-hydrogen atoms.^xxiv All hydrogen atoms were found in the difference Fourier map and refined isotropically. At final convergence, R₁ = 3.88 % and GOF = 1.022 for 581 parameters.

Amino acid analysis of 1 using Marfey's method

Guangomide A (1) (1.4 mg) and 6N HCl (2 mL) were added to MeOH (0.2mL) and heated at 110°C for 14hr in a sealed vial. The cooled reaction mixture was evaporated to dryness. To the residue was added 1M NaHCO₃ (0.1 mL) and 1% Marfey's reagent (FDAA) in acetone (0.1 mL), incubated at 37°C for 0.5 hr. The reaction mixture was quenched with 2N HCl (50 μL), analyzed by reverse phase HPLC. The analysis was performed by the following condition: Alltech altima C18 column (5μ, 250mm × 10mm id.), solvent system acetonitrile/H₂O (4:1) up to (1:1) over 60 min with 1 mL/min flow rate, UV detection at 340 nm. Separately N-methyl-d and l-phenylalanine were derivatized with FDAA in the same manner as that of 1. The HPLC conditions gave the distinguishable retention times for d and l forms (51.0 min and 50.3 min, respectively). The configuration of the N-methylphenylalanine was determined to be d-form based on co-injection of each form of the standard amino acids and 1.

Homodestcardin (3)

White amorphous powder. [α]²⁷_D −143.8 (c 0.9, CHCl₃); λ_max (MeOH)/nm 204 (log ε 3.75); ¹H and ¹³C NMR data are listed in Table 2; HRESIMS: m/z 622.4170 [M+H]⁺ (calcd for C₃₂H₅₆N₅O₇, 622.4174).