Corallomycetellains A–J, Cytotoxic Epipolythiodioxopiperazine Alkaloids Isolated from the Fungi Corallomycetella repens HDN23-0007

Abstract

Ten new epipolythiodioxopiperazine (ETP) alkaloids, named corallomycetellains A–J (1–10), along with one known analogue, haematocin (11), were isolated from the fungi Corallomycetella repens HDN23-0007. Their structures, including absolute configurations, were established by comprehensive spectroscopic data and electronic circular dichroism (ECD) calculations. Compounds 1–2 represent the first two examples of aranotin-type ETPs possessing an aromatic indole moiety. Compounds 2–4 all featured a unique C2-methyl disulfide substituent, whereas compound 4 additionally possessed a C2′-oxomethyl group. In in vitro cytotoxicity assays, compounds 7–10, which contained α–α′ polysulfide bridges, exhibited strong anticancer activity, with IC₅₀ values ranging from 1.1 to 9.3 μM.

1. Introduction

Epipolythiodioxopiperazines (ETPs) represent a class of fungal secondary metabolites characterized by the presence of sulfur atoms within their diketopiperazine scaffold. The structural diversity of epipolythiodiketopiperazine (ETP) derivatives originates from the variability of diketopiperazine scaffolds, distinct patterns of sulfur atom linkage to the 2,5-diketopiperazine core, and extensive post-biosynthetic modifications. These compounds exhibit a wide range of biological activities, including antiviral (e.g., acetylaranotin) [1], antitumor (e.g., gliotoxin and chaetocin) [2,3,4,5], immunomodulatory (e.g., chetoseminudin A) [6], and antibacterial activities (e.g., sirodesmin PL) [7]. Critically, the biological activities of ETPs are primarily determined by the presence and spatial configuration of their characteristic disulfide bridges [8].

During our ongoing search for novel bioactive metabolites from marine-derived fungi [9,10], microbial fermentation extracts were initially analyzed by high-performance liquid chromatography–ultraviolet (HPLC–UV) profile. Among these, Corallomycetella repens HDN23-0007 was prioritized for further investigation due to its notably complex HPLC–UV profile, which displayed a high density of well-resolved, UV-absorbing peaks across a broad retention time range. A comprehensive analysis combining HPLC–UV profiling, liquid chromatography–mass spectrometry (LC–MS), and the Global Natural Products Social Molecular Networking (GNPS) platform (Figure 1) revealed the presence of potential epidithiodioxopiperazine (ETP) derivatives in its metabolome [11,12]. Subsequently, ten new epipolythiodioxopiperazine (ETP) alkaloids, named corallomycetellains A–J (1–10), along with one known analogue, haematocin (11), were isolated (Figure 2). Compounds 7–10 exhibited strong anticancer activity, with IC₅₀ values ranging from 1.1 to 9.3 μM. In this paper, we describe their isolation, structural elucidation, and biological evaluation.

2. Results and Discussion

2.1. Structural Determination

Corallomycetellain A (1) was obtained as a light-yellow amorphous powder. Its molecular formula was determined to be C₂₁H₁₈N₂O₄S, based on the molecular ion at m/z 417.0880 [M + Na]⁺ observed in HRESIMS, indicating 14 degrees of unsaturation. The 1D NMR (

Table 1. ¹H (500 MHz) and ¹³C (125 MHz) NMR data of 1–2 in CDCl₃.

No.	1		2
	δC, Type	δH (J in Hz)	δC, Type	δH (J in Hz)
1	163.7, C		163.6, C
2	76.4, C		78.9, C
3	40.6, CH2	3.10, d (15.8)	39.3, CH2	3.29, s
		3.22, d (15.0)
4	133.9, C		132.9, C
5	120.6, CH	6.03, m	121.0, CH	6.03, m
6	125.3, CH	6.03, m	125.0, CH	6.01, dd (9.7, 3.3)
7	128.5, CH	5.65, d (8.7)	128.9, CH	5.64, d (9.5)
8	74.9, CH	6.31, d (14.0)	74.6, CH	6.17, d (14.0)
9	64.7, CH	5.27, d (14.0)	64.7, CH	5.27, d (14.0)
11	171.4, C		171.0, C
12	21.6, CH3	2.23, s	21.5, CH3	2.24, s
R-2	13.1, CH3	2.09, s	23.5, CH3	2.07, s
1′	155.7, C		155.8, C
2′	128.9, C		129.6, C
3′	115.0, CH	7.45, d (0.8)	114.7, CH	7.46, d (0.6)
4′	129.3, C		129.3, C
5′	122.8, CH	7.70, dt (7.9, 1.0)	122.9, CH	7.71, d (7.8)
6′	125.6, CH	7.41, td (7.7, 1.1)	125.6, CH	7.41, td (8.1, 1.1)
7′	128.2, CH	7.54, ddd (8.3, 7.2, 1.2)	128.2, CH	7.53, ddd (8.4, 7.2, 1.2)
8′	116.4, CH	8.44, dt (8.6, 1.0)	116.1, CH	8.38, m
9′	134.7, C		135.1, C

) and HSQC spectra of 1 revealed the presence of one methylthio group (δ_C/H 13.1/2.09), one acetyl methyl group (δ_C/H 21.6/2.23), one methylene (δ_C/H 40.6/3.10), ten methines (one oxygenated (δ_C/H 74.9/6.31), one nitrogenated (δ_C/H 64.7/5.27), and eight olefinic ones (δ_C/H 115.0/7.45, δ_C/H 116.4/8.44, δ_C/H 120.6/6.03, δ_C/H 122.8/7.70, δ_C/H 125.3/6.03, δ_C/H 125.6/7.41, δ_C/H 128.2/7.54, δ_C/H 128.5/5.65)), one quaternary sp³ carbon (δ_C 76.4), four olefinic nonprotonated carbons (δ_C 128.9, 129.3, 133.9, 134.7), one ketone carbonyl (δ_C 171.4) and two amide carbonyls (δ_C 163.7, 155.7). Nine of the compound’s fourteen degrees of unsaturation were accounted for by the structural features identified so far, suggesting that compound 1 possessed a penta-cyclic core.

A series of the COSY correlations from H-6 (δ_H 6.03)/H-7 (δ_H 5.65) and H-8 (δ_H 6.31)/H-9 (δ_H 5.27) as well as the HMBC correlations from H-7 to C-5 (δ_C 120.6)/C-9 (δ_C 64.7), from H-9 to C-4 (δ_C 133.9) and from H₂-3 to C-2 (δ_C 76.4)/C-4/C-5/C-9 indicated the presence of six membered rings combined with five membered rings in 1. (Figure 3). While the right series of the COSY correlations from H-5′ (δ_H 7.70)/H-6′ (δ_H 7.41) and H-7′ (δ_H 7.54)/H-8′ (δ_H 8.44) as well as the HMBC correlations from H-5′ to C-3′ (δ_C 115.0)/C-9′ (δ_C 134.7)/C-7′ (δ_C 128.2), from H-8′ to C-4′ (δ_C 129.3)/C-6′ (125.6) and from H-3′ (δ_H 7.45) to C-2′ (δ_C 128.9)/C-9′ revealed an extra six membered rings combined with five membered rings (Figure 3). Moreover, the compound was deduced to possess a 6-5-6-5-6 fused diketopiperazine (DKP) skeleton, based on the characteristic chemical shifts of the amide carbonyls (δ_C 163.7, C-1; δ_C 155.7, C-1′) and the molecular weight. Analysis of the NMR data (Table 1) for the left C-11 (δ_C 171.4) and C-12 (δ_C 21.6) of 1 and HMBC correlation from H-12 (δ_H 2.23) to C-11 indicated the presence of an acetyl group. Based on all relevant signals and chemical shifts, we hypothesized that the acetyl group can only form an ester bond with the oxygen at the C-8 (δ_C 74.9) position. Therefore, the planer structure of 1 with the presence of a 6/5/6/5/6 diketopiperazine skeleton was suggested, as shown in Figure 2.

The relative configurations of 1 were established as 2R*,8S*,9S* by the coupling constants analysis of H-8/H-9 (³J_{H-8, H-9} = 14.0 Hz) and the ROESY correlation between SMe-2 (δ_H 2.09) and H-8 (δ_H 6.31). To determine the absolute configurations of 1, ECD calculations were performed at the B3LYP/6-31+G(d) level. The tendencies of the experimental ECD spectrum were in reasonable agreement with the calculated one of (2R*,8S*,9S*)-1 (Figure 4), establishing the absolute configuration as 2R, 8S, 9S in 1.

The molecular formula of corallomycetellain B (2), obtained as a light-yellow amorphous powder, was C₂₁H₁₈N₂O₄S₂, deduced by HRESIMS. Comparison of ¹H and ¹³C NMR spectra (Table 1) of 1 and 2 revealed that they shared the same 6/5/6/5/6 diketopiperazine skeleton with an aromatic indole moiety. The only difference was an existence of a disulfide bond in C-2 (δ_C 78.9) in 2 supported by the HRESIMS results and chemical shift differences of Me-R-2 in 1 and 2 (δ_C 13.1 in 1 while δ_C was 23.5 in 2).

The relative configurations of 2 were established in the same way as 1 by the coupling constants analysis of H-8/9 (³J_{H-8, H-9} = 14.0 Hz) and the comparison of 1D NMR data with 1 (Table 1). Consequently, the absolute configuration of 2 was determined as 2R, 8S, 9S, since the experimental ECD curves of 1 and 2 agreed perfectly (Figure 5). Reported ETPs typically feature a sulfur-bridged diketopiperazine core, frequently integrated with fused oxazine rings. In comparison, compounds 1–2 represent the first two examples of aranotin-type ETPs possessing a rare aromatic indole unit (Figure S2).

Corallomycetellain C (3) and corallomycetellain D (4) were obtained as a light-yellow amorphous powders with the molecular formulae of C₂₄H₂₆N₂O₆S₃ and C₂₄H₂₆N₂O₇S₂, deduced by HRESIMS, respectively. Analysis of the ¹H and ¹³C NMR data of 3 with those of haematocin (11) [13] indicated that 3 has the same skeleton as 11 (

Table 2. ¹H and ¹³C NMR data of 3–4 in CDCl₃.

No.	3 a		4 b
	δC, Type	δH (J in Hz)	δC, Type	δH (J in Hz)
1	165.8, C		166.6, C
2	77.4, C		76.9, C
3	38.3, CH2	2.89, m	38.5, CH2	2.89, m;
		3.69, d (15.9)		3.67, d (16.4)
4	133.7, C		133.2, C
5	120.2, CH	5.94, m	120.4, CH	5.93, m
6	125.1, CH	5.95, m	125.2, CH	5.95, m
7	128.2, CH	5.58, t (6.4)	127.7, CH	5.57, m
8	75.2, CH	6.13, d (14.5)	74.9, CH	6.17, dd (14.6, 2.2)
9	64.6, CH	5.19, d (14.2)	65.1, CH	5.16, d (13.6)
11	170.3, C		170.9, C
12	21.4, CH3	2.12, s	21.4, CH3	2.13, s
Me-R1-2	23.8, CH3	2.47, s	23.9, CH3	2.51, s
1′	164.9, C		163.7, C
2′	74.2, C		95.4, C
3′	39.7, CH2	2.89, m;	41.1, CH2	2.61, m;
		3.00, d (16.0)		3.04, d (15.3)
4′	133.5, C		134.0, C
5′	120.2, CH	5.94, m	119.3, CH	5.91, m
6′	124.8, CH	5.95, m	125.2, CH	5.95, m
7′	128.1, CH	5.58, t (6.4)	128.1, CH	5.57, m
8′	76.0, CH	6.03, d (13.1)	75.3, CH	5.86, d (13.9)
9′	63.8, CH	5.19, d (14.2)	63.4, CH	5.18, d (13.6)
11′	170.5, C		170.5, C
12′	21.4, CH3	2.09, s	21.4, CH3	2.10, s
Me-R2-2′	14.4, CH3	2.23, s	53.3, CH3	3.50, s

^a Data recorded at ¹H (500 MHz) and ¹³C (125 MHz) NMR. ^b Data recorded at ¹H (600 MHz) and ¹³C (150 MHz) NMR.

). The difference in mass of 32 Da between 3 and 11 indicated an addition of a sulfur atom in 3, which is supported by the HRESIMS analysis. The HMBC correlation from S-SMe-2 (δ_H 23.8) to C-2 (δ_C 77.4), along with the differences in chemical shifts of C-2 and S-SMe-2 (δ_C 74.2 and δ_C 14.5 in 11 [13], compared to δ_C 77.4 and δ_C 23.8 in 3), revealed the presence of a disulfide bond at C-2 in 3 (Figure 3). Compound 4 had a methoxy group instead of a methylthio group in C-2′, which was supported by chemical shifts of OMe-2′ (δ_C 53.3 and δ_H 3.50) and C-2′ (δ_C 95.4) as well as an HMBC correlation from OMe-2′ to C-2′ (Figure 3).

The relative configurations of 3–4 were established by ¹H-¹H coupling constants and ROESY data, and the comparison of 1D NMR data with haematocin (11). The coupling constants between H-8/8′ and H-9/9′ were more than 14.0 Hz, which revealed trans relationships of H-8/H-9 and H-8′/H-9′. The ROESY correlations from H-3a/3′a (δ_H 2.89) to H-9/9′ (δ_H 5.19), from H-3b (δ_H 3.69) to S-SMe-2 (δ_H 2.47) and from H-3′b (δ_H 3.00) to SMe-2′ (δ_H 2.23) indicated that S-SMe-2/SMe-2′ and H-9/9′ were located on the opposite side of the pyrrole ring (Figure 6). Therefore, the relative configuration of 3 was suggested. According to the ROESY correlations from H-3a (δ_H 2.89) to H-9 (δ_H 5.16), from H-3b (δ_H 3.67) to S-SMe-2 (δ_H 2.51), from H-8 (δ_H 5.86) to OMe-2′ (δ_H 3.50) (Figure 6) and the coupling constants analysis of H-8′/9′ (³J_{H-8′, H-9′} = 14.0 Hz), the relative configurations of 4 were established to be the same as 3. The experimental ECD curves for 3–4 and haematocin (11) agreed perfectly (Figure 5), which demonstrated that their absolute configurations were 2R,2′R,8S,8′S,9S,9′S.

Corallomycetellain E (5), obtained as yellow oil, had a molecular formula of C₂₀H₂₄N₂O₆S₂ deduced by an HRESIMS peak at m/z 453.1139 [M + H]⁺. Comparison of ¹H and ¹³C NMR data of 5 (

Table 3. ¹H (500 MHz) and ¹³C (125 MHz) NMR data of 5 in DMSO-d₆.

No.	δC, Type	δH (J in Hz)	No.	δC, Type	δH (J in Hz)
1	164.9, C		2′	75.1, C
2	73.0, C		3′	49.4, CH2	2.50, d (8.6)
3	38.8, CH2	2.88, d (16.1)			2.92, d (14.1)
		3.08, d (15.5)	4′	69.8, C
4	133.5, C		5′	65.5, CH	4.07, d (3.9)
5	118.9, CH	5.98, dt (5.5, 2.9)	6′	127.8, CH	5.74, d (1.7)
6	123.3, CH	5.90, ddd (9.8, 4.8, 2.7)	7′	132.0, CH	5.75, d (1.0)
7	130.5, CH	5.62, d (9.9)	8′	67.8, CH	4.40, d (9.6)
8	73.9, CH	4.63, d (13.4)	9′	68.2, CH	3.96, ddd (10.4, 6.9, 3.6)
9	67.6, CH	4.78, d (13.7)	SMe-2′	14.3, CH3	2.13, s
SMe-2	14.0, CH3	2.16, s	OH-8′		4.32, d (7.2)
OH-8		5.27, d (1.7)	OH-4′		5.32, s
1′	167.2, C		OH-5′		4.88, d (4.8)

) to those of rostratazine B [14] revealed that 5 shared the same skeleton with rostratazine B. The difference was the replacement of a hydroxyl group at C-4′ (δ_C 69.8) in 5, supported by the HMBC correlation from OH-4′ (δ_H 5.32) to C-9′ (δ_C 68.2) (Figure 3). Therefore, the planer structure of 5 was proposed.

The relative configuration of compound 5 was assigned based on NOESY data and coupling constant analysis. For the left moiety, the relative configuration of 2R*,8S*,9S* was determined from the following key observations: NOE correlations between H-3a (δ_H 2.88) and H-9 (δ_H 4.78), and between H-3b (δ_H 3.08) and SMe-2 (δ_H 2.16), together with a large coupling constant between H-8 and H-9 (³J_{H-8, H-9} = 13.0 Hz) (Figure 6). For the right moiety, a cis relationship between H-8′ and H-9′ was indicated by their smaller coupling constant (³J_{H-8′, H-9′} = 9.6 Hz). Furthermore, the relative configuration was confirmed by a network of NOE correlations: H-3′a (δ_H 2.92)/OH-4′ (δ_H 5.32), OH-4′ (δ_H 5.32)/OH-5′ (δ_H 4.88), OH-5′ (δ_H 4.88)/OH-8 (δ_H 5.27), H-3′b (δ_H 2.50)/H-9′ (δ_H 3.96), H-9′ (δ_H 3.96)/SMe-2′ (δ_H 2.13) (Figure 6). Therefore, the relative configurations of 5 were proposed as 2R*,2′R*,4′S*,5′R*,8S*,8′S*,9S*,9′S*. Consequently, the absolute configurations of 5 were determined as 2R,2′R,4′S,5′R,8S,8′S,9S,9′S by comparing the experimental electronic circular dichroism (ECD) curve with the calculated one (Figure 4).

Corallomycetellain F (6) was isolated as colorless oil with a molecular formula of C₂₂H₂₄N₂O₄S₂ deduced by a HRESIMS peak at m/z 467.1070 [M + Na]⁺. Comparison of the ¹H and ¹³C NMR data (

Table 4. ¹H (500 MHz) and ¹³C (125 MHz) NMR data of 6 in CDCl₃.

No.	δC, Type	δH (J in Hz)	No.	δC, Type	δH (J in Hz)
1	167.2, C		1′	164.5, C
2	68.4, C		2′	73.5, C
3	39.0, CH2	1.56, m	3′	46.6, CH2	2.95, d (14.3)
		2.58, d (15.8)			3.66, d (14.3)
4	133.6, C		4′	133.5, C
5	119.7, CH	5.79, q (3.5)	5′	130.8, CH	7.15, dd (7.7, 1.8)
6	125.3, CH	5.92, dt (8.5 3.7)	6′	128.7, CH	7.28, m
7	128.0, CH	5.57, dd (9.9, 1.9)	7′	128.0, CH	7.28, m
8	74.9, CH	6.22, d (14.3)	8′	130.8, CH	7.15, dd (7.7, 1.8)
9	65.4, CH	4.90, d (14.3)	9′	128.7, CH	7.28, m
11	171.1, C		SMe-2′	13.7, CH3	2.36, s
12	21.5, CH3	2.19, s	NH		6.70, s
SMe-2	14.6, CH3	2.26, s

) to those of phomazine B revealed that 6 shared the same skeleton with phomazine B [15]. The difference was the replacement of an acetoxyl group at C-6 (δ_C 74.9) in 6, supported by the HMBC correlation from H-12 (δ_H 2.19) to C-11 (δ_C 171.1) (Figure 3).

The relative configuration of 6 was established by ¹H-¹H coupling constants and ROESY data. The large J value (14.3 Hz) between H-9 (δ_H 4.90) and H-8 (δ_H 6.22) indicated the trans-orientations of H-8 and H-9, which was further supported by the observed ROE correlation between H-9 and H-12 (δ_H 2.19). In addition, ROESY correlations between H-9 and H-3a (δ_H 1.56), between H-9 and H-5′/8′ (δ_H 7.15) and between SMe-2 (δ_H 2.26) and H-3b (δ_H 2.58) (Figure 6) allowed the assignment of the two thiomethyl groups as cis to each other and both trans relative to H-9. Hence, the relative of 6 was assigned as 2R*,2′R*,8S*,9S*. To determine the absolute configuration of 6, ECD calculations were performed at the B3LYP/6-31+G(d) level. The overall pattern of the experimental ECD spectrum was in reasonable agreement with the calculated one of (2R,2′R,8S,9S)-6, indicating the absolute configuration of C-2, C-2′, C-8 and C-9 in 6 as 2R,2′R,8S,9S (Figure 4).

Corallomycetellains G–I (7–9), isolated as white powder, have the molecular formulae of C₂₂H₂₀N₂O₆S₂, C₂₄H₂₄N₂O₆S₂ and C₂₂H₂₀N₂O₆S₃, respectively, deduced by HRESIMS. The ¹H and ¹³C NMR spectra of compound 7 revealed 10 proton signals and 11 carbon signals, precisely half of the total counts indicated by their molecular formulae. This observation suggests that 7 possesses a symmetrical structure. Comparative analysis of ¹H and ¹³C NMR data (Table 2 and

Table 5. ¹H (600 MHz) and ¹³C (150 MHz) NMR data of 7 (CDCl₃) and 8 (CD₃OD).

No.	7		8
	δC, Type	δH (J in Hz)	δC, Type	δH (J in Hz)
1	162.8, C		162.9, C
2	78.2, C		78.7, C
3	36.4, CH2	2.88, d (17.8)	35.8, CH2	2.97, m
		3.70, d (17.8)		3.68, m
4	132.3, C		133.2, C
5	119.8, CH	5.98, m	119.1, CH	6.05, m
6	124.5, CH	5.98, m	124.6, CH	6.04, m
7	127.7, CH	5.55, dd (11.6, 4.4)	126.7, CH	5.52, m
8	74.1, CH	6.04, d (12.3)	74.1, CH	5.93, d (13.2)
9	64.3, CH	4.98, d (13.5)	64.1, CH	5.12, d (13.3)
11	170.6, C		171.0, C
12	21.3, CH3	2.15, s	20.0, CH3	2.08, s
1′	162.8, C		162.8, C
2′	78.2, C		78.7, C
3′	36.4, CH2	2.88, d (17.8)	35.8, CH2	2.97, m
		3.70, d (17.8)		3.68, m
4′	132.3, C		133.2, C
5′	119.8, CH	5.98, m	119.1, CH	6.05, m
6′	124.5, CH	5.98, m	124.6, CH	6.04, m
7′	127.7, CH	5.55, dd (11.6, 4.4)	126.8, CH	5.52, m
8′	74.1, CH	6.04, d (12.3)	74.3, CH	5.93, d (13.2)
9′	64.3, CH	4.98, d (13.5)	64.2, CH	5.12, d (13.3)
11′	170.6, C		173.3, C
12′	21.3, CH3	2.15, s	35.9, CH2	2.38, td (7.4 2.8)
13′			17.7, CH2	1.66, h (7.4)
14′			12.7, CH3	0.96, s

) of 7 and 3 confirmed that both share an identical 6/5/6/5/6 diketopiperazine core. The structural distinction was the presence of an additional disulfide bridge between C-2 and C-2′ in 7. The planar structure of 7 was further elucidated based on 2D NMR correlations. Key COSY correlations were observed between H-6 (δ_H 5.98) and H-7 (δ_H 5.55) and between H-8 (δ_H 6.04) and H-9 (δ_H 4.98). Additionally, HMBC correlations from H-3 to C-1 (δ_C 162.8), C-2 (δ_C 78.2), C-4 (δ_C 132.3), C-5 (δ_C 119.8) and C-9 (δ_C 64.3); from H-7 to C-5 and C-9; from H-6 to C-8 (δ_C 74.1); and from both H-8 and H-12 (δ_H 2.15) to C-11 (δ_C 170.6) collectively supported the proposed connectivity, as illustrated in Figure 3.

The key structural difference between 7 and 8 was the presence of a butanoyl group at C-8′ (δ_C 74.3) in 8 instead of an acetoxy group in 7, which was supported by the HMBC correlations from H-13′ (δ_H 1.66) to C-12′ (δ_C 35.9), C-11′ (δ_C 173.3) and C-14′ (δ_C 132.1) (Figure 3). In contrast to 7, compound 9 was proposed to feature a trisulfide bridge. This inference was supported by HRESIMS data and by the distinct chemical shifts of C-2 and C-2′ in 9 (δ_C 78.2 and δ_C 82.4, respectively) compared to those in 7 (both δ_C 78.2).

Corallomycetellain G (7) was first isolated in 2001 from the fermentation broth filtrate of Rhizostilbella sp. [16] However, the relative configurations were not confirmed. Here, we determined the relative configurations of 7 based on the analysis of the coupling constants between H-8/8′ and H-9/9′ (³J_{H-8/8′, H-9/9′} = 13.2 Hz) as well as the ROESY correlations between H-3a (δ_H 3.70) and H-9 (δ_H 4.98) and between H-9 (δ_H 4.98) and H-12 (δ_H 2.15) (Figure 6). Comparison of 1D and 2D NMR spectra between 7 and 8–9 indicated the same relative configurations of them, supported by the coupling constants and chemical shifts (Table 5 and

Table 6. ¹H (600 MHz) and ¹³C (150 MHz) NMR data of 9–10 (in CDCl₃).

No.	9		10
	δC, Type	δH (J in Hz)	δC, Type	δH (J in Hz)
1	163.9, C		161.0, C
2	78.2, C		78.1, C
3	40.0, CH2	2.86, d (16.8);	36.3, CH2	3.00, d (17.9)
		3.28, m		3.82, d (17.7)
4	131.2, C		132.1, CH
5	120.7, CH	5.93, dt (5.2, 2.3)	120.0, CH	6.04, m
6	124.3, CH	5.93, dt (5.2, 2.3)	124.4, CH	6.01, m
7	128.0, CH	5.56, m	127.8, CH	5.58, d (9.8)
8	75.0, CH	6.22, t (14.0, 13.0)	74.2, CH	6.14, d (12.9)
9	64.9, CH	5.32, d (13.3)	64.6, CH	5.05, d (13.0)
11	170.5, C		170.7, C
12	21.5, CH3	2.21, s	21.4, CH3	2.18, s
1′	164.9, C		162.2, C
2′	82.4, C		76.6, C
3′	41.8, CH2	2.92, dd (17.2)	36.1, CH2	3.26, d (18.4)
		3.33, m		4.26, d (18.5)
4′	131.9 C		138.2, C
5′	120.4, CH	5.93, dt (5.2, 2.3)	125.7, CH	7.20, dd (7.2, 6.7)
6′	124.5, CH	5.93, dt (5.2, 2.3)	125.3, CH	7.33, dd (2.8, 2.4)
7′	128.6, CH	5.56, m	128.7, CH	7.33, dd (2.8, 2.4)
8′	74.7, CH	6.19, t (14.0, 13.0)	115.7, CH	7.93, d (8.2)
9′	63.9, CH	5.08, d (14.8)	128.3, C
11′	170.5, C
12′	21.4, CH3	2.16, s

).

The comparison of the ECD curve of 7 [(218 (negative), 239 (negative), 276 (positive) and 359 nm (negative)] with those of emethallcin E [17] [(217 (negative), 237 (negative), 273 (positive), and 328 nm (negative)] confirmed that they had the same configuration around the epidithiodioxopiperazine ring. Therefore, the absolute configuration of 7 was determined as 2R,2′R,8S,8′S,9S,9′S. Based on the highly similar ECD spectra, the absolute configurations of 8 were proposed to be the same as those of 7 (Figure 5), which was further supported by computational calculations (Figure 4). The absolute configuration of 9 was determined as 2R,2′R,8S,8′S,9S,9′S by comparison of its experimental ECD spectrum with the calculated ones (Figure 4).

Corallomycetellain J (10) was obtained as a light-yellow amorphous powder with a molecular formula of C₂₀H₁₆N₂O₄S₂ deduced by an HRESIMS peak at m/z 435.0432 [M + Na]⁺. The ¹H and ¹³C NMR data of 10 revealed that 10 was similar to deoxyapoaranotin [18] (Table 6). The obvious difference was the presence of a 1,3-cyclohexadiene skeleton in ring A, supported by HMBC correlations from H-6 (δ_H 6.01) to C-4 (δ_C 132.1) and C-8 (δ_C 74.2) and from H-7 (δ_H 5.58) to C-5 (δ_C 120.0) and C-9 (δ_C 64.6), as well as a COSY correlation between H-8 (δ_H 6.14) and H-9 (δ_H 5.05) (Figure 3).

The coupling constants between H-8 and H-9 (³J_{H-8, H-9} = 18.0 Hz) indicated the transorientation of H-8 and H-9. Comparison of the 1D NMR data for 10 with those of 7 (Table 5 and Table 6), particularly for positions C-1 through C-9 and C-1′ through C-3′, confirmed that they had the same configuration around the epidithiodioxopiperazine ring. Accordingly, the relative configurations of 10 were proposed as 2R*,2′R*,8S*,9S*. Comparing the ECD curve with the calculated ones (Figure 4), the absolute configuration of 10 was determined as 2R,2′R,8S,9S.

2.2. Biological Assays

The cytotoxicity of compounds 1–10 was evaluated against a panel of human cancer cell lines, including chronic myeloid leukemia (K562), pancreatic adenocarcinoma (ASPC-1 and MIA-PACA-2), small cell lung cancer (NCI-H446 and NCI-H446/EP) and pancreatic cancer (MIA-PACA-2 and MIA-PACA-2AR) as well as the human hepatocyte cell line (L-02) (

Table 7. Cytotoxicity against six cancer cell lines of 3 and 7–10 (IC₅₀, μM).

No.	IC50 (μM)
	K562	L-02	ASPC-1	NCI-H446	NCI-H446/EP	MIA-PACA-2 a	MIA-PACA-2AR a
3	25.8	18.4	>30.0	27.0	>30.0	-	-
7	4.9	2.1	4.2	5.1	9.3	-	-
8	1.4	3.5	1.9	4.0	5.2	4.8	2.7
9	1.9	1.2	2.9	3.4	3.4	-	-
10	1.1	1.8	1.5	2.7	2.3	3.5	2.7
Doxorubicin	0.3	0.3	5.9	0.8	11.3	0.8	0.3

^a Compounds 3, 7 and 9 were not evaluated against the cancer cell lines MIA-PACA-2 and MIA-PACA-2AR.

). Compounds 7–10 possessed an α–α′ polysulfide bridge and exhibited strong anticancer activity with IC₅₀ values ranging from 1.1 to 9.3 μM. Structure–activity relationship (SAR) analysis revealed that Compounds 7–8 and 10, bearing a disulfide bridge, and compound 9, containing a trisulfide bridge between C-2 and C-2′, showed stronger activity (IC₅₀ = 1.1–5.2 μM) than 3 (IC₅₀ = 25.8–27.0 μM), suggesting the crucial role of the α–α’ polysulfide linkage in cytotoxic potency. Among them, compound 10 exhibited the most potent activity, particularly against ASPC-1 and NCI-H446/EP cell lines (IC₅₀ = 1.5–2.3 μM), surpassing the reference drug Doxorubicin (IC₅₀ = 5.9–11.3 μM). Other compounds exhibited no inhibitory activity against the tested cancer cell lines.

3. Materials and Methods

3.1. General Experimental Procedures

Optical rotations were measured with a JASCO P-1020 digital polarimeter (JASCO Corporation, Tokyo, Japan). UV spectra were recorded on a Hitachi 5430 spectrophotometer (Hitachi Ltd., Tokyo, Japan). ECD spectra were measured on a JASCO J-715 (JASCO Corporation, Tokyo, Japan). NMR spectra were collected on Bruker AVANCE NEO 400 MHz spectrometer (Bruker, Karlsruhe, Germany), Agilent 500 MHz DD2 spectrometers (Agilent Technologies, Palo Alto, CA, USA) and a JNM-ECZ600R/S1 (JEOL Ltd., Tokyo, Japan). HRESIMS data were measured on a Thermo Scientific LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Sephadex LH-20 (Amersham Biosciences, Piscataway, NJ, USA) and silica gel (Qingdao Marine Chemical Factory, Qingdao, China) were used as stationary phases in column chromatography (CC). Semipreparative HPLC was performed on an ODS column (YMC-Pack ODS-A, YMC Co., Ltd., Fukuchiyama, Kyoto, Japan).

3.2. Fungal Material

Marine sediment collected from the Yellow River estuary, China (coordinates: 37°57′43″ N, 119°15′55″ E) in July 2023 was pretreated by ultrasonication for one minute. After suspending and diluting with seawater, the sediment was spread on three types of culture medium: PDA (Potato Dextrose Agar) medium, Rose Bengal medium and Sabouraud Dextrose Agar medium and cultured at 28 °C and 15 °C, respectively. The fungal strain HDN23-0007 was isolated from Rose Bengal medium at 28 °C and identified as Corallomycetella repens based on sequencing of the ITS region (GenBank No. PV628721) with 100% similarity. It was deposited at Key Laboratory of Marine Drugs, the Ministry of Education of China, School of Medicine and Pharmacy, Ocean University of China, Qingdao, People’s Republic of China.

3.3. Fermentation and Extraction

The fungal Corallomycetella repens HDN23-0007 was cultured in 1000 mL Erlenmeyer flasks containing 300 mL of malt culture medium (yeast extract 0.4%, malt extract 1%, glucose 0.4%) dissolved in naturally collected seawater (Huiquan Bay, Yellow Sea, Qiangdao, China). Fifty flasks were cultured under static conditions at 28 °C for 30 days. After 30 days of cultivation, the whole broth (15 L) was filtered and separated into supernatant and mycelia. The former was extracted three times with EtOAc, while the latter was extracted three times with methanol. All extracts were combined and evaporated under reduced pressure to give a crude extract (14.0 g).

3.4. Isolation and Purification

The crude extract was chromatographed over ODS, eluting with mixtures of MeOH/H₂O to obtain five fractions (Fr.1 to Fr.5). Fr.1 and Fr.2 were separated on the Sephadex LH-20 column with MeOH to obtain Fr.1.1 and Fr.2.1. Fr.1.1 was further purified by semipreparative HPLC (20% ACN/H₂O) to yield compound 5 (t_R = 9.8 min, 3.0 mg). Fr.2.1 was further separated by semipreparative HPLC (58% MeOH/H₂O) to obtain Fr.2.1.1 and Fr.2.1.2. Fr.2.1.1 was further purified by semipreparative HPLC (62% ACN/H₂O) to yield compound 3 (t_R = 10.5 min 6.0 mg). Fr.2.1.2 was further purified by semipreparative HPLC (50% ACN/H₂O) to yield compound 4 (t_R = 15.0 min 3.0 mg).

Fr.3 was separated by semipreparative HPLC (58% MeOH/H₂O) to obtain Fr.3.1 and Fr.3.2. Fr.3.1 was further purified by semipreparative HPLC (64% ACN/H₂O), semipreparative HPLC (60% ACN/H₂O) and semipreparative HPLC (52% ACN/H₂O), respectively, to yield compounds 7 (t_R = 8.1 min 6.0 mg), 11 (t_R = 10.0 min 8.5 mg), 6 (t_R = 11.0 min 3.0 mg). Fr.3.2 was further purified by semipreparative HPLC (52% ACN/H₂O) to yield compound 9 (t_R = 10.5 min 6.0 mg).

Fr.4 was separated by semipreparative HPLC (60% MeOH/H₂O) to obtain Fr.4.1 and Fr.4.2. Fr.4.1 was further purified by semipreparative HPLC (52% ACN/H₂O) to yield compound 1 (t_R = 13.8 min 2.5 mg). Fr.4.2 was further purified by semipreparative HPLC (58% ACN/H₂O) to yield compound 10 (t_R = 11.2 min 3.0 mg). Fr.5 was separated by semipreparative HPLC (58% MeOH/H₂O) to obtain Fr.5.1 and Fr.5.2. Fr.5.1 was further purified by semipreparative HPLC (64% ACN/H₂O) to yield compound 8 (t_R = 9.8 min 3.0 mg). Fr.5.2 was further purified by semipreparative HPLC (62% ACN/H₂O) to yield compound 2 (t_R = 12.0 min 3.0 mg).

Corallomycetellain A (1): light-yellow amorphous powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ + 86.0 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 249 nm; ECD (MeOH) λ_max 217 (−), 259 (−), 293 (+) nm; ¹H and ¹³C NMR data, Table 1; HRESIMS m/z 417.0880 [M + Na]⁺ (calcd. for C₂₁H₁₈N₂O₄SNa, 417.0885).

Corallomycetellain B (2): light-yellow amorphous powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ + 41.6 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 253 nm; ECD (MeOH) λ_max 221 (−), 266 (−), 308 (+) nm; ¹H and ¹³C NMR data, Table 1; HRESIMS m/z 449.0589 [M + Na]⁺ (calcd. for C₂₁H₁₈N₂O₄S₂Na, 449.0606).

Corallomycetellain C (3): light-yellow amorphous powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 36.0 (c 0.25 mg/mL, MeOH); UV (MeOH) λ_max: 264 nm; ECD (MeOH) λ_max 227 (−), 285 (+) nm; ¹H and ¹³C NMR data, Table 2; HRESIMS m/z 557.0835 [M + Na]⁺ (calcd. for C₂₄H₂₆N₂O₆S₃Na, 557.0851).

Corallomycetellain D (4): light-yellow amorphous powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 48.0 (c 0.25 mg/mL, MeOH); UV (MeOH) λ_max: 263 nm; ECD (MeOH) λ_max 230 (−), 279 (+) nm; ¹H and ¹³C NMR data, Table 2; HRESIMS m/z 541.1078 [M + Na]⁺ (calcd. for C₂₄H₂₆N₂O₇S₂Na, 541.1079).

Corallomycetellain E (5): yellow oil; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 1.5 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 211, 264 nm; ECD (MeOH) λ_max 222 (−), 256 (−), 285 (+) nm; ¹H and ¹³C NMR data, Table 3; HRESIMS m/z 453.1139 [M + H]⁺ (calcd. for C₂₀H₂₅N₂O₆S₂, 453.1154).

Corallomycetellain F (6): colorless oil; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ + 144.0 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 216, 264 nm; ECD (MeOH) λ_max 226 (−), 280 (+) nm; ¹H and ¹³C NMR data, Table 4; HRESIMS m/z 467.1070 [M + Na]⁺ (calcd. for C₂₁H₁₈N₂O₄SNa, 467.1075).

Corallomycetellain G (7): white powder; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 113.6 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 265 nm; ECD (MeOH) λ_max 218 (−), 239 (−), 276 (+) 356 (−) nm. ¹H and ¹³C NMR data, Table 5; HRESIMS m/z 495.0655 [M + Na]⁺ (calcd. for C₂₂H₂₀N₂O₆S₂Na, 495.0660).

Corallomycetellain H (8): white powder; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 102.0 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 264 nm; ECD (MeOH) λ_max 218 (−), 238 (−), 276 (+) 356 (−) nm; ¹H and ¹³C NMR data, Table 5; HRESIMS m/z 523.0594 [M + Na]⁺ (calcd. for C₂₄H₂₄N₂O₆S₂Na, 523.0973).

Corallomycetellain I (9): white powder; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 312.3 (c 0.3 mg/mL, MeOH); UV (MeOH) λ_max: 268 nm; ECD (MeOH) λ_max 211 (−), 230 (+), 261 (−) nm; ¹H and ¹³C NMR data, Table 6; HRESIMS m/z 527.0389 [M + Na]⁺ (calcd. for C₂₂H₂₀N₂O₆S₃Na, 527.0381).

Corallomycetellain J (10): white powder; [α] ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ − 314.8 (c 0.5 mg/mL, MeOH); UV (MeOH) λ_max: 212, 262 nm; ECD (MeOH) λ_max 221 (−), 242 (−), 285 (+) nm; ¹H and ¹³C NMR data, Table 6; HRESIMS m/z 435.0432 [M + Na]⁺ (calcd. for C₂₀H₁₆N₂O₄S₂Na, 435.0449).

Haematocin (11): white powder; ECD (MeOH) λ_max 219 (−), 285 (+) nm; ¹H and ¹³C NMR data, Figures S91 and S92.

3.5. LC–MS/MS and Molecular Networking Analysis

Liquid chromatography–tandem mass spectrometry (LC–MS/MS) analyses were performed on a Dionex Ultimate 3000 UHPLC system (Thermo Scientific) coupled to a Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Scientific). Electrospray ionization was operated in negative-ion mode. The mobile phase comprised 0.1% (v/v) formic acid in ultrapure water (solvent A) and HPLC-grade acetonitrile (solvent B). The gradient was 0–1 min, 10% B; 1–23 min, linear increase to 100% B; 100% B held until 26 min; 26–30 min, return to 10% B and re-equilibration. The flow rate was 0.25 mL/min and the injection volume was 3 µL.

All MS/MS data were converted to mzXML format files by MSConvert software (v3.0.20169, ProteoWizard, Palo Alto, CA, USA). WinSCP as an FTP client was used to upload mzXML format files to GNPS for analysis. The desired files or entire folders were then assigned to different analysis groups (from G1 to G6). The molecular network was constructed using the GNPS data analysis workflow and algorithms. The network visualizations were completed through Cytoscape (v3.10.3; National Resource for Network Biology, San Diego, CA, USA).

3.6. ECD Calculations

Conformational searches were performed using the systematic protocol in Spartan’ 14 [19] with the Merck Molecular Force Field (MMFF). Each MMFF minimum was re-optimized at the B3LYP/6-31+G(d) level of density functional theory using Gaussian09 (D.01, Gaussian, Inc., Wallingford, CT, USA) [20] Optimizations were initiated from multiple starting conformers, and vibrational frequency analyses confirmed that all stationary points are true minima. For each configurational ensemble, the lowest energy conformers with Boltzmann populations > 5% were retained. Excited-state properties for these conformers were computed with time-dependent DFT (TD-DFT) for twenty singlet excited states; solvation in methanol was treated with the polarizable continuum model (PCM). ECD spectra were simulated with SpecDis [21]; each transition was convoluted with a Gaussian band shape (0.26 eV FWHM), and an overall blue or red shift was applied to match the experimental spectra.

3.7. In Vitro Cytotoxicity Assays

The cytotoxicity of compounds 1–10 against K562 cells was evaluated using the MTT assay, while their effects on a panel of adherent cell lines (L-02, ASPC-1, NCI-H446, NCI-H446/EP, MIA-PACA-2, and MIA-PACA-2AR) were assessed using the SRB assay. Doxorubicin was used as the positive control in all assays. The half-maximal inhibitory concentration (IC₅₀) values were calculated by fitting the dose–response curves obtained after 72 h of continuous drug exposure. Data were presented as the mean from three independent experiments. The detailed methodologies for biological testing have been described in previous reports [22,23].

4. Conclusions

Ten new epipolythiodioxopiperazine (ETP) alkaloids, namely corallomycetellains A–J (1–10) and one known compound haematocin (11) were isolated from the marine sediment-derived fungi Corallomycetella repens HDN23-0007. Their structures, including absolute configurations, were elucidated by NMR, MS spectroscopic data and electronic circular dichroism (ECD) calculations. Among them, corallomycetellains A (1) and B (2) are the first reported aranotin-type ETPs containing aromatic indole moieties. In bioassays, corallomycetellains G–J (7–10), contained α–α′ polysulfide bridges, exhibited strong anticancer activity, with IC₅₀ values ranging from 1.1 to 9.3 μM. Among them, corallomycetellain J (10) exhibited the most potent activity, particularly against ASPC-1 and NCI-H446/EP cell lines (IC₅₀ = 1.5–2.3 μM), surpassing the reference drug Doxorubicin (IC₅₀ = 5.9–11.3 μM). Overall, this study expands the chemical diversity of marine fungal metabolites and provides promising leads for the development of novel anticancer agents. It is worth noting that the ETP alkaloids, characterized by their sulfur bridges, often exhibit significant cytotoxicity. The potent activity observed in compounds 7–10 likely stems from the reactive polysulfide linkage, which can engage in redox cycling or cross-link with critical cellular thiols. While these effects contribute to their strong inhibitory impact on cancer cells, the potential for non-specific toxicity towards normal cells remains a challenge for their clinical translation. In terms of anti-angiogenesis, ETP-type molecules are known to disrupt the interaction between the transcription factor HIF-1α and its coactivator p300 by targeting the zinc-binding CH1 domain, suggesting that they could potentially inhibit tumor vascularization. Therefore, to mitigate these risks, future studies should focus on structural optimization, such as the development of prodrugs or targeted delivery systems to enhance tumor specificity.