For open column chromatography applications, silica gel of mesh sizes 100–200 and 200–300 (Qingdao Marine Chemical Inc., Qingdao, China), Lobar LiChroprep RP-18 (40–60 μm, Merck, Darmstadt, Germany), and Sephadex LH-20 (Merck) were the materials of choice. High-Resolution Electrospray Ionization Mass Spectrometry (HRESIMS) experiments, conducted in positive ion mode, utilized a Waters Xevo G2-XS QTof mass spectrometer (Waters, Milford, MA, United States). Nuclear Magnetic Resonance (NMR) spectroscopic data were collected with a Bruker Avance 600 MHz spectrometer, employing tetramethylsilane (TMS) as an internal standard for calibration. Optical rotations were ascertained utilizing an MCP 500 polarimeter instrument manufactured by Anton Paar. Ultraviolet (UV) spectroscopic analyses were carried out on a Shimadzu UV-1800 spectrometer (Shimadzu Co., Ltd., Kyoto, Japan).

C. globosum 2020HZ23, the producing fungal strain, was originally separated from the inner tissues of the marine brown algae Sargassum thunbergii, harvested in September 2020 from Qingdao, China. Morphological attributes along with sequencing of the Internal Transcribed Spacer (ITS) region (GenBank accession no. OR195778) (Figure 1) facilitated the precise identification of this strain as C. globosum 2020HZ23. To clearly indicate the evolutionary position of this fungal strain C. globosum 2020HZ23, a phylogenetic analysis based on its ITS sequence as well as those from other Chaetomium species, was performed. As shown in Figure 1, the fungus C. globosum 2020HZ23 was located at the head position of the entire phylogenomic tree with a high confidence of 99%. This fungus is currently deposited at the Qingdao Hiser Hospital Affiliated of Qingdao University.

The fungal strain underwent fermentation in a static state on a solid rice medium. Each 1 L Erlenmeyer flask contained a concoction of 0.1 g sodium glutamate, 0.1 g corn flour, 0.3 g peptone, 1 g mannitol, 1 g maltose, 2 g D-glucose, 70 g rice, and 100 mL seawater from Qingdao Beach. The pH value was regulated to 6.5 prior to the fermentation process, which lasted 25 days at ambient temperature. Post fermentation, methanol extraction of the liquid was executed, followed by a triple filtration with Whatman filter paper. The methanolic extract was then concentrated under reduced pressure and partitioned between water and ethyl acetate. Further vacuum concentration of the ethyl acetate fraction yielded a 150 g extract. The extract was subjected to silica gel column chromatography with a gradient of petroleum ether and ethyl acetate to yield five fractions (Frations 1–5), which were consolidated based on thin-layer chromatography (TLC) analyses. Fraction 1 was further purified with Sephadex LH-20, producing compound 2 (2.3 mg) and compound 1 (1.8 mg). Fraction 2 was processed through Sephadex LH-20 and reversed-phase HPLC (20–50% MeCN/H₂O, with the timespan of 10.0 min and the flow rate of 10 mL/min), yielding compound 3 (2.0 mg, t_R = 6.413 min).

Compound 1: A red amorphous powder; [α]²⁵_D + 1,270 (c 0.005, MeOH); UV (MeOH) λ_max (log ε): 225 (4.15), 295 (4.12) nm; ECD (c 1.0 mg/mL, MeOH) λ_max (∆ε): 230 (−22.6), 310 (+28.2), 380 (−22.1); ¹H and ¹³C NMR data, shown in Table 1; HRESIMS m/z 510.2022 [M + Na]⁺ (calcd for C₂₇H₃₄ClNO₅Na, 510.2022).

Compound 2: A red amorphous powder; [α]²⁵_D + 1,360 (c 0.005, MeOH); UV (MeOH) λ_max (log ε): 220 (4.24), 295 (4.18) nm; ECD (c 1.0 mg/mL, MeOH) λ_max (∆ε): 230 (−30.2), 310 (+25.2), 375 (−24.2); ¹H and ¹³C NMR data, shown in Table 1; HRESIMS m/z 538.2333 [M + Na]⁺ (calcd for C₂₉H₃₈ClNO₅Na, 538.2336).

Candidate conformers were generated utilizing the Conformer Rotamer Ensemble Sampling Tool (CREST) (Grimme, 2019; Pracht et al., 2020) and subject to Density Functional Theory (DFT) calculations via the Gaussian 16 program (Frisch et al., 2016). Conformers that fell within a 10 kcal/mol energy window were optimized at the B3LYP/6-31G(d) level of theory, implementing Grimme’s D3 dispersion correction. All optimized conformations underwent frequency analysis at the identical theoretical level to ascertain their local minima status on the potential energy surface. The energy values of all optimized conformations were then determined using the M062X/6–311 + G(2d,p) level with D3 dispersion correction. By combining the “Thermal correction to Gibbs Free Energy” from frequency analysis with electronic energies from M062X/6–311 + G(2d,p), Gibbs free energies were calculated for each conformer. Utilizing the Boltzmann distribution law, equilibrium populations at room temperature (298.15 K) were determined. Conformers with population values above 2% underwent additional computations. Electronic Circular Dichroism (ECD) Time-dependent Density Functional Theory (TDDFT) calculations were executed at the CAM-B3LYP/6-311G(d) level of theory, in methanol (MeOH) and with the application of the IEFPCM solvent model. 36 excited states were computed for each conformer (Pescitelli and Bruhn, 2016). The resultant ECD curves were developed using the Multiwfn 3.6 software (Lu and Chen, 2012).

A549 cells were purchased from National Collection of Authenticated Cell Cultures (China) and incubated with RPMI 1640 (Gibco, Beijing, China) with 10% fetal bovine serum (FBS; Gibco). Cells were grown at 37°C in a 5% CO₂ humidified atmosphere.

Cell viability was assessed using the Cell Counting Kit-8 (CCK-8, MCE, United States) according to the guidance of the manufacturer. Cells were seeded into 96-well plates and collected after treatment for 48 h. 10 μL of CCK-8 solution was added into the cultures at 37°C for 1 h. The absorbance at 450 nm was measured with a microplate reader (Spark multimode microplate reader, Tecan, Austria).

A549 cells were seeded in six-well plates until cell confluence reached approximately 100%. The wounds were scratched with 10 μL pipette tips, and cells were washed with PBS. The cells were cultured with 1% FBS medium. The scratch recovery was observed at 0 and 48 h, and the healing rates were estimated with ImageJ software.

The molecular formula of Compound 1, a dark red solid, is confirmed as C₂₇H₃₄ClNO₅ through High-Resolution Electrospray Ionization Mass Spectrometry (HRESIMS), suggesting 11 degrees of unsaturation. The chlorine atom was authenticated through the isotopic peak observation of [M + H]⁺: [M + H + 2]⁺ at a 3:1 ratio. ¹³C, DEPT, and HSQC spectra revealed one disubstituted double bond (δ_C 119.1 and 149.3), two trisubstituted double bonds (δ_C 147.2 and 111.2; δ_C 141.5 and 111.4), two tetrasubstituted double bonds (δ_C 144.9 and 99.9; δ_C 123.0 and 168.8), one ester carbonyl carbon (δ_C 168.8), and two keto carbons (δ_C 181.5 and 201.8) (Table 1). Structural similarities between 1 and the co-isolated chaetoviridin A (3) were noted from the spectral data, with both compounds possessing a tricyclic core with two side chains, an aldol group and a methyl-branched pentenyl. The most noticeable differences were observed in the chemical shifts of C-1 (from δ_C 151.6 in 3 to δ_C 141.5 in 1) and C-3 (from δ_C 157.0 in 3 to δ_C 147.2 in 1). Moreover, four extra resonances in compound 1, indicative of a butyl unit, were observed. Given the chemical shifts for C-1 and C-3 and the overall molecular weight, a nitrogen atom was postulated at the 2 position, bearing a butyl group. This inference was supported by COSY correlations of H₂-1”/H₂-2”/H₂-3”/H₃-4″ and HMBC correlations of H₂-1″ with C-1 and C-3 (Figure 3). The E configuration of the C-9-C-10 double bond was determined via ³J_H9-H10 (15.4) (Table 1). The absolute configurations at C-4′, C-5′, and C-11 were established as 4’R, 5’R, 11S, based on NMR data comparison with the co-isolated 3, as well as the four previously synthesized chaetoviridin A epimers (Makrerougras et al., 2017), considering the same biosynthetic pathway. To determine the stereochemistry of C-7, we conducted ECD calculations on the simplified structures of (7S)-1 (1a) and (7R)-1 (1b), which resulted in the assignment of the C-7 position as S (Figure 4). Additionally, the negative CE at approximately 380 nm was attributed to the electron transition from MO83 (HOMO) to MO84 (LUMO) (Figure 5), in alignment with the ECD spectrum of chaetoviridin A reported by Steyn and Vleggaar (1976), as well as the nitrogenated azaphilones reported by Wang et al. (2020). Consequently, we identified the compound 1 as N-butyl-2-aza-2-deoxychaetoviridin A.

Compound 2, procured as a dark red solid, was assigned a molecular formula of C₂₉H₃₈ClNO₅ through HRESIMS analysis. The chlorine atom was also authenticated through the isotopic peak observation of [M + H]⁺:[M + H + 2]⁺ at a 3:1 ratio. Additionally, ¹³C, DEPT, and HSQC spectra were utilized, which revealed the presence of one disubstituted double bond (δ_C 119.1 and 149.4), two trisubstituted double bonds (δ_C 147.3 and 111.3; δ_C 141.6 and 111.5), two tetrasubstituted double bonds (δ_C 145.0 and 99.9; δ_C 123.1 and 168.6), one ester carbonyl carbon (δ_C 168.6), and two keto carbons (δ_C 181.1 and 201.8) (Table 1). Structural similarities between 2 and the co-isolated 3 were observed in the spectral data. Both compounds possess a tricyclic core with two side chains, an aldol group, and a methyl-branched pentenyl. Upon comparing the NMR data of 2 to that of 1, it was noted that they share the same stereogenic centers, while differences manifest in the side chain attached to N-2. The presence of two additional carbon resonances compared to 1, along with COSY correlations of H₂-1”/H₂-2’/H₂-3”/H₂-4”/H₂-5”/H₃-6″ and HMBC correlations of H₂-1”/C-1 and H₂-1”/C-3, confirmed the attachment of a hexyl group to N-2 (Figure 3). Consequently, following further 2D NMR analysis, the structure of 2 was ascertained to be N-hexyl-2-aza-2-deoxychaetoviridin A.

Compound 3 was ascertained as chaetoviridin A through the comparison of NMR data with those documented in existing literature (Park et al., 2005).

We utilized a CCK8 assay to examine the impact of compounds 1–3 on the viability of the A549 cancer cell line. Both compounds 1 and 2 demonstrated dose-dependent cytotoxicity, with IC₅₀ values of 13.6 and 17.5 μM (Figure 6B), respectively, while compound 3 showed a low degree of cytotoxicity (IC₅₀ > 50 μM, data were not shown). The cytotoxic results suggested that the cytotoxic potency escalates with N-substitution at the C-2 position and the introduction of a side chain. A wound-healing assay was conducted to assess the impact of compounds 1 and 2 on the migration and invasion capabilities of A549 cells. Figures 6A,C illustrate that A549 cells in the control group were able to migrate across the complete wound area within a 48-h period. However, cell migration was significantly curtailed in a dose-dependent manner when treated with specific concentrations (1, 2 and 4 μM) of compounds 1 and 2.