New Phenylspirodrimanes from the Sponge-Associated Fungus Stachybotrys chartarum MUT 3308

Two phenylspirodrimanes, never isolated before, stachybotrin J (1) and new stachybocin G (epi-stachybocin A) (2), along with the already reported stachybotrin I (3), stachybotrin H (4), stachybotrylactam (5), stachybotrylactam acetate (6), 2α-acetoxystachybotrylactam acetate (7), stachybotramide (8), chartarlactam B (9), and F1839-J (10) were isolated from the sponge-associated fungus Stachybotrys chartarum MUT 3308. Their structures were established based on extensive spectrometric (HRMS) and spectroscopic (1D and 2D NMR) analyses. Absolute configurations of the stereogenic centers of stachybotrin J (1), stachybocin G (2), and stachybotrin I (3), were determined by comparison of their experimental circular dichroism (CD) spectra with their time-dependent density functional theory (TD-DFT) circular dichroism (ECD) spectra. The putative structures of seventeen additional phenylspirodrimanes were proposed by analysis of their respective MS/MS spectra through a Feature-Based Molecular Networking approach. All the isolated compounds were evaluated for their cytotoxicity against five aggressive cancer cell lines (MP41, 786, 786R, CAL33, and CAL33RR), notably including two resistant human cancer cell lines (786R, CAL33RR), and compounds 5, 6, and 7 exhibited cytotoxicity with IC50 values in the range of 0.3−2.2 µM.

Drug resistance is currently a major problem in several cancers such as UM (uveal melanoma), RCC (renal cell carcinoma), and HNSCC (head and neck squamous cell carcinoma). Although the 1980s was the decade of general radio-chemotherapy with the use of very toxic antitumor drugs and radiation procedures that resulted in numerous side effects, the 2000s was the decade of targeted therapies due to the development of treatments that specifically target driving mutations [23][24][25]. Despite the improvements, many of the patients were not cured and suffered relapses. During the 2010s, the strategies focused on damaging the microenvironment, particularly blood vessels, in colon, breast, lung, and kidney cancers [26]. However, the effect of anti-angiogenic drugs was shortlived, and relapses were inevitable. The 2020s is definitely the decade of immunotherapies improving patient survival, but only in 20% of patients with various cancers such as UM, RCC, and HNSCC [27][28][29][30]. Therefore, to further improve current treatments, the introduction of new therapies is urgently needed. These breakthrough treatments can improve the quality of life or survival of many patients. The discovery of these new drugs from natural products is one of the most important tasks in medicinal chemistry.

Structure Elucidation
Stachybotrys chartarum MUT 3308 was cultivated, in solid and liquid conditions, using PDA as a medium. For each culture condition, all the fungal material and the medium were extracted with an appropriate mixture of solvents and fractionated by reverse-phase or normal-phase chromatography. The most interesting fractions, based on the HPLC-PDA-ELSD and UHPLC-HRMS/MS metabolic profiles, were purified by RP HPLC to yield, in total, pure compounds 1 (2.1 mg), 2 (3.5 mg), 3 (

6′′
158.6, C ---All this data allowed us to identify compound 1 as a PSD derivative never isolated before, named stachybotrin J (1). Although this compound was semi-synthesized by Steinert et al. in 2022, the absolute configuration of the stereogenic center of the arginine residue was not determined [32]. The experimental CD spectrum of 1 exhibited two negative Cotton effects (CEs) at λmax = 227 nm and λmax = 270 nm. The Boltzmann-averaged TD-DFT calculated ECD spectrum for the most stable conformers of the enantiomer 1a (2′′S, 3R, 5S, 8R, 9R, 10S), performed at the B3LYP/6-311+G(d,p) level of theory, also showed two negative CEs, at λmax = 230 nm and λmax = 308 nm, which reproduced the signs and differences in amplitude of The experimental CD spectrum of 1 exhibited two negative Cotton effects (CEs) at λ max = 227 nm and λ max = 270 nm. The Boltzmann-averaged TD-DFT calculated ECD spectrum for the most stable conformers of the enantiomer 1a (2 S, 3R, 5S, 8R, 9R, 10S), performed at the B3LYP/6-311+G(d,p) level of theory, also showed two negative CEs, at λ max = 230 nm and λ max = 308 nm, which reproduced the signs and differences in amplitude of the experimental CEs. Thus, a 2 S-configuration was determined for 1 ( Figure 3, Tables S11-S14, Figure S67).
The molecular formula of compound 3, C 32 H 39 NO 6 , was deduced from the HRESI(+)MS analysis which showed a pseudo-molecular ion peak at m/z 534.2839 [M + H] + (calcd for C 32 H 40 NO 6 + , 534.2850, 14 degrees of unsaturation). Compound 3 showed very close 1 H and 13 C NMR chemical shifts for its PSD moiety, which were also very close to those of stachybotrylactam (5) (Table S3) [35]. Stachybotrin I (3) (or an isomer) was previously isolated from the culture of a S. atra ST002348 [35]; however, (i) multiplicities of the protons in the 1 H NMR spectrum were not reported, (ii) 13 C chemical shifts were deduced from the HMQC spectrum, and consequently the 13 C chemical shifts were not determined for quaternary carbons, and (iii) the relative/absolute configurations were not established. The CD spectrum of 3 exhibits two negative Cotton effects at 226 nm and 268 nm. Comparison with the calculated ECD spectra for the most stable conformers of the two possible enantiomers of 3 [3a (2 S 3R, 5S, 8R, 9R, 10S) and 3b (2 R, 3R, 5S, 8R, 9R, 10S)] allowed us to deduce a 2 S-configuration for compound 3 (Figure 3, Tables S19-S22, Figure S67).
PSD dimers were scarcely reported in the literature and their origin, natural or artifacts, is still a matter of debate. Recently, Jagels and his group [38] completed the Jarvis hypothesis [36] according to which stachybotrylactam (5) and its N-functionalized derivatives from S. chartarum could be artifacts by showing that isoindolinones production is favored in N-rich media. Several plausible biogenetic pathways for PSDs have been proposed. Structurally, PSDs are mainly polyketide-terpenoid hybrid meroterpenoids [39,40]. Compounds 1-10 could be derived from a common intermediate, ilicicolin B (14), which originates from farnesyldiphosphate (11) and orsellinic acid (12) ( Figure 5). Afterwards, ilicicolin B (14) would undergo a series of reactions, notably oxidations and cyclizations, to yield stachybotrydial (16) [41]. Stachybotrydial (16) could react with a wide range of nucleophiles readily available in the medium, notably amines, for which the complete mechanism is still not fully elucidated [38,42], to give all isoindolinones. For example, ammonia, obtained by the enzymatic conversion of the nitrate present in the medium, would react with stachybotrydial (16) to give stachybotrylactam (5) [38,43]. In an identical way, amino acids, such as glycine, L-phenylalanine, and L-arginine, would react with stachybotrydial (16) to yield stachybotrin H (4), stachybotrin I (3), and stachybotrin J (1), respectively. Very recently, some PSD derivatives, such as stachybotrin J (1), have been obtained by semi-synthesis from stachybotrydial (16) and amino acids to support this hypothesis [32]. However, the absolute configurations of these compounds have not been reported. Thus, stachybocin G (2) could be obtained by reaction of D-lysine with two stachybotrydial (16) units or by C-2 epimerization of stachybocin A (33) ( Figure 5).

Feature-Based Molecular Networking Analysis
Mass spectrometry Feature-Based Molecular Networking (FBMN) analyses were performed to putatively assign further PSD derivatives that could be produced by S. chartarum MUT 3308 when cultivated in solid (F2: CH 3 OH/CH 2 Cl 2 (1:1, v/v) crude extract) and liquid (AcOEt crude extract) conditions [44]. For this purpose, each fungal crude organic extract was (i) analyzed by UHPLC-HRESIMS(/MS), (ii) preprocessed using MZmine 2 [45], and (iii) analyzed by the FBMN approach to also distinguish possible isomers in the network based on their retention time [44]. The graphical representation of the molecular network (depicting the chemical space present in the MS/MS data) of S. chartarum MUT 3308 allowed us to highlight 196 nodes, of which 131 are linked together, which suggests the production of numerous metabolites. The main cluster, dedicated to compounds 1-10 and their derivatives, is constituted by 64 nodes, of which 21 nodes (33%) were common to both cultivation conditions, 34 nodes (53%) that were only observed for the solid cultivation condition, and 9 nodes (14%) that were specific to the liquid cultivation condition ( Figure 6). The FBMN approach allowed us to assign the isolated compounds 1-10 and to annotate seventeen more PSD derivatives. In total, in this cluster, 28 nodes (44%) have been identified ( Figure 6, Table S23). respectively. Very recently, some PSD derivatives, such as stachybotrin J (1), have been obtained by semi-synthesis from stachybotrydial (16) and amino acids to support this hypothesis [32]. However, the absolute configurations of these compounds have not been reported. Thus, stachybocin G (2) could be obtained by reaction of D-lysine with two stachybotrydial (16) units or by C-2′′ epimerization of stachybocin A (33) ( Figure 5).

Feature-Based Molecular Networking Analysis
Mass spectrometry Feature-Based Molecular Networking (FBMN) analyses were performed to putatively assign further PSD derivatives that could be produced by S. chartarum MUT 3308 when cultivated in solid (F2: CH3OH/CH2Cl2 (1:1, v/v) crude extract) and liquid (AcOEt crude extract) conditions [44]. For this purpose, each fungal crude organic extract was (i) analyzed by UHPLC-HRESIMS(/MS), (ii) preprocessed using MZmine 2 [45], and (iii) analyzed by the FBMN approach to also distinguish possible isomers in the network based on their retention time [44]. The graphical representation of the molecular network (depicting the chemical space present in the MS/MS data) of S. chartarum MUT 3308 allowed us to highlight 196 nodes, of which 131 are linked together, which The green subcluster is constituted by 21 nodes, 6 of which were assigned as 2αacetoxystachybotrylactam acetate (7)   .3264 could not be assigned to any known metabolite by manual or GNPS-based dereplication approaches. Figure 6. Feature-Based Molecular Network analysis of the crude solid culture extract and the liquid culture filtrate crude extract of S. chartarum MUT 3308 (common fragment number: 6; similarity score: 0.6). Nodes are shown as pie charts to reflect the relative abundance of each ion in each of the extracts. Node size represents the total sum of the precursor ion intensity in the MS 1 scan. Edge thickness corresponds to relative cosine score similarity between nodes. The annotated cluster is enlarged. Isolated molecules are in blue, new molecules are in red, and proposed molecules in black or purple if several nodes could be matched to them. Figure 6. Feature-Based Molecular Network analysis of the crude solid culture extract and the liquid culture filtrate crude extract of S. chartarum MUT 3308 (common fragment number: 6; similarity score: 0.6). Nodes are shown as pie charts to reflect the relative abundance of each ion in each of the extracts. Node size represents the total sum of the precursor ion intensity in the MS 1 scan. Edge thickness corresponds to relative cosine score similarity between nodes. The annotated cluster is enlarged. Isolated molecules are in blue, new molecules are in red, and proposed molecules in black or purple if several nodes could be matched to them.
The orange subcluster is comprised of 18 nodes and only 1 of which was annotated as stachybotrin J (1) Table S23).
All the putatively annotated compounds could also be proposed as their isomers. The relative and/or absolute configuration cannot be determined unless other appropriate spectroscopic techniques are used.

Biological Assays
Compounds 1-10, isolated in small amounts, were evaluated for their cytotoxicity against five aggressive human cancer cell lines: MP41 (melanoma), 786 (renal carcinoma), 786R (sunitinib-resistant renal cell carcinoma), CAL33 (head and neck carcinoma), and CAL33RR (cisplatin-and radiotherapy-resistant head and neck carcinoma). The above cells were treated with compounds 1-10 for two days and XTT assays were used to assess cell metabolism and proliferative capacity. The IC 50 values (µM) are shown in Table 2. Compounds 5-7 showed a weak toxicity against the MP41 cell line (IC 50 < 1.0 µM) although compounds 1-4 and 8-10 showed almost no cytotoxicity against the MP41 cell line (IC 50 > 50 µM). In addition, compounds 5-7 exhibited better cytotoxic activities against the 786 and CAL33 cell lines (from 3.6 to 2.5-fold less), with IC 50 values in the range of 0.3-1.5 µM, compared to sunitinib (IC 50 = 2.5 ± 0.5) and cisplatin (IC 50 = 1.5 ± 0.3), respectively, which were used as positive controls. Similarly, compounds 5-7 also showed better cytotoxic activities against the two resistant human cancer cell lines, compared to the positive controls, with IC 50 values from 0.8 to 2.2 µM against 786R (sunitinib IC 50 > 10 ± 1) and with IC 50 values ranging from 0.6 to 1.0 µM against CAL33RR (cisplatin IC 50 > 10 ± 1). Compounds 1-4 and 8-10 were almost non-cytotoxic against the MP41, 786, 786R, CAL33, and CAL33RR cell lines (except for compound 8, IC 50 < 20 µM). Consequently, our results clearly suggest that in terms of structure-activity relationships, a non-substituted lactam functionality in PSDs is required for cytotoxicity against the aggressive human cancer cell lines studied, notably the resistant cancer cell lines. PSDs have hardly been studied for their antitumor activities. In the literature, compounds 5, 4, and 8 showed no cytotoxicity against K562 (leukemia), HL60 (leukemia), and Hela (cervical cancer) cell lines (IC 50 > 100 µM) [37], and compounds 5, 3, 7, and 8 showed no cytotoxicity against NIH-3T3 (fibroblast) and HepG2 (liver carcinoma) cell lines (IC 50 > 50 µM) [18]. On the other hand, alternative therapies to metastatic RCC and HNSCC are urgently needed to prevent relapse on current conventional treatments (sunitinib for RCC and cisplatin for HNSCC). Recently, N,N -diarylureas and thioureas with a nitro-benzothiazole moiety, were synthesized and evaluated by our team for their anticancer properties, particularly against the 786 and CAL33 cell lines. Compared to compounds 5-7, the lead compound of this previous study, named C29, exhibited lower cytotoxic activity against cell lines 786 and CAL33 (1.3-to 6.7-fold higher), with IC 50 values of 2 and 4 µM, respectively [50]. Taken together, all these data seem to indicate that small modifications of the PSD skeleton could lead to a significant change in bioactivity and/or selectivity against human cancer cell lines, which is of great interest for the development of new anticancer drugs, especially against resistant cancer cell lines.

Feature-Based Molecular Networking Analysis
The data were processed by using the FBMN method [44]. The data files were converted from the raw data format to mzXML format using MSConvert software (Prote-oWizard package 3.0). All mzxml values were processed using MZmine 2.53 [45]. Mass detection was realized with an MS 1 noise level of 5 × 10 6 and an MS 2 noise level of 5 × 10 3 . The ADAP chromatogram builder was employed with a minimum group size of scans of 5, a group intensity threshold of 5 × 10 6 , a minimum highest intensity of 1.7 × 10 7 , and m/z tolerance of 0.0 (or 10 ppm). Deconvolution was performed with the Baseline cut-off algorithm according to the following settings: minimum peak height of 2.7 × 10 7 , peak duration range of 0.1-2 min, baseline level of 3 × 10 6 , and an auto m/z center calculation. MS/MS scans were paired using a m/z tolerance range of 0.02 Da and RT tolerance range of 0.1 min. Isotopologs were grouped using the isotopic peak grouper algorithm with a m/z tolerance of 0.0 (or 10 ppm) and a RT tolerance of 0.2 min. Peaks were filtered using a feature list row filter, keeping only peaks with MS/MS scans (GNPS). Peak alignment was performed using the join aligner with a m/z tolerance of 0.0 (or 10 ppm), a weight for m/z at 75%, a RT tolerance of 0.2 min and weight for RT at 25%. The MGF file and the metadata were generated using the export/submit to GNPS option [51]. The molecular network was calculated and visualized using Cytoscape software [52]. The parent mass tolerance was 0.02 Da and the MS/MS fragment ion tolerance was 0.02 Da. The edges were filtered to have a cosine score above 0.6 and more than 6 matched peaks.

Computational Analysis
TD-DFT calculations of ECD spectra were performed with the Gaussian 16 program package [53]. Conformer distribution analysis and geometry optimizations for all structures were carried out using the AM1 semi-empirical force field implemented in the Spartan 08 program. For each compound, the minimum energy structures were filtered and checked for duplicity. Then, each conformer was geometrically optimized using the hybrid DFT method B3LYP and the basis set 6-31+G(d,p) (B3LYP/6-31+G(d,p)), with thermochemical parameters and the frequencies at 298 K and 1 atm. The solvation effects of methanol were modelized with the polarizable continuum model (PCM). From the TD-DFT calculations performed on each structure optimized conformer, the calculated excitation energy (in nm) and rotatory strength R, in dipole velocity (Rvel) and dipole length (Rlen) forms, were simulated into an ECD curve by using the following Gaussian function (1): where σ is the width of the band at 1/e height, and Ei and Ri are the excitation energies and rotatory strengths for transition i, respectively. σ = 0.30 eV and Rvel were used. The Boltzmann-averaged ECD spectra were obtained from B3LYP/6-31+G(d,p)optimized structures. All DFT and TD-DFT calculations were performed using HPC resources from Azzurra.

Cell Culture
The human head and neck squamous cell carcinoma (HNSCC) cell line CAL33 (DSMZ, ACC 447) was provided through a Material Transfer Agreement with the Oncopharmacology Laboratory, Centre Antoine Lacassagne (CAL), where it had initially been isolated [54]. CAL33RR cells were generated by chronic exposure to cisplatin and several round of irradiation by 8 gray X-rays [55]. The kidney cancer cell line 786-0 (ATCC, CRL-1932) was purchased from the American Tissue Culture Collection. The 786R cell line was generated by chronic exposure to sunitinib [56]. The uveal melanoma cell line MP41 (ATCC, CRL-3297) was purchased from the American Tissue Culture Collection. The cells were cultured in Dulbecco's Modified Eagle medium (DMEM; Gibco) supplemented with 7% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA, USA).

Cytotoxicity Measurement (XTT)
The cells (5 × 10 3 cells/100 µL) were incubated in a 96-well plate with different concentrations of the drugs for 48 h. Fifty microliters of XTT reagent were added to each well. Each assay was performed in triplicate. The assay is based on the cleavage of the tetrazolium salt 2,3-Bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT, Sigma-Aldrich, Merck KGaA, Saint-Louis, MO, USA) in the presence of an electroncoupling reagent to produce a soluble formazan salt. This conversion only occurs in viable (metabolically active) cells. The number of viable cells is directly correlated with the amount of orange formazan by measuring the absorbance of the dye at 450 nm on a spectrophotometer.

Institutional Review Board Statement: Not applicable.
Data Availability Statement: Not applicable.