Discovery of Bioactive Indole-Diketopiperazines from the Marine-Derived Fungus Penicillium brasilianum Aided by Genomic Information

Identification and analysis of the whole genome of the marine-derived fungus Penicillium brasilianum HBU-136 revealed the presence of an interesting biosynthetic gene cluster (BGC) for non-ribosomal peptide synthetases (NRPS), highly homologous to the BGCs of indole-diketopiperazine derivatives. With the aid of genomic analysis, eight indole-diketopiperazines (1−8), including three new compounds, spirotryprostatin G (1), and cyclotryprostatins F and G (2 and 3), were obtained by large-scale cultivation of the fungal strain HBU-136 using rice medium with 1.0% MgCl2. The absolute configurations of 1−3 were determined by comparison of their experimental electronic circular dichroism (ECD) with calculated ECD spectra. Selective cytotoxicities were observed for compounds 1 and 4 against HL-60 cell line with the IC50 values of 6.0 and 7.9 μM, respectively, whereas 2, 3, and 5 against MCF-7 cell line with the IC50 values of 7.6, 10.8, and 5.1 μM, respectively.


Introduction
In the past decades, marine-derived fungi have attracted more and more attention to medicinal chemists because of their ability to produce biologically active compounds with diverse structures for drug discovery [1]. Especially, one fungus species could generate structurally versatile natural products because it consists of a large number of biosynthetic gene clusters (BGCs) associated with its secondary metabolism [2,3]. For example, several bioactive compounds have been generated by the same fungus from different sources, including antifungal diketopiperazine-type alkaloids [4], antibacterial dimeric diketopiperazines [5], cytotoxic prenyl asteltoxin derivatives [6], and anti-virus highly oxygenated cyclopiazonic acid-derived alkaloids [7]. However, it has been proven that most of the genes encoding secondary metabolites in fungi are cryptic under certain culture conditions [8]. In recent years, due to the advances in genome sequencing and bioinformatics-based predictions of encoded natural products in BGCs, it has been enabled to obtain structurally-unique secondary metabolites from fungi with the assistance of genomic information [8][9][10].
During our investigation focused on discovering bioactive natural products from marine-derived fungi [11][12][13][14], two known cytotoxic diketopiperazine alkaloids, spirocyclic diketopiperazine alkaloid (4) [15] and cyclotryprostatin B (5) [16], were discovered from the marine-derived fungus Penicillium brasilianum HBU-136 by using rice medium in the first large-scaled fermentation. After gaining these two known diketopiperazine alkaloid derivatives, we tried to locate their BGC in the genome of the fungus HBU-136. A search of the genome sequence using antiSMASH 4.2 found that the strain harbors multiple BGCs for non-ribosomal peptide synthetases (NRPS), one of which shows high similarity to the BGC of fumitremorgins A-C in the fungus Aspergillus fumigatus [17][18][19]. By gene-to-gene comparisons, it is disclosed that the main difference between the BGC in HBU-136 (ctp) against in Aspergillus fumigatus (ftm) lies at the tailoring genes (Table S2 and Figure S45), which suggested that the BGC in HBU-136 may having the ability to produce new derivatives. Based on this information, we tried more fermentation conditions with LC-MS analysis, and finally found three new indole-diketopiperazines (1−3) from the rice medium with 1.0% MgCl 2 fermentation. Herein, we report the isolation, structural characterization, putative biosynthetic relation and biological activities of these compounds.
By gene-to-gene comparison, it suggested that the proposed gene clusters were consistent well with the assembly of the dipeptide skeleton for indole-diketopiperazines ( Figure 1). The key biosynthetic step of dipeptide skeleton started from two successive amidation reactions catalyzed by the ctpNRPS gene and led to the formation of brevianamide F [20,21]. Then ctpPT-1 gene mediated the isopentane addition and delivered the product directly to ctpP450-2 gene, who catalyzed the tetrahydropyridine ring formation, or to ctpOMT gene to get the methoxyl first before the ring-formation step [20,21]. The formed cyclotryprostatin skeleton could be further transformed into spirotryprostatin-type by ctpP450-3 gene via a radical formation and migration eliciting semipinacol-type rearrangement mechanism [22]. P450s, the most versatile biocatalysts in nature, have been proved to catalyze such uncommon reactions during skeleton constructing processes in many cases [23]. After the core skeletons constructed, the structural diversity of both spirotryprostatins and cyclotryprostatins mainly come from the tailoring reactions of ctpP450-3 gene to complete hydroxylation reactions, ctpP450-1 gene to catalyze the benzene hydroxylation, and ctpOMT gene to mediate the methylation reaction. 8), including three new compounds, spirotryprostatin L (1) and cyclotryprostatins F-G (2-3), and five known analogs, spirocyclic diketopiperazine alkaloid (4) [15], cyclotryprostatin B (5) [16], 20hydroxycyclotryprostatin B (6) [25], 12β-hydroxy-13α-ethoxyverruculogen TR-2 (7) [26], and an unnamed compound (8) [27], were then obtained ( Figure 2).

Structure Elucidation
Spirotryprostatin L (1), a yellow amorphous powder, was obtained with the molecular formula C22H25N3O7 (12 degrees of unsaturation) by positive HRESIMS. The IR absorption bands at 3447 cm −1 and 1647 cm −1 indicated the presences of hydroxyl and carbonyl groups. The characteristic UV

Structure Elucidation
Spirotryprostatin L (1), a yellow amorphous powder, was obtained with the molecular formula C 22 H 25 N 3 O 7 (12 degrees of unsaturation) by positive HRESIMS. The IR absorption bands at 3447 cm −1 and 1647 cm −1 indicated the presences of hydroxyl and carbonyl groups. The characteristic UV absorption maxima at 220, 249, 287, 387 nm suggested a spirotryprostatin skeleton for 1 [15,16]. The NMR signals in 1 (Tables 1 and 2) were representative of one keto carbonyl (C-3), two amide carbonyls (C-11 and C-17), one 1,3,4-trisubstituted aromatic unit (C-3a, C-4, C-5, C-6, C-7, and C-7a), and three methyls (one oxygenated) (6-OCH 3 , C-22, and C-21). These structural features suggested a prenylated indole-diketopiperazine nucleus of 1, structurally similar to the known compound 4 which was previously obtained from the fungus Aspergillus fumigatus [15]. A detailed analysis for HMBC correlations of 1 (from 12-OH to C-12 and C-13) ( Figure 3) indicated that an additional hydroxyl group [δ H 5.14 (1H, brs, 12-OH) and δ C 90.2 (C-12)] was attached to C-12. Based on the systematic 2D NMR data analyses, the plane structure of 1 was determined.   In order to assign the relative configuration of 1, its NOESY experiment was carried out. The     Cyclotryprostatin F (2) was also obtained as a yellow amorphous powder, whose molecular formula was deduced as C23H27N3O6 from HRESIMS. The NMR data of 2 ( Table 1; Table 2) suggested that it also possessed a diketopiperazine skeleton, which was considered similar to the known compound 5 [16] through detailed NMR analyses of them. Combining the upfield shifted C- Cyclotryprostatin F (2) was also obtained as a yellow amorphous powder, whose molecular formula was deduced as C 23 H 27 N 3 O 6 from HRESIMS. The NMR data of 2 (Table 1; Table 2) suggested that it also possessed a diketopiperazine skeleton, which was considered similar to the known compound 5 [16] through detailed NMR analyses of them. Combining the upfield shifted C-11 (δ C 166.3 in 2; δ C 167.0 in 5) and C-14 (δ C 19.2 in 2; δ C 22.1 in 5), and downfield shifted C-12 (δ C 86.7 in 2; δ C 59.9 in 5) and C-13 (δ C 36.5 in 2; δ C 29.7 in 5) of 2 with the change of the molecular weight from 5 to 2 suggested that 2 was a 12-hydroxylation derivative of 5. This deduction could be confirmed by the key HMBC correlations from H-13 and H-15 to C-12 in 2 ( Figure 3). Similar 1 H-1 H coupling constants, NOESY correlations and ECD Cotton effects ( Figure 6) of 2 and 5 implied that 2 had the 8S,9S,12R,18S absolute configuration, which was also confirmed by TD-DFT ECD calculation of (8S,9S,12R,18S)-2. It is deduced that, the newborn OH-12 in 1 and 3 may be catalyzed by the multi-functional P450 ctpP450-3 ( Figure 1). And the new formed OMe-8 in 23 may be introduced by the methyltransferase encoded gene ctpOMT or other genes outside the reported boundary, or they were formed during our isolation processes.

Instrumentation
Optical rotations were measured on a JASCO P-1020 digital polarimeter (Jasco Corp., Tokyo, Japan). UV spectra were recorded on a Thermo Scientific Multiskan GO microplate Cyclotryprostatin G (3) was also isolated as an analog of 5, deduced from the similar NMR data of them. Compound 3 was the demethylation derivative of 5 at C-6, which was verified by 1 H-1 H COSY (H-4/H-5/H-6/H-7) and HMBC (H-6 and C-4, and H-6 and C-7a) experiments for 3. The stereochemistry of 3 was suggested to be 8S,9S,12S,18S by comparing the NOESY and ECD spectra of 3 to those of 5 ( Figure 6).
It is deduced that, the newborn OH-12 in 1 and 3 may be catalyzed by the multi-functional P450 ctpP450-3 (Figure 1). And the new formed OMe-8 in 23 may be introduced by the methyltransferase encoded gene ctpOMT or other genes outside the reported boundary, or they were formed during our isolation processes.

Biological Activities Screening
It has been reported that many indole-diketopiperazine derivatives displayed a wide range of biological effects, such as cytotoxic [16,28] andantibacterial activities [29]. In the present study, all of Mar. Drugs 2019, 17, 514 7 of 11 the isolated compounds (1-8) were tested for their cytotoxic and antibacterial activities. Among them, 1 and 4 displayed selective cytotoxicities against HL-60 cell line with the IC 50 values of 6.0 and 7.9 µM, respectively. Whereas compounds 2, 3, and 5 exhibited activities against MCF-7 cell line with the IC 50 values of 7.6, 10.8, and 5.1 µM, respectively. However, all of the metabolites appeared to be inactive in antibacterial and antifungal assays (MIC > 25 µM).

Genome Sequencing and Bioinformatics Analysis
The fungus Penicillium brasilianum HBU-136 was cultured in PDB medium for 30 h at 25 • C, then high-quality genomic DNA was isolated from cultured cells using Fungal DNA Kit (OMEGA Bio-Tek, E.Z.N.A. ® , (Norcross, GA, USA). All the purified DNA was used to construct 350 bp DNA library with DNA library preparation kit (Illumina, CA, USA), and was sequenced on HiSeq X10 instrument, employing paired-end 150 base reads (Mega genomics, CN). De novo assembling of the draft genome was performed by using the software ABySS 2.15 (Vancouver, NA, Canada) and SPAdes 3.11 (St. Petersburg, EU, Russia), with multiple k-mers specified as "-k 33,55,77,99,117" and coverage fold of 160X PE clean reads. All the assemble scaffold sequences were tested by quality assessment tool QUAST 4.0, the last draft genome sequence was filtered by over 500 bp and was determined via the largest N50 number. Gene functions were proposed by the online BLAST program (http://blast.ncbi.nlm.nih.gov/) and Conserved Domain Database (CDD) at the NCBI server (https://www.ncbi.nlm.nih.gov/). This Whole Genome Shotgun project of the fungus HBU-136 has been deposited at DDBJ/ENA/GenBank under the accession SSWP00000000. The version described in this paper is version SSWP01000000.

Fungal Material
The fungus Penicillium brasilianum HBU-136 was collected from the Bohai Sea (Huanghua, Hebei Province, China, June 2016). The strain with the NCBI GenBank accession number MH377073 was deposited in the College of Pharmaceutical Sciences, Hebei University, China.

Biological Assay
Cytotoxic Assay. The cytotoxic activity of compounds 1-8 were evaluated in vitro by the MTT method [30], with the concentration of 10 µM. Three human tumor cell lines were used, including human promyelocytic leukemia cell line (HL-60), human colorectal cancer cell line (HCT-116), and human breast cancer cell line (MCF-7), with cisplatinum (DDP) as a positive control.
Antibacterial Assay. The antibacterial activity of the isolated compounds was determined using the conventional broth dilution assay [31] with ciprofloxacin (CIP) as a positive control.

ECD Spectrum Measurement and Calculation
The tested compounds 1-5 were dissolved in chromatographic methanol at the concentration of 0.5 mg/mL and transferred 300 µL to quartz cuvette (1 cm for width) for tested. First tested the ECD spectrum of methanol as the blank control, then tested for compounds 1-5. The wavelength was set at 190-400 nm with the band width of 1 nm, and Nitrogen should be used throughout the experiment at the flow rate of 3 mL/min. The ECD spectrum was calculated by the TDDFT methodology with a larger basis set at the B3LYP/6-311++G(2d,p) level and simulated using SpecDis 1.62 [32] according to Boltzmann distributions.

Conclusions
In summary, three new indole-diketopiperazine derivatives (1-3) showed selective cytotoxicities have been obtained from the marine-derived fungus P. brasilianum HBU-136 with the aid of genomic analysis. ECD quantum chemistry calculations were carried out to assign the absolute configurations of the new compounds 1-3. The present research indicated that it is a powerful approach to discover new natural products from marine-derived fungi by the combination of chemical and bioinformatics analyses.