A New Citrinin Derivative from the Indonesian Marine Sponge-Associated Fungus Penicillium citrinum.

Sponge-associated fungi are attractive targets for the isolation of bioactive natural products with different pharmaceutical purposes. In this investigation, 20 fungi were isolated from 10 different sponge specimens. One isolate, the fungus Penicillium citrinum strain WK-P9, showed activity against Bacillus subtilis JH642 when cultivated in malt extract medium. One new and three known citrinin derivatives were isolated from the extract of this fungus. The structures were elucidated by 1D and 2D NMR spectroscopy, as well as LC-HRMS. Their antibacterial activity against a set of common human pathogenic bacteria and fungi was tested. Compound 2 showed moderate activity against Mycobacterium smegmatis ATCC607 with a minimum inhibitory concentration (MIC) of 32 µg/mL. Compound 4 exhibited moderate growth inhibition against Bacillus subtilis JH642, B. megaterium DSM32, and M. smegmatis ATCC607 with MICs of 16, 16, and 32 µg/mL, respectively. Furthermore, weak activities of 64 µg/mL against B. subtilis DSM10 and S. aureus ATCC25923 were observed for compound 4.

In the COSY spectrum, two spin systems are observed. One of them consists of H-9, H-1, H-2, and H-10, while the other one comprises H-9 , H-1 , H-2 , and H-10 . This is also confirmed by HMBC correlations. For the first molecular fragment, the HMBC correlations range from H-1 to C-3 and C-8, from H-2 to C-3, C-8, and C-10, from H-9 to C-1, C-2, and from H-10 to C-1, C-2 and C-3. Correspondingly, the second fragment shows HMBC correlations from H-1 to C-3 and C-8 , from H-2 to C-3 , C-8 , and C-10 , from H-9 to C-1 , C-2 , and from H-10 to C-1 , C-2 and C-3 .
Further HMBC correlations from H-6 to C-5, C-7, and C-8, as well as the correlations from H-11 to C-3, C-4, C-5, and C-8 established the first monomeric unit as shown in Figure 3 (black). Accordingly, the second monomeric unit (Figure 3, grey) is confirmed by HMBC correlations from H-6 to C-5 , C-7 , and C-8 , as well as the correlations from H-11 to C-3 , C-4 , C-5 , and C-8 .
As can be concluded from the HMBC correlations from H-11 to C-6 , from H-6 to C-3 , C-4 and C-11 , from H-11 to C-6, and from H-6 to C-3, C-4 and C-11, the two monomeric units are connected via bonds between C-4 and C-6 , as well as C-6 and C-4 . Therefore, compound 1 was elucidated as a new natural product, for which we propose the name penicitrinone G.
The relative configuration was determined based on NOE analysis together with the coupling constant. The correlations from H-6 to H-11 , and from H-6 to H-11 indicated that both H-6 and H-6 , as well as H-11 H-11 are in equatorial position, which established the central ring as chair configuration. The relative configuration of both dihydrofuran rings was determined mainly by comparing the coupling constant with the reported compounds and simulated 1 H NMR spectra of trans and cis stereoisomers [24]. The stereochemistry of H-1 and H-2 was determined as trans configuration. This assignment was based on the coupling constant of compound 1 with J 1,2 = 3.6 Hz, which is close to the reported penicitrinone E (J 2 ,3 = 4.3 Hz) as well as the simulated 1 H NMR spectrum (J = 4.0 Hz) [24]. However, the relative configuration of H-1 and H-2 was elucidated to be cis due to the coupling constant of J 1 ,2 = 6.4 Hz, which was more similar to the coupling constant for the simulated cis configuration (J = 8.2 Hz) [24]. The different configurations of those two dihydrofuran rings can also be confirmed by the NOE correlations. For the first dihydrofuran ring NOE correlations range from both H-11 and H-6 to H-2, while for the other dihydrofuran ring, there is a NOE correlation from H-6 to H-10 instead of H-2. Therefore, the relative configuration of compound 1 was assigned as shown in Figure 3. analysis of the COSY, heteronuclear single quantum correlation (HSQC), and heteronuclear multiple bond correlation (HMBC) spectra (Table 1, Figure S3-S5).
In the COSY spectrum, two spin systems are observed. One of them consists of H-9, H-1, H-2, and H-10, while the other one comprises H-9', H-1', H-2', and H-10'. This is also confirmed by HMBC correlations. For the first molecular fragment, the HMBC correlations range from H-1 to C-3 and C-8, from H-2 to C-3, C-8, and C-10, from H-9 to C-1, C-2, and from H-10 to C-1, C-2 and C-3.
As can be concluded from the HMBC correlations from H-11 to C-6', from H-6 to C-3', C-4' and C-11', from H-11' to C-6, and from H-6' to C-3, C-4 and C-11, the two monomeric units are connected via bonds between C-4 and C-6', as well as C-6 and C-4′. Therefore, compound 1 was elucidated as a new natural product, for which we propose the name penicitrinone G.
The relative configuration was determined based on NOE analysis together with the coupling constant. The correlations from H-6 to H-11', and from H-6′ to H-11 indicated that both H-6 and H-6′, as well as H-11 are in equatorial position, which established the central ring as chair configuration. The relative configuration of both dihydrofuran rings was determined mainly by comparing the coupling constant with the reported compounds and simulated 1 H NMR spectra of trans and cis stereoisomers [24]. The stereochemistry of H-1 and H-2 was determined as trans configuration. This assignment was based on the coupling constant of compound 1 with J1,2 = 3.6 Hz, which is close to the reported penicitrinone E (J2′,3′ = 4.3 Hz) as well as the simulated 1 H NMR spectrum (J = 4.0 Hz) [24]. However, the relative configuration of H-1′ and H-2′ was elucidated to be cis due to the coupling constant of J1′,2′ = 6.4 Hz, which was more similar to the coupling constant for the simulated cis configuration (J = 8.2 Hz) [24]. The different configurations of those two dihydrofuran rings can also be confirmed by the NOE correlations. For the first dihydrofuran ring NOE correlations range from both H-11 and H-6′ to H-2, while for the other dihydrofuran ring, there is a NOE correlation from H-6 to H-10' instead of H-2. Therefore, the relative configuration of compound 1 was assigned as shown in Figure 3.

Bioactivity
In order to gain insight into the biological activity of compounds 1-4, they were tested against a panel of eight different bacteria (Bacillus megaterium DSM32, Bacillus subtilis JH642, Bacillus subtilis DSM10, Micrococcus luteus ATCC 4698, Mycobacterium smegmatis ATCC607, Listeria monocytogenes DSM20600, Staphylococcus aureus ATCC25923, Escherichia coli K12), as well as against the yeast Candida albicans FH2173, and the mold fungus Aspergillus flavus ATCC9170. The results of the bioactivity assays are summarized in Table 2. Only compounds 2 and 4 revealed moderate activities against Gram-positive strains, while no effect was observed against Gram-negative bacteria and fungi.

Bioactivity
As can be seen in Table 2, penicitrinol J (4) exhibited moderate activity against B. megaterium, B. subtilis JH642, and M. smegmatis (with corresponding MICs of 16, 16, and 32 µg/mL), whereas penicitrinone A (2) only showed weak activity against M. smegmatis. The activity against M. smegmatis was of the same magnitude for compounds 2 and 4 (MIC of 32 µg/mL, respectively). On the other hand, penicitrinone E (3) and the new compound 1 were inactive against the chosen test strains. Comparing the structures of compounds 2 and 3, the only difference is an additional carboxylic acid group in compound 3, which might imply that the carboxylic acid group slightly decreases the antibacterial activity. A likewise structural comparison of compounds 3 and 4 in relation to their bioactivities, suggests that phenolic systems (4) provide increased antibacterial potency in contrast to chinoid systems (3). In this case, the benefit of the aromatic system even seems to outweigh the effect of the carboxylic acid function. The increased activity of penicitrinol J (4) as opposed to penicitrinone E (3) was previously observed in a paper diffusion assay, but not quantified by determination of the respective MIC values [26]. The idea, that the antibacterial activity might be related to the presence of a phenolic system can be supported by the results of Yang et al. [27], who reported that penicitrinol A shows superior antibiotic activity to penicitrinone A (2) when tested against M. luteus and E. coli. Whether the difference in bioactivity of compounds 3 and 4 is to be attributed to structural factors or to their different redox properties still needs to be investigated. For the related compound citrinin, its redox properties seem to play an important role for its bioactivity, especially for its reported anti-fungal effect [28][29][30]. In contrast to previous reports, in the present study no activity could be detected for penicitrinone A (2) against M. luteus [27], E. coli [27] or B. megaterium [31].The lack of anti-fungal activity determined for compound 2 against C. albicans and A. flavus on the other hand, is in accordance with earlier findings [25].

Biosynthesis
Due to the interesting structure of the new compound 1, we were wondering what type of reactions are involved in its biosynthesis. To hypothesize a biosynthetic pathway for compound 1, first the proposed biosyntheses of penicitrinone and penicitrinol derivatives were analyzed [26,[32][33][34]. In contrast to the biosynthesis of most citrinin derivatives, which include a Diels-Alder reaction, we propose a radical pathway for the formation of compound 1. This assumption is based on a similar mechanism reported for the formation of the structurally related compound dibefurin [35] (Figure 4). We assume that the key precursor for the formation of compound 1 is 2,3,4-trimethyl-5,6,7-trihydroxy-2,3-dihydrobenzofuran. Even though this substance itself is not known to date, it constitutes one of the monomers forming penicitol B [36]. The key intermediate might arise from the oxidation of 2,3,4-trimethyl-5,7-dihydroxy-2,3-dihydrobenzofuran, a compound frequently found in extracts of P. citrinum species [25,26,37]. This substance in turn, could either be produced directly via the polyketide pathway, as shown in Figure 4, or might alternatively be a decomposition product of citrinin as proposed by Clark et al. [24]. Starting point of our hypothesis is the pentaketide I, which is C-methylated by a radical SAM enzyme and then yields after aldol condensation and keto-enol tautomerization the enzyme bound aromatic compound IV. Labelling experiments for the elucidation of the citrinin biosynthesis [33] seem to support our assumption of C-methylation of an unbranched pentaketide. Reductive release from the enzyme [33], yielding the aromatic aldehyde V is also suggested in analogy to the biosynthesis of citrinin. After reduction of the ketone to the alcohol, we speculate that the aldehyde might undergo a Dakine type oxidation, thus producing a phenol (VIII). Condensation of VIII affords the heterocyclic compound IX, which after oxidation furnishes the precursor X. Selective formation of a five-membered ring instead of a six-membered ring might either be ascribed to the fact that formation of five-membered rings is kinetically favoured over the formation of six-membered rings or alternatively an enzyme might promote the selectivity of the reaction. Another route for the biosynthesis of X could proceed by the same pathway outlined for citrinin [33] with the crucial intermediate being 2,4-dihydroxy-5-methyl-6-(3-oxobutan-2-yl)isophthalaldehyde. A twofold Dakin type oxidation of this aromatic dialdehyde, followed by condensation might form X. Yet, as the twofold Dakin type oxidation would lead to a tetrahydroxylated aromatic compound, we deem this pathway less likely. Compound X finally undergoes a radical dimerization process as outlined in detail in Figure 4, with the sequence of oxidation and reduction reactions yielding the isolated compound 1.
methylation of an unbranched pentaketide. Reductive release from the enzyme [33], yielding the aromatic aldehyde V is also suggested in analogy to the biosynthesis of citrinin. After reduction of the ketone to the alcohol, we speculate that the aldehyde might undergo a Dakine type oxidation, thus producing a phenol (VIII). Condensation of VIII affords the heterocyclic compound IX, which after oxidation furnishes the precursor X. Selective formation of a five-membered ring instead of a six-membered ring might either be ascribed to the fact that formation of fivemembered rings is kinetically favoured over the formation of six-membered rings or alternatively an enzyme might promote the selectivity of the reaction. Another route for the biosynthesis of X could proceed by the same pathway outlined for citrinin [33] with the crucial intermediate being 2,4-dihydroxy-5-methyl-6-(3oxobutan-2-yl)isophthalaldehyde. A twofold Dakin type oxidation of this aromatic dialdehyde, followed by condensation might form X. Yet, as the twofold Dakin type oxidation would lead to a tetrahydroxylated aromatic compound, we deem this pathway less likely. Compound X finally undergoes a radical dimerization process as outlined in detail in Figure 4, with the sequence of oxidation and reduction reactions yielding the isolated compound 1.

Fungal Isolation and Purification
The isolation of sponge-associated fungi was done according to Kjer's protocol [39]. Sponges were initially sprayed with sterile natural seawater, cut into three pieces with an approximate size of 1 cm 2 and placed into malt extract agar (MEA) (Himedia, Mumbai, India) medium (30 g malt extract, 5 g mycological peptone, 15 g agar, and 1000 mL sterile natural seawater, hereinafter referred to as marine MEA). The agar plates were incubated at room temperature (25 • C) for three days. The initial selection of fungal colonies was done based on phenotype, e.g., colony morphology and color [40]. In total, 20 fungal colonies were isolated and propagated until axenic cultures were obtained from ten sponges. In an initial screening, using the agar plug method, five strains were active and strain WK-P9 was selected for further investigation due to its prominent activity.

Molecular Identification of the Fungus
DNA amplification was performed using the Toyobo KOD FX Neo kit (Toyobo, Osaka, Japan). Fungus WK-P9 was grown in marine MEA for 3 days; a loop of mycelia was transferred into 2 µL PCR grade water, which served as template for colony PCR. The PCR mixture was set up in a total volume of 50 µL as follows: PCR grade water 9 µL, 2× PCR buffer for KOD FX Neo 25 µL, 2mM dNTPs 10 µL D, primer ITS1 (5'-tccgtaggtgaacctgcgg-3', 10 pmol/µL) 1.5 µL, primer ITS4 (5'-tcctccgcttattgatatgc-3', 10 pmol/µL) 1.5 µL, DNA template 2 µL (from mycelia), KOD FX Neo (1.0 U/µL) 1 µL. On a T100™ Thermal Cycler from Bio-Rad (Feldkirchen, Germany), the PCR program was run as follows: Denaturation initially at 95 • C for 3 min, followed by 35 cycles (denaturation at 95 • C for 3 min, annealing at 55 • C for 45 s, and extension at 72 • C for 1 min), then 72 • C extension for 7 min and cooling to 16 • C. The PCR product was Sanger-sequenced (1st BASE DNA sequencing services). Then, MEGA X was used to generate an alignment and the phylogenetic tree using the sequence data obtained. The phylogenetic tree was constructed using the Maximum likelihood and Neighbor-Joining analysis, with 1000 replications of bootstrap value ( Figure S8). The sequence data was submitted to GenBank (Acc. Number: LC371661.1).

Isolation and Structure Elucidation
Three pieces of P. citrinum WK-P9 (each 1 × 1 cm 2 ), which nearly covered the whole surface of a Petri dish, were inoculated onto malt extract broth (Himedia, Mumbai, India) media diluted by sterilized natural seawater (12 L) under clean-bench conditions, and cultivated for 12 days at 24 • C. The crude extract of P. citrinum WK-P9 (4.0 g) was collected after soaking the medium with ethyl acetate (EtOAc) using the ratio 1:3. The purification was initially performed by Silica (Wako, Japan) vacuum liquid chromatography (VLC) using a gradient composed of three different solvent systems-n-hexan:EtOAc

Antibacterial Susceptibility Tests
The antimicrobial activity of the crude extract and the following sub-fractions was conducted using the agar diffusion method [41]. The respective test bacteria (Escherichia coli K12, Bacillus megaterium DSM32, Bacillus subtilis JH642, Micrococcus luteus ATCC4698) were spread on Luria Bertani (LB) agar plates (10 g peptone, 5 g yeast extract, 5 g NaCl, 15 g agar, mixed with 1 L distilled water). For sample preparation, 15 µL of each crude extract (10 mg/mL dissolved in methanol) or fraction were added to a paper disk. Methanol was used as the negative control and carbenicillin (5 µL of a 50 mg/mL stock solution) (Carl Roth GmbH + Co., Karlsruhe, Germany) was used as positive control. The dried paper disks were subsequently positioned on the agar plate and incubated at 30 • C overnight. The diameter of the resulting inhibition zone was determined.
Determination of the minimum inhibitory concentrations (MIC) of purified compounds 1-4 was carried out by micro broth dilution assays in 96 well plates. All compounds were dissolved in Mar. Drugs 2020, 18, 227 9 of 12 dimethyl sulfoxide (DMSO, Carl Roth GmbH + Co., Karlsruhe, Germany) and tested in triplicate. For B. subtilis DSM10 and S. aureus ATCC25923, an overnight culture (37 • C, 180 rpm) was diluted to 5 × 10 5 cells/mL in cation adjusted Mueller Hinton II medium (Becton Dickinson, Sparks, NV, USA). L. monocytogenes DSM20600 was incubated for 2 days before the assay inoculum was adjusted using the same medium and growth conditions (Mueller Hinton II medium). As positive controls, dilution series of rifampicin, tetracycline and gentamycin (all Sigma Aldrich, St. Louis, MS, USA) were prepared (64-0.03 µg/mL). Cell suspensions without test sample or antibiotic control were used as negative controls. After incubation (18 h and 48 h for L. monocytogenes, 37 • C, 180 rpm, 80% rH) cell growth was assessed by turbidity measurement with a microplate spectrophotometer at 600 nm (LUMIstar®Omega BMG Labtech, Ortenberg, Germany).
The pre culture of M. smegmatis ATCC607 was incubated in Brain-Heart Infusion broth (Becton Dickinson) supplemented with Tween 80 (1.0% (v/v)) for 48 h at 37 • C and 180 rpm before the cell concentration was adjusted in cation adjusted Mueller Hinton II medium. Isoniazid (Sigma-Aldrich) was used instead of gentamycin as a third positive control. Cell viability was evaluated after 48 h (37 • C, 180 rpm, 80% rH) via ATP quantification (BacTiter-Glo™, Promega, Madison, WI, USA) according to the manufacturer's instructions.
C. albicans FH2173 was incubated for 48 h at 28 • C and 180 rpm before the pre culture was diluted to 1 × 106 cells/mL in cation adjusted Mueller Hinton II medium. For A. flavus ATCC9170, a previously prepared spore solution was used to prepare the assay inoculum of 1 × 105 spores/mL. Yeast and mold assays were incubated for 48 h at 37 • C, 180 rpm and 80% rH. For both, tebuconazole (Cayman Chemical Company, Ann Arbor, MI, USA), amphotericin B and nystatin (both Sigma Aldrich) were used as positive control (64-0.03 µg/mL). The readout was carried out by ATP quantification.

Conclusions
In summary, by applying bioassay-guided fractionation, four citrinin derivatives were obtained from the sponge-associated fungus P. citrinum WK-P9. Among those substances, the new derivative 1 was characterized for the first time, revealing a connection of monomers which has so far been unprecedented for citrinin derivatives. To provide an explanation for the formation of this new derivative, a biosynthetic hypothesis was proposed. Furthermore, all isolated compounds were screened for bioactivity. In this respect, penicitrinol J (4) showed moderate antimicrobial activity against B. subtilis JH642, B. megaterium DSM32, and M. smegmatis ATCC607. Even though P. citrinum is a well-investigated fungal species, such strains still have an inherent high possibility to deliver new specialized metabolites, as exemplified by the isolation of the new compound 1. New strains in combination with variation in culture conditions, e.g., using the one strain many compounds (OSMAC) approach, thus represent a promising bioresource for natural product discovery.