Chromone Derivatives and Other Constituents from Cultures of the Marine Sponge-Associated Fungus Penicillium erubescens KUFA0220 and Their Antibacterial Activity

A previously unreported chromene derivative, 1-hydroxy-12-methoxycitromycin (1c), and four previously undescribed chromone derivatives, including pyanochromone (3b), spirofuranochromone (4), 7-hydroxy-6-methoxy-4-oxo-3-[(1E)-3-oxobut-1-en-1-yl]-4H-chromene-5-carboxylic acid (5), a pyranochromone dimer (6) were isolated, together with thirteen known compounds: β-sitostenone, ergosterol 5,8-endoperoxide, citromycin (1a), 12-methoxycitromycin (1b), myxotrichin D (1d), 12-methoxycitromycetin (1e), anhydrofulvic acid (2a), myxotrichin C (2b), penialidin D (2c), penialidin F (3a), SPF-3059-30 (7), GKK1032B (8) and secalonic acid A (9), from cultures of the marine sponge- associated fungus Penicillium erubescens KUFA0220. Compounds 1a–e, 2a, 3a, 4, 7–9, were tested for their antibacterial activity against Gram-positive and Gram-negative reference and multidrug-resistant strains isolated from the environment. Only 8 exhibited an in vitro growth inhibition of all Gram-positive bacteria whereas 9 showed growth inhibition of methicillin-resistant Staphyllococus aureus (MRSA). None of the compounds were active against Gram-negative bacteria tested.


Introduction
The fungi of the genus Penicillium (Family Aspergillaceae) are the most common fungi occurring in a diverse range of habitats from soil to vegetation to various food products, air, indoor environments, and marine environments. They have a worldwide distribution and a large economic impact on human life [1]. The marine-derived Penicillium species can be found to be associated with a variety of marine invertebrates such as marine sponges, corals, and tunicates, as well as with fish, marine algae, mangroves and also from the sediments; although sediments and sponges are their main sources or hosts for producing new marine natural products. Interestingly, marine-derived Penicillium species produce diverse structural classes of secondary metabolites such as polyketides, sterols, terpenoids, alkaloids, among others, and more than half of these metabolites exhibited bioactivities [2].
Thus, in our ongoing search for antibiotics from marine-derived fungi from the tropical sea, we investigated secondary metabolites from cultures of Penicillium erubescens KUFA 0220, which was isolated from the marine sponge Neopetrosia sp., collected from the coral reef at Samaesan Island, Chonburi province, in the Gulf of Thailand.
Compounds 1a-e, 2a, 3a, 4, 7-9 were tested for their antibacterial activity against five reference bacterial strains consisting of three Gram-positive (Staphylococcus aureus ATCC 29213, Enterococcus faecium ATCC 19434 and Enterococcus faecalis ATCC 29212) and two Gram-negative bacteria (Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853), three multidrug-resistant isolates from the environment (MRSA S. aureus 66/1, VRE E. faecium 1/6/63 and E. faecalis B3/101) and a clinical isolate ESBL E. coli SA/2. Some of the isolated compounds were also investigated for their capacity to inhibit biofilm formation in the four reference strains as well as for their potential synergism with the clinically used antibiotics against multidrug-resistant isolates from the environment. Mar. Drugs 2018, 16
Compound 1c was isolated as a white solid (mp 232-233 • C), and its molecular formula C 14 H 12 O 6 was established based on its (+)-HRESIMS m/z 277.0715 [M + H] + , (calculated 277.0712 for C 14 H 13 O 6 ), indicating nine degrees of unsaturation. The IR spectrum showed absorption bands for the hydroxyl (3420 cm −1 ), conjugated ketone carbonyl (1662 cm −1 ), aromatic (1627, 1555 cm −1 ), and ether (1270 cm −1 ) groups. The 13 C NMR spectrum of 1c (Table 1, Figure S11) displayed fourteen carbon signals which, according to DEPTs and HSQC spectra (Table 1, Figure S12), can be classified as one conjugated ketone carbonyl (δ C 174.8), seven quaternary sp 2 (δ C 167.3, 155.2, 152.2, 151.9, 143.6, 111.2, 105.9), three methine sp 2 (δ C 110.7, 106.5, 104.1), two oxymethylene sp 3 (δ C 62.5 and 59.5), and one methoxyl (δ C 56.4) carbon. The 1 H NMR spectrum (Table 1, Figure S10) showed two aromatic singlets at δ H 7.15 and 6.44, another singlet of one olefinic proton at δ H 6.25, two singlets of oxymethylene protons at δ H 5.02 (2H) and 4.41 (2H), and a singlet of methoxyl protons at δ H 3.80 (3H). The general features of the 1 H and 13 C NMR spectra of 1c resembled those of 12-methoxycitromycin (1b), which was previously isolated from the Australian marine-derived and terrestrial Penicillium spp. [5], and also isolated in this work. The only difference between the two compounds is the methyl group in 1b (δ H 2.34, d, J = 0.6 Hz; δ C 19.2) is replaced by a hydroxymethyl group (δ H 4.41; δ C 59.5) in 1c. The position of the methoxyl group was also confirmed by the NOESY correlation from the methoxyl protons to H-13 (δ H 7.15, s) ( Figure S14). Therefore, 1c is 1-hydroxy-12-methoxycitromycin. The literature search revealed that 1c has never been previously reported. The analysis of the 1 H, 13 C NMR (Table 2, Figures S28 and S29) and the (+)-HRESIMS spectra of 3a revealed that its planar structure was the same as that of penialidin F, previously isolated from the culture of Penicillium janthinellum DT-F29, collected from marine sediments [9]. Curiously, even though the authors reported the optical rotation of penialidin F as levorotatory ([α] 25 D −4.13, c = 1.0, MeOH), they did not determine the absolute configuration of its stereogenic carbon (C-2). Similarly, we have also found the optical rotation of the 3a levorotatory, ([α] 25 D −7.5, c = 0.04, MeOH). Since 3a was not isolated as a suitable crystal for X-ray analysis, its calculated ECD spectrum was performed to compare with the experimental ECD spectrum. Therefore, the conformational analysis of 3a by molecular mechanics (MM2 and MMFF95 force fields) focused on combinations of hydroxyl 120 • rotations and two 3,6-dihydro-2H-pyran-2-ol ring conformations. A total of 30 conformations were energetically minimized and ranked using a faster DFT model (smaller basis set, APFD/6-31G). The lowest three of these, representing 99% of the model Boltzmann population, were then further energetically minimized with a larger basis set (APFD/6-311+G(2d,p)). The most stable conformation is depicted in Figure 2 and represents 64% of the Boltzmann population while the other two amount to 25% and 11%.   These three models were then used to calculate the expected Boltzmann-averaged ECD spectrum of 3a's R enantiomer. The good fit between the calculated and experimental ECD spectra shown in Figure 3 is enough evidence to conclude that 3a is the R enantiomer. However, the weak experimental ECD signal of 3a could indicate that this compound does not exist as a pure R enantiomer but as an enantiomeric mixture with an excess of the R enantiomer. These three models were then used to calculate the expected Boltzmann-averaged ECD spectrum of 3a's R enantiomer. The good fit between the calculated and experimental ECD spectra shown in Figure 3 is enough evidence to conclude that 3a is the R enantiomer. However, the weak experimental ECD signal of 3a could indicate that this compound does not exist as a pure R enantiomer but as an enantiomeric mixture with an excess of the R enantiomer.
Compound 3b was isolated as a 1:2 mixture (estimated from the integration of the proton signals in the 1 H NMR spectrum) with myxotrichin C (2b  Figure S22) exhibited fourteen carbon signals which, in combination with DEPTs and HSQC spectra ( Figure S24), can be categorized as one conjugated ketone carbonyl (δ C 173.2), six quaternary sp 2 (δ C 157.9, 152.1, 150. 8, 144.4, 115.4, 113.0), two methine sp 2 (δ C 107.4 and 102.7), one ketal (δ C 97.7), one oxymethylene sp 3 (δ C 57.1), one methylene sp 3 (δ C 37.1), one methyl (δ C 22.4), and one methoxyl (δ C 48.3) carbon. The 1 H NMR spectrum ( Figure S21) displayed two aromatic singlets at δ H 7.26 (H-13) and 6.83 (H-10), two pairs of geminally coupled methylene protons at δ H 4.52, dd (J = 14.9, 0.9 Hz)/4.22, dt, (J = 4.9, 2.1 Hz) and 2.63, dd (J = 17.6, 1.5 Hz)/2.96, dd (J = 1.76, 2.6 Hz), a methyl singlet at δ H 1.44 and a methoxyl singlet at δ H 3.21. Comparison of the 1 H and 13 C data of 3b with those of 3a (Table 2) led to the conclusion that 3b is a methyl ketal of 3a. This hypothesis is confirmed not only by the molecular formula of 3b, which is 14 amu more than that of 3a, but also by the HMBC correlation ( Figure S25) of the singlet of the methoxyl protons to the ketal carbon at δ C 97.7. Therefore, 3b was named penialidin G. Mar. Drugs 2018, 16, x 6 of 24 Compound 3b was isolated as a 1:2 mixture (estimated from the integration of the proton signals in the 1 H NMR spectrum) with myxotrichin C (2b). Based on the (+)-HRESIMS m/z 279.0878 [M + H] + , (calculated 277.0869 for C14H15O6), the molecular formula C14H14O6 was attributed to 3b. The 1 H and 13 C NMR spectra of 3b (a minor compound) resembled those of penialidin F (3a) ( Table 2). The 13 C NMR spectrum of 3b (Table 2, Figure S22) exhibited fourteen carbon signals which, in combination with DEPTs and HSQC spectra ( Figure S24 (Table 2) led to the conclusion that 3b is a methyl ketal of 3a. This hypothesis is confirmed not only by the molecular formula of 3b, which is 14 amu more than that of 3a, but also by the HMBC correlation ( Figure S25) of the singlet of the methoxyl protons to the ketal carbon at δC 97.7. Therefore, 3b was named penialidin G.
Surprisingly, the ECD spectrum of the mixture of myxotrichin C (2b) and 3b did not exhibit any Cotton effects. Consequently, we concluded that 3b is a mixture of both enantiomers.
The biogenesis of 2b, 3a, and 3b can be hypothesized as originated from the hexaketide intermediate (i) (Figure 4). Enzyme-catalyzed nucleophilic addition of the primary hydroxyl group to the ketone carbonyl led to a cyclization to form a 2-methyl-3,6-dihydro-2H-pyran-2-ol ring, through an intermediate (ii), in 3a (2R). Dehydration of the hemiketal in 3a furnished myxotrichin C (2b), which underwent a nucleophilic addition of methanol (chromatographic solvent) at C-2 to form an enantiomeric mixture of 3b. Therefore, 3b can be an artifact and not a natural product. The cooccurrence of 2b and 3b can be a concrete proof of this hypothesis. Surprisingly, the ECD spectrum of the mixture of myxotrichin C (2b) and 3b did not exhibit any Cotton effects. Consequently, we concluded that 3b is a mixture of both enantiomers.
Compounds 1a-e, 2a, 3a, 4, 7-9 were evaluated for their antibacterial activity against Gram-negative and Gram-positive bacteria by disc diffusion method, and the MIC and MBC of several reference strains and multidrug-resistant isolates from the environment were also determined. In the disc diffusion assay, a halo of growth inhibition for all Gram-positive bacteria exposed to 8 (Table 7) and for methicillin-resistant Staphylococcus aureus (MRSA) 66/1 exposed to 9 was detected. However, in the range of concentrations tested, it was only possible to determine MICs for 8 (Table 7), with MIC values of 8 mg/mL for E. faecalis ATCC 29212 and vancomycin-resistant E. faecalis (VRE) B3/101, 16 mg/mL for E. faecium ATCC 19434, and 32 mg/mL for E. faecium 1/6/63 (VRE) and S. aureus ATCC 29213. While it was not possible to determine the MBC for the other Gram-positive strains, the MBC for S. aureus ATCC 29213 was 64 mg/mL (Table 7). These results suggested that 8 might have a bacteriostatic effect. The ability of the tested compounds to prevent biofilm formation was evaluated on four reference strains by measuring the total biomass. For 8, four concentrations ranging from 2 × MIC to 1 4 MIC were tested against E. faecalis ATCC 29212, E. faecium ATCC 19434 and S. aureus ATCC 29213. For the other compounds, since it was not possible to determine their MIC values, the highest concentration tested in the previous assays was used. The results were interpreted using a comparative classification that divides adherence capability of tested strains into four categories: (i) non-adherent, (ii) weakly adherent, (iii) moderately adherent, and (iv) strongly adherent [19]. OThe optical density cut-off value (ODc) for each microtiter plate was defined as three standard deviations above the mean OD of the negative control. The use of this classification, which uses the negative control as the starting point instead of using the positive control as a reference, reduces the risk of inconsistencies due to external factors that influence biofilm production [20]. The tested compounds did not inhibit the biofilm formation of S. aureus ATCC 29213, E. coli ATCC 25922, and P. aeruginosa ATCC 27853. However, the biofilm forming ability of E. faecalis ATCC 29212, which is classified as a strong biofilm producer, was impaired by 8 (MIC and 2 × MIC) and 9 (Table 8). On the other hand, 8 was able to increase the biofilm production of a weak biofilm producer E. faecium ATCC 19434.  [19], Average OD value for negative control was found to be 0.055 ± 0.002, therefore the optical cut-off value (ODc) is equal to 0.055 + (3 × 0.002) = 0.061; 2 × ODc = 0.122; 4 × ODc = 0.244.
The screening of a potential synergy between the tested compounds and clinically relevant antimicrobial drugs revealed a slight synergy, as determined by the disc diffusion assay (Table 9). Compound 1b, in combination with cefotaxime (CTX), resulted in a small synergistic effect, as seen by a small increment in the zone of inhibition when compared to the inhibition halo of CTX alone in E. coli SA/2, an extended-spectrum β-lactamase producer (ESBL). A similar effect was observed for VRE E. faecalis B3/101 when 8 was combined with VAN. These results were confirmed by the checkerboard method or by determining the MIC for each antibiotic in the presence of a fixed concentration of each compound when it was not possible to determine a MIC value for the test compound. In the latter, the concentration of each compound used was the highest concentration tested in previous assays which did not inhibit the growth of the four multidrug-resistant strains under study. The effects observed using the disc diffusion assay were not replicated, however, when VRE E. faecalis B3/101 was exposed to 1d, 3a and 9, there was a two-fold reduction in the MIC of VAN. On the other hand, when ESBL E. coli SA/2 was exposed to 1c and 7, there was at least a two-fold increase in the MIC of CTX. When VRE E. faecium 1/6/63 was exposed to 9, there was a two-fold reduction in the MIC of VAN. On the contrary, when it was exposed to 1e, there was at least a two-fold increase in the MIC of VAN (Table 9). The differences in the results obtained using both techniques may be explained by different diffusion rates of each compound in the agar plates.
Thus, in terms of antibacterial activity, 8 is the most promising. Even though no synergy with VAN or OXA was found, this compound alone exhibited an antibiofilm activity against E. faecalis and antibacterial activity against the reference S. aureus, E. faecalis, and E. faecium strains. Most importantly, 8 showed antibacterial activity against both vancomycin-resistant E. faecalis and vancomycin-resistant E. faecium strains, a pathogen classified by WHO as high priority for the research and development of new antibiotics [21]. These results call for a more in-depth study of this compound.

General Experimental Procedures
The melting points were determined on a Stuart Melting Point Apparatus SMP3 (Bibby Sterilin, Stone, Staffordshire, UK) and are uncorrected. Optical rotations were measured on an ADP410 Polarimeter (Bellingham + Stanley Ltd., Tunbridge Wells, Kent, UK). Infrared spectra were recorded in a KBr microplate in an FTIR spectrometer Nicolet iS10 from Thermo Scientific (Waltham, MA, USA) with a Smart OMNI-Transmission accessory (Software 188 OMNIC 8.3, Thermo Scientific, Waltham, MA, USA). 1 H and 13 C NMR spectra were recorded at ambient temperature on a Bruker AMC instrument (Bruker Biosciences Corporation, Billerica, MA, USA) operating at f300 or 500 and 75 or 125 MHz, respectively. High resolution mass spectra were measured with a Waters Xevo QToF mass spectrometer (Waters Corporations, Milford, MA, USA) coupled to a Waters Aquity UPLC system. A Merck (Darmstadt, Germany) silica gel GF 254 was used for preparative TLC, and a Merck Si gel 60 (0.2-0.5 mm) was used for column chromatography.

Fungal Material
The fungus was isolated from the marine sponge Neopetrosia sp. which was collected, by scuba diving at a depth of 5-10 m, from the coral reef at Samaesan Island (12 • 34 36.64" N, 100 • 56 59.69" E), Chonburi province, Thailand, in April 2014. The sponge was washed with 0.01% sodium hypochlorite solution for 1 min, followed by sterilized seawater three times, and then dried on sterile filter paper under sterile aseptic condition. The sponge was cut into small pieces (5 mm × 5 mm) and placed on Petri dish plates containing 15 mL potato dextrose agar (PDA) medium mixed with 300 mg/L of streptomycin sulfate, and incubated at 28 • C for 7 days. The hyphal tips emerging from sponge pieces were individually transferred onto PDA slants and maintained as pure cultures at Kasetsart University Fungal Collection, Department of Plant Pathology, Faculty of Agriculture, Kasetsart University, Bangkok, Thailand, for further identification. The fungal strain KUFA0220 was identified as Penicillium erubescens, based on morphological characteristics such as colony growth rate and growth pattern on standard media, namely Czapek's agar, Czapek yeast autolysate agar, and malt extract agar. Microscopic characteristics including size, shape and ornamentation of conidiophores and spores were examined under a light microscope. This identification was confirmed by molecular techniques using Internal Transcribed Spacer (ITS) primers. DNA was extracted from young mycelia following a modified Murray and Thompson method [22]. Primer pairs ITS1 and ITS4 [23] were used for ITS gene amplification. PCR reactions were conducted on Thermal Cycler and the amplification process consisted of the initial denaturation at 95 • C for 5 min, 34 cycles at 95 • C for 1 min (denaturation), at 55 • C for 1 min (annealing) and at 72 • C for 1.5 min (extension), followed by final extension at 72 • C for 10 min. The PCR products were examined by agarose gel electrophoresis (1% agarose with 1 × TBE buffer) and visualized under UV light after staining with ethidium bromide. DNA sequencing analyses were performed using the dideoxyribonucleotide chain termination method [24] by Macrogen Inc. (Seoul, Korea). The DNA sequences were edited using the FinchTV software (version 1.4, Geospiza Inc, Seattle, WA, USA) and submitted into the BLAST program for alignment and compared to fungal species in the NCBI database (http://www.ncbi.nlm.nih.gov/). Its gene sequences were deposited in GenBank with accession number KY041867.

Extraction and Isolation
The fungus was cultured for one week at 28 • C in five Petri dishes (i.d. 90 mm) containing 20 mL of potato dextrose agar per dish. The mycelial plugs (5 mm in diameter) were transferred to two 500 mL Erlenmeyer flasks containing 200 mL of potato dextrose broth, and incubated on a rotary shaker at 120 rpm at 28 • C for one week. Fifty 1000 mL Erlenmeyer flasks, each containing 300 g of cooked rice, were autoclaved at 121 • C for 15 min. After cooling to room temperature, 20 mL of a mycelial suspension of the fungus was inoculated per flask and incubated at 28 • C for 30 days, after which 500 mL of ethyl acetate was added to each flask of the moldy rice and macerated for 7 days, and then filtered with Whatman No. 1 filter paper (GE Healthcare UK Limited, Buckinghamshire, UK). The ethyl acetate solutions were combined and concentrated under reduced pressure to yield 160 g of crude ethyl acetate extract which was dissolved in 500 mL of CHCl 3 and then filtered with Whatman No. 1 filter paper. The chloroform solution was then washed with H 2 O (3 × 500 mL) and dried with anhydrous Na 2 SO 4 , filtered and evaporated under reduced pressure to give 112 g of the crude chloroform extract which was applied on a column of silica gel (450 g), and eluted with mixtures of petrol-CHCl 3 and CHCl 3 -Me 2 CO, wherein 250 mL fractions were collected as follow: Frs 1-147 (petrol-CHCl 3 , 1:1), 148-223 (petrol-CHCl 3 , 3:7), 224-230 (petrol-CHCl 3 , 1:9), 231-238 (CHCl 3 ), 239-452 (CHCl 3 -Me 2 CO, combined (40.8 mg) and precipitated in MeOH to give 10 mg of 1e [5]. Sfrs 33-45 were combined (70.1 mg) and purified by TLC (Silica gel G 254 , CHCl 3 :MeOH:HCO 2 H, 9:1:0.01) to give 6 mg of 5. Sfrs 46-56 were combined (65.3 mg) and precipitated in MeOH to give further 7 mg of 5.

Electronic Circular Dichroism (ECD)
Electronic Circular Dichroism (ECD) of 3a and 6 The ECD spectra of 3a and 6 (1.5 mM in methanol) were obtained in a Jasco J-815 CD spectropolarimeter (Jasco, Mary's Court, Easton, MD, USA) with a 0.01 mm cell (40 accumulations for 3a). The dihedral driver and MMFF95 minimizations were done in Chem3D Ultra (Perkin-Elmer Inc., Waltham, MA, USA). All DFT minimizations with model chemistries APFD/6-31G and APFD/6-311+G(2d,p) [25] as well as ECD spectral calculations (TD-APFD) were performed with Gaussian 16W (Gaussian Inc., Wallingford, CT, USA) using an IEFPCM solvation model for methanol. The simulated spectral lines for 3a ( Figure 3) and 6 ( Figure 8) were obtained by summation of Gaussian curves, as recommended in Reference [26]. A line broadening of 0.3 eV was applied to all transitions to generate the calculated line.

X-ray Crystal Structure of 4
A single crystal of 4 was mounted on a cryoloop using paratone. X-ray diffraction data were collected at 290 K with a Gemini PX Ultra equipped with CuK α radiation (λ = 1.54184 Å). The crystal was monoclinic, space group P2 1 /n, cell volume 1245.43(7) Å 3 and unit cell dimensions a = 12.3445(4) Å, b = 7.8088(3) Å and c = 12.9397(5) Å and angle β = 93.165(3) • (uncertainties in parentheses). There are two molecules in the asymmetric unit, one Erubescenschromone A molecule and one water molecule, and the calculated crystal density is 1.495 g/cm −3 . The structure was solved by direct methods using SHELXS-97 and refined with SHELXL-97 [27]. Carbon and oxygen atoms were refined anisotropically. Hydrogen atoms were directly found from difference Fourier maps and were refined freely with isotropic displacement parameters. The refinement converged to R (all data) = 6.32% and wR2 (all data) = 11.26%.
Full details of the data collection and refinement and tables of atomic coordinates, bond lengths and angles, and torsion angles have been deposited with the Cambridge Crystallographic Data Centre (CCDC 1856735).

X-ray Crystal Structure of 5
A single crystal of 5 was mounted on a cryoloop using paratone. X-ray diffraction data were collected at 290 K with a Gemini PX Ultra equipped with CuK α radiation (λ = 1.54184 Å). The crystal was monoclinic, space group P2 1 /c, cell volume 1324.77(16) Å 3 and unit cell dimensions a = 11.6888(8) Å, b = 7.7695(4) Å and c = 14.9560(12) Å and angle β = 102.748(7) • (uncertainties in parentheses). The calculated crystal density was 1.525 g·cm −3 . The structure was solved by direct methods using SHELXS-97 and refined with SHELXL-97 [27]. Carbon and oxygen atoms were refined anisotropically. Hydrogen atoms from one of the methyl groups were placed at their idealized positions using appropriate HFIX instructions in SHELXL and included in subsequent refinement cycles, all the others were directly found from difference Fourier maps and were refined freely with isotropic displacement parameters. The refinement converged to R (all data) = 12.24% and wR2 (all data) = 14.96%.
Full details of the data collection and refinement and tables of atomic coordinates, bond lengths and angles, and torsion angles have been deposited with the Cambridge Crystallographic Data Centre (CCDC 1859409).

Bacterial Strains and Growth Conditions
Gram-positive bacteria included Staphylococcus aureus ATCC 29213, Enterococcus faecium ATCC 19434, Enterococcus faecalis ATCC 29212, methicillin-resistant Staphylococcus aureus (MRSA) 66/1 isolated from public buses [28], and vancomycin-resistant enterococci (VRE) Enterococcus faecium 1/6/63 and Enterococcus faecalis B3/101 isolated from river water [29]. Gram-negative strains comprised Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853 and the clinical isolate SA/2, an extended-spectrum β-lactamase producer (ESBL). All strains were kept in Trypto-Casein Soy agar (TSA-Biokar Diagnostics, Allone, Beauvais, France) slants, at room temperature, in the dark. Before each assay, all strains were cultured in Mueller-Hinton agar (MH-Biokar Diagnostics, Allone, Beauvais, France) and incubated overnight at 37 • C. Stock solutions of the compounds were prepared in dimethyl sulfoxide (DMSO-Alfa Aesar, Kandel, Germany) and kept at −20 • C. With the exception of 1e, 10 mg/mL stock solutions were prepared. Compound 1e was less soluble in DMSO than other compounds, so a 2 mg/mL stock solution was prepared. In the experiments, the final concentration of DMSO in the medium was below 1%, as recommended by the Clinical and Laboratory Standards Institute [30].

Antimicrobial Susceptibility Testing
The antimicrobial activity of the compounds was screened using the Kirby-Bauer method, as recommended by the CLSI [31]: 6 mm blank paper discs (Liofilchem, Roseto degli Abruzzi TE, Italy) were impregnated with 15 µg of each compound, and the blank paper discs impregnated with DMSO were used as negative control. MH inoculated plates were incubated for 18-20 h at 37 • C. The results were evaluated by measuring the inhibition halos. The minimal inhibitory concentration (MIC) was performed in accordance with the recommendations of the CLSI [32]. Two-fold serial dilutions of the compounds were prepared in cation-adjusted Mueller-Hinton broth (CAMHB-Sigma-Aldrich, St. Louis, MO, USA) within the concentration range 64-0.063 mg/L, except for 1e, for which the highest concentration tested was 32 mg/L. Colony forming unit counts of the inoculum were conducted in order to determine the initial inoculum size (which should be approximately 5 × 10 5 CFU/mL). The 96-well U-shaped untreated polystyrene microtiter plates were incubated for 16-20 h at 37 • C and the MIC was determined as the lowest concentration of compound that prevented visible growth. The minimal bactericidal concentration (MBC) was determined by spreading 100 µL of the content of the wells with no visible growth on the MH plates. The MBC was determined as the lowest concentration of compound that killed 99.9% of the initial inoculum after overnight incubation at 37 • C [33]. These assays were conducted for reference and multidrug-resistant strains.

Biofilm Formation Inhibition Assay
The effect of all compounds on biofilm formation was evaluated using the crystal violet method, as follows: the highest concentration of the tested compound in the MIC assay was added to bacterial suspensions of 1 × 10 6 CFU/mL prepared in unsupplemented Tryptone Soy broth (TSB-Biokar Diagnostics, Allone, Beauvais, France) or TSB supplemented with 1% (p/v) glucose [D-(+)-Glucose anhydrous for molecular biology, PanReac AppliChem, Barcelona, Spain] for Gram-positive strains. When it was possible to determine a MIC, four concentrations of compound were tested, i.e., 2 × MIC, MIC, 1 2 MIC and 1 4 MIC. A control with appropriate concentration of DMSO, as well as a negative control (TSB alone), was included. Sterile 96-well flat-bottomed untreated polystyrene microtiter plates were used. After a 24 h incubation at 37 • C, the biofilms were heat-fixed for 1 h at 60 • C and stained with 0.5% (v/v) crystal violet (Química Clínica Aplicada, Amposta, Spain) for 5 min. The stain was solubilized with 33% (v/v) acetic acid (Acetic acid 100%, AppliChem, Darmstadt, Germany) and the biofilm biomass was quantified by measuring the absorbance of each sample at 570 nm in a microplate reader (Thermo Scientific Multiskan ® EX, Thermo Fisher Scientific, Waltham, MA, USA) [20,34]. This assay was performed for reference strains.

Antibiotic Synergy Testing
The potential synergy between the compounds and clinically relevant antimicrobial drugs was screened using the Kirby-Bauer method, as previously described [35]. A set of antibiotic discs (Oxoid, Basingstoke, UK) to which the isolates were resistant was selected: cefotaxime (CTX, 30 µg) for E. coli SA/2, vancomycin (VAN, 30 µg) for E. faecalis B3/101 and E. faecium 1/6/63, and oxacillin (OXA, 1 µg) for S. aureus 66/1. Antibiotic discs impregnated with 15 µg of each compound were placed on seeded MH plates. The controls used included antibiotic discs alone, blank paper discs impregnated with 15 µg of each compound alone and blank discs impregnated with DMSO. Plates with CTX were incubated for 18-20 h and plates with VAN and OXA were incubated for 24 h at 37 • C [30]. The potential synergy was considered when the inhibition halo of an antibiotic disc impregnated with the compound was greater than the inhibition halo of the antibiotic or compound-impregnated blank disc alone. The combined effect of the compounds and clinical relevant antimicrobial drugs was also evaluated by determining the antibiotic MIC in the presence of each compound. Briefly, when it was not possible to determine a MIC value for the test compound, the MIC of CTX (Duchefa Biochemie, Haarlem, The Netherlands), VAN (Oxoid, Basingstoke, UK), and OXA (Sigma-Aldrich, St. Louis, MO, USA) for the respective multidrug-resistant strain was determined in the presence of the highest concentration of each compound tested in previous assays. In the case of 1e, the concentration used was 32 mg/L while it was 64 mg/L for the other compounds. The antibiotic tested was serially diluted whereas the concentration of each compound was kept fixed. Antibiotic MICs were determined as described above. For 7, it was possible to determine the MIC for E. faecalis B3/101 and E. faecium 1/6/63, so the checkerboard method was used instead, as previously described [34]. The fractional inhibitory concentrations (FIC) were calculated as follows: FIC of compound = MIC of compound combined with antibiotic/MIC compound alone, and FIC antibiotic = MIC of antibiotic combined with a compound/MIC of antibiotic alone. The FIC index (FICI) was calculated as the sum of each FIC and interpreted as follows: FICI ≤ 0.5, 'synergy'; 0.5 < FICI ≤ 4, 'no interaction'; FICI > 4, 'antagonism' [36].

Conclusions
Marine-derived fungi have proved to be important sources of bioactive secondary metabolites, many of which exhibit cytotoxic and antibiotic activities. One of the most studied marine-derived fungi is of the genus Penicillium. In the past ten years, the Penicillium species from the marine environment received more attention than other fungal genera since compounds isolated from members of the Penicillium genus accounted for more than 25% of compounds of marine fungal origin. Although polyketides are the major secondary metabolites isolated from marine-derived Penicillium species, other structural classes of secondary metabolites such as alkaloids, terpenoids, and sterols are also isolated. In this work, we have described isolation and structure elucidation of two common fungal sterol derivatives: β-sitostenone and ergosterol 5,8-endoperoxide, fifteen polyketides, five of which have not been previously described, and a macrocyclic ether containing 1,4-disubstituted phenyl and succinamide moiety called GKK1032B, from the culture of the fungus P. erubescens strain KUFA 0220, which was isolated from the marine sponge Neopetrosia sp., collected from the Gulf of Thailand. From the compounds evaluated for their antibacterial activity against Gram-positive and Gram-negative bacteria of reference strains and multidrug-resistant isolates, their capacity to inhibit biofilm formation and synergistic effect, only GKK1032B displayed significant activities in all assays. Although the rest of the compounds, including those which have not been previously described, did not show significant antibacterial activity, it does not mean that they are void of bioactivities. Therefore, it is necessary to test these compounds in other bioassay platforms to explore their potential. Finally, it is worth mentioning that this is the first report of the chemical study of the marine-derived P. erubescens. Funding: This research was funded by Fundação para a Ciências e Tecnologia (FCT) (grant number POCI-01-0145-FEDER-016790) and North Portugal Regional Operational Programme (NORTE 2020)(grant number NORTE-01-0145-FEDER-000035).