Aspermerodione, a novel fungal metabolite with an unusual 2,6-dioxabicyclo[2.2.1]heptane skeleton, as an inhibitor of penicillin-binding protein 2a

Rising drug resistance limits the treatment options infected by methicillin-resistant Staphylococcus aureus (MRSA). A promising solution for overcoming the resistance of MRSA is to inhibit the penicillin-binding protein 2a (PBP2a). A novel terpene-polyketide hybrid meroterpenoid, aspermerodione (1), characterized by an unusual 2,6-dioxabicyclo[2.2.1]heptane core skeleton, and a new heptacyclic analogue, andiconin C (2), were isolated and identified from the liquid cultures of endophytic fungus Aspergillus sp. TJ23. The structures and their absolute configurations of all chiral centers were elucidated via extensive spectroscopic analyses and electronic circular dichroism (ECD) calculations and determined via single-crystal X-ray diffraction analysis. Aspemerodione (1) was found to be a potential inhibitor of PBP2a, and work synergistically with the β-lactam antibiotics oxacillin and piperacillin against MRSA.

the culture medium 14 , we preliminarily investigated the influence on the secondary productions of the endophytic fungus Aspergillus sp. TJ23 in a liquid broth.

Results and Discussion
Structural elucidation of the isolates. Compound 1 was isolated as colorless crystals (in CHCl 3 -MeOH).
The molecular formula of 1 was determined to be C 26 (Table 1) revealed 26 carbon atom resonances corresponding to six sp 3 methyls (including one oxygenated), five methylenes (including one oxygenated), one sp 3 methine at δ C 46.5 (C-5), and seven sp 3 quaternary carbons [including one ketal carbon at δ C 109.3 (C-9)]. Three carbonyl groups and four olefinic carbon atoms, accounting for 5 out of 11 degrees of unsaturation, indicated that 1 possessed a hexacyclic ring system.
All proton resonances were assigned to their respective carbons with the help of HSQC spectrum. Comparison of the HRESIMS and spectroscopic data of 1 with previously reported meroterpenoids 15 , prompted us to consider that compound 1 was an unusual rearranged meroterpenoid comprising two subunits, A and B, which were subsequently ascertained as follows ( Fig. 2 and Fig. S19, Supplementary Information). The key HMBC correlations from H-1, H 3 -14, and H 3 -15 to C-3 and C-5; from H-2, H 3 -14, and H 3 -15 to C-4; from H-5 to C-7, C-10, and C-13; and from H 3 -13 to C-1, C-5, C-9, and C-10, combined with the spin-spin coupling systems of H-1/H-2 and H-5/H 2 -6/H 2 -7 in its 1 H-1 H COSY spectrum, indicated the presence of subunit A. In addition, the subunit B was constructed by the key HMBC correlations from H 3 -9′ to C-2′, C-3′, and C-11; from H 3 -10′ to C-4′, C-6′, and C-12; from H-6′ to C-2′, C-4′, C-8′, and C-10′; from H 2 -12 to C-4′ and C-6′, and from OCH 3 -8′ to C-8′. The fusion of subunits A and B was deduced based on the obvious HMBC (Fig. 2) correlations from H 2 -12 and H 2 -1′ to C-8 and C-9, which displayed the direct carbon-carbon connectivity of C-12 and C-8 and ether bridge linkage between C-1′ and C-9. Given the index of hydrogen deficiency of 11 from its molecular formula, the carbon resonance of C-9 (δ C 109.3) in 1 was observed to significantly shift downfield, to which we assigned the existence of another O-linkage at C-9/C-3′ to generate an unusual 2,6-dioxabicyclo[2.2.1]heptane core scaffold. Thus, the planar structure of 1 was established as shown in Fig. 1.
A suitable crystal of 1 was obtained by slow evaporation of a mixture of CHCl 3 -MeOH (10:1) at room temperature, and subjected to a single-crystal X-ray diffraction experiment using Cu Kα radiation (Fig. 5, CCDC 1554255) 17 , which supported the proposed structure and its absolute configuration. It was given the name       skeleton, except that a couple of double peak olefinic proton signals (both d, J = 10.3 Hz) and a sp 3 methine group in andiconin B were replaced by an oxygenated methine and a singlet olefinic proton in 2. Those variations and one additional degree of unsaturation when compared to andiconin B suggested an oxygen linkage between C-1 and C-11 to form a new ring system and demonstrated the formation of a new double bond with dehydration between C-6′ and C-7′, which was evidenced by the relative downfield shifted carbon resonances of C-6′, C-7′, and C-11, and supported by the significant up-field chemical shifts from 152.6 and 126.5 ppm to 84.1 and 40.0 ppm, respectively, in 2. This deduction was further verified by the key HMBC ( Fig. 2 and Fig. S19, Supplementary Information) correlations from H-1 to C-3, C-5. C-9, C-10, and C-13, and from H-6′ to C-2′, C-4′, C-5′, C-8′, C-10′, and C-12, and further supported by the spin-spin coupling systems of H-1/H 2 -2 and H-9/H-11 in its 1 H-1 H COSY spectrum (Fig. 2). Thus, the planar structure of 2 was elucidated as shown ( Fig. 1). The relative configuration of 2 was determined by the comprehension of the key NOESY spectrum (Fig. 3) and coupling constants. The obvious NOEs of H-5 with H 3 -15 and H 2 -7, H 2 -7 with H 2 -1′, and H 2 -1′ with H 3 -9′ indicated that those protons were all arbitrarily assigned to the α-orientation. Accordingly, the NOESY correlations of H-1/H 3 -14/H 3 -13, H-11/H 2 -12, and of H-9 with H 3 -13 and H-11 suggested that they were in the β configuration.
To our knowledge, aspermerodione (1) represents a novel meroterpenoid with an unusual 2,6-dioxabicyclo[2.2.1]heptane carbocyclic core. The biosynthetic origin for 1 was proposed with the co-isolated 2 as a bio-analogue (Fig. 8). Derived from farnesyl pyrophosphate (FPP) and HDMP 15,16 , a series of alkylation and intramolecular cyclization reactions led to the formation of the crucial biosynthetic intermediate, andiconin. Andiconin subsequently participated in many steps of reactions involving oxidation, methoxylation, hemiketal, and Michael addition reactions to furnish the polycyclic scaffold of 1. In addition, andiconin could also be transformed to 2 via a series of reactions including reduction, dehydration, oxidation, Michael addition, and keto-enol tautomerism.  Antimicrobial activity. Antibacterial activities against MRSA were observed for 1 with MIC values of 32 μg/ ml, whereas compound 2 showed only marginal antimicrobial activity. In addition, compound 1 was found to work synergistically with the β-lactam antibiotics oxacillin and piperacillin with ∑FIC values that were less than 0.5 (Table 2) 14 . To further investigate the antibacterial mechanism of 1, in silico target confirmation assays 13 were employed to forecast its possible targets. Therefore, 1 was screened against all five PBPs (PBP1, PBP2, PBP2a, PBP3, and PBP4) encoded in the MRSA core genome in silico 8 , which were considered to be important or promising targets for staphylococcal β-lactam resistance 9,10,18,19 .
Inverse docking identified the possible antibacterial target. The potent antimicrobial activity of compound 1 prompted us to investigate the underlying mechanism. As mentioned above, some complexes of PBPs (PDB ID: 5TRO, 2ZC3, 4CJN, 3PBR and 3HUM) from S. aureus were applied in virtual screening. The total-score values (Table 3) predicted that compound 1 was most likely to bind to allosteric site of PBP2a with higher total score value. To further define this speculation, a microscale thermophoresis (MST) method, which has previously been used to investigate protein-protein, small organic molecule-protein and antibody-protein interactions, was employed to assay the binding affinity between the compounds and PBPs 12 . As a result, 1 exhibited the strongest binding with PBP2a ( Table 3). The equilibrium dissociation constant (Kd) of 1 was 18.4 ± 1.29 μM (Fig. 9).
Previous research showed the PBP2a as an elongated protein with a transpeptidase domain (residues 327-668) and a nonpenicillin-binding domain (residues 27-326), which includes an N-terminal extension subdomain (residues 27-138). Two major binding sites of PBP2a had been identified: one at the active site, where binding of antibiotic or peptidoglycan is known to take place, and another within the nonpenicillin-binding domain, 60 Å distant from the active site, which was proved to be allosteric site. The further virtual docking revealed that compound 1 possibly bound to the allosteric site of PBP2a, further lead to the opening of the active site, enabling catalysis by PBP2a. This result was consistent with synergistic effect of compound 1 with β-lactam antibiotics. As shown in Fig. 10, key hydrogen-bond interactions were between compound 1 with the residues of Asn104, Asn146 and Lys273. Based on the molecular docking results, importing hydrophilic group substituents placed at the C-6 and C-7 position is a worthy target for further structural modification.
In order to visualize the morphology of the bacteria after treated by the inhibitor, the transmission electron microscopy (TEM) was employed. The evidence of TEM images (Fig. 11) of MRSA treated by compound  1 confirmed an obvious change in the ultrastructure of the strains. In the control strains, intact septa were evident, while the strains treated with MIC of compound 1, appearing irregular membrane damage and disruption (Fig. 11). In summary, aspermerodione (1), a complex polycyclic meroterpenoid containing a rare 2,6-dioxabicyclo[2.2.1] heptane motif and a new biosynthetically related compound, andiconin C (2) were isolated and identified from the liquid culture of the endophytic fungus Aspergillus sp.TJ23. The structures of both compounds were elucidated by the combined implementation of X-ray crystallography and ECD calculations. A plausible biosynthetic pathway of 1, involving its bio-metabolite 2, has been proposed. Our research demonstrates that aspermerodione (1) has anti-MRSA activity and based on the observation of TEM, the anti-MRSA activity of 1 might be tightly related to its effect of damaging bacterial cell wall and/or membrane. Moreover, aspermerodione (1) acts as a potential inhibitor of PBP2a that works synergistically with the β-lactam antibiotics oxacillin and piperacillin on antibacterial activities against MRSA. Since a promising strategy for combatting MRSA is to extend the lifespan and efficacy of our currently employed drugs using combination therapy, we believe that the disclosed novel chemical structure and bioactivity of 1 may provide a promising precursor for the development of a combination treatment regime against MRSA.

Methods
General. Optical rotations were determined using a Perkin-Elmer 341 polarimeter (PerkinElmer, Waltham, MA, USA). UV and IR spectra were obtained using a Varian Cary 50 (Varian, Salt Lake City, UT, USA) and a Bruker Vertex 70 instrument (Bruker, Karlsruhe, Germany), respectively. ECD spectra were recorded on a JASCO-810 spectropolarimeter. NMR spectroscopic data were recorded on a Bruker AM-400 spectrometer (Bruker, Karlsruhe, Germany), and the 1 H and 13 C NMR chemical shifts are expressed in δ, referring to the solvent impurity peaks for CDCl 3 (δ H 7.26 and δ C 77.0). HRESIMS was performed in the positive ion mode on a Thermo Fisher LC-LTQ-Orbitrap XL spectrometer (Thermo Fisher, Palo Alto, CA, USA). Semi-preparative HPLC was conducted on a Dionex HPLC system equipped with an Ultimate 3000 pump (Thermo Fisher, Scientific, Germany), an Ultimate 3000 auto sampler injector, and an Ultimate 3000 diode array detector (DAD) controlled by Chromeleon software (version 6.80), using a reversed-phased C 18 Table 3. Predicted binding free energies of compound 1 with five PBPs of MRSA, which considered to be important or possible media for staphylococcal β-lactam resistance. (Surflx-Dock scores) a . a Docking score/ interaction potential of compounds with targets (kcal/mol). * The docking scores predicted that PBP2a showed significantly higher binding affinity for 1 than other PBPs.   the TLC analysis. Fr.3.2 (0.6 g) was further separated via semipreparative HPLC (MeOH-H 2 O, 85%) to afford compounds 1 (20.8 mg; t R 13.2 min; 2 mL/min) and 2 (14.2 mg; t R 9.0 min; 2 mL/min), respectively.  The crystallographic data for compounds 1 and 2 were collected at 100 K on a Rigaku Oxford Diffraction Supernova Dual Source, Cu at Zero equipped with an AtlasS2 CCD using Cu Kα radiation and were deposited in the Cambridge Crystallographic Data Centre. Copies of the data can be obtained free of charge from the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB21EZ, UK (fax: +44-1223-336-033; or e-mail: deposit@ccdc.cam.ac.uk). The structures were solved by direct methods using Olex2 software 20 , and the non-hydrogen atoms were located from the trial structure and then refined anisotropically with SHELXL-2014 21 using a full-matrix least squares procedure based on F 2 . The weighted R factor, wR and goodness-of-fit S values were obtained based on F 2 . The hydrogen atom positions were fixed geometrically at the calculated distances and allowed to ride on their parent atoms. ECD calculation details. The conformations of 1 and 2 generated by BALLOON 22,23 were subjected to semiempirical PM3 quantum mechanical geometry optimizations using the Gaussian 09 program 24 . Duplicate conformations were identified and removed when the root-mean-square (RMS) distance was less than 0.5 Å for any two geometrically optimized conformations. The remaining conformations were further optimized at the B3LYP/6-31 G(d) level in MeOH with the IEFPCM solvation model using Gaussian 09, and the duplicate conformations emerging after these calculations were removed according to the same RMS criteria above. The harmonic vibrational frequencies were calculated to confirm the stability of the final conformers. The electronic circular dichroism (ECD) spectrum was calculated for each conformer using the TDDFT methodology at the LC-wPBE/6-311++G(d,p)//B3LYP/6-31 G(d) level with MeOH as the solvent via the IEFPCM solvation model implemented in the Gaussian 09 program. The ECD spectrum for each conformer was simulated using a Gaussian function with a bandwidth, σ, of 0.4 eV. The spectra were combined after Boltzmann weighting according to their population contributions, and a UV correction was applied 25 .

Determination of the minimum inhibitory concentrations (MIC).
Determination of the MICs were conducted according to our previously reported broth microdilution method 12,13 . In brief, the inoculum was standardized to approximately 5 × 10 5 CFU/mL. The plates were incubated at 37 °C for 16 h, and the MIC values were recorded as the lowest concentration of antibiotic at which no visible bacterial growths were observed. Each experiment was performed three times.
Combination susceptibility test against MRSA. Determination of the combination susceptibility was conducted according to our previously reported broth microdilution checkerboard assay 12,13 . MHB was inoculated with MRSA (5 × 10 5 CFU/mL), and 100 μL was distributed into each well of a 96-well plate except in well 1 A. Inoculated MHB (200 μL) containing test compound (at 2 × MIC) were added to well 1 A, and 100 μL of the same sample was placed in each of well for 2A-12A. Column A wells were mixed 6 to 8 times, and then 100 μL was withdrawn and transferred to column B. Column B wells were mixed 6 to 8 times, followed by a 100 μL transfer SCiENTifiC RepoRtS | (2018) 8:5454 | DOI:10.1038/s41598-018-23817-1 to column C. This procedure was repeated to serially dilute the rest of the columns of the plate up to column G (column H was not mixed to allow the MIC of the antibiotic alone to be determined). Inoculated media (100 μL) containing the antibiotic at two times, the MIC was placed in wells A1-H1 (row 1) and serially diluted in the same manner as row 11. The plates were incubated for 16 h at 37 °C. The MIC values of both compounds and the antibiotic in the combination were recorded as well as the MIC values of the compound alone (row 12) and antibiotic alone (column H). The ∑FIC values were calculated as follows: ∑FIC = FIC Compd + FIC Antibiotic , where FIC Compd is the MIC of the compound in the combination/MIC of the compound alone, and FIC Antibiotic is the MIC of the antibiotic in the combination/MIC of the antibiotic. The combination is considered synergistic when the ∑FIC value is ≤0.5, accumulative or indifferent when the ∑FIC value is >0.5 but <2, and antagonistic when the ∑FIC is ≥2 27 .
Molecular docking and virtual screening. The compounds were saved as mol2 files and used as an input for docking. The docking was performed using the Surflx-Dock module of the Sybyl softare 28,29 . Ligand binding pocket residues were selected using graphical tools in the Sybyl software to create the boundaries of the docking search. All of the hydrogen atoms were added to define the correct ionization and tautomeric states, and the carboxylate, phosphonate and sulphonate groups were considered in their charged form. In the docking calculation, the default FlexX scoring function was used for exhaustive searching, solid body optimizing and interaction scoring 28,29 . The pose with the lowest-energy and the most favorable score was remained.
Protein expression and purification. Wild-type PBP2a (residues 23-668) and four other PBPs (PBPs 1-4) were cloned into a pET28a vector (EMD Biosciences, Novagen, United States) from the mecA sequence of Staphylococcus aureus ATCC 43300. After confirmation by DNA sequencing, the recombined plasmid was transferred into Escherichia coli strain, BL21 (DE3). The transformed cells were grown in LB medium at 37 °C to an OD 600 (0.8-1.0) and induced with 0.4 mM of isopropyl-D-thiogalactopyranoside (IPTG) at 293 K for 16 h. After they were harvested via centrifugation, the cells were re-suspended on ice in a lysis buffer containing 20 mM Tris (pH 7.5), 200 mM NaCl, and 10 mM imidazole, followed by disruption on a French press. Cell debris was removed via centrifugation at 21,000 rpm for 30 min. The protein was bound to Ni-agarose affinity resin, washed with a buffer containing 20 mM Tris (pH 7.5), 200 mM NaCl, and 10 mM imidazole, and eluted with a buffer containing 20 mM Tris (pH 7.5), 250 mM NaCl, and 150 mM imidazole. Fractions containing PBP2a were pooled and further purified on Sephacryl S-200 size-exclusion resin in 2 mM of Tris (pH 7.5) and 200 mM of NaCl.

Microscale thermophoresis. Recombinant PBP2a and four other PBPs (PBPs 1-4) were labelled with the
Monolith NT TM Protein Labelling Kit, RED (Cat # L001), according to the supplied labelling protocol. Labelled PBPs was used at a concentration of 50 nM. Samples were diluted in a 20 mM HEPES (pH 7.4) and 0.05 (v/v)% Tween-20. We used 5 mM of compounds 1 and 2 as the highest concentration for the serial dilution. Oxacillin and Meropenem were used as positive controls. After 10 min of incubation at room temperature, the samples were loaded into Monolith TM standard-treated capillaries, and the thermophoresis was measured at 25 °C after 30 min of incubation on a Monolith NT.115 instrument (NanoTemper Technologies, München, Germany). The laser power was set to 20% or 40% using 30 seconds on-time. The LED power was set to 100%. The dissociation constrant, Kd, values were fitted by using the NTAnalysis software (NanoTemper Technologies, München, Germany) 30 .
Transmission electron microscopy (TEM). Mid-logarithmic grown MRSA was collected by centrifugation at 10,000 g for 10 min, and fixed with modified Karnovsky's fixative. Then the specimens were examined using an energy-filtering transmission electron microscope (Tecnai G2 20 TWIN, FEI, Oregon, USA) operated at 200 kV accelerating voltage. Statistical analysis. The Graph Pad Prism 5.0 software was employed to analyze the statistical data, with the outcome expressed as the means ± SD. Values were analysed using SPSS version 12.0 software by one-way analysis of variance (ANOVA), and a p < 0.05 was considered statistically significant.