Callistemenonone A, a novel dearomatic dibenzofuran-type acylphloroglucinol with antimicrobial activity from Callistemon viminalis

A new acylphloroglucinol with a novel architecture including an unprecedented dearomatic dibenzofuran core, named callistemenonone A (1), was isolated from the leaves of Callistemon viminalis (Myrtaceae). The structure was fully characterized on the basis of extensive spectroscopic analysis, including UV, HRESIMS, as well as 1D and 2D NMR spectral data (HSQC, HMBC, and ROESY). The deduced structure represents the first example of a natural dibenzofuran with two phenyl moieties coupling through tertiary hydroxy and ketal carbons. A plausible biogenetic pathway involving oxidative coupling and dearomatization as key steps is proposed to account for the biosynthesis of this novel class of dibenzofuran. Moreover, antimicrobial assays, in conjunction with the time-killing and biophysical studies, revealed that 1 exerted potent bactericidal activity against a panel of methicillin resistant pathogenic microbes with a unique mechanism.

The genus Callistemon belongs to the family Myrtaceae, and is comprised of over 30 species with most being woody, aromatic, and ornamental trees or shrubs endemic to Australia 1 . Callistemon viminalis (Sol. ex Gaertn,) G. Don (commonly known as red bottlebrush) is an Australian native green shrub with leaves and stems reputed to possess a variety of biological effects, including antidiabetic 2 , antimicrobial, anti-inflammatory, antistaphylococal, antithrombin [3][4][5][6] , and nematicidal, larvicidal, as well as pupicidal activities 7 . It was introduced into China some decades ago. Presently, it is widely cultivated in southern China as an ornamental tree and is adopted into Chinese folk medicine to treat colds and arthralgia 8 .
Plants in the genus Callistemon are known for their enriched phloroglucinol content, which is responsible for a variety of the biological activities of these herbal medicines 4,5 . Previous phytochemical studies on different parts of C. viminalis showed the presence of flavonoids 9 , triterpenoids 9 , tannins, and phloroglucinol derivatives 10,11 . In a continuation of our efforts to discover new antibacterial constituents from the plants of family Myrtaceae [12][13][14][15][16][17] , the ethanolic extract of C. viminalis was found to exhibit in vitro antibacterial activity with a MIC value of 100 μg/ mL against MRSA JCSC 2172. Antibactewrial activity guided isolation has now led to the isolation of a new potent (MIC 20 μg/mL, anti-MRSA JCSC 2172) acylphloroglucinol with an unprecedented dearomatic dibenzofuran skeleton, designated as callistemenonone A (1) (Fig. 1). Herein, the isolation, structural elucidation, bioassay, and antibacterial mechanism of 1 are described.

Results and Discussion
C. viminalis leaves were extracted at room temperature with 95% EtOH, and the resulting pooled crude extract was concentrated under reduced pressure and then partitioned to afford n-hexane-soluble, EtOAc-soluble, and aqueous fractions. The n-hexane and EtOAc-soluble fraction were combined and subjected to repeated silica gel column chromatography to obtain 1 as a yellowish oil.
Although the HMBC experiment failed to provide direct and conclusive evidence for the linkage between the structural fragments I and II due to the absence of a suitable correlative proton, assembly of the two moieties was achieved through a connection of C-1-O-C-3′ which led to the presence of a furan ring and, together with portions I and II, accounted for the nine double bond equivalents. Compound 1 might be an equimolar mixture of the two enantiomers based on its zero optical rotation value 10 . The relative configuration of 1 was deduced from an analysis of the ROESY correlations (Fig. 3). The key ROE correlations between the 2′-OH/H-4a′ and the 3′-OMe/H-4′a indicated that the 2′-OH and 3′-OMe were cofacial. Therefore, the structure was determined as 1, and given the trivial name callistemenonone A, representing the first member of a new family of phloroglucinol derivatives with an unprecedented dearomatic dibenzofuran core combining an α,β-triketone with a phloroglucinol unit.
A biotransformation pathway ( Fig. 4) was proposed to account for the biogenesis of callistemenonone A (1). Briefly, isovaleryl-CoA (2) and malonyl-CoA (3) would undergo PKS (polyketide synthase)-catalyzed condensation to yield the acylphloroglucinol 4, which has previously been reported from plants of the Myrtaceae family 11,20,21 . Oxidative coupling leads to the putative key precursor 6 through the radical intermediate 5. Further oxidative dearomatization of 6 would generate the intermediate 7, leading to construction of the unique tertiary hydroxyl carbon and the formation of the dibenzofuran architecture in callistemenonone A (1) through Michael addition.
The antimicrobial activity of 1 was evaluated towards a panel of Gram-positive bacteria, including Staphylococcus aureus strain (CMCC 26003), Bacillus cereus strain (CMCC 63302) and several strains of methicillin-resistant S. aureus MRSA JCSC 2172, MRSA JCSC 4469, and MRSA JCSC 4744 using standard MIC method and following the CLSI guidelines. As a result, compound 1 was shown to possess potent antibacterial activity against these Gram-positive bacteria with MICs and MBCs ranging from 5 to 80 μg/mL as presented in Table 2. The potency is 3-60 fold less than that the positive control substance vancomycin (Van). Callistemenonone A (1) was inactive towards the Gram-negative bacteria Escherichia coli (ATCC 8739), presumably due to its relative inability to penetrate the permeability barrier presented by the outer membrane and/or the action of efflux transporters.
In order to understand the actions of callistemenonone A (1) at the morphologic level and to shed light on the mechanism underlying the antimicrobial activity against the clinical MRSA isolate JCSC 2172, a mechanistic investigation was initiated combining morphological, biochemical, and biophysical studies. As a result, time-kill kinetic experiments dosing with 1 × MIC revealed that 1 exhibited rapid, in vitro bactericidal activity (killing > 90% within 4 h) and complete bactericidal effect, with a sharp reduction in CFU of 99.99%, after 24 h exposure     Further biophysical studies using fluorescence probes for membrane potential were performed to determine whether 1 caused cytoplasm membrane depolarization [22][23][24] . As shown in Fig. 5B, after treating the MRSA JCSC 2172 cells with serial concentrations of 1 (0.5 × MIC, 1 × MIC, 2 × MIC, 4 × MIC, 8 × MIC), a sharp increase in the fluorescence was observed, suggesting that the favorable bactericidal action of 1 could be attributed to a significant dissipation of the membrane potential due to depolarization. However, the SYTOX green assay indicated that the bactericidal action of 1 was independent of membrane disruption 25,26 , because there was no obvious increase of fluorescence detected when the MRSA suspension was treated even with 8 × MIC of 1 for up to 4 h. In contrast, the positive control melittin caused an unambiguous increase in fluorescence within 60 s (Fig. 5C) and in the whole concentrations range from 0.5 to 8 MIC after 4 h (Fig. 5D). These results were a strong indication that the bactericidal action of 1 was independent of membrane disruption.
Moreover, the aforementioned results were good in agreement with the morphological study using scanning electron microscopy, wherein 1 failed to disrupt the cytoplasmic membrane structure of MRSA JCSC 2172 cells following treatment with 8 × MIC of 1 for 4 h. As shown in Fig. 6, the morphology of the treated cells remained intact with smooth spheres comparable to those of the untreated control cells. Based on the results of membrane depolarization and membrane disruption assay, in conjunction with the time-killing study (killing > 90% after 4 h), it can be concluded that callistemenonone A (1) acts rapidly as a bactericidal antibiotic by disturbing the bacterial membrane potential without causing membrane disruption. In summary, these experiments have described herein the isolation, structural elucidation, and biological evaluation of a novel acylphloroglucinol named callistemenonone A (1), which was isolated from the leaves of Callistemon viminalis (Myrtaceae). The metabolite represents a new type of dimeric acylphloroglucinol with an unprecedented dearomatic dibenzofuran core which is biogenetically formed through tertiary hydroxy and ketal carbons. The oxidative coupling reaction and dearomatization were proposed to be key biotransformation steps for its biogenesis. The investigation of antibacterial mechanism, in conjunction with the time-killing study, revealed that 1 acted rapidly as a bactericidal agent by disturbing the bacterial membrane potential without causing membrane disruption. This represents a very favorable bactericidal mechanism which differs from any other antibiotic agent available either in the clinic or in development, rendering it a very promising candidate for further extensive investigation through medicinal chemistry, with an aim towards new antibiotic drug discovery. Extraction and isolation. The chopped fresh leaves of C. viminalis (15.0 kg) were extracted with 95% Etch (30 L × 3) at room temperature for 24 hours. The pooled solvents were evaporated under reduced pressure to yield a brown syrupy residue (3.0 kg), which was suspended in H 2 O (3 L) and extracted successively with n-hexane (3 L × 3) and EtOAc (3 L × 3) to afford n-hexane-and EtOAc-soluble fractions. Most of spots of the two parts were similarity, which were detected by Thin layer chromatiography (TLC). So n-hexane-and EtOAc-soluble fractions (2.2 kg) were combined and subjected to silica gel column chromatography eluting with n-hexane/EtOAc with increasing polarity from 20:1 → 0:1 to yield six sub-fractions (A-F). Sub-fraction C (32.0 g) was chromatographed on a silica gel column with a gradient of CHCl 3 /MeOH (100:1 to 20:1) to yield three fractions (C 1 -C 3 ). Sub-fraction C 2 (3 g) was subjected to RP-C 18 column chromatography (diameter: 60 cm, length: 254 mm; particle size: 150-200 mesh) and eluted with MeOH/H 2 O (90:10, 95:5, 100:0) to obtain three further sub-fractions (C 21 to C 23 ). Sub-fraction C 21 (400 mg) was subjected to CC on Sephadex LH-20 (CHCl 3 /MeOH, 1:1, v/v) and further purified by repeated silica gel column chromatography (n-hexane/EtOAc, 20:1 → 1:1) to yield 1 (25 mg). Cytoplasmic membrane depolarization Assay. The effect of callistemenonone A on the membrane potential of MRSA JCSC 2172 cells was investigated using a modified version of the method described by Wu and Hancock 28 . Briefly, cultures of MRSA (JCSC 2172) were grown to exponential phase and then harvested. Bacteria were suspended and washed with buffer solution (5 mM HEPES at pH 7) until an optical density OD 0.1 at 600 nm (OD 600 ) was obtained. The cell suspension was then incubated with 0.4 μM 3,3-dipropylthiacarbocyanine (DiSC 3 -5) for 20 min at 37 °C with shaking. Then, 100 mM KCl was added to the buffer to balance the chemical potential of K + inside and outside the cells. The desired concentration of callistemenonone A (1) was added into a stirred cuvette, and the change in fluorescence was monitored at an excitation wavelength of 660 nm and an emission wavelength of 675 nm. DMSO and vancomycin was used as the negative controls with no effect on the depolarization. Experiments were repeated at least three times and found to be reproducible. Data from an individual experiment are presented. SYTOX green assay. The protocol was based on the method of Rathinakumar et al. 29 . Briefly, MRSA (JCSC 2172) cells were suspended in 0.85% NaCl buffer until an OD 600 of 0.2 was obtained. The bacterial suspension was incubated with serial concentrations of callistemenonone A (1), respectively for 4 h, and then incubated with 3 μM of SYTOX green for 5 min. The fluorescence was measured at an excitation wavelength of 504 nm and an emission wavelength of 523 nm. Melittin, a cell lytic factor, was used as the positive control. Experiments were repeated at least three times and found to be reproducible. Data from an individual experiment are presented.

Experimental
Scanning electron microscopy. For the observations using scanning electron microscopy, exponential-phase bacteria were treated with the compound at 8 × MIC for 4 h at 37 °C. The cells were washed twice, suspended in PBS, and prefixed in 0.1 M phosphate buffer (pH 7.2) containing 2% glutaraldehyde and 2.5% paraformaldehyde overnight. After washing six times with 0.1 M phosphate buffer, the samples were post-fixed in 1% osmium tetroxide for 2 h. After washing another three times with 0.1 M phosphate buffer, the samples were dehydrated through a graded ethanol series and subjected to freeze-drying (JFD-310, JEOL, Tokyo, Japan). Samples were then mounted on stubs and coated with gold in a sputter coater (JFC-1600, JEOL, Tokyo, Japan), and then examined and photographed under a scanning electron microscope (JSM-6360LV, JEOL, Tokyo, Japan). Synergistic assay. The synergy of 1 in combination with melittin was investigated in vitro against MRSA2172 using the Bliss independence model as described in previous reports 30 . Briefly, bacterial strains were incubated with a sub-inhibitory concentration of simvastatin and control antimicrobials for 12 h and the degree of synergy was calculated using the formula: Synergy (S) was calculated using the formula: S = (f A0 /f 00 ) (f 0B /f 00 ) − (f AB /f 00 ).The parameter f AB refers to the optical density of the bacteria grown in the presence of 1 and melittin; parameters f A0 and f 0B refer to the bacterial growth rate in the presence of 1 alone and melittin alone, respectively; the parameter f 00 refers to the bacterial growth in the absence of drugs. Degree of synergy (S) values corresponds to the following cut-offs: Zero indicates neutral, values above zero (positive value) represents synergism, and values below zero (negative values) correspond to antagonism. Drug combinations with higher positive value represents high degree of synergism.