Geographic Variability and Anti-Staphylococcal Activity of the Chrysophaentins and Their Synthetic Fragments

Drug-resistant Staphylococcus aureus is a continuing public health concern, both in the hospital and community settings. Antibacterial compounds that possess novel structural scaffolds and are effective against multiple S. aureus strains, including current drug-resistant ones, are needed. Previously, we have described the chrysophaentins, a family of bisdiarylbutene macrocycles from the chrysophyte alga Chrysophaeum taylori that inhibit the growth of S. aureus and methicillin-resistant S. aureus (MRSA). In this study we have analyzed the geographic variability of chrysophaentin production in C. taylori located at different sites on the island of St. John, U.S. Virgin Islands, and identified two new linear chrysophaentin analogs, E2 and E3. In addition, we have expanded the structure activity relationship through synthesis of fragments comprising conserved portions of the chrysophaentins, and determined the antimicrobial activity of natural chrysophaentins and their synthetic analogs against five diverse S. aureus strains. We find that the chrysophaentins show similar activity against all S. aureus strains, regardless of their drug sensitivity profiles. The synthetic chrysophaentin fragments indeed mimic the natural compounds in their spectrum of antibacterial activity, and therefore represent logical starting points for future medicinal chemistry studies of the natural products and their analogs.

such as vancomycin, indicates that more new compounds are still needed [25,26]. These trends, combined with the knowledge that while incremental changes to lead compound scaffolds that result in increased potency and spectrum generally do not provide a permanent solution to combating resistance, suggest that new antibiotic compounds should differ structurally from earlier generation antibacterials.
Previously, we reported a suite of novel bisdiarlybutene macrocycles with antibacterial activity isolated from the marine chrysophyte alga, Chrysophaeum taylori, from St. John, U.S. Virgin Islands [27]. We found that these compounds have inhibitory activity against three drug-resistant Gram-positive organisms, and several drug-susceptible strains, and we developed a structure-activity relationship to explain their potencies. We made a subsequent collection from three locations on St. John, and have looked at geographic and temporal variability of chrysophaentin production. Furthermore, to address supply limitations of the natural products, we are working towards a total synthesis of the chrysophaentins, and have in hand two fragments from this effort. In this study, we have expanded the antibacterial activity and structure activity relationship of the natural chrysophaentins and two synthetic fragments to two additional S. aureus strains, S. aureus UAMS-1, a clinical osteomyelitis isolate, and CA-MRSA USA300-LAC. It was important to expand our library of S. aureus strains to include one with a more recent clinical, but still drug-susceptible, background, and a drug-resistant strain comprising the predominant and successfully invasive CA-MRSA USA300. We investigated the natural and synthetic chrysophaentin antibacterial activity to determine overall potency of growth inhibition and the mode of inhibition (bacteriostatic or bactericidal).

Chrysophaeum taylori Collection and Chrysophaentin Identification
The original chrysophaentins came from one collection of C. taylori made in Round Bay, located on the southern side of St. John, U.S. Virgin Islands, in July 2007. From this collection, we identified four asymmetrically linked macrocycles, chrysophaentins A-D (1-4); one linear analog, chrysophaentin E (5); and three symmetrically-linked macrocycles, chrysophaentins F-H (8-10) ( Figure 1). The eight chrysophaentins were remarkably similar to one another. All possess an ether linkage between C-1 and C-14′. If present, the second ether linkage was located between either C-14 or C-16 and C-1′, and halogen variability only occurred at C-2 and C-2′. Chrysophaentin H additionally differed in its pentahalogenation, with replacement of the hydrogen at C-12′ with a bromine.
To obtain additional raw material, we returned to St. John in June 2009 and made three collections at Round Bay, Long Bay, and Hawk's Nest ( Figure 2). Each sample was dried and sequentially extracted with hexanes, chloroform, and methanol, and the chrysophaentins were found to be present in the methanol extract. Each methanolic extract from the three new collections plus the original collection were chromatographed by reverse-phase HPLC on a Jupiter Proteo C12 column with a linear gradient of 50-80% methanol in 0.05% TFA-water in fifty minutes. This afforded a chromatographic picture of the chrysophaentins present in each extract ( Figure 3). The two collections from Round Bay ( Figure 3A,C), separated by two years, were remarkably similar, as was the collection from Long Bay ( Figure 3B). However, it was immediately obvious that the methanol extract from Hawk's Nest ( Figure 3D) was different; in fact, it lacked the asymmetrically-linked chrysophaentins A-D. Instead, the linear chrysophaentin E was the most abundant, as well as two newly identified linear chrysophaentins, termed E2 and E3 (Figure 1, compounds 6 and 7). Algae collected from all locations contained the symmetrical chrysophaentins, albeit at much lower abundance relative to compounds A-E.  (red pin). Image was created using the Google Earth software package [28].

Identification of New Acyclic Chrysophaentins
Previously we determined the structure of acyclic chrysophaentin E (5), which has a molecular formula of C 32 H 26 Cl 4 O 8 as determined by HR-MS and NMR. Chrysophaentins E2 (6) and E3 (7) were obtained in insufficient quantities for full NMR characterization. However, through analysis of the respective fragmentation patterns observed in the HR-ESI-MS data, we were able to assign structures for each of these new linear chrysophaentins. HR-ESI-MS showed chrysophaentin E2 to possess a molecular formula of C 32 (6), and two bromines replaced two chlorines in E3 (7), when compared to chrysophaentin E (5). Since positions C-2 and C-2′ were the only sites of halogen variability in the natural chrysophaentins, we reasoned that the dibrominated compound 7 likely contained a bromine at both of these carbons. On the other hand, chrysophaentin E2 must contain bromine at only one of these sites, with chlorine at the other. To identify the respective locations of chlorine and bromine in 6, we compared the mass fragmentation patterns of the three linear analogs ( Figure 4). As seen in Figure 4A Figure 4C). Although MS data on model halogenated biaryl ethers was not available, our data suggest that the presence of bromine ortho to the ether bond is destabilizing as compared to chlorine in the same position, and facilitates cleavage at the ether bond in compound 7. Thus, the nearly identical mass spectral patterns observed for 5 and 6, and the unique fragmentation at the ether bond in the spectrum for 7, allow assignments of the structures of the new linear chrysophaentins E2 and E3, as shown in Figures 1 and 4.

Chrysophaentin Fragment Synthesis
While the extracts from Long Bay and Round Bay afforded more of the asymmetrical chrysophaentins, it became apparent that a synthetic approach to chrysophaentin production was necessary in order to fully examine the biological activity of the structural variants. Accordingly, we devised a convergent synthetic strategy to allow for a total synthesis of the chrysophaentins as well as the investigation of structure-activity relationships. Fragment 12, which, in principle, could be used twice for the assembly of the diether 1, could be constructed by a key Negishi cross-coupling between vinyl iodide 13 and benzylic zinc species 14 (Scheme 1). A chloroiodination of alkyne 15 and a metalation of a halide derived from aldehyde 16 were envisioned to provide access to 13 and 14, respectively. In fact, while some modification of protective groups was required, this strategy could be readily realized.

Scheme 1. Retrosynthetic approach toward chrysophaentin A (1).
The commercially available benzoic acid 17 was reduced with lithium aluminumhydride (LAH) to the benzylalcohol 18 (Scheme 2). Bromination [29] and copper-catalyzed substitution with TMS-acetylene magnesium bromide [30] provided the alkyne 20 in 89% overall yield from 17. After desilylation of 20 with tetrabutylammonium fluoride (TBAF), dimethyl resorcinol 15 was subjected to iodochlorination [31,32] in an attempt to generate 13. However, the desired alkyne addition product was formed to a minor extent, and mainly overiodination products were observed. Therefore, we demethylated 15 with boron tribromide and installed the deactivating pivaloyl esters in 82% yield. Iodochlorination of 21 now proceeded in excellent yield to give a 2:1 mixture of alkene isomers 22 and 23. While this ratio could be further improved in favor of the (E)-alkene 22 using solvents other than dichloromethane, yields were considerably lower. Furthermore, we were unable to separate the mixture at this stage, and the clean conversion in dichloromethane allowed us to carry it forward to the coupling step.
Formylation [33,34] of dimethyl hydroquinone 24 provided aldehyde 16, which was reduced to the alcohol 25 and brominated to give 26 in high yield (Scheme 3). Four equivalents of this bromide were then converted to the corresponding zinc reagent. Removal of excess zinc was necessary prior to combination with 22 and 23 and subsequent microwave heating in the presence of palladium acetate. These cross-coupling conditions [35] generated satisfactory yields of the chloroalkenes 11 and 27 in a 2:1 ratio after methanolysis of the pivaloate esters with cesium carbonate and methanol. An analytically pure sample of 11 was obtained by separation with supercritical fluid chromatography (SFC). Boron tribromide was used to cleave the two methyl ethers on 11 to give the tetraphenol 12, which was obtained as the pure (E)-isomer and further purified by preparative supercritical fluid chromatography (SFC) before submission to biological testing. In a control experiment, the use of a 2:1 mixture of 11 and 27 in the deprotection reaction led to a 2:1 mixture of 12 and its (Z)-diastereomer, as expected. The assignments of the E/Z-configurations of 12 and its (Z)-diasteromer as well as 22 and 23 were mainly based on the available literature data [31,32]. (26) and Negishi coupling to give tetraphenol 12.

Antibacterial Evaluation
With the new fragments in hand, we tested whether their potencies were similar to the parent chrysophaentins. A microbroth dilution assay was used to evaluate the antibacterial activity of these compounds, and the minimum inhibitory concentration that led to a 50% decrease in growth (MIC 50 ) was determined using a model curve fit to the data. The MIC 90 was determined as the lowest concentration where there was no visible growth, and the minimum bactericidal concentration (MBC) was determined as the lowest concentration where a plated aliquot of treated bacteria led to a three logarithmic decrease in colony forming units (CFU) per milliliter relative to the starting inoculum. All compounds were evaluated against three different S. aureus strains from ATCC and two diverse clinical isolates. The strains used in this study included: S. aureus ATCC 25923, a clinical isolate from Seattle, 1945; methicillin-resistant S. aureus ATCC BAA-41, a hospital-acquired strain isolated in New York City in 1994; multidrug-resistant S. aureus (MDRSA) ATCC BAA-44, a hospital-acquired strain isolated in Lisbon, Portugal, with resistance towards ampicillin, methicillin, oxacillin, penicillin, erythromycin, gentamicin, tetracycline, azithromycin, amikacin, clindamycin, cephalothin, ceftriaxone, imipenem, lincomycin, streptomycin, perfloxacin, rifampin, and neomycin [6,36]; S. aureus UAMS-1, a clinical osteomyelitis isolate [37]; and community-associated methicillin-resistant S. aureus, USA300-LAC [38,39].

Minimum Bactericidal Concentrations for Select Natural Chrysophaentins
The MIC 50 for select chrysophaentins was reported previously [27]. Those qualitative antibacterial data have been expanded here to include descriptions of the MIC 90 and the MBC for the chrysophaentins against S. aureus 25923, MRSA BAA-41, and MDRSA BAA-44. These values are summarized in Table 1. We found chrysophaentin A to inhibit the growth of S. aureus 25923, MRSA BAA-41 and MDRSA BAA-44 with MIC 50 values on the order of 1-4 μM, and MIC 90 values ranging from 4-9 µM against all three strains. Chrysophaentin A was bacteriostatic at its MIC 90 , becoming bactericidal only at concentrations four to eight times higher than the MIC 90 . We have found chrysophaentin A to be the most potent compound in all of our antibacterial assays, yet chrysophaentins F and H which bear a symmetrical bisbibenzyl ether skeleton also inhibit S. aureus 25923 and MRSA BAA-41 with single digit micromolar MIC 50 s. While their MIC 90 s were higher than those of chrysophaentin A (17-19 µM), the MBC for chrysophaentin F was still four-fold higher than the MIC 90 . A significant reduction in potency was observed for the acyclic natural product chrysophaentin E (5), which was also bacteriostatic at its MIC 90 . Chrysophaentins D (4) and G (9), both of which contain a bromine atom on the D ring, were the least active compounds, with an 8-to13-fold increase in MIC 50 s values, and an 8-fold increase in MIC 90 s.   (Table 1), and the MIC 90 was only slightly higher at 9 µM. Chrysophaentin A was still bacteriostatic at low micromolar concentrations, and the MBC was only 2 times higher at 19 µM. By comparison, the symmetrical chrysophaentin F (8) was slightly less active with respective MIC 50 and MIC 90 values of 12 and 37 µM, and was also bacteriostatic at the MIC 90 . The MIC 50 s for compounds 11 and 12 were similar to those of 8 as was the MIC 90 for compound 11, while the MIC 90 for compound 12 was higher. For both synthetic compounds, the MBC was two-fold higher than the MIC 90 . Finally, chrysophaentin D (4) was the least active with MIC 50 s between 41 and 49 µM, and MIC 90 s at 65 µM.
Growth curves for chrysophaentins A (1), D (4), and F (8), and compound 12 were graphed as percentage growth of untreated bacteria for S. aureus UAMS-1 ( Figure 5A) and CA-MRSA USA300-LAC ( Figure 5B). These curves clearly demonstrate the relationship between the structure of the chrysophaentins and analogs, halogen composition, and anti-staphylococcal activity.

Discussion
C. taylori is a rare marine chrysophyte alga. It was originally discovered in the Dry Tortugas off the southern coast of Florida [40], and has since been found in several locations worldwide. Bioactive natural products have been discovered from collections made in two locations. The Puerto Rico location yielded hormothamnione [41] and 6-desmethoxyhormothamnione [42], while the collections from St. John, U.S. Virgin Islands have yielded the ten chrysophaentins, as well as hormothamnione and 6-methoxyhormothamnione. As we have described here, analysis of methanol extracts revealed geographic variability within the C. taylori collections from St. John. The original collection coming from Round Bay in 2007 yielded chrysophaentins A-H. A second collection at Round Bay two years later contained a similar profile of compounds, indicating temporal stability in the natural compound production at this location. A 2009 collection at Long Bay contained the same eight chrysophaentins in a similar ratio. Both Long Bay and Round Bay are on the southeastern side of the island, and each of these collections was made in shallow protected waters, where the algae were growing on coral sand. The 2009 collection from Hawk's Nest on the northwestern side of the island yielded a different profile of compounds. It was immediately obvious when comparing the chromatograms that the difference laid in the complete absence of the asymmetrical chrysophaentins in the samples collected at this location. Instead, the most abundant compounds included the acyclic chrysophaentin E, as well as two newly identified linear chrysophaentins, namely chrysophaentins E2 and E3.
Synthetic efforts to obtain a steady supply of chrysophaentin A are under way. An early approach that was employed provided two chrysophaentin fragments that we have shown exhibit similar potencies and spectrum of activity to those of the parent chrysophaentins. Previously, we profiled the anti-staphylococcal activity of several of the natural chrysophaentins towards common laboratory strains of S. aureus. In this study, we sought to expand the structure-activity relationship (SAR) by including the synthetic fragments, and to expand the number and diversity of S. aureus strains. We tested select compounds against five S. aureus strains to measure their potency (MIC 50 and MIC 90 ) and to determine whether the compounds were bacteriostatic or bactericidal. All chrysophaentins, natural and synthetic, were bacteriostatic at their MIC 90 ; but were bactericidal at concentrations ranging from two to eight fold higher than their respective MIC 90 s. Furthermore, the synthetic fragments showed similar potency to the acyclic chrysophaentin E (5).
The synthetic fragments fit perfectly into our proposed structure-activity relationship. The tetrachlorinated macrocycle with the asymmetrical linkage between C-16 and C-1′ (chrysophaentin A) was essential for maximum potency. Compounds with an ether bond between C-14 and C-1′, which lead to a symmetrical macrocycle as seen in chrysophaentin F, show a decrease in overall potency. The presence of a fifth halogen at C-12′ did not decrease the antibacterial activity, even when the additional halogen was a bromine, indicating that the atom type at this position or even this ring is not critical to the observed activity. The absence of a macrocyclic structure, as in chrysophaentin E and the synthetic fragments 11 and 12, further decreased antibacterial activity. However, the presence of the asymmetrical macrocycle did not guarantee potency as halogen diversity on the phenyl rings also affected activity. Chrysophaentin D, which contains two bromines in place of the chlorines on the phenol rings in 1, was the least active compound considered in this study.
We have shown that subtle changes to the chrysophaentin structure and composition lead to differences in antibacterial activity. Knowing that the macrocycle is essential for full antibacterial potency, it is not surprising that the synthetic fragments described here have diminished activity compared to the cyclic natural products. However, we were gratified to find that they are indeed antimicrobial and their activities are comparable to the linear chrysophaentins. Their ease of synthesis and antimicrobial activity provide an excellent starting point not only toward the cyclic chrysophaentins, but also for generating more elaborate synthetic analogs.
The biosynthetic pathway leading to production of the chrysophaentins has not been characterized and a genome sequence for any Chrysophaeum species has yet to be published. Interestingly, the most closely related natural products include the marchantins, bazzanins, and chlorinated isoplagiochins, all of which were isolated from different species of liverworts [43]. While these macrocyclic compounds are reminiscent of the chrysophaentins, each family possesses defining structural characteristics that differ relative to one another and to the chrysophaentins. The marchantins [44] are bisbibenzyl ethers whose phenol rings are connected by an ethyl group, and thus far do not contain halogens. The bazzanins [45,46] and chlorinated isoplagiochins [47], on the other hand are chlorinated, but are bisbibenzyls rather than bisbibenzyl ethers. Like the marchantins and all liverwort metabolites, they also contain ethyl linkages. The presence of a butene linkage is unique to the chrysophaentins. While some information is known about the biosynthesis of the liverwort metabolites [48], the pathways are incompletely characterized; moreover, parallels to chrysophaentin biosynthesis cannot be predicted as the producing organisms are not closely related. The lack of biosynthetic information underlines the importance of a synthetic approach that could supply large quantities of the chrysophaentins for further biological studies.

Biological Material
Samples of the marine algae were collected from Round Bay in July 2007, and Long Bay, Round Bay, and Hawk's Nest in June 2009, St. John, U.S. Virgin Islands (Collection permit VIIS-2008-SCI-0034 National Park Service). Each of the alga samples were observed as fluffy, sulfur yellow colonies growing on coarse sand at shallow depths averaging −20 ft. The samples were identified as Chrysophaeum taylori Lewis and Bryan as described previously [27]. Briefly, the presence of clear stalk like structures made up of branching mucilaginous "streamers" of pear shaped, invaginated cells were visible by light microscopy (4-10×) on freshly collected specimens, as well as those preserved in 2.5% glutaraldehyde in sterile sea water immediately after collection.

Mass Spectrometry and MS/MS Fragmentation
The MeOH extract was partitioned between n-BuOH-H 2 O (1:1), and the organic layer was fractionated on Sephadex LH-20. Fractions containing the unknown chrysophaentins were subjected to HR-ESI-MS and MS in-source fragmentation. High resolution accurate mass data was obtained using a Waters LCT Premiere ESI-TOF mass spectrometer equipped with a Z-Spray electrospray ion source. The instrument was operated in the w-mode at a nominal resolution of 10,000 and an internal standard was used as reference. The ionization mode was negative ion and the ESI capillary voltage was −3000 V and the cone voltage was 30 V. The collisionally activated dissociation (CAD) mass spectra were obtained using a Waters LCT Premiere ESI-TOF mass spectrometer according to the method described by Ren et al. [49] The ion guide 1 voltage was 65 V and the cone voltage was 30 V. The instrument was operated in negative ion mode.

General Synthesis
All moisture sensitive reactions were performed using syringe-septum techniques under an atmosphere of either dry N 2 or dry argon unless otherwise noted. All glassware was dried in an oven at 140 °C for a minimum of 6 h or flame-dried under an atmosphere of dry nitrogen prior to use. Reactions carried out at −78 °C employed a CO 2(s) /acetone bath. Et 2 O and tetrahydrofuran were dried by distillation over sodium/benzophenone under an argon atmosphere. Dry methylene chloride was purified by filtration through an activated alumina column. All degassed solvents were prepared using the freeze/pump/thaw method (3×). Methanol, acetonitrile, and N,N-dimethylformamide were stored over molecular sieves (3 Å). Deuterated chloroform was stored over anhydrous potassium carbonate. Reactions were monitored by TLC analysis (pre-coated silica gel 60 F 254 plates, 250 µm layer thickness) and visualized by using UV lamp (254 nm) or by staining with either Vaughn's reagent (4.8 g of (NH 4 ) 6 Mo 7 O 24 · 4H 2 O and 0.2 g of Ce(SO 4 ) 2 in 100 mL of a 3.5 N H 2 SO 4 ) or a potassium permanganate solution (1.5 g of KMnO 4 and 1.5 g of K 2 CO 3 in 100 mL of a 0.1% NaOH solution). Flash column chromatography was performed with 40-63 µm silica gel (Silicycle). Microwave reactions were performed on a Biotage Initiator microwave reactor. Infrared spectra were measured on a Smiths Detection IdentifyIR FT-IR spectrometer (ATR). Unless otherwise indicated, all NMR data were collected at room temperature in CDCl 3 or (CD 3 ) 2 CO on a 300, 400, 500, 600, or 700 MHz Bruker instrument. Chemical shifts (δ) are reported in parts per million (ppm) with internal CHCl 3 (δ 7.26 ppm for 1 H and 77.00 ppm for 13 C), or internal acetone (δ 2.05 ppm for 1 H and 29.85 ppm for 13 C), as the reference. 1 H NMR data are reported as follows: chemical shift, multiplicity (s = singlet, bs = broad singlet, d = doublet, t = triplet, q = quartet, m = multiplet, dd = doublet of doublets, dt = doublet of triplets, td = triplet of doublets, qd = quartet of doublets, sep = septet), integration, and coupling constant(s) (J) in Hertz (Hz).

Conclusions
The rare marine chrysophyte alga C. taylori has yielded a number of biologically active natural products. While the occurrence of C. taylori has been documented at a growing number of locations globally, thus far only collections obtained from the southeastern side of St. John appear to contain the full suite of chrysophaentins. The geographical variability described here, along with temporal and seasonal changes in the abundance of C. taylori, limited our ability to obtain additional material for study. Nevertheless, early synthetic efforts toward the synthesis of chrysophaentin A in particular have provided fragments of the natural products that have proven useful for biological studies. Here we have expanded the antimicrobial studies of several chrysophaentins and two synthetic fragments to include drug-susceptible and drug-resistant S. aureus strains that are of current clinical interest. Although approximately half the size of the cyclic chrysophaentins, we have shown that the potencies and spectrum of antimicrobial activity of the synthetic fragments are comparable to those of the parent chrysophaentins, in particular to the acyclic compound chrysophaentin E. Moreover, both the natural products and synthetic fragments are equally effective against both drug-resistant and drug-susceptible S. aureus strains. In summary, our results show promise for the notion of synthesizing equally potent chrysophaentin analogs that will be more readily available, and provide new data on the antimicrobial spectrum of this novel class of antibiotics.