Antibacterial Oligomeric Polyphenols from the Green Alga Cladophora socialis

A series of oligomeric phenols including the known natural product 3,4,3′,4′-tetrahydroxy-1,1′-biphenyl (3), the previously synthesized 2,3,8,9-tetrahydroxybenzo[c]chromen-6-one (4), and eight new related natural products, cladophorols B–I (5–12), were isolated from the Fijian green alga Cladophora socialis and identified by a combination of NMR spectroscopy, mass spectrometric analysis, and computational modeling using DFT calculations. J-resolved spectroscopy and line width reduction by picric acid addition aided in resolving the heavily overlapped aromatic signals. A panel of Gram-positive and Gram-negative pathogens used to evaluate pharmacological potential led to the determination that cladophorol C (6) exhibits potent antibiotic activity selective toward methicillin-resistant Staphylococcus aureus (MRSA) with an MIC of 1.4 μg/mL. Cladophorols B (5) and D–H (7–11) had more modest but also selective antibiotic potency. Activities of cladophorols A–I (4–12) were also assessed against the asexual blood stages of Plasmodium falciparum and revealed cladophorols A (4) and B (5) to have modest activity with EC50 values of 0.7 and 1.9 μg/mL, respectively.


■ INTRODUCTION
Aromatic systems are ubiquitous within natural products, reflecting the availability of their biosynthetic precursors in plants, algae, and microorganisms via the shikimate and polyketide pathways and the varied biological effects of such functional groups. 1,2 When incorporated as building blocks in vascular plants, aromatic polymers 3−6 such as lignin provide structural support that ultimately enabled plants to occupy terrestrial ecosystems leading to the evolution of trees. 7 Aromatic natural products conferring structural rigidity were unknown in the marine realm until lignin was reported in the coralline red alga Calliarthron cheilosporioides (phylum Rhodophyta). 8 This startling discovery suggested that an ancestor of green and red algae acquired the capacity to polymerize phenolic monomers but that this feature was silenced in most evolving lineages. Other red algae produce a variety of structurally simple, often halogenated phenolics of mostly unknown biological function, 9 although more complex, aromatic natural products of mixed shikimate and isoprenoid biosynthetic origin have also been observed. 10 Many members of the brown algae (phylum Ochrophyta), which evolutionarily diverged just before land plants, produce polymeric, acetogenic, phenolic compounds (phlorotannins) constituting 1− 20% of dry mass. 11 These oligomers of phloroglucinol protect tissues against UV radiation and herbivory 11,12 and may contribute to structural support. 12 In contrast, significantly fewer phenolic natural products have been reported in green algae (phylum Chlorophyta). Brominated polyphenolic compounds including the feeding deterrent avrainvilleol and the condensation product of two molecules of avrainvilleol, rawsonol, are produced by green algae belonging to the genus Avrainvillea. 13−15 Another example of phenolic compounds from green algae are two vanillic acid analogues (1−2) from an Australian Cladophora socialis that have been found to disrupt insulin cell signaling. 16 The present study describes a series of related oligomeric phenols, cladophorols A−I (4−12), from the green alga C. socialis that attracted our attention for its antibiotic activity against methicillin-resistant Staphylococcus aureus (MRSA). Analysis of mass spectrometric and NMR spectroscopic data enabled structure elucidation despite the high molecular weights and repeating structural patterns of members of this natural product family. We used J-resolved NMR spectroscopy to reduce signal overlap in the 1 H and 13 C NMR spectra and employed picric acid line width-narrowing to reveal the exact linkage pattern of the subunits. 17 In addition, we propose a structural revision of known vanillic acid analogues (1−2) based on a combination of spectroscopic and computational approaches.
Characterization of 4. HRESIMS was used to assign the molecular formula of 4 as C 13 H 8 O 6 . The 1 H NMR spectrum exhibited only four singlets at δ H 7.58, 7.33, 7.32, and 6.73 (Table 1), attributed to two aromatic rings based on HSQC correlations with δ C 115.4, 108.3, 107.4, and 104.4, respectively, and the absence of coupling between the protons.
Correlations observed in HMBC from the signal at δ H 6.73 (H-4) to δ C 111.4 (C-10b), 144.4 (C-2), 146.0 (C-4a), and 148.6 (C-3) established the first ring. The second ring was determined from HMBC correlations from δ H 7.58 (H-7) to δ C 131.7 (C-10a), 147.3 (C-8), 154.8 (C-9), and 164.0 (C-6). Direct linkage between the two aromatic rings was established by HMBC correlations from δ H 7.32 (H-1) to C-10a and from δ H 7.33 (H-10) to C-10b. With connectivity of the carbonyl functionality and the two aromatic rings established, a third  The Journal of Organic Chemistry Article ring was necessary to account for the 10 degrees of unsaturation calculated for this compound.
Since the 1 H and 13 C NMR chemical shifts of 4 were similar to those reported for 1, the structure depicted in Figure 2a was considered. Although it did not conflict with any of the observed NMR spectroscopic features, full consideration of all plausible structures and evaluation of these data to determine the best fit led to a more plausible alternative (Figure 2b). An HMBC spectrum recorded in DMSO-d 6 with picric acid, 17 which reduced the line widths of the phenolic hydroxy region, revealed four exchangeable protons at δ H 10.41, 9.89, 9.74, and 9.23, each correlating with three carbons ( Figure S13). In the originally considered structure (Figure 2a), only three hydroxyl groups are within this range; the carboxylic acid group is expected to shift the fourth labile proton downfield. Additionally, HMBC signals from δ H 9.89 to C-7, C-8, and C-9 indicated an alcohol group at C-8. Since none of the labile protons correlated with carbonyl C-6, we concluded that it was part of the third ring. Compound 4 was thus identified as 2,3,8,9-tetrahydroxybenzo[c]chromen-6-one and named cladophorol A. This compound was previously reported as a synthetic analogue of ellagic acid created to improve potency for DNA gyrase inhibition, but the spectroscopic data were not disclosed at that time. 18 Characterization of 5. Compound 5 was obtained as a pale red amorphous solid with a molecular formula of C 25 H 16 O 9 deduced by HRESIMS. The 1 H and COSY NMR spectra exhibited resonances accounting for four aromatic rings. 1 H− 1 H coupling constants were used to establish substitution patterns including one 1,2,3,5-tetrasubstituted ring, one 1,2,3,4-tetrasubstituted ring, and two 1,2,4-trisubstituted rings. HMBC correlations from H-1 to δ C 123.1 (C-10a), 136.3 (C-4a), and 155.6 (C-2), from H-7 to C-10a, δ C 152.7 (C-9), and 163.4 (C-6), and from H-8 to 114.5 (C-6a) and 144.3 (C-10) indicated the first two rings were connected in a benzo[c]chromen-6-one moiety similar to 4 ( Figure S2), although the location of hydroxy groups differed with substitution at positions C-2, C-4, C-9, and C-10. The two other rings formed a biphenyl group based on HMBC correlations from H-2 A and H-6 A to δ C 133.9 (C-1 B ), from H-5 A to δ C 139.6 (C-1 A ), from H-2 B and H-6 B to C-1 A , and from H-5 B to C-1 B . Despite careful examination of the HMBC spectra, a connecting point between the biphenyl and benzo[c]chromen-6-one could not be identified. Whereas a 4 J correlation was reported between H-5′ and C-4 for 1, 16 a similar correlation was not found in 5, despite recording HMBC spectra in DMSO-d 6 or acetone-d 6 and optimizing for 2 Hz long-range coupling. Adding picric acid to DMSO-d 6 for the NMR acquisition led to six labile proton signals ( Figure  S23) which showed correlations with all carbons bearing an alcohol group and their neighbor ( Figure S25). The four singlets at δ H 11.04, 9.99, 9.68, and 9.57 exhibited HMBC correlations with δ C 106.9 (C-3), 146.6 (C-4), C-4a, 116.4 (C-8), C-9, C-10, C-10a, 116.0 (C-2 B ), 150.1 (C-3 B ), and 144.0 (C-4 B ), clarifying that the benzo[c]chromen-6-one and the biphenyl were linked via an oxygen attached to C-2 and C-4 B . For this reason, 5 was identified as 4,9,10-trihydroxy-2-[(3,3′,4′-trihydroxy-(1,1′-biphenyl)-4-yl)oxy]benzo[c]chromen-6-one and named cladophorol B.
Structure Revision of 1. Since 4 and 5 were well characterized, the validity of the structure previously reported for 1 was questioned. 16 When recorded in DMSO-d 6 , the carboxyl carbons of 4 and 5, which are part of a lactone ring, produced NMR signals at δ C 164.5 and 160.3, respectively. The carboxyl carbon of 1 was reported at δ C 161.1, a value upfield of expectations for a free carboxylic acid. 16 Table 2). The largest discrepancies were due to C-4a and C-6 which are involved in the lactone ring. The hypothetical structures 13 and 14 showed smaller deviations for δ C with rmsd of 1.7 and 1.8 ppm, respectively, but only 13 exhibited a small δ H rmsd (0.08 ppm). When DP4+ probabilities were calculated, structure 13 scored 100% when considering chemical shift data for both 1

The Journal of Organic Chemistry
Article allowed assignment of all chemical shifts and coupling constants (Table 3). Using DQF-COSY and J-resolved spectroscopy, seven AMX aromatic spin systems resulting from 1,2,4-trihydroxyphenyl groups were observed as doublets at δ H 6.67−6.91 and 6.42−6.57 and doublet of doublets at δ H 6.27−6.47. Interestingly, within each of these ranges, a gradient of proton chemical shifts was noted whereby each set of signals from a given ring appeared in a predictable order. This feature was highlighted by simulating each set of protons using the extracted δ C and J from NMR spectra ( Figure 4). When comparing proton signals for H-3, H-5, and H-6, the two signals corresponding to rings Y and Z appeared to be outliers.
Most of the carbon signals were resolved in the 13 C NMR spectrum, although HSQC and HMBC spectra showed large clusters of correlations between protons of rings D−G and carbons averaged at 157.5 (C-4 D−G ), δ C 151.2 (C-2 D−G ), 139.6 (C-1 D−G ), 123.1 (C-6 D−G ), 108.8 (C-5 D−G ), and 106.4 (C-3 D−G ). This prevented a precise assignment of signals but also indicated these rings were similar. Two other spin systems unrelated to the gradient of chemical shifts showed HMBC correlations from H-2 A to δ C 139.9 (C-1 B ) and from H-2 B and H-6 B to δ C 133.9 (C-1 A ) allowing assignment of these two spin systems to a biphenyl unit ( Figure S2). Given all these data, The Journal of Organic Chemistry Article two structures were hypothesized, differing by the position of the free alcohol groups on each hydroxyphenyl ( Figure 5).
Two model structures, 15a and 15b, containing a biphenyl unit and three 1,2,4-trihydroxyphenyl rings ( Figure 5), were computationally tested to determine which arrangement was more probable. Their 1 H and 13 C NMR chemical shifts were predicted (Tables S26 and S27) following DFT computation at the mPW1PW91/6-311+G(d,p) level of theory and compared to the corresponding position of 6. The rmsd and MAE values calculated for both 1 H and 13 C were lower when 15a was considered, indicating that 6a is the correct structure. Based on these results, 6 was identified as α-hydro-ω-[3,4dihydroxyphenyl]octa[oxy(2-hydroxyphen-4-yl)] and named cladophorol C.
Characterization of 7. The molecular formula for 7 was established as C 60 H 42 O 20 based on HRESIMS, accounting for one additional dihydroxyphenyl moiety. The 1 H and 13 C NMR spectra almost perfectly superimposed those of 6, the only differences being three additional proton signals at δ H 6.86 (d, J = 8.8 Hz), 6.50 (d, J = 2.9 Hz), and 6.40 (dd, J = 8.8 Hz) and six additional carbon signals at δ C 157.6, 151.2 3 , 139.5 7 , 123.1 7 , 108.7 7 , and 106.5. As a result, 7 was identified as α-hydro-ω-  Rings Y and Z are the last two rings of the oligomer. b13 C signal can be interchanged with one of the corresponding signals of rings D−G.

The Journal of Organic Chemistry
Article suggesting that 8 was also an oligomeric phenolic with a terminal biphenyl. A limited spread of the signals for the repeating phenol units, as illustrated by the 1 H spectral comparison of H-6 ( Figure 6), suggested an end-capping of the oligomer with a different moiety. A carbonyl, indicated by a peak on the 13 C NMR spectrum at δ C 163.4, and 12 other distinct 13 C NMR signals, agreed with a benzo[c]chromen-6one unit ( Table 4). The substitution pattern was determined as 2,4,8,9-tetrahydroxy-from the presence of two distinct singlets at δ H 7.63 and 7.32 and two doublets at δ H 6.96 and 6.55 (both J = 2.7 Hz) in the 1 H NMR spectrum (Table 5). This assignment was corroborated by HMBC correlations from H-1 and H-3 to δ C 136.2 (C-4a) and from H-1 to δ C 130.7 (C-10a) ( Figure S2). This benzo[c]chromen-6-one group matched revised structures 13 and 14; however, the exact point of attachment of the oligomer needed to be established to determine which configuration was more probable for 8. A comparison of 13 C NMR values of 8 recorded in DMSO-d 6 and those reported for 13 16 (Table S1) showed the greatest differences at C-2, C-4, and C-6, suggesting the oligomer was attached to C-2 as in 14 instead of C-4. Characterization of 9−12. Compounds 9−12 each exhibited 1 H and 13 C NMR spectra similar to that of 8 (Tables 4 and 5). Proton integrations in the three overlapping regions (δ H 6.85−6.98, 6.49−6.56, and 6.39−6.45) indicated three fewer protons for 9 and 3, 6, and 9 additional protons for 10−12. This suggested 9−12 were members of a series with 8,    (1.4−9 μg/mL)). The total number of phenol rings is also an important factor since the most potent compounds in both series (6−7 vs 8−12) have 9−10 rings and the activity weakens as this number increases. Surprisingly, all of the other cell lines, including the other Gram-positive bacterium, VREF, were much less sensitive to this family of natural products. The bioactivities of 4−12 were also evaluated in an assay using intraerythrocytic Plasmodium falciparum. 22 Only 4 and 5 inhibited the parasite with half-maximal effective concentration (EC 50 ) values of 0.7 and 1.9 μg/mL, respectively. In addition, the cytotoxicities of 3−12 were evaluated using immortalized human keratinocytes (HaCaT) and human kidney cells (HEK293T). Although compounds 5−11 exhibited weak toxicity (<25%) at their respective maximum concentrations against HaCaT (Figure 7) and 4 and 5 had weak toxicity toward HEK293T (half maximal cytotoxicity concentration (CC 50 ) of 9.4 and 4.5 μg/mL, respectively), no toxicity was observed at the MRSA MICs for 5−11 or the P. falciparum EC 50 values for 4 and 5. The Journal of Organic Chemistry Article ■ EXPERIMENTAL DETAILS General Experimental Procedures. NMR spectra ( 1 H, HMBC, HSQC, DQF-COSY, and J-resolved) were recorded on an 18.8 T (800 MHz for 1 H and 200 MHz for 13 C) Bruker Avance IIIHD 800 instrument equipped with a 5 mm triple resonance broadband cryoprobe. All spectra were acquired in CD 3 OD or DMSO-d 6 , and chemical shifts were reported in ppm (δ) relative to the residual solvent peaks (δ H 3.31 and δ C 49.00 for CD 3 OD, δ H 2.50 and δ C 39.52 for DMSO-d 6 ). All spectra were processed using MestReNova 11.0. HPLC separations were performed with a Waters 1525 binary pump and a Waters 2487 dual wavelength absorbance detector set at 260 nm, using two different columns: a 9.4 × 250 mm, C 18 silica reversed-phase (Zorbax stable-bond, 5 μm particle size), and a 4.6 × 250 mm phenylhexyl phase (Phenomenex Luna, 5 μm particle size). High-resolution mass spectrometry was conducted on an Orbitrap spectrometer in negative ion mode. The masses of isolated compounds were estimated by qNMR using a capillary filled with benzene-d 6 and calibrated against caffeine. 23 Specimen Collection. Green alga (collection G-1240) was harvested as floating clumps on Dec 3, 2015, at the ocean surface near Titi Island, Viti Levu, Fiji (16°16′26″ S, 179°26′02″ E). This collection was identified as Cladophora socialis Kuẗzing by morphological and 18S rRNA phylogenetic analyses. Voucher specimens were preserved in aqueous formaldehyde and stored at the University of South Pacific. The collection was stored at −80°C until extraction.
Species Identification. The collected green alga (G-1240) was identified by comparing its morphological traits with that of previously described Cladophora species 24 and by sequence analysis of nuclear, small subunit (SSU) rRNA (18S rRNA). Genomic DNA from an ethanol-preserved algal specimen was extracted using the innuPREP plant DNA kit (Analytik Jena, Germany) according to the manufacturer's protocol. The three overlapping 18S rRNA gene fragments from genomic DNA were amplified via the polymerase chain reaction (PCR) in three separate reactions using universal primer pairs, NS3/NS4, NS5/NS6, and NS7/NS8. 25 Each PCR amplification was performed in a 25 μL reaction volume consisting of The Journal of Organic Chemistry . All PCR amplifications were performed in a GeneAmp PCR system 2700 (Applied Biosystems, Foster City, CA) thermocycler using the following temperature cycling parameters: initial denaturation at 94°C for 5 min followed by a total of 40 cycles of amplification in which each cycle consisted of denaturation at 94°C for 40 s, primer annealing at 50°C for 40 s and primer extension at 72°C for 1 min. After amplification, final extension of the incompletely synthesized DNA was carried out at 72°C for 7 min. The PCR fragments were analyzed by agarose gel electrophoresis (1% wt/vol). The gel was stained with ethidium bromide and visualized under a UV transilluminator. All of the PCR fragments were either sequenced with forward or reverse primers, and sequences were manually edited and assembled using CAP3 Sequence Assembly Program. 26 The assembled G-1240 18S rRNA sequence (1334 bp) was submitted to GenBank (accession no. MK127549). The sequence similarity of the assembled contig of G-1240 18S rRNA to other known Cladophora spp. was determined by comparing it with the nonredundant nucleotide database (NCBI) using the blastn program. 27 The top ranked matches according to E values and maximum scores revealed high similarity to C. socialis with 96% identity.
Further, the 18S rRNA sequence of G-1240 was compared with those of other known species of the genus Cladophora and two outgroup taxa Ulva curvata and Ulothrix zonata from the same class but different orders, obtained from GenBank. Phylogenetic analysis was conducted in MEGA X 28 using the Maximum Likelihood method based on the Tamura−Nei model 29 with 1000 bootstrap iterations. The phylogenetic analysis revealed that G-1240 is closely related to C. socialis ( Figure S1). Overall, morphological and phylogenetic analyses are consistent with identification of the green alga G-1240 as C. socialis.
Cladophorol A (4): pale red amorphous solid; 1 H and 13 C NMR data, see Table 1 (1, 13, 14, 15a, and 15b) were subjected to a Monte Carlo conformational search using the MMFF94 force field in Spartan '16. 30 The geometry of all conformers within 21 kJ/mol were optimized using density functional theory (DFT) in the Gaussian D09 package, 31 using the B3LYP/6-31G(d) level of theory. Solvent interactions, either DMSO (1,13,14) or MeOH (15a, 15b), were considered with PCM. Frequencies computation was performed to ascertain true minima were obtained (zero imaginary frequency), and the resulting free energies were extracted to calculate the Boltzmann distribution. All conformers representing more than 2% of the population were considered for the computation of the shielding tensors which was performed at the mPW1PW91/6-311+G(d,p) level of theory. The shielding tensors (GIAO) were averaged using their respective Boltzmann weight, and the chemical shifts were finally derived by scaling the shielding tensors against the experimental values.
Pharmacological Assays. Antibacterial and antifungal assays were performed as previously described 32
The human keratinocyte cell line (HaCaT) was maintained, and cytotoxicity of isolated compounds were assessed using LDH cytotoxicity assay as previously described. 34 Antimalarial activity was assessed using a standard parasite proliferation assay with asexual blood-stage P. falciparum and SYBR Green detection. 22 Screening media (complete media supplemented with 0.05% Albumax II but without human serum) and fresh O + erythrocytes (TSRI Normal Blood Donation) were used for cultures of P. falciparum strain Dd2L (gift from Dr. David Fidock, Colombia University). A Labcyte ECHO acoustic liquid handler transferred compounds into assay plates. The plates were inoculated (Multi-Flo; BioTek, VT) with parasitized erythrocytes and fresh erythrocytes prepared in screening media, resulting in a final parasitemia of 0.3% and hematocrit of 2.5%. A chamber with a low oxygen gas mixture at 37°C was used to culture the assay plates for 72 h with daily gas exchanges. SYBR Green lysis buffer was added to the wells after incubation using a Multi-Flo liquid dispenser, and plates were incubated at room temperature for an additional 24 h to achieve optimal development of the fluorescence signal which was read by an Envision Multimode Reader (PerkinElmer, MA). The positive controls used were atovaquone (EC 50 = 0.0006 μM) and artemisinin (EC 50 = 0.03 μM).

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.8b03218.