Chemical Profiling (HPLC-NMR & HPLC-MS), Isolation, and Identification of Bioactive Meroditerpenoids from the Southern Australian Marine Brown Alga Sargassum paradoxum

A phytochemical investigation of a southern Australian marine brown alga, Sargassum paradoxum, resulted in the isolation and identification of four new (5, 9, 10, and 15) and nine previously reported (1, 2, 6–8, and 11–14) bioactive meroditerpenoids. HPLC-NMR and HPLC-MS were central to the identification of a new unstable compound, sargahydroquinal (9), and pivotal in the deconvolution of eight (1, 2, 5–7, and 10–12) other meroditerpenoids. In particular, the complete characterization and identification of the two main constituents (1 and 2) in the crude dichloromethane extract was achieved using stop-flow HPLC-NMR and HPLC-MS. This study resulted in the first acquisition of gHMBCAD NMR spectra in the stop-flow HPLC-NMR mode for a system solely equipped with a 60 μL HPLC-NMR flow cell without the use of a cold probe, microcoil, or any pre-concentration.


Introduction
Terpenoids and phenolic compounds represent the largest group of secondary metabolites reported from marine brown algae [1]. Marine brown algae belonging to the Sargassum genus (Sargassaceae, Fucales) number in excess of 300 species [2] located throughout most regions of the world [3,4], and are known to produce secondary metabolites such as meroterpenoids [1,5]. Meroditerpenoids are metabolites of mixed biogenesis containing terpenoid and non-terpenoid derived fragments. Meroditerpenoids derived from algae of the genus Sargassum are comprised of a polyprenyl chain attached to either a p-benzoquinone or hydroquinone moiety [6][7][8]. Variations of the meroditerpenoid structures occur mainly in the terpene side chain and include the addition of exocyclic double bonds, carboxylic acids, alcohols, or aldehyde functional groups [1,7,9]. Meroditerpenoids have been reported to display a range of biological activities including antibacterial, antioxidant, and antitumor activities [6,7,[10][11][12].
We recently examined a specimen of the marine brown alga Sargassum paradoxum (R. Brown ex Turner) Hooker and Harvey, collected from Port Phillip Bay, Victoria, Australia. This alga was selected for chemical investigation on the basis of the crude extract antimicrobial activity and that no previous chemistry had been reported from this particular Sargassum species [1,13]. Previous studies conducted by our research group on another closely related Sargassum species (S. fallax) resulted in the isolation of three new meroditerpenoids [7]. This also provided the motivation for studying this Sargassum species in that a comparison of the two species could potentially allow for chemotaxonomic markers to be identified. Herein we report the chemical profiling study conducted using HPLC-NMR and HPLC-MS together with the bioassay-guided isolation and structure determination of one new meroditerpenoid with a hydroquinone moiety paradoxhydroquinone (5) and two new meroditerpenoids possessing a p-benzoquinone moiety, paradoxquinol (10) and paradoxquinone (15). A new unstable meroditerpenoid, sargahydroquinal (9), was identified on the basis of the HPLC-NMR and HPLC-MS studies conducted. Revisions in the proton and/or carbon NMR chemical shift assignments for three of the meroditerpenoids (1, 2, and 13) have also been documented. In all, 14 meroditerpenoids ( Figure 1) were detected in the crude extract of S. paradoxum and, of these, 13 could be identified using a combination of on-line (HPLC-NMR and HPLC-MS) and off-line (conventional isolation and structure elucidation) methodologies.

Results and Discussion
The frozen marine alga was extracted with 3:1 methanol/dichloromethane, evaporated under reduced pressure, and sequentially solvent partitioned (triturated) into dichloromethane-and methanol-soluble fractions, respectively. While both the dichloromethane-and methanol-soluble fractions displayed some antimicrobial activity, the dichloromethane extract was slightly more selective in its activity (Section 2.3). 1 H NMR and analytical HPLC analyses indicated that the dichloromethane extract contained a series of structurally related compounds. On the basis of these observations, only the dichloromethane extract was further examined.

Chemical Profiling (HPLC-NMR & HPLC-MS)
The dichloromethane extract was subjected to both HPLC-NMR and HPLC-MS chemical profiling and a total of 10 peaks (A-J) were detected ( Figure 2). Analysis of the stop-flow WET1D proton NMR spectra showed the presence of characteristic proton NMR signals in the aromatic and olefinic regions, together with various upfield proton NMR signals ( Figure 2). Examination of the UV profile extracted for each of the peaks detected by PDA showed the presence of two distinct UV chromophores at λmax 255 and 290 nm, typical of p-benzoquinones and hydroquinones that are known to occur in this genera of algae [1,13]. The two main peaks that were present in the dichloromethane crude extract of S. paradoxum (peaks A and B in Figure 2) were completely identified and characterized by HPLC-NMR and HPLC-MS. Both compounds displayed a UV maxima at 290 nm, supporting the presence of a hydroquinone moiety [14].
The HPLC-NMR stop-flow WET1D proton NMR spectrum of peak A ( Figure 2) showed two meta-coupled protons at δH 7.22 and 7.25, indicating the presence of an aromatic moiety displaying an AB splitting pattern as well as deshielded proton signals at δH 4.95, 4.88, and 4.87. Complete analysis of the 1D and 2D NMR data (Table 1) obtained via stop-flow HPLC-NMR, and the high resolution  ESI MS obtained for this peak via HPLC-MS analysis (m/z 427.2846 [M − H] − ), allowed for the structure to be unequivocally elucidated as the previously described meroditerpenoid fallahydroquinone (1) [7].
The HPLC-NMR stop-flow WET1D proton NMR spectrum of peak B (Figure 2) supported the presence of a meta-coupled aromatic AB system (δH 7.23 and 7.26). Also present was a deshielded olefinic proton resonance (δH 6.69). The gHMBCAD NMR spectrum obtained in the stop-flow HPLC-NMR mode showed a signal at δC 170.9 ppm, confirming a carboxylic acid functional group (Table 1). Complete analysis of the 1D and 2D NMR data obtained via stop-flow HPLC-NMR and the high-resolution ESI MS obtained via HPLC-MS (m/z 425.2694 [M − H] − ) suggested the previously described meroditerpenoid sargahydroquinoic acid (2) [8]. While a direct comparison of the proton NMR chemical shifts obtained via stop-flow HPLC-NMR to the NMR data reported in the literature is impeded by the expected chemical shifts differences that occur for spectra acquired using off-line conventional deuterated NMR solvents compared to HPLC-NMR solvents, the majority of the proton NMR signals for this compound varied from the literature data by a consistent factor. However, the proton chemical shift for position H-10′ (δH 6.69) differed to that reported for sargahydroquinoic acid (2) (δH 6.02) [8]. To confirm the 10′/11′ double bond configuration, single irradiation nOe NMR experiments were carried out whilst in the stop-flow HPLC-NMR mode. The proton at δH 6.69 was irradiated and a key nOe enhancement observed to the methylene protons at δH 3.02 (H-12′). This provided confirmation that peak B ( Figure 2) has a 10′Z double bond configuration, which corresponds to the structure assigned to sargahydroquinoic acid. Further confirmation of the 10′Z double bond configuration came from comparison to the meroditerpenoids chabrolohydroxybenzaquinone A (3) and C (4), which were isolated as co-occurring double bond isomers from the soft coral Nephthea charoli [15]. These compounds show that there is a characteristic difference in the carbon chemical shift of the C-12' methylene (δC 26.7 ppm and 34.5 ppm for the 10′E and 10′Z isomers, respectively) [15]. This γ effect is particularly valuable for distinguishing E and Z isomers of trisubstitued alkenes. Since the carbon for the C-12′ methylene in (2) displayed a carbon chemical shift at δC 35.3, this provided further confirmation for the 10′Z configuration.
a Carbon NMR assignments made on the basis of gHSQCAD and gHMBCAD NMR experiments; b Signal not detected due to signal suppression; c Signal suppressed.
The HPLC-NMR stop-flow WET1D proton NMR spectra of peaks C and D resembled that of peak A [fallahydroquinone (1)] (Figure 2), both of which eluted closely. Upon closer examination, it was evident that peak D was a meroditerpenoid that contained one alcohol functional group, due to the presence of a deshielded methylene proton at δH 4.84. Examination of the high resolution-ESI MS obtained via HPLC-MS for peaks C and D indicated that these were isobaric and that they both possessed a UV maxima at 290 nm, supportive of a hydroquinone moiety [14]. On the basis of the mass difference and the proton NMR differences it was proposed that peak D contained a terminal alcohol moiety but, unlike fallahydroquinone (1), did not contain a secondary alcohol functional group; this led to peak D being identified as the previously reported compound 2-[11-(hydroxymethyl)-3,7,15-trimethyl-2,6,10,14-hexadecatetraen-1-yl]-6-methyl-1,4-benzenediol (6) [16]. This represents the first report of 6 occurring as a natural product. Due to suppression of the HDO peak, no deshielded proton resonances could be observed for peak C; however, in consideration of the slight elution time difference, and the identical molecular masses, it was suspected that peak C contained an oxymethine group. While in stop-flow HPLC-NMR, the closely eluting compounds showed diffusion, and the oxymethine signal at δH 4.84 could clearly be distinguished via on-flow HPLC-NMR as being present in peak D and absent in peak C (see Supplementary Information file). The structure of peak C was proposed to be the new meroditerpenoid 2-[12-hydroxy-3,7,11, 15-tetramethyl-2,6,10,14-hexadecatetraen-1-yl]-6-methyl-1,4-benzenediol, attributed the trivial name paradoxhydroquinone (5). Although subsequent off-line isolation of peaks C and D afforded an inseparable mixture, identification of these peaks was possible via the chemical profiling approach since individual WET1D proton and gCOSY NMR data could be obtained separately for each peak. This highlights an important feature of HPLC-NMR: typically co-eluting secondary metabolites that are difficult to obtain separately by conventional off-line isolation approaches can be obtained as separate entities via stop-flow HPLC-NMR.
Peak E was determined to be the previously described meroditerpenoid fallaquinone (7) [7], based on the observed UV maxima, together with the stop-flow WET1D proton NMR spectrum and the high-resolution ESI MS obtained via HPLC-MS, which showed the presence of an ion at m/z On the basis of the HPLC-NMR stop-flow WET1D proton NMR spectrum of peak F and the UV profile (λmax 290 nm), this compound also possessed a hydroquinone moiety [14]. The high-resolution ESI MS ion m/z of 409.2745 [M − H] − obtained via HPLC-MS suggested a molecular formula of C27H38O3. The stop-flow WET1D proton NMR spectrum showed a distinct proton signal at δH 10.07, s, confirming the presence of an aldehyde moiety. The MarinLit database indicated that sargaquinal (8), which possesses a p-benzoquinone moiety and an aldehyde functional group, had been previously reported from the Sargassum genus [17]. Peak F was proposed as being the new meroditerpenoid (possessing a hydroquinone moiety), attributed the name sargahydroquinal (9). During subsequent off-line bioassay-guided isolation, sargahydroquinal (9) oxidized into sargaquinal (8) before any off-line data could be obtained. Owing to the low abundance of sargahydroquinal (9) in the crude dichloromethane extract, only the stop-flow WET1D proton NMR and gCOSY data were obtained in the HPLC-NMR mode for sargahydroquinal (9) and are provided in Table 2. By comparison of the off-line data for sargaquinal (8) to peak F, the structure of sargahydroquinal (9) could be established. The configuration of the double bonds for sargaquinal (8) have been reported as 2′E, 6′E, and 10′E [17].
A bsNOESY NMR experiment carried out for sargaquinal (8) revealed nOe enhancements between the olefinic protons at positions H-2′ and H-6′ to the methylene protons at positions H-4′ and H-8′, along with the olefinic proton at position H-10′ showing a nOe enhancement to the aldehyde proton at position H-20′. On biosynthetic grounds, the same configuration was attributed to the double bonds in sargahydroquinal (9). The UV profile (λmax 255 nm) observed for peak G in the HPLC-NMR and HPLC-MS analyses confirmed a p-benzoquinone moiety [14]. The HPLC-NMR stop-flow WET1D proton NMR spectrum of peak G showed close resemblance to fallaquinone (7), except that an aromatic proton NMR resonance was absent. Comparison of the mass of peak G to fallaquinone (7), observed from ESI MS obtained via HPLC-MS analysis, showed a difference of 14 mass units, indicating that one of the aromatic proton signals had been replaced by a methyl substituent. The structure of peak G was therefore proposed to be the new meroditerpenoid 5-[12R-hydroxy-11-(hydroxymethyl)-3,7, 15-trimethyl-2,6,10,14-hexadecatetraen-1-yl]-2,3-dimethyl-1,4-benzoquinone, attributed the trivial name paradoxquinol (10).
Peak I showed close similarities to sargahydroquinoic acid (2) and was identified as the previously reported sargahydroquinoic acid (11) [7,17], the benzoquinone counter part of 2.
Peak J possessed a p-benzoquinone moiety from the UV chromophore at 263 nm [14]. Analysis of the HPLC-NMR stop-flow WET1D NMR data showed close resemblance to 6. On the basis of the molecular ion deduced from the high-resolution ESI MS (409.2742 [M − H] − ) obtained via HPLC-MS, in consultation with a SciFinder search, peak J was identified as previously described meroditerpenoid 2-[11-(hydroxymethyl)-3,7,15-trimethyl-2,6,10,14-hexadecatetraen-1-yl]-6-methyl-1,4-benzoquinone (12) [16,17]. Although subsequent off-line isolation afforded 12 and 15 as an inseparable mixture, identification of these compounds was possible via the chemical profiling approach since individual WET1D and gCOSY NMR data could be obtained separately for 12, which could then be related back to the mixture.

Off-Line Bioassay-Guided Isolation
Following the chemical profiling (HPLC-NMR and HPLC-MS) conducted on the dichloromethane crude extract, a bioassay-guided isolation approach was undertaken to deduce the nature of the bioactive constituent(s) responsible for the crude extract bioactivity. The off-line isolation also allowed for any further identification and characterization of the meroditerpenoids concluded on the basis of the chemical profiling analysis as well as identifying any additional minor constituents not detected by HPLC-NMR.
A portion of the dichloromethane extract was fractionated by silica gel flash chromatography and selected fractions were evaluated for antimicrobial activity. Fractions for further purification were selected based on the observed antimicrobial activity, analysis of the 1 H NMR spectra, and the analytical HPLC chromatograms. Purification was conducted using reversed phase HPLC resulting in the isolation of 12 meroditerpenoids, eight (1, 2, 5-7, and 10-12) of which were previously detected by HPLC-NMR and HPLC-MS analyses, and four of which were additional meroditerpenoids (8 and 13-15) that were not observed in the chemical profiling. Nine known meroditerpenoids (1, 2, 6-8, and 11-14) were isolated and identified by direct comparison of their NMR spectroscopic and mass spectrometry data with that reported in the literature [7,8,[16][17][18][19]. The off-line study also confirmed the structures of 1, 2, 5-7, and 9-12 as deduced by HPLC-NMR and HPLC-MS.
Fallahydroquinone (1) was isolated as an unstable yellow oil. Analysis of the 2D NMR data showed that the carbon NMR chemical shifts for positions C-2 and C-6, reported as occurring at δC 125.4 and 127.8, respectively, should be reassigned to δC 127.8 (C-2) and δC 125.4 (C-6), respectively. In addition, the proton NMR signals for the position H-4′ and H-8′ protons reported at δH 2.01 and 2.08, respectively, should also be reversed. The specific rotations of the previously reported meroditerpenoids fallahydroquinone (1) and fallaquinone (7) were compared to literature values [7]. While the specific rotation for fallahydroquinone (1) was comparable in sign and magnitude, the specific rotation of fallaquinone (7) differed to that reported. The purity of fallaquinone (7) in this study was higher than that obtained in the original isolation. It was concluded that the impurities present were sufficient to cause the specific rotation to vary in sign and magnitude. The accuracy of the specific rotations was considered given the low quantities isolated. The relevant compounds isolated in this study were of high purity, and did not display small specific rotations. That both the specific rotation of both fallahydroquinone (1) and fallaquinone (7) are of the same sign and magnitude is consistent with the fact that the former oxidizes to the latter with retention of configuration.
Comparison of the 2D NMR data of sargaquinone (13) with the literature data indicated discrepancies with some of the carbon NMR chemical shift assignments. The carbon NMR chemical shifts for positions C-3, C-5, C-3′, and C-15′ were either incorrectly assigned or not assigned. These carbon NMR chemical shifts have been unequivocally established as occurring at δC 132.3, 133.2, 140.0 and 131.3, respectively.
Sargahydroquinal (9) and one other unidentified peak (peak H) ( Figure 2) degraded during the off-line isolation, which meant that it was not possible to conduct a complete 2D NMR or antimicrobial activity evaluation on these constituents.

Paradoxquinol (10)
Paradoxquinone ( The compounds isolated from S. paradoxum in this study also occur in many other Sargassum species. However, two of the compounds-fallahydroquinone (1) and fallaquinone (7)-have only been reported from one other Sargassum species (S. fallax) [7]. On this basis, it is proposed that the presence of these two compounds could potentially be used as chemotaxonomic biomarkers for S. paradoxum and S. fallax.
Antimicrobial evaluation of the crude extracts, selected silica column fractions, and the isolated meroditerpenoids was carried out against five bacteria and one fungus. Eleven of the isolated meroditerpenoids (1-2, 5-7, 9-13, and 15), together with a commercially available antibiotic and antifungal agent, were tested at 1 mg/mL, with all displaying varying degrees of zones of inhibition in the antimicrobial assays (Table 5). Compared to the antibiotic ampicillin, the isolated compounds were far less potent against S. aureus and S. pyogenes. However, compounds 2, 7, 8, and 13 were more potent against P. aeruginosa than ampicillin. In recent years, concern has grown around P. aeruginosa with regards to its prevalence and multidrug resistance [40]; therefore, the importance of identifying possible drug leads against these bacteria remains highly relevant. There was no difference in activity between compounds with the hydroquinone or the p-benzoquinone moieties. The activity observed for sargaquinone (13), the simplest of the meroditerpenoids isolated, suggests that the unsubstituted meroditerpenoid skeleton is responsible for the activity against P. aeruginosa. The addition of an alcohol group at position 12′ or 20′ (1, 5, 6, 12, and 15) appears to reduce the activity against P. aeruginosa, but increases the activity against S. pyogenes. Finally, incorporation of a carboxylic acid at position C-20′ (2 and 11) gives rise to activity against S. aureus and S. aureus MRSA.

General Experimental Procedures
All organic solvents used were analytical reagent (AR or GR), UV spectroscopic, or HPLC grades with milli-Q water also being used. Optical rotations were carried out using a 1.5-mL cell on a Rudolph Research Analytical Autopol IV automatic polarimeter, set to the Na 589 nm wavelength. UV/VIS spectra were recorded on an Agilent CARY 60 spectrophotometer, using ethanol. CD spectra were obtained on a Jasco 815 Circular Dichroism spectrometer in ethanol. FTIR spectra were recorded as a film using a NaCl disk on a Perkin-Elmer Spectrum One FTIR spectrometer. 1 H (500 MHz), 13 C (125 MHz), and 1D nOe spectra were acquired in CDCl3 on a 500 MHz Agilent DD2 NMR spectrometer with referencing to solvent signals (δ 7.26 and 77.0 ppm). Two-dimensional NMR experiments recorded included gradient correlation spectroscopy (gCOSY), heteronuclear single-quantum correlation spectroscopy with adiabatic pulses (HSQCAD), gradient heteronuclear multiple-bond spectroscopy with adiabatic pulses (gHMBCAD), and band selective 2D nuclear overhauser enhancement spectroscopy (bsNOESY) experiments. ESI mass spectra were obtained on a Micromass Platform II mass spectrometer equipped with a LC-10AD Shimadzu solvent delivery module (50% CH3CN/H2O at a flow rate of 0.2 mL/min) in both the positive and negative ionization modes using cone voltages between 20 and 30 V. Silica gel flash chromatography was carried out using Davisil LC35Å silica gel (40-60 mesh) with a 20% stepwise solvent elution from 100% petroleum spirits (60-80 °C) to 100% dichloromethane to 100% ethyl acetate and finally to 100% methanol. HPLC-NMR was carried out on a 500 MHz Agilent DD2 NMR spectrometer equipped with a Varian 1 H[ 13 C] pulsed field gradient flow probe with a 60-μL active volume flow cell coupled to a Varian Prostar 210 solvent delivery system, a Prostar 430 Autosampler, and a Prostar 335 PDA detector. The HPLC-NMR analyses were carried out using CORBA communication and operated with VnmrJ software. The 2H resonance observed from the D2O was used to obtain a field-frequency lock. The resonances from the HOD and the methyl of the acetonitrile were suppressed using the water enhanced through transverse gradients (WET) solvent suppression experiment [41]. The residual HOD resonance of D2O was referenced to 4.64 ppm. For both on-flow and stop-flow HPLC-NMR modes, 50-μL injections (4995 µg) of the dichloromethane extract were injected onto an Agilent Eclipse Plus C18 (150 × 4.6) 5-µ column using a solvent composition of 75% CH3CN/D2O at a flow rate of 1 mL/min. In the stop-flow HPLC-NMR mode WET1D, gCOSY, HSQCAD, gHMBCAD, and 1D nOe NMR experiments were acquired. HRESIMS was carried out on an Agilent 6200 Series TOF system (ESI operation conditions of 8 L/min N2, 325 °C drying gas temperature, and 3500 V capillary voltage) equipped with an Agilent 1200 Series LC solvent delivery module (100% CH3OH at a flow rate of 0.3 mL/min) in either the positive or negative ionization modes. The instrument was calibrated using the "Agilent Tuning Mix" with purine as the reference compound and the Hewlett-Packard standard HP0921. HRESILCMS was carried out on the same system and conditions using an Agilent Eclipse Plus C18 (4.6 × 150) 5-µ column using a solvent composition of 75% CH3CN/H2O at a flow rate of 1 mL/min. All analytical HPLC analyses and method development were performed on a Dionex P680 solvent delivery system equipped with a PDA100 UV detector (operated using "Chromeleon" software). Analytical HPLC analyses were carried out using either a gradient method 0-2 min 10% CH3CN/H2O; 14-24 min 75% CH3CN/H2O; 26-30 min 100% CH3CN; and 32-40 min 10% CH3CN/H2O or an isocratic method (either 100%, 95%, 90%, or 85% CH3CN/H2O) on an Alltech Alltima HP C18 (250 × 4.6) 5-µ column at a flow rate of 1.0 mL/min. Semi-preparative HPLC was carried out on a Varian Prostar 210 solvent delivery system equipped with a Prostar 335 PDA detector (operated using "Star Workstation" software) using an isocratic method (either 100%, 95%, 90%, or 85% CH3CN/H2O) and an Alltech Alltima C18 (250 × 10) 5-µ column at a flow rate of 3.5 mL/min.

Biological Evaluation
Crude extracts, enriched fractions, and compounds isolated from the marine alga, together with the standard antibiotic (ampicillin, Sigma-Aldrich, Castle Hill, Australia) and antifungal (carbendazim, Sigma-Aldrich, Castle Hill, Australia) compounds, were evaluated against six microorganisms (five bacteria and one fungus) at concentrations of 50, 25, or 1 mg/mL (see Table 5). Each microorganism was prepared by creating a 0.5 McFarlane solution suspension. Lawn cultures were prepared on either Mueller-Hinton or Brain Heart Infusion Agar (BHIA) (used for S. pyogenes). Then 20 µL of the crude extracts, enriched fractions, pure compounds or the standard compounds (ampicillin and carbendazim) were pipetted onto 6 mm diameter filter paper disks and their solvents evaporated. These disks were then placed onto the prepared lawn cultures and incubated at 37 °C overnight. Active antimicrobial samples displayed a zone of inhibition outside the disk, which was measured in mm as the radius of inhibition for each bacteria/fungi. The five test bacteria were Eschericha coli (ATCC 25922), Staphylococcus aureus (ATCC 25923), Staphylococcus aureus MRSA (344/2-32), Pseudomonas aeruginosa (ATCC 27853), and Streptococcus pyogenes (345/1). The test fungus was Candida albicans (ATCC 10231).

Marine Alga Material
The

Chemical Profiling
Chemical profiling was carried out on the dichloromethane soluble extract of the alga employing HPLC-NMR and HPLC-MS methodologies. Details of these analyses are provided in Section 3.1. The dichloromethane extract (77 mg) was dissolved in HPLC-NMR grade CH3CN (771 μL) and filtered through a 0.45 PTFE membrane filter (Grace Davison Discovery Sciences).