Analysis of Unusual Sulfated Constituents and Anti-infective Properties of Two Indonesian Mangroves, Lumnitzera littorea and Lumnitzera racemosa (Combretaceae)

: Lumnitzera littorea and Lumnitzera racemosa are mangrove species distributed widely along the Indonesian coasts. Besides their ecological importance, both are of interest owing to their wealth of natural products, some of which constitute potential sources for medicinal applications. We aimed to discover and characterize new anti-infective compounds, based on population-level sampling of both species from across the Indonesian Archipelago. Root metabolites were investigated by TLC, hyphenated LC-MS/MS and isolation, the internal transcribed spacer (ITS) region of rDNA was used for genetic characterization. Phytochemical characterization of both species revealed an unusual diversity in sulfated constituents with 3,3’,4’-tri- O -methyl-ellagic acid 4-sulfate representing the major compound in most samples. None of these compounds was previously reported for mangroves. Chemophenetic comparison of L. racemosa populations from different localities provided evolutionary information, as supported by molecular phylogenetic evidence. Samples of both species from particular locations exhibited anti-bacterial potential (Southern Nias Island and East Java against Gram-negative bacteria, Halmahera and Ternate Island against Gram-positive bacteria). In conclusion, Lumnitzera roots from natural mangrove stands represent a promising source for sulfated ellagic acid derivatives and further sulfur containing plant metabolites with potential human health beneﬁts.


Introduction
Mangrove forests represent a unique habitat that comprises salt-tolerant plant species (mostly trees), predominantly bordering tropical and subtropical coastlines [1]. Besides their ecological significance, mangrove plant species have a wide variety of economic uses, such as construction material, fodder or textiles [2,3]. In addition, many mangrove plant species possess medicinal value and have been used traditionally in several regions of the world [4][5][6][7]. Due to the tidal influence, mangrove soils contain high levels of sulfate which is connected to the occurrence of sulfate-reducing microbial communities [8]. Nevertheless, the investigation of bioactive natural sulfur compounds in mangrove species was so far completely neglected. ample, the extracts of endophytic fungi isolated from leaves of ten mangrove species from Thailand, including L. littorea, showed some cytotoxic activity against cancer cell lines [35]. In line with the agenda of discovering new anti-infective and neuroactive constituents while at the same time promoting the protection and sustainable development of mangrove ecosystems, our work was focused on two mangrove species, namely Lumnitzera littorea and L. racemosa. Our aim was to (1) investigate the diversity of natural products present in the roots of L. littorea and L. racemosa from Indonesia, (2) evaluate selected biological effects, and (3) investigate the phylogenetic relationships of the two species as well as the chemophenetic patterns of their natural products across Indonesia. Therefore, we combined a molecular phylogeny of Lumnitzera based on the internal transcribed spacer (ITS) region with phytochemical analyses by hyphenated chromatographic and tandem mass spectrometric techniques. Chromatographic separation using reversed phase HPLC connected to high resolution ESI-MS that allow the determination of accurate mass and elemental composition represents a suitable technique to identify sulfur containing metabolites [36]. For the calculation of the molecular composition of sulfur-containing compounds, the small negative mass defect of sulfur isotopes and the isotopic pattern of sulfur distinct from that of carbon, nitrogen and hydrogen can be applied [36]. Since sulfur in addition to the most abundant isotope 32 S (95%) possesses a 34 S isotope (4.2%), also the larger as usual M+2 peak contributes to the determination of sulfur in compounds or fragments. Nevertheless, for complete structure elucidation, compounds have to be isolated and characterized by NMR. Phylogenetic approaches can be useful for identifying plant lineages with potential medicinal properties [37,38], the interpretation of chemical profiles [39], and might be a powerful tool for discovering novel compounds or novel compound variants [40][41][42], including antibiotic sources [43,44].

Root Sample Extraction
Air-dried samples (root bark and fine roots) were ground using a ball mill (Retsch, MM400) for two minutes. In extraction experiments with n-hexane, ethyl acetate and methanol, the highest extract amount could be obtained with methanol. Therefore, 100 mg of each sample were vortexed with 5 mL methanol in Eppendorf tubes before sonication for an hour. All samples were centrifuged for fifteen minutes using a Megafuge 1.0R (Unity Lab Services) to gain pure solutions. The extracts were aliquoted for analytical investigations and bioassay screening. The crude extracts were directly used for TLC and low-resolution ESI-MS analyses. For LCMS measurements, 250 µL of each extract were purified by an SPE cartridge (RP18, Chromabond, Macherey-Nagel, Düren, Germany), using methanol as eluent, and the concentration was afterwards adjusted to 1 mg/mL.

General Experimental Procedures
Thin layer chromatography (TLC) analyses were done with silica gel 60 F 254 (Merck, Darmstadt, Germany) using the solvent system CHCl 3 :MeOH 6:4. Compound spots were visualized using long-wavelength UV light (366 nm), short-wavelength UV light (254 nm) and spraying with vanillin-H 2 SO 4 reagent, followed by heating. As sample LR15 in TLC displayed stronger spots compared to the others, preparative TLC (thickness 0.5 mm) was performed using the same conditions. The major bands were scraped off and extracted to verify the compound identity by MS investigations.
Low-resolution ESI-MS spectra were obtained from a Sciex API-3200 instrument (Applied Biosystems, Concord, ON, Canada) combined with an HTC-XT autosampler (CTC Analytics, Zwingen, Switzerland). 1 H and 13 C NMR spectra were recorded on an Agilent DD2 400 NMR spectrometer at 399.917 and 100.570 MHz, respectively. Chemical shifts are reported relative to TMS ( 1 H NMR) or peaks of solvent ( 13 C, CD 3 OD 49.0 ppm and DMSO-d 6 39.5 ppm). For samples with low concentration, 1 H and 13 C NMR spectra were recorded on a Bruker Avance Neo 500 NMR spectrometer at 500.234 and 125.797 MHz, respectively, using a 5 mm prodigy probe with the TopSpin 4.0.7 spectrometer software. 2D NMR spectra were recorded on an Agilent VNMRS 600 MHz NMR spectrometer using standard CHEMPACK 8.1 pulse sequences ( 1 H, 13 C gHSQCAD and 1 H, 13 C gHMBCAD) implemented in Varian VNMRJ 4.2 spectrometer software.

UHPLC-ESI-QqTOF-MS and MS/MS
Samples (2 µL) were loaded on an EC 150/2 Nucleoshell RP 18 column (C 18 -phase, ID 2 mm, length 150 mm, particle size 2.7 µm, Macherey Nagel, Düren, Germany) under isocratic conditions (5% eluent B, 2 min), and separated using a linear gradient from 5% to 95% eluent B in 17 min. Separation was performed on an ACQUITY UPLC I-Class UHPLC System (Waters GmbH, Eschborn, Germany) with a flow rate of 0.4 mL/min and 40 • C column temperature. Eluents A and B were 0.3 mmol/L aq. ammonium formate and acetonitrile, respectively. The column effluent was introduced on-line into a TripleTOF 6600 QqTOF mass spectrometer equipped with a DuoSpray ESI/APCI ion source, operating in negative ion SWATH (Sequential Windowed Acquisition of All Theoretical Fragment Ion Mass Spectra) mode and controlled by the Analyst TF 1. potentials were −35 and −10 V, respectively. The product ion spectra (tandem mass spectra, MS/MS) were acquired in the high sensitivity mode (accumulation time 20 ms) in the m/z range of 65-1250 using unit Q1 resolution with mass resolution above 30,000. Collision potential (CE) was set to −35 V, whereas collision energy spread (CES) was 15 V. The data were evaluated by Peak View 1.2.0.3 software (AB Sciex GmbH, Darmstadt, Germany).

RP-UHPLC-ESI-LIT-Orbitrap-MS
Samples (2 µL) were loaded on an ACQUITY UPLC reversed-phase BEH column (C 18phase, ID 1 mm, length 50 mm, particle size 1.7 µm, Waters GmbH, Eschborn, Germany) under isocratic conditions (95% A + 5% eluent B, 2 min), and separated using a linear gradient from 5% to 95% eluent B in 10 min using water and acetonitrile (both containing 0.1% v/v formic acid) as eluents A and B, respectively. The separations were performed with a Dionex Ultimate 3000 UHPLC System (Thermo Fisher Scientific, Bremen, Germany) at the flow rate of 400 µL/min and column temperature of 40 • C. The column effluents were transferred on-line into a hybrid LTQ-Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Bremen, Germany), equipped with a heated electrospray ionization (HESI) source at 300 • C and operated in the negative ion mode. The analysis was performed under ion spray (IS) voltage of 3.8 kV, with nebulizer and auxiliary gases set to 10 and 5 psig, respectively. The capillary temperature was set to 325 • C. The spectra were acquired at the mass to charge ratio (m/z) range of 120-2000 and resolution of 30,000. Tandem mass spectra were acquired with isolation width of 0.5-2 m/z and collision induced dissociation mode (30-35% normalized intensity), activation time 10 ms and activation frequency 250. Spectra evaluation was performed in Xcalibur 2.2 software (Thermo Fisher Scientific).

Extraction and Isolation
62.58 g roots of LR7 were exhaustively extracted with methanol to give 3.86 g of crude extract after evaporation of the solvent.

DNA Extraction, Polymerase Chain Reaction, and Sequencing
Genomic DNA was isolated from leaf samples using the Nucleo Spin Plant II Kit (Macherey-Nagel GmbH & Co. KG, Dueren, Germany), with minor modifications, using buffer PL2 and adding 20 µL RNAse, 30 µL mercaptoethanol and PVP (2%). The yield of DNA extraction was measured using a Qubit ® 3.0 Fluorometer (Thermo Fischer Scientific, Waltham, MA, USA) and DNA bands were visualized with SYBR ® Safe DNA gel stain (Thermo Fisher Scientific) on 1% agarose gels (1× TAE buffer solution) using a GenoPlex VWR ® gel documentation system with GenoCapture version 7.12.07.0 (Synoptics Ltd., Cambridge, UK). We performed polymerase chain reaction (PCR) of the ITS region, which comprises the ITS1 spacer, 5.8S rRNA gene, and ITS2 spacer using primers 17SE_m (5'-CGGTGAAGTGTTCGGATCG-3') and 26SE_m (5'-CGCTCGCCGTTACTAGGG-3') [45], with reaction volumes of 25 µL, including 2 µL of genomic DNA, 0.3 µL of each primer, 0.5 µL BSA, 1µL DMSO and 20.9 µL, and 1 × Dream TaqGreen Master Mix, on a Labcycler Gradient PCR machine (SensoQuest GmbH, Germany). The initial denaturation step was set at 95 • C for 2 min, followed by 35 cycles of denaturation at 95 • C for 1 min, primer annealing at 53 • C for 1 min, extension at 72 • C for 2 min, followed by final extension at 72 • C for 10 min. PCR products were purified using a Nucleo Spin Gel and PCR clean-up kit (Macherey-Nagel GmbH & Co. KG, Dueren, Germany). The purified DNA samples were then measured using an Eppendorf Biophotometer to adjust the DNA concentration before Sanger sequencing at LGC Genomics GmbH, Berlin, Germany, using the same primers as mentioned above.

Phylogenetic Analyses
All sequences were aligned using the MUSCLE algorithm as implemented in Geneious 6.1.8 [46] and corrected by hand. Phylogenetic analyses were done using Maximum Likelihood (ML) with RAxML [47,48], and Bayesian inference with MrBayes v3.2.7 [49]. We used the GTR + Γ substitution model in all analyses to ensure comparability between the Maximum Likelihood and Bayesian analyses. In RaxML, we inferred the maximum likelihood tree with 100 non-parametric bootstrap replicates. In MrBayes, we ran the inference for 5 million generations (sampling every 5000 generations) with four runs and four chains each. The appropriateness of sampling (all Effective Sample Sizes (ESS) >200) was checked in Tracer v1.7.1 [50], before building a majority-rule consensus tree (with 2 million generations excluded as burn-in). In both analyses, Laguncularia racemose (L.) C.F.Gaertn. was set as outgroup.

Anti-infective Bioassays
Crude extracts (50 and 500 µg/mL) were tested in triplicate for antibacterial activity against the Gram-negative Aliivibrio fischeri and the Gram-positive Bacillus subtilis following the procedure described by dos Santos et al. [51]. Chloramphenicol (100 µM) was used as positive control and induced the complete inhibition of bacterial growth. The results are presented as relative values (% inhibition) in comparison to the negative control (bacterial growth in medium containing 1% DMSO without test compound = 0% inhibition). Negative values indicate an increase of bacterial growth, which is common with testing extracts containing further nutrients by nature.
The anthelmintic bioassay for all extracts (500 µg/mL) was performed in triplicate according to the method developed by Thomson and coworkers [52] using the model organism Caenorhabditis elegans (Bristol N2 wild-type strain) that previously was shown to correlate with anthelmintic activity against parasitic trematodes. The solvent DMSO (2%) and the standard anthelmintic drug ivermectin (10 µg/mL, 100% dead worms after 30 min incubation) were used as negative and positive controls, respectively. The number of dead and living nematodes in each sample was counted using the microscope Olympus CKX41. Results are given as percentage of dead worms.
The cytotoxic activity of selected samples (LL3, LL11, LR5, LR15) was evaluated at the concentrations of 0.05 and 50 mg/mL against the human prostate cancer cell line PC3 and the colon adenocarcinoma cell line HT-29 by determining cell viability in MTT and CV assays as described previously [51]. Digitonin (125 g/mL) was used as positive control. The results are given as percentage of control values without treatment (=100%).

Phytochemcial Analyses
The metabolite profiles of roots from 12 accessions of Lumnitzera littorea and 19 accessions of L. racemosa collected across Indonesia were investigated by TLC ( Figure 1) and liquid chromatography, coupled on-line to mass spectrometry (LC-MS) or tandem mass spectrometry (LC-MS/MS). For structure verification, selected compounds were isolated and investigated by nuclear magnetic resonance spectroscopy (NMR).

Phytochemcial Analyses
The metabolite profiles of roots from 12 accessions of Lumnitzera littorea and 19 accessions of L. racemosa collected across Indonesia were investigated by TLC ( Figure 1) and liquid chromatography, coupled on-line to mass spectrometry (LC-MS) or tandem mass spectrometry (LC-MS/MS). For structure verification, selected compounds were isolated and investigated by nuclear magnetic resonance spectroscopy (NMR).  Figure 2). The majority of the analytes could be detected in both species, while compound 17 could only be found in L. littorea, and constituents 7 and 10 are predominantly occurring in L. racemosa (Figure 2). These com-  Figure 2). The majority of the analytes could be detected in both species, while compound 17 could only be found in L. littorea, and constituents 7 and 10 are predominantly occurring in L. racemosa (Figure 2). These compounds were successfully cross-annotated in the root extracts obtained from other accessions by ultra-high-performance liquid chromatography -quadrupole mass spectrometry (RP-UHPLC-Q-MS) with detection in UV and visible (UV-VIS) spectra (Table S1, Figure S2). Table 2. Metabolites annotated in roots of Lumnitzera littorea and L. racemosa by reversed phase ultra-high-performance chromatography-tandem mass spectrometry (RP-UHPLC-MS/MS). The annotated metabolites are numbered according to peak numbers in Figure 2. Individual tandem mass spectra are shown in Figure S1.    The major metabolites detected in samples from both species were ellagic acid derivatives (Figures 1-3, Table 2), mostly O-methylated at different positions. Commonly, ellagic acid derivatives, including ellagitannins, are widespread in Combretaceae: Ellagic acid and several methylated derivatives were previously detected in leaves and twigs of L. racemosa [26,30] and in related species such as, e.g., in Terminalia species which are widely used in traditional medicine [53], in Pteleopsis hylodendron Mildbr. [54] and in Combretum alfredii Hance [55]. In our samples, the broad bands in TLCs along with multiple signals in specific extracted ion chromatograms (XICs) acquired in RP-UHPLC-QqTOF-MS experiments, indicate the presence of several positional isomers, i.e., compounds characterized by the same molecular weight, but featured with different substitution patterns. Such O-methylated species can be assigned by characteristic losses of 15 u, accompanying formation of an odd [M-H-15] − ion [53]. The corresponding signals are characteristic for methyl substitution both in cyclic [56] and aromatic (often in combination with a carbonyl loss) systems [57]. High intensities of such signals in tandem mass spectra of phenolic compounds typically indicate methylation as part of a methoxy group (OCH 3 ) [58]. 551.1027) share common fragments with ellagic acid (6) and trimethylellagic acid (16), respectively, suggesting a structural relationship. Indeed, the mass difference of 208 u between 10 and 16 is, most likely, due to moieties localized outside the aromatic core. Its elemental composition (C7H12O7) might indicate a probable presence of a sugar acid (hexahydroxyheptanoic acid) or a substituted sugar in the structure (Supplementary Figure  S1-11). The most prominent compound 15 present in both species was isolated and identified as 3,3',4'-tri-O-methylellagic acid 4-sulfate by 1D and 2D NMR measurements (Table S5) and comparison to published data [64]. Compound 16 (Table S6) was verified as the corresponding 3,3',4-tri-O-methyl ellagic acid without sulfation [65]. During separation, the sulfation of this compound is reflected in a clear enhancement of polarity visible by lower Rf value on normal phase TLC ( Figure 1) and reduced retention time on reversed phase column ( Figure 2). 1D and 2D NMR allowed the elucidation of compound 13 as 3,4-di-Omethyl ellagic acid (Table S4). The substitution pattern was established by HMBC correlations and a ROESY correlation between 4-OMe and H-5. According to its chromatographic behavior compound 9 can be assigned as the derived sulfate. Compound 6 was verified as ellagic acid by 1 H NMR and comparison of MS data to a reference standard. Furthermore, we could obtain several sulfated phenylbutane derivatives including 4-(4hydroxyphenyl)-2-butanol 2-sulfate (2 , Table S2) previously isolated from roots of Rheum  Figure S1). This neutral loss is well-known for sulfated phenolic compounds [59] which not only occur naturally in plants (reviewed by Correira de Silva and co-workers [60]), and marine organism [61,62] but are also quite typical as an important group of polyphenol metabolites dedicated for kidney-clearance from human plasma [63]. In addition to the characteristic loss of 80 u, the tandem mass spectra of sulfated compounds feature diagnostic signals in the low m/z range, which can be attributed to SO 3 − (m/z 79.9567) and HSO 4 − (m/z 96.9594) ( Table 2 and Supplementary Figure S1). However, to the best of our knowledge, sulfated ellagic acid derivatives were not reported in mangrove species before.
Remarkably, further sulfur containing compounds could be detected (compounds 1-5, 8, 11, 18-21, Table 1, Figure S1). In compounds 20 and 21 the sulfur appears to be integrated in another form than as sulfate, based on the molecular formula and the MS/MS spectra. However, the exact structures of these compounds could not be assigned based on the acquired SWATH tandem mass spectra. Compounds 8 (m/z 487.0179) and 10 (m/z 551.1027) share common fragments with ellagic acid (6) and trimethylellagic acid (16), respectively, suggesting a structural relationship. Indeed, the mass difference of 208 u between 10 and 16 is, most likely, due to moieties localized outside the aromatic core. Its elemental composition (C 7 H 12 O 7 ) might indicate a probable presence of a sugar acid (hexahydroxyheptanoic acid) or a substituted sugar in the structure (Supplementary Figure S1-11).
The most prominent compound 15 present in both species was isolated and identified as 3,3',4'-tri-O-methylellagic acid 4-sulfate by 1D and 2D NMR measurements (Table S5) and comparison to published data [64]. Compound 16 (Table S6) was verified as the corresponding 3,3',4-tri-O-methylellagic acid without sulfation [65]. During separation, the sulfation of this compound is reflected in a clear enhancement of polarity visible by lower Rf value on normal phase TLC ( Figure 1) and reduced retention time on reversed phase column (Figure 2). 1D and 2D NMR allowed the elucidation of compound 13 as 3,4-di-O-methylellagic acid (Table S4). The substitution pattern was established by HMBC correlations and a ROESY correlation between 4-OMe and H-5. According to its chromatographic behavior compound 9 can be assigned as the derived sulfate. Compound 6 was verified as ellagic acid by 1 H NMR and comparison of MS data to a reference standard. Furthermore, we could obtain several sulfated phenylbutane derivatives including 4-(4-hydroxyphenyl)-2-butanol 2-sulfate (2 , Table S2) previously isolated from roots of Rheum maximowiczii [66]. The downfield shift of H-2 and C-2 compared to the aglycon indicates sulfation at this position. However, due to the low amount of isolated compound we could not determine the configuration at C-2. Compound 3 was identified as the previously undescribed 4-(4-sulfoxy-3-methoxy phenyl)-butan-2-one (zingerone sulfate). NMR data (Table S3) are in good agreement with data published for the basic skeleton 4-(4-hydroxy-3-methoxyphenyl)-butan-2-one [67]. Compound 4 represents according to the MS fragmentation pattern most likely the corresponding zingerol sulfate.
Sulfated phenolics are rare natural products in plants and were not described before for Combretaceae. There are only a few reports about the occurrence of sulfated ellagic acid derivatives in plants, e.g., 3-O-methylellagic acid 4-sulfate and 3,3 -di-O-methylellagic acid 4-sulfate were detected previously in Jaboticaba wood from Myrciaria cauliflora (Mart.) O. Berg (Myrtaceae) [68]. The first compound was found to exhibit antioxidant and antiinflammatory activities in cigarette smoke extract-exposed small airway epithelial cells and may be useful for the treatment of COPD [68]. The major metabolite 3,3',4'-tri-Omethylellagic acid 4-sulfate (15) found in this study in most of the Lumnitzera samples was reported to occur in rhizomes of Geum rivale L, [64] and roots of Potentilla candidans Humb. & Bonpl. Ex Nestl. [69], both members of the Rosaceae family.
So far, the biological function of sulfated secondary phenolics in plants is unknown. The presence of a sulfate moiety enhances the water solubility of compounds and increases ion strength in the containing tissue. The accumulation of sulfates may be organ-specific. In case of the investigated Lumnitzera species, they are found in the roots, which normally are in contact with sea water, while they were not reported from studies on leaves and twigs [26,[29][30][31]. In this study, both species showed similar patterns of sulfated metabolites despite L. littorea is predominantly occurring at well-drained sites with less salinity, and typically growing as a tree, while L. racemosa is more resistant to saline conditions and occurs at the margin of bare salt pans [70] often growing as a shrub. It was postulated before that the occurrence of sulfated flavonoids in plants is an ecological trait rather than a systematic feature [64,71] and that these compounds might play a role in adaptation to water-stress in salty soils. Furthermore, natural products with a sulfated scaffold have emerged as antifouling agents with low or nontoxic effects to the environment [72] Thus, a similar function can be assumed for the detected sulfated compounds in Lumnitzera roots.
In general, marine organisms contain a significant number of phenolic metabolites which occur in sulfated form [61,62,73]. Given the high concentration of sulfate in sea water, mangrove soils contain high levels of sulfate and thus, sulfate is easily available for plants. Sulfate-reducing bacteria influence iron, phosphorus and sulfur availability in anoxic mangrove sediments and mangrove species zonation across the intertidal zone [74]. Anaerobic sulfate reducing microbial communities are involved in sulfur cycling in these soils and in the decomposition of mangrove-derived soil organic matter [8,75].

Phylogenetic Analyses
The variation of the metabolite pattern across different populations could be supported by molecular phylogenetic evidence. As Bayesian and ML analyses yielded topologically identical trees, we here only show the results from the Bayesian analysis (with support values from both; Figure 4). Both species form well-supported monophyletic groups. Notably, within L. racemosa, samples from populations LR1, LR2, LR3, LR4 and LR8 form a well-supported clade (Clade 2, Figure 4), which is characterized by sharing the same TLC and mass spectrometry metabolic profile, thus suggesting a different chemotype. These samples completely lacked sulfated and nonsulfated trimethylellagic acids, but were dominated by dimethylellagic acid and its sulfate. In contrast, the phylogenetic tree does not show a geographical pattern, at least not on the level where we have appropriate resolution. In order to confirm these results, we sequenced a second individual from each of these specific populations, except for LR2 (for which only one plant had been found at the location). On the one hand, the infraspecific, interpopulational variation in metabolic profiles calls for caution when selecting cultivated mangrove plants for the purpose of metabolic profiling for medicinal purposes (compare to [76]). On the other hand, the nuclear ITS region seems to constitute a useful guide for selecting individuals for cultivation, at least for Lumnitzera and some other mangrove plant species (this study, and [76]).

Evaluation of Anti-infective Properties
To evaluate the anti-infective potential of the Lumnitzera root extracts, the antibacterial and anthelmintic activities were determined using nonpathogenic model organism as test systems. The samples from different locations varied significantly in their antibacterial activities ( Figure 5). Both species form well-supported monophyletic groups. Notably, within L. racemosa, samples from populations LR1, LR2, LR3, LR4 and LR8 form a well-supported clade (Clade 2, Figure 4), which is characterized by sharing the same TLC and mass spectrometry metabolic profile, thus suggesting a different chemotype. These samples completely lacked sulfated and nonsulfated trimethylellagic acids, but were dominated by dimethylellagic acid and its sulfate. In contrast, the phylogenetic tree does not show a geographical pattern, at least not on the level where we have appropriate resolution. In order to confirm these results, we sequenced a second individual from each of these specific populations, except for LR2 (for which only one plant had been found at the location). On the one hand, the infraspecific, interpopulational variation in metabolic profiles calls for caution when selecting cultivated mangrove plants for the purpose of metabolic profiling for medicinal purposes (compare to [76]). On the other hand, the nuclear ITS region seems to constitute a useful guide for selecting individuals for cultivation, at least for Lumnitzera and some other mangrove plant species (this study, and [76]).

Evaluation of Anti-Infective Properties
To evaluate the anti-infective potential of the Lumnitzera root extracts, the antibacterial and anthelmintic activities were determined using nonpathogenic model organism as test systems. The samples from different locations varied significantly in their antibacterial activities ( Figure 5).  Only two extracts (LL8 and LR7) completely inhibited the growth of the Gram-positive Bacillus subtilis at 500 µg/mL after 16 h ( Figure 5A). The activity against Gram-negative bacteria showed a different pattern ( Figure 5B). Here, samples LL3, LL12 and LR19 completely inhibited the growth of the test organism Allivibrio fischeri at a concentration of 500 µg/mL after 24 h, LL8 and LR9 induced 71% and 81% inhibition, respectively. Especially the extract LR19 exhibited a noteworthy antibacterial potential against the Gramnegative test organism since it induced 100% growth inhibition already at the lower test concentration of 50 µg/mL. At this point, the assignment of the antibacterial properties to a certain (group of) constituents is not possible. The selective antibacterial activity of different accessions of only a few samples of the same plant species suggests, however, that the effects might be connected to associated microorganisms rather than intrinsic plantproduced metabolites.
None of the Lumnitzera crude extracts showed anthelmintic activity against the nematode Caenorhabditis elegans ( Figure S3). This is in contrast to studies on mangrove plant species from other families. For example, anthelmintic activity was found for leaf and stem extracts of Acanthus ilicifolius L. (Acanthaceae) [77,78].
Two samples from each species (LL3, LL11, LR5, LR15) were exemplarily tested for their cytotoxic activity against the human prostate cancer cell line PC3 and the colon adenocarcinoma cell line HT-29. The investigated samples did not influence the viability of Only two extracts (LL8 and LR7) completely inhibited the growth of the Gram-positive Bacillus subtilis at 500 µg/mL after 16 h ( Figure 5A). The activity against Gram-negative bacteria showed a different pattern ( Figure 5B). Here, samples LL3, LL12 and LR19 completely inhibited the growth of the test organism Allivibrio fischeri at a concentration of 500 µg/mL after 24 h, LL8 and LR9 induced 71% and 81% inhibition, respectively. Especially the extract LR19 exhibited a noteworthy antibacterial potential against the Gram-negative test organism since it induced 100% growth inhibition already at the lower test concentration of 50 µg/mL. At this point, the assignment of the antibacterial properties to a certain (group of) constituents is not possible. The selective antibacterial activity of different accessions of only a few samples of the same plant species suggests, however, that the effects might be connected to associated microorganisms rather than intrinsic plant-produced metabolites.
None of the Lumnitzera crude extracts showed anthelmintic activity against the nematode Caenorhabditis elegans ( Figure S3). This is in contrast to studies on mangrove plant species from other families. For example, anthelmintic activity was found for leaf and stem extracts of Acanthus ilicifolius L. (Acanthaceae) [77,78].
Two samples from each species (LL3, LL11, LR5, LR15) were exemplarily tested for their cytotoxic activity against the human prostate cancer cell line PC3 and the colon adenocarcinoma cell line HT-29. The investigated samples did not influence the viability of the cancer cell lines at a concentration of 0.05 µg/mL, however, completely inhibited the cell growth at a concentration of 50 µg/mL indicating a moderate cytotoxic potential ( Figure S4).
Ellagic acid is known to possess a wide range of biological activities based on its antioxidant and chemopreventive potential including antimicrobial, anti-inflammatory, neuroprotective, antihepatotoxic, anticholestatic, antifibrogenic, anticarcinogenic, cytotoxic, and antiviral effects [79,80]. Several of these multiple activities may, however, be related to general tanning properties of polyphenolics, rather than being specific effects [81,82]. Furthermore, ellagic acid and related compounds are potent Aldose reductase inhibitors that could play an important role in the management of diabetic complications [69]. Methylation often reduces these effects whereas the introduction of a sulfate group increases the inhibitory activity [69], but reduces membran permeability. In contrast, the only moderate cytotoxicity of ellagic acid against cancer cell lines is further decreased by the presence of a sulfate moiety [64]. The potassium salts of 3,3'-dimethylellagic acid 4-sulfate and 3,3',4'-trimethylellagic acid 4-sulfate were also found to exhibit moderate antimicrobial potential against Gram-positive Bacillus subtilis and Staphylococcus aureus with MIC values in the range of 22.5-50.8 µg/mL [83]. These compounds were, however, not effective against Gram-negative Escherichia coli. [83].
In our investigations the antibacterial activity is likely not directly connected to the sulfated ellagic acid derivatives or the other sulfated metabolites. These metabolites occur across all samples, but the antibacterial effects are limited to samples from particular locations ( Figure 5). Nevertheless, Lumnitzera roots are a promising source for pharmacologically interesting sulfated ellagic acid derivatives and further sulfated plant metabolites.

Conclusions
In our study, a series of unusual sulfated constituents was characterized in root samples from the mangrove species Lumnitzera littorea and L. racemosa (Combretaceae). Thus, most of the methylated ellagic acid derivatives isolated from both species possess a sulfate moiety. However, L. racemosa samples from North Sumatra (LR1), Aceh (LR2, LR3), East Kalimantan (LR4), and Maluku (LR8), completely lack sulfated and non-sulfated trimethylellagic acid, but instead are dominated by dimethylellagic acid and its sulfate. This phytochemical pattern is corroborated by phylogenetic data, where these specific samples form a well-supported clade in the ITS tree. Interestingly, the occurrence of antimicrobial activity and sulfated ellagic acid derivatives are not connected. Although the ellagic acid derivatives are present within all samples, the antibacterial effects are limited to samples from particular locations in Indonesia, suggesting that other compounds from the plant or root-associated microorganisms might be responsible. The moderate cytotoxic effect, in contrast, can be attributed to the occurrence of the ellagic acid derivatives. In summary, Lumnitzera roots represent a potentially promising source for sulfated ellagic acid derivatives and further sulfur containing plant metabolites.