Ecological and Pharmacological Activities of Polybrominated Diphenyl Ethers (PBDEs) from the Indonesian Marine Sponge Lamellodysidea herbacea

Two known Polybrominated Diphenyl Ethers (PBDEs), 3,4,5-tribromo-2-(2′,4′-dibromophenoxy)phenol (1d) and 3,4,5,6-tetrabromo-2-(2′,4′-dibromophenoxy)phenol (2b), were isolated from the Indonesian marine sponge Lamellodysidea herbacea. The structure was confirmed using 13C chemical shift average deviation and was compared to the predicted structures and recorded chemical shifts in previous studies. We found a wide range of bioactivities from the organic crude extract, such as (1) a strong deterrence against the generalist pufferfish Canthigaster solandri, (2) potent inhibition against environmental and human pathogenic bacterial and fungal strains, and (3) the inhibition of the Hepatitis C Virus (HCV). The addition of a bromine atom into the A-ring of compound 2b resulted in higher fish feeding deterrence compared to compound 1d. On the contrary, compound 2b showed only more potent inhibition against the Gram-negative bacteria Rhodotorula glutinis (MIC 2.1 μg/mL), while compound 1d showed more powerful inhibition against the other human pathogenic bacteria and fungi. The first report of a chemical defense by compounds 1d and 2b against fish feeding and environmental relevant bacteria, especially pathogenic bacteria, might be one reason for the widespread occurrence of the shallow water sponge Lamellodysidea herbacea in Indonesia and the Indo-Pacific.


Introduction
Sponges are known to harbour a huge variety of secondary metabolites, which could potentially be important leads in drug development [1]. Reported compound classes include but are not limited to peptides [2][3][4], pyridines [5], diterpenoids [6], alkaloids [7], We further isolated the major bioactive compounds by reverse-phase SPE fractionation followed by C18 reverse phased high-performance liquid chromatography (HPLC). HPLC purification of the active fractions yielded compounds 1d and 2b at levels of 2.9 mg and 6.8 mg, respectively. Pure compounds were injected at different concentrations to create a calibration curve, which was then used to determine the natural concentrations of the compounds in the sponge tissue. The analysis determined compounds 1d and 2b with concentrations of 1.6 mg and 2.4 mg, respectively, per mL sponge tissue. When tested at these respective concentrations, both PBDEs showed a significant deterrence at natural volumetric concentrations against the tested pufferfish. Compound 2b displayed a high deterrence, with 61.1% of the pellets being rejected (p = 0.008, Fisher's exact test), while compound 1d revealed a somewhat lower but still significant deterrence, with 44.4% of the pellets being rejected (p = 0.041, Fisher's exact test). The observed deterrence of the pure compounds was somewhat lower compared to the tested fractions or crude extract, except for the EtOAc and the water fraction, which were not significantly deterred (p = 0.32, Fisher's exact test).

Antimicrobial Activity against Pathogenic and Non-Pathogenic Environmental Bacteria
An agar diffusion assay was used to determine possible antimicrobial activity against environmental bacteria, using 14 different pathogenic and non-pathogenic environmental bacteria, which had previously been isolated from a reef flat habitat ( Figure 2) [35,36]. The extracts exhibited pronounced antimicrobial activities by inhibiting more than half of the test bacteria, of which four of them were putatively pathogenic. According to the Gram type of the bacteria, the crude extract and liquid-liquid separated fractions inhibited all of the Gram-positive bacteria except Streptomyces sp. (1656), while half of the Gram-negative bacteria were inhibited. Interestingly, the Agar diffusion assay (ADA) showed the same pattern as the one we found in the feeding assay. That is, the crude extract showed the highest activity, followed by its most lipophilic fractions, HEX and DCM. No inhibition We further isolated the major bioactive compounds by reverse-phase SPE fractionation followed by C 18 reverse phased high-performance liquid chromatography (HPLC). HPLC purification of the active fractions yielded compounds 1d and 2b at levels of 2.9 mg and 6.8 mg, respectively. Pure compounds were injected at different concentrations to create a calibration curve, which was then used to determine the natural concentrations of the compounds in the sponge tissue. The analysis determined compounds 1d and 2b with concentrations of 1.6 mg and 2.4 mg, respectively, per mL sponge tissue. When tested at these respective concentrations, both PBDEs showed a significant deterrence at natural volumetric concentrations against the tested pufferfish. Compound 2b displayed a high deterrence, with 61.1% of the pellets being rejected (p = 0.008, Fisher's exact test), while compound 1d revealed a somewhat lower but still significant deterrence, with 44.4% of the pellets being rejected (p = 0.041, Fisher's exact test). The observed deterrence of the pure compounds was somewhat lower compared to the tested fractions or crude extract, except for the EtOAc and the water fraction, which were not significantly deterred (p = 0.32, Fisher's exact test).

Antimicrobial Activity against Pathogenic and Non-Pathogenic Environmental Bacteria
An agar diffusion assay was used to determine possible antimicrobial activity against environmental bacteria, using 14 different pathogenic and non-pathogenic environmental bacteria, which had previously been isolated from a reef flat habitat ( Figure 2) [35,36]. The extracts exhibited pronounced antimicrobial activities by inhibiting more than half of the test bacteria, of which four of them were putatively pathogenic. According to the Gram type of the bacteria, the crude extract and liquid-liquid separated fractions inhibited all of the Gram-positive bacteria except Streptomyces sp. (1656), while half of the Gram-negative bacteria were inhibited. Interestingly, the Agar diffusion assay (ADA) showed the same pattern as the one we found in the feeding assay. That is, the crude extract showed the highest activity, followed by its most lipophilic fractions, HEX and DCM. No inhibition zone appeared with the H 2 O fractions, while only weak inhibition occurred with the EtOAc fraction. All of the tested extracts and fractions showed the highest inhibition activity against WHV1 (Aurantimonas coralicida). Due to the lack of purified PBDEs, no further ADA experiments were performed. zone appeared with the H2O fractions, while only weak inhibition occurred with the EtOAc fraction. All of the tested extracts and fractions showed the highest inhibition activity against WHV1 (Aurantimonas coralicida). Due to the lack of purified PBDEs, no further ADA experiments were performed.

Antimicrobial Assay against Pathogenic Clinical Microbes
The crude extract and fractions also displayed remarkable bioactivities in the antimicrobial assay against clinically relevant, pathogenic microbial, and fungal strains (cf. Table  1). The crude extract strongly inhibited Bacillus subtilis, Staphylococcus aureus, Mucor hiemalis, and Rhodotorula glutinis, whereas its activities against Escherichia coli and Mycobacterium smegmatis were weak. No activities were detected against Pseudomonas aeruginosa and Candida albicans. Pronounced activities were observed against all of the tested microbes with the HEX and DCM fractions. The EtOAc fraction only showed the potent inhibition of B. subtilis and moderate inhibition against S. aureus, R. glutinis, and M. hiemalis. Screening of the two purified compounds, 1d and 2b, also exhibited a broad range of antimicrobial activities against the seven test strains. The compounds were highly active against all tested strains except E. coli (MIC > 66.7 μg/mL). Both compounds strongly inhibited B. subtilis (MIC 0.5 μg/mL), displaying a five times lower MIC than the positive control tetracycline (2.8 μg/mL). Furthermore, compound 1d exhibited bioactivity against the Grampositive bacteria S. aureus that was as potent as the positive control gentamicin (MIC 0.5 μg/mL), whereas compound 2b showed four times less bioactivity. Both compounds 1d and 2b were significantly more active (MIC 4.2 μg/mL and 2.1 μg/mL, respectively) than the positive control nystatin against the aerobic yeast cell R. glutinis.

Antimicrobial Assay against Pathogenic Clinical Microbes
The crude extract and fractions also displayed remarkable bioactivities in the antimicrobial assay against clinically relevant, pathogenic microbial, and fungal strains (cf. Table 1). The crude extract strongly inhibited Bacillus subtilis, Staphylococcus aureus, Mucor hiemalis, and Rhodotorula glutinis, whereas its activities against Escherichia coli and Mycobacterium smegmatis were weak. No activities were detected against Pseudomonas aeruginosa and Candida albicans. Pronounced activities were observed against all of the tested microbes with the HEX and DCM fractions. The EtOAc fraction only showed the potent inhibition of B. subtilis and moderate inhibition against S. aureus, R. glutinis, and M. hiemalis. Screening of the two purified compounds, 1d and 2b, also exhibited a broad range of antimicrobial activities against the seven test strains. The compounds were highly active against all tested strains except E. coli (MIC > 66.7 µg/mL). Both compounds strongly inhibited B. subtilis (MIC 0.5 µg/mL), displaying a five times lower MIC than the positive control tetracycline (2.8 µg/mL). Furthermore, compound 1d exhibited bioactivity against the Gram-positive bacteria S. aureus that was as potent as the positive control gentamicin (MIC 0.5 µg/mL), whereas compound 2b showed four times less bioactivity. Both compounds 1d and 2b were significantly more active (MIC 4.2 µg/mL and 2.1 µg/mL, respectively) than the positive control nystatin against the aerobic yeast cell R. glutinis.

Inhibitory Effects on HCV Infectivity
Next, we investigated the inhibitory effect of the L. herbacea crude extract against HCV infectivity in human liver cells. As depicted in Figure 3, only moderate inhibition of 50% was observed in contrast to the EGCG positive control. Therefore, we did not investigate its liquid-liquid fractions further. Interestingly, we found that the organic extract was not toxic in human liver cells, which was in contrast to previous studies that had reported cytotoxic effects by OH-PBDE compounds [30,37,38].

Inhibitory Effects on HCV Infectivity
Next, we investigated the inhibitory effect of the L. herbacea crude extract against HCV infectivity in human liver cells. As depicted in Figure 3, only moderate inhibition of 50% was observed in contrast to the EGCG positive control. Therefore, we did not investigate its liquid-liquid fractions further. Interestingly, we found that the organic extract was not toxic in human liver cells, which was in contrast to previous studies that had reported cytotoxic effects by OH-PBDE compounds [30,37,38].

Structure Elucidation of Compounds 1d and 2b
As the first step in structure elucidation, the exact masses of the purified compounds were obtained with a UPLC-HRMS (Waters Synapt G2 . Both compounds were compared to compounds in the MarinLit database [39]. The results showed that both sum formulas were related to known PBDEs (Table 2). To locate the positions of the bromine atoms in the ring system of compounds 1d and 2b, we conducted 1 H and 13 C NMR measurements. The results revealed that compound 1d possessed three bromine atoms in the A-ring and two bromine atoms in the B-ring ( Figure 4). In addition, an extra bromine atom in the A-ring was determined for compound 2b. including the standard deviation (SD). MeOH was used as a negative control (NC), and epigallocatechin gallate (EGCG) was used as a positive control.

Structure Elucidation of Compounds 1d and 2b
As the first step in structure elucidation, the exact masses of the purified compounds were obtained with a UPLC-HRMS (Waters Synapt G2 . Both compounds were compared to compounds in the MarinLit database [39]. The results showed that both sum formulas were related to known PBDEs ( Table 2). To locate the positions of the bromine atoms in the ring system of compounds 1d and 2b, we conducted 1 H and 13 C NMR measurements. The results revealed that compound 1d possessed three bromine atoms in the A-ring and two bromine atoms in the B-ring ( Figure 4). In addition, an extra bromine atom in the A-ring was determined for compound 2b. Figure 4. Four constitutional isomers of PBDEs are given to evaluate their 13 C NMR chemical shift deviation (1a)-(1d) and to compare them to compound 1d from our work. Penta-BDE, 3,4,5,6-tetrabromo-2-(2'-bromophenoxy)phenol (2a) that was first reported by Salva and Faulkner [42] is shown as a comparison to compound 2b in our study. to compare them to compound 1d from our work. Penta-BDE, 3,4,5,6-tetrabromo-2-(2 -bromophenoxy)phenol (2a) that was first reported by Salva and Faulkner [42] is shown as a comparison to compound 2b in our study.
The NMR data clearly indicated that the B-ring in both compounds 1d and 2b was identical. Since the A-ring is hexa-substituted in 2b, there are no structural alternatives, whereas for 1d, four constitutional isomers were possible (1a to 1d). In order to elucidate the correct regioisomers, the 1 H and 13 C NMR data were compared to existing NMR data in the literature (Tables 3 and 4). Furthermore, the structures of 1d and 2b were also verified through comparison with predicted 13 C NMR chemical shifts [43][44][45][46] using three different computational approaches: NMRPredict [47], NMRShiftDB [48], and CSEARCH [49].   The 13 C chemical shifts of compound 1d (this work) were compared to predicted values and literature data (the atom numbering of Bowden et al., 2000 [50] was used for all literature data). The columns with the calculated 13 C chemical shift values (columns 3 to 6) using NMRPredict are indicated by an asterisk (*). The carbon atoms C-3 to C-6 were not explicitly assigned for the prediction of the constitutional isomers 1a to 1c due to very similar 13  The averaged deviation of δ( 13 C) for all 12 carbon atoms to the calculated and existing experimental data sets in comparison to our experimental values is given in Figure 5. The results showed that the mean value of 1d was close to the experimental shifts reported for 1d in the literature [49] (Figure 5, bars 6 to 10). Comparison to the calculated values obtained with NMRPredict also preferred constitutional proposals 1d (<1 ppm) over the regioisomers 1a to 1c (>2 ppm). Therefore, 1d was confirmed as 3,4,5-tribromo-2-(2 ,4dibromophenoxy)phenol (Figure 4). In fact, our δ( 13 C) values for compound 1d were 100% identical to the one's published by Suyama et al. [40] (Table 4, not shown in Figure 5). The averaged deviation of δ( 13 C) for all 12 carbon atoms to the calculated and existing experimental data sets in comparison to our experimental values is given in Figure 5. The results showed that the mean value of 1d was close to the experimental shifts reported for 1d in the literature [49] (Figure 5, bars 6 to 10). Comparison to the calculated values obtained with NMRPredict also preferred constitutional proposals 1d (<1 ppm) over the regioisomers 1a to 1c (>2 ppm). Therefore, 1d was confirmed as 3,4,5-tribromo-2-(2′,4′-dibromophenoxy)phenol (Figure 4). In fact, our δ( 13 C) values for compound 1d were 100% identical to the one's published by Suyama et al. [40] (Table 4, not shown in Figure 5). Compound 2b showed very similar 13 C chemical shifts compared to the only reported NMR data from Utkina et al. [41] (Tables 5 and 6). Since we do not have constitutional isomers for the A-ring in compound 2b, we decided to use another compound with the same A-ring in order to determine how well the 13 C chemical shifts can be predicted for that ring. A similar compound, 3,4,5,6-tetrabromo-2-(2'-bromophenoxy)phenol (2a), which was first identified by Salva and Faulkner [50], with one less bromine in the B-ring compared to compound 2b was used for comparison ( Figure 4). The 13 C chemical shifts of the A-ring (C-1 to C-6) of both compounds showed high similarity (Table 6). Therefore, compound 2b was confirmed as 3,4,5,6-tetrabromo-2-(2′,4′-dibromophenoxy)phenol (Figure 4). Compound 2b showed very similar 13 C chemical shifts compared to the only reported NMR data from Utkina et al. [41] (Tables 5 and 6). Since we do not have constitutional isomers for the A-ring in compound 2b, we decided to use another compound with the same A-ring in order to determine how well the 13 C chemical shifts can be predicted for that ring. A similar compound, 3,4,5,6-tetrabromo-2-(2 -bromophenoxy)phenol (2a), which was first identified by Salva and Faulkner [50], with one less bromine in the B-ring compared to compound 2b was used for comparison ( Figure 4). The 13 C chemical shifts of the A-ring (C-1 to C-6) of both compounds showed high similarity (Table 6). Therefore, compound 2b was confirmed as 3,4,5,6-tetrabromo-2-(2 ,4 -dibromophenoxy)phenol ( Figure 4).  [40]. However, chemical shift comparison could not be performed due to the lack of 1 H and 13 C NMR data for the related compounds. Therefore, to achieve a close comparison, we used compound 2a from Salva and Faulkner [42], which is similar to compound 2b with an identical A-ring.
Compound 1d was first characterized and described in 1969 from Dysidea herbacea from Palau, Micronesia, using only proton NMR spectral data. In the following years, the structure was verified by partial synthesis and further proton NMR analysis [20,51]. Both compounds were also the major metabolites in a different specimen of Dysidea [20,21]. Ten years later, the 13 C NMR data of compound 1d were reported for the first time along with new bioactivities, such as the inhibition of inosine monophosphate dehydrogenase (IM-PDH), guanosine monophosphate synthetase, and 15-lipoxygenase [21]. Later, compound 1d was detected together with novel PBDEs [50,52,53]. In 2008, Ortlepp and colleagues also reported the antifouling activity of compound 1d against several fouling organisms, such as the axenic diatom Amphora coffeaeformis, the blue mussel Mytilus edulis as well as antibacterial activities isolated from marine biofilms [56]. However, compound 1d was considered to be a toxic compound to the barnacle cyprids in these assays. Suyama and his co-workers rediscovered compound 1d in a red algae cyanobacteria mix (Leptofaucia sp. and Oscillatoria sp., respectively) from the Grabo Reef in Papua New Guinea [40]. This is not surprising, as Unson and colleagues had previously demonstrated that the 3,5-dibromo-2-(2 ,4 -dibromophenoxy)phenol reported from Dysidea herabcea was not produced by the sponge itself but rather by its cyanobacterial symbiont (filamentous cyanobacterium), which closely resembles Oscillatoria spongeliae [18], indicating that cyanobacteria, especially of the genus Oscillatoria, is one of the main sources of PBDEs in nature. Suyama and colleagues also provided a few corrections on the 13 C chemical shift data that closely match our results using the same CDCl 3 solvent [40].
Compound 2b, on the other hand, was first discovered in the 1980s from the marine sponge Dysidea fragilis using HRMS and NMR analysis [41]. The compound exhibited antimicrobial activity against the clinically relevant Gram-positive bacteria S. aureus and B. subtilis [51]. The compound was reisolated recently with the same spectroscopic data along with compound 1d. The study revealed that adding more bromine positions in the B-ring increased the inhibitory potency for the α-D-galactosidase from the marine bacterium Pseudoalteromonas sp. KMM 701 of the GH36 family [55]. Furthermore, the multifaceted activities of compound 2b were also reported by Handayani et al. [30]. The compound showed potent antibacterial activity against the human pathogenic bacteria B. subtilis with a MIC of 3.1 µg/mL along with high cytotoxicity in the brine shrimp lethality test (BSLT), with an LC 50 of 0.9 µg/mL. Antifungal activity was shown against Cladosporium cucumerinum at 50 nmol (8 mm inhibition zone) and 25 nmol (5 mm inhibition zone). Antiproliferative activity of compound 2b has been reported against MCF-7 human adenocarcinoma breast cancer cells [37]. Unfortunately, NMR chemical data on compound 2b has only ever been incompletely provided in the more recent literature described above.
Information on deterrent activities of extracts from the sponge Dysidea started in the 1990s when Duffy and Paul reported deterrent activities against fishes during field feeding assays with tetra-brominated BDEs from Dysidea sp. collected in Guam, Micronesia [19]. Unfortunately, only the structure of the BDE was reported, and no structural data were reported for the compound, as they instead referred to the original description of the compound 3,5-dibromo-2-(2 ,4 dibromophenoxy)phenol, which Carte and Faulkner [20] described. The compound deterred the fishes in two kinds of feeding assay (paired and multiple-choice assay), where the incorporation of BDE in food pellets resulted in 40-70% rejection rates [19]. In a follow-up study by Pennings et al. [57], the deterrent activity of the compound was again confirmed with general reef fish and the pufferfish C. solandri and the crab Leptodius spp. in aquarium and field assays at 1, 2, and 4% dry mass. Faulkner et al. also confirmed an ecological role of observed BDEs in Dysidea herbacea from Palau, Micronesia. BDEs were abundantly located in needle-like forms and feather-like patterns in the ectosomal layer of the sponge [58]. They conclude that the accumulation of BDEs in the ectosomal layer of the sponge coupled with the feeding deterrent and antimicrobial activities indicates an ecological role of these compounds in defense against bacteria, crab, and fish predators. Further evidence for the defensive role of BDEs was provided by Becerro and colleagues [59], who detected 3,5-dibromo-2-(2 ,4 -dibromophenoxy)phenol in the sponge Dysidea granulosa from Guam and the sponge feeding gastropterids Sagaminopteron nigropunctatum and S. psychedelicum. Quantification of the compound revealed its accumulation by both sponge predators in the gastropterids tissue at over twice the sponge concentration, most likely for defense against fish predators [59].
Our results are the first report of sponge-derived PBDEs, penta-and hexa-brominated diphenyl esters (compounds 1d and 2b), exhibiting significant deterrence in the fish feeding assay and in antimicrobial activities against ecological relevant microorganisms. We also found that the presence of one additional bromine on the phenolic A-ring resulted in a higher fish deterrence but not in antimicrobial activity. However, since PBDEs are the major metabolites in the crude extracts ( Figure S3) and have considerable antibacterial and antifungal activity against clinically relevant microorganisms, one can assume that PBDEs are also responsible for the observed activities against environmental bacteria. Furthermore, they likely serve an important ecological function, as they were especially active against environmental pathogenic strains, such as Vibrio mediterranei and Aurantimonas coralicida, which have been implicated in several diseases of marine organisms [60,61]. In addition, pathogenic bacteria Pantoea eucrina and Rhodococcus corynebacterioides infections have been reported in humans [62,63].
The pronounced bioactivity of the two isolated PBDEs was further confirmed in antimicrobial assays with clinically relevant strains, where both compounds exhibited high inhibitory activity against B. subtilis, that was even five times stronger than the positive control tetracycline (MIC 2.8 µg/mL). Furthermore, compound 1d also inhibited S. aureus at the same concentration as the control, while it showed weak activity against E. coli and A. baumannii. In addition, both compounds also exhibited pronounced antifungal activity against R. glutinis, demonstrating MIC values that were around ten times lower (4.2 µg/mL and 2.1 µg/mL, respectively) than those of the positive control nystatin (33.3 µg/mL). The yeast Rhodoturula is a ubiquitous genus in different ecosystems, colonizing plants, humans, and other mammals [64]. R. glutinis was also found in benthic animals and deep-sea sediments from the northwest Pacific Ocean [65]. The pathogenicity of the genus Rhodoturula towards animals and humans by causing bloodstream infections, short bowel disease, and acute lymphoid leukemia has been reviewed comprehensively [64]. The emerging resistance of R. glutinis and other Rhodoturula species against caspofungin (100%), fluconazole (94.7%), luliconazole (1-8 µg/mL), and voriconazole (74.4%) has been well documented [66] and emphasizes the need for potent antifungal compounds such as 1d and 2b. Both compounds seem to have broad antifungal activity, as seen by their pronounced inhibition of M. hiemalis. Compound 1d also exhibited about 30% stronger activity against P. anomala than the positive control nystatin. Moderate fungicidal activity was also reported for compound 2b against C. cucumerinum [30].
L. herbacea crude extract showed activity in the HCV inhibitory assay. However, inhibition was weak compared to the positive control epigallocatechin gallate (EGCG) and therefore was not followed up upon with pure compounds. These results differed from a previous study that showed novel HCV inhibitory activity from the PBDE 6-hydroxy-2,2 ,4,4 -tetrabromodiphenyl ether from a Japan Dysidea sp [38].
Based on the bioactivity results reported here and previous literature reports, one can assume that PBDEs have an important ecological function in protecting organisms against various fish predators and pathogenic microorganisms. The production of these defensive metabolites might be one reason why L. herbacea is one of the dominant benthic invertebrates in certain areas [67]. In Indonesia, the dominance of Dysidea was recently recorded in the Wakatobi Marine National Park, Sulawesi, particularly in areas with high sedimentation [24,68]. Further research towards the ecological roles of PBDEs and especially their role in competitions with other sessile benthic invertebrates could close the gap of why Lamellodysidea is the prevalent inhabitant of shallow-water coral reefs throughout the Indo-Pacific.

Sample Collection
Lamellodysidea herbacea was collected by hand in Pari Island, Kepulauan Seribu, Indonesia, in August 2017. L. herbacea were taken at 3-8 m depths. The samples were cut with a scalpel and were then transferred into an ambient seawater-filled plastic bag. L. herbacea was morphologically identified as belonging to the Dysideidae family, Dictyoceratida order, and Demospongiae class. Collected samples were transferred to the nearest lab, frozen, and then transported to the Environmental Biochemistry working group, Carl von Ossietzky Universität Oldenburg, Oldenburg, Germany.

Extraction and Isolation
A wet weight sample of the sponge (243 g) was freeze-dried (31.65 g), grounded into powder, and extracted exhaustively with MeOH: EtOAc (1:1; HPLC grade VWR International GmbH, Darmstadt, Germany). Bioassay-guided isolation was conducted using 558 mg of the total organic extracts (4.1 g), which were further polarity partitioned into HEX (96.6 mg), DCM (127.8 mg), EtOAc (8.1 mg), and H 2 O (305.9 mg). The HEX and DCM fractions were chosen to be tested for all bioactivity screening assays except the HCV inhibitory assay. For identification of the potential bioactive compounds fractions were further separated using an HPLC (Agilent) fraction collector coupled with a UPLC-HRMS (MaXis ESI TOF, Bruker Daltonik GmbH, Bremen, Germany) with a BEH C 18 column (Waters ACQUITY, Milford, MA, USA) (1.7 µm 2.1 × 50 mm) [63]. A step gradient was used from 95% H 2 O and 5% MeCN (solvent A) to 5% H 2 O and 95% MeCN (solvent B). The resulting fractions were collected in a microplate and tested in MIC assays with clinically relevant pathogenic bacteria (see 4.5 for details). MS and MS/MS of the active peaks were obtained using UPLC-HRMS (Waters Synapt G2-Si, Milford, MA, USA) with the same gradient and column system described above and the MassLynx 4.2 data analysis program.

Feeding Deterrence
Feeding deterrence assays were conducted using a modified method from Pawlik [69], according to Helber et al. [70]. Thirteen pufferfish, Canthigaster solandri (Richardson, 1845), were chosen as model organisms since they are omnivorous invertebrate predators known to feed on sponges and since they have been used in several previous studies [35,70,71]. The assay was conducted using two kinds of artificial food pastes (0.05 g alginic, 0.075 g powdered squid mantle, one drop food color) consisting of treatment food with natural volumetric concentration of sponge crude extract, fractions or pure compounds added and without sponge chemistry as a control. The paste was made by mixing the ingredients with distilled water that were each loaded into separate 2 mL syringes. A long strand form was created by emptying the syringe content onto the plates containing a 0.25 M calcium chloride solution. The strand was rinsed with seawater several times and was then cut into small pellets to ensure that the puffers could eat them whole.
Thirteen individual puffers were maintained by feeding them with the pellets without the extract for several days before the assays. The assay was started by offering a control and treatment pellet sequentially to each puffer. The puffers were not scored if they did not eat the first control pellet. If the control was eaten, the treated pellet was then offered, and rejection was scored if the fish spat the pellet out or mothed the pellets (see previous studies for details) [70]. Next, another control pellet was offered to confirm that the fish was still hungry. The rejection was not scored if the second control was ignored or rejected. The pellet rejection of the first and second control was excluded from the calculation. The assay was continuously conducted until at least ten puffers participated in the assay. Fisher's exact test was applied for validating significant differences in the palatability between the control and treated pellet [70]. Further feeding assay with the two purified compounds were also conducted testing them at their natural concentration. The concentration was obtained by calculating the compounds' total peak area and by comparing it to a calibration curve ( Figure S2).

Agar Diffusion Assay (ADA) against Pathogenic and Non-Pathogenic Environmental Bacteria
Crude extracts were tested against 14 marine environmental bacteria strains, including pathogenic and non-pathogenic bacteria associated with marine diseases, using a modified version of the disk diffusion method [71]. The strains were identified and used by our colleagues in similar assays [35,72]. Marine broth (MB) agar plates were inoculated with 200 µL of the test strain cultures (24 h cultured in 10 mL liquid broth medium). Subsequently, different crude extracts (conc. 0.5 mg/disk), a negative control (MeOH: EtOAc, 15 µL/disk), and positive control (0.1 M Chloramphenicol, Oxoid ® , Z 15 µL/disk) were pipetted onto sterile paper disks (Ø 6 mm, Whatman) and dried. The impregnated disks were applied upside down onto the agar medium plates at equidistant points. The diameters of inhibition zones were measured to the closest mm after incubation at 37 • C for 24 and 48 h. The assays were conducted in three replicates. The results were expressed by calculating the average of the total inhibition zone surrounding the disks with the crude extract or fractions.

Minimum Inhibition Concentration (MIC) Antimicrobial Assay
To evaluate the bioactivity of the fractions against pathogenic bacteria and fungi, a minimum inhibition concentration (MIC) assay in 96-well microplates was conducted following Okanya et al. [73]. The samples were tested using a panel consisting of Gram-positive Bacteria Bacillus subtilis (DSM 10), Staphylococcus aureus (DSM 346), and Gram-negative bacteria Escherichia coli (DSM 1116), Pseudomonas aeruginosa (PA14), and Mycobacterium smegmatis (ATCC700084). Additional antifungal assays against the filamentous fungus, Mucor hiemalis (DSM 2656), and the yeasts Candida albicans (DSM 1665) and Rhodotorula glutinis (DSM 10134) were conducted. The bacterial strains were added into 20 mL MHB (Mueller Hinton Broth) medium, except for Mycobacterium smegmatis, which was added into Middlebrook 7H9 medium. Yeast and fungal strains were added into 20 mL MYC medium (1.0% phytone peptone, 1.0% glucose, 50 mM HEPES (11.9 g/L) pH 7.0). Assay starting conditions were adjusted to OD600 of 0.01 for the bacteria and to 0.05 for the yeast/fungi.
The MIC test was conducted by first pipetting 20 µL aliquots (10 mg/mL) of the sample into 280 µL of microbial culture in the well of the first row (A). Then, serial dilution steps (1:2) were applied until the eighth row (H). The result was evaluated with visual observation after overnight incubation (for Mycobacterium smegmatis was 48 h) on a microplate-vibrating shaker (Heidolph Titramax 1000, Heidolph Instruments GmbH & Co.KG, Schwabach, Germany) at 600 rpm and at 30 • C (for E. coli, P. aeruginosa, M. smegmatis was 37 • C). Negative and positive controls were also applied ( Table 1). Clear wells indicated no growth. Compounds were considered more potent if higher dilutions still resulted in no growth (the most potent extracts still inhibited test strains in the eighth row (H)).
Fractions were collected every 30 s using an Agilent 1100 HPLC system equipped with a fraction collector. The fractions were collected in 96-well plates and were dried [74], followed by inoculation with the test bacteria. The growth inhibition in the wells was correlated with the chromatogram peak based on the retention time. UPLC-HRMS was further used to analyze the mass and the UV spectrum of the active target compounds.

Inhibitory Effects on HCV Infectivity
Huh7.5 cells, steadily expressing Firefly luciferase (Huh7.5 Fluc; 1 × 10 4 /well) were provided by Charles Rice (Rockefeller University, New York, NY, USA) which were generated by the stable retroviral transduction of the gene encoding for the Firefly luciferase (FLuc) via a transduction approach described elsewhere [75]. The cells were provided by Charles Rice (Rockefeller University, New York, NY, USA). Those were seeded in 96-well plates with Dulbecco's modified Eagle's medium (DMEM, Gibco, Thermo Fisher Scientific, Schwerte, Germany) containing 2 mM glutamine, 1% minimum essential medium nonessential amino acids (MEM NEAA, Gibco, Thermo Fisher Scientific, Schwerte, Germany), 100 µg/mL streptomycin, 100 IU/mL penicillin (Gibco, Thermo Fisher Scientific, Schwerte, Germany), 10% fetal bovine serum, and 5 µg/mL blasticidin. Cells were incubated at 37 • C with a 5% CO 2 supply. The assay was conducted by infecting Huh7.5 cells with the Jc1-derived Renilla reporter viruses in the presence or absence of sponge crude extracts as described elsewhere [72]. Reporter virus infection was determined by Renilla luciferase activity. The cell viability was measured through the determination of Firefly luciferase, which was stably expressed in the target cells. Epigallocatechin gallate (EGCG; was purchased from Sigma-Aldrich, Seelze, Germany) was a positive control. After three days of incubation, the infected cells were washed with PBS, and 50µL H 2 O was added afterward. The infected cells were lysed and then frozen at −80 • C for 1 h following Renilla and Firefly luciferase activity measurements on a Berthold Technologies Centro XS3 Microplate Luminometer (Berthold Technologies GmbH & Co.KG, Bad Wildbad, Germany) as indicators of viral genome replication and cell viability, respectively.

Conclusions
The two isolated PBDEs (1d/2b) from the Indonesian marine sponge L. herbacea were characterized by detailed LC-MS and NMR experiments. Additionally, we used three different chemical shift calculation programs to verify the bromine positions within the PBDE 3,4,5-tribromo-2-(2 ,4 -dibromophenoxy)phenol (1d). Both compounds showed remarkable antimicrobial and feeding deterrent activities. Their ecological role in defense against fish predators and microbial pathogens had so far not been reported and might be a reason for the widespread occurrence of this sponge in shallow water coral reef habitats throughout the Indo-Pacific. This study also provided additional detailed activity results against various clinical pathogenic bacteria and fungi, demonstrating the bioactivity of the investigated PBDEs as potential antimicrobial and antifungal agents.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/md19110611/s1, Figure S1a. The HRMS (a,b) and MS/MS spectra (c) of the compound 1d. Figure S1b. The HRMS (a,b) and MS/MS spectra (c) of the compound 2b. Figure S2. Calibration curves of compounds 1d (a) and 2b (b) showing correlation between observed peak areas and injected compound concentrations (0.1 µg/mL, 1 µg/mL, 10 µg/mL, 100 µg/mL, 1 mg/mL). Figure S3. HPLC chromatogram from the crude extract of Lamellodysidea herbacea. It shows that compounds 1d and 2b are the major metabolites of the extract. Figure S4. UV chromatogram of compounds 1d and 2b (a,c), including their UV spectra and λmax [nm] (b,d). Table S1. 13 C NMR data for compound 1d in CDCl 3 compared to their chemical shift calculations by two NMR prediction programs NMRShiftDB and CSEARCH.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding authors.