Bioactivity of Spongian Diterpenoid Scaffolds from the Antarctic Sponge Dendrilla antarctica

The Antarctic sponge Dendrilla antarctica is rich in defensive terpenoids with promising antimicrobial potential. Investigation of this demosponge has resulted in the generation of a small chemical library containing diterpenoid secondary metabolites with bioactivity in an infectious disease screening campaign focused on Leishmania donovani, Plasmodium falciparum, and methicillin-resistant Staphylococcus aureus (MRSA) biofilm. In total, eleven natural products were isolated, including three new compounds designated dendrillins B–D (10–12). Chemical modification of abundant natural products led to three semisynthetic derivatives (13–15), which were also screened. Several compounds showed potency against the leishmaniasis parasite, with the natural products tetrahydroaplysulphurin-1 (4) and dendrillin B (10), as well as the semisynthetic triol 15, displaying single-digit micromolar activity and low mammalian cytotoxicity. Triol 15 displayed the best profile against the liver-stage malaria parasites, while membranolide (5) and dendrillin C (11) were strong hits against MRSA biofilm cultures.


Introduction
As the pharmaceutical industry has deprioritized research into infectious disease treatments in recent decades, new antimicrobial scaffolds are now being discovered less frequently [1,2]. The recent lull in novel antimicrobial development has steadily reduced the number of effective first-line treatments available, resulting in the gradual increase of drug-resistant infections, with over 2.8 million cases reported in the United States in 2019 [3]. The severity of this problem is predicted to escalate exponentially as resistance continues to surge, with the global death toll attributed to antimicrobial resistant infections estimated to exceed 10 million by the year 2050 [4]. Infectious parasitic diseases carry a significant global disease burden [5] with malaria and leishmania alone contributing to nearly half a million annual deaths. The need for the development of novel classes of infectious disease treatments is critical to the continued advancement of modern medicine [2,6,7].

Chemical Analysis of Dendrilla antarctica
Collections of Dendrilla antarctica were made in three consecutive field seasons, March-June 2016, 2017, and 2018. From the 2016 collection, freeze-dried sponge was extracted with dichloromethane. Purification of the concentrated extract by normal phase medium pressure liquid chromatography (MPLC) afforded 13 fractions, with one containing pure aplysulphurin (2). Other fractions were further purified by HPLC to give known natural products 9,11-dihydrogracilin A (3), tetrahydroaplysulphurin-1 (4), membranolide (5), and the new minor metabolite dendrillin B (10). The subsequent collections from 2017 and 2018 were extracted and fractionated using the same protocol, resulting in minor constituents glaciolide (6), a norditerpene 7, dendrillin A (9), and previously undescribed dendrillin C (11) and D (12). Previously reported compounds were characterized based on comparison of their NMR spectra to the published values. Dendrillins B-D are described here.

Chemical Analysis of Dendrilla antarctica
Collections of Dendrilla antarctica were made in three consecutive field seasons, March-June 2016, 2017, and 2018. From the 2016 collection, freeze-dried sponge was extracted with dichloromethane. Purification of the concentrated extract by normal phase medium pressure liquid chromatography (MPLC) afforded 13 fractions, with one containing pure aplysulphurin (2). Other fractions were further purified by HPLC to give known natural products 9,11-dihydrogracilin A (3), tetrahydroaplysulphurin-1 (4), membranolide (5), and the new minor metabolite dendrillin B (10). The subsequent collections from 2017 and 2018 were extracted and fractionated using the same protocol, resulting in minor constituents glaciolide (6), a norditerpene 7, dendrillin A (9), and previously undescribed dendrillin C (11) and D (12). Previously reported compounds were characterized based on comparison of their NMR spectra to the published values. Dendrillins B-D are described here.
Dendrillin B (10) was obtained as a clear oil with spectral data similar to that of previously reported membranolide (5), but with differences among the NMR signals of key functional groups about the aromatic ring. A formula of C21H28O5 was established from HRESIMS. Key 1 H NMR signals (Table 1), which established similarities to membranolide including gem methyl groups H3-18 (δH 0.36) and H3-19 (δH 0.93), and a singlet methyl H3-20 at δH 1.40 ( Figure 2). H3-18 and H3-19 correlated in the HMBC spectrum to alkyl carbons C-5 (δC 51.1) and C-3 (δC 39.3). The HMBC correlation of H3-20 to C-5, taken with COSY correlations among H2-3 (δH 1.32), H-2α (δH 1.70), H-2β (δH 1.79), H-1α (δH 2.35), and H-1β (δH 1.50), defined the 4,4,10-trimethylcyclohexyl group familiar to membranolide. HMBC correlation of H3-20 to C-9 (δC 154.4) placed the trimethylcyclohexyl group on an aromatic ring defined by a pair of nearly overlapping aromatic protons shifted considerably downfield (H-11, δH 7.78) and (H-12, δH 7.76) and three quaternary carbons, C-8 (δC 136.6), C-13 (δC 125.9), and C-14 (δC 145.5). The lowfield position of the aromatic protons were indicative of conjugation, supported by an HMBC correlation of H-12 to carbonyl C-16 (δC 168.1). C-16 was found to be a carboxylate ester situated in a fused hydroxybenzofuranone ring system via an HMBC correlation from H-15 (δH 6.70) to C-16, of which H-15 is revealed to be a hemiacetal methine through an HMBC correlation between OH (δH 4.28) and C-15 (δC 97.3), as well as observation of coupling between H-15 and the hydroxy proton. The final substituent on the aromatic ring was found to be a methyl propionate ester ortho to the trimethylcyclohexyl group based on HMBC correlations from a doublet methyl H3-17 (δH 1.75; J = 7.5) to C-8 (δC 136.6) and C-6 (δC 176.6), which coupled to a quartet methine H-7 (δH 4.78, J = 7.3). H-7 displayed further HMBC correlations to C-9 and C-14 (δC 145.5). The final confirmation of the position of the terminal methyl ester is supported by an HMBC correlation from H3-21 (δH 3.67) to C-6 ( Figure 2).  The relative configuration of dendrillin B (10) was established as shown in Figure 2, and is predicated on the lack of rotation about C-7 of the propionate ester owing to the steric hindrance from its close proximity to the trimethylcyclohexyl substituent [18]. The aromatic ring is expected to be in the axial position, consistent with other aromatic members of this series, including aplysulphurin [21] and membranoid A [15] (= membranolide B [18]). In this position, H 3 -18 is found in the anisotropic shielding zone of the aromatic ring, exemplified by its highfield shift (δ H 0.36), and placing it in proximity of H 3 -17 to observe an NOE between the two methyl groups. Additional NOEs observed for the trimethylcyclohexane ring are consistent with membranolide (5) and membranoid A [15,18]. The final stereocenter about C-15 was established with the hemiacetal methine in the α-orientation owing to a NOESY correlation between H-15 (δ H 6.70) and H 3 -17.
Dendrillin C (11) was obtained as a clear oil and a formula of C 20 H 32 O 3 was established from HRESIMS. The same trimethylcyclohexane ring was evident from the H 3 -18 (δ H 0.98), H 3 -19 (δ H 0.92), and H 3 -20 (δ H 1.03) shifts (Table 1) and corresponding HMBC correlations ( Figure 3). An additional triplet methyl group H 3 -17 (δ H 1.13) was shown through COSY correlation to H 2 -7 (H-7a (δ H 2.43) and H-7b (δ H 3.20)). That C-7 (δ C 26.7) bears diastereotopic protons supports previous assignments (above) of restricted rotation in that congested ring system. H 3 -17 correlated by HMBC to C-8 (δ C 167.9), and H-7b to C-14 (δ C 134.2), fixing the ethyl group at C-8. Further HMBC correlations of H-11β (δ H 2.01) to C-8 and H-12β (δ H 1.91) to C-14, as well as a COSY correlations between H-11β and H-12β and H-12β to H-13 (δ H 3.60), secured the assignment of the reduced aromatic ring that was found in dendrillin B (10). The presence of an aldehyde at C-15 (δ C 190.5) conjugated with the ∆ 8,14 olefin was established by observation of HMBC correlation of H-15 (δ H 10.18) to C-14. A methyl ester was revealed at C-16 (δ C 174.8) and C-21 (δ C 51.8), which could be positioned about the ring based on an HMBC correlation between H-12β and C-16. 145.5). The lowfield position of the aromatic protons were indicative of conjugation, supported by an HMBC correlation of H-12 to carbonyl C-16 (δC 168.1). C-16 was found to be a carboxylate ester situated in a fused hydroxybenzofuranone ring system via an HMBC correlation from H-15 (δH 6.70) to C-16, of which H-15 is revealed to be a hemiacetal methine through an HMBC correlation between OH (δH 4.28) and C-15 (δC 97.3), as well as observation of coupling between H-15 and the hydroxy proton. The final substituent on the aromatic ring was found to be a methyl propionate ester ortho to the trimethylcyclohexyl group based on HMBC correlations from a doublet methyl H3-17 (δH 1.75; J = 7.5) to C-8 (δC 136.6) and C-6 (δC 176.6), which coupled to a quartet methine H-7 (δH 4.78, J = 7.3). H-7 displayed further HMBC correlations to C-9 and C-14 (δC 145.5). The final confirmation of the position of the terminal methyl ester is supported by an HMBC correlation from H3-21 (δH 3.67) to C-6 ( Figure 2). ) establishing the planar structure of dendrillin C (11).
The relative configuration of dendrillin C (11) was approached by comparison with the other members of this family. Two distinguishing features characterize the planar structure of 11 than influence its conformation; one is the reduced (tetrahydro) central ring and the second is the methine at C-9. Aromatic central ring members of this family of D. antarctica metabolites have a well-characterized configuration with the central ring substituent occupying the axial position of the trimethylcyclohexane chair conformation. This is borne out both by X-ray analysis of, for example, aplysulphurin [26], membranoids C and E [15], and darwinolide [17], and the respective upfield shift of H 3 -19 (δ H~0 .5) owing to its position in the anisotropic cone of the aromatic ring. H 3 -20, in the equatorial position, similarly has a characteristic deshielded shift (~δ H 1.4). Dendrillin C lacks the shielded H 3 -19 (δ H 0.92) and deshielded H 3 -20 (δ H 1.03) indicative of its central ring in an equatorial position on the trimethylcyclohexane ring system ( Figure 4). NOESY correlations from the axial methyl H 3 -20 to equatorial H-5β (δ H 1.46) and axial H 3 -19 and from equatorial H 3 -18 (δ H 0.98) to axial H-5α (δ H 1.35), support the conformation of the trimethylcyclohexane system illustrated in Figure 4. H-7a (δ H 2.43) has a NOESY correlation to H-5β, fixing the ethyl group in the proximity of C-5. The configuration at C-13 was established with the methyl ester in the β-orientation based on a NOESY correlation between H-13 (δ H 3.60) and H-12α (δ H 1.82), placing the H-13 methine in the pseudoequatorial orientation. Dendrillin D (12) was obtained as a clear oil with spectral data unlike that of the previously isolated dendrillins and related spongian diterpenes [27], but with certain motifs conserved, including the trimethylcyclohexane substituent and its attached reduced (tetrahydro) aromatic ring. A formula of C 21 H 32 O 4 was established from HRESIMS. Key 1 H NMR signals (Table 1) established the presence of a C-9 (δ H 2.18) methine, which displayed a COSY correlation to H 2 -11 (δ H 1.74), as well as HMBC correlation to C-14 (δ C 128.0). Further HMBC correlations were observed between H 2 -11/C-12 (δ C 27.3) and H 2 -11/C-13 (δ C 30.7), from a doublet methyl H 3 -17 (δ H 1.45, J = 6.5) to C-8 (δ C 169.3), and from its coupled quartet methine H-7 (δ H 5.21, J = 6.5) to C-14 and a carbonyl C-15 (δ C 172.5). These data suggested the central cyclohexene ring is part of a larger bicyclic tetrahydroisobenzofuranone system, which forms the core of the structure. A final acetoxymethyl substituent was found on the fused ring structure connected at C-13, and is supported in its assignment by a COSY correlation from H-13 (δ H 2.83) to H 2 -16 (δ H 4.21), as well as HMBC correlations from H 2 -16 to C-12, C-14, a carbonyl C-21 (δ C 171.0) and an acetate methyl C-22 (δ C 21.0). The only HMBC correlation found from H 3 -22 (δ H 2.06) is to its adjacent carbonyl C-21, establishing the planar structure as 12 ( Figure 5). The relative configuration of dendrillin D was supported in part by the prior analysis of dendrillin C (11), including the axial nature of H 3 -20 (δ H 0.94) and H 3 -19 (δ H 0.92), which was supported by NOESY correlations ( Figure 5) of both of them to equatorial Hβ-5 (δ H 1.41). NOESY correlation was also observed from equatorial H 3 -18 (δ H 0.99) to axial Hα-5 (δ H 1.11). These data established the trimethylcyclohexane ring as conformationally the same as 11. The orientation of C-7 was assigned through NOESY correlations from H 3 -17 (δ H 1.45) to equatorial H 3 -18, and from H-7 (δ H 5.21) to equatorial Hβ-5 and axial H 3 -20, placing H 3 -17 in the α-orientation and H-7 in the β-orientation. The α-orientation of H-9 (δ H 2.18) was then assigned via a NOESY correlation between H-9 and H 3 -17, as well as H-1α and H-5α, placing C-11 in the β-orientation. The final stereocenter about C-13 is established with the acetoxymethyl in the β-orientation owing to NOESY correlations from H 2  , from a doublet methyl H3-17 (δH 1.45, J = 6.5) to C-8 (δC 169.3), and from its coupled quartet methine H-7 (δH 5.21, J = 6.5) to C-14 and a carbonyl C-15 (δC 172.5). These data suggested the central cyclohexene ring is part of a larger bicyclic tetrahydroisobenzofuranone system, which forms the core of the structure. A final acetoxymethyl substituent was found on the fused ring structure connected at C-13, and is supported in its assignment by a COSY correlation from H-13 (δH 2.83) to H2-16 (δH 4.21), as well as HMBC correlations from H2-16 to C-12, C-14, a carbonyl C-21 (δC 171.0) and an acetate methyl C-22 (δC 21.0). The only HMBC correlation found from H3-22 (δH 2.06) is to its adjacent carbonyl C-21, establishing the planar structure as 12 ( Figure 5). The relative configuration of dendrillin D was supported in part by the prior analysis of dendrillin C (11), including the axial nature of H3-20 (δH 0.94) and H3-19 (δH 0.92), which was supported by NOESY correlations ( Figure 5) of both of them to equatorial Hβ-5 (δH 1.41). NOESY correlation was also observed from equatorial H3-18 (δH 0.99) to axial Hα-5 (δH 1.11). These data established the trimethylcyclohexane ring as conformationally the same as 11. The orientation of C-7 was assigned through NOESY correlations from H3-17 (δH 1.45) to equatorial H3-18, and from H-7 (δH 5.21) to equatorial Hβ-5 and axial H3-20, placing H3-17 in the α-orientation and H-7 in the β-orientation. The α-orientation of H-9 (δH 2.18) was then assigned via a NOESY correlation between H-9 and H3-17, as well as H-1α and H-5α, placing C-11 in the β-orientation. The final stereocenter about C-13 is established with the acetoxymethyl in the β-orientation owing to NOESY correlations from H2-16 (δH 4.21) to β oriented H2-11 (δH 1.74) and axial H3-20.

Semisynthetic Studies
To further explore Dendrilla antarctica diterpene bioactivity, three of the major natural products were subjected to derivatization procedures ( Figure 6). Both 9,11-dihydrogracilin A (3) and tetrahydroaplysulphurin-1 (4) were treated under reductive ozonolysis conditions. The expected 8ketodihydrogracilin (13) [28] (see Figure S22) was recovered from reaction with 3. The same reaction conditions applied to tetrahydroaplysulphurin-1 (4) did not lead to isolation of a di-ketone derivative owing to rapid degradation. However, a product could be purified from this reaction. 1D and 2D NMR analysis, taken with mass spectrometry, found a formula of C22H32O8, bearing three more oxygen atoms than the original compound. The presence of two quaternary carbons at 108.5 and 117.2 ppm, with the remaining atoms assignable to a tetrahydroaplysulphurin-like scaffold, secured the product as the ozonide (14) (Figures S23-27). In contrast to their traditional role as intermediates in oxidative alkene cleavage, 1,2,4-trioxolanes with bulky substituents, as found in 14, can be relatively stable compounds. Such stabilized ozonides have found interest in treatment of malaria based on their resemblance to the natural product artemisinin [29].
The third semisynthetic product prepared for our structure-activity profile was derived by the lithium aluminum hydride (LAH) reduction of membranolide (5), which yielded the corresponding tetrahydroaplysulphurin-1 (4), membranolide (5), and the new minor metabolite dendrillin B (10). The subsequent collections from 2017 and 2018 were extracted and fractionated using the same protocol, resulting in minor constituents glaciolide (6), a norditerpene 7, dendrillin A (9), and previously undescribed dendrillin C (11) and D (12). Previously reported compounds were characterized based on comparison of their NMR spectra to the published values. Dendrillins B-D are described here.
Dendrillin B (10) was obtained as a clear oil with spectral data similar to that of previously reported membranolide (5), but with differences among the NMR signals of key functional groups about the aromatic ring. A formula of C21H28O5 was established from HRESIMS. Key 1 H NMR signals (Table 1)

Semisynthetic Studies
To further explore Dendrilla antarctica diterpene bioactivity, three of the major natural products were subjected to derivatization procedures ( Figure 6). Both 9,11-dihydrogracilin A (3) and tetrahydroaplysulphurin-1 (4) were treated under reductive ozonolysis conditions. The expected 8-ketodihydrogracilin (13) [28] (see Figure S22) was recovered from reaction with 3. The same reaction conditions applied to tetrahydroaplysulphurin-1 (4) did not lead to isolation of a di-ketone derivative owing to rapid degradation. However, a product could be purified from this reaction. 1D and 2D NMR analysis, taken with mass spectrometry, found a formula of C 22 H 32 O 8 , bearing three more oxygen atoms than the original compound. The presence of two quaternary carbons at 108.5 and 117.2 ppm, with the remaining atoms assignable to a tetrahydroaplysulphurin-like scaffold, secured the product as the ozonide (14) (Figures S23-27). In contrast to their traditional role as intermediates in oxidative alkene cleavage, 1,2,4-trioxolanes with bulky substituents, as found in 14, can be relatively stable Mar. Drugs 2020, 18, 327 7 of 14 compounds. Such stabilized ozonides have found interest in treatment of malaria based on their resemblance to the natural product artemisinin [29]. The third semisynthetic product prepared for our structure-activity profile was derived by the lithium aluminum hydride (LAH) reduction of membranolide (5), which yielded the corresponding triol (15) ( Figure 6). Spectroscopic analysis ( Figure S30) was in agreement with that previously reported [19].

Biological Profiling
The full suite of Dendrilla antarctica-derived compounds, with the exception of darwinolide (1) and dendrillin D (12) owing to mass limitations, were tested against Leishmania donovani infected macrophages [15], liver-stage sporozoites in primary human hepatocytes [30], and against MRSA biofilm. Cytotoxicity was evaluated using the mammalian J774A.1 cell line in L. donovani-active compounds, while primary human hepatocytes were evaluated for malaria-active compounds.
For the Plasmodium falciparum infected liver cells, screening was conducted at a fixed concentration of 5 µg/mL for each compound, and activities representing 60% inhibition and 100% inhibition are reported in comparison with the positive control primaquine. The results ( Table 2) indicate modest activity (60% inhibition) at low micromolar concentrations for tetrahydroaplysulphurin-1 (4, ≤13 µM) and its ozonide derivative (14, ≤15 µM), dispelling any boost in activity for the ozonide. The most promising bioactivity (100% inhibition) was found for semisynthetic triol 15, (≤16 µM). Aplysulphurin (2) at ≤13 µM was also active at the highest level. None of the active compounds displayed discernable cytotoxicity towards uninfected primary human hepatocytes. The MRSA biofilm assay was carried out at concentrations of 100, 50, and 25 µg/mL for each compound, and activities representing 50% and 90% eradication are reported. The results (Table 2) show that cadlinolide C (8), dendrillin B (10), and ozonide (14) achieved 90% eradication at 100 µg/mL. 8-Ketodihydrogracilin (13) and dendrillin C (11) exhibited activity at 50 µg/mL, though dendrillin C was also active at the lowest level tested (25 µg/mL), achieving 50% eradication. The most active compound in the screen was membranolide (5), which displayed potent activity (90% eradication) at the lowest concentration tested (25 µg/mL, 58 µM), and in fact achieved nearly 100% eradication at this level, suggesting the minimum inhibitory concentration against MRSA biofilm is actually much lower than reported here.

Conclusions
Dendrilla antarctica is a chemically rich sponge that continues to be a source of new chemistry and promising bioactivities. Following up on previous investigations, our large field collections of this well-studied Antarctic sponge in 2016, 2017, and 2018 have provided the resources for kg-scale dry weight bulk extractions, which resulted in a library of 11 diterpenoids, of which eight were previously described. Glaciolide (6) and 7 are known from northern environments [23,24], but are newly reported here from D. antarctica. Three new natural products have been designated dendrillins B-D (10-12). In the process, noteworthy bioactivities against infectious disease models including pathogens that cause leishmaniasis, malaria, and MRSA infections were found, revealing insights into diverse structural elements that could be meaningful in future drug development efforts.
The aromatic compounds in the library (2, 5, 10), each of which contains a variant of an isobenzofuran system with an ester at position C-6, showed activity below 10 µM against the leishmaniasis parasite. Moreover, the gain of some flexibility seems to not affect the bioactivity, but rather appears to be beneficial as it reduces the cytotoxicity, as shown for tetrahydroaplysulphurin-1 (4) compared with its parent compound, fully unsaturated and fairly rigid, aplysulphurin (2). The same observation is true for the semi-synthetic triol (15) compared with its parent compound membranolide (5); the reduction of the two ester moieties led to a modest improvement in activity (from 9.7 to 4.2 µM), but dramatically decreased the cytotoxicity. In the end, these improvements are translated into a selectivity index ratio of almost 40 for the triol compared with 8 for the natural product. Another interesting feature is revealed by comparison of 4 to cadlinolide C (8), differing only by an acetoxy substitution at C-16. The lack of this structural unit results in a four-fold drop of the bioactivity. Very similarly, membranolide, missing substitution at C-16 position, is less active than the structurally comparable dendrillin B (10), suggesting that the stereochemistry at C-15 position may not be crucial. From this perspective, the activity improvement associated with the gain of flexibility resulting in the lactone opening of membranolide to produce the triol seems to validate the importance of a substituent at C-16 position. Analysis of the top hits in each bioassay (Table 3) could represent a reasonable starting point for further structure/activity (SAR) studies. The high selectivity index (>40) of the new chemical entity dendrillin B in this regard is especially encouraging and warrants continued investigation as it showed no discernable toxicity against the human cell line and was among the most potently active, rivaling the activity seen in the miltefosine control.
Another prospective pharmacophore is revealed in the activity noted against Plasmodium falciparum by structurally similar compounds aplysulphurin (2), tetrahydroaplysulphurin-1 (4), and its ozonide derivative (14), each of which contain a fused furopyranone ring system bearing an acetate at positon C-16 and are the only compounds within the small library to sport this motif. The addition of a ozonide moieties, although known to retain antimalarial activity [30], did not significantly improve the activity. Aplysulphurin (2) and the semisynthetic triol (15) in particular showed the best activity profile, displaying 100% inhibition of the parasite cultures at low micromolar concentrations with no discernable cytotoxicity towards heathy liver cells. This could be owing to the increased rigidity of the furopyranone ring system, which in aplysulphurin is fused to a planar benzene ring and could represent another promising motif practicable for optimization. However, aplysulphurin appears bioactive in many bioassays, including some cytotoxicity assays.  Another prospective pharmacophore is revealed in the activity noted against Plasmodium falciparum by structurally similar compounds aplysulphurin (2), tetrahydroaplysulphurin-1 (4), and its ozonide derivative (14), each of which contain a fused furopyranone ring system bearing an acetate at positon C-16 and are the only compounds within the small library to sport this motif. The addition of a ozonide moieties, although known to retain antimalarial activity [30], did not significantly improve the activity. Aplysulphurin (2) and the semisynthetic triol (15) in particular showed the best activity profile, displaying 100% inhibition of the parasite cultures at low micromolar concentrations with no discernable cytotoxicity towards heathy liver cells. This could be owing to the increased rigidity of the furopyranone ring system, which in aplysulphurin is fused to a planar benzene ring and could represent another promising motif practicable for optimization. However, aplysulphurin appears bioactive in many bioassays, including some cytotoxicity assays.
Several of the Dendrilla antarctica-derived compounds showed promising activity in the MRSA biofilm screen, including the new natural products dendrillin B (10), and more potently dendrillin C (11), which is worth noting owing to their introduction as new chemical entities that could be further optimized, as well as the robust nature of the notoriously difficult to treat MRSA biofilm, and the lack of currently effective treatments [3]. Another interesting feature is revealed by a comparison between 9,11-dihydrogracilin A (3) and its ozonolysis product 8-ketodihydrogracilin (13). The latter displayed activity when its parent compound did not. This structural difference, namely the oxidative cleavage of the double bound leading to a truncated moiety, seems to be favorable for the bioactivity. However, the most encouraging result from this assay is the activity of the known natural product membranolide (5), which displayed remarkable activity even at the lowest concentrations tested. Given the low cytotoxicity against J774A.1 cells (76.8 µ M) and the tendency of MRSA biofilm to form plaques on surfaces, membranolide represents another promising lead in addition to darwinolide (1) that could potentially be developed for topical treatment after further evaluation and structural optimization.
Overall, this investigation has highlighted the robust bioactivity profile of the cold-water sponge D. antarctica. Besides bioactive natural products, semi-synthetic derivatives have informed structure/activity relationships and identified strategies for further study of infectious disease drug targets.  Another prospective pharmacophore is revealed in the activity noted against Plasmodium falciparum by structurally similar compounds aplysulphurin (2), tetrahydroaplysulphurin-1 (4), and its ozonide derivative (14), each of which contain a fused furopyranone ring system bearing an acetate at positon C-16 and are the only compounds within the small library to sport this motif. The addition of a ozonide moieties, although known to retain antimalarial activity [30], did not significantly improve the activity. Aplysulphurin (2) and the semisynthetic triol (15) in particular showed the best activity profile, displaying 100% inhibition of the parasite cultures at low micromolar concentrations with no discernable cytotoxicity towards heathy liver cells. This could be owing to the increased rigidity of the furopyranone ring system, which in aplysulphurin is fused to a planar benzene ring and could represent another promising motif practicable for optimization. However, aplysulphurin appears bioactive in many bioassays, including some cytotoxicity assays.
Several of the Dendrilla antarctica-derived compounds showed promising activity in the MRSA biofilm screen, including the new natural products dendrillin B (10), and more potently dendrillin C (11), which is worth noting owing to their introduction as new chemical entities that could be further optimized, as well as the robust nature of the notoriously difficult to treat MRSA biofilm, and the lack of currently effective treatments [3]. Another interesting feature is revealed by a comparison between 9,11-dihydrogracilin A (3) and its ozonolysis product 8-ketodihydrogracilin (13). The latter displayed activity when its parent compound did not. This structural difference, namely the oxidative cleavage of the double bound leading to a truncated moiety, seems to be favorable for the bioactivity. However, the most encouraging result from this assay is the activity of the known natural product membranolide (5), which displayed remarkable activity even at the lowest concentrations tested. Given the low cytotoxicity against J774A.1 cells (76.8 µ M) and the tendency of MRSA biofilm to form plaques on surfaces, membranolide represents another promising lead in addition to darwinolide (1) that could potentially be developed for topical treatment after further evaluation and structural optimization.
Overall, this investigation has highlighted the robust bioactivity profile of the cold-water sponge D. antarctica. Besides bioactive natural products, semi-synthetic derivatives have informed structure/activity relationships and identified strategies for further study of infectious disease drug targets.
1 Table 3. Structure/activity (SAR) study of top Dendrilla antarctica-derived compounds against three pathogens. Green boxes illustrate promising activity, red indicates characteristics to be improved.   Several of the Dendrilla antarctica-derived compounds showed promising activity in the MRSA biofilm screen, including the new natural products dendrillin B (10), and more potently dendrillin C (11), which is worth noting owing to their introduction as new chemical entities that could be further optimized, as well as the robust nature of the notoriously difficult to treat MRSA biofilm, and the lack of currently effective treatments [3]. Another interesting feature is revealed by a comparison between 9,11-dihydrogracilin A (3) and its ozonolysis product 8-ketodihydrogracilin (13). The latter displayed activity when its parent compound did not. This structural difference, namely the oxidative cleavage of the double bound leading to a truncated moiety, seems to be favorable for the bioactivity. However, the most encouraging result from this assay is the activity of the known natural product membranolide (5), which displayed remarkable activity even at the lowest concentrations tested. Given the low cytotoxicity against J774A.1 cells (76.8 µM) and the tendency of MRSA biofilm to form plaques on surfaces, membranolide represents another promising lead in addition to darwinolide (1) that could potentially be developed for topical treatment after further evaluation and structural optimization.
Overall, this investigation has highlighted the robust bioactivity profile of the cold-water sponge D. antarctica. Besides bioactive natural products, semi-synthetic derivatives have informed structure/activity relationships and identified strategies for further study of infectious disease drug targets.

General Procedures
Solvents were obtained from Fisher Scientific Co. and were HPLC grade (>99% purity) unless otherwise stated. MPLC analysis and fractionation were performed on a Teledyne-Isco CombiFlash system equipped with an evaporative light scattering detector (ELSD). HPLC analysis and fractionation were performed on an Agilent 1200 system equipped with an Agilent 1200 DAD and/or Shimadzu LC20-AT or LC10-AT system equipped with a Sedex 75 ELSD and/or a SPD-10ATvp UV-Vis detector, using analytical (Phenomenex Luna Silica (250 × 4.6 mm, 5 µm)), semi-preparative (Phenomenex Luna Silica (250 × 10 mm, 5 µm)), or preparative (Phenomenex Luna Silica (250 × 21.2 mm, 5 µm)) conditions. Analytical LC/MS was performed on a Phenomenex Kinetex C18 column (50 × 2.1 mm, 2.6 µm) on an Agilent 6230 LC/TOF-MS with electrospray ionization detection. NMR spectra were acquired in CDCl 3 with residual solvent referenced as an internal standard (δ H 7.27 ppm; δ C 77.0 ppm) for 1 H and 13 C NMR spectra, respectively. 1 H NMR spectra were recorded on a Varian 500 MHz or 600 MHz direct-drive instrument equipped with cold-probe detection and 13 C NMR spectra were recorded at 125 MHz. UV absorptions were measured by an Agilent Cary 60 UV/vis spectrophotometer in CH 3 OH, while IR spectra were recorded with an Agilent Cary 630 FTIR. Optical rotations were measured using an AutoPol IV polarimeter at 589 nm utilizing a 10 mm path length cell.

Collection of Dendrilla antarctica
Sponge samples were collected using SCUBA at depths of 10-35 m from various sites within a 3.5 Km radius around Palmer Station, Antarctica (64.7732 • S, 64.0538 • W) in the austral summers of 2016, 2017, and 2018. A specimen examined by Professor Rob van Soest, then at the University of Amsterdam, was originally identified as Dendrilla membranosa, but recently revised to D. antarctica [15]. Samples were cleaned and then frozen at −70 • C, followed by lyophilization, and transported back to the University of South Florida for further processing.
One final round of isolations was made when additional minor constituents (9 and 12) were detected in several sponges collected during the 2018 field season, which were previously analyzed for a separate metabolomics study. While their masses did not contribute significantly to the metabolomic analysis, an attempt was made to isolate these components by combining the worked-up extracts of four sponges, resulting in a combined extract mass of 182.1 mg. This extract was subjected to NP analytical HPLC with a binary solvent gradient starting and holding at 5% EtOAc for the first 10 min, ramping up to 20% EtOAc over the next 10 min, increased to 100% EtOAc over the following 5 min, and finally held at 100% EtOAc for 5 min. This ultimately resulted in the isolations of new natural product dendrillin D (12) (0.7 mg) collected at a retention time of 20.3 min, dendrillin A (9) (1.8 mg) collected at a retention time of 22.0 min, as well as additional amounts of 9,11-dihydrogracilin A (3) (~60 mg) collected at a retention time of 13.8 min. Known structures (3,9) were assigned based on comparison with their published data [20,24].

Ozonolysis of tetrahydroaplysulphurin-1 (4):
Ozone was bubbled through a solution of tetrahydroaplysulphurin (4) (15 mg) in 2.5 mL of CH 2 Cl 2 kept at 0 • C. The reaction was incubated for 5 min, after which 500 µL of dimethylsulfide was added and incubated for 45 min at 0 • C. The solvent was eliminated under nitrogen gas and residues were dissolved in 300 µL of CH 2 Cl 2 and NP-HPLC used to isolate products. Then, 3.8 mg of ozonide (14) corresponding to 23% yield was isolated and used for the spectroscopic characterization. HRESIMS m/z 447.199 [M + Na] + 447.199 calculated for C 22

Leishmania Donovani and J774A.1-Cell Cytotoxicity Assay
The Leishmania donovani cell line infected macrophage assay and cytotoxicity screen were conducted as previously described [15].

Liver-Stage Plasmodium Falciparum Assay
The liver-stage Plasmodium falciparum assay was conducted as previously reported [16,31].

Minimum MRSA Biofilm Eradication Concentration Assay
The MRSA biofilm assay was conducted as previously reported [17].