New Antimicrobial Bromotyrosine Analogues from the Sponge Pseudoceratina purpurea and Its Predator Tylodina corticalis

Bioassay-guided fractionation of extracts from temperate Australian collections of the marine sponge Pseudoceratina purpurea resulted in the isolation and characterisation of two new and six known bromotyrosine-derived alkaloids with antibiotic activity. Surprisingly, a single specimen of the mollusc Tylodina corticalis, which was collected while feeding on P. purpurea, contained only a few of the compounds found in the sponge suggesting selective accumulation and chemical modification of sponge metabolites.

In our continuing effort to isolate and identify new drug candidates from marine sponges [26] we screened temperate water sponges for antimicrobial and herbicidal activity. This revealed that the ethyl acetate partition of the ethanolic extract of Pseudoceratina purpurea (Carter 1880) displayed selective OPEN ACCESS activity against Staphylococcus aureus (35 mm zone of inhibition in a disc diffusion assay with no activity against P. aeruginosa). This sponge has previously been reported to be a rich source of bromotyrosine-derived alkaloids with wide ranges of biological activity [8,12,16,[27][28][29][30][31][32][33][34][35]. However, temperate Australian specimens of this sponge have not previously been investigated and it is a characteristic of this species that different alkaloid profiles have been found from sponges collected at different locations.
In addition we were also encouraged to investigate the natural products of this sponge because we observed, and collected, a single specimen of the opisthobranch Tylodina corticalis (Tate, 1889) feeding on this sponge. As Tylodina is a partially shelled mollusc it would be interesting to determine if this mollusc accumulates defensive chemicals from its diet, lending weight to the theory that chemical defence is a pre-adaptation to loss of the protective shell found in most molluscs [25,36]. Tylodina corticalis has not previously been chemically investigated, but Proksch et al. have published their work on Tylodina perversa from the Mediterranean [5,[37][38][39]. In their study, they found that T. perversa, fed on a diet of Aplysina aerophoba sequestering uranidine, isofistularin-3, aerophobin 1, aerophobin 2, and aplysinamisin-1. Whereas aerophobin 2 and isofistularin-3 made up about one third of the total alkaloids each in the sponge, aerophobin 2 constituted ~70% of the alkaloids found in the mantles, mucus, and egg masses of T. perversa, indicating selective sequestration by the mollusc. Another 20% of the alkaloids found in T. perversa was attributed to aerothionin, which was not detected in the prey sponge suggesting that this compound may have come from a previous diet of a related sponge (e.g., A. cavernicola). Several species of Aplysina have been reported to contain high concentrations of aerothionin [13,[40][41][42][43].
Herein we report the identification of 16 bromotyrosine alkaloids from Pseudoceratina purpurea and Tylodina corticalis by high resolution LC-MS/UV-Vis spectroscopy. Eight of these were isolated preparatively and subject to 2D NMR analysis, revealing two to be new compounds.

Results and Discussion
The ethanol extracts of both the sponge and mollusc appeared as yellow solutions, which quickly turned dark-purple when exposed to air. This suggested the presence of the well-known verongid sponge pigment uranidine [44]. As the crude extract showed selective activity against S. aureus (Supplementary Information), activity against this organism was used for the bioassay-guided isolation. The ethyl acetate partition of the ethanolic extract of P. purpurea was subject to gel filtration (Sephadex LH-20) and fractions (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22), with activity against S. aureus, were combined. A similar extraction of T. corticalis yielded only 12 mg in the ethyl acetate partition so this was not subject to gel filtration but analysed without further fractionation. Comparison of the two extracts ( Figure 1A,B) suggested a similar pattern of metabolites eluting around 20 min but the T. corticalis extract was much simpler in composition. However, more careful examination of this region ( Figure 1C,D) showed that even the major peaks of the sponge and mollusc did not overlap exactly. Using nanospray HRMS in combination with UV-Vis spectra ( Figure S1; Supplementary Information) allowed us to tentatively identify most of the compounds from both organisms (Table 1). While both organisms contained bromotyrosine-derived alkaloids, very little overlap in secondary metabolites between the two organisms existed (highlighted in grey in Table 1), which was surprising considering there was evidence of extensive feeding by T. corticalis on the sponge sample that was collected and extracted in this work.  Preparative HPLC separation of the eight major compounds from the sponge confirmed the tentative structures assigned by HR LC-MS and UV-Vis and that two of the compounds (1 and 2) were new ( Figure 2). The six known compounds were (−)-pseudoceratinine A (3) [45]; (+)-aerophobin 2 (4) [46]; (−)-hexadellin A (5) [47]; aplysamine 2 (6) [48]; 16-debromoaplysamine 4 (7) [49]; (−)-purealin (8) [50] (Figure 2) and their NMR spectral data matched those reported in the literature. The two new compounds were subject to full spectral analysis to determine their structure and stereochemistry.
The signal at δH 8.51 (NH) showed a coupling to δH 3.13 (H-17) and an HMBC correlation to the amide carbonyl (C-14; δ 158.9). Further COSY correlations from H-17 to H-18 (δH 1.73), and in turn H-19 (δH 2.51) supported the conclusion that substructure B was an amide unit connected to a propyl chain (c.f. an ethyl chain in ceratinadin A and B). HMBC correlations between H-19 and C-20 and C-24 ( Figure 4) indicated that the CH2 at δC 21.2 ppm was connected to the imidazole ring. The w-COSY correlation between two broad singlets at δH 11.93 (N-21H) and δH 12.15 (N-23H), and the 2H singlet at δH 7.77 suggested the imidazole is present in the form of an imidazolium salt. The presence of a homohistamine unit is commonly observed in verongid metabolites [51] and was also found in other metabolites from this sponge. For example, aerophobin, pseudoceratinin A, and purealin show similar NMR data for this substructure and provided good evidence for unit B/C. Although there were no correlations to C-22, a quaternary carbon at δ 146.6 was observed, which matches literature values for 2-aminoimidazole. Because of resonance and long T1 relaxation times, it is common not to observe correlations with guanidine carbons.
The HMBC correlations in the trihydroxy quinolinone (uranidine) ring system (H-1′ to C-2′ and C-10′) matched those previously reported for 7-substituted 3,6,9-trihydroxy quinolinone [52]. In addition, H-8′ also showed a correlation to C-10′ (see Figure 4b), which could only be observed with 7-substitution (c.f. from H7′ a 4 JCH coupling would not be observable). If the uranidine moiety was attached at C-8′; 3 JCH couplings should be observed from H-7′ to C-9′ and C-5′. The site of attachment at the imidazole was clearly indicated by strong correlations between C-24 and H-19 and H-8′ (Figure 4). Further evidence for the presence of unit D arose from 1 H-1 H coupling between H-2′ (δH 7.64, d, J = 5. 8 Hz) and H-1′ (δH 11.76, d, J = 5.8 Hz).  The chemical shifts for subunits A, C and D all aligned very well with those described for ceratinadin B [9], a compound similar to the one described here. However, ceratinadin B contains a histamine unit instead of homohistamine (subunit B) and thus a lower molecular weight. Due to the similarity of these two compounds, and with the other ceratinadins, we named this compound ceratinadin D. Finally, the specific rotation of 1 is very similar to that of ceratinadin B strongly suggesting the same absolute stereochemistry (1R, 6S).
These data suggested the compound was aplysamine 4 (9) [10]. However, comparison of the 1 H-NMR spectrum reported for aplysamine 4 differed to our compound (Table 3) in the following ways: In aplysamine 4, the aromatic protons from ring A and B are all at δH ~7.44. We found two different environments for the aromatic ring protons, where H-2 were at δH 7.44 but H-20 were at 7.55. This could be attributed to the use of different solvents but what was also different was the chemical shift of the ethyl and propyl-chain carbons. In aplysamine 4 the propyl group terminated with an ammonium ion (C-20 = 39.0 ppm) whereas in our compound, this carbon was at 36.2 ppm. Similarly the carbon shifts for the ethyl group did not match those of aplysamine 4.  (Table 3) enabled the construction of three substructures ( Figure 5). Inspection of 1 H, 13 C, and 1 H-1 H COSY NMR spectra suggested that the following proton signals belonged to substructure A: δH 12.02 (oxime), 7.44 (sp 2 methine), 3.76 (sp 3 methine) and 3.75 (OMe). HMBC correlations from H-2 to C-2, C-3, H-6 to C-4, and H-7a/b to C-8 and C-11 provided good evidence for the oxime-tyrosine unit A. The connection between substructures A and B was obtained by HMBC correlations from H-7a/b to carbonyl C-11 and from the amide proton H-13 to oxime carbon C-8. C-8 and C-11 could be easily distinguished by their typical chemical shifts as reported in the literature [51]. The geometry of the oxime was determined as E from the up-field 13 C chemical shift of C-7 (δC 27.9), c.f. δC 35.7 for
The amide proton (δH 8.12) showed coupling to a methylene (δH 3.38) assigned to H-14. Further COSY correlations to δH 1.96, and then to δH 3.88 supported the assumption that substructure B was an amide unit connected to a propyl chain, attached to an oxygen (H-16 3.88 ppm; C-16 71.3 ppm from HSQC), which could only be explained by attachment to unit C via an ether linkage. Indeed, a weak correlation between C-18 (151.3 ppm) and H-16 was observed in the HMBC spectrum ( Figure 6) and provided strong evidence for the assigned structure.     (H-20) showed HMBC correlations to sp 2 quaternary carbons (C-18, C-19) and a ROESY correlation to a methylene at δH 2.81/δC 31.5, consistent with attachment of an alkyl group directly to the aromatic ring, in support of substructure C. Direct coupling of δH 2.81 to δH 3.05, and of δH 3.05 to the primary amine protons at δH 7.80 were in good accordance to literature values [51] for a tyramine unit. Aplysamine 8 is thus an isomer of aplysamine 4 and considering the similarity in structure to aplysamines 1-7, no new name is required for this compound.
Putting the pieces together, we conclude that this is a new brominated tyrosine derivative (2) and, due to the similarities with previously described aplysamines 1-7, shall be named aplysamine 8.
Compounds 1-8 were bioassayed against E. coli and S. aureus (MTT microdilution assay; Supplementary Information). Only aplysamine 8 (2) was found to have any notable activity (MIC 125 μg/mL against E. coli and 31 μg/mL against S. aureus) c.f. ampicillin 1.1 μg/mL against S. aureus. Hexadellin (6), aplysamine 2 (7) and 16-debromoaplysamine 4 (8) had mild activity against S. aureus (125-250 μg/mL). While the MTT and disc diffusion assays are not directly comparable, the high antibacterial activity observed for the crude extract was not reflected in the bioassay of individual metabolites. This may be due to a cumulative effect of many weak antibiotics, synergistic or contingent effects as has previously been noted for other natural products [53].
Although opisthobranchs of the genus Tylodina are found in distant regions, they are exclusively associated with sponges of the order Verongida (in particular the family Aplysinellidae) [39]. Tylodina corticalis typically feed on the genus Pseudoceratina [54]. It has also been shown, that Tylodina spp accumulate the secondary metabolites (bromotyrosine alkaloids) from their prey, sequestered and utilised as feeding deterrents for themselves and their eggs [38,39,55,56].
Returning to the apparent lack of similarity between the metabolites positively identified from the sponge and those tentatively identified from T. corticalis, firstly it should be noted that the major difference is that many compounds found at high concentrations in the sponge were not found in the mollusc. This is, however, typically what is observed with other spongiverous molluscs such as nudibranchs, for example [25]. Next, four of the eight major compounds from the sponge were detected in the mollusc, albeit at a lower level and two of the minor/trace metabolites from the sponge were also detected in the mollusc at much higher relative concentrations. The notable difference is the high concentrations of purealidin T and purpuramine J, which were not observed in the sponge extract at all despite a careful search. However, purealidin T is the N-oxide of purealidin Q and purpuramine J is the N-oxide of aplysamine 2, a major alkaloid from the sponge. Taken together these observations suggest that the sponge metabolites are certainly accumulated on a selective basis (purealidin P/Q) but also that the mollusc might be modifying the sponge metabolites (purealidin T, purpuramine J) to N-oxides. Many cases of modification of sponge metabolites by predatory molluscs have previously been reported [57]. Certainly, the major metabolites isolated from T. corticalis are N-oxides closely related to compounds found in the sponge (Figure 7).

Animal Material
The sponge (654 g) was collected by hand at Jervis Bay, NSW, Australia (S 35°08′07″, E 150°43′33″, at 1-4 m depth) under New South Wales Department of Primary Industries collecting permitNo. P08/0006-1.1. The samples were frozen upon collection and stored at −20 °C. It was identified as Pseudoceratina purpurea Carter (phylum Porifera, class Demospongiae, order Verongida, family Aplysinellidae). A voucher specimen (JB-S07) was prepared according to Hooper [58] and deposited at the Department of Chemistry and Biomolecular Sciences, Macquarie University. The opisthobranch Tylodina corticalis (Tate) was found feeding on the sponge Pseudoceratina purpurea. One individual was separated from the sponge, frozen at −20 °C in the field and stored at −80 °C in the laboratory.

Bioassays
Detailed descriptions of the bioassay-guided fractionation, as well as protocols for the disc diffusion and MTT assay can be found in the Supplementary Information.