Sampling was performed at the Bunaken National Park in August 2015, located near Manado, North Sulawesi, Indonesia. In total, 550 samples for phylogenetic, chemical and microbiological studies were obtained under the framework of the “Indobio” Biodiversity and Health project from different sites around the island by snorkeling and scuba diving (Figure 1). Further information about the samples will be published elsewhere. The samples were either directly processed, or cooled until further processing in the lab.

Preliminary identification of slugs and egg masses was performed based on Gosliner et al. (2008, 2015) (Figure 2), identification of sponges based on Atlas of living Australia¹. For barcoding, a small portion of the sampled slugs, egg masses and sponges was stored in 96% EtOH and the DNA was isolated using QIAgen DNeasy Blood and Tissue-Kit (Qiagen, Hilden, Germany). Partial 16S rDNA sequences and partial CO1 gene sequences were amplified using 16Sar-5′ and 16Sbr-3′ primers (Palumbi, 1996) and LCO1490-JJ and HCO2198-JJ primers (Astrin and Stüben, 2008) and sent to Macrogen Inc. (Seoul, Korea). The specimens’ identities were determined by subsequent blastn analysis of the obtained sequences.

Prior to the isolation of bacteria, the surface of the samples (in the following the identification code of each sample is used; further information can be found via the internet portal of Diversity Workbench within the module DiversityCollection². Chan15Bu-2 (Chromodoris annae), Chdi15Bu-3, Chdi15Bu-55 and Chdi15Bu-38 (C. dianae), Chsp3015Bu-4 (Chromodoris sp. 30), Chwi15Bu-2 (C. willani), Chan15Bu-11E (C. annae egg mass), Glst15Bu-1 (Doriprismatica stellata), Glst15Bu-1E (D. stellata egg mass), Glst15Bu-1P (D. stellata sponge), Hesa15Bu-1 and Hesa15Bu-2 (Hexabranchus sanguineus egg mass) and Phpu15Bu-1 (Phyllidiella cf pustulosa) were washed with sterile deionized H₂O, disinfected with 70% ethanol and washed again with sterile deionized H₂O (Milli-Q, Merck, Germany) to remove contaminations from sampling and documentation. The samples were frozen in liquid nitrogen and homogenized using a Potter S Homogenizer (Sartorius, Germany). The homogenisate was brought onto Marine Broth agar plates (37.9 g/L Marine Broth (Difco), 15 g/L agar) or ISP2 agar plates containing 2% NaCl (4 g/L yeast extract, 10 g/L malt extract, 4 g/L glucose, 20 g/L NaCl, 15 g/L agar) using an inoculation loop and the 13 streaks technique to obtain axenic colonies. Axenic colonies were transferred to new plates. For cryo-conservation, axenic cultures were grown in the respective liquid medium, mixed 50:50 with sterile glycerol and stored at -80°C.

Bacteria were grown in 30 mL of the respective medium. Therefrom, 25 mL were transferred to a 50 mL reaction tube, 25 mL ethyl acetate were added and the tube was shaken thoroughly. The organic phase was carefully removed with a pipette and evaporated. The extracts were dissolved to 1 mg/mL in methanol for HR-LCMS and to 10 mg/mL in DMSO for antibiotic testing.

For antibacterial assays, test strains were grown in 200 μL cultures in flat bottom 96 well plates at room temperature. The initial OD₆₀₀ was set to 0.1 and the extracts were added to yield a final concentration of 0.5 μg/μL per culture for the first well and subsequently diluted 1:1 in the following wells to the lowest concentration of 4 ng/μL. 0.5 μg/μL ampicillin served as positive control, pure DMSO as negative vehicle control. The tests were evaluated when the negative control reached an OD₆₀₀ of 1.0. Extracts were tested against Arthrobacter psychrolactophilus and Escherichia coli XL1Blue. For the preparation of Staphylococcus aureus COL and E. coli O-19592 assay plates, cation adjusted Müller Hinton agar was autoclaved and allowed to cool down to 50–55°C. 10⁷ cells/ml of the test strains were seeded into the agar and after solidification, 3 μl of the ethyl acetate extracts were spotted onto the plates. Ampicillin (0.1 μg/ml) and ethyl acetate served as positive and negative controls, respectively. Plates were incubated at 37°C for 18 h and checked for inhibition zones.

Genomic DNA of the different bacterial strains was isolated using the GenElute Bacterial Genomic DNA Kit (Sigma) according to the manufacturer protocol. The genomic DNA was subjected to several PCR screening rounds using degenerated primers (see Supplementary Table 1). Primers were based on literature, or designed based on conserved regions in ketosynthase (KS) domains of PKS and adenylation domains (A-domains) of NRPS. Specific PCR for serine activating A-domains was performed as nested PCR, i.e., the PCR products of a general A-domain PCR with degenerated primers were gel purified and used in a subsequent PCR with AnSerin/JS002 as primer pair. The amplified products of approximately 700bp (KS domains), 700bp (A3–A7 region of A-domains), 1000bp (A2–A8 region of A-domains) and 450bp (serine specific nested PCR) were gel purified and cloned into pGEM-T vector (Promega, Madison, WI, United States), transformed into chemical competent E. coli Alpha Silver Select Efficiency (Bioline, Luckenwalde, Germany) and sequenced. Sequences were analyzed using blastx and blastn and closest protein sequences of were analyzed for amino acid specificity using antiSMASH 3.0 (Weber et al., 2015) and submitted to Genbank. The accession numbers for this project are KY671135 – KY671183 (16S rDNA), KY671132 – KY671134 (KS), and KY671184 – KY671199 (A domains).

The bacterial species were determined by 16S rDNA sequencing. Therefore, genomic DNA was isolated from 2 mL of liquid culture using the GenElute Bacterial Genomic DNA Kit (Sigma) and the 16S rDNA region was amplified by PCR using rD1 and fD1 primers (Weisburg et al., 1991). The resulting amplificate was cloned into pGEM-T vector (Promega, Madison, WI, United States), introduced into E. coli Alpha Silver Select Efficiency cells (Bioline, Luckenwalde, Germany). The plasmids were isolated using the PureYield Miniprep kit (Promega, Madison, WI, United States) and the insert was sequenced from both sides using T7 and SP6 primers (GATC, Konstanz, Germany). Vector fragments in the sequencing result were removed using VecScreen-Blast and the full 16S rDNA was assembled. BLASTn search of the 16S rDNA sequences revealed the closest relatives, thereby enabling determination of the species, due to the fact that identity to known type strains was in the range of 98–100%. The phylogenetic tree of all obtained bacterial isolates was created by aligning the 16S rDNA sequences of the respective type strains using MAFFT (Katoh et al., 2002) and constructing the tree using raxML (Stamatakis, 2014).

Preliminary identification of the slugs and egg masses were confirmed by barcoding. Metadata of the collected macroorganisms will become available via the internet portal of Diversity Workbench within the module DiversityCollection³. 49 bacterial strains were isolated as axenic culture starting from 12 different samples. The highest proportion, of strains (20% each) was isolated from the egg masses Hesa15Bu_1 and Hesa15Bu_2 of H. sanguineus whereby only 2% (equates to 1 strain) were isolated from the processed sponge sample. Most of the strains were recovered using Marine Broth standard medium, while 10 strains were recovered by using ISP2 medium optimized for actinobacteria (see Materials and Methods). For a full table of strains and their respective source see Supplementary Table 2.

Axenic bacterial strains were cultivated in liquid medium and ethyl acetate crude extracts were prepared. These extracts were first tested against the Gram-positive test strain Arthrobacter psychrolactophilus and the Gram-negative test strain E. coli XL1Blue. 35 strains were tested positive in this assay. However, to prioritize the samples, the 10 most active extracts were selected for further testing against relevant pathogens, i.e., S. aureus COL (MRSA) and E. coli O-19592 (EHEC). In this assay, three extracts showed anti-MRSA activity and one extract showed anti-EHEC activity (Figure 3). It can be seen that especially isolates of Pseudoalteromonas strains showed considerable activity. However, the strain showing activity against the Gram-negative EHEC strain was a Marinomonas species.

For all isolated axenic bacteria the 16S rDNA gene was amplified using rD1 and fD1 primers, designed for the specific and long range amplification of eubacterial 16S rDNA. All isolated strains belong to known genera and show high homology (98–100% identity) to sequences stored in the database. The closest homolog type strain of each isolated strain was included in an alignment to obtain a phylogenetic tree (Figure 3).

From the obtained isolates, 35 out of 49 strains (equal to 71%) showed antibacterial activity. All of the 10 isolated Pseudoalteromonas strains showed activity, and 7 thereof revealed strong activity against the Gram-positive test strain. Further, two strains also inhibited methicillin resistant S. aureus (MRSA). Another bacterial group with a high proportion of strains showing antibacterial effects was the Vibrio group. Therein, 14 out of 16 isolated strains showed at least slight activity, 5 strains exhibited strong activity against the Gram-positive test strain, and 1 strain showed anti-MRSA activity.

One isolate even showed activity against enterohemorrhagic E. coli (EHEC). This strain was the only isolate from the Marinomonas clade.

All 49 strains were tested for the presence of DNA sequences known to be associated with biosynthetic gene clusters (BGCs) contributing to the production of specialized metabolites, i.e., degenerated primers were used to screen the genomic DNA for the presence of NRPS sequences. Primers were adapted from literature to target conserved regions of A-domains. Despite many trials, the number of sequences obtained by this approach was very low, i.e., 10 A-domain sequences (6 different from Gordonia terrae Bu15_45, 2 different from Pseudoalteromonas rubra Bu15_31, 1 from Pseudoalteromonas sp. Bu15_09 and 1 from Gordonia sp. Bu15_44). Since oxazole natural compounds are amongst valuable antibiotics, such as streptogramin A, a two-step nested PCR approach for serine specific A-domains was projected by designing a primer between the conserved regions A3 and A7 with specificity to serine incorporating A-domains (AnSerin see Supplementary Table 1). This enabled us to obtain further 6 sequences (1 from Vibrio coralliilyticus Bu15_04, 3 different ones from Pseudoalteromonas sp. Bu15_09, 1 from Pseudoalteromonas sp. Bu15_26 and 1 from Bacillus thuringiensis Bu15_29). All except one proved to be partial A-domain sequences with first blastx hits of A-domains predicted to incorporate serine.

Gordonia sp. Bu15_44, P. rubra Bu15_31, V. coralliilyticus Bu15_01 and Pseudoalteromonas sp. Bu15_09 were tested for the presence of DNA sequences associated with PKS BGCs, i.e., degenerated primers (see Supplementary Table 1) amplifying partial KS fragments were used. Only Gordonia sp. Bu15_44 yielded 3 distinct KS fragments.