Molecular Network and Culture Media Variation Reveal a Complex Metabolic Profile in Pantoea cf. eucrina D2 Associated with an Acidified Marine Sponge

The Gram-negative Pantoea eucrina D2 was isolated from the marine sponge Chondrosia reniformis. Sponges were collected in a shallow volcanic vents system in Ischia island (South Italy), influenced by CO2 emissions and lowered pH. The chemical diversity of the secondary metabolites produced by this strain, under different culture conditions, was explored by a combined approach including molecular networking, pure compound isolation and NMR spectroscopy. The metabolome of Pantoea cf. eucrina D2 yielded a very complex molecular network, allowing the annotation of several metabolites, among them two biosurfactant clusters: lipoamino acids and surfactins. The production of each class of metabolites was highly dependent on the culture conditions, in particular, the production of unusual surfactins derivatives was reported for the first time from this genus; interestingly the production of these metabolites only arises by utilizing inorganic nitrogen as a sole nitrogen source. Major components of the extract obtained under standard medium culture conditions were isolated and identified as N-lipoamino acids by a combination of 1D and 2D NMR spectroscopy and HRESI-MS analysis. Assessment of the antimicrobial activity of the pure compounds towards some human pathogens, indicated a moderate activity of leucine containing N-lipoamino acids towards Staphylococcus aureus, Staphylococcus epidermidis and a clinical isolate of the emerging food pathogen Listeria monocytogenes.


Introduction
The Gram-negative genus Pantoea was only recently established as new genus within the family of Enterobacteriaceae. To date, it comprises about 20 species (https://lpsn.dsmz.de/genus/pantoea) isolated from both the host and non-host environments [1]. The high adaptability to different habitats was also accomplished by a remarkable phenotypic diversity that ranges from the pathogeny of some isolates toward plants and humans-although the actual involvement of Pantoea strains in the insurgence of diseases in both plants and humans is a matter of controversy-to high beneficial biotechnological and therapeutic potential. Pseudomonas sp.
The isolated bacteria were identified through 16S rRNA Sanger sequencing, and sequences were submitted on BLASTn (Nucleotide Basic Local Alignment Search Tool) against GenBank database for species annotation [18]. A phylogenetic tree was built including the strain D2 along with the closest matches from GenBank using MEGA X (Figure 1). These results suggest that D2 is closely related, if not conspecific, to Pantoea eucrina, a poorly studied member of Pantoea genus. This genus was established 30 years ago in the Enterobacteriaceae family. It counts only for 20 isolated species [1, [19][20][21], and certain strains have been found to produce antimicrobial metabolites, many of which peptides [4,5,22]. In addition, a new marine strain of Pantoea was recently found to produce high amounts of an exopolysaccharide [23]. This is the first record of this genus associated with a sponge, in addition, its metabolome has never been deeply investigated.

Isolation and Identification of Bacterial Strains from the Sponge Chondrosia reniformis Living at Naturally Lowered pH Conditions
Chondrosia reniformis, a high microbial abundance (HMA) sponge [15] widespread in the Mediterranean Sea and with a rich microbiome [16], was collected in the marine area around Castello Aragonese, Ischia Island. This species is common in around the acidified area formed by local volcanic CO2 vents [17], suggesting a long term adaptation to OA conditions. The microbiome of this sponge was hence exploited as a reservoir for the isolation and culture of bacterial strains with interesting metabolite profiles. Since the sponges were collected in an acidified environment, bacterial isolation was performed in a decreasing pH gradient. Sponge pieces were brought in a sterile manner to the lab, submersed in 0.22 μm filtered seawater and plated on MB agar plates at pH: 5, 6 and 7 for the microbial isolation. No bacteria grew at pH 5, also considering agar plates frailty at that pH, while most bacteria developed on pH 6 agar plates, resulting in nine morphologically different strains, which were all identified through the 16S rRNA sequencing (Table 1). Table 1. Genera of the bacterial isolates at pH 6 based on their 16S sequences.
The isolated bacteria were identified through 16S rRNA Sanger sequencing, and sequences were submitted on BLASTn (Nucleotide Basic Local Alignment Search Tool) against GenBank database for species annotation [18]. A phylogenetic tree was built including the strain D2 along with the closest matches from GenBank using MEGA X (Figure 1). These results suggest that D2 is closely related, if not conspecific, to Pantoea eucrina, a poorly studied member of Pantoea genus. This genus was established ~30 years ago in the Enterobacteriaceae family. It counts only for 20 isolated species [1, [19][20][21], and certain strains have been found to produce antimicrobial metabolites, many of which peptides [4,5,22]. In addition, a new marine strain of Pantoea was recently found to produce high amounts of an exopolysaccharide [23]. This is the first record of this genus associated with a sponge, in addition, its metabolome has never been deeply investigated.

MS/MS-Based Molecular Networking Analysis of Pantoea eucrina D2 Metabolome Grown under Different Culture Conditions
The principle of "one strain many compounds" (OSMAC) approach is based on the principle that several microbes can potentially produce many more metabolites than they do under determined conditions. A simple and effective way to influence bacterial metabolisms is changing growth parameters such as nutrients, temperature, salinity, aeration, in this way activating or upregulating metabolic pathways which usually silenced [24]. Herein metabolites produced by the sponge-associated P. cf. eucrina D2 in different culture media were explored by modifying the sources of inorganic nitrogen and comparing them with the metabolic profile obtained in the optimal and complex TSB medium. The metabolome was investigated by analyzing the crude extracts through HPLC-MS/MS data dependent analysis. The obtained MS raw data were converted in mzXML format, processed by MZmine [25] in order to remove noise, to filter the isotopes and to align the data. Then, the output files were uploaded to Global Natural Products Social molecular networking (GNPS) [14] and used to build a feature-based molecular network [26], finally the data were exported and visualized on Cytoscape [27].
In this way, a complex web of nodes grouped in several clusters was obtained ( Figure 2). Each node represents an ion found in at least one of the growth cultures, different colors are used to define their presence in different growth media, the edges size among different nodes are directly dependent on their fragmentation spectra similarity (cosine score), while the node size is directly proportional to the precursor ion intensity (sum precursor intensity). This visualization provides a scheme of the different metabolites production in the specific conditions ( Figure 2). The principle of "one strain many compounds" (OSMAC) approach is based on the principle that several microbes can potentially produce many more metabolites than they do under determined conditions. A simple and effective way to influence bacterial metabolisms is changing growth parameters such as nutrients, temperature, salinity, aeration, in this way activating or upregulating metabolic pathways which usually silenced [24]. Herein metabolites produced by the spongeassociated P. cf. eucrina D2 in different culture media were explored by modifying the sources of inorganic nitrogen and comparing them with the metabolic profile obtained in the optimal and complex TSB medium. The metabolome was investigated by analyzing the crude extracts through HPLC-MS/MS data dependent analysis. The obtained MS raw data were converted in mzXML format, processed by MZmine [25] in order to remove noise, to filter the isotopes and to align the data. Then, the output files were uploaded to Global Natural Products Social molecular networking (GNPS) [14] and used to build a feature-based molecular network [26], finally the data were exported and visualized on Cytoscape [27].
In this way, a complex web of nodes grouped in several clusters was obtained ( Figure 2). Each node represents an ion found in at least one of the growth cultures, different colors are used to define their presence in different growth media, the edges size among different nodes are directly dependent on their fragmentation spectra similarity (cosine score), while the node size is directly proportional to the precursor ion intensity (sum precursor intensity). This visualization provides a scheme of the different metabolites production in the specific conditions ( Figure 2).
Analysis of GNPS unique library hits and analog library hits evidenced the presence of three class of metabolites: diketopiperazines, surfactins and lipoamino acids. Herein we focused our attention on two latter classes.  Analysis of GNPS unique library hits and analog library hits evidenced the presence of three class of metabolites: diketopiperazines, surfactins and lipoamino acids. Herein we focused our attention on two latter classes.

Surfactins Molecular Cluster
The analysis of the [M + H] + network in Figure 3 matched on GNPS library for surfactin C (1036 m/z) and through the clusterization suggested the occurrence of related species with m/z at 1050.7041, 1064.7197, 1078.7348, 1092.7509 in addition, another pseudomolecular ion with m/z of 1106.7681 compatible with a surfactin derivative was found in the full-MS spectrum (1000-1200 Da) ( Figure S1) by manual HR data curation.

Surfactins Molecular Cluster
The analysis of the [M + H] + network in Figure 3 matched on GNPS library for surfactin C (1036 m/z) and through the clusterization suggested the occurrence of related species with m/z at 1050.7041, 1064.7197, 1078.7348, 1092.7509 in addition, another pseudomolecular ion with m/z of 1106.7681 compatible with a surfactin derivative was found in the full-MS spectrum (1000-1200 Da) ( Figure S1) by manual HR data curation.
Surfactins are a family of lipopeptides produced by strains of the genus Bacillus. They feature a cycloheptadepsipeptide core with a terminal β-hydroxy fatty acid (β-OH FA). Culture samples of different Bacillus spp. were found to produce complex mixtures of surfactin congeners, which differ either in the length of the fatty acid chain or in the amino acid sequence and composition of the peptide moiety. The extract obtained in MSM C media, which contains all the detected surfactins, was further analyzed to deepen the sequence of surfactins analogs and to reveal the potential presence of isomers.
The fragmentation pattern of surfactins has been extensively investigated. The initial cleavage of the protonated ester bond was followed by sequential loss of amino acid residues giving rise to b+ and y+ series of fragment ions useful to assign the amino acid sequence, except for the discrimination of isomeric Leu and Ile and OMet Asp and Glu.
Although the difference of 129 between b6 and b5 ions could be ascribable both to a Glu residue or to the presence of an OMet-Asp, we proposed here the C16 β-hydroxyl fatty acid chain-Glu1-Val2- Surfactins are a family of lipopeptides produced by strains of the genus Bacillus. They feature a cycloheptadepsipeptide core with a terminal β-hydroxy fatty acid (β-OH FA). Culture samples of different Bacillus spp. were found to produce complex mixtures of surfactin congeners, which differ either in the length of the fatty acid chain or in the amino acid sequence and composition of the peptide moiety.
The extract obtained in MSM C media, which contains all the detected surfactins, was further analyzed to deepen the sequence of surfactins analogs and to reveal the potential presence of isomers.
The fragmentation pattern of surfactins has been extensively investigated. The initial cleavage of the protonated ester bond was followed by sequential loss of amino acid residues giving rise to b+ and y+ series of fragment ions useful to assign the amino acid sequence, except for the discrimination of isomeric Leu and Ile and OMet Asp and Glu.
Finally, two homologs with Glu1-Leu/Ile 2-Leu/Ile3-Leu/Ile4-OMeAsp or Glu5-Leu/Ile6-Leu/Ile7 as aminoacidic sequence and a C17 and C18 β-hydroxyl fatty acid chain were found to be ascribable to the ions with m/z at 1092.7 and 1106.7 (Figure 5d,e) [29,30]. Although the difference of 129 between b6 and b5 ions could be ascribable both to a Glu residue or to the presence of an OMet-Asp, we proposed here the C16 β-hydroxyl fatty acid chain-Glu1-Val2-Leu/Ile3-Leu/Ile4-Val5-OMeAsp6-Val7 structure based on the observation of a ion fragment a 905, likely arising from b6 ion by MeOH loss. To the best of our knowledge, this surfactin isoform is described in this study for the first time.
A peculiar feature of the surfactin analogs produced by P. cf. eucrina D2 seems to be the presence of the four consecutive leucine motifs found in certain derivatives, and the reversed position of Leu5 and Asp6 compared to the canonical surfactin C (Table S1). The production of surfactins with C17/C18 β-hydroxyl fatty acid chain appear a further unusual feature, with few literature precedents [30]. A peculiar feature of the surfactin analogs produced by P. cf. eucrina D2 seems to be the presence of the four consecutive leucine motifs found in certain derivatives, and the reversed position of Leu5 and Asp6 compared to the canonical surfactin C (Table S1). The production of surfactins with C17/C18 β-hydroxyl fatty acid chain appear a further unusual feature, with few literature precedents [30].
To the best of our knowledge, this is the first report of the production of surfactins from a Pantoea strain. This is not surprisingly since the production of surfactins was not observed using conventional growth conditions (i.e., TBS medium) whereas it was triggered in the OSMAC media using inorganic sources of nitrogen NH4Cl or NaNO3. Similarly, the use of inorganic nitrogen sources demonstrated to be an effective tool to simulate the production of the antimicrobial peptide SBR-22 in the marine Streptomyces psammoticus BT-408 strain [31] and of the pigment pyocyanin in Pseudomonas sp. MCCB-103 [32]. As concerning the carbon sources, glycerol was found to enhance the production, when compared to glucose. The use of glycerol as cost-effective substrate for biosurfactants production by wild type and engineered strains of Pseudomonas has been extensively explored and rationalized [33], whereas the use of glycerol in surfactin production has been only sporadically explored [34]. Unfortunately, surfactins derivatives were detected as minor components co-eluting with major lipoamino acids, therefore we were unable to perform an isolation and chemical characterization work.  To the best of our knowledge, this is the first report of the production of surfactins from a Pantoea strain. This is not surprisingly since the production of surfactins was not observed using conventional growth conditions (i.e., TBS medium) whereas it was triggered in the OSMAC media using inorganic sources of nitrogen NH 4 Cl or NaNO 3 . Similarly, the use of inorganic nitrogen sources demonstrated to be an effective tool to simulate the production of the antimicrobial peptide SBR-22 in the marine Streptomyces psammoticus BT-408 strain [31] and of the pigment pyocyanin in Pseudomonas sp. MCCB-103 [32]. As concerning the carbon sources, glycerol was found to enhance the production, when compared to glucose. The use of glycerol as cost-effective substrate for biosurfactants production by wild type and engineered strains of Pseudomonas has been extensively explored and rationalized [33], whereas the use of glycerol in surfactin production has been only sporadically explored [34]. Unfortunately, surfactins derivatives were detected as minor components co-eluting with major lipoamino acids, therefore we were unable to perform an isolation and chemical characterization work. Figure 6 reports the second cluster featuring a huge number of nodes. The node at m/z 404.3149 matched the 2-(14-methylpentadecanoylamino)-3-phenylpropanoic acid, a member of lipoamino acids (also referred as N-acyl amino acids), namely conjugates of one amino acid unit linked via an amide bond to saturated or unsaturated fatty acids. In particular, this compound was recently reported as component of a complex mixture of related lipoamino acids from an entomopathogenic Pantoea sp. strain isolated from an individual insect from the Diaspididae family [24]. This suggested that the members of this cluster could represents lipoamino acid variants differing for the amino acid and/or fatty acid moiety. acids (also referred as N-acyl amino acids), namely conjugates of one amino acid unit linked via an amide bond to saturated or unsaturated fatty acids. In particular, this compound was recently reported as component of a complex mixture of related lipoamino acids from an entomopathogenic Pantoea sp. strain isolated from an individual insect from the Diaspididae family [24]. This suggested that the members of this cluster could represents lipoamino acid variants differing for the amino acid and/or fatty acid moiety. Comprehensive manual dereplication of each node within the cluster was done by determination of the molecular formula, from the analysis of the positive pseudomolecular ion [M + H] + , and from inspection of the fragmentation pattern. The key fragment arising from the cleavage of the amide bond allowed to assign the amino acid portion, and by subtraction, the length of the fatty acid subunit, whereas the presence of double bonds was inferred from the calculated round double bond equivalent (RDBE). Through the above analysis, the annotation of each node within the cluster was done and putative phenylalanine, leucine/isoleucine and valine derivatives were found with fatty acid chains ranging from C10 to C21, some of which containing one or two insaturations or an hydroxyl group, the detailed assignments are summarized in Table S2.

Lipoamino Acids Molecular Cluster
Our results parallel those reported by Tourè et al. [8] in what concerns the qualitative composition of the mixture except for some congeners. Although our Pantoea cf. eucrina D2 was found to produce the phenylalanine conjugates as minor components, whereas in all tested media cultivation conditions the Leu/Ile conjugates are the major components.
Interestingly, we observed that P. cf. eucrina D2 was able to produce lipoamino acids in all tested growing media containing only inorganic nitrogen sources, so showing nitrifying properties. Comprehensive manual dereplication of each node within the cluster was done by determination of the molecular formula, from the analysis of the positive pseudomolecular ion [M + H] + , and from inspection of the fragmentation pattern. The key fragment arising from the cleavage of the amide bond allowed to assign the amino acid portion, and by subtraction, the length of the fatty acid subunit, whereas the presence of double bonds was inferred from the calculated round double bond equivalent (RDBE). Through the above analysis, the annotation of each node within the cluster was done and putative phenylalanine, leucine/isoleucine and valine derivatives were found with fatty acid chains ranging from C10 to C21, some of which containing one or two insaturations or an hydroxyl group, the detailed assignments are summarized in Table S2.
Our results parallel those reported by Tourè et al. [8] in what concerns the qualitative composition of the mixture except for some congeners. Although our Pantoea cf. eucrina D2 was found to produce the phenylalanine conjugates as minor components, whereas in all tested media cultivation conditions the Leu/Ile conjugates are the major components.
Interestingly, we observed that P. cf. eucrina D2 was able to produce lipoamino acids in all tested growing media containing only inorganic nitrogen sources, so showing nitrifying properties.

Isolation and Structure Elucidation of Pure Compounds
In order to solve the Leu/Ile and the configurational ambiguities left by the MS analysis and to assess the antimicrobial activity of the major pure components, a medium scale cultivation (1.8 L) in TBS medium was performed, the crude extracellular extract (1.0 g) was fractionated by reverse Phase MPLC, and the obtained fractions were purified on UPLC equipped with a semipreparative PFP column as described in Material and methods section, affording pure Compounds 1-8 (Figure 7).

Isolation and Structure Elucidation of Pure Compounds
In order to solve the Leu/Ile and the configurational ambiguities left by the MS analysis and to assess the antimicrobial activity of the major pure components, a medium scale cultivation (1.8 L) in TBS medium was performed, the crude extracellular extract (1.0 g) was fractionated by reverse Phase MPLC, and the obtained fractions were purified on UPLC equipped with a semipreparative PFP column as described in Material and methods section, affording pure Compounds 1-8 (Figure 7).
The structures of these compounds were determined by combined MS/MS and 2D NMR analysis and by comparison with literature data. The analysis of 2D COSY, HSQC and HMBC allowed the straightforward assignment of the leucine aminoacyl spin system (Table S5). Particularly diagnostic for the Ile/leu discrimination are the 1 H and 13 C NMR chemical shifts of two γ methyl groups (δH 0.96 and 0.97/δC 21.8 and 23.2).
Structural N-acylamino acid analogs 2-5 were characterized using comparative NMR, HRMS/MS and Marfey's analysis. Compounds 2-4 were already isolated as mixture of enantiomers The analysis of 2D COSY, HSQC and HMBC allowed the straightforward assignment of the leucine aminoacyl spin system (Table S5). Particularly diagnostic for the Ile/leu discrimination are the 1 H and 13 C NMR chemical shifts of two γ methyl groups (δ H 0.96 and 0.97/δ C 21.8 and 23.2).

Isolation and Structure Elucidation of Pure Compounds
In order to solve the Leu/Ile and the configurational ambiguities left by the MS analysis and to assess the antimicrobial activity of the major pure components, a medium scale cultivation (1.8 L) in TBS medium was performed, the crude extracellular extract (1.0 g) was fractionated by reverse Phase MPLC, and the obtained fractions were purified on UPLC equipped with a semipreparative PFP column as described in Material and methods section, affording pure Compounds 1-8 (Figure 7).
The structures of these compounds were determined by combined MS/MS and 2D NMR analysis and by comparison with literature data. The analysis of 2D COSY, HSQC and HMBC allowed the straightforward assignment of the leucine aminoacyl spin system (Table S5). Particularly diagnostic for the Ile/leu discrimination are the 1 H and 13 C NMR chemical shifts of two γ methyl groups (δH 0.96 and 0.97/δC 21.8 and 23.2).
Structural N-acylamino acid analogs 2-5 were characterized using comparative NMR, HRMS/MS and Marfey's analysis. Compounds 2-4 were already isolated as mixture of enantiomers The stereochemistry of the double bond was assigned as Z, based on the 13 C chemical shift of the C8-C11 allylic carbons at δ C 27.4 ppm [35].
Structural N-acylamino acid analogs 2-5 were characterized using comparative NMR, HRMS/MS and Marfey's analysis. Compounds 2-4 were already isolated as mixture of enantiomers and an inseparable mixture together with their isoleucine congeners from Cobetia sp. isolated from the marine hydroid Hydractinia echinata [36].
N-palmitoyl-l-leucine (4) was already isolated by Streptomyces sp. and identified as a late stage inhibitor of m-RNA splicing through a high-throughput screening of 5304 pre-fractions of marine bacterial lysates [37].
Through NMR and MS analysis and by comparison with literature data [8] Compounds 6, 7 and 8 were identified as N-palmitoyl-l-phenylalanine, N-palmitenoyl-l-phenylalanine and N-oleyl-l-phenylalanine, respectively. N-palmitenoyl-l-phenylalanine (7) was already isolated from Pantoea sp., whereas Compounds 6 and 8 were obtained in the same study by synthesis [8]. The configuration of the double bond in the acyl chain of 7 and 8 was assigned as Z on the basis of the 13 C NMR chemical shifts of allylic methylenes, whereas the L-configuration of the phenylalanine residue was assigned by comparison of the experimental positive optical rotation data with those reported in the literature [8].

Minimum Inhibition Concentration (MIC)
The data regarding antimicrobial activity of lipoamino acids are often dated and not widely described, in the specific case, are not reported at all for leucine N-acylamino lipids. Herein the antimicrobial potential of the pure molecules was determined by microdilution method and the MIC value was reported for each pathogen, DMSO at an initial concentration of 2% (v/v) was adopted as negative control and antibiotics at their respective breakpoint concentrations were used as positive controls. The results reported in Table 2 show that this class of molecules is ineffective against the Gram-negative strains (Acinetobacter baumannii 13 and Stenotrophomonas maltophilia ATCC 13637). In addition, the amino acid moiety seems to play an important role in the exertion of antimicrobial properties, in fact, in accordance with a recent report on phenylalanine lipoamino acids [8], Compounds 6-8 are totally ineffective or slightly active towards Staphylococcus aureus and also against Staphylococcus epidermidis, independently from fatty acid chain length. On the other hand, leucine-containing congeners were very active towards Staphylococcus spp., with the best MIC values given by Compounds 1, 3 and 4 with 10 (µg/mL) towards S. aureus and by Compounds 3 and 4 with 10 (µg/mL) towards S. epidermidis. Each experiment was repeated at least three times and the mean value is reported. a Vancomycin was used as positive control for Staphylococcus strains, ampicillin for L. monocytogenes, chloramphenicol for S. maltophilia and gentamicin for A. baumannii.
The activity of these biosurfactants towards foodborne pathogens was observed on certain shorter chain derivatives [38], where N-myristoyl-l-phenylalanine showed a MIC of 34 (µg/mL) towards L. monocytogenes, although longer chain congeners herein isolated seems to be more effective towards the clinical isolated L. monocytogenes strain used in this study. Additionally, it is noticeable that phenylalanine congeners hold a comparable activity with leucine one, with the best activity value obtained for Compounds 3 and 6 (4.0 µg/mL).

General Experimental Procedures
1D and 2D NMR experiments were recorded on Bruker Avance NEO 400 MHz and 700 MHz spectrometers (Bruker, USA) with a RT-DR-BF/1H-5mm-OZ SmartProbe. Chemical shifts were reported in δ (ppm) and were referenced to the residual CDCl 3 as internal standards (δ H = 7.26 and δ C = 77.0 ppm) and CD 3 OD as internal standards (δ H = 3.31 e δ C = 49.0 ppm). All of the recorded signals were in accordance with the proposed structures. Spin multiplicities are given as s (singlet), br s (broad singlet), d (doublet), t (triplet) or m (multiplet).
The LC-HRMS and MS/MS analysis were carried out on an LTQ XL-Orbitrap high-resolution mass spectrometry system (Thermo Scientific) equipped with an Accelera 600 pump HPLC (LCHRMS).
Further fragmentation analysis was carried out on an LTQ XL mass spectrometry system (Thermo Scientific) equipped with a HESI source and connected to an Ultimate 3000 HPLC pump.
The first purification step was run on a User Manual PuriFlash XS 520 Plus equipped with UV detector. Purification of single molecules was performed on an Acquity UPLC H-CLASS connected to a PDA detector (Waters).
The 96-well plates were read on a Biotek ELX800, monitoring the absorbance at 600 nm at room temperature.

Sponge Collection and Bacterial Isolation at Different pH
Specimens of the sponge Chondrosia reniformis were collected by scuba diving at 2-m depth in several spots influenced by CO 2 emissions and low pH close to Castello Aragonese, Ischia, Italy (40 • 43.84 N 13 • 57.08 E). Immediately after the collection some sponge pieces were stored at 15 • C submersed in 0.22 µm filtered seawater and taken to the laboratory for the microbial isolation. Bacterial isolation was performed at different pH values. Sponge pieces were shredded and mixed with 20 mL sterile seawater, then the solution was shaken for 30 min, and finally the bacterial suspension was serially diluted (10 −1 , 10 −2 and 10 −3 in 10 mL) and 100 µL of each dilution was plated on different MB agar plates, previously adjusted with HCL 1 M to the three chosen pH values: 5, 6 and 7.
The petri dishes were incubated at 15 • C for 15 days, which is the same temperature of the seawater registered during the sampling session. After the incubation period, morphologically different colonies were picked by a sterile loop, grown in liquid MB broth and finally pure bacteria cultures were stored at -80 • C in sterile cryovials with glycerol 20% w/w.

16S Sequencing
PCR was carried out in a total volume of 50 µL containing 25 µL of PCR Master Mix 2× (a ready-to-use solution containing TaqPol, buffer, MgCl2 and dNTPs), 0.2 µM of both primers 27F (5 -AGAGTTTGATCCTGGCTCAG-3 ) and 1492R (5 -GGTTACCTTGTTACGACTT-3 ), and 100 ng of DNA. PCR protocol is reported in Table S3. Five microliters of each PCR product were run on 1% agarose gel at 110 V for 45 min to check the quality of DNA and observed under UV light. Then PCR products were purified with the GeneAll kit according to manufacturer's instructions and the obtained amplicons were send to Eurofins Genomics for the sequencing. Finally, the sequences were used as template for a taxonomic annotation via BLASTn tool against GenBank database, using the 16S RNA sequences collection. The contig obtained submitting the forward and the reverse sequences to Prabi CAP3 (http://doua.prabi.fr/software/cap3) was submitted to BLASTn for the affiliation analysis. Evolutionary analyses involved ten nucleotide sequences, our D2 strain along with nine close matching entries according to BLASTn, and were conducted in MEGA X [40]. A phylogenetic tree was inferred using the neighbor-joining method [41]. The evolutionary distances were computed using the Kimura 2-parameter method [42] and are in the units of the number of base substitutions per site. All ambiguous positions were removed for each sequence pair (pairwise deletion option). Bootstrap values were calculated with 1000 resamples.

Bacterial Cultivation, OSMAC Cultures and Extraction Methodologies
To produce the extracts in the complex media TSB, a single CFU of D2 was used to inoculate 3 mL of TSB, in sterile bacteriological tubes. After 24 h of incubation at 20 • C and 210 rpm, the inoculum was employed to inoculate 125 mL of the respective media in a 500-mL flask, at the initial cell concentration of 0.01 OD 600 /mL. The flask was incubated for 5 days at the same conditions.
The OSMAC (one strain many compounds) approach was used to trigger the production of metabolites unexpressed or underexpressed under normal conditions. Different D2 cultures have been set utilizing MSM mod liquid media as base, glucose or glycerol as carbon source, while the inorganic nitrogen salts NH 4 Cl or NaNO 3 were used as sole nitrogen sources at different concentrations. In this case, a different protocol was used to inoculate the strain: a D2 single colony was picked from a pure plate and inoculated in 3 mL of TSB. After 2 days of incubation at 20 • C and 210 rpm, the culture was centrifuged, the medium was discarded, and the cells were resuspended in 1 mL of MSM mod. The absorbance was measured at the spectrophotometer and was inoculated in 125 mL of each different medium in a 500-mL flask at the same conditions previously described.
The medium-scale fermentation was set by inoculating 1.8 L of TSB with the same procedure described above at 20 • C for 5 days to afford 1 g of crude extract.
After the incubation time, all the cultures were centrifuged at 6800× g at 4 • C for 45 min, and the supernatant was extracted with 2 volumes of ethyl acetate, then the organic phase was dried under vacuum at the rotary evaporator to afford the crude extract.

Mass Spectrometry Analysis
The TSB and OSMAC extracts were first subjected to a desalination step by using the Sep-Pak tC18 Plus Short Cartridge 400 mg. Each extract was dissolved in the minimum amount of MeOH and upload on a cartridge, then 3 beds of MQ H 2 O were used to wash the extract and 3 beds of MeOH and MeOH + TFA 0.01% to eluate the metabolites. Then the methanolic extracts were dried under nitrogen flux and subjected to mass spectrometric analysis.
Both low resolution and high resolution data dependent analysis were carried out performing chromatographic separation of samples on a Synergi 2.5 mm

Molecular Networking Building
The MS data were treated with MZmine [25] upon being converted from *.raw extension to *.mzXML extension using the tool MSConvert by ProteoWizard.
Version 2.53 of MZmine was used to process the data and the parameters used in each step are listed in the Table S4, the job was exported as two output data, one quantification table (.csv) and one file containing the MS and MS/MS features (.mgf).
These data were submitted to GNPS website (https://gnps.ucsd.edu) [14] and the network was created with the feature-based molecular networking (FBMN) workflow [26]. The parameters were changed in accordance with the data and the used mass spectrometer, the precursor ion mass tolerance was set to 0.05 Da, the fragment Ion mass tolerance was set to 0.05 Da. The molecular network was created with a maximum of 100 nodes for each cluster, edges were filtered to have cosine score above 0.7 and at least 3 matching peaks, more edges among two nodes were kept only if each of the nodes is, respectively in each other list of the 10 more similar nodes. Analogs for each node were also searched, with a maximum mass difference of 300 Da, here at the same way all the analogs hits had to show a cosine score above 0.7 and at least 3 matching peaks. Finally the network was visualized by Cytoscape [27].

Medium-Scale Cultivation and Isolation of Pure Lipoamino Acids
The medium-scale fermentation was set by inoculating 1.8 L of TSB with the same procedure previously described for the small cultures, at 20 • C for 5 days.
The vial was heated at 50 • C for 1.5 h. The mixture was dried, and the resulting d-FDAA and l-FDAA derivatives were dissolved in MeOH (200 µL) for subsequent analysis. Authentic standards of l-Leu, l-Ile and d-Allo-Ile were treated with l-FDAA and d-FDAA as described above and yielded the l-FDAA and d-FDAA standards. Marfey's derivatives of 3 were analyzed by LC-ESI-MS, and their retention times were compared with those from the authentic standard derivatives. A Luna Omega 3 µm Polar C18 column (100 × 2.1 mm) maintained at 25 • C was eluted at 300 µL/min with 0.1% HCOOH in H 2 O and ACN. The gradient program was as follows: 10% ACN 2 min, 10% → 95% ACN over 10 min, 100% ACN 3 min. Mass spectra were acquired in positive ion detection mode, and the data were analyzed using the Xcalibur suite of programs.

Minimum Inhibition Concentration (MIC) Assessment
The antimicrobial activity of the pure lipoamino acids was determined by microdilution method and MIC values were determined comparing them with appropriate antibiotics, how described by the Clinical and Laboratory Standard Institute (CLSI) [39]. The antimicrobial assay was performed in CAMHB as medium, pure compounds were dissolved in DMSO and were 2-fold serially diluted from in a final volume of 100 µL of CAMHB medium in a 96-well microtiter plate (Sarstedt), obtaining concentration in the range 128-1 µg/mL. DMSO at an initial concentration of 2% (v/v) was adopted as negative control. Each well contained 50 µL of a pure molecule solution at twice the desired final concentration, therefore it was inoculated with 50 µL of bacterial culture grown overnight at 37 • C, resulting in a final inoculum of 4 × 105 CFU/mL in a 100 µL final volume of each well. Then, each plate was incubated for 20 h at 37 • C to allow optimal bacterial growth. The pathogenic strains used in the liquid inhibition assay are: S. aureus 6538P [43], L. monocyogenes MB677 [44], S. epidermidis ATCC 35,984 [45], A. baumannii 13 [46] and, S. maltophilia [47].

Conclusions
In this study, MN and OSMAC integrated strategies were used to study the secondary metabolism of the endophytic strain Pantoea cf. eucrina D2 isolated from the sponge Chondrosia reniformis adapted to OA conditions. The study evidenced the production of otherwise silent surfactin biosurfactants induced by using inorganic nitrogen as sole nitrogen source.
The molecular networking-based approach confirmed the production of lipoamino acids as chemotaxonomic markers of members of Pantoea genus [8] and guided towards the isolation of six new structural variants of this growing family of bacterial secondary metabolites, together with two known derivatives.
All isolated compounds exhibited antimicrobial activity. It is currently suggested that lipoamino acids exert a protective role in the endosymbiotic interaction with the host [8] and that the observed species-specific tight regulation of the composition of lipoamino acid mixtures [48,49] may be the result of the epigenetic regulation of their biosynthesis in response to specific environmental pressures. The leucine-based N-acylamino acids were found to possess a good antimicrobial potential towards Gram-positive strains, in particular towards a clinical isolate of the foodborne pathogen L. monocytogenes. The natural presence of Pantoea cf. eucrina D2 in Chondrosia reniformis may likely afford the sponge with advantages, in what regards defense from other undesirable invasive microbes by out-competition and antimicrobial properties. However, also, the production of exopolysaccharides and surfactant-type chemicals, may create protective envelopes for beneficial endosymbiotic microorganisms (e.g., N-fixing microbes) sensitive to low pH scenarios [1,23].
Further studies will be devoted to the optimization of the production of both biosurfactant families and to the exploration of the biotechnological potential of Pantoea eucrina D2 in bioremediation, biocontrol and food preservation fields.