Spotlight on Antimicrobial Metabolites from the Marine Bacteria Pseudoalteromonas: Chemodiversity and Ecological Significance

This review is dedicated to the antimicrobial metabolite-producing Pseudoalteromonas strains. The genus Pseudoalteromonas hosts 41 species, among which 16 are antimicrobial metabolite producers. To date, a total of 69 antimicrobial compounds belonging to 18 different families have been documented. They are classified into alkaloids, polyketides, and peptides. Finally as Pseudoalteromonas strains are frequently associated with macroorganisms, we can discuss the ecological significance of antimicrobial Pseudoalteromonas as part of the resident microbiota.


Introduction
Last October, we celebrated the 20th anniversary of the genus Pseudoalteromonas having been split from Alteromonas [1]. The genus Pseudoalteromonas includes Gram-negative, heterotrophic, and aerobic bacteria with a polar flagellum and has a GC content comprised between 38% and 50% [2]. It belongs to the order Alteromonadales in the γ-Proteobacteria class. The Pseudoalteromonas strains require a seawater base for growth and are therefore true marine bacteria. Nowadays, 41 species are assigned to this genus and over 3772 Pseudoalteromonas strains are currently listed in the NCBI taxonomy browser. Allowing for rare exceptions, Pseudoalteromonas strains are associated with healthy animals or algae.
To date, few strains are known as pathogenic or opportunistic. Flavobacterium piscicida, reclassified as P. piscicida by Gauthier et al., is involved in flavobacteriosis in farm fish [3], while P. agarivorans NW4327 was recently reported as pathogenic for the sponge Rhopaloeides odorabile [4].
This bacterial genus is of great interest to the scientific community because of (i) its prolific metabolite-producing capacity and (ii) its usual association with macroorganisms [5], leading to a suspected and sometimes documented ecological significance. These two properties may, in fact, be interconnected. According to the hologenome theory [6][7][8][9][10], the holobiont is composed of the host and its associated microbial communities, named microbiota. Such a superorganism gains genetic plasticity and flexibility and therefore appears better equipped to face and adapt to environmental variations. The microbiota is supposed to play a critical role in holobiont homeostasy through its metabolic activities. Moreover, the microbial shielding of the microbiota may defend the holobiont from pathogen settlement and therefore participate in the host protection. The abundance of antimicrobial

Polyketides
Polyketides constitute a rich and diverse family of natural compounds from prokaryotes and eukaryotes. By definition, all polyketides are synthesized by condensation of acetyl (also referred to as ketides) or malonyl units via specific enzymes called PolyKetide Synthases (PKS). These enzymes catalyze the condensation of activated acyl derivatives (acyl-CoA) on an existing ketide linked to the enzyme via a thioester bond [56]. Several categories of PKS are described within bacteria, depending on (i) the linear/iterative functioning of the enzyme; and (ii) the way acyl-coA is incorporated (involving or not an Acyl Carrier Protein ACP) [57]. Hence, to date, three PKS types have been recognized. Type I PKS are sequential multifunctional enzymes while type II PKS are iterative and multienzyme complexes. In contrast to type I and II PKS, type III PKS are iterative but do not involve ACP. Until the late 1990s, type III was thought to be plant-specific but recent developments have highlighted that they are also found in microorganisms [58]. PKS have been shown to yield many bioactive molecules, especially antibiotics such as erythromycin [59] and tetracycline [60]. Within the genus Pseudoalteromonas, only three different species produce nine groups of polyketides exhibiting antimicrobial activity (Table 3). Indeed, polyketides are at the origin of numerous antimicrobial phenols and, in association with NRPS and Fatty Acid Synthase (FAS) systems, of the well-known thiomarinols produced by P. luteoviolacea [61] (Table 3).

Non Ribosomally Synthesized Peptides
Non Ribosomal Peptides Synthetases (NRPS) represent a wide group of multimodular enzymes that polymerize amino acids as well as fatty acids, and α-hydroxy acids [68]. NRPS can also incorporate unusual amino acids (non-proteinogenic) and hence generate a great chemodiversity. Basically, NRPS are divided into modules, each one being responsible for binding one specific amino acid to the peptidic chain. To do so, each module consists of at least three domains. The Adenylation domain (A) activates the amino acid. Then, the Peptidyl Carrier Protein (PCP, also referred to as thiolation domain T) transfers it to the Condensation domain, which incorporates the new amino acid residue into the rest of the peptide via a peptidic bond. For the last module, a final ThioEsterase domain (TE), breaks the NRPS-Peptide bond and releases the product [69]. Furthermore, several tailoring modules can modify the structure of the peptide [70]. As for PKS, the mechanism of synthesis is usually linear but can also be iterative [69,71]. NRPS are frequently associated with PKS and even FAS to generate hybrid molecules, such as the bromoalterochromides produced by several Pseudoalteromonas species (Table 4) [72,73]. Actually, many NRPS products are hybrid, meaning these products can be classified as polyketides and non-ribosomal peptides at the same time.

Bacteriocins and Bacteriocin-Like Inhibitory Substances (BLIS)
Bacteriocins and Bacteriocin-Like Inhibitory Substances (BLIS) are proteinaceous antibacterial compounds. They are ribosomally synthesized and exhibit a narrow spectrum of activity, generally limited to closely related strains. In Pseudoalteromonas, most of them are large proteins (MW > 100 kDa) ( Table 5). Although they fell under this classification, most of them exhibited enzymatic activity ((L)-amino acid oxidase (LAAO)) activity. In many cases, these LAAOs were considered to be a flavin adenine dinucleotide (FAD)-containing homodimeric protein, [76]. LAAOs are widely found in Pseudoalteromonas and provide an important ecological function in marine environments [77].

Uncharacterized Chemistry of Antimicrobial Metabolites Produced by Pseudoalteromonas Species
Numerous Pseudoalteromonas strains have shown an antimicrobial activity but the bioactive compound(s) remain unidentified [55,[85][86][87]. During a global marine research cruise, 15 strains related to P. ruthenica have exhibited antibacterial activity against Vibrio anguillarum [15]. Three bioactive fractions were detected after ethyl acetate extraction and RP-HPLC splitting. The compounds could not be assigned to any known secondary metabolites produced by pigmented Pseudoalteromonas species [15]. Moreover, P. tunicata strains inhibiting a variety of common fouling organisms were shown to produce a polar, heat-stable compound [88] and a heat-sensitive 3-10 kDa compound [89]. A cell-free supernatant of a P. piscicida strain isolated from a crustacean also showed an antifungal activity [90]. More recently, an antimicrobial anionic protein with an 87 kDa molecular weight was isolated from a culture of P. piscida [91]. This hydrophobic compound showed a high content of serine as well as aspartic and glutamic acids, but its amino acid sequence was not defined. Finally, two yellow-pigmented strains of P. citrea and P. aurantia were shown to exert an antimicrobial activity against bacteria, but their chemical nature has not been elucidated yet [92,93].

Genome Mining Strategies as a Tool to Discover Antibiotics in Pseudoalteromonas
The last two decades were marked by the onset of genomic area with cheaper and faster technologies for genome sequencing. To date, 29,000 complete bacterial genomes have been deciphered and another 31,000 bacterial genome sequences are still in progress under way (GOLD database: http://www.genomesonline.org; date of access: May 2016). This has led to the development of genome-based strategies to discover new drugs and antibiotics [94]. As a consequence, bioinformatic tools have been developed to improve the metabolite-pathway analysis, allowing access to putative metabolites and therefore to the investigation of the biotechnological potential of bacterial strains. Thanks to this genome mining approach, the metabolome can be predicted. Furthermore, bioinformatic tools enable the identification of new and potentially novel compounds via the expression of silent genes. Indeed, it is estimated that less than 10% of secondary metabolite gene clusters are expressed in sufficient amount for detection in current lab conditions [95].
As NRPS, PKS, and hybrid NRPS-PKS pathways exhibit repeated motifs in genomic nuclear sequences, they have led to the development of powerful bioinformatics programs such as antiSMASH [96], NapDos [97], SBSPKS [98], or Np.searcher [99]. Moreover, the description of a specificity-conferring code in the NRPS A-domain [100] has resulted in the development of specialised chemical structure prediction tools, e.g., NRPS predictor2 [101] and NRPS substrate predictor [102]. Another tool dedicated to ribosomally-produced peptides (bacteriocins), BAGEL3 [103] can be useful in a genome mining approach focusing on the discovery of new antimicrobial compounds. The reader is referred to the recent reviews focusing on genomics strategies defined to discover microbial natural products [104][105][106][107]. Combined with this, experimental plans have also been established to characterize potential secondary metabolites according to in silico analyses: isotope labelling, gene knockout, heterologous expression, or transcription activation. For more information on these strategies, the reader is referred to [95,106,108,109].
Despite an increasing number of available Pseudoalteromonas genomes and an increased knowledge about their bioactive compounds, only a few studies based on genome mining have been dedicated to the discovery of new antibacterial metabolites [110,111]. A consequent genomic study of 21 antimicrobial marine bacteria, of which seven belonged to the Pseudoalteromonas genus, highlighted the potential of pigmented Pseudoalteromonas strain as producers of secondary metabolites. Genomic analyses corroborated biochemical results e.g., Photobacterium halotolerans S2753 was previously described to produce holomycin [13] and the gene cluster responsible for its biosynthesis was underlined in this study. Concerning Pseudoalteromonas, despite a deeper analysis, no potential gene or gene cluster was identified on the bacterial genome of Pseudoalteromonas ruthenica [110]. As for biochemical approaches, all efforts made into genome analysis failed to characterize anti-Vibrio and Staphylococcus compounds in this strain. Added to this, the same team recently applied a coupled metabolomics and genomics workflow to determine the biosynthetic potential of Pseudoalteromonas luteoviolaceae [112]. Combined methods allowed rapid identification of new antibiotics and their biosynthetic pathways. Papaleo and co-workers conducted a study on four Antartic strains, two of which belonged to the genus Pseudoalteromonas, having inhibitory activity against Burkholderia cepacia complex [111]. As antimicrobial activity appeared to be shared and related to microbial volatile organic compounds, they conducted a genomic comparative analysis to point possible common secondary metabolite producer genes. Few candidates (11) were shown to be involved in secondary metabolite biosynthesis, transport, and catabolism, and further analysis had to be performed to clarify the implication of these genes in the antimicrobial activity.
As shown by the small number of reports, genome mining is in its infancy for the genus Pseudoalteromonas; however, pioneering studies on this topic highlight the use of the genomics approach in bioprospecting. However, as underlined by these same studies, it should be kept in mind that the reality of these virtual metabolites is to be established. They have to be structurally and functionally characterized. Therefore, the genome mining tools should be considered as tools to orientate metabolite discovery and not as an end in themselves. Nevertheless, further analyses should be carried out to characterize the different pathways and/or the natural products. The recent development of generic tools for Pseudoalteromonas genetic manipulation [113], as well as more specific ones [77,[114][115][116], will be helpful in this endeavor.

Ecological Significance in Marine Life
To date, of the 41 Pseudoalteromonas species described, 16 (39%) have been shown to produce antimicrobial compounds. When investigating the ecological niche occupied or the isolation origin of these bioactive Pseudoalteromonas, we observed that these antimicrobial metabolite-producing Pseudoalteromonas were directly associated with macroorganisms except for P. phenolica, isolated from seawater (Table 6). Therefore, there is a great temptation to connect Pseudoalteromonas antimicrobial properties with its host association. The antimicrobial metabolite producing Pseudoalteromonas strains may form a microbial shield and as a result contribute to the protection of their host against pathogens [111]. A plethora of antimicrobial metabolite-producing strains of Pseudoalteromonas has been isolated in association with marine invertebrates ( Table 6). The most documented hosts belong to Porifera (sponges in particular) and Cnidaria. The former are known to house great bacterial diversity. Pseudoalteromonas strains producing antimicrobial metabolite in vitro have been isolated from animals living in tropical [141], temperate [87,142], and cold seawater [111]. In Cnidarians, the role of Pseudoalteromonas in host defense and health has been hypothesized facing the antimicrobial metabolite producing strains of Pseudoalteromonas in corals [6,48,143,144]. Such associations with antimicrobial metabolite producing strains of Pseudoalteromonas strains are less described in the other Phyla except for molluscs (mainly bivalves [84,145,146] and crustaceans [147]).
The suggested role of such bacterial strains has led to the hologenome theory and concept [6][7][8][9][10], in which the holobiont is considered as the true evolutionary unit. Therefore antimicrobial metabolite producing strains of Pseudoalteromonas may play a key role in microbiota shaping and microbial shielding of marine invertebrates. Using next-generation sequencing methods, the impact of biotic (pathogen or probiotic) and/or abiotic stresses onto microbiota may be clarified. Such a strategy was recently applied to hemolymph microbiome of oysters [148]. An abiotic stress (temperature) was shown to provoke significant modifications of the microbiome composition while a biotic one (Vibrio sp. infection) did not. This microbiome stability has supported the hypothetical role of microbiota in host defense. Exploiting antimicrobial-producing Pseudoalteromonas spp. as tools to shape the marine host-associated microbiota along with high-throughput sequencing of host-associated microbiota may elucidate the role of Pseudoalteromonas in host defense.
The molecular dialog between antimicrobial-producing Pseudoalteromonas and the host immune system is another area that should be investigated. Whether in invertebrates, vertebrates, or algae, macroorganisms have Pattern-Recognition Receptors (PRR) at their disposal to detect and recognize microbial components known as Pathogen-Associated Molecular Patterns (PAMP). In Gram-negative bacteria such as Pseudoalteromonas, one of the most potent PAMPs is the LipoPolySaccharide (LPS). Few studies completely defined LPS structures of Pseudoalteromonas species [149]. However, it appears that most Pseudoalteromonas LPS known to date are composed of a pentaacylated Lipid A instead of a hexaacylated one [149][150][151][152][153][154]. Such a structural difference may provide an advantage to evade PRRs-mediated recognition. Indeed, LPS from various Pseudoalteromonas strains were shown to elicit a low immune response and are capable of modulating immune responses in their hosts [150,153].
Finally, in vivo experiments are required to determine the ecological significance of the association between Pseudoalteromonas and their source macroorganisms. The major issue is to determine whether antimicrobial-producing Pseudoalteromonas provide a real benefit to their host, especially in the context of pathogenic events. In such a case, Pseudoalteromonas could stand as a next generation of probiotics for marine aquaculture.

Conclusions
The genus Pseudoalteromonas has a high, if not the highest, proportion of species producing antimicrobial metabolites in the marine bacterial world. However, this bacterial genus is still underexplored at the biotechnological level. Furthermore, the vast majority of Pseudoalteromonas spp. have not been found to exhibit pathogenicity. Almost all known antimicrobial-producing Pseudoalteromonas spp. originate from healthy marine macroorganisms, suggesting that these strains may participate in the host's homeostasis. It would be simplistic to link only the associated microbiota and the ensuing microbial shield with the presence of Pseudoalteromonas strains. Obviously, the metabolite richness and the genetic plasticity of the symbiont-host relationship result from metabolic and therefore microbial diversity. Nevertheless, the involvement of antimicrobial-producing Pseudoalteromonas in microbiota shaping and protection of their host should not be under-estimated, or neglected. In any event, the genus Pseudoalteromonas appears as the first or at least one of the leading antimicrobial providers in the marine microbiota. Therefore, Pseudoalteromonas strains offer real potential to develop the next generation of marine probiotics and their use as probiotics in aquaculture should be further investigated.