Genome analysis of the salt-resistant Paludifilum halophilum DSM 102817T reveals genes involved in flux-tuning of ectoines and unexplored bioactive secondary metabolites

Paludifilum halophilum DSM 102817T is the first member of the genus Paludifilum in the Thermoactinomycetaceae family. The thermohalophilic bacterium was isolated from the solar saltern of Sfax, Tunisia and was shown to be able to produce ectoines with a relatively high-yield and to cope with salt stress conditions. In this study, the whole genome of P. halophilum was sequenced and analysed. Analysis revealed 3,789,765 base pairs with an average GC% content of 51.5%. A total of 3775 genes were predicted of which 3616 were protein-coding genes and 73 were RNA genes. The genes encoding key enzymes for ectoines (ectoine and hydroxyectoine) synthesis (ectABCD) were identified from the bacterial genome next to a gene cluster (ehuABCD) encoding a binding-protein-dependent ABC transport system responsible for ectoines mobility through the cell membrane. With the aid of KEGG analysis, we found that the central catabolic network of P. halophilum comprises the pathways of glycolysis, tricarboxylic acid cycle, and pentose phosphate. In addition, anaplerotic pathways replenishing oxaloacetate and glutamate synthesis from central metabolism needed for high ectoines biosynthetic fluxes were identified through several key enzymes. Furthermore, a total of 18 antiSMASH-predicted putative biosynthetic gene clusters for secondary metabolites with high novelty and diversity were identified in P. halophilum genome, including biosynthesis of colabomycine-A, fusaricidin-E, zwittermycin A, streptomycin, mycosubtilin and meilingmycin. Based on these data, P. halophilum emerged as a promising source for ectoines and antimicrobials with the potential to be scaled up for industrial production, which could benefit the pharmaceutical and cosmetic industries.


Introduction
Among extreme niches, natural and artificial hypersaline habitats were shown to harbor several species of halotolerant and halophilic bacteria (Boujelben et al. 2015;Gibtan et al. 2017). To cope with high salinity, one of the most important strategies used by these salt-resistant and salt-loving bacteria is the transport and/or biosynthesis of organic osmolytes, the so-called compatible solutes (Kempf and Bremer 1998;Bremer and Krämer 2000;Wood et al. 2001). These highly water-soluble organic molecules can protect microorganisms against salt stress, dehydration, heat, oxydative and UV stresses (Schröter et al. 2017;Brands et al. 2019). Among these compatible solutes, ectoines are of particular interest. Beside their primary function of protecting cells against harsh conditions, the most powerful stabilizing properties on biological macromolecules (enzymes, DNA, antibodies, and Donyez Frikha-Dammak and Houda Ayadi have contributed equallyto this work. even whole cells) confer ectoines attractive potentials in the fields of skin caring, food processing, molecular biology, agriculture, biotechnology, and medical values in human diseases Kanapathipillai et al. 2005;Graf et al. 2008;Pastor et al. 2010;Abdel-Aziz et al. 2013;Hahn et al. 2017). This led to the development of the first industrial-scale production process using the salt-tolerant bacterium Halomonas elongata as the host (Schwibbert et al. 2011;Kunte et al. 2014). This strain naturally synthesizes ectoine from the precursor l-aspartate-β-semialdehyde (l-ASA), a central hub in microbial aspartate family amino acid production. The biosynthesis comprises a cascade of three biochemical conversions catalyzed by l-2,4-diaminobutyrate transaminase (EctB), l-2,4-diaminobutyrate acetyltransferase (EctA), and ectoine synthase (EctC) (Schwibbert et al. 2011). This major biosynthesis pathway of ectoine has been clearly characterized to be a highly conservative cluster across halophile species (León et al. 2018;Zhao et al. 2018;Ma et al. 2020;Van Thuoc et al. 2020). Some microorganisms are also able to synthesize a hydroxylated derivative of ectoine, the 5-hydroxyectoine catalyzed by the ectoine hydroxylase (EctD) (Prabhu et al. 2004;Garcia-Estepa et al. 2006;Höppner et al. 2014).
According to the whole-genome organization of halophilic bacteria such as Halomonas elongata, Salinivibrio proteolyticus, and Chromohalobacter salexigens, the three genes (ectA, ectB, ect C) involved in ectoines biosynthesis from l-ASA are typically organized in an operon ect-ABC (Schwibbert et al. 2011;Czech et al. 2018a;Van Thuoc et al. 2020). The fourth gene ectD, responsible for bioconversion of ectoine to hydroxyectoine, can be part of the ectABC ectoine biosynthetic gene cluster (BGC), but it is often found elsewhere in the genome (Widderich et al. , 2016. ect gene clusters may sometimes contain the gene for a specialized aspartokinase (ask_ect) (Reshetnikov et al. 2006;Stöveken et al. 2011). Aspartokinase (Ask), and in selected cases of the aforementioned Ask_Ect enzyme, along with l-aspartate-β-semialdehyde-dehydrogenase (Asd), provide the precursor l-ASA for ectoine biosynthesis (Reshetnikov et al. 2011;Czech et al. 2018b). In addition, various types of transport systems detected in the bacterial genome of halophilic strain, mediate the acquisition of these stress protectants from surrounding media. These include the TeaABC system from H. elongata and Salinivibrio proteolyticus M318, which belong to the periplasmic binding protein-dependent tripartite ATP independent periplasmic transporter family (TRAP-T) (Rosa et al. 2018). Another transport system, named Ehu (ectoine-hydroxyectoine uptake) that belongs to the binding protein-dependent ABC transporter members was also detected in the genome of Sinorhizobium meliloti (Ter Beek et al. 2014).
Beside their ability for osmoprotectants production, salt resistant bacteria, especially actinomycetes, are among the candidates considered to have the potential to produce special or unknown bioactive metabolites. Yet, only limited attention has been given to their secondary metabolite biosynthesis (Manivasagan et al. 2014;Kim et al. 2017). Cultivability dependent methods are not very effective in searching for novel secondary metabolites because most of their biosynthesis genes cannot be expressed, or are expressed at a very low level. Meanwhile, genome sequencing methods have proven to be more effective and a panoply of hidden BGCs were revealed in halophilic bacterial genomes (Doroghazi et al. 2014;Min et al. 2018). In addition, genome mining efforts have also allowed to understand the silencing or activation of biosynthetic pathways in microbes with the development of bioinformatics software, such as antiSMASH (antibiotics and Secondary Metabolite Analysis Shell), SMURF (Secondary Metabolite Unique Regions Finder) and PRISM (PRediction Informatics for Secondary Metabolomes) (Baral et al. 2018). Traditionally, BGCs include non-ribosomal peptide synthase (NRPS), polyketide synthase (PKS), and ribosomally synthesized and post-translationally modified peptide (RiPP) family clusters. For instance, a total of 104 antiSMASH-predicted putative BGCs for secondary metabolites with high novelty and diversity were identified in nine Ktedonobacteria genomes (Zheng et al. 2019). In the study of Hu et al. (2018), seven of the PKS and NRPS gene clusters related to antibiotics compounds, including friulimicin, lobophorin, laspartomycin, colabomycin, borrelidin, pristinamycin and kanamycin, have been discovered in the genome of an actinomycetale streptomyces strain isolated from plant samples collected from a highsalt environment. The prediction of gene clusters involved in the biosynthesis of terpenoid/polyketide synthase by genome and transcriptome sequencing revealed a new family of diterpene cyclases in several bacteria (Yang et al. 2017).
Paludifilum halophilum DSM 102817 T is a salt and heat-resistant bacterial strain from the family of Thermoactinomycetaceae from the phylum Actinobacteria (Jiang et al. 2019). It was isolated from Sfax solar saltern sediment in Tunisia (Frikha-Dammak et al. 2016) and has shown be a high-yield ectoine producer under salt stress conditions (Ayadi et al. 2020). Because of the industrial importance of ectoines and novel antimicrobials, and the performance of the strain, a research project was established to sequence and analyse the genome of P. halophilum DSM 102817 T . This helps to elucidate synthesis and degradation pathways of ectoines, which is presented in this study. Furtheremore, we provide a comparative analysis of the genes involved in ectoine metabolism from this genome with corresponding genes from the industrial producer H. elongata.

Materials and methods
Strain culture conditions P. halophilum DSM 102817 T (strain SMBg3) was routinely cultivated in an optimized mineral sea water (SW) medium (Ayadi et al. 2020) with the following composition (per liter): 5 g glucose, 4.07 g MgSO 4 ⋅7H 2 O, 2.6 g MgCl 2 ⋅6H 2 O, 0.4 g KCl, 47 mg NaBr, 13 mg NaHCO 3, 67 mg CaCl 2 ⋅2H 2 O, 278 mg FeSO 4 ⋅7H 2 O and 1 g aspartic acid. The pH of the growth medium was adjusted to 8.3-8.4 with 4 M NaOH, and the osmolarity of the individual experiments was set-up by adding NaCl. A 100-mL medium in 250 mL erlenmeyer flasks were inoculated with 5 mL of exponentially growing pre-cultures and incubated at 40 °C in a rotary shaker incubator (New Brunswick Scientific, NJ, USA) at 200 rpm. Their growth was monitored by measuring the dry mycelial weight (DMW) from 2 mL of broth culture after incubation of the pellet at 80 °C for 24 h.

Effect of salinity on growth and ectoines identification
To investigate the effects of salinity on the growth of P. halophilum, 5 mL of seed culture was inoculated in 100 mL of glucose SW medium with NaCl concentration ranging from 5 to 25% (w/v). The flasks were incubated as described above and samples were withdrawn to measure the increase of dry mycelial biomass and the growth was evaluated by the determination of maximal growth rate (µ max ) in exponential phase. Ectoines extraction according to the method of Kunte et al. (1993) was done on mycelial cells harvested from a six-day culture in SW-15 medium with 15% NaCl. Briefly, 10 mg dry cell pellets were resuspended in 570 µL of a methanol/chloroform/water mixture (10/5/4, v/v/v) and mixed for 15 min at 37 °C. To precipitate proteins and extract osmolytes, 170 µL of chloroform and 170 µL of water were added. Liquid phase separation was enhanced by gentle centrifugation and the hydrophilic top layer containing compatible solutes was recovered. The identification of ectoines was determined by HPLC-UV based on their retention time in comparison with standard products (sigma Aldrich) on a KNAUER-NH 2 analytical column, with 2 mL/min of flow rate and detection at λ max of 210 nm.

Whole genome sequencing, annotation and bioinformatic analyses
The biomass of P. halophilum for genome analysis was obtained from the culture grown at 40 °C for 6 days in glucose SW-15 mineral broth with 15% salinity. A high quality genomic DNA was prepared using the DNeasy Plant Maxi Kit-Qiagen, following the manufacturer's guidelines (Thermo, USA) (www. qiagen. com). Agarose gel (1.5%) electrophoresis was used for visual assessment of DNA integrity, quantified by the nanodrop method (Garcia-Alegria et al. 2020). Genomic sequencing was carried out using a paired-end sequencing strategy at the University of Neuchatel, using next-generation sequencing technology Illumina MiSeq2000 instrument. The paired-end library had a mean insert size of 500 bp. Reads were assembled using the CLC NGS Cell v. 5.0.4 assembler (CLC bio, Waltham, MA). The genomic DNA base content (mol% GC) was directly calculated from the draft genome data. The draft genome sequence of P. halophilum reported in this study has been deposited at DDBJ/ENA/GenBank under the accession NOWF00000000.1 (https:// www. ncbi. nlm. nih. gov/ nucco re/ NOWF0 00000 00.1); BioProject: PRJNA395604 (https:// www. ncbi. nlm. nih. gov/ biopr oject/ PRJNA 395604); BioSample: SAMN07411558 (https:// www. ncbi. nlm. nih. gov/ biosa mple/ 74115 58). This WGS version of the project consists of sequences NOWF01000001-NOWF01000213.
To ensure the exact phylogenetic position of P. halophilum within the Thermoactinomycetaceae family, multiple genome alignment was generated by Mauve software (http:// asap. ahabs. wisc. edu/ mauve/) (Darling et al. 2010), and a phylogenomic tree was visualized by FigTree and EvolView tools. The online OrthoVenn, available at website http:// www. bioin fogen ome. net/ Ortho Venn/ was used for developing Venn diagram.

Results
Salt tolerance and ectoine biosynthesis potential P. halophilum was selected due to its high salt resistance potential during our primary assay of the Sfax saltern strains library. This strain grows well on Bennett's and ISP2 agar, generating white spores. To determine its resistance potential against environmental salinity, it was grown in modified SW-media supplemented with different concentrations of NaCl and the results are presented in Fig. 1a. No growth was detectable up to a salinity of 5% NaCl, a property expected for a halophile. An increase in the salinity up to 20% NaCl strongly stimulated growth, but further increases to 25% impaired it. In fact, P. halophilum not only depends on a considerable salt concentration for its growth but it can also cope with a broad spectrum of salinities (from 5 to 20% NaCl). To correlate the osmotolerance capacity of the strain with the solute compatible production, we assessed the ability of P. halophilum for ectoines synthesis. According to the results of the HPLC analysis, we found that at an optimum salinity growth of 15% NaCl, the most abundant osmoprotector was ectoine and hydroxyectoine that shared the same retention times of 22.609 and 20.027 min, respectively with standard ectoines (Fig. 1b).

Genome characteristics and phylogeny
Upon sequencing, the chromosome of P. halophilum was assembled from 13 million reads resulting in a total length
To further distinguish P. halophilum from the three closely related thermoctinomycete species, we ran EDGAR analysis and results are shown in Fig. 5. P. halophilum was shown to harbor 43 distinct genes that were not found in the other three closely associated species. Moreover, several distinctive genes were also identified in the other three species, shown in parenthesis: K. eburnean (21), P. fulgidum (26) and M. thermohalophilus (15), respectively (Fig. 5a). Homology searching of genes encoding known proteins involved in secondary metabolism revealed the existence of at least 1638 common gene clusters associated with the biosynthesis of Fig. 3 Pie chart of GO (gene ontology) analysis summarized P. halophilum genes, according to molecular function secondary metabolites among the four strains. P. halophilum yielded 2621 metabolites, K. eburnean showed 2614 genes coding for secondary metabolite, M. thermohalophilus provided 2516 as global genes coding for secondary metabolite, and P. fulgidum 2264 genes coding for a global metabolic secondary production (Fig. 5b).

Genes involved in ectoines biosynthesis and degradation
The second objective of the study was to identify, based on the draft genome of P. halophilum, biosynthetic and catabolic pathways of ectoines under salt stress conditions. While the bacterial genome lacks all the genes for ectoines degradation, it contains the whole canonical ectABCD ectoine/hydroxyectoine BGC, a diaminobutyrate-2-oxoglutarate transaminase (ectB), a l-2,4-diaminobutyric acid acetyltransferase (ectA), an ectoine synthase (ectC), and an hydroxyectoine synthase (ectD) (Fig. 6). The nucleotide blast results extracted from NCBI database showed that genes ectA, ectB, ectC and ectD of P. halophilum shared 100% identity with those of Streptomyces chrysomallus. However, ectoine/hydroxyectoine BGCs often contained other genes involved in either the transcriptional regulation (ectR) of the ect operon, the provision of the precursor l-ASA from aspartyl-P (asd) and the aspartyl-P from aspartate (ask_ect), or sometimes, even a gene for a mechanosensitive channel (mscS) (Reshetnikov et al. 2011;Widderich et al. 2016;Czech et al. 2018b). We found that P. halophilum genome lacks the mentioned ectR, ask_ect, or mscS genes. However, we detected four genes encoding a binding-protein dependent ABC transporter located upstream of the ectABCD genes (Fig. 6). A closer analysis of these genes revealed that the encoded proteins are related to those of the functionally characterized ectoine/hydroxyectoine ABC-type uptake system Ehu ABCD from S. meliloti (Jebbar et al. 2005;Hanekop et al. 2007). To compare the organization of the genomic region encoding the ectoine/hydroxyectoine synthase of P. halophilum with the industrial ectoine producer H. elongata, a Mauve analysis was also used (Fig. 6). Results showed that the four genes in P. halophilum were organized in a canonical ectABCD cluster, while in H. elongata, the ectD gene encoding the hydroxylase for hydroxyectoine synthesis was located apart from the ectABC cluster. In addition, Ehu genes coding for the transport of ectoine/hydroxyectoine were found only in P. halophilum.

Central carbon metabolism related to the synthesis of precursors of ectoines
As a further step, we mapped in P. halophilum and with the aid of KEGG analysis, the central carbon metabolism to investigate its potency to provide building precursors for ectoines biosynthesis (Czech et al. 2018b;Ma et al. 2020). The bacterial genome carries genes for glycolysis, pentose phosphate (PPP), and tricarboxylic acid (TCA) pathways, but missing genes for Entner Doudoroff (ED) pathway (Fig.  S1). We also assessed several key enzymes, namely pyruvate carboxylase (Pc), phosphoénolpyruvae carboxylase (Ppc) and oxaloacetate decarboxylase (Oad). These enzymes Fig. 5 Venn diagram of the number of homologous genes between P. halophilum and K. eburnea, M. thermohalophilus and P. fulgidum with its closest functional relatives, respectively. BioVenn, a web application for the comparison and visualization of biological lists, has been used for Venn diagrams drawing Fig. 6 Genetic organization of the ehu-ect cluster from P. halophilum and comparison with ect cluster of H. elongata. Genes annotation: EhuC, ectoine/hydroxyectoine ABC transporter permease; EhuB, ectoine/hydroxyectoine ABC transporter substrate-binding protein; ectA, diaminobutyrate acetyltransferase; PP_2800, diaminobutyrate-2-oxoglutarate transaminase (ectB); ectC, ectoine synthase; thpD, ectoine hydroxylase (ectD) respectively interconvert pyruvate, phosphoenolpyruvate, and OAA and could have a role in supporting high ectoine biosynthetic fluxes by anaplerotic pathways replenishing OAA needed for the TCA cycle. Interestingly, the genes encoding these enzymes, except oxaloacetate decarboxylase were identified from the genome of P. halophilum (Table 2).
Regarding nitogen metabolism, analysis of the ectoine biosynthesis pathways revealed the importance of glutamate and alanine in directing fluxes through ectoine synthesis pathway (Ono et al. 1999). A number of 13 copies of genes for glutamate synthase (Glt) and 3 copies for glutamate dehydrogenase (Gdh) in the P. halophilum genome were identified, but alanine aminotransferase (Alaat) and l-alanine dehydrogenase (Ald) were not detected ( Table 2). The enzymes specified by these genes are responsible for the reductive transfer of ammonium to 2-ketoglutarate to generate glutamate, which acts as the major ammonium donor in the cell (Magasanik 1982). There were also 8 putative aspartate aminotransferases (AspC), which catalyze the reversible transfer of the amino group from glutamate to oxaloacetate, rendering aspartate and 2-ketoglutarate. This is a key enzyme as it links the TCA cycle with aspartokinase (Ask), the first enzyme of the ectoines synthesis pathway. P. halophilum has only one aspartokinase (Ask) catalyzing the formation of aspartyl phosphate, which is a common metabolic intermediate in the biosynthesis of ectoines and aspartate family of amino acids. Together, these results support that the genome of P. halophilum harbors the genes for high flux of ectoines through the metabolic model shown in Fig. 7.

Genes involved in diverse secondary metabolites biosynthesis
Another interesting genomic trait of strain P. halophilum is the presence of several new gene clusters that have low similarity with known clusters. A total of 18 gene clusters involved in secondary metabolism were predicted by antiSMASH, including 1 NRPS (non-ribosomal peptide synthetase) type, 1 PKS (polyketide synthase) type 3 and 2 hybrid clusters, namely Type 1 PKS-NRPS and NRPSfatty acid type biosynthetic clusters (Table 3). Out of the 18 potential biosynthetic clusters, 8 exhibited some level of similarities with known BGC whereas 10 clusters represented orphan BGCs for which no known homologous gene clusters could be identified. Notably, 6 of the known clusters shared similarity with those for antibacterial compounds including colabomycine-E, fusaricidin-A, zwittermycin-A, streptomycin, meilingmycin, and mycosubtilin, whereas the 2 others shared similarity with those S-layer-glycan or ectoine compounds. However, the levels of similarity were fairly low in most cases, which suggests the novelty of the possible metabolites from those predicted gene clusters. Several other secondary metabolites could be potentially produced by P. halophilum. Among them, one siderophore molecule encoded by cluster 2, one bacteriocin (cluster 16), and two other gene clusters, 8 and 9, are predicted to be responsible for terpene biosynthesis.

Discussion
High salinity is a key determinant for the growth of P. halophilum in the saltern ponds from which it was originally isolated (Frikha-Dammak et al. 2016). Such a challenging habitat requires active measures by this halophile to counteract the outflow of water from the cells and to optimize the solvent properties of the cytoplasm for biochemical reactions and the functionality of cell components (Czech et al. 2018a). Analysis of the physiological response of P. halophilum revealed that this bacterium uses the salt-out strategy through the accumulation of ectoines at relatively high yields (Ayadi et al. 2020). The data presented here support this conclusion and highlight the ability of P. halophilum to cope with a broad spectrum of salinities ranging from 5 to 20%. At an optimal salinity of 15%, the synthesis and accumulation of the effective compatible solute ectoine reached its maximum yield of about 12%.
When the whole genome sequence of P. halophilum was sequenced and taxonomically analyzed using the Mauve software pipeline, it was found to be taxonomically clustered with K. eburnea, P. fulgidum and M. thermohalophilus as shown by construction of a phylogenomic tree (Fig. 2). To further infer the evolutionary relatedness of the strain relative to closely related phylogenetic species, we ran EDGAR analysis and results showed that P. halophilum harbors 43 distinct genes which were not found in the other three closely associated species (Fig. 5). The 43 distinct genes found in P. halophilum, made up approximately 1.7% of the total genome size of the strain. These distinct gene determinants were likely salt resistance proteins, transcriptional regulators, transporter proteins, and for sporulation proteins (Table S2). The comparative genomic analysis of the strain P. halophilum with other three strains confirms a strong adaptation potential possessed by the strain. Genomics-based approaches have been developed to unveil biosynthetic pathways of ectoines from P. halophilum. The four genes (ectA, ectB, ectC, and ectD) responsible for ectoines synthesis from l-ASA were identified from the bacterial genome (Fig. 6a). The genetic organization of the ectABCD operon from the strain is in canonical arrangement (Schwibbert et al. 2011;Widderich et al. 2014Widderich et al. , 2016 as all genes were located in the same cluster. The presence and the arrangement of these fourth genes in an operon is also found in some other halophilic bacteria such as the industrial strain H. elongata, C. salexigens, Streptomyces coelicolor, and Nocardiopsis gilva Schwibbert et al. 2011;Han et al. 2018;Piubili et al. 2018). However, a different genetic organization of the ectABCD genes in the genome sequence of P. halophilum was observed. These are positioned next to a gene cluster (ehuABCD) encoding a binding-proteindependent ABC transport system that serves to carry ectoine and 5-hydroxyectoine to the surrounding medium (Jebbar et al. 2005). This Ehu transport system in the P. halophilum genome seems to be primarily involved in the adjustment of represented by dashed lines. ectA, l-2,4-diaminobutyrate acetyltransferase; ectB, l-2,4-diaminobutyrate transaminase; ectC, ectoine synthase; ectD, ectoine hydroxylase. aspC, ask, and asd encoding aspartate aminotransferase, aspartokinase, and l-aspartate-semialdehyde-dehydrogenase, respectively the intracellular ectoines concentration by its secretion in extracellular space instead of its degradation as in H. elongata (Schwibbert et al. 2011) or C. salexigens (Vargas et al. 2006). Indeed, previous studies have revealed that the genetically engineered H. elongata strain lacked ectoine catabolic genes (Schwibbert et al. 2011;Schulz et al. 2017) and their osmotically controlled ectoine/hydroxyectoine-specific tripartite ATP-independent periplasmic (TRAP) transport system TeaABC (Grammann et al. 2002) was deleted. It excretes considerable amounts of ectoines into the growth medium and allows their recovery in a highly purified form on the scale of several tons per annum (Kunte et al. 2014). This genetic arrangement of ehu-ect gene cluster is also found in the Nitrospina sp. SCGC AAA 799_C22 isolate, a bacterium that lives in the polyextreme interfaces of Red Sea brines (Ngugi et al. 2016), and also in the thermo-halotolerant Gram-positive bacterium Paenibacillus lautus (Richter et al. 2019). High-level ectoines impose a biosynthetic burden on cells. For this purpose, we assessed a number of routes related to central metabolism leading to precursors of ectoines. Genome analysis revealed that the central catabolic network in P. halophilum comprises the pathways of glycolysis, pentose phosphate, and tricarboxylic acid cycle that provide OAA, acetyl-coASH and NADPH2 required for ectoines production (Fig. S1). In addition, P. halophilum possesses specific genes encoding for pyruvate carboxylase (Pc) and phosphoenolpyruvate carboxylase (Ppc) involved in anaplerotic replenishment of the TCA cycle (Table 2). These enzymes convert respectively pyruvate and phosphoenolpyruvate to OAA and could have a role in replenishing the precursor l-ASA, the starting metabolite of the pathway for biosynthesis of ectoines and aspartate family amino acids (threonine, methionine, lysine). These anaplerotic replenishment pathways were also detected in C. salexigens (Pastor et al. 2013) and H. elongata (Schwibbert et al. 2011). Since ectoines are nitrogen compounds and their biosynthesis also depends on a good supply of glutamate used as a substrate of EctB in the first step of l-2,4-diaminobutyrate biosynthesis, we analyzed key enzymes involved in nitrogen assimilation. Interestingly, we found 3 copies of glutamate dehydrogenase, and 13 copies of glutamate synthase in P. halophilum genome. These enzymes are involved in anaplerotic replenishment of intracellular glutamate pool, which in turn directs the cellular metabolism toward high-levels of ectoines. It is known that glutamate dehydrogenase (Gdh) catalyses a reaction between ammonium and 2-oxoglutarate to directly build glutamate in an NADPH-dependent reaction, while glutamate synthase (Glt) assimilates ammonium and glutamate into glutamine, and subsequently, the glutamate synthase converts glutamine and 2-oxoglutarate to form two molecules of glutamate (Magasanik 1982). These two glutamates supply cellular pathways in P. halophilum were previously identified in the model strain of actinomycete, Streptomyces coelicolor and showed to be activated  (Tiffert et al. 2008;Shao et al. 2015). In addition, aspartate aminotransferase catalyzes the reversible transfer of the amino group from glutamate to oxaloacetate and links the TCA cycle with aspartokinase, the first enzyme of the ectoines synthesis pathway. Together, these results suggest that P. halophilum can be considered as a new natural cell factory for highlevel production of ectoines. Indeed, our preliminary batch fermentation in lab-scale 7.5 L bioreactor showed that the conversion yield of ectoines on glucose reached the value of 37% with a yield of 0.63 g/g dry cellular weight, threefold higher than that of H. elongata and half of which was found in the fermentation supernatant (unpublished data). Another interesting genomic trait of P. halophilum, the presence of several new gene clusters coding for new natural products that may be undetected under standard fermentation conditions (Winter et al. 2011). In P. halophilum, several new gene clusters that have low similarity with known clusters have been detected (Table 3). As the typical PKSI, the BGC of zwittermycin-A (ZmA) identified in P. halophilum has been first identified in Bacillus cereus UW85, including the condensation of five precursors (Kevany et al. 2009). Two of these precursors, l-Ser and malonyl-CoA, would be readily available for ZmA biosynthesis, being common primary metabolites. The three remaining precursors would be unique to ZmA biosynthesis and would require specific enzymes for their formation. The PKS gene cluster of ZmA from Bacillus cereus UW85 contains zmaA, zmaB, zmaK, zmaM, zmaO and zmaQ and encodes transcriptional regulator, aminotransferase, UDPkanosamine hydrolase, UDP-glucose C3 dehydrogenase and kanosamine transporter (Kevany et al. 2009). As the typical PKS III, meilingmycin detected in P. halophilum, a macrolide antibiotic structurally and biologically similar to avermectin was already found in Streptomyces nanchangensis (Sun et al. 2002). meilingmycin differs from avermectin in the fact that it has no α-l-oleandrose attached at position 13 of the macrolide ring but has an isopantenoic acid moiety at position 4, which is probably derived from valine. We also identified in P. halophilum genome a biosynthetic cluster showing 50% similarity with known fusaricidin biosynthetic cluster (BGC0001152_c1). Fusaricidin is a lipopeptide antibiotics previously produced by Paenibacillus polymyxa (formerly Bacillus polymyxa) and consists of a guanidinylated β-hydroxy fatty acid linked to a cyclic hexapeptide including four amino acid residues in the D-configuration (Kuroda et al. 2000). Besides, P. halophilum harbors the ectoine biosynthetic pathway that shows 100% similarity with ectione biosynthetic cluster of Streptomyces chrysomallus (BGC0000853_c1). Finally, the availability of the genome sequence of P. halophilum SMBg3 provides a framework for biotechnological analysis and characterization of new natural products.

Conclusions
This work presents the first insight into the genome of the highly salt-resistant bacterium P. halophilum DSM 102817 T . Determining the complete genome sequence of the strain, allowed us to access and understand the metabolic pathways of ectoines, and paves the way for post-genome technologies to create more efficient producer strains. Our data revealed that the metabolism of P. halophilum under salt stress is well adapted to support high fluxes biosynthetic toward ectoines through an ehuect gene cluster. The existence of multiple routes for the flux-tuning of ectoines precursors, the hability to release ectoines to the surrounding medium, and the inability to degrade ectoines, might allow the strain to attain high total yields of products. Based on these genomic data, a flux balance model of ectoines production from glucose and ammonia has been developed. In addition, the prediction of BGCs for secondary metabolites suggested that P. halophilum has the potential to produce various novel natural products, which could be of industrial or scientific significance. Currently, our work is in progress to develop at pilot-scale, a process for efficient production of ectoines by P. halophilum coupling feed-batch fermentation with the downstream extraction processing.