The complete genome sequence of Ensifer meliloti strain CCMM B554 (FSM-MA), a highly effective nitrogen-fixing microsymbiont of Medicago truncatula Gaertn

Strain CCMM B554, also known as FSM-MA, is a soil dwelling and nodule forming, nitrogen-fixing bacterium isolated from the nodules of the legume Medicago arborea L. in the Maamora Forest, Morocco. The strain forms effective nitrogen fixing nodules on species of the Medicago, Melilotus and Trigonella genera and is exceptional because it is a highly effective symbiotic partner of the two most widely used accessions, A17 and R108, of the model legume Medicago truncatula Gaertn. Based on 16S rRNA gene sequence, multilocus sequence and average nucleotide identity analyses, FSM-MA is identified as a new Ensifer meliloti strain. The genome is 6,70 Mbp and is comprised of the chromosome (3,64 Mbp) harboring 3574 predicted genes and two megaplasmids, pSymA (1,42 Mbp) and pSymB (1,64 Mbp) with respectively 1481 and 1595 predicted genes. The average GC content of the genome is 61.93%. The FSM-MA genome structure is highly similar and co-linear to other E. meliloti strains in the chromosome and the pSymB megaplasmid while, in contrast, it shows high variability in the pSymA plasmid. The large number of strain-specific sequences in pSymA as well as strain-specific genes on pSymB involved in the biosynthesis of the lipopolysaccharide and capsular polysaccharide surface polysaccharides may encode novel symbiotic functions explaining the high symbiotic performance of FSM-MA.


Introduction
To secure their nitrogen supply, legumes such as alfalfa, pea, (soy−/faba-)bean establish an endosymbiotic interaction with soil bacteria collectively called rhizobia that can reduce atmospheric nitrogen gas and produce reduced nitrogen molecules metabolizable by the plants. This symbiosis between legumes and rhizobia is of ecological and economic importance because of its contribution to the global nitrogen cycle, its impact on sustainable agriculture and its biotechnological potential to ensure nitrogen supply in agriculture [1].
The reduction of atmospheric nitrogen by rhizobia takes place in a specific niche, within the cells of de novo formed organs called nodules found usually on the roots and in some cases on the stem of the plants. Nodule development is initiated when flavonoids released by the plants induce the expression of the bacterial nodulation (nod) genes resulting in the production of the lipo-chitooligosaccharide signal molecules, the Nod factors. Nod factors cause a change in the direction of polar growth in developing root hairs and simultaneously induce cell division in the root cortex cells. As a result, a nodule primordium is formed that turns into meristematic tissue to produce the cells of the nodule and bacteria become entrapped in the curled root hair where they form an infection pocket. From the site of the infection pocket, a tubular structure, called infection thread, is formed in the root hair that grows toward the cells of the developing nodule. In the infection thread, bacteria multiply and finally they are released into the cytoplasm of the nodule cells via a mechanism resembling endocytosis resulting in organelle-like structures called symbiosomes. Symbiosomes have a membrane of plant origin which surrounds one or more bacteria. After bacterial release, the cells of both partners differentiate into mature symbiotic cells. The nodule cells become enlarged polyploid cells which host several tens of thousands of bacteria that are themselves differentiated into a nitrogenfixing form called bacteroid [2][3][4]. Interestingly, in Medicago and closely related species like Pisum and Vicia, the host imposes a terminal differentiation on the bacterial partner that is accompanied by the increase in the DNA content and size of the bacteroids and results in the loss of their cell division capacity [5]. This terminal differentiation is orchestrated by nodule-specific cysteine-rich peptides that are expressed exclusively in the infected cells of the nodule [6,7].
To effectively investigate these interactions, two genetic model legume species, Lotus japonicus (Regel) K. Larsen (bird's-foot trefoil) and Medicago truncatula Gaertn. (barrel clover/barrel medic) have been chosen for which structural and functional genomics tools and databases have been developed [8,9]. M. truncatula is a diploid, selfpollinating annual plant belonging to the Medicago genus, which contains species that are among the most extensively cultivated forage and pasture plants. Medicago plants establish symbiosis only with a limited number of bacterial species, mainly with Ensifer (synonym Sinorhizobium) meliloti and Ensifer medicae, and with certain Ensifer fredii strains and Rhizobium mongolense [10][11][12]. However, some combinations of wild-type plants (species, sub-species and ecotypes) and bacterial strains of the most-studied bacterial species, E. meliloti and E. medicae, often lead to incompatible interactions [13][14][15][16][17], i.e. nodule formation is initiated but bacteria cannot invade nodules or cannot persist and fix nitrogen in the symbiotic organ. The incompatibility can be caused by functions/proteins encoded by genes in the accessory genome of the bacteria [14] such as the strain-specific HrrP peptidase [18], strain specific exopolysaccharide production [19] and/or allelic variants of the host genes like the NFS1 and NFS2 genes encoding NCR peptides in M. truncatula [20,21]. Strikingly, the model bacterium E. meliloti strain 1021 (with the reference genome and most of the available mutants) is poorly matched for nitrogen fixation with the most widely used M. truncatula accessions Jemalong A17 and M. truncatula ssp. tricycla R108 [22,23].
E. meliloti strain FSM-MA (first catalogued as E. arboris strain CCMM B554, also known as LMG-R33403 and MR372) was isolated from the nodules of Medicago arborea L. (moontrefoil/tree medic) in Maamora Forest between Rabat and Meknes, Morocco, and is stored in The Moroccan Coordinated Collections of Microorganisms as CCMM B554. Recently, Kazmierczak et al. [22] identified E. meliloti strain FSM-MA as a highly effective symbiotic partner of the two most widely used M. truncatula ecotypes, A17 and R108, as well as all tested Medicago sativa L. (alfalfa) cultivars. To gain the potential to identify novel bacterial symbiotic genes and genes associated with FSM-MA's exceptional symbiotic performance, we sequenced the genome of the strain FSM-MA. Here we present a summary classification and a set of general features for E. meliloti strain FSM-MA, together with a description of its genome sequence and annotation.

Organism information
Classification and features E. meliloti FSM-MA is a motile, non-sporulating, Gramnegative strain (Fig. 1) in the order Rhizobiales of the class Alphaproteobacteria. This fast growing strain forms colonies within 3 days on YEB agar plates [22] at 30°C. The colonies (Fig. 1a, b) are light beige colored on YEB plates, slightly doomed, mucoid and have a smooth margin. The rod shaped free-living form (Fig. 1c, d) has dimensions of 1.0-2.0 μm in length and approximately 0.5 μm in width, while bacteroids in M. truncatula Jemalong A17 nodules (Fig. 1e, f ) have the same width and are elongated to 5-8 μm. A summary of the classification is provided in Table 1.

Extended feature descriptions
Phylogenetic analysis of E. meliloti strain FSM-MA was performed by aligning the 16S rRNA sequence to the 16S rRNA sequences (consensus sequence length of 1346 basepairs (bp)) of other Ensifer strains (Fig. 2). The FSM-MA 16S rRNA sequence has 100% sequence identity with those of the widely used E. meliloti strains such as strain 1021 or Rm41, while four mismatches can be observed with the E. medicae strain WSM419 sequence. Moreover, there are five mismatches between the 16S rRNA sequence fragments of strain FSM-MA and E. arboris strain LMG14919 T . A Multilocus Sequence Analysis (Additional file 1: Figure S1) using 14 chromosomal genes further confirmed FSM-MA as an E. meliloti strain and clearly separated it from E. arboris strain LMG14919 T , E. medicae strain WSM419 and the E. fredii strains NGR234, USDA257 and HH103. Among the E. meliloti strains, strain FSM-MA is most closely related to strains BO21CC and BL225C which were isolated from M. sativa nodules in Lodi, Italy [24]. Finally, the two-way average nucleotide identity [25] was calculated between genomes using the default settings of the ANI calculator. The genome of strain FSM-MA showed 99,42% identity with the genome of E. meliloti strain 1021 and only 90,09% identity with the genome of E. arboris strain LMG14919 T , 87,09% identity with the genome of E. medicae strain WSM419 and 83,16% and 83,31% identity with the genomes of E. fredii strains NGR234 and HH103, respectively. Once more this analysis showed that FSM-MA is an E. meliloti strain and not an E. arboris strain, considering a cut-off for species delineation at 95% identity [25].

Genome project history
This organism was selected for sequencing on the basis of its superior symbiotic performance [22] with the most widely used accessions (A17 and R108) of the model legume M. truncatula. The genome project and the sequence of the three replicons are deposited in the National Center for Biotechnology Information (NCBI; accession numbers: CP019584, CP019585, CP019586). Genome sequencing and sequence assembling were performed at the University of Malaya (Kuala Lumpur, Malaysia) and at the Seqomics Biotechnology Ltd. (Mórahalom, Hungary). Annotation was carried out at Seqomics Biotechnology Ltd. A summary of the project information can be found in Table 2.
Growth condition and genomic DNA preparation E. meliloti strain FSM-MA was grown on solid YEB medium (0.5% beef extract; 0.1% yeast extract; 0.5%peptone; 0.5% sucrose; 0.04% MgSO 4 .7H 2 O; pH 7.5) for 3 days and a single colony was used to inoculate 3 ml YEB broth medium. The culture was grown for 24 h on a gyratory shaker at 225 rpm at 30°C, then 0.5 ml of the starter culture was used to inoculate 50 ml YEB broth medium. The culture was grown at 30°C at 225 rpm until OD 600 = 0.6 was reached. DNA was isolated from the cells with the MasterPure Complete DNA and RNA Purification Kit (Epicentre). The integrity of the extracted genomic DNA was analyzed by 0.7% agarose gel electrophoresis. The final concentration of the DNA, estimated with the help of a Qubit Fluorometer (ThermoFisher Scientific), was 0.45 mg ml −1 .

Genome sequencing and assembly
The genome sequence of E. meliloti strain FSM-MA was generated using Pacific BioScience (PacBio) and Illumina technologies. An Illumina Mate Paired library (average insert length 7 kbp) was constructed and sequenced using the Illumina MiSeq platform, which generated 3,387,162 reads. Similarly, a PacBio SMRTbell library was constructed and sequenced on the PacBio RS II platform to generate 254,443 filtered reads (N50 value at 8643 bp and total bases at 1,726,776,880 bp). Assembly was then carried out using HGAP version 3 [26] yielding three contigs with an average coverage of 186.71×. Subsequently, Illumina reads were aligned to the PacBio assembly with the help of the CLC Genomics Workbench version 9.5 and the observed 17 InDels were corrected. The final assembly contains three circular contigs corresponding to the three replicons (the chromosome and the pSymA and pSymB megaplasmids) totaling 6,703,999 bp and total input read coverage was at 249.2×.

Genome annotation
Genes were identified and annotated using the NCBI Prokaryotic Genome Annotation Pipeline. The NCBI non-redundant database, UniProt, TIGR/Fam, Pfam, PRIAM, KEGG, COG, and InterPro databases were used to analyse the predicted coding sequences after translation. HMMER [27] and tRNAscan-SE [28] were used to identify the rRNA and tRNA genes, respectively.

Genome properties
The genome is 6,703,999 bp and comprised of three replicons (Table 3) with the size of 3,641,423 bp (chromosome), 1,422,736 bp (pSymA) and 1,639,840 bp (pSymB). The average GC content is 61.93%. Three rRNA operons, 67 RNA only genes were identified and 6583 protein coding genes were predicted in the genome. Five thousand thirty-two protein-coding genes were assigned a putative function and 1551 genes were predicted to code for hypothetical proteins ( Table 4). The distribution of genes in COG functional categories is presented in Table 5.

Insights from the genome sequence
The genome size of FSM-MA falls within the expected size range of 6.65-8.94 Mbp observed in the 33 Evidence codes -IDA: Inferred from Direct Assay; TAS: Traceable Author Statement (i.e., a direct report exists in the literature); NAS: Non-traceable Author Statement (i.e., not directly observed for the living, isolated sample, but based on a generally accepted property for the species, or anecdotal evidence). These evidence codes are from the Gene Ontology project [46] (http://geneontology.org/page/guide-go-evidence-codes) sequenced E. meliloti genomes that have been deposited in the Integrated Microbial Genomes (IMG) database. The genome of all E. meliloti strains is composed of a circular chromosome and two megaplasmids/chromids, however, certain strains harbour additional replicons too. In strain FSM-MA, however, no additional plasmid was detected. The strain contains three rRNA gene clusters as other E. meliloti strains do. Similarly to other Ensifer strains, the highest number of genes is assigned to the COG functional categories amino acid transport and metabolism (9.46%), carbohydrate transport and metabolism (8.65%) and transcription (7.76%). An enrichment of the COG functional categories amino acid transport and metabolism, transcription and signal transduction mechanisms is observed in pSymA, while carbohydrate transport and metabolism and cell wall/membrane/envelope biogenesis are overrepresented on pSymB (Table 5).

Extended insights
Comparing the FSM-MA genome structure with that of other E. meliloti strains using the Mauve software [29] revealed high co-linearity of the chromosomes and the pSymB megaplasmids in contrast to the pSymA plasmids that are highly variable. For example, the average sequence identity between FSM-MA and strain 1021 is 99.4% and their chromosomes and pSymB plasmids are essentially co-linear. The major differences between the chromosomes originated from the insertion of three putative prophages/insertion elements into the FSM-MA genome at genes coding for tRNAs (SMB554__06910: tRNA-Thr, SMB554_09150: tRNA-Lys, SMB554_16265: tRNA-Met). These inserted elements are of approximately 48, 43 and 44 kbp and contains 70, 54 and 34 predicted ORFs, respectively. In the putative prophages at tRNA-Thr and tRNA-Lys, among hypothetical  [30]. The tree with the highest log likelihood (−2208.71) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. Evolutionary analyses were conducted in MEGA7 [31] proteins, a number of phage related functions such as terminase, phage portal and capsid proteins (both prophages) as well as ORFs encoding endonucleases, transcriptional regulators, site-specific integrase, DNA ligase, peptidase or peptidoglycan-binding protein are encoded (prophage at tRNA-Lys). The inserted sequence at tRNA-Met seems to contain genes coding for type I restriction-modification system elements, an N 6 -DNAmethylase, chromosome segregation and AAA family ATPases as well as transcriptional regulators among hypothetical proteins. On the other hand, one putative prophage in the 1021 genome at a tRNA-Ser_CGA gene and the SMc01989-SMc02032 gene cluster coding for transcriptional regulators, membrane transporter and oxido-reductase elements are missing from the FSM-MA genome. The differences between the pSymB plasmids are mainly attributed to mobile genetic elements (IS elements, transposons) that are associated with strain-specific genes, essentially coding for proteins involved in the biosynthesis and transport of strainspecific LPS (lipopolysaccharide) and K-antigen (capsular polysaccharide) surface polysaccharides (discussed later). The pSymA plasmidsthat are the carriers of major symbiotic functions such as genes encoding Nod factor biosynthesis and the nitrogenase enzyme and co-factor biosynthesishave a number of co-linear blocks but have about 80 kbp size difference (FSM-MA > 1021), and more than 200 kbp (>1/7) of the sequences are absent in the other strain.
As the FSM-MA strain is interesting from the symbiotic point of view, we analysed those genes that are important for the development and functioning of the nitrogen-fixing symbioses. The initiation of the symbiotic interaction requires the production of Nod factors with proper chemical structure via the activity of the socalled Nod, Noe and Nol proteins. The FSM-MA genome contains all the known nod, noe and nol genes described in E. meliloti. The nif and fix genes code for the structural elements of the nitrogenase complex (nitrogenase, nitrogenase reductase, electron transport proteins) performing the reduction of atmospheric nitrogen as well as for proteins required for the biosynthesis of co-factors and the assembly of the   complex. All these genesincluding the ones that are present in multiple copies such as the three fix-NOQP operonscan be found in the FSM-MA genome. Notably, despite the high diversity of the E. meliloti pSymA plasmids harbouring these symbiotic genes, the arrangement and the genomic environment of the nodulation and nitrogen fixation genes in FSM-MA and strain 1021 are the same. Surface polyasaccharides play an essential role during the infection process [4] when bacteria enter the cells of the developing nodules via the infection threads. In the Medicago -E. meliloti symbiosis, the production of the succinoglycan exopolysaccharide is required for the continuous growth of the infection threads and its lack can be suppressed by the production of galactoglycan or certain capsular polysaccharides. Lipopolysaccharides might also affect both the infection and bacteroid differentiation processes. The organization and genomic environment of genes for the production and transport of the species-specific polysaccahrides EPS I (exo and exs genes) and EPS II (exp genes) as well as of the conserved part (lipidA and O-antigene core) of LPS (chromosomal and pSymB-born genes) and the KPS transporters are the same in the two E. meliloti strains. In contrast, the genes responsible for the production of the strainspecific polysaccharide moieties of LPS and KPS (Additional file 2: Figure S2), located on pSymB, are unique for the given strains.

Conclusions
The genome sequence of FSM-MA is of particular interest because the strain is highly effective with the most widely used ecotypes, Jemalong and R108 of the model legume M. truncatula. Comparative genomics with less and similarly effective strains as well as the creation and