Comparative Genomics between Two Xenorhabdus bovienii Strains Highlights Differential Evolutionary Scenarios within an Entomopathogenic Bacterial Species.

Bacteria of the genus Xenorhabdus are symbionts of soil entomopathogenic nematodes of the genus Steinernema. This symbiotic association constitutes an insecticidal complex active against a wide range of insect pests. Within Xenorhabdus bovienii species, the X. bovienii CS03 strain (Xb CS03) is nonvirulent when directly injected into lepidopteran insects, and displays a low virulence when associated with its Steinernema symbiont. The genome of Xb CS03 was sequenced and compared with the genome of a virulent strain, X. bovienii SS-2004 (Xb SS-2004). The genome size and content widely differed between the two strains. Indeed, Xb CS03 had a large genome containing several specific loci involved in the inhibition of competitors, including a few NRPS-PKS loci (nonribosomal peptide synthetases and polyketide synthases) producing antimicrobial molecules. Consistently, Xb CS03 had a greater antimicrobial activity than Xb SS-2004. The Xb CS03 strain contained more pseudogenes than Xb SS-2004. Decay of genes involved in the host invasion and exploitation (toxins, invasins, or extracellular enzymes) was particularly important in Xb CS03. This may provide an explanation for the nonvirulence of the strain when injected into an insect host. We suggest that Xb CS03 and Xb SS-2004 followed divergent evolutionary scenarios to cope with their peculiar life cycle. The fitness strategy of Xb CS03 would involve competitor inhibition, whereas Xb SS-2004 would quickly and efficiently kill the insect host. Hence, Xenorhabdus strains would have widely divergent host exploitation strategies, which impact their genome structure.

Introduction

Bacteria are found in all environments, from the soils to the inside of eukaryotic cells. Selective pressures exerted by the ecological niche shape the size and the content of bacterial genomes. For example, host-associated bacteria, both mutualistic and parasitic, usually undergo genome reduction, due to several evolutionary forces. In nonobligate pathogenic bacteria, purifying selection leads to the inactivation of genes encoding proteins interfering with bacterial virulence (i.e., antivirulence genes) (Bliven and Maurelli 2012; Prosseda et al. 2012). In the nutrient-rich and stable host ecological niches of obligate intracellular bacteria, relaxed selective pressure often leads to the pseudogenization resulting in degradation and loss of metabolism and regulation genes. Moreover, the populations of highly host-restricted bacteria are usually very small. In this context, both genetic drift and reduced horizontal genetic transfer (HGT) opportunities bias genomic evolution toward reduction (Ochman and Moran 2001; Moran and Plague 2004; Ochman and Davalos 2006).

Environmental species, such as soil or marine bacteria, have to deal with nutrient-poor environments. If this environment is stable, strong purifying selection favors minimization of cell and genome size, an evolutionary process named streamlining (Giovannoni et al. 2014). In contrast, the genome of bacteria living in changing and nutrient-poor environments contain several regulators, stress-resistance genes, and metabolism genes, which generates different sets of genes adapted to different sets of constraints (Ochman and Davalos 2006; Sanchez-Perez et al. 2008; Bratlie et al. 2010). Bacteria living in mixed-species environments also are subject to a strong competition in the nutrient exploitation. Genomes may hence harbor a great number of genes involved in secondary metabolism playing a role in synthesis of antimicrobial compounds. NRPS-PKS enzymes (nonribosomal peptide synthetases and polyketide synthases) have a central role in such metabolism (Challis and Hopwood 2003; Hibbing et al. 2010). As a consequence, within a given genus, the genomes from environmental bacteria are usually larger than the genomes of host-associated bacteria. As an example, unlike Mycobacterium tuberculosis or Mycobacterium leprae, Mycobacterium marinum, a marine bacterium and fish pathogen, has a large genome encoding several stress-resistance genes and a striking number of NRPS-PKS loci acquired during HGT events (Stinear et al. 2008).

How do genomes evolve when bacteria alternate between different ecological niches that exert opposite evolutionary pressures? The Enterobacteriaceae Xenorhabdus nematophila alternates between host-associated lifestyles, mutualistic, and pathogenic in the living invertebrate hosts, and in a mixed-species environment in the insect cadaver. Indeed, X. nematophila is an obligate symbiont of the entomopathogenic nematode Steinernema carpocapsae, which lives into the soils. The bacteria–nematode pair is pathogenic for a wide range of insects (Thomas and Poinar 1979; Herbert and Goodrich-Blair 2007). When the nematodes penetrate into the hemocoel (general cavity) of an insect larva, they release the X. nematophila bacteria from a specialized intestinal organ, the receptacle (Snyder et al. 2007; Chaston et al. 2013) into the insect hemolymph (equivalent of the blood). Xenorhabdus nematophila bacteria multiply into the hemolymph and kill the insect by septicemia and toxemia (Boemare 2002). The bacteria then degrade the insect tissues, enabling nematode maturation, and reproduction, while competing with soil microorganisms for nutrient acquisition in the insect cadaver. When the nutrient resources are depleted, the juvenile nematodes switch to the infective juvenile (IJ) stage, which reassociates with the bacteria, leaving the cadaver in search for a new host (Poinar 1990; Goodrich-Blair and Clarke 2007; Stock and Goodrich Blair 2008). Under laboratory conditions, the direct injection of X. nematophila into the hemocoel kills the insect host within 30 h (Sicard et al. 2004; Herbert and Goodrich-Blair 2007). The genome of X. nematophila consists of a medium-sized chromosome (4.43 Mb) and a 155 kb megaplasmid, and encompasses many genes potentially involved in the synthesis of insecticidal and in antimicrobial compounds (Ogier et al. 2010; Chaston et al. 2011). Abundant signs of recent genomic rearrangements suggest that alternation of ecological niches maintains an elevated genomic plasticity (Ogier et al. 2010).

In the same genus, the bacterial species Xenorhabdus poinarii displays similar cyclic and alternative way of life, but it is attenuated in virulence when directly injected into the insect hemolymph (Akhurst 1986; Ogier et al. 2014). Interestingly, the X. poinarii strain G6 harbors only a small 3.66 Mb chromosome and displays decay of isolated genes and excision of some genomic loci. These typical features of attenuated free pathogenic bacteria or facultative mutualistic bacteria growing exclusively within hosts are general traits for the whole species X. poinarii (Ogier et al. 2014). Therefore, the evolutionary scenario in the X. poinarii species is likely a reductive genomic evolution, the result of a high dependence on the associated nematode for some virulence-related functions and of a more advanced process of host restriction (Ogier et al. 2014).

In this study, we focused on the Xenorhabdus bovienii species, whose biological features are less homogeneous than the X. nematophila and X. poinarii ones. First, the monophyletic bacterial clade X. bovienii interacts with at least nine different nematode species from the Steinernema genus, belonging to two phylogenetic clades (Lee and Stock 2010). Second, the Steinernema spp.—X. bovienii pairs and the bacteria alone show highly variable virulence in lepidopteran insects (Bisch et al. 2015). Nevertheless, recent genomic analyses confirm the species status of the X. bovienii clade according to current standard criteria in microbial systematics (Murfin et al. 2015). These data suggest different evolutionary pressures within the X. bovienii species. Two Steinernema spp.—X. bovienii pairs have been recently studied. The Steinernema jollieti/X. bovienii strain SS-2004 pair was isolated in United States (Spiridonov et al. 2004). The 4.23 Mb chromosome of X. bovienii SS-2004 (Xb SS-2004) has been sequenced and shares numerous features with the X. nematophila one (Ogier et al. 2010; Chaston et al. 2011). The Steinernema weiseri—X. bovienii strain CS03 pairs was isolated in Czech Republic (Mráček et al. 2003). Both the bacteria–nematode pair and the bacteria alone show attenuated virulence in lepidopteran insects (Bisch et al. 2015). The bacteria–nematode pair and the bacteria alone displayed attenuated virulence and no virulence, respectively, in lepidopteran insects (Bisch et al. 2015). This suggests that the nematode S. weiseri is responsible for certain virulence functions, whereas, in other Steinernema–Xb associations, the bacteria are virulent on their own.

In this study, we sequenced the X. bovienii CS03 (Xb CS03) genome. We compared the genomes of Xb SS-2004 and Xb CS03. Considering the differences in genome size, and in pseudogene and gene content between the two strains, we suggest that Xb SS-2004 specializes in exploitation of the insect host, whereas Xb CS03 inhibits its competitors.

Materials and Methods

Bacteria, Nematodes, and Growth Conditions

Xb CS03 was isolated in Czech Republic (Europe) from the Steinernema weiseri 583 nematode (Mráček et al. 2003; Tailliez et al. 2006). The Xb SS-2004 used for the genome comparison was isolated in Missouri (Northern America) from the Steinernema jollieti nematode (Spiridonov et al. 2004; Chaston et al. 2011). Xb strains were grown at 28 °C. Serratia marcescens, Corynebacterium xerosis, Bacillus megatherium, Enterobacter cloacae, Micrococcus luteus, Ochrobactrum intermedium, Proteus vulgaris, and Pseudomonas putida were grown at 37 °C. All bacteria were routinely grown in Luria–Bertani (LB) broth or on nutrient agar (Difco). Kanamycin was added at 40 µg/ml when needed. Bacterial strains were stored at −80 °C with 16% glycerol (vol/vol).

Xenorhabdus Phenotype Assays

The antimicrobial activity of Xb CS03 and Xb SS-2004 against bacterial strains and their hemolytic activity were assessed as previously described by Boemare et al. (1997). Briefly, to assess antimicrobial activity, Xenorhabdus bacteria were grown for 48 h on nutrient ager plates and then killed by exposing the plates to chloroform for 20 min. The plates were left for 10 min under a sterile air flow to allow the chloroform to evaporate. Overnight cultures of bacteria were diluted (20%; OD540nm ∼ 0.1) in sterile soft agar (nutrient broth, 5.0 g; bacto-agar, 7.0 g; water, 1L) and poured onto the prepared plates. When Xenorhabdus and Photorhabdus strains, which are killed by warm sterile soft agar, were used, overnight cultures were diluted in LB broth. The diameter of the inhibition zone was recorded after 24 h of incubation at 37 °C (environmental bacteria) or 28 °C (Xenorhabdus and Photorhabdus). To assess their hemolytic activity, the Xenorhabdus bacteria were grown on trypticase soy (bioMérieux) with 5% (vol/vol) defibrinated sheep blood (bioMérieux). Hemolysis was defined as the observation of a clear halo around bacteria grown on standard sheep blood agar plates.

Sequencing and Assembly of Xb CS03 Genome

The complete genome of Xb CS03 was sequenced as previously described (Ogier et al. 2014) with 6531 Sanger reads, a 20-fold coverage of 454 GSflx reads, a 14-fold coverage of mate-paired 454 GSflx reads (library insert around 3 kbases) and a 110-fold coverage of Illumina reads (36 bp). For the finishing phases, primer walking of clones, polymerase chain reactions (PCRs), and in vitro transposition technology was used generating 844, 286, and 3,028 additional reads.

Genomic Analysis

Functional annotation of the Xb CS03 genome and analysis of the regions of genomic plasticity (RGP) of Xb CS03 and Xb SS-2004 were carried out as previously described (Ogier et al. 2010, 2014). We used the RGPFinder web tool implemented on the MaGe annotation platform to identify genomic regions displaying breaks in synteny between the Xb CS03 and Xb SS-2004 genomes. If the regions displayed characteristics typical of foreign DNA acquired by HGT, such as compositional bias (GC% deviation, codon adaptation index) or tRNA, IS, integrase genes, and genetic elements involved in DNA mobility, they were classified as genomic islands (GI) or prophages (P), if identified as such by Prophinder (Lima-Mandez et al. 2008). Regions without such features were classified as RGPsensu stricto. We used the SiLiX program (Miele et al. 2011) of the MicroScope platform (Vallenet et al. 2013) to describe the pan-genome, core genome, and flexible genomes inside the Xenorhabdus genus (homology constraints: 50% amino acid identity and 80% alignment coverage) or inside the Xb species (homology constraints: 80% amino acid identity and 80% alignment coverage). Strain-specific genes were manually classified according to their annotations.

Pseudogene Detection

Pseudogenes were identified using the Gene Fission tool of the MicroScope platform. This tool provides a list of candidate genes potentially involved in a fission event (predicted pseudogenes). Fission events are computed from the synteny results obtained from the genomes available in the PkGDB database. Predicted pseudogenes were manually inspected to remove false positives on the basis of the gene size and identity compared to reference sequences. Pseudogenes were classified as specific (i.e., pseudogenized only in the reference genome) or nonspecific (i.e., pseudogenized in both Xb). Functional categories were manually assigned to pseudogenes according their annotation.

NRPS-PKS Genes

NRPS-PKS loci were detected using the “2metDB” method (Bachmann and Ravel 2009) implemented in the MicroScope platform and manually curated on the basis of their annotation. We considered that a NRPS/PKS was conserved in Xenorhabdus and other bacterial species when they shared ≥50% amino acid identity. A locus was considered complete if the alignment coverage was at least 80% and partial otherwise.

Reverse Transcription-PCR

RNA extraction and cDNA synthesis were performed as described by Jubelin et al. (2013). Briefly, total RNA was extracted from a stationary phase culture (OD540nm∼ 1.5) of Xb CS03 with an RNeasy Protect Bacteria miniprep kit (Qiagen), including DNase I incubation, in accordance with the manufacturer's recommendations. The absence of DNA contamination was checked by carrying out a control amplification of the 16S rRNA gene. The quantity and quality of total and messenger RNA were assessed with a NanoDrop 2000 spectrophotometer (Thermo Scientific) and an Agilent 2100 Bioanalyzer with the RNA 6000 Nano LabChip kit (Agilent), respectively. The cDNA was synthesized from 1 µg of total RNA, with Super Script II Reverse Transcriptase (Invitrogen) and random hexamers (100 ng.µl−1, Applied Biosystems).

The primers used to amplify each xaxAB fragment are listed in supplementary table S1, Supplementary Material online. The PCR on Xb CS03 cDNA was realized with Taq polymerase (Invitrogen) according to the manufacturer recommendations, using a hybridization temperature of 57 °C and an elongation time of 90 s.

Construction of Recombinant Xb CS03

A XbaI-SalI fragment containing the xaxAB from X. nematophila F1 was hydrolyzed from the pBBxaxAB plasmid built in a previous study (Vigneux et al. 2007). The xaxAB fragment was then ligated into the medium copy plasmid pBBR1-MCS2 (Kovach et al. 1995) hydrolyzed with SalI and XbaI. The resulting plasmid, pBBMCS2-xaxAB, was checked by sequencing (Millegen, Labège, France). In this construction the xaxAB locus is under the control of the P promoter, whose transcriptional activity is constitutive in Xenorhabdus. Finally, pBBMCS2-xaxAB was introduced into Xb CS03 by mating as described in Givaudan and Lanois (2000).

Results

General Genomic Features of Xb CS03

Xb CS03 contains a 4,635,301 bp chromosome encoding 4,757 coding sequences (table 1). This is the largest chromosome described for a completely assembled Xenorhabdus strain to date, the size of the other chromosomes ranging from 3,659,522 to 4,432,590 bp (Chaston et al. 2011; Ogier et al. 2014). Xb CS03 also contains a 177 kb megaplasmid and an 8 kb plasmid (supplementary table S2, Supplementary Material online). In the Xb CS03 plasmid, 8 out of the 11 coding sequences are identical to coding sequences from the Xenorhabdus doucetiae plasmid (Ogier et al. 2014). Xb CS03 megaplasmid harbors 197 coding sequences, some of which encoding putative extracellular enzymes, putative antibiotic resistance cassettes, toxin–antitoxin systems, and NRPS/PKS. This content is largely different from the X. nematophila ATCC 19061 megaplasmid previously described (Chaston et al. 2011). The repA gene of the Xb CS03 megaplasmid (XBW1_mp0001) is homologous to the RepFIB replicon present in the large enterobacterial plasmids of the IncF and IncP incompatibility groups (Gibbs et al. 1993). In contrast, the repA gene of the X. nematophila megaplasmid (XNC1_p0159) is homologous to the RepAC replicon from plasmids belonging to the IncA/C incompatibility group (Llanes et al. 1996). These data indicate that these two plasmids have different origins. In contrast, Xb CS03 megaplasmid shares 61, 57, and 44 orthologous coding sequences with the chromosome of X. nematophila ATCC19061, Xb SS-2004, and Xb CS03, respectively (best bidirectional hit, 35% amino acid identity, 80% alignment coverage). This suggests an elevated intragenomic flow between megaplasmid and chromosome, rather than an entire megaplasmid horizontal genetic transfer between Xenorhabdus strains.

Table 1

Genome Features of X. bovienii CS03 (Xb CS03) Compared with X. bovienii SS-2004 (Xb SS-2004)

	Xb CS03			Xb SS-2004
Feature	Chromosome	Megaplasmid	Plasmid	Chromosome
Size (bp)	4,635,301	177,250	8,117	4,225,498
G+C content (%)	44.77	43.89	43.87	44.97
Coding sequences	4,757	197	11	4,362
Coding density (%)	83.63	82.82	62.95	84.92
Average CDS length (bp)	843.62	799	502.36	843.05
Average intergenic length (bp)	162.37	145.51	422	158.16
Repeated regions (%)	14.69	7.24	0	16.27
Pseudogenes	99	3	0	65
Insertion sequencesa	298	29	0	369
Phagic genesb	454	119	0	450
Prophage locib	10	1	0	8
rRNA operons	7	0	0	7
tRNAs	76	0	0	83
Accession number	FO818637	FO818638	FO818639	FN667741

aidentified with the IS Saga tool (Varani et al. 2011).

bidentified with Prophinder (Lima-Mendez et al. 2008).

We previously compared four Xenorhabdus genomes. Pan and core genomes were composed of 7,250 and 1,904 gene families, respectively (Ogier et al. 2014). The addition of the Xb CS03 genome in this analysis (homology constraints: 80% amino acid identity and 80% alignment coverage) does not modify drastically the size of the Xenorhabdus core genome (1,673 gene families; fig. 1). However, the pan genome size increased with the Xb CS03 genome (10,600 gene families; fig. 1), which confirms the diversity of the flexible genome (pan genome minus core genome) of the Xenorhabdus genus (Ogier et al. 2010).

F

Venn diagram showing numbers of shared gene families (80% amino acid identity, 80% alignment coverage) in the genomes of X. bovienii CS03, X. bovienii SS-2004, X. nematophila ATCC19061, X. doucetiae FRM16, and X. poinarii G6.

Genomic Comparison within the Xb Species: General Features

Comparison of general genomic features between Xb CS03 and Xb SS-2004 genomes highlights two main differences (table 1). The Xb SS-2004 genome is slightly richer (24%) in insertion sequences than the Xb CS03 genome, and the number of pseudogenes in the Xb CS03 genome is 52% higher than in the Xb SS-2004 genome. Interestingly, among the 99 Xb CS03 pseudogenes, 40 matched with intact genes from Xb SS-2004, whereas among the 65 Xb SS-2004 pseudogenes, only 17 matched with intact genes in Xb CS03 (supplementary table S3, Supplementary Material online). Hence, while the pseudogenes of Xb SS-2004 mainly belong to the Xb core-genome (i.e., degraded also in Xb CS03), 40% of Xb CS03 pseudogenes are strain-specific. We classified the pseudogenes into nine functional categories (Cell division, DNA/RNA metabolism, Host/Environment interaction, Metabolism, Phage/Recombination, Protein biosynthesis, Regulation, Transport, and Unknown function) (fig. 2). Xb SS-2004 is particularly rich in pseudogenes from the “Phage/Recombination” category (32% of the pseudogenes). Phages and transposases may undergo a selective pressure which leads to their pseudogenization and deletion in bacterial genomes (Lawrence et al. 2001). This phenomenon could account for the pseudogenization of such genes in Xb SS-2004. However, the main difference is in the “Host and Environment Interaction” category, which represents 39% of the total pseudogenes content in Xb CS03 and only 26% of the Xb SS-2004 pseudogenes.

F

Functional classification of pseudogenized genes in X. bovienii CS03 and X. bovienii SS-2004. (A) Functional classification of X. bovienii CS03 pseudogenes. (B) Functional classification of X. bovienii SS-2004 pseudogenes.

Xb Core and Flexible Genome

We compared the Xb CS03 and Xb SS-2204 genomic contents. The Xb core-genome and the Xb specific core-genome (which excludes the genes also found in X. nematophila ATCC19061, X. doucetiae FRM16 and X. poinarii G6) were composed of 2,673 and 1,000 genes, respectively (fig. 3A and supplementary table S4, Supplementary Material online). Xb CS03 and Xb SS-2004 flexible genomes contained 1,815 and 1,267 strain-specific genes, respectively (fig. 3A and supplementary table S4, Supplementary Material online).

F

Xenorhabdus bovienii pan, core, and flexible genomes. (A) Venn diagram showing numbers of shared gene families (80% amino acid identity, 80% alignment coverage) in the genomes of X. bovienii CS03 and X. bovienii SS-2004. The Xb core genome (2,673 gene families) includes the Xenorhabdus core genome (1,673 gene families common to X. nematophila ATCC19061, X. doucetiae FRM16, and X. poinarii G6) and the specific X. bovienii core genome (1,000 gene families). (B) Functional classification of coding sequences in X. bovienii SS-2004 and X. bovienii CS03 flexible genomes. The coding sequences of each flexible genome have been classified into nine functional categories (Cell division, DNA/RNA metabolism, Host/Environment interaction, Metabolism, Phage/Recombination, Protein biosynthesis, Regulation, Transport, and Unknown function) according to their annotation. (C) Specific host and environment interaction genes functions for the X. bovienii SS-2004 (red) and X. bovienii CS03 (green) flexible genomes. The genes belonging to the flexible genome of Xb CS03 and Xb SS-2004 and encompassed within the Host and environment interaction category were classified according to their annotation. In the histogram, the pseudogenes are indicated in lighter color. RM: Restriction-Modification; TA: toxin-antitoxin; Rhs: Rearrangement hotspots.

We examined the structure of Xb CS03 and Xb SS-2004 flexible genome by searching for RGP. These regions were identified as GI, prophage loci (P), or uncharacterized regions (RGPsensu stricto), according to the criteria used in a previous study (Ogier et al. 2010, 2014). We identified more RGP in Xb CS03 flexible genome (69 chromosomal RGP encompassing 14 GI, 10 P, and 45 RGPsensu stricto; 25% of the genome) than in Xb SS-2004 flexible genome (55 chromosomal RGP encompassing 11 GI, 8P, and 36 RGPsensu stricto; 20% of the genome; supplementary table S5, Supplementary Material online). One additional prophage was located on the Xb CS03 megaplasmid. We identified one of the Xb CS03 prophages (XBW1_1585-1616) as similar to the xenorhabdicin-encoding phage xbp1 of Xb SS-2004 (Morales-Soto et al. 2012) and named it xbwp1 (supplementary tables S5 and S6, Supplementary Material online). One or several of the RGP may be embedded in an integrative conjugative element (ICE; (Ogier et al. 2010; Guglielmini et al. 2011). Four ICEs were identified in Xb CS03, whereas two ICEs had previously been described in Xb SS-2004 (Chaston et al. 2011; Ogier et al. 2014). Only ICE1 was complete (supplementary table S7, Supplementary Material online). In summary, these results show that mobile genetic elements make up a large proportion of the flexible gene content of Xb CS03, substantially contributing to its larger genome size compared with Xb SS-2004. Classification of the flexible gene content into the functional categories defined above resulted in similar distributions for Xb CS03 and Xb SS-2004 with a slight difference in the “Host/Environment interaction” and “Unknown function” classes (fig. 3B).

The Xb CS03 genome therefore displays two remarkable features by comparison with Xb SS-2004: a large flexible genome and an important content of pseudogenes specific to Xb CS03 (belonging to the flexible genome). Furthermore, the coding sequences that underwent pseudogenization in the Xb CS03 genome are frequently annotated as potentially involved in the host and environment interactions. We thus focused our follow-up analyses on the flexible gene content falling into the functional class Host and environment interactions.

Xb CS03 Wealth in Antimicrobial Compounds Synthesis Loci Correlates with a Strong Antimicrobial Activity

The nature of the flexible genome coding sequences classified in the Host and environment interaction category is largely different between Xb CS03 and Xb SS-2004 (fig. 3C and supplementary table S8, Supplementary Material online). First, Xb CS03 contains more coding sequence potentially involved in resistance to DNA invasion, such as restriction–modification systems for resistance to bacteriophage invasion (Kobayashi 2001) and toxin–antitoxin systems for resistance to mobile genetic elements, such as GIs and plasmids (Schuster and Bertram 2013).

The flexible genome of Xb CS03 is also particularly rich (42% of the coding sequences) in loci involved in the synthesis of secondary metabolites with potential antimicrobial activity (fig. 3C and supplementary table S8, Supplementary Material online). A major part of those loci encode NRPS/PKS (58% of the Host and environment interaction genes contained in the flexible genome of Xb CS03, vs. 10% in Xb SS-2004). Among the NRPS-PKS loci specific to Xb CS03, three (XBW1_0736-0742, XBW1_1481-1490, and XBW1_3836-3844) belonged to RGPsensu stricto (respectively, RGPs 12, 24, and 60) and one was an isolated gene (XBW1_3328). Those NRPS/PKS loci were present in other Xb strains as well as in other Xenorhabdus species (supplementary table S9, Supplementary Material online), which indicates that they could have been excised from the genome of Xb SS-2004.

We compared the antimicrobial activity of Xb CS03 and Xb SS-2004 against several potential bacterial competitors: Xenorhabdus and Photorhabdus strains, but also P. vulgaris, Se. marcescens, C. xerosis, B. megatherium, E. cloacae, O. intermedium, and Ps. putida, bacterial species that have been found in association either with the soil, with the insect G. mellonella or with entomopathogenic nematodes Steinernema (Lysenko and Weiser 1974; Boemare et al. 1983; Aguillera and Smart 1993; Gouge and Snyder 2006). Micrococcus luteus was used as reference strain to assess antimicrobial activities (Boemare et al. 1997). When tested against strains from the closely related genera Xenorhabdus and Photorhabdus, both Xb CS03 and Xb SS-2004 were able to inhibit bacterial growth, although the patterns of inhibition were different (table 2). When tested against phylogenetically distant bacteria, Xb CS03 displayed antimicrobial activity against more than half of the tested strains (C. xerosis, B. megatherium, E. cloacae, O. intermedium, and M. luteus), whereas Xb SS-2004 was active only against the highly sensitive strain M. luteus (table 2).

Table 2

Antimicrobial Activity of X. bovienii CS03 and SS-2004 against Bacterial Species Associated with the Insect or the Nematode

	Xb CS03	Xb SS-2004
Xb CS03	0	0
Xb SS-2004	14–17	0
Xenorhabdus bovienii TR03	0	10–25
Xenorhabdus bovienii FR44	0	15–20
Xenorhabdus bovienii TB10	0	0
Xenorhabdus nematophila ATCC19061	15	16–18
Xenorhabdus doucetiae FRM16	0	0
Xenorhabdus poinarii G6	0	0
Photorhabdus luminescens TT01	28–30	0
Proteus vulgaris CIP 5860	0	0
Serratia marcescens	0	0
Corynebacterium xerosis	22–25a	0
Bacillus megatherium	41	0
Enterobacter cloacae	10–20	0
Micrococcus luteus	55	15
Ochrobactrum intermedium LMG 3301	0	0
Pseudomonas putida	10–15a	0
Stenotrophomonas maltophilia	0	0

Note.—At least two independent experiments were performed, except for Bacillus megatherium and Stenotrophomonas maltophilia (one experiment). Target strains come from our laboratory collection. Antimicrobial activity is given as the diameter of the clearing halo around the X. bovienii colony (in mm).

aPresence of colonies into the clearing halo.

Xb CS03 also contains several loci encoding non-NRPS-PKS enzymes, which may catalyze the production of antifungal metabolites (fig. 3C and supplementary table S8, Supplementary Material online). For example, a locus displaying similarity to subregions of the Streptomyces tendae nikkomycin synthesis locus (XBW1_3291-XBW_3307) was present only in Xb CS03 (fig. 4). Nikkomycin has antifungal activity against Saccharomy cescerevisiae and the pathogenic fungi Candida albicans and Cryptococcus neoformans (Cabib 1991; Hector 1993). In Xb CS03, the nikkomycin-like locus did not have the typical features of a GI, but was located in an RGPsensu stricto (RGP56). As the nikkomycin locus is absent from Xb SS-2004, but also from X. nematophila ATCC19061, X. doucetiae FRM16, and X. poinarii G6 genomes, it was probably acquired by the Xb CS03 genome.

F

Xenorhabdus bovienii CS03-specific nikkomycin-like locus. The figure shows the nikkomycin-like locus of Xb CS03 (bottom), located between two genes of the core genome (XBW1_3290-3309). In the Xb SS-2004 genome (top), no genomic locus is inserted between the orthologous genes (XBJ1_0800-0801). The locus responsible for nikkomycin biosynthesis in Streptomyces tendae contains 22 genes (nikA-V). In Xb CS03, only subregions are present: the nikIJKLMO genes responsible for the synthesis of aminohexuronic acid, and the nikS gene encoding an enzyme responsible for the loading of the precursor of the hydroxypyridylhomothreonine amino acid to the bidomain protein NikT. Nikkomycin-like locus genes are shown in green, putative regulators in orange, transposases and insertion sequences in yellow, genes of unknown function in gray, and core-genome genes in white.

Pseudogenization of Genes Encoding Putative Virulence Factors in the Xb CS03 Flexible Genome

At first glance, Xb CS03 flexible genome displayed less coding sequences involved in bacterial invasion and iron metabolism. Additionally, one could note that several Xb CS03 genes encoding transcriptional regulators but also extracellular enzymes, proteins involved in the host invasion, and toxins were pseudogenized (fig. 3C).

As a first example, Xb CS03 tc loci were shuffled, and either highly degraded or interrupted by transposases. Consequently, no complete tc locus was found in Xb CS03. By contrast, there were three and one whole tc locus (encoding A, B, and C components) in X. nematophila ATCC19061 and Xb SS-2004, respectively (fig. 5). The tc loci of Xenorhabdus encode families of insecticidal toxins conserved in several other entomopathogenic bacteria (Hinchliffe et al. 2010). Some proteins of the Tc complex display high oral toxicity in insects (Bowen et al. 1998), in which they disrupt the insect gut epithelium (Sheets et al. 2011) and inhibit the phagocytosis of insect hemocytes (Lang et al. 2010).

F

The tc toxin loci of X. nematophila ATCC19061 (A), X. bovienii SS-2004 (B), and X. bovienii CS03 (C). Genes encoding the A component of the Tc toxin are colored in brown, the B component in blue, the C component in red; transposases and insertion sequences are colored in yellow, genes of unknown function in gray, and core-genome genes in white. Gene fragments are indicated by hatching.

The xaxAB locus is also an interesting example of pseudogenization. This locus was previously described in X. nematophila and encodes the hemolysin XaxAB, a binary pore-forming toxin with apoptotic and necrotic activity toward insect cells and probably required during the degradation of the insect cadaver in X. nematophila (Jubelin et al. 2011). In Xb SS-2004, the genomic locus xaxAB (XBJ1_1710-1711) displayed the same organization as in the X. nematophila genome and the products share, respectively, 74% and 69% of identity with the XaxA and XaxB proteins. In contrast, in Xb CS03, both genes were pseudogenized: xaxA (XBW1_1010-1032) was interrupted by a prophage insertion and xaxB (XBW1_1009-1010) was split in two fragments (fig. 6A). In vitro, XaxAB displays a hemolytic activity against sheep erythrocytes (Brillard et al. 2001). We tested the hemolytic activity of Xb CS03 and Xb SS-2004 on sheep red blood agar plates (fig. 6B). Xb CS03, contrarily to Xb SS-2004, did not display a hemolysis halo around the bacteria.

F

The xaxAB locus and the hemolytic activity of X. bovienii CS03. (A) The xaxAB loci of X. bovienii SS-2004 (top) and X. bovienii CS03 (bottom). For each xaxAB gene remnant of X. bovienii CS03, the size is indicated. The xaxAB genes are colored in red, the prophage in purple, transposases and insertion sequences in yellow, genes of unknown function in gray, and core-genome genes in white. (B) Hemolytic activity of X. bovienii SS-2004 (bottom) and X. bovienii CS03 (top). The indicated strains were streaked onto sheep red blood agar plates and incubated at 28 °C. Zones of clearing were observed over a 15-h period. (C) Electrophoretic separation of the products of reverse transcription-PCR on gene remants of the X. bovienii CS03 xaxAB locus. The amplified fragments were separated on a 1% agarose gel. The lane numbers correspond to the fragments described in figure 6. Lane 1: 3′- fragment of xaxA; lane 2: 5′-fragment of xaxA; lane 3: 3′-fragment of xaxB; lane 4: 5′-fragment of xaxB. (D) Hemolytic activity of X. bovienii CS03 pBBR1-MCS2 (top) and X. bovienii CS03 pBBMCS2-xaxAB (bottom). Aliquots (20 µl) of cultures of the indicated strains were spotted onto sheep red blood agar plates and incubated at 28 °C. Zones of clearing were observed over a 15-h period.

In order to test if the observed pseudogenization had an impact on the transcription of the gene remnants, we conducted reverse transcription-PCR with primers targeting each of the four remnants of the Xb CS03 xaxAB locus on stationary phase cultures. The 5′-fragments of both xaxA and xaxB remained transcribed, whereas the 3′-fragments were not (fig. 6C). To test the ability of Xb CS03 to express and secrete a functional XaxAB hemolysin, we introduced the pBBMCS2-xaxAB plasmid, which constitutively expresses the xaxAB locus from X. nematophila F1, in Xb CS03. The hemolytic activity of the Xb CS03/pBBMCS2- xaxAB transconjugant was complemented (fig. 6D). Altogether, these data suggest that the pseudogenization of the xaxAB locus is recent.

Discussion

The genomes of bacteria undergo changes due to the selective pressures exerted by their lifestyles, either host-associated or environmental. In this study, we explored the genome structure and content of Xb CS03. As in other Xenorhabdus strains, the life cycle of Xb CS03 alternates between close association with the nematode and the insect, and environmental steps in which it must compete with other microbial species. However, unlike many Xenorhabdus strains, Xb CS03 alone and Xb CS03 associated with its nematode S. weiseri were nonvirulent and displayed attenuated virulence, respectively, in lepidopteran larvae (Bisch et al. 2015).

Xb CS03 has the largest genome of any of the Xenorhabdus strains sequenced to date. Large genomes are generally observed in species or strains retaining a capacity for environmental survival and/or acting as pathogens in a broad range of hosts, such as My. marinum, which has a larger genome than the pathogen My. tuberculosis (Stinear et al. 2008). Nevertheless, their weak virulence toward lepidopteran insects (Bisch et al. 2015) is not consistent with a broad host spectrum for Xb CS03, or for the S. weiseri–Xb CS03 pair. The Xb CS03 strain is also characterized by a large flexible genome, which may be highly dynamic, as already described for X. nematophila and X. bovienii, and for the phylogenetically close entomopathogenic bacterial genus Photorhabdus (Ogier et al. 2010; Murfin et al. 2015). The Xb CS03 genome contains more pseudogenes than the genome of Xb SS-2004, the only other available whole assembled genome for the Xenorhabdus bovienii species. This feature is surprising because Xb CS03 has a very large genome, whereas pseudogene proliferation is generally considered to be a hallmark of the initial stages of genomic reduction (Moran and Plague 2004). These genomic structure data reveal considerable differences from the genome of X. poinarii G6, another strain with attenuated virulence. X. poinarii G6 has a small genome with rare pseudogenes or insertion sequences. The observed genome reduction in this strain was mediated by the excision of genomic blocks from the flexible genome of the X. poinarii ancestor genome (Ogier et al. 2014). The pathologic and genomic properties of Xp G6 are common to all X. poinarii strains described to date (Akhurst 1986; Ogier et al. 2014). Furthermore, the association of X. poinarii with its nematode, Steinernema glaseri, is virulent (Akhurst 1986; Converse and Grewal 1998; Rosa et al. 2002; Ansari et al. 2003). In light of these data, we conclude that the evolutionary history of the genome of Xb CS03 differs radically from that of the species X. poinarii.

We investigated the content of the Xb CS03 genome further, by comparing genes encoded by Xb CS03 and Xb SS-2004. Xb CS03 was found to contain more genes potentially involved in interactions with other microbial competitors than Xb SS-2004: loci encoding restriction–modification systems, toxin–antitoxin systems, enzymes catalyzing the production of secondary metabolites, including numerous NRPS-PKS enzymes (Bode 2009; Singh et al. 2015). However, both strains have an xnp1-like locus encoding a tail-phage bacteriocin responsible for antimicrobial activity against bacteria phylogenetically close to Xenorhabdus (Morales-Soto et al, 2011). These genomic data were consistent with the pattern of antimicrobial activity in Xb CS03, which targets phylogenetically distant bacterial species associated with the soil and insects rather than phylogenetically close bacteria. Competitor inhibition is an important strategy in the life cycle of Steinernema–Xenorhabdus pairs. If Xenorhabdus is able to persist for several weeks in the cadaver of the insect before reassociating with new generations of IJs and to protect the insect cadaver from competing soil microorganisms, it must produce numerous antimicrobial compounds (Nielsen-LeRoux et al. 2012). As S. weiseri emerges more slowly than other Steinernema–Xenorhabdus pairs (Bisch et al. 2015), competitor inhibition is probably crucial for the successful reassociation and emergence of the S. weiseri-Xb CS03 pair.

The second main difference between the Xb SS-2004 and Xb CS03 genomes is that the former contains a larger number of genes potentially involved in the exploitation of host resources. These genes include genes encoding virulence factors, as described in the virulence database (VFDB, http://www.mgc.ac.cn/VFs/): extracellular enzymes, secretion systems, toxins, adhesins, invasins, and iron-acquisition systems (Chen et al. 2005). In Xb CS03, several of these virulence loci (e.g., tc and xaxAB) have been pseudogenized by the insertion of transposons or prophages. This pattern of inactivation probably results from a relaxation of positive selection (Mira et al. 2001; Ochman and Davalos 2006).

According to the transmission-virulence trade-off hypothesis, the evolution of pathogens is driven by a positive link between virulence and the capacity to contaminate new hosts (Alizon and Michalakis 2015). A trade-off between efficient exploitation of the insect and inhibition of the competitive microorganisms in the insect cadaver may occur during the Steinernema–Xenorhabdus life cycle. Based on the differences between the genomes of Xb SS-2004 and Xb CS03 described here and previous descriptions of differences in their pathogenesis (Bisch et al. 2015), we suggest that this trade-off is heavily biased in Xb SS-2004 and Xb CS03. The fitness strategy of Xb SS-2004 seems to be based on efficient host exploitation, whereas that of Xb CS03 seems to rely more heavily on the inhibition of competitors.

What kind of evolutionary history could explain the specialization of Xb CS03 in competitor inhibition? We propose two nonmutually exclusive hypotheses. First, within the nematode–bacterium pair, the nematode partner S. weiseri 583 might be responsible for some of the virulence functions. For example, Xb CS03 is sensitive to antimicrobial peptides (Bisch et al. 2015). Steinernema weiseri may protect its bacterial symbiont by counteracting this component of the insect immune system, as has been demonstrated for other Steinernema species (Wang and Gaugler 1999; Brivio et al. 2006; Li et al. 2009; Liu et al. 2012). In this configuration, Xb CS03, the bacterial partner of the pair, would be able to specialize in the acquisition of genomic resources involved in microbial competition. Second, Xb CS03 may be evolving toward a saprophytic lifestyle. Indeed, Steinernema–Xenorhabdus pairs occasionally infest dead insects (San-Blas and Gowen 2008; Půza and Mrácek 2010). If the S. weiseri 583-Xb CS03 pair has developed the ability to multiply in insect cadavers, then outcompeting other microorganisms by producing a wide range of antimicrobial compounds may be crucial to its success. For both these hypotheses, an inability to kill the insect would be an evolutionary consequence. Indeed, maintaining a large set of synthetic enzymes involved in antimicrobial compound production is demanding in terms of energetic expenses, and probably leads to strong purifying selection on genes not essential for saprophytic microorganisms, such as the genes encoding virulence factors.

In conclusion, our analysis suggests that, within the Xb species, strains have had very different evolutionary genomic histories. To date, only a few genomic scenarios have been described for bacteria alternating between different ecological niches exerting opposite evolutionary pressures. We suggest that this configuration generates a multiplicity of genomic patterns and, consequently, a multiplicity of strategies for interaction with hosts.

Supplementary Material

Supplementary tables S1–S9 are available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).