Genomic Features and Assembly Metrics of the SC 0516 strain
In the present study, bacterial strain SC 0516 was isolated from the hemolymph of infected larvae as well as from a macerate of infective juvenile Steinernema carpocapsae larvae. In addition, the bacterial genome of SC 0516 was sequenced and assembled de novo. The draft genome of strain SC 0516 was 4,179,879 bp in length, with 142-fold genome coverage and an approximately 43.5% G+C content. A total of 103 contigs were generated, with an N50 contig length of 73 255 bp and an L50 value of 16 contigs. No plasmids were detected in the analysis.
To functionally analyze the genome of strain SC 0516, the contigs were subjected to subsystem annotation on the RAST server. A total of 4174 protein-coding genes, including 51 tRNAs, 5 rRNAs and 5 noncoding RNAs, were assigned to 336 annotated subsystems, which can be defined as biological processes that are components of metabolism or structural complexes supported by a set of functional roles. Most of these genes were predicted to be involved in the metabolism of amino acids and derivatives (309 genes), proteins (189 genes) and carbohydrates (150 genes).
This whole-genome shotgun project has been deposited in GenBank under accession number JACDOS000000000.1. The version described in this article is the first version.
Pangenome and phylogenetic analysis of Xenorhabdus nematophila SC 0516
A pangenome consists of core genes (i.e., genes found in all strains of a genus or species), flexible genes (found in more than one strain but not all strains), and single genes (found in only one strain, also known as accessory genes). We included 77 drafts and 14 complete genomes of Xenorhabdus and Photorhabdus bacteria in the analysis (Supplementary Data 1). As a selection criterion, genomes that were highly fragmented (> 300 contigs) were excluded. The consensus pangenomic matrix of Xenorhabdus and Photorhabdus from our dataset consisted of 23,603 gene clusters (Figure 1A). More than half of the complete set of genes constituting the pangenome (15,662 genes, 66%) were found to be uniquely present (referred to as a cloud genome), meaning that each strain contributed approximately 172 new genes on average to the pangenome. The shell genome (genomes ≥ 50%) and the soft core (genomes= 95%) consisted of 6,311 and 1,630 gene clusters, respectively (Figure 1A). This analysis showed that the pangenome of both genera has an open form (Figure 1B). This occurs when the number of new gene families continues to increase in a taxonomic lineage and this increase does not appear to be asymptotic, regardless of how many new genomes are added to the pangenome; higher rates of gene gains by horizontal gene transfer (HGT) are characteristic of these lineages (16, 17). The core genome was characterized as the set of genes present in all the genomes analyzed. We established that the core genome contained approximately 348 genes that were present in all 91 genomes studied, representing 1.4% of the pangenome (Figure 1C). However, the core genome presented a negative trend due to the addition of new strains, as the probability of gene sharing between strains decreased as new strains were incorporated into the study sample (Figure 1D). This was consistent with studies revealing a general negative relationship between pangenome size and the proportion of core genes, where larger "open" pangenomes have a lower proportion of core genes (18). Overall, the Xenorhabdus-Photorhabdus pangenome showed a high level of genome variability, with only 1% of its genome being constant. Thus, the remaining 66% of the pangenome is presented as a variable piece of DNA composed of a wide repertoire of genes and molecular functions.
To our knowledge, this was the first pangenomic analysis of the Xenorhabdus and Photorhabdus genera in which the core genomes were examined using BDBH, COG and OMCL strategies. Based on most of the genes that are part of the core genome grouping according to the NCBI Clusters of Orthologous Groups (COGs) database, a high percentage of the genes are involved in the generation and conversion of energy processes (29.35%), translation, ribosome structure and biogenesis (19.18%), amino acid transport and metabolism (12.40%) and replication, recombination and repair (11.61%), as shown in Figure 2.
The taxonomic position of strain SC 0516 was assessed using a maximum likelihood core genome phylogeny calculated from the 281 highest-scoring markers selected by the GET_PHYLOMARKERS pipeline from the 348 consensus groups calculated by GET_HOMOLOGUES. Three complete genomes of Yersinia pestis were included in the phylogenetic analysis. The core genome phylogeny showed that our strain, SC 0516, was clustered with the species X. nematophila, as shown in Figure 3. Strain SC 0516 is closely related to X. nematophila YL001 (GenBank accession number CP032329.1), which was isolated from nematodes from a soil sample collected in the locality of Shanxi, China. The strength of the phylogenetic tree obtained using this core genome-based approach resolved clades with maximum support.
In addition, this analysis allowed us to locate genus-specific genes of Xenorhabdus and Photorhabdus, highlighting those genes involved in pathogenicity and virulence. Overall, we found that the genus Xenorhabdus did not possess unique elements related to pathogenicity and virulence. However, the distinctive genes present in the genus included genes associated with tellurium resistance and polyamine transport. On the other hand, the genus Photorhabdus presented 29 unique elements associated with virulence and pathogenicity related to the type III secretion system, pilus structures and fimbriae (Supplementary Data 2). This analysis was consistent with previous observations that the genus Photorhabdus possesses genes related to the type III secretion system that are absent in Xenorhabdus; these genes play an important role in host insect invasion as well as the secretion of toxins and various virulence factors (10, 11, 19, 20). These results reveal that, although the members of these genera are phylogenetically related bacteria with similar lifestyles, they may differ drastically in their molecular mechanisms of pathogenicity in the same host, and Xenorhabdus probably uses different effectors and secretion systems, which could be reflected in the differences in virulence capacity between the two genera.
Evaluation of pathogenicity in Galleria mellonella
To evaluate the pathogenicity of our strain, X. nematophila SC 0516, survival experiments were performed in G. mellonella by injecting different doses of colony forming units (CFUs). For a comparative analysis of pathogenicity, we used a bacterial strain previously characterized in our working group, identified as Photorhabdus luminescens HIM3 (21). Bioassays showed that both bacteria were pathogenic to G. mellonella larvae, as demonstrated by the mortality rate visualized in Kaplan–Meier survival curves, which differed significantly from the control (E. coli DH5α) (X2= 457.636, df= 6, and p <0.001), as shown in Figure 4. This analysis revealed that X. nematophila SC 0516 was more virulent than P. luminescens HIM3 at all doses used (Figure 4). This difference was also illustrated by the median survival times (the time for which 50% of the larvae survive) estimated from the Kaplan–Meier survival analysis. In the case of X. nematophila SC 0516, the median times were 26.4 and 24 hours for doses of 102 and 103 CFUs, respectively, and 30.4 hours for a dose of 101 CFUs. However, P. luminescens HIM3 showed median times of 37.2 and 36 hours for doses of 102 and 103 CFUs, respectively, and 41 hours for a dose of 101 CFUs (Table 1). We also observed differences in the phenotypes of the dead larvae. In the larvae treated with X. nematophila SC 0516, a slight darkening of the body was observed after 48 hours of treatment, while those treated with P. luminescens HIM3 developed a dark reddish color at the same time. No external symptoms were observed in the control larvae (treated with E. coli DH5α) (Supplementary Figure S1).
Table 1
Survival time (h) of Galleria mellonella larvae injected with different doses of colony forming units (CFUs) of X. nematophila SC 0516 and P. luminescens HIM3.
|
101 CFU
|
102 CFU
|
103 CFU
|
E.coli DH5α
|
-
|
-
|
48 A
|
P.luminescens HIM3
|
41 ± 0.771 B
|
37.2 ± 0.469 B
|
36 B
|
X. nematophila SC 0516
|
30.4 ± 0.779 C
|
26.4 ± 0.625 C
|
24 C
|
Log-Rank test:457.636, df: 6, p <0.001.
Survival time on the same column followed by the same letter are not significantly different (Holm-Sidak test, p>0.05)
|
To explain these differences, we looked for unique virulence- and pathogenicity-related elements present in the genomes of both bacteria. Our analysis showed that P. luminescens HIM3 presented a higher number and diversity of pathogenicity- and virulence-associated components than X. nematophila SC 0516 in almost all categories (Figure 5). The only unique elements presented by X. nematophila SC0516 were the mcf (makes caterpillars floppy) toxin (reported in the literature as a highly insecticidal toxin) and a higher number of genes associated with the type IV secretion system (T4SS) and the type I toxin-antitoxin system (Figure 5). An et al., 2009 compared the gene expression of Photorhabdus temperata and Xenorhabdus koppenhoeferi in vivo in the insect Rhizotrogus majalis and found that more than 60% of the genes were uniquely induced in one of the two bacteria (22). However, in Xenorhabdus bacteria, the mechanisms of toxin delivery to the insect are not completely known, although some mechanisms of secretion through the flagellar apparatus or by vesicular systems of the outer membrane have been suggested (23, 24). The pathogenicity of some Gram-negative bacteria depends on their ability to secrete virulence factors into the mammalian host through the release of outer membrane vesicles (OMVs). Some virulence factors of OMVs include proteases, hemolysins, phospholipids and lipopolysaccharides (25, 26). The analysis of OMV proteins led to the identification of a 58 kDa GroEL homolog as a major component of a complex that has been characterized as a virulence factor with insecticidal activity in some Xenorhabdus species, including X. nematophila (27, 28). These factors, termed moonlighting proteins, escaped our pangenome analysis, as was the case for the GroEL chaperonin, and could explain the differences in pathogenicity and virulence found between the two bacteria.
Evaluation of the insecticidal activity of GroEL proteins
In this case, we decided to clone and express the GroEL proteins from both bacteria, and we named them Cpn60-Xn and Cpn60-Pl (Supplementary Figure S2). It should be noted that genomic data obtained from these bacteria show that they possess only a single copy of the groEL gene. The biological activity of the two purified proteins was evaluated by a direct hemolymph injection method. This method introduced the protein directly into the insect, mimicking the release of toxins by bacteria, a phenomenon that occurs shortly after a nematode infects a target insect.
The bioassay results indicated that only the Cpn60-Xn protein killed a high percentage of G. mellonella larvae and that the mortality rate depended on the concentration of protein used. The relationship between larval death and the protein concentration was assessed by semilogarithmic linear regression analysis, and the slope values (m) representing toxicity were 42.73 and 16.87 for Cpn60-Xn and Cpn60-Pl, respectively; thus, the difference between the slopes was highly significant (F= 31.50, DFn= 1, DFd= 26, p<0.001). (Figure 6). The 50% lethal concentration (LC50) of the purified Cpn60-Xn protein was found to be 102.34 ng/larvae. No external symptoms or mortality were observed in control larvae infiltrated with PBS or BSA (2000 ng), or simply punctured. GroEL toxicity has been reported in Enterobacter aerogenes (EnGroEL) and some Xenorhabdus species. EnGroEL was shown to be a paralytic toxin that ultimately killed cockroaches of the genus Blatella when injected into the hemolymph at a minimum dose of 2.7±1.6 ng (29). Later, a GroEL protein from X. nematophila was purified and found to show insecticidal activity against larvae of the cotton bollworm, Helicoverpa armigera, after oral administration but showed no effect when injected into the hemolymph of this insect (30). Two other proteins from X. budapestensis (HIP57) and X. ehlersii (XeGroEL) were shown to be toxic in G. mellonella larvae when injected into the hemolymph, with LC50 values of 206.81 and 0.76 ± 0.08 ng/larva, respectively (31, 32). In the genus Photorhabdus, there is an initial report that GroEL is a protein component of an 860 kDa complex found in the toxic fraction of the P. luminescens W-14 extract (33); however, no direct evidence supporting this finding has been reported. Our work provides the first evaluation of the activity of this protein.
A global alignment of these two protein sequences showed that they were very similar, sharing 89.86% identity of 548 residues. Cpn60-Xn showed 35 substitutions relative to Cpn60-Pl, which were distributed throughout the protein. A three-dimensional homology-generated model of Cpn60-Xn showed that 13 of these substitutions were located in the apical domain, 2 in the intermediate domain and 20 in the equatorial domain (Supplementary Figure S2). These point substitutions could be responsible for conferring the greater toxicity of Cpn60-Xn relative to Cpn60-Pl.