Skip to main content
BMC Genomics logoLink to BMC Genomics
. 2024 Nov 15;25:1087. doi: 10.1186/s12864-024-10858-2

Analyses of Xenorhabdus griffiniae genomes reveal two distinct sub-species that display intra-species variation due to prophages

Jennifer K Heppert 1,#, Ryan Musumba Awori 2,#, Mengyi Cao 3, Grischa Chen 3, Jemma McLeish 1, Heidi Goodrich-Blair 1,
PMCID: PMC11566119  PMID: 39548374

Abstract

Background

Nematodes of the genus Steinernema and their Xenorhabdus bacterial symbionts are lethal entomopathogens that are useful in the biocontrol of insect pests, as sources of diverse natural products, and as research models for mutualism and parasitism. Xenorhabdus play a central role in all aspects of the Steinernema lifecycle, and a deeper understanding of their genomes therefore has the potential to spur advances in each of these applications.

Results

Here, we report a comparative genomics analysis of Xenorhabdus griffiniae, including the symbiont of Steinernema hermaphroditum nematodes, for which genetic and genomic tools are being developed. We sequenced and assembled circularized genomes for three Xenorhabdus strains: HGB2511, ID10 and TH1. We then determined their relationships to other Xenorhabdus and delineated their species via phylogenomic analyses, concluding that HGB2511 and ID10 are Xenorhabdus griffiniae while TH1 is a novel species. These additions to the existing X. griffiniae landscape further allowed for the identification of two subspecies within the clade. Consistent with other Xenorhabdus, the analysed X. griffiniae genomes each encode a wide array of antimicrobials and virulence-related proteins. Comparative genomic analyses, including the creation of a pangenome, revealed that a large amount of the intraspecies variation in X. griffiniae is contained within the mobilome and attributable to prophage loci. In addition, CRISPR arrays, secondary metabolite potential and toxin genes all varied among strains within the X. griffiniae species.

Conclusions

Our findings suggest that phage-related genes drive the genomic diversity in closely related Xenorhabdus symbionts, and that these may underlie some of the traits most associated with the lifestyle and survival of entomopathogenic nematodes and their bacteria: virulence and competition. This study establishes a broad knowledge base for further exploration of not only the relationships between X. griffiniae species and their nematode hosts but also the molecular mechanisms that underlie their entomopathogenic lifestyle.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-024-10858-2.

Keywords: Xenorhabdus griffiniae, Nematode-bacterium symbiosis, Prophage, CRISPR loci, Pangenome, Bacterial subspeciation, Insect toxins, Entomopathogenic bacteria

Background

Buried in soils across the world is living gold i.e., a rich but as-yet underutilized bioresource: Steinernema nematodes. These insect-killing roundworms have been found in 51 countries to date [16] and are profitable commercial products for the control of insect crop pests. In addition, they are colonized by microbes, including obligate symbiotic bacteria from the genus Xenorhabdus, that produce a battery of useful biomolecules [7]. To date, 35 Xenorhabdus species have been described [79, https://lpsn.dsmz.de/genus/xenorhabdus] found in association with Steinernema nematodes, and the two species work in tandem to infect and kill insects and exploit the nutrient-rich cadaver for the reproductive stage of their shared lifecycle.

Mechanistically, the nematode’s Xenorhabdus bacterium gut symbionts potentiate their insect-killing trait and serve as the primary food source for the nematode. In the non-feeding, host-seeking infective juvenile (IJ) stage of the nematode’s lifecycle, Steinernema nematodes house their Xenorhabdus symbionts in a specialized tissue of the anterior intestine known as the receptacle. After an IJ successfully enters an insect via natural openings such as spiracles, it defecates into the haemolymph, its Xenorhabdus bacteria, which by their secretion of insect toxins, immunosuppression and growth in the haemolymph, kill the insect [7]. The resultant insect cadaver is an enclosed nutrient-rich niche that both nematode and bacterium leverage to reproduce. Nematode fecundity is enhanced by the consumption of Xenorhabdus [10], and Xenorhabdus defends the niche by secreting bacteriocins, antimicrobials and scavenger deterrents, which antagonize both microbial and invertebrate competitors [1113]. Before their exit from the insect cadaver and entry into the surrounding soil, nascent IJ nematodes are specifically colonized by Xenorhabdus in the anterior intestine and ultimately, the receptacle [14].

Each Steinernema nematode is colonized in the receptacle by a specific Xenorhabdus species. However, some Xenorhabdus species associate with multiple Steinernema nematode hosts [7] suggesting a relatively fluid partnership landscape in which the molecular determinants of host-symbiont specificity across the two genera are still being defined. Key to this understanding is a robust comparative analysis of the phylogenetic relatedness and genomic content of Xenorhabdus isolates. Genome assemblies that are lowly contaminated, with high levels of completeness and greater than 50 × coverage, are sufficient for bacterial species delineation [15], comparative genomics and identification and analyses of genes-of-interest. Indeed, insights into the Xenorhabdus-Steinernema symbiosis, bacterial speciation, and entomopathogenicity have been gained through such analyses of Xenorhabdus genomes. For example, comparing the degree of genetic similarity across whole genomes using digital DNA-DNA hybridization (dDDH) has led to the delineation of five novel Xenorhabdus species to date [8, 9, 16, 17]. Also, a comparative analysis of Xenorhabdus bovienii CS03 and SS-2004 genomes revealed that CS03 is more adapted to destroying microbial competitors than is SS-2004 but encodes fewer genes associated with entomopathogenicity [18]. Pangenome analyses of X. bovienii strains revealed intra-species content variation including of prophage origin, suggestive of strain adaptation to specific host environments [19]. The biosynthetic gene cluster that encodes the production of GameXpeptides was found in 72% of a sample of 29 Xenorhabdus genomes [20], which suggests that many strains use this peptide to suppress insect immune pathways [21].

Of the > 100 Steinernema nematode species described to date, only Steinernema hermaphroditum is a self-fertilizing hermaphrodite [22, 23]. This makes S. hermaphroditum particularly well-suited for rigorous nematode genetic studies, including those on the dynamics of transmission of Xenorhabdus bacteria from one generation of their host nematode to another [24, 25]. To lay the groundwork for such studies [26], we aimed to comprehensively analyse genomes of Xenorhabdus griffiniae, gut symbionts of S. hermaphroditum [22, 27]. We hypothesize that insights into the X. griffiniae-S. hermaphroditum symbiosis are attainable through detailed comparative genome analyses of X. griffiniae strains and their close phylogenetic relatives. In this study we delineated two species of Xenorhabdus among the analysed strains, reconstructed their phylogenetic relationships with the rest of the Xenorhabdus genus, and analysed their pangenome and unique loci including prophages, CRISPR, and those encoding phage tail-like structures, secondary metabolites and insect toxins, the last of which was substantiated through insect mortality assays.

Methods

Bacterial genome sequencing

X. griffiniae HGB2511 and Xenorhabdus sp. TH1 (respective Genbank accession numbers 16s rDNA: MZ913116-MZ913125 and OR047834) were isolated from infective juvenile nematodes of Steinernema hermaphroditum CS34 and Steinernema adamsi, respectively, as previously described [16]. Briefly, Galleria mellonella larvae were exposed to infective juvenile nematodes, died, and emergent infective juveniles were isolated by White traps in sterile water. These nematodes were then surface sterilized, ground, plated on LB agar plates supplemented with pyruvate (0.1%), and incubated at 30 °C. X. griffiniae ID10 was purchased from BacDive (DSM 17911). Cultures were inoculated in lysogeny broth (LB) stored in the dark (dark LB:0.5% yeast extract, 1% tryptone, 0.5% NaCl) and grown overnight under agitation at 30 °C. Genomic DNA was prepared for sequencing by phenol extraction and spooling (X. griffiniae HGB2511), or by using the Qiagen DNeasy Kit per the manufacturer’s instructions with minor modification, (X. griffiniae ID10), or by both methods (Xenorhabdus sp. TH1). The Qiagen DNeasy Kit protocol was modified to prevent the viscous Xenorhabdus cell lysate from clogging the DNA-binding column by diluting the lysate with 10 mL of Qiagen QBT buffer before running it through the DNA-binding column by gravity. Purified DNA was sequenced using both short-read and long-read sequencing at the Millard and Muriel Jacobs Genetics and Genomics Laboratory, at the California Institute of Technology (for X. griffiniae HGB2511 and Xenorhabdus sp. TH1) or Novogene (for X. griffiniae ID10). For the X. griffiniae ID10 genome, short-read sequencing was performed using Illumina NovaSeq 150 base pairs (bp) paired-end short-read sequencing with library construction consisting of genomic DNA fragmentation, Illumina adapter ligation and PCR amplification, followed by size selection and purification, resulting in 33 gigabases (GB), of which 842 megabases (MB) were used for assembly. For primer-free long-read sequencing, library preparation consisted of size selection, adapter ligation and purification using Beckman Coulter AMPure XP beads. Sequencing was performed on an Oxford Nanopore (ONP) PromethION platform with base calling performed using Guppy software [28] with standard parameters. Prior to assembly, the ONP long-reads were filtered using Filtlong v0.2.1 resulting in 1.13 GB that was used in the assembly. Short- and long-reads were assembled using Unicycler v0.5.0 [29], resulting in a coverage of 432X. Sequencing of the TH1 and HGB2511 was done via a similar workflow. For each, 1.13 GB and ~ 0.74 GB of bases were obtained from the Illumina and ONP runs, respectively. Likewise, a similar hybrid assembly method was also used, resulting in assemblies that were 406x and 496x for HGB2511 and TH1, respectively. Genome characteristics were determined via the PATRIC platform [30] (Table 1). Coverage was calculated by taking the total number of base pairs used in the assembly and dividing that by the genome size [31]. EvalG was used to determine the quality and completion of the assemblies [32]. GenBank accessions for ID10, HGB2511 and TH1 genomes are CP147737.1, CP147738.1 and CP147734.1, respectively. Names and GenBank accession numbers of other genomes used in this study are found in Supplementary sheet S17 in Additional file 2.

Table 1.

Characteristics of genomes assembled in this study. Characteristics were identified using analysis tools on the Bacterial and Viral Bioinformatics Resource Center platform

Strain Total lengtha (bp) Completeness (contamination) GC content (%) Total CDS (with functional assignments) tRNA genes rRNA genes Depth
ID10 4,767,870 100% (0.4%) 43.91 4688 (70%) 80 22 432x
HGB2511 4,594,889 100% (0.2%) 43.78 4369 (72%) 81 22 406x
TH1 3,770,596 99.6% (0.7%) 43.48 3469 (78%) 81 22 496x

aThe total length was equal to the Contig N50 (bp) since all three chromosome genome assemblies circularised

Tree generated using bayesian inference

Phylogenetic analyses were performed as described previously [33]. Briefly, select Xenorhabdus (30) and Photorhabdus (1) species for which genomic sequences were publicly available were analysed using MicroScope MaGe’s Gene Phyloprofile tool [34] to identify homologous open reading frames (ORF) sets (homologs with at least 50% identity with synteny) which were conserved across all assayed genomes. Putative paralogs were excluded from the downstream analysis to ensure homolog relatedness, resulting in 1235 homologous sets (one-to-one orthologs). Homolog sets were retrieved via locus tag indexing using Python v3.8.0, and nucleotide sequences were individually aligned using Muscle v3.8.31 [35], concatenated using Sequence Matrix v1.9 [36], and trimmed of nucleotide gaps using TrimAL v1.4 [37]. A General Time Reversible + γ variation substitution model was used for maximum likelihood and Bayesian analysis. Maximum likelihood analyses were performed via RAxML v8.2.10 [38] using rapid bootstrapping and 1,000 replicates and were visualized via Dendroscope v3.6.2 [39]. Nodes with less than 40% bootstrap support were collapsed. Bayesian analyses were performed via MrBayes v3.2.7a in BEAGLE [40] on the CIPRES Science Gateway platform [41]. A total of 100,000 Markov chain Monte Carlo (MCMC) replicates were performed. Twenty-five per cent were discarded as burn-in, and posterior probabilities were sampled for every 500 replicates. Two runs were performed with three heated chains and one cold chain. The final average standard deviation of split frequencies was 0.052352. Bayesian trees were visualized with FigTree v1.4.4 [42]. Posterior probabilities are 100% except where otherwise indicated.

Digital DNA-DNA hybridisation and pangenome analyses

To determine pairwise digital DNA-DNA hybridisation (dDDH) values among 31 strains (ID10, VH1, XN45, TH1, HGB2511, BMMCB, BG5, Kalro, 97 and 22 validly published/26 validly published Xenorhabdus type strains), their fasta formatted genomes were uploaded to the TYGS server [43] and analysed as previously described [16]. Genomes of six X. griffiniae strains plus that of Xenorhabdus sp. TH1 were comparatively analysed using a pangenome approach in Anvio 7.1 [44]. Briefly, fasta formats of the seven genomes were reformatted to simplify the definition lines, then converted to anvio contig databases. On these, hidden Markov models (HMMs) and genes were identified using HMMER [45] and Prodigal [46], respectively. The functions of these genes were then predicted, based on orthology, using the Cluster of Orthologous Genes (COG) database [47] as a reference. These annotated contig databases were then used to construct a pangenome with the anvi-pan-genome program under the following parameters –use NCBI-BLAST, MCL inflation 10, minbit 0.5, –exclude-partial-gene-calls. The anvi-display-pan program was used to both display the pangenome as a sunburst chart and subsequently create selected bins. To obtain sub-pangenomes of only type VI secretion system-associated orthologous gene clusters (GC), all GCs annotated with “type VI” were identified using the search functions feature and binned. The anvi-split program was then used to obtain a pangenome of only Type VI secretion system-associated GCs (Supplementary Sheet S3 in Additional file 2). For each analysis, pangenomes were represented both as sunburst charts and tabulated gene lists. To calculate average nucleotide identities (Supplementary Sheet S13 in Additional file 2), the fastANI [48] program was used within an Anvio environment. The average alignment fraction and fragments were 0.85 and 1425 respectively.

To identify the mobilome, all genes annotated with the COG category “X” were extracted from the main pangenome and used to create a sub-pangenome of the mobilome only. This was then used to calculate the total number of genes annotated as phage, transposase, and plasmid-related in each genome. For each genome, these totals were correlated with proteome size using Pearson’s adjusted (due to the small sample size) r-square at a 0.05 alpha level.

Identification and analysis of unique genes within the accessory genome

Using the tabulated output of the Anvio pangenome analysis and Microsoft Excel, core genes and genes unique to a given strain were identified using the number of genomes in which the gene cluster has hits identifier, where the value equalled 7 for core genes (19,196) and 1 for unique genes for the HGB2511 (340), ID10 (411) and TH1 (454) strains. Because the genomes of the sub-species clade which includes X. griffiniae Kalro are so similar, there were fewer than ten unique genes per strain [XN45(8), VH1(7), xg97(4), Kalro (0)]. Thus, to gain a better understanding of what genes might be unique within the sub-species clade, we compared the Kalro genome alone to the HGB2511, ID10 and TH1 genomes and found 454 unique genes for further analysis. The COG annotations assigned as part of the pangenome analysis were used to elucidate the functional categories that might be enriched among those genes we found to be unique to a given species. The number of unique genes per COG category was plotted as a percentage of the total number of unique genes with a COG designation for each strain and the core genome using GraphPad PRISM v10.1.1. As expected, a much larger fraction of genes with no COG designation was found in each of the unique gene categories (approximately 50% in each case) compared to the core genome (6%).

Elucidation and analysis of prophages in X. griffiniae genomes

To identify prophages in bacterium genomes, we used VIBRANT 1.2.1 [49] under default parameters. Each prophage sequence was then separately reannotated with Pharokka [50] under “meta mode” and Bakta [51]. Genomad [52] was used to both taxonomically classify prophages and assess their quality (through CheckV [53]) and completion (Supplementary Sheet S14 in Additional file 2). Resultant gene lists of the annotated prophages are available in (Supplementary Sheets S8-S11 in Additional file 2).

To identify similar prophages across strains, we used progressive Mauve [54] to identify collinearity blocks between prophage sequences. Considerably similar pairs were selected, pairwise aligned, and visualised as dot plots using Geneious 8.1.9 [55].

To determine the effect of prophages on the dDDH values among X. griffiniae strains, all prophage sequences were deleted from their corresponding bacterium genomes in Geneious 8.1.9. Then, dDDH analyses were rerun using both original and “phageless” X. griffiniae bacterium genomes (Supplementary Sheet S12 in Additional file 2).

To identify strain-specific and subspecies-specific genes that are from prophages, a pangenome approach was used. Briefly, their entire prophage sequence from a strain —ID10, Kalro, TH1, xg97 and HGB2511— were merged into a fasta file. For example, all prophages from the ID10 genome were merged into a single fasta file “ID10_prophages.fasta”. These five prophage fasta files, plus those of genomes of ID10, HGB2511, VH1, XN45, xg97 and Kalro, were used to create a pangenome as aforementioned. Using the anvio-display-pan program, strain-specific or subspecies-specific GCs that were of prophage origin were visually identified and binned. Using the anvio-summary program, the number, annotation, and aa sequences of these strain-specific or subspecies-specific genes of prophage origin were obtained from resultant gene lists (Supplementary Sheet S14 in Additional file 2). This was then used to calculate what proportion of strain-specific and subspecies-specific genes were from prophages.

To identify genomic loci encoding complete type VI secretion systems (T6SS) within genomes, we used Secret6 [56] under default parameters. Geneious was used to create multiple sequence alignments of T6SS-encoding loci and to calculate pairwise nucleotide sequence identities.

To infer gene gain and loss events within the X. griffiniae clade, we used the above described anvi’o workflow to create aa pangenome of the seven strains plus Xenorhabdus sp. BG5; BG5 was included to specifically infer the evolution of gene content for strain TH1, as it is the strain that just diverges prior to the emergence of TH1 (Fig. 1). We then used the tabulated version of the pangenome to manually construct a phyletic pattern (Supplementary Sheet S6 in Additional file 2). A dDDH-based phylogenomic tree of the eight strains only, was reconstructed via the TYGS pipeline. Both this tree and the phyletic pattern were used as input data in COUNT [57]. The evolution of gene content was inferred using Wagner parsimony with a gain penalty of two.

Fig. 1.

Fig. 1

Phylogenomic reconstruction of type strains of Xenorhabdus and other strains closely related to Xenorhabdus griffiniae. A This neighbour joining tree was reconstructed using genomic distances calculated with the same formula (Genome Distance BLAST Phylogeny distance formula d5) used for species delineation by digital DNA-DNA hybridisation (dDDH) analyses. Orange and aqua lines correlate with dDDH boundaries for species (> 70%) and subspecies (> 80%), respectively. Strains of X. griffiniae formed two subspecies, those from India-Indonesia (red) and those from Kenya (green). B Bayesian phylogenetic tree created using one-to-one orthologs from Xenorhabdus type strains, X. griffiniae strains, and Photorhabdus asymbiotica as an outgroup. Posterior probabilities are equal to 1 (100%) unless otherwise indicated at a given node. Strains of X. griffiniae formed two separate clades, those isolated from S. hermaphroditum nematodes found in Indonesia (ID10) and India (HGB2511) (red) and those isolated from nematodes found in Kenya (green)

Identification and analysis of defence systems and CRISPR-Cas loci in X. griffiniae genomes

Anti-phage or anti-plasmid defence systems present within genomes were identified using the DefenseFinder bioinformatics search tool and results were interpreted using the associated knowledge base [5860] (Table S4 in Additional file 3, Supplementary Sheet S22 in Additional file 2). To identify CRISPR-Cas systems in X. griffiniae genomes, we used BlastN on the Magnifying Genome platform to search for CRISPR repeats with the previously identified X. nematophila CRISPR repeats, XnCRISPR-E and XnCRISPR-G as queries [61]. Regions with similar or identical sequences were extracted and manually curated for repeat-spacer content. All results were then verified using CRISPRDetect [62]. The only distinction between the two approaches was for ID10, for which the manual approach had suggested five potential repeat sites of one spacer only, whereas CRISPRDetect did not identify any. We therefore conducted a second manual annotation of the repeats in this genome (Additional file 3). Based on conservation with consensus repeats called by CRISPRDetect in both upstream and downstream repeats, we considered ID10 1ai and ID10 1bii regions to be bona fide single spacer CRISPR loci. Cas protein-encoding genes were identified based on their annotation in the Magnifying Genomes platform and confirmed using CasFinder-3.1.0 [63]. To identify protospacers, we searched for full-length identical sequences using each identified spacer as a BlastN query against the X. griffiniae, TH1, and BMMCB genomes in the Magnifying Genome platform. Protospacers were identified in both other strains and within the same genome (self-targeting protospacers). To gain a broader view of the non-self-protospacers, we used CRISPRTarget (http://crispr.otago.ac.nz/CRISPRTarget/crispr_analysis.html) to search the spacer sequences found by CRISPRDetect for TH1, ID10, HGB2511, and BMMCB genomes, against subsets of the available databases (ALCAME genes, Genbank-Phage, RefSeq-Archaea, RefSeq-Plasmid, and RefSeq-Viral) for high confidence sequence matches to the potential protospacers [62, 64]. All putative protospacers for a given spacer were listed (Table S6 in Additional file 3) and the top, annotated hit for each is listed in Table 2.

Table 2.

Annotated spacer-protospacer content among X. griffiniae and related strains

graphic file with name 12864_2024_10858_Tab2_HTML.jpg

Elucidation of biosynthetic gene clusters

To predict which secondary metabolites can be produced by X. griffiniae, a fasta format of the ID10 genome was uploaded to the AntiSMASH [65] webserver (https://antismash.secondarymetabolites.org) as data input. Default parameters (i.e., relaxed detection strictness) and the use of all extra analyses were selected. Output data comprised detailed bioinformatic analyses of 23 biosynthetic gene clusters (BGCs) in the ID10 genome. Each of the 23 BGCs was then manually inspected in two main ways. First, the clusterBLAST feature was used to identify known homologous BGCs by comparing the gene synteny and sequence similarity of an ID10 BGC to those of known BGCs found in the MiBIG database [66]. An ID10 BGC was considered homologous to a known BGC if it contained > 80% genes in perfect synteny, with each gene having BLASTp sequence similarity > 47%, sequence coverage > 40%, and Evalue < 2.78E-19 with the corresponding gene in the known BGC. Second, for ID10 non-ribosomal peptide synthetase (NRPS) BGCs, their Stachelhaus codes [67] for adenylation domains, as well as their epimerisation/dual condensation domains [68] were analysed to determine the amino acid sequence of the linearised non-ribosomal peptide (NRP) the NRPS was predicted to biosynthesize. Predicted NRPs were then compared to known NRP to identify NRPS BGCs that encode the biosynthesis of novel derivatives/peptides. This workflow was similarly applied for the analysis of BGCs in the HGB2511 genome. Chemical structures of the compounds whose production was predicted to be encoded by the known ID10 BGCs were obtained from the Natural Product Atlas [69] and edited in Chemdraw and Inkscape [70].

Putative toxin, secretion system, cas protein, and restriction modification system identification

Putative toxins predicted to impact insect or nematode virulence were identified using multiple approaches. The first approach used the loci of previously characterized known or suspected toxins [71]. BlastP was performed on the Magnifying Genomes platform, and the BLAST query accession proteins are listed in the putative toxins table (Table 3), this procedure was repeated for other novel putative toxins identified as well. The PathoFact software package [72] was used to identify novel putative toxins in the ID10, HGB2511, Kalro, TH1 and X. nematophila 19061 genomes with the standard settings and chromosomal genomes downloaded from the Magnifying Genomes platform. The toxin library outputs from PathoFact were compared and toxins potentially unique to each genome were further examined (Supplementary sheets S2-S3 in Additional file 4). This resulted in the identification of a potential hydrogen cyanide synthetase locus (XTH1_v2_1430-1432) which appears to be unique to the TH1 genome among those analysed. Further, this search revealed zonula occludens toxin proteins (QL128_RS10325, QL128_RS10250, QL128_RS10260) which appear to be unique to the Kalro genome (homologs of were also found in VH1, XN45, and xg97). To both confirm and expand the list resulting from the combination of the above analyses, the search term “toxin” was used on the Magnifying Genomes platform to further query the gene annotations for the HGB2511, ID10, TH1 and Kalro genomes. This search resulted in the identification of a protein annotated as an insecticidal toxin (XGHID_v1_0629) in the ID10 genome and a homolog (XGHIN1_v1_3228) was subsequently found in the HGB2511 genome via BlastP. The products of toxin-antitoxin systems were generally not considered in our analysis. The highest confidence and most interesting results are summarized in the Putative Toxins Table (Table 3).

Table 3.

Putative entomotoxin-encoding genes in Xenorhabdus griffiniae and TH1 genomes

Putative toxin/toxin gene BLAST sequence HGB2511 ID10 Kalro TH1
mcf XNC1_2265 XGHIN1_v1_3249 XGHID_v1_0647 QL128_RS15575 XTH1_v2_2642
MARTX XNC1_1381 XGHIN1_v1_2845 XGHID_v1_1587 QL128_RS13505 XTH1_v2_2371
prtA XNC1_4025 XGHIN1_v1_3782 XGHID_v1_0525 QL128_RS17950 XTH1_v2_551
HIP57 (GroEL) XNC1_3605 XGHIN1_v1_3711 XGHID_v1_1038 QL128_RS17730 XTH1_v2_2959
xaxAB XNC1_ 3766, XNC1_3767 none found XGHID_v1_0927, XGHID_v1_0928 QL128_RS17245, QL128_RS17240 none found
xhlA and B, tpsA and B XNC1_4556, XNC1_4555 XGHIN_v1_0252, XGHIN_v1_0253 XGHID_v1_0280, XGHID_v1_0281 QL128_RS01115, QL128_RS01110 XTH1_v2_0208, XTH1_v2_0209
xhlA hemolysin XNC1_3177 XGHIN1_v1_1428 XGHID_v1_3009 QL128_RS06870 XTH1_v2_1082
stx1a XBI1v2_2730004 none found none found none found none found
xptA2 XNC1_2569 none found none found none found none found
Tc toxins (A) XNC1_2333 + 2334, XNC1_2560 + 2561, XNC1_2566, XNC1_2569, XNC1_3020 + 3021 + 3022 + 3023 + 3024, and XNC1_2187 none found none found none found none found
Tc toxins (B) XNC1_2186, XNC1_2335, XNC1_2568 none found none found none found none found
Tc toxins (C) XNC1_2188, XNC1_2336, XNC1_2567 none found none found none found none found
Chitinase XNC1_2562 none found none found none found none found
Putative chitinases XGHIN1_v1_0490 XGHIN1_v1_0490 XGHID_v1_3568 QL128_RS18035 none found
Putative chitinases XGHIN1_v1_3307 XGHIN1_v1_3307 XGHID_v1_0504 none found none found
Putative chitinases XGHIN1_v1_4103 XGHIN1_v1_4103 none found none found none found
Pir toxins XNC1_1142, and XNC1_1143 XGHIN1_v1_1121 XGHID_v1_3432 QL128_RS04800 XTH1_v2_858
TPS-Fha (txp40) XNC1_1129 none found none found none found none found
Xenocin XNC1_1221–1223 XGHIN1_v1_1178 XGHID_v1_3165 QL128_RS05125 XTH1_v2_0927
Insecticidal toxin (as annotated in ID10 genome) XGHID_v1_0629 XGHIN1_v1_3228 XGHID_v1_0629 none found none found
hcnABC XTH1_v2_1430-1432 none found none found none found XTH1_v2_1430-1432
Zonula occludens toxin QL128_RS10325 none found none found QL128_RS10325 none found
Zonula occludens toxin QL128_RS10250 none found none found QL128_RS10250 none found
Zonula occludens toxin QL128_RS10260 none found none found QL128_RS10260 none found

To systematically predict the types and numbers of secretion systems encoded by the X. griffiniae genomes (HGB2511, ID10, Kalro, VH1, XN45, xg97), TXSScan application from the Macsyfinder 2.0 [73] program was run on the Galaxy [74] platform specifying an ordered, circular replicon, diderm bacteria and with default HMMER options. A combination of the summary output for each genome is found in Table S1 in Additional file 1. Alignment of MARTX regions was performed using MUSCLE in MegAlignPro (DNA Star).

To confirm the type and numbers of Cas proteins encoded by the X. griffiniae genomes (HGB2511, ID10, Kalro, VH1, XN45, xg97), CasFinder application from the Macsyfinder 2.0 [73] program was run on the Galaxy Pasteur platform specifying an ordered, circular replicon, diderm bacteria and with default HMMER options. The summary table is compiled from the software summary output and the additional sheets contain the best solution predictions from the Cas Finder output (Supplementary Sheet S20 in Additional file 2) and include the gene names and locus tags from the different Cas systems found (Table S5 in Additional File 3).

Restriction modification systems were identified by searching the BMMCB, TH1, ID10, HGB2551 and Kalro genomes for restriction enzymes, anti-restriction proteins, and restriction enzyme-associated methylases-based gene annotations on the MAGE Microscope platform. These genomes were further searched using BLAST 2.15.0 + to compare a list of “Gold Standard” methyltransferase and endonuclease protein sequences from the New England Biolabs’ REBASE with the protein sequences contained within each genome [75]. An E value cut-off of ≤ 1 × 10−5 and 75% coverage were used to generate a list of high confidence candidates. This candidate list was compared with the initial list of restriction system proteins generated using MAGE, and redundant sequences were removed. Methyltransferase homologs that were not found to be near predicted restriction endonucleases were excluded from further analysis, though the presence of many ‘orphan’ methyltransferases may indicate a need to protect the bacterial chromosome from restriction modification systems [76]. Putative restriction enzyme loci were examined to identify or confirm neighbouring methylases in the case of Type I, II and III restriction enzymes, or lack thereof, in the case of the Type IV and the HNH restriction endonucleases identified [77]. Of the putative restriction modification genes examined, only three loci from the BMMCB genome (LDNM01_v1_10020, LDNM01_v1_400040, LDNM01_v1_1980001) were predicted to encode restriction endonucleases from Type I or III but were not observed to encode a proximal methyltransferase. Because of their incomplete nature, these loci were excluded from the final table and count of restriction systems. However, each locus was near a contig break, so a less fragmented BMMCB genome assembly may reveal these to be complete predicted restriction endonuclease loci.

Insect larvae rearing and preparation

Eggs of Manduca sexta, the tobacco hornworm, were purchased from Carolina Biological Supply (North Carolina, USA) and were reared to the fifth instar according to a previously described protocol [78]. Briefly, eggs were sterilized with 0.6% (v/v) bleach solution on arrival and then transferred to a four-ounce plastic container with a Gypsy moth diet (MP Biomedicals, Ohio, USA). The eggs were then incubated at 26 °C in a humidified insect incubator with a 16-h light: 8-h dark photoperiod. Once hatching was complete, the larvae were transferred to a new four-ounce container for two days and then transferred individually to a new two-ounce container. Feeding and cleaning were performed every two to three days until the larvae reached the fifth instar stage. On the day of the experiment, the larvae were examined and sorted by weight ranging from 0.67 g to 3.5 g, and the larvae were randomly distributed across the conditions. Larvae (n = 10, with n = 5 for phosphate-buffered saline (PBS) control) were placed into individual two-ounce plastic containers. Groups of 10 larvae were carefully injected with 10 µL of various doses (10–5, 10–4, and 10–3) of bacteria between the first set of abdominal prolegs using a 30 gauge Hamilton syringe. Following injections, the larvae were incubated at 26 °C in a humidified insect incubator with a 16-h light: 8-h dark photoperiod and monitored for survival over 72 h.

Preparation of bacteria for infection of M. sexta

Bacterial strains were streaked from –80 °C freezer stocks onto dark LB agar and cultivated at room temperature in the dark for two days. Broad streaks including multiple colonies were used to inoculate LB medium (5 mL) and incubated on a rotating wheel at 30 °C for approximately 8 h. At 8 h the optical density (OD) at 600 nm was measured (OD600), and cultures were all near an OD600 of 1.0. Cultures were normalized to an OD600 of 1.0 by adjusting the volume of culture taken (e.g., 500 µL OD600 of 1.0). Cells were spun down for 1 min at max speed in Eppendorf tubes and washed twice with 1 mL of sterile PBS. After the final wash, cells were resuspended in 500 µL of PBS. The washed cells were then diluted ten-fold, six times in sterile PBS in a 96-well plate. For each dilution, 10 µL was inoculated onto LB agar plates (LBP) to quantify the number of colony-forming units (CFU) in each dilution. To test for sterility, PBS was also inoculated onto LBP. Based on a previous study [71], we estimated that 100, 1,000, and 10,000 CFU/10 µL was suitable for an observable virulence (insect mortality) and dose–response, and we further calculated that 10–5, 10–4, and 10–3 dilutions of OD600 of 1.0 culture would yield approximately that number of cells per 10 µL injected. M. sexta larvae were raised, injected as described above, and then observed over 72 h. Abbott's formula was used to correct for death in the PBS control condition across the groups. 

Results

Strains ID10 and HGB2511 belong to one of two subspecies of Xenorhabdus griffiniae while Xenorhabdus sp. TH1 is a novel species

An X. griffiniae bacterial isolate (HGB2511) and its nematode host, a strain of Steinernema hermaphroditum [22] are being developed as a genetically tractable model for interrogating bacteria-host interactions [22, 25, 79]. To clarify the relationship between HGB2511 and other related Xenorhabdus isolates, we sought to comparatively analyse their genomes. We first sequenced the genome of three strains: HGB2511 [22], ID10, the X. griffiniae type strain isolated from an Indonesian strain of S. hermaphroditum [27], and TH1 isolated from Steinernema adamsi from Thailand [80]. For all three strains, we obtained circularised genome assemblies of the bacterial chromosome that had < 0.7% contamination, > 99.6% completion and > 50 × depth (Table 1).

We leveraged these high-quality genomes to conduct a wide range of comparative genome analyses, the first of which was taxonomic species delineation via both dDDH and phylogenomics. These analyses both showed the same relationships among the X. griffiniae and closely related species (Fig. 1 & Fig. S1 in Additional file 1). The type strain designation of strain ID10 was corroborated by its lack of > 70% pairwise dDDH values, the threshold for conspecific strains [81], with any of the other type strains of Xenorhabdus (Fig. 1A). Five additional strains were delineated as members of the X. griffiniae species, as they each had pairwise dDDH values with strain ID10 that were above the 70% threshold. Among the six X. griffiniae strains, two subspecies were evident due to intragroup pairwise dDDH values that were all above the 80% threshold [82]. Strains from Kenya XN45, VH1, xg97, and Kalro, belonged to one subspecies (referred to as the Kenyan subspecies) while strains ID10 and HGB2511 from Indonesia and India, respectively, belonged to another subspecies (referred to as the India-Indonesia subspecies)(Fig. 1A and 1B). Although Xenorhabdus sp. Although TH1 was the most closely related strain to X. griffiniae HGB2511 and Xenorhabdus sp. BG5, it is an undescribed species of the genus as it lacked pairwise dDDH values, with any of the type strains, which were above 70% (Fig. 1A). Indeed, in a pangenome of the seven strains, TH1 had the highest number of strain-specific genes, and none of the ANI pairwise values between strain TH1 and six of its closest relatives met the 95% threshold [48] for conspecific strains (Fig. 2A, Supplementary Sheet S13 in Additional file 2). Hence, Xenorhabdus sp. TH1 is a novel species of the Xenorhabdus genus and not a strain of X. griffiniae as was stated in the description of its nematode host [80].

Fig. 2.

Fig. 2

Graphical representations of a pangenome of the Xenorhabdus griffiniae clade. A Sunburst chart of a pangenome of seven close known relatives of the X. griffiniae type strain coupled to a heatmap of their average nucleotide identities (ANI). The pangenome contained 27,337 genes that were clustered into 5,113 groups of orthologs known as gene clusters (GC). The genomes from which each of the GCs was constituted are depicted in the sunburst chart as follows. Each concentric ring represents a genome and each radius represents a GC. For each radius, a dark shade across a concentric ring denotes that the GC is composed of genes from that genome. For the ANI heatmap, shades of red represent pairwise ANI values between 95% (blush) and 100% (rose), the threshold values for conspecific strains. B Loci encoding type six secretions systems (T6SS) found only in X. griffiniae genomes, of those analysed here. Each genome encoded two different T6SS, XG1-T6SS and XG2-T6SS. The core T6SS-encoding genes are indicated (tssA-M). Other genes are in grey. Pairwise percentage nucleotide identities for shown genomic loci that encode XG1-T6SS and XG2-T6SS ranged between 98–100% and 80–100%, respectively

X. griffiniae species encode type six secretion systems with subspecies-specific effectors

The pangenome of the six X. griffiniae strains plus Xenorhabdus sp. TH1 (Fig. 2), the most closely related known species to the X. griffiniae clade (Fig. 1), had a total of 27,337 genes. These were grouped into 5113 groups of orthologs, which we termed gene clusters (GC). Out of these, 2347 GCs were core in that they contained orthologs from every genome within the pangenome. On the other hand, a total of 369, 334, 282, 8, 4 and 7 GCs were unique to genomes of TH1, ID10, HGB2511, XN45, xg97 and VH1 respectively. Strain Kalro lacked unique GCs as all its GCs were present in the xg97 genome, even though their respective nematode hosts, Steinernema sp. Kalro and Steinernema sp. 97 are likely two different undescribed species [83]. Accessory GCs, each of which was composed of orthologs from between two and n-1 genomes, were 1739. Among these were 313 and 448 GCs, which were unique to genomes from the India-Indonesia and Kenyan subspecies, respectively. GCs that encode traits that define an X. griffiniae strain likely fall among the 319 GCs that were unique to X. griffiniae genomes, of which, only 202 (63%) had known functions. Among these, T6SS function was most enriched as it represented 12% of all X. griffiniae-specific genes (Sheet S2 in Additional file 2). For the T6SS GCs, those specifically encoding core components tssA-M were highly conserved as they had 95–100% combined homogeneity indices—this is an anvi’o pangenome metric for estimating the similarity of orthologs within a GC calculated from sequence similarities and gap penalties derived from a multiple sequence alignment (MSA) of amino acid sequences [44]. The higher the value the more the positions with identical residues and no gaps within the MSA (Sheet S2 in Additional file 2). Upon deeper investigation, we found that all X. griffiniae genomes encode two complete T6SS that we designated XG1-T6SS and XG2-T6SS (Fig. 2B). XG1-T6SS loci were almost identical across the six genomes, as the pairwise nucleotide percentage identities for this locus were between 98–100% (Sheet S16 in Additional file 2). Moreover, they were found in roughly the same chromosomal location (Fig. 3) in the four circularised genomes (xg97, Kalro, HGB2511 and ID10). For XG2-T6SS, none of the six corresponding genomic loci had pairwise nucleotide sequence identities that were less than 80% (Sheet S16 in Additional file 2). However, the ID10 XG2-T6SS encoding locus uniquely lacked tssK, tssJ and two other genes that were directly downstream of tssJ (Fig. 2). Like XG1-T6SS, the XG2-T6SS loci were also found in a similar chromosomal region across the four circularised genomes (Fig. 3). Based on the high pairwise nucleotide identities, we identified homologs of XG1-T6SS and XG2-T6SS-encoding loci in X. szentirmaii US123, X. doucetiae T, X. cabanillasi T, X. hominickii ANU, X. nematophila T, X. poinarii T and X. bovienii SS-2004 (Sheet S16 in Additional file 2). The strain SS-2004 homologs were those identified by Chaston et al. [84] and designated T6SS-1 and T6SS-2, respectively, by Kochanowsky et al. [85].

Fig. 3.

Fig. 3

Loci of prophages and gene clusters encoding complete type six secretion systems (T6SS) and xenorhabdicin (xnp1) in complete genomes of four Xenorhabdus griffiniae strains. Genomes of X. griffiniae and XN45 and VH1 were not included in this analysis as they were not circularised, which often results in a prophage locus being split over multiple contigs

We further identified T6SS-associated GCs that are subspecies-specific. The Kenyan and India-Indonesia subspecies have eight and nine subspecies-specific GCs, respectively. The majority of these are predicted to encode spike proteins annotated as VgrG or PAAR-domain-containing Rhs proteins (Sheet S3 in Additional file 2). One Kenyan subspecies GC encodes a tssF that was not part of the two complete T6SS-encoding loci. We analysed the genes in the neighbourhood of the subspecies-specific PAAR-encoding loci for genes that encode T6SS effector proteins and their cognate immunity proteins. We found four such loci that are specific to the India-Indonesia subspecies and that share similar gene content and synteny (Fig. S4 in Additional file 1). For each of these four loci, their encoded PAAR proteins are highly similar since their amino acid sequences had combined homogeneity indices between 94–100% (Sheet S3 in Additional file 2). These findings indicate that in X. griffiniae, the T6SS spike and its cognate effector proteins may contribute to intraspecific traits. We extended these findings by predicting other putative secretion systems encoded across the sequenced genomes using TXSScan (Table S1 in Additional file 1). The number of type I systems varied across the genomes analysed, and at least one copy of flagellum, Type 4a pilus (T4aP), Type 5a Secretion system (T5aSS), and Type 5b secretion system (T5bSS) were identified in all the genomes analysed.

Prophages mediated the acquisition of both subspecies-specific and strain-specific genes

We hypothesized that horizontal gene transfer was a major driver of subspeciation in X. griffiniae, since the mobilome constituted the largest fraction of functionally annotated, strain-specific genes (Fig. 4, sky blue = mobilome: prophages, transposons, plasmids), and strain-specific genes result in speciation when they confer ecologically useful traits [86].

Fig. 4.

Fig. 4

Stacked bar charts depicting the Clusters of Orthologs Groups of proteins (COGs) of the core and unique genes among four closely related Xenorhabdus strains. The core pangenome bar shows the COGs of genes that are common to all genomes in the pangenome. The subsequent bars show the COGs of the unique genes from the HGB2511, ID10, Kalro, and TH1 genomes derived from the previously described pangenome analysis of X. griffiniae strains and Xenorhabdus sp. TH1 (Fig. 2A). The circularised Kalro genome was used to represent the XN45, VH1 and Xg97 genomes as these four genomes are highly similar (> 99.83% pairwise ANI values). The numbers on top of each bar are the total number of genes in each category, followed by the percentage of those genes without COG designation in parentheses

We first investigated this by inferring the evolution of gene content among the six strains of X. griffiniae and Xenorhabdus sp. TH1. Strain BG5 was included in this analysis since it is most closely related known species that diverged just before TH1 in our phylogenomic reconstructions (Fig. 1). We inferred that net gene loss likely drove the speciation of Xenorhabdus sp. TH1 (Fig. 5), consistent with its smaller genome size when compared to those of X. griffiniae strains (Table 1).

Fig. 5.

Fig. 5

Neighbour-joining phylogenomic tree depicting the evolution of gene content among strains of Xenorhabdus griffiniae. Three species are depicted in this tree, Xenorhabdus sp. BG5, Xenorhabdus sp. TH1 and X. griffiniae. For X. griffiniae, its India-Indonesia and Kenya subspecies are in red and green, respectively. The emergence of both subspecies was likely associated with net gene gains. Gene content analysis was conducted in COUNT applying Wagner parsimony

Conversely, net gene gains possibly resulted in the formation of the two X. griffiniae subspecies. We addressed the question of whether horizontal gene transfer (HGT) may have mediated these gene gains by conducting a preliminary pangenome analysis of 49 Xenorhabdus strains. We found that the total number of phage-related genes accounted for 55% of the variation in the proteome sizes among Xenorhabdus genomes. In this analysis, the total number of phage-related genes accounted for 48.69% of the variation in proteome sizes among the seven strains (adjusted r2 = 0.48691, p = 0.04899983). Similar correlations for transposable elements (adjusted r2 = 0.23241, p = 0.15) and plasmid-related genes (adjusted r2 = 0.1762, p = 0.19) were insignificant. Based on this, we focused on the identification of subspecies-specific and strain-specific genes that were linked with prophages.

We first identified prophages in TH1, ID10, HGB2511, Kalro and xg97 (Fig. 3 and Fig. S2 in Additional file 1, Sheets S8-S11 and S15 in Additional file 2). Genomes of VH1 and XN45 were excluded as they were too fragmented to yield robust results. Taxonomically, all identified prophages in the five genomes belonged to the family Caudoviricetes (Sheet S15 in Additional file 2). The genomes of HGB2511, ID10, Kalro, 97 and TH1, had ten, nine, seven, seven, and three prophages, respectively.

To determine how these prophage numbers compared to those found in other strains, we similarly identified prophages in ten other Xenorhabdus strains whose chromosomal genomes were each assembled into one contig (Sheet S7 in Additional file 2). We found a mean of seven prophages per genome, ranging from 3 to 13, indicating that HGB2511 and ID10 harbour higher-than-average numbers of prophages in their genomes. Three prophage loci were similar across genomes: 1) ID10 prophage 3 and HGB2511 prophage 3 which had 73% pairwise nucleotide percentage identities; 2) ID10 prophage 9 and HGB2511 prophage 5 with 61% pairwise nucleotide percentage identities (Fig. S3 in Additional file 1); and 3) the locus (xnp1) [87], whose conserved and variable regions (Fig. 6) had 95–96% and 51% average pairwise nucleotide percentage identities, respectively. These regions were previously elucidated in strains of X. nematophila and X. bovienii and shown to encode xenorhabdicin, an antimicrobial R-type pyocin, or tailocin structure [88]. The xnp1 locus, including genes essential for xenorhabdicin production and release, were detected in all seven genomes (Fig. 3; Fig. 6; Fig. S2 in Additional file 1).

Fig. 6.

Fig. 6

Genomic loci encoding phage tail-like particles from Xenorhabdus griffiniae. A The xenorhabdicin-encoding genomic loci (xnp1) in six strains of X. griffiniae and Xenorhabdus sp. TH1 and a cartoon of the corresponding xenorhabdicin particle. B Genomic loci (pvc) encoding an extracellular contractile injection system, the Photorhabdus Virulence Cassette (PVC) in P. asymbiotica ATCC43949, X. griffiniae ID10 and HGB2511. The genome coordinates of the loci are shown for the circularised genome assemblies. For XN45 and VH1, the shown coordinates are for contigs JACWFC010000129 and JADEUF010000065, respectively

Comparative analysis of xnp1 loci revealed that in ID10 it has been split into two loci located in different parts of the genome (Fig. 3 & Fig. 6), probably due to transposition events. The TH1 xnp1 locus included genes that encode O-antigen acetylase and glycosyltransferases (Fig. 6) which may be involved in conferring immunity to xenorhabdicins through modification of its likely receptor, lipopolysaccharide O antigen [89].

To identify the specific genes contained in the identified prophage regions, we took ID10 prophages as an example (Fig. 7). Genes predicted to encode viral replication and hypothetical proteins constituted 35 and 15 percent, respectively (Sheet S19 in Additional file 2) whereas ‘cargo’ genes with non-virus annotations constituted the remaining half. These annotated cargo genes encoded diverse products including toxin-antitoxin systems that have wide-ranging effects on bacterial physiology and mobile genetic elements within genomes [90]; Importin-11, predicted to encode a nuclear transport receptor that presumably would be delivered for modulation of animal host cell physiology [91]; and diguanylate cyclase, predicted to be part of a signal transduction cascade mediated through the second messenger cyclic-di-GMP [92] (Fig. 7).

Fig. 7.

Fig. 7

Graphical representation of predicted products of genes contained within prophage loci of Xenorhabdus griffiniae ID10. The genes could be categorized into three broad categories; annotated cargo genes, viral replication genes and those whose products are unknown

The identified prophages contained 45 and 38% of strain-specific genes in the HGB2511 and ID10 genomes, respectively. Likewise, 39 and 47% of India-Indonesia and Kenyan subspecies-specific genes were from prophages (Fig. 5). Indeed, removal of prophage regions from the four genomes elevated their pairwise dDDH values: Pairwise dDDH values between ID10 and strains HGB2511, Kalro, TH1 and xg97 rose by 3–3.2 percentage points when identified prophages were removed from all genomes (Sheet S12 in Additional file 2). These findings demonstrate that, in X. griffiniae, a considerable proportion of both subspecies and strain-specific genes were gained through prophages.

Comparative analysis of CRISPR-Cas, protospacer, and anti-CRISPR content

The high prevalence of prophages and prophage-mediated gene gains in X. griffiniae genomes suggests that these bacterial symbionts relatively frequently encounter phage-related foreign DNA. This prompted us to investigate the presence or absence of defence systems in three X. griffiniae strains HGB2511, ID10, and Kalro (representative of the XN45, VH1, and Xg97 genomes), as well as TH1 and BMMCB. Consistent with the diversity of defence systems observed among bacterial genomes, including between strains, each of the five strains analysed here encoded a unique repertoire of defence systems relative to the others [93] (See Table S4 in Additional file 3). Each genome had at least one type of restriction-modification system, which is described in more detail below, as well as at least one type of toxin/antitoxin system, predicted to encode a stable toxin and an unstable anti-toxin. All of the examined strains contained a copy of the Type II toxin-antitoxin system, MazEF that can elicit cell death in response to phage infection in a process known as abortive infection [94]. Abortive infection, which can be triggered by diverse defence pathways, kills the infected cell before phage reproduction can occur, thereby protecting the rest of the population [95]. All of the strains also encoded a Class 1-Subtype-I-E CRISPR-Cas immunity system [90]. Because CRISPR-Cas systems can yield insights into strain diversity and prior history of phage exposure [96], we further explored the content of these loci.

The genomes of each of the analysed X. griffiniae genomes and the close relative TH1 encode a syntenic locus containing a full set of Class1-Subtype-I-E cas genes encoded adjacent to a gntR homolog (Fig. 8A; Table S5 in Additional file 3). BMMCB, has an incomplete set of cas genes on a single contig, with casD and cas2 lacking. However, since the BMMCB genome is fragmented, we cannot rule out the possibility that these genes are encoded elsewhere. HGB2511, ID10, and TH1 also have a second set of Class1-Subtype-I-E genes (but lacking cas1 and cas2) encoded adjacent to an eda homolog (Fig. 8B). All strains except ID10 and BMMCB also have three CRISPR arrays, comprising conserved repeats and variable targeting spacers that are predicted to be transcribed and cleaved into non-coding, small (61 nt), targeting CRISPR RNAs (crRNAs): array 1a, adjacent to the full set of cas genes at the gntR locus (Fig. 8A,D), array 1b, adjacent to the eda homolog (Fig. 8B,E), and array 2, adjacent to a dsbB gene (Fig. 8C,F). BMMCB lacks the CRISPR arrays adjacent to eda and dsbB. Instead, BMMCB has a second CRISPR array with just two spacers in another region of the genome (at coordinate LDNM01000107.1:164548-16833, not shown in the figure) that encodes phage-related genes. This array falls at the edge of a contig break, so may not be an accurate reflection of the repeats that might be present at this locus.

Fig. 8.

Fig. 8

CRISPR-Cas regions of Xenorhabdus strains. A-C Comparison of three CRISPR-Cas related genomic regions in selected strains, anchored for synteny in the diagram using gntR (A), eda (B), or nhaB (C). ORFs are indicated by solid block arrows (blue for cas genes, with the locus tag number of cas3 provided) with annotated gene names indicated above, and identical colours indicating homology. CRISPR repeat arrays (red vertical stripe block arrows) for each region (1a, 1b, 2) had variable number of repeats (noted in parentheses). Start coordinates for each are shown underneath. Degenerate repeat arrays are indicated by brackets. A conserved sequence [1aii; vertical red line] containing a repeat, a spacer, and a degenerate repeat) was identified at the end of cas2, in all strains except BMMCB, which lacks this gene (A). Another [1bi] was apparent at the end of cas3 in the eda locus of HGB2511, ID10, and TH1 (B). D-F The spacer sequences of each CRISPR array found at gntR (D), eda (E), or nhaB (F) were compared for identity to each other or other loci among the analysed strains. Each box represents a spacer, and different colours indicate different sequences. Spacers represented by black boxes and white lettering have 100% identity to “target” loci outside of the array, either within the same genome or within one of the other genomes analysed here. gntR: DNA-binding transcriptional repressor; bleo/gloA: bleomycin resistance/glyoxalase; tn: transposase; lpp: lipoprotein; mem: membrane protein: DUF4056: domain of unknown function 4056 gene; eda: 4-hydroxy-2-oxoglutarate aldolase; nhaB: Na( +):H( +) antiporter NhaB; dsbB: protein thiolquinone oxidoreductase; cdiA: Deoxyribonuclease CdiA; palA/fhbA: filamentous hemagglutinin; luxR: LuxR family transcriptional regulator; cro/CI: HTH cro/C1-type domain-containing protein; symE: Type I addiction module toxin, SymE family; potA: spermidine preferential ABC transporter ATP binding subunit; araC: AraC family transcriptional regulator; fyuA: Putative TonB-dependent siderophore receptor; mfs: putative MFS transporter, signal transducer; cbbBC: Molybdopterin-binding oxidoreductase; res: Type III restriction endonuclease subunit R; mod: site-specific DNA-methyltransferase (adenine-specific). The circularised Kalro genome was used to represent the XN45, VH1 and Xg97 genomes as these four genomes are highly similar (> 99.83% pairwise ANI values)

Since spacer sequences are acquired in a directional manner in response to active infection by foreign nucleic acid material (e.g., phages or plasmids), comparisons of spacer content across related strains can be used to infer their shared life histories and prior exposure to such threats [97, 98]. The spacer contents of CRISPR arrays 1a, 1b and 2, are variable in number and sequence across the strains (Fig. 8D-F). Consistent with the close relatedness of the Kenyan subspecies strains xg97, Kalro, XN45, and VH1, their CRISPR spacer content is identical, except for the absence in XN45 of a duplicated spacer found in array 1a of the other strains (Fig. 8D). Otherwise, there is no overlap in spacer identity among CRISPR arrays of the different strains, indicating their divergence prior to the acquisition of existing spacer content. Compared to the other X. griffiniae and the close relative TH1, the CRISPR arrays in ID10 appear to have a limited number of targeting spacers. Based on the presence of conserved repeat sequences, ID10 encodes two (1ai and 1bii) bona fide, single repeat CRISPR loci at locations syntenic with regions 1a and 1b of the other strains. Three other loci (1aii, 1bi, and 2) appear to be remnants, with only a single clear left repeat, a spacer, and a degenerate right repeat (Fig. 8A-C; Additional file 3).

To gain insights into the types of threats encountered by X. griffiniae and related strains, we searched for putative target sequences (known as protospacers) based on their identity with CRISPR array spacer sequences. We found some spacers have 100% identity to protospacers either within the same genome (self-targeting) or within one of the other genomes analysed here (Table 2; Table S6 in Additional file 3). In many cases, these protospacers were within phage-related, conjugation machinery, and restriction modification systems, in line with the role of CRISPR systems in defending against these types of mobile genetic elements [99, 100]. Consistent with their close relationship, the Kenya clade demonstrated identical protospacer content in genes throughout the genome, including several predicted to be targeted by CRISPR small RNAs from HGB2511 (Table 2; Table S6 in Additional file 3).

Additional functional genes with protospacer sequences that could be targeted by crRNA included those predicted to encode filamentous hemagglutinin (Kenya clade spacer 1b-1), an ABC transporter (Kenya clade spacer 1b-3), and the enzymes FolD (TH1 spacer 1a-5) and GcvP (BMMCB spacer 1a-3). Curiously, the palA/fhaB filamentous hemagglutinin gene with self-identity to the Kenya clade spacer 1b-1 is encoded in the dsb locus, in proximity to the Kenya clade spacer region 2 (Fig. 8C). Since spacer self-identity would presumably result in self-targeting, we hypothesize the genome also encodes an anti-CRISPR immunity mechanism such as anti-CRISPR proteins known as Acr. These proteins are difficult to predict with sequence similarity because they vary widely [100]. We manually searched for such loci in the selected genomes using a “guilt-by-association" approach of putative Acr by identifying small open reading frames in proximity to the protospacer-containing gene and a helix-turn-helix (HTH) domain-containing gene, which is predicted to be the anti-CRISPR regulator. Of the self-targeting protospacers we detected, only the one in palA/fhaB of the Kenyan subclade had a promising candidate based on these criteria (Fig. 8C; Additional file 3). The putative Acr is a DUF2247 domain-containing protein (e.g., QL128_RS09645) which is predicted to encode a protein of 171 aa and is encoded near an HTH cro/C1-type domain-containing protein (e.g., QL128_RS09655) that may be a putative Aca transcriptional regulator [100].

Restriction modification systems

In addition to CRISPR arrays, restriction-modification systems resist the introduction of foreign DNA, including phage infection, by detecting and cleaving non-chromosomal DNA. Restriction-modification systems can be classified based on their structure, cofactor requirements, DNA recognition site, and relative cleavage locations [101, 102]. Type I, II, and III all encode both a restriction endonuclease and a methyltransferase, whereas Type IV endonucleases cleave modified DNA (such as 5-hydroxymethylcytosine) and variably encode an adjoining methyltransferase [102]. There are also anti-restriction proteins which inhibit restriction modification systems by various mechanisms. We predicted the number and type of complete restriction-modification systems and anti-restriction proteins in the BMMCB, TH1, HGB2511, ID10 and Kalro genomes (Table S2 in Additional file 3). Type I and II systems were the most prevalent across the genomes. Only the HGB2511 genome encoded a complete Type III system, while ID10, Kalro, and TH1 genomes each encoded Type IV restriction endonucleases. The genomes of all X. griffiniae and Xenorhabdus sp. TH1 encoded at least one anti-restriction protein, with ID10 appearing to encode seven anti-restriction proteins, by far the largest complement (Table S2 in Additional file 3).

X. griffiniae encode the biosynthesis of diverse natural products

A Xenorhabdus bacterium occupies the uncanny ecological role of both a mutualist of soil-dwelling nematodes and a parasite of diverse insects. A means it uses to juggle this Jekyll and Hyde lifestyle is the production of potpourri of peptides, polyketides, siderophores, aminoglycosides, and toxin proteins and complexes. Hence, Xenorhabdus genomes often comprise loci, known generally as biosynthetic gene clusters (BGCs), which are responsible for the biosynthesis of some of these natural products. A BGC can include many genes, often under the control of one promoter, which collectively encode the production pathway of a single natural product and its derivatives. Commensurate with its genus, which ranks among those that produce the most diverse set of natural products [103], the X. griffiniae ID10 genome contained over 21 biosynthetic gene clusters that were predicted to encode the production of over ten different types of natural products (Fig. 9; Table S3 in Additional file 1).

Fig. 9.

Fig. 9

Genomic loci of known biosynthetic gene clusters in the X. griffiniae ID10 genome and predicted chemical structures of the natural products whose biosynthesis they encode. The paxABC, darABCDE and xabABCD BGCs were predicted to encode the production of potentially novel derivatives of PAX peptides (8), darobactin (5) and xenoamicin (4), respectively, that differed from known structures in amino acid building blocks at positions highlighted in red. Safracin (1), gameXpeptide C (2), benzobactin (3), photoxenobactin (6), type 2 bovienimide (7), rhabduscin (9), 3-isopropyl-4-oxo-2-oxetanecarboxylic acid (10), arylpolyene (11)

Fourteen of these ID10 biosynthetic gene clusters (BGCs) are predicted to encode the biosynthesis of either known compounds or their derivatives (Fig. 9; Table S3 in Additional file 1). For example, the lpcS and isnAB,GT BGCs in the ID10 genome are predicted to encode the production of group IIA bovienimides [104] and rhabduscin [105], respectively, both of which are insect immunity suppressors, the pxb BGC encodes the production of the insecticidal photoxenobactins [20], and safA-H, ioc/leu, xvbA-J, bcmA-G, and ape BGCs respectively encode the production of safracin antibiotics, 3-isopropyl-4-oxo-2-oxetanecarboxylic acid (IOC), benzobactins, bicyclomycins, and aryl polyenes [20, 106108]. The gxpS BGC was predicted to encode the synthesis of GameXPeptide C [109], as their predicted peptide sequence was DVal-LVal-DLeu-LVal-LLeu. In contrast, each member of the X. griffiniae India-Indonesia subspecies had a unique set of BGCs. Specifically, the HGB2511 strain lacked BGCs that encoded the production of rhabduscin, benzobactin, bicyclomycin and actinospectacin, all of which were present in the ID10 genome (Table S3 in Additional file 1).

Notably, the ID10 genome contained known BGCs but the predicted biosynthetic products are previously unknown derivatives. For example, the ID10 paxABC BGC, which encodes the biosynthesis of PAX peptides, is predicted to encode an heptapeptide backbone of LSer-LLys-LLys-DLys-DLys-DLys-DLys, which differs from those of X. nematophila [110] and X. khoisanae [111] by having LSer at position one instead of Gly, since the respective Stachelhaus code was DVWHLSLIDK and not DILQIGLIWK. The xabABCD BGC is predicted to encode the biosynthesis novel xenoamicins that incorporate DIle in lieu of DVal [112] at position eight of the tetradodecapeptide backbone. However, since predictions based on Stachelhaus codes are prone to inaccuracies, only the chemical structure elucidation of these peptides can determine whether they are indeed novel derivatives. The synthesis of novel derivatives is also predicted for BGCs that encode the biosynthesis of the known ribosomally-synthesized and post-translationally modified peptide (RiPP), darobactin [113], since the ID10 darABCDE BGC was predicted to encode the biosynthesis a core peptide with the sequence Trp-Asn-Trp-Ser-Lys-Gly-Phe and not Trp-Asn-Trp-Ser-Lys-Ser-Phe.

X. griffiniae encode entomotoxins and are insecticidal to Manduca sexta

An essential part of the Steinernema-Xenorhabdus entomopathogenic lifecycle, is the ability of the host-symbiont pair to infect and kill insect hosts, and Xenorhabdus produce virulence factors that target other microorganisms competing for the nutritious insect cadaver [7]. To better understand the toxic potential of the X. griffiniae bacteria, we mined the HGB2511, ID10, Kalro and TH1 genomes for toxin-domain-containing loci. Using a list of known toxins found in other Xenorhabdus [19, 71] we identified homologs of genes encoding the known insecticidal toxins Mcf “makes caterpillars floppy” and PirAB in each of the strains [114116], along with proteins homologous to the MARTX toxin family (Table 3). Similar to X. innexi HGB1681, the MARTX proteins in HGB2511, ID10, Kalro and TH1 each lack four of the A repeats at the N-terminus of the protein (A ∆3–7), leaving nine repeats compared to the 14 found in X. nematophila 19061, X. bovienii SS-2004 (Jollieti), and Vibrio species [71, 117] (Fig. S5 in Additional file 1). It remains unclear how these differences in repeat structure might impact MARTX protein function in X. griffiniae.

Insecticidal Tc toxins are three-part toxin complexes (TcA, TcB, and TcC type) commonly found in entomopathogenic bacteria, including in some Xenorhabdus species [118]. Tc toxin family proteins were notably absent from the X. griffiniae HGB2511, ID10 and Kalro genomes and the Xenorhabdus sp. TH1 genome (Table 3). No evidence was found of homologs of Shiga toxin (stx1a) related genes, such as those found in some X. bovienii strains (Table 3) [19]. Within the ID10 genome we identified a putative insecticidal toxin (XGHID_v1_0629), a homolog of which was also found in the HGB2511 genome (XGHIN1_v1_3228). These genes are homologs of a Photorhabdus asymbiotica gene (PAU_03337) (Fig. 6B) that encodes Photorhabdus dNTP pyrophosphatase 1 (Pdp1), a cytotoxic protein that not only kills immune cells by reducing their intracellular deposits of deoxynucleotide triphosphates (dNTPs) but is also an effector protein of the extracellular contractile injection system (eCIS) known as Photorhabdus virulence cassette (PVC) [119]. Indeed, analysis of genes upstream pdp1 revealed that both ID10 and HGB2511 encode PVCs (Fig. 6B). Although PVCs are phage tail-like particles that are structurally similar to xenorhabdicin (Fig. 6B), they differ by having within their tube, effector proteins that are translocated into the target cell, upon tail fibre-mediated binding and subsequent tail sheath contraction [120]. Notably, when loci that encode eCIS and T6SS were found on the same genome, core genes often were lacking from T6SS encoding loci [121], probably explaining why tssK, tssJ are absent from the XG2-T6SS encoding locus of ID10. The N-terminus (50aa) of the Photorhabdus asymbiotica Pdp1 acts as a signal peptide for secretion through the PVC [122]. Amino acid alignment of the ID10 and HGB2511 Pdp1 proteins with the Photorhabdus Pdp1 and two non-PVC secreted homologs [122] revealed amino acids at the N-terminus of the Xenorhabdus proteins that could act as a signal peptide for PVC secretion (Fig. S6, Additional file 1).

A de novo search for other toxin homologs using PathoFact software on our genomes of interest revealed two strain-specific loci of particular interest [72]. The Xenorhabdus sp. TH1 genome contains a complete hydrogen cyanide synthase locus (hcnABC) (Table 3). hcnABC is found in plant-associated and entomopathogenic bacteria [123] where it plays a role in insect killing. Notably, hcnABC was recently identified in the genome of a steinernematid-associated Pseudomonas piscis bacterium [83]. In the X. griffiniae Kalro genome, three proteins with zonula occludens toxin (zot) domains were identified (QL128_RS10325, QL128_RS10250, QL128_RS10260) (Table 3). Zonula occludens toxin (Zot) domain-containing proteins target the eukaryotic cell cytoskeleton and compromise the structure of intercellular tight junctions, leading to a permeabilization of epithelia [124, 125]. Homologs of the three Zot domain-containing proteins found in Kalro were also identified in xg97, VH1 and XN45, and BLASTp revealed other Zot domain-containing proteins in other Xenorhabdus species (including X. bovienii, X. khoisanae, X. eapokensis, X. ehlersii and X. innexi). QL128_RS10325 and QL128_RS10260 are each predicted to encode a transmembrane helix and to be membrane embedded, whereas QL128_RS10250 is a considerably shorter peptide lacking both transmembrane domains and secretory signals. These Zot domain-containing proteins may affect the insect midgut as part of an oral route of infection [126], or destroy insect epithelial tissues when the bacteria are released into the hemocoel.

The diversity of toxin coding potential within the analysed Xenorhabdus genomes suggest possible differences in their entomopathogenicity. To begin to interrogate this possibility, we assessed the survival of fifth instar Manduca sexta insect larvae over a 72 h period after injection with five Xenorhabdus strains at three concentrations. We aimed to compare the strains at an inoculum of ~ 1000 cells because at that dosage our controls, X. nematophila (19061) and X. innexi (HGB1681) were previously shown to induce near to 100% lethality and < 10% lethality, respectively [71], and indeed, these trends were recapitulated in our study (Fig. 10). Each bacterial strain displayed a dose-dependent survival response, with the highest inoculum resulting in the greatest mortality (Fig. S7, Additional file 1). Insects injected with ~ 1000 cells of X. griffiniae ID10 displayed robust survival, like X. innexi, whereas greater than 50% of the animals injected with approximately the same number of X. griffiniae HGB2511 and Xenorhabdus sp. TH1 cells succumbed within 72 h of injection, like the level observed for X. nematophila (Fig. 10). These results reveal that the isolates tested have different levels of virulence against the Lepidopteran insect Manduca sexta.

Fig. 10.

Fig. 10

Comparative entomopathogenicity of Xenorhabdus bacteria to Manduca sexta larvae. Lines represent survival curves for n=10 insect larvae injected with approximately 1000 cells of each strain. Values have been corrected with Abbott's formula within the table below

Discussion

As part of their symbiotic and entomopathogenic lifecycle, all Xenorhabdus must colonize and be transported by a nematode host, suppress insect immunity, establish a community within and consume the cadaver, and support reproduction of the nematode to ensure future transport [10]. Many will also compete or cooperate with other resident or transient microbial community members and respond to variations in abiotic factors and higher-order trophic interactions. Here we conducted a comparative genomics analysis to gain insights into the consequences of such variable selective pressures on the evolution of Xenorhabdus genome content. Our analysis centred around seven related Xenorhabdus strains, chosen based on their close phylogenetic proximity to X. griffiniae. Strains of this Xenorhabdus species are particularly relevant to the development of laboratory model systems used to study nematode-bacteria symbiosis because they are symbionts of S. hermaphroditum, currently the most genetically tractable steinernematid [22, 25, 127].

To allow detailed comparative genome analyses, we assembled new high-quality, circularised genomes for three strains: X. griffiniae ID10, HGB2511, and TH1 (Table 1) [22, 27, 80]. High-quality genomes were already available for four other strains previously identified as X. griffiniae: xg97, Kalro, XN45 and VH1 [16, 83]. HGB2511 and the Kenyan isolates were verified as X. griffiniae species, as defined by a genome with pairwise values for dDDH and ANI that are greater than the 70% [128] and 95% [48] respective species thresholds, with that of the type strain ID10 (Fig. 1) [27]. Consistent with previous findings [16], our analysis confirmed that strain BMMCB is not an X. griffiniae strain as originally designated [129]. Instead, we found that strain BMMCB is likely conspecific with Xenorhabdus sp. SF857 (Sheet S1 in Additional file 2), the recently described type strain of a novel species X. bakwenae [8]. Strain TH1 is not conspecific to ID10 nor any other type strain, making it a novel species within the Xenorhabdus genus. These conclusions were further supported by phylogenetic reconstructions which revealed that Xenorhabdus sp. TH1 does not cluster with X. griffiniae but shares a last common ancestor with the progenitor of the X. griffiniae clade (Fig. 1). From our findings on the evolution of gene content, we speculate that TH1 diverged, primarily through gene losses, from the progenitor of the X. griffiniae clade to ultimately form its species (Fig. 5), similar to the speciation observed in Bordetella pertussis [130].

Phage-related genes are known drivers of genome variation between closely related strains [131] and previously were implicated as drivers of the differences in gene content among other close Xenorhabdus relatives of X. griffiniae [16]. The X. griffiniae genomes we analysed were found to be enriched in mobilome content, with phage-related genes specifically driving the diversification of subspecies and strains. We found that the India-Indonesia and Kenyan subspecies arose from the net gain of genes, of which 39–47% were from prophages (Fig. 4&5). Further, we found that the proportion of X. griffiniae strain-specific genes of prophage origin is 35–48%. This is especially high, considering that for example, of 30 Bifidobacterium strains examined, the highest proportion of strain-specific genes of prophage origin observed was only between 0.03–35.4% [132]. Moreover, X. griffiniae prophage regions reduced pairwise dDDH values among them by three percentage points. In Salmonella enterica prophage sequences have been shown to be highly variable and differentially conserved among strains, making them key drivers of genome diversification and useful markers for serovar typing [133, 134], Similarly, we conclude that in X. griffiniae prophages possibly are major drivers of subspeciation and strain differentiation.

An essential component of the Xenorhabdus lifestyle is interaction with other organisms, including competing microbes, the mutualistic nematode host, and the prey insect host. Among the molecular machines that facilitate bacterial manipulation of other organisms is the T6SS, which delivers effectors into target (non-self) cells [135]. In this study, we found a high sequence similarity of corresponding X. griffiniae T6SS-encoding loci, but gene content variability in loci encoding concomitant effector proteins. This indicates that effector proteins possibly contribute to traits that vary within a species. One of these traits is nematode host specificity, which varies within both X. griffiniae and X. bovienii species [7]. In X. bovienii, the inactivation of vgrG in XG2-T6SS loci in strain SS-2004 resulted in the near loss (200-fold decrease) of the bacterium's capacity to colonise its nematode host [136]. Although we found that the presence and sequence of VgrG does not appear to vary across X. griffiniae (or other Xenorhabdus), the genes encoding the effector proteins transported by VgrG do vary among the X. griffiniae species. Therefore, we speculate that one or more X. griffiniae VgrG-associated effector proteins may determine a strain’s capacity to naturally colonise its specific nematode host, similar to the conclusion reached about the T6SS function in X. bovienii SS-2004 [136].

CRISPR content and spacer identities support the conclusion that the strains studied here are diversifying due to phage pressure and reflect the taxonomic relationships we observed. The Kenyan subclade has nearly identical CRISPR-Cas loci, consistent with their very close relationship. The only difference is that the XN45 CRISPR array 1a appears to lack a repeated spacer that the others have. Considering the Kenyan subclade as a single group, all the genomes were distinct from each other with respect to CRISPR spacer content, indicative of their unique histories in exposure to, and successful defence from phages and mobile genetic elements. Consistent with this, non-self-targeting protospacers identified within the group could be found within prophage genes and conjugation machinery in the genomes of the other strains. Our analysis identified instances of potential self-targeting, which offered the opportunity to search for anti-CRISPR genes, which are a key aspect of the co-evolution of phage and defence systems and have potential utility in applications of CRISPR technologies [100]. In the Kenyan subclade, we identified one clear candidate for such an anti-CRISPR locus comprising an HTH-domain Aca regulatory candidate and a small DUF2247 domain-containing protein ORF of 171 amino acids (aa), adjacent to the protospacer-containing gene palA/fhaB, which is also in the same region as a CRISPR array. However, DUF2247 proteins are also known as “imm38” and their presence within polymorphic toxin loci, such as palA/fhaB, has implicated them as immunity proteins to these toxins [137]. Consistent with this possibility, members of the Kenyan clade are the only strains of those analysed here that appear to have both a DUF2247 ORF and full-length palA/fhaB genes in that locus.

The X. griffiniae and Xenorhabdus sp. TH1 genomes we compared all bear hallmarks of the entomopathogenicity characteristic of the Xenorhabdus genus. However, when cultures were injected into M. sexta insect larva, the ID10 strain displayed attenuated virulence when compared with HGB2511 and TH1. The magnitude of this difference was directly comparable to the difference in virulence observed between X. innexi and X. nematophila species [71]. Members of the X. bovienii species group have demonstrated a similar range of virulence phenotypes when injected in the absence of the vectoring nematodes [138]. These differences may be due to genomic variation between the closely related species. Notably, we identified strain specific toxin loci, such as the hcnABC locus in TH1 and the zot domain containing toxins in the Kenyan clade that may underly different mechanisms or levels of virulence in X. griffiniae and its close relatives (Fig. 10; Table 3). It is also possible that the lack of virulence observed for ID10 is due to gene expression programs which control phenotypic variation locking the isolate in a state of attenuated virulence [139, 140]. If so, we predict that other isolates of the ID10 strain may retain high levels of virulence, similar to what has been observed for other Xenorhabdus species [139, 141, 142]. Alternatively, the ecological insect host range of ID10 may be distinctive enough from other X. griffiniae that it has lost the ability to infect the Lepidopteran insect Manduca sexta that we tested here, or the nematode may carry most of the virulence potential between the host-symbiont pair [143].

Conclusion

This study yielded three complete genome assemblies, which were of X. griffiniae ID10, X. griffiniae HGB2511 and Xenorhabdus sp. TH1. Xenorhabdus sp. TH1 is a novel bacterial species and putative type strain with the temporary designation Candidatus Xenorhabdus lamphunensis, while X. griffiniae contained two subspecies. Both CRISPR loci and loci encoding T6SS effector proteins divided along these X. griffiniae subspecies lines. Intraspecific variation, including subspeciation, was largely driven by prophages. In terms of biosynthetic potential, X. griffiniae genomes encoded the production pathways of diverse and biotechnologically useful natural products such as antibacterials, antiprotozoals, and insecticidal toxins. Intraspecific variation in biosynthetic potential was observed, which we substantiated by the different levels of entomopathogenicity, among X. griffiniae strains, to M. sexta. Ultimately, these genome assemblies and genomic insights are foundational for continuing studies into the symbiosis between X. griffiniae and its self-fertilizing nematode host, S. hermaphroditum.

Supplementary Information

12864_2024_10858_MOESM1_ESM.docx (1.6MB, docx)

Additional file 1. Word document containing supporting data for the detection and analysis of secretion systems, maximum likelihood phylogenies, prophages and type six effector protein loci identification, MARTX locus alignments, and dose response assays for strains used in insect virulence assays. Table S1: Comparison of the number of putative secretion systems that are present Xenorhabdus griffinae strains. Table S2. Loci encoding putative restriction modification systems identified by homology and genome annotation searches. Table S3: Biosynthetic gene clusters (BGCs) in the X. griffiniae ID10 genome. Figure S1: Maximum likelihood phylogenetic tree created using RAxML with one-to-one orthologs from Xenorhabdus type strains, closely related X. griffiniae strains, and Photorhabdus asymbiotica as an outgroup. Figure S2: Loci of prophages and xnp1 in the complete Xenorhabdus sp. TH1 genome. Figure S3: Dotplots of prophage loci that were considerably similar between strains. Figure S4: Schematic of subspecies-specific type six secretion system (T6SS) effector-encoding loci in six strains of Xenorhabdus griffiniae. Figure S5: MARTX protein multiple sequence alignments Figure S6: Pdp1 protein multiple sequence alignment Figure S7: Percentage survival of Manduca sexta post injection with multiple concentrations of Xenorhabdus bacteria strains tested.

12864_2024_10858_MOESM2_ESM.xlsx (11.1MB, xlsx)

Additional file 2. Excel workbook containing raw data and results of all genome analyses. Data tables which collectively contain raw data and values from pangenome, dDDH, ANI, biosynthetic gene cluster, prophage analyses; defence systems; phyletic patterns; genome accession numbers and names of strains used.

12864_2024_10858_MOESM3_ESM.docx (12.8MB, docx)

Additional file 3. Word document with additional details of the detection of Xenorhabdus griffiniae defense systems and analysis of CRISPR loci. Table S4. Summary of defense systems in X. griffiniae and related strains. Table S5. Locus tags and coordinates of CRISPR-Cas features. Table S6. CRISPR spacers, protospacers, concomitant annotation and genome wherein they are located. Methods and results for X. griffiniae CRISPR repeats, protospacers and self-targeting immunity.

12864_2024_10858_MOESM4_ESM.xlsx (82KB, xlsx)

Additional file 4. Excel workbook of putative insect toxins encoded by Xenorhabdus strains. These sheets contain raw data of the toxin domain-containing proteins that are unique among the X. griffiniae and Xenorhabdus sp. TH1 species. Also contained herein, are raw data from the comparison between the combined toxin library outputs for the X. griffiniae and Xenorhabdus sp. TH1 species with the output for X nematophila 19061. Also provided are the raw “toxin library” summary outputs from the PathoFact software used in the analyses.

Acknowledgements

The authors would like to thank Dr. Apichat Vitta, Department of Microbiology and Parasitology, Faculty of Medical Science, Naresuan University, Thailand, for isolating the Steinernema adamsi nematodes, the host of the TH1 bacterial symbiont, Candidatus Xenorhabdus lamphunensis. We also thank the Dillman lab (UC-Riverside) for bringing S. adamsi to the United States, and sharing the strain with permission from Dr. Vitta. We thank our collaborators in the P. Sternberg (CalTech), A. Dillman, and E. Schwarz (Cornell) labs for feedback on the project throughout the process. The Sternberg lab provided resources for bacterial genome sequencing and was supported by the Millard and Muriel Jacobs Genetics and Genomics Laboratory at California Institute of Technology. The authors thank Zachary Burcham for assistance running the PathoFact software used in the toxin analysis and Sarah Kauffman for assistance with insect injections.

Authors’ contributions

J.K.H and R.M.A are co-first authors and contributed equally to the development of the manuscript. J.K.H, R.M.A, and H.G.-B. collected, and analysed the data, and wrote the original draft of the manuscript. M.C. and G.C. performed DNA extraction, sequencing, and assembly for HGB2511 and TH1 genomes. J.K.H. performed all other “bench” experiments, extractions and assemblies, with help from J.M. on the insect virulence assay. H.G.-B. supervised and provided resources for the study. All authors reviewed the manuscript.

Funding

This work was supported by a National Science Foundation EDGE grant (IOS-2128266).

Data availability

The datasets generated during the current study are available in the GenBank repository under BioProject ID: PRJNA1085699. Previously reported data are available from public repositories (see Methods), and additional data are provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Jennifer K. Heppert and Ryan Musumba Awori contributed equally to this work.

References

  • 1.Bhat AH, Chaubey AK, Askary TH. Global distribution of entomopathogenic nematodes, Steinernema and Heterorhabditis. Egyptian J Biol Pest Control. 2020;30:31. [Google Scholar]
  • 2.Edmunds C, Wilding CS, Rae R. Pathogenicity and environmental tolerance of commercial and UK native entomopathogenic nematodes (Steinernema and Heterorhabditis spp.) to the larvae of mosquitoes (Aedes aegypti and Ochlerotatus detritus). Int J Pest Management. 2021;67:232–40. [Google Scholar]
  • 3.Kovtun A. New locality records for Steinernema and Heterorhabditis (Nematoda: Rhabditida: Steinernematidae, Heterorhabditidae) fauna of Ukraine. Quarantine Plant Protect. 2023;2:39–45.
  • 4.Ali M, Allouf N, Ahmad M. First report of entomopathogenic nematode Steinernema affine (Nematoda: Steinernematidae) in Syria and its virulence against Galleria mellonella L. (Lepidoptera: Pyralidae). Egyptian J Biol Pest Control. 2022;32:101. [Google Scholar]
  • 5.Kizi NKS. First report on local entomopathogenic nematode Steinernema feltiae in Uzbekistan. Int J Adv Sci Res. 2023;3:225–35. [Google Scholar]
  • 6.Mackiewicz JP, Kramarz PE, Rożen A. Thermal sensitivity of Xenorhabdus bovienii (Enterobacterales: Morganellaceae) isolated from Steinernema feltiae (Rhabditida: Steinernematidae) originating from different habitats. Appl Entomol Zool. 2022;57:347–55. [Google Scholar]
  • 7.Awori RM. Nematophilic bacteria associated with entomopathogenic nematodes and drug development of their biomolecules. Front Microbiol. 2022;13:993688. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Ritter CL, Malan AP, Dicks LMT. Xenorhabdus bakwenae sp. n., associated with the entomopathogenic nematode Steinernema bakwenae. Nematol. 2023;25:1169–79. [Google Scholar]
  • 9.Machado RAR, Bhat AH, Castaneda-Alvarez C, Askary TH, Půža V, Pagès S, et al. Xenorhabdus aichiensis sp. nov., Xenorhabdus anantnagensis sp. nov., and Xenorhabdus yunnanensis sp. nov., Isolated from Steinernema entomopathogenic nematodes. Curr Microbiol. 2023;80(9):300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Mucci NC, Jones KA, Cao M, Wyatt MR 2nd, Foye S, Kauffman SJ, et al. Apex predator nematodes and meso-predator bacteria consume their basal insect prey through discrete stages of chemical transformations. mSystems. 2022;7:e0031222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Murfin KE, Ginete DR, Bashey F, Goodrich-Blair H. Symbiont-mediated competition: Xenorhabdus bovienii confer an advantage to their nematode host Steinernema affine by killing competitor Steinernema feltiae. Environ Microbiol. 2018. 10.1111/1462-2920.14278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Bashey F, Young SK, Hawlena H, Lively CM. Spiteful interactions between sympatric natural isolates of Xenorhabdus bovienii benefit kin and reduce virulence. J Evol Biol. 2012;25:431–7. [DOI] [PubMed] [Google Scholar]
  • 13.Gulcu B, Hazir S, Kaya HK. Scavenger deterrent factor (SDF) from symbiotic bacteria of entomopathogenic nematodes. J Invertebr Pathol. 2012;110:326–33. [DOI] [PubMed] [Google Scholar]
  • 14.Chaston JM, Murfin KE, Heath-Heckman EA, Goodrich-Blair H. Previously unrecognized stages of species-specific colonization in the mutualism between Xenorhabdus bacteria and Steinernema nematodes. Cell Microbiol. 2013;15:1545–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Chun J, Oren A, Ventosa A, Christensen H, Arahal DR, da Costa MS, et al. Proposed minimal standards for the use of genome data for the taxonomy of prokaryotes. Int J Syst Evol Microbiol. 2018;68:461–6. [DOI] [PubMed] [Google Scholar]
  • 16.Awori RM, Waturu CN, Pidot SJ, Amugune NO, Bode HB. Draft genomes, phylogenomic reconstruction and comparative genome analysis of three Xenorhabdus strains isolated from soil-dwelling nematodes in Kenya. Access Microbiol. 2023;5:000531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Castaneda-Alvarez C, Prodan S, Zamorano A, San-Blas E, Aballay E. Xenorhabdus lircayensis sp. nov., the symbiotic bacterium associated with the entomopathogenic nematode Steinernema unicornum. Int J Syst Evol Microbiol. 2021;71:71. [DOI] [PubMed] [Google Scholar]
  • 18.Bisch G, Ogier J-C, Médigue C, Rouy Z, Vincent S, Tailliez P, et al. Comparative genomics between two Xenorhabdus bovienii strains highlights differential evolutionary scenarios within an entomopathogenic bacterial species. Genome Biol Evol. 2016;8:148–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Murfin KE, Whooley AC, Klassen JL, Goodrich-Blair H. Comparison of Xenorhabdus bovienii bacterial strain genomes reveals diversity in symbiotic functions. BMC Genomics. 2015;16:889. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Shi Y-M, Hirschmann M, Shi Y-N, Ahmed S, Abebew D, Tobias NJ, et al. Global analysis of biosynthetic gene clusters reveals conserved and unique natural products in entomopathogenic nematode-symbiotic bacteria. Nat Chem. 2022;14:701–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Jin G, Hrithik MTH, Lee D-H, Kim I-H, Jung J-S, Bode HB, et al. Manipulation of GameXPeptide synthetase gene expression by a promoter exchange alters the virulence of an entomopathogenic bacterium, Photorhabdus temperata, by modulating insect immune responses. Front Microbiol. 2023;14:1271764. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Stock SP, Griffin CT, Chaerani R. Morphological and molecular characterisation of Steinernema hermaphroditum n. sp. (Nematoda: Steinernematidae), an entomopathogenic nematode from Indonesia, and its phylogenetic relationships with other members of the genus. Nematol. 2004;6:401–12.
  • 23.Bhat AH, Chaubey AK, Shokoohi E, William MP. Study of Steinernema hermaphroditum (Nematoda, Rhabditida), from the West Uttar Pradesh, India. Acta Parasitol. 2019;64:720–37. [DOI] [PubMed] [Google Scholar]
  • 24.Cao M, Schwartz HT, Tan CH, Sternberg PW. The entomopathogenic nematode Steinernema hermaphroditum is a self-fertilizing hermaphrodite and a genetically tractable system for the study of parasitic and mutualistic symbiosis. Genetics. 2022;220(1):iyab170. 10.1093/genetics/iyab170. [DOI] [PMC free article] [PubMed]
  • 25.Schwartz HT, Tan C-H, Peraza J, Raymundo KLT, Sternberg PW. Molecular identification of a peroxidase gene controlling body size in the entomopathogenic nematode Steinernema hermaphroditum. Genetics. 2024;226: iyad209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Alani OS, Cao M, Goodrich-Blair H, Heppert JK. Conjugation and transposon mutagenesis of Xenorhabdus griffiniae HGB2511, the bacterial symbiont of the nematode Steinernema hermaphroditum (India). MicroPubl Biol. 2023. 10.17912/micropub.biology.000772. [DOI] [PMC free article] [PubMed]
  • 27.Tailliez P, Pagès S, Ginibre N, Boemare N. New insight into diversity in the genus Xenorhabdus, including the description of ten novel species. Int J Syst Evol Microbiol. 2006;56:2805–18. [DOI] [PubMed] [Google Scholar]
  • 28.Sherathiya VN, Schaid MD, Seiler JL, Lopez GC, Lerner TN. GuPPy, a Python toolbox for the analysis of fiber photometry data. Sci Rep. 2021;11:24212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Wick RR, Judd LM, Gorrie CL, Holt KE. Unicycler: Resolving bacterial genome assemblies from short and long sequencing reads. PLoS Comput Biol. 2017;13: e1005595. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Olson RD, Assaf R, Brettin T, Conrad N, Cucinell C, Davis JJ, et al. Introducing the bacterial and viral bioinformatics resource center (BV-BRC): a resource combining PATRIC, IRD and ViPR. Nucleic Acids Res. 2022;51:D678–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Lander ES, Waterman MS. Genomic mapping by fingerprinting random clones: A mathematical analysis. Genomics. 1988;2:231–9. [DOI] [PubMed] [Google Scholar]
  • 32.Parrello B, Butler R, Chlenski P, Olson R, Overbeek J, Pusch GD, et al. A machine learning-based service for estimating quality of genomes using PATRIC. BMC Bioinformatics. 2019;20:486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Grossman AS, Mauer TJ, Forest KT, Goodrich-Blair H. A widespread bacterial secretion system with diverse substrates. mBio. 2021;12:10.1128/mbio.01956-21. [DOI] [PMC free article] [PubMed]
  • 34.Médigue C, Calteau A, Cruveiller S, Gachet M, Gautreau G, Josso A, et al. MicroScope-an integrated resource for community expertise of gene functions and comparative analysis of microbial genomic and metabolic data. Brief Bioinform. 2019;20:1071–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004;32:1792–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Vaidya G, Lohman DJ, Meier R. SequenceMatrix: concatenation software for the fast assembly of multi-gene datasets with character set and codon information. Cladistics. 2011;27:171–80. [DOI] [PubMed] [Google Scholar]
  • 37.Capella-Gutiérrez S, Silla-Martínez JM, Gabaldón T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics. 2009;25:1972–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Kozlov AM, Darriba D, Flouri T, Morel B, Stamatakis A. RAxML-NG: a fast, scalable and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics. 2019;35:4453–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Huson DH, Scornavacca C. Dendroscope 3: an interactive tool for rooted phylogenetic trees and networks. Syst Biol. 2012;61:1061–7. [DOI] [PubMed] [Google Scholar]
  • 40.Ayres DL, Cummings MP, Baele G, Darling AE, Lewis PO, Swofford DL, et al. BEAGLE 3: improved performance, scaling, and usability for a high-performance computing library for statistical phylogenetics. Syst Biol. 2019;68:1052–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Miller MA, Pfeiffer W, Schwartz T. Schwartz. Creating the CIPRES Science Gateway for inference of large phylogenetic trees. In: 2010 Gateway Computing Environments Workshop (GCE). 2010. p. 1–8.
  • 42.Rambaut A. Figtree. 2019. https://github.com/rambaut/figtree/releases.
  • 43.Meier-Kolthoff JP, Göker M. TYGS is an automated high-throughput platform for state-of-the-art genome-based taxonomy. Nat Commun. 2019;10:2182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Eren AM, Kiefl E, Shaiber A, Veseli I, Miller SE, Schechter MS. Community-led, integrated, reproducible multi-omics with anvi’o. Nat Microbiol. 2021. [DOI] [PMC free article] [PubMed]
  • 45.Eddy SR. Accelerated profile HMM searches. PLoS Comput Biol. 2011;7:1–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Hyatt D, Chen G-L, LoCascio PF, Land ML, Larimer FW, Hauser LJ. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics. 2010;11: 119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Galperin MY, Kristensen DM, Makarova KS, Wolf YI, Koonin EV. Microbial genome analysis: the COG approach. Brief Bioinform. 2019;20:1063–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Jain C, Rodriguez-R LM, Phillippy AM, Konstantinidis KT, Aluru S. High throughput ANI analysis of 90K prokaryotic genomes reveals clear species boundaries. Nat Commun. 2018;9:5114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Kieft K, Zhou Z, Anantharaman K. VIBRANT: automated recovery, annotation and curation of microbial viruses, and evaluation of viral community function from genomic sequences. Microbiome. 2020;8:90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Bouras G, Nepal R, Houtak G, Psaltis AJ, Wormald P-J, Vreugde S. Pharokka: a fast scalable bacteriophage annotation tool. Bioinformatics. 2023;39:btac776. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Schwengers O, Jelonek L, Dieckmann MA, Beyvers S, Blom J, Goesmann A. Bakta: rapid and standardized annotation of bacterial genomes via alignment-free sequence identification. Microb Genom. 2021;7:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Camargo AP, Roux S, Schulz F, Babinski M, Xu Y, Hu B, et al. Identification of mobile genetic elements with geNomad. Nat Biotechnol. 2023. 10.1038/s41587-023-01953-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Nayfach S, Camargo AP, Schulz F, Eloe-Fadrosh E, Roux S, Kyrpides NC. CheckV assesses the quality and completeness of metagenome-assembled viral genomes. Nat Biotechnol. 2021;39:578–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Darling AE, Mau B, Perna NT. progressiveMauve: multiple genome alignment with gene gain, loss and rearrangement. PLoS ONE. 2010;5: e11147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, et al. Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics. 2012;28:1647–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Zhang J, Guan J, Wang M, Li G, Djordjevic M, Tai C, et al. SecReT6 update: a comprehensive resource of bacterial Type VI Secretion Systems. Sci China Life Sci. 2023;66:626–34. [DOI] [PubMed] [Google Scholar]
  • 57.Csűös M. Count: evolutionary analysis of phylogenetic profiles with parsimony and likelihood. Bioinformatics. 2010;26:1910–2. [DOI] [PubMed] [Google Scholar]
  • 58.Tesson F, Hervé A, Mordret E, Touchon M, d’Humières C, Cury J, et al. Systematic and quantitative view of the antiviral arsenal of prokaryotes. Nat Commun. 2022;13:2561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Neron B, Denise R, Coluzzi C, et al. MacSyFinder v2: Improved modelling and search engine to identify molecular systems in genomes. bioRxiv; 2022. 10.1101/2022.09.02.506364.
  • 60.Tesson F, Planel R, Egorov A, Georjon H, Vaysset H, Brancotte B, et al. A comprehensive resource for exploring antiphage defense: DefenseFinder Webservice, Wiki and Databases. 2024:2024.01.25.577194.
  • 61.Veesenmeyer JL, Andersen AW, Lu X, Hussa EA, Murfin KE, Chaston JM, et al. NilD CRISPR RNA contributes to Xenorhabdus nematophila colonization of symbiotic host nematodes. Mol Microbiol. 2014;93:1026–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Biswas A, Staals RHJ, Morales SE, Fineran PC, Brown CM. CRISPRDetect: A flexible algorithm to define CRISPR arrays. BMC Genomics. 2016;17:356. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Couvin D, Bernheim A, Toffano-Nioche C, Touchon M, Michalik J, Néron B, et al. CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and integrates search for Cas proteins. Nucleic Acids Res. 2018;46:W246–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Biswas A, Gagnon JN, Brouns SJJ, Fineran PC, Brown CM. CRISPRTarget: bioinformatic prediction and analysis of crRNA targets. RNA Biol. 2013;10:817–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Blin K, Shaw S, Augustijn HE, Reitz ZL, Biermann F, Alanjary M, et al. antiSMASH 7.0: new and improved predictions for detection, regulation, chemical structures and visualisation. Nucleic Acids Res. 2023;51:W46–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Terlouw BR, Blin K, Navarro-Muñoz JC, Avalon NE, Chevrette MG, Egbert S, et al. MIBiG 3.0: a community-driven effort to annotate experimentally validated biosynthetic gene clusters. Nucleic Acids Res. 2022;51:D603–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Stachelhaus T, Mootz HD, Marahiel MA. The specificity-conferring code of adenylation domains in nonribosomal peptide synthetases. Chem Biol. 1999;6:493–505. [DOI] [PubMed] [Google Scholar]
  • 68.Rausch C, Hoof I, Weber T, Wohlleben W, Huson DH. Phylogenetic analysis of condensation domains in NRPS sheds light on their functional evolution. BMC Evol Biol. 2007;7: 78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.van Santen JA, Poynton EF, Iskakova D, McMann E, Alsup TA, Clark TN, et al. The Natural Products Atlas 2.0: a database of microbially-derived natural products. Nucleic Acids Res. 2021;50:D1317–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Bah T. Inkscape: guide to a vector drawing program. New Jersey: Prentice Hall Press; 2011. 
  • 71.Kim I-H, Aryal SK, Aghai DT, Casanova-Torres ÁM, Hillman K, Kozuch MP, et al. The insect pathogenic bacterium Xenorhabdus innexi has attenuated virulence in multiple insect model hosts yet encodes a potent mosquitocidal toxin. BMC Genomics. 2017;18:927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.de Nies L, Lopes S, Busi SB, Galata V, Heintz-Buschart A, Laczny CC, et al. PathoFact: a pipeline for the prediction of virulence factors and antimicrobial resistance genes in metagenomic data. Microbiome. 2021;9:49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Abby SS, Denise R, Rocha EPC. Identification of protein secretion systems in bacterial genomes using MacSyFinder version 2. Methods Mol Biol. 2024;2715:1–25. [DOI] [PubMed] [Google Scholar]
  • 74.Jalili V, Afgan E, Gu Q, Clements D, Blankenberg D, Goecks J, et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2020 update. Nucleic Acids Res. 2020;48:W395–402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Papoulis SE, Wilhelm SW, Talmy D, Zinser ER. Nutrient loading and viral memory drive accumulation of restriction modification systems in bloom-forming cyanobacteria. mBio. 2021;12. 10.1128/mbio.00873-21. [DOI] [PMC free article] [PubMed]
  • 76.Kobayashi I. Behavior of restriction–modification systems as selfish mobile elements and their impact on genome evolution. Nucleic Acids Res. 2001;29:3742–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 77.Teklemariam AD, Al-Hindi RR, Qadri I, Alharbi MG, Ramadan WS, Ayubu J, et al. The battle between bacteria and bacteriophages: a conundrum to their immune system. Antibiotics. 2023;12: 381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 78.Hussa E, Goodrich-Blair H. Rearing and injection of Manduca sexta larvae to assess bacterial virulence. JoVE. 2012;(70):e4295. 10.3791/4295. [DOI] [PMC free article] [PubMed]
  • 79.Rodak NY, Tan CH, Sternberg PW. An improved solid medium-based culturing method for Steinernema hermaphroditum. MicroPubl Biol. 2024;2024:10.17912/micropub.biology.001110. 10.17912/micropub.biology.001110. [DOI] [PMC free article] [PubMed]
  • 80.Baniya A, Subkrasae C, Ardpairin J, Anesko K, Vitta A, Dillman AR. Steinernema adamsi n. sp. (Rhabditida: Steinernematidae), A new entomopathogenic nematode from Thailand. Journal of Parasitology. 2024;110:22–39. [DOI] [PubMed] [Google Scholar]
  • 81.Meier-Kolthoff JP, Klenk HP, Göker M. Taxonomic use of DNA G+C content and DNA-DNA hybridization in the genomic age. Int J Syst Evol Microbiol. 2014;64 Pt 2:352–6. [DOI] [PubMed] [Google Scholar]
  • 82.Meier-Kolthoff JP, Hahnke RL, Petersen J, Scheuner C, Michael V, Fiebig A, et al. Complete genome sequence of DSM 30083T, the type strain (U5/41T) of Escherichia coli, and a proposal for delineating subspecies in microbial taxonomy. Stand Genomic Sci. 2014;9: 2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 83.Awori RM, Hendre P, Amugune NO. The genome of a steinernematid-associated Pseudomonas piscis bacterium encodes the biosynthesis of insect toxins. Access Microbiol. 2023;5(10):000659–v3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 84.Chaston JM, Suen G, Tucker SL, Andersen AW, Bhasin A, Bode E, et al. The entomopathogenic bacterial endosymbionts Xenorhabdus and Photorhabdus: convergent lifestyles from divergent genomes. PLoS ONE. 2011;6:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 85.Kochanowsky RM, Bradshaw C, Forlastro I, Stock SP. Xenorhabdus bovienii strain jolietti uses a type 6 secretion system to kill closely related Xenorhabdus strains. FEMS Microbiol Ecol. 2020;96:fiaa073. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 86.Konstantinidis KT, Tiedje JM. Genomic insights that advance the species definition for prokaryotes. Proc Natl Acad Sci. 2005;102:2567–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 87.Morales-Soto N, Forst SA. The xnp1 P2-like tail synthesis gene cluster encodes xenorhabdicin and is required for interspecies competition. J Bacteriol. 2011;193:3624–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 88.Morales-Soto N, Gaudriault S, Ogier J-C, Thappeta KRV, Forst S. Comparative analysis of P2-type remnant prophage loci in Xenorhabdus bovienii and Xenorhabdus nematophila required for xenorhabdicin production. FEMS Microbiol Lett. 2012;333:69–76. [DOI] [PubMed] [Google Scholar]
  • 89.Carim S, Azadeh AL, Kazakov AE, Price MN, Walian PJ, Lui LM, et al. Systematic discovery of pseudomonad genetic factors involved in sensitivity to tailocins. ISME J. 2021;15:2289–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 90.Jurėnas D, Fraikin N, Goormaghtigh F, Van Melderen L. Biology and evolution of bacterial toxin–antitoxin systems. Nat Rev Microbiol. 2022;20:335–50. [DOI] [PubMed] [Google Scholar]
  • 91.Lee J-Y, Rahman FU, Kim E-K, Cho S-M, Kim H-R, Lee K, et al. Importin-11 is essential for normal embryonic development in mice. Int J Med Sci. 2020;17:815–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 92.Yu Z, Zhang W, Yang H, Chou SH, Galperin MY, He J. Gas and light: triggers of c-di-GMP-mediated regulation. FEMS Microbiol Rev. 2023;47:fuad034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 93.Georjon H, Bernheim A. The highly diverse antiphage defence systems of bacteria. Nat Rev Microbiol. 2023;21:686–700. [DOI] [PubMed] [Google Scholar]
  • 94.Engelberg-Kulka H, Hazan R, Amitai S. mazEF: a chromosomal toxin-antitoxin module that triggers programmed cell death in bacteria. J Cell Sci. 2005;118(Pt 19):4327–32. [DOI] [PubMed] [Google Scholar]
  • 95.Lopatina A, Tal N, Sorek R. Abortive infection: Bacterial suicide as an antiviral immune strategy. Annu Rev Virol. 2020;7:371–84. [DOI] [PubMed] [Google Scholar]
  • 96.Lemay M-L, Horvath P, Moineau S. The CRISPR-Cas app goes viral. Curr Opin Microbiol. 2017;37:103–9. [DOI] [PubMed] [Google Scholar]
  • 97.Hampton HG, Watson BNJ, Fineran PC. The arms race between bacteria and their phage foes. Nature. 2020;577:327–36. [DOI] [PubMed] [Google Scholar]
  • 98.Thompson CP, Doak AN, Amirani N, Schroeder EA, Wright J, Kariyawasam S, et al. High-resolution identification of multiple Salmonella serovars in a single sample by using CRISPR-SeroSeq. Appl Environ Microbiol. 2018;84(21):e01859–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 99.Kushwaha SK, Narasimhan LP, Chithananthan C, Marathe SA. Clustered regularly interspaced short palindromic repeats-Cas system: diversity and regulation in Enterobacteriaceae. Future Microbiol. 2022;17:1249–67. [DOI] [PubMed] [Google Scholar]
  • 100.Makarova KS, Wolf YI, Koonin EV. In silico approaches for prediction of anti-CRISPR proteins. J Mol Biol. 2023;435: 168036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 101.Bickle TA, Krüger DH. Biology of DNA restriction. Microbiol Rev. 1993;57:434–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 102.Roberts RJ, Belfort M, Bestor T, Bhagwat AS, Bickle TA, Bitinaite J, et al. A nomenclature for restriction enzymes, DNA methyltransferases, homing endonucleases and their genes. Nucleic Acids Res. 2003;31:1805–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 103.Gavriilidou A, Kautsar SA, Zaburannyi N, Krug D, Müller R, Medema MH, et al. Compendium of specialized metabolite biosynthetic diversity encoded in bacterial genomes. Nat Microbiol. 2022;7:726–35. [DOI] [PubMed] [Google Scholar]
  • 104.Li J-H, Cho W, Hamchand R, Oh J, Crawford JM. A conserved nonribosomal peptide synthetase in Xenorhabdus bovienii produces citrulline-functionalized lipopeptides. J Nat Prod. 2021;84:2692–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 105.Crawford JM, Portmann C, Zhang X, Roeffaers MBJ, Clardy J. Small molecule perimeter defense in entomopathogenic bacteria. Proc Natl Acad Sci. 2012;109:10821–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 106.Grammbitter GLC, Schmalhofer M, Karimi K, Shi Y-M, Schöner TA, Tobias NJ, et al. An uncommon type II PKS catalyzes biosynthesis of aryl polyene pigments. J Am Chem Soc. 2019;141:16615–23. [DOI] [PubMed] [Google Scholar]
  • 107.Vior NM, Lacret R, Chandra G, Dorai-Raj S, Trick M, Truman AW. Discovery and biosynthesis of the antibiotic bicyclomycin in distantly related bacterial classes. Appl Environ Microbiol. 2018;84:e02828. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 108.Velasco A, Acebo P, Gomez A, Schleissner C, Rodríguez P, Aparicio T, et al. Molecular characterization of the safracin biosynthetic pathway from Pseudomonas fluorescens A2–2: designing new cytotoxic compounds. Mol Microbiol. 2005;56:144–54. [DOI] [PubMed] [Google Scholar]
  • 109.Nollmann FI, Dauth C, Mulley G, Kegler C, Kaiser M, Waterfield NR, et al. Insect-specific production of new GameXPeptides in Photorhabdus luminescens TTO1, widespread natural products in entomopathogenic bacteria. ChemBioChem. 2015;16:205–8. [DOI] [PubMed] [Google Scholar]
  • 110.Fuchs SW, Proschak A, Jaskolla TW, Karas M, Bode HB. Structure elucidation and biosynthesis of lysine-rich cyclic peptides in Xenorhabdus nematophila. Org Biomol Chem. 2011;9:3130–2. [DOI] [PubMed] [Google Scholar]
  • 111.Dreyer J, Rautenbach M, Booysen E, van Staden AD, Deane SM, Dicks LMT. Xenorhabdus khoisanae SB10 produces Lys-rich PAX lipopeptides and a Xenocoumacin in its antimicrobial complex. BMC Microbiol. 2019;19:132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 112.Zhou Q, Grundmann F, Kaiser M, Schiell M, Gaudriault S, Batzer A, et al. Structure and biosynthesis of Xenoamicins from entomopathogenic Xenorhabdus. Chemistry – A European Journal. 2013;19:16772–9. [DOI] [PubMed] [Google Scholar]
  • 113.Imai Y, Meyer KJ, Iinishi A, Favre-Godal Q, Green R, Manuse S, et al. A new antibiotic selectively kills Gram-negative pathogens. Nature. 2019;576:459–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 114.Daborn PJ, Waterfield N, Silva CP, Au CPY, Sharma S, Ffrench-Constant RH. A single Photorhabdus gene, makes caterpillars floppy (mcf), allows Escherichia coli to persist within and kill insects. Proc Natl Acad Sci. 2002;99:10742–7. [DOI] [PMC free article] [PubMed]
  • 115.Waterfield NR, Daborn PJ, Dowling AJ, Yang G, Hares M, Ffrench-Constant RH. The insecticidal toxin makes caterpillars floppy 2 (Mcf2) shows similarity to HrmA, an avirulence protein from a plant pathogen. FEMS Microbiol Lett. 2003;229:265–70. [DOI] [PubMed] [Google Scholar]
  • 116.Wang H-C, Lin S-J, Wang H-C, Kumar R, Le PT, Leu J-H. A bacterial binary toxin system that kills both insects and aquatic crustaceans: Photorhabdus insect-related toxins A and B. PLoS Pathog. 2023;19: e1011330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 117.Satchell KJF. Multifunctional-autoprocessing repeats-in-toxin (MARTX) Toxins of Vibrios. Microbiol Spectr. 2015;3:3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 118.Roderer D, Raunser S. Tc toxin complexes: assembly, membrane permeation, and protein translocation. Annu Rev Microbiol. 2019;73:247–65. [DOI] [PubMed] [Google Scholar]
  • 119.Wang X, Cheng J, Shen J, Liu L, Li N, Gao N, et al. Characterization of Photorhabdus Virulence Cassette as a causative agent in the emerging pathogen Photorhabdus asymbiotica. Sci China Life Sci. 2022;65:618–30. [DOI] [PubMed] [Google Scholar]
  • 120.Jiang F, Li N, Wang X, Cheng J, Huang Y, Yang Y, et al. Cryo-EM structure and assembly of an extracellular contractile injection system. Cell. 2019;177:370–383.e15. [DOI] [PubMed] [Google Scholar]
  • 121.Chen L, Song N, Liu B, Zhang N, Alikhan N-F, Zhou Z, et al. Genome-wide identification and characterization of a superfamily of bacterial extracellular contractile injection systems. Cell Rep. 2019;29:511–521.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 122.Jiang F, Shen J, Cheng J, Wang X, Yang J, Li N, et al. N-terminal signal peptides facilitate the engineering of PVC complex as a potent protein delivery system. Sci Adv. 2022;8: eabm2343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 123.Flury P, Vesga P, Péchy-Tarr M, Aellen N, Dennert F, Hofer N, et al. Antimicrobial and insecticidal: cyclic lipopeptides and hydrogen cyanide produced by plant-beneficial Pseudomonas strains cHA0, CMR12a, and PCL1391 contribute to insect killing. Front Microbiol. 2017;8:100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 124.Fasano A, Fiorentini C, Donelli G, Uzzau S, Kaper JB, Margaretten K, et al. Zonula occludens toxin modulates tight junctions through protein kinase C-dependent actin reorganization, in vitro. J Clin Invest. 1995;96:710–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 125.Pierro MD, Lu R, Uzzau S, Wang W, Margaretten K, Pazzani C, et al. Zonula occludens toxin structure-function analysis: Identification of the fragment biologically active on tight junctions and of the zonulin receptor binding domain. J Biol Chem. 2001;276:19160–5. [DOI] [PubMed] [Google Scholar]
  • 126.Gao M, Dong S, Hu X, Zhang X, Liu Y, Zhong J, et al. Roles of midgut cadherin from two moths in different Bacillus thuringiensis action mechanisms: correlation among toxin binding, cellular toxicity, and synergism. J Agric Food Chem. 2019;67:13237–46. [DOI] [PubMed] [Google Scholar]
  • 127.Cao M. CRISPR-Cas9 genome editing in Steinernema entomopathogenic nematodes. bioRxiv [Preprint]. 2023:2023.11.24.568619. 10.1101/2023.11.24.568619.
  • 128.Hugenholtz P, Chuvochina M, Oren A, Parks DH, Soo RM. Prokaryotic taxonomy and nomenclature in the age of big sequence data. ISME J. 2021;15:1879–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 129.Mothupi B, Featherston J, Gray V. Draft whole-genome sequence and annotation of Xenorhabdus griffiniae strain BMMCB associated with the South African entomopathogenic nematode Steinernema khoisanae strain BMMCB. Genome Announc. 2015;3:e00785–815. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 130.Linz B, Ivanov YV, Preston A, Brinkac L, Parkhill J, Kim M, et al. Acquisition and loss of virulence-associated factors during genome evolution and speciation in three clades of Bordetella species. BMC Genomics. 2016;17:767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 131.Kirchberger PC, Schmidt ML, Ochman H. The ingenuity of bacterial genomes. Annu Rev Microbiol. 2020;74:815–34. [DOI] [PubMed] [Google Scholar]
  • 132.Lugli GA, Milani C, Turroni F, Tremblay D, Ferrario C, Mancabelli L, et al. Prophages of the genus Bifidobacterium as modulating agents of the infant gut microbiota. Environ Microbiol. 2016;18:2196–213. [DOI] [PubMed] [Google Scholar]
  • 133.Mottawea W, Duceppe M-O, Dupras AA, Usongo V, Jeukens J, Freschi L, et al. Salmonella enterica prophage sequence profiles reflect genome diversity and can be used for high discrimination subtyping. Front Microbiol. 2018;9:836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 134.Yu J, Xu X, Wang Y, Zhai X, Pan Z, Jiao X, et al. Prophage-mediated genome differentiation of the Salmonella Derby ST71 population. Microb Genom. 2022;8:8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 135.Singh RP, Kumari K. Bacterial type VI secretion system (T6SS): an evolved molecular weapon with diverse functionality. Biotechnol Lett. 2023;45:309–31. [DOI] [PubMed] [Google Scholar]
  • 136.Pothula R, Lee M-W, Patricia SS. Type 6 secretion system components hcp and vgrG support mutualistic partnership between Xenorhabdus bovienii symbiont and Steinernema jollieti host. J Invertebr Pathol. 2023;198: 107925. [DOI] [PubMed] [Google Scholar]
  • 137.Zhang D, de Souza RF, Anantharaman V, Iyer LM, Aravind L. Polymorphic toxin systems: Comprehensive characterization of trafficking modes, processing, mechanisms of action, immunity and ecology using comparative genomics. Biol Direct. 2012;7: 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 138.McMullen JG, McQuade R, Ogier J-C, Pagès S, Gaudriault S, Patricia SS. Variable virulence phenotype of Xenorhabdus bovienii (γ-Proteobacteria: Enterobacteriaceae) in the absence of their vector hosts. Microbiology (Reading). 2017;163:510–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 139.Park Y, Herbert EE, Cowles CE, Cowles KN, Menard ML, Orchard SS, et al. Clonal variation in Xenorhabdus nematophila virulence and suppression of Manduca sexta immunity. Cell Microbiol. 2007;9:645–56. [DOI] [PubMed] [Google Scholar]
  • 140.Sugar DR, Murfin KE, Chaston JM, Andersen AW, Richards GR, deLéon L, et al. Phenotypic variation and host interactions of Xenorhabdus bovienii SS-2004, the entomopathogenic symbiont of Steinernema jollieti nematodes. Environ Microbiol. 2012;14:924–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 141.Hussa EA, Casanova-Torres ÁM, Goodrich-Blair H. The global transcription factor Lrp controls virulence modulation in Xenorhabdus nematophila. J Bacteriol. 2015;197:3015–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 142.Cao M, Goodrich-Blair H. Xenorhabdus nematophila bacteria shift from mutualistic to virulent Lrp-dependent phenotypes within the receptacles of Steinernema carpocapsae insect-infective stage nematodes. Environ Microbiol. 2020;22:5433–49. [DOI] [PubMed] [Google Scholar]
  • 143.Bisch G, Pagès S, McMullen JG, Stock SP, Duvic B, Givaudan A, et al. Xenorhabdus bovienii CS03, the bacterial symbiont of the entomopathogenic nematode Steinernema weiseri, is a non-virulent strain against lepidopteran insects. J Invertebr Pathol. 2015;124:15–22. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

12864_2024_10858_MOESM1_ESM.docx (1.6MB, docx)

Additional file 1. Word document containing supporting data for the detection and analysis of secretion systems, maximum likelihood phylogenies, prophages and type six effector protein loci identification, MARTX locus alignments, and dose response assays for strains used in insect virulence assays. Table S1: Comparison of the number of putative secretion systems that are present Xenorhabdus griffinae strains. Table S2. Loci encoding putative restriction modification systems identified by homology and genome annotation searches. Table S3: Biosynthetic gene clusters (BGCs) in the X. griffiniae ID10 genome. Figure S1: Maximum likelihood phylogenetic tree created using RAxML with one-to-one orthologs from Xenorhabdus type strains, closely related X. griffiniae strains, and Photorhabdus asymbiotica as an outgroup. Figure S2: Loci of prophages and xnp1 in the complete Xenorhabdus sp. TH1 genome. Figure S3: Dotplots of prophage loci that were considerably similar between strains. Figure S4: Schematic of subspecies-specific type six secretion system (T6SS) effector-encoding loci in six strains of Xenorhabdus griffiniae. Figure S5: MARTX protein multiple sequence alignments Figure S6: Pdp1 protein multiple sequence alignment Figure S7: Percentage survival of Manduca sexta post injection with multiple concentrations of Xenorhabdus bacteria strains tested.

12864_2024_10858_MOESM2_ESM.xlsx (11.1MB, xlsx)

Additional file 2. Excel workbook containing raw data and results of all genome analyses. Data tables which collectively contain raw data and values from pangenome, dDDH, ANI, biosynthetic gene cluster, prophage analyses; defence systems; phyletic patterns; genome accession numbers and names of strains used.

12864_2024_10858_MOESM3_ESM.docx (12.8MB, docx)

Additional file 3. Word document with additional details of the detection of Xenorhabdus griffiniae defense systems and analysis of CRISPR loci. Table S4. Summary of defense systems in X. griffiniae and related strains. Table S5. Locus tags and coordinates of CRISPR-Cas features. Table S6. CRISPR spacers, protospacers, concomitant annotation and genome wherein they are located. Methods and results for X. griffiniae CRISPR repeats, protospacers and self-targeting immunity.

12864_2024_10858_MOESM4_ESM.xlsx (82KB, xlsx)

Additional file 4. Excel workbook of putative insect toxins encoded by Xenorhabdus strains. These sheets contain raw data of the toxin domain-containing proteins that are unique among the X. griffiniae and Xenorhabdus sp. TH1 species. Also contained herein, are raw data from the comparison between the combined toxin library outputs for the X. griffiniae and Xenorhabdus sp. TH1 species with the output for X nematophila 19061. Also provided are the raw “toxin library” summary outputs from the PathoFact software used in the analyses.

Data Availability Statement

The datasets generated during the current study are available in the GenBank repository under BioProject ID: PRJNA1085699. Previously reported data are available from public repositories (see Methods), and additional data are provided within the manuscript or supplementary information files.


Articles from BMC Genomics are provided here courtesy of BMC

RESOURCES