The Xenorhabdus nematophila nilABC Genes Confer the Ability of Xenorhabdus spp. To Colonize Steinernema carpocapsae Nematodes

Abstract

Members of the Steinernema genus of nematodes are colonized mutualistically by members of the Xenorhabdus genus of bacteria. In nature, Steinernema carpocapsae nematodes are always found in association with Xenorhabdus nematophila bacteria. Thus, this interaction, like many microbe-host associations, appears to be species specific. X. nematophila requires the nilA, nilB, and nilC genes to colonize S. carpocapsae. In this work, we showed that of all the Xenorhabdus species examined, only X. nematophila has the nilA, nilB, and nilC genes. By exposing S. carpocapsae to other Xenorhabdus spp., we established that only X. nematophila is able to colonize S. carpocapsae; therefore, the S. carpocapsae-X. nematophila interaction is species specific. Further, we showed that introduction of the nilA, nilB, and nilC genes into other Xenorhabdus species enables them to colonize the same S. carpocapsae host tissue that is normally colonized by X. nematophila. Finally, sequence analysis supported the idea that the nil genes were horizontally acquired. Our findings indicate that a single genetic locus determines host specificity in this bacteria-animal mutualism and that host range expansion can occur through the acquisition of a small genetic element.

Microbial associations with plant and animal hosts are ubiquitous in nature and tend to show specificity with regard to the taxonomy of each partner. For example, Salmonella enterica serotype Typhi (29), Helicobacter pylori (5), Neisseria gonorrhoeae (28), and Haemophilus influenzae (22) are pathogens with specificities for human or primate hosts, while Rhizobium leguminosarum forms mutualistic nodules only with Phaseolae legumes (30). The molecular basis underlying host range specificity is well understood in the mutualism between the Rhizobiaceae and their plant hosts: specificity is achieved by cross talk involving bacterial and plant signaling factors and receptors (15). In contrast, species-specific interactions between bacteria and animal hosts are not well understood at the molecular or genetic level. Although myriad traits have been proposed to contribute to host range specificity (16, 21, 36), none of these have been directly demonstrated to play a role in dictating the animal host range. The increasing global concern over new human pathogens emerging through host range expansion makes understanding the genetic basis of host range specificity in bacteria-animal associations essential (6, 27).

Insect parasitic nematodes of the genus Steinernema are mutualistically associated with bacteria of the Xenorhabdus genus. This association is a natural and tractable model for understanding the ecology, evolution, and molecular foundations of bacterial interactions with animal hosts. The soil-inhabiting infective stage of a Steinernema nematode is colonized by symbiotic Xenorhabdus bacteria, which it carries and releases into an insect host. Xenorhabdus bacteria provide activities that suppress insect immunity, kill the insect, and enzymatically degrade the cadaver to support nematode reproduction. When the insect cadaver is depleted, Xenorhabdus bacteria colonize progeny Steinernema nematodes, which emerge from the spent insect cadaver to hunt for new insect prey (14).

Field and phylogenetic studies indicate that specific pairs of Xenorhabdus and Steinernema species occur in nature (10, 11, 35). Furthermore, several studies have demonstrated that certain Steinernema-Xenorhabdus associations are exclusive in that noncognate pairs will not associate during experimental mixing (1, 31). Therefore, it is likely that Xenorhabdus bacteria have evolved specificity for their cognate Steinernema nematode hosts and vice versa. In nature, only X. nematophila has been found to be associated with S. carpocapsae nematodes, although a thorough investigation of the range of Xenorhabdus spp. that can colonize S. carpocapsae has not been reported.

The X. nematophila nilA, nilB, and nilC genes were previously identified in a signature-tagged mutagenesis screen that was designed to elucidate bacterial genes necessary for colonization of the S. carpocapsae nematode host (17). In this initial work, it was also revealed that the nilABC gene cluster is absent from two other Xenorhabdus species, X. poinarii and X. beddingii. In the present study, we assessed the role of the nilA, nilB, and nilC genes in species specificity.

MATERIALS AND METHODS

Strains, plasmids, media, growth conditions, and molecular biological methods.

The bacterial strains and plasmid constructs used in this study are listed in Tables 1 and 2, respectively. Except where noted, the bacteria were grown in Luria-Bertani (LB) broth (25) at 30°C that for the Xenorhabdus spp. was either stored in the dark or supplemented with 0.1% sodium pyruvate (39). Antibiotics were used at the following concentrations: ampicillin (Ap), 30 μg/ml; chloramphenicol, 30 μg/ml; kanamycin (Km), 50 μg/ml; and erythromycin (Erm), 200 μg/ml.

TABLE 1.

Xenorhabdus strains used in this study

Straina	Relevant characteristics	Source or reference
HGB007	Sequenced ATCC 19061 X. nematophila wild type	ATCC
HGB777	HGB007 parent, Δ(nilA-nilC)7-Kmr (XnΔSR1)	7
HGB1186	HGB777 parent, attTn7::miniTn7 from pEVS107	This study
HGB778	HGB777 parent, attTn7::miniTn7 from pTn7/SR1	7
HGB1182	HGB777 parent, attTn7::miniTn7 from pTn7/SR1/AM1Z (nilA21 M1Z)	This study
HGB1183	HGB777 parent, attTn7::miniTn7 from pTn7/SR1/BM1Z (nilB20 M1Z)	This study
HGB779	HGB777 parent, attTn7::miniTn7 from pTn7/SR1/CM1Z (nilC19 M1Z)	7
HGB1184	HGB777 parent, attTn7::miniTn7 from pTn7/SR1/TnM1Z	This study
HGB340	HGB007 parent, chromosomal integration of pECM20	23
HGB003	ATCC 35271 X. bovienii unsequenced wild-type isolate	ATCC
HGB1055	X. bovienii sequenced wild-type isolate	Monsanto
HGB1166	HGB003 parent, attTn7::miniTn7 from pEVS107	This study
HGB1167	HGB003 parent, attTn7::miniTn7 from pTn7/SR1	This study
HGB1168	HGB003 parent, attTn7::miniTn7 from pTn7/SR1/AM1Z	This study
HGB1169	HGB003 parent, attTn7::miniTn7 from pTn7/SR1/BM1Z	This study
HGB1170	HGB003 parent, attTn7::miniTn7 from pTn7/SR1/CM1Z	This study
HGB1171	HGB003 parent, attTn7::miniTn7 from pTn7/SR1/TnM1Z	This study
HGB1180	HGB003 parent, attTn7::miniTn7 from pTn7/rrs/GFP	This study
HGB1172	HGB003 parent, attTn7::miniTn7 from pTn7/SR1/rrs/GFP	This study
HGB086	ATCC 49121 X. poinarii wild-type isolate	ATCC
HGB1173	HGB086 parent, chromosomal integration of pTn7/rrs	This study
HGB1174	HGB086 parent, chromosomal integration of pTn7/SR1/rrs	This study
HGB1175	HGB086 parent, chromosomal integration of pTn7/SR1/AM1Z/rrs	This study
HGB1176	HGB086 parent, chromosomal integration of pTn7/SR1/BM1Z/rrs	This study
HGB1177	HGB086 parent, chromosomal integration of pTn7/SR1/CM1Z/rrs	This study
HGB1178	HGB086 parent, chromosomal integration of pTn7/SR1/TnM1Z/rrs	This study
HGB1181	HGB086 parent, chromosomal integration of pTn7/rrs/GFP	This study
HGB1179	HGB086 parent, chromosomal integration of pTn7/SR1/rrs/GFP	This study
HGB836	DSM 16338 X. szentirmaii wild-type isolate	20; A. Fodor
HGB1322	HGB836 parent, attTn7::miniTn7 from pEVS107	This study
HGB1323	HGB836 parent, attTn7::miniTn7 from Tn7/SR1	This study
HGB084	ATCC 49542 X. beddingii wild-type isolate	ATCC
HGB833	DSM 16342 X. budapestensis wild-type isolate	20; A. Fodor
HGB834	DSM 16337 X. ehlersii wild-type isolate	20; A. Fodor
HGB835	DSM 16336 X. innexi wild-type isolate	20; A. Fodor

^aHGB stock number of Xenorhabdus strain.

TABLE 2.

Plasmids used in this study

Plasmid	Relevant propertiesa	Source or reference
pECM9	Source of PaphA-gfp amplicon; Apr	23
pECM20	mobRP4, oriR6K integrates into X. nematophila chromosome to express gfp; Cmr	23
pEVS107	oriR6K, mobRP4, mobilizable suicide miniTn7-Erm delivery vector; Ermr Kmr	34
pUX-BF13	Mobilizable Tn7 transposition helper plasmid that expresses Tn7 transposase in trans; Apr	2
pTn7/SR1	pEVS107 parent, delivers miniTn7-Erm/nil	7
pTn7/SR1/AM1Z	pTn7/SR1 parent with nilA21 ATG to TAG point mutation	This study
pTn7/SR1/BM1Z	pTn7/SR1 parent with nilB20 ATG to TAG point mutation	This study
pTn7/SR1/CM1Z	pTn7/SR1 parent with nilC19 ATG to TAG point mutation	7
pTn7/SR1/TnM1Z	pTn7/SR1 parent with tn2 ATG to TAG point mutation	This study
pTn7/rrs	pEVS107 parent, integrates into X. poinarii rrs	This study
pTn7/SR1/rrs	pTn7/SR1 parent, integrates into X. poinarii rrs	This study
pTn7/SR1/AM1Z/rrs	pTn7/SR1/AM1Z parent, integrates into X. poinarii rrs	This study
pTn7/SR1/BM1Z/rrs	pTn7/SR1/BM1Z parent, integrates into X. poinarii rrs	This study
pTn7/SR1/CM1Z/rrs	pTn7/SR1/CM1Z parent, integrates into X. poinarii rrs	This study
pTn7/SR1/TnM1Z/rrs	pTn7/SR1/TnM1Z parent, integrates into X. poinarii rrs	This study
pTn7/rrs/GFP	pTn7/rrs parent, integrates into Xenorhabdus rrs, constitutively expresses GFP	This study
pTn7/SR1/rrs/GFP	pTn7/SR1/rrs parent, integrates into Xenorhabdus rrs, constitutively expresses GFP	This study

^aCm, chloramphenicol.

Construction of the nilA, nilB, and tn2 start codon mutations was done as previously described for the nilC start codon mutation (7). Briefly, the genes were mutated by amplifying pTn7/SR1 (SR1 stands for “symbiosis region 1,” the original nomenclature used for the nilABC region [17]) plasmid DNA with the primers noted in Table 3 and Platinum Pfx (Invitrogen). The amplified plasmid was subsequently digested with DpnI to cut parental methylated DNA and was then transformed into Escherichia coli S17-1(λpir). Plasmids were conjugated from E. coli S17-1(λpir) into Xenorhabdus spp. through triparental conjugations with a pUX-BF13 helper plasmid (2) as previously described (7). The correct insertion of Tn7 constructs into the attTn7 site of the X. nematophila chromosome was confirmed by using PCR with AttTn7EXT and ErmAnch1 primers (Table 3). The transposition of Tn7 was unsuccessful in X. poinarii; thus, a 1,477-bp EcoRI fragment of X. poinarii chromosomal DNA (carrying a 16S rRNA gene [rrs] region; GenBank accession no. D78010) was first cloned into the EcoRI site of pBluescript (Stratagene) and then directionally subcloned into the EcoRI/PvuII sites of all miniTn7 plasmid constructs by digestion with EcoRI and EcoRV (Table 2), thereby allowing integration into the chromosome with homologous recombination.

TABLE 3.

Oligonucleotides used in this study

Oligonucleotide	5′ to 3′ sequencea	Purpose
AttTn7Ext	TGTTGGTTCACATCC	Tn7 insert amplification
ErmAnch1	TACTTATGAGCAAGTATTGTC	Tn7 insert amplification
AMtoZfornew	GACAAAATGACTCTAGATTTCTAATTATTATCCTAATCTGC	Mutagenesis
AMtoZrevnew	GCAGATTAGGATAATAATTAGAAATCTAGAGTCATTTTGTC	Mutagenesis
BMtoZfor	TGGATTTGATTTTTTTCTAAGAGTTTTCATTGGTAATG	Mutagenesis
BMtoZrev	TACCAATGAAAACTCTTAGAAAAAAATCAAATCCATCG	Mutagenesis
TnMtoZfor	TTTATGGATGTATTCGCTCTACCATTTAGTTTACGCCGC	Mutagenesis
TnMtoZrev	GCGTAAACTAAATGGTAGAGCGAATACATCCATAAAAGC	Mutagenesis
pKmGFPEcoFor	GAATTCGTTGTGTCTCAAAATCTCTG	gfp amplification
pKmGFPEcoRev	GAATTCGGATATAGTTCCTCCTTTCAGC	gfp amplification

^aUnderlined regions indicate ATG to TAG null mutations.

To create X. poinarii and X. bovienii strains expressing green fluorescent protein (GFP), miniTn7 derivatives were engineered to constitutively express GFP by amplifying the P_aphA-GFP fragment from plasmid pECM9, using primers pKmGFPEcoFor and pKmGFPEcoRev, and cloning the EcoRI-digested PCR product into the similarly cut plasmids pTn7/rrs and pTn7/SR1/rrs to yield pTn7/rrs/GFP and pTn7/SR1/rrs/GFP, respectively. These plasmids were integrated into the X. poinarii chromosome by homologous recombination or integrated into the attTn7 site of the X. bovienii chromosome by Tn7-mediated transposition, as described above. X. nematophila strains were engineered to express GFP by using chromosomal integration of plasmid pECM20 as previously described (23).

The complete genome sequences of X. nematophila and X. bovienii (Jollieti) are available at http://www.xenorhabdus.org. The presence or absence of the nil locus in Xenorhabdus strains was determined by Southern hybridization, using pTn7/SR1 as a probe labeled with the ECF random prime labeling and amplification system (Amersham Pharmacia, Pittsburgh, PA) as previously described (17), and by BLASTn analysis of the X. bovienii (Jollieti) complete genome sequence.

Microscopy and nematode colonization assays.

All microscopic observations of nematodes and in vitro nematode colonization assays were performed as previously described (7, 23). Briefly, lawns of the bacterial strains to be tested were inoculated onto lawns of lipid agar (37), sterile S. carpocapsae eggs were added, and plates were incubated at room temperature. After approximately 2 weeks, infective juvenile stage S. carpocapsae nematodes were harvested from White's water traps (38), surface sterilized with 0.5% sodium hypochlorite for 3 min, and washed with sterile water. The number of surface-sterilized nematodes was then adjusted to 10,000 nematodes in 1 ml of water by comparison to known standards. One milliliter of LB broth was then added, and the surface-sterilized nematodes were sonicated for 1 min in a sonicating water bath to release the colonizing bacteria. The number of CFU/ml was calculated by plating dilutions of the sonicates onto LB plus 0.1% pyruvate. In vivo nematode colonization assays were performed by injecting Galleria mellonella greater wax moth larvae (five replicates of three larvae per replicate; Vanderhorst Wholesale) with 12.5 μl of a nematode and bacteria mixture per larva. The mixture contained ∼2,500 sterile first- and second-instar juvenile S. carpocapsae nematodes resuspended in an overnight culture of the test strain. The emergent nematodes were determined to be apoxenic or monoxenic by performing colonization assays as described above and by plating sonicates on LB agar plates containing 0.1% pyruvate and 50 μg/ml ampicillin. The bacterial colonies were verified to be Xenorhabdus spp. by testing for a lack of catalase activity and by examining colony pigmentation: X. nematophila colonies are cream-colored, X. bovienii colonies are yellow, and X. poinarii colonies are rust colored.

Fractionation and NilC immunoblotting.

Strains carrying the nil locus (HGB007, HGB1167, and HGB1174) were washed in phosphate-buffered saline and normalized to an optical density of 10.0. Twofold dilutions of the boiled lysates were separated by sodium dodecyl sulfate-12% polyacrylamide gel electrophoresis (10 μl of gel per lane), transferred to nitrocellulose, and immunodetected with a 1:5,000 dilution of anti-NilC antibody (7) followed by a 1:8,000 dilution of anti-rabbit- horseradish peroxidase-conjugated secondary antibody (Amersham, Pittsburgh, PA). Detection was performed with ECL luminescent substrate (Amersham, Pittsburgh, PA). Similarly, the NilC levels in strains carrying nilA and nilB start codon point mutations (HGB1182, HGB1183, HGB1168, HGB1169, HGB1175, and HGB1176) were compared to those of strains of the same species carrying the native nil locus, using twofold dilutions.

Strains carrying the nil locus (HGB007, HGB1167, and HGB1174) were washed in phosphate-buffered saline, normalized to an optical density of 10.0, and lysed by French pressure lysis. One milliliter of each cell-free lysate was fractionated into the soluble and insoluble fractions by ultracentrifugation. The insoluble fraction was resuspended in 1 ml of 1% sodium dodecyl sulfate to maintain its original ratio relative to the soluble fraction proteins in the lysate.

G-plus-C content sequence analysis.

The G-plus-C content of the circular X. nematophila genome sequence (http://www.xenorhabdus.org) was analyzed at every position with a sliding window of 3,463 nucleotides (nt), the size of the entire nil locus, or 1,783 nt, which corresponds to the size of the nilAB portion of the nil locus, using MATLAB version R2007a (The MathWorks, Natick, MA). Of 4,521,243 sequences, 216,994 had a G-plus-C content that was less than that of the nil locus and 57,504 sequences had a G-plus-C content that was less than that of nilAB. To avoid correlations due to overlapping, we also sampled 1,000 random probes of 1,783 or 3,463 nt from the genome and generated two data sets of 100 independent samplings each. The means and standard deviations of these two data sets were nearly identical to those of the X. nematophila genome sampled with a sliding window at every position.

RESULTS

The X. nematophila-S. carpocapsae interaction is species specific.

To explore the basis of host range specificity in a Xenorhabdus-Steinernema interaction, we focused on the nematode S. carpocapsae, as the relationship between it and its bacterial symbiont, X. nematophila, is well studied at the molecular level (13). Because nonbiological factors such as geographic isolation of hosts and/or microbes can give the appearance of microbe-host specificity where it does not exist, we sought to experimentally establish that this interaction is, in fact, species specific. We mixed S. carpocapsae nematodes with eight different Xenorhabdus species [X. nematophila, X. beddingii, X. bovienii (ATCC 49109), X. bovienii (Jollieti), X. budapestensis, X. ehlersii, X. innexi, X. poinarii, and X. szentirmaii], including two X. bovienii isolates from distinct nematode species (4, 11, 20, 33). Except for X. szentirmaii and X. bovienii (Jollieti), which did not support nematode development and reproduction, all species tested supported S. carpocapsae development into adults that produced progeny. Progeny nematodes derived from each strain were assessed for the presence of bacteria in the colonization site by sonication and dilution plating (see Materials and Methods; limit of detection = 0.0002 CFU/nematode). Only the native symbiont, X. nematophila, colonized S. carpocapsae nematodes; progeny nematodes that developed from lawns of the other species did not have any bacteria associated with them. Thus, the X. nematophila-S. carpocapsae interaction is species specific and is likely maintained in nature by the exclusive ability of X. nematophila to colonize S. carpocapsae nematodes.

X. nematophila nilABC colonization genes are absent from other Xenorhabdus species.

We previously identified a 3.5-kb region of the X. nematophila chromosome that encodes four genes: nilA, nilB, nilC, and tn2 (collectively referred to as the nil locus) (Fig. 1) (17). Deletion of the nil locus from the X. nematophila chromosome results in the inability of the bacterium to colonize S. carpocapsae nematodes but in no other known phenotypic changes (7). The nilA, nilB, and nilC genes are predicted or known to encode membrane proteins that may function in adherence (7, 17). Through Southern blot analysis using the nil locus as a probe, we found that of all the Xenorhabdus species tested, only X. nematophila harbors the nil locus (data not shown). This observation is supported by the lack of any sequences similar to the nil locus in the complete genome sequence of X. bovienii (Jollieti) (http://www.xenorhabdus.org).

FIG. 1.

Map of the X. nematophila nil locus. Each line arrow indicates the orientation and length of the gene named below it. The percentages of G-plus-C content of the portions indicated by brackets are shown.

The X. nematophila nil locus enables colonization of S. carpocapsae by Xenorhabdus species.

The critical role of the nil locus in X. nematophila colonization of S. carpocapsae nematodes (7, 8, 17), coupled with its absence in other Xenorhabdus species, suggests that it may encode proteins that function as S. carpocapsae host range specificity determinants. To test this, we introduced the nil locus onto the chromosomes of X. bovienii strain ATCC 49109, X. poinarii strain ATCC 49121, and X. szentirmaii DSM16338 (representing three distinct Xenorhabdus clades [35]) to create X. bovienii Tn7-nil, X. poinarii Tn7-nil, and X. szentirmaii Tn7-nil, respectively (see Materials and Methods). X. szentirmaii Tn7-nil did not support nematode development, indicating that the nil locus is not involved in this aspect of host range specificity (data not shown). However, X. bovienii Tn7-nil and X. poinarii Tn7-nil colonized S. carpocapsae nematodes, while the corresponding control strains lacking the nil locus did not (Fig. 2 and Table 4). Therefore, the nil locus is necessary and sufficient for Xenorhabdus bacterial colonization of the S. carpocapsae nematode, indicating that it contributes to defining the host specificity of X. nematophila.

FIG. 2.

The nilA, nilB, and nilC genes are required by X. nematophila (A), X. bovienii (B), and X. poinarii (C) for S. carpocapsae nematode colonization. For trials where no bacteria were detected, the colonization level is shown as the limit of detection, 0.005 CFU/nematode. Results are shown as means ± standard deviations (n = 4 or 5 replicates per trial). The bars represent the colonization levels of wild-type X. nematophila (w), X. nematophila with the nil region deleted (Δ), and Xenorhabdus spp. with an integrated Tn7 that is empty (v), carries the nil region with no mutations (nil), or has start-to-stop codon mutations in nilA (A⁻), nilB (B⁻), or nilC (C⁻). The different letters above the bars indicate values that are significantly different (P < 0.001) from one another by analysis of variance and Tukey's test with a 95% confidence interval.

TABLE 4.

S. carpocapsae nematode colonization by Xenorhabdus species

Species	Presence of nil locus	Mean CFU/nematodea	Proportion of nematodes colonizedb	Mean CFU/nematodec	Mean CFU/colonized nematoded
X. nematophila	−	<0.005	<0.001	<0.005
X. nematophila	+	65.5 ± 9.7	0.97	37.7 ± 4.4	38.8
X. bovienii	−	<0.005	<0.001	<0.005
X. bovienii	+	0.89 ± 0.2	0.026	0.6 ± 0.2	24.0
X. poinarii	−	<0.005	<0.001	<0.005
X. poinarii	+	1.2 ± 0.3	0.121	1.1 ± 0.3	9.1

^aMean CFU/nematode for strains lacking the nil locus (X. nematophila Tn7-nil deletion strain HGB777 or X. bovienii or X. poinarii strains with a chromosomal control vector) or carrying the nil locus (wild-type X. nematophila, X. bovienii Tn7-nil, or X. poinarii Tn7-nil). The data represent the means and standard deviations of five replicates.

^bColonization frequency was determined by epifluorescence microscopic examination of 1,000 nematodes from duplicate populations reared in cocultures with the indicated species carrying the nil locus and expressing GFP.

^cMean CFU/nematode of the indicated species carrying the nil locus and expressing GFP. The data represent the means and standard deviations of five replicates.

^dCalculated by dividing the mean CFU/nematode in column 4 by the proportion of nematodes colonized in column 3.

Although X. bovienii Tn7-nil and X. poinarii Tn7-nil colonize S. carpocapsae nematodes to levels that are far greater than those of their counterparts that do not have the nil gene (Fig. 2; Table 4), they do not colonize to the same level as X. nematophila. To determine if this was due to improper expression or localization of the nil gene products, we monitored the expression of NilC protein by immunoblotting. X. bovienii Tn7-nil cells express NilC at a level equal to that expressed by X. nematophila, whereas X. poinarii Tn7-nil cells express approximately twofold less than that expressed by X. nematophila (Fig. 3A). In both X. bovienii Tn7-nil and X. poinarii Tn7-nil, NilC was localized to the membrane fraction, as it is in X. nematophila (7) (Fig. 3B). NilC was not detected in X. bovienii or X. poinarii carrying a vector control, nor were its levels affected by point mutations in nilA or nilB (data not shown). Therefore, the difference in absolute colonization levels is not likely due to differences in the expression or localization of NilC. The colonization levels of different Xenorhabdus species vary considerably across different Xenorhabdus-Steinernema species pairs. For example, a recent study reported average colonization levels of 43.8, 1.0, and 0.7 CFU/nematode (determined by crushing 500 nematodes) for X. nematophila, X. cabanillasii, and X. japonica, respectively, in their cognate nematode hosts (32). Therefore, the colonization levels of X. bovienii Tn7-nil and X. poinarii Tn7-nil in S. carpocapsae may simply reflect an inherent limited colonization proficiency of X. bovienii and X. poinarii relative to that of X. nematophila. Alternatively, X. nematophila may possess additional unique genetic factors that contribute to its ability to reproduce within S. carpocapsae.

FIG. 3.

Expression and localization of NilC in three Xenorhabdus spp. containing the nil gene. (A) Twofold serial dilutions of stationary-phase X. nematophila (lanes 1 to 3), X. bovienii Tn7-nil (lanes 4 to 6), and X. poinarii Tn7-nil (lanes 7 to 9) were immunoblotted for NilC. The X. nematophila and X. bovienii Tn7-nil samples were diluted with one additional twofold dilution prior to loading. Triplicate samples showed results identical to those shown. (B) Stationary-phase X. nematophila (lanes 1 to 3), X. bovienii Tn7-nil (lanes 4 to 6), and X. poinarii Tn7-nil (lanes 7 to 9) were washed and lysed by French pressure lysis. The cell-free supernatants (lanes 1, 4, and 7) were separated into soluble (lanes 2, 5, and 8) and insoluble (lanes 3, 6, and 9) fractions and immunoblotted for NilC. As can be seen, NilC is found in the insoluble fractions in all Xenorhabdus spp. tested and is therefore likely to be properly membrane associated.

Individual nil locus genes play genetically similar roles in host colonization by X. nematophila and in cross-species colonization by noncognate Xenorhabdus spp.

To assess the individual contributions of the four nil locus genes to colonization and host range, we engineered four separate null mutation constructs, each with a null mutation in a nil locus gene. We previously found that X. nematophila nilB and nilC transposon mutants are completely defective and a nilA transposon mutant is partially defective in colonizing S. carpocapsae nematodes. The tn2 gene, which encodes a putative transposase, is not likely to encode a colonization factor, although its involvement in nematode colonization has not been directly examined (17). X. nematophila, X. bovienii Tn7-nil, and X. poinarii Tn7-nil carrying null mutations in either nilB or nilC were unable to colonize S. carpocapsae nematodes (Fig. 2), demonstrating that each of these two genes is essential for normal and cross-species colonization. A null mutation in nilA, however, had variable effects on colonization in different organisms. X. nematophila carrying a nilA-null mutation colonized to the same extent as a nilA transposon mutant (17), approximately 2.5% of that of the wild type. However, a nilA-null mutation did not significantly affect cross-species colonization by X. bovienii Tn7-nil and prevented cross-species colonization by X. poinarii Tn7-nil (Fig. 2), indicating that the importance of nilA in colonization and host range specificity is variable. As predicted, a null mutation in tn2 had no significant effect on nematode colonization by any of the three species (Fig. 2).

The interactions between noncognate Xenorhabdus bacteria that carry the nil locus and S. carpocapsae nematodes are biologically relevant.

In the experiments presented above, bacteria and nematodes were cocultivated on a synthetic growth medium rather than in their natural host, an insect cadaver. To test if the nil locus enables cross-species colonization within an insect cadaver, X. bovienii Tn7-nil, X. poinarii Tn7-nil, and control strains lacking the nil locus were coinjected with sterile S. carpocapsae nematodes into Galleria mellonella insect larvae. The emergent nematodes were colonized by X. bovienii and X. poinarii only in trials where the nil locus was present (Fig. 4), confirming that the ability to colonize S. carpocapsae is species specific in a natural setting and that the nil locus confers this specificity.

FIG. 4.

X. bovienii (X. bov.) and X. poinarii (X. poi.) colonize S. carpocapsae nematodes in vivo in the presence (black bars) of the nil locus, but not in its absence (gray bars). For trials in which no colonizing bacteria were detected, the colonization level is shown as the limit of detection, 0.0002 CFU/nematode. Results are shown as means ± standard deviations (n = 5 replicates per trial). The letters (a and b) above the bars indicate values that are significantly different (P < 0.001) from each other by analysis of variance and Tukey's test with a 95% confidence interval.

Xenorhabdus bacteria colonize a specific region, termed the vesicle, of their Steinernema nematode hosts (Fig. 5) (3). To determine if X. bovienii Tn7-nil and X. poinarii Tn7-nil colonize the S. carpocapsae vesicle, these strains were engineered to express GFP and examined by using epifluorescence microscopy. X. bovienii Tn7-nil and X. poinarii Tn7-nil bacteria, but not control strains lacking the nil locus, were observed exclusively within the S. carpocapsae vesicle (Fig. 5). This result indicates that cross-species-colonizing bacteria exhibit the same tissue tropism as X. nematophila, the native S. carpocapsae symbiont.

FIG. 5.

X. bovienii Tn7-nil and X. poinarii Tn7-nil colonize the vesicle of S. carpocapsae nematodes. Visible light (A, C, and E) and fluorescence (B, D, and F) micrographs of S. carpocapsae nematodes colonized by GFP-expressing X. nematophila (A and B), X. bovienii Tn7-nil (C and D), and X. poinarii Tn7-nil (E and F). All images were taken at a magnification of ×600, and a 10-μm scale bar is shown. Dashed lines indicate the borders of the vesicles. The images are oriented with each nematode's digestive tract below the vesicle (apparent as white autofluorescence).

Colonization levels of cross-species-colonizing bacteria and their distribution within host populations.

Typically, X. nematophila colonizes >90% of the S. carpocapsae nematodes in a population (31, 37); however, this is not the case for all Xenorhabdus-Steinernema pairs. For example, X. poinarii colonizes only 50% of the nematodes in populations of its native host, S. glaseri (1, 3, 9). We examined the proportion of colonized nematodes, using epifluorescence microscopy, and found that 2.6% and 12.1% of nematodes were colonized by X. bovienii Tn7-nil and X. poinarii Tn7-nil, respectively (Table 4). Therefore, it is possible that species-specific characteristics of Xenorhabdus bacteria determine their colonization initiation efficiency regardless of host. Alternatively, these results could indicate a requirement for additional colonization factors. Next, by calculating the average number of bacteria per colonized nematode, we determined that X. bovienii Tn7-nil and X. poinarii Tn7-nil colonize S. carpocapsae nematodes to levels that are approximately 62% and 24%, respectively, of those achieved by X. nematophila (Table 4).

The nil locus encodes proteins involved in initiating colonization and either bacterial outgrowth or colonization maintenance.

The X. nematophila colonization process is thought to have both initiation and outgrowth stages. Newly formed nematodes have few bacteria that grow to fill the colonization site, and the final population of bacteria in a mature, fully colonized nematode (Fig. 5) represents 1 to 2 individual clones (23). These findings have led to the working model of colonization in which very few individual cells initiate colonization (although the process of initiation is not understood) and then divide to fill the colonization site. Our data demonstrate that the nil locus enables colonization of S. carpocapsae nematodes by Xenorhabdus bacteria but do not reveal if NilA, NilB, and NilC function in the initiation and/or outgrowth stages of nematode colonization, either or both of which could contribute to the specificity of the natural interaction.

To determine if the nil locus gene products function in colonization initiation and/or outgrowth within nematodes hosts, we examined the distribution of S. carpocapsae nematodes colonized by an X. nematophila nilA mutant expressing GFP. nilA mutants are attenuated but not completely deficient in colonization and therefore allow these aspects of colonization to be determined. Of 500 nematodes examined (24), 65% were empty and 35% were visibly colonized, in contrast to the distribution for wild-type X. nematophila expressing GFP, where 3.8% of nematodes were observed to be empty and 96% were visibly colonized. The finding that a nilA mutant colonizes a lower percentage of nematodes than the wild type indicates that nilA functions in colonization initiation or in the very early stages of colonization, with uncolonized nematodes representing those in which initiation failed. Compared to the gross colonization levels for nil-null mutants (Fig. 2 and above), this result also suggests that the nilA mutant is defective in outgrowth, since the proportion of nematodes colonized by the nilA mutant is over 10-fold greater (35% colonized) than that which would be expected from its overall colonization levels (∼2.5% of wild-type levels). In other words, if the nilA mutant were defective only in initiation, one would expect that only 2.4% of all nematodes would be colonized. Instead, each of the 35% nilA-colonized nematodes must contain fewer bacteria (due to defective outgrowth) than wild-type-colonized nematodes to explain the relatively low average of nilA CFU/nematode.

The nil locus was likely acquired through a horizontal gene transfer event.

The absence of nil genes in other Xenorhabdus species and the presence of sequences similar to nilB in other bacterial genera (17) prompted us to investigate whether some or all of the nil locus may have been horizontally acquired. In support of this idea, both the entire nil locus (35.7% G+C) and the nilAB portion of the nil locus (30.7% G+C) are of lower G-plus-C content than average, compared to that of the entire X. nematophila genome (43.3% G+C) (Fig. 1). To gauge the likelihood that the entire nil locus or the nilAB portion of the nil locus was horizontally acquired, we calculated the G-plus-C content of all positions within the X. nematophila genome, using a sliding window that was identical in size to that of the entire nil locus or the nilAB portion of the nil locus. Indeed, 95.2% of all similarly sized fragments had a higher G-plus-C content than that of the nil locus, and 98.7% of all similarly sized fragments had a higher G-plus-C content than that of the nilAB portion of the nil locus.

To assess if the nil genes are part of a larger genetic island, we used the MaGe (Magnifying Genomes microbial genome annotation system; https://www.genoscope.cns.fr/agc/mage/) genomic island search tool to identify regions of the X. nematophila genome greater than 5 kb that contain open reading frames (ORFs) lacking reciprocal best hits in comparison genomes: Xenorhabdus bovienii (Jollieti), Photorhabdus luminescens TT01, Yersinia pestis CO92, Yersinia pseudotuberculosis IP32953, Salmonella enterica LT2, and Escherichia coli K-12. This analysis identified a 20-kb region that includes the nil genes, three transposase-like elements, and five other ORFs. None of these five ORFs have homologs in either Pasteurella multocida (str Pm70) or Haemophilus influenzae (86-028NP), two bacteria with NilB homologs. The region is flanked by conserved genes predicted to encode succinylornithine transaminase (left side; GenBank accession no. AAL79613) and phosphotransacetylase (right side) but not by insertion sequences or tRNA elements (12). Based on the atypical GC content, the presence of mobile elements, and the presence of genes involved in symbiosis, the nil region can be considered part of the flexible gene pool and a genomic island (12).

DISCUSSION

We have expanded the host ranges of animal-associated bacteria by supplying a single genetic locus. Although the nil locus is not present in any other Xenorhabdus spp. examined, sequences similar to nilB occur in several pathogens with distinct host range specificities, including Moraxella catarrhalis, Actinobacillus pleuropneumoniae, Haemophilus spp., and Neisseria spp. (17). Based on our results, which demonstrate that nilB is a species specificity determinant in a mutualistic relationship between an animal and a bacterium, it will be of interest to determine if the genes of these pathogens that are similar to nilB also contribute to the host range in a homologous manner. If so, it would lend further support to the idea that common molecules mediate interactions between hosts and microbes, regardless of whether the outcome is pathogenesis or mutualism (19).

Our data demonstrate that a relatively small genetic locus can significantly alter the ability of specific mutualistic partners to recognize each other, even to the exclusion of other strains or species. Similarly, species specificity in the relationship between Rhizobiaceae bacteria and Phaseolae legumes can be manipulated through minor genetic alterations (15). However, the specificity of the Rhizobia is dictated through minor variations in genes that are uniformly present among members of the genus. In contrast, in the Xenorhabdus genus, nil genes appear to be restricted to X. nematophila, indicating either that this locus was horizontally acquired or that other species of Xenorhabdus have lost nil gene homologs. Given the comparatively low G-plus-C content of the nil locus and nilAB portion of the nil locus relative to that of the X. nematophila genome, we believe that the former hypothesis is more likely correct. Although the tn2 gene is not required for nematode colonization (Fig. 2), its presence within the nil locus suggests that a transposition event may have mediated the horizontal gene transfer event that introduced the nilAB genes, or perhaps the entire nil locus, into the X. nematophila genome from an unrelated organism.

A similar horizontal transfer of a single gene may have contributed to the evolution of Yersinia pestis. Unlike its close relative Yersinia pseudotuberculosis, Yersinia pestis is able to colonize the flea midgut, enabling it to be transmitted to humans through the bite of a flea. Two genetic elements are known to be involved in Y. pestis-flea interactions: hms, necessary for production of an extracellular matrix, and ymt, which encodes a murine toxin with phospholipase D activity that protects the bacteria in the midgut of the flea from blood meal-derived antimicrobial activity. Although Y. pseudotuberculosis has functional hms genes, it appears that Y. pestis acquired ymt through a horizontal gene transfer. Thus, acquisition of a small genetic element may have had a profound impact on the environmental niche that can be occupied by this bacterium (18). Therefore, it is prudent to consider the impact that a similar transfer of host range specificity determinants could have on global health (6, 27).

In the mutualism between X. nematophila and S. carpocapsae, a successful association appears to require contributions from both the host and the microbe. We have shown here that X. nematophila requires specific factors to colonize its nematode host, while the nematode may control levels of colonizing bacteria (23). Similarly, the initiation and maintenance of Euprymna scolopes-Vibrio fischeri mutualism is characterized by bacterial factors necessary for initiation (40) and host mechanisms to control bacterial population size (26). This suggests that animal-microbe mutualisms in general may be characterized by host-imposed restrictions that select for specific bacteria and host-imposed controls that prevent the unchecked growth of the colonizing mutualist.