Friend and foe: the two faces of Xenorhabdus nematophila

Key Points

• The Gram-negative bacterium Xenorhabdus nematophila is a mutualist of a specific nematode species and a pathogen of insects, and is used as a model to study microorganism–host interactions and the similarities and differences underlying these symbioses.

• The initial encounter between X. nematophila and its hosts is hypothesized to require nutrient and surface sensing by bacterial regulators, in addition to synthesis of bacterial effector proteins that can be delivered to host cells by outer membrane vesicles.

• X. nematophila can survive and cause disease within insects by suppressing both cellular and humoral insect immunity. X. nematophila can suppress transcription of insect genes encoding antimicrobial peptides. X. nematophila also kills insect blood cells (haemocytes).

• X. nematophila has an extensive array of virulence factors, including toxins, haemolysins, proteases, lipases and fimbriae.

• In common with other mutualistic relationships between microorganisms and invertebrate hosts, the mutually beneficial relationship between X. nematophila and Steinernema carpocapsae nematodes involves nutrient exchange. Analysis of metabolic mutants suggests that the nematode might provide necessary nutrients to X. nematophila during growth within the nematode.

• Transcriptional regulators of X. nematophila that contribute to virulence and mutualism have been described. These might function to sense the external environment and alter transcription of various surface and secreted proteins in response to these environmental stimuli.

Abstract

Comparisons of mutualistic and pathogenic relationships are necessary to decipher the common language of microorganism–host interactions, as well as the subtle differences in dialect that distinguish types of symbiosis. One avenue towards making such comparisons is to study a single organism that speaks both dialects, such as the γ−proteobacterium Xenorhabdus nematophila. X. nematophila inhabits and influences the lives of two host animals, helping one to reproduce optimally while killing the other.

Main

Microorganisms can directly support or compromise the development, immunity and nutrition of plants and animals, resulting in beneficial (mutualistic) or detrimental (pathogenic) outcomes for the host^1,2,3,4,5. Pathogenic associations between microorganisms and hosts have been intensively studied and the molecular mechanisms that underlie some of these associations are known in detail. In recent years, there has been a significant effort to understand molecular aspects of beneficial microorganism–host interactions. This research indicates that microbial mutualism and pathogenesis share common molecular features^6,7,8. The challenge now is to determine how these relationships differ and what tips the balance towards either outcome. Dissecting these relationships will have an impact on our understanding of host range, evolution of virulence and mutualism, disease reservoirs and vectors of disease. For example, such studies should provide insights into host–symbiont specificity, microbial adaptations to new hosts and the selective pressures that influence the development of a pathogenic or mutualistic relationship.

All symbioses, regardless of host outcome, typically progress through several stages — initiation (recognition and entry), microbial growth (survival and multiplication) and maintenance (persistence and transmission) — during which host and microorganism recognize, communicate with and manipulate each other. Pathogenesis and mutualism are, at a gross level, very different from the perspective of the host. However, it is becoming increasingly evident that the molecular and cellular events underlying these types of interactions are similar. For example, derivatives of the ubiquitous microbial compound peptidoglycan induce disease symptoms during pathogenesis but also induce the squid, Euprymna scolopes, to produce a mucus that is necessary for colonization by its bacterial symbiont Vibrio fischeri ⁷ (Table 1). Insights into the similarities and differences among mutualistic and pathogenic relationships are likely to result from using systems that allow the direct study of both interactions. One model mutualist and pathogen, Xenorhabdus nematophila , is emerging as an invaluable tool for elucidating the molecular basis of microorganism–host interactions, owing to the ease with which its pathogenic and mutualistic interactions can be separated and independently studied in the laboratory. In this Review, recent advances in the study of X. nematophila symbioses are discussed.

Table 1 Microorganism–animal associations with similarity to the Xenorhabdus nematophila–Steinernema carpocapsae association

Friend and foe: the X. nematophila –host model

X. nematophila colonizes the entomopathogenic nematode Steinernema carpocapsae^9,10 in a mutualistic association, and is also a pathogen of insects¹¹. The mutualistic relationship between X. nematophila and S. carpocapsae is not obligate, as both partners can survive in the absence of the other; however, X. nematophila is required for S. carpocapsae nematodes to reproduce efficiently during their lifecycle^1,10,12. S. carpocapsae nematodes are either found in insect hosts or in the soil. The soil-dwelling vector stage, called the infective juvenile (IJ) (Fig. 1), is encased in a double cuticle, and is non-feeding owing to its closed mouth and anus¹³. Prior to the IJ stage, ingested X. nematophila bacteria colonize S. carpocapsae at a discrete intestinal location known as the vesicle. The IJ nematode then serves as a vector, carrying X. nematophila into a susceptible insect (Fig. 2), in which it is released from its nematode vector (Fig. 1) and rapidly kills the insect^14,15,16. X. nematophila is capable of killing insects in the absence of S. carpocapsae by direct injection of X. nematophila cells in the laboratory. Although all three organisms are capable of independent survival in the laboratory, it is unclear whether X. nematophila is capable of long-term survival in any reservoir outside its animal hosts in nature¹⁷ and, therefore, may rely on its nematode vector for transmission. The insect carcass provides nutrients for the propagation of both nematode and bacterium. In response to a signal, possibly nutrient deprivation or space limitation¹⁸, X. nematophila re-associates with the nematode, and the pair leave in search of a new insect host to repeat the cycle (Fig. 2).

Figure 1: Xenorhabdus nematophila in the infective juvenile stage of Steinernema carpocapsae.

A differential interference contrast image of a S. carpocapsae infective juvenile nematode is shown at the top. The schematic diagram indicates the approximate positions of relevant structures, including the vesicle that contains X. nematophila bacteria. Owing to the transparency of the S. carpocapsae nematode, X. nematophila strains expressing green fluorescent protein can be visualized using fluorescence microscopy. Stages are shown schematically at the bottom, together with image panels (a). Following exposure to insect haemolymph, X. nematophila cells are released from the vesicle by passage through the intestine (b, c). Occasionally, infective juveniles do not lose their outer cuticle and released X. nematophila cells accumulate in the inter-cuticular space (d). Lower panels reproduced with permission from Ref.16 © (2004), BRILL.

Figure 2: The Xenorhabdus nematophila lifecycle.

The infective juvenile (IJ) nematode containing X. nematophila bacteria (nematode–bacteria complex) enters a susceptible insect host through natural openings that include the mouth, anus and spiracles. After entering the insect blood system, the nematode releases X. nematophila and develops into a fourth stage juvenile. Together, the nematode and bacteria overcome insect immunity and kill the insect. The insect cadaver is used as a nutrient source and is protected from opportunistic infection and scavenging by metabolites produced by X. nematophila. Within this environment, Steinernema carpocapsae nematodes reproduce sexually and progeny develop through four juvenile (J) stages. Some nematodes develop into infective juveniles after being recolonized by X. nematophila. The pair then exit the depleted insect carcass in search of a new host.

The X. nematophila–host model system is simple (relative to associations involving microbial consortia) and tractable. Genetic, biochemical and molecular techniques can be used to study X. nematophila¹⁶, and each animal that X. nematophila associates with is amenable to multiple experimental techniques. Furthermore X. nematophila maintains a species-specific interaction with its cognate nematode host, S. carpocapsae^9,19, yet can kill an extensive range of insect larvae hosts¹¹, making it a useful model for understanding the molecular basis of host range. Additionally, X. nematophila is a vectored pathogen and so can provide insights into the transition from one host environment to another.

X. nematophila commonly infects insects of the order Lepidoptera, many of which are significant agricultural pests¹¹. Insects in this order are cheap and easily reared and manipulated in the laboratory, and their size facilitates bacterial injections and extraction of immune cells (haemocytes) from their blood (haemolymph). Insect immunity closely parallels vertebrate innate immunity, and involves conserved signal transduction cascades that link pathogen recognition to the subsequent induction of cellular and humoral immune responses²⁰ (Box 1). This conservation of mechanisms, coupled with the experimental tractability of insects, brings insects into the spotlight as models to understand how the human innate immune system responds to pathogens²¹. Insect–microorganism relationships are of further relevance, given the recent recognition that insects can function as reservoirs of emerging human pathogens, such as Bacillus anthracis and Yersinia pestis (Table 1), which might have evolved from insect-associated ancestors^21,22. S. carpocapsae nematodes are easily propagated; several hundred thousand IJs can be generated from the infection of a single insect host or from lawns of X. nematophila and can be stored in water or buffer for weeks^15,23,24,25. The well-studied nematode Caenorhabditis elegans facilitates comparative molecular analyses²⁶ and provides a foundation for the application of new methodologies for the study of S. carpocapsae gene function. Unlike many animals associated with bacterial mutualists (Table 1), S. carpocapsae nematodes are viable in the absence of X. nematophila. Therefore, axenic nematodes can be obtained and assessed for responses to diverse bacterial and environmental stimuli^10,12,27.

First impressions: initiating interactions

A key element of symbiosis initiation is the ability of organisms to detect and identify associates and assess their potential threat or benefit. The initial encounter between X. nematophila and its insect host occurs in the insect blood system, which it gains access to through its nematode vector (Fig. 2) or by experimental injection. During natural infection, the colonized IJ nematode (Fig. 1) is exposed to the insect's gut contents, triggering a loss of the outer cuticle, which opens the mouth and anus¹⁵. Subsequently, the nematode migrates into the insect blood system and begins ingesting haemolymph. This causes release of X. nematophila by defecation, a process associated with the expansion of the intestinal lumen and migration of individual bacterial cells down the intestine and through the anus^1,15,16 (Fig. 1). Nematode exposure to insect gut contents alone is not sufficient to cause release, suggesting the release-triggering compound(s) is a component of haemolymph¹⁵. Similarly, the entomopathogenic bacterium Photorhabdus luminescens is released from its nematode host in response to a low molecular weight, heat- and protease-stable haemolymph compound of unknown identity²⁸.

Nutrient sensing. X. nematophila must detect its transition from the nematode into its insect host. Temperature change, which is sensed by vectored mammalian pathogens such as Y. pestis and Borrelia burgdorferi (reviewed in Ref. 29; Table 1), is unlikely to be a vector-to-host transition signal for X. nematophila, as both its hosts are ectotherms. Instead, the transition signal probably relates to differences between components in haemolymph and the nematode vesicle lumen. Haemolymph supports vigorous growth of X. nematophila (0.41 doublings per hour in haemolymph versus 0.62 doublings per hour in Luria-Bertani medium; E. E. H. and H. G.-B., unpublished data)³⁰ even during release from the nematode¹⁶. By contrast, the maximum X. nematophila growth rate observed in the nematode vesicle was 0.1 doublings per hour²⁴, indicating that this environment is comparatively nutrient limiting. Therefore, nutrient upshift might signal entry into the insect environment. X. nematophila opp mutants, which are defective in oligopeptide transport, show delayed growth in haemolymph, indicating that oligopeptides might be used by X. nematophila as a nutrient signal³⁰ (Fig. 3).

Figure 3: Xenorhabdus nematophila putative effectors of host interactions.

X. nematophila produces a number of secreted and surface-localized protein effectors that have been implicated in mediating host interactions. A change in nutrient status, including the presence of oligopeptides, might signal bacterial entry into the insect host, inducing the production of virulence effectors. Bacterial outer membrane vesicles that bleb off from the cell surface are toxic to haemocytes and contain the cytotoxic fimbrial subunit (blue) and lipopolysaccharide (LPS, green). LPS and secreted toxins and haemolysins contribute to virulence towards insect larvae through haemocyte destruction and lipases that are secreted through the flagellar apparatus could provide nutrients to bacteria and nematodes through bioconversion of the insect carcass. Mechanisms of secretion of haemolysins and toxin complex are unknown. X. nematophila OpaB is predicted to be a cell-surface adhesin, with similarity to the Yersinia virulence adhesin Ail. Although the role of X. nematophila OpaB in host interactions is undetermined, opaB expression is positively correlated with virulence toward insects. NilA, B and C are membrane proteins that are necessary for nematode colonization and might function in signalling or adhesion (NilB). PixA and IP2 are intracellular inclusion proteins that are not required for nematode colonization, but are hypothesized to provide nutritional benefit to the nematode host.

Further support for the nutrient-sensing hypothesis comes from studies of the transcription factor Lrp (leucine responsive regulatory protein), which is necessary for virulence towards Manduca sexta insects and mutualistic nematode colonization^31,32. Lrp proteins are ubiquitous among bacteria and archaea and are members of the feast or famine regulatory protein family³³ that regulate adaptation between nutrition states. X. nematophila lrp mutants isolated from nematode vesicles or defined medium grow more slowly on rich solid medium relative to cells pre-grown in haemolymph or rich, liquid medium³¹, which might indicate that Lrp is necessary for adaptation from nutrient limiting to nutrient rich conditions. Taken together, these findings support a model in which X. nematophila experiences and senses — through Lrp — nutrient upshift during bacterial release into haemolymph and presumably nutrient limitation after re-association with the nematode (based on the nematode colonization defect of lrp mutants). X. nematophila Lrp is a global regulator and regulates numerous known or putative virulence and nematode colonization factors (presumably in response to nutrient signals)^31,34,35,36. Examination of changes in the Lrp-dependent regulon in response to changing nutrient conditions should shed light on specific genes contributing to adaptation in each host environment.

Although X. nematophila Lrp is the first protein in the Lrp family with a demonstrated role in virulence, Lrp homologues have been linked to the expression of virulence determinants in several pathogens, including Salmonella enterica , Proteus mirabilis and pathogenic Escherichia coli (reviewed in Ref. 37). In many organisms, Lrp regulates one (or a few) genes, and X. nematophila Lrp is only the second demonstrated example, after E. coli Lrp, of an Lrp that participates in global regulation³¹. Global regulatory activity of Lrp homologues might be restricted to those Lrps that are encoded by enteric species, including Salmonella and Yersinia species, which encounter diverse environments. Similar to X. nematophila, Y. pestis adapts to transitions between a vector (the flea) and the mammalian host that it infects (Table 1), and Lrp homologues, of which Y. pestis has four³⁷, might have a crucial role in this adaptation and the expression of virulence determinants.

Surface sensing. Another feature of the host environment that is probably sensed by X. nematophila early during infection is insect cell surfaces, particularly those of blood cells (haemocytes) or connective tissues¹⁵. Interactions with such surfaces are typically mediated by adhesins, surface proteins and structures with binding capacity. One candidate X. nematophila adhesin is OpaB, an Ail (attachment and invasion locus)-family outer membrane protein³⁸ (Fig. 3) the expression of which is positively correlated with virulence among X. nematophila variants³⁸. As with Yersinia enterocolitica Ail, X. nematophila OpaB might mediate attachment to host surfaces. Another candidate mediator of host-cell interactions is the type I fimbriae (surface appendage)³⁹ (Fig. 3), encoded by the mrx (mannose-resistant Xenorhabdus) fimbrial operon³⁴. The specific adhesive qualities of fimbriae are attributed to an adhesin tip protein homologue⁴⁰ which, in X. nematophila, is encoded by mrxH³⁴. Current evidence suggests that the X. nematophila fimbrial structure and adhesin are expressed and therefore function during mutualism with the nematode⁴¹, whereas the X. nematophila pilin subunit, encoded by mrxA, interacts with the insect independently of the fimbrial structure. The expression of mrxA is Lrp-dependent³⁴, and purified MrxA agglutinates mammalian blood cells³⁹ and surprisingly has pore-forming cytotoxic activity against insect haemocytes^42,43,44. Furthermore, MrxA is present in the toxic outer membrane vesicles produced by X. nematophila⁴⁵ (Fig. 3). The studies reviewed above suggest a model in which X. nematophila detects the transition into an insect environment, partly through changes in nutritional content, and interacts with insect host cells through the production of outer membrane vesicles that deliver cytotoxic fimbrial subunits to haemocytes.

Host-cell-surface cues are also likely to have an important role in X. nematophila– nematode mutualism. The X. nematophila population within an IJ nematode is founded by between one and two individual bacterial cells²⁴ that grow to fill the vesicle, which is the lumen between two nematode epithelial cells at the anterior end of the intestine⁴⁶ (Figs 1,4). Similar to specificity in the E. scolopes–V. fischeri mutualism, only X. nematophila can colonize the S. carpocapsae vesicle; other Xenorhabdus species do not colonize S. carpocapsae IJs^9,19. The anterior vesicle is connected to the oesophagus through a forward connection, and its posterior section abuts the intestine, but remains closed until the release process triggers opening (S. Forst and S. P. Stock, personal communication). The specificity of the X. nematophila colonization might be a result of its exclusive ability to enter the forward connection, adhere to a structure in this region, survive host-imposed stress, or some combination of these processes. Experiments comparing the locations and survival of X. nematophila and non-colonizing strains or species during early stages of colonization should help distinguish among these possibilities.

Figure 4: Xenorhabdus nematophila colonization of Steinernema carpocapsae nematodes.

a | A developing S. carpocapsae nematode ingests X. nematophila bacteria (expressing green fluorescent bacteria). Scale bar represents 20 μm. In response to high population densities or low nutrient availability, S. carpocapsae nematodes develop into the infective juvenile stage, a process that includes initiation of colonization by X. nematophila. b | Newly formed, colonized, non-feeding infective juveniles contain few bacteria (rods) in the vesicle, a state termed oligo-colonization. These bacteria are often seen attached to the intravesicular structure (IVS) (grey circles) which is thought to be associated with glycan-containing material (not visible in micrographs). c | Nematodes that develop into infective juveniles in the absence of colonization-competent bacteria still form a vesicle and an IVS. d | As colonized infective juvenile nematodes mature, the bacteria grow to fill the vesicle, and this population derives from 1–2 founding bacterial cells. e | Some metabolically impaired X. nematophila mutants fail to grow, and instead become spheroplasts that eventually disappear from the population. In b–e, the left panel shows a bright field micrograph of extruded nematode intestines (samples in b–d were stained with crystal violet).

Adherence is suggested by the fact that in colonized nematodes, X. nematophila cells are attached to the intravesicular structure (IVS), an untethered cluster of spheres present in the vesicle lumen of both colonized and un-colonized IJs⁴⁶ (Fig. 4). Similar to X. nematophila cells, the IVS could be released from nematodes after exposure to haemolymph, and fluorescent staining revealed that the individual spheres are surrounded by a mucus-like substance that contains glycans based on its lectin reactivity⁴⁶. Although it is possible the released IVS accumulated this substance while descending through the intestine, the favoured hypothesis is that the mucus is present within the vesicle (Fig. 4). Indeed, electron micrographs reveal that within the vesicle X. nematophila is embedded in an amorphous matrix of host origin (as it is present in un-colonized nematodes) (S. P. Stock, personal communication)⁴⁷. Owing to their structural diversity, glycans play an important part in mediating the specificity of both pathogenic and mutualistic host–microorganism interactions⁴⁸. For example host-derived mucus helps mediate specificity in the mutualism between E. scolopes and V. fischeri⁴⁹. Furthermore, germ-free zebrafish express glycans at the oesophageal–intestinal junction, the portal for colonizing intestinal microorganisms². In both squid and zebrafish, the introduction of specific microorganisms causes downregulation of the mucus^2,49. In C. elegans, the glycan composition of the outer cuticle and intestinal epithelia can affect pathogen adherence and resistance to the Bacillus thuringiensis pore-forming toxin^50,51. Therefore, in addition to their potential role in promoting beneficial microbial adherence, intestinal surface glycans in S. carpocapsae might also mediate resistance to the pathogenic properties of X. nematophila, such as pore-forming toxins.

The data described above indicate that nematode-derived glycans are present in, or possibly around, the vesicle and therefore might have a role in the colonization process. If so, X. nematophila lectins probably bind to such glycans. One potential lectin is the type I fimbrial adhesin homologue encoded by mrxH but so far the colonization phenotype of an mrxH mutant has not been reported. Another possible lectin is NilB (nematode intestine localization B), an outer membrane protein (A. Bhasin and H. G.-B., unpublished data) essential for mutualistic nematode colonization^32,52 (Fig. 3). Homologues of nilB are present in several pathogens that colonize mucosa, including Neisseria meningitidis and Haemophilus influenzae ³², and NilB might serve to facilitate glycan interactions in the mucus environment that these pathogens colonize.

Developing relations: post-initiation events

Manipulating immunity. In both pathogenesis and mutualism, once the identity and intentions of the host and microorganism are established, each partner initiates strategies to manipulate the other. In the pathogenic X. nematophila–insect relationship, the insect host uses cellular and humoral immunity (Box 1) to clear X. nematophila from its blood system, but this defensive response is countered by X. nematophila-mediated suppression of immunity. Although in vitro studies indicate that haemocytes are capable of phagocytosing Xenorhabdus cells⁵³, evidence from in situ experiments indicates that, by the late stages of infection, only ∼10% of X. nematophila cells are degraded in haemocytes and the remainder are extracellular¹⁵. Extracellular X. nematophila cells are not attached to haemocytes¹⁵ or trapped in nodules^15,54. In fact, insects infected with X. nematophila have fewer blood-cell aggregates and nodules than insects infected with E. coli⁵⁴.

Several studies have revealed the mechanisms underlying the ability of X. nematophila to avoid elimination by insect cellular immunity. X. nematophila kills 50% of haemocytes by 3 hours post-injection⁵⁵, probably through the toxic effects of lipopolysaccharide (LPS)⁵⁶, cytolysins^{36,57,58,59,60}, toxins⁶¹ and the pore-forming fimbrial subunit^42,43,44 (Fig. 3). This amount of insect cell death might be sufficient to prevent the entrapment of X. nematophila by insect cells, although it is also possible that X. nematophila directly inhibits phagocytosis. X. nematophila directly inhibits nodulation and aggregation using a secreted compound that suppresses phospholipase A₂, which is the host enzyme that is responsible for activation of aggregation and nodulation through the eicosanoid pathway^54,62,63 (Box 1).

Insects that are infected with X. nematophila also have impaired humoral immunity (Box 1). X. nematophila cells, or purified X. nematophila LPS injected into insect haemolymph, suppress phenoloxidase activation^64,65. Additionally, the expression of insect antimicrobial peptides, which is induced after injection with S. enterica or heat-killed X. nematophila, is not induced after injection with live X. nematophila^38,66. In fact, when S. enterica and X. nematophila are injected together, the expression of antimicrobial peptide is low, suggesting X. nematophila actively suppresses their induction³⁸. The mechanism(s) of suppression has not been established yet, but might include inhibition of the cellular response noted above, as in Drosophila melanogaster , phagocytosis and degradation of microorganisms by haemocytes is necessary for induction of at least one type of antimicrobial peptide⁶⁷.

The data discussed above show that X. nematophila independently suppresses immunity when injected into insect haemolymph. However, its mutualistic nematode vector host can also contribute to suppression of immunity. The surface cuticular lipids of Steinernema feltiae nematodes (colonized by Xenorhabdus bovienii bacterial symbionts) deactivate haemolymph proteins that are necessary for the induction of antimicrobial peptides⁶⁸. B. burgdorferi, a spirochaete that is transmitted by ticks, causes a chronic inflammatory disease in humans. However, initial acute infection is not associated with inflammation and can be asymptomatic, suggesting initial immune responses are inhibited⁶⁹ (Table 1). As with X. nematophila, Borrelia spp. are independently capable of resisting immunity, but their vector (tick) also contributes protective functions: tick saliva and salivary gland extract inhibit macrophage killing of spirochaetes and downregulate inflammatory signals⁷⁰. Thus, in addition to providing a means of transmission, pathogen vectors also contribute to the establishment of infection.

Toxins. Once X. nematophila controls insect immunity it causes insect death within 48 hours post-infection. X. nematophila secretes multiple products that have long been presumed to have a role in virulence or decomposition of the insect cadaver¹⁴ (Fig. 3). In the past 5 years there has been much new information regarding the genes that encode these factors, allowing a direct examination of their role in virulence (Table 2). Among the most effective weapons in the X. nematophila arsenal are secreted toxins, including the 'toxin complex' (Tc/Xpt)^61,71,72,73 and the C1 cytotoxin (XaxAB)⁵⁷. In common with the toxins associated with cholera, tetanus and diphtheria, which can elicit many of the symptoms of disease in their purified form⁷⁴, X. nematophila toxins are lethal to insects when expressed in E. coli and administered orally^{71,72,75,76,77}. Tc toxins and the C1 cytotoxin are found in Xenorhabdus species and a diverse range of bacteria, including Yersinia spp. and Pseudomonas spp.^57,61 The role(s) of these toxins in the lifecycle of these bacteria remains unclear. An interesting observation is that the CO92 strain of Y. pestis, the flea-vectored cause of bubonic plague⁷⁸, has a mutated tc locus, which led to the hypothesis that inactivation of the insecticidal toxin was necessary for Y. pestis to colonize the flea midgut⁷⁹. Contrary to this idea, the other sequenced strain of Y. pestis (KIM) has an intact tc locus⁶¹. Furthermore, this hypothesis provides no framework for understanding the role of the Tc toxins in Y. enterocolitica, which is not known to associate with insects. Mutations in Y. enterocolitic a strain T83 tc homologues cause a defect in intestinal infection in mice⁸⁰, suggesting that the role of Tc toxins in pathogen physiology extends beyond their originally assigned function as insecticidal toxins.

Table 2 Xenorhabdus nematophila genes and host interaction activities

Maintaining balance in mutualism

The X. nematophila–S. carpocapsae nematode mutualism can be characterized by each partner's struggle to control the other. X. nematophila nematode colonization can be separated into two stages: initiation (occurring before the nematode develops its closed mouth and anus) and outgrowth (occurring after the IJ ceases ingestion)²⁴ (Fig. 4). Quantitative measurements of bacterial populations during outgrowth show that, although overall bacterial numbers increase over time, growth is punctuated by periodic decreases in the average number of bacterial cells per nematode, which might indicate bacterial cell death²⁴. Furthermore, X. nematophila metabolic mutants that are capable of initiating colonization but are defective in outgrowth (see below) often have a round, spheroplast phenotype, and such cells disappear from the population over time, indicating that they are non-viable and are cleared⁸¹ (Fig. 4e). In addition, as IJ nematodes age, their bacterial symbiont population size decreases^23,82,83. Possible causes of observed X. nematophila cell death and reductions in population size in the vesicle include nutrient depletion and accumulation of toxic compounds. An additional possibility is that the nematode host controls the quality and size of its X. nematophila bacterial symbiont population through selective killing. E. scolopes squid select for and control the population size of symbiont V. fischeri bacteria through the imposition of oxidative stress and by daily expulsion of ∼90% of V. fischeri cells from the light organ (reviewed in Ref. 84; Table 1). The colonized IJ nematode is encased in an outer cuticle and therefore cannot 'vent' X. nematophila cells into the environment. However, it is feasible that the nematode could release antimicrobial molecules into the colonization site, and subsequently degrade and absorb any dead cells. Indeed, the immune response of the free-living bacteriovorous nematode C. elegans includes microorganism-inducible signal transduction cascades together with genes that are predicted to encode antimicrobial factors (reviewed in Ref. 85). One essential component of this model is that a subset of colonizing X. nematophila must survive antimicrobial activity during outgrowth, in order to maintain the mutualistic relationship.

Similarly, Y. pestis has adapted to survive in the midgut of its vector host, the flea (Table 1). In this environment other species of Yersinia become spheroplasts, presumably owing to damage from host–environment factors, and are eliminated. Survival in the flea midgut is conferred by Ymt (Yersinia murine toxin), a bacterial cytoplasmic protein released after bacterial cell lysis (but YMT is not required for Y. pestis to cause disease in mammals)⁸⁶. This protein is uniquely conserved in Y. pestis relative to other Yersinia species and it has been proposed that Ymt is necessary to protect against, or deactivate, antimicrobial activity in the flea midgut (reviewed in Ref. 87). Thus, in three disparate systems (X. nematophila, V. fischeri and Y. pestis) bacterial persistence in the vector or mutualist host seems to be characterized by specific survival of challenges encountered in the host environment.

Sharing a meal: nutrition and development

Periodic reductions in the X. nematophila population size might also help X. nematophila to survive because dead bacterial cells release nutrients. The presence of nutrients in the vesicle was inferred from the ability of X. nematophila to grow in this niche²⁴, and was further supported by studies assessing the ability of X. nematophila metabolic mutants to colonize the vesicle. Metabolic mutants that are defective in the biosynthesis of the vitamins para-aminobenzoate and pyridoxine, and the amino acids methionine and threonine, fail to grow in the vesicle (Fig. 4), indicating that these nutrients might be limiting in the vesicle^81,83. By contrast, X. nematophila mutants that were unable to synthesize serine, histidine, cysteine, isoleucine, valine and pantothenate were able to colonize as well as wild-type, suggesting that these nutrients are sufficiently plentiful in the vesicle to support growth⁸¹.

The nutrients that support X. nematophila growth in the vesicle might therefore be derived from dead sibling bacteria. However, several studies hint that the nematode host contributes (directly or indirectly) to the supply of nutrients to resident bacteria. Although X. nematophila is required for S. carpocapsae nematodes to efficiently complete their lifecycle^1,10,12, maintenance of this mutualism apparently comes at a cost incurred by the IJ vector stage of the nematode. This idea is based on the inverse correlation between bacterial load and IJ survival that has been suggested by several studies^10,12,27. Thus, the non-feeding IJ stage of S. carpocapsae nematodes might divert nutrient stores away from their own use to help maintain a viable bacterial symbiont population.

Nutrient exchange is a common theme among mutualistic symbioses, although in most relationships the bacterial symbiont provides nutrition for its host. For example, mutualistic Buchnera bacteria (Table 1) provide amino acids to aphids (reviewed in Ref. 88), and intestinal microorganisms harvest energy for their vertebrate hosts (reviewed in Ref. 89). Although the evidence implies host-to-microorganism nutrient flow in the X. nematophila–S. carpocapsae relationship, reciprocal nutrient exchange might also occur. X. nematophila produces two proteins (PixA and IP2) (Fig. 3) that form intracellular inclusions, which, unlike the B. thuringiensis crystal proteins⁹⁰, have no apparent insecticidal activity. It was proposed, therefore, that the inclusion proteins have a role in bacterial or nematode nutrition⁹¹. A mutation in the pixA gene did not reduce X. nematophila virulence towards insects, or affect colonization or survival in the IJ nematode vesicle. In fact, the pixA mutant had a competitive advantage over wild-type X. nematophila for colonization of the IJ nematode vesicle²³. This finding implies that X. nematophila expresses the methionine-rich PixA protein in the vesicle but derives no direct colonization benefit from it. One possibility, therefore, is that X. nematophila expresses PixA for the benefit of its nematode host. One test of this hypothesis would be to compare the survival and metabolic activities of IJ nematodes colonized by wild-type and pixA mutant X. nematophila.

Development and protection

The fitness cost experienced by the IJ vector stage of the nematode is balanced by the positive contributions of X. nematophila to nematode reproduction within the insect cadaver¹⁰. Early studies indicated that X. nematophila is necessary for the development of sexual organs in the nematode¹. The X. nematophila factors that contribute to development have not yet been identified, but such factors might be controlled by the Lrp transcription factor, as lrp mutants support the production of fewer nematode progeny than do wild-type X. nematophila³¹. Many parasitic filarial nematodes, the causative agents of human diseases such as elephantiasis and river blindness, contain symbiotic Wolbachia bacteria. Treatment of filarial nematodes with tetracycline causes death of the Wolbachia and concomitant sterility in adult female nematodes, as well as defects in embryogenesis and moulting (reviewed in Ref. 3). Therefore, like S. carpocapsae, filarial nematodes rely on their bacterial symbionts for normal development and reproduction. Continued comparison of these two systems could reveal underlying paradigms of the bacterial contributions to animal reproduction and development, as well as new targets for drug design to control parasitic nematodes.

In addition to contributing to nematode fecundity, X. nematophila helps to protect the insect cadaver from microbial infection^12,92 and predation by arthropods⁹³, as well as competition from other Steinernema nematodes and microorganisms that can reduce the fitness of S. carpocapsae^19,94,95. This protection is attributed to the fact that X. nematophila produces antibiotics that are effective against other Xenorhabdus species, Gram-positive and Gram-negative bacteria, and yeast (reviewed in Ref. 96). X. nematophila is therefore also part of an emerging paradigm of microbial symbionts in mutualistic associations producing secondary metabolites that help to protect their host from pathogens^97,98. The X. nematophila genome encodes numerous non-ribosomal peptide and polyketide synthase homologues, which are predicted to synthesize secondary metabolites (H. Bode and B. Goldman, personal communication). The identification and characterization of such metabolites is not only relevant to understanding X. nematophila biology, but is also a potential source for the discovery of novel therapeutic activities.

Regulation of host interactions

The success of X. nematophila in its mutualistic and pathogenic relationships relies on its ability to express relevant factors only in the appropriate host. To date, several regulators have been implicated in host interactions, including the global regulator Lrp. The absence of Lrp causes pleiotropic defects, complicating interpretations of its role in host interactions. However, Lrp controls specific factors involved in mutualism and pathogenesis³¹ (Fig. 5).

Figure 5: Xenorhabdus nematophila regulatory hierarchies control host interactions.

X. nematophila regulators that are necessary for mutualism with Steinernema carpocapsae nematodes (left circle) and pathogenesis in Manduca sexta insects (right circle) are shown. Arrows and blunt arrows indicate positive and negative influence (direct or indirect) on expression respectively, revealed by phenotype and expression analyses. Lrp (leucine responsive regulatory protein) is necessary for both mutualism and pathogenesis and therefore occupies both circles. Similarly, the Lrp-dependent gene mrxA (mannose resistant Xenorhabdus fimbriae A), encoding the pilin subunit, has proposed functions in both mutualism and pathogenesis. Lrp and NilR (nematode intestine localization R) synergistically repress nilA, B and C, which are each necessary for mutualism. Because Lrp is necessary for mutualistic colonization, it is likely that it positively regulates as-yet-unidentified colonization functions. Similarly, RpoS, which encodes the stress-responsive sigma factor, is necessary for mutualism, but the genes it regulates have not yet been identified. Lrp positively regulates the LysR homologue transcription factor LrhA, which is necessary for pathogenesis, but not mutualism. LrhA positively regulates the master flagellar regulator FlhDC. FlhDC is necessary for the expression of the flagellar sigma factor (FliA), which in turn activates genes encoding various exoenzymes (prtA, encoding a protease; xaxAB, encoding the Xenorhabdus C1 α–X haemolysin; xhlBA, encoding Xenorhabdus haemolysin and its secretion system; and xlpA, encoding Xenorhabdus lipase). FlhDC also regulates the expression of the flagellar secretion apparatus, through which XlpA lipase is secreted. The flagellar regulon is negatively regulated by the EnvZ/OmpR two-component regulatory system, which also negatively regulates the expression of antibiotic activity. Because the virulence defect of an lrhA mutant is more profound than those of flhDC mutants, it is inferred that LrhA also regulates other as-yet-unidentified virulence determinants.

Lrp is necessary for wild-type expression of lrhA (LysR homologue A), which encodes a member of the LysR-type transcriptional regulator family (Fig. 5). Of the regulatory mutants described so far (Table 2), an lrhA mutant has the most severe virulence defect, killing at most 10% of insects compared with the 80–100% killed by wild-type, indicating that LrhA has a crucial role in controlling virulence. LrhA most closely resembles the HexA (hyperexpression of exoenzymes A) subfamily of LysR regulators but differs from other characterized members of this family in that it activates, rather than represses, virulence determinants and exoenzymes (G. R. Richards and H. G.-B., unpublished data).

LrhA positively affects (directly or indirectly) the expression of flhD, which encodes the flagellar master regulator (Fig. 5). Motility per se is not necessary for virulence, as a fliC mutant (lacking a structural component of the flagellum, and non-motile) is fully virulent (G. R. Richards, E. E. H. and H. G.-B., unpublished data). However, the flagellar apparatus does have a measurable role in virulence⁹⁹ that could be attributed to the transcriptional control of virulence determinants, secretion of virulence determinants through the flagellar export apparatus, or a combination of these¹⁰⁰ (Fig. 3). In common with outer membrane vesicles, secretion through the flagellar apparatus might be an important mechanism of effector delivery in pathogens that lack a type III secretion system, including X. nematophila¹⁰¹ and Campylobacter jejuni , a food-borne pathogen that causes enteritis¹⁰².

The X. nematophila FlhDC regulon is repressed by the OmpR/EnvZ two-component system, which also represses antibiotic activity and the expression of a non-ribosomal peptide synthetase operon¹⁰⁰ (Fig. 5). The OmpR/EnvZ system consists of the membrane receptor EnvZ, which perceives and transduces environmental signals to affect the activity of the transcription factor OmpR. The signal perceived by X. nematophila EnvZ is currently unknown but it has been suggested that EnvZ/OmpR repression of flhDC is relieved after the insect is dead, when de-repression of motility would facilitate distribution of X. nematophila within the insect cadaver¹⁰⁰. This hypothesis suggests that exoenzymes controlled by the flagellar regulatory pathway, such as xlpA, prtA and xaxAB, which encode lipase, protease and C1 haemolysin, respectively^57,100, might function to convert the insect biomass into nutrients to support nematode development, rather than as virulence factors¹⁰⁰ (Fig. 3).

So far, lrp is the only regulatory gene reported that functions in both pathogenesis and mutualism (Fig. 5). An lrp mutant colonizes IJ nematode vesicles at 23% of wild-type X. nematophila levels. This might be due to Lrp repression of the nil (nematode intestinal locus) genes. As described above, nilB encodes an outer membrane protein and was discovered in a screen for X. nematophila mutants that are defective in colonization. This screen also revealed two additional membrane-protein-encoding genes necessary for colonization: nilA, predicted to encode a small inner membrane protein³², and nilC, encoding an outer membrane, periplasmically orientated lipoprotein³⁵. nilA, B and C are genetically linked: nilA and nilB are co-transcribed whereas nilC is divergently orientated³².

Lrp represses transcription of nilA, B and C synergistically with NilR, a small helix-turn-helix-containing protein^31,35,52 (Fig. 5). This repression was unexpected, as Lrp is also necessary for colonization. An initial hypothesis to explain this finding was that inappropriately high levels of nil gene expression in the lrp mutant could be detrimental to colonization³⁵. Contrary to this idea, nilR mutants have similarly high levels of nil gene expression, but do not have a colonization defect⁵². Ectopic expression of nilR caused a ∼60-fold decrease in colonization levels, raising the possibility that this strain fails to de-repress nil gene expression sufficiently during colonization⁵². Repression of the Nil proteins, including the outer membrane protein NilB, might provide a selective advantage under some conditions, possibly during insect infection. The spirochaete B. burgdorferi also regulates its outer membrane protein profile in response to the transition between its tick vector and mammals. B. burgdorferi requires ospA to colonize tick midguts and OspA is the predominant surface protein expressed in this environment. When the tick begins to feed, OspA is replaced by another surface protein, OspC, which is necessary for B. burgdorferi interactions in other environments (reviewed in Ref. 103). Like the X. nematophila nil genes, ospA is regulated by repression through the sigma factor RpoS¹⁰⁴. Therefore, in both X. nematophila and B. burgdorferi, surface proteins necessary for interacting with the vector host are repressed and the signals that trigger de-repression should yield insights into how these bacteria sense the host-to-vector transition.

Conclusions

X. nematophila has emerged as a relevant and tractable system to elucidate and compare mechanisms of pathogenesis and mutualism. Studies have established the discrete stages, regulatory hierarchies and effectors that define X. nematophila interactions with each of its hosts. These studies have revealed that X. nematophila senses, in part through the global regulator Lrp, its transition between hosts by a change in nutritive status, resulting in the appropriate regulation of host-interaction effectors. Current evidence suggests that X. nematophila mutualistic colonization of its nematode host might include bacterial lectin–host glycan interactions. Using this model as a framework, molecular analysis of the responses of S. carpocapsae nematodes to X. nematophila colonization and continued characterization of X. nematophila colonization factors will provide detailed information regarding the initiation, specificity and maintenance of this mutualism. Meanwhile, examination of the X. nematophila– insect pathogenic interaction has established mechanisms for robust microbial suppression of host immunity. The current challenge is to elucidate the X. nematophila effectors that mediate suppression of susceptible immune response targets.

By comparing the details of the interactions that X. nematophila makes with different hosts to those gleaned from the study of other bacteria–host relationships we should be able to produce a guide to understanding common and unique aspects of communication between friends and foes.

Box 1 | Insect immunity

Microbial invaders are detected by surveillance factors termed pattern recognition receptors (PRRs) that recognize conserved microbial molecules. Manduca sexta PRRs protect larvae from bacterial infection^105,106,107 and are transcriptionally induced in response to bacterial infection^106,107,108. PRR activation results in haemocyte activities (the cellular response) and the release of soluble molecules with antimicrobial activities (the humoral response). The cellular response of insects includes nodulation and phagocytosis. When triggered by microbial challenge, ENF family cytokines stimulate aggregation and adherence of haemocytes^109,110. Haemocyte aggregation leads to nodulation, in which haemocytes and bacteria form a complex trap of cell debris and extracellular material¹¹⁰. Nodules become melanized through humoral activation of the enzyme phenoloxidase (PO)¹¹⁰. Haemocyte aggregation and nodulation are activated in part by eicosanoids, a class of fatty acids that are generated from membrane phospholipids by the activity of phospholipase A₂ (Ref. 111).

The humoral prophenoloxidase (PPO) cascade is activated by a series of proteolytic activities and is similar to the thio-ester regions of C3 and C4 complement proteins of vertebrates¹¹². Once initiated, the PO cascade results in the creation of immunologically active compounds such as opsonins and cytotoxins and can induce changes in haemocyte behaviour¹¹³. Antimicrobial peptides (and proteins such as lysozyme) function by permeabilizing microbial membranes and contribute to disease resistance^106,114. Transcripts encoding Drosophila antimicrobial peptides are induced in response to lipopolysaccharide or bacteria by conserved signal transduction cascades²⁰. Transcripts for antimicrobial peptides and proteins are similarly induced in M. sexta^{38,106,108,115}. However, the M. sexta signal transduction cascades linking detection to antimicrobial peptide transcript induction have not been elucidated.