Isolation of naturally associated bacteria of necromenic Pristionchus nematodes and fitness consequences

Nematodes and bacteria are among the most numerous organisms on Earth with numbers of nematodes thought to exceed 1 million per m²(Floyd et al., 2002) and numbers of bacterial cells in 1 g of soil thought to be approximately 10¹⁰ (Faegri et al.,1977). Many nematodes use bacteria only for food but some species have more specialized interactions, ranging from mutualism to parasitism. For example, entomopathogenic nematodes of the genera Steinernema and Heterorhabditis exhibit a close symbiotic relationship with the bacteria Xenorhabdus and Photorhabdus, respectively(Forst et al., 1997) and rely on these bacteria to cause mortality to insect hosts. The slug parasitic nematode Phasmarhabditis hermaphrodita is thought to depend on the bacterium Moraxella osloensis to kill slugs and snails(Wilson et al., 1995a; Wilson et al., 1995b; Tan and Grewal, 2001; Tan and Grewal, 2002). Filarial nematodes such as Brugia malayi require the endosymbiotic bacterium Wolbachia for development, fertility and survival(Taylor et al., 2005). The combination of the pine wood nematode, Bursaphelenchus xylophilus,and strains of Pseudomonas fluorescens are thought to be responsible for causing an increase in mortality of pine trees(Han et al., 2003). Marine nematodes from the family Stilbonematinae have a mutualistic relationship with thiotrophic ectosymbiotic bacteria(Nussbaumer et al., 2004).

Although the interaction of the aforementioned nematodes with specific bacteria has been well documented, little is known about laboratory nematode model organisms, such as Caenorhabditis elegans and Pristionchus pacificus. C. elegans is a model organism for many areas of biology (see The C. elegans Research Community, 2005), whereas P. pacificus has been established as a satellite organism in evolutionary developmental biology(Hong and Sommer, 2006a). Both species have – at least in part – been selected as model organisms because they can be cultured in the laboratory using artificial Escherichia coli OP50 as food(Brenner, 1974; Sommer et al., 1996).

Recent studies started to investigate the environment in which Pristionchus can be found in nature. Several field studies revealed that Pristionchus nematodes have close associations with scarab beetles and the Colorado potato beetle (Leptinotarsa decemlineata)(Herrmann et al., 2006a; Herrmann et al., 2006b). For example, P. pacificus was isolated from the oriental beetle(Exomala orientalis) in Japan and the United States(Herrmann et al., 2007). Biological surveys of beetle-associated Pristionchus species have concentrated on Europe, North-America, Japan and South Africa. In total, more than 1200 Pristionchus isolates have been obtained from more than 15 000 surveyed beetles. These isolates fall into 18 distinct species with a specific biogeographic pattern (Mayer et al., 2007). P. pacificus currently represents the only cosmopolitan species (Zauner et al.,2007).

Pristionchus nematodes show a high species specificity with certain beetles. For example, the two European species Pristionchus maupasi and Pristionchus entomophagus are found on cockchafers(Melolontha sp.) and dung beetles (Geotrupes sp.),respectively (Herrmann et al.,2006a). Similarly, the Colorado potato beetle, which lives in Europe and North America, is highly infested with Pristionchus uniformis (Herrmann et al.,2006b). Chemoattraction studies have shown that different Pristionchus species display unique chemoattraction profiles towards insect pheromones and plant volatiles(Hong and Sommer, 2006b)demonstrating the utility of such assays for probing the nematodes'environment under laboratory conditions. Pristionchus chemoattraction is highly diverse and is presumably involved in shaping the specific interaction with host beetles.

In general, nematode–insect associations can be categorized as phoretic, necromenic or parasitic (Kiontke and Sudhaus, 2006). Pristionchus has a necromenic association with beetles whereby the infective juvenile nematodes enter an insect, wait for the death of the host and then feed on bacteria and fungi that proliferate on the insect carcass. Necromenic associations are typically more specific than phoretic associations, in which nematodes use insects or other invertebrates for transport but not as food. It has been suggested that necromeny represents a pre-adaptation for the evolution of true parasitism because the nematode is exposed to low oxygen levels, high temperatures and toxic host enzymes (Weischer and Brown,2000).

In the context of the different life-style of Pristionchusnematodes, it is important to note that these nematodes show major morphological and physiological adaptations with respect to feeding when compared with C. elegans and other rhabditids. C. eleganshas a grinder in the terminal bulb of the pharynx, which disrupts food bacteria (such as E. coli OP50), and under laboratory conditions bacteria are completely lysed (Fig. 1A,B). By contrast, Pristionchus nematodes have a pharynx with a metacorpus and a terminal bulb typical for rhabditid nematodes but do not have a grinder (Fig. 1C)(Fürst von Lieven and Sudhaus,2000; Chiang et al.,2006). Pristionchus worms do not completely lyse bacteria and intact cells can be found in the intestine as revealed by transmission electron microscopy (TEM; Fig. 1D). Interestingly, it has been suggested that Pristionchus nematodes might be actively involved in bacterial dissemination in the wild (Chantanao and Jensen, 1968; Poinar, 1983).

Here, we begin an analysis of the tritrophic interactions of Pristionchus nematodes with bacteria associated with beetles and soil. To gain insight into this relationship we chose the following aims: (1)to identify (using metagenomic and microbiological techniques) bacteria that are associated with Pristionchus nematodes isolated from soil and beetle hosts, i.e. P. maupasi (from cockchafers), P. entomophagus (from dung beetles) and P. pacificus (from the oriental beetle); (2) to investigate the effect these bacteria have on a range of Pristionchus species using chemotaxis, survival and fecundity assays; (3) to assess survival of P. pacificus to human(Pseudomonas aeruginosa and Staphylococcus aureus) and insect (Xenorhabdus nematophila, Xenorhabdus sp. and Photorhabdus luminescens) pathogens; (4) to examine whether physiological functions are variable between nematode species and are affected by bacteria.

Nematode strains used were: Pristionchus pacificus (PS0312), P. entomophagus (RS0144), P. maupasi (RS5015) and Caenorhabditis elegans (Bristol N2). Nematodes were grown at 20°C on nematode growth medium (NGM) seeded with E. coli OP50 before use in experiments. Bacteria were isolated from beetle nematodes and soil nematodes. Bacteria were maintained on LB agar plates at 30°C. In all experiments bacteria were grown overnight in LB broth at 30°C. Bacteria used in this study were: Pseudomonas aeruginosa PA14, Staphylococcus aureus Newman, Photorhabdus luminescensTT01Tn7GFP (associated with the nematode Heterorhabditis bacteriophora), Xenorhabdus sp. (associated with Steinernema scapterisci), Xenorhabdus nematophila (associated with Steinernema carpocapsae). Phase variation in X. nematophila,Xenorhabdus sp. and P. luminescens was monitored using NBTA agar(nutrient agar supplemented with Bromothymol Blue and triphenyltetrazolium chloride) and only phase 1 bacteria were used, which are the most virulent. All equipment was sterilized before use and bacteria were only isolated under a laminar flow. To rule out any airborne contaminants 10 LB plates were placed in the laminar flow for 5–8 h and then sealed and stored at 30°C overnight and monitored for colonies the next day. No bacterial colonies or contaminants were recorded, proving that all bacteria isolated were solely from the nematodes.

Cockchafers (Melolontha hippocastani) were collected from Karlsruhe, Germany, Oriental beetles (Exomala orientalis) were collected from Carver, Massachusetts and dung beetles (Geotrupes sp.)were collected from the Schönbuch forest, near Tübingen, Germany. Beetles were cut in half transversely with scissors and then placed on 6 cm NGM plates and stored at room temperature. Beetles were inspected daily for 7–14 days for nematodes moving or reproducing on the cadavers. Any nematodes that resembled Pristionchus were removed, washed in M9 buffer for 1–2 min and placed on separate LB plates and incubated at room temperature. After 48 h the individual nematodes were removed, lysed and sequenced (see below for methods) to confirm whether the nematodes were the correct species of Pristionchus (according to Mayer et al., 2007). Bacteria growing on LB plates were isolated, subcultured and then prepared for sequencing for species identification using PCR amplification of 16S ribosomal RNA genes (Lane, 1991). The bacteria isolated from nematodes from beetles will be referred to as`beetle-derived' bacteria, hereafter.

We also extracted one adult P. entomophagus from a soil sample from the Schönbuch forest, Tübingen and isolated the bacteria that were subsequently excreted from the nematode gut and removed from the nematode cuticle and grew them on an LB plate. The bacteria were then identified (as described below) and will be referred to as `soil-derived' bacteria.

Bacteria of each species were grown overnight in LB broth and DNA was extracted using Aqua Pure genomic DNA kit (Bio-Rad, Hercules, CA, USA). Polymerase chain reaction (PCR) amplification of bacterial 16S rRNA genes was carried out in 20 μl reactions using primer set 27f(5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492r(5′-TACGGYTACCTTGTTACGACTT-3′)(Lane, 1991). Thermal cycling conditions were as follows: 3 min at 95°C followed by 35 cycles of 15 s at 95°C, 30 s at 55°C, 1.5 min at 72°C, and a final step of 8 min at 72°C. A typical reaction contained 2 μl 10× PCR buffer, 2 μl 2 mmol l^–1 dNTPs, 1 μl 10 μmol l^–1 27f,1 μl 10 μmol l^–1 1492r, one unit of Taq DNA polymerase,12.8 μl H₂O and 1 μl of bacterial DNA. PCR amplicons were visualized by standard agarose gel electrophoresis(Sambrook et al., 1989) and bands were excised using a clean scalpel. DNA was extracted from bands using QIAquick gel extraction kit (Qiagen, Valencia, CA, USA).

After isolation of bacteria, nematodes were removed from the LB plate and identified using the small subunit rRNA gene. Genomic DNA from single nematodes was isolated using the NaOH digestion method of Floyd et al.(Floyd et al., 2002). Briefly,single worms were added to 20 μl of 0.25 mol l^–1 NaOH and incubated at 25°C overnight. The worm mixture was then heated to 99°C for 3 min before the addition of 4 μl of 1 mol l^–1 HCl, 10μl of 0.5 mol l^–1 Tris-HCl (pH 8.0) and 5 μl of 2%Triton X-100. The mixture was then heated to 99°C for 3 min, frozen to–20°C and then heated for a further 3 min at 99°C. Two microlitres of the extract were then used for PCR. DNA was amplified using the primers SSU18A (5′-AAAGATTAAGCCATGCATG-3′) and SSU26R(5′-CATTCTTGGCAAATGCTTTCG-3′). PCR was carried out in 25 μl reactions containing 2.5 mmol l^–1 MgCl₂, 0.16 mmol l^–1 each deoxynucleoside triphosphate, 0.5 μmol l^–1 each primer, 2 μl lysate, 2 units Taq DNA polymerase(Amersham Biosciences, Piscataway, NJ, USA). The mixture was then subjected to the following PCR conditions: 2 min at 95°C, 35 cycles including, 15 s at 95°C, 15 s at 50°C, 2 min at 72°C, followed by 7 min at 72°C. PCR products were then diluted 10–20-fold and added to the Big Dye terminator sequencing mix (Applied Biosciences, Foster City, CA, USA), which contained the sequencing primer SSU9R(5′-AGCTGGAATTACCGCGGCTG-3′). For bacteria we required a minimum length of 200 base pairs for the query sequence (16S rRNA). Gene sequences of nematodes and bacteria were aligned using Seqman (DNA Star, Madison, WI, USA),compared with GenBank database sequences using Blastn searches using sequence similarity matches at 90%.

Soil samples were taken from the Schönbuch forest, Tübingen and were added to 9 cm NGM plates and stored at room temperature. The plates were then checked for presence of nematodes every day for the next 7 days. In total, four P. entomophagus and four P. lheritieriindividuals were isolated, washed in M9 buffer and placed in single worm lysis buffer. The resultant suspension was then used for nematode identification (as described above) and bacterial cloning. Bacterial DNA was amplified using conditions and reagents as described above and was then cloned using Topo cloning kit (Invitrogen, Carlsbad, CA, USA) following the manufacturer's guidelines. The full-length gene was then ligated into pCR4-TOPO (Invitrogen)and transformed into Top 10 chemically competent bacterial cells for sequence analysis. In total, 683 clones were picked and screened for bacterial inserts using PCR primers T7 (5′-TAATACGACTCACTATAGGG-3′) and T3(5′-ATTAACCCTCACTAAAGGGA-3′). Bacterial inserts were then sequenced using the same methods as described above.

Chemotaxis assays were modified from previous studies(Zhang et al., 2005; Hong and Sommer, 2006b). Briefly, 25 μl of overnight bacterial suspension was placed 0.5 cm away from the edge of a 9 cm Petri dish filled with NGM medium. The same amount of E. coli OP50 was placed on the opposing side and acted as the counter attractant. Approximately 50–200 J4/adult stage Pristionchusindividuals were placed between the two bacterial spots. All nematodes used were previously fed on E. coli OP50. Plates were then sealed with Parafilm® and stored at room temperature in the dark. After 24 h the number of nematodes found in each bacterial spot was recorded. A chemotaxis index was used to score the response of the nematodes, which consisted of:number of nematodes in the test bacteria – numbers of nematodes in control bacteria/total number of nematodes counted(Zhang et al., 2005). This gave a chemotaxis score ranging from –1.0 (total revulsion from test bacteria) to 1.0 (total attraction towards test bacteria). A score of around 0 means there were equal numbers of nematodes in each bacterial spot. Five plates were used per replicate, and the procedure was repeated five times for each bacterium (a total of 25 individual assays).

Chemotaxis experiments were as follows: (1) E. coli OP50 versus soil-derived bacteria; (2) E. coli OP50 versus insect-derived bacteria; (3) B. thuringiensis or Bacillus sp. 1 versus insect-associated bacteria –this was used to examine the effect of removing E. coli from the analysis and using more ecologically relevant controls; (4) E. coliOP50 versus insect and human pathogenic bacteria; (5) P. luminescens versus human and insect pathogenic bacteria.

Liquid cultures of all insect- and soil-derived bacteria as well as human and insect pathogens were grown overnight at 30°C. Bacterial suspensions(200 μl) were spread evenly on 6 cm Petri dishes with NGM medium and incubated overnight. Twenty J4 P. pacificus were added to each plate and stored at 25°C. Survival of worms was monitored daily for 8 days. Nematodes were transferred every 2 days to fresh plates to prevent misidentification of original worms from offspring. Mortality was determined by prodding worms with a metal pick and nematodes that did not respond were considered dead. One hundred P. pacificus were exposed to each of the soil-derived and insect-derived bacteria as well as the human pathogens(P. aeruginosa and S. aureus), the insect pathogens (X. nematophila and Xenorhabdus sp. and P. luminescens) and E. coli OP50 was used as the control. C. elegans was also exposed to each bacterium as a comparison to P. pacificus. Nematodes were only exposed to bacteria at the phase 1 stage which is the most virulent and were transferred to fresh plates every 2 days to ensure nematodes would be only exposed to phase 1. Nematodes were well fed on E. coli OP50 before addition to pathogenic bacteria to avoid starving. Any mortality observed was due to bacterial pathogenicity and no starvation or bagging behaviour was observed.

Twenty microlitres of overnight bacteria suspension was placed on separate Petri dishes with NGM agar and left to dry. Five single virgin hermaphrodites of P. pacificus were individually placed on separate dishes in the bacterial spot and were stored at 25°C. The numbers of live offspring produced by each worm was recorded daily. Worms were transferred to fresh plates daily for 4–5 days.

Liquid cultures of E. coli OP50, Xenorhabdus sp. and Bacillus sp. 2 were cultured overnight at 30°C. Bacteria were mixed with red fluorescent 0.5 μm carboxylate-modified polystyrene latex beads (Sigma Aldrich, St Louis, MO, USA) to a final bead concentration of 0.8%. Standard 6 cm NGM agar plates were seeded with 20 μl of the bacteria–bead mixture and when dry, ten J4 P. pacificus or C. elegans larvae were added to separate dishes and allowed to feed overnight at 20°C. Cycle length was recorded by timing the intervals between defecations using a dissecting microscope with a halogen light source. For each species, ten defecations of ten individual worms were recorded.

Residence time was assessed by allowing worms to feed overnight on the bacteria–bead mixture (using methods as described above). Individual worms were transferred to fresh bacterial plates and the total time taken to clear the intestine of the fluorescent bead mixture was recorded as well as the total number of defecations. We took fluorescent images of worm faeces to determine the number of defecations with beads still present (until no further fluorescence could be observed). Ten worms were analyzed for each treatment.

Chemotaxis scores, fecundity and survival (after 7 days) were compared using one way analysis of variance (ANOVA) and differences between treatments were determined using the Bonferroni multiple comparison test using Small Stata, 9.2 (StataCorp, College Station, TX, USA). Mean defecation cycles and residence times were compared using Student's t-test for means. The effect of bacteria over time was analyzed using Kaplan–Meier and log rank tests.

A total of eight Pristionchus individuals (four P. entomophagus, four P. lheritieri) were isolated from soil samples from the Schönbuch forest, Tübingen (Germany). From those animals, we sequenced a total of 683 bacterial clones and identified 292 unique sequences with at least 40 different bacterial 16S sequences per individual nematode. The majority of these bacteria belonged to the Pseudomonadales, Burkholderiales, Flavobacteria and Xanthomonadales and also included the plant pathogens Erwinia and Agrobacterium and human pathogens such as Bordetella, Burkholderia and Microbacterium (see Tables S1 and S2 in supplementary material). Other spore-forming bacteria, such as Bacillus sp., cannot be detected with this approach because bacterial spores escape single worm lysis. This non-saturated bacterial sequencing approach clearly indicates that Pristionchus nematodes ingest an enormous diversity of bacteria. However, from this type of analysis it remains unknown if the nematodes can digest all of these bacteria as a food source.

We grew beetle- and soil-derived Pristionchus nematodes on rich medium and isolated a total of 23 bacterial strains(Fig. 2). Specifically, we obtained bacteria derived from each nematode–beetle system, i.e. bacteria from P. pacificus from the oriental beetle (collected in Carver, MA, USA), P. maupasi from the cockchafer (Karlsruhe, Germany)and P. entomophagus from dung beetles (Tübingen, Germany). Together, these bacterial isolates represent all major groups previously identified in the metagenomic sequencing approach. Bacterial isolates were sequenced for species designation and represent single species isolates.

In chemotaxis experiments using soil-derived bacteria (from adult P. entomophagus) we found that Pristionchus species were equally attracted to all available Pseudomonas strains (P>0.05; Fig. 3A). Pristionchusspecies were weakly attracted to the Pseudomonas strains and scored chemoattration indices of 0.1–0.3. By contrast, Bacillus sp. 1 was highly unattractive to the Pristionchus nematodes. More specifically, P. pacificus was significantly more averse to Bacillus sp. 1 than P. maupasi and P. entomophagus(P<0.001). When exposed to Bacillus sp. 2, the response of the Pristionchus nematodes varied (P<0.001). P. entomophagus was more attracted to this bacterium than P. pacificus and P. maupasi, which showed aversion.

Next, we analyzed the chemotaxis behaviour of three Pristionchusspecies exposed to beetle-derived bacteria from three settings. In the first system (cockchafer and P. maupasi) we found no significant differences between the chemotaxis index scores of P. pacificus, P. maupasi and P. entomophagus when exposed to Pseudomonas fluorescens, Pseudomonas aurantiaca, Bacillus megaterium, Stenotrophomas maltophila, Serratia sp. or Bacillus sp. (P>0.05; Fig. 3B). However, when exposed to Bacillus thuringiensis, Enterobacter sp. and Pseudomonas chlororaphis there were significant differences between each nematode species (P<0.05). Most strikingly, P. pacificus was strongly averse to B. thuringiensis. From the second system (dung beetle and P. entomophagus), we isolated five bacteria that are usually found in animal and insect faeces(Holt et al., 2000). P. entomophagus was significantly more attracted to Serratia sp., Enterobacter amnigenus and Proteus vulgaris than P. pacificus and P. maupasi (P<0.05)(Fig. 4A). By contrast, P. maupasi and P. pacificus responded poorly to the Serratia sp., Ochrobactrum sp., Enterobacter sp. and P. vulgaris. In the third system (oriental beetle and P. pacificus) four bacterial species were isolated from P. pacificus emerging from oriental beetles including Serratia sp., Achromobacter sp., Pantoea agglomerans and P. fluorescens. These bacteria are commonly found in soil, water and the digestive tract of insects (Holt et al.,2000). All nematodes responded poorly to oriental beetle-derived bacteria, except for P. fluorescens, which was similarly attractive to all Pristionchus species (Fig. 4B). Taken together, the chemotaxis studies indicate that (1) Pristionchus nematodes never favour one bacterial strain completely over all others, (2) they strongly avoid certain Bacillus strains and(3) there is only a low Pristionchus species-specificity of bacteria in the chemotaxis assays.

P. pacificus grown on Bacillus sp. 1 and Bacillus sp. 2 (from soil-derived nematodes) reduced brood size significantly compared to E. coli OP50 (P<0.001; Fig. 3A). This is also true for B. thuringiensis, P. aurantiaca and Serratia sp. from cockchafers (P<0.05; Fig. 3B), Ochrobactrum, P. vulgaris and Serratiaisolated from the dung beetle (P<0.05; Fig. 4A) and S. marcescens,P. agglomerrans and Achromobacter sp. from oriental beetle(P<0.05; Fig. 4B).

The only bacteria that affected the survival of P. pacificus were Serratia sp. from cockchafers, and S. marscens, P. agglomerrans and Achromobacter sp. from the oriental beetle(P<0.05).

Next, P. pacificus was given the choice of either E. coliOP50, B. thuringiensis or Bacillus sp. 1 as a counter attractant to a range of insect-associated bacteria as an attractant. The nematode consistently avoided both Bacillus sp. 1 and B. thuringiensis and scored chemotaxis indices between 0.96–0.99 and 0.85–0.96, respectively (Fig. 5). The response to the Bacillus species was significantly different from the response to the E. coli OP50 control(P<0.001). These results demonstrate the revulsion Pristionchus has to these Bacillus species and provides a first indication that these nematodes might be able to avoid bacteria that have a potential harmful effect on their fitness.

We assessed survival of P. pacificus and C. elegans to human and insect pathogens. P. pacificus was highly susceptible to P. luminescens, X. nematophila and Xenorhabdus sp. but was not susceptible to S. aureus and P. aeruginosa(Fig. 6A). After 24 h, only 1±0.55 P. pacificus were alive compared with 12±0.87 and 19±0.19 P. pacificus exposed to X. nematophilaand Xenorhabdus sp., respectively. P. luminescens is therefore the most pathogenic of the entomopathogenic bacteria to P. pacificus. The number of surviving P. pacificus exposed to S. aureus and P. aeruginosa was similar to the numbers surviving E. coli OP50 exposure (P>0.05). C. elegans was highly susceptible to both human and insect pathogens(P<0.001; Fig. 6B)which is in stark contrast to the immunity P. pacificus has to S. aureus and P. aeruginosa.

In chemotaxis assays P. pacificus was exposed to either E. coli OP50 (as counter-attractant) or X. nematophila, Xenorhabdussp., P. aeruginosa or S. aureus in separate experiments. We found that P. pacificus strongly avoided the killer species of bacteria (X. nematophila and Xenorhabdus sp.) and the human pathogens (S. aureus and P. aeruginosa; Fig. 6C). There was no difference between the responses of the nematode to these bacteria(P>0.05). When E. coli OP50 was replaced with P. luminescens as the counter-attractant, P. pacificus was attracted to the human pathogens. The nematode found it difficult to distinguish between X. nematophila and P. luminescens, as well as Xenorhabdus sp. and P. luminescens, which was expected as the nematodes were faced with a choice of two unpleasant toxic species (Fig. 6B).

We wanted to determine whether the absence of the grinder in P. pacificus affected defecation time and residence time compared with those in C. elegans. When P. pacificus is fed on E. coliOP50 the mean defecation time was 106±6.7 s compared to 48±1.9 s for C. elegans (P<0.001; Fig. 7A). When P. pacificus was fed on Bacillus sp. 2 the defecation cycle was significantly longer than when fed E. coli OP50(P<0.001). There was no difference in defecation cycle when fed toxic Xenorhabdus sp. or E. coli OP50 (P>0.05). Conversely, the defecation cycle of C. elegans did not alter when fed Bacillus sp. 2 but was longer when fed Xenorhabdus(P<0.05). Videos demonstrating physiological differences in defecation cycles, as well as step-by-step muscle contractions can be seen in Movie 1 and Fig. S1 in supplementary material.

We defined the residence time as the total time recorded for a feeding worm to clear the gut of the OP50–fluorescent bead mixture after being transferred to a bacterial spot without beads. When E. coli OP50 was replaced with Bacillus sp. 2 the residence time for P. pacificus was significantly longer (P<0.05), whereas the residence time for C. elegans did not differ significantly(P>0.05; Fig. 7B).

We began an analysis of the tritrophic interactions of Pristionchus nematodes with bacteria associated with beetles and soil. We identified bacteria from Pristionchus nematodes using metagenomics and microbiological techniques and investigated the effect these bacteria have on a range of Pristionchus species using chemotaxis,survival and fecundity assays. We assessed survival of P. pacificusto human and insect pathogens and examined whether physiological functions are affected by bacterial species.

The interactions between Pristionchus and bacteria must be considered in the context of the necromenic life-style of these nematodes. On the death of the beetle infected with Pristionchus dauer larvae,bacteria and fungi rapidly colonize the beetle's body. This in turn creates a toxic `beetle soup' consisting of large numbers of competing microorganisms originally present in the beetle intestine and the local environment. Around this time Pristionchus is thought to exit from the resistant dauer stage and develop into the J4 stage and then into adult form, with the main purpose to feed and reproduce. By distinguishing between an array of pathogenic and non-pathogenic bacteria and choosing the correct species to feed on, Pristionchus can lower the possibility of eating bacteria that cause low brood size, slow development and mortality. Once the food supply is depleted, the remaining nematodes turn into dauer larvae and search for potential beetle hosts in the surrounding soil. It is not known whether Pristionchus retains the bacteria it encounters before turning into the dauer form. Insect parasitic nematodes such as S. carpocapsaestore their symbiotic bacteria X. nematophila in an intestinal vesicle (Martens et al., 2003)whereas H. bacteriophora stores P. luminescens in the intestine (Ciche and Ensign,2003). Fedorko and Stanuszek(Fedorko and Stanuszek, 1971)observed P. uniformis dauer larva carrying bacteria cells in the gut but this observation has yet to be confirmed for Pristionchus species living on scarab beetles.

In general, nematodes show an enormous range of feeding strategies that allows them to occupy innumerable ecological niches(Munn and Munn, 2002). Predatory and plant parasitic nematodes have stylets, whereas soil-dwelling rhabditids, such as C. elegans, have a grinder in the terminal bulb of the pharynx. In contrast to C. elegans, P. pacificus lacks a grinder. Under laboratory conditions, bacterial lysis in the gut is incomplete and bacteria can survive the passage through the P. pacificus gut. We speculate that in order to gain any nutrition from these bacteria Pristionchus will have to increase time taken for digestion. Also, Pristionchus might be actively involved in bacterial dissemination in the wild, a claim that has already been made several decades ago (Chantanao and Jensen, 1968; Poinar,1983).

The metagenomic analysis revealed that Pristionchus harbours a huge diversity of bacteria within its gut and on its cuticle including plant pathogenic and opportunistic human pathogens. Although mechanisms of immunity have been discovered in C. elegans in response to human pathogenic bacteria, C. elegans is susceptible to a range of naturally occurring Gram-negative and -positive bacteria and fungi(Ewbank, 2002). We have shown that Pristionchus associates with many different pathogenic bacteria in nature and P. pacificus is resistant to a number of bacteria that C. elegans is susceptible to, such as S. aureus, P. aeruginosa and P. fluorescens. Although future studies will reveal the exact molecular basis behind this resistance, the analysis of the P. pacificus genome provides the first insight into potential underlying mechanisms. When compared to C. elegans, the P. pacificus genome shows a large expansion of genes encoding cytochrome P450 enzymes, glucosyl transferases, ABC transporters and other proteins that are thought to be involved in the degradation of xenobiotic compounds (C. Dieterich, S. W. Clifton, L. Schuster, A. Chinwalla, K. Delehaunty, I. Dinkelacker, R. Fulton, J. Godfrey, P. Minx, M. Mitreva et al., manuscript in revision). Future research will focus on genetic studies to elucidate methods of innate immunity, pathogenicity of associated bacteria and methods of detection and avoidance when in contact with pathogenic bacteria.

This study is one of the first to utilize both microbiological and metagenomic techniques to isolate bacteria from nematodes. Studies of other nematodes such as the slug parasitic nematode P. hermaphrodita have isolated 13 bacterial species from dauer larvae, culture medium and infected slugs (Wilson et al., 1995a; Wilson et al., 1995b). Bacterial communities associated with invertebrates have also been assessed using metagenomic tools. For example the bacterial communities in the hindgut paunch of a wood-feeding termite recorded 1750 bacterial 16S rRNA gene sequences that represented 12 phyla and 216 phylotypes(Warnecke et al., 2007). We did not expect to find a wealth of bacteria present in the nematode gut and on the cuticle. The function and relationship of many of these bacteria to Pristionchus remains unknown but from these results it can be seen that the nematodes associate with a huge diversity in nature.

In chemotaxis assays P. pacificus avoids species that cause ill health but the mechanism for sensing the causal properties of these bacteria currently remains unknown. We exploited this behaviour by using different Bacillus strains as a control and making other bacteria more attractive. Thus, the P. pacificus response to bacteria in chemotaxis assays differs from that of C. elegans in the type of attraction (or repulsion). Similar differences have also been observed for the P. pacificus and C. elegans responses to pure insect-associated chemicals (Hong and Sommer,2006b). Most likely, all of these differences are part of the adaptive forces that shape the tripartite interactions between bacteria,nematodes and beetle hosts in the wild.

C. elegans protects itself from potential pathogenic bacteria by avoidance behaviour and innate immunity pathways(Pujol et al., 2001; Nicholas and Hodgkin, 2004; Kurz and Ewbank, 2003). Avoidance behaviour of C. elegans has been recorded for B. thuringiensis, S. marcescens, P. aeruginosa, P. luminescens and M. nematophila using different molecular mechanisms(Pujol et al., 2001; Pradel et al., 2007; Yook and Hodgkin, 2007; Zhang et al., 2005; Beale et al., 2006; Schulenburg and Müller,2004; Sicard et al.,2007; Hasshoff et al.,2007). In our experiments, bacteria that cause mortality to P. pacificus tend to score low in the chemotaxis index, ranging from–0.09±0.04 for Serratia sp. (from the cockchafer) to 0.37±0.05 for S. marcescens (from the oriental beetle). Unlike previous studies (Zhang et al.,2005) that demonstrated that C. elegans grown on a mixture of OP50 and P. aeruginosa or S. marcescens would avoid the two human pathogens in chemotaxis experiments, we have shown that OP50-raised Pristionchus species avoid Bacillus species without any training. Nematodes exposed to Bacillus species,particularly Bacillus sp. 1 and B. thuringiensis had significantly lower brood size than those grown on E. coli OP50. Previous studies have demonstrated that exposure of P. pacificus to purified Cry 5B crystal protein from B. thuringiensis causes a significant reduction in brood size and affects development(Wei et al., 2003).

P. pacificus is highly susceptible to P. luminescens and X. nematophila and Xenorhabdus sp. These bacteria are commonly found in entomopathogenic nematodes (Steinernema and Heterorhabditis) and are responsible for causing insect mortality 24–48 h after nematode penetration(Forst et al., 1997). As Pristionchus has a strong relationship with a number of beetle hosts that entomopathogenic nematodes can also infect, e.g. Steinernema scarabaei, which has been isolated from oriental beetles(Stock and Koppenhöfer,2003), the chances of co-infection with Pristionchus and entomopathogenic nematodes and their associated bacteria are high. Our studies suggest that P. pacificus can recognize and avoid highly pathogenic bacteria such as X. nematophila. The ecological interaction between Pristionchus and entomopathogenic nematodes and bacteria clearly warrants further research.

P. entomophagus was significantly more attracted to dung beetle bacteria than to P. maupasi and P. pacificus. This was the only nematode–beetle system for which bacterial specificity was recorded, as generally the three Pristionchus species tested responded similarly when exposed to bacteria isolated from the respective beetle host. From this study the reasons behind nematode–beetle host specificity has not be discovered but it is not due to bacteria harboured in the beetle gut. As Pristionchus nematodes show a high species specificity with cockchafers, dung beetles and Colorado potato beetles(Herrmann et al., 2006a; Herrmann et al., 2006b) and display unique chemoattraction profiles towards insect pheromones and plant volatiles (Hong and Sommer,2006b) other reasons apart from bacteria species must be considered. The behavioural response of the entomopathogenic nematode S. carpocapsae is correlated with nematode-induced mortality and number of infective juveniles produced on each host species(Lewis et al., 1996; Lewis, 2002). Other reasons for host specificity may include increased production of males, higher reproduction rate or better health. Also as well as bacteria living in the beetle, fungi and other parasitic or phoretic nematodes are present, perhaps these organisms contribute to nematode–beetle specificity. Further studies are needed to elucidate the exact mechanism behind these complex nematode–beetle associations.

Taken together, we have analyzed the tritrophic interactions of Pristionchus nematodes with various bacteria from soil and beetles. We found a range of different interactions from bacterial dissemination by the worm to reduction in brood size, longer defecation cycles and nematode mortality caused by certain bacterial strains. Pristionchus can recognize, respond to and avoid bacteria that cause poor health. The ability to discriminate between bacteria is important for success and survival in the soil ecosystem. Given the genetic and genomic toolkit available in P. pacificus, this nematode interaction with its living environment can in the future be investigated at the molecular level to provide mechanistic insight into the adaptive and non-adaptive forces that shape this nematodes ecosystem.

We are grateful to Elissa Hallem and Todd Ciche for P. luminescens, Heidi Goodrich Blair for X. nematophila, Jane Roberts and Jeremy Pearce at Becker Underwood, UK for strains of entomopathogenic bacteria. We thank members of the Sommer lab for discussions and Drs R. Hong and D. Bumbarger for comments of the manuscript. Also we thank Paul Robbins for collection of oriental beetles. This research was funded by the Max-Planck Society.