SEARCH

SEARCH BY CITATION

Keywords:

  • bacteriocin;
  • evolution of virulence;
  • growth inhibition;
  • kin selection;
  • parasite migration rate;
  • Xenorhabdus nematophila

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Competition among different parasite genotypes within a host is predicted to affect virulence. The direction of this effect, however, depends critically on the mechanisms that parasites use to compete or to cooperate with each other. One mechanism that bacteria use to compete with each other is via the production of bacteria-killing toxins, called bacteriocins. This warfare among parasites within a host is predicted to reduce the rate of host exploitation, resulting in lower virulence. By contrast, if parasites within a host are highly related, there could be a reduction in within-host conflict, increasing virulence. We examined this idea by allowing an insect-parasitic nematode (Steinernema carpocapsae) and its symbiotic bacteria (Xenorhabdus nematophila) to evolve for 20 passages under two different migration treatments (low and high). We found that host mortality rates were higher in the low-migration treatment when compared with the high-migration treatment. In addition, bacteria isolated from the same insect host inhibited each other’s growth, but only in the high-migration treatment. These results show that population structure and interactions among parasites within hosts can be critical to understanding virulence.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

The expected relationship between the genetic similarity of co-infecting parasites (relatedness) and the detrimental effect they cause their host (virulence) has undergone a paradigm shift in recent years (Brown et al., 2002; Foster, 2005). The long-standing prediction was that when parasites co-infecting a host are highly related, kin selection should favour prudent host exploitation and reduced virulence relative to a co-infection comprised of unrelated parasites (Hamilton, 1972; Frank, 1992, 1994, 1996). The reasoning being that within-host competition would drive unrelated parasites to overexploit their host, leading to a ‘tragedy of the commons’ and greater virulence. This view assumes that the competition among parasite strains is mediated through resource utilization. However, several recent studies have shown that parasites can engage in other forms of interactions (Turner & Chao, 1999; Griffin et al., 2004; Massey et al., 2004). For example, parasites may cooperate in attacking their host by producing a public good, which may benefit individuals other than the producing parasite. In this case, virulence increases rather than decreases with high relatedness as fewer freeloaders are maintained (Chao et al., 2000; West & Buckling, 2003). Alternatively, resource competition among co-infecting parasites can lead to interference competition or antagonistic interactions between them, which can also affect the relationship between virulence and relatedness (Gardner et al., 2004).

One of the most intriguing means of interference competition is a form of chemical warfare engaged in by bacteria. All lineages of bacteria have a diverse arsenal of weapons, called bacteriocins, which are bacteriocidal toxins thought mainly to affect closely related strains and species (Riley & Chavan, 2007). Individual clones are immune to their own bacteriocin, but they pay a cost for production, as lysis is often involved in bacteriocin release (Riley & Gordon, 1999). Despite their ubiquity, the roles of bacteriocins in nature are virtually unknown. While they have long been assumed to be important in intraspecific competition, only recently has it been demonstrated that colicins (bacteriocins produced by Escherichia coli) are effective mediators of intraspecific competition in vivo (Kirkup & Riley, 2004). Additionally, some bacteriocins are involved in quorum sensing (Gillor, 2007), and evidence suggests a possible role in community-level interactions (Tait & Sutherland, 2002; Riley et al., 2003).

Bacteriocins are also thought to influence virulence. Massey et al. (2004) showed that host death rate was lower when hosts were co-infected with two different genera of bacteria, which were capable of inhibiting each other’s growth in vitro. They suggested that bacteriocin-based interactions might cause virulence in mixed infections to be lower than in single-strain infections. Moreover, a phenotypic model developed by Gardner et al. (2004) predicted that allocation to bacteriocin production, and hence virulence, might evolve in response to the average relatedness of con-specific competitors. These studies motivated us to examine the link between relatedness, virulence and bacteriocin activity.

We manipulated parasite population structure to examine its effect on virulence evolution, using the insect-parasitic nematode Steinernema carpocapsae and its symbiotically associated bacteria Xenorhabdus nematophila. This species of bacteria produces a wide array of insect and microbial toxins (Forst & Nealson, 1996) as well as a phage tail-like bacteriocin called xenorhabdicin. Xenorhabdicin has been shown to inhibit the growth of other Xenorhabdus species and some species in closely related genera (Boemare et al., 1992; Thaler et al., 1995; Sicard et al., 2005), but no intraspecific inhibition has been previously shown.

In our study, we allowed parasites (nematodes and their bacteria) to evolve under two different migration treatments. In the low-migration treatment, parasites had a 90% probability of co-infecting a host with parasites emerging from the same insect host in the previous passage; whereas, in the high-migration treatment, parasites were mixed at random with every passage. Our previous work on these experimental lines, which focused on the nematodes, indicated faster host exploitation in the low-migration treatment (Bashey et al., 2007). As both the nematodes and the bacteria produce insect toxins, and both potentially engage in interference competition within the host, faster host exploitation and greater virulence in the low-migration treatment could arise from a greater contribution to insect-toxin production or due to less fighting among the parasites. Here, we focus on the role of antagonistic interactions among co-infecting bacteria. Thus, we compared the treatments, after 20 passages through insect hosts, for differences in virulence and in the level of inhibition observed between bacterial clones isolated from infected insects. We found reduced rates of mortality of insect hosts, and higher levels of inhibition among bacteria, when insects were infected with parasites from the high-migration treatment.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Parasite life cycle and source

Infective juveniles of S. carpocapsae are free-living, nonfeeding and developmentally dormant in the soil (Poinar & Leutenegger, 1968). Each infective juvenile nematode carries approximately 100 X. nematophila cells in a specialized part of its intestine (Martens et al., 2003; Sicard et al., 2003). Upon entering an insect host, the nematode releases its bacteria and resumes development (Poinar, 1966). Both the nematode and the bacteria contribute to killing the insect (Burman, 1982; Dunphy & Webster, 1988; Simoes, 2000) and each species reproduces separately inside the insect carcass. Xenorhabdus nematophila contributes to nematode fecundity (Poinar & Thomas, 1966; Han & Ehlers, 2000; Sicard et al., 2003) and protects the insect carcass from other organisms (Thaler et al., 1997; Zhou et al., 2002; Sicard et al., 2005). When resources become limiting (Popiel, 1989), the juvenile nematodes become colonized by one or two bacteria cells (Martens et al., 2003) and emerge from the insect.

Steinernema carpocapsae infects a broad range of insects (Peters, 1996). In our study, we used the greater wax moth caterpillar, Galleria mellonella, as the insect host. Parasites (the nematode S. carpocapsae and its symbiotically associated bacteria X. nematophila) were obtained from three commercial sources: Integrated Biocontrol Systems Inc. (Greendale, IN, USA: S.c. strain ‘Sal’), Mellinger’s Inc. (N. Lima, OH, USA: S.c. an unidentified strain) and Biocontrol Network (Brentwood, TN, USA: S.c. strain ‘All’) and equal concentrations of each were combined to create the initial parasite source population.

Experimental lines

From this initial source population, 10 experimental lines were established and separately maintained from each other for 20 host passages. In each passage, parasites were propagated by one of two migration treatments through a novel set of the insect host. In the five lines comprising the high-migration treatment, parasites were propagated by infecting a given host with a mixture of parasites that emerged from eight hosts. Each insect was placed in a 55 × 20 mm2 Petri dish lined with filter paper and infected with 200 nematodes in 0.5 mL of de-ionized water. Infected hosts were kept at 26 °C and transferred to modified White traps for collection of emerging parasites (Bashey et al., 2007). In the five lines comprising the low-migration treatment, 90% of the infective dose came from one host and 10% came from a mixture of eight hosts. Thus, in both treatments, parasites were propagated through eight hosts with each passage; however, the treatments differed in the probability that parasites emerging from a given host would co-infect a new host together. Within the low-migration lines, this protocol created eight sublines that represent groups of parasites that were kept together each passage.

Mortality assay

After 20 experimental passages, we compared the insect mortality rates of the low- and high-migration treatments (Fig. 1). Two hundred G. mellonella were infected with parasites (nematodes with their bacteria) from each experimental line and checked for mortality eight times over 72 h postinfection. Each caterpillar was infected in a manner similar to how the experimental lines were propagated: using a dose of 200 nematodes that emerged from eight different hosts (high-migration) in the previous passage or only nematodes emerging from a single host (low-migration). Caterpillars were scored as alive if they moved spontaneously or in response to probing. We performed Cox proportional hazards regression with the TPHREG Procedure in sas/stat®software (SAS Institute Inc., Cary, NC, USA) to estimate the survival probability at each census and to test for an overall difference in survival curves between treatments while accounting for the nested effect of experimental line.

image

Figure 1.  Schematic of the experiment. (i) Insect hosts were infected with parasites (nematodes and their symbiotic bacteria) from one of five experimental lines from either the (a) high- or (b) low-migration treatment. In the high-migration treatment, parasites emerging from eight hosts were randomly mixed at every passage, whereas, in the low-migration treatment, 90% of the infective dose was comprised of parasites emerging from the same host. A total of 200 insects were infected and observed for mortality. (ii) Bacteria were extracted from two hosts per line and (iii) up to eight bacterial colonies were isolated per host resulting in an 8 × 8 set of within-host inhibition assays for each host. In (a), the two hosts sampled from each high-migration line represent replicate infections. In (b), each sampled host was infected with parasites that emerged from different hosts in previous passages, i.e. two different sublines with each low-migration line. (iv) Between-host inhibition assays tested for inhibition between colonies from these different sublines.

Download figure to PowerPoint

Bacteria sampling

To assay inhibition among bacterial clones, 20 dead hosts from the mortality assay described above were sampled for bacteria at 66 h postinfection. The two hosts from each high-migration line represent replicate infections, whereas the two hosts from each low-migration line were infected with parasites that emerged from different hosts in the previous passages (Fig. 1). Each dead host was injected with 1 mL of Luria–Bertani broth (LB) and 1 mL of LB, the haemolymph, and the bacteria it contained were collected. After centrifugations, bacteria were diluted in LB and plated on nutrient agar supplemented with 0.004% (w/v) triphenyltetrazolium chloride and 0.0025% (w/v) bromothymol blue (NBTA). Up to eight colonies were isolated from each insect host, preserved in glycerol and used in the growth inhibition assays. Xenorhabdus nematophila colonies were sampled based on their adsorption of blue dye when grown on NBTA. Species identity was confirmed by sequencing of 16S rDNA (Tailliez et al., 2006).

Growth inhibitions assays

Assays were performed between colonies isolated from the same host (both treatments) and between colonies isolated from different hosts in the same experimental line (low-migration treatment only, see Fig. 1). These clones were grown in liquid culture for 72 h at 28 °C shaking (300 rpm) in 5 mL of LB broth. To test the sensitivity of a clone, molten Soft (MS) agar (0.6% agar) was sowed with 2% (v/v) of its liquid culture. Then, supernatant from the liquid culture of a potential producer clone was filtered through a 0.45-μm membrane, and 15 μL was spotted onto the surface of an MS agar plate containing the potential sensitive clone. Plates were incubated for 48 h, at which time the inhibition of growth of the bacteria in the soft agar below the 15 μL of supernatant could be visualized as a clear zone of inhibition (Fig. 2). Chi-squared tests were performed to see if the total number of inhibitions differed between the sets of inhibition assays.

image

Figure 2.  Photographs showing (a) inhibition or (b) no inhibition of the growth of one bacterial clone caused by the supernatant of another. Line indicates 5 mm.

Download figure to PowerPoint

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Hosts infected by parasites (nematodes and bacteria) from the low-migration treatment died significantly earlier than hosts infected by parasites from the high-migration treatment (inline image = 9.52, P = 0.002, Fig. 3). This difference was most pronounced from 40 to 54 h postinfection, where hosts infected by parasites from the high-migration treatment were more than twice as likely to be alive.

image

Figure 3.  Average survivorship (±1 SE) for insect hosts infected with parasites (nematodes and their bacteria) from either the low- or high-migration experimental lines. Two hundred Galleria mellonella were individually exposed and monitored for 72 h. Survivorship was significantly lower for hosts infected with parasites from the low-migration treatment.

Download figure to PowerPoint

The higher virulence in the low-migration lines was associated with the absence of inhibition between bacterial clones sampled from the same host (Table 1). On the contrary, in the high-migration treatment, inhibitions were observed between bacterial clones isolated from the same host in 75% of the lines. In addition, almost one-fourth of the clones isolated from the high-migration treatment were capable of inhibiting the growth of at least one other clone isolated from the same insect (Table 1). These differences between treatments were highly significant when analysed at the level of clone or cross-test (Table 1) and remained significant when more conservative tests, accounting for the fact that several clones were sampled from the same insect host and experimental line, were performed (Table 1).

Table 1.   Comparisons between migration treatments of the within-host inhibition assays.
 Number showing within-host inhibitionsTotal examined
  1. *A total of 100 Xenorhabdus nematophila colonies were isolated and cross-tested with at least one other clone isolated from the same insect host (χ2 = 14.19; P < 0.001).

  2. †A total of 600 within-host inhibition assays were performed (χ2 = 16.01; P < 0.0001).

  3. ‡Multiple clones were successfully isolated from 15 insect hosts (Fisher’s exact, P < 0.0256).

  4. §Bacteria was successfully isolated from nine of the 10 experimental lines used in the mortality assay (Fisher’s exact, P < 0.0476).

Clones*
 Low migration051
 High migration1249
Cross-tests†
 Low migration0294
 High migration19306
Insect hosts‡
 Low migration08
 High migration47
Experimental lines§
 Low migration05
 High migration34

Despite the lack of inhibition between clones isolated from the same insect host in the low-migration treatment, these clones still possessed the ability to inhibit the growth of other X. nematophila. In fact, when bacteria from one host were cross-tested against bacteria from a host infected with parasites from a different subline (Fig. 1, iv), inhibitions were observed at a level comparable with those found in the high-migration treatment (Table 2).

Table 2.   Comparisons of between-host inhibitions in the low-migration treatment with within-host inhibitions in the high-migration treatment.
 NumberTotal examined
  1. *Number of clones capable of inhibiting at least one other clone (χ2 = 3.52; P = 0.0605).

  2. †Number of clones whose growth was inhibited by at least one other clone (χ2 = 1.81; P = 0.1783).

  3. ‡Number of inhibitions observed (χ2 = 1.72; P = 0.2790).

  4. §Number of lines showing inhibitions (Fisher’s exact; P = 1.000).

Producer clones*
 Low migration (between host)2047
 High migration (within host)1249
Sensitive clones†
 Low migration (between host)847
 High migration (within host)1449
Cross-tests‡
 Low migration (between host)22256
 High migration (within host)19306
Experimental lines§
 Low migration (between host)34
 High migration (within host)34

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Our results confirm the importance of parasite population structure on patterns of virulence. We observed significantly faster rates of host death (higher virulence) in the low-migration lines than in the high-migration lines (Fig. 3). We also observed significantly fewer inhibitions among bacterial clones isolated from the same host in the low-migration lines (Table 1). Hence, our data suggest that low migration led to increased relatedness among bacteria within a host, reducing interference competition among them, thus increasing their ability to overcome the host.

Within the low migration lines, we also tested for inhibition among bacterial clones isolated from different hosts. We found that such inhibitions were as common as the inhibitions observed among clones isolated from the same hosts in the high-migration treatment (Table 2). This result demonstrates that the low-migration treatment did not result in a loss of the ability to produce a growth-inhibiting agent. Rather, it suggests that the low-migration treatment was effective in structuring the population. Thus, bacteria isolated from the same host were immune to each other’s attacks, but they were not immune to attacks of bacteria isolated from other hosts (i.e. different subpopulations) in the same treatment.

Overall, our results suggest that there was variation among bacterial clones at the origin of the experiment, and that this variation became partitioned among hosts differently in the two migration treatments. Most likely it took several experimental passages before the low-migration lines became structured, as no significant differences in host mortality rate were seen across treatments prior to passage 12 (Bashey et al., 2007). Additionally, it is possible that new variation arose during the course of the experiment by mutation or that our treatments resulted in a difference in investment in bacteriocin production (as modelled by Gardner et al., 2004). Unfortunately, we do not currently have the data to address these possibilities. Furthermore, deriving predictions for our migration treatments based on Gardner’s model is difficult as low migration can both increase kinship and decrease the scale of competition (Taylor, 1992a,b; Wilson et al., 1992; West et al., 2002), which would impose conflicting selective pressures on investment in bacteriocin production. Nevertheless, our results unambiguously show that differential migration can affect virulence, and the frequency of within-host inhibitions.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Our results show that parasite migration can affect the outcome of interactions between hosts and parasites, even in the absence of changes to host population densities that were critical in other studies (Kerr et al., 2006; Boots & Mealor, 2007). Further, we provide the first documentation of intraspecific growth inhibition in X. nematophila. We also show that high parasite migration rates can lead to more inhibitions between bacterial clones isolated from the same host and to reduced host mortality rates. Therefore, this study provides support for the view that bacteriocins play an important role in mediating intraspecific interactions; and underscores the importance of parasite population structure to our understanding of virulence.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

We thank C. Fuqua, A. Bhasin, A. Countryman, S. Rangel and C. Reynolds for advice and logistical assistance. This research was supported by NSF DEB-0515832 and by the Indiana METACyt Initiative of Indiana University, funded in part through a major grant from the Lilly Endowment, Inc.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References