Elevated resource availability sufficient to turn opportunistic into virulent fish pathogens

Authors

  • C. Wedekind,

    Corresponding author
    1. Department of Ecology and Evolution, Biophore, University of Lausanne, 1015 Lausanne, Switzerland
    2. Department of Aquatic Ecology, Eawag: Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland
    Search for more papers by this author
  • M. O. Gessner,

    1. Department of Aquatic Ecology, Eawag: Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland
    2. Institute of Integrative Biology (IBZ), ETH Zurich, 8600 Zurich, Switzerland
    Search for more papers by this author
  • F. Vazquez,

    1. Department of Surface Waters, Eawag: Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland
    Search for more papers by this author
  • M. Maerki,

    1. Department of Surface Waters, Eawag: Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland
    Search for more papers by this author
    •  Present address: Kantonale Verwaltung, Abteilung für Umwelt, 5001 Aarau, Switzerland.

  • D. Steiner

    1. Department of Aquatic Ecology, Eawag: Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland
    Search for more papers by this author

  • Corresponding Editor: D. M. Tompkins.

Abstract

There is mounting evidence that organic or inorganic enrichment of aquatic environments increases the risk of infectious diseases, with disease agents ranging from helminth parasites to fungal, bacterial, and viral pathogens. The causal link between microbial resource availability and disease risk is thought to be complex and, in the case of so-called “opportunistic pathogens,” to involve additional stressors that weaken host resistance (e.g., temperature shifts or oxygen deficiencies). In contrast to this perception, our experiment shows that the link between resource levels and infection of fish embryos can be very direct: increased resource availability can transform benign microbial communities into virulent ones. We find that embryos can be harmed before further stresses (e.g., oxygen depletion) weaken them, and treatment with antibiotics and fungicides cancels the detrimental effects. The changed characteristics of symbiotic microbial communities could simply reflect density-dependent relationships or be due to a transition in life-history strategy. Our findings demonstrate that simple microhabitat changes can be sufficient to turn “opportunistic” into virulent pathogens.

Introduction

The availability of resources shapes communities and ecosystems (Elser et al. 2000, Smith and Schindler 2009) including host–pathogen systems (Johnson et al. 2007, Hall et al. 2009). Although the link between resource availability and disease is not well understood, organic or inorganic enrichment of aquatic environments commonly leads to increased risk of infection. This trend can be observed in various kinds of pathogenic agents, including helminth, myxozoan, bacterial, fungal, and viral pathogens (McKenzie and Townsend 2007). The effects of increased resource availabilities are often complex and may involve responses in intermediate or vector hosts, often in combination with stressors that weaken host resistance to infection (Dror et al. 2006, Bosch et al. 2007, Flemming et al. 2007, Rohr et al. 2008b, Sahu et al. 2008, Karraker and Ruthig 2009). However, the importance of inherent pathogen characteristics and host genetics relative to host condition is rarely clear (Flemming et al. 2007, Sahoo et al. 2008, Takahashi et al. 2008), and recent findings challenge some of the previously assumed simplifications about microbial infections and host condition (Rohr et al. 2008a) or about the social dynamics within pathogen communities (Köhler et al. 2009).

Pathogens are ubiquitous, even if no primary infection symptoms can be observed. Animals and plants are associated with symbiotic microbial communities that typically outnumber host cells and whose roles range from mutualism to commensalism to parasitism (Rosenberg et al. 2007). Microbes that are typically benign but have the potential to become virulent to immunologically compromised hosts are called “opportunistic pathogens” (Baquero et al. 2008). Thermal or oxygen stress, high host density, or mechanical injuries are among the possible stressors that affect host susceptibility to such infectious diseases. In exceptional cases, acute stress may induce a general immune response that reduces the prevalence of certain pathogens (Wedekind and Little 2004). However, the typical pattern is that chronic or acute stress increases host susceptibility (Aeby and Santavy 2006, Bruno et al. 2007), especially so in the case of “opportunistic” microbial pathogens (Lesser et al. 2007).

Here we examine the interaction between microbial communities of fish embryos with their host. We chose the whitefish Coregonus suitieri, i.e., a salmonid that is an ecological keystone species in lake ecosystems and, at the embryonic and larval stages, a well-established laboratory model. The primary causes of embryo mortality both in the wild and in hatcheries are often unclear, but microbial pathogens appear to play a major role (Wedekind et al. 2001). Accordingly, whitefish embryos show adaptations such as precocious hatching to avoid infection (Wedekind 2002) and early immunological defense (von Siebenthal et al. 2009), which can be linked to allelic diversity on the major histocompatibility complex (Wedekind et al. 2004). Here we test for the effects of resource availability on symbiotic microbial communities and their virulence to their fish host.

Methods

Large-type lake whitefish (C. suitieri) were caught from their spawning sites in Lake Lucerne (Switzerland) during the spawning season in December. Their gametes were used for in vitro fertilization, and the eggs were raised in a constant flow of lake water (60 L/h) at 8.5°C. A total of 1080 healthy-looking embryonated eggs were taken at random 54 days after fertilization, washed (von Siebenthal et al. 2009), distributed to 2-mL wells of 24-well plates, and incubated in a climate chamber at 6°C. Five eggs were used per well, resulting in 216 independent experimental units. The water used for washing the eggs and for incubations was lake water that had been passed over a sand filter, autoclaved, cooled to ambient temperature, and aerated before use. We added to each well 100 μL of a growth medium (5 g peptone [Difco, Lawrence, Kansas, USA] and 3 g meat extract [Fluka, Basel, Switzerland]) dissolved in 1000 mL distilled water, pH adjusted to 7.0). This medium was diluted in sterile lake water to obtain four different concentrations (0, 0.001, 0.01, 0.1) before adding it to the wells either alone or in combination with one of two mixtures of antimicrobial compounds administered at either high or low concentration. “Mix A high” contained penicillin, streptomycin, and amphotericin B (100×; catalogue number 15240-062; GIBCO, Carlsbad, California, USA); “mix A low” was a 10-fold diluted solution of “mix A high”; “mix B high” was composed of 10 mg/L chloramphenicol + 300 mg/L ampicillin + 20 mg/L rifampicin; and “mix B low” was a 10-fold diluted solution of “mix B high.”

An Ox 25 oxygen microelectrode (Unisense Ltd., Aarhus, Denmark) fixed to a custom-made micromanipulator with a tip diameter <100 μm was used to measure oxygen concentrations in the medium at a distance of 5 mm to the egg membrane. These measurements were taken daily in about half of the replicate wells that had received antimicrobial compounds and in 24 of the non-treated replicates. The calibrated range of the electrode signal at 6°C was −15.9–8.1 mV, corresponding to oxygen concentrations of 0–11.2 mg/L as determined by the Winkler method.

Bacterial growth was determined periodically during the course of the experiment in replicate wells not treated with antimicrobial compounds. Embryos were killed by an overdose of tricaine methanesulfonate (MS-222). Eggs and water were then separated by centrifugation and stored separately for a few days at −20°C in 1.5-mL tubes with 1 mL of 50% ethanol/PBS (phosphate buffered saline, 130 mmol/L NaCl, 7 mmol/L Na2HPO4, 3 mmol/L NaH2PO4, pH 7.2–7.4) per 1–5 eggs or 0.1 mL incubation water, respectively. For bacterial counts, the samples were centrifuged at 5000g for 10 minutes and the supernatant discarded. Eggs were mixed on a vortex mixer for 30 s with 1 mL 0.1% sodium pyrophosphate, pH 7.2. Following the methods of Hahn et al. (1992), 10 μL of the pyrophosphate extract were applied to gelatine-coated (0.1% gelatine, 0.01% KCr(SO4)2) 8-well slides (Eri Scientific, Portsmouth, New Hampshire, USA; 8 mm well diameter) and allowed to air dry. After dehydration in 50, 80 and 100% ethanol for 3 minutes, 10 μL of 0.0001% DAPI solution was added to each sample and incubated for 5–10 minutes at room temperature, rinsed with distilled water, air dried, mounted with Citifluor solution (Citifluor, Canterbury, UK), and examined at 1000× magnification with a Zeiss Axioskop 2 microscope fitted for epifluorescence with a high-pressure mercury bulb (100W) and filter set 02 (G365, FT395, LP420; Zeiss, Feldbach, Switzerland). Twenty-five randomly selected fields (0.015 mm2 each) were examined per sample.

Since infected eggs appear increasingly cloudy and opaque in transmitted light, we photographed all eggs daily under standardized conditions and measured their optical density using the public domain program ImageJ (available online).7 This optical density was used as a potential additional measure of microbial density.

Embryo mortality and hatching rates were determined daily. Embryos were considered dead if no blood flow could be observed under a bright-field microscope. Hatched larvae and hatching debris (i.e., egg membranes) were immediately removed from the wells. At day 3 after the start of the experiment, all remaining eggs were stored at −20°C for later analysis of fungal biomass. Fungal biomass was measured as ergosterol, a membrane constituent essentially restricted to fungi and absent from oomycetous protists (Gessner and Newell 2002). The eggs were freeze dried, weighed to the nearest 0.01 mg (average sample dry mass = 7.0 mg), and placed in 10 mL of 0.8% KOH in methanol. Lipids were extracted from the samples for 30 minutes in tightly capped pressure-resistant tubes with constant stirring and temperature maintained at 80°C. The primary extract was cleaned by solid-phase extraction (Gessner and Schmitt 1996) and the volume of the eluate reduced to ≤200 μL to maximize sensitivity of the assay. Ergosterol was purified and quantified by high-pressure liquid chromatography (HPLC). A LichroSpher 100 RP-18 column (0.46 × 25 cm; Merck, Zug, Switzerland) connected to a Jasco AS-950 autosampler and PU-980 liquid chromatograph was used (Jasco, Gross-Umstadt, Germany). Sample volumes of 100 μL were injected. The mobile phase was methanol at a flow rate of 1.5 mL/min, with a column temperature of 33°C. Ergosterol eluted after 8 minutes under these conditions. It was detected with a Jasco diode array detector (MD 2010 plus) and quantified based on the area of absorbance peaks at 282 nm. A newly purchased batch of ergosterol from Fluka (>98% purity) was used as the standard. Ergosterol values were converted to fungal biomass by assuming a concentration of 5.5 μg per mg of fungal dry mass (Gessner and Chauvet 1993).

Individual wells of cell-culture plates (each with five eggs at the start) were used as independent replicates for statistical analyses. For example, we used the average optical density of up to five eggs per well or the average bacterial number per egg in a well for statistical analyses. The two factors resource level and antimicrobial mixtures were administered in a full-factorial design to allow for two-way ANOVAs. The experiment started with six replicates per experimental cell with antimicrobials and with 30 replicates per cell without antimicrobials; the replicate number was successively reduced in the latter case by daily sampling for bacterial density determinations. Effects of time were tested by repeated-measures ANOVAs (H-F corrected), or with time as a grouping factor in a two-way ANOVA for its effect on bacterial density (because the daily sampling was destructive; the other grouping factor was resource level). Hatching rates were square-root arcsine transformed before statistical testing. Graphical inspection of the other residuals suggested that the model assumptions for parametric statistics were not notably violated. All reported p-values refer to two-tailed tests. Analyses were performed in JMP 7.0 (SAS Institute 2007).

Results and Discussion

During the course of the experiment, oxygen concentration gradually declined in wells without antimicrobial compounds but remained high throughout in wells that received either antimicrobial mix A or B (Fig. 1a; effects of time, F3, 144 = 3.5, P = 0.02; resource concentration, F3,48 = 0.4, P = 0.77; antimicrobial treatment with high and low concentrations combined, F2,48 = 8.3, P < 0.001; time × treatment, F6, 144 = 16.8, P < 0.0001; all other interactions, F < 1.6, P > 0.12). In parallel, the abundance of bacteria, both suspended in the well water and associated with eggs, increased over time (Fig. 1b; in water, F2,57 = 5.0, P = 0.01; associated with eggs, F2,57 = 5.8, P = 0.006). Because large-type whitefish eggs have an average diameter of 2.7 mm (SD = 0.06; Wedekind and Müller 2005), corresponding to a volume of 0.01 mL, the bacterial density was at all times much higher in the eggs than in the well water (Fig. 1b).

Figure 1.

Changes in oxygen conditions, microbial growth, and fish hatching over the course of the experiment. (a) Average oxygen concentration in replicate wells treated with antimicrobial mixtures A or B (stars and pointed line) and non-treated wells receiving different levels of microbial resources (0.2 ≤ SE ≤ 0.5); (b) bacterial abundance per embryonated egg (solid symbols; average egg volume ≈ 0.01 mL) and incubation medium (open symbols; reported per 0.1 mL to facilitate comparison with bacterial colonization of eggs), for replicates not treated with antimicrobial mixtures (square symbols; means ± SE) and those treated with antimicrobial solution B at high concentration (stars); (c) average optical density (as measured in ImageJ on a relative scale from 0 [white] to 255 [black]; see inset) of eggs not treated or treated with either antimicrobial solution A or B at either low or high concentration (0.6 < SE < 2.8; the inset shows some embryonated eggs of different optical densities); and (d) average hatching rate in wells that were not treated or treated with microbial mix A or B at high or low concentration (0.01 < SE < 0.04).

The constantly high oxygen concentrations in wells receiving antimicrobial compounds (Fig. 1a) was in accordance with greatly reduced bacterial counts three days after the start of the experiment (Fig. 1b; in water and attached to eggs, F1,30 > 98.6, P < 0.0001). The average optical density of embryonated eggs was also reduced (Fig. 1c) and turned out to be a useful proxy of bacterial counts (correlation between number of bacteria per egg and optical density, r = 0.54, N = 80, P < 0.0001). However, comparison of Fig. 1b, c suggests that direct bacterial counts produced more pronounced differences than measurements of optical density. Hatching rates were higher in the wells containing antimicrobial compounds than in non-treated wells (Fig. 1d; effect of time, F2, 170 = 17.4; antimicrobial treatment, F4, 171 = 13.1; interaction term, F8, 340 = 5.8, P always < 0.0001), indicating that the negative effect of elevated resource supply for microbes on larval hatching success was at least partly canceled by the antimicrobial treatments. However, it is not clear from these results whether the negative effect of elevated resource levels was directly caused by microbial growth (Fig. 1b, c), or by oxygen depletion as a consequence of enhanced microbial growth (Fig. 1a), or by both.

To further examine whether resource availability directly affected embryo survival, we focus on the first 24 hours of the experiment (shaded area in Fig. 1a). During these first 24 hours, oxygen concentrations did not significantly decline in wells without antimicrobial compounds (effects of time, F1,20 = 0.001, P = 0.98; resource concentration, F3,20 = 0.5, P = 0.69; interaction term, F3,20 = 0.5, P = 0.69). However, hatching rates after this first day were already affected by treatment with antimicrobial mix B (Fig. 1d; F4, 211 = 4.6, P = 0.002). Moreover, embryo mortality during that time was significantly affected by high resource concentration when microbial growth was unrestricted (Fig. 2; two-way ANOVA; effect of resource concentration, F3,80 = 5.4, P = 0.002; effect of antimicrobial treatment, F2,81 = 11.7, P = 0.04; interaction term, F6,77 = 2.6, P = 0.025). Resource concentration had no significant effect on embryo mortality if eggs were treated with either of two antimicrobial mixtures, i.e., if the non-treated replicates were excluded from the model (effect of resource concentration, F3,44 = 0.7, P = 0.57; Fig. 2b). This demonstrates that elevated resources levels for microbes did not directly affect embryo survival during this initial period.

Figure 2.

Effects of resource enrichment level on embryo mortality (mean + SE) during the first 24 hours in replicates that received (a) no antimicrobial treatment and (b) high doses of antimicrobial mixture A (left histograms in each pair) or B (right histograms in each pair).

Microbial growth on embryonated eggs was directly related to the availability of resources (Fig. 3). Bacterial density increased with increasing concentrations of resources for growth (F3,56 = 3.8, P = 0.015) and so did concentrations of ergosterol, a measure of fungal biomass (F3,55 = 4.9, P = 0.004). The microbial biomass at the start of the experiment did not seem to influence its outcome, as the optical density of eggs at day 0 did not significantly correlate with later bacterial counts in the untreated wells (r = −0.09, N = 24, P = 0.68).

Figure 3.

Effects of resource enrichment level on ergosterol as a measure of fungal biomass (histograms) and bacterial counts (per mL of incubation water) after 72 hours of incubation without antimicrobial compounds. Means ± SE (N = 14–16 for each nutrient concentration and measurement type combination).

Resources availability in ecosystems varies over time and space and is often linked to environmental stresses such as pollution (Smith and Schindler 2009). Sewage effluents, for example, contain high loads of labile organic compounds. In this and other situations, microbial growth in aquatic environments can severely reduce local oxygen levels (Even et al. 2004) and may also contribute to clogging of sediments (Battin and Sengschmitt 1999). These factors can be extremely stressful to exposed whitefish embryos (Ventling-Schwank and Müller 1991), but they did not play a role during at least the first 24 hours of our experiment. Oxygen concentrations dropped only after elevated resource levels had allowed microbes to grow already to a degree that directly affected the embryos, leading to increased mortality. Moreover, average oxygen concentrations in our experiment never reached levels known to impede egg or embryo survival of whitefish (Wedekind and Müller 2004). Other possible stressors related to the elevated resource levels for microbes clearly did not play a significant role in influencing our results, either, given that addition of antimicrobial compounds cancelled any apparent effects of elevated resource concentration on embryo mortality. The latter result was found although our antimicrobial treatments were not 100% effective at preventing microbial growth.

Microhabitat changes may frequently affect important characteristics of symbiotic microbial communities. The effects we observed could be linked to density-dependent relationships or to a transition in microbial life-history strategies. Resource enrichment is likely to change not only growth patterns but also the social environment of microbes. Many microbes are known to be phenotypically plastic with respect to such changes (Diggle et al. 2007, Brown and Buckling 2008, Kümmerli et al. 2009). One possible mechanism is quorum sensing, i.e., the accumulation of signaling molecules that enable bacteria to sense cell density. Quorum sensing has been shown to affect gene expression and virulence of, for example, the “opportunistic” pathogen Pseudomonas aeruginosa (Köhler et al. 2009). Such virulence effects may be enhanced on immune-compromised hosts. However, our results demonstrate that changes in the ecological environment of microbes, and possibly linked changes in their social environment, can be sufficient to turn benign microbe communities into virulent ones.

Acknowledgments

We thank M. Bia, J. Muggli, R. Müller, and K. Zepp-Falz for assistance or discussion, two anonymous reviewers for constructive comments, and the Swiss National Science Foundation for financial support.

Ancillary