C. elegans N2 was purchased from the Caenorhabditis Genetics Center (University of Minnesota, Minneapolis, MN, United States), propagated for two generations using standard culture methods, and then kept frozen in liquid nitrogen (-196°C) until required (Stiernagle, 2006). In preparation, aliquots of the frozen population were thawed and propagated for another two generations on Escherichia coli OP50 (obtained from the Caenorhabditis Genetics Center, University of Minnesota, Minneapolis, MN, United States) for use as a heterogeneous culture. Nematodes were removed from Nematode Growth Medium (NGM) with a 5 mL stream of M9 solution (Stiernagle, 2006) and gentle scraping with a spreader of the agar surface. The runoff containing C. elegans N2 was collected in a sterile 50 mL conical centrifuge tube (Falcon, Fisher Scientific, United Kingdom) and the process of adding M9 solution and gentle scraping of the surface of the NGM was repeated twice. The nematodes were left to settle in the centrifuge tubes for 10 min and then 200 μL of the solution was pipetted from the bottom of the tube and added to 30 mL M9 solution in a second 50 mL centrifuge tube. The washing of C. elegans N2 by leaving them to settle in the conical centrifuge tube and transfer of 200 μL into new solution was repeated at least three times, before C. elegans N2 density was established using a Nematode Counting Chamber (Chalex Corporation, Wallowa, OR, United States). Fresh M9 solution was added or removed to reach a total of 1100 to 1450 nematodes per mL.

For the laboratory trials, avirulent B. anthracis Sterne 34F2 expressing the pSW4-GFPmut1 plasmid was used as a surrogate for fully pathogenic B. anthracis (see Sastalla et al., 2009). GFP expression is driven by the pagA promoter which is highly expressed during vegetative growth. Spores from this strain trap GFP produced during vegetative growth into the body of the spore and are fluorescent for at least 14 days post sporulation (Sastalla et al., 2009). As such, spore stocks for relevant experiments in the current study were at least 2 weeks old.

Bacterial cultures were maintained as frozen spore stock or as chilled working stock. Spores were produced in triplicate using the quantitative three-step method described by Tomasino et al. (2008). For standard maintenance, bacteria were grown on TSA for a minimum of 18 h at 37 ± 1°C and for bacterial enumeration in soil during experiments either Luria Bertani (LB) agar (Fisher Scientific, United Kingdom) or both TSA and PLET agar (both Nanologix, United States) were used. In brief, individual bacterial colonies from a working stock slope were inoculated into 10 mL of nutrient broth and incubated for 24 ± 2 h at 37 ± 1°C in a shaking incubator. The following day, 500 μL of culture was inoculated onto nutrient agar plates containing 5 μg/mL manganese sulfate monohydrate and spread evenly across the plate. Spore cultures were then incubated for 12–14 days at 37 ± 1°C before bacterial spores were harvested by adding 10 mL of 2–5°C sterile distilled water to the agar plates. Sterile spreaders were used to remove growth from plates and the resulting suspension was pipetted into 50 mL conical centrifuge tubes before spores were washed three times in sterile distilled water by centrifugation at 5000 rpm for 10 min at room temperature and stored at 2–5°C. Spore starting concentration for experiments in solution was ∼1 × 10⁶ CFU mL^-1 or a 10-fold dilution thereof, which was achieved by diluting the spore stock with an appropriate volume of sterile distilled water. For soil microcosm experiments, the spore concentration was ∼1 × 10⁶ CFU g^-1 in soil (or a 10-fold dilution thereof). Due to the large amount of time and number of consumables required for the protocols experiments were not repeated; instead, a nested approach was used in which three different spore stocks with inoculates from the same master bacterial culture stock were produced for the three replicates within one treatment group (for each day) to ensure reproducibility within the experiment. For microcosm studies in Turkey, soil collected at animal burial sites and containing virulent B. anthracis was spiked with additional B. anthracis Sterne 34F2 spores to bring the level of bacteria in the soil up to ∼10³ CFU g^-1 in Microcosm Experiment 3 (to ensure detectable levels of CFUs at least in the control treatment) and up to ∼10⁶ CFU g^-1 in Microcosm Experiment 4 (to ensure comparability with Microcosm Experiment 5).

All reagents and chemicals were obtained from Fisher Scientific (Loughborough, United Kingdom) or Sigma–Aldrich (Dorset, United Kingdom) unless otherwise stated in the text. Soil was collected in a forest near Gwaelod-Y-Garth, Wales, United Kingdom (51° 32′ 19.91″N, 3° 16′ 9.55″W) for Microcosm Experiments 1, 2, and 5. Soil used for experiments with fully virulent B. anthracis was collected from field sites A, Turkey (Microcosm Experiment 3) and D (Microcosm Experiment 4; coordinates of Turkish field sites are not provided for security reasons). Both Turkish field sites are known burial sites of anthrax positive animals and usually contain ∼10¹–10³ CFU g^-1.

Soil was collected fresh for each experiment, meaning that soil moisture content differed between experiments. Moisture content was measured by drying a portion of soil for 24 h at 60°C and comparing the overall weight before and after the drying period. The Welsh soil was comprised of free draining acid loam over rock (pH = 5.5–6.0); in contrast the Turkish soil was taken from a region rich in calcium due to the presence of limestone, with a pH > 6.0 and which is subjected to an annual cycle of local flooding due to seasonal snow melt, factors which have all been linked to the long term survival of virulent B. anthracis¹ (Hugh-Jones and Blackburn, 2009).

Caenorhabditis elegans N2 was added to GFP-expressing B. anthracis Sterne 34F2 lawns on NGM and the population was left to develop for 3–4 days. The nematodes were then harvested as described above, immobilized for visualization using CytoFix fixation buffer (BD Cat. No. 554655, San Jose, CA, United States) and visualized under an IN Cell Analyzer 2000 (General Electric, Marlborough, MA, United States) at 395 nm from 10 to 100× magnification.

Early observations of C. elegans N2 maintained on B. anthracis Sterne 34F2 indicated that populations developed better on their standard laboratory food source of E. coli OP50. This was further investigated experimentally by inoculating NGM with 50 μL B. anthracis Sterne 34F2 (n = 5) or E. coli OP50 (n = 5) overnight liquid culture. Plates were incubated overnight at 37 ± 1°C before five adult, hermaphroditic C. elegans N2 were added to the NGM. Plates were incubated at 37 ± 1°C for 9 days to allow the C. elegans N2 to develop, which were then recovered from NGM as described above. C. elegans N2 suspensions were adjusted to a total volume 50 mL with M9 solution. The suspension was injected into a Nematode Counting Chamber in 1 mL aliquots, and then frozen for 10 min at -20°C to immobilize the nematodes before counting.

An overview of soil microcosm experiments, their aims and how these build on each other is shown in Figure 1. Soil microcosms were prepared with 3 ± 0.1 g of non-sterile soil in 50 mL Nunc^® EZ Flip^TM conical centrifuge tubes. For Microcosm Experiment 1 (Recovery of B. anthracis Sterne 34F2 from autoclaved and artificially contaminated soil) and 2 (Recovery of B. anthracis Sterne 34F2 from artificially contaminated soil), soil samples were collected near at the Welsh field site with an overall moisture content of >90%.

Moisture content of site A soil used in Microcosm Experiment 3 (Effect of germinant treatment on cultivable soil bacteria in Turkish soil) was not determined as appropriate security measures to dry soil were not in place at the time of the experiment in the Turkish laboratories. Initial moisture content was 38.5% in site D soil (Microcosm Experiment 4: Effect of combined germinant and nematode treatment on B. anthracis spore numbers in contaminated soil from an animal burial site in Turkey) and 39.4% for Welsh soil (Microcosm Experiment 5: Effect of combined germinant and nematode treatment on B. anthracis Sterne 34F2 numbers in artificially contaminated soil from a Welsh field site).

At each time point (30 min after spiking for Microcosm Experiments 1 and 2; Days 1, 2, 3, 4, and 6 for Experiment 3; Days 0, 1, 2, 3, 4, and 8 for Experiments 4 and 5), all microcosms were extracted using the following method: 10 mL elution buffer (PBS, 28 mL L^-1 Tween 80, 0.15 g L^-1 L-Cysteine), formulated to promote spore recovery and minimize the carryover of antibacterial factors in the soil such as polyphenols (Beyer et al., 1995; BS EN 1276, 2009; Bishop, 2014; BS EN 13697, 2015), was added to each microcosm, which was then manually agitated until homogeneous, and two 1 mL aliquots were taken from each tube. One aliquot was used to determine the total viable cell count (non-heat-shock), and the other to determine the spore load (heat-shocked at 65°C for 1 h). Viable counts were obtained by 10-fold serial dilution in PBS and spread on LB agar plates (Fisher Scientific, United Kingdom; in duplicate, Microcosm Experiment 3) or TSA and PLET (in triplicate, Microcosm Experiments 1, 2, 4, and 5). LB and TSA plates were incubated overnight at 37°C, whereas PLET plates were incubated for two nights.

Observation and accurate recording of B. anthracis colonies for Microcosm Experiments 1, 2, 4, and 5 was simplified by adopting two types of ‘morphology controls’: one for pure B. anthracis Sterne 34F2 and a second for the endogenous soil microflora (ESMF). Both morphology controls were prepared on both PLET and TSA. The first was a streak plate of pure B. anthracis Sterne 34F2; the second was a series of 10-fold dilutions from the soil solution of non-spiked soil microcosms (undiluted to 10^-4). These morphology controls provided a reference for visual identification of B. anthracis Sterne 34F2 colonies amongst ESMF, although such an approach will of course not completely negate the possibility of counting false positives particularly at the lower levels of detection when the ratio of B. anthracis to ESMF favors the latter and the ESMF presents a strong confounder (see also Tables 1, 2).

The soil samples were separated into two aliquots, one of which was sterilized by autoclaving for 2 h at 121°C (for Microcosm Experiment 1). Soil was then divided among the microcosms and spiked with 10-fold increases in spore concentrations (10¹–10⁶ spores g^-1) at 50 μL and manually shaken to disperse spores. The low spiking volumes of spore suspension was chosen due to high soil moisture content (>90%) after soil collection as otherwise the soil would have been completely saturated with water.

Germinants in a well-defined mixture known to be highly effective for inducing germination (Ireland and Hanna, 2002) were added in a multi-step treatment at Day 0 (300 mM L-alanine, 15 mM inosine) and Day +2 (500 mM L-alanine, 25 mM inosine) to soil microcosms (∼10²–10³ naturally occurring B. anthracis spores g^-1). The multi-step treatment aimed to keep germination sustained to ensure vegetative bacteria were available for nematode ingestion. A total volume of 2 mL liquid was introduced into each microcosm. Specifically, an extra 0.5 mL of B. anthracis Sterne 34F2 suspension (∼1 × 10³ CFU mL^-1) was added to each microcosm and mixed to disperse spores evenly throughout the matrix to ensure a detectable level of bacteria at least in the control treatment. Either germinant (1 mL) or PBS (of variable volume depending on how much moisture had already been added with the germinant or spore suspension) was then added (Day 0) and the microcosms were mixed. At Day +2, an additional 100 μL of PBS (control) or germinant was added. Soil microcosms were in sealed 50 mL conical tubes to prevent evaporation, and stored at ambient temperature (22.0 ± 2.0°C).

Microcosms were spiked with B. anthracis Sterne 34F2 at approximately 1 × 10⁶ spores g^-1 (Microcosm Experiment 4 – Site D soil: ∼3.5 × 10⁵ CFU g^-1 naturally occurring virulent B. anthracis spores which was unusually high compared to other sampling years and other known contaminated B. anthracis spores sites) and germinant solution was added in a multi-step treatment at Day 0 (300 mM L-alanine, 15 mM inosine) and Day +2 (500 mM L-alanine, 25 mM inosine). A heterogeneous nematode population consisting of all age ranges of C. elegans N2 suspended in either the germinant solution or PBS on Day 0 was added at a total volume of 1100–1400 nematodes mL^-1. A total volume of 600 μL liquid was added to each microcosm. Specifically, 100 μL of spore suspension (∼3 × 10⁷ CFU mL^-1) was added to each microcosm and mixed to disperse spores throughout the matrix. Either 330 μL of PBS (control), germinant, or germinant-nematode suspension was then added (Day 0) and the microcosms mixed. At Day +2, an additional 170 μL of PBS (control) or germinant was added. Soil microcosms were sealed in conical tubes to prevent evaporation, and stored at ambient temperature (Turkish laboratory: 22.0 ± 2.0°C; Welsh laboratory: 20.8 ± 1.6°C).

Data of viable counts for Microcosm Experiments 3, 4, and 5 were log₁₀ (+1) transformed and analyzed as dependent variables using generalized linear models (GLM). For Microcosm Experiments 3 and 5, model fit was best with a Gaussian family and identity link function. For Microcosm Experiment 4, a quasipoisson family and an identity link function were used. Due to a higher number of native microorganisms in Turkish soil, the detection limit for B. anthracis was at 10⁵ CFU g^-1 soil and highly variable, as opposed to a conservative 10² CFU g^-1 in Welsh soil. The higher limit of detection resulted in more noise in the data set, which could not be normalized using standard approaches such as a further transformation of data and adjustments with error and link functions within the GLM. Because results equal to zero were likely to be artifacts (i.e., we could not identify B. anthracis below the lower limit of detection due to high ESMF), these data points were removed for statistical analysis of the results of Microcosm Experiments 3 and 5, resulting in completely normally distributed data.

For GLM on cultivable microorganisms on Turkish soil (Microcosm Experiment 3), treatment (control, and germinants only), day and viable count type (total or spore viable count) were used as independent variables with interactions terms between treatment: day, treatment: viable count type and day: viable count type. For both GLMs on Turkish and Welsh decontamination datasets (Microcosm Experiments 4 and 5), treatment (control, germinants only, nematodes only, and a combination of germinants and nematodes), day, viable count type (total or spore viable count), and media (PLET, TSA) were treated as independent variables with interactions terms as in Experiment 3. A step-wise model reduction process was applied and standardized residuals from each model were first checked visually for normality and homogeneity of variance using a histogram, Q–Q plots, and fitted values. All analyses were performed in R 3.1.1 statistical package (R Development Core Team, 2014).

Caenorhabditis elegans N2 cultures provided with GFP-tagged B. anthracis Sterne 34F2 clearly show that the nematodes ingest both vegetative bacteria and spores (Figures 2, 3). The images indicate that while spores pass through the gut intact, the vegetative bacteria are ingested confirming our initial hypothesis that C. elegans N2 can digest and neutralize the actively growing bacterium, but not the spore form of B. anthracis. Nematodes were visually observed to become lethargic and bloated after feeding on B. anthracis for an extended period. However, in terms of population development, no long-term negative effects were observed among nematodes that ingested the human pathogen: the nematodes survived on B. anthracis Sterne 34F2 cultures equally well as those feed E. coli OP50, their standard laboratory food source (Figure 4). Therefore, to decontaminate B. anthracis-contaminated soil, the use of nematodes is feasible if the bacteria are present in their vegetative form and not as a spore.

First, the recovery of B. anthracis Sterne 34F2 from autoclaved soil microcosms using selective (PLET) and non-selective (TSA) media was assessed. The lower limit of detection was 40 to 50 colony forming units (CFU) g^-1 (Table 1); below this level, experimental variation caused results to be inconsistent. Although the number of colonies on PLET agar were slightly lower (40) than those seen using TSA (50), they were still within the same logarithmic range, suggesting that both media are equally effective for recovering B. anthracis from autoclaved soil.

Following the recovery of B. anthracis Sterne 34F2 from autoclaved soil, the effect of the ESMF on the ability to recover B. anthracis Sterne 34F2 was determined. Various concentrations of B. anthracis Sterne 34F2 were added to the non-sterile soil collected from Garth Mountain, southeast Wales, to test recovery success. B. anthracis Sterne 34F2 colonies were detected on TSA in samples taken from microcosms spiked with 10¹ B. anthracis spores g^-1 soil (Table 2); however, for an accurate determination of CFU g^-1 soil in non-sterile microcosms, a minimum of 10⁴ CFU g^-1 soil is desirable to ensure repeatability of results. The higher starting load has the advantage of reducing the overall background of ESMF from sample cultures (Table 2). Hence, to determine whether a combined germinant and nematode treatment could effectively reduce the B. anthracis spore load in contaminated soil, it was essential to work with detectable levels of B. anthracis in further trials to ensure repeatability across experiments.

Our initial decontamination experiment assessed the impact of treatment with L-alanine and inosine as germinants on the total number of viable bacteria recovered from Turkish soil naturally contaminated with low levels of B. anthracis (∼10³ B. anthracis spores g^-1). Exposure to the L-alanine and inosine germinant mixture on Day 0 and Day 2 of the experiment resulted in a 3.5 log increase in the soil microflora cultivable on Luria Bertani (LB) agar over the time course of the experiment (Figure 5B). In contrast, only a 0.5 log increase was observed after treatment with phosphate buffered saline (PBS) solution (Figure 5A; Generalized Linear Model [GLM from hereon]: total/spore recovery^∗treatment interaction: F_1,212 = 11.00, P = 0.001; GLM: treatment^∗day interaction: F_4,212 = 171.89, P < 0.001; GLM: total/spore recovery^∗day interaction: F_4,212 = 10.15, P < 0.001). Unfortunately, the low levels of B. anthracis relative to the ESMF in the soil meant that the increase of the target bacterium due to the germination and replication of the pathogen could not be determined.

Due to the low numbers of B. anthracis spores present in contaminated soil from the animal burial site (∼3.5 × 10⁵ CFU g^-1), we artificially increased the B. anthracis spore load to a final concentration of ∼10⁶ CFU g^-1. As expected, the number of bacteria in soil treated with PBS did not change significantly over the course of the study (Figure 6A). In contrast, soil treated with germinants demonstrated 0.6 log germination by Day 1 and a 3 log reduction in total viable count by the end of the trial (Figure 6C). These results suggest that, rather than replicating like the rest of the ESMF (Figure 6C), B. anthracis numbers actually declined when treated with germinants.

To determine if we could further reduce the number of viable B. anthracis in artificially contaminated soil, separate microcosms were treated with a combination of germinants and C. elegans N2. While treatment with nematodes alone had no effect on B. anthracis numbers (Figure 6B) a combination of germinants and nematodes led to a 3.5 log reduction by Day 8 which differed significantly from other treatments (Figure 6D; GLM: treatment^∗day interaction: F_18,298 = 15.16, P < 0.001). Total and spore recovery from microcosms differed also across days (Figures 6C,D; GLM: total/spore recovery^∗day interaction: F_6,263 = 4.85, P < 0.001). There was no significant difference in recovery between TSA and PLET agar types (GLM: P > 0.05). For the germinant and nematode treatment, viable counts on Day 3 could not be determined because of high levels of ESMF.

To determine if soil composition had an effect on the efficiency of germination and the ability of the nematodes to consume their food source, we repeated the same experiment as in Microcosm Experiment 4 using soil collected from a site on Garth Mountain in southeast Wales that had no recorded history of B. anthracis spore contamination. The results in this soil (Figure 7) were similar to results seen in the Turkish soil (Figure 6).

Spore numbers significantly decreased with application of L-alanine and inosine plus nematodes. While treatment with nematodes alone had no effect on B. anthracis numbers (Figure 7B), a 4.6 log reduction was observed when nematodes were used in combination with germinants (Figure 7D) which was significantly different from that achieved with germinants alone over the duration of the experiment (3.5 log; GLM: treatment^∗day interaction: F_18,298 = 15.16, P < 0.001). The total number of viable bacteria and spores recovered was dependent on treatment and differed over time (Figure 7; GLM: total/spore recovery^∗day interaction: F_6,298 = 27.51, P < 0.001; GLM: total/spore recovery^∗treatment interaction: F_3,298 = 6.94, P < 0.001).

During the last few days of the experiments for both Welsh and Turkish soil, enumeration of B. anthracis colonies was considerably more challenging from microcosms which had been treated with germinants. As B. anthracis colonies declined and the limit of detection was reached, the number of ESFM colonies stayed the same or increased (consistent with results from Microcosm Experiment 3). Particularly, samples that were not autoclaved to determine the overall B. anthracis colonies had to be diluted further than autoclaved samples (for spore enumeration) so increasing the error in the collected data and increasing the likelihood of false negatives. The different agar types (PLET and TSA) did not affect recovery (GLM: P > 0.05).