Identification of environmental bacteriophages specific for B. anthracis

For this study, we used two distinct methods to collect eight novel bacteriophages that specifically infect B. anthracis (and not closely related B. cereus strains). For one method, we adapted a standard screening protocol to directly identify plaque-forming units (PFUs) or “free” phage particles within >150 different environmental extracts from an array of soil, fresh water and marine samples, as well as herbivore feces, rhizosphere and phylloplane washings, decaying organic matter, and soil invertebrate intestines. The B. anthracis laboratory strain ΔSterne was used as the susceptible host in these experiments because it lacks the inducible prophages that reside in many B. anthracis strains which could hamper our identification procedure. Ultimately, we identified PFUs from only three environments, including the gut of the earthworm Eisenia fetida (yielding Worm intestinal phages, Wip1 and Wip2), fern root systems (yielding the Fern rhizosphere phage, Frp1), and commercial potting soil (yielding the Soil lysogenic phage, Slp1). These phages were purified, amplified, and shown to infect B. anthracis ΔSterne and Sterne strains, but not the related B. cereus strain ATCC 14579 (Table 1). Partial genomic sequences were determined (and submitted to GenBank) for each phage to confirm their unique nature. M13 fingerprinting was also performed to distinguish each phage (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Growth of environmental phages on B. anthracis and B. cereus strains.

https://doi.org/10.1371/journal.pone.0006532.t001

A second method for bacteriophage identification focused on the inducible prophages of B. anthracis-like environmental organisms. Briefly, environmental samples were plated on non-selective BHI agar and the resulting colonies were ultimately screened for phenotypes and genotypes commonly associated with B. anthracis and B. cereus (Table S1). Notable B. cereus s.l. strains were then subjected to a prophage induction protocol, based on 24 hours of growth in BHI broth supplemented with fosfomycin, a soil antibiotic and inducer of lyosgenic phages from B. cereus s.l. [46]. Induced culture supernatants were then tested for PFUs on B. anthracis ΔSterne. In this manner, we identified Wip4 and Wip5 from earthworm gut bacteria, Frp2 from a fern rhizosphere bacterium and Htp1 from a human tonsil bacterium. As with the “free” phages identified above, the induced phages were found to be both genetically unique and specifically infective toward B. anthracis (Table 1).

Characteristics of B. anthracis bacteriophages

The nature and variety of B. anthracis phages identified here were examined by electron microscopy. For comparison, we included the well described lysogenic B. anthracis bacteriophage, Wβ [46]. Ultimately, we observed three distinct morphotypes representing three viral families. The Wip2, Wip4, and Frp2 phages (Figure 2C, 2D and 2G) were most similar to Wβ (Figure 2A); their icosahedral heads and long, flexible, non-contractile tails are hallmarks of the Siphoviridae family of bacteriophages. Here, Wip2 is notable, not for an average head size of 60±1.5 nm (mean±SD, n = 50), but for a long tail length of 420±36 nm (n = 25) when compared to the average Siphoviridae length of 150 nm [54]. The Wip5 and Frp1 phages (Figure 2E and 2F) form a distinct group classified as Myoviridae, based on their icosahedral heads and short, thick, contractile tails. Wip1 and Htp1 are tail-less phages (Figure 2B and 2H) with isometric heads and are Tectiviridae; as with such phages, both Wip1 and Htp1 produced tail-like tubes after adsorption or chloroform treatment (data not shown). Our findings indicate that each particular worm gut has multiple distinct phage types as illustrated in Figure 2J, with a micrograph of bacteria-free (sterile-filtered) gut fluid containing undefined extracellular phages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Transmission electron micrographs of bacteriophages negatively stained with 2% uranyl acetate.

The bacteriophages infecting B. anthracis include, (A) Wβ, (B) Wip1, (C) Wip2, (D) Wip4, (E) Wip5, (F) Frp1, (G) Frp2, (H) Htp1, and (I) Bcp1. An extract from the gut of the earthworm Eisenia fetida is shown (J) with two distinct and uncharacterized phages indicated by arrows. Scale bars represent either 25 nm (A–I) or 50 nm (J).

https://doi.org/10.1371/journal.pone.0006532.g002

The impact of lysogeny on colony morphology

To discern a role for environmental phages in the B. anthracis lifecycle, we infected ΔSterne with several phages from Table 1 and evaluated phenotypes of the resulting lysogens. All of the phages formed turbid plaques and were thus expected to be temperate phages capable of lysogeny. Indeed, lysogens recovered from plaques obtained with each phage were quite stable. In an analysis of spontaneous prophage curing, the loss of each phage was observed at a frequency ≤10⁻⁴ (Table S2), which is similar to that observed for stable λ phage lysogens [55]. The presence of putative phage integrase- and recombinase-encoding loci in the genomic sequences of several phages in this study suggests that lysogeny in most cases is associated with phage integration.

The colony morphology of ΔSterne lysogens was varied, based on the lysogenic phage. After a 24 hour incubation at 30°C, the Wip4, Wip5, Frp1, and Htp1 lysogens all presented flat, large, and opaque colonies that were quite distinct from a smaller, white, and matte morphology seen here with ΔSterne alone and the Wip1, Wip2, Frp2, Slp1, and Wβ lysogens. We confirmed this difference by re-generating all the lysogens using a liquid culture infection method. Here, lysogeny was readily detected and again the Wip4, Wip5, Frp1 and Htp1 lysogens all formed large and opaque colonies (examples seen in Figure S1A and S1B).

An interesting exception among the phages in this study concerns Bcp1 (for B. cereus phage), a member of the Myoviridae family (Figure 2I) that forms plaques on B. cereus and B. anthracis Sterne, but not B. anthracis ΔSterne (Table 1). Bcp1 nonetheless adsorbed to B. anthracis ΔSterne (Figure S2A and S2B, Table S3) and eventually yielded lysogens. The ΔSterne/Bcp1 lysogens were, however, only found in biofilm-like adherences at the liquid-air interface of 3 month infection cultures. It is likely that over long-term incubation, rare phage-sensitive derivates became stably infected and formed the observed biofilms. Non-lysogens were not observed in these structures. After subculture to BHI agar, the ΔSterne/Bcp1 lysogen also produced the large-colony phenotype (distinct from ΔSterne).

The impact of lysogeny on sporulation

Since the distinctly larger colonial phenotype of Wip4, Wip5, Frp1, Htp1 and Bcp1 lysogens after 24 hours of growth on agar medium could be attributable to a faster growth rate, we followed growth in liquid M9 minimal medium. Indeed, each of these lysogens doubled in 37–39 minutes while ΔSterne and the Wip1, Wip2, Frp2, Slp1, and Wβ lysogens doubled in 61–64 minutes. Fast growing B. anthracis derivatives that form large and opaque colonies also appear after long-term storage [56] or growth at high temperature [47] and correspond to non-sporulating or asporogenous (Spo⁻) forms. To determine whether the ΔSterne lysogens described here were also Spo⁻, we examined sporulation in 24 hour Leighton-Doi (LD) cultures grown with aeration at 37°C. As expected, the large-colony Wip4, Wip5, Frp1, Htp1, and Bcp1 lysogens did not sporulate in conditions that yielded >10⁸ spores ml⁻¹ for ΔSterne and the small-colony Wip1, Wip2, Frp2, Slp1, and Wβ lysogens (Table 2). Interestingly, when these Spo⁻ lysogens were manipulated to lose (or cure) their respective prophages, the resulting phage-free strains reverted to Spo⁺ (Table 2). These findings indicate that certain B. anthracis prophages can act to block sporulation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Formation of mature spores by B. anthracis ΔSterne and its lysogens under varying temperature and aeration conditions.

https://doi.org/10.1371/journal.pone.0006532.t002

This ability of infecting phages to induce the Spo⁻ phenotype was further examined. Infections of ΔSterne with Wip4 were performed over a range of MOIs and the resulting colonies were analyzed for both lysogeny and the Spo⁻ phenotype. All resulting lysogens were Spo⁻ and they were pronounced even at the lowest MOI (Figure S1C). Furthermore, only live and infective phages (not heat-inactivated forms) resulted in Spo⁻ colonies, suggesting that lysogeny and not a physical interaction with the bacterial surface is needed. We next performed a more detailed comparison of lysogens obtained immediately after infection with the “Spo⁺” (Wβ and Wip2) and “Spo⁻”phages (Wip4, Wip5, and Frp1). While lysogeny was readily detected for both classes, the Wβ and Wip2 lysogens were invariably Spo⁺ and the Wip4, Wip5, and Frp1 lysogens were invariably Spo⁻ (Table S4). These results, taken with findings that phage-curing restores the Spo⁺ phenotype, suggest that lysogeny drives the asporogenous phenotype.

A second distinct sporulation phenotype was also identified in this study. The entire group of Spo⁺ lysogens (Wip1, Wip2, Frp2, Slp1 and Wβ isolates) sporulated in LD cultures grown at 24°C with poor aeration, whereas the parental ΔSterne non-lysogen alone did not (Table 2). Furthermore, the parental phenotype (no sporulation at 24°C) was restored by phage curing. To investigate this phenotype, we looked at the sporulation kinetics of Wip1, Wip2 and Frp2 lysogens in conditions of varying temperature and aeration. In LD broth at 30°C with aeration at 75 rpm there was no real difference in the timing of sporulation for the lysogens and ΔSterne (Figure 3A–C). When, however, the growth temperature was 24°C, the Spo⁺ lysogens triggered sporulation at least four days earlier than ΔSterne and ultimately yielded much higher spore titers (Figure 3D–F). Without aeration this distinction was more profound and lysogens sporulated up to ten days earlier (Figure 3G–I) and by one year retained viabilities of >10⁶ spores and/or vegetative cells ml⁻¹ while ΔSterne cultures were barely viable (<1×10³ cells ml⁻¹). This enhanced sporulation phenotype was also observed in soil-extract medium (Figure 3J–L), but not in rich media like BHI or LB (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Certain lysogens of B. anthracis sporulate rapidly.

Total bacterial cells (vegetative cells and spores; dotted lines) and spores (solid lines) in liquid sporulation cultures were determined over time. Values are mean averages (n = 3) and error bars represent standard deviations. Strains include ΔSterne (open squares) and its lysogens obtained with either Wip1 (open diamonds in A, D, G, and J), Wip2 (open triangles in B, E, H, and K), Frp2 (open circles in C, F, I, and L) or Wip4 (cross-hatches in L). A Wip2-cured lysogen (star symbol in K) and a Wip2-cured lysogen re-infected with Wip2 (open circles in K) are included as well. Sporulating cultures were analyzed under the following conditions: (A–C) LD medium incubated at 30°C with agitation at 75 rpm; (D–F) LD medium incubated at 24°C with agitation at 75 rpm; (G–I) LD medium incubated at 24°C with no agitation; and (J–L) soil medium incubated at 24°C with agitation at 75 rpm.

https://doi.org/10.1371/journal.pone.0006532.g003

Our findings show that lysogenic phages can manipulate the capacity of B. anthracis ΔSterne to sporulate. For several phages, lysogeny yields higher density and longer lived populations. For still other phages, the effect is to block sporulation; this inhibition is perhaps akin to capacity of the B. anthracis virulence plasmids to repress sporulation under certain conditions [35]. Interestingly, the Spo⁻ lysogens cannot be considered poor survivors compared to Spo⁺ forms, considering their long-term survival (>4 months) in soil-extract (Figure 3L).

Lysogeny favors exopolysaccharide synthesis and multicellular behavior

The observation that Bcp1 lysogens form biofilm-like structures led to a more detailed analysis of multicellular behavior. Biofilms are the preferred environmental state for many organisms, consisting of complex and adherent multicellular assemblages maintained within exopolysaccharide-rich matrices [57]. Previously, B. anthracis strain Sterne (itself a φ20 lysogen [48]) was shown to form biofilms using an in vitro flow cell method [25]. Here, we used a modified version of a published protocol for biofilm formation [58], whereby BHI cultures were incubated for 3 months without aeration. Adherent biofilms at the liquid-air interface were recovered, plated for viability, and visualized by microscopy (Figure 4A). Here, ΔSterne yielded only poorly viable debris at the bottom of each culture tube, while the lysogens formed robust biofilms comprised of viable vegetative cells (for Spo⁻ lysogens Bcp1 and Wip4) or vegetative cell/spore mixtures (for Spo⁺ lysogens Wip1 and Frp2). By microscopy, the lysogen biofilms were dramatic bundles of parallel or convoluted vegetative filaments (of indeterminate length) in matrices bound by GFP-PlyG^BD. GFP-PlyG^BD is a fluorescent B. anthracis exopolysaccharide-specific binding agent [46]. While the structure and biosynthetic pathway of this exopolysaccharide are known [59], [60], a role for this molecule in biofilm formation has not been previously described. Here, we have found a pronounced increase in the binding of GFP-PlyG^BD to the exopolysaccharide of all lysogens, based on microscopic and quantitative binding studies of mid-log phase bacteria labeled with GFP-PlyG^BD (Figure 4A). We also show this in parallel labeling studies with a second exopolysaccharide binding protein [61], Alexa Fluor-tagged wheat germ agglutinin. In conditions that support little surface-labeling of ΔSterne, there is a strong fluorescent signal for each lysogen (Figure 4A). The B. anthracis vegetative exopolysaccharide may thus promote the formation of biofilms in a manner akin to that previously described for the Eps exopolysaccharide of B. subtilis [62].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Phenotypic analysis of B. anthracis and its lysogens.

(A) Biofilms formed at the liquid-air interface of 3 month BHI cultures grown without aeration at 24°C (scale bars are 0.5 cm). The total number of bacteria (vegetative cells and spores) and spores alone in each adherent biofilm were enumerated and visualized at 200X and 2000X magnification. GFP-PlyG^BD-labeled cells are shown in 2000X fluorescence images taken with 0.3 second exposures. A non-colorized image is shown for the GFP-PlyG^BD-labeled Wip1 lysogen. 2000X fluorescence images (0.3 second exposures) of mid-log phase BHI cultures labeled with GFP-PlyG^BD or WGA^AF are also shown. Relative fluorescence units (RFUs) corresponding to ∼1×10⁸ mid-log phase cells suspended in buffer are shown as averages of three experiments. (B) Lysogeny alters B. anthracis spore architecture. Transmission electron micrographs (TEM) show that the hallmark single-layer exosporium of ΔSterne actually consists of two distinct layers in >30% of Wip1 lysogen spores. Arrows show the surface nape structure for each exosporium. 45,000X magnifications are shown. Phase contrast images (2000X) show the Wip1 lysogen with an unusual surface structure indicated by an arrow. Closer inspection by scanning electron microscopy (SEM) reveals the surface structure to be an extended, grainy matrix (indicated by an arrow). Scale bars represent 1 µm.

https://doi.org/10.1371/journal.pone.0006532.g004

The Spo⁺ lysogens formed particularly complex biofilms. With ΔSterne/Wip1 as an example, the biofilms were heterogeneous with respect to observed cell-types. Extended bundles of vegetative cells were observed adjacent to regions of highly clumped spores and short vegetative rods, all contained in an exopolysaccharide-rich matrix (Figure 4A, Figure S3A). Interestingly, this biofilm was identical to that formed by either of two environmental B. cereus lysogens, RS423 and RS421, identified for this study (Figure S3C and S3D, Table S1). Both RS421 and RS423 formed adherent biofilms at 3 months, comprised of vegetative bundles, spores, and short rods in a matrix bound by GFP-PlyG^BD.

The static, nutrient-poor, and long-term incubation conditions required to analyze biofilm formation here, could be favoring the accumulation of adaptive mutations in the host chromosome as previously described [63]. Thus, the long-term survival phenotypes observed here could reflect the cumulative effects of lysogeny and adaptive mutation. To examine this, we recovered ΔSterne::Wip2 lysogens from biofilms at 120 days and subjected them to heat-treatments that induced prophage loss. The resulting phage-cured derivative was then re-lysogenized with Wip2; both the cured and re-lysogenized forms were then examined in long-term survival assays. For the cured derivative, we observed complete reversion to the poor survival phenotype observed with ΔSterne (Figure 3K); thus, this strain did not encode mutations that alone promoted survival. For the re-lysogenized derivative, we observed survival kinetics identical to that of the original Wip2 lysogen (Figure 3K); thus, there was no adaptive mutation that enabled lysogen survival. The prophage alone is responsible for the long-term survival phenotype.

Lysogeny favors changes in spore structure

The changes in vegetative cell structure observed here (i.e., filamentous bundles and increased exopolysaccharide) also led us to also investigate changes in spore structure. Spores were prepared from ΔSterne/Wip1 and subjected to an ultrastructural analysis that, indeed, identified two major changes (Figure 4B). First, thin-section TEMs revealed that over 31% of 500 ΔSterne/Wip1 spores had two exosporial layers; normally this is a single layer, with two layers appearing in <1% of ΔSterne spores. An unexplained double exosporium layer has previously been reported for B. megaterium [64] and B. anthracis (M. Fazzini, unpublished observations). The second major change concerned a novel external filamentous structure attached to each spore (Figure 4B). In SEMs, these structures are grainy matrices that appear to mediate adherence to other spores and the grid surface. These matrices did not bind GFP-PlyG^BD and were DNase-insensitive, thus they did not consist of the B. anthracis exopolysaccharide or DNA (data not shown).

Phage-encoded RNA polymerase sigma factors drive phenotypic alterations

We investigated the mechanism by which B. anthracis phages mediate phenotypic changes. Assuming that phage-encoded factor/s are responsible, we constructed and screened both Bcp1 and Wip4 genomic expression libraries for clones disrupting the Spo⁺ phenotype of ΔSterne on LD sporulation agar. Phage genomic fragments were generated by Tsp509I digestion and cloned into a self-replicating B. anthracis vector; after transformation into ΔSterne, the expression of library-encoded phage genes was driven only by their native promoters (i.e., no plasmid-encoded promoter were used). For each library, 0.5% of the resulting clones were Spo⁻ on LD plates, while <0.01% were Spo⁻ in vector-only controls. Sequence analysis revealed that each Bcp1 clone encoded the bcp25,26 locus, while each Wip4 clone encoded wip38,39 (Figure 5A). The deduced amino acid sequences of Bcp25, Bcp26, and Wip39 were very similar to that of several families of bacterial sigma factors, which are RNA polymerase subunits that confer promoter specificity to the transcription complex [65]. The strongest similarity was to σ^F and σ^G of B. cereus s.l. organisms, which are factors that direct sporulation gene expression [66]. Similarity to σ^B, a sigma factor not involved in sporulation but nonetheless similar to σ^F, was not detected. σ^F of B. anthracis Ames was 34%, 24%, and 33% identical to Bcp25, Bcp26, and Wip39, respectively (Figure S4). Wip38 was similar only to a hypothetical B. anthracis phage protein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Phage-encoded effectors of phenotypic change in B. anthracis.

(A) Tsp509I genomic fragments of Bcp1 (1,948 bp) and Wip4 (1,287 bp) identified in an expression library screen of phage-encoded loci that disrupt the Spo⁺ phenotype of ΔSterne. (B) GFP-PlyG^BD-labeled mid-log-phages cells are shown in 2000X fluorescence images with 0.3 second exposures. ΔSterne/pASD2 derivatives are used with and without the indicated promoter-bearing Bcp1 and Wip4 loci. RFUs corresponding to ∼1×10⁸ mid-log phase cells suspended in buffer are displayed as averages of three experiments. (C) Biofilms formed at liquid-air interfaces of 3 ml LD cultures grown without aeration in 35 mm dishes for 1 week. Average numbers of organisms per ml are shown from three experiments for each condition, indicating total viability (vegetative cells and spores) and spore counts. For each strain, 200X and 2000X images of GFP-PlyG^BD-labeled cells are shown with and without fluorescence emission.

https://doi.org/10.1371/journal.pone.0006532.g005

In the next series of experiments, we wanted to confirm that expression of bcp25,26 and wip38,39 was specifically driving the Spo⁻ phenotype. First, we used RT-PCR to show that these loci were indeed expressed from the Bcp1 and Wip4 prophages in ΔSterne during sporulation (Figure S5A and S5B). Next, the bcp25,26 and wip48,49 loci were taken out of the prophage context and cloned into plasmids as fragments that either included or excluded their respective 353-bp and 258-bp upstream promoters. After transformation of these vectors into ΔSterne, RT-PCR analysis confirmed that expression of bcp25,26 and wip38,39 in these backgrounds during sporulation still specifically required their respective upstream promoter sequences (Figure S5C and S5D). While the promoter-less bcp25,26 or wip48,49 clones sporulated at wild-type-like levels, the promoter-bearing clones (i.e., those expressing bcp25,26 or wip48,49) (in the absence of any other phage components) were completely unable to sporulate (Table 2). These findings thus support a mechanism by which Bcp1- and Wip4-encoded sigma factors act alone (in the absence of other phage-encoded elements) to block sporulation. In agreement with this, we also found that insertional inactivation of either bcp25 or wip39 in the Bcp1 or Wip4 lysogen backgrounds, respectively, restores the Spo⁺ phenotype (Table 2) of these formerly Spo⁻ lysogens.

It was very interesting to find that bcp25,26 and wip38,39 were expressed during vegetative growth as well as sporulation (Figure S5E). For this reason, we investigated a role for these loci in changing vegetative cell phenotypes including exopolysaccharide and biofilm production. Ultimately, we found that expression of bcp25,26 and wip38,39 from their native promoters in pASD2 did specifically induce high-level, GFP-PlyG^BD-mediated fluorescence in vegetative ΔSterne (Figure 5B). This exopolysaccharide expression was also accompanied by the formation of biofilms at the liquid-air interface of 1 week-old cultures (Figure 5C). Here, convoluted ropy masses and flaky, sheets of cells were observed, consisting of viable vegetative cells (and no spores) in a matrix bound by GFP-PlyG^BD. The parental ΔSterne strain produced no biofilm whatsoever. Expression of bcp25,26 and wip38,39 therefore drives the vegetative phenotypes associated with Bcp1 and Wip4 infection.

We infer from these findings that Bcp1 and Wip4 prophages mediate host phenotypic modifications via sigma factors encoded by bcp25,26 and wip48,49. While it is interesting that B. anthracis phages are using transcriptional regulatory proteins to effect host cell changes, it not surprising considering the pleiotropic effects such regulators should enable. Nonetheless, such a mechanism has been seen only for proteins like NucC and RecC of Serratia marcescens [67], [68], which are cryptic prophage-encoded transcriptional activators of extracellular nuclease and bacteriocin production. NucC and RecC are, however, not sigma factors, and roles for phage-encoded sigma factors have only been described in phage gene transcription [69], [70].

Analysis of environmental B. cereus s.l. lysogens

The phenotypic changes associated with B. anthracis have to this point been based on analysis of lysogens generated in laboratory infections. In order to determine whether these changes also occur with naturally occurring lysogens, we examined environmental B. cereus s.l.-like strains bearing inducible prophages that are at least infective toward B. anthracis. We used the B. cereus strain ATCC 25621 (a cow feces isolate), and a group of three strains that were identified for this study including RS1045 (a worm gut strain), RS1255 (a human tonsil strain), and RS1557 (a fern rhizosphere strain). Each of these strains exhibited various B. anthracis-like genotypes and/or phenotypes (Table S1), and, in particular, was lysogenic for B. anthracis phages. These isolates most likely represent B. cereus s.l. members that are particularly closely related, genotypically, to B. anthracis [45], [71], [72], [73].

Phenotypic analysis of ATCC 25621, RS1045, and RS1255 revealed that each strain was asporogenous in conditions that support sporulation of ΔSterne (Figure 6A). Focusing on RS1045, we also observed both the induction of high-level exopolysaccharide surface-expression and the formation of biofilms that were highly enriched for extended vegetative filaments (Figure 7A and 7B). For both the Spo⁻ defect of RS1255 and the exopolysaccharide and biofilm formation of RS1045, phage-curing was found to completely reverse the observed effect. Hence, RS1255^CURED could sporulate (Figure 6A) and RS1045^CURED produced a ΔSterne-like level of exopolysaccharide (Figure 7A) and yielded only adherent spores at the liquid-air interface of glass cover-slips (Figure 7B). Here, RS1045 and RS1255 are particularly interesting, since they are the strains from which Wip4 and Htp1, respectively, were originally isolated. The phenotypes associated with both Wip4 and Htp1 infection of their native B. cereus parental strains, therefore match that observed in experimental infections of B. anthracis ΔSterne seen in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Sporulation phenotypes of natural B. cereus s.l. lysogens.

Kinetic analyses of sporulation were performed in LD medium. All values are mean averages (n = 3) and error bars are standard deviations. (A) Sporulation of environmental B. cereus. Total viability (dashed lines) and spores alone (solid lines) are shown for growth conditions that include 37°C at 180 rpm (RS1045 and ATCC 25621), 30°C at 180 rpm (RS1255) or 24°C at 75 rpm (RS1557). B. anthracis ΔSterne (squares), the indicated environmental isolates (triangles), and phage-cured variants RS1255^CURED and RS1557^CURED (diamonds) are shown. (B) Sporulation for environmental B. anthracis strains. Cultures were grown at 37°C and aerated at 180 rpm. Strains include RS1615 or RS1046 (triangles) and RS1615^CURED or RS1046^CURED (diamonds). An electron micrograph of the inducible phage from RS1615 is shown (scale bar in 50 nm). (C) Sporulation for B. anthracis Sterne. Cultures were grown at 37°C and aerated at 180 rpm. Included are Sterne (circles), Sterne^CURED (diamonds), and the Wip1 or Wip4 lysogens of Sterne (closed circles). An electron micrograph shows the inducible Sterne phage (scale bar is 50 nm).

https://doi.org/10.1371/journal.pone.0006532.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Exopolysaccharide and biofilm phenotypes of natural B. cereus s.l. lysogens.

(A) Mid-log phase cells from BHI cultures labeled with GFP-PlyG^BD. 2000X fluorescence images are 0.2 second exposures. For ΔSterne, a corresponding phase image is shown. RFUs for ∼1×10⁸ mid-log phase cells suspended in buffer are shown. (B) Biofilms formed by environmental B. cereus (RS1045) and B. anthracis (RS1615) backgrounds. For RS1045 strains, sections at the liquid-air interface of crystal violet-stained biofilms are shown. The RS1045 field is dominated by filamentous forms, while that of the phage-cured for is dominated by spores. Liquid-air interfaces are shown (along with total bacteria ml⁻¹ for each culture) for the RS1615 strains. Additionally, 200X images of GFP-PlyG^BD-stained biofilm (RS1615) or bottom-settled material (RS1615^CURED) are shown.

https://doi.org/10.1371/journal.pone.0006532.g007

The Spo⁺ environmental strain RS1557 was also distinctive in that it displayed the “rapid” sporulation phenotype and sporulated well in conditions that caused RS1557^CURED to lag by several days (Figure 6A). The prophage of RS1557 is Frp2 (Table S1), which also induced the rapid sporulation of ΔSterne in Figure 3F. This finding, taken with that of the Spo⁻ environmental lysogens above, confirms that all the lysogeny-mediated phenotypes in ΔSterne can also be observed in environmental B. cereus s.l. lysogens, often with the same phages.

Analysis of “natural” B. anthracis lysogens

To complete this phenotypic study, we next analyzed a group of naturally occurring B. anthracis lysogens. Included is the well-studied lab isolate Sterne, which harbored an inducible prophage (seen in Figure 6C) similar to that previously described [48]. The RS1615 and RS1046 environmental strains were also used, and were identified here as bona fide B. anthracis isolates based on the detection of chromosomal and/or virulence plasmid markers (Table S1). RS1615 is a soil isolate that encodes the B. anthracis-specific phage φ1615 (seen in Figure 6B), while RS1046 is a worm isolate that encodes Wip5 (seen in Figure 2E).

In a comparison of the sporulation kinetics of Sterne, RS1615, or RS1046 to that of their respective phage-cured derivatives (Figure 6B and 6C), the lysogens consistently sporulated faster and more efficiently (i.e., they yielded higher spore titers). These findings are particularly noteworthy, considering that well aerated cultures were used here (37°C and 180 rpm) which normally yielded no differences between ΔSterne and its Spo⁺ lysogens. Thus, for these B. anthracis lysogens, the prophage state induced a rapid sporulation phenotype. This phenotype, unlike that seen with the Spo⁺ ΔSterne lysogens, was observed in all conditions and was not temperature- and aeration-dependent. Further analyses focused on RS1615 and showed that both high-level exopolysaccharide expression (Figure 7A) and the formation of adherent biofilms enriched with filamentous bacteria (Figure 7B) was absolutely dependant on lysogeny.

The effect of poly-lysogeny was investigated using the Sterne strain. Sterne (a φ20 lysogen) was stably infected with either Wip1 or Wip4. The dominant phenotype of each poly-lysogen was found to be dictated by the new phage (Figure 6C). Thus, the φ20/Wip1 poly-lysogen had a rapid sporulation phenotype like that of ΔSterne/Wip1 in Figure 3D, and the φ20/Wip4 poly-lysogen was asporogenous like ΔSterne/Wip4 in Table 2. In all, these findings confirm that phenotypes observed in ΔSterne, are applicable to other B. anthracis backgrounds.

B. anthracis lysogens colonize soil microcosms

Lysogeny of B. anthracis induces several phenotypes that should impact survival in the environment. Considering that B. anthracis is a member of a lineage of soil organisms, we proceeded to assess the impact of lysogeny on the ability of ΔSterne and environmental B. anthracis and B. cereus strains to survive long-term in conditions that mimic the soil.

Microcosms were developed whereby ∼1×10⁹ vegetative bacteria (with <1×10¹ spores) were inoculated into sterile potting soil, and at time points over 24 weeks were recovered and enumerated. For ΔSterne, viability of both Spo⁺ (Wip1, Slp1, and Wip2) and Spo⁻ (Wip4, Bcp1, and Frp1) lysogens remained around 10⁷ CFU g⁻¹ of soil for 6 months (Figure 8A and Figure S6A). Environmental B. anthracis strains RS1615 and RS1046 were similarly durable (Figure 8C and Figure S6D), as was the environmental B. cereus strain RS1045 (Figure S6C). So the lysogens did exhibit long-term survival in artificial soil conditions. Indeed, the only requirement for survival was lysogeny, as ΔSterne, RS1615^CURED, RS1046^CURED, and RS1045^CURED yielded few or no viable organisms by 6 months (Figure 8A and 8C, Figure S6C and S6D). Microscopic analyses at 3 months showed the lysogens forming large exopolysaccharide-rich biofilms (Figure S7) akin to seen here in laboratory media. Thus, lysogeny-driven biofilm formation is closely associated with, and perhaps required for, long-term colonization of a soil habitat.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Impact of lysogeny on survival in soil and earthworms.

Survival is shown as colony forming units (CFUs) per gram of recovered soil or worm guts (solid lines). Shed phages (dashed lines) are shown as plaque-forming units (PFUs) extracted per gram of soil or worm guts. Total B. anthracis viability (vegetative cells and spores; squares), spore counts (diamonds), and phages (triangles) are shown. Values are mean averages (n = 5) and error bars are standard deviations. (A) ΔSterne/pASD2 (non-lysogen) and its Wip1 and Wip4 lysogens in soil. (B) ΔSterne/pASD2 (non-lysogen) and its Wip1 and Wip4 lysogens in earthworm intestines. (C) Survival of environmental B. anthracis strain RS1615 and its phage-cured derivative (RS1615^CURED) in soil and earthworm intestines. (D) Infection of B. anthracis during co-culture with lysogens or free Wip1 particles in soil or earthworm intestines. Strains ΔSterne/pASD2 and Sterne/pASD2 were either inoculated alone or with strains ΔSterne/Wip1 or RS1615 into each microcosm. At indicated times, ΔSterne/pASD2 and Sterne/pASD2 were recovered and scored by PCR for infection with Wip1 or φ1615. Survival of ΔSterne/pASD2 and Sterne/pASD2 inoculated alone (squares) and derivatives infected with Wip1 (closed circles) or φ1615 (open circles) are shown. For the spiking with Wip1, strain ΔSterne/pASD2 was inoculated alone or with 1×10⁸ Wip1 phages. Free Wip1 (triangles) is shown with total viability (squares) and spore counts (diamonds) for ΔSterne/pASD2 lysogenized with Wip1.

https://doi.org/10.1371/journal.pone.0006532.g008

B. anthracis lysogens colonize the gut of Eisenia fetida, the invertebrate redworm

As part of his seminal studies into the cause of infectious disease in animals, Louis Pasteur noted an abundance of earthworms at buried anthrax carcasses and proceeded to identify B. anthracis spores in the guts of these worms [53]. While he postulated that earthworms were important in the B. anthracis lifecycle, there has been no further study of this interaction. Nevertheless, the worm gut is a complex microbial habitat [74] that includes B. cereus and B. thuringiensis [7], [75]. These are the reasons why we initially looked for B. anthracis phages in the worm gut and why we next proceeded to study colonization.

Earthworm microcosms were developed based on a combination of published protocols [75], [76] whereby non-sterile soil containing individual animals were inoculated with ∼5×10⁹ bacteria and incubated for 1 week, prior to 22 subsequent weeks marked by repeated passages through “clean” soils (i.e., no B. anthracis). At 1, 12, and 24 week time points after inoculation into clean soil, gut contents were plated on selective media to enumerate B. anthracis derivatives. For ΔSterne, the viability of both Spo⁺ (Wip1, Slp1, and Wip2) and Spo⁻ (Wip4, Bcp1, and Frp1) lysogens remained steady around 10⁴ CFUs g⁻¹ of worm gut over 24 weeks (Figure 8B and Figure S6B). The B. anthracis strains RS1615 and RS1046 and B. cereus strain RS1045 were stably recovered as well (Figure 8C and Figure S6D). For strains that lacked lyosgenic phages, including ΔSterne, RS1615^CURED, RS1046^CURED, and RS1045^CURED, only transient colonization was noted with few recoverable bacilli at 24 weeks. B. anthracis therefore colonizes the intestinal tract of earthworms in a process that requires lysogeny. Presumably, latent endosymbiotic capabilities of B. anthracis are being activated in the worm gut.

Phage shedding in environmental microcosms

Bacteriophage-host interactions are often viewed only in terms of lytic behavior and integrated prophages (i.e., lysogeny). Nonetheless, a complicated range of other conditions may also exist, including chronic states marked by phage budding or extrusion and persistent infections marked by episomal prophages, low-level bacterial lysis, phage shedding, and prophage curing [77]. Such behaviors are influenced by growth conditions and other factors, and are believed to be common for environmental lysogens.

Here, we investigated the viral-host state in B. anthracis and found evidence of pseudolysogeny. Pseudolysogeny occurs when phages multiply in a small fraction of a population, shedding infective particles, and maintaining environmental levels of virus in the presence of infected bacterial populations [77], [78], [79]. In studies of the 24 week soil and earthworm microcosms, we observed considerable viral shedding by otherwise stable B. anthracis and B. cereus lysogen populations (Figure 8A–C, Figure S6A–D). For each lysogen, we observed roughly the same number of free phages as viable bacteria in each microcosm throughout the experiment. No PFUs were observed with non-lysogenic strains. The shedding of phage here is consistent with pseudolysogeny. We similarly checked shedding in sheep blood, BHI, and LD media and observed free phages released from most lysogens, with the exception of Wip1, which only shed in blood.

Lysogeny and the acquisition of novel traits in environmental microcosms

The shedding of phages by B. anthracis and B. cereus lysogens could reflect a mechanism for DNA transfer (and thus niche expansion) among these organisms. To investigate this, a series of co-cultures were established in soil microcosms and worms, in which ΔSterne or Sterne strains (transformed with pASD2 to provide antibiotic resistance and enable selective recovery) were co-inoculated with B. anthracis and B. cereus phage “donor” strains. Phages shed by the donors may infect the susceptible ΔSterne or Sterne “recipients”, which can then be selectively recovered and analyzed. Identical experiments were also established whereby free Wip1 phages were used in lieu of a donor lysogen.

For the soil microcosms, recipient ΔSterne/pASD2 or Sterne/pASD2 organisms were inoculated one day before the addition of “donor” lysogens or phages. The recipients were recovered at indicated time points, enumerated, and examined to confirm lysogeny. When ΔSterne/pASD2 was incubated without a phage source as a control, little or no survival was apparent at 24 weeks (Figure 8D, Figure S6E). This effect is not as severe with Sterne/pASD2 alone, possibly because it is a φ20 lysogen [48] which has been heavily passaged in the laboratory. Nonetheless, when ΔSterne/pASD2 or Sterne/pASD2 were co-cultured with any of the donor lysogens (or free phage), there was a pronounced improvement in survival, with 10⁶–10⁸ CFUs detected per gram of soil (Figure 8D, Figure S6E) over the experiment. In each case, the surviving recipient strains had all become lysogenized with Wip1, φ1615, φ1046, or Wip4. Thus, B. anthracis phages that are either shed or added as free particles can lysogenize susceptible strains and confer the long-term survival phenotype in the soil milieu.

For the worm experiments, “donor” lysogens (ΔSterne/Wip1, RS1615, RS1046, or RS1045) were used to colonize the worm gut for one week, before the animals were recovered, washed, and passaged through clean soil (i.e., no added bacteria). Again, ∼1×10⁸ free Wip1 particles were added in lieu of the donor lysogens in some experiments. After one week, the worms were moved to new microcosms and exposed to “recipient” bacilli before they were again recovered, washed, passaged for 24 weeks through clean soil, and plated for enumeration. For ΔSterne/pASD2 and Sterne/pASD2 incubated without a phage source as a control, few if any organisms were recovered from the worm gut at 24 weeks (Figure 8D). In the presence of either the lysogens or free phages, however, persistent survival of 10³–10⁴ CFUs per gram of worm gut was routinely observed, including both vegetative and spore forms. In each case, recipient strains had all become stable lysogens of Wip1, φ1615, φ1046, or Wip4. Phages shed by stable worm gut populations of B. anthracis can, therefore, infect incoming non-lysogens and confer the novel worm-survival phenotype.

Having observed phage shedding and concomitant lysogeny in soil and worm microcosm co-cultures, we proceeded to a similar analysis in other environments that support phage shedding. Sheep blood, BHI and LD cultures were all co-inoculated with donor lysogens and the ΔSterne/pASD2 recipient. Recipient strains were recovered after 2 days and examined to confirm lysogeny. In most cases where we previously observed viral shedding (Figure S8), infection of the recipient strain was detected (Table S5). The exception was Wβ, which is shed in all conditions, yet only lysogenizes ΔSterne/pASD2 in the sheep blood.

Phage-induced expression of host loci in the soil and earthworm intestine

To understand how prophages activate a phenotype like worm gut colonization, we sought B. anthracis loci expressed in this milieu. Two methods were used to identify bacterial promoters that, when fused to a GFP gene, yield colonies that are: 1) fluorescent in a Wip4 lysogen on worm-extract agar; 2) non-fluorescent in ΔSterne (non-lysogen) on worm-extract agar; and 3) non-fluorescent in a Wip4 lysogen on BHI agar. This differential regulation should reflect a gene that is induced by virtue of lysogeny in the worm gut.

For one method we constructed a promoter-probe library by fusing ΔSterne genomic fragments to a promoterless GFP gene. Ultimately, only one fusion exhibited differential regulation – fluorescent in a Wip4 lysogen on worm agar (Figure 9A) but not BHI agar (data not shown) and non-fluorescent in a ΔSterne background in any condition (data not shown). Sequence analysis revealed the promoter of BA3436. BA3436 is a 317 residue protein that is >40% identical to luciferase proteins of Gram-positive and Gram-negative bacteria (Conserved Domain Database entry cd01096), and is expressed as the promoter-proximal gene of a likely tricistronic locus that includes BA3435 and BA3434.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 9. Phage-regulated loci in B. anthracis.

(A) Promoter-gfp fusions expressed on worm agar. Bacteria were plated for 16 hours at 30°C and resulting colonies are shown in 200X phase-contrast and fluorescence images (0.5 second exposures). Strains are ΔSterne or its Wip4 lysogen (“/Wip4”) transformed with pASD2 encoding gfp alone or indicated promoter fusions (P). (B) Expression of P^BA3443-gfp on BHI agar in either the ΔSterne/Wip4 background or ΔSterne co-transformed with the compatible vector pWH1520 (either without and insert or that bearing bcp25,26 or wip48,49). Bacteria were plated for 16 hours at 30°C and colonies are shown in 200X phase-contrast or fluorescence images with 0.5 second exposures. 2000X fluorescence images of representative bacteria are shown with exposure times. RFUs are averages of three experiments for ∼1×10⁸ mid-log phase cells grown in BHI and suspended in buffer. (C) BA3443 and earthworm colonization. One month after colonization, bacteria were recovered, visualized, and enumerated. Strains are ΔSterne/Wip4 transformed with indicated vectors or the ΔSterne/Wip4 mutant ΔBA3443. Hind-guts were examined by phase-contrast or fluorescence microscopy at 200X or 2000X magnification. Exposure times are indicated in 2000X images. Average CFUs per gram of gut is shown from three experiments, indicating total viability (vegetative cells and spores) and spores alone. (D) B. anthracis-encoded loci required in the soil. Bacteria were recovered and enumerated at the indicated times from inoculated soil microcosms. Strains include ΔSterne/Wip4 with pASD2 alone or the indicated mutations. Values are mean averages (n = 3) and error bars are standard deviations.

https://doi.org/10.1371/journal.pone.0006532.g009

Three additional differentially regulated promoters were identified based only on the predicted functions of their ORFs. The presence of BA3443, BA0672, and BA1295 in the B. anthracis genome has previously been described as a remnant of ancestral endosymbiotic or insect-pathogenic lifestyles [30]. BA3443 is a homolog of the enhancin metalloprotease family (Pfam entry 03272), that includes many B. cereus s.l. proteins with predicted roles in mucus degradation and gut wall interactions with invertebrates [80]. BA0672 and BA1295 are homologs of the immune inhibitor A metalloprotease family (Pfam entry 05547) that includes several B. cereus s.l. proteins with predicted roles in degrading antibacterial peptides and insect gut survival [81]. As these proteins could mediate worm-gut survival for B. anthracis, we created transcriptional fusions of BA3443, BA0672, and BA1295 to gfpmut2 and screened them for differential fluorescence in a ΔSterne/Wip4 background. Using BA3443 as an example here, the promoters of all three loci directed GFP-mediated fluorescence in bacterial colonies on worm agar (Figure 9A), but not BHI agar (Figure 9B). In a ΔSterne background, no fluorescence was observed on worm agar (Figure 9A). Thus, BA3443, BA0672, and BA1295 are specifically induced in the worm milieu in lysogen-dependent manner.

Remaining with BA3443 as the model for phage-induced genes here, we sought to determine whether the bcp25,26 and wip48,49 products could drive this up-regulation. Promoterless versions of bcp25,26 and wip48,49 were cloned into pWH1520 and introduced into ΔSterne also harboring plasmid pASD2::P^BA3433-gfp. Resulting strains were plated on BHI agar, which is a condition that does not normally support BA3443-directed GFP fluorescence in either the ΔSterne or ΔSterne/pWH150 backgrounds (Figure 9B). When bcp25,26 or wip48,49 was present and expressed from pWH1520, we saw brightly fluorescent colonies on BHI agar that also yielded over 100-fold higher RFU levels (compared to controls) in BHI liquid (Figure 9B). BA3443 expression is therefore phage gene-dependant. This effect is likely to reflect a mechanism whereby phage-encoded sigma factors (including Bcp25, Bcp26, and Wip49) direct RNA polymerase to host promoters like that of BA3443.

Next, we confirmed that expression of BA3443 does occur in the worm gut. Worms were colonized for one month with ΔSterne/Wip4 lysogens encoding gfp alone or that fused to the BA3443 promoter. Hind guts were extracted and plated for viability to first confirm colonization with both strains (Figure 9C). Next, gut contents were examined microscopically and found to contain abundant masses of vegetative cells that were highly fluorescent only for the P^BA3443-gfp-bearing strain. The BA3443 locus is therefore expressed by B. anthracis lysogens colonizing the worm gut.

Finally, we evaluated the importance of these phage-induced genes in the colonization process. Toward this end, we insertionally inactivated BA3443, BA0672, and BA1295 in a ΔSterne/Wip4 background and inoculated each mutant into the worm microcosm. After 1 month, only the ΔBA3443 mutant exhibited a severe defect and was barely recoverable (Figure 9C and data not shown). Since BA0672 and BA1295 encode immune inhibitor A homologs that are 67% identical to each other, they may be redundant activities in the worm gut. To see if these findings extend to soil-survival conditions, we similarly examined each mutant in 3 month soil microcosms. While the control strains (with either pASD2 or a ΔBA4109 mutation) survived to 3 months, the BA3443, BA0672, and BA1295 mutants were only transiently detected and therefore unable to survive (Figure 9D). These findings confirm that phage-regulated B. anthracis loci are required for long-term colonization of the soil and earthworm gut.

Bacterial strains and plasmid constructs

The B. anthracis strains used in this study are described in Table S6. Sources of environmental B. anthracis and B. cereus strains are listed in Table S1. Plasmid pWH1520 is an E. coli-Bacillus spp. shuttle vector. Plasmid pASD2 is an E. coli-B. anthracis shuttle vector with a temperature-sensitive replicon that cannot support plasmid replication at growth temperatures above 37°C [87]. For construction of the pASD2::gfp transcriptional fusion vector, gfpmut2 [88] was amplified by PCR with indicated primers (Table S7) and the product was cloned into the KpnI-SmaI sites of pASD2; this creates a unique primer-encoded EcoRI site upstream of gfpmut2 for promoter-probe fusions. Promoter-bearing sequences were amplified by PCR with the indicated primers to generate a 543-bp fragment that encoded 444-bp of the BA3443 promoter, an 881-bp fragment that encoded 740-bp of the BA0672 promoter, a 992-bp fragment of that encoded 147-bp of the BA1295 promoter, and a 444-bp fragment that encoded 356-bp of the BA3436 promoter. For insertional mutagenesis, internal fragments of BA3443, BA0672, BA1295 and BA4109 were amplified by PCR with indicated primers and cloned into the KpnI site of pASD2. For expression studies, the bcp25,26 locus was PCR-amplified from a Bcp1 DNA template with indicated primers yielding either a 1948-bp fragment that includes 353-bp of promoter sequence or a 1541-bp fragment that includes no promoter. The wip38,39 locus was amplified from a Wip4 template as either a 1287-bp fragment that includes 258-bp of promoter sequence or a 965-bp fragment that encodes no promoter. Resulting bcp25,26 and wip38,39 products were cloned into the SmaI sites of pASD2 or pWH1520. Promoter-bearing clones are referred to in the text with the prefix “P-“. The PlyG binding domain (PlyG^BD) was amplified by PCR from a Wβ phage template with indicated primers and cloned in-frame with the 3′ end of gfpmut2 in plasmid pBAD24 as described previously [46].

Bacterial manipulations and growth conditions

Bacterial strains were grown in Luria broth (LB), brain-heart infusion broth (BHI), or Leighton-Doi broth [89] according to standard protocols; plates were made by adding Bacto^TM agar to a final concentration of 1.6%. Soil-extract and cellulose medium were prepared as described [90], [91]. Worm-extract plates were prepared as follows: 150 g of Eisenia fetida earthworms (New York Worms, Long Island, NY) were flash frozen, thawed, suspended in 1 liter of dH₂0 and autoclaved. Agar (1.6%) was then added and the solution was re-autoclaved, cooled to 55°C, and poured into 150 mm wide plastic Petri dishes. When necessary, antibiotics were added to liquid or solid media at the following concentrations: tetracycline, 10 µg ml⁻¹; kanamycin, 50 µg ml⁻¹; spectinomycin, 250 µg ml⁻¹, and ampicillin 100 µg ml⁻¹. Activity of the P^BAD promoter in pBAD24 was induced during growth by the addition of L-arabinose to a final concentration of 0.2%. All Bacillus strains were grown at 30°C unless otherwise indicated. Aside from the production of B. anthracis mutants, all strains bearing pASD2 were maintained at 30°C to support its freely replicating, plasmidial form. Bacterial doubling times were determined using a standard laboratory technique.

Plasmids were introduced into electrocompetant B. anthracis using 0.4-cm gap cuvettes (BioRad) and conditions of 25 µF, 400Ω, and 2.5 kV in a GenePulser electroporation apparatus (Bio-Rad, Inc.). Plasmid pASD2 was used to insertionally inactivate B. anthracis ΔSterne loci according to a previously described method [87]. The general method used to construct Bacillus lysogens (unless otherwise stated) required mid-exponential phase cultures (grown in 5 ml BHI at 30°C with aeration at 150 rpm in 50 ml Falcon^TM tubes) to be infected with 1×10⁸ phage particles for 30 minutes, washed, and plated (undiluted and three 100-fold serial dilutions thereof) on BHI agar for 16 hours at 30°C. Representative colonies were subcultured on BHI agar and screened by PCR with phage-specific primers (including those in Table S7) to confirm lysogeny. In the standard 1 day infection protocols, frequencies of lysogeny ranged from 0.01 to 0.3. For Bcp1, ΔSterne lysogens were only obtained after infection with 5×10³ phage particles and incubation in the dark without aeration for 1 month. After the 1 month infection, all resulting colonies were lysogenic and assembled into biofilms. Phage curing was induced using a previously described heat treatment method [92]. Spontaneous phage curing was analyzed by plating overnight BHI cultures on BHI agar and screening 1×10⁴ resulting colonies for the rare colony morphology associated with loss of the indicated prophage. Putative cured derivatives were examined by PCR with phage-specific primers to confirm prophage loss.

Phage identification and amplification

Environmental samples were prepared, according to standard protocols, from soil, salt and fresh water, herbivore feces, bat guano, landfill, silage, marine sludge, phylloplane washings, arthropod guts, and other sources. For solid materials, 1–5 g was added to 5–10 ml phosphate-buffered saline (PBS) and gently agitated at 4°C for 16 h. After centrifugation, phage-containing supernatants were recovered, passed through 0.2 µM filters, and 5 ml of filtrate was added to 5 ml of late-exponential phase ΔSterne. For Bcp1 identification, strain B. cereus T used instead. After overnight incubation at 30°C with aeration, supernatants were recovered, filtered, and used again to infect late-exponential phase bacterial cultures. After overnight incubation at 30°C with aeration, sterile supernatants were recovered, plated (undiluted and three serial 100-fold dilutions thereof) on fresh bacterial lawns on BHI plates, and incubated overnight at 30°C. Resulting phage plaques were recovered, plaque-purified, and, ultimately, used to infect 5 ml exponential phase bacterial cultures. Every two days, phage supernatants were collected and used to infect larger culture volumes.

Earthworms that yielded B.anthracis-active phage for this study were isolated as either free phages or induced prophages from forest leaf litter in Stroudsburg, PA (Wip1 and Wip4) and a compost heap in Southold, New York (Wip2 and Wip5). Other phages and their sources are as follows: Frp1 and Frp2, fern rhizosphere from Stroudsburg, PA; Htp1, a human tonsil from New York, NY: Bcp1, landfill soil in Port Washington, NY; and Slp1, from commercial potting soil (Miracle-Gro^®). The Wβ phage was originally obtained as a lyosgenic phage of ATCC 11950 and is described elsewhere [46]. Total genomic DNA was obtained from each phage and subjected to agarose gel electrophoresis to confirm that the genome sizes were similar to that previously described [54] for members of the family Tectiviridae (∼16 kb), Siphoviridae (∼45 kb) and Myoviridae (∼140 kb). Partial genomic sequences ranging in size from one to five kilobases were then determined for each phage using a previously described metagenomic method [93]. Portions of the Wip1, Wip2, Wip4, Frp2, Htp1, and Bcp1 sequences are available in GenBank under accession numbers GQ214700, GQ214701, GQ221690, GQ214703, GQ214702, and EU930824, respectively. In addition to the sequence analysis, the distinct nature of each phage was confirmed by M13 fingerprinting analysis [94] Protein sequences for the recombination repair protein of phage Bcp1 and site-specific recombinase of phage Wip4 are have been submitted to GenBank under the names Bcp90 (GQ338829) and Wip21 (GQ338829), respectively.

Spontaneous phage shedding in culture

Five ml overnight BHI cultures were washed twice with PBS, resuspended in 5 ml of PBS and used to inoculate 5 ml of either defibrinated sheep blood (Cleveland Scientific, Inc.), BHI, or LD medium at a dilution of 1∶100. After 24 hours of growth at 30°C with aeration, cultures were pelleted by centrifugation and the supernatant was recovered and filtered with 0.2 µM membranes. Sterile supernatants were titered on B. anthracis ΔSterne and resulting plaques were enumerated. Infections during co-culture are described in Table S5.

Fluorescence studies

The GFP-PlyG^BD fusion protein was expressed and purified as described [46]. Alexa Fluor^®-labeled wheat germ agglutinin (AF^WGA) was obtained from Molecular Probes (Eugene, OR) and used according to the manufacturer's protocol. Biofilm samples were labeled with a GFP-PlyG^BD solution (1 mg ml⁻¹) in PBS for 2 minutes at room temperature, rinsed with PBS, and visualized by microscopy. For the analysis of exponential phase bacteria, strains were grown for three hours at 30°C before 1 ml aliquots were removed, washed with PBS, and labeled with either GFP-PlyG^BD (1 mg ml⁻¹ in PBS) or AF^WGA. Samples were labeled for 2 minutes at room temperature, washed in PBS and then visualized by microscopy or subjected to an endpoint fluorescence analysis using a SpectraMax^® M5 microplate reader (Molecular Devices, Inc.). Relative fluorescence units (RFUs) are arbitrary values determined in black 96-well plates using excitation and emission wavelengths of 485 nm and 538 nm, respectively. GFP-PlyG^BD binds B. anthracis-specific neutral polysaccharide and WGA^AF is a carbohydrate-binding protein that recognizes N-acetylglucosaminyl sugars.

Microscopy

Phase-contrast and fluorescence microscopy were performed with an Eclipse E400 microscope (Nikon) using the QCapture Pro^® version 5.1 imaging software. The TEM and SEM analyses were performed at The Rockefeller University Bio-Imaging Resource Center essentially as described [95], [96], [97]. Images were assembled using Adobe Photoshop 10.0.

Sporulation studies

To assess the sporulation phenotype, overnight cultures were initially grown overnight in BHI supplemented with 0.2% glucose. The cultures were then washed twice, resuspended in LD, and used to inoculate 7.5 ml LD cultures at a 1∶100 dilution (in 50 ml Falcon^TM tubes). The cultures were incubated with indicated temperature and aeration conditions and at the indicated time points, aliquots were removed and either plated on BHI agar (for total viability) or heated for 15 min at 95°C, cooled for 5 min at 4°C, and plated for viability on BHI agar (for heat-resistant spore counts). The sporulation defects described in Table 2 were each separately confirmed using an assay for chloroform resistance [98]. For the analysis of sporulation in biofilms, culture tubes were vortexed to resolve all aggregated material prior to the heat resistance assay. For the analysis of spore ultrastructure, vegetative B. anthracis cells were first induced to sporulate by growth at 30°C for 4 days on either LD agar (for TEM) or cellulose agar (for SEM) in the dark. Spores were then purified to homogeneity using a water-washing method [98] and stored in dH₂0 at 4°C in the dark.

Construction and analysis of Bcp1 and Wip4 expression libraries

Phage libraries were generated essentially as described [99] with the exception that Tsp509I digested fragments in the 1.0–3.0 kb size range were cloned into pASD2. Complex plasmid pools of >1×10⁴ distinct clones were electroporated into B. anthracis ΔSterne. Spc- and kan-resistant clones were obtained at 30°C on BHI plates and replica plated to LD sporulation agar (with antibiotics) on glass plates. The glass plates were incubated for two days at 37°C, exposed to chloroform vapors for 30 min, and incubated again for 2 days at 37°C. Chloroform kills everything except spores, thus Spo⁻ colonies cannot re-grow after exposure and are easily distinguished from the majority background of Spo⁺ colonies. Colonies from the master plates, corresponding to dead Spo⁻ colonies on glass plates, were recovered and plasmid was prepared and sequenced. Five and seven clones were eventually identified from the Bcp1 and Wip4 libraries, respectively, with sporulation defects specifically assigned to plasmid insert expression. While insert sizes varied, each encoded either bcp25,26 or wip38,39. The DNA sequences of bcp25,26 and wip38,39 are submitted to GenBank.

Construction and analysis of the ΔSterne promoter-probe library

The chromosomal DNA of ΔSterne was isolated using Qiagen Genomic-tips according to the manufacturer's protocol. After partial digestion with Tsp509I, fragments in the 1- to 3-kb range were gel purified using the QIAquick gel purification system (Qiagen, Inc.) and ligated into the EcoRI site of dephosphorylated pASD2::gfp. The mix was transformed into E. coli XL10-Gold (Stratagene, Inc.) and ∼10,000 colonies were scraped up and used for plasmid purification with the Qiagen Midi kit. The library was then electroporated into B. anthracis ΔSterne/Wip4 and plated on worm-extract agar containing kanamycin and spectinomycin. After two days at 30°C the library consisting of ∼5000 clones was screened for fluorescence using a hand-held fluorescent lamp. Fifty clones displaying various fluorescence intensities were identified and subcultured to BHI agar for rescreening. Only one clone was then observed to not be fluorescent on BHI (but was on worm agar). Plasmid DNA was then recovered and ultimately used to transform ΔSterne. In the ΔSterne background, fluorescence was not observed on either worm or BHI agar. Sequence analysis revealed a 581-bp insert encoding both the 5′ end of BA3436 as 356-bp of its upstream promoter region.

Isolation and analysis of environmental Bacillus strains

The environmental isolates used in this study and the sources from which they were identified are described in Table S1. After recovery from the environment, samples were suspended in PBS to create a slurry, vortexed for 1 minute and plated (undiluted and three 100-fold serial dilutions thereof) on BHI agar. After overnight incubation at 30°C, the plates were screened for colony morphologies common to either B. anthracis (flat and matte) or B. cereus (shiny, gray-white). Colonies of interest were then subjected to the phenotypic and genotypic analyses shown in Table S1. A strain was considered B. cereus if it was β-hemolytic, fosfomycin-sensitive, and PCR-positive using primers directed against plcR, BC5101, BC5449, and BC0442 of ATCC 14579. A strain was considered B. anthracis if it was not hemolytic, resistant to fosfomycin, and was PCR-positive with primers for virulence plasmid loci (ggt, lef, cya, and pagA), each of the chromosomal prophage (BA4067, BA5353, BA0443, and BA3760), and the marker Ceb-Bams 30 [100]. Surface binding to GFP-PlyG^BD and biofilm formation was assessed as described here. Prophages from the indicated strains in Table S1 were either spontaneously released into culture (Htp1) or induced by growth in BHI cultures supplemented with 150 µg ml⁻¹ of fosfomycin [46]. Plaque-forming units were identified on a ΔSterne reporter strain. Strain RS1255 was isolated from a human tonsil that was acquired as part of an Internal Review Board (IRB)-approved protocol.

RNA isolation and RT-PCR

RNA was extracted using RNeasy mini kits according to the manufacturer's protocol with the exception that the PlyG lysin [99] was added during lysis to improve RNA yield. An on-column DNase digest was performed using the RNase-free DNase set (Qiagen), followed by a second treatment with RNase-free DNase (Ambion, Inc.). 0.5 µg of RNA was then subject to a first strand cDNA synthesis using the Superscript III First Strand Synthesis Kit (Invitrogen, Inc.). The cDNA was analyzed with primer pairs indicated in Table S7.

Biofilm formation and analysis

The B. anthracis biofilms were established from overnight 5 ml BHI cultures (with 0.2% glucose) that were washed twice in PBS, resuspended in 5 ml of PBS, and diluted 1∶1000 into 10 ml of BHI. After incubation for 3 months at room temperature in the dark without aeration, the cultures were photographed (Nikon CoolPix 5400) and processed for analysis. First, small biofilm sections were removed with a 200 µl pipet-tip, stained with GFP-PlyG^BD on a glass microscope slide, and analyzed by microscopy. Next, the liquid phase was carefully removed and biofilms were recovered into PBS and vortexed to resolve aggregated material. The samples were then plated for viability both before and after incubation at 95°C for 15 minutes.

The formation of biofilms by B. cereus RS1045 and its phage-cured derivative was examined using a different method. Five ml cultures were established in BHI containing 0.2% glucose and incubated overnight at 30°C with aeration. The next day, cultures were washed twice, resuspended in 5 ml LD medium, and used to inoculate, at a dilution of 1∶1000, 3 ml LD cultures in 12-well multi-well plates (non-treated polystyrene, BD Falcon^TM). Sterile 22 mm² glass coverslips (Fisherbrand) were then added to each well and incubated for three weeks without aeration in the dark. The coverslips were removed, washed gently with PBS, stained with crystal violet, and examined by microscopy.

Soil microcosms

For the preparation of soil microcosms, 200 g of potting soil (Miracle-Gro^®) was first added to 1 liter of dH₂0 and autoclaved twice. Ten ml aliquots of the resulting slurry were transferred to 50 ml Falcon™ tubes and the indicated bacterial strains (5×10⁸ stationary-phase cells suspended in 0.3 ml dH₂0) were added. All strains used here, including the lysogens, were transformed with pASD2 to facilitate recovery. The inoculated soil columns were then incubated at room temperature in the dark without aeration. At the indicated time points, 1 ml samples were removed, vortexed for 2 minutes and plated on BHI with antibiotics for viability both before and after heat-treatment at 95°C for 15 minutes. For the analysis of phage shedding, vortexed samples were also pelleted by centrifugation and supernatants were removed, filtered with 0.2 µM membranes, and titered on ΔSterne to determine PFUs ml⁻¹ of culture.

For the analysis of phage infection in co-cultures, soil microcosms were prepared and inoculated with 5×10⁸ of the indicated B. anthracis lysogens (phage “donor” strains). None of the donor strains were antibiotic resistant. For some microcosms, 1×10⁸ Wip1 phages were added in lieu of donors. After 24 hours, 5×10⁸ each of the B. anthracis strains Sterne/pASD2 or ΔSterne/pASD2 (phage “recipient” strains) were added. The co-inoculated microcosms were incubated at room temperature in the dark without aeration. At the indicated times, samples were pelleted, resuspended in 5 ml PBS, and plated on BHI agar with antibiotics. Where indicated, samples were plated both before and after heat-treatment at 95°C for 15 minutes. Resulting colonies were subcultured and analyzed by PCR with phage-specific primers to evaluate infection with the indicated phages. Free phages in soil aliquots were determined by titering sterile-filtered supernatants on ΔSterne to determine PFUs ml⁻¹ of culture.

Earthworm microcosms

For earthworm colonizations, animals were washed twice in dH₂0 and individually transferred using sterile forceps into separate 50 ml tubes containing ∼20 g potting soil (Miracle-Gro^®). The worms were incubated at 24°C for 2 days, before 5×10⁹ stationary phase bacteria were added as suspensions in 5 ml dH₂0. All strains used here, including the lysogens, were transformed with pASD2 to facilitate recovery. After 2 weeks, the worms were recovered from infected microcosms, washed twice in dH₂0, and added to fresh soil. The worms were subsequently washed and placed in new soil every two weeks for 6 months. Ten separate microcosms were established for each strain. At indicated time points, worms were recovered, washed in dH₂0, and their intestinal contents were forced out with sterile Pasteur pipettes. Their contents were suspended in 1 ml PBS, vortexed, and plated for viability on BHI with antibiotics both before and after heat-treatment at 95°C for 15 minutes.

For the analysis of phage infection during co-culture, earthworm microcosms were established and inoculated with B. anthracis lysogens (donors). None of the donor strains were antibiotic resistant. After 2 weeks, the worms were recovered, washed twice with dH₂0, and returned to fresh soil. After one week, the worms were again washed and cycled through fresh soil for another week. Worms were then introduced into fresh soil and inoculated with either of the phage recipient B. anthracis strains ΔSterne/pASD2 and Sterne/pASD2 or 1×10⁸ Wip1 phages. After one week, the worms were then subjected to two week cycles of recovery, washing and re-introduction into fresh soil for a period of 6 months. At the indicated times, worms were recovered and the intestinal contents were removed and examined as above.