Overnight cultures of P. larvae ATCC 9545 grown in 1/2 × liquid porcine brain heart infusion (PBHI) (Acumedia, Lansing, MI, United States) media were grown in a shaking incubator at 37°C. The optical density of the culture at 600 nm was used to estimate the number of cells per milliliter as we have reported previously (Brady et al., 2017). A total of approximately 10⁴ bacterial cells were spread onto tryptic soy agar plates with glass beads and allowed to incubate at 37°C and 5% CO₂ for 8 days. Incubated plates were doused with five ml of cold sterile ddH₂O and allowed to sit for 15 min. Colonies on the plates were gently scraped off the plate and into suspension with sterile loops. The suspensions from eight plates were combined into a 50 ml tube and centrifuged at 12,000 × g for 20 min. Supernatant was poured off and the pellet was resuspended in 40 ml of sterile ddH₂O and centrifuged again as a wash step. The pellet was washed two more times. After the last wash step, the pellet was resuspended in 80% EtOH.

Spores were removed from the ethanol immediately prior to any experiment. Spores suspended in ethanol were centrifuged at 12,000 × g for 5 min to pellet the spores and the supernatant containing alcohol and the lysed debris from dead vegetative cells were discarded with the supernatant. The pellet was washed three more times using sterile 1/2 × PBHI broth and then suspended to a concentration of 10⁴ colony forming units (CFU) per ml. Spore purity was also confirmed using the Schaeffer-Fulton staining method: briefly, samples were heat fixed, stained with 5% malachite green for 5 min over heat, and counterstained with 0.2% safranin (Schaeffer and Fulton, 1933). Spore viability and purity tests were done with samples in triplicate. The malachite green and safranin staining was done to verify spore presence and identify any residual vegetative bacteria in the sample. At least 100 cells from each sample were observed and counts taken for the number of spores, spores-in-mother cells, and vegetative cells.

Phages specific for P. larvae were previously isolated and confirmed to infect and lyse only P. larvae (Brady et al., 2017; Stamereilers et al., 2018) and not B. laterosporus (Brady et al., 2017; Berg et al., 2018; Brady et al., 2018). Phages were tested on field isolates of P. larvae. Each field isolate for phage characterization was identified as P. larvae by confirming gram test, catalase test, 16s sequencing, and by KAT PCR and ERIC PCR. Primers and citations for these methods are indicated in Table 1. Protocols are published in our previous works (Brady et al., 2017; Berg et al., 2018; Brady et al., 2018). The phage, PL.Ph-1, was previously isolated from a dead feral beehive in Utah that appeared to have died from AFB. PL.Ph-1 forms lytic plaques on the laboratory standard P. larvae, ATCC 9545, as well as on 55 of 59 verified field isolates of P. larvae as previously reported (Brady et al., 2017). Phage lysate of PL.Ph-1 [the phage also indicated as phage 1 in Brady et al. (2017)] was prepared for these studies by reconstituting a freezer stock sample of the phage in 500 μL of overnight P. larvae and plating the solution in 1/2 × PBHI top agar and left to incubate at 37°C. After 24 h, visible plaques were selected from the plate, suspended in 25 ml of 1/2 × PBHI broth to which was added 1 × 10⁶ CFU of P. larvae and incubated, shaking at 37°C. After 16 h the lysate was filtered through a 0.22 μM filter (VWR, Radnor, PA, United States). Lysate concentration was determined using standard titration techniques as previously described to determine plaque forming units (pfu) per ml (Brady et al., 2017). Titered lysate of PL.Ph-1 was used for all studies in this paper.

Unconjugated FITC was added to a high titer phage lysate (10¹¹ pfu) suspended in 1× HEPES solution (pH 7.4) to obtain a concentration of 31.25 μg/ml FITC and was allowed to incubate for 1 h. The high titer lysate was ultracentrifuged at 25,000 × g for 1 h to pellet the phages. The supernatant containing unbound FITC was poured off and the pellet resuspended in HEPES solution to a FITC concentration of 15.625 μg/ml to avoid background staining of bacteria as indicated in the results.

For flow cytometry analysis, bacterial samples were loaded into a 96-well plate containing approximately 5 × 10⁴ CFU in each well. Each well received 200 μL of FITC-labeled phages. Cell fluorescence was measured by a Cytoflex S flow cytometer and a minimum of 50,000 cells were counted per sample. Three experiments were run on separate weeks with three replicates for each sample in each experiment.

Beckman Coulter CytExpert software was used to analyze the flow cytometry data collected on the Beckman Coulter Cytoflex S flow cytometer. Three gates were individually set using unstained samples of each bacterial type using channels for Forward Scatter (FSC), Side Scatter (SSC) and FITC. The gates were set on FSC × SSC to exclude debris, FSC-HxFSC-A to isolate single cells, and on the FITC channel to identify positive or negative samples.

Vegetative P. larvae and P. larvae spores (5 × 10⁵ CFU) were resuspended in 1 ml of 3 × 10⁹ pfu/ml high titer lysate and allowed to incubate for 1 h. The phage-treated spores were pelleted at 8,000 rpm for 6 min. The supernatant was poured off and the pellet was resuspended in 40 μL of 1× HEPES solution.

Phage/spore samples were incubated with formvar coated copper grids for 60 s and then incubated with 50 μL of 2% uranyl acetate (pH 7) for 60 s. Moisture was wicked away from the grids and then allowed to air dry prior to imaging. Electron micrographs were taken by the BYU Microscopy Center on a Verios STEM machine (Abràmoff et al., 2004).

The Brady Binding Assay was developed in our lab to identify the ability of a phage to reversibly bind to a test material (such as a spore or unrelated bacteria) and to remain viable against its original target. In brief, the assay is setup by incubating the phage with the test bacterium, transferring the sample onto a filter, and then rinsing the trapped bacteria to remove un-bound phages. The trapped, rinsed, bacteria are incubated with bacteria of the original phage target and a standard plaque assay is done. Brady binding assays for this study were setup as follows: overnight cultures of P. larvae ATCC 9545, B. laterosporus field isolate B-2, Sinorhizobium meliloti strain B100, and P. larvae ATCC 9545 spores were each diluted to 10⁴ CFU/ml. The Brady binding assay was setup each time using an MOI of 10⁴ with each sample in triplicate. The bacteria were pelleted, supernatant discarded, and the pellets resuspended in 1 ml of phage lysate at a titer of 10⁸ pfu/ml, control samples of phage lysate were resuspended in 1 ml of sterile 1/2 × PBHI broth without bacteria, and all samples were incubated for 30 min at room temperature. Each solution was filtered using single-use 0.22 μM vacuum filter to remove all bacteria. The filters were then rinsed with 1 L of 1× phosphate buffered solution (PBS) to release any phages that were not bound to bacteria. The filters were removed, placed in tubes containing 1 ml of 1/2 × PBHI broth, and vortexed for 1 h to dislodge bacteria from the filter. After vortexing, a standard plaque assay was setup using 100 μL of sample (or diluted sample) to incubate 5 or less minutes with 500 μL of overnight P. larvae ATCC 9545, plate in 1/2 × PBHI top agar, and incubate overnight. The resulting plaques were counted and data reported as the average ± standard error of the mean. Data was converted into percentages to compare with flow cytometry results by first dividing the average number of plaques from the positive control group in the Brady assay by the percentage of FITC-positive cells in the corresponding positive control group from flow cytometry. The resultant number was used as the denominator to convert the average number of plaques from each group in the Brady assay into a percentage.

Data were analyzed using SAS software (SAS Institute Inc., Cary, NC, United States) and the Mixed Procedure method to generate p-values, standard deviation, standard error and to determine statistical significance (for Figures 4, 8). For direct count statistics in 3.1, we used Jeffery’s 95% confidence interval (Brown et al., 2001) for binomial proportions. For all experiments α = 0.05.

Figure 1 includes an electron micrograph of phage PL.Ph-1 and images of plates demonstrate the lytic nature of the phage used in these studies. PL.Ph-1 is a siphovirus with a prolate head and is the same phage labeled #1 in studies by Brady et al. (2017) which was used in the phage cocktail for hive treatments in that same paper. This phage was selected for further study because, as previously reported, it was the only phage that readily formed plaques on more than 90% of the field isolates to which it was challenged compared to the other 38 phages that formed plaques on an average of only 39.4 ± 2.4% of field isolates (Brady et al., 2017). Challenge of the PL.Ph-1 phage on multiple strains of B. laterosporus did not yield plaques, which was not surprising since none of the other 38 isolated P. larvae phages formed plaques on B. laterosporus when tested in our lab.

In order to study phage binding to spores, the P. larvae spores to be used for the study were rigorously tested to ensure purity of the prepared spores. A pure spore sample is free of vegetative cells and composed only of spores with few spores-in-mother cells. No vegetative cells were identified in the spore samples using this method and free endospore comprised 92 ± 2.1% of the cells. The samples contained 8.0 ± 2.6% of spores that had not released from their mother cell.

Spore samples were incubated in 1/2 × PBHI broth to further confirm the spore purity of the samples. The purpose of this study was to verify that the spore sample used for subsequent studies did not contain vegetative cells and would not germinate spores in the time span of the experiments. Results of this study are presented in Figure 2. Positive controls of vegetative bacteria were incubated with starting concentrations of 10⁶, 10⁵, and 10⁴ CFU/ml. Spores had an approximate concentration of 10⁵CFU/ml. The optical density of the spore samples did not change significantly over 17 h in comparison to the vegetative P. larvae samples at 10⁶, 10⁵, and 10⁴ CFU/ml over the same amount of time. Spores generated from strain ATCC 9545 do not germinate in 1/2 PBHI liquid media without at least previously being heat activated for 30 min at 70°C (Alvarado et al., 2013); therefore, any increase in optical density of an incubated sample in our studies would result from vegetative bacterial growth in the sample and not from spore germination. These data indicate that the spore samples did not contain any detectable amounts of vegetative bacteria. All studies were conducted with 17 or fewer hours of incubation.

The ethanol-wash treatment used to prepare spores was also tested on vegetative P. larvae to verify that the ethanol treatment had killed any surviving vegetative cells in the spore samples. No colonies formed from ethanol wash samples. Spore samples were plated for germination to ensure their viability. After a 48-h incubation, colonies formed on spore-inoculated plates and the colonies were confirmed to be P. larvae.

Use of flow cytometry to measure phage binding has not previously been reported. We anticipated that phage binding could be quantified using flow cytometry because others have demonstrated that phages could be stained with FITC fluorochrome and then visualized by confocal microscopy (Kelly et al., 2006; Puapermpoonsiri et al., 2010). A flow cytometer takes a reading of each cell in a suspension and reports the fluorescence intensity of each individual cell. Flow cytometry can produce rapid, quantitative results from thousands of individual bacterial cells in each sample while ignoring free-floating phages. We first needed to establish the highest concentration of FITC stain that would not generate background staining on the bacteria. Bacterial samples were incubated with different concentrations of FITC stain to be used for phage staining (Figure 3). These results indicated that the bacteria should not be exposed to FITC solutions of 31.3 μg/ml or greater, otherwise background staining of the bacteria would be observed in the flow data and could complicate interpretation in samples containing stained phages. The experiment proceeded with phages in a solution containing 15.6 μg/ml of FITC to be incubated with the test bacteria.

As depicted in Figure 4, data analysis of flow cytometry results utilized a series of three plots to identify whether or not the fluorescent phages bound to bacteria. The first two plots were used to select the population to analyze (Figures 4A,B) and the third plot reported the number of cells with each intensity of FITC staining (Figure 4C is a negative sample and Figure 4D is a positive sample). On the histogram of the FITC reading, cells that are positive indicate bacteria with phages attached.

FITC-stained phages were mixed with vegetative P. larvae, B. laterosporus, S. meliloti, and spores of P. larvae. The average percentage of FITC-positive bacterial cells for each is reported in Figure 5. P. larvae, B. laterosporus, and P. larvae spores treated with labeled phages are significantly different (p < 0.0001) from untreated samples where S. meliloti treated with phages did not have a statistical difference from an untreated sample (p = 0.3725). Table 2 contains the p-values for statistical analysis of data from the last four columns in Figure 5 to compare bacterial sample containing phages with each other. Phage binding to S. meliloti and to B. laterosporus is significantly different compared to phage binding to P. larvae, regardless of whether it is vegetative or spore forms of P. larvae. Phage binding to S. meliloti is not statistically different from phage binding to B. laterosporus. Phage binding is also not statistically different when comparing vegetative and spore forms of P. larvae to each other. These data support the hypothesis that P. larvae phages bind to spores.

Electron microscopy was used to visualize phage-binding to spores. Figure 6 includes electron micrographs of vegetative P. larvae incubated with phages. Phages were observed in horizontal and vertical positions with respect to the bacterium. Most commonly, the phages were attached by the end of the tail fiber to the side of the vegetative bacterium. Visible damage occurred to the bacterial wall upon infection, and late-stage infection was observed as an eruption from the bacterial wall.

Figure 7 includes electron micrographs of P. larvae spores incubated with phages. Phages were observed in horizontal and vertical positions on the spores. Spores that had not yet released from the mother cell were equally prone to have phages attached, and the phages appeared equally able to attach to the spore side as to the mother cell side (Figure 6A). Phages were observed in both horizontal and vertical positions on spores. Some phages clearly attached to the spore by the end of the tail fiber (Figure 6B). Phages appeared to be attached in higher abundance on vegetative cells than on spores as observed in electron micrographs. Our imaging methods did not distinguish between empty or full capsids, so no interpretation can be made regarding whether or not a phage injected its DNA into a spore or not.

The Brady Binding Assay includes a short incubation period with phages and a long rinse of the bacteria prior to a plaque assay (Figure 8). Since the thoroughly rinsed bacteria are used as the source of phages for the plaque assay, plaques indicate that binding to the challenge bacteria occurred during the initial short incubation period. Phages can produce plaques in two ways: (1) the phages are able to productively infect challenge bacteria during the first step and then produce plaques on their intended host in the plaque assay, or (2) the phages can exhibit reversible binding wherein phages bind to the challenge bacteria in the first step and release to infect and produce plaques on their intended host in the plaque assay.

Results of the Brady assay are presented in Figure 9 and Table 3. The “phages only” control indicates the resultant number of phages that become mechanically trapped on the filter during the Brady assay. A statistically significant number of plaques formed from all test bacteria samples compared to the phage only samples. By comparison, P. larvae spores formed plaques significantly higher than the unrelated S. meliloti and statistically lower than vegetative P. larvae. The number of plaques from B. laterosporus was surprisingly not statistically different from that of vegetative P. larvae, nor was it statistically different from that of P. larvae spores (α = 0.01). These results indicate that P. larvae phages bind with significant abundance to vegetative P. larvae, P. larvae spores and to vegetative B. laterosporus.

Results from the Brady assay were converted into percentages so that the data could be compared between the two quantitative methods used in our research. Table 4 provides a comparison between flow cytometry data and Brady assay results and summarizes a proposed interpretation of these data.