Isolation, host range, and morphology

JG068 (vB_BceP_JG068) was isolated from a sewage processing plant using an uncharacterized strain of Burkholderia dolosa. Additional JG068 hosts were identified using spot tests and soft-agar overlays of BCC strains. A total of 32 strains were tested and seven were found to support lytic propagation of JG068 (with efficiency of plating [EOP] values in parentheses): Burkholderia multivorans ATCC 17616 (10^-4), Burkholderia cenocepacia K56-2 (10⁰), J2315 (10^-2), and PC184 (10⁰), Burkholderia stabilis LMG 14294 (10^-1), and Burkholderia dolosa AU0158 (10⁰) and CEP021 (10⁰). Excluding the soil isolate ATCC 17616, each of these strains was identified as a CF clinical isolate [19–21]. Furthermore, the three susceptible B. cenocepacia strains and AU0158 have all been linked to epidemic spread among CF patients [19, 21]. This host range is relatively broad, clinically relevant, and distinct compared to the tropism of BCC phages that we have previously characterized [10, 11, 22–26]. On K56-2, JG068 forms large clear plaques, 1–3 mm in diameter.

Electron microscopy of JG068 virions (Figure 1) shows that this phage has a short tail and belongs to the C1 morphotype of the order Caudovirales and family Podoviridae[29]. This morphology is relatively rare for a BCC phage as the only Burkholderia podoviruses identified to date are the BPP-1-like BcepC6B and the Bcep22-like DC1, Bcep22, BcepIL02, and BcepMigl [14]. The JG068 capsid is icosahedral and 60.93 ± 2.83 nm in diameter (Figure 1).

Genome sequencing and assembly

In order to determine if the genome sequence of JG068 was novel, several EcoRI genomic DNA fragments were cloned into pUC19 and sequenced. Resulting reads were analyzed using BLASTN and found to be similar (but not identical) to a variety of sequences, including those from previously characterized podoviruses. The genome sequence was completed using ion semiconductor technology on the Ion Torrent platform. A total of 1.07 × 10⁵ sequence reads were aligned into a single contig with over 400-fold coverage. Regions of ambiguity and genome ends were analyzed using Sanger sequencing.

The JG068 genome is 41,604 bp in length and has a 60.7% GC content. Based on the similarity of the sequence to characterized ϕKMV-like Autographivirinae genomes (discussed below), the chromosomal termini of JG068 are likely to be short direct terminal repeats (DTRs) as canonically found in T7 [30]. To identify these termini, JG068 DNA was directly sequenced using primers that bind the following four loci: within the left DTR, extending to the left terminus; immediately outside of the left DTR, extending to the left terminus; within the right DTR, extending to the right terminus; and immediately outside of the right DTR, extending to the right terminus. Primers that bind within the left DTR will amplify both the left and right ends of the genome simultaneously (as the primer binding site is in both DTRs). A reduction to one-half intensity in the sequence chromatogram represents the arrest of half of the reads (those originating from the left end of the genome) at the left chromosome terminus [31]. Primers that bind immediately outside of the left DTR will amplify only the left end of the genome (as the primer binding site is unique) and the sequence chromatogram will stop abruptly at the left chromosome terminus [30]. The same reasoning is true for primers binding within or near the right DTR. Using these methods, it was determined that the chromosomal termini of JG068 are 216 bp DTRs with an identical sequence on each end. The length of these repeats is shorter than that of other ϕKMV-like Autographivirinae such as LKD16 (428 bp), ϕKMV (414 bp), LKA1 (298 bp), and LIMElight (277 bp) [32–34].

In ϕKMV, genome replication is bidirectional with the origin and terminus in the DNA polymerase gene and an internal virion protein gene, respectively [33]. Using GenSkew analysis, a global co-minimum GC skew was identified in the JG068 DNA polymerase gene 18 and a global maximum skew was identified in the internal virion protein gene 38, suggesting that the replication process of JG068 is similar to that of other ϕKMV-like Autographivirinae. In addition to the chromosomal termini and the putative replication origin, a third commonality with respect to the DNA of this genus of phages is that the JG068 genome lacks recognition sites for many common restriction enzymes, including BamHI, BglII, PstI, SacI, SmaI, and XhoI. As in other phages, including those that are ϕKMV-like, these sites tend to be lost as the phage evolves to avoid host restriction systems [32, 33].

Relatedness

When the sequence of JG068 is analyzed using BLASTN, the most similar sequences (with E-values of 2e^-87 or less) are those of Ralstonia phage ϕRSB1, a genetic locus of blood disease bacterium R229, a genetic locus of Burkholderia pseudomallei 1710b chromosome 1, and Caulobacter phage Cd1. In the BLASTN results, JG068 shows similarity to several members of the Autographivirinae subfamily and ϕKMV-like phages genus, including ϕRSB1, Cd1, and Pseudomonas phage Bf7. Members of this subfamily (with enterobacteria phage T7 being the best characterized) belong to one of several genera, including the T7-like phages, SP6-like phages, ϕKMV-like phages, and novel genera that have not yet been named [35, 36]. These podoviruses all encode a single subunit RNA polymerase, as is found in JG068 [35]. To determine if JG068 belongs to this subfamily (and, if so, to what genus), we used CoreGenesUniqueGenes (CGUG) analysis to compare the genomes of JG068, T7 (NC_001604.1), SP6 (NC_004831.2), and ϕKMV (NC_005045.1) [37]. Using T7, SP6, or ϕKMV as the reference genome, the proteins of JG068 were 15.0%, 23.1%, or 44.9% similar, respectively. As ≥40% similarity indicates a genus-level relationship [35], we can conclude that JG068 is a novel member of the Autographivirinae subfamily and the ϕKMV-like phages genus. As noted above, the few BCC podoviruses that have been previously characterized are either Bcep22-like or BPP-1-like [14], thus JG068 is the first BCC Autographivirinae phage to be identified.

Genome annotation

The JG068 genome contains 49 putative open reading frames (ORFs) (Figure 2, Table 2). Similar to T7 and other Autographivirinae, all of the genes are transcribed in the forward direction [33, 38]. Based on BLASTP analysis (with an E-value cutoff of 0.01), 19 JG068 proteins show no similarity to other proteins in the database (Table 2). The remaining 30 proteins have percent identities between 29% (Rz protein gp48, similar to hypothetical protein Bpse9_41836 of B. pseudomallei 91) and 64% (hypothetical protein gp8, similar to a hypothetical protein of Ralstonia phage RSB2) (Table 2), thus showing low-to-moderate similarity to other sequences at the protein level.

A module commonly identified in other BCC phages that is notably absent from the JG068 genome is that for lysogeny. Similar to other obligately lytic phages, JG068 forms clear plaques and does not encode either an integrase or a repressor. Most other characterized members of the Autographivirinae are also virulent (although putative prophage elements described as being T7-like have been identified in Xanthomonas axonopodis, Pseudomonas putida, and B. pseudomallei) [39, 40]. In order to verify experimentally that JG068 was not capable of lysogenizing BCC strains, we isolated JG068-insensitive K56-2 from a JG068/K56-2 lysate. To determine if these cells were insensitive due to either receptor mutation or superinfection immunity, we attempted to lyse these isolates using a second putatively virulent BCC-specific phage, KS12. Previous assays have shown that KS12 uses LPS as a receptor [Juárez Lara, unpublished data]. If JG068-insensitive K56-2 isolates are also insensitive to KS12 infection, they are likely LPS mutants, whereas if they are sensitive to KS12 infection, they are likely JG068 lysogens. Over 100 JG068-insensitive isolates were screened and all were found to be insensitive to both phages compared to wildtype K56-2. Twelve resistant colonies were PCR-screened using JG068-specific primers and each isolate was found to be amplification-negative. This evidence shows that resistance arises due to receptor mutation and not due to JG068 lysogeny.

In Autographivirinae, transcription is first initiated from host promoters found at the far left-hand side of the genome, followed by initiation from phage RNA polymerase-specific promoters found throughout the genome [33, 38]. JG068 promoters putatively recognized by BCC RNA polymerase were identified using Neural Network Promoter Prediction. Using a cutoff of 0.95 and limiting the results to sequences found intergenically, three promoters were identified in the first 650 bp downstream of the left DTR (Figure 2 and Figure 3A). Phage promoters lack the conserved structure observed in bacterial promoters with -10 and -35 regions and are instead described as short consensus sequences that vary among different phages [41]. PHIRE was used to identify a strongly conserved 16 bp consensus sequence (Figure 3B) found six times at five intergenic loci in the JG068 chromosome: overlapping the first host promoter sequence, upstream of the DNA polymerase gene 18 (2 promoters), upstream of the hypothetical protein gene 28 (downstream of the DNA-binding module), upstream of the capsid gene 34, and upstream of the tail fiber gene 40 (Figure 2). Putative rho-independent terminators were identified using TransTermHP. Although multiple sequences were reported in the output, those shown in Figure 3C were chosen because they were intergenic (based on GeneMark predictions) and had a ΔG value of -10 kcal/mol or less. Similar to LIMEzero, the putative terminators are found upstream of the DNA-binding module and downstream of the capsid protein (Figure 2) [34].

Module analysis

Lysis proteins

The lysis proteins of JG068, encoded on the far right end of the genome, include a putative pinholin, signal-arrest-release (SAR) endolysin, Rz, and Rz1. Although the lysis genes in most phages (including ϕKMV) tend to be arranged in a single block with the holin gene followed by the endolysin, Rz, and Rz1 genes [43], this order is not maintained in JG068. Here, the putative pinholin gene 42 is found upstream of the SAR endolysin (47), Rz (48), and Rz1 (49) genes, separated by genes encoding two DNA packaging proteins and two hypothetical proteins (Figure 2). Within the Autographivirinae subfamily, this gene arrangement is also observed in SP6 [44].

Although only two of the four JG068 lysis proteins are similar to those of ϕKMV based on CGUG analysis (the SAR endolysin gp47 and Rz gp48), these two phages are likely to use similar lysis mechanisms. The gp42 pinholin has analogous features to that of ϕKMV: ~60 amino acids in length (60 for JG068 and 66 for ϕKMV), two transmembrane domains (as predicted by TMHMM analysis), and positively-charged arginine and lysine residues at the C-terminus [43]. The gp48 Rz inner membrane protein has a single N-terminal transmembrane domain based on TMHMM analysis. Gene 49, encoding the Rz1 outer membrane lipoprotein, overlaps with gene 48 in the +1 reading frame and extends downstream of the 48 stop codon by 148 base pairs (Figure 2). LipoP analysis predicts a signal peptidase II cleavage site in gp49 between residues 18 (alanine) and 19 (cysteine).

The SAR endolysin of JG068 has a very similar organization to that of ϕKMV. SignalP 3.0 predicts an N-terminal signal sequence probability of 0.97 for gp47 (later versions of SignalP do not make the same prediction, likely because the C-region is absent). The gp47 N-terminus contains two signal sequence regions followed by three putative catalytic residues of the lysozyme domain. Residues 1–7, including a positively-charged histidine, lysine, and arginine, make up the N-region. Residues 8–25 make up the hydrophobic glycine- and alanine-rich H-region, also predicted to be a transmembrane domain by OCTOPUS analysis. Using CD Search and BLASTP alignment with the ϕKMV SAR endolysin, the putative lysozyme catalytic residues were identified as 27E, 36D, and 45T, all found immediately downstream of the H-region. As in ϕKMV, because the C-region is absent, the protein will not be cleaved by a signal peptidase upon secretion and will instead remain associated with the inner membrane until release into the periplasm [43].

Using inducible expression of the ϕKMV SAR endolysin in E. coli, it was shown that this protein could decrease culture absorbance in the absence of a pinholin, but only if the N-terminal signal sequence of the endolysin was present [43]. We performed a similar experiment using the JG068 SAR endolysin to further characterize the export mechanism and biological activity of this protein. To allow for tightly controlled inducible expression, we cloned gene 47 into pET22b (with [gp47] or without [gp47ΔSS] the putative signal sequence in the first 25 residues) and transformed these plasmids into E. coli BL21(DE3)pLysS. When these cells (or a pET22b blank control) are subcultured, their growth rates prior to IPTG induction are similar based on optical density measurements at 600 nm (OD₆₀₀; up to 3 h in Figure 4). However, following induction, expression of gp47 is lethal to the cells as the OD₆₀₀ decreases from ~0.6 at 3 hours to ~0.45 at 8 hours (black squares, Figure 4). A very different trend is observed for the blank control, gp47ΔSS, and uninduced gp47 strains, where the OD₆₀₀ increases from ~0.6 at 3 hours to ~0.9 at 8 hours, double that of the induced gp47 OD₆₀₀ (Figure 4). As the lytic activity is dependent upon expression of not only the lysozyme domain but also the signal sequence, we can conclude that JG068 gp47 acts as a typical SAR endolysin in Gram-negative bacteria. While the classical endolysins of Bcep781/Bcep43 and BcepC6B have been functionally characterized [15, 45], this is the first experimental evidence for SAR endolysin activity in a BCC phage. As these proteins have also been identified in the Bcep22-like viruses [25, 46], similar experiments may be used to confirm the activity of SAR endolysins in these phages as well.

In vivo activity

The Galleria mellonella (greater wax moth) larvae model is commonly used to assess both strain virulence and phage therapy efficacy for members of the BCC, particularly B. cenocepacia[10, 11, 18]. Although this model is less complex than a mammalian system, it shows positive correlation with both mouse and rat models of infection [18]. To determine if JG068 possesses lytic activity against B. cenocepacia in vivo, we infected G. mellonella larvae with 3.7 × 10³ colony forming units (CFU) of K56-2 and treated with endotoxin-removed JG068 at a multiplicity of infection (MOI) of 350 (1.3 × 10⁶ plaque forming units [PFU]). For infected, untreated controls, no larvae survived after 72 hours (Figure 5, left). Those larvae that received only JG068 remained healthy over the course of the experiment (Figure 5, centre), indicating that phage treatment produced no harmful effects. For infected, treated larvae, JG068 administration significantly increased survival to an average of 77% (from 0% in untreated controls) (Figure 5, right). In a previous study, the putatively virulent phage KS12 was the most effective for immediate treatment of K56-2 infection in the G. mellonella model, resulting in 93 ± 12% survival (MOI = 5000) or 57 ± 6% survival (MOI = 500) [10]. Based on the G. mellonella data presented in Figure 5, JG068 is almost as effective as a significantly larger dose of KS12, indicating that JG068 is highly active in vivo.

Bacterial strains and culture conditions

BCC strains were grown aerobically at 30°C overnight on half-strength Luria-Bertani (½ LB) solid agar or agarose plates or in ½ LB broth. Strains used for host range testing and phage propagation are members of the original and updated BCC experimental strain panels [19, 20]. K56-2 LPS mutants [27, 28] were grown at 30°C on ½ LB plates or in ½ LB broth containing 100 mg/l trimethoprim. E. coli DH5α transformed with pUC19 or pET22b was grown at 37°C on LB plates containing 100 mg/l ampicillin. E. coli BL21(DE3)pLysS was grown at 37°C in LB broth containing 25 mg/l chloramphenicol or, if transformed with pET22b, 25 mg/l chloramphenicol and 100 mg/l ampicillin.

Phage propagation, host range analysis, lysogeny screen, and electron microscopy

JG068 was isolated from a sewage plant in Steinhof near Braunschweig in Germany following an enrichment protocol previously described [47]. For propagation of JG068, 100 μl high-titre phage stock and 100 μl BCC liquid culture were combined and incubated 20 minutes at room temperature, mixed with 3 ml 0.7% ½ LB agar or 0.6% ½ LB agarose, overlaid on a ½ LB plate, and incubated at 30°C overnight. High-titre stocks were made in modified suspension medium (modified SM; 50 mM Tris–HCl [pH 7.5], 100 mM NaCl, 10 mM MgSO₄). Phage plates were overlaid with 3 ml modified SM and incubated >1 h at room temperature on a platform rocker. The supernatant was recovered, pelleted by centrifugation for 2 min at 10,000 × g, filter-sterilized using a Millex-HA 0.45 μm syringe-driven filter unit (Millipore, Billerica, MA), and stored at 4°C.

Spot testing was used for preliminary host range analysis. Soft-agar overlays containing 100 μl BCC liquid culture were allowed to solidify for ≥10 minutes at room temperature, spotted with 10 μl drops of high-titer JG068 stock, and assayed for clearing and/or plaque formation after incubation at 30°C overnight. Strains that showed phage susceptibility in this assay were re-tested in soft-agar overlays containing both the strain and JG068. The host ranges for LPS mutants of K56-2 [27, 28] and PC184 [Abdu, unpublished data] were tested similarly. Efficiency of plating (EOP) values were determined as previously described [46].

To assay for JG068-lysogenized K56-2, JG068 and K56-2 were plated in agar overlays at a multiplicity of infection of 10⁰ and incubated at 30°C overnight. To recover surviving cells, plates were overlaid with 3 ml sterile water and incubated 1 h at room temperature on a platform rocker. Cells were pelleted by centrifugation for 2 min at 10,000 x g, washed and resuspended in sterile water, and plated to obtain isolated colonies. Colonies were struck out onto ½ LB plates and spotted with separate 10 μl drops of high-titer JG068 stock and high-titer KS12 stock. Isolates were scored as sensitive or insensitive compared to a sensitive wildtype K56-2 control that was lysed by both phage stocks. A subset of resistant isolates was replated and PCR-screened for lysogeny using JG068-specific primers (9F: GAACATCGGTAACGTCGTCAAGG; 9R: GGCGTGACGAACAGCTTGGC). TopTaq DNA polymerase and buffers (Qiagen, Hilden, Germany) were used according to the manufacturer’s instructions (template: 1 μl overnight culture incubated 5 min at 100°C).

For electron microscopy, phage stocks were prepared as described above with the following modifications: agarose plates and soft agarose were used for overlays, sterile water was used in place of modified SM, and a 0.22 μm filter was used for syringe-driven filtration. A carbon-coated copper grid was incubated with lysate for 5 min and stained with 2% phosphotungstic acid for 30 s. Transmission electron micrographs were captured using a Philips/FEI (Morgagni) transmission electron microscope with charge-coupled device camera at 80 kV (University of Alberta Department of Biological Sciences Advanced Microscopy Facility). The capsid diameter (mean ± standard deviation) was calculated using Microsoft Excel based on measurements from nine individual virions.

Phage DNA isolation, RFLP analysis, and sequencing

Phage DNA was isolated using a modified version of a Wizard DNA Clean-Up protocol [48]. Ten milliliters of filter-sterilized JG068 lysate (propagated using K56-2 on agarose medium) was treated with 10 μl DNase I (Thermo Scientific, Waltham, MA), 100 μl 100x DNase I buffer (1 M Tris–HCl, 0.25 M MgCl₂, 10 mM CaCl₂), and 6 μl RNase (Thermo Scientific) and incubated 1 h at 37°C to degrade the bacterial nucleic acids. Four hundred microliters of 0.5 M EDTA and 25 μl of 20 mg/ml proteinase K (Applied Biosystems, Carlsbad, CA) were added and incubated 1 h at 55°C to inactivate DNase I. After cooling to room temperature, the lysate was added to 8.4 g of guanidine thiocyanate and briefly mixed. One milliliter of warmed, resuspended Wizard DNA Clean-Up Resin (Promega Corporation, Madison, WI) was added to the mixture and mixed by inverting for 15 minutes to allow for release of the phage DNA and binding to the resin. The mixture was pelleted by centrifugation at room temperature for 10 min at 5,000 x g and the supernatant was drawn off until ~5 ml remained. This mixture was resuspended by swirling, transferred into an empty syringe barrel attached to a Wizard Minicolumn (Promega Corporation), and pushed into the column. The column was then washed with 2 ml 80% isopropanol and dried by centrifugation for 2 min at 10,000 x g. JG068 DNA was eluted from the column following addition of 100 μl of 80°C nuclease-free water (Integrated DNA Technologies, Coralville, IA), incubation for 1 min, and centrifugation for ≥20 s at 10,000 x g.

Phage and plasmid DNA were quantified using a NanoDrop 2000c spectrophotometer (Thermo Scientific). For shotgun cloning, JG068 EcoRI fragments were purified using a GENECLEAN III kit (MP Biomedicals, Santa Ana, CA), ligated into pUC19, and transformed into E. coli DH5α (Invitrogen, Carlsbad, CA). Plasmids were purified using the GeneJET Plasmid Miniprep kit (Thermo Scientific) and sequenced using a 3730 DNA Analyzer (Applied Biosystems) by the University of Alberta Department of Biological Sciences Molecular Biology Service Unit (MBSU). Sequences were edited and assembled using Geneious [49].

The complete JG068 genome sequence was determined with the assistance of the MBSU using an Ion Torrent PGM (Applied Biosystems) and assembled using SeqMan NGen (DNASTAR, Madison, WI). Ambiguous regions in the assembly were resequenced using Sanger sequencing from EcoRI clones (for internal fragments) or PCR products (for terminal fragments). Chromosomal termini were identified using primers within the left direct terminal repeat (DTR), extending to the left end (IL: CAACCCTGTACAGCCGACCC), outside of the left DTR, extending to the left end (OL: CCTTGCTCTATCTACCATGTTCCGC), within the right DTR, extending to the right end (IR: TGTGGATAGGGCGAAGTCTGAAGC), and outside of the right DTR, extending to the right end (OR: CTCCGACGAAGCATCCGC). Primers were used to directly sequence the JG068 DNA and sequence alignment was used to identify the loci where sequence intensity dropped by 50% (for primers IL and IR) or 100% (for primers OL and OR). The JG068 sequence has been deposited in GenBank [GenBank: KC853746].

Bioinformatics analysis

Open reading frames (ORFs) were identified using GeneMark.hmm for prokaryotes [54] and annotated using BLASTP [55], HHpred [56], and CD-Search [57]. Gene 49, which has a start codon within the last third of gene 48, was identified using ORF Finder [58]. The replication origin and terminus were predicted using a GC-skew plot generated by GenSkew [59]. Genome restriction profiles were predicted using NEBcutter [60]. Whole genome comparisons were performed using CoreGenesUniqueGenes (CGUG) with a cutoff score of 75 [35, 37]. Screening for bacterial toxins was performed using BTXpred [61]. Bacterial promoters, phage promoters, and rho-independent terminators were identified using prokaryotic Neural Network Promoter Prediction with a cutoff of 0.95 [50], PHIRE [51], and TransTermHP [53] on the PePPER server [62], respectively. Sequence logos were constructed using WebLogo [52]. Transmembrane domains were predicted using TMHMM [63] and OCTOPUS [64]. Lysis protein signal sequences were identified using SignalP 3.0 [65] and LipoP [66].

SAR endolysin expression

To assess the activity of the gp47 SAR endolysin, gene 47 including the signal sequence (bp 40,232 – 40,753; NdeI-SS-F: ATAATAACATATGCACCCCATCGTCAAGCGAG; HindIII-R: AAAAAGCTTCTATCGCCGGATACCAGCAACG) or excluding the signal sequence (bp 40,307 – 40,753; NdeI-ΔSS-F: TAATAACATATGGACGAGGGTATCCGGAACGTC; HindIII-R: as above) was PCR amplified using KAPA HiFi HotStart ReadyMix (Kapa Biosystems, Woburn, MA) and ligated in-frame into pET22b. These constructs were separately transformed into E. coli DH5α, sequenced to verify that the inserts were correct, and transformed into E. coli BL21(DE3)pLysS using a standard protocol [67]. A pET22b blank control plasmid was isolated from E. coli MC4100 and transformed similarly into BL21(DE3)pLysS. To assay endolysin activity in E. coli, 5 ml liquid cultures of the three strains (in triplicate) were each subcultured into six wells of a 96 well plate by adding 10 μl culture to 190 μl LB broth containing 25 mg/l chloramphenicol and 100 mg/l ampicillin. The plate was then covered with plastic wrap and incubated with shaking at 37°C. Optical density measurements at 600 nm (OD₆₀₀) were measured at 1 h intervals with a Wallac 1420 VICTOR² multilabel counter (PerkinElmer, Waltham, MA). At 3 h (OD₆₀₀ ~ 0.6), half of the samples were induced with IPTG (1 mM final concentration) and optical density measurements continued at 1 h intervals up to 8 h. Averages and standard deviations were calculated using Microsoft Excel.

Galleria mellonella assays

G. mellonella infection and treatment assays were performed as described previously [10] with modifications. One milliliter of a 16 h K56-2 overnight culture was pelleted by centrifugation for 2 min at 10,000 × g, resuspended in 1 ml MgSO₄-ampicillin solution (10 mM MgSO₄, 1.2 mg/ml ampicillin), and diluted in this solution to 1:10⁴. High-titre JG068 lysate was passaged through a Detoxi-Gel Endotoxin Removing Column (Thermo Scientific) and supplemented with ampicillin. Larvae (RECORP Inc., Georgetown, ON) were stored at 4°C and warmed to room temperature prior to injection with a 250 μl syringe with repeating dispenser (Hamilton Company, Reno, NV). In the left hindmost proleg, 5 μl of the diluted K56-2 suspension (3.7 × 10³ CFU/5 μl) was injected into ten infected/treated larvae and ten infected/untreated larvae and 5 μl of MgSO₄-ampicillin solution was injected into ten uninfected/treated larvae. In the adjacent left proleg, 5 μl of the JG068 lysate (1.3 × 10⁶ PFU/5 μl) was injected into ten infected/treated larvae and ten uninfected/treated larvae and 5 μl of MgSO₄-ampicillin solution was injected into ten infected/untreated larvae. Larvae were placed into petri dishes, incubated aerobically at 30°C for 72 h, and scored for survival. Experiments were repeated in triplicate. Averages and standard deviations were calculated using Microsoft Excel.