Leftover hemoculture samples that were culture-positive for B. pseudomallei were taken from Srinagarind Hospital, Faculty of Medicine, Khon Kaen University. Retrievals of patients’ hemoculture samples were reviewed and approved by the Khon Kaen University Ethics Committee for Human Research based on the Declaration of Helsinki and the ICH Good Clinical Practice Guidelines (approval number: HE651180). In addition, this research project was approved by the Institutional Biosafety Committee (IBC), Faculty of Medicine, Siriraj Hospital, and Faculty of Tropical Medicine, Mahidol University. The approval certificate numbers are SI 2022–013 and MU 2023–002.

Bacterial strains used in this study are listed in Table 1. B. thailandensis strain E174 was chosen as a propagation strain for phage isolation and purification. Bacteria were cultured on Luria–Bertani (LB) agar (Hardy Diagnostics, United States) and incubated at 37°C for 18–24 h. To obtain mid-log phase bacteria, 30 μL of overnight-cultured bacteria was sub-cultured into 3 mL of LB broth (Thermo Fisher Scientific, United States) and incubated at 37°C for approximately 4 h.

Hemoculture bottles that were positive for B. pseudomallei were collected and kept at 4°C until phage isolation. After centrifugation at 8,000× g for 10 min, 10 mL of the hemocultures was aspirated and filtered through a 0.45-μm membrane filter (Whatman, United Kingdom) to remove cell debris. Then, the filtrated samples were stored at 4°C until phage screening by spot assay, following the method previously described by Clokie and Kropinski (2009) with some modifications. Essentially, 10 μL of the filtrate hemoculture samples were spotted onto the LB top agar (0.35%) containing B. thailandensis strain E174 before incubation at 37°C for 16–18 h. The next day, a single clear plaque was picked by using a sterile end-cut-yellow pipette tip. Then, the pipette tip was put into a microcentrifuge tube containing 1 mL of SM buffer (100 mM NaCl, 8 mM MgSO₄·7H₂O, 50 mM Tris–HCl, pH 7.5) and swirled to release the piece of agar into the buffer. The collected sample was maintained at 4°C until the next step of purification by double agar overlay plaque assay.

A single clear zone collected from spot assays was purified by double agar overlay plaque assay (Clokie and Kropinski, 2009) for at least 10 rounds. Briefly, a mixture of 100 μL of mid-log bacterial culture, 100 μL of the filtered sample, and 5 mL of LB top agar (0.35% agar) was poured on to an LB bottom agar (1% agar) and incubated overnight at 37°C. Phage propagation was performed by double agar overlay plaque assay with approximately 30 plates per sample. SM buffer was added to each plate before plate lysates were collected, centrifuged, and filtered through a 0.45-μm filter membrane. To concentrate, the filtrated lysate was mixed with 10% (w/v) PEG8000 (Sigma-Aldrich, Inc., United States) and 0.1 M NaCl (Sigma-Aldrich, Inc., United States). After centrifugation, the pellet was dissolved in SM buffer and collected at 4°C for further characterization.

Phages were negatively stained and observed morphology under transmission electron microscopy (TEM), according to the previously described study (Imam et al., 2019). Briefly, phages (10⁸ particles/mL) were negatively stained with 2% uranyl acetate on a Formvar–carbon-coated copper grid (200 mesh). After staining, each sample was imaged using the JEM 1400 transmission electron microscope (JEOL Co., Japan) with an accelerating voltage of 80 kV. Images of randomly selected phage particles (typically, 8–12 per sample) were then analyzed using ImageJ (version 1.49) software to estimate the approximate sizes of the phage capsid and tail.

The host range of the isolated phages was determined using spot assays. Ten microliters of high-titer phages were spotted on various bacterial strains, including 18 strains of B. pseudomallei, 41 strains of non-encapsulated B. thailandensis, 20 strains of encapsulated B. thailandensis (B. thailandensis capsule variant; BTCV), 1 strain of B. oklahomensis, 1 strain of B. vietnamensis, 1 strain of B. multivorans, 4 strains of B. cepacia, and other Gram-negative bacteria, including 3 strains of Acinetobacter baumannii, 3 strains of Salmonella enterica, and 1 strain of E. coli (Table 1). Independent experiments were repeated three times.

For thermal stability determination, phage suspension was incubated at 25°C, 37°C, 45°C, and 55°C, respectively. For pH stability determination, phages were mixed with SM buffer at a pH range of 2–11. After incubation for 30 min, 1, 3, 5, and 24 h, the thermal and pH-treated phage samples were taken for 10-fold serial dilution with SM buffer before phage titers were determined by spot assay. Independent experiments were repeated three times.

Phage genomic DNA (gDNA) was extracted by the phenol–chloroform–isoamyl alcohol (PCI) extraction protocol (Sambrook and Russel, 2001). Briefly, phage suspension (10⁸ PFU/mL) was mixed with 1 M MgCl₂ (Sigma-Aldrich, Inc., United States), DNAseI, and RNAseA (Thermo Fisher Scientific, United States) before incubation for 30 min at room temperature. Then, Proteinase K (Thermo Fisher Scientific, United States), 0.5 M EDTA, and 10% SDS were added. After incubation at 55°C for 1 h, the mixture was treated with phenol:chloroform:isoamyl alcohol (25:24:1) and then centrifuged. Only the top aqueous layer was collected and mixed with chilled, equal amounts of 95% ethanol and 3 M sodium acetate (Sigma-Aldrich, Inc., United States). The mixture was then incubated overnight at −20°C before centrifugation, and the pellets were washed with 70% ethanol before dissolving the dried pellets in nuclease-free water and stored at −20°C or − 80°C for long-term storage.

Genomic libraries of phages were prepared using the Illumina DNA Library Prep kit (Illumina, Inc., United States) before sequencing using an Illumina MiSeq. Low-quality bases were trimmed using sickle v1.33 and Trimmomatic v0.39 (Joshi and Fass, 2011). Then, the trimmed FASTQ files were assembled using SPAdes v.3.6.0 (Bankevich et al., 2012) and Unicycler v0.4.8 (Wick et al., 2017). Phage genomes were reordered and oriented by CONTIGuator (Galardini et al., 2011) and manually using the Geneious program. Sequences were identified as ORFs and annotated by Prokka v1.12 (Seemann, 2014) using a custom “Caudovirales” database. Putative functions were assigned and searched against BLASTP¹ and Pfam (Mistry et al., 2021). Phage genome maps were then generated and visualized using Artemis version 17.0.1 (Rutherford et al., 2000).

For comparative genomics, average nucleotide identity (ANI) was calculated using FastANI v1.3 (Jain et al., 2018). To determine phylogenetic relationships among these viruses, phage whole genomes were aligned using Multiple Alignment Fast Fourier Transform (MAFFT) v7.505 (mafft --auto input > output) to build nucleotide alignments based on the aligned amino acid sequence (Katoh et al., 2002). Additionally, phage genome sequences were uploaded into ViPTree (Nishimura et al., 2017) to analyze phylogenetic phage proteomes. The evolutionary history was inferred by using the neighbor-joining method (Saitou and Nei, 1987). Then, the phylogenetic trees were visualized using the Interactive Tree of Life v6.5.8 (iTOL) programs (Letunic and Bork, 2021). Comparative genome contents were visualized and compared using Clinker v0.0.25 (Gilchrist and Chooi, 2021).

Statistical analysis was performed using GraphPad Prism 8.0 software (GraphPad, CA, United States). The one-way ANOVA was applied to compare the variance in the group means. p-values less than 0.05 were considered statistically significant (*p < 0.05, **p < 0.01, and ***p < 0.001, ns; not significant).

To see if phages could be isolated from melioidosis patients, 40 hemoculture-positive cultures of B. pseudomallei were obtained from the Srinagarind Hospital, Faculty of Medicine, Khon Kaen University. The cultures were filtered, and the filtrates were spotted on the lawns of 41 B. thailandensis strains (Table 1). Initially, optimization showed that B. thailandensis strains E174 and E332 were sensitive to hemoculture-isolated phages; these two strains were then selected as phage isolation strains. Interestingly, 12 samples from 40 hemoculture-positive cultures were positive for phages on these strains, as indicated by a spot assay on a lawn of B. thailandensis (30.0% isolation rate). These results were then confirmed by the double agar overlay plaque assay. Several phage plaque morphologies were observed on bacterial lawns.

Two of the phages that produced plaques with distinct morphologies were selected for further study and named vB_HM387 and vB_HM795. After several rounds of single plaque purification, vB_HM387 formed small turbid plaques (approximately 1–3 mm in diameter) with rough edges and a clear spot at the center (bull eyes’ morphologies), whereas vB_HM795 formed small turbid plaques (approximately 1–2 mm in diameter) with no edges on B. thailandensis strain E174 lawn (Figure 1). This is the first report showing that phages can be isolated from hemoculture-positive cultures from melioidosis patients and that these phages can have distinct plaque morphologies on B. thailandensis lawns.

To investigate if human blood triggers prophage release from bacterial genomes, B. thailandensis was spiked into fresh normal human venous blood and PBS, followed by assessing viable bacteria and free phages. We found that free phages were detected in B. thailandensis-spiked human venous blood, but no free phages were detected in B. thailandensis-spiked PBS (Supplementary material). Additionally, after incubation, no viable B. thailandensis was detected in B. thailandensis-spiked human venous blood, but B. thailandensis remained viable in PBS. These findings imply that Burkholderia prophages could be induced by interactions of bacteria with human blood.

To observe the morphology of the hemoculture-isolated phages, TEM analysis was undertaken. Based on the TEM assessment of phage tail morphologies, two selected phages (vB_HM387 and vB_HM795) are Myoviruses in the order Caudovirales (Figure 2). These phages are characterized by the presence of hexagonal capsids (approximately 102.6–107.5 and 63.6–68.2 nm width) and long contractile tails (approximately 120.8–130.0 nm and 109.1–117.5 nm in length) for vB_HM387 and vB_HM795, respectively.

The lytic activities of vB_HM387 and vB_HM795 phages on different strains of Burkholderia species and other related Gram-negative bacteria were assessed using spot assays. The results revealed that phages vB_HM387 and vB_HM795 could lyse 6/19 and 11/19 isolates of B. pseudomallei and 35/61 and 53/61 isolates of B. thailandensis, respectively (Table 2). Among six clinical B. pseudomallei strains isolated from Thailand, each of the phages vB_HM387 and vB_HM795 could lyse 2/6 different B. pseudomallei strains and shared the ability to lyse the other 2/6 different strains of B. pseudomallei. Thus, there were a total of four B. pseudomallei strains that could be lysed by each phage. Differently, among five soil B. pseudomallei strains, phage vB_HM795 could lyse 4/5 strains, whereas phage vB_HM387 could lyse 2/5 strains, with two strains overlapping (Supplementary Table S2). Among 61 B. thailandensis isolates composed of non-capsulated and capsulated B. thailandensis or B. thailandensis that express a B. pseudomallei-like capsule (BTCV), the infection rates against non-capsulated B. thailandensis were 80.49% (33/41) and 82.93% (34/41) for vB_HM387 and vB_HM795, respectively. For the ability of hemoculture-isolated phages to infect BTCV (Sim et al., 2010), we found that vB_HM387 could infect only two out of 20 tested BTCV strains, whereas strikingly, vB_HM795 could infect 19 of 20 BTCV strains. These results indicated that phages vB_HM387 and vB_HM795 had a broad spectrum to infect B. thailandensis, but phage vB_HM795 had a wider range to infect BTCV strains than vB_HM387. Neither phage could infect other related species of Burkholderia or Gram-negative bacterial strains (Table 2), suggesting the specificity of two isolated phages against B. pseudomallei and B. thailandensis.

Importantly, vB_HM387 and vB_HM795 could infect both clinical and environmental B. pseudomallei strains isolated from Thailand. Interestingly, one hemoculture-isolated phage (vB_HM795) from a melioidosis patient from Thailand could infect and cause lysis of a B. pseudomallei strain isolated from Australia (Table 2). Taken together, there are differences in biological properties between vB_HM387 and vB_HM795, in which vB_HM795 showed a broader spectrum than vB_HM387 to infect BTCV and Australian B. pseudomallei strains.

Assessing the effects of pH and temperature on the stability of phage virions may provide an indication of how phages can withstand the stress imposed by human blood. To determine pH stability, phages were exposed to varying pH ranges from 2 to 11 before being assessed for viability by spot assays. The results showed that phage vB_HM387 was inactivated at very acidic conditions (pH 2–3) but was stable at pH 5–9. Additionally, the viability of vB_HM387 was only slightly reduced following incubation at the strong alkaline (pH 11) condition for 24 h (T24; Figure 3A). Whereas, phage vB_HM795 was sensitive to strongly acidic (pH 2–3) and strong alkaline (pH 11), but it was stable at pH 5–7, and its viability only slightly reduced at 5-h post-treatment, and a further reduction in viability was observed after 24-h incubation at pH 9 (Figure 3B). In conclusion, the major difference between these two phages is that vB_HM387 was more resistant to alkaline conditions than vB_HM795.

To determine the thermal tolerance of the phages, they were incubated at varying temperatures (25, 37, 45, and 55°C) before assessing their viability by spot assay. As shown in Figures 3C,D, both phages vB_HM387 and vB_HM795 were stable at 25 and 37°C. However, the titers of these two phages were significantly reduced after exposure to 45°C. At the higher temperature (55°C), phage vB_HM387 remained viable until T5 but was completely inactivated by T24, whereas phage vB_HM795 could not tolerate 55°C, even for a short period of incubation. In conclusion, these two phages showed different thermal stability, as phage vB_HM387 showed more tolerance at high temperatures than vB_HM795.

To investigate the genetic contents of the isolated phages vB_HM387 and vB_HM795, their genomes were sequenced and analyzed (deposited in GenBank accession numbers: OR990504 and OR990505, respectively). Analyses of the genomes of phages vB_HM387 and vB_HM795 indicated that they consist of double-stranded DNA genomes of 36.3 and 44.0 kb in size and 65.3 and 63.5% GC contents, respectively. No tRNAs were detected in either of the phages, and 43 and 51 putative open reading frames (ORFs) were predicted in the genomes of vB_HM387 and vB_HM795, respectively.

The circular genome map of phage vB_HM387 is shown in Figure 4, with 9 of 43 genes (20.93%) encoding hypothetical proteins. The remaining 34 ORFs were functionally annotated, as predicted by RAST and BLASTP. Genome analysis revealed five major functional groups of genes/proteins: a packaging module, a phage structure module, a lysis module, a DNA metabolism module, and an additional module. The structural module comprised 20 genes encoding proteins required for host recognition and structural assembly: tail-associated proteins, baseplate-associated proteins, and head-associated proteins. Phage vB_HM387 encodes a major tail sheath (gp25) and major tail tube proteins (gp26), which are normally found in Myoviruses. Predicted morphology was confirmed by observation under transmission electron microscopy (Figure 2A). The lysis gene module comprised four genes required for host cell lysis (two holins, the N-acetylmuramidase family protein, and LysB). A gene encoding integrase, which is required for lysogeny, was identified, indicating that phage vB_HM387 was an obligate lysogenic phage. Interestingly, the type II toxin–antitoxin (TA) addition module genes were found in this phage genome.

A circular genome map of phage vB_HM795 is shown in Figure 5, with 23 of 50 genes (46.00%) encoding hypothetical proteins. The products of 27 ORFs were functionally annotated, and the genome analysis revealed five major functional groups as described above. The structural module comprised 16 genes. Phage vB_HM795 also encodes a major tail sheath (gp17) and major tail tube proteins (gp18), suggesting that these phages are Myoviruses (Figure 2B). The lysis gene module comprised three genes required for host cell lysis (holin, lysozyme, and spanin). A gene encoding integrase was also identified, indicating that this phage is a temperate phage. Interestingly, genomic analysis of the phage vB_HM795 revealed a gene encoding a known virulence-associated protein E (vapE) of B. pseudomallei (CAJ2861095). This finding is correlated with a previous report that prophages frequently harbor gene-encoding virulence-associated proteins (vap), which play roles in bacterial pathogenesis (Canchaya et al., 2003).

In order to study the phylogenetic relationships between vB_HM387, vB_HM795, and other phages, whole-genome sequences of phages vB_HM387 and vB_HM795 were uploaded into ViPTree to construct phylogenetic trees based on the whole-genome encoded amino acid sequences. Figure 6A shows the phylogenetic tree of phage vB_HM387. This revealed that vB_HM387 belongs to the same species as seven other Burkholderia phages in the NCBI database, including having 95.05% nucleotide identity to phage phiE094 (90% coverage, NC_055911.1), 98.93% nucleotide identity to phiBP82.2 (89% coverage, NC_070912.1), 98.29% nucleotide identity to phiE202 (86% coverage, NC_009234.1), 96.33% nucleotide identity to phiE52237 (80% coverage, NC_007145.2), 96.25% nucleotide identity to phiBEK (80% coverage, CP008753.1), and 96.14% nucleotide identity to phiX216 (80% coverage, JX681814.1). To confirm these findings, the most similar phages were selected for a comparative genomic synteny and similarity analysis using Clinker (Figure 6B). Phage vB_HM387 shows a high similarity to phages E094 and BP82.2, with some sequence variability in genes encoding phage capsid scaffolding protein, site-specific DNA-methyltransferase, and baseplate assembly proteins.

Moreover, the vB_HM387 phage is also conserved in the bioinformatically identified prophage-like elements GI15 of B. pseudomallei K96243 and PI-E264-2 of B. thailandensis E264 (Ronning et al., 2010). Searching the NCBI database using the BLAST algorithm revealed that phage vB_HM387 shows high identity to various B. pseudomallei strains (93% coverage, >98% identity) isolated from various regions, including Thailand, e.g., strain 1026b (CP002833.1), Malaysia, e.g., UKMH10 (LR595892.1), Japan, e.g., strain Tokushima (AP028081.1), and Australia strains, e.g., MSHR5864 (CP017048.1), and MSHR6755 (CP017046.1). These results indicated that the phage vB_HM387 can exist as a prophage in many strains of B. pseudomallei isolated from multiple countries.

Figure 7A shows the phylogenetic tree of phage vB_HM795. Our analysis revealed that phage vB_HM795 represents a unique cluster, far from the other previous Burkholderia phages in the NCBI database. An NCBI BLASTN search revealed that phage vB_HM795 shared only 3% coverage with other phages. To confirm this, the most similar Burkholderia phages (Mica and ST79) were selected, and a comparative genomic similarity was carried out using Clinker (Figure 7B). Similar to the phylogenetic tree analysis, vB_HM795 shows less homology to other Burkholderia phages, including vB_HM387, indicating that phage vB_HM795 can be considered a novel phage. Phage vB_HM795 has unique DNA packaging, head and tail biosynthesis, host lysis, lysogeny, or DNA replication genes. However, some genes show similar to but low percentage identity to the most similar Burkholderia phages, Mica and ST79, including large terminase, baseplate assembly protein V, baseplate assembly protein, baseplate J/gp47 family protein, major tail sheath protein, and contractile injection system protein.

Searching the NCBI database using the BLAST algorithm revealed that phage vB_HM795 shows a high identity to a fragment of the bacterial chromosome of B. pseudomallei strain HNBP001 isolated from China (CP038805.1) with 97% coverage and 98.01% identity. Additionally, phage vB_HM795 also showed some similarity to other B. pseudomallei strains (>83% coverage and > 98%), such as the Chinese isolated strain 14 M0960418 (CP019042.1), the Japan-isolated strain GTC3P0054 (AP028075.1), the Singapore isolated strain AW17-22 (CP073737.1), the Ecuador isolated strain 7,894 (CP009535.1), and the India isolated strain VBP399 (CP071757.1). Searching vB_HM387 and vB_HM795 against the Burkholderia group (taxid:119060) deposited in the GenBank database with a maximum number of 5,000 target sequences revealed that phages vB_HM387 and vB_HM795 showed high sequence similarity (>60% coverage, >90% identity) to 98 genomes of B. pseudomallei strains. Specifically, vB_HM387-like and vB_HM795-like prophages were found in 40/98 (40.82%) and 30/98 (30.61%) of these genomes, respectively, indicating that 30–40% of B. pseudomallei genomes contained DNA similar to either vB_HM387 or vB_HM795 DNA, with vB_HM387-like DNA being more prevalent. Furthermore, 28/98 (28.57%) B. pseudomallei strains carried both vB_HM387-like and vB_HM795-like prophage DNA in their genomes. Taken together, hemoculture-isolated phages vB_HM387 and vB_HM795 are lysogenic phages that are highly similar to prophages harbored in the genomes of several B. pseudomallei strains that had been deposited in the NCBI database. Furthermore, our data indicated that some B. pseudomallei strains carried both vB_HM387-like and vB_HM795-like prophages in their genomes.