Pseudomonas aeruginosa is a Gram-negative bacterium causing opportunistic infections in humans and animals. Some P. aeruginosa strains cause intractable chronic infections because of their ability to form biofilms and due to their drug resistance. Bacteriophage (phage) therapy is the therapeutic use of specific bacterial viruses (i.e., phages) to treat infections caused by pathogenic bacteria [1]. While phage therapy has a long history of clinical use in the Eastern world, effective antibiotics have been mainly used in the Western world. However, phage therapy has recently been revitalized as one of the alternative therapeutic measures to chemotherapy for treating intractable P. aeruginosa infections in humans and animals in the Western world [2, 3], and clinical trials against P. aeruginosa infections have been carried out [4,5,6].

The combined use of phage with a certain antibiotic has shown increased antimicrobial effects over the single use of phage or antibiotics [7,8,9,10,11]. The combined use of phage with sublethal concentrations of antibiotics has been shown to synergistically increase antibacterial effects—so-called phage–antibiotic synergy (PAS)—and has been shown effective against bacteria including Klebsiella pneumoniae, Staphylococcus aureus, P. aeruginosa, Escherichia coli, and Burkholderia cepacia complex [8, 12,13,14,15]. In addition, the combined use of different antimicrobial agents such as phage and antibiotics is expected to decrease the chances of resistant bacteria emerging [16, 17]. Thus, the combined use of phage with antibiotics is believed to be a promising therapeutic strategy in phage therapy. However, phages do not show PAS with all antibiotics [18, 19], and the combination of phage and antibiotic that shows PAS against P. aeruginosa needs to be fully characterised.

In the present study, PAS was first screened with the antibiotic-disc-embedded double-layered agar method using different groups of P. aeruginosa phages and various antibiotics. Next, the PAS of phage KPP22, classified in the family Myoviridae genus Pbunavirus, was examined using the same method and a collection of different clinical strains of P. aeruginosa. The effect of anti-Pseudomonas drugs such as piperacillin and ceftazidime was then analyzed by measuring the plaque size, the bacterial density in liquid culture, and inhibition of biofilm formation.

In this study, Pseudomonas viruses KPP21, KPP22, KPP23, and KPP25 were used [20,21,22,23], which are all tailed phages and taxonomically belong to different families and/or genera (KPP21: family Podoviridae genus Luz7virus, KPP22: family Myoviridae genus Pbunavirus, KPP23: family Siphoviridae, and KPP25: family Podoviridae genus Kpp25virus). Their genetic backgrounds are distinctly different [20,21,22,23]. Testing their host range using 30 clinically isolated P. aeruginosa strains, these phages had different host specificities (Supplementary Table S1), which highlighted their different biological properties. P. aeruginosa strains PAO1, PA3, PA4, PA5, PA23, and PA29 were used as host bacteria [24]. The strain PAO1 was donated from Professor Tetsuya Matsumoto, Department of Microbiology, Tokyo Medical School, Japan. The strains PA3, PA4, PA5, PA23, and PA29 were clinically isolated from different patient specimens in Kochi University Hospital, and were previously serotyped [24]. Phage and bacteria were cultured at 37 °C in this study. LB medium (Miller) (Sigma-Aldrich, St. Louis, MO) was used as culture media for bacteria and phage, unless otherwise stated. All antibiotics, chemicals, and reagents were purchased from Nacalai Tesque (Kyoto, Japan) or Wako Pure Chemical Industries (Osaka, Japan), unless otherwise stated. In this study, the data were statistically analyzed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria) [25].

PAS can be examined by observation of plaque sizes using the antibiotic-disk-embedded double-layered agar method, as described elsewhere [12, 13, 15]. Briefly, 100 μl of the phage suspension, diluted to be 3.0–8.0 × 102 plaque-forming units (pfu)/ml with LB broth (Miller) (Sigma-Aldrich) and 200 μl of the overnight-cultured bacterial suspension (1.0–8.0 × 108 colony-forming units (cfu)/ml) were mixed in 0.5% of the melted agar on a 1.5% agar plate. After solidifying the top 0.5% of agar, an antibiotic disc (BBLTM Sensi-DiscTM Susceptibility Test Discs; Becton, Dickinson and Company, Franklin Lakes, NJ) was embedded on the center of the double-layered agar plate containing phage and bacteria. Twenty-five antibiotic discs were tested, including 16 antibiotics with inhibition of cell wall synthesis, five antibiotics with inhibition of protein synthesis, one antibiotic with alteration of cell membrane, two antibiotics with inhibition of nucleotide synthesis, and one antibiotic with antimetabolite activity. P. aeruginosa strains PA5 and PAO1 were used as indicator hosts of phages KPP21 and KPP22, respectively. P. aeruginosa strain PA3 was used as an indicator host of phages KPP23 and KPP25. These phages formed adequate plaques (i.e., size and morphology) using the antibiotic-disk-embedded double-layered agar method, and so these bacterial strains were used for the PAS examination. PAS was semi-quantitatively examined by observing the plaque sizes around the inhibitory rings of the antibiotic disc, and comparing them with the plaque sizes without an antibiotic disc. The plaques with unchanged sizes and those with enlarged sizes were scored as 1 and 2, respectively (Fig. 1a). The experiments were performed at least twice to validate the results.

Fig. 1
figure 1

Examination of phage–antibiotic synergy (PAS) using the antibiotic-disc-embedded double-layered agar method. (a) Criteria for plaque size and score. The photos from left to right respectively show three types of plaques: plaques as a control that were not tested with an antibiotic disc, plaques around an antibacterial halo that are the same size as the control, and plaques around an antibacterial halo that are larger than the control. In each photo, the bar indicates 10 mm. Plaques with the same size as the control and plaques larger than the control are scored as 1 and 2, respectively. The criteria applied to examine PAS are as shown in (b) and (c). (b) PAS of four phages KPP21, KPP22, KPP23, and KPP25 with various antibiotics. As the indicator host bacteria, P. aeruginosa strains PA5, PAO1, PA3, and PA3 were used for phages KPP21, KPP22, KPP23, and KPP25, respectively. (c) PAS of phage KPP22 with various antibiotics among four P. aeruginosa strains. P. aeruginosa strains PAO1, PA4, PA8, PA23, and PA29 were used as indicator host bacteria. In (b) and (c), the names of antibiotics are abbreviated as follows: cefepime [CFPM], cefozopran [CZOP], cefoperazone [CPZ], sulbactam/cefoperazone [SBT/CPZ], ceftazidime [CAZ], cefotaxime [CTX], cefpodoxime [CPDX], latamoxef [LMOX], flomoxef [FMOX], cefotiam [CTM], cefmetazole [CMZ], piperacillin [PIPC], meropenem [MEPM], imipenem [IPM], aztreonam [AZT], fosfomycin [FOM], gentamicin [GM], tobramycin [TOB], amikacin [AMK], minocycline [MINO], chloramphenicol [CP], colistin [CL], ciprofloxacin [CPFX], levofloxacin [LVFX], and sulfamethoxazole/trimethoprim [ST]. The antibiotics inhibiting cell wall synthesis include CFPM, CZOP, CPZ, SBT/CPZ, CAZ, CTX, CPDX, LMOX, FMOX, CTM, CMZ, PIPC, MEPM, IPM, AZT, and FOM. Those inhibiting protein synthesis include GM, TOB, AMK, MINO, and CP. The antibiotics which alter the cell membrane include CL. The antibiotics which inhibit nucleotide synthesis include CPFX and LVFX. The antibiotic with antimetabolite activities includes ST

Of the 25 antibiotics tested, phages KPP21, KPP22, and KPP23 showed PAS with five, 13, and three antibiotics, respectively. Phage KPP25 showed no PAS with any antibiotics (Fig. 1b). The antibiotics showing PAS predominantly had antimicrobial activities targeting inhibition of cell wall synthesis (16/25) or protein synthesis (5/25). The remaining antibiotics with other modes of action showed no PAS in the experiments. Of the 16 tested antibiotics that inhibited cell wall synthesis, four (25%) of the antibiotics with phage KPP21, 10 (62%) of the antibiotics with phage KPP22, and two (13%) of the antibiotics with phage KPP23 showed PAS. Of the five tested antibiotics that inhibited protein synthesis, one (20%) of the antibiotics with phage KPP21, three (60%) of the antibiotics with phage KPP22, and one (20%) of the antibiotics with phage KPP23 showed PAS. Although antibiotics inhibiting cell wall synthesis have been reported to show PAS [8, 12,13,14,15], these results indicate that antibiotics inhibiting protein synthesis also show PAS. In addition, phage KPP22 was found to have PAS with the widest range of antibiotics among the four tested phages, particularly antibiotics inhibiting cell wall synthesis and protein synthesis.

Moreover, because of the interest in the use of phages classified in the family Myoviridae genus Pbunavirus with antibiotics [2, 21, 22, 26], PAS of phage KPP22 was also examined with 25 antibiotics on the clinical strains PA4, PA23, and PA29, using the antibiotic-disc-embedded double-layered agar method (Fig. 1c). The strains PA4, PA23, and PA29, which have been isolated from different patients, showed different efficiencies of plating but were not resistant to phage KPP22 (Supplementary Fig. S1). The results were interpreted with the PAS screening of phage KPP22 on the strain PAO1 as described above. The antibiotics inhibiting cell wall synthesis tended to show PAS on strains PA4, PA23, and PA29, although there were differences between them. The antibiotics acting on cell wall synthesis, such as aztreonam, cefepime, cefozopran, ceftazidime (CAZ), and piperacillin (PIPC), showed PAS on all four tested bacterial strains PA4, PA23, PA29, and PAO1. The antibiotics that inhibited protein synthesis showed no PAS on strains PA4, PA23, and PA29, except for the effect of phage KPP22 with amikacin on strain PA4. The antibiotics with other modes of action showed no PAS on these strains, unlike strain PAO1. Thus, the selection of antibiotics for strain characteristics appears to be important in exhibiting the maximal effects in phage and antibiotic combination therapy.

PIPC and CAZ are used as anti-Pseudomonas drugs [27]. It is of concern that the combined use of phage KPP22 with PIPC or CAZ exhibits a synergistic increase of the antimicrobial effects on the strain PAO1 in a liquid culture. The bacterial overnight culture (ca. 3.0 × 109 cfu/ml) and phage suspension (ca. 1.0 × 105 pfu/ml) were prepared. Subsequently, 5 μl of bacterial suspension and 5 μl of phage suspension were mixed with 140 μl of LB broth (Miller) (Sigma-Aldrich) containing antibiotic in each well of a 96-well flat-bottom microtiter plate (Thermo Fisher Scientific, Waltham, MA). The plate was incubated with shaking at 37 °C and the turbidity was measured every 15 min at 600 nm, using a HiTS incubation reader (Scinics Corporation, Tokyo, Japan). The data were collected from two wells for each treatment, and the experiments were performed in triplicate.

The combined use of phage KPP22 and antibiotic (i.e., 1.25 μg/ml of PIPC or 0.31 μg/ml of CAZ) showed lower bacterial turbidity than the groups with no treatment and with the treatment of phage or antibiotic (Fig. 2a). Analysis of the turbidity data by two-way factorial analysis of variance showed that the combined use of phage KPP22 with PIPC and CAZ exhibited significant synergistic effects after 4.5 and 2.5 h incubation, respectively (P < 0.05; Fig. 2a). Thus, since the combined use of phage KPP22 with PIPC or CAZ was considered to have PAS in the liquid culture, as seen on the double-layered agar plate assay, it can be applied as one of the therapeutic strategies in combined antibiotic–phage therapy.

Fig. 2
figure 2

Anti-Pseudomonas effect, with the combined use of phage KPP22 with an antibiotic such as PIPC (left) and CAZ (right). Statistical analysis was performed by two-way factorial analysis of variance to examine if the effects are additive (P ≥ 0.05) or synergistic (P < 0.05) in the interaction of phage and antibiotic treatments. Synergistic interactions (P < 0.05) are shown as “*” in the graphs. (a) Change in bacterial turbidity in the presence of phage KPP22 with PIPC or CAZ. Strain PAO1 (3.0 × 106 cfu/ml) was inoculated in the presence/absence of phage KPP22 (3.0 × 104 pfu/ml) with or without PIPC (1.25 μg/ml) or CAZ (0.31 μg/ml) in LB broth, and the turbidity change was measured over time. (b) Inhibition of biofilm formation by treatment of phage KPP22 with PIPC and CAZ. After incubating the strain PAO1 with phage KPP22 (106, 104, 102, or 0 pfu/ml) and antibiotic (100, 50, 25, 12.5, 7.25, 3.63, 1.81, or 0 ng/ml of PIPC or CAZ), the stained biofilm was quantified by spectrocolorimetry under an optical density of 595 nm (OD595)

In P. aeruginosa infections, biofilm formation may lead to prolonged or chronic infections. Therefore, biofilm formation inhibition by PAS of phage KPP22 with PIPC and CAZ was examined. P. aeruginosa was cultured with phage with PIPC or CAZ in the medium specialized for biofilm formation, following a method described elsewhere with slight modifications [28]. Briefly, the overnight-cultured bacterial suspension was diluted 50-fold with M63 minimal medium supplemented with MgSO4, glucose, and casamino acids (M63MGS broth). Subsequently, 50 μl of bacterial suspension in the M63MGS broth was supplemented with 50 μl of M63MGS broth containing phage and/or antibiotics in a 96-well U-bottom microtiter plate (AS ONE, Osaka, Japan). The final phage concentrations in the M63MGS broth were ca. 1.0 × 106, 1.0 × 104, 1.0 × 102, and 0 pfu/ml. The final concentrations of PIPC and CAZ in the M63MGS broth were 100, 50, 25, 12.5, 7.25, 3.63, 1.81, and 0 ng/ml. Six wells were prepared in each treatment. After incubation at 37 °C for 24 h, the wells were washed twice with tap water, and then the biofilms were stained by 0.1% crystal violet. After rinsing the wells three times with tap water, 125 μl of 30% acetic acid was added and incubated for 15 min at room temperature. 100 μl of the sample was then transferred to a 96-well flat-bottom microtiter plate (AS ONE), and the optical density was measured at 595 nm with a Multiskan JX microplate reader (Thermo Fisher Scientific). In this assay, P. aeruginosa strain PAO1 was used, because this showed the highest biofilm quantity among four strains PAO1, PA4, PA23, and PA29 (Supplementary Fig. S2). The experiments were performed in duplicate.

The biofilm formation was examined in the presence or absence of phage KPP22 with or without PIPC or CAZ (Fig. 2b). First, in the absence of the phage, as the concentrations of PIPC increased, the biofilm quantity increased. The phage did not inhibit biofilm formation at 102 pfu/ml with any concentration of PIPC. In contrast, in the presence of phage at 104 and 106 pfu/ml, biofilm formation was suppressed in the presence of PIPC at more than 12.5 ng/ml. Biofilm quantity was also measured after incubation of P. aeruginosa with phage KPP22 and CAZ. In the absence of phage KPP22, the biofilm quantity increased up to 12.5 ng/ml and then gradually decreased at higher concentrations of CAZ. The phage did not inhibit biofilm formation at 102 pfu/ml with any concentration of CAZ. In contrast, in the presence of phage at 104 and 106 pfu/ml, biofilm formation was suppressed in the presence of CAZ at more than 3.25 ng/ml. Thus, phage KPP22 with PIPC or CAZ also showed synergistic inhibitory effects on biofilm formation.

Stanley showed that plaque size increases as the bacterial concentration in the inoculum decreases [29]. Based on this finding, we hypothesized that the plaque diameter increased as the bacterial concentration in the inoculum decreased with antibiotic treatment, as seen for phage KPP22 with PIPC and CAZ in P. aeruginosa strain PAO1. To test the hypothesis, plaque diameters were examined using the double-layered agar method, in which 100 μl of the phage suspension at 1.0–8.0 × 102 pfu/ml was mixed with 200 μl of the bacterial suspension in 0.5% of melted agar on a 1.5% agar plate. First, to confirm the plaque diameter enlargement is dependent on the antibiotic concentrations, different concentrations of PIPC (50, 5, 0.5, and 0.05 μg/ml) and CAZ (10, 1, 0.1, and 0.01 μg/ml) were supplemented in both 0.5% and 1.5% culture media. An overnight bacterial culture (ca. 3.0 × 109 cfu/ml) was used as the inoculum bacterial host. Second, to examine whether the plaque diameter changed depending on the inoculum bacterial concentration, overnight bacterial culture, and bacterial suspensions, the bacterial overnight culture was serially diluted two-fold, and used as the inoculum bacterial host. After incubation of the double-layered agar plates for 24 h, the plaque diameters were measured.

The plaque diameters were confirmed to become larger as the concentration of PIPC or CAZ increased. PIPC and CAZ significantly increased the plaque sizes 2–3 times, comparing the plaque diameters on the double-layered agar plates containing 5 μg/ml of PIPC and 1 μg/ml of CAZ with those on the double-layered agar plates without antibiotics (one-way analysis of variance and Tukey’s post hoc test, P < 0.01; Fig. 3a). Measuring the plaque diameters on the double-layered agar plates inoculated with different bacterial concentrations (Fig. 3b), to the plaque diameter enlargement showed the inoculum bacterial concentration had decreased. The plaque diameters at the inoculum bacterial concentrations of 1.9 × 107 to 3.0 × 108 cfu/ml were significantly larger than those at the inoculum bacterial concentrations of 2.5 × 109 cfu/ml (one-way analysis of variance and Tukey’s post hoc test, P < 0.01; Fig. 3b). This enlargement ratio of plaque diameter is almost double that in the experiment, and was similar to that in the treatment of phage KPP22 with PIPC and CAZ. In addition, analyzing the data by Pearson’s correlation coefficient, a moderate-to-strong association between plaque diameter and the inoculum bacterial concentration was observed (r = –0.61, P < 0.001). Thus, the treatment of PIPC and CAZ was considered to decrease the bacterial concentration, and consequently increased phage KPP22 lytic activity. Considering the change in plaque size of phage KPP22 on strain PAO1 was dependent on the inoculum bacterial concentration, the reduction of bacterial concentration by antibiotic could support phage lytic activity, which probably contributed to PAS.

Fig. 3
figure 3

Study of PAS using phage KPP22 with P. aeruginosa strain PAO1, using the double-layered agar method. Thirty plaque diameters were measured, and their means with standard deviations are plotted with error bars on the graph. The data were analyzed by one-way analysis of variance and Tukey’s post hoc test. Significant differences are indicated by “*” (P < 0.01). (a) Examination of PAS of phage KPP22 with PIPC and CAZ. In the left and right panels, the results of phage KPP22 with PIPC and CAZ are shown, respectively. (b) Change in plaque diameters depending on inoculum bacteria concentrations. Plaque diameters tended to become enlarged as the concentration of inoculum bacteria decreased

Research has recently been conducted on PAS against infections caused by P. aeruginosa. First, combined use was shown to strongly limit bacterial growth recovery for a long period, which consequently constrains drug-resistant bacteria and tends to reduce bacterial virulence when using phage against P. aeruginosa [14, 30, 31]. Second, PAS can also eliminate biofilms. The combined use showed PAS on P. aeruginosa biofilm, using a phage mixture (containing viruses classified within the Podoviridae and Myoviridae) with ciprofloxacin or CAZ [18]. Phage with amikacin or meropenem were shown to eradicate P. aeruginosa biofilm [32]. Third, sequential treatment, in which phage treatment occurred prior to antibiotic treatment, can also significantly increase the antibacterial effects with some antibiotics, such as gentamicin and tobramycin, compared to simultaneous treatment [18]. Fourth, in addition to the synergistic antimicrobial effects of phages in phage cocktails [33], ciprofloxacin achieved successful treatment with PAS in an in vivo setting [34]. Fifth, in the present study, the selection of phage was considered to be important because different phages showed PAS with a different range of antibiotics. Among four different phages, phages classified within the family Myoviridae genus Pbunavirus showed PAS with the widest range of antibiotics, and showed PAS with PIPC or CAZ that was able to inhibit biofilm formation. Taken together, the combined use of phage and antibiotics offers a possible opportunity for developing treatment strategies for infections caused by P. aeruginosa, with further optimization required regarding the optimal selection and application timing of phages and antibiotics.

We believe that thorough examination of the PAS of phages with an optimal antibiotic in vitro, together with in vivo efficacy experiments, will enhance the successful combined use of phage and antibiotics against P. aeruginosa infections.