In total, 103 A. baumannii isolates from our laboratory collection were tested for evaluating bacterial susceptibility to phages (Table S1). All isolates were of clinical origin, collected in 10 different hospitals from different regions in Serbia. The majority of the isolates (n = 89) were carbapenem-resistant, whereas 14 were carbapenem-sensitive. Among the carbapenem-resistant isolates, three isolates were also identified as colistin-resistant.

All isolates were tested for their resistance to antibiotics using VITEK 2 automated system (BioMérieux, Marcy l'Étoile, France). Minimum inhibitory concentrations for antibiotics were determined in accordance with the guidelines of the Clinical and Laboratory Standards Institute.

All isolates were propagated either in Luria–Bertani (LB) or Mueller–Hinton (MH) broth at 37°C with aeration [180 revolutions/min (rpm)]. Stocks were made in LB supplemented with 15% (vol/vol) glycerol and kept at −80°C.

Pulsed-field gel electrophoresis (PFGE) analysis of the A. baumannii isolates was performed as previously described (28). In situ DNA was digested with ApaI (Thermo Fisher Scientific, 20 U/agarose block in 100 μL enzyme reaction buffer) for 3 h at 30°C. Gels (1.2% agarose) were run at LKB hexagonal electrode array PGFE system using 2015 Pulsafor unit (LKB Instruments, Bromma, Sweden) for 18 h at 300 V in 0.5 × Tris-Borate-EDTA at 9°C and stained with ethidium bromide (SERVA Electrophoresis GmbH). Pulse times were linearly increased from 8 to 18 s during electrophoresis. Gel images were captured under UV light and analyzed using Bionumerics 7.6 software (Applied Maths NV). Cluster analysis, using Pearson correlation coefficient with a 1.0% optimization and a hierarchic UPGMA algorithm, was used to generate a dendrogram describing the relationship among A. baumannii pulsotypes. Given the sensitivity of Dice coefficient, Pearson correlation coefficient was applied to obtain clustering without influence of minor variations in band detection. As there is no consensus on the percentage of minimum similarity needed for clustering, a minimum similarity of 60% was defined for the present data set as the cutoff value for clustering, with a minimum of two members per cluster.

Acinetobacter baumannii isolates of interest were analyzed by typing using a previously described multiplex polymerase chain reaction (PCR) protocol (29).

Phages were isolated by standard enrichment method (30) from Belgrade sewage wastewaters, using A. baumannii 6077/12 (31) as the host bacterium. Sewage water was centrifuged at 4,500 × g for 10 min at room temperature (RT) (Eppendorf 5804) and filtered through 0.45-μm filters (Filtropur S 0.45; Sarstedt, Nümbrecht, Germany). Phage enrichment was achieved by mixing 10 mL of filtrate with double strength LB in a 1:1 ratio and adding 200 μL of overnight culture of A. baumannii 6077/12. The mixture was incubated for 16 h at 37°C with aeration (180 rpm), followed by another round of centrifugation and filtration. Then, the filtered supernatant was collected and used for plaque assay with double-layer agar plates. Briefly, 100 μL of the filtered supernatant was mixed together with 10 μL of an overnight culture of 6077/12 in 10 mL of melted LB top agar (0.75% wt/wt agar), vortexed briefly, and poured into a Petri dish over a thin 1.5% LB agar layer.

Four rounds of single plaque purification were performed for samples giving rise to plaques. Plaques were collected using pipette tips and resuspended in SM buffer [100 mM NaCl, 8 mM MgCl₂ × 7H₂O, 50 mM Tris-HCl (pH 7.5)] and filtered through 0.45-μm filters. Filtrates were used for sequential plaque assay. After the final plaque assay, working phage preparations were made by inoculation of single plaque into 10 mL of mid-exponential-phase culture of strain 6077/12 followed by overnight incubation. Lysate was centrifuged and supernatant filtered as previously described.

A culture of A. baumannii strain 6077/12 was grown to mid-exponential phase in a volume of 500 mL and was infected with 1 mL of previously obtained phage suspension [titer 10⁹ plaque forming units (PFUs)/mL]. After overnight incubation, the lysate was centrifuged in Sorvall RC3B centrifuge at 13,689.2 × g for 30 min in rotor GS3 to pellet bacterial debris. The supernatant was collected and treated with DNaseI and RNase A (Thermo Fisher) for 2 h at 37°C. Afterward, NaCl and polyethylene glycol (PEG 8000; Sigma) were added (to obtain 1 M and 10% final concentrations, respectively), stirred, and incubated overnight at +4°C to precipitate the phages. The precipitate was centrifuged at 13,689.2 × g for 30 min at +4°C (Sorvall RC3B centrifuge, rotor GS3), and the pellet was resuspended in 4 mL of SM buffer. CsCl was added to reach 0.75 g/mL density. The samples were ultracentrifuged at 131,980.8 × g for 24 h in Beckman Coulter Optima L-80 XP ultracentrifuge in rotor SW55 Ti, after which the distinct phage-containing bluish band was visible at the midpoint of the tube. The phages were extracted from tube by using a 22-gauge needle (PrecisionGlide™; Becton Dickinson, Dubline, Ireland) and dialyzed overnight against 2 L of SM buffer at +4°C. Purified phages were filtered through 0.22-μm filters (Filtropur S 0.45; Sarstedt) and used in subsequent experiments.

Phages purified by CsCl gradient centrifugation were dialyzed against phage buffer [20 mM Tris-HCl (pH 7.2), 10 mM NaCl, 20 mM MgSO₄] [20-min microdialysis on 0.025 μm VSWP membrane filters (Merck Millipore, Darmstadt, Germany)]. Ultrathin carbon films (~3 × 3 mm in size) were floated from mica-sheets into a drop of phage lysate (100 μL), and after an adsorption time of 20 min, samples were transferred into a drop of 1% (vol/vol) of electron microscopy grade glutaraldehyde (20 min) for fixation. Negative staining with 2% uranyl acetate and subsequent transmission electron microscopy (TEM) (Tecnai 10; FEI Thermo Fisher Scientific, Eindhoven, the Netherlands) was done as described earlier at an acceleration voltage of 80 kV (32). Micrographs were captured with a MegaView G2 charge-coupled device camera (Emsis, Muenster, Germany).

DNA of both phages ISTD and NOVI was isolated either by Phage DNA Isolation Kit (Norgen Biotek) or by phenol–chloroform DNA extraction method (33). The kit was used for DNA isolation from phage lysate according to manufacturer's recommendation, whereas phenol–chloroform extraction method was performed on phages purified and concentrated by banding on the block gradient of CsCl. DNA was run on a 1% agarose gel, stained with ethidium bromide (SERVA Electrophoresis GmbH), and visualized under UV light, as previously described. Restriction enzyme digestion was performed with several enzymes: EcoRI, HindIII, ApaI, PstI, SalI, DraI, XbaI, and DpnI (Thermo Fisher Scientific) according to manufacturer instructions. Restriction enzyme digestions were repeated three times.

An adsorption test was performed to determine the number of phages that adsorb to A. baumannii 6077/12 host cells during a certain time period. The host strain was incubated in LB broth until exponential phase was reached. The culture was transferred into five sterile microtubes, 1 mL per tube, and 10 μL of 10⁸ PFUs/mL ISTD or NOVI phage suspension was added. Mixed suspensions were incubated for 1, 2.5, 10, 20, and 30 min at 37°C without aeration, to allow adsorption of phage particles to the bacterial cell surface. Suspensions were then centrifuged for 1 min at the speed of 15,000 × g (Eppendorf 5418 R), so the bacterial cells with phages attached would be harvested in a pellet. The supernatant was transferred into sterile microtubes and was used in a plaque assay for determination of the unadsorbed phage titer. Based on the starting concentration of phages added and the concentration of unadsorbed phages, the percentage of adsorbed phages was calculated by the formula of Geller et al. (34):

The life cycle of phages ISTD and NOVI was investigated by a one-step growth experiment, and “burst size” was also determined as the number of virions released from one cell. Acinetobacter baumannii 6077/12 was inoculated in 10 mL of LB broth and incubated at 37°C, with aeration (180 rpm), until it reached exponential phase. Culture volume of 1 mL was transferred to a sterile microtube and was infected with 10 μL of ISTD or NOVI phage suspension [stock concentration 10⁸ PFUs/mL, multiplicity of infection (MOI) 0.01]. The mix was incubated for 5 min at 37°C, to allow the adsorption of phages to the bacterial cell surface, and afterward diluted 100 × in LB broth, in order to avoid contact of the unadsorbed or newly liberated phages with new cells. During further 60-min incubation, starting from the moment of dilution, a 100-μL sample of the suspension was collected every 10 min, and phage titer was determined. Burst size was calculated as the ratio between the number of total released phages and the number of infected bacterial cells.

Host range of ISTD and NOVI was investigated by testing their lytic activity on 103 A. baumannii isolates in total. Sensitivity toward the phage was tested by double agar layer spot assay. Briefly, 10 μL of overnight culture of each isolate was inoculated into 10 mL of melted soft agar. After vortexing, soft agar containing bacteria was overlaid on LB agar in Petri dish. Upon solidifying, 10 μL of phage suspension was spotted on top of the agar. After 16 h of incubation at 37°C, plates were inspected for appearance of growth inhibition zones surrounded by halos. Isolates that were sensitive to phages were additionally inspected after 24, 48, and 72 h at RT for enlargement of halos.

The relation between the spreading of halo and spreading of phages through the agar was investigated using similar approach as described in the work of Cornelissen et al. (35). The abundance of phages, as well as the bacteria, was evaluated inside the plaque, in the plaque vicinity, and in more distant parts of halo formed with a regular spot assay. The values were addressed as PFUs and colony-forming units (CFUs) per gram of agar. To extract the cells and phages, parts of soft agar were sliced with a sterile scalpel, measured for weight, and resuspended in 1 mL of saline solution by vigorous vortexing. From this mixture, serial dilutions were made and plated for CFU counting. For PFU counts, a drop of chloroform was first added to the solution in order to kill the residual bacteria, and then the serial dilutions were made. Obtained results were pondered according to the weight of the agar pieces. Statistical analysis of bacterial and phage titers was done by one-way analysis of variance followed by Tukey post-hoc test (Statistica Software).

Additionally, the ability of phages to form halos on pregrown bacteria was tested. Bacteria were added to melted soft agar, poured, and grown overnight. The following day, 10 μL of phage preparation was dropped on the surface of the lawn; the plate was incubated at 37°C for additional 24 h, after which the plates were inspected for zone or halo formation.

Characterization of phage-planktonic bacteria kinetics was performed as follows: overnight culture of bacterial strain A. baumannii 6077/12, grown in MH, was adjusted to density of 0.5 McFarland scale and further on diluted 100 times into 200 μL MH broth without (growth control) or with ISTD phage suspension in a 96-well-microtiter plate (Tissue Culture Plate; Sarstedt). Phages were applied at MOIs 0.01, 0.1, 1, and 10. At 3-, 6-, 9-, and 24-h time points after treatment, samples were taken, serially diluted in saline, and plated on MH agar. Viable cell counts were calculated from triplicate assays.

Biofilm formation assay was performed according to the method described previously (36). Briefly, overnight cultures of A. baumannii isolates were adjusted to density of 0.5 McFarland scale, and 20 μL was transferred to wells of 96-well-microtiter plates filled with 180 μL LB broth medium. Microtiter plates were incubated aerobically for 48 h at 37°C. After incubation, attached bacteria were washed three times with phosphate-buffered saline (PBS; pH 7.2) and fixed by drying 30 min at 65°C. For staining and visualization of biofilm, 0.1% crystal violet (HiMedia Labs Pvt. Ltd., India) was used (30 min, RT). The stain was rinsed by washing three times with PBS and then resolubilized with mixture of ethanol (96%) and acetone (4:1). Quantification of biofilm formation was done by measuring absorbance at 595 nm using Plate Reader Infinite 200 pro (MTX Lab Systems, Austria) from decuplicate assays.

Antibiofilm activity of phages was evaluated against biofilm formed on porous glass beads having a diameter 4 mm, pore size 60 μm, and surface area of ~60 cm² (VitraPor; ROBU, Hattert, Germany) as described previously (37, 38). Briefly, overnight culture of A. baumannii 6077/12 was diluted 100 times in MH and incubated with glass beads (1 bead: 1 mL) for 24 h, at 37°C without aeration. Then, beads were washed three times with saline solution, and each bead was transferred in a separate well of a 24-well-plate (Tissue Culture Plate; Sarstedt) for treatment. Each treatment was done in triplicate. Beads were incubated with phages at two MOIs–0.1 and 100—in MH broth at 37°C. An untreated control of biofilm without phages was also prepared. After 6- and 24-h treatments, CFUs of biofilm-embedded bacteria were counted. Each bead was washed three times with saline and transferred into tubes with 1 mL of sterile saline. Tubes with beads were vortexed for 30 s, sonicated in an ultrasound water bath for 60 s, and again vortexed for further 30 s to dislodge bacterial cells from bead surface, as previously described (39). Further on, serial dilutions were made and spotted on MH agar plates. Cells were counted from biofilm before treatment, biofilm after phage treatments, and biofilm with fresh medium. After 24 h of incubation, the plates were inspected, and numbers of colonies were counted.

Phages ISTD and NOVI were isolated from two sewage wastewater samples, using A. baumannii 6077/12 as the host bacterial strain, given its overall characteristics (31) and good biofilm-forming ability (Figure S2). Using phage purification methods, concentration, and titration, a pure, high-titer (10¹¹ PFUs/mL) stock of both A. baumannii phages was obtained. Plaques produced by both phages were small, <1 mm in diameter, surrounded by large (5–6 mm) halos (Figure 1), which were constantly enlarging 2–3 mm/d.

Based on the TEM micrographs, both phages have rigid, thick, contractile tails of similar size but different head morphotypes, that is, phage ISTD with prolate heads, and phage NOVI with isometric heads (Figure 2). Their morphology suggests that both phages belong to the order Caudovirales, family Myoviridae. For phage ISTD, long tail fibers are typically folded upward along the tail sheath (Figure 2Ia) but are occasionally also visible in random extended positions with typical knee-like bends (Figures 2Ib,c). A distinct ring structure is also visible on the tail (Figures 2Ia,b). For phage NOVI, short (flexible) tail appendages are visible at the bottom of the phage tail (Figures 2IIa–c). Both phages are also shown with contracted tail sheaths and protruding tail tubes in Figures 2Ic, IIc, respectively. The relevant phage dimensions are summarized in the legend of Figure 2.

All tested A. baumannii isolates and their characteristics are given in Table S1. Pulsed-field gel electrophoresis analysis was performed for all 103 A. baumannii isolates. Using the ApaI macrorestriction patterns generated, cluster analysis was performed using Bionumerics software, and the results demonstrated high heterogeneity among the analyzed isolates. Thirteen groups designated A–M could be identified, along with 15 strains producing unique pulsotypes, which could not be clustered within the mentioned groups (Figure S1). Groups B and F were predominant, comprising 22 and 20 isolates with differences up to 37.8 and 39.6%, respectively. Other groups consisted of relatively few isolates (A: 3, C: 5, D: 8, E: 2, G: 3, H: 4, I: 2, J: 7, K: 2, L: 7, M: 3).

Phage ISTD lysed 37 isolates (~36%), whereas NOVI lysed 23 isolates (~22%). Interestingly, phage-sensitive and phage-resistant isolates could be found in almost all groups, suggesting that sensitivity to these phages is not associated with the genetic relatedness of the isolates. However, it should be noted that groups G, J, and K (12 isolates in total) included only phage-sensitive isolates, whereas groups H, L, and M (14 isolates in total) consisted only of phage-resistant isolates.

In addition, 37 ISTD-sensitive isolates were tested using multiplex PCR to establish their sequence type groups. Typing revealed that 29 of these isolates belonged to sequence type Group 1, which corresponded to European clonal complex II, whereas eight belonged to sequence type Group 3, which corresponded to European clonal complex III (Table S1). In total, 23 isolates were sensitive to both phages, mostly belonging to ECC II.

Moreover, it was demonstrated that isolates with the same PFGE profile have different sensitivity toward the phage ISTD. For instance, pairs 8709–11462, 1865/12–8778, GN228–GN385 have the same pulsotype (ApaI macrorestriction profile), but exhibit different phenotype regarding phage ISTD resistance/sensitivity.

All sensitive isolates produced lysis zones of the same size and clearness. Halos appeared around all plaques; however, the timing of the appearance of the halo differed. With several isolates (16F, 12655, 13056, 8761, 10061, 6000, 8709, 9977, 11587, GN227), the halo from both phages did not appear after the first 24 h of incubation, but did so after prolonged incubation. These isolates have not been grouped together according to PFGE or sequence type. Among carbapenem-resistant isolates, two colistin-resistant isolates (GN1106 and GN1107) were also sensitive to both NOVI and ISTD phages.

In order to determine the genetic basis of the phenotypic differences of these two phages, total DNA was isolated and analyzed. DNA of both phages was successfully isolated, and its integrity was confirmed with agarose gel electrophoresis (data not shown). However, none of the eight tested restriction enzymes used were able to cut either DNA. Also, four attempts to sequence the genomes using different platforms of next-generation sequencing (Illumina HiSeq/MiSeq and PacBio's RS II) failed to generate any sequence data.

In Figure 3, the adsorption rates of both phages are reported. After 1 min of phage–bacteria incubation, 79 and 73% of ISTD and NOVI phage particles were attached to host cells, respectively, and after 30 min, more than 95% of particles of both phages were adsorbed.

To identify the different phases of the phage infection process, a one-step growth curve of both phages on A. baumannii 6077/12 was determined (Figure 4). The latent period (the period of time that passes from the initial infection to the first burst of new phages) is 20 min for ISTD and 30 min for NOVI. After an exponential rise of PFUs/mL, ISTD reaches a plateau after 50 min and NOVI after 60 min. ISTD has a burst size of 113 ± 7, and NOVI of 52 ± 6 phage particles released per bacterial cell.

As mentioned, both phages produced turbid halos around the plaques on the majority of tested sensitive isolates, indicating depolymerizing activity of phage enzymes on surface exopolysaccharides of the given isolate. It was noted that exopolysaccharides were efficiently degraded, considering that halo increment could be monitored visually. To establish whether free phages or diffusing enzymes are responsible for this phenomenon, the halo was investigated for phage presence and titer. In parallel, the number of viable bacteria was also monitored.

The halo around the zone of lysis produced by a regular spot assay expanded over the incubation period, with the diameter of the halo increasing by 2–3 mm each day. These “new” layers of halo, named halo1, halo2, and halo3 (Figure 5), were analyzed over three consecutive days, as the halo was enlarging. Free phages could only be detected in halo1, that is, in proximity of the spot, whereas halo2 and halo3 did not contain free phages. The number of phages found in halo1 was relatively high, as, for example the titer of ISTD amounted 6.3 × 10⁴ PFUs/mL compared to 3.5 × 10⁵ PFUs/mL in the zone of lysis. To determine if this phage presence in halo1 interfered with the present bacteria, bacterial number was also monitored. Viable cell count in halo1 was 6.3 × 10⁵ CFUs/mL compared to 4.4 × 10⁶ CFUs/mL in the unaffected bacterial lawn (Figure 6). The titers of both phages and bacteria remained unchanged during the next 2 days, except for the slight decrease in CFUs, which was also noted in control (data not shown). Comparing the performance of two phages inside halo1, it seems that ISTD diffuses more vigorously and results in higher reduction of bacterial cell number than phage NOVI.

Following the results obtained in the halo analysis, the possibility of lysis zone or halo forming was analyzed on pregrown bacterial cells, on stationary phase of growth. The experiment was set as usual double-layer spot test, and phages were spotted on top of the soft agar immediately after it solidified but also on the following day, after the bacterial lawn was already formed by growth overnight at 37°C. The phages spotted immediately after solidifying of agar produced the usual zone of lysis and the halo, which spread with time, but phages spotted on a preformed, 24-h bacterial lawn were also able to produce the spreading halo, without a visible zone of lysis (Figure 7). This result was confirmed by additional experiments, where it was shown that phages, even when applied 4 days after the formation of the bacterial lawn, do not produce zones of lysis but can produce a halo, which spreads with time (data not shown).

Given the shorter latent period, higher burst size, and broader host range, ISTD phage was selected for further analysis of its killing efficiency on bacterial cells during either planktonic or biofilm lifestyle. The activity of ISTD phage against planktonic A. baumannii 6077/12 cell culture was evaluated. Acinetobacter baumannii 6077/12 cells in early exponential phase of growth were incubated with different titers of phage, at MOIs of 0.01, 0.1, 1, and 10, respectively. Interestingly, all MOIs applied yielded the same results regarding the host lysis kinetics (Figure 8). After 3 h, the number of planktonic cells was almost 100-fold lower in comparison to the inoculum size of cells, and a reduction of 3.5 log units in CFU number was observed when the phage-treated samples were compared with the untreated controls. By 6 h, the cell count started recovering, but the reduction was still 3 log units compared to untreated growth control. However, after 9 h, the difference between treated and untreated bacteria decreased to 2.3 log units, whereas after 24 h, there was no difference in the CFU count between the infected cultures and growth control.

Phage ISTD bactericidal activity was also evaluated against 6077/12 cells embedded in biofilm formed on porous glass beads. Phages at MOIs 0.1 and 100 were incubated with bead biofilms, and the CFUs representing cells from biofilm were counted at 6 and 24 h posttreatment. After 6 h of bacteria incubation with ISTD phages at MOIs 0.1 and 100, respectively, the number of viable cells dislodged from biofilms was 0.6 and 2 log units lower in comparison to the untreated control, respectively (Figure 9). However, after 24 h, no difference in CFUs number was observed between treated and untreated controls.