A qPCR Based Screening Platform for Exploratory Assessment of Phage Training Outcomes in Enterobacter cloacae and Stenotrophomonas maltophilia

1. Introduction

The rise of antibiotic resistance has made bacteriophages (phages) an increasingly promising alternative for therapeutic applications [1]. Phages are specific, widely available, and can be utilized as additions or alternatives to antibiotics [2]. However, phage therapy faces several challenges and one of them is the rapid emergence of bacterial resistance during treatment [3]. To address this issue, phage therapy is often administered using a cocktail of different phages, which target the same bacterial strain [4]. This approach aims to minimize the likelihood of bacteria developing resistance to the various phages used, as developing resistance to each of them can impose greater fitness costs for the bacteria [5].

While the use of cocktails can help mitigate the emergence of bacterial resistance, identifying multiple phages that are specific to the target pathogenic bacterial strain remains labor-intensive and is not always feasible [6]. An alternative approach to combat bacterial resistance is the ‘pre-training’ of phages. This process involves co-evolving phages with the target bacteria, which includes incubating phages with the bacteria over several passages [7]. The goal of this training process is to produce phage populations that are already adapted to phage-resistant bacteria that are likely to emerge during phage treatment.

This straightforward process can be modified to improve the characteristics of the resulting phages. In recent years, various factors have been explored [8,9]. For instance, research has demonstrated that adding evolutionarily naïve bacteria allows phages to explore novel mutations, which can help them evade bacterial resistance [10,11]. This may lead to an expansion of the phage’s host range, albeit at the expense of their growth rate [12]. Other studies have indicated that the presence of competing phages in co-evolution experiments may have either antagonistic or synergistic effects on the phages obtained [8,13]. Competing phages may inhibit bacterial growth more effectively and assist phages in evading CRISPR-Cas immunity [14]; however, they can also accelerate the emergence of resistant bacteria or reduce the efficiency of phages, particularly if they compete for the same attachment sites [15]. In addition, some other factors have been studied over the past twenty years [16]. Selecting the appropriate factors presents a challenge, as identifying optimal conditions and monitoring the lytic activity of the obtained phages against emerging bacterial populations can be labor-intensive and time-consuming.

Traditional plaque assays, while standard, are low-throughput and labor-intensive when screening numerous phage-bacteria combinations. As a potential solution, real-time PCR (qPCR) serves as a valuable tool for addressing several challenges associated with the study of phages. It enables researchers to scale investigations of various phage characteristics [17]. However, it does have limitations, primarily related to its inability to distinguish between living organisms and DNA remnants [18,19].

Here, we present the use of qPCR as a tool to screen a pool of phage-bacteria interactions to determine the most effective training strategy for phages and the optimal phage population obtained after ten passaging. We tested four training strategies of four different phages across two experiments (biological replicates) over 5 and 10 passages: (A) training the phage with its bacterial host, (B) introducing naïve hosts to maintain a source for phage propagation and avoid their extinction, (C) incorporating a competing phage to promote broad adaptation, (D) combining the use of naïve hosts and a competing phage.

We studied phages that infect two different bacterial species (Stenotrophomonas maltophilia and Enterobacter cloacae), and we asked whether any of the four phages would show directional trends toward maintained or broadened infectivity under these training scenarios. While prior work has explored individual training factors [10], we focused on identifying consistent patterns across phages and on using a qPCR-based screening platform as a scalable tool for such phage-training assessments. In addition to the phenotypic screening; genomic analysis of evolved populations of phage EC152 was conducted.

2. Materials and Methods

2.1. Bacterial Strains and Phages Used and Growth Conditions

Bacterial strains used in this study were obtained from the Collection of Extremophilic Microorganisms and Type Cultures (CEMTC) of the Institute of Chemical Biology and Fundamental Medicine, Siberian Branch of the Russian Academy of Sciences (ICBFM SB RAS). The following bacteria were used: E. cloacae strain CEMTC 2064 and S. maltophilia strain CEMTC 2355. The bacteria were cultivated in Luria–Bertani broth (Thermo Fisher Scientific, Waltham, MA, USA) aerobically at 37 °C. Bacterial growth was monitored by measuring optical density at 600 nm (OD600), an OD600 of 1.0 corresponded to 8 × 10⁸ cells/mL. The following phages from the CEMTC ICBFM SB RAS were used: Enterobacter phage EC151, a weakly lytic phage [20], Enterobacter phage EC152, a lysogenic phage, both infect E. cloacae CEMTC 2064. In addition. Stenotrophomonas phage StM171, a weakly lytic phage [21], and lytic Stenotrophomonas phage StenM174 [22], both specific to S. maltophilia CEMTC 2355 were applied. The characteristics of these phages are summarized in Table 1. And the genomic, phylogenetic and biological characteristics of EC152 are described in more details in supplementary file S1.

2.2. Adapting qPCR for Characterization of Phage Biological Characteristics and Phage Infection Yields

Phage infection efficiency was assessed by measuring changes in the number of phage particle after incubation with the bacterial host. To assess infection efficiency, qPCR was performed according to the protocol presented in Table 2, using oligonucleotides targeting unique genes of the phages under study (Table 3). Raw fluorescence data were processed using the regression algorithm in CFX Manager software.

Phage samples used for qPCR were prepared as follows: 50 µL of each sample was incubated at 100 °C for 30 minutes to disrupt the capsids. An equal volume of chloroform was then added; the mixture was vigorously vortexed for 10 seconds and centrifuged at 14,000 rpm for 5 minutes. Five microliters of the supernatant fraction containing phage DNA was collected for qPCR analysis. Then, the extracted DNA was diluted 5-fold with sterile water to reduce the potential inhibitory effect of bacterial components.

Serial 10-fold dilutions of the phage were used to construct a standard curve. The detection limit was set at the lowest concentration with a Cq value above background noise and within the linear range (~100% efficiency; ΔCq ≈ 3.3 per log dilution; Figure 1A). Two types of controls were used in each PCR run: controls #1 were used to assess the ability of the phage to infect bacteria. The phage Cq value after 24 hours of incubation with bacteria was compared with its Cq value before infection. This was done by including a phage sample diluted to its original concentration before infection (Figure 1B). Controls #2 were used to quantify phage particles in the samples and determine the efficiency of the PCR reaction; three serial dilutions with a known number of PFU/mL of the test phage were included as standards in each run (Figure 1C).

Reproducibility of the results was evaluated by calculating the mean and standard deviation of the Cq values obtained for control samples #2 for each of the four tested phages in different runs. We also examined the range of reaction efficiency and the slope across runs for the same sample.

2.3. Co-Evolution Experiment and Studied Scenarios

Using the optimized qPCR method, we investigated how phage adaptation affects their ability to infect clones derived from their original host. We studied two phages infecting E. cloacae CEMTC 2064: EC151 and EC152, and two phages infecting S. maltophilia CEMTC 2355, StM171 and StenM174.

Phage adaptation was achieved by passaging them with their respective hosts, this was done in two independent biological repeats lasting 10 passages under four different scenarios. The scenarios were as follows (Table 4):

1. Scenario A: phage co-evolution with the bacterial host; the experiment was performed for all four phages, each of which was passaged separately with the host strain.

2. Scenario B: similar to scenario A, but with daily addition of an evolutionarily naive (ancestral) host (OD600 = 0.3); for phages EC151 and EC152, E. cloacae CEMTC2064 was added, for phages StM171 and StenM174, the original S. maltophilia strain CEMTC2355 was added.

3. Scenario C: co-evolution of phage with host in the presence of a competing phage; StenM174 and StM171 were prepared as a cocktail at equal titers before the start of the co-evolution experiment and passaged together, and the same was done for EC151 and EC152.

4. Scenario D: A combination of scenarios B and C; the same cocktails were prepared with daily additions of evolutionarily naive bacteria.

Each experiment began with growing bacteria to reach OD600 0.3, followed by the addition of phages at an MOI of 0.1. The total volume of the phage-bacteria mixture was 700 µl; the suspension was incubated at 37 °C with shaking at 180 rpm for 24 hours. At each passage, 35 µl of this mixture was transferred to a new tube containing 665 µl of LB medium. For scenarios B and D, an additional 10 µl of naive bacteria grown the same day to an OD600 of 0.3 were added to each passage, and the tubes from the previous passage were then stored at 4 °C.

Phages and bacteria collected after the 5th and 10th passages were separated by centrifugation at 5000 rpm for 5 minutes. Bacterial pellets obtained after the 5th and 10th passages (scenarios A and C) were resuspended and grown on LB agar in Petri dishes. The mixture of grown individual colonies was transferred to new Petri dishes three more times and analyzed by PCR to confirm the absence of residual phages. Bacteria from scenarios B and D were not used for further analysis due to the presence of evolutionarily naive bacteria added at each passage.

The centrifuged suspension containing the phage particles was transferred to new centrifuge tubes. A portion of the suspension was used to isolate phage DNA, and the number of phage genomes was determined using qPCR by comparison with control samples #2. Thus, for each phage, by the end of each experiment, eight different phage populations were obtained (adapted according to scenarios A, B, C, and D after passages 5 and 10) and six bacterial populations (adapted against one of the phages, adapted against a competing phage, and adapted against a mixture of both phages, after passages 5 and 10). Those phage and bacterial populations were used in the subsequent experiment in addition to the original (ancestral/naïve) phages and bacteria.

2.4. Assessing the Infectivity of Passaged Phages

Passaging a bacterial population with two different phages, either individually or in a mixture, resulted in the formation of different bacterial clones. To assess the success of phage adaptation, we examined two metrics: 1) whether the adapted phage population could infect all variable bacterial clones in addition to the original host; 2) whether an adaptation scenario increased phage infection efficiency in comparison to the ancestral phage.

To answer these questions, a portion of the suspension volume containing the phage populations obtained in each scenario (A, B, C, D) after the 5th and 10th passages in addition to the original phage was used to infect the bacterial populations obtained during the experiment from scenarios A and C after the 5th and 10th passages in addition to the original bacterial host. Bacterial cultures were grown to an optical density (OD600) of 0.3, and phages were added at a constant multiplicity of infection (MOI = 0.1). Incubation was performed overnight without mixing in 96-well plates at 37 °C in LB medium. Phage DNA samples for real-time PCR were then prepared as described previously (section 2.2). Thus, more than 500 phage-bacteria interactions were studied using qPCR, with each qPCR run including two technical replicates.

Phage infectivity was calculated as the ratio of the number of phage genomic particles after bacterial infection (calculated as titer) to the known titer of the control sample #1 (which represents the same phage before the adaptation experiment), after which this ratio was expressed as log10 PFU/mL. Thus, a significant increase in phage abundance after infection was considered as an indicator of phage ability to infect bacteria. The threshold value established for determining the ability of an adapted phage to infect bacteria was 0.5, which represents a 3-fold increase in the number of phage particles after infection of bacteria compared to the initial concentration, or 2.3 Cq according to qPCR.

$$Phageinfectionability={\mathit{log}}_{10}{\displaystyle \frac{qunatityofphageparticlesafterinfection}{quantityofphageparticlesbeforeinfection(control\#1)}}$$

The increase in infectivity was calculated by comparing the ability of the adapted phage to infect a bacterial population with the ability of the original phage to infect the same bacterial population. We considered only significant increases greater than 1 (i.e., a 10-fold increase in the number of phage particles infectable by the adapted phage compared to the initial phage, or 3.3 Cq according to qPCR). Finally, we also considered an adapted phage to have enhanced infectivity against a bacterial population, if the ancestral phage was unable to infect this population, while the adapted phage was

$$Increaseininfectionabilityofadaptedphage={\displaystyle \frac{adaptedphageinfectionabilityagainstabacterialpopulation}{ancestralphageinfectionabilityagainstthesamebacterialpopulation}}$$

Following the calculations, the observed outcomes are reported for each replicate individually.

2.5. DNA Isolation and Sequencing of Adapted Phages

Adapted populations of phage EC152 were propagated until a high titer of 10¹⁰ PFU/mL was obtained. Then phage DNA was isolated as described previously [20] and fragmented using Covaris S2 Focused-ultrasonicator (PerkinElmer U.S. LLC, Washington, USA) with target fragments length of approximately 500 bp. Fragmented DNA was purified using VAHTS DNA purification beads (Nanjing Vazyme Biotech Co.,Ltd, Jiangsu Province, China) and used for preparation of pair-end libraries with the NadPrep DNA Library Preparation Kit v2 and NadPrep Universal Stubby Adapter (UDI) Module (Nanjing Biotechnology Co., Ltd., China). DNA sequencing was performed using the GeneMind FASTASeq 300 sequencer and FCH flow cell PE150 (GeneMind Biosciences Co., Ltd., Shenzhen, China).

The reads obtained were filtered by quality using Trimmomatic v.0.39 [23] (accessed on 20 February 2026). The genomes were assembled de novo using SPAdes Genome Assembler v.3.15.5 [24] (accessed on 20 February 2026).

2.6. Bioinformatic Analysis of Phages’ Genomes

The ancestral phages’ complete genomes were aligned using BLAST+ 2.12.0 tool [25] on Proksee [26], their tail fiber genes were aligned using MEGA 11 [27]. The assembled EC152 genomes of the adapted populations were aligned manually to the reference genome of the ancestral phage EC152 (GenBank PP681140) using UNIPRO UGENE software [28]. Additionally, the resulting sequences of EC152 were compared to the ancestral phage using BLASTn [29], the proteins containing mutations were further analysed using HHpred [30] and PSIPRED [31].

3. Results

3.Optimization of Sample Preparation with Phage DNA as a Template for qPCR

The reproducibility of the qPCR data was confirmed by analysing the same phage samples control #2 in several runs. Apart from phage StM171, reaction efficiencies ranged from 95% to 105%, and the maximum standard deviation and Cq value variation for the same samples across runs were considered acceptable (≤0.3) (Table 5). No amplification of the negative controls (water) was observed.

Approximate phage titers were estimated by comparing the Cq values of diluted samples with known titers from the same runs determined by the double agar method. The detection limit of phages was determined, and we found that real-time qPCR can reliably quantify the concentrations of phages down to ~10⁴ PFU/mL, with sensitivity decreasing below 10³ PFU/mL.

3.2. Results of Adaptation of Stenotrophomonas Phages StM171 and StenM174

To assess the success of phage adaptation under different co-evolution scenarios, we evaluated two key metrics: (1) the ability of evolved phage populations to infect a panel of bacterial clones adapted to the phage itself or to a competing phage or to a cocktail of both phages; and (2) whether any adaptation scenario increased infection efficiency compared to the ancestral phage. Although the results were phage specific, several general trends were noted regarding the conditions that favoured the rise of generalist phages, capable of infecting most, if not all, bacterial clones. Phages obtained after ten passages were consistently more effective than those obtained after five passages. For S. maltophilia phages, the presence of an evolutionarily naïve host was more beneficial than co-adaptation with a competing phage.

Among the four studied phages, StM171 showed a marked loss of infectivity across most evolved bacterial clones. In both experiments, adapting StM171 with the competing phage StenM174 (scenarios C and D) abolished detectable infectivity against all bacterial clones tested (Figure 2aand Figure 2b). The one exception occurred when StM171 was adapted alone (scenario A) or in the presence of naïve host (scenario B) for ten passages. In one of these two experiments, these conditions yielded phage populations capable of infecting most bacterial clones, including the ancestral bacterial strain and the clones adapted to StM171 or StenM174 separately for five or ten passages. However, these successful populations could not infect any bacteria that adapted to a cocktail of both phages (Figure 2a).

Phage StenM174 adapted better than StM171. While resistance toward StenM174 still arose in the bacterial clones across both experiments, we observed that longer adaptation (ten passages) in the presence of a naïve host (scenario B) yielded phage populations that infected all the tested bacterial clones in both experimental replicates. StM171 had beneficial effects on the adaptation of StenM174 in two ways. First, in the first experiment, an apparent cross-sensitization was observed, where bacteria evolved against StM171 became sensitive to StenM174 (Figure 2c). Second, in the second experiment, adapting phage StenM174 within a cocktail with StM171, both with or without the presence of a naïve host (scenarios C and D), produced phage populations capable of infecting all bacterial clones (Figure 2d). Regarding enhanced infectivity, scenario B (adapting the phage alone in the presence of an evolutionarily naïve host) in experiment one, and scenario D (adapting the phage with the competing phage StM171 in the presence of an evolutionarily naïve host) in experiment two were the only ones where StenM174 populations displayed an increase in genome copies relative to the ancestral phage against all bacterial clones. The detailed values of the increase in phage quantity following the infection are provided in table S1.

3.3. Results of Adaptation of Enterobacter Phages EC151 and EC152

In contrast, for the E. cloacae phages, both the naïve host and the competing phage were beneficial for obtaining generalist phages with enhanced infectivity. For phage EC151 after ten passages, all adapted populations infected every bacterial clone in the first experiment; in the second experiment, scenario C produced phages that infected five of seven clones and showed an increase in genome yields. (Figure 3a). In the second experimental repeat, scenario C again produced the most effective phages among other scenarios; the phages produced could infect five out of seven bacterial clones and showed enhanced infectivity (Figure 3b). Scenario B (adaptation with a naïve host) was the next most successful, yielding a phage population able to infect all bacterial clones in the first experimental repeat and four out of seven clones in the experiment two, although enhanced infectivity was only observed in experiment two (Figure 3).

Phage EC152 displayed the broadest infectivity profile of the four phages. After ten passages, EC152 populations adapted in scenario A (no naïve host and no competing phage) were able to infect most of the bacterial clones in both experiments. Furthermore, populations from scenarios B, C and D (presence of a naïve host, a competing phage, or both) were able to infect all of the obtained bacterial clones in both repeats with enhanced infectivity compared to the ancestral phage (Figure 2c and Figure 2d).

3.4. Results of Sequencing of the EC152 Adapted Populations

Since EC152 was the only phage, which had consistent enhanced outcomes for all four scenarios after ten passages, we checked for mutations in the genome of the adapted EC152 populations. Four populations obtained after 10 passages were sequenced from the two experimental repeats. Complete genome sequencing with >200x coverage of EC152 populations from the two independent experiments after 10 passages revealed that genomes were largely (>99%) identical to the reference genome of the ancestral phage (GenBank PP681140), with seven SNPs identified via BLASTn alignment. The mutations included three nonsynonymous mutations within ORFs and four intergenic mutations. The distribution of the mutations among the phages varied, 6/8 of the phages included the intergenic mutations, and 7/8 included the mutation P14T in the transcriptional regulator protein, only phages trained in the presence of the competing phage EC151 (scenario C and D) in the second experiment included additional mutations: T26A in the baseplate protein and G342R in the tail fiber protein (Table 6).

The P14T mutation could affect DNA-binding affinity or promoter recognition, thereby optimizing lytic gene expression under evolutionary pressure. T26A substitution removes a hydroxyl and slightly changes local polarity and packing of the protein, this could affect folding stability, interactions with neighboring baseplate subunits, or how the tail fiber docks. However, it is less likely to change receptor binding, compared to the G342R substitution which replaces the small, neutral glycine residue with bulky, positively charged arginine residue (pKa ~12.5), potentially enhancing electrostatic interactions with negatively charged host receptors (e.g., LPS) at the C-terminal receptor-binding domain, possibly improving adsorption or host range (Figure 4).

Finally, the intergenic mutations were co-occurring, two intergenic mutations, PP681140: 4 C->T and PP681140:23 G->T, were located between the CDS of hypothetical proteins WZX10556.1 and WZX10819, and the other two intergenic mutations, PP681140:11597 C->T and PP681140:11603 C->G were located between the CDS with hypothetical functions WZX10585.1 and WZX10585.1.

4. Discussion

Phage-bacteria co-evolution is a process that can be used to enhance the therapeutic potential of phages, especially when countering resistance after administration. We tested phage adaptation across four scenarios: (A) phage alone, (B) phage + naive host, (C) phage and a competing phage without naive host, (D) phage and a competing phage + naive host. In previous research, the addition of a naïve host and a competing phage were found to be beneficial in some cases [32,33,34]. The adaptation was done over 10 passages and adapted phages and bacterial population were tested after the fifth and tenth passages. Adapted phages were used against the bacterial panels evolved under single phage or two phages’ pressures.

In our two biological replicates, adding an evolutionary naïve host and extending adaptation to 10 passages was associated with detection of a broader adaptability for all four phages. For EC151 and EC152, the presence of a competing phage or a naïve host trended toward higher infectivity in most tested conditions. These phages perhaps benefitted from their encoded counter anti-phage defence systems to maintain infectivity, namely preQ0 modification in EC151 [20] and NAD+ reutilization pathway in EC152.

In contrast, StM171 proved refractory to adaptation under most conditions tested here, especially in the presence of the competing phage StenM174. The poor performance of StM171 could be associated to the absence of its own DNA and RNA polymerases, or the high evolutionarily pressure caused by StenM174 with its higher lytic ability and faster adsorption time. The only instance where StM171 adapted was in the presence of an evolutionarily naïve host in scenario B; however, this result was not replicated. StenM174 benefitted from the presence of evolutionarily naïve host, and cross-sensibilization was shown in one experiment, bacteria resistant to StM171 became more vulnerable to StenM174. This result is similar to the collateral sensitivity in Klebsiella when bacteria adapted to one phage by losing the capsular polysaccharide, which increased their vulnerability to LPS-binding phages [35].

The observed inter-replicate variability underscores the stochastic nature of experimental evolution. Differences in the initial pool of spontaneous mutations may dictate divergent adaptive trajectories, a limitation that can only be addressed with a substantially larger number of replicate populations. Despite this, the mutations observed in the case of the EC152 phage, were consistent regardless of the scenario to which the phage adapted. In 6 out of 8 phages the same co-occurring intergenic mutations repeated and in 7 out of 8 phages the same mutation in the transcription regulator was observed. These mutations could be responsible for the high adaptability of EC152 under different stressors. Previous study has shown that mutations in transcription regulators like cro and cI genes of phage lambda, influence the decision between lytic and lysogenic life cycles [36]. Additionally, intergenic mutations could also affect transcription, as in lambda-like phages, which have been shown to affect infection efficiency [37].

While this study has a clear limitation of deploying only two biological replicates per condition. The results showed repeated patterns regarding the duration of the experiment or the addition of evolutionary naïve host regardless of the phage studied. Consequently, the results could be regarded as hypothesis-generating and descriptive and provide plausible directions for future work. We emphasize that the primary contribution of this work lies in demonstrating the feasibility and throughput of the qPCR-based screening platform, which can now be employed in larger, statistically powered evolutionary experiments. Lastly, two applicable outcomes could be utilized in future investigations: (1) qPCR screening platform applied to a complex experimental design can efficiently screen 500 interactions in a short period of time (1-2 days), allowing for the quicker selection of adapted phages for therapy. (2) The importance of creating bacterial panels for testing phages, evolving bacterial clones under diverse pressures (single phage, cocktails) creates better screening tools mimicking in vivo mutants.