Introduction

Antibiotic resistance poses a significant threat to global public health, emphasizing the urgent need for innovative strategies to combat bacterial infections(Luong et al., 2020). Phage therapy has emerged as a promising alternative, yet the rapid emergence of phage-resistant bacterial mutants challenges its effectiveness(Oechslin, 2018; Salmond & Fineran, 2015; Strathdee et al., 2023).

Beyond resistance, antibiotic tolerance presents a critical concern, enabling bacteria to survive high concentrations of antibiotics (Balaban et al., 2019a; Balaban et al., 2004; Brauner et al., 2016), which can further promote the evolution of resistance (Levin-Reisman et al., 2019; Levin-Reisman et al., 2017; Liu et al., 2020). This raises an important question: Could analogous persistence and tolerance arise in response to phage infections, thereby contributing to the development of phage resistance?

Results

Bacterial Persistence Against Phage Infection

To address this question, we investigated the killing dynamics of Klebsiella pneumoniae under treatment with the lytic phage Kp11 (GenBank: ON148528.1). In analogy with the concept of antibiotic persistence, the majority of bacteria were rapidly eliminated following lytic phage treatment, while a small fraction (~10⁻⁵ to 10⁻⁶) survived, resulting in a bi-phasic killing curve (Fig. 1b). However, the bacteria survived likely comprised both persisters and resistant mutants. To quantitatively distinguish between resisters and persisters, colonies of surviving bacteria post-phage treatment were individually plated on media with and without phage. The fraction of persisters was estimated by subtracting the number of resisters from the total colony-forming units (CFU) (Fig. 1a, 1b). Subsequent killing experiments on the descendants of persisters revealed a comparable survival fraction to the ancestral population, suggesting that this survival was not attributable to genetic mutation but rather to phenotypic heterogeneity, i.e., phage persistence (Fig. 1c).

Figure 1

Canonical antibiotic persistence is characterized by an extended lag time or slow growth rate(Balaban et al., 2004; Brauner et al., 2016). However, previous studies showed that the high antibiotic persistence Escherichia coli strain hipA7 did not exhibit enhanced survival when exposed to lytic phages compared to its sensitive ancestor(Pearl et al., 2008). As there currently lacked a well-defined high persistence variant of K. pneumoniae, we turned to the E. coli strain KLY and its high-persistence derivative metG* and challenged these two well-studied strains with T4 phage, which was known for its lytic activity against E. coli. The observed similar survival rates suggested that increased lag time does not confer protection against phage infection (Fig. 1d and Supplementary Fig.SI).

Phage tolerance induced by heat stress

Given that antibiotic persistence can be triggered by stresses such as starvation(Betts et al., 2002; Nguyen et al., 2011), we explored whether phage persistence might similarly be affected or not. K. pneumoniae was subjected to various stress conditions prior to phage treatment. Remarkably, pre-treatment at elevated temperatures significantly reduced K. pneumonia’s susceptibility to lytic phage infection (Fig. 2a). Killing assays revealed that heat-treated bacteria exhibited a reduced killing rate following phage exposure (Fig. 2b), with effects being both time- and temperature-dependent (Supplementary Fig.S2). Moreover, infection curves indicated that heat-treated bacteria relapsed more frequently, suggesting a higher probability of resistance mutant emergence (Fig. 2c).

Figure 2

To determine whether heat treatment directly conferred increased phage resistance or not, we quantified the fraction of phage-resistant mutants before and after heat treatment. Our analysis revealed a negligible presence of resistant mutants in both groups (Fig. 2d). Consistent with this, mutation rates estimated using Delbrück-Luria tests(Luria & Delbruck, 1943) confirmed no significant difference (Fig. 2e). Furthermore, although the survival rate of heat-treated bacteria substantially increased after 2 hours of phage infection, the proportion of phage-resistant mutants remined comparatively small (Fig. 2d). These findings indicated that the enhanced defense against phage infection observed following heat treatment was not due to an increase in phage-resistant mutants, but rather reflected a phenotypic boost in survival analogous to antibiotic persistence(Brauner et al., 2016). Therefore, we propose that the observed increase in resistance frequency is due to the cumulative effect of a net increase in survival, similar to how antibiotic tolerance promotes resistance.

Inspired by this concept, we conducted experiments using a very small bacterial inoculum (~10³) to minimize the impact of preexisting spontaneous resistant mutants, ensuring that any observed resistance would stem from division during phage infection. As expected, untreated bacteria were completely eradicated (Fig. 2f), whereas turbidity was observed in wells containing heat-treated bacteria co-cultured with lytic phages, indicating the emergence of resistance mutants during treatment (Fig. 2f).

Microscopic observations further revealed that heat-treated bacteria could divide under phage pressure, while untreated bacteria were lysed by phages before dividing (Fig. 2g and Movie 1 and 2). This indicates that the tolerance induced by heat treatment facilitates the emergence and evolution of phage resistance, as heat-treated bacteria underwent more division events under phage pressure, ultimately increasing the likelihood of developing phage resistance(Levin-Reisman et al., 2017; Liu et al., 2020).

Heat-Induced Phage Tolerance challenges phage therapy

We next assessed the infection of K. pneumoniae by phages Kp7, Kp9, Kp10, and Kp11, as well as E. coli by phages T1 and T4, following pre-heat treatment. Our results revealed a consistent tolerance response across these experiments (Fig. 3a and Supplementary Fig.S3a, 3b). This observation was further validated using clinical K. pneumoniae strains K359 and K381, both of which exhibited a marked reduction in phage sensitivity post-heat treatment (Fig. 3b and Supplementary Fig.S3c)

Figure 3

Phage therapy often involves combining multiple phages to minimize the risk of resistance evolution(Chan & Abedon, 2012; Leon & Bastias, 2015). In light of this, we extended our investigation to phage combinations and found that heat-treated bacteria exhibited reduced sensitivity even to combined phage treatments (Fig. 3d). Notably, we identified a Kp7-resistant strain, R^Kp7, which remained susceptible to Kp11 (Supplementary Fig.S4a) and was also sensitive to the combination of Kp7 and Kp11 (Supplementary Fig.S4b). This approved the effectiveness of combination therapy against spontaneous resistance-conferring mutations to a single phage(Leon & Bastias, 2015). Intriguingly, however, post heat-treatment, the combination had a much higher propensity to fail due to the evolution of resistance to both phages (Fig. 3d). These findings highlight the necessity of considering phage tolerance as a crucial factor in the design of effective phage therapies.

Reduced Phage DNA Entry and Altered Envelope Stress Response in Heat-Treated Bacteria

To investigate the mechanisms underlying phage tolerance in heat-treated bacteria, we first focused on the early stages of phage infection: adsorption and DNA injection. Our analysis revealed that heat treatment did not alter the phage adsorption rate (Fig. 4a). Additionally, the production and structure of the bacterial capsule, the primary receptor for phage Kp11, remained unchanged (Supplementary Fig.S5). However, we observed a significant impairment in the entry of phage DNA into heat-treated bacteria (Fig. 4b), suggesting a hindrance in the DNA injection process.

Figure 4

To explore this further, we employed liquid chromatography-mass spectrometry (LC-MS) for total protein analysis and identified that the down-regulated proteins were predominantly associated with the bacterial envelope (Fig. 4c, 4d). In addition to a primary receptor, phages typically require a secondary receptor, often an outer membrane component(Sueki et al., 2020; Sun et al., 2006), to establish adsorption and initiate DNA injection. Based on these findings, we hypothesized that heat treatment may induce the loss or modification of certain envelope components, thereby inhibiting phage DNA injection and contributing to heat-induced phage tolerance.

The bacterial envelope serves as a barrier against environmental stresses(Dehinwal et al., 2024; Mitchell & Silhavy, 2019; Silhavy et al., 2010).We noticed that heat stress influenced not only the components mediating phage infection but also broader bacterial envelope dynamics, suggesting that overall envelope stress responses could be altered(Eberlein et al., 2018). To investigate this, we examined the response of heat-treated bacteria to various envelope stresses, including unfavorable conditions such as pH changes and exposure to toxic compounds (metals, bile salts, and polymyxins), and found substantial changes in susceptibility (Fig. 4e). Interestingly, heat-treated bacteria exhibited increased survival against polymyxins, similar to phage tolerance (Fig. 4e). Notably, heat-treated bacteria showed significantly reduced lipopolysaccharide (LPS) production, known to mediate polymyxin adsorption(Yin et al., 2020) (Fig. 4f). Further experiments using the LPS inhibitor CHIR-090 confirmed that bacterial survival under polymyxin treatment increased with LPS reduction (Fig. 4g). Given that polymyxin is a last-resort antibiotic for treating carbapenem-resistant K. pneumoniae, this crosstolerance with phages might present a significant challenge to already scarce treatment options.

Phage Resistance in Heat-Evolved pspA Mutant

Heat treatment enhances bacterial tolerance to lytic phages, prompting us to investigate whether mutants surviving heat treatment could also withstand subsequent phage infections or not. To address this, we conducted a serial evolutionary experiment under selective heat stress (Fig. 5a). In line with our hypothesis, we identified a mutant that exhibited significantly enhanced insusceptibility under both heat treatment and phage challenge (Fig. 5b, 5c). Moreover, this strain maintained growth under phage treatment, indicating that the mutation conferred phage resistance (Fig. 5d, 5e).

Figure 5

Whole-genome sequencing of the evolved strains identified a truncation in the gene pspA as responsible for the reduced sensitivity to both heat and phage treatment (Fig. 5f). The absence of similar phenotypes in the pspA knockout strain (Supplementary Fig.S6) suggested that this was a gain-of-function mutation in the pspA gene. This was further supported by the failure to restore the phenotypes upon reintroduction of the wild-type pspA gene into the mutant strain; however, overexpression of the mutated pspA gene in the ancestral strain exhibited a similar phenotype (Supplementary Fig.S6).

Phage shock proteins, encoded by the psp operon, are reported to be induced by both heat stress and phage infection(Flores-Kim & Darwin, 2016; Model et al., 1997). However, the underlying mechanism has remained elusive. Our findings led us to hypothesize that the psp operon might function through envelope biogenesis, thereby linking responses to heat and phage treatment.

To test our hypothesis, we first investigated whether the phage life cycle in the pspA* mutant was disrupted at an early stage, building on previous results regarding heat-induced phage tolerance. Adsorption assays revealed a significant impairment in phage adsorption to pspA*, suggesting alterations in receptors availability (Fig. 6a). Further analysis corroborated this observation, demonstrating a reduction in capsular production (Fig. 6b) and a complete loss of capsular structure (Fig. 6c).

Figure 6

Figure 7

Given the critical role of K. pneumoniae capsules in evading host immunity during septicemia(Huang et aL, 2022b), we assessed the immune evasion capacity of the pspA* mutant. The results showed that pspA* was rapidly cleared from mice blood, demonstrating a significant reduction in immune evasion capacity (Fig. 6d and Supplementary Fig.S7). A mild reduction in immune evasion was also observed in heat-treated bacteria (Fig. 6d). Notably, the pspA* mutant exhibited increased survival to polymyxin treatment (Fig. 6e) without a corresponding increase in the minimum inhibitory concentration (MIC) (Fig. 6f), suggesting a tolerant phenotype mediated by a reduction in LPS. Mass spectrometry analysis indicated that many down-regulated proteins in pspA* were envelope-associated (Fig. 6g), implying an important role for the pspA gene in maintaining envelope integrity.

Discussion

Previous reports have demonstrated that antibiotic persistence confers a distinct advantage to lysogenic phages but does not protect bacteria from lytic phages(Pearl et al., 2008). Here, we identified bacterial persistence against lytic phages, where bacteria survived extended phage treatment without developing resistant mutations. This introduces a novel strategy for bacterial defense against phages. Notably, our research revealed that heat stress enhances phage tolerance, subsequently promoting the emergence of resistance, aligned with observations in antibiotic research(Balaban et aL, 2019b; Levin-Reisman et al., 2019; Levin-Reisman et aL, 2017). This parallel emphasizes the necessity of cautious and strategic design of phage therapy.

Heat, as a universal stress, induces protein misfolding and aggregation, despite the activation of heat shock proteins that mitigate these effects(Feder & Hofmann, 1999; Mogk et aL, 2019). The accumulation of misfolded envelope proteins can become toxic, leading to cell death(Balchin et aL, 2016; Konovalova et aL, 2016; Takai et aL, 1998). Correspondingly, the relatively low efficiency of translocation and assembly of envelope proteins(Bakkes et aL, 2010; Dalbey & Kuhn, 2012; Vigh et aL, 2007) renders them more vulnerable under heat treatment(Schiffrin et aL, 2017), resulting in a decrease in abundance. This reduction in envelope protein might be an adaptive response to elevated temperatures, as evidenced by the observations following heat treatment. Consequently, this would lead to a decrease in phage receptors (such as capsule and outer membrane protein FhuA) and the receptor for polymyxin (LPS), thereby conferring tolerance.

This phenomenon may also occur in the heat-tolerant mutant pspA*, where the reduction in envelope components resulted in insusceptibility to phages and polymyxins, highlighting a potential regulatory balance governed by PspA. PspA, a key regulator in the phage shock protein system, functions as part of the envelope stress response system in bacteria, preventing membrane depolarization and ensuring the envelope stability(Flores-Kim & Darwin, 2016; Horstman & Darwin, 2012; Kleerebezem & Tommassen, 1993; Yamaguchi et aL, 2010). Recent studies have shown that PspA is analogous to ESCRT-III, a component of the ESCRT machinery in eukaryotic cells, which plays a crucial role in membrane remodeling(Liu et al., 2021),(Junglas et al., 2021).

The bacterial envelope acts both as a defense barrier against external stresses and as a conduit for invaders. While it reduces the invasion of harmful substances and maintains intra-extracellular balance(Arts et al., 2015; Raivio, 2014; Wood, 2015), it also provides contacts for invaders; for example, capsules serve as receptors for phages(Latka et al., 2021), and LPS interacts with polymyxins(Manioglu et al., 2022). Thus, heat-induced alteration to the envelope could be a double-edged sword. Clinically, heat stress would not only impact phage susceptibility but also alters the interaction between K. pneumoniae and the host immune system. This dual impact underscores the complexity of bacterial adaptive responses and suggests that environmental factors can significantly influence the outcome of phage therapy. Furthermore, increased sensitivity to acidic conditions and bile salts induced by heat stress challenges bacterial survival, particularly relevant to K. pneumoniae colonization sites in the human body(Gasink et al., 2009; Martin et al., 2016).

Lastly, the envelope stress responses induced by heat in bacteria lead us to consider the role of fever. We found that heat-induced phage tolerance is both time- and temperature-dependent; even moderately elevated temperatures, such as 42°C, can induce phage tolerance with prolonged treatment, indicating alterations in the bacterial envelope. This suggests that elevated body temperature, akin to fever, may affect bacterial survival and colonization by modulating their responses to the host immune system and antimicrobial agents.

Materials and Methods

Bacteria, Phages and Plasmids

K. pneumoniae and E. coli strains used in this work are listed in Table S1. E. coli strain DH5a was used for DNA cloning(Seidman & Struhl, 2001). The phages used in this work are also listed in Table SI. Plasmid constructions were performed in E. coli strain DH5a using standard methodologies and are listed in Table S2.

Phage purification & concentration

Overnight cultured bacteria were inoculated into 200 mL of fresh LB medium and grown to an OD₆₀₀ of ~ 0.3. Then, 100 μL of phage stock was added to the cultures for a few hours, after which the cultures were centrifuged at 4°C, at 4000 rpm when cultures turn to transparency. The supernatants were filtered using a 0.22 pm filter to wipe off bacteria, and the phage cultures were concentrated using ultracentrifuge tubes (100 KD). Filtered SM buffer was added to the concentrated phage cultures until a stable volume was achieved, and the final phage stocks were stored at 4°C(Bonilla et al., 2016).

Killing dynamic of phage infection and phage persisters definition

Killing dynamic of phage infection to K. pneumoniae were performed by treating cells (~ 10⁶ CFU/mL) with lytic phage Kp11 (MOI ~ 100) for 120 minutes while shaking at 220 rpm in a 15 mL tube. Samples were taken at indicated time points, washed three times with PBS buffer and CFUs were evaluated by plating on LB agar plates before and after phage infection. To determine the persistence fraction and obtain phage persisters, a single colony from post-phage treatment was inoculated into fresh LB broth and grown at 37°C with shaking at 220rpm for 6 hours. Then each culture was collected and dropped on solid LB agar plates supplemented with and without phages. Phage persisters were identified as those solely capable of growth on phage-free plates, with the fraction of persisters estimated by subtracting the number of resistant mutants from the total CFUs. After identification, descendants of persisters were diluted to ~ 10⁶ CFU/mL and infected with phage Kp11 for 40 minutes, with CFUs evaluated by plating before and after phage treatment over several cycles.

Plaque assays

Phage titers and efficiency of plating were determined using the small drop plaque assay method(Goldfarb et al., 2015). 200 μL of bacterial culture was mixed with 5 mL of melted 0.7% soft LB agar medium and overlaid on 1.5% solid LB plates to form bacterial lawn. After allowing the bacterial lawn to dry for 15 minutes, 5 μL of 10-fold serial dilutions of phages in SM buffer were dropped to the lawn. Once the drops had dried, the plates were inverted and incubated at 37°C overnight. Plaque forming units (PFUs) were determined by counting the derived plaques after incubation, with lysate titers calculated as PFUs per mL. If no individual plaques could be observed, a faint lysis zone across the drop area was considered equivalent to 10 plaques. The efficiency of plating (EOP) was measured by comparing results from control and treated bacteria.

ScanLag analysis

Bacteria were diluted serially and plated on solid LB agar medium. The plates were placed in a ScanLag(Levin-Reisman et al., 2010) setup at 37°C, which consists of an array of office scanners that automatically capture images of the plates every 20 minutes, thus monitoring the appearance of colonies. An automated image analysis application extracts the distribution of appearance time of the colonies.

Growth dynamic of phage infection culture

In this work, bacterial cultures with and without heat treated were diluted to 10⁵ − 10⁶ CFU/mL in LB medium and infected with high concentration of phages at an MOI of 100. Optical density measurements at a wavelength of 600nm were taken every 10 minutes using a TECAN Infinite 200 plate reader in a 96-well plate.

Fluctuation test for measuring mutation rate (modified from Delbruck-Luria Fluctuation test)

To measure the mutation rate of heat-treated bacteria to phage, several replicas of culture (~ 1000 bacteria per well) were grown overnight in a 96-well plate to 10⁹ CFU/mL in 200 μL and treated with and without heat stress. Then each culture was plated onto solid agar plates supplemented with 10⁹ PFUs of phages and incubated at 37°C. The fraction of empty plates was used to calculate the mutation rate from five biological repeats of 14 replicas(Luria & Delbruck, 1943).

Resistance emergence fraction experiments

Bacteria culture was seeded in 96-well plate to achieve an approximate total bacterial load of 1000 CFU per well, minimizing the impact of spontaneous resisters and ensuring that any observed resistance was due to division during phage infection. Bacteria in each well were incubated with phages at a various MOI for 24 hours. Given the exceptionally low total bacterial load utilized in this experiment, the potential for phage-resistant bacteria due to stochastic mutations would be minimized. The high MOI guaranteed that each bacterium was likely to be infected by at least one phage particle. Due to the rapid infection and lysis in this system, the probability of observing turbid wells after 24 hours was minimal. The fraction of turbid wells was used to calculate the phage resistance emergence from three biological repeats of 14 replicates.

Time-lapse microscopy and image acquisition

To observe bacterial dynamics under phage treatment, time-lapse microscopy was performed using an FV3000 with a temperature-controlled stage. K. pneumoniae cultures with and without heat treatment were exposed to lytic phage Kp11 at an MOI of 100. For time-lapse imaging, bacterial cultures were transferred to a 35 mm glass-bottom dish (MatTek Corporation) and incubated at 37°C in a humidified atmosphere. The microscope was set to capture images at regular intervals (e.g., every 2 minutes for un-treated bacteria and 10 minutes for heat-treated bacteria). A 100x oil immersion objective was used to obtain high-resolution images. Images were captured using Ti2-E (Nikon, Japan) with a live-cell system FV3000 RS (Olympus, Japan). The resulting image stacks were analyzed using ImageJ.

Phage adsorption assay

The indicated bacterial cultures were diluted to fresh LB medium and infected with phages at different MOIs (0.1, 1, 10). 500μL of samples were collected at 3- and 10 minutes post-infection and centrifuged for 1 min at 12000 rpm. 500μL of LB broth with 50 μL of chloroform were added to the supernatant, which was then placed on ice for 10 min to remove bacterial body. Phage adsorption to the indicated bacteria was measured by titrating the free phages present in supernatant(Kim et al., 2019).

Transmission Electron Microscopy

The samples were fixed in a mixture of 2% paraformaldehyde and 2.5% glutaraldehyde, followed by four washes with 0.1 M phosphate buffer (PB). Post-fixation was performed using 1% osmium tetroxide and 1.5% tetrapotassium hexacyanoferrate trihydrate for 1 hour at 23°C. Samples were then dehydrated through graded ethanol solutions (50%, 70%, 80%, 90%, 100%, 100%, 100%) for 15 minutes each. Next, they were infiltrated twice with 1,2-epoxypropane for 15 minutes each, followed by gradient infiltration with a mixture of 1,2-epoxypropane and Epon 812 resin for 8 hours (SPI, America). The samples were then embedded in pure Epon 812 twice and polymerized in an oven at 60°C. Blocks of polymerized resin were sectioned using a Leica EM UC7 ultramicrotome (Wetzlar, Germany), yielding ultra-thin sections (70 nm) mounted on coated copper grids. Sections were stained on-grid with 2% uranyl acetate for 25 minutes and lead citrate for 5 minutes. Imaging was conducted using an H-7650B transmission electron microscope (Hitachi, Tokyo, Japan).

Phage DNA injection and replication kinetics monitoring

Indicated bacteria were infected with phages at an MOI of 10. At indicated time points, cells were harvested, and total DNA (including both bacterial and phage DNA) was extracted. qPCR was performed using primers specific to the phage genome (shown in Table S2), allowing quantification of phage DNA levels by comparing qPCR results.

Mass spectrometry

10 mL of bacteria for each sample were used for total protein extraction by using Bacterial Protein Extraction Kit (C600596, Sangon) and subjected to liquid chromatography-mass spectrometry (LC-MS). Briefly, proteins were disulfide reduced with 25 mM dithiothreitol (DTT) and alkylated with 55 mM iodoacetamide. In-gel digestion was performed using sequencing-grade modified trypsin in 50 mM ammonium bicarbonate at 37°C overnight. The peptides were extracted twice with 1% trifluoroacetic acid in 50% acetonitrile for 30 min. The peptide extracts were then centrifuged using a SpeedVac. For LC-MS/MS analysis, peptides were separated by a 60-minute gradient elution at a flow rate 0.300 μL/min with a Thermo-Dionex Ultimate 3000 HPLC system, directly interfaced with the Thermo Orbitrap Fusion mass spectrometer. The analytical column was a homemade fused silica capillary column (75 μm ID, 150 mm length; Upchurch, Oak Harbor, WA) packed with C-18 resin (300 A, 5 pm; Varian, Lexington, MA). Mobile phase A consisted of 0.1% formic acid, and mobile phase B consisted of 100% acetonitrile and 0.1% formic acid. The Orbitrap Fusion mass spectrometer was operated in the data-dependent acquisition mode using Xcalibur 3.0 software and there is a single fullscan mass spectrum in the Orbitrap (350-1550 m/z, 120,000 resolution) followed by 3 seconds data-dependent MS/MS scans in an Ion Routing Multipole at 30% normalized collision energy (HCD). The MS/MS spectra from each LC-MS/MS run were searched against the selected database using the Proteome Discovery searching algorithm (version 1.4).

Evolutionary protocol of cyclic exposure to heat stress

The cyclic evolution protocol (Fig. 5) consists of two steps: killing by heat stress at 50°C for 90 minutes and 24 hours growth. Three replicate populations were evolved, each initiated from a single colony isolated from the ancestral glycerol stock in 2 mL of fresh LB medium, which was shaken at 220 rpm at 37°C for 24 hours. The cultures were transferred to an elevated temperature of 50°C for 90 minutes. Then the cultures were resuspended in 2 mL of fresh LB medium and grown for 24 hours at 37°C with shaking. Half of the regrown cultures were used for analysis. After each cycle, samples from each culture were diluted serially and plated on solid LB agar medium to determine the survival.

Mouse Infection

Bacteremia modeling in mice followed the approach described by Huang et al(Huang et aL, 2022a). Female ICR mice (6-8 weeks old) were intravenously (i.v.) injected with K. pneumoniae. Bacteria were collected from pre-counted frozen stocks and resuspended in Ringer’s solution to the desired concentration for injection (100 μL). To calculate actual infectious doses, bacterial CFUs were plated one day prior to infection. Bacterial levels in the blood were measured by orbital hemorrhage at set time points post-infection, with the initial bacterial load (0 min) determined by dividing inoculum CFUs by blood volume (0.07 mL/g of body weight). Bacterial counts in organs were measured similarly, using tissue homogenates and estimating residual blood volume at 0.1 mL per gram of organ. The total bacterial burden was calculated by summing the CFU counts in the blood and organs, and survival rate was determined by dividing the total surviving bacteria by the initial inoculum CFUs.

Genomic extraction and whole-genome sequencing

Genomic DNA was extracted using the Tianamp Bacteria DNA Kit (Tiangen). Sequencing was performed by Sangon using the Illumina-HiSeq with 150 bp paired-end reads, at an average coverage of 100-200. Identification of single nucleotide polymorphisms (SNPs) was performed by aligning the raw Illumina reads to the genome of the reference strain ATCC 43816 using bwa version 0.7.18.

Statistical Analysis

All experiments were performed with at least three independent biological replicates unless otherwise specified. Bar graphs display the mean ± standard deviation (s.d.). Unpaired, two-tailed t-tests were used to calculate P values for comparisons of survival rates, EOP, and LPS production, with statistical significance set at P < 0.05. For the protein profiling experiments, differential protein abundance between the heat-treated and control groups were analyzed using Student’s t-tests for each protein. Multiple comparisons were adjusted using the Benjamini-Hochberg false discovery rate (FDR) correction to control for Type I errors. Proteins with a corrected P-value < 0.05 were considered significantly differentially expressed.

Data and materials availability

All data needed to evaluate the conclusions in the paper are present within the paper and/or the Supplementary Materials.

Acknowledgements

We thank Quan-Jiang Ji (ShanghaiTech University) for providing plasmids used in K. pneumoniae genome editing and Nathalia. Q. Balaban (Hebrew University of Jerusalem) for the E. coli strains KLY and metG*. We are grateful to Guan-Xiang Liang, Yao-Yu Yang, and Gao-Pu Zhang (Tsinghua University) for the phages, and Jun-Hao Zhu (Institute of Microbiology, Chinese Academy of Sciences) for manuscript comments. We acknowledge Hai-Teng Deng and Xian-Bin Meng (Proteomics Facility, Technology Center for Protein Sciences, Tsinghua University) for assistance with protein MS analysis, and Jing-Jing Wang (Cell Biology Facility, Center for Biomedical Analysis, Tsinghua University) for technical support. We also thank the Cell Biology Facility (Center for Biomedical Analysis, Tsinghua University) for equipment support.

Additional information

Funding:

National Key Research and Development Program of China (2022YFC2303202)

Tsinghua-Peking Joint Center for Life Sciences (20111770319)

Tsinghua University Dushi Program (20231080044).

Author contributions:

Conceptualization: F. Zhang and J. -F Liu

Additional files

Supplementary Video 1. Untreated bacteria were lysed by phages before cell division could occur. Shown is the rapid lysis of untreated K. pneumoniae ATCC43816 infected with phage Kp11 at an MOI of 100. The bacteria are lysed before they can undergo cell division, demonstrating the efficiency of phage-induced lysis at this infection level. Watch as bacteria are quickly disrupted following phage exposure.

Supplementary Video 2. Heat-treated bacteria were able to undergo cell division despite exposure to the lytic phage. Shown is the infection of heat-treated K. pneumoniae ATCC43816 by phage Kp11 at an MOI of 100. Unlike the untreated bacteria, these bacterial cells can be observed undergoing division before lysis occurs. The delayed lysis highlights the increased tolerance of heat-treated bacteria to phage infection under similar conditions. This increased frequency for bacterial division facilitates the evolution of phage resistance, highlighting the impact of heat treatment on bacterial survival under phage infection.

Supplementary Materials

Supplementary Figures

Supplementary Fig.S1

Supplementary Fig.S2

Supplementary Fig.S3

Supplementary Fig.S4

Supplementary Fig.S5

Supplementary Fig.S6

Supplementary Fig.S7

Supplementary Tables

Table S1

Table S2

Table S3