A Pseudomonas plant pathogen uses distinct modes of stationary phase persistence to survive bacteriocin and streptomycin treatments

Antimicrobial treatment of bacteria often results in a small population of surviving tolerant cells, or persisters, that may contribute to recurrent infection. Antibiotic persisters are metabolically dormant, but the basis of persistence to membrane-disrupting biological compounds is less well-understood. We previously found that the model plant pathogen Pseudomonas syringae pv. exhibits persistence to tailocin, a membrane-disrupting biocontrol compound with potential for sustainable disease control. Here we compared physiological traits associated with persistence to tailocin and to the antibiotic streptomycin, and established that both treatments leave similar frequencies of persisters. Microscopic proling of treated populations revealed that while tailocin rapidly permeabilizes most cells, streptomycin treatment results in a heterogeneous population of redox and membrane permeability states. Sorting cells according to redox reporter intensity identied streptomycin persisters among the low-redox fraction, but tailocin persisters were only cultured from the fraction with intermediate redox activity. Cells from culturable fractions were able to infect host plants, while nonculturable redox-active cells were not. Tailocin and streptomycin were effective in eliminating all persisters when applied sequentially, in addition to eliminating cells in other viable states. This study identies distinct redox states associated with antibiotic persistence, tailocin persistence, and virulence, and demonstrates that tailocin is highly effective in eliminating dormant cells. infectious. In linking redox states to heterogeneous phenotypes of tailocin persistence, streptomycin persistence, and infection capability, this work will inform the search for mechanisms and markers for each phenotype.


Introduction
The phenomenon of bacterial physiological tolerance to antibiotics is a long-standing problem in treating infection. The small fraction of cells in a bacterial population that survive after sustained lethal doses of antibiotics, termed persister cells, are a potential source of recurrent infections (1) or new resistance mutations (2,3). Unlike genetically resistant cells, persisters occupy a low-metabolic state that is both nonheritable and reversible, where both growth and susceptibility are regained following antibiotic removal. The persister state may be induced in response to stress or stochastic variation in gene expression, and occurs at increased frequency in stationary phase populations due to nutrient starvation (4). Persistence is just one of several states of bacterial dormancy proposed in the literature (5), and researchers have observed phenotypic overlaps between persisters and viable but nonculturable (VBNC) cells, de ned as living cells which are not revivable in standard media (6,7). Single cell observation after live/dead vitality staining has been a useful strategy to phenotype heterogeneous populations, previously demonstrating that both antibiotic persisters and starvation-induced VBNC cells occupy a low-redox state (8), and that VBNC cells outnumber persister cells after antibiotic treatment (9). The low metabolic rate of persisters and VBNC cells confers protection from antibiotics that target active processes (10).
Physiological tolerance may be an important issue in controlling bacterial plant diseases, which cause signi cant economic losses (11). Plant pathogenic bacteria face a wide variety of stresses including antimicrobial treatments, extremes in temperature, dessication, nutrient starvation, and host redox defenses. VBNC cells are well-documented in plant pathogens; nonculturable populations arise in response to plant or environmental conditions and can revive to initiate novel infections (12)(13)(14)(15)(16). The role of culturable persisters in plant disease is not well understood, but at least two phytopathogen species have been observed to form persisters to the aminoglycoside antibiotic streptomycin and to tetracycline (17,18). Streptomycin and other antibiotics have been used for plant disease prevention since the 1950s, and extension records indicate that 8 antibiotic classes are currently applied to crops in different regions of the world (19,20). Antibiotic use is now restricted to a few speci c applications in the US, but the more than 40 tons of aminoglycosides annually applied to plants still far exceeds the amount used in clinical medicine (20,21). The prevalence of persistence, and its potential impact on disease recurrence and management, is an important question in phytopathology.
A promising strategy for eradicating antibiotic persisters is combination treatment with compounds that attack static structures rather than growth processes, including membrane-disrupting polymyxin antibiotics such as colistin (10). Candidate biological control treatments including phage hydrolases, bacteriocins, and antimicrobial peptides also disrupt bacterial membranes, and are also effective at killing antibiotic persisters (22)(23)(24)(25). Bacterial subpopulations can also survive membrane disrupting treatments in a conditional or nonheritable fashion (25)(26)(27)(28). Investigating the frequency and biological basis of phenotypic tolerance to membrane disruptors will be important to understand the limits and extend the durability of these biocontrol strategies. Tailocins, or phage tail-like particles that disrupt the membranes of the target cell, are a class of bacteriocins with promise for highly speci c control of plant disease (29). A tailocin puri ed from Pseudomonas syringae pv. syringae strain B728A e ciently kills the model bean pathogen P. syringae pv. phaseolicola strain 1448A (also referred to as P. savastanoi pv. phaseolicola in the literature, henceforth Pph). This and other tailocins prevent disease when applied prophylactically to plants (30)(31)(32). We recently found that a small fraction of the Pph population escapes lethal doses of tailocin via a nongenetic persister-like mechanism, and heritable tailocin resistance repeatedly arises in culture from the surviving population (30). We termed this "tailocin persistence" in light of its biphasic killing pattern.
In this study, we asked whether tailocin persistence in Pph is distinct from persistence to antibiotics, and hypothesized that tailocin persisters could be eliminated through combination therapy with antibiotics. We rst established that Pph exhibits persistence to streptomycin, then used redox and membrane integrity reporter dyes to determine that tailocin treatment is much more e cient than streptomycin in rapidly eliminating viable cells from the unculturable population. Cell sorting analysis determined that culturable tailocin persisters have a higher level of redox activity compared with those surviving streptomycin, and combination treatment with streptomycin and tailocin eliminated both culturable and nonculturable viable cells. Moreover, we found that culturable persisters to each treatment were able to infect host plants, while nonculturable redox-active cells were not. This study demonstrates a distinct physiological state of persistence to an effective membrane disrupting biocontrol agent, and establishes a foundation for future studies toward identifying persister eradication mechanisms.

Streptomycin and tailocin treatments yield similar frequencies of Pph persisters
We previously found that tailocin exposure results in a stable population of genetically susceptible Pph survivors, which we de ned as tailocin persisters (30). In this study, we asked whether Pph exhibits persistence to the antibiotic streptomycin, and whether streptomycin persistence is distinct from tailocin persistence. Pph growth curves were performed to establish the timing of early and late stationary phase (Fig. S1). Survival to 5X MIC streptomycin and tailocin was characterized through kinetic killing curve assays at log phase, early stationary phase (20h), and late stationary phase (96h, Figure 1). In both early and late stationary phase cultures, CFU counts declined to 0.04% of initial values within 3h of streptomycin treatment, remaining stable at subsequent timepoints (Fig. 1A). Similarly, 0.06% of early or late stationary phase Pph cells remained culturable after tailocin exposure, consistent with our previous observations, with the majority of killing occuring within a few minutes (Fig. 1B). Adding either treatment in log phase resulted in a lower proportion of survivors. For each assay, colonies from the surviving population were con rmed to be susceptible upon re-exposure to streptomycin and tailocin. These results demonstrate that stationary phase Pph populations form similar proportions of culturable persisters to both streptomycin and tailocin, although tailocin killing is far more rapid. Subsequent experiments in this study were performed on early stationary phase Pph unless otherwise noted, using the same doses and treatment durations as used for the experiments in Figure 1.

Streptomycin and tailocin treatments have distinct effects on population physiology
Having established that tailocin and streptomycin both eliminate most culturable cells, we next sought to compare their e ciency in eliminating the viable population, inclusive of nonculturable cells. We rst measured treatment-induced changes in total cell concentration and culturable frequency. Hemocytometer readings revealed that streptomycin did not cause a reduction in the total concentration of cells in stationary phase or log phase (Fig 2A, Fig. S2A). Tailocin caused a slight reduction in the number of cells, although this was only statistically signi cant in log phase (Fig 2A, Fig. S2B). Dilution plating con rmed that treatments reduced the proportion of culturable cells from 46% to 0.04% for streptomycin ( Fig. 2A) and to 0.05% for tailocin (Fig. 2B), consistent with our earlier measures of survival over the T0 population (Fig. 1). To compare the physiological states of streptomycin-and tailocinexposed Pph populations, we imaged cells on an agarose pad after staining with a combination of three uorescent dyes: the vitality indicator Redox Sensor Green (RSG), the red membrane permeability indicator Propidium Iodide (PI), and the blue membrane permeant nucleic acid stain Hoescht 33342. This strategy allowed imaged cells to be classi ed into ve categories (Fig. 2C): 1) redox-active with intact membranes (green/blue), 2) redox-active with compromised membranes (green/red/blue), 3) redoxinactive with compromised membranes (red/blue), 4) redox-inactive with intact nuclear material and membranes (blue), and 5) unstained "ghost" cells with no nucleic acid content, visible in phase-contrast only. The method was rst tested on log-phase and ethanol-killed cells to rule out signal interference or overlap between stains (Fig. S3), and we con rmed that the staining combination did not affect culturable rate of streptomycin-treated cells. In preliminary experiments we noted that all Pph cells showed permeabilization and loss of redox activity starting after two hours on the agarose pad, thus all imaging was performed within twenty minutes after placement on the pad.
In untreated stationary phase cultures, over 80% of the population was composed of redox active, or Category 1, cells ( Fig. 2D-E, Table S1). After streptomycin treatment, Pph cultures were evenly distributed through Categories 2-4, each comprising an average of 19-25% of the population. Notably, nearly half the treated population were redox-active cells with permeable membranes (Category 2) or redox-inactive cells with intact membranes (Category 4), two states not distinguished by common live-dead staining methods. Cells in Category 2 had a lower green signal intensity than those in Category 1, indicating that membrane-damaged active cells were associated with reduced redox activity (Fig. S4A). A Category 4-like state was previously associated with persister and VBNC cells in E. coli, which were also characterized by increased cell roundness (7); we determined that Category 4 individuals also had signi cantly increased average roundness compared with the four other categories of treated Pph cells (Fig. S4B).
Unlike streptomycin, tailocin treatment converted most of the stationary phase population to Category 3 (membrane compromised, inactive) within three minutes (Fig. S5). After four hours, only 3% of the remaining cells were in Category 1 (Fig. 2E, Table S1), and Category 4 cells were extremely rare. When the experiments were performed on cultures in log phase, streptomycin and tailocin-induced changes were similar to those in stationary culture, although there was an apparently reduced proportion of redox-active Category 2 cells after tailocin treatment compared to the log phase results (Fig. S2D). Together, the results indicate that streptomycin treatment shifts the majority of the Pph population into diverse physiological states, while tailocin treatment rapidly compromises redox activity and membrane integrity in the vast majority of the population. They also show that after either treatment, the proportion of redoxactive cells and other intact cells far exceeds that of culturable persisters.
Streptomycin and tailocin culturable persisters occupy distinct physiological states While microscopic studies were useful for pro ling redox and permeability changes following either treatment, this approach could not determine the culturability of each staining category. Therefore, we applied uorescence-assisted cell sorting (FACS) to determine whether the culturable and infectious fractions of streptomycin and tailocin-treated populations could be separated according to redox staining characteristics. Because propidium iodide had stained some redox-active Pph cells (i.e., Category 2 cells), we rst sought an alternate permeability stain that could provide a distinct live-dead separation in twocolor sorting studies. DRAQ7 is a far-red membrane permeant dye that has been validated in eukaryotic cell culture studies (33), but is not widely used for determining viability in bacteria. In microscopic analysis on Pph, DRAQ7 stained redox-inactive cells, but unlike PI, was not observed to co-stain with RSG ( Fig. S6). This indicated that DRAQ7 does not permeate cells with redox activity. A triple staining experiment using DRAQ7 instead of PI was performed on treated and untreated Pph to con rm that the stain yielded similar estimates of redox-inactive membrane compromised cells to PI (Table S2). persister induction treatment, we also treated stationary cells with the protonophore CCCP, a highly e cient inducer of multidrug-tolerant persisters in Pseudomonas aeruginosa (34). A killing curve assay demonstrated that 3h CCCP treatment (5x MIC, or 100 µg mL -1 ) resulted in a stable culturable population representing 13% of the initial count (Fig. S7). CCCP treated cultures showed slightly increased RSG staining in intact cells, with many cells permeabilized to DRAQ7 (Fig. 3G). Histogram analysis supported the nding that streptomycin and CCCP treatments generated populations with increased redox signal, while tailocin treatment largely abolished redox activity (Fig. 3H). For unknown reasons, ethanol-killed cells had a higher level of green uorescence in cell sorting than tailocin-permeabilized cells (Fig. 3H).
Cells were separated according to physiological state to determine the culturability of each fraction.
Optimization assays con rmed that no culturable cells could be recovered from DRAQ7-staining fractions, so we focused on the region of low DRAQ7 intensity. Cells were gated into fractions G1, G2, and G3, corresponding to the highest to lowest green uorescence intensity ( Fig. 3A and 3D-G). Cells intact after streptomycin, tailocin, and CCCP treatments primarily fell into the G1, G3, and G2 gates, respectively (Fig. 3E-G and Table S1). Cells collected from each sorting gate were plated on culture media. In untreated cultures, colonies were recovered from all fractions (Fig. 4A). After streptomycin treatment, over 99.5% of colonies recovered came from the low-redox G3 fraction (Fig. 4B), even though this fraction represented only 13% of gated cells (Fig. 3). In contrast, in the tailocin-treated culture over 99% of colonies were recovered from the G2 fraction ( Fig. 4C), despite this fraction containing only 0.2% of the total gated cells (Fig. 3). Colonies were cultured from both G2 and G3 fractions after CCCP treatment ( Fig.   4D), but similarly to the other two treatments, nothing was cultured from the G1 fraction. These results demonstrate that while streptomycin persisters occupy a low-redox state consistent with dormancy, tailocin persisters are associated with a state of moderate redox activity. Additionally, diverse treatments resulted in a lack of culturability in cells with high redox activity.
We next asked whether the culturability or redox activity of sorted fractions was associated with infectious capacity of the pathogen. Due to the low volume of sorted inoculum, pathogenicity of the fractions was assessed in a qualitative bean pod inoculation assay, and symptoms of watersoaking or necrosis were observed after ve days (Fig 4E-H). No symptoms developed after inoculation from the high-redox G1 fraction of any culture, even without antimicrobial treatment (Fig. 4E). For cultures treated with streptomycin, tailocin, or CCCP, symptoms were observed at sites inoculated with any fraction with a signi cant culturable population (~10 4 or greater CFU mL -1 , 4F-H). Symptoms were weakest in the tailocin treated cultures, but this may be attributable to the low number of culturable cells obtained through sorting. Notably, for streptomycin and tailocin-treated cultures, the fractions associated with the largest number of membrane-intact cells (G1 and G3, respectively) did not cause symptoms ( Fig. 3E-F and 4F-G). These results demonstrate that culturability in media is associated with infection capacity in antimicrobial stressed Pph, and that the highest redox fractions of all cultures were noninfectious.
Streptomycin and tailocin-treated cells colonize the host at the same rates as untreated cells Because streptomycin persisters were associated with low activity, we hypothesized that streptomycin persisters might colonize the plant at a slower rate than the more active tailocin persisters. Sorting did not yield a su cient number of cells to perform timepoint analysis, so we instead compared colonization rates of treated and untreated cultures that had been adjusted to contain the same concentration of culturable cells (2.5 ×10 4 CFU mL -1 ). Because streptomycin causes a vast decline in culturable cells, the streptomycin-treated inoculum contained roughly 700-fold more RSG-staining cells than the untreated inoculum. The tailocin inoculum contained a similar number of RSG-staining cells to the untreated inoculum, but a far greater number of permeabilized cells. Despite differing viable population sizes and physiologies, streptomycin and tailocin-treated cultures colonized bean leaves at the same rate as untreated cells (Fig. 5A) and were able to cause normal symptoms of leaf spot and chlorosis at 8 days (Fig. 5B). This nding indicates that the distinct physiological states of streptomycin and tailocin persisters do not delay their ability to colonize a susceptible host. It also suggests that in the streptomycin-stressed inoculum, the large populations of high-redox nonculturable cells may not make a signi cant contribution to early infection, or at least not enough to speed colonization of a susceptible host.
To test the latter hypothesis, we performed a second experiment in which streptomycin and untreated cultures were adjusted to contain the same proportion of redox-active cells, regardless of intensity or culturability. Inocula adjustments were based on microscopic observations of mean RSG staining from Figure 2. In this experiment, both inocula contained 7×10 4 visibly RSG-staining cells per mL, but the untreated culture contained an estimated 450-fold greater concentration of culturable cells than the streptomycin-treated inoculum. The streptomycin-treated population started growing in the leaf much more slowly than the untreated inoculum, with the population only increasing after a two-day lag (Fig.  5C). This further supports the hypothesis that in a physiologically heterogeneous antibiotic-stressed Pph population, the high-redox unculturable cells do not signi cantly contribute to early host colonization.

Streptomycin eradicates Pph persisters of tailocin
Having determined that Pph tailocin persisters exist in a distinct physiological state from streptomycin persisters, we next hypothesized that streptomycin could eliminate tailocin persisters and vice-versa. Antibiotic persisters often exhibit multidrug tolerance, so we also asked whether tailocin persisters could be eliminated by two other antibiotics, the bacteriostatic translational inhibitor tetracycline or the DNA replication inhibitor cipro oxacin. Cross-survival rates of streptomycin persisters were rst tested with a sequential treatment of tailocin, tetracycline, or cipro oxacin. 30 to 70% of the streptomycin persistent population remained culturable after tetracycline or cipro oxacin treatment, while only 1% remained culturable after tailocin treatment (Fig. 6A). Conversely, when tailocin persisters were washed and treated with tetracycline or cipro oxacin, means of 7.2 and 11.5% remained culturable after treatment, respectively, while no colonies could be recovered after streptomycin treatment (Fig. 6B). To determine how this compares to whole population survival rates to these antibiotics, we treated stationary phase Pph cultures with tetracycline or cipro oxacin alone, and measured CFU recovery at 6.5±2.3% and 0.06±0.02% of the initial population, respectively. Thus, compared to untreated Pph cells, tailocin persisters exhibited no survival to streptomycin, a similar survival rate to tetracycline, and a 178fold increased rate of survival to cipro oxacin. CCCP-treated cells were also highly multidrug tolerant, but fewer than 0.1% survived tailocin treatment (Fig. 6C).
To further check for elimination of viable unculturable cells, we concentrated and microscopically examined the streptomycin-treated tailocin persisters, and found that all cells stained with PI only or were unstained (Fig. S8). No Category 1, 2, or 4 cells were observed. To rule out the possibility of rare live cells reviving to colonize the host, bean leaves were inoculated with concentrated cultures after the combination treatment. No symptoms developed on leaves (Fig. S8), and no Pph colonies were recovered in leaves collected immediately after inoculation or at days 1-5. In summary, cells surviving antibiotic and CCCP treatments have a high propensity to survive treatment with other antibiotics, but are mostly eliminated by tailocin. Streptomycin treatment is highly effective at eliminating culturable tailocin persisters, while tetracycline and cipro oxacin are less effective.

Discussion
Membrane-disrupting treatments have long shown promise for disease control and for eradication of antibiotic persisters, and there remains an enormous trove of membrane-disrupting biological compounds still to be discovered (35). Understanding the basis and management of population-level tolerance to these compounds will be important to maximize their e cacy. Here, a study of physiological heterogeneity in the model plant pathogen Pph demonstrated that tailocin and streptomycin treatments have vastly different physiological consequences. The small fraction of culturable cells surviving each treatment exhibited distinct redox phenotypes, and corresponded closely with the fractions capable of causing infection. The study shows that that streptomycin and tailocin could be a potent combination treatment for sterilization of Pph cultures, including the elimination of viable nonculturable cells. The study also links redox states to heterogeneous persistence and virulence phenotypes, which could inform the search for associated mechanisms and markers.
Tailocins are triggered by recognition of speci c lipopolysaccharides (LPS) on the target cell surface, after which the tail is driven into the membrane using energy stored in its contractile structure (36). The consequences are rapid membrane depolarization and ATP depletion, as well as transcriptional and translational arrest, from as little as one particle per target cell (37,38). How then, can persisters survive and maintain redox activity? The nding of nondormancy would be consistent with an active mechanism of tailocin survival, and the rapid timeframe of killing suggests that this trait is expressed in a portion of the planktonic population rather than being tailocin-induced. The metabolic activity of tailocin persisters is reminiscent of conditional tolerance to the membrane-destabilizing peptide colistin in Pseudomonas aeruginosa and other animal pathogens. Colistin tolerance was associated with increased expression of Pmr proteins that modify LPS to reduce colistin a nity, meaning that only transcriptionally active cells avoid membrane destabilization (27,28,39). Tailocin sensitivity is also linked to the composition of LPS (40), which can vary according to changes in environment and gene expression (41). We recently found that an LPS cluster gene of unknown function affects the frequency of tailocin persisters without impacting tailocin susceptibility or host tness (30). One hypothesis consistent with our ndings would be that persistence derives from active tailocin avoidance, potentially through population heterogeneity in tailocin recognition targets or other susceptibility factors. Further molecular analysis of the persistent population will be needed to pinpoint the underlying mechanisms. The streptomycin sensitivity and cipro oxacin tolerance we observed in tailocin persisters also echoes recent work on colistin, which was found to be much more e cient in eliminating aminoglycoside persisters than eliminating cipro oxacin persisters in E. coli (10). The authors hypothesized that colistin works synergistically with aminoglycosides due to the membrane damage exerted by both treatments. Our study suggests that aminoglycoside synergy, and perhaps uoroquinolone cross-tolerance, could be common themes of LPS destabilizing antimicrobials.
This study employed complementary methods of ow cytometry and uorescence microscopy to pro le antimicrobial-induced changes in Pph populations. Microscopy was useful in distinguishing intermediate viability categories, demonstrating that a third of redox-active cells were permeable to PI after streptomycin treatment. Membrane-damaged live cells were previously found to self-repair and resuscitate from VBNC populations of Pseudomonas and Shewanella sp. (42), and have been observed in nonstressed growing populations of other bacteria (43). This study indicates that DRAQ7 provides a more con dent indication of fatal membrane damage for Pseudomonas. Streptomycin also induced a high proportion of "Category 4" cells retaining intact membranes and increased roundness but no redox signal, a state similar to one associated with persisters, VBNC cells, and newly dead cells in E. coli (7,8). We hypothesize that Category 4 includes the fraction associated with streptomycin persistence in sorting experiments, which similarly lacked redox signal above that of dead cells. Tailocin reduced the proportion of Category 4 cells in Pph cultures by 99.9%, consistent with its ability to target dormant cells. However, the abundance of cells in each microscopy category far exceeded the abundance of persisters, illustrating that each phenotype contains heterogeneity and that persistence levels cannot be anticipated by staining phenotype alone.
Streptomycin treatment shifted the majority of the intact Pph population to a state of increased RSG staining intensity. Antibiotics stimulate the production of reactive oxygen species (ROS) in bacteria (44), and antibiotic-induced increases in RSG intensity were recently associated with accumulation of ROS-protective reductases in Campylobacter jejuni (45). Thus we suspect that the high-redox fraction in Pph similarly re ects a reductase response to intracellular ROS production, although ROS-speci c methods would be needed to determine this conclusively. If so, the non-culturability of this fraction could be consistent with ndings that ROS avoidance is a marker of persistence and post-antibiotic culturability (46,47), although in our study even the fraction of moderate RSG intensity was unculturable after streptomycin removal. All intact cells would be counted as live or VBNC cells using common permeabilitybased quanti cation methods (48), but the high-redox cells we observed are distinct from the VBNC cells induced by long-term starvation, reported as being dormant and persister-like (6,7). This study demonstrates that the large live but unculturable fraction does not revive in a susceptible host or greatly contribute to short-term infection in a mixed population.
Even in the absence of antimicrobial treatment, the high-redox fraction of Pph did not cause symptoms on the host. In recent years it has become clear that pathogenicity on plants is often a heterogeneously expressed trait, with essential virulence factors produced in a population bistable manner in P. syringae and other plant pathogens (49,50). The virulent state is associated with a suppression of genes involved in active growth processes (51,52). This study links one avirulent Pph subpopulation to an increased redox signal. It is striking that ve hours of exposure to a long relied-upon disease control treatment did not greatly reduce the number of intact and active cells, but rather shifted much of the population toward a non-infectious state. A more complete understanding of how antibacterial treatments affect pathogen physiology, both in the lab and eld, will be essential in tailoring disease control strategies that are more effective in reduction of pathogen inoculum.

Declarations
Zhao Zhao at the Yale Flow Cytometry Facility at the Yale School of Medicine for performing Fluorescence Assisted Cell Sorting, and Dr. Ann Haberman for helpful advice.

Con icts of Interest
The authors declare that they have no competing interests affecting this work.

Materials And Methods
Bacterial strains, plant lines, and culture conditions Tailocin was prepared from cultures of P. syringae pv. syringae (Psy) strain B728a. Experiments were performed using P. syringae pv. phaseolicola (Pph) strain 1448A. Cultures were grown from a single colony in King's B medium (53) at 28° C, 200 rpm shaking, unless otherwise indicated. Common bean (Phaseolus vulgaris) variety Kentucky Wonder (Seed Savers' Exchange, Decorah, Iowa) were grown in disposable plastic pots (8 × 6 cm and 8 cm deep) in PRO-MIX growing medium (BX, M) and maintained at 23°C with 70% relative humidity and 16 hours daylength in a Conviron growth chamber. For bean pod inoculations, plants were grown in a greenhouse (24 to 26 °C) in large pots (12 cm diameter and 12cm deep). Pods were collected from 50 to 55 day-old bean plants.

Tailocin preparation
Tailocin was prepared and quanti ed from supernatants of Psy B728a as previously described (32,54). Overnight cultures of B728a were diluted 1:100 in King's B, grown for 3 hours at 28 °C, and tailocin production was induced by addition of mitomycin C (MP Biomedical LLC, Solon, Ohio) to a concentration of 0.5 µg mL -1 . After 24h induction, supernatants were collected by centrifugation. Residual live cells were killed by treating supernatant with chloroform. The aqueous phase was collected by centrifugation, then amended with NaCl and Polyethylene glycol 8000 to nal concentrations of 1M and 10% w/v, respectively. After 1h incubation on ice, the supernatant mixture was centrifuged at 16,000 x g for 30 minutes at 4 °C. The resulting tailocin pellet was dissolved in 10 mM Tris (pH 7.0) and 10 mM MgSO 4 .
Residual PEG 8000 was removed by two extractions with equal volumes of chloroform. The activity of prepared tailocin was evaluated by spotting 5 µl serial dilution onto soft agar overlay plates seeded with Pph. Tailocin activity was expressed in activity units (AU) derived from the highest dilution factor resulting in a visible inhibition zone (55).

Minimum inhibitory concentration (MIC)
The MICs of streptomycin (MP Biomedical LLC, Solon, Ohio), tetracycline (MP Biomedical LLC, Solon, Ohio) and cipro oxacin (Acros Organics, Fair Lawn, New Jersey) to Pph were determined by evaluation of turbidity using a previously described method (56) with some modi cations. An overnight culture of Pph was diluted to an OD 600 of 0.1 in King's B medium, and 20 µL of the cell suspension was added to 180 µL of King's B amended with antibiotics to achieve nal antibiotic concentrations of 25, 12.5, 6.25, 3.12, 1.56, 0.78, 0.39, 0.19 and 0 µg mL -1 in 200 µL volume. Growth was assessed by measuring the OD 600 over 20 hours using an absorbance plate reader (Bio-Tek). The MIC of each antibiotic was the lowest concentration at which no increase in turbidity was measured across at least three independent cultures. MIC for tailocin was similarly determined in activity units (AU), starting with nine 1:2 serial dilutions of the initial tailocin preparation.
Killing curve of Pph after treatment with streptomycin and tailocin To prepare stationary phase cultures of Pph, a single colony was inoculated into 5 mL King's B broth, grown for 20h at 28 °C, diluted 1:100, and grown for 18 h (to a typical OD 600 of 1.3) or 4 days. To prepare exponential phase cultures, a 20 h culture was diluted 1:50 in King's B medium and incubated for 2.5 h (OD 600 = 0.15). To perform killing curve experiments, streptomycin was added to the cultures to reach a concentration of 16 µg mL -1 (5x MIC), followed by shaking incubation at 28 °C for 5 hours. 1 mL samples were collected prior to streptomycin addition (T=0) and hourly for 5 hours. Samples were centrifuged two minutes at 13,000 rpm and resuspended two times in sterile saline (0.8 % NaCl) and enumerated by dilution plating on King's B agar. Colonies were counted at 48h.
Tailocin killing curves were generated as previously described (30), with modi cations. Log, stationary, or 4-day Pph cultures were diluted to OD 600 = 0.1 in 0.8 % NaCl, and tailocin was added to a concentration of 250 AU mL -1 , which represents 5X MIC for Pph. Samples were removed before and immediately after addition of tailocin and then each hour for 4 hours. Samples were washed twice with saline and enumerated by serial dilution. Addition of the rst wash was typically completed in under 3 minutes from sample collection, therefore the sample removed immediately after tailocin addition was termed t=0.05.

Microscopic cell physiology analysis of Pph
Staining of Pph with redox sensor green (RSG) and propidium iodide (PI) was performed using the BacLight RedoxSensor Green Vitality Staining Kit (with additional staining with Hoechst 33342 (both from Thermo Fischer Scienti c, USA). Pph cultures at log and stationary phase were treated with streptomycin (5h) or tailocin (4h) as described above. Total cells were enumerated by hemocytometer count under phase contrast microscope. Culturable cells were enumerated by dilution plating, after washing the cells twice with saline (0.8% NaCl). Agarose pads (1.5%) were prepared on glass slides as previously described (57). Treated and untreated cells (10 µL) were amended with 0.1 µL RSG (1 mM), 0.1 µL PI (20 mM) and 0.15 µL Hoechst 33342 (1 mM) and incubated 10 m in the dark. 1 µL culture was placed on the middle of the agarose pad. Images were collected using a Zeiss Axio Imager M1 uorescence microscope within 15 minutes of placement on the pad. Multichannel images were captured using FITC, rhodamine, and DAPI lter sets in Zen 2.6 (blue edition) software. For each of four independent experiments, 10 elds were imaged across at least three different slides per treatment. All cells were counted in each image, totaling a range of 1000 to 1700 cells for each treatment and timepoint in each experiment, except for the tailocin 4h timepoint, for which there were 650-700 cells in the 10 elds. Single cells were classi ed into ve staining categories (green/blue, red/green/blue, red/blue, blue, or unstained) by visual comparison of the same cell under three different channels and in phase contrast. Cell counts were recorded by clicking on each cell using the Cell Counter plugin in Fiji (58). Intensity and roundness of selected cells were measured using the MicrobeJ plugin in Fiji (59).

Flow Cytometry
The physiological state of streptomycin and tailocin-treated cultures were evaluated through redox and cell-integrity staining followed by ow cytometric analysis. Four treatments were selected for ow cytometry: stationary phase Pph culture, and stationary phase treated with streptomycin, tailocin, and CCCP as described above Based on 50,000 sorted events, regions of green and red uorescence intensity de ned by stationary phase, ethanol-killed, and unstained Pph cells were used to de ne three gates associated with bright redox signal intensity (G1), medium intensity (G2), and low intensity (G3) among intact cells. Cultures were aseptically sorted into tubes until 10 7 events were collected, or for low-density gates, until the entire suspension was sorted. Collected fractions were adjusted to a nal concentration 0.1 % sterile peptone buffer, a common diluent used in Pph enumeration (60), in a nal volume of 1 mL. Isolated dots outside the polygon were not included in the analysis.
For culturing studies, sorted fractions were centrifuged at 13,000 rpm for two minutes and pellets were resuspended in 50 µL PBS. 20 µL aliquots of the suspension were serially diluted and 5 µL was spotted in triplicate on King's B agar plates. Colonies were enumerated after 48h incubation at 28 °C.
Sorted fractions were inoculated on detached bean pods according to the method of Bozkurt and Soylu (61). In brief, mature bean pods (from 50-day old plants) were collected, washed in distilled water, surface sterilized in 70% ethanol, and pierced using sterile 10 µL pipette tips. 30 µL of the 50 µL concentrated sorted fraction were placed on the wound. Inoculated pods were stored in sterile plastic containers lined with moist Whatman lter paper and incubated in a 28 °C chamber. Disease symptoms were recorded at 5 days after incubation.

Pph inoculation to bean plants
Fifteen-day-old bean plants were inoculated with untreated, streptomycin-treated, tailocin-treated, or streptomycin and tailocin-treated stationary phase cultures of Pph. Treated cultures were prepared using the methods and ending timepoints described for stationary phase killing curves, with streptomycin/tailocin combination treatments incubated for 4 hours. Untreated cultures were diluted in PBS to OD=0.0001, and single antibiotic-treated cultures were diluted to achieve the same concentration of culturable cells as the untreated inoculum (Fig. 6A experiment) or the same concentration of RSGstaining cells as the untreated inoculum (Fig. 6B experiment). Concentration adjustments were made based on observations from repeated prior experiment. 200 μL inoculum was in ltrated into the underside of the primary leaves of bean plants using 1 ml BD syringes. Samples from in ltrated areas were collected at 0, 1, 2, and 5 dpi using a 1 cm cork borer. Leaf discs were collected into a 1.5 ml tube containing 200 μL 10 mM MgCl 2 and homogenized using disposable pellet pestles (Fischer Scienti c).
Homogenates were serially diluted on King's B agar supplemented with 50 μg mL -1 nalidixic acid, to which Pph is genetically resistant, and CFUs were enumerated after 48h incubation at 28 °C.

Antibiotic cross-tolerance experiments
To measure cross-tolerance of streptomycin-tolerant Pph cultures against other antibiotics and tailocin, stationary phase Pph cultures were washed and resuspended to OD600= 0.1 in saline and treated with 16 µg mL -1 streptomycin for 5h. Cells were washed twice and resuspended in saline to remove streptomycin, then treated for 4h with tailocin (250 AU mL -1 ), tetracycline (8 μg mL -1 ), or cipro oxacin (8 μg mL -1 ) before washing again and serially diluting. Low-CFU samples were resuspended in a reduced volume of saline (50 µL) after the nal wash. The same procedure was followed to determine the tolerance level of carbonyl cyanide m-chlorophenylhydrazone (CCCP)-treated or tailocin-treated stationary phase cultures to other antibiotics, with the following modi cations: cultures were treated with CCCP (Sigma Aldrich) at a concentration of 100 µg mL -1 for 3h, or with 250 AU of tailocin for 1h.

Statistical Analysis
Differences in total and culturable cells were assessed using a Student's T-test (two-tailed distribution with two-sample, equal variance calculations). Multiple comparisons were performed with one-way ANOVA. Means were separated using Tukey's Honest Signi cant Difference Test at p = 0.05. Statistical analyses were performed in R version 4.0.3.

Data Availability
Images of elds used to generate Figure 2, Figure S2, and Supplemental Tables 1 and 2   immediately after tailocin addition (0.05h, or 3 minutes) and hourly. Error bars represent the standard deviation of the mean for three replicate cultures. The experiment was performed three independent times.

Figure 2
Distinct physiological states in stationary phase Pph cultures after treatment with streptomycin or tailocin. Cultures were treated with streptomycin (A) or tailocin (B) and enumerated by hemicytometer (grey bars) or by dilution plating (blue bars). Asterisks represent signi cant difference compared to T0 (p<10-6). Tailocin-treated cultures averaged 20% fewer cells at T4 than T0, but this was not statistically      Letters on the box plots denote statistical groups within timepoints (p ≤ 0.05).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.