The Tradeoffs Between Persistence and Mutation Rates at Sub-Inhibitory Antibiotic Concentrations inStaphylococcus aureus

Antibiotics, minimum inhibitory concentration, pharmacodynamics, population biology, 28 antibiotic resistance mutation rate, bacterial persistence 29 30 This PDF file includes: 31 Main Text 32 Figures 1 to 5 33 Tables 1 to 2 34 35 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 3

36 The rational design of the antibiotic treatment of bacterial infections employs these drugs to reach 37 concentrations that exceed the minimum needed to prevent the replication of the target bacteria. However, 38 within a treated patient, spatial and physiological heterogeneity promotes antibiotic gradients such that the 39 concentration of antibiotics at specific sites is below the minimum needed to inhibit bacterial growth. Here, 40 we investigate the effects of sub-inhibitory antibiotic concentrations on three parameters central to bacterial 41 infection and the success of antibiotic treatment, using in vitro experiments with Staphylococcus aureus 42 and mathematical-computer simulation models. Our results, using drugs of six different classes, 43 demonstrate that exposure to sub-inhibitory antibiotic concentrations not only alters the dynamics of 44 bacterial growth but also increases the mutation rate to antibiotic resistance and decreases the rate of 45 production of persister cells thereby reducing the persistence level. Understanding this trade-off between 46 mutation rates and persistence levels resulting from sub-inhibitory antibiotic exposure is crucial for 47 optimizing, and mitigating the failure of, antibiotic therapy. 48 49 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 4

50 In the rational design of antibiotic therapy, drugs are administered such that the concentration of the 51 treating drug exceeds the threshold needed to prevent the replication of the target pathogen [1]. However, 52 in a treated individual, the concentration of an antibiotic within the body varies across different anatomical 53 regions due to factors such as variations in vascularization and the pharmacokinetics (PK) of the treating 54 antibiotic [2]. Notably, even though antibiotics are administered such that the concentration of the drug in 55 the serum exceeds the minimum inhibitory concentration (MIC), they are often present at sub-inhibitory 56 concentrations over time throughout the body [3]. Despite this, almost all studies on the pharmacodynamics 57 (PD) of antibiotics focus on super-inhibitory concentrations, ignoring the effects of sub-inhibitory 58 concentrations of antibiotics on bacteria. 59 60 In this study, we utilize a laboratory strain of the clinically significant pathogen Staphylococcus aureus [4] 61 to examine the impact of exposure to sub-inhibitory concentrations of six antibiotic classes on growth 62 dynamics, mutation rates, and the level of persistence. Persistence is the fraction of quiescent bacterial cells 63 that survive treatment with a super-inhibitory concentration of an antibiotic [5]. In a previous study with 64 Escherichia coli, we have demonstrated that exposure to sub-inhibitory concentrations of antibiotics results 65 in a decrease in the growth rate along with the maximum bacterial density achieved, as well as an increase 66 in the lag phase (the time before the bacterial population begins to replicate) [6]; we confirm the generality 67 of those findings here. Moreover, other studies have established that super-inhibitory antibiotic 68 concentrations can elevate the mutation rate for resistance to other drugs [7] we have found that this 69 phenomenon extends to sub-inhibitory antibiotic concentrations as well. Finally, we provide evidence that 70 pre-exposure to sub-inhibitory concentrations of antibiotics decreases the level of persistence. 71 72 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 5

73 The Effects of Sub-inhibitory Concentration of Antibiotics on Bacterial Growth Dynamics 74 To determine the effects of exposure to sub-inhibitory concentrations of antibiotics on the growth dynamics 75 of bacteria, we follow the changes in the optical densities of Staphylococcus aureus Newman exposed to 76 sub-inhibitory concentrations of antibiotics from six different classes in Fig. 1 [8]. The growth dynamics of 77 S. aureus Newman vary among the drugs for all six antibiotics, however, there is a clear concentration-78 dependent variation in the maximum growth rate (Fig. S1), the maximum optical density (Fig. S2), and the 79 lag time (Fig. S3). These results are consistent with those previously observed for Escherichia coli [6], 80 demonstrating that the results obtained previously are not restricted to Gram-negative bacteria. 81 82 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 6 83 Fig. 1. Growth dynamics of S. aureus with varying antibiotics and concentrations. Changes in the 84 optical density at 600nm (OD600) exposed to different concentrations of six classes of drugs. Lines 85 represent the average of five technical replicates. Each concentration is given as a fraction of the MIC 86 shown in Table S1: 1x (light blue), 0.5x (light purple), 0.25x (pink), 0.125x (green), 0.06x (blue), 0.03x 87 (purple), 0.016x (red), 0.008x (light green), 0.004x (dark blue), with a drug-free control shown in black. 88 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 7 The Effects of Exposure to Sub-inhibitory Concentration of Antibiotics on the Mutation Rate 89 Null Model of Mutation Rate 90 To explore the intrinsic variation in the estimation of mutation rates, we use a mathematical-computer 91 simulation model that employs the Monte Carlo process to generate mutants (Supplemental Text and 92 Supplemental Equations 1-4) [9]. Shown in Table 1 are five independent runs of this model each with 20 93 independent replicates. Though there is variation in the estimated mutation rate between runs, this variation 94 is not statistically significant. 95 Table 1. Variation in Mutation Rates Estimated from a Monte Carlo Simulation of Random 96 Mutation. 97 Null Model Mutation Rate Predictions Trial 1 2.96x10-9 Trial 2 3.44x10-9 Trial 3 3.43x10-9 Trial 4 2.20x10-9 Trial 5 2.98x10-9 98 Changes in the Mutation Rate due to Sub-inhibitory Drug Pre-exposure 99 To determine the effect sub-inhibitory pre-exposure has on the mutation rate to antibiotic resistance, we 100 exposed S. aureus Newman to the concentration of the six drugs above that did not change the maximum 101 stationary phase density. After 24 hours of pre-exposure, we performed a Luria-Delbruck fluctuation test to 102 determine the mutation rate to streptomycin resistance (Table 2) [10]. Notably, pre-exposure to sub-103 inhibitory concentrations of antibiotics significantly increased the mutation rate to streptomycin resistance, 104 a result unanticipated by the null model. 105 106 To elucidate the contribution of the generalized bacterial stress response, known as the SOS response, to 107 the increase in mutation rate, we repeated the above experiments with a strain lacking recA, the major 108 constituent of the SOS response [11]. When this knockout strain was pre-exposed to the same fraction of 109 the MIC of each drug, there was no evidence of a significant increase in the mutation rate (Table 2). The 110 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 8 recA knockout was resistant to fosfomycin, ceftriaxone, and azithromycin; thus, these antibiotics could not 111 be used for pre-exposure of this strain (Table S1). The background strain for this knockout, JE2, was found 112 to have a higher baseline mutation rate than Newman (4.01x10-8 ± 8.87x10-9). However, when pre-exposed 113 to sub-inhibitory concentrations of antibiotics, JE2 still exhibited a 10-fold increase in the mutation rate to 114 streptomycin (p=0.008, n=20). 115 116 Streptomycin was the only drug used to estimate the mutation rate, although other antibiotics were tested. 117 For this experiment, the mechanism of resistance must be a single point mutation, which significantly limits 118 the classes of drugs that could be used. Tobramycin, another aminoglycoside, was found to have an 119 extremely high baseline mutation rate (due to its inability to be enumerated on a fluctuation test), and thus 120 any increase in the rate could not be observed. The fluoroquinolones were found to have too low of a 121 baseline mutation rate, such that it was below the limit of detection. Interestingly, S. aureus Newman was 122 found to be heteroresistant to the quinolone nalidixic acid, while it was not heteroresistant to the 123 fluoroquinolone ciprofloxacin (Fig. S4). 124 125 Table 2. Mutation Rates to Streptomycin Resistance in S. aureus Pre-Exposed to Different 126 Antibiotics. 127 S. aureus Newman JE2 Δ recA Control 5.05x10-9 ± 8.98x10-10 3.53x10-8 ± 4.08 x 10-9 Rifampin 4.96x10-8 ± 1.25 x10-8 * 3.45x10-8 ± 6.35 x 10-9 Vancomycin 4.29x10-8 ± 1.44 x10-8 ** 3.17x10-8 ± 6.88 x 10-9 Fosfomycin 4.13x10-8 ± 7.5 4 x10-9 ** - Ceftriaxone 2.32x10-8 ± 6.24 x10-9 ** - Azithromycin 6.98x10-8 ± 9.1 1 x10-9 *** - Gentamicin 1.80x10-8 ± 3.5 1 x10-9 ** 4.17x10-8 ± 7.44 x 10-9 128 *p< 0.05, **p< 0.005, ***p< 0.0005 129 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 9 The Effects of Exposure to Sub-inhibitory Concentrations of Antibiotics on the Level of Persistence 130 Null Model of Persistence 131 To determine the effect that the rate of persistence generation has on the final level of persistence, we 132 employed a mathematical-computer simulation model of persistence with differing rates of persister cell 133 generation (Supplemental Text and Supplemental Equations 5-8). In Fig. 2, we show that a higher rate of 134 persister cell generation results in a higher level of persistence at six hours, such that in a time-kill 135 experiment the total number of surviving cells would be higher in a rate-dependent manner. 136 137 Fig. 2. Predicted changes in the total cell density of a bacterial population capable of producing 138 persister cells to a bactericidal antibiotic . These simulations assume all parameters are equal between 139 runs except for the parameter x, the rate constant of persister cell generation. The other parameters used for 140 this simulation are A = 5.0, vS = 2.0, vP = 0, vMIN = -3.0, e = 5x10-7, MIC = 1.0, and r = 1000. 141 142 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 10 Changes in the Level of Persistence Due to Sub-inhibitory Drug Pre-exposure 143 To determine the effect that sub-inhibitory pre-exposure has on the level of persistence, we first had to 144 select drugs for which S. aureus Newman shows persistence—which is shown on time-kill curves as cells 145 that survive super-inhibitory drug exposure but do not replicate and do not have an increased MIC. In Fig. 146 S5, we show that daptomycin and tobramycin, two highly bactericidal antibiotics, both have differing levels 147 of persistence, whereas ciprofloxacin, tetracycline, and streptomycin do not exhibit clear evidence for 148 persistence at the tested concentrations[12]. We chose 6x MIC for tobramycin and 4x MIC for daptomycin 149 to perform subsequent time-kill curves to maximize the difference in the levels of persistence. To ensure 150 the drug-exposed survivors were due to persistence and not some other phenomenon such as tolerance or 151 resistance, single colonies from the last time point of the time-kills were selected, and the time-kill was 152 repeated. The time-kill curves with these colonies were qualitatively and quantitatively similar to those in 153 Fig. S5, showing that the surviving cells were indeed persisters (Fig. S6). MICs were performed on the 154 cells surviving the time-kills and their MIC was found to be the same as the parental strain, providing 155 evidence for persistence rather than resistance. 156 157 To elucidate the effects sub-inhibitory pre-exposure has on the level of persistence, we performed time-kill 158 experiments with the drugs and concentrations selected above. Cultures were pre-exposed for 24 hours to 159 the six antibiotics used in Fig. 1 at sub-inhibitory concentrations which were shown not to reduce the 160 stationary phase densities. As shown in Fig. 3 and Fi g. 4, pre-exposure to sub-inhibitory concentrations of 161 the six antibiotics decreased the levels of persistence to both tobramycin and daptomycin. Variation in the 162 initial density occurred due to the reduced densities generated by exposure to sub-inhibitory concentrations 163 of the drugs. 164 165 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 11 166 Fig. 3. Time-kill Experiments with Tobramycin. Six-hour time-kill curves were performed with 6x the 167 MIC of tobramycin (Table S1). Cultures were either pre-exposed for 24 h or not pre-exposed to sub-168 inhibitory concentrations of one of the six antibiotics; from there either the cultures were allowed to grow 169 in the absence or presence of tobramycin. Lines represent: no pre-exposure, no tobramycin (black); no pre-170 exposure, tobramycin (purple); pre-exposure, no tobramycin (pink); and pre-exposure, tobramycin (green). 171 172 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 12 173 174 Fig. 4. Time-kill Experiments with Daptomycin. Six-hour time-kill curves were performed with 4x the 175 MIC of daptomycin (Table S1). Cultures were either pre-exposed for 24 h or not pre-exposed to sub-176 inhibitory concentrations of one of the six antibiotics; from there either the cultures were allowed to grow 177 in the absence or presence of daptomycin. Lines represent: no pre-exposure, no daptomycin (black); no pre-178 exposure, daptomycin (purple); pre-exposure, no daptomycin (pink); and pre-exposure, daptomycin 179 (green). 180 181 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 13 Changes in Metabolic Activity Due to Sub-inhibitory Drug Pre-exposure 182 183 Persister cells enter a state of dormancy in which they reduce their metabolic activity. Accordingly, if 184 metabolism is increased, persistence levels will decrease [13]. To evaluate the effect that the pre-exposure 185 to sub-inhibitory concentrations of antibiotics has on bacterial metabolic activity, we measured the 186 intracellular amount of ATP via a luminescence assay. In Fig. 5 we show that pre-exposure to sub-187 inhibitory concentrations of the selected antibiotics increased the ATP levels, indicating a higher metabolic 188 rate that may account for the results in Fig. 3 and Fig. 4. 189 190 Fig. 5. ATP determination. Cultures were either pre-exposed or not pre-exposed to sub-inhibitory 191 concentrations of one of the six antibiotics: From there the amount of ATP in these cultures was 192 experimentally estimated after 24 hours of pre-exposure via luminescence at 560 nm. ** p< 0.005, ***p< 193 0.0005, ****p< 0.00005. 194 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 14

195 Antibiotics are prescribed to patients at concentrations designed to exceed the minimum concentration 196 necessary to prevent the replication of the target pathogen [14]. Therefore, the minimum inhibitory 197 concentration (MIC) is the dominant and often the unique pharmacodynamic parameter used to design 198 antibiotic treatments [15]. However, i n vivo conditions introduce significant variab ility in factors such as 199 local bacterial concentration at the infection site, replication rate, nutrient availability, and the immune 200 response [16]. Moreover, though the antibiotic is administrated at super-inhibitory concentrations, this 201 concentration may not be reached in all, or even most, locations of the body, including the infection sites 202 [17]. This means treatment occurs at gradients of antibiotic concentrations throughout the body, including 203 antibiotic concentrations insufficient to kill or prevent the replication of the infecting bacteria [18]. 204 205 Previous studies have revealed that exposing E. coli to sub-inhibitory concentrations of antibiotics leads to 206 decreasing both maximum growth rate and maximum optical density while increasing the lag phase of 207 growth [6]. Our results here confirm this phenomenon applies to S. aureus as well. These changes are 208 consistent through all six classes of drugs tested where a concentration-dependent response is observed; as 209 the concentration of the antibiotic increases, so does the degree of impairment of the growth dynamics. 210 These results show that significant antibacterial activity occurs at sub-inhibitory concentrations, in some 211 cases exceptionally lower than the MIC, suggesting that antibiotic may have clinical utility at sub-212 inhibitory concentrations. This may explain why infections can be successfully treated despite being 213 located in sites where super-inhibitory antibiotic concentrations are not achieved. Apart from locational 214 heterogeneity, sub-inhibitory antibiotic concentrations can also obtain due to suboptimal dosing, extending 215 the time between doses, and using partially inactivated drugs due to inappropriate storage. 216 217 Along with the changes in growth dynamics, sub-inhibitory exposure may lead to physiological changes in 218 the bacteria [19]. When bacteria are exposed to super-inhibitory concentrations of antibiotics, resistant 219 mutants in the population will be able to survive and replicate in the presence of this selective pressure due 220 to mutations [20, 21]. Mutation rates, including those of antibiotic resistance, are not fixed. One pathway 221 that modulates mutation rates is the SOS response which is nearly ubiquitous in bacteria [22]. This 222 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 15 response plays a vital role in DNA repair and enables survival under physiological stress. Several external 223 factors can lead to the activation of the SOS response [23]. Our results illustrate one of these factors is 224 exposure to sub-inhibitory concentrations of antibiotics (Table 2). The major regulator of the SOS response 225 is RecA [11]. In S. aureus, there are two major pathways involved in this response: the LexA dependent 226 pathway which results in the expression of the UmuC error prone polymerase, and the RexAB dependent 227 pathway which results in the formation of small colony variants [24-26]. Taken together, activation of both 228 pathways results in an increase in the mutation rate of at least one order of magnitude, as we have shown in 229 Table 2. As expected, when recA is knocked out, these pathways cannot be activated, and pre-exposure to 230 sub-inhibitory concentrations of antibiotics does not lead to a change in the mutation rate. 231 232 Another phenomenon that could arise from pre-exposure to sub-inhibitory concentrations of antibiotics is 233 changes in persistence levels. Persistence is a temporary phenotypic change in which the majority of the 234 population is susceptible to antibiotics and a minority population is capable of surviving exposure to 235 antibiotics without an increase in the MIC [27, 28]. Persister cells can survive antibiotic treatment by 236 entering a dormant or slow-growing state, due to several possible mechanisms [29, 30]. Different 237 environmental factors can change the frequency of generation of persister cells in a bacterial population; 238 our results here show that one of these factors is the exposure to antibiotics—in this case, sub-inhibitory 239 levels of six distinct antibiotics. When bacteria are confronted with sub-inhibitory levels of antibiotics 240 before encountering super-inhibitory concentrations of other drugs, it triggers metabolic changes which 241 decrease the rate of generation of these persister cells. Our results further demonstrate that these metabolic 242 changes occur due to exposure to sub-inhibitory concentrations of antibiotics which is shown by a higher 243 intracellular ATP concentration (Fig. 5). This increase in metabolic activity opposes the dormancy that 244 defines persistence, therefore leading to a lower rate of persister cell formation when the bacterial 245 populations are then exposed to super-inhibitory concentrations of other drugs. Unexplored, but testable, 246 implications also arise from this increase in metabolic activity. Conceivably, toxins and other virulence 247 factors are also upregulated by exposure to sub-inhibitory concentrations of antibiotics. 248 249 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 16 These results contribute to our understanding of the interaction between bacterial mutation, persistence, and 250 antibiotics as an academic matter; however, there are serious clinical implications that follow these findings 251 as well [31]. The administration of a first line of antibiotic therapy will create a gradient of antibiotic 252 concentrations within the body. If this first treatment fails, and a secondary line of treatment is 253 administered, the increase in mutation rate produced in response to the sub-inhibitory concentrations in 254 different body locations could lead to the generation of resistant mutants which could then result in 255 treatment failure that would not otherwise have occurred. On the other hand, we show that persistence 256 would be reduced wherever there was pre-exposure to antibiotics. This ab ility to persist is an important 257 attribute for bacterial populations when conditions are unfavorable for their survival. As a result, pre-258 exposure to sub-inhibitory concentrations of antibiotics reducing persistence levels could enhance second-259 line treatment efficacy, improving the effectiveness of super-inhibitory concentrations of the antibiotic used 260 in therapy and therefore reducing the risk of recurrent infections [32]. These results are especially salient in 261 chronic and recurrent infections such as those involving biofilms [33]. Ultimately, these findings boil down 262 to one important trade-off that has real-world impacts in the clinic, that is a trade-off between higher 263 mutation rates and lower persistence levels resulting from previous exposure to sub-inhibitory 264 concentrations of antibiotics. 265 266 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 17

267 Growth media 268 All experiments were conducted in Muller Hinton II (MHII) Broth (90922-500G) obtained from Millipore. 269 All bacterial quantification was done on Lysogeny Broth (LB) agar (244510) plates obtained from BD. E-270 tests were performed on MH agar plates made from MH broth (M391-500g) with 1.6% agar obtained from 271 HiMedia. 272 273 Growth Conditions 274 Unless otherwise stated, all experiments were conducted at 37°C with shaking. 275 276 Bacterial strains 277 All experiments were performed with Staphylococcus aureus Newman obtained from Bill Schafer of 278 Emory University. Je2 ΔrecA and Je2 from the Nebraska Transposon Mutant Library [34] were obtained 279 from Joanna Goldberg of Emory University. 280 281 Antibiotics 282 Streptomycin (S6501), sulfamethoxazole (S6377), vancomycin (V1130), ceftriaxone (C5793), fosfomycin 283 (P5396), and daptomycin (D2446) were all obtained from Sigma-Aldrich. Tobramycin (T1598) was 284 obtained from Spectrum. Azithromycin (3771) was obtained from TOCRIS. Ciprofloxacin (A4556) was 285 obtained from AppliChem Panreac. Gentamicin (BP918-1) and rifampin (BP2679-1) were obtained from 286 Fisher. Nalidixic acid (KCN23100) was obtained from PR1MA. Tetracycline (T17000) was obtained from 287 Research Products International. All E-test strips were obtained from Biomérieux. 288 289 Sampling bacterial densities 290 The densities of bacteria were estimated by serial dilution in 0.85% saline and the total density of bacteria 291 was estimated on LB plates with 1.6% agar. 292 293 294 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 18 Growth rate estimation 295 Exponential growth rates were estimated from changes in optical density (OD600) in a Bioscreen C For 296 this, 24-hour overnight cultures were diluted in MHII to an initial density of approximately 10 5 cells per 297 mL. Five technical replicates were performed for each condition in a 100-well plate. The plates were 298 incubated at 37°C and shaken continuously. Estimates of the OD600 were made every 5 minutes for 299 24/i1hours. Normalization was performed and then means and standard deviations of the maximum growth 300 rate, lag time, and maximum OD were found using an R Bioscreen C analysis tool accessible at 301 https://josheclf.shinyapps.io/bioscreen_app. 302 303 Minimum inhibitory concentration estimation via broth microdilution 304 MICs were determined according to the CLSI guidelines, deviating only in the choice of media [35]. 305 Briefly, 96-well plates with two-fold dilutions of antibiotics in MHII media were prepared and inoculated 306 with 10 5 bacteria per mL. An extended gradient was created by combining three sets of two-fold serial 307 dilutions from three starting antibiotic concentrations. The plates were incubated at 37°C with conditions 308 shaking and the optical density (OD600) was measured after 24 hours. 309 310 Fluctuation tests 311 Independent overnights of S. aureus Newman, Je2 ΔrecA, and Je2 were either exposed to sub-inhibitory 312 concentrations of rifampin at 0.5x MIC, vancomycin at 0.5x MIC, fosfomycin at 0.25x MIC, ceftriaxone at 313 0.25x, gentamicin at 0.25x, azithromycin at 0.25x, or grown without antibiotic and then plated on LB agar 314 plates containing 5x MIC of streptomycin. Experiments were performed with 20 biological replicates and 315 the mutation rates were calculated as in [36, 37] with BZrates.com. 316 317 Time kill experiments 318 Cultures of 10 7 S. aureus Newman were either exposed overnight to sub-inhibitory concentrations of 319 rifampin, vancomycin, fosfomycin, ceftriaxone, azithromycin, and gentamicin at the above concentrations 320 or were grown without antibiotics. After this overnight incubation, all cultures were diluted in fresh MHII 321 to 10 7 cells per mL. The cultures were then exposed to super-MIC concentrations of streptomycin, 322 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 19 daptomycin, tetracycline, tobramycin, or ciprofloxacin at varying concentrations, and viable cell density 323 was estimated at 0, 2, 4, 5, and 6 hours. 324 325 Population analysis profile test 326 PAP tests were performed as in [38, 39]. Briefly, a gradient of nalidixic acid or ciprofloxacin 327 concentrations was added to LB plates. The concentrations were 0, 0.5, 1, 2, 4, 8, 16, and 32 xMIC. 328 Multiple dilutions of S. aureus Newman (100-10-7) were then plated on every concentration. Colonies were 329 enumerated after 48 hours and the highest dilution with colonies present was recorded. The frequency of 330 surviving cells was calculated by dividing the highest density of cells at each concentration by the number 331 of surviving cells on plates with no antibiotics. 332 333 Numerical solutions (simulations) 334 For our numerical analysis of the mathematical models detailed in the Supplemental Text, we used 335 Berkeley Madonna, using parameters in the ranges estimated for S. aureus Newman. Copies of the 336 Berkeley Madonna program used for these simulations are available at www.eclf.net. 337 338 Statistical Analysis 339 Statistical significance analysis was carried out by paired t-tests using GraphPad Prism (version 10.2.0). 340 341 ATP Assay 342 ATP determination kits were obtained from ThermoFisher Scientific (A22066). To perform the ATP 343 determination, the manufacture’s provided protocol was followed with the following changes. Overnight 344 cultures either pre-exposed to the antibiotics or not exposed were pelleted and the pellets washed with 345 saline. Cultures were resuspended in saline and sonicated with a Branson Needle-Tip Sonicator. Post-346 sonication, cells were centrifuged, and the supernatants were placed in a black 96-well plate and incubated 347 at room temperature for 30 minutes. After incubation, luminescence was then read at 560 nm. 348 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 20 Acknowledgments 349 We thank the U.S. National Institute of General Medical Sciences for their funding support via R35 GM 350 136407 and the U.S. National Institute of Allergy and Infectious Diseases for their funding support via U19 351 AI 158080. FB acknowledges the support of CIBERESP (CB06/02/0053) from the Carlos III Institute of 352 Health of Spain. The funding sources had no role in the design of this study and will not have any role 353 during its execution, analysis, interpretation of the data, or drafting of this report. The content is solely the 354 responsibility of the authors and do not necessarily represent the official views of the National Institutes of 355 Health nor those of the Carlos III Institute of Health of Spain. 356 357 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 21

358 1. Velkov T, Bergen PJ, Lora-Tamayo J, Landersdorfer CB, Li J. PK/PD models in 359 antibacterial development. Curr Opin Microbiol. 2013;16(5):573-9. Epub 20130718. doi: 360 10.1016/j.mib.2013.06.010. PubMed PMID: 23871724; PubMed Central PMCID: 361 PMCPMC3834155. 362 2. Andes D, Anon J, Jacobs MR, Craig WA. Application of pharmacokinetics and 363 pharmacodynamics to antimicrobial therapy of respiratory tract infections. Clin Lab Med. 364 2004;24(2):477-502. doi: 10.1016/j.cll.2004.03.009. PubMed PMID: 15177850. 365 3. Landersdorfer CB, Nation RL. Limitations of Antibiotic MIC-Based PK-PD 366 Metrics: Looking Back to Move Forward. Front Pharmacol. 2021;12:770518. Epub 367 20211029. doi: 10.3389/fphar.2021.770518. PubMed PMID: 34776982; PubMed Central 368 PMCID: PMCPMC8585766. 369 4. Mulani MS, Kamble EE, Kumkar SN, Tawre MS, Pardesi KR. Emerging 370 Strategies to Combat ESKAPE Pathogens in the Era of Antimicrobial Resistance: A 371 Review. Front Microbiol. 2019;10:539. Epub 20190401. doi: 10.3389/fmicb.2019.00539. 372 PubMed PMID: 30988669; PubMed Central PMCID: PMCPMC6452778. 373 5. Balaban NQ, Merrin J, Chait R, Kowalik L, Leibler S. Bacterial persistence as a 374 phenotypic switch. Science. 2004;305(5690):1622-5. Epub 20040812. doi: 375 10.1126/science.1099390. PubMed PMID: 15308767. 376 6. Berryhill BA, Gil-Gil T, Manuel JA, Smith AP, Margollis E, Baquero F, et al. 377 What's the Matter with MICs: Bacterial Nutrition, Limiting Resources, and Antibiotic 378 Pharmacodynamics. Microbiol Spectr. 2023;11(3):e0409122. Epub 20230503. doi: 379 10.1128/spectrum.04091-22. PubMed PMID: 37130356; PubMed Central PMCID: 380 PMCPMC10269441. 381 7. Nagel M, Reuter T, Jansen A, Szekat C, Bierbaum G. Influence of ciprofloxacin 382 and vancomycin on mutation rate and transposition of IS256 in Staphylococcus aureus. 383 Int J Med Microbiol. 2011;301(3):229-36. Epub 20101105. doi: 384 10.1016/j.ijmm.2010.08.021. PubMed PMID: 21115395. 385 8. Eyler RF, Shvets K. Clinical Pharmacology of Antibiotics. Clin J Am Soc 386 Nephrol. 2019;14(7):1080-90. Epub 20190312. doi: 10.2215/cjn.08140718. PubMed 387 PMID: 30862698; PubMed Central PMCID: PMCPMC6625637. 388 9. Metropolis N, Ulam S. The Monte Carlo Method. Journal of the American 389 Statistical Association. 1949;44(247):335-41. doi: 10.1080/01621459.1949.10483310. 390 10. Lang GI. Measuring Mutation Rates Using the Luria-Delbrück Fluctuation Assay. 391

Mol Biol. 2018;1672:21-31. doi: 10.1007/978-1-4939-7306-4_3. PubMed 392 PMID: 29043614. 393 11. Maslowska KH, Makiela-Dzbenska K, Fijalkowska IJ. The SOS system: A 394 complex and tightly regulated response to DNA damage. Environ Mol Mutagen. 395 2019;60(4):368-84. Epub 20190107. doi: 10.1002/em.22267. PubMed PMID: 30447030; 396 PubMed Central PMCID: PMCPMC6590174. 397 12. Dawan J, Wei S, Ahn J. Role of antibiotic stress in phenotypic switching to 398 persister cells of antibiotic-resistant Staphylococcus aureus. Annals of Microbiology. 399 2020;70(1):1. doi: 10.1186/s13213-020-01552-1. 400 13. Aduru Sai V, Szenkiel K, Rahman A, Ahmad M, Fabozzi M, Smith Robert P, et 401 al. Sub-inhibitory antibiotic treatment selects for enhanced metabolic efficiency. 402 Microbiology Spectrum. 2024;12(2):e03241-23. doi: 10.1128/spectrum.03241-23. 403 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 22 14. Kowalska-Krochmal B, Dudek-Wicher R. The Minimum Inhibitory 404 Concentration of Antibiotics: Methods, Interpretation, Clinical Relevance. Pathogens. 405 2021;10(2). Epub 20210204. doi: 10.3390/pathogens10020165. PubMed PMID: 406 33557078; PubMed Central PMCID: PMCPMC7913839. 407 15. Udekwu KI, Parrish N, Ankomah P, Baquero F, Levin BR. Functional 408 relationship between bacterial cell density and the efficacy of antibiotics. J Antimicrob 409 Chemother. 2009;63(4):745-57. Epub 20090213. doi: 10.1093/jac/dkn554. PubMed 410 PMID: 19218572; PubMed Central PMCID: PMCPMC2654042. 411 16. Baquero F, Negri MC. Selective compartments for resistant microorganisms in 412 antibiotic gradients. Bioessays. 1997;19(8):731-6. doi: 10.1002/bies.950190814. PubMed 413 PMID: 9264256. 414 17. Baquero F, Coque TM. Widening the Spaces of Selection: Evolution along 415 Sublethal Antimicrobial Gradients. mBio. 2014;5(6):10.1128/mbio.02270-14. doi: 416 doi:10.1128/mbio.02270-14. 417 18. Lagator M, Uecker H, Neve P. Adaptation at different points along antibiotic 418 concentration gradients. Biol Lett. 2021;17(5):20200913. Epub 20210512. doi: 419 10.1098/rsbl.2020.0913. PubMed PMID: 33975485; PubMed Central PMCID: 420 PMCPMC8113895. 421 19. Kong KF, Schneper L, Mathee K. Beta-lactam antibiotics: from antibiosis to 422 resistance and bacteriology. Apmis. 2010;118(1):1-36. doi: 10.1111/j.1600-423 0463.2009.02563.x. PubMed PMID: 20041868; PubMed Central PMCID: 424 PMCPMC2894812. 425 20. Luria SE, Delbrück M. Mutations of Bacteria from Virus Sensitivity to Virus 426 Resistance. Genetics. 1943;28(6):491-511. doi: 10.1093/genetics/28.6.491. PubMed 427 PMID: 17247100; PubMed Central PMCID: PMCPMC1209226. 428 21. Gullberg E, Cao S, Berg OG, Ilbäck C, Sandegren L, Hughes D, et al. Selection of 429 resistant bacteria at very low antibiotic concentrations. PLoS Pathog. 430 2011;7(7):e1002158. Epub 20110721. doi: 10.1371/journal.ppat.1002158. PubMed 431 PMID: 21811410; PubMed Central PMCID: PMCPMC3141051. 432 22. Crane JK, Alvarado CL, Sutton MD. Role of the SOS Response in the Generation 433 of Antibiotic Resistance In Vivo. Antimicrob Agents Chemother. 2021;65(7):e0001321. 434 Epub 20210617. doi: 10.1128/aac.00013-21. PubMed PMID: 33875437; PubMed Central 435 PMCID: PMCPMC8373240. 436 23. Podlesek Z, Žgur Bertok D. The DNA Damage Inducible SOS Response Is a Key 437 Player in the Generation of Bacterial Persister Cells and Population Wide Tolerance. 438 Front Microbiol. 2020;11:1785. Epub 20200804. doi: 10.3389/fmicb.2020.01785. 439 PubMed PMID: 32849403; PubMed Central PMCID: PMCPMC7417476. 440 24. Kahl BC, Becker K, Löffler B. Clinical Significance and Pathogenesis of 441 Staphylococcal Small Colony Variants in Persistent Infections. Clin Microbiol Rev. 442 2016;29(2):401-27. doi: 10.1128/cmr.00069-15. PubMed PMID: 26960941; PubMed 443 Central PMCID: PMCPMC4786882. 444 25. Clarke RS, Bruderer MS, Ha KP, Edwards AM. RexAB is essential for the 445 mutagenic repair of Staphylococcus aureus DNA damage caused by co-trimoxazole. 446 Antimicrob Agents Chemother. 2019;63(12). Epub 20191007. doi: 10.1128/aac.00944-447 19. PubMed PMID: 31591116; PubMed Central PMCID: PMCPMC6879246. 448 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 23 26. Clarke RS, Ha KP, Edwards AM. RexAB Promotes the Survival of 449 Staphylococcus aureus Exposed to Multiple Classes of Antibiotics. Antimicrob Agents 450 Chemother. 2021;65(10):e0059421. Epub 20210726. doi: 10.1128/aac.00594-21. 451 PubMed PMID: 34310219; PubMed Central PMCID: PMCPMC8448105. 452 27. Wood TK, Knabel SJ, Kwan BW. Bacterial persister cell formation and 453 dormancy. Appl Environ Microbiol. 2013;79(23):7116-21. Epub 20130913. doi: 454 10.1128/aem.02636-13. PubMed PMID: 24038684; PubMed Central PMCID: 455 PMCPMC3837759. 456 28. Kussell E, Kishony R, Balaban NQ, Leibler S. Bacterial persistence: a model of 457 survival in changing environments. Genetics. 2005;169(4):1807-14. Epub 20050131. doi: 458 10.1534/genetics.104.035352. PubMed PMID: 15687275; PubMed Central PMCID: 459 PMCPMC1449587. 460 29. Roghanian M, Van Nerom K, Takada H, Caballero-Montes J, Tamman H, Kudrin 461 P, et al. (p)ppGpp controls stringent factors by exploiting antagonistic allosteric coupling 462 between catalytic domains. Mol Cell. 2021;81(16):3310-22.e6. doi: 463 10.1016/j.molcel.2021.07.026. PubMed PMID: 34416138. 464 30. Wang X, Wood TK. Toxin-antitoxin systems influence biofilm and persister cell 465 formation and the general stress response. Appl Environ Microbiol. 2011;77(16):5577-466 83. Epub 20110617. doi: 10.1128/aem.05068-11. PubMed PMID: 21685157; PubMed 467 Central PMCID: PMCPMC3165247. 468 31. Huemer M, Mairpady Shambat S, Brugger SD, Zinkernagel AS. Antibiotic 469 resistance and persistence-Implications for human health and treatment perspectives. 470 EMBO Rep. 2020;21(12):e51034. Epub 20201208. doi: 10.15252/embr.202051034. 471 PubMed PMID: 33400359; PubMed Central PMCID: PMCPMC7726816. 472 32. Prestinaci F, Pezzo tti P, Pantosti A. Antimicrobial resistance: a global 473 multifaceted phenomenon. Pathog Glob Health. 2015;109(7):309-18. Epub 20150907. 474 doi: 10.1179/2047773215y.0000000030. PubMed PMID: 26343252; PubMed Central 475 PMCID: PMCPMC4768623. 476 33. Waters EM, Rowe SE, O'Gara JP, Conlon BP. Convergence of Staphylococcus 477 aureus Persister and Biofilm Research: Can Biofilms Be Defined as Communities of 478 Adherent Persister Cells? PLOS Pathogens. 2016;12(12):e1006012. doi: 479 10.1371/journal.ppat.1006012. 480 34. Fey PD, Endres JL, Yajjala VK, Widhelm TJ, Boissy RJ, Bose JL, et al. A genetic 481 resource for rapid and comprehensive phenotype screening of nonessential 482 Staphylococcus aureus genes. mBio. 2013;4(1):e00537-12. Epub 20130212. doi: 483 10.1128/mBio.00537-12. PubMed PMID: 23404398; PubMed Central PMCID: 484 PMCPMC3573662. 485 35. Jorgensen JH, Ferraro MJ. Antimicrobial susceptibility testing: a review of 486 general principles and contemporary practices. Clin Infect Dis. 2009;49(11):1749-55. doi: 487 10.1086/647952. PubMed PMID: 19857164. 488 36. Jones ME, Thomas SM, Rogers A. Luria-Delbrück fluctuation experiments: 489 design and analysis. Genetics. 1994;136(3):1209-16. doi: 10.1093/genetics/136.3.1209. 490 PubMed PMID: 8005425; PubMed Central PMCID: PMCPMC1205875. 491 37. Stewart FM. Fluctuation tests: how reliable are the estimates of mutation rates? 492 Genetics. 1994;137(4):1139-46. doi: 10.1093/genetics/137.4.1139. PubMed PMID: 493 7982567; PubMed Central PMCID: PMCPMC1206060. 494 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint 24 38. El-Halfawy OM, Valvano MA. Antimicrobial heteroresistance: an emerging field 495 in need of clarity. Clin Microbiol Rev. 2015;28(1):191-207. doi: 10.1128/cmr.00058-14. 496 PubMed PMID: 25567227; PubMed Central PMCID: PMCPMC4284305. 497 39. Satola SW, Farley MM, Anderson KF, Patel JB. Comparison of detection 498

for heteroresistant vancomycin-intermediate Staphylococcus aureus, with the 499 population analysis profile method as the reference method. J Clin Microbiol. 500 2011;49(1):177-83. Epub 20101103. doi: 10.1128/jcm.01128-10. PubMed PMID: 501 21048008; PubMed Central PMCID: PMCPMC3020420. 502 503 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 1, 2024. ; https://doi.org/10.1101/2024.04.01.587561doi: bioRxiv preprint