High Salt Cross-Protects Escherichia coli from Antibiotic Treatment through Increasing Efflux Pump Expression

Environmental stresses often co-occur when bacteria confront antibiotic treatment. We provide a clear example that a natural stress condition (high salt) can cross-protect bacteria from antibiotic treatment by triggering the bacterial stress response program (elevated AcrAB-TolC efflux pump expression). Our study highlights the importance of taking the co-occurrence of bacterial environmental stresses into consideration when investigating antibiotic susceptibility and applying antimicrobial treatment.

hyperosmotic stress on bacteria. When confronting a high salt concentration, bacteria must initiate a stress response to tackle the severe loss of water and turgor pressure (2,(17)(18)(19)(20). Cellular stress response enables Escherichia coli to accumulate osmolytes such as potassium, glutamate, and trehalose to maintain the balance of external and internal osmolarity (21). This process helps E. coli regain water and turgor pressure to support growth.
Though a high salt concentration is proposed to significantly affect microbial physiology, it remains unclear whether a high salt concentration can affect the antibiotic susceptibility of bacteria. Here, we find that a high salt concentration cross-protects E. coli from antibiotic treatment, which results from increased AcrAB-TolC efflux pump expression under high-salt conditions.

RESULTS AND DISCUSSION
Decreased antibiotic susceptibility of E. coli under high-salt conditions. To investigate the effect of high-salt conditions on antibiotic susceptibility, we measured the antibiotic growth inhibition curve of E. coli cells under both normal conditions (0.1 M NaCl in glucose minimal medium) and high-salt conditions (0.4 M NaCl in glucose minimal medium). Two well-known ribosome-targeting antibiotics, tetracycline (binds to the 30S ribosomal subunit) and chloramphenicol (binds to the 50S ribosomal subunit), were used to study drug susceptibility (22). The exponential growth rate of E. coli cells was plotted as a function of the antibiotic concentration in the medium. As shown in Fig. 1A, the bacterial growth rate dramatically drops with increasing concentrations of tetracycline under normal conditions (red symbols). Under high-salt conditions, though the growth rate is much lower than that under normal conditions in drug-free medium (0.52/h at 0.4 M NaCl versus 0.96/h at 0.1 M NaCl), it drops very slowly with increasing concentrations of tetracycline (purple symbols). The growth rate under high-salt conditions exceeds that under normal conditions when the tetracycline concentration is Ͼ1 M. This result demonstrates that high salt can lead to crossprotection against antibiotics. For a direct comparison, we made the inhibition curve of the relative growth rate (the growth rate is normalized by the growth rate in drug-free medium, / 0 ) (Fig. 1B). The tetracycline susceptibility of E. coli under high-salt conditions is indeed much weaker than that under normal conditions. The bacterial growth rate drops by~60% at 1 M tetracycline under normal conditions, while under high-salt conditions, it drops by only~35%, even at 4 M tetracycline. Similarly, high salt also reduces the chloramphenicol susceptibility of E. coli ( Fig. 1C and D). The crossprotection effect against chloramphenicol is especially remarkable in glucose-6phosphate minimal medium, where 4 M chloramphenicol has no effect on the bacterial growth rate under high-salt conditions (Fig. S1).
Significant alterations in efflux pump and porin protein expression levels under high-salt conditions. To investigate the origin of cross-protection against antibiotics mediated by high salt, we postulated that high salt could change the level of some specific proteins, which ultimately reduced the intracellular drug concentration. In principle, intracellular drug accumulation can be affected by drug permeation and drug efflux (23). Porin proteins OmpF and OmpC are among the most abundant outer membrane proteins and are proposed to be responsible for controlling drug influx into cells. On the other hand, AcrAB-TolC is the major multidrug efflux pump system that pumps a drug out of the cell to reduce its effective intracellular concentration (23)(24)(25). We applied quantitative mass spectrometry and real-time quantitative PCR (qPCR) to measure relative ompF, ompC, acrA, acrB, and tolC expression in normal and high-salt media. As shown in Fig. 2, the relative acrA, acrB, and tolC expression levels all strongly increase under high-salt conditions. The expression levels of two porin proteins exhibit the opposite trend under high-salt conditions. ompF expression strongly decreases while ompC expression increases remarkably under high-salt conditions. Overall, on the basis of the above-described results, the reduced antibiotic susceptibility that occurs under high-salt conditions may be attributed to increased drug efflux, decreased drug influx (mediated by OmpF), or both.
Disappearance of cross-protection effect in efflux pump mutants. We next characterized the antibiotic susceptibility of E. coli mutants deficient in either efflux pump or porin proteins ( Fig. 3 and 4). Both the acrB and tolC mutants exhibit remarkably higher susceptibility to tetracycline and chloramphenicol than wild-type cells under both normal and high-salt conditions, confirming the important role of the AcrAB-TolC efflux pump in bacterial drug resistance. Strikingly, unlike the case of wild-type cells, the growth rates of acrB and tolC mutant cells under high-salt conditions also dramatically drop with increasing concentrations of tetracycline and chloramphenicol ( Fig. 3A and C). The disappearance of the cross-protection effect is explicitly demonstrated in the relative growth rate inhibition curve; the relative growth rate inhibition curves of the acrB and tolC mutants under both normal and high-salt conditions almost completely overlap ( Fig. 3B and D). This result indicates that the AcrAB-TolC efflux pump is related to the cross-protection against antibiotics mediated by high-salt conditions. The growth inhibition curve of the ompC mutant was almost the same as that of wild-type cells (Fig. 4), indicating that the change in ompC expression is not related to the cross-protection effect under high-salt conditions. The ompF mutant is slightly less susceptible to tetracycline treatment than wild-type cells under both normal and high-salt conditions (Fig. 4A). However, the growth rate of the ompF mutant still exhibits a strong dependence on the tetracycline concentration under normal condi-   (Fig. 4A). Moreover, the chloramphenicol susceptibility of the ompF mutant is similar to that of the ompC mutant and the wild-type strain under both normal and high-salt conditions (Fig. 4B). This indicates that the reduction of ompF expression under high-salt conditions contributes only marginally to cross-protection against antibiotics. Decreased antibiotic susceptibility of efflux pump overexpression strain. To further confirm that the cross-protection effect originates from the increased AcrAB-TolC efflux pump expression level, we studied whether artificial AcrAB-TolC efflux pump overexpression could also lead to decreased antibiotic susceptibility under normal conditions. We constructed an artificial AcrAB-TolC efflux pump overexpression vector (Fig. 5A). The acrAB operon was placed downstream of the P tac promoter. In addition, the tolC gene, together with its native ribosome-binding site (RBS), was directly inserted downstream of the acrAB operon to form a single acrAB-tolC cistron. Therefore, the expression of acrAB-tolC is driven by the inducible P tac promoter, which is under the regulation of a placI q -lacI cassette. The AcrAB-TolC overexpression vector pZE-acrAB-tolC was transformed into acrB and tolC mutants to obtain the FL15 and FL16 strains, respectively. We then measured the antibiotic growth inhibition curves of the FL15 and FL16 strains under normal conditions. As shown in Fig. 5 and S2, elevated AcrAB-TolC efflux pump expression (20 M isopropyl-␤-D-thiogalactopyranoside [IPTG]) indeed reduces the antibiotic susceptibility of E. coli under normal conditions, mimicking the pattern of the growth inhibition curve of wild-type cells under high-salt conditions. Even in the absence of an inducer, the leaky AcrAB-TolC efflux pump expression driven by the P tac promoter has mildly reduced the antibiotic susceptibility compared with that of wild-type cells. On the contrary, the FL30 and FL31 strains, which overexpress green fluorescent protein, used as the control are still as highly susceptible to antibiotic treatment as the acrB and tolC mutants. Overall, the above results strongly support the notion that the antibiotic cross-protection effect of high-salt conditions results from increased AcrAB-TolC efflux pump expression. The functional AcrAB-TolC multidrug efflux pump requires all three proteins (25); therefore, the increased antibiotic tolerance should, in principle, require increased expression of all three efflux pump proteins, as found under high-salt conditions (Fig.  2). In support of this assumption, both the FL17 (acrB deficient and overexpressing AcrA and AcrB) and FL18 (tolC deficient and overexpressing TolC) strains fail to obtain higher antibiotic tolerance than wild-type cells (Fig. S3 and S4). This result indicates that it is the increase in all three efflux pump proteins (functional AcrAB-TolC efflux pump) that is indispensable for the antibiotic cross-protection effect of high-salt conditions.
Conclusion. Our study demonstrates that a high salt concentration, a common environmental stress condition encountered by bacteria inside the human body, can cross-protect them from antibiotic treatment by triggering the stress response program. It was known before that nutrient limitation can also enhance bacterial antibiotic tolerance by increasing the fraction of persister cells in the bacterial population (26)(27)(28)(29). The increased fraction of persister cells originates from the induction of toxin-antitoxin activity by ppGpp signaling under nutrient-limited conditions. The persister cells represent a subpopulation that can survive antibiotic treatment while they remain largely metabolically dormant. Instead, here we find that high-salt conditions can reduce bacterial susceptibility to antibiotics at the whole-population instead of the subpopulation level. Our study highlights the importance of taking the co-occurrence of environmental stressors into consideration when investigating antibiotic susceptibility and applying antimicrobial treatment. The antibiotic cross-protection mediated by environmental stresses inside the human body may accelerate the spread of bacterial drug resistance, which will further complicate the current severe drug resistance situation and should be taken into consideration in future clinical studies.

MATERIALS AND METHODS
Strains. The strains used in this study included the wild-type K-12 NCM3722 strain (22,30) and its derivatives including acrB, tolC, ompF, and ompC mutants and AcrAB-TolC overexpression strains FL15 and FL16. To construct the efflux pump-and porin protein-deficient forms of strain NCM3722, the acrB::kan, tolC::kan, ompF::kan, and ompC::kan alleles in related mutant strains from the Keio collection (31) were transferred into NCM3722 through P1 transduction to generate the four mutant strains, respectively.
To construct the AcrAB-TolC overexpression vector pZE-acrAB-tolC, a placII q -lacI cassette was first inserted into the AatII/XhoI sites of the pZE11-luc vector (32); a P tac promoter was PCR amplified and cloned into the XhoI/KpnI sites to replace the P LtetO-1 promoter of the native vector; the coding sequence of the acrAB operon was then inserted downstream of the P tac promoter at the KpnI/XbaI sites, resulting in pZE-acrAB; and finally, the coding sequence of the tolC gene, together with its RBS, was inserted into the XbaI sites of pZE-acrAB to form a single cistron with the acrAB operon, generating the pZE-acrAB-tolC vector. The pZE-acrAB-tolC vector was then transformed into the acrB and tolC mutants to generate the FL15 and FL16 strains, respectively. The PCR reagents used in this study were Golden PCR mix (green) and T5 super PCR mix (Tsingke BioTech Co., China).
The pZE-acrAB vector was transformed into the acrB::kan mutant strain, resulting in the FL17 strain.
scription with TranScript cDNA Synthesis SuperMix (Tiangen, China). The qPCRs were performed with a SuperReal Premix SYBR green Plus kit (Yeasen Biotech, Shanghai, China) in accordance with the manual. The qPCRs were carried out in a Bio-Rad CFX96 Touch real-time PCR system with the following protocol: 95°C for 15 min, followed by 40 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. The ftsZ housekeeping gene was used as the internal reference. Measurement of protein abundance. The abundances of porin and efflux pump proteins were measured by quantitative mass spectrometry as described by Hui et al. and Dai et al. (5,20).