Antibiotics Used in Empiric Treatment of Ocular Infections Trigger the Bacterial Rcs Stress Response System Independent of Antibiotic Susceptibility

The Rcs phosphorelay is a bacterial stress response system that responds to envelope stresses and in turn controls several virulence-associated pathways, including capsule, flagella, and toxin biosynthesis, of numerous bacterial species. The Rcs system also affects antibiotic tolerance, biofilm formation, and horizontal gene transfer. The Rcs system of the ocular bacterial pathogen Serratia marcescens was recently demonstrated to influence ocular pathogenesis in a rabbit model of keratitis, with Rcs-defective mutants causing greater pathology and Rcs-activated strains demonstrating reduced inflammation. The Rcs system is activated by a variety of insults, including β-lactam antibiotics and polymyxin B. In this study, we developed three luminescence-based transcriptional reporters for Rcs system activity and used them to test whether antibiotics used for empiric treatment of ocular infections influence Rcs system activity in a keratitis isolate of S. marcescens. These included antibiotics to which the bacteria were susceptible and resistant. Results indicate that cefazolin, ceftazidime, polymyxin B, and vancomycin activate the Rcs system to varying degrees in an RcsB-dependent manner, whereas ciprofloxacin and tobramycin activated the promoter fusions, but in an Rcs-independent manner. Although minimum inhibitory concentration (MIC) analysis demonstrated resistance of the test bacteria to polymyxin B and vancomycin, the Rcs system was activated by sub-inhibitory concentrations of these antibiotics. Together, these data indicate that a bacterial stress system that influences numerous pathogenic phenotypes and drug-tolerance is influenced by different classes of antibiotics despite the susceptibility status of the bacterium.


Introduction
Serratia marcescens is a Gram-negative pathogen from the order Enterobacterales that causes contact lens-associated keratitis in healthy patients [1][2][3] and a wide variety of nosocomial infections in the immune compromised, such as ventilator-associated pneumonia and sepsis in adults and neonates [4,5]. S. marcescens isolates are typically resistant to antibiotics of the macrolide, tetracycline, β-lactam, and narrow spectrum cephalosporin classes due to expression of efflux pumps and β-lactamases [6]. However, they are generally susceptible to aminoglycoside, third generation cephalosporin, and fluoroquinolone antibiotics [6,7].
The Rcs stress response system has been found in bacteria from the Enterobacterales including, but not limited to, numerous pathogens, such as Escherichia coli, Klebsiella species, Proteus mirabilis, Salmonella enterica, and Yersinia pestis [8]. The core Rcs system ( Figure 1) is a complex signal transduction cascade composed of a variety of components that include Figure 1. Model for antibiotic activation of the Rcs system. This simplified depiction of the core Rcs system shows the major components required for Rcs function. The Rcs system is a complex phosphorelay signal transduction system that regulates the transcription of many genes through control of the RcsB transcription factor. The IgaA/GumB inner membrane protein blocks Rcs activity under non-stressful conditions. Envelope stress by antibiotics, transmitted by RcsF, prevents IgaA/GumB inhibition of RcsC-D and allows RcsB-mediated transcription. Mutation of igaA/gumB constitutively derepresses the Rcs transcriptional cascade, and mutation of rcsB prevents Rcs system function. This model predicts that Rcs activation by antibiotics can stimulate pathogenesis and antibiotic tolerance phenotypes. OM: outer membrane; PG: peptidoglycan; IM: inner membrane. Importantly, the Rcs system has been shown to contribute to antibiotic tolerance by a number of bacteria, with Rcs system-defective mutants being more susceptible to penicillin and cephalosporin antibiotics for E. coli [12] and polymyxin B for E. coli [13] and S. enterica [10]. Similarly, induced expression of rcsB or expression of alleles that increase Rcs activity conferred increased tolerance to β-lactam and cephalosporin antibiotics for E. coli [12,14]. A major mechanism used by bacteria to increase antibiotic tolerance is biofilm formation [15]. The Rcs system plays a positive role in S. marcescens biofilm formation under high sheer conditions by promoting capsular polysaccharide synthesis [16]. A similar role for the Rcs system in E. coli and S. enterica biofilm formation has been described [17]. Beyond antibiotic tolerance, a recent study by Smith et al. suggests that Rcs plays a role in the acquisition of genetic elements by Serratia sp. 39006 that may contribute to horizontal gene transfer and antibiotic resistance [18].
The S. marcescens Rcs system has been shown to regulate synthesis of the ShlA cytolysin [19,20], where it is also a key regulator of capsular polysaccharide and flagella synthesis, as well as the production of a hemolytic biosurfactant [16,21]. Importantly, the Rcs system was shown to be a major regulator of S. marcescens ocular pathogenesis [22].
The goal of this study was to evaluate whether antibiotics commonly used topically for empiric treatment of ocular infections activate the bacterial Rcs pathway. In this study, we used antibiotics recommended for the empiric treatment of bacterial keratitis by the IM OM PG RcsB

IgaA/GumB RcsC RcsD
Rcs regulon toxins flagella biofilm capsule P Antibiotics Figure 1. Model for antibiotic activation of the Rcs system. This simplified depiction of the core Rcs system shows the major components required for Rcs function. The Rcs system is a complex phosphorelay signal transduction system that regulates the transcription of many genes through control of the RcsB transcription factor. The IgaA/GumB inner membrane protein blocks Rcs activity under non-stressful conditions. Envelope stress by antibiotics, transmitted by RcsF, prevents IgaA/GumB inhibition of RcsC-D and allows RcsB-mediated transcription. Mutation of igaA/gumB constitutively derepresses the Rcs transcriptional cascade, and mutation of rcsB prevents Rcs system function. This model predicts that Rcs activation by antibiotics can stimulate pathogenesis and antibiotic tolerance phenotypes. OM: outer membrane; PG: peptidoglycan; IM: inner membrane.
Importantly, the Rcs system has been shown to contribute to antibiotic tolerance by a number of bacteria, with Rcs system-defective mutants being more susceptible to penicillin and cephalosporin antibiotics for E. coli [12] and polymyxin B for E. coli [13] and S. enterica [10]. Similarly, induced expression of rcsB or expression of alleles that increase Rcs activity conferred increased tolerance to β-lactam and cephalosporin antibiotics for E. coli [12,14]. A major mechanism used by bacteria to increase antibiotic tolerance is biofilm formation [15]. The Rcs system plays a positive role in S. marcescens biofilm formation under high sheer conditions by promoting capsular polysaccharide synthesis [16]. A similar role for the Rcs system in E. coli and S. enterica biofilm formation has been described [17]. Beyond antibiotic tolerance, a recent study by Smith et al. suggests that Rcs plays a role in the acquisition of genetic elements by Serratia sp. 39006 that may contribute to horizontal gene transfer and antibiotic resistance [18].
The S. marcescens Rcs system has been shown to regulate synthesis of the ShlA cytolysin [19,20], where it is also a key regulator of capsular polysaccharide and flagella synthesis, as well as the production of a hemolytic biosurfactant [16,21]. Importantly, the Rcs system was shown to be a major regulator of S. marcescens ocular pathogenesis [22].
The goal of this study was to evaluate whether antibiotics commonly used topically for empiric treatment of ocular infections activate the bacterial Rcs pathway. In this study, we used antibiotics recommended for the empiric treatment of bacterial keratitis by the American Academy of Ophthalmology [23]. Given the role of the Rcs system in promoting antibiotic tolerance and the regulation of virulence factors, it is possible that activation of Antibiotics 2021, 10, 1033 3 of 14 this system could influence clinical outcomes for patients infected by the Enterobacterales. To that end, we developed luminescent reporter plasmids for Rcs activity and used them in a keratitis isolate of S. marcescens with antibiotics from several classes that are recommended for the treatment of ocular infections.

Generation of Luminescent Reporter Plasmids for Rcs System Activity
In order to conveniently measure Rcs activation, luminescent reporter plasmids were made using Rcs-responsive promoters. GumB, an IgaA ortholog, is a negative regulator of Rcs activity, such that a gumB deletion mutant has a highly activated Rcs system [16,19]. Transcriptomic analysis of a ∆gumB mutant was used to identify genes that were more highly expressed than in the wild type (to be described elsewhere). Three promoters were cloned upstream of the luxCDABE operon on a broad-host range low-copy vector ( Figure 2A and Figure S1). The promoters were for the SMDB11_1637, SMDB11_2817, and SMDB11_1194 open reading frames. All of these previously uncharacterized open reading frames bear high similarity to Rcs-regulated genes in other bacteria. SMDB11_1637 is similar to osmotically inducible lipoprotein B (osmB), which is positively regulated by the Rcs system in Erwinia amylovora [24], E. coli [25,26], P. mirabilis [27], S. enterica [28], and Yersinia pseudotuberculosis [29]. SMDB11_1194 is highly similar to umoD, which is Rcs-regulated in P. mirabilis [27], as is its ortholog YPO1624 in Y. pseudotuberculosis. SMDB11_2817 has similarity to yaaX from E. coli with the DUF2502 domain of unknown function and was identified as an RcsB-regulated gene in E. coli [25].
In addition, the nptII promoter from the Tn5 transposon was used as a constitutive control promoter to test for theoretical physiological conditions that could interfere with luminescence.
To validate the Rcs system activation of these promoters, they were moved into a contact lens-associated keratitis wild-type (WT) isolate of S. marcescens, strain K904, and isogenic mutants with manipulated Rcs systems that confer high (∆gumB) or no (∆rcsB and ∆gumB ∆rcsB) Rcs activity. Strains and plasmids are listed in Table S1 (Supplementary Materials). Bacteria were grown overnight, and the luminescence was determined as a function of culture optical density ( Figure 2). The test strains were previously shown to achieve similar growth levels over the tested time frame [30].
The results ( Figure 2) suggested that the promoter activity for each of the genes, except the control nptII promoter, was highly increased (>6000-fold) in the Rcs-activated mutant background (∆gumB). Furthermore, a clear reduction in luminescence was observed in the ∆gumB ∆rcsB double mutant, confirming that the increase observed in the ∆gumB mutant was Rcs dependent. There was some expression in the absence of Rcs activity (see ∆rcsB mutant), indicating that there is some Rcs-independent expression from these promoters (i.e., other transcription factors may regulate some of these promoters; see discussion). Importantly, the nptII promoter showed only a minor but significant change (~2-fold) among the various mutant backgrounds, suggesting that the Rcs system status has little to no impact on the ability of the luminescent reporter system to function. Together, these results indicate that we have identified and cloned three Rcs system-responsive promoters and created reporter constructs to analyze compounds that may influence Rcs-activity. . Schematic diagram of a promoter transcriptional fusion to the luminescence-producing luxCDABE operon that was cloned into a broad-host range medium-copy plasmid. Four different promoters were evaluated by moving them into S. marcescens with normal (WT), hyper-activated (∆gumB), or defective (∆rcsB, ∆gumB ∆rcsB) Rcs-systems. (B-E). Transcription from the four promoters was measured using a luminometer after the bacteria were grown for 20 h in LB medium (n = 4-6). The luminescence values were normalized by optical density, which was similar for each genotype. The nptII promoter is an E. coli promoter that was used as a constitutive control. The PSMDB11_1637, PSMDB11_2817, and PSMDB11_1194 promoters were Rcs responsive. The asterisks indicate that the ∆gumB group is statistically different than the other groups, p < 0.01. WT: wild type.

Antibiotics Targeting the Cell Envelope Activate the S. marcescens Rcs System Regardless of the Antibiotic Susceptibility Status of the Bacterium
The antibiotic susceptibility of four antibiotics used in treatment of ocular infections was analyzed: polymyxin B, cefazolin, ceftazidime, and vancomycin (Tables 1 and 2). These target either the peptidoglycan cell wall or the bacterial membrane. Minimum inhibitory concentrations (MICs) for the antibiotics to inhibit S. marcescens strain K904 were determined ( Table 2). The isolate was susceptible to ceftazidime, but was able to grow at the highest tested concentrations of polymyxin B, cefazolin, and vancomycin. This was a typical pattern for keratitis isolates of S. marcescens [7]. Nevertheless, prior to identification of the infecting microbe, any of the antibiotics other than polymyxin B are candidates for empiric therapy for keratitis. . Schematic diagram of a promoter transcriptional fusion to the luminescence-producing luxCDABE operon that was cloned into a broad-host range medium-copy plasmid. Four different promoters were evaluated by moving them into S. marcescens with normal (WT), hyper-activated (∆gumB), or defective (∆rcsB, ∆gumB ∆rcsB) Rcs-systems. (B-E). Transcription from the four promoters was measured using a luminometer after the bacteria were grown for 20 h in LB medium (n = 4-6). The luminescence values were normalized by optical density, which was similar for each genotype. The nptII promoter is an E. coli promoter that was used as a constitutive control. The PSMDB11_1637, PSMDB11_2817, and PSMDB11_1194 promoters were Rcs responsive. The asterisks (*) indicate that the ∆gumB group is statistically different than the other groups, p < 0.01. WT: wild type.

Antibiotics Targeting the Cell Envelope Activate the S. marcescens Rcs System Regardless of the Antibiotic Susceptibility Status of the Bacterium
The antibiotic susceptibility of four antibiotics used in treatment of ocular infections was analyzed: polymyxin B, cefazolin, ceftazidime, and vancomycin (Tables 1 and 2). These target either the peptidoglycan cell wall or the bacterial membrane. Minimum inhibitory concentrations (MICs) for the antibiotics to inhibit S. marcescens strain K904 were determined ( Table 2). The isolate was susceptible to ceftazidime, but was able to grow at the highest tested concentrations of polymyxin B, cefazolin, and vancomycin. This was a typical pattern for keratitis isolates of S. marcescens [7]. Nevertheless, prior to identification of the infecting microbe, any of the antibiotics other than polymyxin B are candidates for empiric therapy for keratitis.   [37]. c At least one promoter was activated in the wild type, but none in the ∆rcsB mutant.
Polymyxin B was previously shown to activate the Rcs system of E. coli and S. enterica [11,13]. Unlike these bacteria, the vast majority of S. marcescens isolates are resistant to polymyxin B due to a 4-aminoarabinose modification of the lipid A portion of the lipopolysaccharide molecules that populate the outer leaflet of the outer membrane [38]. The K904 strain was evaluated for polymyxin B susceptibility and found to be resistant (MIC > 1024, Table 2). The induction of the Rcs system by polymyxin B in a resistant bacterial species has not been evaluated.
Polymyxin B did not activate the nptII promoter in the WT bacteria, as expected ( Figure 3A); however, the Rcs-dependent promoters were activated in an antibiotic dosedependent manner, up to 5-10 fold above the absence of antibiotic ( Figure 3B-D). To ensure that the effect was Rcs-dependent, the reporters were tested in an isogenic ∆rcsB mutant strain. While there was a less than 2-fold increase in luminescence correlating with the presence of antibiotics, it was not dose dependent in the ∆rcsB mutant ( Figure 3B-D). These suggest that polymyxin B activates the Rcs system even in a resistant bacterium.
The identical approach was used for three different classes of cell wall-targeting antibiotics. A β-lactam antibiotic, cefazolin, is used to treat Gram-positive bacteria (Table 1). S. marcescens strain K904 was resistant to cefazolin ( Table 2). Experiments indicated very little induction except in the SMDB11_1194 promoter ( Figure 4). Similarly, S. marcescens K904 was resistant to the glycopeptide vancomycin (Tables 1 and 2) and was activated by the three Rcs-dependent promoters in the WT but not the ∆rcsB mutant ( Figure 5). By contrast, S. marcescens was susceptible to the cephalosporin ceftazidime (Tables 1 and 2).
Two of the Rcs-dependent promoters were activated by ceftazidime in the WT but not the ∆rcsB mutants ( Figure 6).
Antibiotics 2021, 10, x FOR PEER REVIEW 6 of 16 ensure that the effect was Rcs-dependent, the reporters were tested in an isogenic ∆rcsB mutant strain. While there was a less than 2-fold increase in luminescence correlating with the presence of antibiotics, it was not dose dependent in the ∆rcsB mutant ( Figure 3B-D). These suggest that polymyxin B activates the Rcs system even in a resistant bacterium. The identical approach was used for three different classes of cell wall-targeting antibiotics. A β-lactam antibiotic, cefazolin, is used to treat Gram-positive bacteria ( Table  1). S. marcescens strain K904 was resistant to cefazolin ( Table 2). Experiments indicated very little induction except in the SMDB11_1194 promoter ( Figure 4). Similarly, S. marcescens K904 was resistant to the glycopeptide vancomycin (Tables 1 and 2) and was activated by the three Rcs-dependent promoters in the WT but not the ∆rcsB mutant ( Figure 5). By contrast, S. marcescens was susceptible to the cephalosporin ceftazidime (Tables 1 and 2). Two of the Rcs-dependent promoters were activated by ceftazidime in the WT but not the ∆rcsB mutants ( Figure 6).    Relative luminescence values were determined by dividing luminescence by optical density after 4 h of antibiotic challenge. The nptII promoter (A) was unaffected by cefazolin. Only the Rcs-dependent SMDB11_1194 promoter (C) was activated to a greater extent in the WT than the Rcs-defective ∆rcsB mutant. Mean and standard deviation are shown (n = 6-9 are shown). Asterisks (*) indicate statistical differences between groups at the indicated concentrations, p < 0.05.  Relative luminescence values were determined by dividing luminescence by optical density after 4 h of antibiotic challenge. The nptII promoter (A) was unaffected by ceftazidime. The Rcs-dependent promoters (B-D) were activated to a greater extent in the WT than the Rcs-defective ∆rcsB mutant. Mean and standard deviation are shown (n = 6-9 are shown). Asterisks (*) indicate statistical differences between groups at the indicated concentrations, p < 0.05.

Non-Cell Envelope-Targeting Antibiotics Activated the Test Promoters in an Rcs-Independent Manner
The same approach used for envelope-targeting antibiotics was used for two nonenvelope-targeting antibiotics. Ciprofloxacin is a fluoroquinolone that targets DNA metabolism and is highly effective against Gram-negative ocular pathogens such as Pseudomonas aeruginosa and S. marcescens (Tables 1 and 2). Figure 7 demonstrates that the three test promoters were highly activated by low concentrations of ciprofloxacin in the WT. However, similar, or even higher levels of expression, were observed in the Rcsdefective mutant, indicating that the activation of the test promoters was Rcs-independent and suggesting that ciprofloxacin does not activate the Rcs system. Relative luminescence values were determined by dividing luminescence by optical density after 4 h of antibiotic challenge. The nptII promoter (A) was unaffected by ceftazidime. The Rcs-dependent promoters (B-D) were activated to a greater extent in the WT than the Rcs-defective ∆rcsB mutant. Mean and standard deviation are shown (n = 6-9 are shown). Asterisks (*) indicate statistical differences between groups at the indicated concentrations, p < 0.05.

Non-Cell Envelope-Targeting Antibiotics Activated the Test Promoters in an Rcs -Independent Manner
The same approach used for envelope-targeting antibiotics was used for two nonenvelope-targeting antibiotics. Ciprofloxacin is a fluoroquinolone that targets DNA metabolism and is highly effective against Gram-negative ocular pathogens such as Pseudomonas aeruginosa and S. marcescens (Tables 1 and 2). Figure 7 demonstrates that the three test promoters were highly activated by low concentrations of ciprofloxacin in the WT. However, similar, or even higher levels of expression, were observed in the Rcs-defective mutant, indicating that the activation of the test promoters was Rcs-independent and suggesting that ciprofloxacin does not activate the Rcs system. The ribosome-targeting aminoglycoside antibiotic tobramycin is used to treat o bacterial pathogens (Tables 1 and 2). Data in Figure 8 indicate very little inducti promoter activity by tobramycin except by low induction of the SMDB11_1194 prom Slightly higher expression of the promoters was observed in the ∆rcsB mutant, sugge that the promoter transcriptional activation was Rcs-independent. The ribosome-targeting aminoglycoside antibiotic tobramycin is used to treat ocular bacterial pathogens (Tables 1 and 2). Data in Figure 8 indicate very little induction of promoter activity by tobramycin except by low induction of the SMDB11_1194 promoter. Slightly higher expression of the promoters was observed in the ∆rcsB mutant, suggesting that the promoter transcriptional activation was Rcs-independent.

Discussion
The major impetus behind this study was to test whether the Rcs system was activated by antibiotics used in topical treatment of keratitis. The results show that several of the antibiotics that are widely used for this purpose indeed do activate the Rcs system. A limitation of the study is that the ocular surface antibiotic pharmacokinetics differ from those in the microplate. While topical antibiotics use very high concentrations, the combined action of blinking and the tears wash away most topical antibiotics in a short time frame. Similarly, antibiotic concentrations reduce over time after application, which could lead to levels that activate the Rcs or other stress response systems. Nevertheless, patients with keratitis are given multiple doses of topical antibiotics each day, and although there are limited studies, data demonstrate measurable quantities of the antibiotics accumulate in the corneal tissue [32,36]. Furthermore, experimental studies with rabbits have shown that concentrations of topically applied antibiotics that mimic clinical treatment regimens are able to kill bacteria in the cornea and even to achieve concentrations sufficient to eliminate bacteria that are considered resistant by systemic standards [39][40][41]. Therefore, the combination of the highly sensitive promoters and large antibiotic concentration gradients used in this study likely reflects the antibiotic concentrations that bacteria experience during antibiotic therapy for ocular infections.
Additional differences between this in vitro study and the ocular environment include a lack of the innate immune system components that could influence the activity of the antibiotics through synergistic effects or produce envelope stress through other means, such as envelope-targeting defensins and enzymes such as lysozyme and phospholipase A [42,43]. These potential effects will be analyzed in future studies.
Data from this study indicate that the promoters for SMDB11 ORF 1194, 1637, and 2817 are Rcs responsive, given the several log10-fold increase in the ∆gumB mutant that Relative luminescence values were determined by dividing luminescence by optical density after 4 h of antibiotic challenge. The nptII promoter (A) was unaffected by tobramycin. The experimental promoters were expressed to equal or greater extent in the ∆rcsB mutant than the WT. Mean and standard deviation are shown (n = 6-9 are shown). Asterisks (*) indicate statistical differences between groups at the indicated concentrations, p < 0.05.

Discussion
The major impetus behind this study was to test whether the Rcs system was activated by antibiotics used in topical treatment of keratitis. The results show that several of the antibiotics that are widely used for this purpose indeed do activate the Rcs system. A limitation of the study is that the ocular surface antibiotic pharmacokinetics differ from those in the microplate. While topical antibiotics use very high concentrations, the combined action of blinking and the tears wash away most topical antibiotics in a short time frame. Similarly, antibiotic concentrations reduce over time after application, which could lead to levels that activate the Rcs or other stress response systems. Nevertheless, patients with keratitis are given multiple doses of topical antibiotics each day, and although there are limited studies, data demonstrate measurable quantities of the antibiotics accumulate in the corneal tissue [32,36]. Furthermore, experimental studies with rabbits have shown that concentrations of topically applied antibiotics that mimic clinical treatment regimens are able to kill bacteria in the cornea and even to achieve concentrations sufficient to eliminate bacteria that are considered resistant by systemic standards [39][40][41]. Therefore, the combination of the highly sensitive promoters and large antibiotic concentration gradients used in this study likely reflects the antibiotic concentrations that bacteria experience during antibiotic therapy for ocular infections.
Additional differences between this in vitro study and the ocular environment include a lack of the innate immune system components that could influence the activity of the antibiotics through synergistic effects or produce envelope stress through other means, such as envelope-targeting defensins and enzymes such as lysozyme and phospholipase A [42,43]. These potential effects will be analyzed in future studies.
Data from this study indicate that the promoters for SMDB11 ORF 1194, 1637, and 2817 are Rcs responsive, given the several log 10 -fold increase in the ∆gumB mutant that required a functional rcsB gene. However, it is clear that the selected promoters could also be strongly activated by ciprofloxacin in an Rcs-independent manner. This is not unexpected, as several envelope stress response systems, beyond Rcs, are conserved among the Enterobacterales. For example, in Salmonella, the promoter of the osmB gene (similar to SMDB11_1687) is activated by both the Rcs and the RpoS stress response systems [28], suggesting that individual stress response genes are controlled by multiple regulatory systems. The use of the ∆rcsB strain in addition to the WT enabled clear identification of Rcs-dependent activation of the reporters by ocular antibiotics.
Remarkably, even antibiotics that S. marcescens strain K904 was highly resistant to, such as polymyxin B and vancomycin, elicited strong activation of the Rcs system. These results suggest that the antibiotics are still capable of perturbing the envelope, even if they are not able to prevent growth. In general, the three different promoters reacted similarly to each antibiotic, with the notable exception of cefazolin, which only activated the SMDB11_1194 promoter. This may be due to differential promoter elements that make this promoter more sensitive than the others to Rcs function. Polymyxin B has been shown to activate the Rcs system in polymyxin B susceptible S. enterica at subinhibitory levels, and this was postulated to be driven by polymyxin B's selective permeabilization of the outer membrane to hydrophobic compounds at low concentrations [11,44]. Several other antibiotics that directly or indirectly affect membrane permeability, including β-lactam, fluoroquinolone, and macrolide antibiotics, are likely capable of the same effect [45].
Of interest, ciprofloxacin appeared to activate some of these promoters to a greater extent in the ∆rcsB mutant. This suggests that Rcs may actively inhibit other stress response systems under normal situations. Consistent with this observation, previous studies have demonstrated a complex interplay between the Rcs system and other envelope stress response systems [46][47][48]. Beyond Rcs, there are other envelope sensing stress response systems in the Enterobacterales, including RpoS, the Cpx system, the phage response proteins, EnvZ/OmpR, and others (reviewed by [46][47][48]). Very few studies have evaluated the roles of these proteins in Serratia species; however, studies have demonstrated pleiotropic roles for Cpx, OmpR, and RpoS in the control of pathogenesis-relevant phenotypes, such as biofilm formation, and secreted enzymes and cytotoxic secondary metabolite production in Serratia species [49][50][51][52]. The activation of the Rcs system, as noted above, is correlated with changes that drive virulence-associated phenotypes, such as biofilm formation [16,17]. The ability of antibiotics to promote these phenotypes through the Rcs system during ocular infections will be evaluated in subsequent studies.
During the course of this study, another group reported on the production of a Rcsdependent fluorescent reporter system for E. coli [9]. This was subsequently and cleverly used to screen small molecule libraries for activators of the Rcs system, with the concept that the identified molecules may be evaluated and developed as envelope-targeting antimicrobials [53]. Therefore, Rcs-reporter systems can be used for both basic biomedical research and applied studies, and the reporters generated in this study could be useful to a variety of researchers.

Bacterial Growth and Media
Bacteria (Table S1) were maintained in glycerol stocks at −80 • C and streaked out on lysogeny broth (LB) agar [54] before use. Single colonies were grown in LB broth with aeration on a tissue culture rotor (New Brunswick Tc-7, New Brunswick, NJ, USA). Gentamicin (10 µg/mL) was used to maintain plasmids. Plasmids were moved into S. marcescens by conjugation, and tetracycline (10 µg/mL) was used for selection against donor E. coli [55], as previously described. Antibiotics were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise noted.

Generation of Luminescence Reporters
The pigA promoter on plasmid pMQ713 [56] was replaced with the SMDB11_1194, SMDB11_1637, and SMDB11_2817 using yeast homologous recombination, as previously described [57,58]. Plasmids are listed in Table S1. The pMQ713 plasmid was linearized by restriction enzyme digestion with EcoR1 and Sal1 (New England Biolabs, Ipswich, MA, USA). DNA for the three promoter regions were synthesized as linear double-stranded DNA fragments (Integrated DNA Technologies, Coralville, IA, USA) that include DNA for the promoter region and for site-directed recombination with pMQ713 that places the luxCDABE reporter under transcriptional control of the respective promoter (listed in Table S2). The lengths of the cloned promoters were 338 bp for SMDB11_1194, 354 bp for SMDB11_1637, and 337 bp for SMDB11_2817. To generate the nptII-driven luxCDABE plasmid, the tdtomato gene from pMQ414 was digested with BamH1 and EcoR1 enzymes, and the luxCDABE operon was amplified by PCR from pMQ670 [59] using primers 3805 and 3806 via PrimeSTAR DNA polymerase (Takara Bio, San Jose, CA, USA). The linearized plasmid and luxCDABE amplicon were combined as above. The plasmids were isolated, and the cloned promoter region was sequenced to validate the constructs.

Luminescent Reporter Assays
Strains of S. marcescens bearing luminescent reporter plasmids were taken from a −80 • C freezer and grown on LB agar with tetracycline (10 µg/mL) and gentamicin (10 µg/mL) for 18 h at 30 • C. Single colonies were grown in LB broth with gentamicin in test tubes, which were aerated on a tissue culture rotor for 18-20 h at 30 • C. For reporter verification experiments, the cultures were measured for growth by evaluating optical density at λ = 600 nm (OD 600 ) and luminescence at the 527 nm setting from 200 µL samples in black-sided, clear-bottomed 96-well plates (ThermoFisher, Waltham, MA, USA, product 165305) using plate readers (Molecular Devices SpectraMax M3 and L, San Jose, CA, USA). Relative luminescence units (RLU) values were determined by dividing the raw luminescence values by optical density values.
For antibiotic effect on promoter activity experiments, cultures were normalized by measuring optical density at OD 600 across a 1-cm path length cuvette with a spectrophotometer (Molecular Devices SpectraMax M3). The assay was conducted in 96-well black-sided, optical bottom plates as above. Two-fold serial dilutions of the antibiotics were performed with a multichannel pipette, and the bacteria were then added to a final concentration of OD 600 = 0.05 (~9 × 10 7 CFU/mL). The plate was incubated for 4 h at 37 • C in a plastic bag with a dampened paper towel. At 0 and 4 h, luminescence and OD 600 values were obtained as above. To obtain RLU values, luminescence values were divided by optical density and normalized to RLU values from the no antibiotic challenge control wells.

Minimum Inhibitory Concentration Analysis
MIC values were determined by Epsilometer (E-test) testing (bioMérieux Inc., Durham, NC, USA) for cefazolin, ceftazidime, vancomycin, tobramycin, gentamicin, polymyxin B, and ciprofloxacin. In brief, an overnight growth of bacteria was adjusted to a turbidity standard of 0.5 McFarland (~1.2 × 10 8 CFU/mL) and overlayed with swab streaking on Mueller Hinton agar. E-test strips were placed onto the agar and allowed to incubate for 24 h at 37 • C. The MIC gradients were visually determined and recorded after incubation following the manufacturer's guidelines.

Statistical Analysis
Tests were performed using Prism software (GraphPad, San Diego, CA, USA). Oneway ANOVA with Tukey's post-test was used to compare multiple groups and Student's t-test was used to compare between pairs. For this study, p-values of less than 0.05 were considered significant.

Conclusions
In this study, luminescence reporters for Rcs-stress system activation were generated for use in bacteria of the Enterobacterales order. This stress system induces major transcriptional changes in response envelope stresses that result in increased capsule production and biofilm formation. Using these reporters, the Rcs response of the ocular pathogen S. marcescens to antibiotics used for the treatment of keratitis was evaluated. Several classes of antibiotics used to treat keratitis induced the Rcs system even when the test bacterium was highly resistant to the respective antibiotic. These data suggest that topical treatment of ocular infections with antibiotics may lead to Rcs-dependent phenotypic changes that aid in bacterial antibiotic tolerance.