1. Introduction
A need to create new antibiotics and search for ways to enhance efficiency of existing antimicrobials in the face of a rapid increase in resistant pathogenic strains requires a comprehensive study of bacterial physiological responses under stress induced by antibiotics [
1]. Unlike bacteriostatic drugs, which predominantly inhibit growth of bacteria, bactericidal antibiotics kill bacterial cells, thereby preventing a disease from becoming chronic and reducing the likelihood of the appearance of resistant strains. In 2007, the Collins group proposed a hypothesis according to which antibiotics with different intracellular targets kill bacteria by a single mechanism by increasing production of highly deleterious hydroxyl radicals during the Fenton reaction [
2]. It is hypothesized that H
2O
2 involved in this reaction is formed in the respiratory chain upon acceleration of electron transport, and the concentration of free iron increases due to the destruction of FeS clusters during toxic changes in metabolism after the primary drug–target interactions. The hypothesis caused a wide resonance, since such a mechanism of bactericidal activity makes it possible to increase efficiency of existing antibiotics and design novel drugs capable of both influencing pathways of generation of reactive oxygen species (ROS) and altering levels of activity of antioxidant systems. However, the ROS-dependent mechanism of bactericidal activity has been challenged by several scientific groups, which have shown the absence of oxidative stress under the action of antibiotics [
3,
4,
5]. Subsequent work by the Collins group provided further evidence for ROS involvement in bacterial killing, including analysis of respiratory activity and changes in H
2O
2, glutathione (GSH) and antioxidant gene expression [
6,
7,
8]. They concluded that in the killing process ROS may be synergistic with the damage directly caused by the antibiotic in the primary target. In this case, factors that stimulate formation of ROS should increase sensitivity of bacteria to antibiotics. Endogenous and exogenous H
2S has been reported to protect bacteria from bactericidal antibiotics and H
2O
2, supposedly through H
2S-mediated free iron sequestration and prevention of the Fenton reaction [
9,
10,
11]. Inactivation of H
2S-producing enzymes, either by mutation or by inhibitors, made those bacteria highly susceptible to a variety of antibiotics [
9,
11].
Fluoroquinolone ciprofloxacin used in our work belongs to bactericidal antibiotics. It kills bacteria by damaging their DNA through direct binding with DNA gyrase and/or topoisomerase IV, which results in the formation of double-strand DNA breaks (DSBs), replication arrest, chromosome fragmentation and cell death [
12]. ROS have been reported to be involved in the mechanism of quinolone-mediated rapid lethality [
2,
13]. However, in our own studies, we did not reveal an increase in sensitivity of
E. coli to ciprofloxacin in mutants with defective redox systems of glutathione and thioredoxin, as well as in mutants lacking the general stress response regulator RpoS [
14,
15,
16]. The effect of mutations and additives that change the redox situation in cells on bacterial survival was inversely related to their effect on the growth rate [
16,
17]. We did not observe ciprofloxacin-dependent stimulation of respiration, H
2O
2 formation, GSH oxidation and induction of antioxidant genes [
15,
16]. We also found that the effect of chloramphenicol and high concentrations of ciprofloxacin on
E. coli was accompanied by a transient generation of H
2S due to appearance of excess cysteine as a result of inhibition of protein synthesis [
18,
19,
20]. An excess of cysteine is toxic to cells. Because of its high redox activity, cysteine is susceptible to autooxidation with ROS formation, and can also reduce Fe
3+ to Fe
2+, which potentiates the Fenton reaction in the presence of H
2O
2 [
21,
22]. Generation of H
2S together with export of excess cysteine and its incorporation into glutathione are mechanisms of cysteine homeostasis that maintain its cytoplasmic concentration at a low level [
19]. However,
E. coli lacking cysteine synthetase B encoded by the
cysM gene was equally sensitive to ciprofloxacin as the wild-type strain, despite the absence of H
2S production in this mutant [
16]. Discrepancies between our data and results of other researchers described above may be due to a strong dependence of ROS formation on cultivation conditions. All of our experiments were performed using M9 minimal glucose medium, while in most of the works of other authors rich LB medium was used. Differences in media composition can affect not only growth and metabolic rates, but also production of ROS. In contrast to minimal media, where cells need to synthesize cysteine using sulfate, an LB medium constituent cystine is directly imported into cells, which may affect the ability of
E. coli to maintain cysteine homeostasis. Therefore, in this work, we aimed to compare changes in physiological parameters and H
2S production in
E. coli treated with ciprofloxacin or chloramphenicol in media with different cystine content.
2. Materials and Methods
2.1. Bacterial Strains and Growth Conditions
A strain of
Escherichia coli BW25113 (wild-type) used in this study was from the Keio collection. A strain carrying the transcriptional gene fusion
sulA(
sfiA)::
lacZ was created by transduction with P1 phage from strain DM4000 [
23]. Three types of media were used to grow bacteria: (1) M9 minimal medium (Na
2HPO
4 12H
2O—15.13 g/L, KH
2PO
4—3 g/L, NH
4Cl—1 g/L, NaCl—0.5 g/L, MgSO
4 7H
2O—0.246 g/L, CaCl
2—0.011 g/L) with glucose (0.2%) [
24]; (2) M9 medium without sulfate, but with glucose (0.2%), casamino acids (CA, 0.2%) and feeding 15 µM cystine every 45 min; (3) Luria–Bertani (LB Miller) medium (pepton—10 g/L, yeast extract—5 g/L, NaCl—10 g/L). Cultures were grown overnight, then centrifuged, diluted in 80 mL fresh medium to an initial optical density at 600 nm (OD
600) of about 0.05 and grown in 250 mL flasks with shaking (150 rpm) at 37 °C. When OD
600 reached 0.4 (mid-exponential phase), chloramphenicol (Cam, 25 μg/mL) or ciprofloxacin (CF) at concentrations of 0.03 (2 × MIC), 0.3, 3 and 10 μg/mL was added to the medium and incubation continued for two hours. The specific growth rate (µ) was calculated by equation µ = Δln OD
600/Δt, where
t is the time in hours.
2.2. Real-Time Monitoring of Dissolved Oxygen (dO2), pH and Extracellular K+ and Sulfide
Dissolved oxygen and pH in E. coli cultures were continuously measured directly in the flasks using a Clarke oxygen electrode InPro 6800 (Mettler Toledo, Greifensee, Switzerland) and a pH electrode ESC-10601/7 (“IT” Company, Moscow, Russia), respectively. The dO2/pH controller of a BioFlo 110 fermentor (New Brunswick Scientific Co., Edison, NJ, USA) was used for data recording.
Extracellular sulfide was continuously recorded directly in the flasks using a system of sulfide-specific ion-selective chalcogenide XC-S2-001 (operating pH range 6–12) (Sensor Systems Company, St. Petersburg, Russia) and reference electrodes and a computer pH/ion meter cpX-2 (IBP, Pushchino, Russia).
Changes in levels of extracellular K+ were registered using a system of K+-selective (ELIS-121K) and reference electrodes. For a sensitive determination of K+ during E. coli growth in M9 and M9 + CA + cystine media, potassium concentration was reduced to 0.2 mM. Synchronous processing of all primary data from the sensor system was carried out using the RS-232 and Modbus protocols and the Advantech OPC Server v3.0 software package.
2.3. Determination of ATP, NAD+/NADH Ratio and Membrane Potential
ATP concentration was measured using a luciferin–luciferase ATP determination kit (Molecular Probes). Samples for analysis were removed before antibiotic addition and then every 30 min for 2 h. Then, 50 µL of the cell suspension was added to 450 µL of cell disruption reagent dimethyl sulfoxide (DMSO) to extract ATP. After 5 min, ATP concentration was determined according to the manufacturer’s protocol.
NAD
+ and NADH pools and NAD
+/NADH ratio were determined by the recycling assay as described previously [
25]. Samples were removed immediately before drug addition (time zero) and after 25 min of exposure. Sample preparation and analysis were performed as described previously [
16].
Membrane potential changes (∆ψ) were assessed using the ∆ψ-sensitive fluorescent dye DiBAC
4(3), which stains only depolarized cells [
26], as described elsewhere [
16]. Samples of log-phase cells treated with protonophore carbonylcyanide
m-chlorophenylhydrasone (CCCP, 20 µM) were used as positive control. Fluorescent cells were counted using a Leica DM2000 microscope as earlier described [
16]. Total cell numbers were counted in transmitted light. All experiments were conducted 3–6 times on separate days.
2.4. Determination of H2S in the Gas Phase
Formation of gaseous H2S was assessed using paper strips soaked with lead acetate [Pb(Ac)2], which specifically reacts with H2S to form a brown stain of lead sulfide. The paper strips were fixed above the surface of the liquid culture. Spots were scanned and quantified using ImageJ software. The mean values of histograms of color intensity of each paper strip were measured. Data were presented as the ratio of the value before antibiotic treatment to the values obtained after 2 h of exposure, expressed as a percentage. All experiments were performed 3–6 times on separate days.
2.5. Study of Cell Viability and β-Galactosidase Activity
For colony-forming studies, culture samples taken before and 0.5, 1 and 2 h after antibiotic addition were washed (centrifuged for 2 min at 15,000 g and resuspended in an equal volume of 0.9% NaCl), serially diluted in 0.9% NaCl solution, mixed with molten soft LB-agar (0.8%) at 42 °C and poured onto plates with solid LB-agar (1.5%). Colonies were counted after 24 h of incubation at 37 °C.
The genotoxic effect of ciprofloxacin was assessed by the degree of expression of the
sulA gene, which belongs to the SOS regulon and is induced in response to DNA damage [
27]. Changes in
sulA expression were determined in
E. coli strain carrying the
sulA::
lacZ gene fusion by measuring β-galactosidase activity by Miller’s method [
24].
2.6. Statistical Analysis of the Data
Each result is indicated as the mean value of three to six independent experiments ± the standard error of the mean (SEM). Significant difference was analyzed by Student’s t-test. A p value of 0.05 was used as the cut-off for statistical significance. Results were analyzed by means of the program packet Statistica 8.0.360 (StatSoft Inc., Tulsa, OK, USA, accessed on 27 August 2007). The data obtained using the electrodes represent the results of one of at least three independent experiments.
4. Discussion
Taken together, the data obtained in this work showed that the main difference in physiological response of E. coli to bacteriostatic and bactericidal antibiotics is the rate of inhibition of metabolic processes, regardless of the culture medium used. In all the media studied, the protein synthesis inhibitor chloramphenicol caused a rapid inhibition of growth and respiration, slowed K+ uptake and acidification of the medium associated with glucose consumption and also increased the NAD+/NADH ratio and ATP level. In contrast to M9, in M9 + CA + cys and LB media, chloramphenicol significantly reduced membrane potential measured with DiBAC4(3) fluorescent dye.
Fluoroquinolone ciprofloxacin binding to both gyrase and DNA and stabilizing the gyrase-DNA-cleaved complex was shown to inhibit DNA supercoiling and relaxation [
12,
29]. Apparently being not rapidly lethal at this stage, it inhibited replication and transcription and induced SOS response. However, it was found that, after gyrase removing from DNA, double-strand DNA breaks were formed, which could lead to chromosome fragmentation and cell death. In our study, ciprofloxacin inhibited
E. coli growth, respiration, acidification and K
+ uptake, and increased ATP levels and the NAD
+/NADH ratio in a dose-dependent manner. Degree of this inhibition depended on the medium used: in the richer M9 + CA + cys and LB media, higher concentrations of ciprofloxacin were needed to completely inhibit metabolism than in the minimal M9 medium. The inhibition of respiration caused by ciprofloxacin was biphasic in all media. At ciprofloxacin concentrations not exceeding OBC, a phase was observed in which the cells continued totake up oxygen. The duration of this phase coincided with the time during which the cells maintained a high growth rate. Concentrations of ciprofloxacin above OBC caused a rapid inhibition of respiration already in the first phase. The onset of the second phase of inhibition of respiration depended on the dose of ciprofloxacin and earlier was associated with the onset of the SOS response late phase (“programmed cell death”) and was absent in
recA mutant [
15,
16]. The beginning of this phase coincided with the complete cessation of growth and consumption of the energy source, the release of K
+ from the cells, and a drop in the membrane potential and the ATP pool. We did not observe an acceleration of oxygen consumption in any of the studied media under the action of all concentrations of ciprofloxacin. This is consistent with our previous data for ciprofloxacin and the results obtained by Liu and Imlay for the fluoroquinolone norfloxacin [
4,
15,
16,
20], but is not consistent with the data of the Collins group, which reported an increase in respiration rate under the action of norfloxacin [
7]. However, under the action of bactericidal ciprofloxacin, high metabolic activity persisted longer than under the action of a bacteriostatic chloramphenicol. The greatest similarity in the physiological responses of bacteria to ciprofloxacin and chloramphenicol was observed at doses of ciprofloxacin above the OBC.
At the same concentration of ciprofloxacin,
E. coli maintained a higher rate of growth and respiration in LB than in M9 medium. The specific growth rate is an integral parameter, which reflects the activity of core metabolic processes and, accordingly, the number of active targets for antibiotics [
31]. Therefore, there are more ciprofloxacin targets in cells and more DNA damage (stalled transcriptional complexes and replication forks and DSBs) accumulates in LB than in M9 medium at a given dose of ciprofloxacin. The SOS system can be involved in reparation of damage induced by quinolones, and as
E. coli can successfully repair only up to four simultaneous DSBs, slowly growing cells with fewer replication forks and reduced transcriptional activity experience fewer breaks and are able to repair them and survive [
29]. Fast growing cells receive more damage and die, which explains the inverse relationship between CFU and specific growth rate and the decrease in the number of CFU by 2–3 orders of magnitude in LB compared to M9 medium. This is consistent with our earlier data that, regardless of the causes affecting the growth rate, even a small increase in μ is accompanied by a significant increase in ciprofloxacin-induced cell death [
17]. In addition, the maintenance of respiratory activity during drug-induced stress may promote the formation of ROS, which can potentially damage cellular structures and contribute to the lethal activity of ciprofloxacin. That is, a paradox arises: the less sensitive the metabolic processes of cells in liquid culture are to the action of ciprofloxacin, the stronger the lethality in the CFU test. The fast phase of cell death on the killing curves obtained by counting CFU corresponds to a period of high metabolic activity of cells in liquid culture. From this point of view, the late phase of the SOS response, which completely stops metabolism and transfers cells to a dormant state, may be an adaptive response that increases the number of persisters. Our data are consistent with Hong et al.’s findings that most quinolone-induced cell death seems to occur not during antibiotic exposure, but at the growth recovery stage on LB agar plates in the absence of antibiotics [
13], although the dominant role of ROS in this process needs further confirmation.
Another consequence of the slower inhibition of metabolism in rich media was a shift in the optimal bactericidal concentration towards higher doses of ciprofloxacin (from 0.3 µg/mL in M9 to 3 µg/mL in LB). The biphasic dependence of bactericidal activity on drug concentration, which is characteristic of quinolones [
30], disappeared in the
recA mutant [
15]. In this work, we have shown that the maximum expression of the
sulA gene, i.e., the maximum induction of the SOS response, corresponds to the OBC for ciprofloxacin in different media. It is possible that the dramatic inhibition of growth and respiration and the resulting reduction in DNA damage and increased survival at quinolone concentrations above OBC are controlled by the SOS system. Previously, it was suggested that the increased bactericidal activity of moderate concentrations of quinolones compared to their high doses may be associated with a decrease in ROS production at high doses of antibiotics [
32]. When respiration is inhibited, ROS production may decrease. However, at the same time, under the action of high doses of quinolones, there is a decrease in replication and transcription, and, accordingly, the number of active targets for antibiotics decreases, which can block antibiotic action.
Here and in our previous work [
18,
19,
20], we have shown that treatment of exponentially growing
E. coli with chloramphenicol or high doses of ciprofloxacin (3–20 µM/mL) in M9 medium is accompanied by transient H
2S generation. Hydrogen sulfide formation has also been observed during starvation stress caused by glucose depletion or valine supplementation (isoleucine starvation) [
18,
19]. We found that the production of H
2S is a marker of the appearance of excess free cysteine in the cytoplasm during a sharp inhibition of protein synthesis [
18,
19]. Sulfide generation, as well as the acceleration of glutathione synthesis and the export of excess cysteine from cells, contributed to the reduction of cysteine level in the cytoplasm and can be considered as mechanisms of its homeostasis. Maintaining a low intracellular level of cysteine is very important for cells because cysteine is capable of autoxidation to form ROS, and can also reduce Fe
3+ to Fe
2+, potentiating the Fenton reaction to form deleterious hydroxyl radicals [
21,
22]. It can be assumed that under the action of bactericidal antibiotics, ROS formed when excess intracellular cysteine appears will contribute to the lethality of the antibiotic. However,
E. coli transiently generated H
2S in M9 medium only when exposed to ciprofloxacin concentrations above OBC, when growth inhibition was severe. Under these conditions, inhibition of H
2S production in the
cysM mutant encoding cysteine synthase B did not affect survival compared to the parental strain [
16]. In contrast to the M9 medium, when
E. coli was grown in LB or M9 + CA + cys media, H
2S was generated in the absence of external influences. Under these conditions, cysteine synthesis is almost completely inhibited [
33], and
E. coli cells are supplied with cysteine through cystine transport via CysB-controlled TcyP and TcyJLN importers [
22,
34]. Massive uptake of cystine by cells can create an excess of intracellular cysteine and trigger the mechanisms of its homeostasis, in particular H
2S production and cysteine efflux. Thus, the effect of antibiotics on
E. coli in LB and M9 + CA + cys media occurs in the background of increased intracellular cysteine and the already running mechanisms of cysteine homeostasis, which may significantly modify the stress response to antibiotics. We observed an increased expression of the
sulA gene in the M9 + CA + cys medium before treatment of bacteria with ciprofloxacin, which could be a consequence of ROS-mediated DNA damage with an excess of intracellular cysteine. An alternative reason for increased
sulA expression may be a change in the activity of the SOS response regulator RecA, which contains essential SH groups [
35], during cystine-induced disulfide stress. The produced H
2S is an anti-inducer of CysB and can inhibit the cysteine regulon [
33]. In addition, low micromolar levels of sulfide have been shown to inhibit cytochrome
bo oxidase and switch respiration to
bd oxidase [
36,
37]. It was also found that H
2S reduced susceptibility of bacteria to bactericidal antibiotics and H
2O
2 in the LB medium, presumably by reducing the level of intracellular Fe
2+ and preventing the Fenton reaction [
9,
10]. Excess cysteine exported from cells can be oxidized to cystine in the presence of oxygen and then again uptake by the cells. Therefore, in the presence of cystine in the medium, a futile import/reduction/export cycle occurs that consumes a large amount of cellular energy [
22,
34], which may also affect antibiotic tolerance. In light of the above, the mechanisms of cysteine homeostasis are a convenient target for the development of new drugs that increase the efficiency of bactericidal antibiotics. Recently, inhibitors of bacterial H
2S biogenesis have been created that reduce resistance and tolerance to antibiotics [
11]. Our data on H
2S production in different media indicate that such inhibitors will be more effective in rich media containing cystine/cysteine and less effective in minimal media where H
2S production is short-lived.