Reversible Induction of ATP Synthesis by DNA Damage and Repair in Escherichia coli

Early metabolic events in Escherichia coli exposed to nalidixic acid, a topoisomerase II inhibitor and an inducer of the SOS system, were investigated by in vivoNMR spectroscopy, a technique that permits monitoring of bacteria under controlled physiological conditions. The energetics of AB1157 (wild type) and of its isogenic, SOS-defective mutants, recBC,lexA, and ΔrecA, were studied by31P and 19F NMR before, during, and after exposure to nalidixic acid. The content of the NTP in E. coli embedded in agarose beads and perfused at 36 °C was found to be 4.3 ± 1.1 × 10−18 mol/cell, yielding a concentration of ∼2.7 ± 0.7 mm. Nalidixic acid induced in the wild type and mutants a rapid 2-fold increase in the content of the NTP, predominantly ATP. This induction did not involve synthesis of uracil derivatives or breakdown of RNA and caused cell proliferation to stop. Removal of nalidixic acid after 40 min of treatment rescued the cells and resulted in a decrease of ATP to control levels and resumption of proliferation. However, in ΔrecA cells, which were more sensitive to the activity of the drug, ATP elevation could not be reversed, and ATP content continued to increase faster than in control cells. The results ruled out association between the elevation of ATP and the induction of the SOS system and suggested involvement of a process reminiscent of apoptosis in the stimulation of ATP synthesis. Thus, the presence of the RecA protein was found to be essential for reversing the ATP increase and cell rescue, possibly by its function in repair of DNA damage.

Several antitumor and antibacterial drugs were shown to induce a rapid elevation in the total pool of the NTPs in prokaryotes and eukaryotes (1)(2)(3)(4)(5)(6)(7). In Escherichia coli, bleomycin and UV radiation, which are also bacterial SOS activators, were shown to induce a rapid, transient, and 2-fold increase in ATP concentration (1,2). The pronounced elevation of ATP was shown to be independent of oxidative phosphorylation and was therefore attributed to an unknown intracellular phosphorylation pathway (1). In addition, with UV a delayed and substantially smaller increase in ATP was found to occur in the recBC mutant (1), whereas a complete inhibition of this increase occurred in a recBC mutant treated with bleomycin (2). In wild type, Within ϳ30 min of induction of ATP, the bacteria entered a recovery phase, and ATP rapidly declined to baseline levels (1)(2)(3). However, in the mutants with impaired SOS activity, such as recA13 (defective RecA protein), recA430 (deficient RecA protease activity), and lexA1 (cleavage resistant LexA repressor), recovery was impeded, and ATP levels remained elevated (1)(2)(3). These results indicated involvement of the SOS pathway in the return to normal NTP levels, i.e. the recovery phase (3).
The rapid increase in ATP after damage to DNA is not unique to prokaryotic cells and also occurs in mammalian cancer cells in response to chemotherapy (4 -7). For example, in T47D clone 11 human breast cancer cells, adriamycin induced a fast transient increase of 30 -50% in the NTP levels before cell death (4,8). The adriamycin-induced increase in the NTP levels was also observed in small cell lung carcinoma cultures (5). Other chemotherapeutic drugs such as cisplatin in human ovarian carcinoma (6) and hypoxanthine with methotrexate in rat lymphoma cells and in L1210 mouse leukemia cells (7) also induced an increase in ATP levels.
The activity of many anticancer and antibacterial drugs appears to be related to their capacity to inhibit DNA topoisomerase II (9,10). Type II topoisomerases catalyze changes in the topology of DNA by passing one segment of duplex DNA through a transient break in a second DNA segment. The reaction intermediary consists of a topoisomerase covalently linked, via a phosphotyrosine bond, to each 5Ј terminus at the site of the DNA break (11). Drugs such as adriamycin affect the DNA breakage and reunion reaction of mammalian DNA topoisomerase II by stabilizing the cleavable topoisomerase-DNA complex. It was previously suggested that inhibition of DNA topoisomerase II by chemotherapeutic drugs may cause an increase in ATP concentrations (11).
The bactericidal action of nalidixic acid serves as a model for the study of the mechanism of cytotoxicity by topoisomerase II-targeting antitumor drugs (11). In the presence of the RecBCD nuclease, nalidixic acid is a strong inducer of SOS repair functions (12,13), with the nuclease unwinding activity being the critical property (13). In this work, the effects of nalidixic acid on E. coli metabolism were characterized noninvasively by NMR studies, along with other standard methods. Although NMR spectroscopy was previously applied to investigate metabolic activities in suspensions of E. coli (14 -17), monitoring metabolic changes in vivo over extended periods required the development of improved methods of cell encapsulation and perfusion that ensured controlled physiological conditions. These methods enabled us to follow intracellular metabolites during growth and after exposure to nalidixic acid. 31 P NMR was applied to monitor the phosphate metabolites, and 19 F was applied to monitor fluoronucleotides derived from 5-fluorouracil (FUra) 1 incorporation. The involvement of the SOS response in the observed metabolic alterations was investigated by using ⌬recA, lexA, and recBC mutants in which the SOS system cannot be induced. Nalidixic acid was shown to induce a rapid increase in the content of the NTPs, predominantly ATP. This induction was found to be independent of the SOS response and appeared to be associated with an apoptoticlike process.

EXPERIMENTAL PROCEDURES
E. coli Mutants-The bacterial strains used here were derivatives of E. coli K-12 (Table I). A deletion of recA was introduced into AB1157 cells by P1 transduction using WBM535 (20) as the donor strain (21). The presence of the recA genotype was verified by the extreme sensitivity of the mutant to UV irradiation at 254 nm.
Cell Cultures-The standard growth medium for culturing the bacteria contained 1.45 mM Na 2 HPO 4 , 0.73 mM KH 2 PO 4 , 82.6 mM Tris base, 100 mM NaCl, 19 mM NH 4 Cl, 20 mM KCl, 10 mg/ml thiamine (B1), 1 mM MgSO 4 , 0.1 mM CaCl 2 , 0.012 mM FeCl 3 , 11 mM glucose, and amino acid solution (100 ml/liter; Sigma). The pH of this medium was adjusted to 7.5. The concentration of P i in this medium was low, to bring the intensity of the P i signal close to the intensities of the intracellular phosphate signals of the 31 P NMR spectra. We verified that this concentration was sufficient to maintain the same growth rate as in the high phosphate medium (generation time, ϳ45 min for wild type, lexA3, and rec A mutants and 60 min for the recBC mutant). In experiments with FUra (Sigma), cells were were cultivated for several generations in the standard growth medium supplemented with 10 -100 M FUra.
Survival Assay-E. coli in mid-logarithmic growth (A ϭ 0.5 at 595 nm) were exposed for 40 min to low concentrations of nalidixic acid (Sigma): 0.4 g/ml and 1.4 g/ml, equivalent to 1.6 ϫ 10 Ϫ9 and 5 ϫ 10 Ϫ9 g/cell, respectively, or a high concentration of 40 g/ml, equivalent to 1.6 ϫ 10 Ϫ7 g/cell. Aliquots (100 l) of the treated and untreated cultures were plated on Luria agar plates. The surviving colonies were counted after an overnight incubation at 37°C.
Transmission Electron Microscopy-E. coli was grown in Luria Bertani medium to the mid-logarithmic phase of growth. Bacteria were treated with nalidixic acid (40 g/ml) for 40 min. Control and treated bacteria were washed twice with Luria Bertani medium and then centrifuged. The pellets were fixed in 2.5% glutaraldehyde in cacodylate buffer (pH 7.4) and post-fixed in 1% osmium tetraoxide for 2 h. The samples were then dehydrated in graded ethanol solutions and embedded in Epon (Polarbed 812 resin, Polaron, Watford, United Kingdom). Ultra thin sections were constructed with uranyl acetate and lead citrate and were examined with a Jeol JEM 1200 EXII microscope at 80 kV.
NMR Experiments-Encapsulation of E. coli for NMR experiments: This method was adapted from the encapsulation method of the algae Dunalliella (22). A day before an NMR experiment, ϳ9 ϫ 10 10 cells were harvested in the mid-logarithmic phase of growth, resuspended in 3 ml of fresh growth medium and incubated briefly (ϳ5 min) at 37°C. The suspension was then mixed thoroughly with 3 ml of a 6% low gelling agarose (Sigma) solution and added rapidly to stirred paraffin oil (100 ml), all of which were maintained at 37°C and under sterile conditions. This procedure produced a uniform population of spherical agarose drops (300 -500 m diameter) containing the bacteria. Cooling of the paraffin oil to 10°C yielded gelled spherical beads with entrapped cells (ϳ1.5 ϫ 10 10 cells/ml beads). These beads were separated from the paraffin oil and washed several times with fresh growth medium and kept at 4°C overnight. The next day, 2.5 ml of the gelled agarose beads were transferred to a 10-mm NMR tube. The sample was placed in the spectrometer, and the cells embedded in the agarose beads were perfused at 36 Ϯ 1°C under sterile conditions, as described below.
Medium for NMR-The standard growth medium described above was slightly modified for NMR use, containing a higher concentration of glucose (90 mM) and Tris base (165 mM) and a lower concentration of NaCl (57 mM). We verified that the growth rate in this modified medium was the same as in the standard medium with a generation time of ϳ45 min. The increased glucose supply and increased buffer capacity (because of Tris) ensured adequate nutrition and a stable pH during the NMR experiments. In NMR experiments with FUra, the perfusion medium also contained 15 M of the drug.
Nalidixic acid was added at a concentration of 140 g/ml, which was equivalent to a dose of ϳ9.3 ϫ 10 Ϫ9 g/cell. Below this concentration no change in the spectra could be observed, whereas a higher dose (ϳ50%) yielded similar results.
Perfusion System-A perfusion system, previously developed in our laboratory (22), was adapted for the NMR studies (Fig. 1). The cells were perfused unidirectionally at a flow rate of 3 ml/min to ensure an adequate supply of glucose and oxygen. A bubble trap was added in the delivery line to prevent bubbles from entering the tube and perturbing the homogeneity of the magnetic field. The temperature in the NMR tube was adjusted to 36 Ϯ 1°C during the NMR experiments, using the variable temperature control unit (Bruker). The medium container outside of the spectrometer was saturated with humidified oxygen (100%) and kept at 39 Ϯ 1°C. Medium was changed by replacing the whole container with a new one containing the desired medium (e.g. with a fresh medium or one containing the drug).
NMR Measurements-NMR measurements were performed with a Bruker AM-500 spectrometer equipped with a quadro-nuclei ( 1 H, 31 P, 13 C, and 15 N) probe or a Bruker AMX-400 spectrometer equipped with a triple-tuned ( 1 H, 31 P, and 19 F), software-controlled probe. 31 P spectra recorded at 202.5 MHz were obtained by collecting 760 transients within 13 min, applying 60°pulses with a 1-s repetition time, and proton decoupling with a composite pulse decoupling sequence of ϳ1 watt. 31 P spectra recorded at 164 MHz were obtained by collecting 1900 transients within 8.8 min, applying 55°pulses with a 0.42-s repetition time, and composite pulse decoupling during acquisition as above. 19 F spectra recorded at 376 MHz were obtained by collecting 1400 transients within 8.5 min, applying 60°pulses with a 0.24 s repetition time, and composite pulse decoupling during acquisition. The 19 F and 31 P spectra were recorded sequentially. In both spectrometers, the temperature of the sample was maintained at 36 Ϯ 1°C with a Bruker variable temperature control unit.
Analysis of NMR Data-The 31 P and 19 F chemical shifts were determined in reference to the ␣NTP signal at Ϫ10.04 ppm and to the FUra signal at Ϫ171 ppm, respectively. Changes in the areas or the intensities of each signal were directly proportional to changes in the content of the corresponding metabolite. The areas of the signals were determined either by line shape simulation using the Bruker GLINFIT program or by applying the integral mode of the spectrometer. A comparative analysis of the relative changes in areas or intensities of the NTP signals yielded similar results that were within experimental error. NTP content per cell was estimated from the signal area of ␥NTP in the 31 P spectra of perfused cells (for example, see Fig. 2) in reference to the signal area of P i attributable predominantly to the 2.18 mM inorganic phosphate in the medium, taking into account saturation effects as a result of the NMR acquisition parameters and the differences in the T 1 relaxation rates of ␥NTP (T 1 ϭ 0.7 s) and P i (T 1 ϭ 5.4 s). 1 The abbreviations used are: FUra,5-fluorouracil; NDP, nucleoside diphosphate. In this study, the high cell density in the NMR sample and the high metabolic activity of the cells required a rapid and continuous supply of nutrients and oxygen, as well as a fast removal of inhibitory products and prevention of medium acidification. These demands were met by perfusing cells embedded in agarose beads unidirectionally, at a high rate (3 ml/min), with medium saturated with oxygen, containing a high concentration of glucose (90 mM) and Tris base (165 mM) to achieve high buffer capacity.
The optimal concentrations of glucose and the buffer as well as the rate of perfusion were determined by modifying these parameters and measuring the NTP levels of the perfused cells by 31 P NMR, aiming to reach maximal and reproducible levels within consecutive recordings (each for 12 min). The initial number of cells in the beads was measured before the encapsulation. However, it was not possible to determine exactly the number of cells at the end of the experiments, because isolation of the cells from the beads led to substantial cell death. At the start of each NMR experiment, the intensities of the phosphate signals were very low, close to the noise level. However, within the first 1-2 h of adaptation to the NMR perfusion conditions (at 36°C), the intracellular phosphate signals including the three NTP resonances became detectable. From the area of the NTP signal relative to that of the external P i signal of the medium (2.18 mM), taking into account saturation effects, the amount of NTP per cell was found to be 4.3 Ϯ 1.1 ϫ 10 Ϫ18 mol/cell, yielding a cellular NTP concentration of ϳ2.7 Ϯ 0.7 mM. In control experiments, under constant perfusion conditions for ϳ7 h, a large increase in the NTP signals was observed (Figs. [2][3][4]. This increase can be attributed only to an increase in the number of cells, because cellular NTP concentration remains the same under constant conditions (e.g. constant supply of nutrients and oxygen, constant pH, and constant temperature). The increase in NTP was logarithmic and reached saturation levels as expected for cell growth curves. From this logarithmic increase a generation time of ϳ100 min was determined for E. coli entrapped in beads. This generation time was twice as long as that measured in suspension (ϳ45 min), presumably because of the high density of the bacteria in the agarose beads.
The 31 P NMR spectra of cells perfused at 36°C exhibited in addition to the NTP signals phosphomonoester signals, P i signal predominantly attributable to the medium P i (2.18 mM), and NAD and uridine diphosphosugar signals (Fig. 2). The signals of ␣NDP and ␤NDP overlap with the signals of ␣NTP and ␥NTP, respectively. The area of the ␥NTP plus ␤NDP signal was the same, within experimental error, as that of ␤NTP. We can therefore conclude that the content of NDP under the NMR perfusion conditions is below detection level, namely, [NDP] is Ͻ20% of [NTP]. The uridine diphosphosugar concentration in the cells was similar to the concentration of NTP.
Changes in the phosphate metabolites induced by exposure to nalidixic acid were studied by 31 P NMR in wild type, and the recBC, lexA, and ⌬recA mutants. In each experiment, two separate samples of the same cell preparation were monitored one after the other. The first sample was treated transiently with nalidixic acid (140 g/ml), whereas the second one was continuously perfused with nalidixic acid-free medium and served as a control. Administration of nalidixic acid (for 40 min or 3 h) induced a rapid increase of ϳ2-fold in the NTPs of all the strains tested (Figs. 3 and 4). Specifically, the extent in NTP elevation was 2.00 Ϯ 0.20 (n ϭ 6) for E. coli wild type, 2.10 Ϯ 0.08 (n ϭ 2) for recBC, 2.2 (n ϭ 1) for lexA, and 1.7 Ϯ 0.03 (n ϭ FIG. 2. Representative 31 P NMR spectrum and increase in NTP during growth of E. coli in the NMR spectrometer. AB1157 cells embedded in agarose beads were perfused in the NMR spectrometer at 36 Ϯ 1°C as described under "Experimental Procedures." The spectrum was recorded 100 min after initiation of the NMR experiment. 760 transients were accumulated during 12 min. Exponential multiplication with a line broadening of 15 Hz was applied before the Fourier transformation. A.U., arbitrary unit; UDPS, uridine diphosphosugar; PME, phosphomonoesters. 3) for ⌬recA, and substantially higher than the increase attributable only to the growth of the bacteria (Figs. 3 and 4, insets). The main induction occurred during the first 20 min of exposure to the drug. In the cells treated for 3 h with nalidixic acid, the initial 2-fold increase was followed by an additional but slower increase for 2 more h (Fig. 4).
Removal of nalidixic acid from the perfusion medium after 40 min caused in E. coli wild type, recBC, and lexA a reduction in the induced NTP content to a level similar to that of untreated, control cells. This recovery was then followed by a gradual increase in the NTPs caused by normal cell growth. Exceptionally, in ⌬recA cells, despite removal of the nalidixic acid, the level of the NTPs remained high and did not return to control level, and a further increase occurred thereafter, two to three times faster than in ⌬recA control cells (Fig. 3D). Thus E. coli ⌬recA cells were not able to recover from the nalidixic acid insult, and the induction of ATP synthesis could not be reversed. Long term treatment with nalidixic acid of E. coli wild type also inhibited rescue of the cells (Fig. 4). The cells did not recover; on the contrary, despite removal of nalidixic acid after 3 h of exposure, a drastic ϳ4 fold decrease in NTP had occurred, indicating cell death (Fig. 4). Thus, it appears that during the longer exposure to the drug, the cells entered a process from which they could not be rescued anymore.
Labeling of the NTP pool can provide a tool to monitor the source of the induction as a result of nalidixic acid administration. It was previously shown that when E. coli were grown in the presence of FUra, ϳ70 -95% of the FUra was incorporated into RNA (23,24). Herein the incorporation of FUra was used as a tool to monitor changes in FUra derivatives during elevation of NTPs by nalidixic acid. The FUra derivatives included the soluble nucleotides F-UMP, F-UDP, and F-UTP, all of which can be monitored intracellularly by 19 F NMR. In the presence of a low concentration of FUra, the growth of E. coli wild type was only slightly inhibited (Fig. 5). For the NMR experiments, wild type (n ϭ 2) and ⌬recA (n ϭ 2) were cultivated in the presence of a low concentration (15 M) of FUra for 2-3 generations. The soluble nucleotides of the cells were then monitored by recording sequentially both 31 P spectra, which measured the total NTP pool, and 19 F spectra, which measured the content of fluorinated derivatives (Fig. 6). 31 P spectra of wild-type cells demonstrated the transient induction in NTP level by nalidixic acid (Figs. 6A and 7A) as was found for FUra-free cells. The 19 F spectra of these FUralabeled cells exhibited two signals at 165.3 and 165.45 ppm, which were assigned, based on their chemical shift, to F-uridine diphosphosugar and F-UTP, respectively (Fig. 6B). These two separate, but close, signals increased during cell growth before the addition of nalidixic acid, in parallel to the increase in NTPs. However, in the presence of nalidixic acid no change in either of the two fluorinated derivatives was observed, even though there was a substantial increase in NTP (Figs. 6B and 7B). After removal of the nalidixic acid, the level of the NTPs decreased to that of controls and was then followed by a gradual increase. In parallel, the incorporation of FUra was restored, and a gradual increase in the fluorinated uracil derivatives was observed (Fig. 7B). Similar experiments with the ⌬recA mutant also showed that during the nalidixic acid-mediated induction of NTPs, no change in the fluorinated uracil derivatives was observed.
Nalidixic acid is known to act on E. coli as a bacteriocidic agent (25). However, to rule out the possibility that during the elevation in the ATP content nalidixic acid induced a faster cell growth, we have studied the effect of nalidixic acid on cell growth and survival. The ability of wild type and the ⌬recA mutant to survive in the presence of varying doses of nalidixic acid was monitored in cell suspensions, after treatment of the cells with the drug for 40 min, as in the NMR experiments. At the low doses of nalidixic acid (0.4 g/ml ϭ 1.6 ϫ 10 Ϫ9 g/cell, and 1.4 g/ml ϭ 5 ϫ 10 Ϫ9 g/cell) the growth of wild-type bacteria was not affected, and the number of cells increased as in untreated cells (Fig. 8), whereas the growth of ⌬recA cells was arrested, and only a marginal increase in cell number had occurred. At the high dose (40 g/ml ϭ 1.6 ϫ 10 Ϫ7 g/cell) the growth of wild-type bacteria was arrested, with no change in the initial cell number (Fig. 8), whereas a decrease in cell number indicating cell death had occurred in ⌬recA cells (Fig.  8). The dose applied in the NMR experiments was 140 g/ml ϭ 9 ϫ 10 Ϫ9 g/cell. In terms of amount of drug per cell, it was closer to the lower doses in suspension (1.4 -5 ϫ 10 Ϫ9 g/cell) than to the high one (1.6 ϫ 10 Ϫ7 g/cell). Indeed, the NMR studies of cells treated with nalidixic acid for 40 min did not indicate cell death (e.g. decline of ATP) during the experiments, suggesting that the survival under the NMR conditions was similar to that of cells in suspension treated with the low dose of nalidixic acid. However, during the longer duration of treatment (3 h) with nalidixic acid, the bacteriocidic capacity of this drug was exhibited by a marked decrease in NTP (Fig. 4).
In addition to the survival of cells in suspension, we also investigated morphological changes induced by nalidixic acid using transmission electron microscopy. Control E. coli demonstrated the typical elongated structure surrounded by an external envelope. The cytoplasm exhibited a bright nonconfined area in the center of the bacterium that contained a nucleoid with a dispersed appearance. The peripheral cytoplasm displayed an increased density with a granular appearance attributable to the large number of ribosomes (Fig. 9, A and C). After treatment with nalidixic acid, the cells appeared more dense with the nucleoid less dispersed and more localized (Fig. 9, B  and D-F). Also, the difference between the density of the nucleoid and the density of the area with the ribosomes was less pronounced. In the majority of the nalidixic acid-treated cells we also observed inclusion bodies in the cytoplasm (Fig. 9, E and F). These inclusion bodies exhibited a crystalline-like structure and were generally linked to the cell membrane. DISCUSSION Changes in the energetics of E. coli cells after exposure to the SOS activator nalidixic acid were monitored in vivo using 31 P and 19 F NMR spectroscopy. The results demonstrated that nalidixic acid induced a rapid 2-fold increase in the content of the intracellular NTPs, predominantly ATP. A similar elevation of ATP level was previously demonstrated in E. coli treated with UV radiation (1, 3), as well as after treatment with bleomycin (2). However, we were not able to verify the effect of bleomycin in vivo, because concentrations up to 100 g/ml did not cause a significant increase in NTP (data not shown).
The experimental procedure used previously to determine ATP concentration in E. coli was based on boiling the cells and FIG. 6. 31 P and 19 F NMR spectra of AB1157 cells before and during exposure to nalidixic acid. Cells were grown before the NMR experiment with 15 M 5-fluorouracil as described under "Experimental Procedures." During the NMR experiment the cells were perfused continuously with medium containing 15 M 5-fluorouracil. A, 31 P spectra before (upper trace) and immediately after (0 -9 min; lower trace) exposure to nalidixic acid. Each spectrum was accumulated within 8.8 min and was processed with line broadening of 15 Hz. B, 19 F spectra before (upper trace) and after (9 -17 min; lower trace) exposure to nalidixic acid. Each spectrum was accumulated within 8.5 min and was processed with line broadening of 20 Hz. UDPS, uridine diphosphosugar; PME, phosphomonoesters. Measurements in the presence of 40 g/ml nalidixic acid were termed HIGH concentration. The initial cell density of AB1157 ranged between 0.8 ϫ 10 Ϫ8 and 1.6 ϫ 10 Ϫ8 cells/ml, and that of ABE10 ranged between 3 ϫ 10 Ϫ7 and 5 ϫ 10 Ϫ7 cells/ml. Each column corresponds to 3-5 independent experiments and is given as the mean Ϯ S.E. extracting the ATP. A calculation of the concentration of ATP in the cells, based on the results presented previously (1-3), indicated a very low content of ATP of ϳ4 ϫ 10 Ϫ20 mol/cell, equivalent to a concentration of ϳ0.025 mM. Thus, it appears that this extraction method measured a small fraction of the total ATP pool, estimated to be ϳ3 mM (26). In contrast, we have determined by the NMR in vivo method an NTP content of 4.3 ϫ 10 Ϫ18 mol/cell, which is equivalent to 2.7 mM. The concentration of ATP is ϳ70% of the total NTP concentration, 1.9 mM, close to previous determinations (26).
We therefore attribute the discrepancy between our results and the results reported previously (1)(2)(3) to the difference in the methodology of the measurement.
It is well known that the induction of the SOS regulatory network is controlled by a complex circuitry that involves the RecA and LexA proteins (27). Thus, the nalidixic acid-induced increase in NTP in ⌬recA and lexA mutants clearly indicated that the elevation of NTP is independent of the SOS activity of the drug. The nalidixic acid-mediated induction of NTP in recBC-defective strains further supports this conclusion, be-cause it is known that nalidixic acid is unable to induce an SOS response in the absence of the RecBCD enzyme (12,13). When the recBC-defective mutant was exposed to UV irradiation or bleomycin, which also damaged DNA and induced the SOS response, the intracellular ATP did not increase markedly and rapidly (1,2). This disagreement with our results could arise from variations in the sites of the DNA damage exerted by nalidixic acid and by UV irradiation.
In wild type and lexA and recBC mutants treated transiently for 40 min with nalidixic acid, the level of the NTP pool declined to that of controls after removal of nalidixic acid. It was previously reported that in the lexA1 mutant the ATP level, with exposure to UV radiation or bleomycin, did not decline and remained elevated (2). As in the induction phase of NTP, our findings differ in the recovery phase as well and show a decline in lexA mutant after removal of nalidixic acid. Thus, our in vivo studies do not support the hypothesis previously proposed by Barbé and colleagues (1, 2) that the decrease in ATP during the recovery phase is attributable to cleavage of the LexA repressor by the RecA protease. This discrepancy, as the others described above, might be related to differences in the site of the DNA damage.
We have also followed in vivo changes induced by nalidixic acid in FUra-derived nucleotides using 19 F NMR. It was previously demonstrated that in the presence of FUra the derived F-UTP is readily incorporated into all types of RNA, replacing mainly uracil (28). The incorporation of FUra into RNA over several generations provides a method for observing fluoronucleotides derived from RNA breakdown. At a relatively high dose of FUra (100 M), incorporation into RNA was shown to modify cellular metabolism (28,29). In the present study, a high dose of FUra retarded cell growth. However, at a low dose (ϳ10 -20 M) the rate of cell growth was only slightly reduced, and the energetics, exhibited by the phosphate profile of the cells, was not altered. Using this low dose of FUra enabled us to find out whether uracil nucleotides are involved in the NTPs induction by nalidixic acid and whether the NTPs originate from breakdown of RNA. The results clearly demonstrated that although nalidixic acid induced an increase in the NTPs, the F-UTP remained constant, and the FUra uptake and phosphorylation were totally inhibited. This further indicated that an enhanced synthesis of UTP from its soluble precursors, obtained either from breakdown of RNA or from uptake of FUra, is not involved in the nalidixic acid-induced increase of the NTPs. Previous results showing that nalidixic acid did not induce breakdown of RNA but rather caused DNA degradation (25) are in agreement with our results.
The initial response of ⌬recA cells to nalidixic acid was similar to that of the wild-type cells and the other mutants. However, the ⌬recA cells, unlike the other cells, could not recover after the removal of nalidixic acid, and the increase in ATP could not be reversed. Similar findings were previously reported for a mutant defective in RecA protein and for another mutant with deficient RecA protease activity (1,3). It is known that RecA function is required for homologous recombination and for repair of DNA damage caused by DNA-damaging drugs and by UV radiation (30,31). Furthermore, the recA gene product promotes recognition of homology and strand exchange between two homologous DNAs during both recombination and repair (32,33). Thus, the inability of the ⌬recA mutant to recover from the DNA damage appeared to be related to the repair function of RecA; namely, in the absence of RecA the failure to repair DNA damage is responsible for the continuous irreversible increase in ATP.
It is interesting to note that the mammalian homologue of RecA is Rad51, which regulates the recombination and double- stranded DNA repair in mammalian systems (34,35). RecA and Rad51 proteins were shown to be similar in both structure and function; the primary sequences of the two proteins showed significant homology, and both displayed ATP-dependent DNA binding (36). The failure to properly repair DNA damage in the absence of Rad51 appears to be in the pathway of the p53 checkpoint, leading to cell cycle arrest or apoptosis of mammalian cells (34,37).
It has been recently recognized that programmed cell death is a well established process in the microbial world too (38). For example, binding of microcin B17 and CcdB to DNA gyrase (E. coli topoisomerase II) can cause double-stranded DNA breaks (39) and inhibition of the gyrase (40). These changes triggered the SOS response followed by cell death, reminiscent of apoptosis (40). It is important to note that DNA gyrase, the cellular target of microcin B17 and CcdB protein, is also the target of nalidixic acid, and therefore DNA degradation and cell death induced by nalidixic acid might be likened to apoptosis too. Moreover, CcdB-induced cell killing does not require the host enzymes RecA and RecBC, which are needed for the SOS response (41), a situation similar to that of the nalidixic acid induction of NTP. The morphological changes induced by nalidixic acid showing condensation of the cytoplasm (Fig. 9) could reflect a process reminiscent of apoptosis as the condensation of the cytoplasm in apoptotic mammalian cells (42). Furthermore, the inclusion bodies with a crystalline structure inside the cytoplasm (Fig. 9, E and F), similar to those observed in various bacterial cells in response to stress conditions (43,44), may also suggest the presence of a specific pathway of cell death.
In summary, we have shown that nalidixic acid, a topoisomerase II inhibitor, induces in E. coli wild type and in mutants defected in the SOS response a rapid and marked increase in ATP. This induction is reversible when the treatment with nalidixic acid is transient (40 min), except in ⌬recA cells. The mechanism associated with the specific elevation in ATP level appears to be unrelated to the SOS response but might be related to an apoptotic-like process.