ABSTRACT
In bacteria, SOS is a global response to DNA damage, mediated by the recA-lexA genes, resulting in cell cycle arrest, DNA repair, and mutagenesis. Previously, we reported that Escherichia coli responds to DNA damage via another recA-lexA-mediated pathway resulting in programmed cell death (PCD). We called it apoptosis-like death (ALD) because it is characterized by membrane depolarization and DNA fragmentation, which are hallmarks of eukaryotic mitochondrial apoptosis. Here, we show that ALD is an extreme SOS response that occurs only under conditions of severe DNA damage. Furthermore, we found that ALD is characterized by additional hallmarks of eukaryotic mitochondrial apoptosis, including (i) rRNA degradation by the endoribonuclease YbeY, (ii) upregulation of a unique set of genes that we called extensive-damage-induced (Edin) genes, (iii) a decrease in the activities of complexes I and II of the electron transport chain, and (iv) the formation of high levels of OH˙ through the Fenton reaction, eventually resulting in cell death. Our genetic and molecular studies on ALD provide additional insight for the evolution of mitochondria and the apoptotic pathway in eukaryotes.
IMPORTANCE The SOS response is the first described and the most studied bacterial response to DNA damage. It is mediated by a set of two genes, recA-lexA, and it results in DNA repair and thereby in the survival of the bacterial culture. We have shown that Escherichia coli responds to DNA damage by an additional recA-lexA-mediated pathway resulting in an apoptosis-like death (ALD). Apoptosis is a mode of cell death that has previously been reported only in eukaryotes. We found that E. coli ALD is characterized by several hallmarks of eukaryotic mitochondrial apoptosis. Altogether, our results revealed that recA-lexA is a DNA damage response coordinator that permits two opposite responses: life, mediated by the SOS, and death, mediated by the ALD. The choice seems to be a function of the degree of DNA damage in the cell.
INTRODUCTION
Programmed cell death (PCD) is an active process resulting in cell suicide. Today, any form of cell death mediated by an intracellular death program, no matter what triggers it, is called PCD (1). All eukaryotes have well-regulated PCD pathways built into their genes, which help sculpt organs during development and scuttle infected cells before they get out of control (1–4). Apoptosis, the classical form of eukaryotic PCD, is specifically characterized by DNA fragmentation and membrane depolarization (5–8).
As in many bacteria, the chromosome of Escherichia coli bears the toxin-antitoxin module mazEF (9). In E. coli, mazEF is responsible for nonapoptotic bacterial PCD (10). The mazF gene specifies the stable toxin MazF, and the mazE gene specifies the labile antitoxin MazE, which is degraded by the ClpPA protease (9). Various stressful conditions, including damage to the DNA, induce E. coli MazF (11–13), a sequence-specific endoribonuclease cleaving at ACA sites in mRNAs (14). Furthermore, the induction of MazF leads to the generation of a unique alternative translation machinery composed of leaderless mRNA and ribosomes in which the 16S rRNAs are missing their last 43 nucleotides (15). This alternative translation machinery is responsible for the selective synthesis of specific proteins, some of which, like YfiD, SlyD, ClpX, and YgcR, are involved in cell death (16).
Recently, we showed that when the mazEF-mediated cell death pathway is triggered by severe DNA damage, the operation of an additional cell death pathway in E. coli is prevented (10). This second death pathway is mediated by recA-lexA and, like mazEF, is also triggered by severe DNA damage. However, unlike mazEF-mediated cell death, it is characterized by membrane depolarization and DNA fragmentation, two hallmarks of apoptosis (5–8). Therefore, we called this E. coli novel death pathway apoptosis-like death (ALD) (10).
A well-known cellular response to DNA damage that is mediated by recA-lexA is the extensively studied SOS system (reviewed in references 17 to 21). In an uninduced cell, the lexA gene product, LexA, acts as a repressor of more than 40 genes (22, 23), including recA and lexA, by binding to operator sequences (called an SOS box) upstream from each gene or operon. Damage to the DNA creates regions of single-stranded DNA; single-stranded DNA promotes the conversion of RecA to an active form that can facilitate the otherwise latent capacity of LexA (and some other proteins like UmuD and the λCI repressor) to autodigest (18, 20–22, 24).
Here, we show that the E. coli mazEF pathway inhibits not only the ALD pathway but also the SOS response. In addition, we have more fully characterized ALD and the differences between the SOS response and ALD. Accordingly, ALD, unlike SOS, occurs only under severe DNA damage, and it has the following additional characteristics of apoptosis: (i) rRNA degradation by the endoribonuclease YbeY, (ii) upregulation of a unique set of genes that we called extensive-damage-induced (Edin) genes, (iii) a decrease in the activities of complexes I and II of the electron transport chain, and (iv) the formation of high levels of OH˙ through the Fenton reaction, eventually resulting in cell death. Thus, our results revealed that LexA-RecA is a DNA damage response coordinator that permits two opposite responses: life, mediated by SOS, and death, mediated by ALD. We found that the mechanism for choosing between life and death is related to the concentration of the DNA-damaging agent that we used to treat the cells. Thus, the choice seems to be a function of the degree of DNA damage in the cell.
RESULTS
The SOS response is inhibited by the mazEF pathway.Previously, we reported that the mazEF pathway inhibits the recA-lexA-dependent ALD pathway (10). Since the SOS response is known to be recA-lexA dependent, we asked whether the E. coli mazEF pathway also inhibits the SOS response. To study the SOS response, we used plasmid pL(lexO)-gfp (25), which bears gfp, the gene for green fluorescent protein (GFP), under the control of the lexA operator, lexO. In this system, under uninduced conditions, LexA represses gfp transcription by binding to the SOS box in the gene operator, lexO. Under the stressful condition of DNA damage, RecA becomes activated and acts as a coprotease that stimulates the inactivation of LexA by autocleavage (18), allowing the gfp gene to be transcribed and its fluorescence to be detected. Thus, in this system, fluorescence is a reporter for the RecA-dependent SOS response. Using this fluorescence reporter system, we caused moderate DNA damage by adding 10 µg/ml nalidixic acid (NA) to the cultures. We observed no fluorescence in the wild-type (WT) strain; we did observe fluorescence in the strain lacking mazEF (Fig. 1A). Thus, the SOS response did not take place in the E. coli WT strain MC4100relA+ but only in strain MC4100relA+ΔmazEF, from which the mazEF genes had been deleted (Fig. 1A). Note that the SOS response can be observed after treatment with 10 µg/ml NA. We found similar results when we caused DNA damage by adding norfloxacin (see Fig. S1A in the supplemental material) or by treatment with UV irradiation (see Fig. S1B). Our results suggest that the SOS response cannot take place in the presence of mazEF.
Figure S1
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Effects of the mazEF module, and genes downstream from mazEF, on the SOS response. For the effect of the death genes, we compared E. coli WT strain MC4100relA+ and its ΔmazEF (A), ΔyfbU (B), ΔslyD (C), ΔyfiD (D), ΔclpP (E), and ΔygcR derivatives (F). For the effect of the survival genes, we compared the WT ΔdeoC (G) and ΔelaC (H) (marked in blue) derivatives with MC4100relAΔmazEF. All of these strains harbored plasmid pL(lexO)-gfp. After growing the cells in M9 medium with added ampicillin (100 µg/ml) and with shaking at 37°C to an OD600 of 0.5 to 0.6, we treated (time zero) them (or not) with 10 µg/ml NA. We measured fluorescence units (FU) using a fluorometer over a period of 4 h (240 min). All of the values shown are relative to those of the control cells not treated with NA.
Previously, we found that the nutritional starvation signal (p)ppGpp, responsible for the stringent response (26), is involved in the mazEF-mediated cell death (9). Here, we asked whether the SOS response was permitted in E. coli strains defective in (p)ppGpp production. To this end, we first used E. coli strain MC4100relA1 in which the relA gene is impaired by an insertion (relA1) (27). Indeed, we did observe the SOS response in the MC4100relA1 strain (see Fig. S2 in the supplemental material). This result explains the reason for the induction of the SOS response in MC4100relA1, although this strain carries a mazEF module (26).
Figure S2
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Previously, we also reported that the induction of the mazEF-mediated death pathway activates the selective synthesis of two groups of proteins: the products of genes yfbU, slyD, yfiD, clpP, and ygcR, which participate in the death process (the “death genes”), and the products of genes elaC and deoC, which lead to cell survival (the “survival genes”) (16). Here, we found that the SOS response took place in cells from which we deleted any one of the individual death genes (Fig. 1B to F); however, as in the WT strain, the SOS response did not take place in cells from which we deleted either of the survival genes deoC and elaC (Fig. 1G and H). These results suggest that the SOS response requires the absence of either mazEF or one of the other death genes.
In addition to the use of plasmid pL(lexO)-gfp, we also studied SOS by following the increase in mRNA levels of dinB and umuC, which are known to be induced as part of the SOS response (21). As in the case of pL(lexO)-gfp, under the condition of SOS induction with 10 µg/ml NA, the levels of both dinB and umuC are not increased in the WT strain MC4100relA+, but they are significantly elevated in its ΔmazEF derivative (see Fig. S3 in the supplemental material).
Figure S3
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
ALD-mediated DNA fragmentation occurs only under conditions of severe DNA damage.Previously, we showed that one of the characteristics of E. coli ALD is membrane depolarization, a hallmark of eukaryotic apoptosis. This membrane depolarization takes place only under conditions of severe DNA damage (above 50 µg/ml NA) (10). We wondered if DNA fragmentation, the other characteristic common to both eukaryotic apoptosis and E. coli ALD, might also occur only under conditions of severe DNA damage. To this aim, we treated E. coli strain MC4100relA+ (WT) and its ΔmazEF derivative with either a low (10-µg/ml) or a high (100-µg/ml) concentration of NA. We found that DNA fragmentation occurred only in the absence of mazEF (in strain MC4100relA+ΔmazEF) and only when the strain was exposed to a high concentration of NA (100 µg/ml) (Fig. 2E), not a low concentration of NA (10 µg/ml) (Fig. 2F). Thus, with a low concentration of NA, which we have shown to induce the SOS response (Fig. 1), DNA fragmentation does not take place. Thereby, we could distinguish the SOS response from ALD as a function of the concentration of the agent causing the DNA damage.
DNA fragmentation as a function of treatment with three concentrations of nalidixic acid (detected by TUNEL assay). WT E. coli MC4100relA+ (A to C) or MC4100relA+ΔmazEF (D to F) was grown to an OD600 of 0.6. Each culture was divided into three samples that were treated with the addition of no NA, 100 µg/ml NA, or 10 µg/ml NA. DNA fragmentation was determined using the Apo-Direct kit (10). The location of the horizontal threshold line between TUNEL-negative cells (highlighted in red) and TUNEL-positive cells (highlighted in blue) was determined using untreated, unstained MC4100relA+ cells. These data represent the results of one of three similar experiments. FITC, fluorescein isothiocyanate; FSC, forward scatter.
rRNA degradation is an additional characteristic of ALD.In metazoans, apoptosis involves not only irreversible DNA damage but also cleavage in specific places on several RNA species, including 28S rRNA, U1 snRNA, and Ro RNP-associated Y RNAs (28). In addition, when the yeast Saccharomyces cerevisiae was exposed to a variety of stimuli known to induce apoptosis, the two rRNAs 25S and 5.8S were degraded extensively (29). Here, we asked whether RNA degradation is involved in E. coli ALD. As in our previous experiments, we treated the WT E. coli strain MC4100relA+ and its ΔmazEF derivative with 100 µg/ml NA. Then, we extracted the RNA and determined the integrity of the RNA. We found that massive degradation of the 16S and 23S rRNAs occurred only in the ΔmazEF derivative and only when it was treated with a high concentration (100 µg/ml) of NA, the characteristic conditions for ALD (Fig. 3A). However, the degradation of the rRNAs does not occur under a low concentration of NA (10 µg/ml) (Fig. 3A), which we showed to induce the SOS response (Fig. 1). In addition, after treating the ΔmazEF strain with a high concentration (100 µg/ml) of NA, we followed the progress of 16S rRNA and 23S rRNA degradation over time (Fig. 3B). After half an hour, we observed some degradation of the 23S rRNA; this degradation continued gradually over 4 h, at the end of which time we found that most of the 23S rRNA was completely degraded. The degradation of the 16S rRNA started later, only after 1.5 h of treatment, and by 4 h, most of the 16S rRNA was degraded (Fig. 3B).
E. coli ALD is also characterized by rRNA degradation. (A) rRNA degradation occurs only at high NA concentrations and only in the ΔmazEF strain. E. coli MC4100relA+ (WT) or its ΔmazEF derivative was grown to an OD600 of 0.5. Each culture was divided into three samples that were treated with the addition of 100 µg/ml NA or 10 µg/ml NA or not treated. To detect RNA degradation, after 4 h of incubation without shaking at 37°C, RNA was extracted and a Bioanalyzer was used to determine the integrity of the RNA. (B) rRNA degradation in ΔmazEF cells over time. MC4100relA+ΔmazEF was grown, treated, and incubated with 100 µg/ml NA as in panel A. At the times indicated, RNA integrity was determined as described for panel A. (C) The ybeY gene but not the rnr gene is involved in the degradation of rRNA in the E. coli ΔmazEF derivative. WT MC4100relA+ and its ΔmazEF, ΔmazEFΔrnr, or ΔmazEFΔybeY derivative were grown and treated with 100 µg/ml NA as described for panel A. RNA was extracted, and its integrity was determined as described for panel A.
We asked which enzyme is responsible for the degradation of the rRNA under conditions permitting ALD. Recently, Jacob and colleagues (30) showed that YbeY, a previously unidentified endoribonuclease, and the exonuclease RNase R (encoded by rnr) act together to degrade defective 70S ribosomes, sparing properly matured 70S ribosomes and intact individual subunits. Here, from strain MC4100relAΔmazEF, we deleted each of these RNases separately. Deleting the rnr gene did not affect RNA degradation (Fig. 3C). However, deleting the ybeY gene clearly prevented the degradation of the 23S rRNA, and instead of the 16S rRNA, we found two other products: one of them seems to be a 17S rRNA and the second is smaller than the 16S rRNA (Fig. 3C). Since Jacob and colleagues (30) showed that YbeY is involved in the critically important processing of the 3′ terminus of the 16S rRNA, we think that this 17S rRNA might be a precursor of the 16S rRNA. We do not have yet a clue to the nature of the other product, which is smaller than the 16S rRNA. Our results show that, under conditions leading to ALD, the product of the ybeY gene, but not that of the rnr gene, was involved in the degradation of rRNA. Thus, we have shown that ALD was characterized not only by membrane depolarization and DNA fragmentation (10) but also by rRNA degradation. Moreover, like the other properties of ALD, RNA degradation occurred only when cells were treated with a high concentration of NA. Based on these results, we suggest that treatment with a high concentration of NA is the condition for ALD; treatment with a low concentration of NA is the condition for SOS.
Viability assays distinguish between ALD and SOS.To further distinguish between ALD and the SOS response, we measured the cell survival (CFU) of E. coli WT strain MC4100relA+ and its ΔmazEF derivative under the condition of extensive DNA damage (100 µg/ml NA) and under the condition of moderate DNA damage (10 µg/ml NA). When we applied 100 µg/ml NA, the mazEF death pathway in the WT strain caused a reduction in cell survival of about 5 orders of magnitude. Under the same condition, in the strain lacking mazEF (ΔmazEF), the ALD pathway caused a reduction in cell survival of about 4 orders of magnitude (see Fig. S4A in the supplemental material). In contrast, when we applied 10 µg/ml NA, we observed a reduction in viability of about 3 orders of magnitude only in the WT strain in which the mazEF death pathway was functioning; under this condition, cell survival was only slightly affected in the ΔmazEF strain (Fig. S4B), suggesting that ALD was not functioning. It seems likely that the killing effect was spared by the SOS response, which we have found to be induced under such conditions (Fig. 1A). In summary, our results suggest that in the absence of the mazEF pathway (ΔmazEF), the cellular response depends on the intensity of the DNA damage: moderate DNA damage led to the SOS response and cell survival (Fig. S4B), and severe DNA damage activated the ALD pathway, leading to cell death (Fig. S4A). Thus, we found previously that ALD leads to membrane depolarization and DNA fragmentation (10), and we found here that ALD led also to RNA degradation (Fig. 3) and eventually to cell death (Fig. S4A).
Figure S4
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
LexA degradation is increased under the conditions of ALD.Previously, we showed that, like the better-known SOS response, ALD is recA-lexA dependent (10). Here, we hypothesize that we can distinguish between the ALD response and the SOS response by the degree of RecA coprotease activity. We determined RecA coprotease activity by measuring LexA degradation. After various time periods, under conditions of ALD (100 µg/ml NA) or SOS (10 µg/ml NA), we extracted proteins from the WT strain and its ΔmazEF derivative. We quantified LexA degradation by subjecting the proteins to Western blot analysis using anti-LexA antibody. As lexA3 encodes a noncleavable LexA repressor (31), as a control we used a lexA3 mutant of E. coli strain MC4100relA+ΔmazEF (10). As expected, under all conditions tested, LexA was not degraded in either the WT or the lexA3 mutant: the LexA protein remained intact (Fig. 4). In contrast, in the ΔmazEF strain, LexA was partially degraded under the SOS conditions (10 µg/ml NA) and totally degraded under the ALD conditions (100 µg/ml NA) (Fig. 4). This degradation was visible within 1 h of incubation with 100 µg/ml NA. LexA degradation also took place in MC4100relA+ mutants from which we deleted not mazEF itself but rather the gene yfiD or slyD downstream of mazEF (Fig. 4). Here also, LexA degradation can be observed mainly under the ALD conditions (100 µg/ml). However, in this experiment, it can minimally be observed under the SOS conditions (10 µg/ml). Altogether, we see that (i) in the WT strain, in the presence of the mazEF system, RecA is not active as a coprotease, and (ii) in the absence of the mazEF system, RecA is more active as a coprotease under ALD conditions (treatment with 100 µg/ml NA) than under the condition of the SOS response (treatment with 10 µg/ml NA).
LexA degradation is increased under high concentrations of nalidixic acid in the absence of the mazEF pathway. The degree of LexA degradation was reflected by the level of RecA activity. WT E. coli strain MC4100relA+ and each of its ΔmazEF, ΔyfiD, ΔslyD, and ΔmazEFlexA3 derivatives were grown to an OD600 of 0.5. Each culture was divided into three aliquots to which appropriate concentrations of nalidixic acid (NA) were added. These samples were incubated for up to 4 h. At the times indicated, cells were lysed and the proteins were analyzed by Western blot analysis using rabbit polyclonal antibody to the LexA DNA binding region.
The ALD pathway includes the expression of a new set of genes of the E. coli oxidative respiratory system.Having found that LexA degradation was significantly higher under ALD than under SOS conditions, we hypothesized that additional genes tightly regulated by LexA would be transcribed under ALD conditions. We used microarray analysis to test the transcription pattern of upregulated genes in ALD. We added no NA (no treatment [n.t.]), 100 µg/ml NA, or 10 µg/ml NA to cultures of the WT strain MC4100relA+ or its ΔmazEF or ΔmazEFlexA3 derivatives. We extracted the RNA and compared the gene expression of treated samples with that of the untreated sample, resulting in 6 lists of upregulated genes (see Fig. S5A in the supplemental material, lists 1 to 6). We wished to focus on the ALD pathway genes that were specifically upregulated only in the ΔmazEF derivative and only after treatment with 100 µg/ml NA. To separate the nonspecifically upregulated genes (Fig. S5A, list 1, and Fig. S5B, blue oval area marked “1”) from the specifically upregulated genes (ΔmazEF + 100 µg/ml NA), we eliminated all of the expressed genes that had been induced in the other 5 lists (Fig. S5B and C), resulting in a list of 190 specifically upregulated genes (Fig. S5A, bottom row). We saved only one copy of genes that appeared no more than once, leaving 162 genes of the ALD pathway (see Table S1). After eliminating 42 of these genes whose functions are not yet known, we divided the remaining 120 genes by function (Fig. S5D and Table S1). In addition to the 8 known SOS genes, we were amazed to find 5 tricarboxylic acid (TCA) cycle genes, 7 respiratory electron transport chain genes, and 12 Fe-S cluster genes, all of which were specifically upregulated.
Figure S5
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Table S1
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Under ALD conditions, the activities of respiratory electron transport chain complexes I and II are decreased.In our microarray results, the genes upregulated only under ALD conditions included genes of the electron transport chain system, suggesting that ALD conditions may affect the activity of that system. The upregulated genes included nuoG, sdhA, and sdhB. nuoG encodes NADH dehydrogenase I (NuoG), which transfers electrons from NADH to respiratory system complex I. SdhA and SdhB, products of sdhA and sdhB, are two subunits of succinate dehydrogenase (SDH), which transfers electrons in the electron transfer chain from succinate to flavin adenine dinucleotide (FAD) in complex II. We tested NADH dehydrogenase I (complex I) and succinate dehydrogenase (complex II) activities. Since the microarray analysis revealed upregulation of genes in the respiratory chain, we expected increases in the activities of both complexes I and II. Instead, under ALD conditions (4 h with 100 µg/ml NA), although complex II activity increased in the WT strain, it drastically decreased (about 90%) in the ΔmazEF strain (Fig. 5B). Under the same conditions, the activity of complex I decreased significantly more in the ΔmazEF strain (about 56%) than in the WT strain (about 28%) (Fig. 5A). Note that under SOS conditions (4 h with 10 µg/ml NA), complex II activity increased in both the WT and its ΔmazEF derivative (see Fig. S6B in the supplemental material). In strain MC4100relA+lexA3ΔmazEF, in which the LexA repressor is not cleaved, under ALD conditions, the activities of complexes I and II did not decrease (Fig. 5A and B), indicating that the decreases were recA-lexA dependent. Under ALD conditions, these decreases in activity might be dependent on 9 Fe-S clusters in complex I and 4 Fe-S clusters in complex II. This possibility is supported by our measurements of the activity of E. coli NADH dehydrogenase II (Fig. S6C), which, instead of Fe-S clusters, contains a thiol-bound Cu(I) and a putative copper binding site (32). Under ALD conditions, we observed no change in NADH dehydrogenase II activity (Fig. S6C). Our combined results suggest that, under ALD conditions, mainly complex II activity was decreased.
Figure S6
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Complex I and complex II activities are decreased by high concentrations of NA in the ΔmazEF derivative of MC4100relA+. E. coli strains MC4100relA+, MC4100relA+ΔmazEF, and MC4100relA+lexA3ΔmazEF were grown to an OD600 of 0.5. Each culture was divided into two samples to which either no NA (n.t.) or 100 µg/ml NA was added. After 4 h of incubation, total proteins were extracted. Enzymatic activities were determined as described in Materials and Methods. (A) Complex I activity (NADH dehydrogenase I) was determined by measuring the reduction levels of deamino-NADH. (B) Complex II activity (succinate dehydrogenase) was determined by measuring the reduction levels of dichlorophenol indophenol. These data represent the results of three experiments.
ALD conditions induce high levels of OH• formation.In eukaryotic cells, mitochondrial membrane depolarization causes reactive oxygen species (ROS) production, important for inducing apoptosis (33). In mitochondrial apoptosis, NDUFS1 protein from complex I (an NADH dehydrogenase of the electron transport chain) is cleaved by caspases, leading to the production of ROS (34). Under ALD conditions, we observed decreased activity in the electron transport chain (especially in complex II activity) (Fig. 5; see also Fig. S6 in the supplemental material), so we asked whether in E. coli, in the absence of mazEF, ROS would be formed under the ALD conditions. The hydroxyl radical (OH˙), which is highly hazardous to the cell, is the final product of the Fenton reaction. We examined OH production under ALD conditions. We grew cells and treated them with NA (100 µg/ml). We measured the formation of OH using hydroxyphenyl fluorescein (HPF), which is nonfluorescent until it reacts with OH. While we detected no OH formation in the WT strain (Fig. 6A), we detected relatively high levels of OH formation in the ΔmazEF strain (Fig. 6B). Under the ALD condition, we also detected OH˙ formation (see Fig. S7A and B), as well as membrane depolarization (see Fig. S7C and D), in strain MC4100relA1, which carries WT mazEF but an impaired relA (27). As was shown by us above, this strain permits also the SOS response (Fig. S2), probably because the function of mazEF is prevented due to impairment in the production of (p)ppGpp. In addition, the OH˙ levels in strain MC4100relA+ΔmazEF were higher under ALD conditions (100 µg/ml NA) than under SOS conditions (10 µg/ml NA) (Fig. 6B). OH˙ formation could also be detected in the derivatives deleted in yfiD or slyD (acting downstream of the mazEF gene) but only under ALD conditions (100 µg/ml NA) (see Fig. S8A and B). We also asked if OH˙-neutralizing agents would prevent ALD. To neutralize OH˙, we used thiourea (an OH˙ scavenger) or dipyridyl (a Fe2+ chelator and thereby an inhibitor of the Fenton reaction). We studied four ALD hallmarks: membrane depolarization, RNA degradation, LexA degradation, and reduction in viability. The addition of either thiourea or dipyridyl prevented membrane depolarization (Fig. 6C) and RNA degradation (Fig. 6D) and reduced RecA coprotease activity (reflected by LexA cleavage [see Fig. S9A in the supplemental material]). Furthermore, adding dipyridyl or thiourea to the growth medium reduced cell killing by NA (100 µg/ml) by about 2 orders of magnitude (see Fig. S9B). Note that adding either thiourea or dipyridyl also prevented HPF staining, supporting the notion that the HPF staining is specific to OH˙ formation (see Fig. S9C).
Figure S7
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Figure S8
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Figure S9
Copyright © 2014 Erental et al.This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
ALD involves high levels of OH˙ formation and is prevented by Fe2+ chelator and by the mazEF pathway. (A and B) OH˙ formation was determined by HPF staining. E. coli WT strain MC4100relA+ (A) or its ΔmazEF derivative (B) was grown to an OD600 of 0.5. Each culture was divided into three samples that were treated with the addition of 100 µg/ml NA or 10 µg/ml NA or not treated. After 4 h of incubation, the cells were stained with HPF to detect the formation of OH˙. The fluorescence intensity (FI) of the HPF in the cells was determined by FACS. These data represent the results of one of three similar experiments. (C) Thiourea and dipyridyl prevented membrane depolarization. E. coli strain MC4100relA+ΔmazEF was grown to an OD600 of 0.5. The culture was divided into 4 aliquots. To one (control) aliquot, no additions were made (n.t.); to each of the three other aliquots, we added 100 µg/ml nalidixic acid (NA); to one of these, we added 100 mM thiourea scavenger OH˙, and to another, we added 500 µM dipyridyl to chelate Fe2+ ions. After 4 h of incubation, the cells were stained with DiBAC4. The intensity of the fluorescence (FI) of DiBAC4 in the cells was determined by FACS as described previously (10). Cells that exhibited a DiBAC4 FI value higher than 103 were considered to have undergone membrane depolarization. (D) Thiourea or dipyridyl prevented RNA degradation. E. coli strain MC4100relA+ΔmazEF was grown to an OD600 of 0.5. After 4 h of incubation with drugs (100 µg/ml NA and either 500 µM dipyridyl or 100 mM thiourea), RNA was extracted and RNA integrity was determined as described in the legend to Fig. 3. (E and F) Deleting the iscS gene led to a significant reduction in HPF staining. HPF staining was done as described for panels A and B. (G and H) Deleting the iscS gene prevented membrane depolarization. Membrane depolarization was done as described for panel C. These data represent the results of one of three similar experiments.
Gene iscS codes for ISC cysteine desulfurase, responsible for Fe-S cluster formation; deleting iscS abolishes the Fenton reaction (35). Since our results suggest that the Fenton reaction is involved in ALD, we wished to clarify the involvement of iscS under ALD conditions. Indeed, deleting iscS from the ΔmazEF strain prevented OH˙ formation (compare Fig. 6E and F) and membrane depolarization (compare Fig. 6G and H), making the formation of high levels of OH˙ another hallmark of ALD and probably the cause of its lethal effect on E. coli. Note that under SOS conditions the levels of OH˙ in the ΔmazEF derivative were much lower than those under ALD conditions (Fig. 6B).
DISCUSSION
Though both are mediated by RecA-LexA, E. coli ALD and SOS appear to be two different responses.In the 1970s, Radman first described the bacterial response to DNA damage, calling it the “SOS response” (19). So far, it is the largest, most complex, and best-understood bacterial DNA damage-inducible network (17–21, 24). In the SOS regulatory network, gene expression is controlled by a complex circuit involving RecA and LexA (21). Inducing SOS leads to the generation of regions of single-stranded DNA. In the presence of nucleotide triphosphates, when RecA binds to these regions of single-stranded DNA a nucleotide filament is generated that converts RecA to an active form. Activated RecA facilitates the otherwise latent capacity of LexA (and additional proteins) to undergo self-cleavage (18, 20–22, 24).
Previously, we showed that E. coli apoptosis-like death (ALD) is also mediated by recA-lexA. We showed that ALD shared two hallmarks with eukaryotic mitochondrial apoptosis: membrane depolarization and DNA fragmentation. Based on our previous work showing that membrane depolarization occurs only under a high concentration of NA (above 50 µg/ml), we suggested that ALD is an SOS response to extreme DNA damage (10).
Here, we found that ALD shares additional hallmarks with eukaryotic mitochondrial apoptosis, including rRNA degradation (Fig. 3) and the formation of high levels of OH˙ (Fig. 6). Moreover, we found that ALD is also characterized by the induction of the expression of a new set of genes that we called extensive-damage-induced (Edin) genes (see Fig. S5 in the supplemental material). Finally, we found that ALD also involved a decrease in the activities of respiratory system complexes I and II (Fig. 5; see also Fig. S6). All of these ALD hallmarks, including membrane depolarization and DNA fragmentation as we have previously reported (10), are inhibited by the mazEF pathway (Fig. 3, 5, and 6). We also found that these ALD hallmarks occurred only when the cells were treated with a high concentration of NA (Fig. 2, 3, 4, and 6; see also Fig. S4); treatment with a low concentration of NA induced the SOS response (Fig. 1). Our results support the hypothesis that E. coli ALD and SOS are two responses, both mediated by recA-lexA but activated under different conditions. ALD is activated when the cell experiences high concentrations of DNA-damaging agents; SOS is activated when the cell experiences low concentrations of DNA-damaging agents. Thus, it seems likely that the degree of the DNA damage is what leads to the choice between SOS and ALD. Our results also revealed that, as in the case of the ALD pathway (10), the SOS response is also inhibited by the mazEF-mediated pathway (Fig. 1). In both ALD and the SOS response, in addition to mazF, the death genes downstream from mazF also contribute in some way to this inhibition (Fig. 1B to F).
A model for the intertwined pathways: ALD, the SOS response, and the mazEF death pathway.Our model for the ALD pathway (Fig. 7A) is based on the idea that the level of RecA activation in the cell is a function of the level of damaged DNA. Recall that activated RecA acts as a coprotease that stimulates the inactivation by autocleavage of the LexA repressor molecule (18). Indeed, we found that LexA was degraded more extensively under ALD conditions (high NA concentration) than under SOS conditions (low NA concentration) (Fig. 4). Under ALD conditions, strong RecA activation led to high levels of LexA degradation, leading to the transcription of a new set of genes that we have called Edin (see Table S1 in the supplemental material). We hypothesize that Edin genes are tightly regulated by LexA and that, under ALD conditions, at least some of them are responsible for the decreases in the activities of complexes I and II of the respiratory electron transport chain (Fig. 5; see also Fig. S6). In fact, using the lexA3 allele, which encodes an uncleavable LexA repressor, we showed that the decrease in the activities of complexes I and II depends on the ability of RecA to cleave the LexA repressor: in strain MC4100relA+ΔmazEFlexA3, we observed no decreases in the activities of complexes I and II (Fig. 5), suggesting that one or more of the products of the Edin genes, usually repressed by LexA, may be responsible for the decrease in complex I and complex II activities. It is also likely that the profound decrease in complex II activity might have led to the formation of high levels of OH (Fig. 6). Thus, under ALD conditions, we suggest that the formation of OH˙ is as follows: when complex II activity is decreased, fewer free electrons are transferred in the respiratory chain, resulting in the formation of superoxides, which damage iron-sulfur clusters, making ferrous iron available for oxidation by the Fenton reaction. Finally, this leads to OH˙ formation that causes double-strand breaks (DSBs) in the DNA (Fig. 2E) and, possibly, cell death (see Fig. S4).
Model for the two recA-lexA pathways, ALD (A) and the known SOS response (B), and their inhibition by the EDF-mazEF pathway. The EDF-mazEF-mediated pathway is activated under both severe (A, right) and moderate (B, right) DNA damage. Under both conditions, we observed a difference only in the level of bacterial survivors (A, right, versus B, right). The EDF-mazEF-mediated pathway inhibits the recA-lexA-mediated pathway. Therefore, in the absence of a functional EDF-mazEF-mediated pathway, the recA-lexA pathway is permitted. Under severe DNA damage, strong activation of RecA occurs, leading to the ALD pathway (A, left). On the other hand, under moderate DNA damage, only a moderate activation of RecA occurs, leading to the SOS response (B, left). For additional details regarding the ALD and SOS pathways, see the Discussion. For further details regarding the EDF-mazEF-mediated pathway, see references 10, 15, and 16.
That the ALD pathway may require free ferrous ions oxidized through the Fenton reaction was supported by our results showing that neutralizing the ferrous ions with dipyridyl prevented ALD (Fig. 6; see also Fig. S9 in the supplemental material). The iscS gene is responsible for Fe-S cluster formation and the Fenton reaction (35). Deleting iscS from the ΔmazEF strain prevented the formation of OH˙ and membrane depolarization (Fig. 6E to H). Since our experiments show that under the ALD condition, OH˙ scavengers can overcome LexA cleavage (see Fig. S9A) and membrane depolarization (see Fig. S9C), we suggest that OH˙ production is upstream of membrane depolarization (Fig. 7A). It seems also possible that under ALD conditions, in addition to mediating cell death ALD might also amplify RecA activity. Under these conditions, high levels of OH˙ formation (Fig. 6B) probably damage the DNA (Fig. 2E) and are causing double-strand breaks (DSBs) (Fig. 7A). These DSBs would eventually result in the formation of single-stranded DNA, which in turn would amplify RecA activity (21). This hypothesis is supported by our results from studying the effect on RecA activity of LexA cleavage, which show that inhibition of OH˙ formation with dipyridyl or thiourea decreases RecA activity (see Fig. S9A).
We paid special attention to the Edin genes, which were upregulated under ALD conditions. In our model (Fig. 7A), upregulated Edin genes may be responsible for the decrease in the activity of the electron transport chain and thus play a part in mediating ALD. Alternatively, rather than being ALD mediators themselves, under ALD conditions, the Edin genes may be upregulated in response to cell damage. Recall that the gene nuoG encodes NADH dehydrogenase I, part of complex I, and that sdhA and sdhB encode succinate dehydrogenase, which is part of complex II. The upregulation in the expression of these three genes may represent a cell response either to the decrease in complex I and II activities or, alternately, to damage to the protein molecules of these complexes. Further studies are required to elucidate whether Edin products are ALD mediators or are produced as a cell response to ALD damage.
We also created a model for the SOS response (Fig. 7B). In contrast to ALD, we found that the SOS response occurred only under conditions of moderate DNA damage (10 µg/ml NA) (Fig. 1). Under these conditions, RecA is only moderately activated, leading to only partial LexA degradation (Fig. 4). Since all din genes are repressed by LexA, partial LexA degradation may permit transcription of these din genes, leading to cell cycle arrest and DNA repair activation and, finally, to cell survival (21) (Fig. 7B).
The results of our study suggest that RecA-LexA may be DNA damage response coordinators, permitting one of two opposite responses: life (mediated by the SOS response) (Fig. 7B) and death (mediated by the ALD pathway) (Fig. 7A). The choice between life and death seems to be a function of the degree of DNA damage resulting from the concentration of the DNA-damaging agent to which the cells are exposed (Fig. 2).
Note that both the SOS response and ALD are inhibited by the mazEF-mediated pathway (10) (Fig. 1 to 6). This inhibition was not caused directly by MazF activity, since it was also prevented in MC4100relA+ derivatives from which we deleted additional genes that act downstream from MazF (10) (Fig. 1, 2, 4, 5, and 6). Until now, we have characterized only 15 out of the 50 genes that act downstream from MazF (16), so we still do not know the exact mechanism of this inhibition. It will be important to elucidate this mechanism, because, as we have suggested previously (10), ALD may be a backup system for the mazEF-mediated cell death program. Mutation(s) in one or more of the components of the extracellular death factor (EDF)-mazEF-mediated pathway would lead to the formation of EDF-mazEF “cheaters,” so that the EDF-mazEF-mediated pathway would be inactivated. Should that happen, bacterial cell death would occur through the ALD pathway (10).
Evidence for apoptosis-like death in bacteria.Recent research has revealed evidence of apoptotic death in bacteria (summarized in reference 36). Here, we will only specifically describe two of the reports on death induced by DNA damage. Independently of our previous research (10), apoptosis-like death in E. coli was also described by Dwyer et al. (37). In addition to membrane depolarization and DNA fragmentation, they found that this apoptosis-like death is also characterized by phosphatidylserine exposure to the outer leaflet of the plasma membrane and chromosome condensation. In particular, they found that RecA can bind peptide sequences that serve as the substrates for eukaryotic caspases, suggesting that RecA acts as a caspase. In contrast to our results showing that ALD is induced only by DNA-damaging agents (10), as they described it (37), apoptosis-like death is induced not only by DNA-damaging agents but also by various antibiotics, including spectinomycin, ampicillin, and gentamicin. Their results (37) may differ from ours (10) because they used LB growth medium while we used M9 and they used E. coli strain MG1655 while we used MC4100relA+ΔmazEF. In earlier work, we showed that MG1655 did not produce the extracellular death factor (EDF) (38) that is required for the enhancement of MazF activity (39) and for the inhibition of ALD (10). Thus, MG1655 behaves like a ΔmazEF strain (10). Therefore, it seems to us that, even without deleting the mazEF module, Dwyer and colleagues (37) have discovered bacterial apoptosis-like death because the bacterial strain that they used was lacking EDF.
As a whole, our experiments have revealed that the operation of the whole EDF-mazEF-mediated pathway is required for the inhibition of ALD. It is possible that, as in E. coli strain MG1655, which is deficient in EDF production (38), and MC4100relA1, which is impaired in the formation of ppGpp (26), ALD (as well as SOS) will be permitted in some other E. coli strains, although they carry the mazEF module on their chromosome.
Apoptosis-like death under conditions of DNA damage has also been described in Caulobacter crescentus (40), where the executor protein is BapE, which perturbs chromosome integrity by cleaving supercoiled DNA in a sequence-nonspecific manner. The resulting DNA fragmentation, chromosome condensation, and membrane depolarization are characteristic of eukaryotic apoptosis. As far as we know, there is no BapE homologue in E. coli.
Evolutionary aspects of bacterial ALD.E. coli ALD, having biochemical and morphological hallmarks of apoptosis, may provide additional insight into the evolution of mitochondria and their apoptotic pathway in eukaryotes. It is possible that mitochondria and their apoptotic machinery were evolved during endosymbiosis, when the prokaryotic protomitochondrion was introduced into the primitive protoeukaryotic cell. A relationship between E. coli ALD and mitochondria is already indicated in our present study, in which we show that one of the ALD hallmarks is rRNA degradation in which E. coli YbeY is involved (Fig. 3). Intriguingly, the mammalian counterpart of YbeY is localized in the mitochondria (41), hinting at a parallel between ALD in E. coli and mitochondrial apoptosis. Moreover, YbeY is one of 19 proteins predicted to be involved in complex I disease biology (41). Based on our work here, one might hypothesize that YbeY has a role in eukaryotic mitochondrial apoptosis. Finally, the decrease in complex I and II activities (Fig. 5) led to the formation of high levels of OH˙ (Fig. 6). E. coli complex I activity is carried out by NuoG (NADH dehydrogenase I), which is strikingly similar to the eukaryotic mitochondrial NDUFS1 (42). Cleavage of NDUFS1 by caspases leads to the formation of OH˙ and eventually to apoptotic death (34).
Further investigation of ALD and other described apoptosis-like pathways (36, 37, 40) may provide more evidence for the origin of mitochondrial apoptosis. One intriguing scenario suggested that eukaryotes developed a control system that would avoid PCD by the endosymbiont machinery. By doing so, in parallel, the ancient eukaryotes made use of them when required (43).
MATERIALS AND METHODS
Bacterial strains and plasmids.We used the following E. coli strains: MC4100relA+ (WT) (44) and its derivatives MC4100relA+ΔmazEF (44), MC4100relA+ΔslyD (16), MC4100relA+ΔyfiD (16), MC4100relA+lexA3ΔmazEF (10), and MC4100relA1 (27). Using the method of Datsenko and Wanner (45), we also constructed E. coli strains MC4100relA+ΔmazEFΔrnr, MC4100relA+ΔmazEFΔybeY, and MC4100relA+ΔmazEFΔiscS. We also used plasmid pL(lexO)-gfp (25).
Materials and media.Bacterial cultures were grown in liquid M9 minimal medium containing 1% glucose and a mixture of 100 µg/ml of each amino acid except tyrosine and cysteine (38). Nalidixic acid (NA), norfloxacin, thiourea, and 2,2′-dipyridyl were obtained from Sigma (St. Louis, MO, USA). Bis-1,3-dibutylbarbituric acid trimethine oxonol (DiBAC4) and hydroxyphenyl fluorescein (HPF) were purchased from Invitrogen (Carlsbad, CA, USA). To detect DNA fragmentation, we used the Apo-Direct kit purchased from Calbiochem (San Diego, CA, USA).
Growth conditions and viability.For the experiments on membrane depolarization, DNA fragmentation, RNA degradation, microarray analysis, OH˙ detection, and cell viability, we grew cells in 10 ml M9 minimal medium to an optical density at 600 nm (OD600) of 0.5 to 0.6. Then, we divided each culture into 500-µl aliquots to which we added the appropriate concentration of NA. We incubated each aliquot at 37°C for 4 h (or as described in the figure legends) and then washed them twice with phosphate-buffered saline (PBS) (pH 7.2). We carried out viability assays on LB plates as we have described previously (16). When we used thiourea or 2,2′-dipyridyl, cells were grown as described above and then incubated with 2,2′-dipyridyl (500 µM) or thiourea (100 mM) at 37°C for 10 min. NA was added as described in the figure legends.
Using a GFP reporter to measure the SOS response.To E. coli cultures harboring plasmid pL(lexO)-gfp grown and treated as described above, we added 100 µg/ml ampicillin. Using a 485- ± 15-nm excitation filter and a 530- ± 15-nm emission filter, we measured the fluorescence 25 times in each well at intervals of 10 min. The fluorophore was excited with 1,000-cW lamp energy, and the fluorescence in each well was measured for 1 s (FLUOstar Galaxy; BMG Labtechnologies). For UV irradiation experiments, we grew cells harboring plasmid pL(lexO)-gfp as described in the legend to Fig. 1. At an OD600 of 0.5 to 0.6, we transferred 1 ml of the culture to a 100-mm petri dish and irradiated the cultures with 104 µJ/cm2 UV light inside a UVC 500 cross-linker (Hoefer) for 4 s. Then, we placed 250-µl samples into each well of a 96-well plate. We measured fluorescence (fluorescence units [FU]) with a fluorometer over a period of 4 h. All of the values shown are relative to those of control cells not irradiated with UV.
DNA fragmentation assay using FACS analysis.DNA fragmentation was detected by the terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assay (46). To perform the TUNEL assay, we used the Apo-Direct kit and detected fluorescence in individual DNA fragmented cells by flow cytometry as we have previously described (10). We determined the intensity of the fluorescence by fluorescence-activated cell sorting (FACS) analysis using the LSRII FACS machine and FCS Express V3 software, as we have described previously (10).
Determining RNA integrity.To detect RNA degradation, we grew cells and treated them with the appropriate materials as described above. We extracted RNA using RNAprotect bacterial reagent (Qiagen) and the RNeasy minikit (Qiagen) combined with the RNase-free DNase set (Qiagen) according to the manufacturer’s instructions. To determine RNA integrity, we analyzed RNA samples (200 to 400 ng) using an Agilent Technology 2100 Bioanalyzer.
Microarray analysis.For microarray analysis, cells were grown, incubated with antibiotics, and washed as described above. RNA extraction was carried out as described above. The RNA was subjected to cDNA synthesis, fragmentation, and end terminus biotin labeling according to the standard Affymetrix prokaryotic GeneChip expression protocol with the following modifications: cDNA synthesis was performed with SuperScript III reverse transcriptase (Molecular Probes, Invitrogen [Carlsbad, CA, USA]), and cDNA reaction mixtures were incubated at 42°C overnight. Finally, labeled cDNA samples were hybridized to Affymetrix GeneChip E. coli arrays. Hybridized arrays were stained and washed using the Affymetrix Fuidic Station 450. After staining, arrays were scanned with a GC3000 scanner. For CEL file generation, we used the Affymetrix GeneChip Command Console software (AGCC).
Statistics for microarray analysis.Robust multiarray average (RMA) normalization was performed on CEL files using the Partek Genomic Suite 6.5. Additional statistical analysis was carried out using the Spotfire software package (Somerville, MA, USA) and custom MatLab (R2010B) routines. Gene selection was done using the false discovery rate (FDR) correction procedure for adjusting genes with P values smaller than 0.05 and fold changes greater than 2.
Real-time PCR.For real-time PCR analysis, cells were grown, incubated with antibiotics, and washed as described above. RNA extraction was carried out as described above. cDNA was synthesized by using the ProtoScript Moloney murine leukemia virus (M-MuLV) first-strand cDNA synthesis kit (New England Biolabs) with random primers (kit component). idnT RNA transcript levels were used as an endogenous control for RNA levels in the samples. To amplify dinB cDNA, the following primers were used: forward primer GAGCGTAGTCAGGGGATTGA and reverse primer TCGCTTCACATTCAGACCAG. To amplify umuC cDNA, the following primers were used: forward primer GCTTTACGCAGACATGAGCA and reverse primer CGCAGAATGCCTCATCAATA. To amplify idnT cDNA, the following primers were used: forward primer GTGGGTTTTGTCCTGCTGTT and reverse primer ACAGAGAGCGCTGCTACCAT. Real-time analysis was than conducted using Fast SYBR Green master mix (Applied Biosystems) in a 7500 Fast real-time PCR system (Applied Biosystems).
Membrane depolarization assays using FACS analysis.Cells were grown, incubated with antibiotics, and washed as described above. Samples were diluted 1:100 in PBS (pH 7.2). To stain the cells, we added 1 µl DiBAC4 (1 mg/ml in ethanol) to 1 ml of the diluted cells and incubated them at room temperature for 15 min. To analyze the intensity of the DiBAC4 fluorescence, we used an LSRII FACS machine as we have described previously (10).
Detection of OH˙ formation.To detect OH˙, we used the molecule hydroxyphenyl fluorescein (HPF), which fluoresces only when it reacts with OH˙. We grew the cells, treated them, and washed them twice with PBS (pH 7.2) as described above. Then, we added HPF to the samples to a final concentration of 10 µM and incubated the cells, protected from light, at room temperature for 1 h, after which we washed them twice with PBS (pH 7.2). We analyzed them for staining intensity using an LSRII FACS machine with a 488-nm argon laser for excitation and a 530- ± 15-nm emission filter; we analyzed the results using FCS Express V3 software.
Quantifying LexA degradation by Western blot analysis.We grew cells in 10 ml M9 minimal medium, as described above, and treated them with various concentrations of NA for different periods of time as described in the figure legends. To lyse the cells, we centrifuged samples for 2 min and then suspended the cell pellets in 50 µl of Bugbuster master mix (Novagen), incubating them with vigorous shaking at room temperature for 10 min. Then, we centrifuged them at 4°C for 10 min and transferred the supernatants containing the crude extract to fresh tubes. We determined protein concentrations using the Bradford assay (Bio-Rad, Hercules, CA, USA). For Western blot analysis, we used rabbit polyclonal antibody to the LexA DNA binding region as the primary antibody (Abcam) and donkey polyclonal antibody to rabbit IgG (horseradish peroxidase [HRP]) as the secondary antibody (Abcam).
Measuring enzyme activities.We grew cells to an OD600 of 0.5, treated them with NA at the indicated concentrations, incubated them for the indicated times, and collected them by centrifugation. To lyse the cells, we suspended them in 50 mM morpholineethanesulfonic acid (MES) buffer (pH 6.0) with 10% glycerol and 1 mg/ml lysozyme and incubated them with vigorous shaking at room temperature for 5 min; we then sonicated them, for 10 s, twice at 30-s intervals. We froze aliquots of whole-cell extracts in liquid N2 medium stored at −80°C. We measured complex II activity as the succinate dehydrogenase (SDH) activity based on the succinate-mediated phenazine methosulfate reduction of dichloroindophenol at 600 nm as described previously (47). We measured complex I activity as deamino-NADH dehydrogenase activity by monitoring the reduction of deamino-NADH at 340 nm as described previously (47). We measured NADH dehydrogenase II activity by monitoring the general reduction of NADH at 340 nm and by reducing the activity of deamino-NADH dehydrogenase activity from the total NADH dehydrogenase activity.
Microarray data accession number.All CEL files are available from GEO under accession number GSE56251.
ACKNOWLEDGMENTS
This research was supported by the Israel Science Foundation (ISF) (grant 66/10) and the United States Army (grant W911NF-13-1-0371).
We thank F. R. Warshaw-Dadon (Jerusalem, Israel) for her critical reading of the manuscript.
FOOTNOTES
- Received 3 June 2014
- Accepted 16 June 2014
- Published 15 July 2014
- Copyright © 2014 Erental et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported license, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
