DNA, Cell Wall and General Oxidative Damage Underlie the Tellurite/Cefotaxime Synergistic Effect in Escherichia coli

The constant emergence of antibiotic multi-resistant pathogens is a concern worldwide. An alternative for bacterial treatment using nM concentrations of tellurite was recently proposed to boost antibiotic-toxicity and a synergistic effect of tellurite/cefotaxime (CTX) was described. In this work, the molecular mechanism underlying this phenomenon is proposed. Global changes of the transcriptional profile of Escherichia coli exposed to tellurite/CTX were determined by DNA microarrays. Induction of a number of stress regulators (as SoxS), genes related to oxidative damage and membrane transporters was observed. Accordingly, increased tellurite adsorption/uptake and oxidative injuries to proteins and DNA were determined in cells exposed to the mixture of toxicants, suggesting that the tellurite-mediated CTX-potentiating effect is dependent, at least in part, on oxidative stress. Thus, the synergistic tellurite-mediated CTX-potentiating effect depends on increased tellurite uptake/adsorption which results in damage to proteins, DNA and probably other macromolecules. Our findings represent a contribution to the current knowledge of bacterial physiology under antibiotic stress and can be of great interest in the development of new antibiotic-potentiating strategies.


Introduction
Multi-antibiotic resistance in bacteria has become a public health concern and also an important veterinary problem worldwide. Scientific and pharmaceutical efforts have been devoted to look for new antibiotics and to synthesize or modify existing ones. In spite of these efforts only one novel antibiotic has been introduced in the market during the last 50 years [1]. In addition, even if new antibiotics become available the emergence of resistant strains is only a matter of time.
A few alternative strategies to potentiate the antibacterial effect of known antibiotics have been proposed. Increasing antibiotic toxicity in a synergistic manner combining those with chemical compounds or metals having different cellular targets are among the most promising ones. This strategy significantly reduces the probability of emergence of strains that are resistant simultaneously to both antibacterials. This is the case of tomatidine, a natural compound that acts synergistically with aminoglycosides against multiresistant strains of Staphylococcus aureus [2]. In the same line, while bismuth thiols have been used to improve tobramycin effects [3], the combined administration of the complex desferrioxamine-gallium and gentamicin increased the toxicity against Pseudomonas aeruginosa [4].
In spite of recent contributions in this field, much remains to be explored regarding the molecular mechanisms underlying antibiotic toxicity and synergistic effects. The picture supporting a relevant role for oxidative stress in general toxicity of bactericidal antibiotics [5] has been recently challenged [6,7].
To the best of our knowledge, studies regarding the elucidation of molecular mechanisms involved in antibiotic potentiation have not been published to date. Recently, our group reported the successful use of non-lethal concentrations of the tellurium oxyanion, tellurite (TeO 3 22 ), to potentiate antibiotic toxicity. A tellurite dose-dependent effect on the toxicity of several antibiotics was observed in both clinical and laboratory strains. Particularly interesting was the synergistic effect observed with TeO 3 22 and the widely used third-generation cephalosporin, cefotaxime (CTX) [8].
At very low (nM) concentrations, tellurite acts as a strong toxicant for bacteria exerting its toxic effects essentially through the generation of oxidative stress that in turn oxidizes proteins, membranes, depletes the reduced-thiol pool and affects metabolic enzymes [9].
Although tellurite has been used as selective agent for decades in routine microbiological culture media, information about tellurite damage/toxicity in cells of superior eukaryotes is scarce. Only a couple of representative reports are available including murine hepatocarcinoma transplantable liver tumor (TLT) cells [10] and whole animal assays (rats) [11]. Thus, the combined approach tellurite/antibiotic seems particularly attractive for antibioticpotentiating strategies.
In this work we investigated the molecular mechanism(s) underlying the potentiating effect of TeO 3 2 on CTX in E. coli. Global transcriptional profiling using microarrays and cellular damage assessment indicated that in the presence of CTX increased uptake of tellurite takes place with the concomitant oxidative damage of proteins and DNA. These findings may represent the basis for new strategies to potentiate the antibacterial effect of current antibiotics as well as a source of putative targets for the development of new ones.
In order to identify the cell pathways that are affected in response to tellurite, CTX and/or tellurite/CTX exposure, induced and repressed genes in each condition were grouped using Gene Ontology (GO) terms. In all cases the three gene categories most represented in the subset of differentially expressed genes were stress responses, transport and biosynthetic processes ( Fig. S1 A, B, C).
Induced genes in tellurite-exposed cells belonged mainly to stress responses, biosynthetic and metabolic processes, transcription and RNA catabolism and protein metabolism categories. Most highly expressed genes in this condition included well recognized oxidative stress markers, confirming previous results. Even though the concentration of tellurite used in this study was ten-fold lower than that in previous experiments [12], exposure to the toxicant still resulted in soxS gene induction (GeneID: 948567, 4.7 fold-change), along with that of others such as marR and dnaK (GeneIDs:945825 and 944750, each with almost 3 fold-change), related to metal response and oxidative stress, respectively [13,14] (Table S2 A). Induction of genes related with protein folding and RNA degradation was also observed (Fig. S1 A, Table S2 A) suggesting that tellurite exposure results in oxidative damage to protein and RNA components.
Upon CTX-exposure most induced genes were related to transport, including yhdA, yaaH, ydgK, yjgP, nikC, ygeD, secG, smpA and ynfM that encode different conserved and predicted inner membrane proteins, nickel transporter subunit and other transporters ( Fig. S1 B, Table S2 B). In addition, induction of genes involved in DNA repair such as uvrA and pol [both participating with an oxidative damage-dependent DNA repair system [15] was also observed (Table S2 B). Interestingly, intZ (CPZ-55 prophage predicted integrase) and ycfK (e14 prophage predicted protein) were also induced, suggesting the activation or induction of a putative phage excision mechanism, as suggested previously [16]. The results presented above strongly suggest that CTX produces oxidative damage to DNA. This is most probably mediated by hydroxyl radicals, in a similar manner to that described previously for bactericidal antibiotics [5].
As with tellurite, simultaneous exposure of E. coli to tellurite/ CTX results in the induction of genes (soxS, marR and dnaK) that are associated with the response to oxidative stress [12,13,14], but also others related with transport and biosynthetic processes (Fig.  S1 C). Induction of genes related to protein folding and [Fe-S] clusters repair also supports the idea that part of tellurite/CTXmediated damage results from oxidative stress. To facilitate data analysis of tellurite-and CTX-regulated genes, Venn diagrams were constructed (Fig. S4). In general, array data were validated by qRT-PCR as described in Methods (Fig. S3).
Tellurite Concentration in Antibiotic-exposed E. coli Cultures Since the canonical target of CTX is peptidoglycan, a possible explanation for the observed synergistic effect is that the destabilization of the cell wall caused by CTX may result in increased intracellular tellurite concentrations. Then, the simultaneous presence of both toxicants might cause exacerbated oxidative stress. To test this hypothesis, the cell wall integrity was assessed indirectly in E. coli exposed or not to the antibiotic through quantifying remaining tellurite in the culture medium as described in Methods. A ,35% decrease of extracellular tellurite was determined in culture supernatants of cells exposed to lethal or non-lethal CTX concentrations ( Fig. 1), providing evidence for an increase in tellurite entrance/adsorption to the cell in these conditions. In this context, preliminary experiments using FACE (Fluorescence Assisted Carbohydrate Electrophoresis [17]) with tellurite/CTX-exposed E. coli suggest enhanced activity of murein hydrolases, indicative of alterations in the regulation of normal peptidoglycan turnover (not shown).

ROS Generation in Toxicant-exposed Cultures
Given that i) bactericidal antibiotics apparently generate hydroxyl radicals [5], ii) tellurite generates superoxide [12] and iii) the expression profile of tellurite/CTX cells suggest the occurrence of an exacerbated oxidative stress status, the generation of ROS in toxicant-exposed cells was evaluated. Preliminary experiments using the non-specific ROS probe H 2 DCFDA showed a slight increase of total ROS in cells exposed to sublethal CTX concentrations in the presence of tellurite (not shown).
Superoxide generation in cells exposed simultaneously to tellurite and different CTX concentrations was determined by flow cytometry using the superoxide-specific probe dihydroethidine. In turn, the OH N -specific probe hydroxyphenyl fluorescein (HPF) was used to assess OH N generation in treated cells.
As shown in Fig. S2, in the presence of both toxicants superoxide levels are equivalent to those observed in cells exposed to tellurite alone (compare blue and green lines), demonstrating that superoxide anion is formed at short exposure times. Also, a significant increase in OH N levels was observed in E. coli after 3 h exposure to tellurite/CTX (Fig. 2) mimicking the effects of exposure to CTX alone. These results demonstrate that tellurite contribution to the tellurite/CTX combined effect does not proceed through OH N generation, but rather through superoxide. The observed effects are independent of the CTX concentration ( Fig. 2 B). Conversely, CTX contribution to the tellurite/CTX effect seems to proceed through OH N generation in a dose dependent manner (Fig. 2 B).

ROS-induced Damage to Cell Macromolecules
Since ROS generated as consequence of tellurite/CTX exposure can randomly affect cell macromolecules, damage to both proteins and DNA was evaluated. Protein damage was assessed by carbonyl group formation. Enhanced levels of oxidized proteins were observed in cells exposed to tellurite/CTX but not in cells exposed to either toxicant alone (Fig. 3A). In turn, DNA damage was assessed by real time PCR. In order to cover a representative portion of the bacterial chromosome, 5 randomlyselected genes were amplified. A representative experiment using the dctA gene is shown in Fig. 3B. Higher C p values were observed in CTX-and tellurite/CTX-treated cells as compared to controls, indicating a direct OH N -mediated injury to the DNA molecule (Fig. 3B).

Discussion
The continuous emergence of antibiotic-resistant bacteria which affects not only public health but also veterinary and food industries has become an important scientific concern during the last years. Aiming to explore new alternative antibacterial treatments, combination of different antibiotics [18] or the administration of defined molecules acting as adjuvants have been proposed [19,20]. In this context, the antibiotic-potentiating effect of potassium tellurite at concentrations that are not toxic to eukaryotic cells has been reported [8]. To elucidate the molecular mechanism(s) involved in this phenomenon, global transcriptomic changes in E. coli exposed to tellurite, CTX or TeO 3 22/ CTX were evaluated.

Tellurite-induced Changes in Gene Expression
One of the most obvious results of gene expression profile analysis upon toxicant treatment was the induction of stressresponsive genes, many of which play a pivotal role in the oxidative stress response. For instance, induction of soxS and fumA was observed (Table S1), which agrees with previous results from our laboratory [21]. Since FumA, the gene product of the fumA gene, is an oxidative stress-sensitive relevant TCA cycle enzyme [21,22,23], its induction suggests that compensatory effects to restore ROS-damaged enzymes are in place. In addition, induction of dnaK, encoding the DnaK chaperone that participates in different stress response pathways including heat shock [14] and antioxidant response against H 2 O 2 [24], yfhQ, which has been directly related to DNA repair under ROS-mediated stress in Bacillus subtilis [25] and marR, a transcriptional regulator that has been associated with oxidative stress and response to antibiotics was also observed [13]. The induction of these -and-other genes may be related to the tellurite-induced superoxide stress.
On the other hand, the analysis of expression data showed that most highly expressed and repressed genes were related to diverse biosynthetic processes (Fig. S1). One example is the repression of dcuS, glpB and narG genes, all related to anaerobiosis in E. coli, which suggests a preference for aerobic metabolism in the presence of tellurite. In turn, upregulation of acnA, fumC and zwf, encoding aconitase A, fumarase C and glucose 6 phosphate dehydrogenase, respectively (a consequence of soxS induction), also supports a preference for aerobic metabolism in tellurite-exposed cultures.
Repression of rseB, encoding a negative regulator of rpoE (sE), was also observed. Since induction of rpoE is consequence of the erroneous folding of periplasmic proteins [26], the observed rseB repression suggests that correct folding of these proteins may be accomplished by tightly regulating rpoE expression, as previously reported [27].
Changes in the global transcription profile of E. coli exposed to 0.5 mg/ml tellurite were also assessed by DNA microarray analysis. Surprisingly induction of soxS, gmk, marR and cspA was still observed at this tellurite concentration (10-fold higher than those used in most experiments shown in this work). The E. coli transcriptional response to tellurite-mediated stress seems to be dependent on the oxyanion concentration, as a dose-dependence of the log-fold change value was observed for the referred genes (Table S3).
Despite cells being exposed to a sublethal tellurite concentration (0.05 mg/ml), resulting transcriptional changes suggest that pleiotropic damage occurs and affects mainly the central metabolism, most probably via the generation of oxidative stress.
Some work on transcriptional profiling in tellurite-exposed E. coli has been recently published [28]. Comparable changes to those reported here were observed by these authors regarding carbohydrate and amino acid metabolism. However, lack of other similarities is most probably due to the different experimental conditions used. For instance, E. coli EC3 instead the BW25113 strain was their model organism and tellurite concentrations were 100-fold higher [28] than those used here.

CTX-induced Changes in Gene Expression
Global gene expression analysis in E. coli suggest that CTX exposure results in the induction of genes involved in the stringent response, oxidative stress-mediated DNA damage repair and in de novo synthesis of membrane proteins. CTX exposure resulted in the induction of spoT, ygjM and guaC genes which are related to the stringent response. This is a condition of amino acid starvation, also triggered by other kinds of stresses that result in the inhibition of several metabolic processes, thus allowing bacterial survival. It is characterized by a rapid accumulation of guanosine 39-59bispyrophosphate (ppGpp), a nucleotide that regulates positively or negatively the expression of different gene sets [29,30,31].
Expression of several genes involved in ppGpp accumulation, s E -mediated envelope stress, heat shock, osmotic and oxidative damage have been shown to be induced upon exposure to cell wall-acting antibiotics in Streptomyces coelicolor [32]. These results are in agreement with our findings in differential expression (Table S2; Fig. S1B). Moreover, induction of uvrA (nuclease A) and polA (DNA polymerase I) genes, involved in the nucleotide excision DNA repair mechanism (UvrABC) [33], was observed in CTX-treated cells. This result is supported by the observation of Kohanski and coworkers regarding hydroxyl radical generation by bactericidal antibiotics [5].
Notably, most induced genes were grouped in the transport functional category, suggesting that the antibiotic-induced damage to the cell wall causes not only its destabilization but could also result in membrane protein misfolding. A similar envelope effect was reported for thioridazine-treated Mycobacterium tuberculosis, where induction of membrane protein genes occurred simultaneously with cell wall damage [34].

Changes of Gene Expression in the Presence of Tellurite/ CTX
As with tellurite alone, soxS, marR, dnaK and yfhQ genes (all related to the oxidative stress response) were also found to be induced in tellurite/CTX-exposed cells. This result was not unexpected since tellurite exposure results in the generation of an oxidative stress status in E. coli [9,12]. In addition, induction of clpB (encoding the caseinolytic peptidase B) and dnaK (encoding the DnaK chaperone) also occurred in the presence of both toxicants, again suggesting that a certain degree of protein misfolding is taking place in this condition. DnaK, along with IbpA and IbpB proteins, forms a triad that is responsible for solubilizing thermally-denatured protein aggregates in vivo and in vitro [35]. On the other hand, the induction of the yaeL gene, encoding the extracytoplasmic stress-responsive s E -activating proteinase [36], suggests that these cells are also facing periplasmic stress. This would be a consequence of CTX-mediated damage to the cell wall and/or oxidative damage to membrane and Figure 2. Hydroxyl radical generation in E. coli exposed to CTX or Te/CTX. E. coli was exposed for 3 h to tellurite/CTX (A) and to different antibiotic concentrations (B). As positive control, cells were exposed for 3 h to ampicillin. Units are expressed in mg ml 21 . The data correspond to a representative result of 3 independent trials. doi:10.1371/journal.pone.0079499.g002 membrane proteins. Functional assignment analysis of genes expressed in tellurite/CTX treated cells clearly suggests that pleiotropic damage provoked by exacerbated oxidative stress is occurring in these conditions.
As shown in Tables S1-S4 the number of tellurite-, CTX-and tellurite/CTX-regulated genes is low (especially evident in Fig.  S4). Our interpretation is that this observation is due most probably to the strict cut criteria used to interpret the microarray analysis data.

Identification of Tellurite/CTX Cell Targets
As in the case of other cephalosporin antibiotics, CTXmediated damage to the cell wall is well known and in this scene, increased uptake/adsorption of tellurite may be expected. Indirect evaluation of tellurite uptake in tellurite/CTX-exposed cells -using a previously reported strategy [37]-proved this to be the case although this effect did not depend on CTX concentration (Fig. 1).
Higher levels of intracellular tellurite in the presence of CTX are thus likely to exacerbate the production of ROS and induced ROS-mediated damage to macromolecules. ROS were produced in both tellurite-and tellurite/CTX-treated cells as assessed through the use of the nonspecific H 2 DCFDA probe. Based on these results, the identity of the specific oxygen radical involved in the tellurite/CTX synergy phenomenon was determined. Using DHE, superoxide was detected even at tellurite concentrations that were 10-fold lower than those used previously [12]. Interestingly, superoxide is not generated in CTX-exposed cells, suggesting that only tellurite is responsible for early ROS generation in tellurite/CTX-exposed E. coli. On the other hand, hydroxyl radical production was investigated in tellurite/CTXtreated cultures. After 3 h of exposure to the toxicants, OH N levels were similar in CTX-and tellurite/CTX-challenged cells (Fig. 2A). Also, they did not change significantly when cells were exposed to different antibiotic concentrations, although a slight displacement of the fluorescence peak could be appreciated (Fig. 2B). Thus, tellurite and CTX seemed to be the main responsible for superoxide and hydroxyl radical, respectively. Surprisingly, fluorescence in CTX-or tellurite/CTX-exposed cells was higher than that observed with ampicillin-treated cells (positive control). This should not be a consequence of the total number of antibiotic molecules since ampicillin concentration was 50-fold higher than that of CTX. Although these 2 bactericidal antibiotics share the same cell target, these results suggest that both have different OH Ngenerating mechanisms.
To evaluate superoxide-and OH N -induced cell injury, direct damage to proteins and DNA was then assessed. Increased levels of protein carbonylation were observed exclusively in tellurite/ CTX-exposed cells (Fig. 3A), showing higher protein oxidation than that observed with tellurite or CTX alone. In turn, higher qPCR crossing points (Cp values) were obtained in CTX-and tellurite/CTX-exposed cultures, indicating that oxidative injury to the cell's DNA occurred (Fig. 3B).
Finally and although the OH?-detecting methodology was challenged recently [38], our trancriptomics and DNA damage data support the notion that hydroxyl radicals are being produced upon tellurite/CTX exposure.

Concluding Remarks
Results from this work show that the tellurite/CTX-challenge alters the global gene expression profile and produces cellular damage. Increased levels of intracellular superoxide and OH N produced by tellurite and CTX, respectively, generate direct damage to DNA (and probably RNA) and to the protein pool, which may include a number of membrane transporters. This would favor further entry of tellurite and CTX, and sustain oxidative stress in time. All these effects would explain, at least in part, the synergistic CTX/tellurite phenomenon in E. coli. Intracellular ROS also trigger the expression of oxidative and envelope stress-related genes which result in the synthesis of DNA repair enzymes and membrane proteins (Fig. 4). Experiments to shed further light to the synergistic action of CTX and tellurite in E. coli are in progress in our laboratory.

DNA Microarray Experiments
Total RNA was extracted from E. coli cultures exposed for 15 min to tellurite, CTX or tellurite/CTX using RNeasy Mini kit (Qiagen), as recommended by the manufacturer. RNA concentration and purity was determined using a Nanodrop 2000c spectrophotometer (Thermo).
Labeled cDNA probes were generated by reverse transcription using 20 mg of total RNA, SuperScript II (Invitrogen) and Alexa 555 and 647 dyes (Invitrogen) by conventional methodologies. Dye incorporation into cDNA was measured spectrophotometrically and generated probes were used to hybridize DNA microarrays slides purchased to Microarrays Inc as described earlier [40], with minor modifications. Slides contained 9,308 oligonucleotides representing four E. coli varieties and three plasmids associated to this bacterium. In particular, printed oligonucleotides corresponded to 4,269, 5,306, 5,251 and 5,366 open reading frames from E. coli K12, 0157:H7 (EDL933), 0157:H7 (Sakai) and CFT073 strains, respectively. In addition, 3, 10 and 97 genes from plasmids OSAK1, pO157_Sakai and pO157_EDL933, respectively, were also present in the slides. These were scanned in a ScanArray G X (Perkin Elmer) scanner and image analysis was conducted using GenePix Pro v6.0 software. To discount the background signal we utilized the Bioconductor software and values were normalized using the LOESS method. T-test was used to identify those genes whose expression changed significantly upon treatment. Finally, 3 criteria to identify genes with differential expression (M value, A value and p value from t-test) were used. Only those genes that exhibited values of M $1 (induction), M # 21 (repression); A $8 and p#0.05 were considered [41]. Microarray data were deposited as GSE39696 series in GEO (Gene Expression Omnibus) at the NCBI's database.
Microarray data was validated by qRT-PCR using fadD as the housekeeping gene (C p values did not change in all conditions tested). Specific primers used are indicated in Table S1. A 300 bp DNA fragment of each gene was amplified using Kapa SybrH Fast kit (KapaBiosistems) in a Rotorgene Q (Qiagen) thermocycler. Data generated was analyzed using the relative expression software tool (REST) [42].

Protein Carbonylation
Protein oxidation levels were assessed by determining spectrophotometrically the content of protein carbonyl groups as described previously [44].

DNA Damage Determination
Total DNA was extracted from cells exposed to tellurite, CTX and TeO 3 22/ CTX during 3 h by conventional phenol:chlorophorm extraction methodologies [45] and 10 mg of the obtained material was used as template for real time PCR reactions using Kapa SyberH Fast kit (KapaBiosistems) as reported [46].  Figure S3 Validation of microarray data. Gene induction in E. coli exposed to tellurite (A), CTX (B) or tellurite/CTX (C) as determined by qRT-PCR. Relative expression values were determined using the REST software [42]. Statistical significance was assessed using the ttest. *p,0.05, **p,0.01, ***p,0.001. (TIF) Figure S4 Differential gene expression in E. coli exposed to the indicated compounds. Venn diagrams showing the number of induced (A) and repressed (B) genes that coincide between the indicated experimental conditions. CTX, cefotaxime. (TIF)