Induction of the soxRS Regulon of Escherichia coli by Superoxide*

The soxRS regulon orchestrates a multifaceted defense against oxidative stress, by inducing the transcription of ∼15 genes. The induction of this regulon by redox agents, known to mediate O·̄2 production, led to the view that O·̄2 is one signal to which it responds. However, redox cycling agents deplete cellular reductants while producing O·̄2, and one may question whether the regulon responds to the depletion of some cytoplasmic reductant or to O·̄2, or both. We demonstrate that raising [O·̄2] by mutational deletion of superoxide dismutases and/or by addition of paraquat, both under aerobic conditions, causes induction of a member of the soxRS regulon and that a mutational defect in soxRS eliminates that induction. This establishes that O·̄2, directly or indirectly, can cause induction of this defensive regulon.


rectly or indirectly, can cause induction of this defensive regulon.
The soxRS (superoxide response) regulon positively controls ϳ15 genes in Escherichia coli. The inductions of this regulon by redox cycling agents, such as paraquat, plumbagin, and phenazine methosulfate, which are capable of mediating O 2 .1 production, led to the view that this regulon is capable of responding to O 2 . (1)(2)(3). This conclusion was strengthened by the observation that H 2 O 2 , heat shock, or ionizing irradiation did not induce the soxRS regulon (1)(2)(3)(4). Moreover FumC was induced by paraquat more strongly in a sodA sodB strain than in its SOD-replete parent (5), and its induction in the parental strain was eliminated by mutational deletion of the soxRS response (6), thus indicating that O 2 . could cause induction of soxRS. However it was also noted (6) that marked overproduction of SodA did not diminish induction of members of this regulon such as fumarase C and glucose-6-phosphate dehydrogenase, an indication that O 2 . -independent induction was also a reality. In accord with this view was the finding that NADPH could diminish in vitro transcription/translation of the sodA gene (7). Nitric oxide has also been shown to induce soxRS and to do so in the absence of dioxygen (8).
There is strong evidence that the SoxR protein, which is the sensor of the soxRS regulon (4,9,10), occurs in oxidized and reduced forms and that the oxidized form is the activator of soxS transcription (11)(12)(13)(14). The balance between the oxidized and reduced forms of SoxR within E. coli can undoubtedly be influenced in multiple ways. For example either by oxidation of

MATERIALS AND METHODS
Paraquat was obtained from Sigma and malate from ICN. Bactotryptone, casamino acids, and yeast extract were from Difco. The strains of E. coli used were as follows: GC4468 ϭ parent (16); DJ 901 ϭ GC4468 ⌬ (soxR-Zjc2204) Zjc2205::Tn10 K m (provided by B. Demple) (2); QC1799 ϭ GC4468 ⌬ sodA3, ⌬ sodB-kan (16); and QC1817 ϭ GC4468 ⌬ sodA3 , ⌬ sodB-kan, ⌬ sox8::cat (obtained by transduction of the ⌬ sox 8::cat mutation into QC1799). (The soxRS deletion was provided by B. Weiss (3).) Strains were grown overnight at 37°C, with shaking in air, in LB, or in M9CA media containing 50 g/ml kanamycin and/or 30 g/ml chloramphenicol where required. These cultures were diluted as described in the figure legends into media not containing antibiotics, and paraquat was added after 1 h, and incubation was continued for 75 min. Cells were then harvested, washed in 50 mM potassium phosphate, 0.1 mM EDTA at pH 7.8, and then resuspended in this buffer and lysed in a French press. The extracts were clarified by centrifugation, and protein (17) and fumarase C (5, 6) were assayed. One unit of fumarase was taken to be the activity that converted 1 mol/min of L-malate to fumarate using ⌬E 250 nm ϭ 1.62 mM Ϫ1 cm Ϫ1 . The initial concentration of L-malate was 50 mM, and the assay buffer was 50 mM sodium phosphate, pH 7.3, at 25°C. . and on the redox status of the cell will be greater in an sodA sodB mutant than in its SOD-replete parent. Therefore, we should expect that paraquat should induce a member of the soxRS regulon such as FumC more strongly in an sodA sodB strain than in the parental strain. Bars 1, 2, and 3 in Fig. 1  The induction of FumC caused by deletion of SodA and SodB was greater in cells that had been grown in M9CA rather than in the richer LB medium. This is made apparent by comparison of bars 1 and 2 in Fig. 2 with bars 1 and 4 in Fig. 1. Thus there was a ϳ3-fold induction, caused by the deletion of SOD activity, in the LB-grown cells and a 7-fold induction in the M9CA-grown cells. Bar 3 in Fig. 2 shows that soxRS was as essential for the induction of FumC in the M9CA-grown cells as it was in the LB-grown cells. The experiment shown in Fig. 2 was repeated under dioxygen-depleted conditions. This was done by placing 0.2% inocula, in fresh M9CA medium in a BBL gas pack jar, which was then incubated for 5.5 h before the cells were harvested and extracts prepared for FumC assay. The gas pack jars were not evacuated before incubation so hypoxic, rather than anoxic, conditions prevailed. The sodA sodB extracts were found to have 0.026 units/mg protein of FumC activity, whereas the parental extracts had 0.014 units/mg. Thus dioxygen depletion diminished the ratio of FumC in the sodA sodB extracts from ϳ7-fold to ϳ2-fold, as compared with the parental extracts.

DISCUSSION
Because the induction of FumC, whether by addition of paraquat or by deletion of SodA and SodB, was dependent on soxRS and dioxygen, it follows that O 2 . can induce soxRS. The induction of the soxRS regulon depends upon the oxidation of the reduced form of SoxR, because the oxidized SoxR is the transcriptional activator of soxS. There must be a pathway for the reduction of oxidized SoxR, and the steady state will depend on the balance between the rates of oxidation of reduced . ] in the sodA sodB strain is ϳ20-fold higher than in the parental strain (22), and this caused an ϳ3-fold induction of FumC. Gort and Imlay (15), by using a strain in which the level of SOD could be modulated, reported that a 10-fold diminution of [SOD] was a threshold for induction of FumC and resulted in  . (V f ) must be equal to the sum of its rates of consumption by SOD (V SOD ) and by all other targets and Application of Eq. 4 would require several difficult measurements and/or estimations so another approach is useful, from Eq. 1, as follows. and When V SOD ϭ V T , one-half of all the O 2 . flux is being scavenged by SOD and in analogy to the classical assay for SOD activity (23) in which SOD competes with cytochrome c for the flux of O 2 . , we can define this amount of SOD activity as 1 biological unit. We have previously found that wild type E. coli contains 19 biological units of SOD on the basis of its inhibition of lucigenin luminescence (22). Hence in these cells V T ϭ 0.05 V f , whereas in sodA sodB cells V T ϭ V f . Fig. 3 presents (100) V T /V f as a function of the number of biological units, which is the ratio V SOD /V T . This plot ignores changes in biological units because of enzyme inductions and changes in V T because of consumption of targets. A  . will be at ϳ7 ϫ 10 Ϫ10 M. Variation of these numbers will, of course, occur as growth conditions change.
Thus the ratio of V SOD /V T appeared to be approximately 40/1 when the cells were suspended in 0.25% glucose but was much less when they were suspended in LB or in succinate (22); we therefore used 19/1 as an average approximation. Several papers (15,24,25) allow estimation that V SOD /V T lies in the range 10 -20, in agreement with our present estimate.