Bacterial defenses against oxidative stress

~EVIEWS

Reactive oxygen species, collectively referred to as oxidative stress, have been implicated in several degenerative diseases such as cancer, aging and arthritis. The oxidants, including the superoxide anien (Oz.-), hydrogen peroxide (HzOz), and thc hydroxyl radical (HO'), can be generated by the incomplete reduction of oxygen to water during respiration, by exposure to radiation, light, metals or oxidation-reduction (redox) active drugs such as paraquat, plumbagin and menadione, or by release from stimulated macrophages (Table 1). While the chemical reactions involving oxidams and lipid peroxidation in mammalian cells have been studied in great detail, little is known about eukaryotic gene expression in response to oxidation. In contrast, the response to oxidative stress has been well characterized in bacteria. The following review summarizes what has been learned about oxidative stress from genetic studies of Escherichia coli and Salmonella typhimurium. When might bacteria encounter oxidative stress? Although enteric bacteria are most often found in the intestinal tracts of mammals where they live under predominantly anaerobic conditions, bacteria have also been isolated from sewage and soil. In these latter environments the bacteria depend on aerobic respiration and consequently must be able to defend against oxidants. Even within host organisms, if bacteria move outside the intestinal tract, they encounter an oxidative burst of superoxide radicals and hydrogen peroxide released by macrophages when the macrophages encounter bacteria. Both E. coil and S. typhimurium can defend against and adapt to oxidative stress that occurs during aerobic growth or after an oxidative burst. If these bacteria are pretreated with low doses of hydrogen peroxide, they adapt to subsequent lethal doses 1.2. Similarly, E. coil cells pretreated with low doses of superoxidegenerating compounds become resistant to subsequent higher doses of the compounds3. These adaptive responses require the induction of protein synthesis since cells treated with low doses of the oxidants in the presence of protein synthesis inhibitors are no longer resistant to the subsequent lethal challenges. Many of the proteins and the corresponding genes that are required for the bacterial defense against oxidative damage have been identified (listed in Table 2). Some of the activities can eliminate oxidants directly. Two forms of superoxide dismutase, which catalyse the conversion of superoxide anion to hydrogen peroxide and water, have been identified in E. coli and are encoded by the sodA and sodB genes 4. E. coil also

Bacterial defenses against oxidative stress G~EI.A STORZ, LOUISA. TARTAGLIA,SPENCERB. FARR AND BRUCEN. AMES Bacteria treated with low doses of oxidants such as hydrogen peroxide adapt to subsequent high doses of these oxidants by inducing the expression of numerous genes. The study of these genes and the roles they play in defending bacteria against oxidative damage has given general insights into what oxidants are hazardous to cells, what cell constituents are damaged by oxidants, and how cells sense and respond to oxidative stress. possesses two forms of catalase that break down hydrogen peroxide to oxygen and waterS. The catalase consisting of 80 kDa subunits (HPI) is encoded by katG, and the second catalase, which has a higher molecular mass (HPII), is encoded by katE. While bacteria do not appear to have the glutathione peroxidase activity present in many mammalian cells, both S. typbimurium and E. cob possess a glutathioneindependent alkyl hydroperoxide reductase activity capable of reducing hydroperoxides such as thymine hydroperoxide and linoleic hydroperoxide to their corresponding alcohols 6. This activity is made up of two components encoded by abpC and abpF. Bacteria do possess glutathione reductase (g,orA), which may play a role in protecting against oxidation by maintaining a pool of reduced glutathione, which in turn can serve to maintain the reduced state of cellular proteins. In addition to the defense enzymes that destroy oxidants, several activities appear to repair damage caused by oxidants. Exonuclease III, the product of the xtbA gene, is the major apurinic/apyrimidinic endonuclease (AP endonuclease) activity found in E. coil, and has been shown to remove replication blocks from the 3' termini of oxidized DNA in vitro. A second AP endonuclease capable of removing 3' replication blocks is endonuclease IV (nfo) 7. The roles of the two AP endonucleases in protecting against oxidative damage were illuminated by the findings that strains with mutations in the genes encoding the activities are sensitive to hydrogen peroxide and other oxidants. Several other DNA repair activities are likely to be important for a defense against oxidative stress, as inferred partly from the hydrogen peroxide sensitivity of mutant strains: endonudease III (ntb), DNA

Tsm~ 1. Reactions involving reaclx~e oxygen species (1) 02 -¢ 02.- ---> H202 --¢ HO" + H20 --> H 2 0 (2) paraquat ÷ enzymatic reduction --¢ reduced paraquat reduced paraquat ÷ 02 --¢ paraquat ÷ 02--

(Stepwise one-electron reduction of 02)

(3) O2"-+ 2H2 --'>H202 + 02

(Reaction catalysed by superoxide dismutase)

(4) 2H202 ~ 02 + 2H20 (5) Fe2+ + H202 + H+ ~ Fe3+ + HO" + H20

(Reaction catalysed by catalase)

(Reaction caused by redox active drugs)

(Fenton reaction)

TIG NOVEMBER1990 VOL.6 NO. 11 © 1 9 9 0 Elsevier S c i e n c e P u b l i s h e r s Ltd (UK) 0168 - 9 4 7 9 / 9 0 / $ 0 2 . 0 0

1

~k~EVIEWS polymerase I (polA), and the excision nuclease and exonuclease activities associated with uvrAB and recBC, respectively. The sensitivity or resistance of other mutant strains has also identified genes encoding regulators that help to coordinate the defenses against oxidative damage. Strains with mutations in recA, a regulator of the SOS response (a response induced by several types of DNA damage) are sensitive to hydrogen peroxideS. Mutations that confer increased resistance to menadione have defined a locus, soxR, that regulates a subset of genes induced by superoxide9,i0, and mutations that confer increased resistance to hydrogen peroxide have defined a locus, oxyR, that regulates the expression of a number of genes induced by hydrogen peroxide a. The katF gene, encoding a regulator of katE (HPII catalase) expression, was first identified by mutations that eliminated catalase expressionS. The study of strains with mutations in the genes encoding these defense, DNA repair and regulatory activities, as well as some strains with defects in general metabolism, has given general insights into what constitutes oxidative stress and how cells can defend against it.

ReaOdve o.xy~n species that constitute oxidative stress Superoxide anion (02.-) Although superoxide anion, the consequence of a single electron reduction of oxygen, is only marginally reactive in an aqueous environment, the existence of enzymatic activities that can eliminate the superoxide anion argues that this oxidant is deleterious to the cell.

Studies by Carlioz and Touati on the properties of strains carrying mutations in the genes (sodA, sodB) encoding the superoxide dismutases, further emphasize that superoxide anion is harmful to cells and point to cellular components that are especially susceptible to superoxide damage 4. Carlioz and Touati found that while neither of the single mutations affected growth, sodA sodB double mutants were unable to grow on minimal medium under aerobic conditions. The double mutant strains were viable only if the minimal medium was supplemented with amino acids, in particular the branched chain amino acids leucine, isoleucine and valine. These results demonstrate that superoxide dismutases are essential under aerobic, nutrient-limiting conditions. The observations also suggest that the synthesis of amino acids, especially branched chain amino acids, is especially sensitive to damage by superoxide radicals.

Hydrogen peroxide (H202) Hydrogen peroxide is formed upon the detoxification of superoxide by superoxide dismutases or by the net two-electron reduction of oxygen. Again, since hydrogen peroxide is fairly stable, its contribution to oxidative stress was unclear. Studies by Imlay ei al. on the sensitivity of numerous E. coil strains to killing by different concentrations of hydrogen peroxide have helped to elucidate what damage is caused by hydrogen peroxideSA 1. Imlay and Linn found that a wildtype E. coli strain was sensitive to both low doses of hydrogen peroxide (1-3 ram, 'mode 1') and high doses of hydrogen peroxide (>20 mm, 'mode 2'), but less

T~LE 2. Oxidative defense genes in E. coU Activity Defense enzymes Superoxide dismutase Catalase Alkyl hydroperoxide reductase Glutathione reductase DNA repair enzymes Apurinic/apyrimidinic endonuclease

Gene Manganese Iron HPI HPII C22, F52 "

sodA sodB katG kate ahpC, ahpF gorA

Exonuclease lII Endonuclease IV

xthA

Map position a 88 36 89.2 38 13 77-78

nth polA uvrA uvrB recB recC

38 47 36 87 92 18 61 61

Metabolic enzymes NADH dehydrogenase Glucose-6-phosphate dehydrogenase

ndh zwf

22 41

Regulators SoxR OxyR KatF RecA

soxR oxyR katF recA

92 89.6 59 58

Endonuclease III DNA polymerase Excision nuclease Exonuclease

~¢o

aTaken from Ref. 36 and references contained in this review. T1GNOVEMBER1990 VOL.6 NO. 11

EVIEWS sensitive to intermediate concentrations. Imlay et al. characterized mode 1 killing by studying the sensitivity of different mutant strains to low concentrations of hydrogen peroxide. Strains compromised in their ability to repair DNA damage - xth, recA, polA and nfo xth double mutants - were all very sensitive to killing by low doses of hydrogen peroxide. Second, ndh mutants which lack NADH dehydrogenase (and therefore have abnormally high levels of reducing equivalents) were especially sensitive to oxidation, while starved cells (with low levels of reducing equivalents) were resistant to hydrogen peroxide. These observations, together with studies of an in vitro Fenton reaction (a reaction of hydrogen peroxide with iron, Table 1) capable of generating DNA strand breaks at the same concentrations of hydrogen peroxide causing mode 1 killing, showed that hydrogen peroxide constitutes a significant stress in vivo by generating DNAdamaging radicals in the presence of metals and reducing equivalents. Hydroxyl radical (HO') Hydroxyl radicals or ferryl radicals (hydroxyl radicals complexed with iron), which are generated by the Fenton reaction described above, are extremely reactive and probably react with and damage a nearby molecule as soon as the radicals are generated. Because the radicals are short-lived, they are difficult to study in vivo. However, studies of in vitro reactions, such as the experiments by Imlay et al.8.11 described above, suggest that much of the toxicity of superoxide anion and hydrogen peroxide is due to the conversion of these oxidants to the extremely reactive HO. radicals.

Damage caused by oxidative stress DNA damage The finding that strains with mutations in DNA repair enzymes are hypersensitive to oxidants argues that oxidative stress contributes to DNA damage. Evidence that mutations are generated as a consequence of oxidative DNA damage during aerobic growth has come from studies on the frequencies of spontaneous mutagenesis in bacterial strains that are compromised in their ability to detoxify oxidants. As shown in Table 3, strains carrying mutations in both sodA and sodB have significantly increased frequencies of mutagenesis when assayed for forward mutation to thymine auxotrophy lz. This high frequency is not seen in strains lacking exonuclease III (xtbA) activity, is seen only under aerobic conditions and is increased further by exposure to increased oxygen or treatment with plumbagin. Strains carrying deletions of oxyR, a positive regulator of catalase and alkyl hydroperoxide reductase expression, also have significantly increased frequencies of mutagenesis13, vt. Again, this increased mutation frequency is most pronounced under aerobic conditions. When the mutations generated in the oxyR~ and the oxyR deletion strains were characterized, it was found that the type of mutation whose frequency was most elevated was T.A to A'T transversions, the type of mutation most frequently caused by oxidative mutagens. The high frequency of mutagenesis in oxyR

TAnLE3. Increased frequencies of spontaneous mutagenesis in E. coli and & t y p h i m u r i u m strains with mutations in oxidative defense genes Strain

Number of mutants/platea,b

E. coli sodA+sodB~ sodA- sodBxthsodA- sodB- xtbsodA- sodB- xth-l xth+ S. typhimurium oxyR~ oxyRa2 oxyR+/pKM101c oxyRA2/pKM101

23 120 10 33 280 6 76 57 3102

aThe frequency of mutagenesis in the sodA, sodB and xth mutant strains was assayed by the forward mutation of Thy+ (trimethoprim-sensitive) to Thy- (trimethoprimresistant)12. bThe frequency of mutagenesis in the oxyR mutant strains was assayed by the reversion of His- auxotrophy to His+ prototrophy 13. cpKM101 carries mucA and mucB (analogs of the E. coli umuC and umuD genes), which make strains more susceptible to mutagenesis by a number of mutagens. deletion strains was suppressed by multicopy plasmids expressing high levels of catalase (katG), alkyl hydroperoxide reductase (abpCt9 or superoxide di~mutase (sodA) activityl3 and by suppressor mutatior~s that caused increased expression of katG and abpCF genes TM. These observations provide evidence that activities capable of eliminating oxidants play an important role in protecting against oxidative DNA damage during aerobic growth, damage that is otherwise converted into mutations. Membrane damage While the chemistry of lipid peroxidation is well established, there are few demonstrations that oxidative membrane damage leads to altered membrane function. One example has come from studies on the uptake of labelled metabolites by E. coil strains after treatment with hydrogen peroxide l~. Cells treated with 5 mm hydrogen peroxide showed a rapid loss of both proton motive force-dependent and -independent transport within 5 minutes after treatment with hydrogen peroxide. However, if the cells were first pretreated with lower doses (35 p i ) of hydrogen peroxide, transport recovered rapidly. This adaptation was not seen in strains with mutations in oxyR or katG, suggesting that increased expression of hydrogen peroxidescavenging activities is required to protect cells from membrane damage by oxidative stress. Protein damage The significance of protein damage due to oxidative stress in vivo has not yet been established; however, several in vitro studies have shown that oxidized bacterial proteins are sensitive to degradation. In

TIG NOVEMBER1990 VOL.6 NO. 11

Ik~EVIEWS particular, the ribosomal subunit L12, glutamine synthase and dihydroxyacid dehydrase are degraded more rapidly upon oxidation in vitro m-is, The instability of oxidized dihydroxyacid cehydrase, an enzyme required for the synthesis ¢~f branched chain amino acids, may explain why sodA sodB double mutants require branched chain amino acids for growth on minimal medium. Regulation of genes induced by oxidative stress Superoxide-inducible genes Two-dimensional protein gels of E. coli cells treated with paraquat, plumbagin or menadione have shown that treatment with these superoxide-generating compounds induces the synthesis of a distinct group of proteinsl9, z0. Several of the corresponding superoxideinducible genes have been identified9,m,al-z3. Touati found that the expression of sodA-lacZ fusion constructs was induced by oxTgen and by redox cycling compounds under aerobic but not anaerobic conditionsZl. Two other activities found to be induced by superoxide-generating conditions are endonuclease IVZZ and glucose-6-phosphate dehydrogenase (oxidatively stressed cells may have an increased requirement for NADPH, which is generated by the action of glucose-6-phosphate dehydrogenase)9,10. In addition, Kogoma et al. isolated three lacZ gene fusions whose expression was induced by paraquat and plumbagin ( soi-1 7::lacZ, soi-19. :lacZ, soi-28: :lacZ)23. A regulator, soxR, of these superoxide-inducible genes has been defined by tF,,: isolation of mutants that show increased resistance to menadione, constitutive expression of a subset of superoxide-inducible

proteins as seen on two-dimensional gels, and elevated expression of reporter genes fused to the promoters of the superoxide-inducible genes9.10. Deletion mutations at the soxR locus prevent expression of the fusion constructs, suggesting ~hat soxR functions to activate transcription of superoxidc-inducible genes. Characterization of the soxR locus and additional mutations that affect the expression of the superoxide-inducible genes will undoubtedly help to define how cells respond to superoxide anion.

Hydrogen peroxide-inducible genes When E. coil and S. typbimurium cells are treated with low doses (60 lXM)of hydrogen peroxide, which allow them to adapt to higher doses (10 mM), they induce the synthesis of approximately 30 proteins as seen on two-dimensional gels24, 25. The expression of nine of the proteins that are induced within the first 10 minutes after treatment with hydrogen peroxide is regulated by oxyR. Strains carrying deletions of the oxyR gene are hypersensitive to hydrogen peroxide and cannot induce the expression of the nine proteins. Other oxyR mutan,t strains (oxyR1 and oxyR2) are more resistant to hydrogen peroxide than wild-type strains and constitutively overexpress the nine proteins, including HPI catalase, glutathione reductase and the alkyl hydroperoxide reductase. The sequence of the oxyR gene showed that OxyR is homologous to an expanding family of bacterial regulators including LysR, an E. coil regulator of lysine biosynthesis, and NodD, a Rbizobium regulator of genes involved in the formation of nodulesa0--29. All of the above regulators serve as activators of heterologous genes, and many, including OxyR, apparently negatively regulate their own expression. Primer extension studies of total reduced OxyR cellular RNA isolated from oxy/i~ and oxyR1 'constitutive' mutant strains showed that the levels of the katG and abpC messages are elevated in the oxyR1 mutant strains~0. 35 -10 Sequences upstream of the abpC and katG genes determined to be important for oxyR activation were found to be protected from DNase I digestion by purified OxyR protein. Intriguingly, the sequences protected by OxyR show very little similarity30. oxidative stress The mechanism by which the cell senses hydrogen peroxide and activates transcription was illuminated by studies on the ability of OxyR-overproducing extracts and purified OxyR to activate the transcription of -35 -10 h.._ the oxyR-regulated abpCF and katG genes in vitro31. Initially, the finding that extracts and purified OxyR protein from untreated cells could activate transcription was suroxidized OxyR prising. However, subsequent experiments showed that OxyR was activated solely by FIGH Schematic model of the binding of reduced and oxidized OxyRto an the release from the reducing environment OxyR-regulated promoter before and after oxidative stress. Both the of the E. coil cell. If the extracts from the reduced and oxidized forms of OxyRbind to the promoter, but only overproducing strains were prepared under the oxidized form activates transcription. The footprints of the reduced and reducing conditions, OxyR no longer oxidized forms of OxyR are different, suggesting that a distinct activated expression, though exposure to conformational change in the OxyR protein is associated with its transition oxidizing conditions immediately converted from an inactive to an active form31. OxyR into a transcriptional activator TIG NOVEMBER1990 VOL.6 NO. 11

~'~EVIEWS

superoxide-inducible

h y d r o g e n peroxide-inducible

soxR-regulated sodA nfo zwf soil 7,soil 9 soi28

heat shock-inducible

stationary-inducible ?

FIGR Responses to oxidative stress. Treatment with superoxide-generatingcompounds or with hydrogen peroxide induces the synthesis of groups of proteins. Some of these proteins are also induced by heat shock or by other types of stress. Three regulators have been characterized that modulate the expression of subsets of the stress proteins, induced only by superoxide-induced stress (soxR), hydrogen peroxide-induced stress (oxyR), or oxidative stress occurring during stationary phase (katl9.

(Fig. 1). The reducing conditions did not denature the OxyR protein, since OxyR was still capable of repressing its own expression under the reducing conditions and still bound the katG and ahpC promoter regions. The DNase I protection patterns for purified OxyR under oxidizing and reducing conditions are significantly different, suggesting that OxyR undergoes a conformational change. These findings indicate that OxyR senses oxidative stress directly by becoming oxidized, and that oxidation brings about a conformational change by which OxyR transduces the oxidative stress signal to RNA polymerase. Overlapping regulation As well as the fact that there appear to be at least two distinct pathways for the activation of genes required to defend cells against oxidative stress, there

are additional complexities (shown schematically in Fig. 2). First, when cells are treated with low doses of hydrogen peroxide they become cross-resistant to heat shock and other types of stressZ. Second, some of the proteins induced by hydrogen peroxide or superoxide radicals overlap with heat shock and other stress proteinslg,20,24,25,3z. For example, glucose- or nitrogenstarved cultures of E. coli show increased resistance to high doses (15 mi) of hydrogen peroxide and some of the proteins induced by starvation overlap with hydrogen peroxide-inducible proteins3Z. The finding that strains with mutations in recA have increased sensitivity to hydrogen peroxide suggests that induction of the SOS response is also important for the defense against oxidative stress, though the induction may be a secondary consequence in response to the DNA damage generated by oxidative stress. Consistent with this

WIGNOVEMBER1990 VOL 6 NO. 11 k

~EVIEWS 7 Demple, B., Johnson, A. and Fung, D. (1986) Proc. Natl Acad. Sci. USA 83, 7731-7735 8 Imlay, J.A. and Linn, S. (1988) Science 240, 1302-1309 9 Greenberg, J.T. et al. Proc. NatlAcad. Sci. USA (in press) 10 Tsaneva, I.R. and Weiss, B. (1990)J. BacterioL 172, 4197-4205 11 Imlay, J.A., Chin, S.M. and Linn, S. (1988) Science 240, 640-642 12 Farr, S.B., D'Ari, R. and Touati, D. (1986) Proc. Natl Acad. Sci. USA 83, 8268-8272 Storz, G., Christman, M.E, Sies, H. and Ames, B.N. (1987) Proc. Natl Acad. Sci. USA 84, 8917-8921 14 Greenberg, J.T. and Demple, B. (1988) EMBOJ. 7, 2611-2617 15 Farr, S.B., Touati, D. and Kogoma, T. (1988)J. BacterioL 170, 1837-1842 16 Brot, N. and Weissbach, H. (1983) Arch. Biochem. Biophys. 223, 271-281 17 Levine, R.L., Oliver, C.N., Fulks, R.M. and Stadtman, E.R. (1981) Proc. Natl Acad. Sci. USA 78, 2120-2124 18 Kuo, C.E, Mashino, T. and Fridovich, I. (1987) J. Biol. Chem. 262, 4724-4727 19 Walkup, L.K.B. and Kogoma, T. (1989) J. Bacteriol. 171, 1476-1484 20 Greenberg, J.T. and Demple, B. (1989)J. Bacteriol. 171, 3933-3939 21 Touati, D. (1988)J. BacterioL 170, 2511-2520 22 Chan, E. and Weiss, B. (1987) Proc. NatlAcad. Sci. USA 84, 3189-3193 23 Kogoma, T., Farr, S.B., Joyce, K.M. and Natvig, D.O. (1988) Proc. Natl Acad. Sci. USA 85, 4799-4803 24 Morgan, R.W. et ai. (1986) Proc. Natl Acad. Sci. USA 83, 8059--8063 25 VanBogelen, R.A., Kelley, P.M., and Neidhardt, F.C. (1987) J. Bacteriol. 169, 26-32 26 Christman, M.E, Storz, G. and Ames, B.N. (1989) Proc. Natl Acad. Sci. USA 86, 3484-3488 2 7 BOlker, M. and Kahmann, R. (1989) FdVlBOJ. 8, 2403-2410 28 Tao, K. etal. (1989) Mol. Gen. Genet. 218, 371-376 29 Warne, S.R., Varley, J.M., Boulnois, G.J. and Norton, M.G. (1990) J. Gen. Microbiol. 136, 455--462 30 Tartaglia, L.A., Storz, G. and Ames, B.N. (1989)J. Mol. Biol. 210, 709-719 31 Storz, G., Tartaglia, L.A. and Ames, B.N. (1990) Science 248, 189-194 32 Jenkins, D.E., Schultz, J.E. and Matin, A. (1988) J. Bacteriol. 170, 3910-3914 33 Goerlich, O., Quillardet, P. and Hofnmlg, M. (1989) J. Bacteriol. 171, 6141---6147 34 Sak, B.D., Eisenstark, A. and Touati, D. (1989) Proc. Natl Acad. Sci. USA 86, 3271-3275 35 Mulvey, M.R. and Loewen, P.C. (1989) Nucleic Acids Res. 17, 9979-9991 3 6 Bachmann, B.J. (1990) Microbiol. Rev. 54, 130-197

interpretation, SOS induction was found to occur at much lower hydrogen peroxide concentrations in oxyR deletion strains than in the wild-type parent strains33. The overlapping regulation of these different stresses is not well understood, but further characterization of the overlaps as well as the characterization of additional regulators will lead to better understanding of how cells perceive stress. One additional aspect of the bacterial response to oxidative stress is noteworthy. E. coil cells possess enzyme pairs for a number of different defense activities, including two superoxide dismutase, catalase and exonuclease III activities'l,5, a. For the catalase and exonuclease III activities, it appears that one of the enzymes is expressed predominantly during exponential growth while the other is expressed predominantly after the cells are in a stationary phase. Sak et al. have recently shown that the katF gene product regulates the expression of the forms of catalase (HPII, katE) and exonuclease III (xthA) that are expressed during the stationary phase34. Intriguingly, the katF-encoded protein is homologous to a transcription factors, suggesting that KatF acts to direct RNA polymerase to the k a t e and xthA promoters after cells have stopped growing35. In summary, the study of genes important for the bacterial response to oxidative stress has given several insights into cellular defenses against oxidative stress. Superoxide radical and hydrogen peroxide, acting directly or indirectly through the production of hydroxyl radical species, are deleterious to bacterial cells. Cells require superoxide dismutase, catalase and other enzymes capable of reducing oxidants, in order to prevent mutagenesis, loss of membrane transport, and degradation of essential enzymes. The regulation of the genes induced in response to oxidative stress is intricate. There are at least three distinct responses to oxidative stress as well as several overlaps with other types of stress responses. The expression of a subset of hydrogen peroxide-inducible genes is regulated by a transcriptional activator that is capable of sensing oxidative stress directly. Clearly, many questions about the bacterial response to oxidative stress remain, and comparisons with oxidative stress responses in eukaryotic systems will also be interesting. It is likely that, as for many other genetic responses, many characteristics of the bacterial response to oxidative stress will have parallels in eukaryotic systems.

Acknowledgements We thank Jean Greenberg, Dan Voytas, and a reviewer for helpful comments on the manuscript.

References I Demple, B. and Halbrook, J. (1983) Nature 304, 466-468 2 Christman, M.F., Morgan, R.W., Jacobson, ES. and Ames, B.N. (1985) Cell41, 753-762 3 Farr, S.B., Natvig, D.O. and Kogoma, T. (1985)J. Bacteriol. 164, 1309-1316 4 Carlioz, A. and Touati, D. (1986) EMBOJ. 5, 623-630 5 Loewen, P.C., Switala, J. and Triggs-Raine, B.L. (1985) Arch. Biochem. Biophys. 243, 144-149 6 Jacobson, F.S., Morgan, R.W., Christman, M.E and Ames, B.N. (1989)J. Biol. Chem. 264, 1488--1496

G. STORE IS IN THE LABORATORYOF MOLECULAR BIOLOGY, NATIONAL CANCER INSIIIIri~ BETHESD~ MD 20892, USA," TARTAGLIA IS 1N GENENTECHINC., 460 PT SAN BRUNO Btv~ SOrWHSAN F~CaSO~ CA 94080, US&"S.R F,w a Is IN THELABORATORYOF TOXICOLOGY,HARVARDSCHOOLOFPUBUC HEALTH, 665 HUNTINGTON AVENUF~ BOSTON, HA 02115, US&" AND ~ N . AMES IS IN THE.DMSION OF BIOCHEMISTRY AND MOLECULARBIOLOGY, 401 BARKERHAl1, UNIVERSITgOF CALWOPJVV~BERg£LEY, CA 94 720, USA.

T1G N')VEMBER1990 VOL.6 NO. 11 I

~6~