Regulation of the Salmonella typhimurium Flavohemoglobin Gene A NEW PATHWAY FOR BACTERIAL GENE EXPRESSION IN RESPONSE TO NITRIC OXIDE*

Flavohemoglobins, a family of two-domain proteins with homology to vertebrate hemoglobins, are found in a variety of prokaryotic and eukaryotic microorgan-isms. Recent studies suggest a role for these proteins in nitrogen oxide metabolism. We now show that nitric oxide donors positively regulate a chromosomal flavohemoglobin ( hmp )/ lacZ operon fusion in Salmonella typhimurium. hmp gene expression in the presence of NO z is independent of the SoxS, OxyR, and FNR transcription factors and instead relies on inactivation of the iron-de-pendent Fur repressor. Other Fur-repressed promoters in S. typhimurium are also activated by an NO z donor. In contrast to the wild-type strain, an hmp 2 mutant requires markedly lower concentrations of NO to induce the hmp/lacZ fusion, whereas its response to iron chelation is equivalent to wild type. These data unveil a new pathway for NO-dependent gene expression in S. typhimurium.

In addition to its function as a regulatory molecule in organisms ranging from humans to slime molds, nitric oxide possesses potent and broad spectrum antimicrobial activity (1). The cellular targets responsible for the cytocidal and/or cytostatic action of NO 1 include lipids, thiols, DNA, and transition metals. A growing body of evidence indicates that bacteria are capable of inducing protective enzymes in the presence of NO and other reactive nitrogen intermediates, a process recently termed the nitrosative stress response (2). The SoxRS system, which is activated by superoxide generators, will stimulate antioxidant genes in reaction to NO ⅐ and defend Escherichia coli from the NO-dependent bactericidal activity of macrophages (3). OxyR, a transcription factor involved in stimulation of peroxide detoxification genes, is directly modified by NO via S-nitrosylation and assists in protecting the bacterium from the NO donor S-nitrosocysteine (2). Alkyl hydroperoxide reductase subunit C, one of the enzymes influenced by OxyR, has been shown to protect Salmonella typhimurium from another NO donor, S-nitrosoglutathione (GSNO) (4).
Deletion of flavohemoglobin, a two-domain protein with Nterminal homology to hemoglobins and C-terminal homology to oxidoreductases, also results in hypersensitivity of S. typhimurium to nitrosative stress (5). The mutant strain is equiva-lent to wild-type in sensitivities to superoxide and hydrogen peroxide, suggesting the existence of a nitrosative stress response that is independent of the oxygen-related stress pathways. Flavohemoglobins have been isolated from phylogenetically distant organisms, including Saccharomyces cerevisiae, Bacillus subtilis, and E. coli (6 -8). Although the promoter activities of some flavohemoglobins are influenced by oxygen availability (7,9), nitrogen oxides are known to induce the transcription of these proteins in other organisms. Flavohemoglobin in the fungus Fusarium oxysporum is present only during denitrification, an anaerobic electron transport pathway that reduces nitrate to dinitrogen via nitrogen oxide intermediates (10). A flavohemoglobin mutant strain of the denitrifying bacterium Alcaligenes eutrophus is deficient in the transient production of nitrous oxide (N 2 O), an intermediate that is immediately downstream of NO ⅐ in the denitrification pathway (11). Many other flavohemoglobin-containing organisms do not have the denitrifying capability of F. oxysporum or A. eutrophus. Some, such as E. coli and B. subtilis, will reduce nitrite to ammonia or ammonium rather than nitric oxide (12,13). Nevertheless, purified NO ⅐ was found to be a major inducer of flavohemoglobin (hmp) promoter activity in E. coli (14). An hmp/lacZ operon fusion was stimulated approximately 20-fold by 20 M nitric oxide, while requiring 8 mM nitrite and 40 mM nitrate to give similar results. Unlike the nitrite and nitrate effects, which are escalated anaerobically, the NO ⅐ induction is largely independent of O 2 concentration. The B. subtilis hmp is also induced by comparable levels of nitrite, but the influence of nitric oxide was not reported (7). Recently, purified E. coli flavohemoglobin was demonstrated to oxidize NO ⅐ to the less toxic nitrate (15), thus assigning an enzymatic activity to the nitrosative stress protection shown by phenotypic analyses (5).
The elements responsible for the NO ⅐ -dependent expression of flavohemoglobins are not known. Although the ResDE twocomponent system accounts for some of the anaerobic expression of hmp in B. subtilis, factors contributing to the nitrite induction were not elucidated (7). E. coli strains harboring null alleles of narL and narP, which regulate nitrate and nitrite reductases, do not substantially alter hmp expression (14). FNR, a transcription factor known to positively regulate a flavohemoglobin homolog in the bacterium Vitreoscilla as well as many denitrification enzymes in other species, actually represses anaerobic expression of the E. coli hmp (14,16,17). The NO ⅐ expression is also largely independent of the SoxRS system in E. coli (14). In addition to nitrogen oxides, the iron chelator 2,2Ј-dipyridyl significantly enhances E. coli hmp expression. This induction was tentatively attributed to deactivation of FNR, which requires iron for function (14).
In this study, we have found that the S. typhimurium hmp gene expression is induced by the NO donors and that the transcriptional repressor Fur is the primary factor responsible for hmp regulation. Nitric oxide causes a general derepression of other Fur-regulated genes, suggesting a new mechanism of NO action on bacterial gene expression.
Bacterial Strains and Plasmids-The strains in this study are listed in Table I. To obtain an integrated hmp/lacZ operon fusion, a 1-kilobase pair SalI/SmaI DNA fragment containing approximately 950 base pairs of hmp upstream sequence was isolated from pMC71 and ligated into the SmaI site of pFUSE, a suicide plasmid that harbors the promoterless lacZYA operon (19). The resultant plasmid was propagated in E. coli strain S17-1 lpir (20) and transformed into S. typhimurium 14028s, making MCS38. Integration at the hmp locus was confirmed by Southern blot (not shown). Transductions of various transcription factor null alleles into MCS38 were performed using bacteriophage P22 (21). Details on the construction of plasmids pSKO1/2 and pSKO3/4, used in the Fur titration assay, were described previously (5). The fur mutants were confirmed by siderophore production on chrome azurol S plates (22).
Culture Conditions-All liquid cultures were grown in LB broth. When necessary, antibiotics were added at the following concentrations: 100 mg/ml ampicillin, 50 mg/ml kanamycin, 25 mg/ml chloramphenicol, 15 mg/ml tetracycline. For SPER/NO growth curves, 250 ml of bacteria from an overnight culture was inoculated into a 125-ml culture flask containing 25 ml of medium. The bacteria were then allowed to grow at 37°C for 1 h with 225 rpm shaking before the introduction of 1 mM SPER/NO. Turbidities were followed using a Klett colorimeter.
For gene expression studies, overnight cultures were diluted 1/100 into test tubes containing 2 ml of medium and were shaken at 37°C for 1.5 h. Unless otherwise indicated, 1 mM SPER/NO or 0.2 mM 2,2Јdipyridyl was introduced, and the cultures were allowed to incubate for another 2 h. To stop gene expression, 300 mg/ml spectinomycin was added 5 min before harvesting.
␤-Galactosidase Assays-Cells were pelleted, resuspended in Z buffer (23), and permeabilized with sodium dodecyl sulfate and chloroform. ␤-Galactosidase activities, which were assayed in triplicate, are expressed as Miller units (23). All experiments were performed at least three times with two independently isolated clones. For the Fur titration assay, bacteria from overnight cultures were streaked onto Mac-Conkey agar with ampicillin and incubated overnight at 37°C (24).

RESULTS
The NO Donor Spermine NONOate Is Able to Induce hmp through a Novel Pathway-To facilitate gene expression studies on hmp, we used an S. typhimurium strain harboring a single copy of an hmp/lacZ operon fusion. The expression construct contains 920 base pairs of hmp upstream sequence fused to the promoterless lacZYA operon in the suicide plasmid pFUSE (19). As confirmed by Southern as well as by phenotypic analyses using NO donors (not shown), the endogenous hmp gene was not affected by this integration event. Spermine NONOate (SPER/NO), a well characterized NO ⅐ donor (25,26), was employed to determine sensitivities of wild-type and hmp Ϫ strains as well as hmp promoter activity in response to nitric oxide. Using 1 mM SPER/NO, a growth delay is observed for both strains but is much more pronounced in the hmp Ϫ mutant (Fig. 1). Addition of SPER/NO induces the hmp promoter ap-FIG. 1. Growth of the hmp ؊ strain is hypersensitive to spermine NONOate. Representative growth curves of exponentially growing wild-type (WT) and hmp Ϫ strains grown in LB with or without the addition of 1 mM spermine NONOate (SPER/NO), which was added 90 min after the initial bacterial inoculation (indicated by the arrow). This experiment was repeated three times, yielding similar results. proximately 20-fold (Fig. 2). We examined the transcription factors known to respond to nitric oxide (SoxS and OxyR) or to influence hmp transcription in E. coli (Fnr) for their contribution to aerobic hmp expression in S. typhimurium (2,15,27). Strains harboring null alleles of soxS, oxyR, and fnr are still able to respond to SPER/NO (Fig. 2), indicating that a previously uncharacterized pathway is responsible for the NO ⅐ -dependent expression of flavohemoglobin. (14). Nitric oxide is known to react with transition metals and may therefore perturb cellular iron levels (28). To determine whether the SPER/NO effect on the hmp promoter is specific or due to a more general phenomenon of iron metabolism, we also tested iroA and iroC reporter fusions, which were originally isolated because of their induction under low-iron conditions (19,29). In addition to induction by the iron chelator 2,2Ј-dipyridyl, the ␤-galactosidase activity driven by these promoters is also stimulated by 1 mM SPER/NO (Fig. 3). Therefore, NO appears to cause widespread alteration of iron-regulated gene expression in S. typhimurium.

Low Iron-induced Promoters Are Stimulated by NO-Low iron conditions increase expression of E. coli hmp
The hmp Promoter Is Repressed by Fur-The iroA and iroC promoters are repressed by the Fur protein, a global transcriptional regulator that is active when bound to ferrous iron (19,29). We tested whether a fur null allele could influence the hmp/lacZ fusion. Complete derepression occurs in the fur background, which cannot be augmented with the addition of 2,2Ј-dipyridyl and/or SPER/NO (Fig. 4). To determine if the hmp promoter is capable of binding Fur, a fur titration assay (FURTA) was performed (24). In this assay, a high copy number plasmid carrying a putative Fur-binding element is introduced into a strain that carries a Fur-repressed lacZ fusion. If the plasmid is capable of binding Fur, a titration of this repressor will occur, leading to expression of the lacZ reporter. A

FIG. 5. Fur is capable of binding the hmp promoter in vivo.
A Fur titration assay (24) was performed using high copy number plasmids harboring the hmp promoter region (pSKO1) (a) or a portion of the hmp open reading frame transformed into AJB27 (b), which harbors the Fur-repressed iroC/lacZ fusion. The resultant strains (MCS32 and MCS33) are plated on MacConkey agar with ampicillin, which will turn red (dark) upon the expression of ␤-galactosidase. c, to confirm that the color observed is dependent on the integrated reporter and not on the plasmid, pSKO1 was also transformed into wild-type S. typhimurium 14028 (MCS34). plasmid containing 213 base pairs upstream of the hmp start codon is able to derepress the iroC/lacZ fusion (Fig. 5) on MacConkey plates, whereas a plasmid with an insert containing a portion of the hmp open reading frame is not positive by FURTA. There is a sequence (5Ј-TCTAATGATGTATATCAAA-3Ј) found over the transcription start site in the hmp promoter that can be aligned to the consensus "Fur box" (5Ј-GATAAT-GATAATCATTATC-3Ј) (24). This level of identity to the consensus (10/19 matches) is equivalent to that of another FUR-TA-positive clone (pFTE-1) found during a general screen for Fur-regulated genes in S. typhimurium (30).
The hmp Mutant Is Hypersensitive to NO and Not to Iron Chelation-The finding that Fur, well characterized for its regulation of iron scavenging (31), also influences the expression of hmp raises the possibility that flavohemoglobin may function in the processing of intracellular iron. Consequentially, we tested the effects of 2,2Ј-dipyridyl on growth and hmp/lacZ activity in the hmp Ϫ strain. No differences in growth rates between wild-type and hmp Ϫ strains could be demonstrated for any concentration of the iron chelator (not shown). Also, the expression of the hmp/lacZ fusions as a function of 2,2Ј-dipyridyl concentration is equivalent in both wild-type and hmp Ϫ strains (Fig. 6a). In contrast, induction of the hmp/lacZ reporter in the hmp Ϫ strain is more sensitive to SPER/NO than wild type (Fig. 6b). S-Nitrosoglutathione (GSNO), another nitric oxide donor that causes an accentuated growth deficiency in the hmp Ϫ strain (5), is able to induce the hmp/lacZ fusion and provides a more dramatic (approximately 10-fold) decrease in the quantity needed for hmp gene induction in the hmp mutant background when compared with wild type (Fig. 6c). GSNO also causes a greater relative growth impairment in the hmp Ϫ strain than SPER/NO (not shown).
With increasing concentrations of NO donors, expression begins to decline after maximal gene induction (Fig. 6). This phenomenon is likely the result of toxicity to the strains, since cell growth diminishes at these concentrations. It is interesting to note that GSNO does not achieve the induction levels seen with SPER/NO. This may be due to the more complex chemistry of the nitric oxide species coming from GSNO. Unlike SPER/NO, which should provide a relatively pure source of the NO ⅐ radical (32), GSNO will donate both NO ⅐ as well as the nitrisonium ion (NO ϩ ) (33). Therefore, NO ⅐ may be the major form of nitric oxide regulating hmp gene expression. DISCUSSION In addition to its role in iron acquisition, Fur is postulated to regulate general metabolic processes as well as adaptation to acid stress (24, 29). The results presented here on the regulation of S. typhimurium flavohemoglobin demonstrate that the influence of Fur extends to the nitrosative stress response. We show that nitric oxide will derepress other iron-regulated promoters. The role of NO in control of iron metabolism has precedent in mammalian systems, although at a different level of regulation (34,35). The translation factors IRP-1 and IRP-2, which are thought to sense iron levels through the gain or loss of Fe-S clusters, control protein expression from several mRNAs involved in the uptake of iron. Nitric oxide is thought to mimic or cause iron deficiency by destruction of these clusters or by complexing with free iron, leading to depletion of the cellular iron pool (36). A similar situation may occur in a bacterial cell under nitrosative stress. NO could induce the conversion of Fe-Fur to an inactive Fur by modification of the iron or protein moiety. Alternatively, iron can react with nitric oxide to form dinitrosyl iron complexes, which may effectively sequester cellular iron in an unusable form. The formation of these complexes occurs in macrophages upon induction of nitric oxide synthase (37,38) and is thought to assist in creating the low iron environment found in these cells (39). We have shown that the hmp promoter is capable of binding Fur in vivo, and a putative Fur box is found over the start site of hmp transcription. We were, however, unable to find anything resembling a Fur box in either the E. coli or B. subtilis hmp promoters. FIG. 6. Hmp has no effect on the gene expression response to iron chelation but mutes the response to NO. Expression of the hmp/lacZ fusion in either a wild-type (MCS38) or hmp Ϫ (MCS39) background was monitored after introduction of increasing concentrations of 2,2Ј-dipyridyl (a), SPER/NO (b), or GSNO (c). Shown is a representative experiment done in triplicate. Similar results were obtained in subsequent experiments. Therefore, if Fur is influencing hmp expression in these organisms, it might be through a novel binding site or by indirect means.
Sensitivity to nitric oxide (or NO donors) is the only known growth or gene expression difference between wild-type and hmp Ϫ strains (5,15). The hmp Ϫ strain is no more sensitive than wild type to iron chelation, bolstering the hypothesis that Hmp is involved with nitrosative stress protection rather than with general iron maintenance. Control of iron uptake, however, is an elaborate and complex process (40). We currently cannot rule out the possibility that flavohemoglobin is involved with an aspect of iron metabolism that is particularly sensitive to nitric oxide. Like flavohemoglobin, the expression of superoxide dismutase, an enzyme involved in oxidative stress protection, is regulated by Fur (41). Therefore, in addition to SoxRS and OxyR (2, 3), nitrosative and oxidative stresses share a third pathway toward the induction of protective genes in enteric bacteria. Although these stresses share many common targets within the cell (2), an intriguing disparity exists in their interaction with iron. Whereas reactive oxygen intermediates will undergo Fenton chemistry with iron to create highly toxic metabolites (42), nitric oxide will simply bind iron, with the resultant complex creating unknown consequences for the cell (28). Despite the induction of superoxide dismutase, the E. coli fur mutant is hypersusceptible to oxygen metabolites and cannot grow aerobically without an efficient DNA repair mechanism (43). It will prove interesting to assess the contribution of intracellular iron levels to nitrosative stress and the role that Fur and Hmp play in this process.