Three Pseudomonas putida FNR Family Proteins with Different Sensitivities to O2*

Background: FNR proteins are O2-responsive bacterial transcription factors. Results: Pseudomonas putida possesses three FNR proteins with iron-sulfur clusters that have different sensitivities to O2. Conclusion: The mechanism of the iron-sulfur cluster reaction with O2 is conserved among FNR proteins. Significance: Differential sensitivity of multiple FNR proteins extends the range of O2-responsive gene expression within a single bacterium.

The Escherichia coli fumarate-nitrate reduction regulator (FNR) protein is the paradigm for bacterial O 2 -sensing transcription factors. However, unlike E. coli, some bacterial species possess multiple FNR proteins that presumably have evolved to fulfill distinct roles. Here, three FNR proteins (ANR, PP_3233, and PP_3287) from a single bacterial species, Pseudomonas putida KT2440, have been analyzed. Under anaerobic conditions, all three proteins had spectral properties resembling those of [4Fe-4S] proteins. The reactivity of the ANR [4Fe-4S] cluster with O 2 was similar to that of E. coli FNR, and during conversion to the apo-protein, via a [2Fe-2S] intermediate, cluster sulfur was retained. Like ANR, reconstituted PP_3233 and PP_3287 were converted to [2Fe-2S] forms when exposed to O 2 , but their [4Fe-4S] clusters reacted more slowly. Transcription from an FNR-dependent promoter with a consensus FNR-binding site in P. putida and E. coli strains expressing only one FNR protein was consistent with the in vitro responses to O 2 . Taken together, the experimental results suggest that the local environments of the iron-sulfur clusters in the different P. putida FNR proteins influence their reactivity with O 2 , such that ANR resembles E. coli FNR and is highly responsive to low concentrations of O 2 , whereas PP_3233 and PP_3287 have evolved to be less sensitive to O 2 .
Fumarate-nitrate reduction regulator (FNR) 2 proteins are a major subgroup of the cyclic-AMP receptor protein family of bacterial transcription regulators (1). The major function of FNR proteins is the reprogramming of gene expression to coordinate the switch from aerobic to anaerobic metabolism when facultative anaerobes like Escherichia coli are starved of O 2 (2)(3)(4)(5)(6)(7). The paradigm for O 2 -sensing transcription factors is the E. coli FNR protein. The N-terminal region of FNR contains four essential cysteine residues that coordinate an O 2 -sensitive [4Fe-4S] cluster (8,9). In the absence of O 2 , the [4Fe-4S] cluster is stable, and FNR exists as a homodimer that is capable of high affinity, site-specific DNA binding to an FNR box (TTGATNNNNATCAA) (9,10). When bound to target DNA, FNR activates the expression of genes encoding proteins required for anaerobic metabolism and represses those utilized under aerobic conditions (2,4), such that when O 2 is available, anaerobic metabolism is shutdown in favor of the more energetically efficient aerobic respiratory metabolism. Molecular oxygen reacts with the FNR [4Fe-4S] cluster in a series of steps that ultimately yields the apo form of the protein (Equations 1-3) (11)(12)(13).
͓4Fe-4S͑CysS͒ 4 ͔ 2Ϫ ϩnO 2 ¡ ͓2Fe-2S͑CysS͒ 2 ͑CysSS͒ 2 ͔ 2Ϫ ϩ Fe 2ϩ ϩ Fe 3ϩ ϩreduced O 2 species (Eq. 4) The retention of cluster sulfide (as CysSS) permits facile repair of the FNR [4Fe-4S] cluster in the presence of Fe 2ϩ and a reducing agent (14). Molecular oxygen-dependent conversion of the [4Fe-4S] cluster to the persulfide-ligated [2Fe-2S] causes con-formational rearrangements at the dimer interface that result in a charge-clash in the dimerization helices (15). Consequently, FNR monomerizes and is no longer able to bind DNA or to regulate transcription (10). Upon prolonged exposure to O 2 in vitro and in vivo, the dominant FNR species is the monomeric apo form, which is capable of acquiring a [4Fe-4S] cluster via the action of the Isc (iron-sulfur cluster) biosynthetic machinery, such that FNR continually monitors the cytoplasm for the availability of O 2 (16 -18).
Unlike many bacteria, Pseudomonas putida and Burkholderia spp. possess multiple FNR family proteins that retain the characteristic cluster of cysteine residues in the N-terminal sensory domain. Transcriptomic analysis of the opportunistic pathogen Burkholderia cenocepacia revealed the presence of a 50-gene low oxygen-activated (lxa) locus that was associated with persistence of this obligate aerobe under anaerobic conditions (19). The lxa locus includes the FNR protein BCAM0287, which was induced 17-fold under low O 2 (microaerobic) conditions. In addition, two other FNR protein encoding genes were induced during growth under a 6% O 2 atmosphere, BCAM0049 (induced 77-fold, compared with aerobic conditions) and BCAM1483 (induced 3.5-fold) (19). Although an FNR box-like motif was associated with many genes induced under microaerobic conditions, the functions of the multiple FNR regulators in B. cenocepacia are poorly defined. Similarly, the properties and functions of the three FNR proteins of P. putida KT2440 are poorly understood. Here for the first time, three FNR proteins (PP_3233, PP_3287, and PP_4265, the last of which is also known as ANR) from a single bacterial species, P. putida KT2440, have been isolated, and their responses to O 2 in vivo and in vitro have been assessed.
Gel Filtration, Protein Reconstitution, and Total Amino Acid Analysis-The oligomeric state of ANR was estimated from the elution volume of a sample (50 l of 250 M protein) applied to a calibrated Superdex 200 column. The column was equilibrated with 25 mM Tris-HCl, pH 7.5, containing 500 mM NaCl and 2 mM EDTA, and the standards used to calibrate the column were blue dextran, hemoglobin, ovalbumin, cytochrome c, and aprotinin.
The ANR, PP_3233, and PP_3287 proteins were reconstituted overnight under anaerobic conditions at 25°C in 25 mM HEPES, 100 mM NaCl, 100 mM NaNO 3 , pH 7.5, to which 0.5 mM L-cysteine, 12.5 mM DTT, an 8-fold molar excess of (NH 4 ) 2 Fe(SO 4 ) 2 and 0.2 M NifS cysteine desulfurase were added. Reconstituted proteins were purified on a heparin column (GE Healthcare) and eluted in 25 mM Tris-HCl containing 500 mM NaCl, pH 7.5 (21). Iron content was determined as previously described (21). Total amino acid analysis was carried out by Alta Bioscience (University of Birmingham, Birmingham, UK) following complete acid hydrolysis of ANR protein samples that had previously had the protein content estimated by the Bio-Rad protein assay (20).
UV-visible and CD Spectroscopy-Sealed anaerobic cuvettes containing reconstituted FNR proteins were injected with increasing amounts of air-saturated buffer as indicated and were incubated at 25°C for 10 min followed by spectroscopic analysis. Absorbance measurements were made with a Cary UV-visible spectrophotometer. Changes in absorbance at 405 nm were used to monitor the conversion of the clusters. The extinction coefficient for the E. coli FNR iron-sulfur cluster (⑀ 406 nm ϭ 16,200 M Ϫ1 cm Ϫ1 ) was used to calculate the amount of [4Fe-4S] 2ϩ cluster in the reconstituted proteins. The spectra shown are typical of at least three experiments. CD measurements were made with a Jasco J-810 spectropolarimeter. Aliquots of ANR (680 l) were diluted to 29.8 M iron-sulfur cluster under anaerobic conditions for initial measurements before treating with oxygenated buffer (ϳ220 M O 2 ) to give ϳ2-fold molar excess O 2 and then incubated for 15 min at room temperature prior to further measurements.
Kinetic Measurements-Reactions were initiated by the injection of air-saturated buffer (final concentration, ϳ100 M O 2 ) into sealed anaerobic cuvettes containing reconstituted ANR, PP_3233, or PP_3287 proteins (final concentration, ϳ6-9 M [4Fe-4S]) at 25°C. The dead time of mixing was ϳ5 s. Changes in absorbance at 420 nm were used to monitor the conversion of the clusters. The A 420 nm decay data were fitted to a single or double exponential function (as appropriate) using the program Origin (version 8; OriginLab). Where a double exponential function was fitted, the higher rate constant was assumed to correspond to the initial reaction with O 2 . Reported rate constants are mean values with standard errors from three repeats.
Construction of Plasmids and Bacterial Strains-To investigate the responses of the three P. putida FNR proteins in vivo, it was necessary to create P. putida KT2440 strains that only expressed one of the three FNR proteins encoded by the genome. Two different strategies were used to create unmarked deletion mutants. The P. putida gene PP_4265 encoding ANR was deleted using sacB counter selection and FLP recombinase excision as described by Hoang et al. (22). The primer pairs oAS23 (5Ј-GGAATTCAGCCAGATCGGCGACCTGTA-3Ј), oAS24 (5Ј-CGGGATCCTGTAGGCCAGTGTGCGCGAT-3Ј), oAS25 (5Ј-CGGGATCCACCTTGGCCTGGCGGTA-GAA-3Ј), and oAS26 (5Ј-GCTCTAGACTGTCGGCATGCA-CTTCCAG-3Ј) containing engineered EcoRI, BamHI, and XbaI restrictions sites (as indicated by underlining) were used to amplify 511-and 533-bp DNA fragments flanking the PP_4265 gene. The fragments were cloned into the suicide vector pEX18Ap flanking the gentamicin resistance cassette from plasmid pPS858 and used to generate the unmarked P. putida PP_4265 mutant strain (22).
RNA Isolation and qRT-PCR-Cultures of E. coli JRG6348 transformed with the pBAD-HisB-derivatives pGS2350, pGS2351, pGS2352, or pGS2353 (encoding FNR, PP_4265 (ANR), PP_3233, and PP_3287) all expressed under the control of the pBAD promoter to eliminate any differential transcriptional control over the production of the regulators (see Table  1) were grown under anaerobic conditions (sealed tubes filled to the neck) in M9 minimal medium supplemented with L-broth (5%, v/v), glycerol (0.4%, v/v), trimethylamine N-oxide (20 mM), sodium fumarate (20 mM), and ampicillin (100 g ml Ϫ1 ) at 37°C until the A 600 reached ϳ0.2 (2). Aliquots were removed, and mRNA was stabilized by the addition of 0.4 volume of ice-cold ethanol-phenol (95%:5%) at pH 4.5. The cultures were then exposed to air by shaking, and incubation was continued for 20 min before taking further samples for total RNA preparation using the RNeasy RNA purification kit (Qiagen) according to the manufacturer's instructions. RNA was quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific). Relative lacZ RNA quantities were determined for triplicate cultures as previously described (25).  (41). The cysteine residues that coordinate the FNR iron-sulfur cluster (bold type), residues adjacent to cluster ligating cysteine residues that are substituted compared with FNR (white type on black), the DNA recognition helix (shaded gray), Asp-154 of FNR (bold black type on gray), residues that are identical in all four proteins (*), residues with strongly similar properties (:), and residues with weakly similar properties (.) are indicated.

Results and Discussion
P. putida-KT2440 possesses three FNR proteins: PP_3233, PP_3287 and PP_4265 (hereafter ANR). Compared with the E. coli FNR protein, ANR is 53% identical (76% similar over 226 amino acid residues), PP_3233 is 46% identical (67% similar over 225 amino acid residues), and PP_3287 is 41% identical (58% similar over 224 amino acid residues). The four cysteine residues that coordinate the [4Fe-4S] cluster that is essential for the function of E. coli FNR are conserved, and thus all three P. putida FNR proteins were predicted to contain cysteineligated [4Fe-4S] clusters; however, the amino acid residues in the vicinity of the clusters differ ( Figs. 1 and 2). Previous studies have shown that replacement of amino acid residues adjacent to cluster coordinating cysteine residues can have profound effects on the reactivity of the E. coli FNR iron-sulfur cluster with O 2 (12,26). These observations suggested that the three P. putida FNR proteins might have evolved different sensitivities to O 2 .
The Reaction of the P. putida ANR Iron-Sulfur Cluster with O 2 Resembles That of E. coli FNR-ANR was released from a GST-ANR fusion by "on-column" treatment with the protease thrombin. Application of the resulting apo-ANR protein to a calibrated gel filtration column indicated that unlike apo-FNR, which is monomeric (10), apo-ANR was dimeric, despite retaining Asp-154 (FNR numbering) that is proposed to cause a charge clash preventing dimerization of apo-FNR ( Fig. 1 and Ref. 15). This suggests that additional residues in the dimer interface also contribute to determine the oligomeric state of ANR and FNR. After anaerobic iron-sulfur cluster reconstitution, the iron content of ANR was 4.1 Ϯ 0.3 iron atoms per subunit (n ϭ 3), based on protein estimation by total amino acid analysis. The anaerobic UV-visible spectrum of ANR was characteristic of a [4Fe-4S] protein (⑀ 405 nm ϭ ϳ18,000 M Ϫ1 cm Ϫ1 ), and upon addition of O 2 the spectrum changed to resemble that of a [2Fe-2S] protein, with broad absorbance bands at 320, 420, and 550 nm (Fig. 3A). Upon prolonged (16 h) exposure to air, the [2Fe-2S] form was degraded to the apo-ANR protein. Titration of reconstituted ANR with aerobic buffer revealed a progressive decrease in absorbance in the 400 -420-nm region associated with conversion of the [4Fe-4S] form to the [2Fe-2S] form (Fig. 3A). The CD spectrum of reconstituted [4Fe-4S] ANR exhibited positive bands at 296, 325, 375, and 420 nm, reminiscent of [4Fe-4S] FNR (21). Following exposure to O 2 (ϳ2-fold molar excess), these bands were replaced by a broad spectrum with two positive bands at 325 and 450 nm and one negative band at 375 nm, similar to the [2Fe-2S] form of FNR (Fig. 3B) (21). Treatment of the [4Fe-4S] form of ANR with 2 molar equivalents of O 2 for 15 min followed by analysis of the resulting [2Fe-2S] form by LC-MS revealed the presence of up to five sulfur adducts, with one and two additional sulfurs as the major species (Fig. 3C). Thus, it was concluded that the reaction of the ANR [4Fe-4S] cluster with O 2 proceeds via the same mechanism as that described for FNR, including the retention of cluster sulfide (14). The retention of cluster sulfide as S 0 has implications for the repair of [4Fe-4S] clusters (14). Anaerobic incubation of [2Fe-2S] ANR with a 4-fold molar excess of ferrous ions in the presence of the reducing agent DTT regenerated the [4Fe-4S] form, as judged by the UV-visible spectrum of the protein (Fig. 3D). Thus, the mechanism of [4Fe-4S] repair proposed for E. coli FNR is likely to be a common feature of this family of regulators and probably other iron-sulfur proteins (14).

Reactions of the [4Fe-4S] Clusters of PP_3233 and PP_3287 with O 2 Result in Conversion to [2Fe-2S] Forms-Several
attempts were made to overproduce the P. putida PP_3233 and PP_3287 proteins, but they were consistently found as insoluble aggregates when expressed at high levels, except when fused to the C terminus of the chaperone Trigger factor (Tig). Therefore, PP_3233 and PP_3287 were isolated as Tig fusions, and a Tig fusion of ANR was also generated to permit direct comparisons.
Anaerobic reconstitution of the iron-sulfur clusters of the three Tig fusion proteins resulted in UV-visible spectra characteristic of [4Fe-4S] proteins, with a broad absorbance at 400 -420 nm (Fig. 4, A-C). Titration with aerobic buffer resulted in spectral changes that were consistent with conversion from     (Fig. 4, A-C). The response of the Tig-ANR fusion (Fig. 4A) was similar to that of the untagged ANR protein (Fig. 3A) of ϳ13), the A 420 nm decays for ANR (both ANR and the Tig-ANR fusion), Tig-PP_3233, and Tig-PP_3287 were measured (Fig. 5). For ANR and Tig-ANR, the data were best fitted to a double-exponential function with observed rate constants (k obs ) for the first reaction of 0.034 Ϯ 0.007 s Ϫ1 for ANR and 0.028 Ϯ 0.0015 s Ϫ1 for the Tig-ANR fusion (Fig. 5A). This again indicates that fusion to Tig did not significantly affect the reactivity of the ANR iron-sulfur cluster, and thus it was assumed that a Tig tag would not affect the reactivity of the PP_3233 and PP_3287 clusters. For the Tig-PP_3233 and Tig-PP_3287 fusion proteins, the data were best fitted to a singleexponential function yielding k obs values of 0.0038 Ϯ 0.0002 s Ϫ1 for Tig-PP_3233 and 0.0055 Ϯ 0.0001 s Ϫ1 for Tig-PP_3287 (Fig.  5, B and C) (11,12).
The Responses of PP_3233 and PP_3287 to Increased Culture Aeration Are Weaker than That of ANR-To determine whether ANR, PP_3233, and PP_3287 act as O 2 sensors in vivo, three double mutant strains of P. putida were created in which two of the three genes encoding FNR proteins were deleted ( Table 1). Cultures of these strains were grown in 50-ml conical flasks at 30°C with shaking (200 rpm). For aerobic cultures, the flasks contained 10 ml of medium; for O 2 -limited cultures, the flasks contained 40 ml of medium. Strains that lacked anr exhibited impaired growth under O 2 -limited conditions, and strains that lacked either PP_3233 or PP_3287 were impaired under aerobic conditions (Fig. 6). This is consistent with relative O 2 sensitivities of the ANR, PP_3233, and PP_3287 ironsulfur clusters.
For E. coli FNR, it has been shown that Glu-209, Ser-212, and Arg-213 in the DNA recognition helix make the major interactions with the FNR box: TTGATCTAGATCAA (FF site). The amino acid sequences of the DNA recognition helices of the P. putida FNR proteins are very similar to those of E. coli FNR (PP_3287 has Cys in place of Ser), suggesting that ANR, PP_3233, and PP_3287 will recognize an FNR box (Fig. 1). Preliminary electromobility shift assays indicated that all three P. putida FNR proteins could bind at the FF site under anaerobic conditions. Therefore, the P. putida mutants were transformed with plasmid pGS810, which carries the FNR-dependent FF-41.5 (pFF-41.5) promoter fused to lacZ (Table 1). Cultures were grown under O 2 -limited conditions (50 ml of medium in a 50-ml shaking conical flask) and aerobic conditions (10 ml of medium in a 50-ml shaking conical flask). Measurement of ␤-galactosidase activity for P. putida JRG6723 (expresses only anr) cultures showed that ANR activity decreased ϳ5-fold in response to enhanced aeration (Fig. 7, A  and B). Measurement of the anr transcript by qRT-PCR and ANR protein by Western blotting with anti-serum raised against E. coli FNR for cultures grown in 50-ml shake flasks containing 50, 40, 30, 20, or 10 ml of medium, resulting in increasing O 2 transfer to the cultures as the volume of medium decreased, showed that the amounts of anr transcript and ANR protein were similar in all the cultures (Fig. 7B). However, measurement of ANR-dependent transcription in cultures grown under these same conditions showed that increased aeration progressively lowered FF-41.5 promoter activity (Fig. 7B). Thus, it was concluded that ANR has properties similar to E. coli FNR and is a bona fide O 2 -responsive gene regulator in P. putida. However, ␤-galactosidase activities for the P. putida strains lacking the anr gene (i.e. P. putida JRG6722 expresses only PP_3233, and P. putida JRG6721 expresses only PP_3287) showed only a small decrease in response to increased aeration and low levels of ␤-galactosidase activity (128 Ϯ 6 and 184 Ϯ 3 Miller units, respectively, under O 2 -limited conditions compared with 8550 Ϯ 54 Miller units for ANR) (Fig. 7A). This observation could result from poor expression of the PP_3233 and PP_3287 proteins. Therefore qRT-PCR was used to determine whether PP_3233 and PP_3287 were expressed in JRG6722 and JRG6721 under the conditions tested. The data indicated that the level of expression of PP_3233 was ϳ10-fold lower, and PP_3287 was ϳ5-fold lower than anr. Low levels of PP_3233 and PP_3287 mRNA were consistent with the hypothesis that expression of the three P. putida FNR proteins is likely to be temporally and/or spatially distinct. Therefore, to increase expression of PP_3233 and PP_3287, these genes and their respective promoter regions were ligated into the broad host range vector pBBR1-MCS-5 ( Table 1). The resulting expression plasmids were used to transform P. putida JRG6722 and JRG6721 carrying the pGS810 reporter plasmid creating strains that only expressed PP_3233 or PP_3287. Cultures expressing PP_3233 from pGS2508 (ANR Ϫ , PP_3233 ϩϩ , PP_3287 Ϫ ) or PP_3287 from pGS2509 (ANR Ϫ , PP_3233 Ϫ , PP_3287 ϩϩ ) were grown under O 2 -limited and aerobic conditions, and qRT-PCR showed that the level of PP_3233 and PP_3287 expression was increased by ϳ10and ϳ5-fold compared with the expression of chromosomal anr. Unfortunately, the corresponding increase in PP_3233 and PP_3287 proteins could not be determined because the E. coli FNR anti-serum did not cross-react with these proteins. Nevertheless, for both overexpression strains, FNR-dependent ␤-galactosidase activity decreased significantly with increased aeration (Fig. 7A). Thus,

TABLE 1 Bacterial strains and plasmids used in this work
Amp R , ampicillin resistance; Cm R , chloramphenicol resistance; Gm R , gentamicin resistance; Kan R , kanamycin resistance; Tet R , tetracycline resistance. To confirm the in vivo O 2 responsiveness of the P. putida FNR proteins, a heterologous reporter system consisting of an E. coli fnr, lac mutant (JRG6348) with a chromosomal copy of the FNR-dependent FF-41.5 promoter fused to lacZ was transformed with plasmids expressing FNR, ANR, PP_3233, or PP_3287 under the control of the pBAD promoter ( Table 1). Measurement of the decrease in lacZ transcript abundance by qRT-PCR after exposure of anaerobic cultures to O 2 for 20 min showed that the activities of E. coli FNR and all three P. putida regulators decreased, with FNR-and ANR-dependent transcription showing the greatest responses (Fig. 7C). The weaker responses of PP_3233 and PP_3287 suggested that these proteins were less sensitive to O 2 compared with FNR and ANR, consistent with the in vitro data presented above.

Relevant characteristics Source
Signal Specificity-Transcription factors that utilize iron-sulfur clusters as sensory modules have been shown to respond to O 2 (e.g. FNR), redox state (e.g. SoxR), nitric oxide (e.g. NsrR), and iron-sulfur cluster/iron homeostasis (e.g. IscR) (27,28). Some of these transcription factors respond to more than one of these signals. Hence, the E. coli FNR and SoxR proteins respond to nitric oxide in addition to their primary signals, O 2 and redox cycling, respectively. In vitro kinetic measurements with the [4Fe-4S] form of FNR indicated that it is much more sensitive to nitric oxide than it is to O 2 . However, in vivo, FNR is only nitrosylated when the major nitric oxide sensors (e.g. NsrR and NorR) and detoxification systems (e.g. NorVW, NrfA, and Hmp) are overwhelmed. Thus, FNR serves primarily as an O 2 sensor with a secondary nitric oxide sensing role (25). By contrast, the iron-sulfur clusters of regulators that are primarily nitric oxide sensors (e.g. NsrR and Wbl proteins) or redox sensors (e.g. SoxR) are generally stable for several hours in the presence of O 2 (29 -31). The data described above show that the three P. putida FNR proteins respond to O 2 in vitro and in vivo, suggesting that they are primarily O 2 sensors. To determine whether they also share the nitric oxide-or redox-responsive characteristics of E. coli FNR and SoxR, respectively, anaerobic cultures of P. putida expressing only one of the three FNR proteins and carrying the FNR-dependent pFF-41.5 fused to lacZ were supplemented with the nitric oxide donor NOC-7; in addition, aerobic cultures were exposed to the redox cycling agent paraquat. The responses of PP_3233 and PP_3287 were similar, nitric oxide had little or no effect under anaerobic conditions, and paraquat had no effect under aerobic conditions (Fig. 8). However, for ANR, nitric oxide significantly inactivated anaerobic reporter gene expression, whereas paraquat again had no effect under aerobic conditions (Fig. 8). Thus, the response of ANR was similar to that reported previously for E. coli FNR, further confirming the similarities between these two proteins, but PP_3233 and PP_3287 were less responsive with both O 2 and nitric oxide compared with ANR (25). Nevertheless, in all cases the greatest responses were provoked by culture aeration, and therefore, it was concluded that O 2 is the major modulator of the activity of all three P. putida FNR proteins.
Conclusions-The research described here suggests that the three FNR proteins of P. putida have evolved to fulfill distinct but overlapping roles. All three regulators, ANR, PP_3233, and PP_3287, acquired [4Fe-4S] clusters under anaerobic conditions and were converted to [2Fe-2S] forms upon exposure to O 2 in vitro. ANR has the least number of nonconservative amino acid substitutions in the vicinity of the cluster-ligating cysteine residues compared with E. coli FNR and hence was expected to exhibit similar cluster reactivity to FNR (Fig.  1) (14) in which the [2Fe-2S] 2ϩ cluster is ligated by one (Equations 5 and 6) or two (Equations 5 and 7) cysteine persulfides (CysSS).
Furthermore, ANR resembled FNR in exhibiting a secondary response when cultures were exposed to micromolar levels of nitric oxide (Fig. 8). Thus, the observations reported here are consistent with P. putida ANR acting as an E. coli-type O 2 sensor regulator, in accordance with its ability to regulate the expression of multiple terminal oxidases of the P. putida respiratory chain and the ability of the closely related P. aeruginosa anr gene (encoded protein 88% identical, 94% similar over 244 amino acids) to complement the anaerobic growth phenotype of an E. coli fnr mutant (32,33). The [4Fe-4S] clusters of PP_3233 and PP_3287 also underwent conversion to [2Fe-2S] clusters upon exposure to O 2 , but these reactions were slower than that of ANR, and the responses of these proteins when cultures were exposed to nitric oxide were weaker than that observed for ANR (Fig. 8). The kinetic data for the reaction of the PP_3233 and PP_3287 [4Fe-4S] clusters with O 2 fitted well to a single-exponential function rather than a double-exponential function, implying that, unlike FNR and ANR, the initial cluster oxidation step to generate the [3Fe-4S] 1ϩ intermediate (Equation 5) was much slower than the subsequent decay of the [3Fe-4S] 1ϩ to the [2Fe-2S] 2ϩ form (Equations 6 and 7). Thus, it is suggested that the mechanism for [4Fe-4S] to [2Fe-2S] cluster conversion in PP_3233 and PP_3287 was similar to that described for FNR (11) and ANR, but the PP_3233 and PP_3287 [4Fe-4S] clusters appear to be more stable when bacteria are exposed to air or nitric oxide. The relative rates of ANR, PP_3233, and PP_3287 cluster reactions with O 2 results in differential responses to O 2 availability.
The in vivo properties of ANR, PP_3233, and PP_3287 were consistent with the observed reactivities of the [4Fe-4S] clusters with O 2 . Previous work with E. coli FNR showed that replacement of Ser-24, which is located immediately adjacent to the cluster ligand Cys-23, by Arg resulted in significant aerobic FNR activity, indicative of stabilization of the FNR-S24R ironsulfur cluster (12). Interestingly, PP_3287 has Arg in the position equivalent to Ser-24 in FNR ( Figs. 1 and 2), and thus this amino acid substitution could at least partially account for the lower reactivity of PP_3287 with O 2 . On the other hand, PP_3233 resembles E. coli FNR by retaining a Ser residue at the equivalent of position 24 ( Figs. 1 and 2); however, amino acid substitutions in other locations are known to influence the reactivity of the E. coli FNR iron-sulfur cluster with O 2 (26,34). Like S24R, another amino acid substitution that promoted aer-FIGURE 7. Responses of P. putida FNR proteins to O 2 in vivo. A, the output from an FNR-dependent promoter decreases in response to enhanced aeration of P. putida cultures expressing only one of the three FNR proteins. All the strains were transformed with the FF-41.5-lacZ reporter plasmid pGS810. The rate of culture aeration was increased by decreasing the volume of medium in the shaking conical flasks (50 ml of medium for O 2 -limited cultures and 10 ml of medium for aerobic cultures). Cultures were grown at 30°C for 3 h, at which point samples were taken for measurement of ␤-galactosidase activity. The ␤-galactosidase activities of the aerobic cultures were divided by those of the O 2 -limited cultures. The error bars are the standard deviation from the mean values of the aerobic:O 2 -limited ratios (n ϭ 4). ANR, PP_3233, and PP_3287 indicate chromosomal expression of the corresponding genes; PP_3233 ϩϩ and PP_3287 ϩϩ indicate expression of the corresponding genes from a multicopy plasmid. B, concentration of cytoplasmic ANR does not respond to changes in culture aeration. Shown are Western blots developed with antiserum raised against E. coli FNR for cell samples from P. putida cultures that express only anr grown in shaking 50-ml conical flasks containing 10, 20, 30, 40, or 50 ml of medium (lanes 1-5) to impose an increasing degree of O 2 limitation on the cultures. The equivalent region of a Coomassie Blue-stained gel is shown as a loading control (M indicates protein standard markers: 37, 25, and 20 kDa, top to bottom). The outputs from the pFF-41.5 reporter (pGS810) for cultures grown as described above are shown below each lane (mean values Ϯ standard deviation, n ϭ 3). C, inactivation of FNR proteins upon exposure of anaerobic cultures to air. Cultures of E. coli JRG6348 expressing either no FNR (vector), E. coli FNR, P. putida ANR, PP_3233, or PP_3287, as indicated, were grown under anaerobic conditions and the abundance of FNR-protein-dependent lacZ transcription was measured by qRT-PCR. The cultures were exposed to air for 20 min, and then the abundance of the lacZ transcript was measured again. The relative abundance of lacZ mRNA after transfer to aerobic conditions is shown. The error bars are the standard deviation from the mean (n ϭ 3).

Responses of Three P. putida FNR Proteins to O 2
obic FNR activity was also located immediately adjacent to Cys-23, but this time on the other flank (D22G) (34). The equivalent position in PP_3233 is occupied by Ala (Figs. 1 and 2), and thus by analogy, replacement of the acidic Asp residue might alter the redox properties of the PP_3233 iron-sulfur cluster, such that it is less O 2 reactive.
Although the observations reported here resolve several aspects of the properties of the three FNR proteins possessed by P. putida KT2440, many questions remain, including: (i) What are the conditions encountered by P. putida that induce PP_3233 and PP_3287 target gene expression? (ii) Do the three P. putida FNR proteins control distinct but overlapping regulons, perhaps by making productive interactions with additional transcription factors or alternative sigma factors? (iii) What is the imperative for employing multiple FNR proteins to extend the range of O 2 -responsive gene expression? Further detailed biochemical and physiological studies are now required to address these questions and in so doing discern the mechanism of the observed differential sensitivities to O 2 of these closely related proteins and the broader implications for the control of gene expression in P. putida.