Properties and significance of apoFNR as a second form of air-inactivated [4Fe-4S]ÆFNR of Escherichia coli Stephanie Achebach 1, *, Thorsten Selmer 2, * and Gottfried Unden 1 1 Institut fu ¨ r Mikrobiologie und Weinforschung, Johannes Gutenberg-Universita ¨ t Mainz, Germany 2 Laboratorium fu ¨ r Mikrobiologie, Philipps-Universita ¨ t, Marburg, Germany Fumarate nitrate reductase regulator (FNR) is a cyto- plasmic O 2 sensor and regulator present in Escherichia coli and other bacteria [1–7]. The protein is required for transcriptional regulation of many genes of faculta- tive anaerobic metabolism in response to O 2 availabil- ity [8] and responds to O 2 concentrations in the medium as low as 1–5 mbar O 2 (about 1–5 lm O 2 ) [7,9,10]. The rate of O 2 diffusion into the bacteria is high compared with the low rates of O 2 consumption at the cytoplasmic membrane or within cytoplasm [3,7,9]. Therefore O 2 tension within the bacterial cyto- plasm appears to be similar to that in the external medium under oxic and microoxic conditions. Active FNR containing a [4Fe-4S] cluster, is found under anoxic conditions. In the presence of O 2 , [4Fe-4S]ÆFNR is converted to [2Fe-2S]ÆFNR resulting in monomerization and inactivation of FNR as a gene regulator. The [4Fe-4S] ⁄ [2Fe-2S] conversion is respon- sible for the anaerobic⁄ aerobic switch of FNR and has been demonstrated in vivo and in vitro [11–14]. Recently it became clear that the FeS cluster of [2Fe-2S]ÆFNR is also labile in vivo and in vitro in the presence of air and that FNR loses the [2Fe-2S] cluster [15,16]. The chemical reactions of O 2 and of peroxide causing cluster destruction have been studied [15,17]. ApoFNR, which is devoid of FeS clusters, is a further form of FNR and obtained during isolation of FNR and by prolonged exposure of [4Fe-4S]ÆFNR to air [18–20]. It is not known, however, whether apoFNR is a preparatory artifact, or whether its formation Keywords apoFNR; cysteine disulfides; Escherichia coli; fumarate nitrate reductase regulator; oxygen sensing Correspondence G. Unden, Johannes Gutenberg-Universita ¨ t Mainz, Institut fu ¨ r Mikrobiologie und Weinforschung, Becherweg 15, 55 099 Mainz, Germany Fax: +49 6131 3922695 Tel: +49 6131 3923550 E-mail: unden@uni-mainz.de *These authors contributed equally to the work (Received 1 June 2005, revised 24 June 2005, accepted 1 July 2005) doi:10.1111/j.1742-4658.2005.04840.x The active form of the oxygen sensor fumarate nitrate reductase regulator (FNR) of Escherichia coli contains a [4Fe-4S] cluster which is converted to a [2Fe-2S] cluster after reaction with air, resulting in inactivation of FNR. Reaction of reconstituted [4Fe-4S]ÆFNR with air resulted within 5 min in conversion to apoFNR. The rate was comparable to the rate known for [4Fe-4S]ÆFNR ⁄ [2Fe-2S]ÆFNR cluster conversion, suggesting that apoFNR is a product of [2Fe-2S]ÆFNR decomposition and a final form of air-inacti- vated FNR in vitro. Formation of apoFNR and the redox state of the cysteinyl residues were determined in vitro by alkylation. FNR contains five cysteinyl residues, four of which (Cys20, Cys23, Cys29 and Cys122) ligate the FeS clusters. Alkylated FNR and proteolytic fragments thereof were analyzed by MALDI-TOF. ApoFNR formed by air inactivation of [4Fe-4S]ÆFNR in vitro contained one or two disulfides. Only disulfide pairs Cys16 ⁄ 20 and Cys23 ⁄ 29 were formed; Cys122 was never part of a disulfide. The same type of disulfide was found in apoFNR obtained during isolation of FNR, suggesting that cysteine disulfide formation follows a fixed pattern. ApoFNR, including the form with two disulfides, can be reconsti- tuted to [4Fe-4S]ÆFNR after disulfide reduction. The experiments suggest that apoFNR is a major form of FNR under oxic conditions. Abbreviations CM, carboxymethyl; DTNB, 5,5¢-dithiobis(2-nitrobenzoate); FNR, fumarate nitrate reductase regulator; GnHCl, guanidinium hydrochloride; GST, glutathione S-transferase; TFA, trifluoroacetic acid. 4260 FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS is of functional relevance. The recent demonstration that [2Fe-2S]ÆFNR disappears within few minutes from cells of E. coli under exposure to air [15,16], could be an indication that significant amounts of apo- FNR are found in bacteria. The assumption prompted our studies to test whether apoFNR is formed at rates comparable to the air-induced [4Fe-4S] ⁄ [2Fe-2S] clus- ter conversion and the loss of [2Fe-2S] from FNR. Comparable rates would indicate that apoFNR is the product formed from [2Fe-2S]ÆFNR in vivo [15,16] and in vitro. Therefore the conditions and kinetics for the formation of apoFNR from active FNR were analyzed. Recently it has been shown that various forms of apoFNR exist in vitro which differ in the redox state of the cysteinyl residues and their availability for reconstitution to [4Fe-4S]ÆFNR [18]. Three of the cys- teine ligands of the FeS cluster of FNR are located at the N-terminal end (Cys20, Cys23 and Cys29), the fourth (Cys122) is close to the centre of the protein [21–23]. The fifth residue (Cys16) is not essential. The redox state of the cysteinyl residues of apoFNR should be relevant for function and properties of the protein. Therefore the redox state of the cysteinyl residues of the various forms of apoFNR – ‘aerobic’ and ‘anaer- obic’ apoFNR, and apoFNR obtained by air-induced inactivation of [4Fe-4S]ÆFNR – was determined in order to identify cysteinyl residues which are sensitive to oxidation and disulfide formation. ‘Aerobic’ and ‘anaerobic’ apoFNR are obtained by the isolation of the protein under oxic or anoxic conditions [18,22,24,25]. The cysteine disulfides in the three forms of apoFNR showed a distinct pattern of disulfides, whereas [4Fe-4S]ÆFNR (and [2Fe-2S]ÆFNR) contained no disulfides. Thus the disulfides in apoFNR can be used as a specific indicator for apoFNR formation from [4Fe-4S]ÆFNR during air inactivation. In this way it was possible to demonstrate that reaction of [4Fe-4S]ÆFNR with air generates apoFNR in vitro with rates comparable to [4Fe-4S] ⁄ [2Fe-2S] cluster conver- sion [16,17,26] suggesting that apoFNR is the final form of FNR under oxic conditions. Results Redox state of the cysteinyl residues in apoFNR and in reconstituted [4Fe-4S]ÆFNR ApoFNR that was devoid of Fe and acid-labile sulfur (not shown) was obtained by isolation of FNR from E. coli under oxic and anoxic conditions (‘aerobic’ or ‘anaerobic’). In SDS gel electrophoresis without redu- cing agents, anaerobic apoFNR formed one band with the typical molecular mass of FNR ( M r 30 000). Aero- bic apoFNR showed an additional band of M r 27 000 (not shown) as described earlier [22]. The latter form disappeared when the sample was incubated with dithiothreitol or 2-mercaptoethanol, whereas the M r 30 000 protein was not affected. In aerobic and anaerobic apoFNR 2.8 and 4.0 from the total of five cysteinyl residues reacted with 5,5¢-dithiobis-nitro- benzoate (DTNB) (Table 1) or iodoacetate after dena- turing by guanidinium hydrochloride (GnHCl), and were thus in the thiol state. This suggests that aerobic apoFNR contains a portion with an (additional) disul- fide which has a more condensed structure and higher mobility in SDS gel electrophoresis. After reaction with iodoacetic acid, the carboxy- methyl derivatives were analyzed by MALDI-TOF mass spectroscopy. Aerobic and anaerobic apoFNR showed a molecular mass of 28 112 ± 14 Da before modification (not shown) which is close to the predic- ted molecular mass of the recombinant FNR of 28112 Da (FNR plus N-terminal Gly-Ser extension). The alkylated aerobic apoFNR produced three major peaks of 28 169 Da, 28 284 Da, and 28 398 Da (Fig. 1A), corresponding to one-, three- and fivefold alkylated FNR, as each carboxymethyl (CM) residue increases the molecular mass by 58 Da. No unmodified apoFNR was detected. Threefold alkylated apoFNR showed the most intense signal. The MALDI-TOF spectrum of anaerobic apoFNR consisted of one major signal at 28 408 Da after alkylation equivalent to five- fold alkylated apoFNR, and a minor signal of three- fold alkylated FNR (Fig. 1B). Therefore, aerobic and anaerobic apoFNR are mixtures of FNR with one, three and five, and five and three cysteine thiol resi- dues, respectively. In reconstituted [4Fe-4S]ÆFNR up to four cysteinyl residues reacted with DTNB (Table 1). The residues Table 1. Thiol and disulfide cysteinyl residues in apoFNR (aerobi- cally or anaerobically prepared), reconstituted [4Fe-4S]ÆFNR, and air- inactivated [4Fe-4S]ÆFNR. The contents of cysteine thiols were determined with DTNB or [ 14 C]iodoacetate. The number of cystei- nyl residues present in the oxidized state (‘Disulfide S’) were calcu- lated as the difference to the total of five cysteinyl residues present in FNR. The thiol contents are the mean of three experi- ments (max. deviation 25%). Thiol (mol ⁄ mol FNR) Disulfide S Aerobic apoFNR 2.8 2.2 Anaerobic apoFNR 4.0 1.0 Reconstituted [4Fe-4S]ÆFNR a 4.0 1.0 Air-inactivated [4Fe-4S]ÆFNR a 2.2 2.8 a Fig. 7. S. Achebach et al. Disulfides of apoFNR FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 4261 were accessible to labeling only after denaturing of the protein. In MALDI-TOF analysis, the alkylated pro- tein gave one major signal of 28 395 Da corresponding to fivefold alkylated FNR (FNR-CM 5 , theoretical molecular mass 28 402 Da) (not shown). Four- and threefold alkylated FNR was found only in traces. In this way, [4Fe-4S]ÆFNR (and presumably [2Fe- 2S]ÆFNR) differ clearly from apoFNR by the lack of cysteinyl disulfides which can be used for differenti- ation of apoFNR and [4Fe-4S]ÆFNR in vitro by labe- ling of cysteinyl thiol residues with iodoacetate after denaturing the protein. Kinetics of cysteinyl residue oxidation during inactivation of [4Fe-4S]ÆFNR by air. Reconstituted [4Fe-4S]ÆFNR was exposed to air, and after various times samples were analyzed for the number of reduced cysteinyl residues by reacting the protein with [ 14 C]iodoacetate (Fig. 2). GnHCl was included at the same time to destabilize the FeS clusters and to make the cysteinyl residues accessible to alkylation. In anae- robic, reconstituted FNR about four cysteinyl residues were labeled by this method. After addition of air, the amount of reactive cysteinyl residues increased slightly for about 2 min. Then the amount decreased to about two residues within the next 3 min, corresponding to an average of three cysteinyl residues in the oxidized state. Under anoxic conditions no change in the num- ber of reactive cysteine residues was observed. Exten- ded incubation with air gave no further decrease of reactive thiols. The result demonstrates that, after a lag phase of about 2 min, air causes formation of disulfides in FNR which is finished within about 3 min. Formation of the disulfides has to be accom- panied by the loss of the FeS cluster of FNR due to the loss of two or four ligands of the FeS cluster. This type of kinetics was only obtained when excess Fe ions were removed by EDTA, presumably due to reaction of Fe ions with cysteinyl residues. Defined cysteine disulfides in aerobic and anaerobic apoFNR Alkylated forms of aerobic and anaerobic apoFNR were digested by proteases, and the peptides were iso- lated and characterized by MALDI-TOF. Modified trypsin cleaves the protein at the C-terminal site of arginine and lysine, and the tryptic digest of FNR is predicted to contain the four N-terminal Cys residues in the peptide comprising amino acids 11–48 (Pep11 ⁄ 48). Cys122 is found in a peptide of amino acids 78–135 (Pep78 ⁄ 135). The tryptic digest of aerobic apoFNR contained unmodified Pep11 ⁄ 48 (4258 Da) and two derivatives with two and four carboxymethyl groups, respectively (Fig. 3A). Pep78 ⁄ 135 with Cys122 Fig. 1. MALDI-TOF spectra of aerobically (A) and anaerobically (B) prepared and carboxymethylated apoFNR. The samples of apoFNR were alkylated with 10 m M iodoacetate under denaturing conditions in 4 M GnHCl, purified by solid phase extraction and subjected to MALDI-TOF MS analysis. The spectra were taken using sinapinic acid as matrix in the linear mode of the instrument (for details see Experimental procedures). Fig. 2. Measurement of reactive cysteinyl residues during air-inacti- vation of reconstituted [4Fe-4S]ÆFNR. Aerobic apoFNR was reconsti- tuted, transferred to a Petridish and incubated under air (0 min). At the given time-points, samples were removed and denatured in a solution of GnHCl (4 M)and[ 14 C]iodoacetate (12 mM and 43 BqÆ mmol )1 iodoacetate). The protein was precipitated and washed with methanol + chloroform [31] and used for measurement of pro- tein and radioactivity (mean of three independent experiments). Disulfides of apoFNR S. Achebach et al. 4262 FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS was found only as the alkylated form. No adduct of Pep11 ⁄ 48 with Pep78 ⁄ 135 was present, ruling out the presence of a disulfide bridge between both peptides. Thus aerobic apoFNR is a mixture of proteins with zero, one or two disulfide bonds within the four N-terminal cysteinyl residues. When anaerobic apoFNR was digested with trypsin, only four- and twofold alkylated, but no unmodified Pep11 ⁄ 48 was found (Pep11 ⁄ 48-CM 4 and Pep11 ⁄ 48-CM 2 ) (Fig. 3B). Protease AspN cleaves N-terminal to aspartate By this cleavage, Pep1 ⁄ 21 containing Cys16 and Cys20, Pep22 ⁄ 39 with Cys23 and Cys29, and Pep102 ⁄ 129 con- taining Cys122 were generated (Table 2). When the peptides were produced from aerobic apoFNR after alkylation, Pep22 ⁄ 39 was found as the unmodified and, in small amounts, also as the twofold alkylated species. Pep1 ⁄ 21 was present in the unmodified state, as well as modified with two alkyl residues. Thus Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29 formed disulfides in aerobic apoFNR. Pep102 ⁄ 129, on the other hand, was found only in the alkylated form, demonstrating again that Cys122 was not involved in disulfide formation. No peptides corresponding to the combination of Pep1 ⁄ 21, Pep22 ⁄ 39 or Pep102 ⁄ 129 were found, and thus only disulfides Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29 were present within aerobic apoFNR. Pep1 ⁄ 21 and Pep22 ⁄ 39 were detected from anaerobic apoFNR in the unmodified as well as the twofold alkylated form and, therefore, Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29 could also be responsible for apoFNR with one disulfide. Cys23 ⁄ 29, however, was found at significantly lower levels than in aerobic apoFNR, sug- gesting that the Cys16 ⁄ 20 disulfide is formed preferen- tially to the Cys23 ⁄ 29 disulfide. Quantification of disulfide and thiol containing peptides after HPLC separation The peptides of the AspN digest with the cysteinyl residues in the disulfide or thiol state were quantified Fig. 3. MALDI-TOF spectra of peptides derived from carboxymeth- ylated aerobic (A) and anaerobic (B) apoFNR (trypsin digest). The samples (carboxymethylated tryptic digest of aerobic or anaerobic apoFNR) were purified by solid phase extraction and analyzed by MALDI-TOF MS in the reflector mode of the instrument using a-cyano-4-hydroxycinnamic acid as matrix. Masses (m ⁄ z)of cysteinyl residues containing tryptic peptides Pep11 ⁄ 48 (with Cys16, Cys20, Cys23, Cys29) and Pep78 ⁄ 135 (with Cys122): Pep11 ⁄ 48-CM 0 4258 in (A) (predicted 4259); Pep11 ⁄ 48-CM 2 4375 in (A) and 4374 in (B) (predicted 4375); Pep11 ⁄ 48-CM 4 4494 in (A) and 4493 in (B) (predicted 4491); Pep78 ⁄ 135-CM 1 6248 in (A) and 6244 in (B) (predicted 6244). The predicted masses were obtained from the mass of the peptide plus the mass increase of 58 after alkylation per cysteinyl residues by iodoacetic acid. Table 2. Mass of fragments of aerobically and anaerobically prepared apoFNR after treatment with iodoacetic acid and digestion with prote- ase AspN. The numbering of the peptides gives the first and the last amino acid residue according to numbering in FNR. The mass of the fragments and of their alkylated forms were determined by MALDI-TOF spectroscopy. Pep1 ⁄ 21 (with Cys16 and Cys20) contains in addition the two N-terminal Gly-Ser residues derived from the GST¢-¢FNR fusion. Pep22 ⁄ 39 contains Cys23 and Cys29, Pep102 ⁄ 129 Cys122. The cal- culated masses were obtained from the mass of the peptide and a mass increase of 58 after alkylation by iodoacetic acid per cysteinyl resi- due. ND, not detected. Cys-containing fragments Thiol Disulfide Mass for alkylated peptides of apoFNR Calculated (m ⁄ z) Aerobic (m ⁄ z ± 3) Anaerobic (m ⁄ z±3) Pep1 ⁄ 21-CM 2 2 0 2671 2674 ND a Pep1 ⁄ 21-CM 0 0 1 2555 2556 ND a Pep22 ⁄ 39-CM 2 2 0 2179 2181 2181 Pep22 ⁄ 39-CM 0 0 1 2063 2062 2062 Pep102 ⁄ 129-CM 1 1 0 3047 3049 3050 a Fragments can be detected after reversed phase HPLC (Table 3). S. Achebach et al. Disulfides of apoFNR FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 4263 after alkylation and proteolytic digestion. The peptides were separated by reversed-phase HPLC chromatogra- phy (Fig. 4), and the peptides in the peaks were identi- fied by MALDI-TOF. The amount of peptide was quantified from the peak area of the HPLC eluate (Fig. 4). The peptides Pep11 ⁄ 21 (Cys16, Cys20) and Pep22 ⁄ 39 (Cys23, Cys29) from aerobic and anaerobic apoFNR were found in the noncarboxymethylated (disulfide) form as well as with two carboxymethyl groups (Fig. 4). The peptide with Cys16 ⁄ Cys20 was present to 65% in the oxidized (disulfide) form in aero- bic and in anaerobic apoFNR (Table 3). The 23 ⁄ 29 disulfide in Pep22 ⁄ 39 was significantly increased in aerobic compared with anaerobic apoFNR (Table 3) which represented the only distinct differences between both forms of apoFNR. Generally, a smaller portion of the Cys23 ⁄ Cys29 than of the Cys16 ⁄ Cys20 residues was in the disulfide state, suggesting that the Cys16 ⁄ 20 disulfide is formed preferentially. Redox state of cysteinyl residues during inactivation of [4Fe-4S]ÆFNR with air The redox state of the cysteinyl residues of [4Fe-4S]ÆFNR after inactivation by air (Fig. 2) was determined. After reaction with iodoacetate, FNR of molecular mass 28 262 Da was found which most prob- ably represents FNR with one disulfide and three thiol residues (theoretical molecular mass 28 286 Da). Devia- tion of experimental and expected molecular mass by 24 Da is due to a variation in absolute mass (± 28 Da) by nonspecific absorption of ions. Using [ 14 C]iodoace- tate, FNR with two labeled cysteine thiols was identified after inactivation with air (Fig. 2 and Table 1). Alto- gether, the experiments show that air-inactivated [4Fe-4S]ÆFNR is a mixture of apoFNR with one disul- fide plus three cysteine thiol residues, and FNR with two disulfides and one cysteine thiol. The air inactivated and alkylated FNR of the experi- ment from Fig. 2 was digested with AspN or trypsin and analyzed by MALDI-TOF for the presence of alkylated peptides. In AspN digests, peptides Pep1 ⁄ 21 and Pep22 ⁄ 39 with 2555 and 2062 Da were found which are characteristic of the presence of disulfides Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29, respectively (not shown). In the tryptic digest, peptides with masses indicative for disulfides were found (Fig. 5), suggesting that up to two disulfides are formed by air-inactivation of FNR in vitro. This again shows that the air-inacti- vated FNR is a mixture of apoFNR with one and two disulfides, and the disulfides are the same as in aerobic apoFNR (Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29). Discussion Significance of apoFNR [4Fe-4S]ÆFNR is converted rapidly to apoFNR after exposure to air and a lag phase of about 2 min. The lag phase presumably includes the [4Fe-4S] ⁄ [2Fe-2S] conversion, which has no direct effect on the redox state of the cysteinyl residues. The rate of apoFNR formation is similar to that of [4Fe-4S] ⁄ [2Fe-2S] con- version, indicating that apoFNR is a significant prod- uct of FNR inactivation by air. Fig. 4. Peptides of aerobic apoFNR after carboxymethylation, tryp- sin and AspN digestion and reversed-phase chromatography. The sample buffer was changed by gel filtration on a Nap10 column and then subjected to trypsin digestion. The digest was separated by gel filtration on a Superdex Peptide column. The fraction contain- ing peptide 11 ⁄ 48 (containing Cys16, Cys20, Cys23, Cys29) was digested with AspN and separated by reversed phase chromatogra- phy. To identify the peptides in the individual chromatographic frac- tions, each fraction was subjected to MALDI-TOF analysis. The corresponding peptides in the fractions are indicated. Table 3. Quantitative evaluation of thiol and disulfide containing pep- tides of aerobically and anaerobically prepared apoFNR. Aerobically or anaerobically prepared apoFNR were incubated with GnHCl + iodoacetate and digested with trypsin, and after separation on Sepha- dex Peptide column, by AspN. The desalted peptides were separated by reversed phase HPLC chromatography (AquaPore RP300) (Fig. 4). Relevant peak fractions were subjected to MALDI-TOF to identify the peptides and the degree of alkylation. The relative contents of the disulfide and the dithiol forms of one peptide from one chromato- gram was evaluated from the absorption of the corresponding peaks and the peak areas from three independent experiments and samples. The data are the mean of three independent experiments (max deviation 10%). Peptide Aerobic apoFNR Anaerobic apoFNR Pep-SS Pep-CM 2 Pep-SS Pep-CM 2 Pep11 ⁄ 21 (Cys16, Cys20) 65 35 65 35 Pep22 ⁄ 39 (Cys23, Cys29) 36 64 22 78 Pep102 ⁄ 129 (Cys122) 0 100 0 100 Disulfides of apoFNR S. Achebach et al. 4264 FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS [4Fe-4S]ÆFNR shows a complex response to the pres- ence of air (Fig. 6 gives an overview), and the response differs in vitro and in vivo. The reactions of the FeS cluster-containing forms of FNR were studied by Mo ¨ ssbauer spectroscopy, whereas information on apo- FNR was obtained via the redox state of the cysteine residues. [4Fe-4S]ÆFNR is converted to [2Fe-2S]ÆFNR under oxic conditions within few minutes [13,16,17,26], and the slower reaction in vivo is due to experimental conditions [13]. It has been estimated from expression studies that inactivation of FNR by air (which compri- ses the [4Fe-4S]ÆFNR ⁄ [2Fe-2S]ÆFNR conversion) takes place within 4 min 30 s [16], which supports the more rapid in vitro data. [2Fe-2S]ÆFNR was shown to be rather stable in vitro in the presence of air, whereas H 2 O 2 caused a rapid disintegration of the cluster [15,16; Fig. 6]. In vivo, the [2Fe-2S] cluster disappeared within 15 min (or less) in the presence of air. As a result, aerobically grown E. coli contains no [2Fe- 2S]ÆFNR or ([4Fe-4S]ÆFNR) [16]. Aerobically grown E. coli contains FNR protein in amounts which are similar to, or even higher than, anaerobically grown bacteria [15,27]. It is suggested that the FNR protein present under these conditions consists mainly of apo- FNR. In aerobically grown E. coli,O 2 tension appears to be similar to that in the external growth medium [7,9,10]. Therefore, with respect to O 2 supply and effective O 2 tension, the in vivo conditions are compar- able to the in vitro conditions used here for FNR inactivation. Fig. 5. Redox state of Cys residues in Pep78 ⁄ 135 (A) and Pep11 ⁄ 48 (B) of recon- stituted [4Fe-4S]ÆFNR after air-inactivation trypsin digestion. Reconstituted and subse- quently air-inactivated [4Fe-4S]ÆFNR was denatured in 4 M GnHCl and alkylated with iodoacetate. After solid phase extraction the protein was digested by trypsin, separated by gel filtration, and analyzed by MALDI- TOF MS. (A) Pep78 ⁄ 135, containing Cys122, with one CM-group (predicted mass 6244 Da) (B) Pep11 ⁄ 48 in the unmodi- fied form (predicted mass 4259 Da) contain- ing Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29 as the disulfides. Peptide of mass 4411.9 corres- ponds to Pep10 ⁄ 48 which contains Arg10 in addition (predicted mass 4416) which is obtained with low yield under some conditions. Fig. 6. Reaction steps and time required for conversion of [4Fe-4S]ÆFNR to apoFNR. The scheme shows species of FNR involved in the conversion to apoFNR, species not directly meas- ured for the respective reaction are given in broken lines. The quoted times are either t 1 ⁄ 2 values or the times required for con- version between both forms. The corresponding references are given in brackets. S. Achebach et al. Disulfides of apoFNR FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 4265 ApoFNR is a product of the reaction of ([4Fe-4S]ÆFNR) with air The in vitro rate for the conversion of [4Fe-4S]ÆFNR to apoFNR (via [2Fe-2S]ÆFNR) is either similar to the rates shown for the cluster conversions or more rapid (Fig. 6). ApoFNR formation is the counterpart to the disappearance of [2Fe-2S]ÆFNR in the bacteria [16]. The presence of significant amounts of apoFNR in the bacteria has to be demonstrated, but preliminary results indicate that aerobically grown E. coli contain significant amounts of apoFNR (S. Achebach and G. Unden, unpublished results). ApoFNR containing high levels of disulfides can be used successfully for reconstitution of [4Fe-4S]ÆFNR in vitro [18], further supporting the physiological significance of apoFNR in bacteria. Formation of apoFNR represents a second step of FNR inactivation and transfers FNR to a functional state which is less sensitive to reactivation by short- term or partial anoxic conditions than [2Fe-2S]ÆFNR. Desensitizing of FNR would be significant for E. coli growing on a long-term basis under oxic conditions to avoid production of anaerobic metabolic systems in response to short-term air depletion. Different forms of apoFNR The forms of apoFNR obtained during aerobic or anaerobic preparation and by in vitro inactivation by air differed by the redox state or contents of Cys16 ⁄ Cys20 and Cys23 ⁄ Cys29 disulfides (Fig. 7). The cysteinyl residues were not oxidized by air to sulfenic or sulfonic acids, as peptides with the corresponding mass increases were not detected. Formation of sulfen- ic and sulfonic acid to a larger extent is also unlikely as the protein can be reconstituted to active FNR using sulfhydryl reducing agents like dithiothreitol. The cysteinyl residues play a central role in the func- tion of FNR, and their redox state was tested by chemical modification and MALDI-TOF mass spec- troscopy which gave consistent results. The cysteine disulfides could be used as indicators for apoFNR formation, for which other good biochemical markers are not available. In the presence of reducing agents such as dithiothreitol, or glutathione as in the cyto- plasm of E. coli, the cysteine disulfides are expected to exist in the thiol state. Knowledge of the type of cysteinyl disulfides in FNR provides important information about the bio- chemistry and properties of the protein, the function of which relies mostly on the cysteine residues and the FeS clusters bound by the cysteine residues. Cysteinyl disulfides were formed only between distinct pairs of residues (Cys16 ⁄ 20 and Cys23 ⁄ 29), whereas Cys122 was never part of a disulfide (Fig. 7). This could be related to a specific role of Cys122 which is also essential for function and cluster assembly [28]. FNR-SS-(SH) 3 , which contains one disulfide, was a mixture of proteins with either the Cys16 ⁄ Cys20 or the Cys23 ⁄ Cys29 disulfide. The Cys16 ⁄ Cys20 disulfide is formed preferentially, but there is no clear order in the formation of the two disulfides. Formation of the pairs might reflect suitable spatial vicinity, lacking accessibil- ity of Cys122, or functional differences of the cysteinyl residues. Aerobic apoFNR with high disulfide content [presumably due to the presence of FNR-(SS) 2 -SH] was shown earlier [18] to have an extended lag phase Fig. 7. Schematic presentation of FNR and of cysteinyl thiol and disulfide residues in different forms of FNR. The scheme shows the cysteine thiols and disulfides in different forms of FNR, and their approximate contri- bution to different functional forms of FNR (active [4Fe-4S]ÆFNR, air-inactivated FNR, aerobic and anaerobic apoFNR). The cystei- nyl residues (Cys16, Cys20, Cys23, Cys29, Cys122) are presented by their thiol (–SH) or disulfide (–SS–) residues; FNR-SS-(SH) 3 , e.g. stands for apoFNR with one disulfide and three cysteine thiol residues. –, not pre- sent; +, present; ++, major component. Disulfides of apoFNR S. Achebach et al. 4266 FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS in the reconstitution of [4Fe-4S]ÆFNR from apoFNR in vitro. The increased lag phase could be overcome by the addition of dithiothreitol and protein disulfide reductases. Experimental procedures Isolation and reconstitution of FNR FNR was produced and isolated as a GST¢-¢FNR fusion protein from E. coli CAG627pMW68 [24]. GST¢-¢FNR bound to glutathione-Sepharose 4B (1.5 mL bed volume) was digested for 2 h at 20 °C with 20 U thrombin in 1 mL buffer C (50 mm Tris ⁄ HCl, pH 7.6) and then eluted from the column in 3 mL buffer C (without addition of glutathi- one). For the isolation of anaerobic apoFNR, anoxic buff- ers prepared and maintained in an anaerobic chamber were used and the whole procedure starting with incubation of the bacteria was performed under anoxic conditions [24]. Aerobic apoFNR was prepared in the same way, but all buffers were air saturated and all steps were performed under air. FNR obtained from GST-FNR in this way con- tains a Gly-Ser extension in front of the N-terminal Met- Ile-Pro of wild-type FNR. Azotobacter vinelandii (NifS AV ) was isolated from E. coli BL21(DE3) containing a nifS AV expression plasmid for NifS AV production [24,29]. [4Fe-4S]ÆFNR was reconstituted under anoxic conditions [3,24] in a mixture containing isolated apoFNR, 0.3 mm Fe(II), 2 mm cysteine and NifS AV (1 lg ⁄ 20 lg FNR). The number of thiol groups was determined with 5,5¢-dithiobis- nitrobenzoate (DTNB) [30] or by the incorporation of [ 14 C]iodoacetate. Carboxymethylation of FNR For alkylation of anaerobic apoFNR, 250–350 lg FNR were anoxically incubated for 30 min with 4 m guanidinium hydrochloride (GnHCl) and 10 mm iodoacetate, buffer (25 mm Tris ⁄ HCl, pH 7.6), at room temperature in the dark. Prior to MALDI-TOF MS, low molecular mass con- taminations were removed by solid phase extraction. Alky- lation of aerobic apoFNR was performed in the same way, but all buffers were air saturated and the experiments were done under air. AspN digest Carboxymethylated protein was desalted by gel filtration using Nap10 columns (Amersham Pharmacia Biotech, Pis- cataway, NJ, USA) equilibrated with 1 m GnHCl in 50 mm ammonium acetate, pH 8.0. The sample was added and frac- tions of 500 lL each were collected. Fractions containing the protein were pooled and digested with 1% (w ⁄ w) endo- proteinase AspN (sequencing grade, Roche, Indianapolis, IN, USA) for 5 h at 37 °C which cleaves at the N-terminal end of aspartyl residues. The samples were desalted by solid phase extraction and subsequently dried by vacuum centrifugation. Trypsin digest The carboxymethylated protein was desalted by gel filtra- tion on a Nap10 column equilibrated with 10% (v ⁄ v) aceto- nitrile in 50 mm ammonium acetate, pH 8.6. The eluted protein was combined and digested with 7–9% (w ⁄ w) modi- fied trypsin (sequencing grade, Roche) for 8 h at 37 °C which cleaves at the C-terminal end of arginyl or lysyl resi- dues. After digest, the peptides were dried by vacuum cen- trifugation and stored at )20 °C. Time dependent cysteinyl accessibility during air-inactivation of [4Fe-4S]ÆFNR Aerobically prepared apoFNR was reconstituted under anoxic conditions as described above. Reconstitution was followed spectroscopically at 420 nm using the absorbance at 280 nm as reference. After complete reconstitution, 5mm EDTA was added. After 30–50 min the sample (0.9 mL) was placed into a Petri dish which was transferred to an atmosphere of air with gentle shaking. Samples (50 lL) were withdrawn prior and after 2 min, 5 min, 10 min and 20 min of aeration, mixed immediately with 50 lLof8m GnHCl (final concentration 4 m) and [ 14 C]iodoacetate (final concentration 12 mm, specific radio- activity 43 BqÆmmol )1 , etc.). After 10 h the protein was pre- cipitated and washed carefully with methanol–chloroform [31]. The protein concentration was determined using the Bradford assay [32] and the radioactivity was measured in a scintillation counter. Radioactivity was corrected for not incorporated radioactivity by measuring and subtracting the radioactivity in identically treated samples lacking apo- FNR. Solid phase extraction Protein or peptide samples were applied to C 18 -Sep Vac 1cc 50 mg cartridges (Waters, Milford, MA, USA) equilibrated with 0.1% trifluoroacetic acid (TFA). The columns were washed with 0.1% TFA and proteins or peptides were eluted with 60% (v ⁄ v) acetonitrile in 0.1% (v ⁄ v) TFA. HPLC separation and quantification of the peptides The carboxymethylated protein was subjected to solid phase extraction, eluted, dried by vacuum centrifugation and dissolved in 0.1% (v ⁄ v) TFA and 4 m GnHCl. The samples were desalted by gel filtration on NAP 10 columns S. Achebach et al. Disulfides of apoFNR FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 4267 equilibrated with 10% (v ⁄ v) acetonitrile in 50 mm ammo- nium acetate, pH 8.6. The protein was then digested with 5% (w ⁄ w) modified trypsin (sequencing grade, Roche) for 11 h at 37 °C. The trypsin digest was subjected to HPLC separation over a gel filtration column (Superdex TM Peptide 10 ⁄ 30) operated in 0.1% (v ⁄ v) TFA in 10% (v ⁄ v) acetonit- rile in order to separate the large (4 and 6 kDa) cysteine- containing from the bulk peptides (< 3 kDa). Fractions containing these peptides were identified by mass spectro- metry, pooled and dried. The peptides were re-dissolved in 1 m GnHCl in 20 mm Tris ⁄ HCl, pH 8, containing 2 lgof endoproteinase AspN. The samples were digested for 9 h at 37 °C. The resulting peptides were separated by reversed phase HPLC on an Aquapore RP 300 C 8 column (2.1 · 100 mm) and monitored at 215 nm. Individual pep- tides were collected manually and analyzed by MALDI- TOF MS. MALDI-TOF MS Mass spectra of proteins and peptides were collected using a Perkin-Elmer Voyager-RP ⁄ DE mass spectrometer (Per- septive Biosystems, Wiesbaden, Germany). In order to avoid an interfering binding of salts to FNR, the protein was carefully desalted by solid-phase extraction prior to MALDI-TOF MS. Samples thus obtained were either dissolved in 67% (v ⁄ v) acetonitrile in 0.1% (v ⁄ v) TFA or matrix solutions yielding final concentrations of 1–10 pmolÆ lL )1 protein or peptides. Equal volumes of samples and saturated solutions of either sinnapinic acid (proteins) or a-cyano-4-hydroxycinnamic acid in 0.1% (v ⁄ v) TFA, 67% (v ⁄ v) acetonitrile were mixed on the sample slide and dried in a stream of air. Proteins were measured in the positive linear mode of the instrument at an acceleration voltage of 25 kV using a grid voltage of 89% and a delay time of 300 ns. Peptide spectra were collected in the positive reflec- tor mode at an acceleration voltage of 20 kV using a grid voltage of 58% and a delay time of 100 ns. Acknowledgements The authors thank R. Thauer for the access to MALDI-TOF MS at the Max-Planck-Institut fu ¨ r Ter- restrische Mikrobiologie, Marburg. The work was sup- ported by grants of Deutsche Forschungsgemeinschaft and of the Fonds der Chemischen Industrie. References 1 Spiro S & Guest JR (1990) FNR and its role in oxygen- regulated gene expression in Escherichia coli. FEMS Microbiol Rev 75, 399–428. 2 Unden G, Bongaerts J, Becker S, Holighaus G, Schi- rawski J & Six S (1995) O 2 -sensing and O 2 -dependent gene regulation in facultatively anaerobic bacteria. Arch Microbiol 164, 81–90. 3 Schmitz R, Achebach S & Unden G (2004) Analysis of FNR as O 2 sensor in Escherichia coli.InMethods in Enzymology, Oxygen sensing, Vol. 381 (Sen CK & Semenza GL, eds), pp. 628–644. Academic Press Inc, New York. 4 Guest JR, Green J, Irvine A & Spiro S (1996) The FNR modulon and FNR-regulated gene expression. In Regu- lation of Gene Expression in Escherichia coli (Lin ECC & Lynch AS, eds), pp. 317–342. Chapman & Hall, New York. 5 Gunsalus RP (1992) Control of electron flow in Escheri- chia coli: coordinated transcription of respiratory path- way genes. J Bacteriol 174, 7069–7074. 6Ko ¨ rner H, Sofia HJ & Zumft WG (2003) Phylogeny of the bacterial superfamily of Crp-Fnr transcription regu- lators: exploiting the metabolic spectrum by controlling alternative gene programs. FEMS Microbiol Rev 27, 559–592. 7 Unden G & Schirawski J (1997) The O 2 responsive transcriptional regulator FNR of Escherichia coli: The search for signals and reactions. Mol Microbiol 25, 205– 210. 8 Salmon K, Hung SP, Mekjian K, Baldi P, Hatfield GW & Gunsalus RP (2003) Global gene expression profiling in Escherichia coli K-12: the effects of oxygen availabil- ity and FNR. J Biol Chem 278, 29837–29855. 9 Becker S, Holighaus G, Gabrielczyk T & Unden G (1996) O 2 as the regulatory signal for FNR-dependent gene reg- ulation in Escherichia coli. J Bacteriol 178, 4515–4521. 10 Tseng CP, Albrecht J & Gunsalus RP (1996) Effect of microaerophilic cell growth conditions on expression of the aerobic (cyoABCD and cydABC) and anaerobic (narGHJI, frdABCD, and dmsABC) respiratory pathway genes in Escherichia coli. J Bacteriol 178, 1094–1098. 11 Kiley PJ & Beinert H (1998) Oxygen sensing by the glo- bal regulator FNR: the role of the iron–sulfur cluster. FEMS Microbiol Rev 22, 341–352. 12 Khoroshilova N, Popescu C, Mu ¨ nck E, Beinert H & Kiley PJ (1997) Iron–sulfur cluster disassembly in the FNR protein of Escherichia coli by O 2 : [4Fe4S] to [2Fe2S] conversion with loss of biological activity. Proc Natl Acad Sci USA 94, 6087–6092. 13 Popescu CV, Bates DM, Beinert H, Mu ¨ nck E & Kiley P (1998) Mo ¨ ssbauer spectroscopy as a tool for the study of activation ⁄ inactivation of the transcription regulator FNR in whole cells of Escherichia coli. Proc Natl Acad Sci USA 95, 13431–13435. 14 Green J, Bennett B, Jordan P, Ralph ET, Thomson AJ & Guest JR (1996) Reconstitution of the [4Fe4S] cluster in FNR and demonstration of the aerobic– anaerobic transcription switch in vitro. Biochem J 316, 887–892. Disulfides of apoFNR S. Achebach et al. 4268 FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 15 Sutton VR, Stubna A, Patschkowski T, Mu ¨ nck E, Beinert H & Kiley PJ (2004) Superoxide destroys the [2Fe-2S] 2+ cluster of FNR from Escherichia coli. Biochemistry 43, 791–798. 16 Sutton VR, Mettert EL, Beinert H & Kiley PJ (2004) Kinetic analysis of the oxidative conversion of the [4Fe- 4S] 2+ cluster of FNR to a [2Fe-2S] 2+ cluster. J Bacteriol 186, 8018–8025. 17 Crack J, Green J & Thomson AJ (2004) Mechanism of oxygen sensing by the bacterial transcription factor fumarate–nitrate reduction (FNR). J Biol Chem 279, 9278–9286. 18 Achebach S, Tran QH, Vlamis-Gardikas A, Mu ¨ llner M, Holmgren A & Unden G (2004) Stimulation of Fe-S cluster insertion into apoFNR by Escherichia coli glu- taredoxins 1, 2, and 3 in vitro. FEBS Lett 565, 203–206. 19 Achebach S, Zeuner Y & Unden G (2003) Control of the O 2 -sensor ⁄ regulator FNR of Escherichia coli by O 2 and reducing agents in vivo and in vitro.InRegulatory Networks in Prokaryotes (Du ¨ rre P & Friedrich B, eds), pp. 93–99. Horizon Scientific Press, Norfolk, UK. 20 Unden G, Achebach S, Holighaus G, Tran HQ, Wack- witz B & Zeuner Y (2002) Control of FNR function of Escherichia coli by O 2 and reducing conditions. J Mol Microbiol Biotechnol 4, 263–268. 21 Melville SB & Gunsalus RP (1990) Isolation of an oxygen-sensitive FNR protein of Escherichia coli: inter- action at activator and repressor binding sites of FNR-controlled genes. J Biol Chem 265, 18733–18736. 22 Green J, Sharrocks AD, Green B, Geisow M & Guest JR (1993) Properties of FNR proteins substituted at each of the five cysteine residues. Mol Microbiol 8, 61–68. 23 Sharrocks AD, Green J & Guest JR (1990) In vivo and in vitro mutants of FNR the anaerobic transcriptional regulator of E. coli. FEBS Lett 270, 119–122. 24 Tran QH, Arras T, Becker S, Holighaus G, Ohlberger G & Unden G (2000) Role of glutathione in the forma- tion of the active form of the oxygen sensor FNR ([4Fe4S]ÆFNR) and in the control of FNR function. Eur J Biochem 267, 4817–4824. 25 Trageser M & Unden G (1989) Role of cysteine residues and metal ions in the regulatory functioning of FNR, the transcriptional regulator of anaerobic respiration in E. coli . Mol Microbiol 3, 593–599. 26 Lazazzera BA, Beinert H, Khoroshilova N, Kennedy MC & Kiley PJ (1996) DNA-binding and dimerization of the Fe-S containing FNR protein from Escherichia coli are regulated by oxygen. J Biol Chem 271, 2762– 2768. 27 Unden G & Duche ´ ne A (1987) On the role of cyclic AMP and the FNR protein in E. coli growing anaerobi- cally. Arch Microbiol 147, 195–200. 28 Ralph ET, Scott C, Jordan PA, Thomson AJ, Guest JR & Green J (2001) Anaerobic acquisition of [4Fe-4S] clusters by the inactive FNR (C20S) variant and restor- ation of activity by second-site amino acid substitution. Mol Microbiol 39, 1199–1211. 29 Zheng L, White RH, Cash VL, Jack RF & Dean DR (1993) Cysteine desulfurase activity indicates a role for NifS in metallocluster biosynthesis. Proc Natl Acad Sci USA 90, 2754–2758. 30 Ellman GL (1959) Tissue sulfhydryl groups. Arch Bio- chem Biophys 82, 70–77. 31 WesselD&Flu ¨ gge UI (1984) A method for the quanti- tative recovery of protein in dilute solution in the presence of detergents and lipids. Anal Biochem 138, 141–143. 32 Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein util- izing the principle of protein–dye-binding. Anal Biochem 72, 248–254. S. Achebach et al. Disulfides of apoFNR FEBS Journal 272 (2005) 4260–4269 ª 2005 FEBS 4269 . in aerobic and anaerobic apoFNR Alkylated forms of aerobic and anaerobic apoFNR were digested by proteases, and the peptides were iso- lated and characterized. MALDI-TOF spectra of aerobically (A) and anaerobically (B) prepared and carboxymethylated apoFNR. The samples of apoFNR were alkylated with 10 m M iodoacetate