A periplasmic aldehyde oxidoreductase represents the first molybdopterin cytosine dinucleotide cofactor containing molybdo-flavoenzyme from Escherichia coli Meina Neumann 1 , Gerd Mittelsta ¨ dt 1 , Chantal Iobbi-Nivol 2 , Miguel Saggu 3 , Friedhelm Lendzian 3 , Peter Hildebrandt 3 and Silke Leimku ¨ hler 1 1 Institute of Biochemistry and Biology, University of Potsdam, Germany 2 Laboratoire de Chimie Bacterie ` nne, IFR 88 Institut Biologie Structurale et Microbiologie, CNRS, Marseille, France 3 Max-Volmer-Laboratories, Institut fu ¨ r Chemie, Technische Universita ¨ t Berlin, Germany Molybdoenzymes are involved in a large number of enzymatic reactions in the nitrogen, carbon and sulfur cycles. They occur in all the kingdoms of life. With the exception of nitrogenase, all molybdoenzymes carry the molybdenum cofactor (Moco), where the molybde- num atom is coordinated to the unique dithiolene moiety of a conserved tricyclic pyranopterin cofactor called molybdopterin (MPT). Depending on the remaining ligands of the molybdenum center, molyb- doenzymes are classified into three families: (a) the xanthine oxidase (XO) family, characterized by a cyanolyzable equatorial sulfur ligand coordinated to the molybdenum atom; (b) the sulfite oxidase family, with two oxo ligands at the molybdenum center; and Keywords aldehyde oxidoreductase; aromatic aldehyde; MCD; molybdenum; molybdo-flavoenzyme Correspondence S. Leimku ¨ hler, Institute of Biochemistry and Biology, University of Potsdam, D-14476 Potsdam, Germany Fax: +49 331 977 5128 Tel: +49 331 977 5603 E-mail: sleim@uni-potsdam.de (Received 2 February 2009, revised 10 March 2009, accepted 11 March 2009) doi:10.1111/j.1742-4658.2009.07000.x Three DNA regions carrying genes encoding putative homologs of xanthine dehydrogenases were identified in Escherichia coli, named xdhABC, xdhD, and yagTSRQ. Here, we describe the purification and characterization of gene products of the yagTSRQ operon, a molybdenum-containing iron– sulfur flavoprotein from E. coli, which is located in the periplasm. The 135 kDa enzyme comprised a noncovalent (abc) heterotrimer with a large (78.1 kDa) molybdenum cofactor (Moco)-containing YagR subunit, a med- ium (33.9 kDa) FAD-containing YagS subunit, and a small (21.0 kDa) 2 · [2Fe2S]-containing YagT subunit. YagQ is not a subunit of the mature enzyme, and the protein is expected to be involved in Moco modification and insertion into YagTSR. Analysis of the form of Moco present in YagTSR revealed the presence of the molybdopterin cytosine dinucleotide cofactor. Two different [2Fe2S] clusters, typical for this class of enzyme, were identified by EPR. YagTSR represents the first example of a molyb- dopterin cytosine dinucleotide-containing protein in E. coli . Kinetic charac- terization of the enzyme revealed that YagTSR converts a broad spectrum of aldehydes, with a preference for aromatic aldehydes. Ferredoxin instead of NAD + or molecular oxygen was used as terminal electron acceptor. Complete growth inhibition of E. coli cells devoid of genes from the yagTSRQ operon was observed by the addition of cinnamaldehyde to a low-pH medium. This finding shows that YagTSR might have a role in the detoxification of aromatic aldehydes for E. coli under certain growth conditions. Abbreviations ICP-OES, inductively coupled plasma optical emission spectroscopy; MCD, molybdopterin cytosine dinucleotide; MGD, molybdopterin guanine dinucleotide; Moco, molybdenum cofactor; MPT, molybdopterin; Tat, twin arginine protein transport; XDH, xanthine dehydrogenase; XO, xanthine oxidase. 2762 FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS (c) the dimethylsulfoxide reductase family, where one molybdenum atom is coordinated by two dithiolene groups [1–3]. Whereas in eukaryotes Moco is present solely in its Mo-MPT form, in prokaryotes Moco can be further modified by the addition of mononucleo- tides such as GMP, CMP, IMP or AMP to the phosphate group of the MPT [4–7]. In general, Moco formed in Escherichia coli is further modified by covalent addition of GMP to the phosphate atom on C4¢ of MPT via a pyrophosphate bond, a reaction catalyzed by the mobAB gene products. Furthermore, two of the molybdopterin guanine dinucleotide (MGD) moieties are ligated to the molybdenum atom via the dithiolene group of MPT, forming bis-MGD. So far, only the YedY protein has been shown to bind the Mo-MPT form of Moco in E. coli; however, the physiological role of this protein still remains unclear [8]. In general, prokaryotic members of the XO family have been shown to bind the molybdopterin cytosine dinucleotide (MCD) form of Moco, containing CMP added to the terminal phosphate group of the pterin side chain [9]. The XO family of molybdoenzymes comprises a number of different enzymes in prokaryotes and eukaryotes, transferring oxygen derived from water to their substrate. Most enzymes of the XO family are well characterized as purine-oxidizing and ⁄ or alde- hyde-oxidizing enzymes with broad substrate specifici- ties, but several more specific enzymes, such as carbon monoxide dehydrogenase and nicotine dehydrogenase, have been described [10,11]. Well-characterized enzymes with aldehyde-oxidizing activity are Desulfo- vibrio gigas aldehyde oxidoreductase and mammalian aldehyde oxidases [12]. Mammalian aldehyde oxidases are expressed at high levels in the liver and in the lung, and have been implicated in the detoxification of envi- ronmental pollutants and xenobiotics [13]. Bacterial aldehyde oxidases and aldehyde dehydrogenases have been identified in different bacteria, including Methylo- coccus sp., Pseudomonas sp., Streptomyces moderatus [14], Amycolatopsis methanolica [15], and Pseudo- monas testosteroni [16]. In addition, xanthine dehydro- genases (XDHs) capable of oxidizing various purine and aldehyde substrates have been characterized in bacteria such as Rhodobacter capsulatus [17], Pseudo- monas putida 86 [18,19], and Veillonella atypica [20,21]. With the exception of R. capsulatus and Pseudomo- nas aeruginosa XDH [22,23], which bind Mo-MPT, all bacterial XDHs characterized to date bind the MCD form of Moco. The molecular masses of these XDHs range from 140 to 300 kDa, and different subunit structures have been observed, such as a 2 in Strepto- myces cyanogenus [24], abc in V. atypica [21], a 3 in P. putida [25], a 2 b 2 in R. capsulatus [17], a 2 b 2 in Coma- monas acidovorans [26,27], and a 4 b 4 in P. putida 86 [19]. However, in general, enzymes of the XO family possess the same overall architecture [28], with two dis- tinct [2Fe2S] clusters bound to the N-terminal domain or subunit, an FAD bound to a central domain or subunit (with the exception of D. gigas aldehyde oxi- doreductase, in which the FAD-binding domain is absent [29]), and the Moco-binding domain at the C-terminus. As part of the E. coli K-12 genome project [30], three DNA regions carrying genes encoding putative homologs of XDHs were identified, named xdhABC, xdhD, and yagTSRQ [17,31]; however, none of these proteins has been characterized at the biochemical level to date, and their physiological functions remain as yet unknown. Previous amino acid sequence alignments with the individual domains of the well-characterized bacterial XDH from R. capsulatus revealed amino acid identities of 24–43% between single protein domains; however, the organization of the genes was found to be different to that in R. capsulatus xdhA, xdhB and xdhC and their putative E. coli counterparts [17]. Alignments of two of the annotated operons in E. coli, xdhABC and xdhD, showed higher homologies to XDHs than to aldehyde oxidases [17]. Genetic approaches suggested a role in the purine salvage path- way for these enzymes [31]. The third operon, yagTSRQ, encodes a putative aldehyde oxidase. In the yagTSRQ operon, YagT contains a twin arginine pro- tein transport (Tat) leader peptide for translocation to the periplasm. Reporter protein fusion assays revealed that the YagT signal peptide leads to export to the periplasm and cleavage after amino acid 49, so YagTSRQ was predicted to contain a periplasmic protein complex [32]. YagT shares homologies with the FeS cluster-containing subunit of the class of molyb- do-flavoenzymes, YagS with the FAD-binding subunit, and YagR with the Moco-containing subunit [17]. YagQ was shown to share amino acid homology of 40% to R. capsulatus XdhC, a protein that has been shown to be involved in Moco binding, addition of the terminal sulfido ligand of Moco, and insertion of sulfurated Moco into the XdhB subunit of R. capsula- tus XDH [33]. The yagTSRQ operon is not essential for E. coli under standard growth conditions, as a gene region containing the yagTSRQ operon is deleted in the E. coli K-12-based laboratory strain MC4100 and its derivatives [34]. As the yagTSRQ operon contains an ORF for a protein homologous to R. capsulatus XdhC, and is the least characterized operon at the genetic level of members of the XO family, we analyzed the role of the yagTSRQ operon in E. coli. M. Neumann et al. An MCD-binding aldehyde oxidoreductase from E. coli FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS 2763 In the present work, we describe the purification and characterization of gene products of the yagTSRQ operon in addition to the characterization of their physiological role in E. coli. YagTSR was shown to be a periplasmic aldehyde oxidoreductase that oxidizes a broad spectrum of aldehydes. As complete growth inhibition of E. coli cells devoid of genes from the yagTSRQ operon was observed by the addition of cinnamaldehyde to a low-pH medium, we suggest that a periplasmic aldehyde oxidoreductase might play a role in the detoxification of aldehydes to avoid cell damage in E. coli. Results Purification of YagTSR expressed in the presence or absence of YagQ YagTSRQ and YagTSR were expressed in E. coli TP1000 (DmobAB) cells from plasmids pMN100 and pMN111, respectively. Expression of the proteins results in an N-terminal His6-fusion to YagT, with a deletion of the 49 N-terminal amino acids of the Tat leader peptide. TP1000 (DmobAB) is a derivative of the E. coli MC4100 strain [35], which carries a deletion of the gene region encompassing the yagTSRQ operon. Thus, no endogenous yagQ was present to interfere with our analyses. After purification by Ni 2+ –nitrilotriacetic acid affinity chromatography, eluted fractions from the two different expression constructs displayed three bands on Coomassie Brilliant Blue R-stained SDS ⁄ PAGE gels, corresponding to molecular masses of 78.1, 33.9 and 21.0 kDa, respectively (Fig. 1A). Coex- pressed yagQ was not identified as a subunit of the purified YagTSR enzyme expressed from pMN100. Densitometric analysis of the Coomassie-stained SDS ⁄ PAGE gel revealed comparable peak densities for the YagT and YagS subunits from both expressions; however, the peak density of YagR was reduced to 37% in the strain containing YagTSR expressed in the absence of YagQ (Fig. 1A). The protein expressed from the yagTSRQ operon eluted with a size of 135 kDa from a Superdex 200 size exclusion chromatography column, corresponding to the YagTSR trimer (Fig. 1B, solid line). The protein expressed in the absence of YagQ ()YagQ) displayed two peaks after Superdex 200 size exclusion chromatography, the 135 kDa peak cor- responding to the YagTSR trimer, and a 55 kDa peak A B Fig. 1. Purification of YagTSR after expression in the presence and absence of YagQ. (A) 12% SDS ⁄ PAGE of purification stages. Lane I: 1 lLofE. coli TP1000 · pMN100 (yagTSRQ) extract after cell lysis. Lane II: 10 lL of YagTSR with an OD 445 nm of 0.09 after expression in the presence of YagQ. Lane III: 1 lLofE. coli TP1000 · pMN111 (yagTSR) extract after cell lysis. Lane II: 10 lL of YagTSR with an OD 445 nm of 0.09 after expression in the absence of YagQ. (B) Size exclusion chromatography of YagTSR. Two hundred microliters of YagTSR (+YagQ, solid line, )YagQ, dotted line) with an OD 445 nm of 0.26 was analyzed by analytical size exclusion chromatography in 50 m M Tris and 200 mM NaCl (pH 7.5) using a Superdex 200 column. YagTSR (+YagQ) was additionally purified on a Q-Sepharose column prior to size exclusion chromatography. Inset: plot of the standard proteins. Size exclusion chromatography markers (Bio-Rad): c-globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B 12 (1.3 kDa). An MCD-binding aldehyde oxidoreductase from E. coli M. Neumann et al. 2764 FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS containing to the YagTS dimer (Fig. 1B, dotted line). Analysis of the specific activities of both proteins with vanillin revealed that YagTSR expressed in the absence of YagQ ()YagQ) was completely inactive, whereas YagTSR expressed in the presence of YagQ (+YagQ) exhibited an activity of 27.9 ± 0.2 UÆmg )1 (Table 1). Cofactor analysis of the purified YagTSR proteins To analyze the cofactor content of YagTSR expressed in the presence or absence of YagQ, the molybdenum and iron contents were quantified by inductively cou- pled plasma optical emission spectroscopy (ICP-OES) (Table 1). Iron contents of 4.04 ± 0.15 molecules per YagTSR (+YagQ) and 3.84 ± 0.08 molecules per YagTSR ()YagQ) were identified, corresponding to two [2Fe2S] clusters per protein trimer. The inactive YagTSR ()YagQ) protein was shown to contain no molybdenum bound to the protein, whereas the active YagTSR (+YagQ) trimer was saturated to 58 ± 3% with molybdenum. To analyze the Moco content of YagTSR (+YagQ) and YagTSR ()YagQ), the proteins were incubated for 30 min at 95 °C in the presence of acidic iodine, which oxidizes released MPT to its fluorescent deriva- tive, Form A. Whereas Form A was readily detected from YagTSR (+YagQ) after separation on a reversed-phase C18 column (Fig. 2A), no Form A was released from YagTSR ()YagQ) (Fig. 2B). In contrast, after overnight incubation of YagTSR (+YagQ) with acidic iodine at room temperature, and purification using a Q-Sepharose column, no Form A was detected (Fig. 2C). Instead, a nucleotide derivative of Form A was eluted from the Q-Sepharose column (Fig. 2D). As YagTSR (+YagQ) was purified from the E. coli TP1000 strain (DmobAB) in an active form, the pres- ence of bis-MGD bound to YagTSR could be excluded. To identify the nucleotide bound to Moco in YagTSR (+YagQ), the protein was incubated for 15 min at 95 °C in the presence of 5% sulfuric acid, which released AMP from FAD and the nucleotide of Table 1. Cofactor content of YagTSR expressed in the presence or absence of YagQ. Specific enzyme activity (unitsÆmg )1 ) is defined as the oxidation of 1 lmol vanillinÆmin )1 Æmg )1 in phosphate ⁄ citrate buffer (pH 6.0) at room temperature, using ferricyanide as electron acceptor. Molybdenum (l M molybdenum ⁄ lM YagTSR) and iron (lM 2 · [2Fe2S] ⁄ lM YagTSR) contents were determined by ICP-OES (see Experimental procedures) and related to a fully saturated enzyme. Nucleotide content (l M CMP or AMP ⁄ lM YagTSR) content was analyzed after release of CMP from MCD and AMP from FAD by heat treatment under acidic conditions, as described in Experimental procedures. The concentra- tion of the terminal sulfur ligand of Moco (l M SCN ) ⁄ lM YagTSR) was determined spectrophotometrically as an iron–thiocyanate complex at 420 nm as described in Experimental procedures. Potassium thiocyanate was used as a standard curve. ND, none detectable; –, not deter- mined. Expression strain Activity (unitsÆmg )1 ) Mo (%) Fe (%) AMP (%) CMP (%) Cyanolyzable sulfur (%) TP1000 · yagTSRQ a 27.9 ± 0.2 58 ± 3 101 ± 4 106 ± 8 62 ± 6 58 ± 4 TP1000 · yagTSR b ND ND 96 ± 2 97 ± 2 ND – a YagTSR expressed from plasmid pMN100 (yagTSRQ) was purified by Ni 2+ –nitrilotriacetic acid and Q-Sepharose chromatography as described in Experimental procedures. b YagTSR expressed from plasmid pMN111 (yagTSR) was purified solely by Ni 2+ –nitrilotriacetic acid chromatography as described in Experimental procedures. Fig. 2. Moco analysis of YagTSR expressed in the presence or absence of YagQ. Analysis of the fluorescent derivatives of Moco from YagTSR. Form A was produced from (A) 2 l M YagTSR (+YagQ) and (B) 2.4 l M YagTSR ()YagQ) after 30 min of oxidation with acidic iodine at 95 °C. Form A was separated on a C18 RP-HPLC column with 85% 5 m M ammonium acetate and 15% methanol at an isocratic flow rate of 1 mLÆmin )1 . (C) Form A was produced after overnight oxidation in acidic iodine at room tempera- ture. Released Form A was applied to a Q-Sepharose column, eluted with 10 m M acetic acid, and applied to a C18 RP-HPLC column in 85% 5 m M ammonium acetate and 15% methanol at an isocratic flow rate of 1 mLÆmin )1 . (D) The dinucleotide form of Form A was produced after overnight oxidation in acidic iodine at room temperature. Released Form A dinucleotide was applied to a Q-Sepharose column, eluted with 50 m M HCl, and applied to a C18 RP-HPLC column in 97% 5 m M ammonium acetate and 3% metha- nol at an isocratic flow rate of 1 mLÆmin )1 . Fluorescence was deter- mined by excitation at 383 nm and emission at 450 nm. M. Neumann et al. An MCD-binding aldehyde oxidoreductase from E. coli FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS 2765 the MPT dinucleotide cofactor. Nucleotide analysis using the respective standard nucleotides revealed the presence of AMP and CMP (Table 1). The AMP content could be related to complete saturation of YagTSR (+YagQ) and YagTSR ()YagQ) with FAD (Table 1). CMP was present in YagTSR (+YagQ) at a saturation level of 62 ± 6% per trimer, but was not identified in YagTSR ()YagQ) (Table 1). The CMP content corresponded well to the molybdenum content of YagTSR (+YagQ), and showed that the protein was saturated to 60% with the MCD cofactor. As YagTSR belongs to the XO family of molyb- doenzymes, characterized by a terminal sulfido ligand at Moco, it was of interest to determine the saturation of the sulfur ligand in YagTSR (+YagQ). As shown in Table 1, after incubation of YagTSR (+YagQ) with cyanide, the content of the cyanolyzable sulfur was determined to be 58 ± 4%. This result showed that the MCD cofactor present in YagTSR (+YagQ) was completely saturated with the sulfido ligand, and no demolybdo or desulfo form of the protein was purified. Steady-state kinetics of YagTSR, determined using different aldehydes and purines as substrates For kinetic characterizations, YagTSR was purified by Ni 2+ –nitrilotriacetic acid chromatography and Q-Sepharose ion exchange chromatography (Fig. 3A). The yield of purified YagTSR was 1.8 mg Æ L )1 of E. coli culture. The visible absorption spectrum of YagTSR (+YagQ) was similar to those of other molybdo-flavoenzymes, and showed the presence of FeS and FAD as prosthetic groups (Fig. 3B). Reduc- tion of YagTSR (+YagQ) with benzaldehyde showed that the protein was reduced to a level of 60% under anaerobic conditions. For complete reduction, sodium dithionite was added to a final concentration of 20 mm (Fig. 3B). During the reduction of YagTSR with either benzaldehyde or dithionite, the production of the flavin semiquinone was not observed. Steady-state kinetics with YagTSR showed a broad substrate spectrum with aromatic and aliphatic alde- hydes, whereas purines were not oxidized (Table 2). In total, aromatic aldehydes were converted with higher k cat and lower K m values than those for aliphatic alde- hydes. The lowest K m values of 69.8 and 63.1 lm were obtained for benzaldehyde and cinnamaldehyde, respectively. Both substrates also showed the highest catalytic efficiency, with k cat ⁄ K m ratios of 1.39 and 1.32 lm )1 Æs )1 , respectively. In comparison, the K m of vanillin of 131.8 lm was about twice as high, with a concomitant increase in k cat to 124.6 s )1 . Phenylacetic aldehyde showed the lowest k cat , of 7.0 s )1 . None of the tested aliphatic aldehydes showed a K m below 400 lm, indicating that these are not likely to be physi- ological substrates of YagTSR. Retinalaldehyde, an aromatic aldehyde with a long aliphatic side chain, was not oxidized by YagTSR. As enzymes of the XO family are known to catalyze the conversion of a vari- ety of purines, xanthine, hypoxanthine and caffeine were additionally analyzed as substrates for YagTSR. In summary, no activity was detected with all purines tested during the steady-state kinetic analyses. Additionally, YagTSR showed no detectable nicotine AB Fig. 3. Purification and UV–visible absorption spectra of YagTSR. (A) Twelve percent SDS ⁄ PAGE of purification stages of YagTSR (+YagQ). Lane I: molecular weight marker. Lane II: 1 lLofE. coli TP1000 · pMN100 (yagTSRQ) extract after cell lysis. Lane III: 12 lg of YagTSR after Ni 2+ –nitrilotriacetic acid affinity chromatography. Lane IV: 12 lgof YagTSR after Q-Sepharose ion exchange chromatography. (B) Characterization of purified YagTSR (+YagQ) by UV–visible absorption spectroscopy. Spectra of 7 l M air-oxidized YagTSR (solid line), of 7 lM YagTSR incubated with 500 lM benzaldehyde (dashed line), and of 7 l M YagTSR reduced with 20 m M dithionite (dotted line). Spectra were recorded in 50 m M Tris and 1 mM EDTA (pH 7.5) under anaerobic conditions. An MCD-binding aldehyde oxidoreductase from E. coli M. Neumann et al. 2766 FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS dehydrogenase activity. Thus, YagTSR was identified as an aldehyde-oxidizing enzyme. To identify the phys- iological electron acceptor, nonphysiological electron acceptors, such as 2,6-dichlorophenol-indophenol and ferricyanide, in addition to cytochrome c, NAD + , molecular oxygen or ferredoxin, were analyzed. As shown in Table 3, no activity was observed when cyto- chrome c, NAD + or O 2 was used as terminal electron acceptor. Although the nonphysiological electron acceptors dichlorophenol-indophenol and ferricyanide were suitable electron acceptors, spinach ferredoxin was also able to accept electrons from reduced YagTSR. This indicates that an E. coli ferredoxin might act as a physiological electron acceptor (Table 3). We additionally analyzed the pH optimum and the temperature stability of the enzyme (Fig. 4). YagTSR showed a pK a of 5.5 with vanillin as substrate and ferricyanide as electron acceptor (Fig. 4A), revealing the pK a of the active site glutamate (Glu692), which acts as a base catalyst. The exchange of Glu692 for glutamine resulted in an inactive enzyme (the rate of reaction was at least 10 7 -fold slower than seen for the wild-type enzyme; data not shown), underlining the importance of this residue in the base-catalyzed reac- tion. The protein was stable up to a pH of 4; however, at lower pH, the protein was denatured and release of FAD was observed (data not shown). The effect of temperature on the enzyme activity revealed that the protein was heat stable at temperatures of up to 95 °C in short-term incubations (data not shown). In long- term incubations at higher temperatures, YagTSR retained 92% of its activity after 15 min at 50 °C, and 73% of its activity after 15 min at 70 °C (Fig. 4B). EPR spectroscopy of the YagTSR FeS clusters Figure 5 shows the EPR spectra obtained from YagTSR at low temperatures; these were obtained after reduction of the samples with either benzaldehyde or sodium dithionite. Proteins from the XO family usually exhibit signals from two different [2Fe2S] clus- ters, which can be clearly distinguished by EPR spec- troscopy, owing to their different g-values and temperature behaviors [2,36,37]. The signal showing more axial symmetry, which is also visible at higher temperatures (e.g. 60 K), is assigned to the FeSI clus- ter, and the second signal, which is only visible at tem- peratures below 40 K, is assigned to the FeSII cluster. In YagTSR, a more rhombic signal at 2.005, 1.943 and 1.916 could be seen after sodium dithionite reduc- tion at a temperature of 60 K (for simulation parame- ters, see Table 4). The g av was 1.95 and, typically for a [2Fe2S] center, the linewidth was 1.4 mT. This signal can be attributed to the FeSI cluster. When the tem- perature was decreased to below 40 K, the signal of FeSII appeared to be superimposed on the FeSI signal (Fig. 5B). The corresponding g-values of FeSII were Table 3. Analysis of different electron acceptors for YagTSR. Spe- cific enzyme activity (unitsÆmg )1 ) is defined as the oxidation of 1 lmol of vanillinÆmin )1 Æmg )1 of enzyme in buffer containing 100 m M Tris (pH 6.8). ND, none detectable. Electron acceptor Specific activity (UÆmg )1 ) Ferredoxin a 0.27 ± 0.01 Cytochrome c b ND NAD +c ND Oxygen d ND Ferricyanide e 8.63 ± 0.03 2,6-Dichlorophenol-indophenol f 2.73 ± 0.08 a Reduction of 0.2 mgÆmL )1 spinach ferredoxin was determined by following the absorbance change at 420 nm. b Reduction of 0.65 mgÆmL )1 cytochrome c was determined by following the absorbance change at 550 nm. c Reduction of 1 mM NAD + was determined by following the reduction at 340 nm; 500 l M benzalde- hyde was used as substrate to avoid the overlap in absorption at 340 nm of vanillin. d Activity was measured using either oxygen- saturated phosphate ⁄ citrate buffer (pH 6.0) or Tris buffer (pH 6.8) by following the oxidation of vanillin at 340 nm. e Reduction of 1m M ferricyanide was determined by following the absorbance change at 420 nm. f Reduction of 200 lM 2,6-dichlorophenol-indo- phenol was determined by following the absorbance change at 600 nm. Table 2. Steady-state kinetics parameters of YagTSR with different aldehyde and purine substrates. Steady-state kinetics were deter- mined in phosphate ⁄ citrate buffer (pH 6.0) using 1 m M ferricyanide as electron acceptor at substrate concentrations of approximately 0.5–2 K m . K m and k cat values were obtained after nonlinear fitting using ORIGIN 6.0 software (Microcal; GE Healthcare). ND, none detectable; –, not determined. Substrate K m (lM) k cat (s )1 ) k cat ⁄ K m (lM )1 Æs )1 ) Cinnamaldehyde 63 ± 10 84 ± 5 1.32 Vanillin 132 ± 20 125 ± 2 0.95 Benzaldehyde 70 ± 7 97 ± 3 1.39 Phenylacetic aldehyde 132 ± 6 7.0 ± 0.6 0.05 2,4-Dihydroxybenzaldehyde 428 ± 55 55 ± 8 0.13 Valeraldehyde 429 ± 73 6.8 ± 1.0 0.02 Heptaldehyde 426 ± 42 11 ± 1 0.03 Acetic aldehyde 1150 ± 30 30 ± 1 0.03 Xanthine ND ND – Hypoxanthine ND ND – Nicotine ND ND – Caffeine ND ND – Pyridoxal ND ND – All-trans-retinaldehyde ND ND – M. Neumann et al. An MCD-binding aldehyde oxidoreductase from E. coli FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS 2767 2.07, 1.96, and 1.92. The linewidth was 2.4 mT. The g av value of 1.98 is similar to that for FeSII from other organisms [38]. The linewidth of FeSI at low tempera- tures was slightly increased, especially on the g x -com- ponent. This broadening probably resulted from slight heterogeneity in the local structure of FeSI, which averaged out at 60 K. Alternatively, it may have resulted from a coupling between FeSI and FeSII, which consequently is expected to lead to a small split- ting at low temperatures. The linewidth of both clus- ters was smaller than, for example, that of the [2Fe2S] clusters from R. capsulatus XDH [36]. In addition to reduction with sodium dithionite, the sample was treated with the substrate benzaldehyde (Fig. 5C). The resultant spectrum was identical to the dithionite reduced spectrum (Fig. 5B), showing that the FeS clusters were indeed reduced by electron trans- fer in the course of the catalytic reaction. The spec- trum at higher temperatures, which only reflects FeSI, was identical to the spectrum in Fig. 5A (data not shown). The electron transfer seemed to be quite fast, and no signal attributable to a flavin semiquinone radical could be found at pH 7.5. However, when reduction at pH 10 was performed, the reaction was considerably slower, and the flavin semiquinone radical became visible (data not shown). Growth of E. coli wild-type and different mutant strains in the presence and absence of cinnamaldehyde To determine the physiological role of YagTSR in E. coli, yagR ) , yagT ) and yagQ ) cells were grown in 0.1 · LB medium at pH 4.0 in the presence or absence of cinnamaldehyde. As shown in Fig. 6A, in the absence of cinnamaldehyde the growth of the strains A B Fig. 4. Analysis of the pH optimum and temperature stability of YagTSR. (A) The pH optimum of YagTSR was determined by analysis of the specific activity (unitsÆmg )1 ) in phosphate ⁄ citrate buffer in a pH range from 4 to 8, with ferricyanide as electron acceptor. (B) For the analysis of the temper- ature stability of YagTSR, the enzyme was incubated for 15 min at different tempera- tures. The specific cinnamaldehyde ⁄ ferricya- nide activity was determined, and related to the corresponding enzyme activity before the heat treatment step. Fig. 5. EPR spectra of YagTSR. X-Band EPR spectra of reduced YagTSR. Experimental spectra are shown as solid black lines and corresponding simulations as dotted black lines. (A) Sodium dithio- nite reduced YagTSR at T = 60 K with simulation of FeSI. (B) Sodium dithionite reduced YagTSR at T = 20 K with simulation of FeSII (upper trace) and simulation of complete spectrum (lower trace). (C) Treated with benzaldehyde (substrate) at T = 20 K. The obtained spectrum is identical to the spectrum shown in (B). Exper- imental conditions: 1 mW microwave power (A), 0.25 mW micro- wave power (B, C); 1 mT modulation amplitude, 12.5 kHz modulation frequency, 9.56 GHz microwave frequency. An MCD-binding aldehyde oxidoreductase from E. coli M. Neumann et al. 2768 FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS with mutations in the yagTSRQ operon was not affected in comparison to the corresponding wild-type strain. Cell growth was observed after a lag phase of approximately 4 h, and the stationary phase was initi- ated after approximately 15 h. The addition of 800 lm cinnamaldehyde to the 0.1 · LB medium (pH 4.0) led to impaired cell growth of the wild-type, with a lag phase of 7 h. The maximal attenuance obtained in the stationary phase reached only 35% of the value in the absence of cinnamaldehyde. In comparison, mutations in the yagR, yagT or yagQ genes resulted in complete impairment of cell growth in the presence of 800 lm cinnamaldehyde (Fig. 6B). The phenotype is directly linked to the low pH in the medium, as at higher pH values of 6–7, no effect of the gene disruptions on cell growth in comparison to the wild-type strain was observed (data not shown). In addition, the growth of E. coli was also inhibited by the addition of 10 mm vanillin or 5 mm benzaldehyde to the 0.1 · LB medium at pH 4.0 (data not shown). Discussion In this report, the molybdenum-containing iron–sulfur flavoprotein YagTSR from E. coli was purified and characterized. It was shown to be an aldehyde oxidore- ductase that oxidizes aldehydes to their respective acids. The 135 kDa enzyme is a noncovalent hetero- trimer with a large (78.1 kDa) Moco-containing YagR subunit, a medium (33.9 kDa) FAD-containing YagS subunit, and a small (21.0 kDa) 2 · [2Fe2S]-containing YagT subunit. The YagT protein contains a 49 amino acid Tat leader peptide that allows the export of the active heterotrimer to the periplasm (data not shown). Tat substrates are matured in the cytoplasm prior to their translocation. Here, we wanted to distinguish the translocation event from the Moco insertion into the YagTSR trimer. Thus, we expressed the YagT protein without the Tat leader, resulting in a cytoplasmic and active YagTSR protein complex. YagTSR purified from the cytoplasm contained 0.58 atoms of moly- Table 4. g-Tensor principal values as obtained by simulation of experimental spectra. g av =(g x + g y + g z ) ⁄ 3. Sample Cluster g-Values Linewidth (mT) g x g y g z g av YagTSR FeSI (60 K) 2.005 1.943 (0.006) a 1.916 (0.003) a 1.95 1.4 FeSI (20 K) 2.005 1.943 1.916 1.95 1.7 FeSII (20 K) 2.07 1.96 1.92 1.98 2.4 XDH b FeSI 2.017 1.921 1.921 1.95 2.8 FeSII 2.07 1.97 1.90 1.98 4.0 a The numbers in parentheses denote the g-strain used for simulation. b R. capsulatus XDH, values taken from [36]. Fig. 6. Growth curves of E. coli BW25113, JW0278 (yagR ) ), JW0280 (yagT ) ) and JW0277 (yagQ ) ) in the absence and presence of 800 l M cinnamaldehyde. E. coli strains BW25113 (filled squares), JW0278 (yagR ) , open squares), JW0280 (yagT ) , filled circles) and JW0277 (yagQ ) , open circles) were inoculated at an attenuance at 600 nm of 0.1 in medium containing 1 gÆL )1 tryptone, 0.5 gÆL )1 yeast extract and 1 gÆL )1 NaCl (0.1 · LB) at pH 4.0, and incubated at 37 °C and 150 r.p.m. in the absence (A) and presence (B) of 800 l M cinnamaldehyde. M. Neumann et al. An MCD-binding aldehyde oxidoreductase from E. coli FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS 2769 bdenum, four atoms of iron, one molecule of FAD, 0.58 atoms of acid-labile sulfur, and 0.62 molecules of CMP, showing that YagTSR binds the MCD form of Moco. This is the first enzyme identified in E. coli that binds the MCD. So far, almost all characterized E. coli molybdoenzymes belong to the dimethylsulfox- ide reductase family of molybdoenzymes, and have been shown to bind bis-MGD. The only enzyme that does not belong to this class of molybdoenzymes is the E. coli YedY protein, which belongs to the sulfite oxi- dase family and binds the Mo-MPT form of Moco [8]. Genetic investigations including the E. coli xdhABC, xdhD and yagTSRQ operons were performed by Koz- min and Schaaper [39], who characterized the resis- tance to N-hydroxylated base analogs in E. coli. They classified the E. coli xdhABC, xdhD and yagTSRQ operons as putative family members of the XO family. However, without investigating the cofactor present in these proteins, we believe that the authors erroneously concluded that the enzymes bind the Mo-MPT form of Moco [39]. It is likely that the gene products of the xdhABC and xdhD operons also bind the MCD form of Moco, as an amino acid sequence alignment of the Moco-binding subunits of E. coli YagR, XdhA and XdhD showed high amino acid identities to D. gigas MOP [12] and Oligotropha carboxidovorans CoxL [40], two structurally characterized subunits that bind MCD (Fig. S1). In particular, the amino acids involved in CMP binding are highly conserved in the three E. coli proteins (Fig. S1). We have additionally demonstrated that the activity of YagTSR is independent of MobA, as an active enzyme was purified from the mobAB-deficient E. coli TP1000 strain. Thus, we conclude that MCD biosyn- thesis in E. coli requires a so far unidentified gene product that catalyzes the attachment of CMP to MPT. Our investigations further showed that the yagQ gene product is required for the production of an active, Moco-containing YagTSR trimer. When YagTSR was expressed in the absence of YagQ, the protein was inactive and devoid of Moco, and the YagR subunit was shown to be unstable and was rap- idly degraded during expression ⁄ purification. There- fore, YagQ has a stabilizing effect on YagR and is required for the insertion of Moco into the subunit. These characteristics are similar to those of the R. capsulatus XdhC protein, which is essential for the insertion of Moco into XDH. XdhC was reported to be directly involved in the insertion of the sulfido ligand of Moco while bound to XdhC, and, further- more, to transfer the sulfurated cofactor to Moco-free apo-XDH by direct interaction with the XdhB subunit [33,41]. XdhC seems to perform a ‘quality’ control of Moco, as only sulfurated Moco is inserted into XDH in R. capsulatus. We predict that YagQ performs a similar role for YagTSR in E. coli, as YagTSR was shown to bind the sulfurated form of MCD. However, it needs to be clarified whether Moco bound to YagQ is sulfurated before or after the attachment of the CMP moiety. The EPR spectra of YagTSR were found to be very similar to those of R. capsulatus XDH, showing an almost axial EPR signal for FeSI and a broader, strongly rhombic signal for FeSII [36]. Only subtle dif- ferences in the g-values and linewidths, in particular for the FeSI center (more rhombic g-tensor in YagTSR), are observed. The values for g av are similar. Neverthe- less, the overall close similarity of the EPR parameters indicated the presence of the same ligands and similar geometries of the two redox centers in YagTSR and R. capsulatus XDH. The flavin semiquinone was not visible at pH 7.5; thus, it is not stabilized in YagTSR under physiological pH conditions. This is similar to the situation with XOs, where the semiquinone is also not formed during reductive titrations, whereas in XDHs the flavin semiquinone is stabilized in the absence of NAD + [42,43]. Here, the binding of NAD + destabilizes the flavin semiquinone by increasing the redox potential of FADH ⁄ FADH 2 [43]. As YagTSR does not use NAD + as electron acceptor, the stabiliza- tion of the flavin semiquinone seems to be less favor- able, which in this respect makes YagTSR more similar to XOs. Production of superoxide by YagTSR was also not observed, consistent with the observation that YagTSR does not use molecular oxygen as electron acceptor. Thus, the physiological electron acceptor of YagTSR seems to react with fully reduced FAD. Non- physiological electron acceptors such as 2,6-dichlor- ophenol-indophenol or ferricyanide were used by YagTSR. However, as spinach ferredoxin was also used as an electron acceptor, an E. coli ferredoxin seems to be a possible electron acceptor for this enzyme. Analysis of the substrate specificities of YagTSR showed a broad substrate spectrum with a preference for aromatic aldehydes, whereas purines were not oxi- dized by YagTSR. The protein was shown to be stable at low pH values between 4 and 5. Investigations of the phenotypes of single mutations in yagT, yagR and yagQ showed that YagTSR is essential under low-pH conditions in 0.1 · LB medium in the presence of the aromatic aldehydes cinnamaldehyde, vanillin, or benz- aldehyde. Aromatic aldehydes occur ubiquitously in nature as flavoring and coloring molecules of plants, and during acidic and enzymatic degradation of lignin [44–46]. Vanillin and cinnamaldehyde were shown to have antimicrobial activity at high doses [47–49]; for An MCD-binding aldehyde oxidoreductase from E. coli M. Neumann et al. 2770 FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS example, cinnamaldehyde is inhibitory to E. coli strains at concentrations of 250–500 lgÆmL )1 [47]. It was reported previously that cinnamaldehyde at a con- centration of 500 lgÆmL )1 damaged the surface struc- ture of E. coli and thus resulted in impaired growth [47]. In addition, lower pH values (3.5–4.0) increased the sensitivity to aromatic aldehydes of E. coli [48]. Several potential mechanisms for the toxicity of alde- hydes have been investigated, including damage from chemical reactivity, direct inhibition of glycolysis and fermentation, and plasma membrane damage. Alde- hydes are chemically reactive, and can form products with many classes of biological molecules [50], includ- ing nucleic acids, proteins, and lipids. Our results show conclusively that the periplasmic aldehyde oxidoreduc- tase YagTSR is involved in the detoxification of aro- matic aldehydes to their less toxic acids in the periplasm of E. coli, and cell damage under extreme growth conditions is thus avoided. Experimental procedures Bacterial strains, plasmids, media, and growth conditions Escherichia coli TP1000 (DmobAB) [35] was used for homol- ogous expression of E. coli yagTSRQ and yagTSR. Cells were grown aerobically in LB medium at 22 °C in the pres- ence of 150 lgÆmL )1 ampicillin. Sodium molybdate was added at a concentration of 1 mm. E. coli strains BW25113, JW0278 (yagR ) ), JW0280 (yagT ) ) and JW0277 (yagQ ) ) [51] were used for growth experiments. Cloning, expression and purification of E. coli YagTSRQ DNA fragments containing the coding regions for E. coli yagTSRQ were amplified by PCR from total DNA obtained from E. coli K-12. The flanking restriction sites NdeI and SacI were introduced by the PCR primers, and the amplified yagTSRQ operon was cloned without the N-terminal Tat leader sequence into the NdeI–SacI sites of the expression vector pTrcHis [52]. The resulting plasmid pMN100 expresses the yagTSRQ operon as an N-terminal His6-tag fusion to YagT for affinity purification from the cytoplasm. A plasmid expressing only the yagTSR operon in pTrcHis, named pMN111, was obtained by deletion of the yagQ gene in plasmid pMN100. For production of His6-tagged YagTSR, E. coli TP1000 cells were transformed with plasmids pMN100 and pMN111, respectively. One liter of LB supplemented with 1mm sodium molybdate and 10 lm isopropyl thio-b-d- galactoside was inoculated with 2 mL of an overnight culture and incubated for 24 h at 22 °C and 100 r.p.m. The cells were harvested by centrifugation at 9600 g for 5 min. The cell pellet was resuspended in phosphate buffer (50 mm NaH 2 PO 4 , 300 mm NaCl, pH 8.0). Complete cell lysis was achieved by three passages through a TS Series Benchtop cell disruptor (Constant Systems, Daventry, UK) at 1350 bar in the presence of DNaseI (1 lgÆmL )1 ). The cleared lysate was applied to 0.5 mL of Ni 2+ –nitrilotriacetic acid (Qiagen, Hilden, Germany) per liter of culture. The column was washed with 2 · 20 column volumes of phosphate buf- fer containing 10 and 20 mm imidazole each. Protein was eluted with phosphate buffer containing 250 mm imidazole, and the buffer was changed to 50 mm Tris and 1 mm EDTA (pH 7.5) by either dialysis or PD10 gel filtration chromatography (GE Healthcare, Munich, Germany). For further purification, the YagTSR was applied to a Q-Sepha- rose column (GE Healthcare) and eluted with a linear gra- dient of 0–1 m NaCl in 50 mm Tris and 1 mm EDTA (pH 7.5). Size exclusion chromatography using 0.3 mg of YagTSR was performed using a Superdex 200 column (GE Healthcare) with a bed volume of 24 mL equilibrated in 50 mm Tris and 200 mm NaCl (pH 7.5). The size of YagTSR was determined by using a gel filtration standard (BioRad, Hercules, CA, USA). To determine the purity of YagTSR, densitometric analysis of Coomassie Brilliant Blue-stained SDS ⁄ PAGE gels was performed using quantityone 4.6 software (BioRad). The YagTSR concen- tration of the purified enzyme was determined from the absorbance at 445 nm, using an extinction coefficient of 23 686 m )1 Æcm )1 for the native enzyme. The extinction coefficient was determined on the basis of FAD content after trichloroacetic acid precipitation [53]. Metal analysis Metal analysis was performed using PerkinElmer Optima 2100DV ICP-OES (Fremont, CA, USA). Protein samples were wet-ashed overnight in a 1 : 1 mixture with 65% nitric acid (Suprapur; Merck, Darmstadt, Germany) at 100 °C. Samples were diluted with 4 mL of H 2 O prior to their injection onto the ICP-OES apparatus. As reference, the multielement standard solution XVI (Merck) was used. Moco ⁄ MPT analysis To determine the MCD content of YagTSR, the samples were incubated overnight at room temperature in the pres- ence of acidic iodine to convert MCD to Form A CMP. Form A was separated from Form A using a 400 lL Q-Sepharose ion exchange column (GE Healthcare), which was equilibrated in H 2 O. The oxidized samples were loaded, and Form A was eluted with 0.8 mL of 10 mm ace- tic acid and analyzed as described previously [41]. Form A was eluted with 0.6 mL of 50 mm HCl and directly applied to a C18 RP-HPLC column (4.6 · 250 mm ODS Hypersil, particle size 5 lm; Thermo Scientific, Karlsruhe, Germany), M. Neumann et al. An MCD-binding aldehyde oxidoreductase from E. coli FEBS Journal 276 (2009) 2762–2774 ª 2009 The Authors Journal compilation ª 2009 FEBS 2771 [...]... Schweiger A (2006) EasySpin, a comprehensive software package for spectral simulation and analysis in EPR J Magn Reson 178, 42–55 Supporting information The following supplementary material is available: Fig S1 Amino acid sequence alignment of E coli YagR, XdhA, XdhD, the CoxL chain of the O carboxidovorans CODH and the D gigas aldehyde oxidoreductase MOP This supplementary material can be found in the online... Crystal structure of the xanthine oxidase-related aldehyde oxido-reductase from D gigas Science 270, 1170–1176 Calzi ML, Raviolo C, Ghibaudi E, de Gioia L, Salmona M, Cazzaniga G, Kurosaki M, Terao M & Garattini E (1995) Purification, cDNA cloning, and tissue distribution of bovine liver aldehyde oxidase J Biol Chem 270, 31037–31045 Yasuhara A, Akiba-Goto M, Fujishiro K, Uchida H, Uwajima T & Aisaka K... iodine and analyzed as described previously [41] Nucleotide analysis temperature was controlled with an Oxford ITC4 controller, and the sample was placed in an Oxford ESR900 helium flow cryostat For accurate g-value determination, the magnetic field was calibrated with an external standard (Li particles in LiF matrix), where the g-value is known to be 2.002293 [55] The exact microwave frequency was measured... ST, Rajagopalan KV & Handler P (1967) Purification and properties of xanthine dehydrogenase from Micrococcus lactilyticus J Biol Chem 242, 4108–4117 21 Gremer L & Meyer O (1996) Characterization of xanthine dehydrogenase from the anaerobic bacterium Veillonella atypica and identification of a molybdopterin- cytosine- dinucleotide- containing molybdenum cofactor Eur J Biochem 238, 862–866 22 Noriega C, Hassett... Molybdenum -cofactor- containing enzymes: structure and mechanism Annu Rev Biochem 66, 233–267 4 Johnson JL, Bastian NR & Rajagopalan KV (1990) Molybdopterin guanine dinucleotide: a modified form of molybdopterin identified in the molybdenum cofactor of dimethyl sulfoxide reductase from Rhodobacter sphaeroides forma specialis denitrificans Proc Natl Acad Sci USA 87, 3190–3194 5 Johnson JL, Indermaur LW & Rajagopalan... in human lymphocytes Mutat Res 337, 9–17 51 Baba T, Ara T, Hasegawa M, Takai Y, Okumura Y, Baba M, Datsenko KA, Tomita M, Wanner BL & Mori H (2006) Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection Mol Syst Biol 2, 2006.0008, doi:10.1038/msb4100050 52 Temple CA, Graf TN & Rajagopalan KV (2000) Optimization of expression of human sulfite oxidase and its... were inoculated at an attenuance at 600 nm of 0.1 in the absence and presence of 800 lm cinnamaldehyde in medium containing 1 gÆL)1 tryptone, 0.5 gÆL)1 yeast extract and 1 gÆL)1 sodium chloride at pH 4.0, cells were incubated at 37 °C and 150 r.p.m., and growth was followed at an attenuance of 600 nm Quantification of the cyanolyzable sulfur Acknowledgements Nine hundred microliters of 10 lm YagTSR in... measured with an EIP frequency counter from Microwave Inc (Milpitas, CA, USA) The spectra were simulated using the pepper function of easyspin [56], which works on the basis of the spin-Hamiltonian formalism For nucleotide analysis, AMP and CMP were released from FAD and MCD, respectively, by 15 min of incubation at 95% in the presence of 5% (v ⁄ v) sulfuric acid AMP and CMP produced during the reaction were... Stanich K (2006) Effect of vanillin on the fate of Listeria monocytogenes and Escherichia coli O157:H7 in a model apple juice medium and in apple juice Food Microbiol 23, 169–174 49 Zaldivar J, Martinez A & Ingram LO (1999) Effect of selected aldehydes on the growth and fermentation of ethanologenic Escherichia coli Biotechnol Bioeng 65, 24–33 50 Singh NP & Khan A (1995) Acetaldehyde: genotoxicity and...An MCD-binding aldehyde oxidoreductase from E coli M Neumann et al using 97% 5 mm ammonium acetate and 3% methanol at a flow rate of 1 mLÆmin)1 In-line fluorescence was monitored with an Agilent 1100 series detector with excitation at 383 nm and emission at 450 nm (Boblingen, Germany) To ¨ directly convert MCD to Form A, samples were incubated at 95 °C for 30 min in the presence of acidic iodine and . A periplasmic aldehyde oxidoreductase represents the first molybdopterin cytosine dinucleotide cofactor containing molybdo-flavoenzyme from Escherichia coli Meina. assays revealed that the YagT signal peptide leads to export to the periplasm and cleavage after amino acid 49, so YagTSRQ was predicted to contain a periplasmic protein