Interpretation of the constantly expanding body of genomic information requires that the function of each gene be established. Here we report the genomic analysis and structural modelling of a previously uncharacterized redox-metabolism protein UrdA (SO_4620) of Shewanella oneidensis MR-1, which led to a discovery of the novel enzymatic activity, urocanate reductase. Further cloning and expression of urdA, as well as purification and biochemical study of the gene's product UrdA and redox titration of its prosthetic groups confirmed that the latter is indeed a flavin-containing enzyme catalysing the unidirectional reaction of two-electron reduction of urocanic acid to deamino-histidine, an activity not reported earlier. UrdA exhibits both high substrate affinity and high turnover rate (Km << 10 μM, kcat = 360 s−1) and strong specificity in favour of urocanic acid. UrdA homologues are present in various bacterial genera, such as Shewanella, Fusobacterium and Clostridium, the latter including the human pathogen Clostridium tetani. The UrdA activity in S. oneidensis is induced by its substrate under anaerobic conditions and it enables anaerobic growth with urocanic acid as a sole terminal electron acceptor. The latter capability can provide the cells of UrdA-containing bacteria with a niche where no other bacteria can compete and survive.
Success in modern genomics results in a huge number of DNA texts, the vast majority of which remain without functional assignment (Lander et al., 2001; Claverie, 2005). Many computer algorithms utilizing genome database analysis have been used for the identification of protein-encoding genes and prediction of the function of their respective products (Schweikert et al., 2009). However, even for bacterial genomes such prediction can be performed only for c. 50% of all known genes, and it is not always accurate (Pennisi, 2003).
The genome sequencing of non-fermentative, facultatively anaerobic bacterium Shewanella oneidensis MR-1 was completed in 2002 and revealed 4758 genes (Heidelberg et al., 2002). One of these genes (so_4620) was predicted to encode fumarate reductase, SO_4620 (GenBank accession number AAN57580, UniProt accession number Q8CVD0_SHEON). Based on sequence similarity SO_4620 was described as a periplasmic lipoprotein consisting of two putative flavin-binding domains. One of the postulated domains is the C-terminal non-covalent FAD-binding domain present in the databases as ‘FAD_binding_2’, or Pfam domain PF00890 (Punta et al., 2012). This domain is homologous to the FAD-binding domains of succinate dehydrogenase and L-aspartate oxidase, both from Escherichia coli (Mattevi et al., 1999), and periplasmic fumarate reductase (SO_0970, Fcc3) from S. oneidensis MR-1 (Fig. S1). The second postulated domain is the N-terminal covalent FMN-binding domain known as ‘FMN_bind’, or Pfam domain PF04205 (Yeats et al., 2003), which is characterized by the very unusual covalent attachment of FMN to a Thr or Ser residue via the phosphoester bond distinguishing it from other flavin-binding domains (Hayashi et al., 2001; see for review: Verkhovsky and Bogachev, 2010, and references therein) (Fig. S2). This domain can also be found in Na+-translocating NADH:quinone oxidoreductase complex (NQR, in the NqrC subunit), Rhodobacter nitrogen fixing complex (RNF, in the RnfG subunit), and in nitric oxide reductase transcription regulator (NosR). As SO_4620 has never been characterized as a protein, its genomics-based assignment to fumarate reductase needs experimental verification.
In this paper we show that this previous assignment is in fact wrong and report a novel enzyme (urocanate reductase, UrdA) whose discovery was based on a prediction of the chemical nature of its substrate using the structural modelling and genomic context methods.
Results and discussion
Characterization of the recombinant target protein
The entire so_4620 gene was amplified from genomic DNA of S. oneidensis MR-1 and cloned into an expression vector p15BADc. The p15BADc system contained an arabinose-inducible promoter (araBAD), and the vector used provided insertion of the 6×His tag at the C-terminus of the expressed protein. Expression of so_4620 in E. coli cells resulted in a significant reduction of the cell growth and even partial lysis of the cells. Subsequent purification of the 6×His-tagged protein (SO_4620-6His) from the induced E. coli cells yielded a highly fragmented polypeptide containing no flavin prosthetic groups (data not shown). Thus, expression of the gene in E. coli did not result in production of properly folded holoprotein and was accompanied by proteolysis of the produced polypeptide. It is noteworthy that the expression of nqrC from Vibrio cholerae in E. coli also failed to produce the flavinylated protein (Barquera et al., 2001), which might indicate that the flavin attachment to the threonine residue requires additional factors not present in E. coli.
To produce the properly folded SO_4620-6His holoprotein, V. cholerae was chosen as a host bacterium, as this microorganism had been successfully used earlier for production of the proteins containing covalent FMN-binding domain (Barquera et al., 2001; Backiel et al., 2008; Fadeeva et al., 2008; Casutt et al., 2010). The expression of so_4620 in V. cholerae did not cause any negative effect on the cell growth. After the induction, the protein was purified separately either from soluble or membrane fractions of V. cholerae cells. In both cases the purification yielded bright yellow protein. The electronic absorption spectra of SO_4620-6His purified from both fractions were characteristic for flavoproteins indicating the presence of SO_4620-bound flavins (for the spectrum of the protein isolated from the membrane fraction see Fig. 1). The SO_4620-6His content in the membrane fraction was typically more than fourfold larger than in the soluble fraction (data not shown), indicating the association, but not too tight, of SO_4620-6His with the membrane. As the purified SO_4620-6His protein from the membrane fraction was of larger yield and of higher purity and stability than from the soluble fraction, the former was used for the further analysis.
SDS-PAGE of the purified SO_4620-6His showed one ∼60 kDa polypeptide (Fig. 2), which was in agreement with the calculated theoretical Mr of the mature 6×His-tagged SO_4620 protein (60.2 kDa). The electrophoresis data also allowed testing the presence of the predicted covalently bound flavin (Zhou et al., 1999; Hayashi et al., 2001): as shown in Fig. 2, a fluorescent band with Mr ≈ 60 kDa was detected under the UV illumination.
As mentioned above, the SO_4620 protein consists of two putative domains: (i) N-terminal FMN_bind (Pfam domain PF04205) and (ii) C-terminal FAD_binding_2 (Pfam domain PF00890) (Punta et al., 2012). Thus, it was likely that SO_4620 contains two flavin prosthetic groups typical for these domains: a covalently bound FMN residue and a non-covalently bound FAD. To test this assumption, the content ratio between the covalently and non-covalently bound flavins was determined in the recombinant SO_4620-6His protein using the method described in Barquera et al. (2002). The latter method is based on the denaturation of the studied protein by guanidine chloride, separation of the covalently and non-covalently bound prosthetic groups by centrifugation on a filter with appropriate cut-off value, and subsequent spectrophotometrical determination of the flavins in the fractions. The SO_4620-6His protein originally contained ∼28 nmol of flavins per mg of protein (100%). After the centrifugation, ∼55% of the flavins were found in the low-molecular-mass fraction, while the protein fraction retained ∼42% of the flavins (the total flavin content was less than the expected 100%, apparently because of a small protein fraction adsorbed on the filter and lost during concentration). These data indicated that the ratio between the covalently and non-covalently bound flavins in SO_4620-6His was about 1:1, which was in full agreement with the prediction made.
The non-covalently bound flavin was identified in SO_4620-6His using HPLC analysis. The acid extract of the protein contained only FAD, while FMN and riboflavin were not detected (data not shown). Thus, we concluded that SO_4620-6His contains two flavin prosthetic groups: one non-covalently bound FAD and one covalently bound flavin (most likely a FMN residue).
The presence of the flavin prosthetic groups in the protein clearly pointed to an oxidoreductase activity. Despite the prediction of SO_4620 as a fumarate reductase, the former revealed, to our surprise, no fumarate reductase or succinate dehydrogenase activity. These data were in agreement with Maier et al. (2003), where it was shown that another protein Fcc3 (SO_0970) is the sole physiological fumarate reductase in S. oneidensis MR-1. Testing SO_4620-6His with other substrates, both electron donors and acceptors (various L- and D-amino acids, sugars, organic acids and other metabolites, as well as complex mixtures of organic substances such as yeast extract, tryptone, fatty acids mixtures, and fermentation products of different bacteria) did not reveal any dehydrogenase or reductase activity either.
Structural modelling of the putative catalytic centre of SO_4620
Another approach to determine a putative enzymatic activity of SO_4620 was to predict the chemical nature of its substrate based on the structural modelling of its catalytic centre (Lukk et al., 2012). This could be done using the catalytic centre of periplasmic fumarate reductase (Fcc3, SO_0970) from S. oneidensis MR-1 as a template, as the FAD-binding domain of SO_4620 is very similar to the FAD-binding domain of Fcc3 (39% identity, 58% similarity, Fig. S1). Fcc3 is a well-characterized enzyme with its 3D structure for the enzyme–substrate complex and the catalytic mechanism known (Leys et al., 1999; Reid et al., 2000). In the catalytic centre of Fcc3, the substrate fumarate (or product succinate) is bound in close proximity to FAD (Fig. 3A). Five residues form contacts with the fumarate carboxylic groups: His364 and Thr376, with the C1-group, and Arg401, His503 and Arg544, with the C4-group. The short distance between FAD and fumarate enables hydride ion transfer from the flavin to the fumarate C2-atom to reduce the C=C bond (Leys et al., 1999; Reid et al., 2000). The hydride transfer is followed by the protonation of the C3-atom through the proton transfer pathway formed by Glu377, Arg380 and Arg401 side-chains, thus resulting in the formation of succinate (Leys et al., 1999; Reid et al., 2000; Pankhurst et al., 2006).
In the structural model of the putative catalytic centre of SO_4620 (Fig. 3B) the positions of Arg560, Arg411 and His520 are identical to the positions of respective residues responsible for binding the fumarate C4-carboxylic group in Fcc3 (see also: Fig. S1). This indicated that the hypothetical substrate of SO_4620 contains a carboxylic group. In SO_4620 there also exists the proton transfer pathway (consisting of Arg252, Asp388 and Arg411) analogous to the one in Fcc3, suggesting that the SO_4620 substrate (or product) also contains a C=C bond. However, the residues involved in the binding of the C1-carboxylic group of the substrate in Fcc3 are not conserved in SO_4620: His364 is replaced by Tyr373, and Thr376 by Ile386 respectively. This argued against the presence of the second carboxylic group in the hypothetical SO_4620 substrate. Thus, we could propose that the putative substrate of SO_4620 is either R−CH=CH−COOH (if the enzyme is a reductase) or R−CH2−CH2−COOH (if the enzyme is a dehydrogenase), where R is not a carboxylic group. This analysis allowed us to restrict significantly the area of the possible candidates for the SO_4620 substrate. However, the diversity of natural compounds with formulas R−CH=CH−COOH or R−CH2−CH2−COOH was still too high; it was tempting to determine, which of these two classes of compounds is the physiological substrate of SO_4620 (or, in other words, whether this enzyme is a reductase or dehydrogenase in vivo) and thus to restrict the search area even further.
Redox titration of SO_4620
Since the major factor defining the directionality of the reaction catalysed by Fcc3-like enzymes (that is, whether the enzyme is a reductase or a dehydrogenase) is the midpoint redox potential (Em) of their FAD cofactor (Leys et al., 1999; Turner et al., 1999), we performed the spectroelectrochemical redox titration of the studied protein. A typical set of spectral changes upon equilibrium redox titration of SO_4620-6His at pH 7.5 is shown in Fig. 4A. The redox titration curve for the first derivative of the optical density with respect to redox potential (ΔA/ΔEh, Fig. 4B) showed a sharp peak characteristic for the two-electron transition of one of the flavins (Fl1) and a broad, asymmetric wing indicating two successive one-electron transitions of the other flavin (Fl21 and Fl22 respectively). The redox titration profiles (Fig. 4C) and their respective spectral components (Fig. 4D) of the three transitions were found using the global fit of the optical changes in the whole spectral and redox potential regions (see: Experimental procedures). Component Fl1 had a low-potential, n= 2 transition (Em = −280 mV) with a spectrum characteristic for the two-electron reduction of a flavin species. The two components of Fl2 had, for the first component (Fl21), a spectrum characteristic for a flavin anion semiquinone emerging with Em = −225 mV and, for the second component (Fl22), a spectrum reflecting the formation of the two-electron reduced form with Em = −250 mV. The two latter Em values for the two consecutive one-electron reduction steps of a flavin gave the stability constant for the flavin semiquinone, KS = 2.7. The components Fl1 and Fl22 had similar amplitudes at 450 nm (Fig. 4D) supporting our conclusion that SO_4620 contains two flavins in equimolar stoichiometry.
It is noteworthy that both spectral properties of Fl1 (the characteristic shoulder at 477 nm) and the two-electron nature of its oxidoreduction were identical to the FAD properties of periplasmic oxidoreductases containing the FAD_binding_2 domain (Mikoulinskaia et al., 1999; Turner et al., 1999). At the same time the optical spectra, one-electron nature of oxidoreduction, and stabilization of the anionic form of flavosemiquinone resembled the properties of covalently bound FMN residue in the NqrC subunit of Na+-translocating NADH:quinone oxidoreductase (Bogachev et al., 2009). Moreover, redox-titration of a cytoplasmic variant of the studied protein (SO_4620_cyt, see Supporting information) exhibiting lower content of the covalently bound flavin resulted in the proportional decrease of the amplitudes of both redox transitions of Fl2 without an effect on the amplitude of the Fl1 transition (Fig. S3). These data let us identify Fl1 as non-covalently bound FAD and Fl2 as covalently bound FMN residue respectively.
All three midpoint potential values found for SO_4620 were significantly lower than the Em value [∼0 mV vs. NHE (Sato et al., 1999)] expected for the hypothetical reaction:
This indicated that in vivo, SO_4620 should function as a reductase, and therefore its natural substrate is a compound with structure:
Identification of the substrate of SO_4620 by genomic context methods
The diversity of natural unsaturated organic acids with structure (1) is not too high, but it was possible to restrict the search area even further by using the genomic context methods. Bacterial genes are known to be often clustered according to their function (Lawrence, 1999) and this fact can be used for the protein function prediction (Ramazzina et al., 2008; Rentzsch and Orengo, 2009). In the S. oneidensis MR-1 chromosome, the two genes flanking so_4620 encode a Fe/S biogenesis protein (so_4619) and a putative nucleoside-specific channel-forming protein (so_4621), respectively, neither of which are directly related to a putative SO_4620 substrate. Therefore, we searched for the SO_4620 homologues with the similar structure of the catalytic centre among various bacterial genomes. There are dozens of such genes in genome databases found mostly in Shewanella spp., Fusobacterium spp. and Clostridium spp., including the human pathogen Clostridium tetani (NP_782428). In the chromosomes of Shewanella woodyi and Shewanella pealeana the two so_4620 homologue-flanking genes are swoo_3913 and spea_3575, both encoding putative phenylalanine/histidine ammonia lyases. These enzymes catalyse deamination of phenylalanine (tyrosine) or histidine and produce cinnamic (p-coumaric) or urocanic acid respectively (Fig. 5A) (Mehler and Tabor, 1953; MacDonald and D'Cunha, 2007). The three latter compounds have structure (1) and could serve as the SO_4620 substrates; therefore, SO_4620 could be a reductase of cinnamic (p-coumaric) or urocanic acid. We found that SO_4620-6His was unable to reduce cinnamate, but showed very high reductase activity in the presence of urocanate (see: below).
Enzymatic activity of SO_4620
Analysis of the catalytic properties of SO_4620-6His revealed that it reduces urocanic acid with high rate (kcat = 360 s−1 at pH = 7), the value well comparable with the data for fumarate reduction by periplasmic fumarate reductases (Morris et al., 1994; Dobbin et al., 1999). It was also established that SO_4620-6His has very high affinity to urocanate. The apparent Km for urocanate was << 10 μM (a method limitation prevented from more precise determination of the value). On the other hand, SO_4620-6His showed no activity with other unsaturated organic acids tested (acrylic, cinnamic, crotonic, fumaric and orotic acids). These data allowed us to conclude that urocanate is indeed the natural substrate of SO_4620.
Thus, the activity of SO_4620 appeared as the reduction of C2=C3 bond of urocanate yielding deamino-histidine (Fig. 5B). Using paper chromatography, we found that the urocanate reductase activity of SO_4620-6His is accompanied by disappearance of urocanate and accumulation of deamino-histidine in the reaction medium (data not shown). At the same time there was no dehydrogenase activity of SO_4620-6His at a detectable level in the presence of deamino-histidine. This clearly showed that SO_4620 catalyses an essentially unidirectional reaction, i.e. it works as a molecular diode (Sucheta et al., 1992). This conclusion was also consistent with the very low midpoint potential found for the FAD cofactor in SO_4620 (Fig. 4).
We note that the physiological electron donor for SO_4620 remains yet unknown. We can hypothesize that in vivo the urocanate reductase SO_4620 [like fumarate reductase Fcc3 (Gao et al., 2010)] oxidizes the membrane-bound tetrahaem cytochrome c (CymA, SO_4591), and the covalently bound FMN residue of SO_4620 serves as an electron carrier between the cytochrome c and non-covalently bound FAD.
Physiological role of SO_4620
Similarity in structure and in cellular localization of Fcc3 and SO_4620 may indicate that these two enzymes also play similar roles in the bacterial metabolism. Fcc3 is a terminal enzyme of the electron transport chain of S. oneidensis MR-1, and it enables anaerobic growth using fumarate as the electron acceptor (Maier et al., 2003). We proposed that S. oneidensis MR-1 is also able to use urocanate as the electron acceptor for anaerobic growth. The high affinity of SO_4620 to urocanate can facilitate this function making the latter periplasmic protein a scavenger for this extracellular electron acceptor.
Shewanella oneidensis cells are unable to use lactate as the sole source of carbon and energy for anaerobic growth, and the addition of fumarate initiates growth (Maier et al., 2003; Hunt et al., 2010). In accordance with the proposed function of the SO_4620 protein, addition of urocanate instead of fumarate also promoted the anaerobic growth with similar yield (Fig. 6). These results altogether indicated that S. oneidensis cells are able to use urocanate as a sole terminal electron acceptor for anaerobic growth. In order to compare the efficiency of urocanate as the terminal electron acceptor with other, more traditional substrates of anaerobic respiration, we tested growth curves of S. oneidensis MR-1 cells with different electron acceptors. The anaerobic growth rate of S. oneidensis MR-1 with urocanate was similar to one with fumarate, TMAO, DMSO and NaNO3 and was considerably higher compared with iron(III) citrate (Fig. 6B). However, the urocanate-supported growth exhibited a prolonged lag-phase (∼2 h), which lacked when the other electron acceptors were used, pointing out on a possible urocanate-specific induction of certain enzymatic activities (see: below).
To prove the vital role of SO_4620 for anaerobic growth of S. oneidensis MR-1, a strain (A21) with disrupted synthesis of this protein was constructed. In the SO_4620-deficient strain the urocanate reductase activity measured in the crude extract from cells anaerobically grown on a mixture of urocanate and fumarate (or urocanate and nitrate) was about 20-fold lower than in the wild type (Fig. 7), indicating that SO_4620 is the major or even sole urocanate reductase present in S. oneidensis at the conditions studied. As shown in Fig. 6, the absence of SO_4620 in the A21 strain did not affect its ability to grow in the presence of fumarate. At the same time, the A21 strain failed to grow anaerobically on urocanate, indicating that SO_4620 is essential for the ability to use this compound as the sole terminal electron acceptor. Introduction of a plasmid bearing the so_4620 gene (p15Cm_LF1) into the S. oneidensis A21 strain resulted in a recovery of the urocanate reductase activity (0.45 ± 0.07 μmol min−1 mg prot.−1) as well as in the ability of S. oneidensis A21/p15Cm_LF1 cells to grow anaerobically using urocanate as a terminal electron acceptor (Fig. 6A). This rules out the possibility of polar effects of the so_4620 lesion in S. oneidensis A21.
Apparently, histidine should serve as a natural source of urocanate for anaerobic bacterial growth, as this amino acid is a component of the majority of proteins and some peptides and it is widespread in the environment. Urocanate can be produced from histidine by histidine ammonia-lyases; the latter enzymes are encoded by the genes adjacent to the gene of urocanate reductase in genome of S. woodyi and S. pealeana (see: above). It is noteworthy that according to the predictions made by SignalP or PSORTb programs, these histidine ammonia-lyases (Swoo_3913 and Spea_3575 respectively) are periplasmic proteins; such location is rather unusual for the enzymes of the histidine degradation pathway. Thus, it can be proposed that the following reaction sequence:
is carried out in periplasm of these bacteria allowing reoxidation of NADH formed during the anaerobic metabolism.
We note that the S. oneidensis MR-1 genome also contains at least one gene encoding periplasmic phenylalanine/histidine ammonia-lyase (so_3299) adjacent to the gene for the SO_3301 protein, the latter similar to the FAD-binding domain of SO_4620 (Heidelberg et al., 2002). However, as can be seen in Fig. 6, histidine failed to substitute urocanate as an electron acceptor for anaerobic growth of this microorganism. It can be proposed that either the growth conditions used are ineligible for histidine ammonia-lyase reaction, or S. oneidensis MR-1 cannot produce sufficient amounts of urocanate from histidine and the activity of other bacteria is necessary for this production. Further experiments are required to clarify this question.
It is noteworthy that S. oneidensis MR-1 also contains enzymes of the canonical histidine degradation pathway (Bender, 2012), including histidine ammonia-lyase HutH (SO_0098), urocanate hydratase HutU (SO_0097) and imidazolonepropionase HutI (SO_0095), catalysing the following sequence of reactions:
In contrast to the above mentioned periplasmic pathway , the enzymes involved in the histidine degradation pathway (2) are located in cytoplasm and show no homology to the SO_4620 protein.
As pointed out above, the prolonged lag-phase of the anaerobic growth of S. oneidensis MR-1 with urocanate might indicate induction of certain enzymatic activities obligatory under the given conditions; the off-hand guess is urocanate reductase. To test the latter assumption, we measured the level of urocanate reductase activity in S. oneidensis MR-1 under various growth conditions. In the absence of urocanate under aerobic or anaerobic conditions, the urocanate reductase activity in the cell lysates was rather low (0.01–0.04 E mg prot.−1, Fig. 7), whereas in the presence of urocanate, a significant induction of the urocanate reductase activity was observed in the anaerobically grown cells (up to 0.65–0.75 E mg prot.−1). We note that the urocanate-stimulated induction of urocanate reductase activity did not depend on the presence or absence of any other anaerobic electron acceptor but was fully arrested under aerobic growth conditions (Fig. 7). The replacement of urocanate by histidine did not lead to the induction of urocanate reductase activity (data not shown). Thus, we concluded that both anaerobic growth conditions and the presence of urocanate are required for the induction of urocanate reductase activity in S. oneidensis MR-1. As shown in Fig. 7, this anaerobic induction was totally absent in the S. oneidensis A21 strain with disrupted synthesis of SO_4620 suggesting that the effect is indeed provided by the activity of SO_4620. Induction of the SO_4620 activity by urocanate confirms the conclusion that urocanate is a natural substrate for this protein.
Thus, our in vivo experiments showed that S. oneidensis MR-1 is able to use urocanate as a sole terminal electron acceptor for the anaerobic growth and that this ability was due to the activity of SO_4620. Based on this conclusion, we propose to introduce a new name UrdA for this novel enzyme. The urocanate reductase activity can provide the UrdA-expressing bacteria with a niche where no other bacteria can compete and survive.
In this work we showed that the UrdA (SO_4620) protein of S. oneidensis MR-1 is a urocanate reductase. To the best of our knowledge, no enzyme with such catalytic activity has been reported earlier (see: e.g. Arkhipova and Akimenko, 2005). Moreover, the ability of bacteria to use urocanate as a terminal electron acceptor for anaerobic growth was never recognized. Based on the structural modelling, redox titration and genomic context methods, we predicted the catalytic activity of this enzyme and its physiological role in bacterial cells. The subsequent experiments confirmed this prediction.
Bacterial strains and growth conditions
The bacterial strains used in this study are listed in Table S1. E. coli, V. cholerae and S. oneidensis MR-1 cells were grown aerobically in LB medium at 37°C, 32°C and 28°C respectively. Antibiotics for E. coli were ampicillin (125 μg ml−1), kanamycin (50 μg ml−1) and tetracycline (10 μg ml−1), for V. cholerae, ampicillin (150 μg ml−1), kanamycin (50 μg ml−1) and streptomycin (50 μg ml−1), and for S. oneidensis MR-1, kanamycin (50 μg ml−1), rifampicin (10 μg ml−1) and tetracycline (5 μg ml−1).
To test different electron acceptors, S. oneidensis MR-1 cells were grown anaerobically at 28°C in a medium containing 0.225 g l−1 K2HPO4, 0.225 g l−1 KH2PO4, 0.46 g l−1 NaCl, 0.225 g l−1 (NH4)2SO4, 0.117 g l−1 MgSO4·7H2O, 20 mM L-lactate, 0.05% yeast extract, 100 mM HEPES/NaOH (pH 7.2) and 20 mM of an electron acceptor (Hunt et al., 2010).
Cloning and expression of so_4620 from S. oneidensis MR-1
The so_4620 gene was amplified from genomic DNA of S. oneidensis MR-1 using PCR with primers shew_frd_dir and shew_frd_rev (Table S1). The amplified 1774 bp fragment was cloned into a TA-cloning pGEM-T vector (Promega) resulting in the pG_FRD8 plasmid. Further, we subcloned so_4620 into a vector, allowing expressing of the gene with C-terminus 6×-His tag fusion in different bacteria (E. coli, V. cholerae and S. oneidensis MR-1). To construct such a vector, the Acc16I–BstZI fragment from pACYC177 was cloned into pBAD_MycC-His (Invitrogen) between its Acc16I and BstZI sites resulting in the p15BADc plasmid (a pBAD_MycC plasmid bearing p15A ori instead of pUC ori). The direct cloning of so_4620 into the p15BADc vector using the introduced by PCR NcoI and EcoRI sites was hampered by the intrinsic sites for these endonucleases in the so_4620 sequence. Therefore, we performed this task by several consecutive subcloning procedures. The internal AfeI–BstZI fragment (1411 bp) of the cloned so_4620 was deleted from pG_FRD8 resulting in the pG8Δ2 plasmid. Then the NcoI–EcoRI part from pG8Δ2 was subcloned into the p15BADc vector resulting in the p15BADc_Fl6 plasmid. After this, the EheI–EcoRV fragment from pG_FRD8 was inserted into p15BADc_Fl6 between its EheI and EcoRV sites and the p9D-209 plasmid with the correctly oriented insert was selected. Proper reconstitution of so_4620 in the p9D-209 plasmid was verified by sequencing. Finally, the kanamycin-resistance cassette from pUC4K was inserted into p9D-209 at the SphI site resulting in the p9D-209Km1 plasmid.
Construction of a gene for soluble, cytoplasmic variant of SO_4620
The gene for the mature part of the SO_4620 with Gly20Met and Cys21Gly substitutions was amplified from the p15BADc_Fl6 plasmid using PCR with primers 9DL_frd_dir and pBAD_rev. The amplified 418 bp fragment was cloned into a pGEM-T vector, resulting in the pG_9DL6 plasmid. Then the NcoI–EcoRI fragment from pG_9DL6 was subcloned into a pBAD/Myc-His vector using the NcoI and EcoRI restriction sites, resulting in the pBADc_246 plasmid. The AflII–EheI fragment from the p9D-209Km1 plasmid was replaced by the AflII–EheI (1234 bp) part of pBADc_246, resulting in the p9DL plasmid bearing the truncated form of the so_4620 gene.
Construction of a gene for soluble, periplasmic variant of SO_4620
To replace the signal peptide of SO_4620 (amino acid residues 1–21) with the signal peptide of alkaline phosphatase from E. coli (amino acid residues 1–25 of PhoA), the 64–390 bp part of so_4620 was sequentially extended from the 5′-end by three successive PCR reactions with forward primers dir1, dir2 or dir3 and the reverse primer pBAD_rev. The final PCR product was cloned into a pAL-TA cloning vector (‘Evrogen’) resulting in the pAL_sintB plasmid and verified by sequencing. Then, the NcoI–EcoRI fragment of this plasmid was subcloned into pBAD/Myc-His using the NcoI and EcoRI restriction sites, resulting in the pB_sint plasmid. Finally, the AflII–EheI part of the pB_sint plasmid was used for the replacement of the AflII–EheI part of p9D-209Km1 resulting in the pBSF2 plasmid, bearing the gene for SO_4620 protein with the signal peptide of alkaline phosphatase from E. coli.
The constructed plasmids were transferred into V. cholerae O395N1 toxT::lacZ cells by electroporation.
Isolation of a rifampicin-resistant strain of S. oneidensis MR-1
Spontaneous rifampicin-resistant strain of S. oneidensis MR-1 was isolated by selection of RfR colonies on agar plates with gradually increasing rifampicin concentrations. As a result, the S. oneidensis RIF strain was selected, which is able to grow at 10 μg ml−1 rifampicin. It was shown previously (Myers and Myers, 2000) that rifampicin resistance phenotype of S. oneidensis strains may lead to decreased menaquinone levels and as a result to a hampered anaerobic growth on fumarate. To test this possibility, the anaerobic growth curves on fumarate for the wild type strain as well as for the obtained RfR strain of S. oneidensis were examined. As shown in Fig. S4 the growth curves for these two strains were very similar, which argued against possible side-effects of the RfR selection used in this study.
Construction of an SO_4620-deficient S. oneidensis strain
A kanamycin-resistance cassette was ligated into the AgeI site of the pG_FRD8 plasmid and a Km-containing plasmid (pG_FRD_Km) bearing so_4620 together with the unidirectionally transcribing kanamycin-resistance cassette was selected. The SacII–SalI part of so_4620::Km from pG_FRD_Km was subcloned into a suicide vector pKNOCK-Tc having R6K γ-origin (Alexeyev, 1999), which cannot replicate in S. oneidensis MR-1 (Thompson et al., 2002). Description of the pKnFRD construction is shown as a diagram in Fig. S5. The resulting pKnFRD plasmid was introduced into S. oneidensis RIF by conjugative mating using E. coli SM10λpir as a donor. Initial selection was performed on the plates with LB medium containing rifampicin and kanamycin. Further, more than 100 clones obtained by the initial selection were tested to tetracycline resistance. It was found that all the obtained clones have RfR KmR TcR phenotype, i.e. they are the result of a single-crossover integration of the pKnFRD plasmid. All further attempts to obtain a double-crossover-introduced mutation were unsuccessful. As a truncated from both 5′- and 3′-end version of so_4620 was used for mutagenesis, even a single-crossover integration of the pKnFRD plasmid should lead to the gene inactivation. One such clone (A21) was further used as a SO_4620-deficient S. oneidensis strain. Proper localization of pKnFRD in the S. oneidensis A21 chromosome was verified by PCR analysis. It was also found that the introduced mutation is stably maintained in this strain of S. oneidensis.
Complementation of S. oneidensis A21 strain with plasmid-coded SO_4620 protein
The so_4620-containing DNA fragment was amplified by PCR with the Tersus PCR kit and primers FRD_BamHI_dir and FRD_SphI_rev (see Table S1) using genomic DNA of S. oneidensis MR-1 as the template. The amplified 1.9 kb fragment was cloned into the pAL-TA vector, resulting in the pATLF15 plasmid. The BamHI–SphI fragment of this plasmid was subcloned into the pACYC184 vector digested by the same restriction nucleases, resulting in plasmid p15Cm_LF1. S. oneidensis A21 cells were transformed by the p15Cm_LF1 plasmid using electroporation.
The 6×His-tagged SO_4620 variants were isolated from E. coli or V. cholerae O395N1 cells bearing the appropriate plasmid. For so_4620 induction, the cells were grown to the mid-exponential phase (OD600 = 0.3–0.4). Then the growth medium was supplemented with 0.05% of L-arabinose and the cells were grown further in this medium for 3 h. The cells were harvested by centrifugation (10 000 g, 10 min) and washed twice with medium containing 300 mM NaCl, 10 mM Tris-HCl and 5 mM MgSO4 (pH 8.0). The cell pellet was suspended in medium containing 300 mM NaCl, 20 mM Tris-HCl, 5 mM MgSO4 and 5 mM imidazole (pH 8.0), and the suspension was passed twice through a French pressure cell (16 000 psi). Unbroken cells and cell debris were removed by centrifugation (12 000 g, 10 min). The obtained crude cell extract was separated on membrane fraction and soluble fractions by centrifugation at 200 000 g (60 min). Both fractions were used for further protein purification.
His-tagged SO_4620 (SO_4620-6His) was purified using affinity chromatography. To purify SO_4620-6His from the membrane fraction, the membrane vesicles were suspended in solution A (350 mM KCl, 5 mM imidazole, 20 mM Tris-HCl, pH 8.0), solubilized with 1% LDAO, and centrifuged at 200 000 g for 60 min. The supernatant was loaded onto a Ni-NTA column equilibrated with solution A containing 0.2% LDAO. The column was successively washed with solution A containing 0.2% LDAO and 10 mM imidazole, and with solution A containing 0.05% n-dodecyl-β-d-maltoside (DDM) and 20 mM imidazole. Then SO_4620 was eluted from the column with solution A containing 0.05% DDM and 100 mM imidazole.
To purify SO_4620-6His, SO_4620_cyt and SO_4620_peri from the soluble fraction, the same procedure was used, but without addition of detergents to any of the column buffers. The proteins obtained were concentrated and frozen at −80°C until use.
The products of urocanate reductase reaction as well as urocanic acid and deamino-histidine standards were separated by paper chromatography in solvent consisting of the isobutanol–acetone–formic acid–water mixture (160:160:1:39, vol. by vol.) as described (Sen et al., 1962). Imidazolyl derivatives were detected by spraying the chromatograms with the Pauly diazo reagent. The Rf values for the urocanic acid and deamino-histidine standards were 0.32 and 0.60 respectively.
Redox titration of the SO_4620 variants was carried out using an optically transparent, thin-layer electrode cell as described (Bogachev et al., 2006). Potentials within the range of −360 to 0 mV versus NHE were set with ±20 mV steps during both oxidative and reductive titration using a PAR263A potentiostat (Princeton Applied Research). Optical absorption spectra were recorded in the spectral range of 350–600 nm at a succession of redox potentials. At each potential step, the onset of equilibrium on the working electrode was determined, as the changes in the cell current and optical density at 450 nm became no longer significant (20–35 min/step). The titrations were performed at 21°C. The set-up was PC-controlled in a fully automated regime.
To accelerate redox equilibrium between the working electrode surface and the protein, cobalt sepulchrate (Em = −350 mV), pentaamminechlororuthenium (Em = −130 mV) and hexaammineruthenium (Em = +50 mV) were used as redox mediators. No optical contribution from the mediators was detected in the studied spectral range. All redox potentials quoted refer to NHE.
For the spectroelectrochemical experiments SO_4620_peri and SO_4620_cyt samples were diluted with buffer [150 mM KCl, 50 mM HEPES/Tris (pH 7.5), 200 μM cobalt sepulchrate, 400 μM pentaamminechlororuthenium(III) chloride and 200 μM hexaammineruthenium(III) chloride] and concentrated on centrifugal Amicon filters (30 kDa cut-off, Millipore) up to ∼150 μM SO_4620; the mixture (∼40 μl) was used to fill the electrochemical cell.
To determine the thermodynamic and optical properties of the flavin cofactors of SO_4620, the spectroelectrochemical titration data sets were fitted, taking into account explicitly that the protein contains two bound flavins: one, undergoing two-electron transition and the second, two successive one-electron transitions. For the two-electron flavin (Fl1) the redox potential-dependence of its absorbance changes can be expressed as:
where ΔAλ is the absorbance change at wavelength λ, Eh is the ambient redox potential (mV), is the extinction coefficient for the transition Fl1 ↔ Fl1H2 at wavelength λ, is the midpoint redox potential (mV) of Fl1. For the second flavin (Fl2) two components contribute to its absorbance changes, reflecting the appearance of flavin semiquinone between and and two-electron reduced flavin at lower potentials:
where and are the extinction coefficients for the transitions Fl2 ↔ Fl2● and Fl2 ↔ Fl2H2 at wavelength λ respectively. and are the midpoint redox potentials (mV) of one-electron and two-electron reduction of Fl2 respectively.
Determination of enzymatic activities
The reductase activity of SO_4620-6His was determined at 606 nm by following the oxidation of reduced methyl viologen [ε606 = 13.7 mM−1 cm−1 (Watanabe and Honda, 1982)]. The assay was performed at 25°C in a 3.2 ml anaerobic cuvette. The standard assay mixture contained 100 mM NaCl, 50 mM HEPES/Tris pH 7.0, 0.025% DDM, 10 μM–1 mM of an electron acceptor, and 1 mM methyl viologen. Methyl viologen was reduced with sodium dithionite until an absorbance at 606 nm of about 1.5 was obtained. One unit of the enzyme activity was equivalent to the oxidation of 2 μmol of methyl viologen per minute.
The dehydrogenase activity of SO_4620-6His was measured at 600 nm in the presence of phenazine methosulphate and DCPIP as electron acceptors (Ells, 1959). The assay mixture contained 100 mM NaCl, 50 mM HEPES/Tris (pH 7.0), 0.025% DDM, 1 mM of a substrate studied, 2 mM phenazine methosulphate and 25 μM DCPIP. The assay was performed at 25°C.
The optical measurements were performed using a Hitachi 557 spectrophotometer.
Identification of non-covalently bound flavin
Non-covalently bound flavin was extracted from purified SO_4620-6His as described previously (Zhou et al., 1999). The extracted flavins were separated by reverse-phase HPLC using a Prontosil 120-5C18AQ column (dimensions: 75 × 2 mm) (Bogachev et al., 2006). A linear gradient of methanol (total sweep from 10 to 70% by vol., 25 min) in 5 mM MES-Tris, pH 6.0, with the flow rate of 0.1 ml min−1 was used. Flavins were detected at 360 nm. The retention times for FAD, FMN and riboflavin standards were 10.1, 11.9 and 14.8 min respectively.
Extinction coefficients of SO_4620
To determine the extinction coefficient of SO_4620-6His, a sample with the protein at known concentration was denaturated by SDS (1%, 5 min incubation at 80°C). The optical spectra of the sample were taken before and after denaturation. Using known extinction coefficients for FAD and FMN in solution, the extinction coefficient for the native SO_4620-6His protein was determined as ε452–600 = 26.3 mM−1 cm−1.
Protein content was determined by the bicinchoninic acid method using bovine serum albumin as a standard.
Prediction of protein localization in a bacterial cell was made using TMHMM (Krogh et al., 2001) and PSORTb (Yu et al., 2010) programs. Properties of signal peptides were analysed by SignalP (Petersen et al., 2011) and LipoP (Rahman et al., 2008) servers. Domain architecture was examined using Pfam database (Punta et al., 2012). Sequence alignments were performed by ClustalW (Larkin et al., 2007) program. Structural modelling of SO_4620 catalytic centre was performed using the MODELLER software (Eswar et al., 2006).
This work was supported by Biocentrum Helsinki (project number 7919028), the Sigrid Jusélius Foundation (project number 4700827), the Academy of Finland (project number 115108) and the Russian Foundation for Basic Research (project number 10-04-00352). We thank Vivek Sharma for his assistance with the application of the MODELLER program and Dr C.C. Häse for providing us with the V. cholerae O395N1 toxT::lacZ strain.