Correspondence: Gabriele Diekert, Institut für Mikrobiologie, Friedrich-Schiller-Universität Jena, Lehrstuhl für Angewandte und Ökologische Mikrobiologie Philosophenweg 12, 07743 Jena, Germany. Tel.: +49 3641 949300; fax: +49 3641 949302; e-mail: firstname.lastname@example.org
The O-demethylases of anaerobes are corrinoid-dependent, ether-cleaving methyltransferase enzyme systems consisting of four components. The interaction of the O-demethylase components of the acetogenic bacterium Acetobacterium dehalogenans was studied by protein mobility on native PAGE, far-Western blot analysis and yeast two-hybrid screen. Using native PAGE and far-Western blot, the interaction of the activating enzyme (AE) with its substrate, the corrinoid protein (CP), could be observed. The interaction occurred with four different CPs of A. dehalogenans and a CP from Desulfitobacterium hafniense DCB-2, all involved in ether cleavage. In the corrinoid reduction assay, the AE reduced all CPs tested. This result indicates a broad substrate specificity of the AE of A. dehalogenans. In addition, an interaction of the A. dehalogenans CP of the vanillate-O-demethylase with the two methyltransferases of the same enzyme system was observed. The interaction of the ether-cleaving methyltransferase with the CP appeared to be significantly less pronounced than that reported for the homologous methanol and methylamine methyltransferase systems of methanogenic archaea.
In corrinoid-dependent methyltransferase reactions, the cobalt of the corrinoid cofactor has to be in its super-reduced [CoI]-state. Because the redox potential of cob(II)alamin/cob(I)alamin is usually extremely low (ESHE < −500 mV), autoxidation of the corrinoid cofactor to the inactive [CoII]-form may occur in the cells even under strictly anoxic conditions (Daas et al., 1993; Jarrett et al., 1998; Menon & Ragsdale, 1999; Siebert et al., 2005). Hence, a ‘repair’ mechanism is required to reduce the cobalt for sustaining the catalytic reaction cycle of the methyltransferase reaction in archaea and bacteria (Banerjee & Ragsdale, 2003). In methyltransferase systems, the corrinoid reduction is mediated in an ATP-dependent reaction by reductive activators of corrinoid-dependent enzymes (RACE proteins) that are members of the COG3894 protein family (Schilhabel et al., 2009). The characterization of such an activator was described for the ether-cleaving O-demethylase enzyme systems of Acetobacterium dehalogenans (Schilhabel et al., 2009) and Desulfitobacterium hafniense DCB-2 (Studenik et al., 2012), for the methylamine methyltransferases of Methanosarcina barkeri (Ferguson et al., 2009) and for the corrinoid-iron/sulfur protein (CoFeSP) of Carboxydothermus hydrogenoformans (Hennig et al., 2012).
O-demethylases are the key enzymes of the methylotrophic phenyl methyl ether metabolism of acetogens (Kaufmann et al., 1997; for a review, see Drake et al., 2006) and were also found in Desulfitobacteria (Neumann et al., 2004; Studenik et al., 2012). They catalyze the first reaction of the phenyl methyl ether conversion, namely the ether bond cleavage and the transfer of the substrate's methyl group to tetrahydrofolate (FH4). O-demethylase enzyme systems consist of four different proteins: two methyltransferases (MT I and MT II), a corrinoid protein (CP), and an activating enzyme (AE) (Kaufmann et al., 1997). MT I mediates the cleavage of the substrate's ether bond and the transfer of the methyl group to the super-reduced corrinoid cofactor of CP. In a second reaction step, catalyzed by MT II, the methyl group is transferred from CP to FH4 yielding [CoI]-CP and methyl-FH4 as an intermediate of acetate formation in the Wood–Ljungdahl pathway. To reduce inadvertently oxidized corrinoid cofactor, the RACE protein AE is required. In methanogenic methyltransferase systems, a similar mechanism is involved in methane formation from methyl substrates (Burke & Krzycki, 1997; Ferguson & Krzycki, 1997; Sauer & Thauer, 1997; Tallant & Krzycki, 1997; Ferguson et al., 2000).
A prerequisite for completion of the catalytic cycle of the O-demethylase reaction is the interaction of the four components, which, however, could only be purified as separate proteins so far. The CP plays a crucial role, because it has to interact with the three other components, namely in its super-reduced state with MT I, in its methylated form with MT II, and in its inactive [CoII]-form with AE. In this study, the interaction of the O-demethylase components of A. dehalogenans, especially that of AE and CP, was analyzed.
Materials and methods
All chemicals and gases needed for the growth of microorganisms and protein purification were of highest available purity and were purchased from AppliChem GmbH (Darmstadt, Germany), Carl Roth GmbH (Karlsruhe, Germany), IBA GmbH (Göttingen, Germany), Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany), and Linde AG (Pullach, Germany). Enzymes for molecular biology, if not stated otherwise, were purchased from Thermo Fisher Scientific, Inc. (Waltham, MA).
Heterologous expression of the genes of the vanillate-O-demethylase in Escherichia coli
The production of recombinant proteins was performed in LB medium supplemented with required antibiotics. Gene expression was induced by addition of isopropyl β-D-thiogalactopyranoside (IPTG) or anhydrotetracycline, as described by Schilhabel et al. (2009). After induction, cells were harvested by centrifugation (10 min; 10 000 g) and were stored at −21 °C.
Construction of expression cassettes, cloning and heterologous expression of the genes encoding the CPs of the syringate- and guaiacol-O-demethylase
Besides the known O-demethylases of A. dehalogenans (vanillate- and veratrol-O-demethylase; van and ver), two more O-demethylase enzyme systems designated syringate- (syr) and guaiacol-O-demethylase (gua) were identified (unpublished data). The genes encoding the corresponding corrinoid proteins, CPsyr (Locus Tag A3KSDRAFT_03825; IMG system; Markowitz et al., 2012) and CPgua (Locus Tag A3KSDRAFT_01236), were cloned and heterologously expressed in E. coli. The expression cassettes as Strep-tag fusions (at the 3' end) and the restriction sites NdeI and BamHI for cloning into pET11a (Agilent Technologies, Waldbronn, Germany) were constructed from PCR products, as described previously (Schilhabel et al., 2009). The primers are listed in Supporting Information, Table S1. The plasmids obtained were checked for the insert by sequencing (GATC Biotech AG, Konstanz, Germany). Escherichia coli BL21 (DE3) was used as expression strain. The production of recombinant protein was performed in LB medium (10 g NaCl, 10 g tryptone, 5 g yeast extract per liter) supplemented with ampicillin (100 μg mL−1). Cultivation was performed at 28 °C. Gene expression was induced by the addition of 0.25 mM IPTG. After 4 h of induction, cells were harvested by centrifugation (10 min; 10 000 g) and were stored at −21 °C.
Purification of heterologously expressed O-demethylase components
Recombinant proteins were purified from cell extracts by affinity chromatography on Strep-Tactin according to the manufacturer's protocol. Soluble protein was obtained by disruption of E. coli cells in a French pressure cell followed by centrifugation (15 min; 16 000 g) at 10 °C. Because the expression strain E. coli BL21 (DE3) is not able to synthesize vitamin B12 derivatives de novo (Roth et al., 1996; Jeong et al., 2009), only the cofactor-free apoprotein of CP was produced upon induction. Protein purification and incorporation of hydroxocobalamin as corrinoid cofactor was performed according to Schilhabel et al. (2009).
Gel shift experiments
Recombinant AE and the four different recombinant corrinoid proteins CPvan, CPver, CPgua, and CPsyr (without and with corrinoid cofactor) were used for gel shift experiments. AE/CP mixtures with AE/CP ratios (pmol : pmol) of 30 : 10, 10 : 10, 10 : 30 were prepared in 50 mM Tris-HCl pH 7.5 containing 2 mM DTT. After two hours of incubation at 10 °C, the mixtures were separated by native PAGE (8.5%) using a Tris–glycine buffer system. After PAGE, proteins were stained with silver according to Schägger (2006).
Dot far-Western blot
The purified recombinant proteins (prey proteins) were spotted on a PVDF membrane (GE Healthcare, Munich, Germany) pre-incubated in PBST (140 mM NaCl, 10 mM KCl, 6.4 mM Na2HPO4, 2 mM KH2PO4, 0.05% (v/v) Tween 20). After drying, the membrane was blocked with 3% (w/v) skim milk in PBST for 1 h. Overnight, the rinsed membrane was incubated with the purified bait proteins (0.1 mg mL−1) at 18 °C. Unbound protein was washed off with PBST buffer (3x, 10 min). For the detection of proteins, the membrane was incubated with specific primary antibodies against the baits (3 h, room temperature). After another washing step, the secondary antibody (appropriate AP-conjugated antibody) was applied. After a final washing step, signals were developed using NBT/BCIP.
Plasmid construction for yeast two-hybrid assays
For the yeast two-hybrid screen, the genes or gene fragments encoding OdmA (corrinoid protein, CPvan), OdmB (methyltransferase I, MT Ivan), OdmC (AE), and OdmD (methyltransferase II, MT IIvan) of the vanillate-O-demethylase of A. dehalogenans were cloned into the plasmids pGADT7 and pGBKT7 using the NdeI and BamHI restriction sites with the exception of the odmC gene and its fragments. The gene fragment encoding AE aa101-176 was cloned into the vectors using the EcoRI and BamHI recognition sites. For cloning of the full-length gene of AE and its other fragments into pGBKT7, the restriction sites NcoI and BamHI were used. For cloning of odmC and the fragments encoding AE aa1-133 and AE aa134-598 into pGADT7, NdeI and EcoRI were used as recognition sites. The primers are listed in Table S2.
Yeast two-hybrid assays
The yeast two-hybrid analyses were performed using the Matchmaker GAL4 Two-Hybrid System (Takara Bio Europe/Clontech, Saint-Germain-en-Laye, France). Constructs of pGADT7 and pGBKT7 were transformed into the yeast strains Y187 and AH109, respectively. After transformation, the yeasts were mated according to the method described in Causier & Davies (2002). The diploid yeast cells were suspended in 100 μL TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA) and appropriate dilutions were spotted onto SD Leu- Trp- plates to check yeast growth and onto SD Leu− Trp− His− plates (containing 3 mM 3-aminotriazole) to test for protein–protein interactions. Plates were incubated for 3–7 days at the temperatures indicated in the text. An interaction was valued as positive if yeast growth was observed on the test plates and if β-galactosidase activity was detected. All interactions were tested at least in triplicate.
Corrinoid reduction assay
The corrinoid reduction activity catalyzed by the AE was determined in a final volume of 100 μL in anaerobic quartz cuvettes according to a method of Schilhabel et al. (2009).
Results and discussion
Corrinoid-dependent interaction of AE and CP demonstrated by gel shift experiments
Four different O-demethylase enzyme systems from Acetobacterium dehalogenans have been characterized: the vanillate- (Kaufmann et al., 1998), the veratrol- (Engelmann et al., 2001), the syringate-, and the guaiacol-O-demethylase (unpublished data). The corresponding corrinoid proteins CPvan, CPver, CPsyr, and CPgua were heterologously expressed in E. coli. Because no corrinoid cofactor is synthesized by the expression strain, the recombinant proteins were produced as cofactor-free apoproteins. The incorporation of the corrinoid cofactor was achieved by addition of hydroxocobalamin (Schilhabel et al., 2009). Preliminary studies revealed an interaction of recombinant AE (rAE) and cofactor-containing recombinant CPvan (rCPvan) (Schilhabel et al., 2009). In the study presented here, gel shift experiments were performed in the presence of cofactor-containing recombinant CPver, CPsyr, and CPgua. DTT (2 mM) was used to maintain the cob(II) state of the CPs. Mixtures with different molar ratios of rAE and rCP were separated by native PAGE (Fig. 1). At a molar ratio of 1 : 1, almost exclusively AE/CP protein complex bands at about 180 kDa were observed for each AE/CP mixture, indicating that AE is able to interact with different CPs of A. dehalogenans. With molar ratios other than 1 : 1, the band of the excess protein was visible. These finding suggests a 1 : 1 stoichiometry of the AE/CP complexes formed and indicate that AE may serve as activator for different CPs. The latter conclusion is in accordance with the finding that the gene encoding AE is not part of any of the O-demethylase gene clusters and is located elsewhere in the genome (Schilhabel et al., 2009).
Complex formation of AE with cofactor-free CP was negligible (Fig. 2, lanes −B12). The band at approximately 130 kDa corresponds to the dimeric form of AE. Only weak or no CP bands were visible, because cofactor-free CP did not exhibit a defined band on native gels (data not shown).
The redox state of the corrinoid cofactor may also affect the interaction of AE and CP. An interaction of AE and CP is expected only for the [CoII]-species, because [CoII]-CP is the substrate of AE. Due to the low redox potential of the cob(II)/cob(I)alamin couple of the four CP tested in this communication (ESHE < −550 mV), it was not possible to obtain and apply a stable cob(I) species. After AE-mediated corrinoid reduction to [CoI]-CP, the protein complex has to dissociate to prevent an electron back flow from super-reduced CP to the iron–sulfur cluster of AE. A study on the complex formation of an activator with CP in different redox states has been reported for RACo, the reductive activator of the corrinoid-iron/sulfur protein (CoFeSP) of Carboxydothermus hydrogenoformans (Hennig et al., 2012). The [CoI]-species of CoFeSP could be obtained by addition of sodium dithionite (which was not possible for the CPs tested here). The results revealed that an interaction occurred exclusively with CoFeSP in its [CoII]-state.
The AE activity was assayed by spectrophotometrically recording the corrinoid cofactor reduction of CPvan, CPver, CPsyr, and CPgua at two different CP concentrations. AE catalyzed the reduction of the corrinoid cofactor of all four CPs; all activities were in the same order of magnitude (about 2–7 nkat mg−1; Table 1). A relationship between the specific activity of AE and the sequence identities of the four CPs (78–92%) is not detectable. A CP of D. hafniense DCB-2 (Dhaf_4611) that exhibits about 70% sequence identity compared with the CPs of A. dehalogenans was also shown to be a substrate of AE (Studenik et al., 2012). The specific activity of AE with Dhaf_4611 as a substrate was about 50% compared to that with CPvan and CPsyr. In gel shift experiments, AE and Dhaf_4611 reconstituted with the corrinoid cofactor formed a complex; however, a significant part of AE and CP remained as single proteins (data not shown). The results indicated a low substrate specificity of AE of A. dehalogenans toward CPs of O-demethylases. The ability to reduce different CPs was also reported for RamA, the reductive activator of corrinoid-dependent methylamine methyltransferases of Methanosarcina barkeri (Ferguson et al., 2009). The corrinoid reduction activity with MtmC, the monomethylamine CP, was approximately 0.3 nkat mg−1 (Ferguson et al., 2009, 2012). This activity is similar to that of RACo (about 0.1 nkat mg−1 as calculated from the rate constant in the presence of 2 mM ATP; Hennig et al., 2012) and Dhaf_2573, a reductive activator found in D. hafniense DCB-2, with the CP Dhaf_4611 (about 0.5 nkat mg−1, Studenik et al., 2012).
Table 1. Corrinoid reduction activity of the AE with different CPs. The enzyme activity was determined using two different CP concentrations at 25 °C. CPx = corrinoid protein of the vanillate- (van), veratrol- (ver), syringate- (syr) or guaiacol-O-demethylase (gua)
Corrinoid reduction activity (nkat/mg)
Studies on the interaction of AE and CP using far-Western blot analyses and yeast two-hybrid screens
The interaction of AE and CPvan was further investigated using far-Western blot. rAE or rCPvan (without or with hydroxocobalamin as corrinoid cofactor, ±B12) were spotted onto a PVDF membrane. After incubation with the corresponding bait protein, bait-specific antibodies were used for detection. Membrane strips without bait protein treatment were used as negative controls or, using the antibodies directed against the prey, as a control for protein binding to the PVDF membrane. An interaction was evaluated as positive if the signals on the test strips were significantly stronger than those of the controls. An interaction of AE and CP was observed for all combinations tested, namely AE and CP as baits or preys in the absence or presence of the corrinoid cofactor (Fig. 3). Almost no difference in signal strength between the CP apoprotein (−B12; Fig. 3a) and reconstituted CP (+B12; Fig. 3b) was detectable. These results indicate that under the experimental conditions, the corrinoid cofactor does not have a major impact on the interaction. It is feasible that the cofactor-containing CP exhibits a stronger binding to AE than the apoprotein. This might lead to a separation of AE and the CP apoprotein in native PAGE, where the protein complex is exposed to an electric field in contrast to the far-Western blot analysis. Another possible explanation would be a conformational change of the cofactor-free CP in the gel shift experiment that prevents the interaction with AE.
In addition, a yeast two-hybrid assay was performed to test for the interaction of CP and AE. Because yeast does not produce corrinoids, a cofactor-free CP is synthesized. Using this assay, no interaction of AE and CP was observed independent on the GAL4 domains the fusion partners were bound to (Fig. S1; combinations 2F, 6B). It is feasible that upon fusion of CP and/or AE with the GAL4 domains, the interaction site(s) are not accessible for the interaction partner. Experiments to detect an interaction between AE and CP in the bacterial two-hybrid system did not yield reproducible results and were therefore abandoned.
The only interaction observed in the yeast two-hybrid assay was that between AE and AE as reflected by growth of the corresponding clones on the test plates (Figs S1 and S2; combination 2B). To specify the binding region of AE and AE, three protein fragments, AE 1-133, AE 134-598, and AE 101-177, were used as fusion partners for the GAL4 domains in different combinations. The numbers correspond to the amino acid positions. AE 1-133 harbors the iron/sulfur cluster binding domain. According to structure prediction (Ambrish et al., 2010), AE 101-177 corresponds to the RACo middle domain, which was shown to be involved in the dimerization of the latter protein (Hennig et al., 2012). Growth on the test plates was obtained for full-length AE in combination with AE 1-133 rather than with AE 101-177 (Figs S1 and S2; combinations 2C, 3B), indicating the involvement of the amino acid residues 1-133 in the oligomerization of AE. Yeast growth was also observed when AE 1-133 and AE 134-598 were combined pointing to a possible rearrangement of the full-length protein. The strength of interaction was estimated using the β-galactosidase assay on the test plates (Fig. S2). The results are summarized in Table 2. The strongest signals were observed for AE 1-133 in combination with AE 134-598 at 22 °C. The signals for the other interactions pairs of AE were of lower intensity (Table 2). The oligomerization of AE, which was shown in native PAGE as dimer formation and in the yeast two-hybrid screen as growth of the corresponding yeast clones on the test plates, is in accordance with the results obtained for RACo (Hennig et al., 2012). In contrast, RamA is a monomer in solution (Ferguson et al., 2009).
Table 2. Interaction of AE and its fragments observed by yeast two-hybrid screen. The strength of interaction was estimated from the test plates shown in Fig. S2
Interaction of protein fragments could only be studied in yeast, because expression of these fragments in E. coli was not possible. They seemed to be unstable and/or were produced only in small amounts close to the detection limit. As for the full-length proteins, no interaction of AE or AE fragments with CP or any of the CP fragments tested (CP 1-78, CP 79-209; the latter carries the corrinoid-binding motif DXH100XXG) was observed (Fig. S1).
Interaction of the other O-demethylase components
All experiments described in this section were performed with components of the vanillate-O-demethylase. Gel shift experiments did not reveal any further interaction pairs besides AE and CP (data not shown). In yeast two-hybrid screens, the only interaction besides AE and AE was observed for the CP-MT I pair as indicated by cell growth and β-galactosidase activity on test plates (Figs S1 and S2; combination 6I). The attempt to identify the interaction sites using CP and MT I fragments did not yield any result. The fragment MT I 88-326 (in pGBKT7), which represents the catalytic domain of this enzyme (Kreher et al., 2010; Studenik et al., 2011), led to an auto-activation of the system detected by yeast growth on the test plates in combination with the nonmodified pGADT7 plasmid and with all pGADT7 constructs tested (Fig. S1, combinations 1K-12K).
Far-Western blot analyses revealed a moderate interaction of CP with both methyltransferases only when MT I or MT II were spotted onto the membrane (as prey) and CP was in solution (as bait). The corrinoid cofactor had no influence on the signal strength (data not shown). It is feasible that upon binding of CP to the PVDF membrane, the interaction site is not accessible to the methyltransferases. The interaction of CP with MT I is in accordance with the role of these proteins in O-demethylation. However, compared with the methyltransferase systems of methanogenic archaea for which a copurification or complex formation of CP and MT I is commonly reported (Burke & Krzycki, 1997; Ferguson & Krzycki, 1997; Sauer & Thauer, 1997; Tallant & Krzycki, 1997; Ferguson et al., 2000), the interaction of the corresponding O-demethylase components appears to be much weaker under the conditions applied.
The rather weak interactions of the O-demethylase components described in this study indicate that the enzyme system is not present as a multienzyme complex in the cells and that, other than described for some corresponding methanogenic methyltransferases, CP is a separate protein rather than a subunit of MT I. Because AE appears to be a ‘universal’ activator for different CPs and is present in substoichiometric amounts compared with CP (Kaufmann et al., 1997), it appeared unlikely to be present as part of a multienzyme complex. Nonetheless, the activation of CP obviously requires an interaction with AE, which was demonstrated by the gel shift experiments and far-Western blot analysis.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (JSMC project GSC 214/1). The authors like to thank Yvonne Greiser for skillful technical assistance.