Heavy metal ions inhibit molybdoenzyme activity by binding to the dithiolene moiety of molybdopterin in Escherichia coli

Authors


S. Leimkühler, 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

Abstract

Molybdenum insertion into the dithiolene group on the 6-alkyl side-chain of molybdopterin is a highly specific process that is catalysed by the MoeA and MogA proteins in Escherichia coli. Ligation of molybdate to molybdopterin generates the molybdenum cofactor, which can be inserted directly into molybdoenzymes binding the molybdopterin form of the molybdenum cofactor, or is further modified in bacteria to form the dinucleotide form of the molybdenum cofactor. The ability of various metals to bind tightly to sulfur-rich sites raised the question of whether other metal ions could be inserted in place of molybdenum at the dithiolene moiety of molybdopterin in molybdoenzymes. We used the heterologous expression systems of human sulfite oxidase and Rhodobacter sphaeroides dimethylsulfoxide reductase in E. coli to study the incorporation of different metal ions into the molybdopterin site of these enzymes. From the added metal-containing compounds Na2MoO4, Na2WO4, NaVO3, Cu(NO3)2, CdSO4 and NaAsO2 during the growth of E. coli, only molybdate and tungstate were specifically inserted into sulfite oxidase and dimethylsulfoxide reductase. Other metals, such as copper, cadmium and arsenite, were nonspecifically inserted into sulfite oxidase, but not into dimethylsulfoxide reductase. We showed that metal insertion into molybdopterin occurs beyond the step of molybdopterin synthase and is independent of MoeA and MogA proteins. Our study shows that the activity of molybdoenzymes, such as sulfite oxidase, is inhibited by high concentrations of heavy metals in the cell, which will help to further the understanding of metal toxicity in E. coli.

Abbreviations
FeVco

iron–vanadium cofactor

hSO

human sulfite oxidase

hSO-MD

human sulfite oxidase Moco domain

ICP-OES

inductively coupled plasma-optical emission spectrometry

MGD

molybdopterin guanine dinucleotide cofactor

Moco

molybdenum cofactor

MPT

molybdopterin

Wco

tungsten cofactor

Molybdenum is the only second-row transition metal that is required by most living organisms, and the few species that do not require molybdenum use tungsten, which lies immediately below molybdenum in the Periodic Table [1,2]. Molybdenum- and tungsten-containing enzymes often catalyse the same types of reaction; while molybdenum-containing enzymes are found in all aerobic organisms, including humans, tungsten-containing enzymes are found only in obligate, typically thermophilic, anaerobic bacteria and archaea [3,4]. The molybdenum cofactor (Moco) is the essential component of a group of redox enzymes which catalyse a variety of transformations at carbon, sulfur and nitrogen atoms. More than 40 molybdenum and tungsten enzymes have been identified in bacteria, archaea, plants and animals to date [5,6]. Some of the better known Moco-containing enzymes include sulfite oxidase, xanthine dehydrogenase and aldehyde oxidase in humans [6], assimilatory nitrate reductase in plants [7] and dissimilatory nitrate reductase, dimethylsulfoxide reductase and formate dehydrogenase in bacteria [8–10].

Moco biosynthesis has been studied extensively in Escherichia coli using a combination of biochemical, genetic and structural approaches [11,12], and additional insights have been provided by studies in eukaryotes [13]. The biosynthesis of Moco can be divided into three general steps in all organisms: (a) formation of precursor Z from GTP; (b) formation of molybdopterin (MPT) by insertion of two sulfur atoms into precursor Z; (c) insertion of molybdenum into the dithiolene group of MPT, thus forming Moco. In bacteria, such as E. coli, an additional modification of Moco occurs with the attachment of GMP to the phosphate group of Moco, forming the MPT guanine dinucleotide cofactor (MGD) [14,15]. In total, more than 10 genes are involved in the biosynthesis of Moco in E. coli, and highly conserved proteins have been identified in other organisms.

Molybdenum enters the cell as the soluble oxyanion molybdate, for which high-affinity molybdate transporters have been described in bacteria [11,16] and in higher eukaryotes [17,18]. In the step of molybdenum insertion into MPT, the gene products of moeA and mogA are involved in E. coli. It has been observed that MoeA mediates molybdenum ligation to newly synthesized MPT at low concentrations of molybdate, and MogA helps to facilitate this step in vivo in an ATP-dependent manner via an MPT-adenylate intermediate [19,20]. The crystal structure of the Arabidopsis thaliana Cnx1 G protein, a homologue of E. coli MogA, shows that copper is bound to the MPT dithiolene sulfurs of the MPT–AMP complex [21]. It has been proposed that copper binding to MPT–AMP on Cnx1 is physiologically relevant and that, in vivo, copper may serve to protect the dithiolene moiety prior to the binding of molybdenum. However, as several metal ions are known to bind tightly to sulfur-rich sites, a more recent report by Morrison et al. [22] investigated the effect of copper-limiting reaction conditions on molybdoenzymes in E. coli and Rhodobacter sphaeroides. Their results demonstrated that the activities of dimethylsulfoxide reductase and nitrate reductase were not repressed under copper starvation, showing that copper is not strictly required for the biosynthesis of Moco in bacteria [22].

In addition to copper, various metals are known to bind tightly to sulfur-rich sites, leading to the question of whether other metal ions can bind to the dithiolene moiety of Moco and be inserted into molybdoenzymes. Some of the most common environmental toxins are cadmium and arsenic. Arsenic is ubiquitous in the environment and is most commonly found in an insoluble form associated with rocks and minerals [23,24]. In soluble form, arsenic occurs as trivalent arsenite [As(III)] and pentavalent arsenate [As(V)]. Arsenate, a phosphate analogue, can enter cells via the phosphate transport system, and is toxic because it can interfere with normal phosphorylation processes by replacing phosphate [25]. The competitive substitution of arsenate for phosphate can lead to rapid hydrolysis of the high-energy bonds in compounds such as ATP. Arsenite has recently been demonstrated to enter cells at neutral pH by aqua-glyceroporins (glycerol transport proteins) in bacteria, yeast and mammals [26], and its toxicity lies in its ability to bind sulfhydryl groups of cysteine residues in proteins, thereby inactivating them. Arsenite is considered to be more toxic than arsenate and can be oxidized to arsenate chemically or microbially [27,28].

In contrast, cadmium is soluble as its bivalent cation Cd2+, and Cd2+ ions are readily taken up by bacterial and eukaryotic cells, presumably by the Mn2+ uptake system [29]. Cadmium toxicity may be caused by binding to zinc binding proteins, e.g. proteins that contain zinc finger protein structures [30]. Zinc and cadmium are in the same group in the Periodic Table, contain the same common oxidation state (+2) and, when ionized, are almost the same size. As a result of these similarities, cadmium can replace zinc in many biological systems, in particular systems that contain sulfur ligands [31]. Cadmium can be bound up to 10 times more strongly than zinc to certain biological systems, and is thus difficult to remove [31]. In addition, cadmium can replace magnesium and calcium in certain proteins [30].

Vanadium is chemically similar to molybdenum, and can replace molybdenum in its role in nitrogenase, forming the iron–vanadium cofactor (FeVco) [32]. Like molybdenum, vanadium is available in anionic and cationic forms, the most common being, under physiological conditions, vanadate (H2VO4) and vanadyl (VO2+) [33]. Vanadate can act as a competitor to phosphate (HPO42−), or as a transition metal ion that competes with other metal ions in coordination with biogenic compounds. Because of the low molecular weight of VO3, like phosphate, the VO3 ion is able to permeate plasma membranes and the intestinal wall in humans with relative ease [34–36]. Vanadate ions also mimic most of the rapid actions of insulin in the cell [37].

In this report, we examined the in vivo and in vitro incorporation of metal ions of molybdenum, tungsten, vanadium, copper, cadmium and arsenic into the MPT cofactor. For in vivo studies, we heterologously produced human sulfite oxidase (hSO) and R. sphaeroides dimethylsulfoxide reductase in the presence of metal ions in E. coli and studied the incorporation of metal-bound MPT into these enzymes. Our results demonstrate that only molybdate and tungstate are specifically inserted into MPT and MGD in E. coli. Bivalent copper and cadmium ions and trivalent arsenite can be inserted nonspecifically into MPT, which is inserted into hSO, thus inhibiting enzyme activity. However, copper, cadmium and arsenite are not inserted into bis-MGD containing dimethylsulfoxide reductase in E. coli. Our results suggest that enzymes containing the MPT form of Moco can easily be inhibited by copper, cadmium, arsenic and other metal ions binding to sulfur-rich sites, whereas bis-MGD-containing enzymes are rather protected from nonspecific metal insertion.

Results

Investigation of metal ion insertion into hSO during heterologous expression in E. coli

The expression of hSO in an E. coli modC strain [38] results in an MPT-containing form of hSO, which is free of metal ions at the MPT site (data not shown). It has been shown previously that the addition of 100 μm molybdate is sufficient to complement the modC phenotype [39,40]. Thus, the production of hSO in the E. coli modC strain is an ideal system to study the nonspecific insertion of metal ions, other than molybdate, into hSO. In addition, both the E. coli moeA and mogA strains [38,41] were used for comparative studies, as MoeA and MogA have been shown to be involved in the specific insertion of molybdate into MPT [19,20]. Comparative studies of the metal contents of hSO after production in E. coli modC, mogA and moeA strains should make it possible to distinguish whether the metal ions are inserted specifically or nonspecifically into hSO. For metal insertion into hSO, the protein was produced in the E. coli modC, moeA and mogA strains in the presence of 100 μm of Na2MoO4, Na2WO4, NaVO3, Cu(NO3)2, CdSO4 or NaAsO2. For metal analysis, hSO was purified after expression. In addition, the uptake of metal ions was analysed in E. coli cell extracts. As shown in Fig. 1A, the addition of 100 μm molybdate during expression in the modC strain resulted in a 65% molybdenum-saturated hSO. Surprisingly, with the exception of vanadate, all other metal ions were also readily detectable in hSO. Although the saturation levels of hSO with copper and arsenite were only 27% and 23%, respectively, the saturation level for tungstate was 44% and, for cadmium, 36% (Fig. 1A). In contrast, the levels of MPT saturation in hSO were in the range 55–74% when cells were grown in the presence of Na2MoO4, Cu(NO3)2, Na2WO4 or NaAsO2 (Fig. 1A). Cells grown in the presence of CdSO4 or NaVO3 showed lower levels of MPT saturation of 37% and 39%, respectively. However, this underlines the fact that metal-free MPT is inserted into hSO independent of the availability of metals.

Figure 1.

 Metal and MPT saturation of purified hSO. Purified hSO was analysed after expression from plasmid pTG818 in E. coli strain RK5202 (modC) (A), E. coli strain AH69 (moeA) (B) and E. coli strain RK5206 (mogA) (C). The following metals were added at a concentration of 100 μm to the growth medium: I, Na2MoO4; II, Cu(NO3)2; III, Na2WO4; IV, NaAsO2; V, CdSO4; VI, NaVO3. Dark grey bars, metal contents (μm metal·μm−1 hSO) were determined by ICP-OES (see Experimental procedures) using multielement standards. Light grey bars, the MPT content of hSO was quantified after its conversion to Form A. 100% metal or MPT saturation is related to a fully active hSO in a 1 : 1 ratio. White bars, hSO activity (units·mg−1) defined as an absorbance change of 1.0 AU·min−1·mg−1 protein monitoring the reduction of cytochrome c at 550 nm. ND, none detectable.

In comparison, the metal insertion pattern was different in the E. coli moeA strain (Fig. 1B). When using the moeA strain, neither molybdate nor tungstate was detected in hSO, showing that both metals are specifically inserted into MPT by the MoeA protein. In contrast, saturation levels of copper, arsenite and cadmium were found to be in the region of 21%, 39% and 60%, respectively. Again, no vanadate was detected in hSO. Analysis of the uptake of the metal ions showed that all metals were present in the E. coli extract (data not shown). Thus, E. coli is able to take up vanadate, but vanadate is not inserted into MPT; this was a rather surprising observation, as vanadium is known to functionally replace the molybdenum atom in nitrogenase. The MPT saturation of hSO after expression in the moeA strain in the presence of different metal ions was comparable, and varied between 71% and 88%.

Furthermore, we also examined metal insertion during production in the E. coli mogA strain (Fig. 1C). It has been described previously by Miller and Amy [39] that a molybdate concentration of 1 mm is sufficient to reverse the mogA phenotype, restoring nitrate reductase activity. Thus, at a concentration of 100 μm, neither molybdate nor tungstate reconstituted the cofactor of hSO. In contrast, saturation levels of copper, arsenite and cadmium were found at 41%, 40% and 65%, respectively. As observed after expression in the modC and moeA strains, no vanadate was detected in hSO. The MPT saturation of hSO after production in the mogA strain in the presence of different metal ions was comparable, and varied between 62% and 73%.

Competitive insertion of metal ions into hSO during expression in E. coli modC cells

To analyse the specificity of the insertion of different metal ions into MPT, competition experiments were performed. E. coli modC cells were grown in the presence of 100 μm Na2MoO4 and 100 μm Na2WO4, NaAsO2, Cu(NO3)2 or CdSO4. The same competition experiment was performed with Cu(NO3)2 and all other metal ions. As shown in Table 1, the MPT saturation of hSO was comparable, and varied between 69% and 92%. The results in Table 1 clearly show that the presence of equal amounts of molybdate during growth was sufficient for the specific insertion of molybdenum into MPT. When Na2MoO4 and Na2WO4, Cu(NO3)2, CdSO4 or NaAsO2 were present during growth, the molybdate saturation in hSO was comparable, and varied between 84% and 70%, whereas only about 5% of the competing metal was inserted (Table 1). This result clearly shows the high specificity of molybdate for insertion into MPT. A different pattern was obtained for the insertion of tungstate, arsenite or cadmium in the presence of equal amounts of copper during growth. When Na2WO4 and Cu(NO3)2 were added to the medium, hSO was saturated with 48% tungstate and 11% copper; equal amounts of about 30% copper and arsenite were inserted when NaAsO2 and Cu(NO3)2 were added, and more cadmium (67.8%) than copper (22.2%) was inserted in the presence of CdSO4 and Cu(NO3)2 in the cell (Table 1). Thus, the affinity of MPT for cadmium is higher than that for copper, but, in general, there is no preference for the nonspecific insertion of any other metal ion than molybdate.

Table 1.   Activity, metal content and MPT saturation of purified hSO heterologously produced in E. coli modC cells. Metals were added during growth at a concentration of 100 μm each. hSO activity (units·mg−1) is defined as an absorbance change of 1.0 AU·min−1·mg−1 protein monitoring the reduction of cytochrome c at 550 nm. The MPT content of hSO was quantified after its conversion to Form A in comparison with a fully active hSO. 100% MPT saturation is related to a fully active hSO in a 1 : 1 ratio. Metal contents (μm metal·μm−1 hSO) were determined by ICP-OES (see Experimental procedures) using multielement standards. 100% metal saturation is related to a fully MPT-saturated enzyme in a 1 : 1 ratio. ND, none detectable.
hSO activity and metal/MPT saturationMetals added to the growth medium
Mo/CuMo/WMo/AsMo/CdCu/WCu/AsCu/Cd
hSO activity (units·mg−1)621 ± 29822 ± 54719 ± 59725 ± 25NDNDND
MPT (%)83.7 ± 2.174.5 ± 2.482.3 ± 2.289.5 ± 1.781.7 ± 4.468.6 ± 7.291.6 ± 7.7
Mo (%)78.7 ± 3.070.2 ± 6.677.2 ± 6.184.3 ± 1.2NDNDND
Cu (%)4.5 ± 0.4NDNDND11.3 ± 0.732.9 ± 0.722.2 ± 1.7
W (%)ND5.4 ± 1.4NDND48.2 ± 0.7NDND
As (%)NDND5.2 ± 0.7NDND31.2 ± 1.7ND
Cd (%)NDNDND4.1 ± 1.5NDND67.8 ± 3.3

Analysis of metal insertion into R. sphaeroides dimethylsulfoxide reductase

Most molybdoenzymes in E. coli contain the bis-MGD cofactor. For the biosynthesis of MGD, GMP is attached to the phosphate group of MPT by the MobA protein [42]. It has been shown previously that MGD formation is catalysed by MobA only after the coordination of a molybdenum atom to the dithiolene moiety of MPT [19]. To determine whether molybdate can be replaced by other metals for the biosynthesis of MGD, R. sphaeroides dimethylsulfoxide reductase was heterologously produced in the presence of different metal ions at a concentration of 100 μm in the E. coli modC strain, and analysed for its metal content. Metal analysis of purified dimethylsulfoxide reductase revealed that only tungstate and molybdate were inserted into the enzyme (Fig. 2). Consistent with a previous report [43], dimethylsulfoxide reductase was active in the tungsten-bound form. Copper, cadmium, arsenite and vanadate were not present in the purified dimethylsulfoxide reductase (Fig. 2, data not shown). In addition, the dimethylsulfoxide reductase proteins were devoid of bis-MGD, showing that metal ions such as copper or cadmium cannot replace molybdate in the biosynthesis of MGD, whereas tungstate is able to substitute for this role. Thus, GMP is only added to molybdenum- or tungsten-containing MPT.

Figure 2.

 Activity, metal and MPT saturation of R. sphaeroides dimethylsulfoxide reductase (DMSOR). Dimethylsulfoxide reductase was purified after expression from plasmid pJH820 in E. coli RK5202 (modC) cells. The following metals were added during growth at a concentration of 100 μm: I, Na2MoO4; II, Cu(NO3)2; III, Na2WO4; IV, CdSO4. Dark grey bars, metal contents (μm metal·μm−1 hSO) were determined by ICP-OES (see Experimental procedures) using multielement standards. Light grey bars, the MPT content of dimethylsulfoxide reductase was quantified after the conversion of bis-MGD to Form A. 100% metal or MPT saturation was related to a fully MPT-saturated dimethylsulfoxide reductase in a 1 : 1 or 1 : 2 ratio, respectively. White bars, dimethylsulfoxide reductase activity (units·mg−1) defined as the reduction of 1 μmol of dimethylsulfoxide·min−1·mg−1 protein. ND, none detectable.

Metal insertion into purified hSO in vitro

It has been shown previously that Moco-free hSO can be reconstituted with nascent MPT and molybdate in vitro [44]. At the molybdate concentrations used in the assay (> 1 mm), MoeA and MogA were not required for the generation of the active form of Moco, showing that the ATP-dependent activation of MPT and molybdenum is not required for the in vitro ligation of molybdate. It was of interest to examine the affinity of free MPT for other metal ions, and to determine whether MPT chelated with other nonspecific metals can also be inserted into hSO in vitro. For these experiments, the molybdenum-free human sulfite oxidase Moco domain (hSO-MD) was used as MPT source [45], and extracted MPT was incubated with 100 μm of Na2MoO4, Na2WO4, NaVO3, Cu(NO3)2, CdSO4 or NaAsO2, before purified apo-hSO was added to the mixture. As shown in Fig. 3, the amount of MPT inserted into hSO was about the same in all incubation mixtures, independent of the added metal, with an MPT saturation in the range 43–47%. Analysis of the metal content of hSO revealed that, at a metal concentration of 100 μm, only the bivalent copper and cadmium ions were inserted into hSO (Fig. 3A). As shown previously, molybdate is not inserted into MPT at a concentration of 100 μm [44]. As tungstate, arsenite and vanadate were not inserted into MPT at a concentration of 100 μm, higher concentrations of added metals were analysed. However, the addition of these metals at concentrations of up to 1 mm to the reaction mixture did not result in their insertion into MPT (data not shown). In addition, we tested the insertion of metal ions into MPT-containing hSO. None of the metals was inserted into hSO (Fig. 3B), which is consistent with the previous data for the insertion of molybdate into molybdenum-free MPT-hSO [44], and makes nonspecific binding of the metals to the protein surface unlikely. This shows that, after MPT insertion, hSO adopts a conformation that is not competent for the insertion of metals. Thus, any nonspecific metal insertion into MPT has to occur before the insertion of MPT into hSO.

Figure 3.

In vitro reconstitution of hSO with MPT and metal ions. (A) 20 μm MPT extracted from hSO-MD was incubated with 100 μm of Na2MoO4 (I), Cu(NO3)2 (II), Na2WO4 (III), NaAsO2 (IV), CdSO4 (V) or NaVO3 (VI) for 10 min at 4 °C. Subsequently, 10 μm of purified apo-hSO was added to the mixture and incubated for 20 min at 4 °C, before unbound MPT or metal ions were removed by gel filtration. Dark grey bars, metal contents (μm metal·μm−1 hSO) were determined by ICP-OES (see Experimental procedures) using multielement standards. Light grey bars, the MPT content of hSO was quantified after its conversion to Form A. 100% metal or MPT saturation is related to a fully active hSO in a 1 : 1 ratio. (B) 40 μm MPT-containing hSO was incubated with 100 μm of Na2MoO4 (I), Cu(NO3)2 (II), Na2WO4 (III), NaAsO2 (IV), CdSO4 (V) or NaVO3 (VI) for 30 min at room temperature, before unbound MPT or metal ions were removed by gel filtration. Light grey bar, MPT content of hSO. Metal content (μm metal·μm−1 hSO) of MPT-hSO was below detection limit. ND, none detectable.

Release of MPT bound to MPT synthase by metal ions

It has been suggested that, during the biosynthesis of Moco, the cofactor and its intermediates remain protein bound until insertion into the specific target protein, as oxygen seems to be a major factor of free Moco inactivation [46]. Our results showed that metal ions, such as copper, cadmium or arsenite, might be inserted nonspecifically into MPT, without the involvement of the MogA or MoeA proteins. As MPT is produced by MPT synthase, it was of interest to analyse the effect of metal ions on the release of MPT bound to MPT synthase. For our studies, we chose to compare the MPT release after metal addition to MPT-containing R. capsulatus MPT synthase in comparison with E. coli MPT synthase. The comparison of the two MPT synthases from different sources was of particular interest, as the phenotype of the R. capsulatus moeA strain has been shown to be repairable with 1 mm molybdate [47], which is not the case for the E. coli moeA strain [39]. Thus, we first analysed the effect of high molybdate concentrations on MPT-saturated MPT synthase from the two different sources. The results in Fig. 4 show that, under the same assay conditions, 85.5% of MPT remained bound to E. coli MPT synthase when 1 mm molybdate was added, whereas only 12.3% of MPT remained bound to R. capsulatus MPT synthase. This result shows that E. coli MPT synthase binds MPT more tightly; however, the rate of conversion of precursor Z to MPT was the same in both MPT synthases (data not shown).

Figure 4.

 Comparison of the effect of 1 mm molybdate on MPT-saturated MPT synthase from E. coli and R. capsulatus. 33.2 μm of MPT-saturated MPT synthase from either E. coli (I, II) or R. capsulatus (III, IV) was incubated in the presence (II, IV) or absence (I, III) of 1 mm molybdate. The MPT content of MPT synthase was quantified after its conversion to Form A.

To analyse the effect of other metal ions on MPT release by MPT synthase, we chose the R. capsulatus MPT synthase, which binds MPT less tightly; thus, an effect of metal ions on MPT release should be detected easily. R. capsulatus MPT synthase was saturated with MPT before the addition of Na2WO4, NaVO3, Cu(NO3)2, CdSO4 or NaAsO2. The incubation mixtures were subjected to gel filtration and the MPT synthase fraction was analysed for its MPT content. The results in Fig. 5A show that, under the assay conditions, MPT remained bound to MPT synthase when metals other than molybdate were added. As metal analysis of MPT synthase revealed that none of the metals was inserted into MPT (Fig. 5A), the dithiolene group of MPT must be protected in MPT synthase, and, apparently, only molybdate is able to trigger the release of MPT. This result clearly shows that the nonspecific insertion of metal ions into MPT occurs in a step after the release of MPT from MPT synthase. The reverse experiment showed that MPT was only bound to MPT synthase in its metal-free form, and not in the presence of metal ions (Fig. 5B).

Figure 5.

 Analysis of the effect of added metals on MPT binding or MPT release in purified R. capsulatus MPT synthase. (A) 33.2 μm MPT-saturated MPT synthase was incubated with 1 mm of the indicated metal ions, and unbound material was removed by gel filtration. The MPT content of MPT synthase was quantified after its conversion to Form A. (B) 1 mm of metal ions was incubated with 1 mm of MPT prior to the addition of 21.3 μm MPT synthase. Unbound MPT and metal ions were removed from MPT synthase by gel filtration. The MPT content of MPT synthase was quantified after its conversion to Form A. (A, B) Metal contents were determined by ICP-OES (see Experimental procedures) using multielement standards. All metal contents were below the detection limit. 1 mm of Cu(NO3)2, Na2WO4, NaAsO2 or CdSO4 was added. ND, none detectable.

Discussion

Homeostasis of metal ions is a highly regulated complex process in the cell [48]. As a defence against metal toxicity, organisms have developed systems for metal detoxification, including specific export systems, as found in E. coli [49–51]. Our results show that, at high concentrations of metal ions and in the absence of molybdate ions, copper, cadmium and arsenite are inserted into Moco found in hSO. In addition to the metals investigated in this report, other metals ions known to bind tightly to sulfur-rich sites, such as zinc and cobalt, are inserted into MPT (data not shown). However, only molybdate and tungstate are specifically inserted into hSO, requiring the catalytic activity of the MoeA protein. Although the synthesis of the dioxo Moco found in hSO seems to be more susceptible to nonspecific metal insertion, the biosynthesis of the bis-MGD cofactor present in dimethylsulfoxide reductase provides an additional control step. In dimethylsulfoxide reductase, only molybdate and tungstate are found to be coordinated to bis-MGD, and the insertion of either metal results in an active form of dimethylsulfoxide reductase. This result shows that MobA is only able to add GMP to the molybdate- or tungstate-substituted form of MPT, and not to other metal-substituted forms of MPT. The chemical and physical similarities of tungstate and molybdate are a result of their equal atomic and ionic radii and similar electronegativity and coordination characteristics, contributing to the discrimination of these metals in biological systems [2,52]. In general, tungstate can replace molybdenum in molybdoenzymes in selected organisms, forming the tungsten cofactor (Wco) [3]. Although molybdenum-containing enzymes are found in all aerobic organisms, tungsten-containing enzymes are generally found only in obligate, typically thermophilic, anaerobes. Tungsten may have been the first of these elements to be acquired by living organisms. However, when the atmosphere became more aerobic, the oxygen sensitivity of tungsten compounds made them less available, and the water solubility of high-valent molybdenum oxides may have become more advantageous [3]. Because tungsten and molybdenum have similar chemistry, it is possible that, initially, as the transition to an oxygen-rich environment occurred, the latter substituted for the former in enzyme active sites. The results on dimethylsulfoxide reductase show that tungsten is still able to replace molybdenum at the bis-MGD cofactor, resulting in an active enzyme. However, in R. capsulatus, the expression of dimethylsulfoxide reductase has been shown to be regulated by the availability of molybdenum [53]; therefore, it is unlikely that, during normal growth conditions, tungsten replaces molybdenum to produce an active dimethylsulfoxide reductase in this organism.

Our studies also show that bivalent copper and cadmium ions and trivalent arsenite ions can be inserted nonspecifically into MPT without the catalytic activity of the MoeA or MogA protein. Copper and cadmium also show a higher affinity than molybdate for the dithiolene group of MPT, as revealed by the in vitro insertion of metal-substituted MPT into hSO, as these metals were already inserted into MPT at concentrations of 100 μm. Our results show that the nonspecific insertion of metal ions into MPT occurs at a step beyond the MPT synthase reaction. In MPT synthase, the dithiolene group seems to be protected in a manner that makes it inaccessible for the insertion of nonspecific metal ions. Only in the presence of high molybdate concentration is MPT released and molybdate inserted. This result may imply that, under conditions of high molybdate availability in the cell, the activation of MPT by adenylation may not be required, as molybdate is directly inserted into released MPT after its completion by MPT synthase, as shown by the molybdate repairable phenotype of the E. coli mogA strain [39]. The comparison of MPT synthase from E. coli and R. capsulatus shows that MPT is more easily released from R. capsulatus MPT synthase when 1 mm molybdate is added, implying that MPT is less tightly bound in R. capsulatus MPT synthase in comparison with the protein from E. coli. This result also explains the phenotype of the R. capsulatus moeA strain, which can be complemented by the addition of 1 mm molybdate [47], which is not the case for the E. coli moeA strain [39]. High molybdate concentrations result in the release of MPT from MPT synthase, and thus MogA or MoeA are not required under these conditions.

Our investigations analysed the insertion of metal ions into MPT beyond the reaction by the MogA and MoeA proteins. The insertion of copper, cadmium and arsenite is independent of the MogA and MoeA proteins. Even if the copper-containing MPT–AMP intermediate is formed in E. coli, copper can be replaced in this intermediate by other metal ions without the catalytic activity of the MoeA protein. Our results suggest that copper is not required as an intermediate in Moco biosynthesis in E. coli. In addition to the known toxic effects, the toxicity of copper, cadmium and arsenic in the environment may also be caused by an inhibition of molybdoenzyme activity in E. coli. We present a model for nonspecific metal insertion during Moco biosynthesis in E. coli (Fig. 6). Under physiological molybdate concentrations (1–10 μm), the MogA and MoeA proteins are required in E. coli to form an MPT–AMP intermediate, facilitating molybdate insertion and Moco formation in the cell (full lines). Using the same pathway, tungstate can be specifically inserted into MPT to form Wco. Under high molybdate concentrations (> 1 mm), MPT–AMP formed by MogA is not required and molybdate can be directly inserted into MPT by the aid of the MoeA protein. All other metal ions, when present at high concentrations, are inserted nonspecifically into MPT, not requiring the MogA and MoeA proteins (broken lines). However, this nonspecific reaction can be outcompeted by the presence of molybdate, showing that this is the specific pathway in the cell. Nonspecifically formed Cu–MPT, Cd–MPT or As–MPT can be inserted into MPT-binding molybdoenzymes, such as sulfite oxidase, but not into bis-MGD-containing enzymes, such as dimethylsulfoxide reductase. Here, the E. coli provides an additional ‘quality control step’ by the MobA protein, which only forms the bis-MGD cofactor when molybdenum or tungsten is inserted into MPT. However, the nature of this quality control step and the details of bis-MGD formation are not yet known. In our experiments, we were unable to show a copper-containing MPT–AMP intermediate. Thus, for Moco biosynthesis in E. coli, copper is not required and is rather an inhibitor of molybdoenzymes, inhibiting enzyme activity when inserted nonspecifically.

Figure 6.

 Model for the specific and nonspecific insertion of metal ions during Moco biosynthesis in E. coli. Specific reactions are marked by full lines, nonspecific reactions by broken lines. Details of the reactions are given in the text.

Experimental procedures

Bacterial strains, plasmids, media and growth conditions

E. coli BL21(DE3) cells were used for the heterologous expression of the R. capsulatus moaE and moaD1 genes. For metal incorporation experiments, E. coli RK5202 (chlD202::Mu cts [modC]), RK5206 (chlG206::Mu cts [mogA]) [38] and AH69 (moeA113 ΔzbiK-Km [moeA]) cells [41] were used for the production of hSO from plasmid pTG718 [54]. R. sphaeroides dimethylsulfoxide reductase was expressed in E. coli RK5202 cells from plasmid pJH820 [42] under the same conditions as hSO. Moco-free apo-hSO was obtained after expression in E. coli RK5200 (chlA200::Mu cts [moaA]) [38]. MPT-containing hSO-MD and hSO were expressed from pTG818 [54] or pTG718 in E. coli RK5202 cells. Dimethylsulfoxide reductase, hSO and hSO-MD were purified as described by Temple et al. [54]. E. coli MoaE was expressed from plasmid pGG110 in E. coli BL21(DE3) cells (Novagen, La Jolla, CA, USA), cotransformed with plasmid pREP4 (Qiagen, Hilden, Germany) and purified as described previously [55]. E. coli MoaD was expressed from plasmid pMW15aD in E. coli BL21(DE3) cells and purified as described previously [56]. Cell strains containing expression vectors were grown aerobically in Luria–Bertani medium at 30 °C in the presence of either 150 μg·mL−1 ampicillin and/or 25 μg·mL−1 kanamycin. Metals were added as indicated at a concentration of 100 μm.

Cloning, expression and purification of R. capsulatus MoaE and MoaD1

DNA fragments containing the coding regions for R. capsulatus moaE and moaD1 were amplified by PCR, and flanking restriction sites were introduced. The moaE gene was cloned into the NdeI-BamHI sites of expression vector pET16b (Novagen) and moaD1 into the NdeI-XhoI sites of pET28a (Novagen), resulting in plasmids pSL241 and pMN67, respectively. For the production of His6-tagged MoaE and MoaD1, E. coli BL21(DE3) cells were transformed with plasmid pSL241 or pMN67. One litre of Luria–Bertani medium was inoculated with 10 mL of an overnight culture and incubated at 30 °C until an absorbance (A) at 600 nm of 0.3–0.5. The expression was induced with 100 μm isopropyl thio-β-d-galactoside, and cells were harvested after an additional growth of 5 h. The cell pellet was resuspended in phosphate buffer (50 mm NaH2PO4, 300 mm NaCl, pH 8.0). Cells were lysed by several passages through a French pressure cell, and the cleared lysate was applied to 0.5 mL nickel-nitrilotriacetate resin (Qiagen) per litre of culture. The column was washed with 20 column volumes of phosphate buffers, one containing 10 mm and the other 20 mm of imidazole. Proteins were eluted with buffer containing 250 mm imidazole and, after concentration, the proteins were applied to a PD10 column (GE Healthcare, Munich, Germany) exchanging the buffer for 100 mm Tris, pH 7.2.

Metal analysis by inductively coupled plasma-optical emission spectrometry (ICP-OES)

Metal analysis was performed using a Perkin-Elmer Optima 2100DV inductively coupled plasma-optical emission spectrometer (Perkin-Elmer, Fremont, CA, USA). Protein samples were incubated overnight in a 1 : 1 mixture with 65% nitric acid (Suprapur, Merck, Darmstadt, Germany) at 100 °C. Samples were filled to a 10-fold volume with water prior to ICP-OES analysis. As reference, the multielement standard solutions XII and XVI (Merck) were used.

Moco/MPT analysis

The Moco and MPT contents of purified hSO were quantified after conversion to Form A-dephospho, as described previously [57]. To determine the MGD content of dimethylsulfoxide reductase, the protein was incubated at 95 °C for 30 min in the presence of acidic iodine to convert MGD to Form A. Released Form A was detected as described previously [57].

Enzyme assays

The activity of hSO (units·mg−1) was determined as described previously [58] by monitoring the reduction of cytochrome c at 550 nm, and is defined as an absorbance change of 1.0 AU·min−1·mg−1 protein.

Dimethylsulfoxide reductase activity (units·mg−1) was assayed as described by McEwan et al. [59] with dithionite-reduced benzyl viologen as the electron donor, and is defined as the reduction of 1 μmol of dimethylsulfoxide·min−1·mg−1 protein.

In vitro incorporation assays

Free MPT was obtained from MPT-containing hSO-MD as described previously [57]. For in vitro metal incorporation into apo-hSO, 20 μm of extracted MPT was incubated with 100 μm of Na2MoO4, Na2WO4, NaAsO2, Cu(NO3)2, CdSO4 or NaVO3. Subsequently, 10 μm of purified apo-hSO was added to the mixture, incubated at 4 °C for 20 min, and free MPT and metal ions were subsequently removed by gel filtration. For the determination of metal incorporation into MPT after insertion into hSO, 40 μm of MPT-containing hSO was incubated at 4 °C for 30 min with 100 μm of Na2MoO4, Na2WO4, NaAsO2, Cu(NO3)2, CdSO4 or NaVO3. Unbound metals were removed by gel filtration using a Nick column (GE Healthcare), and metal incorporation into hSO was determined by ICP-OES analysis.

MPT synthase in vitro assays

To determine the release of MPT from MPT synthase, separately purified R. capsulatus or E. coli MoaE and MoaD1 were assembled and incubated with excess MPT prior to the removal of unbound MPT by gel filtration. Subsequently, 1 mm of Na2MoO4, Na2WO4, NaAsO2, Cu(NO3)2 or CdSO4 was added to 33.2 μm MPT-loaded MPT synthase. After a further incubation step of 10 min, released MPT was removed by an additional gel filtration step and the protein fraction was analysed for MPT and metal content.

To determine the binding of metal-containing MPT to MPT synthase, 1 mm of Na2MoO4, Na2WO4, NaAsO2, Cu(NO3)2 or CdSO4 was added to 60 μm MPT prior to the addition of 21.3 μm MPT synthase. After a further incubation step of 10 min, free MPT and metal ions were removed by gel filtration and the protein fraction was analysed for MPT and metal content.

Acknowledgements

We thank K. V. Rajagopalan (Duke University, Durham, NC, USA) for helpful discussions, and for providing pTG718, pTG818, pMW15aD and pJH820. This work was supported by Deutsche Forschungsgemeinschaft Grant LE1171/3-3 and the Fonds der Chemischen Industrie (FCI).

Ancillary