Selective release and function of one of the two FMN groups in the cytoplasmic NAD+-reducing [NiFe]-hydrogenase from Ralstonia eutropha

Authors


S. P. J. Albracht, Swammerdam Institute for Life Sciences, Biochemistry, University of Amsterdam, Plantage Muidergracht 12, NL-1018 TV Amsterdam, the Netherlands.
Fax: + 31 20 5255124, Tel.: + 31 20 5255130,
E-mail: asiem@science.uva.nl

Abstract

The soluble, cytoplasmic NAD+-reducing [NiFe]-hydrogenase from Ralstonia eutropha is a heterotetrameric enzyme (HoxFUYH) and contains two FMN groups. The purified oxidized enzyme is inactive in the H2-NAD+ reaction, but can be activated by catalytic amounts of NADH. It was discovered that one of the FMN groups (FMN-a) is selectively released upon prolonged reduction of the enzyme with NADH. During this process, the enzyme maintained its tetrameric form, with one FMN group (FMN-b) firmly bound, but it lost its physiological activity – the reduction of NAD+ by H2. This activity could be reconstituted by the addition of excess FMN to the reduced enzyme. The rate of reduction of benzyl viologen by H2 was not dependent on the presence of FMN-a. Enzyme devoid of FMN-a could not be activated by NADH. As NADH-dehydrogenase activity was not dependent on the presence of FMN-a, and because FMN-b did not dissociate from the reduced enzyme, we conclude that FMN-b is functional in the NADH-dehydrogenase activity catalyzed by the HoxFU dimer. The possible function of FMN-a as a hydride acceptor in the hydrogenase reaction catalyzed by the HoxHY dimer is discussed.

Abbreviations
SH

soluble NAD+-reducing hydrogenase

BV

benzyl viologen

EPR

electron paramagnetic resonance

FTIR

Fourier-transform infrared

The facultative lithoautotrophic Knallgas bacterium Ralstonia eutropha H16 contains three different [NiFe]-hydrogenases: a membrane-bound enzyme [1–3], a soluble, cytoplasmic hydrogenase (SH) which reduces NAD+[1,4,5] and a protein functional in a H2-sensing, multicomponent regulatory system [6–9]. The subject of this report is the SH, a heterotetrameric [NiFe]-hydrogenase with subunits HoxF (67 kDa), HoxH (55 kDa), HoxU (26 kDa) and HoxY (23 kDa) [4,10]. The SH comprises two functionally different, heterodimeric complexes [4,5]. The HoxFU dimer constitutes an enzyme module termed diaphorase or NADH-dehydrogenase. It is involved in the reduction of NAD+ and holds one FMN group and several Fe-S clusters. The HoxHY dimer forms the hydrogenase module within the SH.

Minimally, [NiFe]-hydrogenases consist of two subunits of different size [11–13]. The larger subunit accommodates the active Ni-Fe site: a (RS)2Ni(µ-RS)2Fe(CN)2(CO) centre (where R = Cys) [14–22]. The smaller subunit contains at least one [4Fe-4S] cluster situated close to the active site (proximal cluster). In many enzymes the latter subunit harbours two more clusters. The [NiFe]-hydrogenase enzyme from Desulfovibrio gigas contains a second cubane cluster (distal) and a [3Fe-4S] cluster (medial) situated between the two cubanes [14,15]. The SH of R. eutropha belongs to a subclass of [NiFe]-hydrogenases where the polypeptide of the small hydrogenase subunit ends shortly after the position of the fourth Cys residue co-ordinating the proximal cluster [4]. The large HoxH subunit in the SH contains all conserved amino acid residues for binding of the Ni-Fe site [23,24]. Hence, the amino acid sequence suggests that the hydrogenase module in this enzyme only contains the Ni-Fe site and the proximal cluster as prosthetic groups. Fourier-transform infrared (FTIR) studies on the SH indicated that the Ni-Fe site contains two more CN ligands than the active site in standard hydrogenases, and is a (RS)2(CN)Ni(µ-RS)2Fe(CN)3(CO) centre [25]. In contrast to standard hydrogenases, the SH is not sensitive towards oxygen and carbon monoxide and shows no redox changes of the Ni-Fe site. The Fe-S clusters in the HoxFUY subunits and the flavin in the HoxF subunit are all considered to be functional in the intramolecular electron transfer during the H2-NAD+ reaction.

It was shown recently that the protein content of SH preparations is considerably overestimated by the routine colourimetric protein-determination methods. This led to the finding that the SH contains two FMN groups and one NADH-reducible [2Fe-2S] cluster [26]. In the present paper we have investigated the possible role of the two FMN groups. It was found that one of the two groups could be selectively released upon reduction of the SH. The H2-NAD+ activity was thereby lost, but the NADH-dehydrogenase activity was not affected. During this process the enzyme maintained its tetrameric form with one FMN group firmly bound.

Materials and methods

Enzyme purification

R. eutropha cells were cultivated heterotrophically at 30 °C in a mineral medium [27] and stored at −70 °C. The SH was purified at 4 °C in air as described [28] with omission of the cethyltrimethyl-ammoniumbromide treatment. The purified SH was dissolved in 50 mm Tris/HCl pH 8.0 and stored in liquid nitrogen. Unless specified otherwise, this buffer was used in all experiments. The purity of the samples was examined by SDS/PAGE [29]. Protein concentrations were routinely determined by the Bradford method [30] using bovine serum albumin as a standard.

Activity measurements

Hydrogenase activities were routinely measured at 30 °C in a 2.1 mL cell with a Clark electrode (type YSI 5331) for polarographic measurement of H2 (Yellow Springs Instruments, Yellow Springs, OH, USA) [31]. For H2-consumption measurements under aerobic conditions the cell was filled with aerobic buffer, 5–10 µL enzyme and H2-saturated water to a final H2 concentration of 36 µm. Subsequently, NADH (5 µm) was added to activate the enzyme, followed by either benzyl viologen (BV, 1 mm) or NAD+ (5 mm) as electron acceptor. When anaerobic conditions were used, all solutions were flushed with Ar before use. To remove residual oxygen, glucose (50 mm) plus glucose oxidase (9 U·mL−1) were added to the reaction medium 3 min before the addition of NADH. Hydrogen was passed over a palladium catalyst (Degussa, Hanau, Germany; type E236P) and Ar through an Oxisorb cartridge (Messer-Griesheim, Düsseldorf, Germany) to remove oxygen. NADH-dehydrogenase activity with K3Fe(CN)6 as electron acceptor was measured aerobically in buffer at 30 °C. The absorption decrease at 420 nm was monitored using a Zeiss M4 QIII spectrophotometer (ε = 1 mm−1·cm−1 for K3Fe(CN)6 at 420 nm). NADH (1.25 mm) and 5 µL sample were added and 3 min later the reaction was started by the addition of 1 mm K3Fe(CN)6.

The specific hydrogenase activities with both NAD+ and BV as acceptors of enzyme, purified from different cell batches varied considerably (17–84 and 12–63 U·mg−1, respectively; 1 U = 1 µmol·min−1). The NADH-K3Fe(CN)6 activities (125–175 U·mg−1) and the intensity of the electron paramagnetic resonance signal from the [2Fe-2S]+ cluster in NADH-reduced enzyme preparations varied much less. The relative decrease in activity observed upon reduction was, however, the same for all enzyme samples used in this study. As outlined in the present paper, the variable hydrogenase activities can be ascribed in part to the lack of FMN-a in a portion of the enzyme molecules.

Electron paramagnetic resonance (EPR) spectroscopy

EPR measurements were carried out as before [32]. The enzyme concentration was determined by double integration of a good-fitting simulation of the EPR signal of the [2Fe-2S] cluster in NADH-reduced enzyme.

FMN determination

Acid-extractable flavin was determined fluorimetrically [33], using FMN (synthetic from Sigma) as a standard, in a Shimadzu RF-5001PC spectrofluorimeter (Kyoto, Japan). The concentration of the standard (in a buffer solution of pH 6.9) was calculated from the difference in absorption at 450 nm before and after addition of excess dithionite using an extinction coefficient of 11.2 mm−1·cm−1[34]. The FMN content of the preparations used in this study was between 1.51 and 1.84 FMN per EPR-detectable [2Fe-2S] cluster. For kinetic measurements of the release of FMN, a Spex Fluorolog III spectrofluorimeter was used (Spex Industries, Edison, NJ, USA). Experiments were performed aerobically in buffer at room temperature. In this case the concentration of released FMN was calculated from the fluorescence of a series of known FMN additions.

Determination of the apparent molecular mass by size-exclusion chromatography

This was performed on a Pharmacia FPLC machine fitted with a Superdex-12 (HR 10/30) column. Ribonuclease A (13.7 kDa), chymotrypsinogen A (25 kDa), ovalbumin (43 kDa), bovine serum albumin (67 kDa) and glucose oxidase (183 kDa) were used as molecular markers. Enzyme was eluted with buffer containing 100 mm NaCl with additions mentioned in the text.

Results and discussion

Effect of reduction of the SH on its H2-NAD+ and H2-BV activities

When SH was incubated anaerobically with H2 and NADH, the H2-NAD+ activity dropped, within 4 min, to a steady level (Fig. 1A). The decrease in activity was most pronounced at low enzyme concentrations.

Figure 1.

Effect of reduction on the SH activity. Glucose (50 mm) and glucose oxidase (9 U·mL−1) were added to the enzyme in buffer in a closed H2-reaction cell at 30 °C. After 3 min, which allowed for the consumption of residual O2, H2 (36 µm) and NADH (5 µm) were added. Subsequently, either 5 mm NAD+ (A) or 1 mm BV (B) were added at the indicated times and the H2 uptake activity was measured. The experiment was carried out with 27 nm (▴), 6.8 nm (▪) or 1.7 nm (•) enzyme. Data are averages of three experiments. The H2-NAD+ activity of untreated enzyme was 31 U·mg−1.

The H2-BV activity, however, was hardly affected by this treatment (Fig. 1B). These results are in agreement with previous observations [1].

When the experiment was performed aerobically a different result was obtained (Table 1). Both the H2-NAD+ and H2-BV activities decreased considerably.

Table 1. The effect of air on the reductive inactivation of the SH. In a closed H2-reaction cell, H2 (36 µm) and NADH (5 µm) were added to enzyme (4.2 nm) in aerobic buffer at 30 °C. The anaerobic control experiment was performed as in Fig. 1. The rate of reduction of NAD+ (5 mm) or BV (1 mm) was measured either directly after the addition of H2/NADH or 8 min later. Data are the minimal and maximal values of three measurements. Experiments with two other enzyme preparations gave similar results.
ReactionActivity (U·mg−1)
AerobicAnaerobic
t = 0t = 8 mint = 0t = 8 min
H2-NAD+50.8–64.513.3–15.874.0–78.414.0–18.5
H2-BV40.7–43.06.9–7.736.4–52.341.6–47.5

Release of FMN from the reduced enzyme

We discovered that the reduced SH released 0.6–0.8 mol FMN per mol enzyme (Table 2). About 0.9 mol FMN per mol enzyme remained bound to the SH. The H2-NAD+ activity decreased dramatically (not shown). However, the NADH-K3Fe(CN)6 activity did not change. It is concluded that the diaphorase dimer was not affected and fully retained its FMN. Release of FMN was also observed upon reduction with dithionite in the presence of H2.

Table 2. Release of FMN upon reduction of the SH and effect on the NADH-K3Fe(CN)6 activity. A mixture of 100 µL enzyme (23 µm as determined by EPR), 100 µL 5 mm NADH and 1.8 mL buffer was dialyzed (cut-off size 30 kDa) against 98 mL H2-saturated buffer in a capped serum bottle under a H2 atmosphere. The contents of the bottle were gently stirred at 30 °C in the dark. Two controls were run also, one with 30 µm FMN instead of enzyme and the other with buffer alone. After 3 h, a sample of the solution outside the dialysis bag was aerated for 3 min and then assayed for FMN. The solution inside the dialysis bag was tested for NADH-K3Fe(CN)6 activity and acid-labile FMN. The experiment has been performed with three different preparations. Data for each preparation are the minimal and maximal values of three measurements. NADH-K3Fe(CN)6 activity is the specific activity compared to that of untreated enzyme. Bound, acid-labile FMN from the protein inside the dialysis bag, corrected for the contribution of the free FMN in the sample volume; Free, free FMN in the buffer outside the dialysis bag; ND, not determined.
PreparationFMN (mol per mol SH)NADH-K3Fe(CN)6 activity (%)
TotalBoundFree
ANDND0.72–0.90+0.4
B1.77–1.870.85–0.96ND−4.1
C1.43–1.550.79–0.920.56–0.63+2.0

It has been reported [35] that dilution of oxidized, aerobic enzyme would lead to an increased fluorescence due to loss of FMN. In this study the oxidized SH was stable under aerobic conditions and did not lose any FMN upon dilution.

In the following we will refer to the FMN released upon reduction as FMN-a and the one located in the HoxF subunit as FMN-b.

Kinetics of the release of FMN induced by reduction with NADH

When an aerobic enzyme solution was monitored in a fluorimeter at excitation and emission wavelengths specific for free oxidized FMN, no change in fluorescence was observed during 15 min after addition of 80 µm H2 (not shown). An immediate increase in fluorescence occurred, however, after the addition of 10 µm NADH (Fig. 2, trace A). The presence of H2 did not alter this effect (Fig. 2, trace B).

Figure 2.

Release of FMN upon reduction of the SH as observed by fluorescence. (A) Enzyme (12.5 nm) and NADH (10 µm) were added as indicated. (B) Enzyme (12.5 nm), H2 (27 µm) and NADH (10 µm) were added as indicated. The experiment was performed in aerobic buffer at room temperature. Changes of FMN fluorescence were monitored in a fluorimeter (excitation at 450 nm; emission at 530 nm). The H2-NAD+ activity of the untreated enzyme was 41 U·mg−1. E, enzyme.

We ascribe this to the release of the reduced FMN-a group from the protein. Once in solution the reduced flavin is auto-oxidized in the aerobic buffer giving rise to a strong fluorescence. The fluorescence reached a plateau ≈ 150 s after the addition of NADH. The traces represent a zero-order reaction with a half time of about 30 s. If the protein concentration was decreased, the relative amount of released FMN increased, but the half time of the event did not change. For example, when 3.1 nm enzyme was used, 0.79 mol FMN per mol enzyme was released into the medium, as calculated from the change in fluorescence.

Recovery of the H2-NAD+ activity by addition of FMN

The previous experiments showed that reduction of the SH by NADH leads to a rapid decrease of the H2-NAD+ activity, presumably due to the release of the FMN-a group. Figure 3 shows that addition of a thousand-fold excess FMN (10 µm) to enzyme, previously reduced by NADH plus H2 for 7 min, reconstituted the H2-NAD+ activity instantaneously.

Figure 3.

The stimulatory effect of FMN on enzyme pretreated by reduction. Enzyme (3.5 nm, H2-NAD+ activity 20.7 U·mg−1) in aerobic buffer was incubated for 7 min at 30 °C with 5 µm NADH plus 36 µm H2. The H2-NAD+ activity was then measured by the addition of 5 mm NAD+ (5.3 U·mg−1). After 2 min, 10 µm FMN was added, resulting in an increase in activity (23.1 U·mg−1). A similar decrease and restoration of activity was obtained if H2 was added after the incubation period of 7 min.

If the reduced enzyme was first oxidized, then FMN had no immediate effect on this activity. Addition of 10 µm FMN to untreated enzyme did not result in H2 uptake in the presence of H2 + 5 µm NADH (not shown), excluding FMN as electron acceptor at this concentration. The experiment in Fig. 3 also shows that upon addition of FMN, the activity (23.1 U·mg−1) increased beyond the original activity (20.7 U·mg−1). Apparently, some enzyme molecules were originally deficient in FMN-a and could now pick up added FMN. Such a stimulatory effect of FMN, but not of FAD or riboflavin, has been noticed earlier [36,37].

Figure 4 shows the effect of the FMN concentration on the reconstitution of the activity of the reduced SH. Addition of about 80 nm FMN induced half maximal activity.

Figure 4.

Effect of the FMN concentration on the H2-NAD+ activity of enzyme, which was first reduced in aerobic buffer. Enzyme (3.5 nm, H2-NAD+ activity 16.8 U·mg−1) in aerobic buffer was incubated for 7 min at 30 °C with 5 µm NADH plus 36 µm H2. The H2-NAD+ activity was then measured by addition of 5 mm NAD+. Two minutes later, variable amounts of FMN were added and the effect on the rate was measured by the method depicted in Fig. 3. With low FMN concentrations a steady-state activity was only obtained some time after the addition of FMN. This time interval decreased with increasing amounts of FMN. For the FMN concentrations used; 10, 25, 100, 250 and 1000 nm (and 10 µm; not shown), these times were 122, 105, 79, 52, 13 (and <2) seconds, respectively (data not shown). Data are averages of three measurements.

As before, the maximal activity obtained upon FMN addition (20.9 U·mg−1 with 10 µm FMN added) was 25% higher than the original activity (16.8 U·mg−1), indicating that part of the original enzyme molecules did not contain FMN-a.

Integrity of the SH during the release of FMN-a

Our experiments show that both the extent of the drop in activity as well as the amount of released FMN were dependent on the enzyme concentration, suggesting a dissociation–association reaction. It has been suggested, but not shown [35,38], that the SH from R. eutropha can dissociate into the NADH-dehydrogenase module (HoxFU) and the hydrogenase module (HoxHY). Dissociation such as this has been clearly demonstrated for the related NAD+-reducing hydrogenase from Rhodococcus opacus[39–41]. We have tried to verify this for the R. eutropha SH by gel-filtration experiments under different conditions (Table 3).

Table 3. Apparent molecular mass of the SH determined by size-exclusion chromatography under various elution conditions. Apparent mass, the used enzyme had a H2-NAD+ activity of 84 U·mg−1; Activity, specific activity in the H2-NAD+ assay as determined after elution; Activity reconstituted with FMN, specific activity in the H2-NAD+ assay as determined after elution but with 100 µm FMN added after the H2, NADH and NAD+ additions; ND, not determined.
ConditionApparent Mass (kDa)Activity (%)Activity reconstituted with FMN (%)
  1. a Protein fractions from condition B were collected, pooled, rebuffered in aerobic buffer with 25 µm K3Fe(CN)6 and rerun. b A preincubation (5 min, 30 °C) with H2, 10 µm NADH and 100 µm FMN was required for optimal activity.

A – Aerobic buffer, 25 µm K3Fe(CN)616494ND
B – Anaerobic buffer, 5 µm NADH, 0.8 mm H2192 0121
C – As B, plus oxidative treatmenta; aerobic buffer, 25 µm K3Fe(CN)6159 0116 b

Untreated enzyme in aerobic buffer containing 25 µm K3Fe(CN)6 eluted with an apparent mass of about 164 kDa. A higher value (192 kDa), but not a lower one, was obtained when the elution buffer was reducing (Table 3; condition B). When enzyme, eluted under reducing conditions, was reoxidized the apparent mass was 159 kDa (Table 3; condition C). The presence of FMN (1.3 µm) did not affect the mass of the SH under the different conditions (not shown).

The SH activity was not affected by gel filtration under oxdizing conditions, but under reducing conditions all activity was lost (Table 3; conditions A, B). This indicates that all FMN-a could be removed upon reduction of the enzyme. At the same time, however, no apparent dissociation of the tetrameric enzyme into the individual diaphorase and hydrogenase modules could be observed. It is concluded that reduction by NADH opens up the enzyme such that the FMN-a group is released.

The role of the FMN-a group in activation of the SH

The H2-NAD+ activity of the enzyme after gel-filtration under reducing conditions could be restored (121%) by addition of 100 µm FMN to the activity assay (Table 3; condition B). For the enzyme treated as in condition C, the activity could not be restored in this way. Instead an anaerobic preincubation for 5 min at 30 °C in the presence of NADH (10 µm), H2 and FMN was required to recover the activity (116%).

The specific H2-BV activity of the enzyme after gel filtration under reducing conditions (Table 3; condition B) was 92% of the original activity. This is in line with the experiments in Tables 1 and 2, and supports the notion that FMN-a is not required for this reaction. Subsequent gel filtration under oxidizing conditions (Table 3; condition C), however, resulted in the total loss of this activity when assayed in the usual way, i.e. after addition of H2, a catalytic amount of NADH and the subsequent addition of BV. Restoration of this activity (to 88%) also required the 5 min preincubation procedure mentioned above.

These observations can be explained as follows. The enzyme devoid of FMN-a and oxidized with K3Fe(CN)6 in air has a Ni-Fe site which cannot react with H2. We propose that this is due to the occupation of the sixth coordination site on nickel by an oxygen species (presumably OH). The 6th ligand must be removed and it is proposed that this is induced by supplying reducing equivalents (from 5 µm NADH or chemical reductants). The mechanism of this reductive activation is not understood. In untreated enzyme, this leads to an instantaneous activation whereupon the reaction with H2 commences. Our experiments show that when FMN-a is missing, such a rapid activation cannot occur, not even in the presence of excess FMN. Apparently, bound FMN-a is required for this to happen. The experiments demonstrate that the release or re-binding of flavin at the FMN-a binding site occurs only in reduced enzyme and that FMN-a is essential for the NADH-induced activation of the Ni-Fe site in the SH, as well as for the H2-NAD+ reaction.

Conclusions

The SH contains two FMN groups [26,37]. The experiments presented here demonstrate, for the first time, that upon reduction of the enzyme by NADH, one of the two FMN groups (FMN-a) is specifically released, while the other FMN group (FMN-b) remains bound.

In contrast to the behaviour of the enzyme from R. opacus[39–41], no apparent dissociation of the SH could be observed under oxidizing or reducing conditions. The oxidized SH did not release FMN when diluted in aerobic buffer; this observation is at variance with a previous report [35].

The experiments lead us to the following conclusions and proposals about the reduction-induced changes in the SH (the current working model is depicted in Fig. 5): (a) FMN-a can be specifically released upon reduction of the enzyme by NADH via the HoxFU module. It is proposed that the SH undergoes a conformational change such that the FMN-a group can dissociate from the enzyme. (b) FMN-a is essential for the H2-NAD+ activity, but not for the H2-BV activity. (c) Reconstitution of the H2-NAD+ activity of enzyme deficient in FMN-a can only occur by adding FMN to the reduced enzyme, but not to the oxidized enzyme. (d) FMN-a is essential for the rapid activation of the Ni-Fe site induced by reducing equivalents from NADH. (e) The SH in crude extracts and in the purified form lacks part of the bound FMN-a (up to 40%). This explains the increase of the H2-NAD+ activity when FMN is added to the reduced enzyme (this work and [36,37]). (f) It is proposed that the FMN-a is bound to the inwards-pointing end of the flavodoxin fold in the HoxY subunit. Such a flavodoxin fold is conserved in the small subunit of all [NiFe]-hydrogenases [42]. It is hypothesized that FMN-a is positioned close to the Ni moiety of the Ni-Fe site. (g) In standard [NiFe]-hydrogenases, where the valence state of the nickel ion can change, it is presently assumed that the Ni3+ ion is transiently reduced to a monovalent state by the hydride, produced after the heterolytic cleavage of H2. Subsequently one electron is rapidly transferred to the proximal Fe-S cluster and nickel oxidizes to Ni2+[13]. The Ni-Fe site in the SH shows, however, no apparent redox changes [25]. We therefore propose that FMN-a in the SH functions as a two-to-one electron converter between the hydride, produced by the heterolytic cleavage of H2 at the 6th coordination site on Ni, and the Fe-S clusters in the SH. Our current hypothesis involves a direct hydride transfer from a Ni2+-hydride intermediate to FMN-a. Future experiments are required to verify this tentative idea. (h) As electron transfer from the hydride (formed at nickel) to the Fe-S clusters is hampered by the absence of the FMN-a, it is unlikely that BV obtains electrons from any of the Fe-S clusters during the H2-BV reaction. The release of FMN-a upon reduction of the SH by NADH indicates that the enzyme opens up. It is hypothesized that in this state BV is able to directly react with the active site (Fig. 5).

Figure 5.

Current model of the reduction-induced changes in the tetrameric, soluble, NAD+-reducing [NiFe]-hydrogenase from R. eutropha. The reason for the specific arrangements of the subunits was provided in [43]. In the Desulfovibrio gigas enzyme, the two hydrogenase subunits are bound via a large hydrophobic contact surface [15]. The protein part that holds two of the Fe-S clusters in the small subunit of the D. gigas enzyme is missing in HoxY of the SH, suggesting an exposure of part of the hydrophobic surface of the HoxH subunit. It was therefore proposed that the HoxU subunit of the NADH-dehydrogenase module binds to this surface. This would enable a smooth electron transfer from the hydrogenase module to the NADH-dehydrogenase module. (A) The aerobic enzyme, as isolated, has a (RS)2(CN)(OH)Ni(µ-RS)2Fe(CN)3(CO) centre (R = Cys) [25]. Both nickel and iron are six coordinated and hence the centre cannot react with H2. The two flavins are firmly bound. (B) Reduction of the prosthetic groups in the HoxF, HoxU and HoxY subunits (reduction indicated as dashed pattern) by NADH (via FMN-b) induces an instantaneous activation of the Ni-Fe site (removal of the OH ligand from nickel) enabling the interaction with H2. As long as FMN-a is present, this flavin can accept the hydride produced upon heterolytic cleavage of H2 at nickel. At the same time, however, the reduced enzyme opens up and slowly (minutes) releases reduced FMN from the FMN-a site. The open enzyme configuration allows BV to approach the active site. (C) Once FMN-a is lost, hydride transfer from dihydrogen activated at the Ni-Fe site is no longer possible. The dissociation of FMN is reversible and so addition of external FMN restores the H2-NAD+ reaction. FMN-a is not required for the H2-BV reaction. If enzyme deficient in FMN-a is inactivated by O2, creating an OH ligand at nickel, it cannot be reactivated by NADH and so both the H2-NAD+ and H2-BV reactions are absent.

Acknowledgements

Dr J. Zwier (Institute of Molecular Chemistry, University of Amsterdam) is acknowledged for the use of the Spex Fluorolog III fluorimeter. This work was supported by the Netherlands Organization for Scientific Research (NWO), the Deutsche Forschungsgemeinschaft, the Fonds der Chemischen Industrie, EU-project BIO4-98-0280 and the European Union Cooperation in the field of Scientific and Technical Research (COST), Action-818.

Ancillary