A two-component flavin-dependent monooxygenase involved in actinorhodin biosynthesis in Streptomyces coelicolor

The two-component flavin-dependent monooxygenases belong to an emerging class of enzymes involved in oxidation reactions in a number of metabolic and biosynthetic pathways in microorganisms. One component is a NAD(P)H:flavin oxidoreductase, which provides a reduced flavin to the second component, the proper monooxygenase. There, the reduced flavin activates molecular oxygen for substrate oxidation. Here, we study the flavin reductase ActVB and ActVA-ORF5 gene product, both reported to be involved in the last step of biosynthesis of the natural antibiotic actinorhodin in Streptomyces coelicolor. For the first time we show that ActVA-ORF5 is a FMN-dependent monooxygenase that together with the help of the flavin reductase ActVB catalyzes the oxidation reaction. The mechanism of the transfer of reduced FMN between ActVB and ActVA-ORF5 has been investigated. Dissociation constant values for oxidized and reduced flavin (FMNox and FMNred) with regard to ActVB and ActVA-ORF5 have been determined. The data clearly demonstrate a thermodynamic transfer of FMNred from ActVB to ActVA-ORF5 without involving a particular interaction between the two protein components. In full agreement with these data, we propose a reaction mechanism in which FMNox binds to ActVB, where it is reduced, and the resulting FMNred moves to ActVA-ORF5, where it reacts with O2 to generate a flavinperoxide intermediate. A direct spectroscopic evidence for the formation of such species within ActVA-ORF5 is reported.


SUMMARY
The two-component flavin-dependent monooxygenases belong to an emerging class of enzymes involved in oxidation reactions in a number of metabolic and biosynthetic pathways in microorganisms. One component is a NAD(P)H:flavin oxidoreductase which provides a reduced flavin to the second component, the proper monooxygenase. There, the reduced flavin activates molecular oxygen for substrate oxidation. Here, we study the flavin reductase ActVB and ActVA-ORF5 gene product, both reported to be involved in the last step of biosynthesis of the natural antibiotic actinorhodin in Streptomyces coelicolor. For the first time, we show that ActVA-ORF5 is a FMN-dependent monooxygenase which together with the help of the flavin reductase ActVB catalyze the oxidation reaction. The mechanism of the transfer of reduced FMN between ActVB and ActVA-ORF5 has been investigated.
Dissociation constant values for oxidized and reduced flavin (FMN ox and FMN red ) with regard to ActVB and ActVA-ORF5 have been determined. The data clearly demonstrate a thermodynamic transfer of FMN red from ActVB to ActVA-ORF5, without involving a particular interaction between the two protein components. In full agreement with these data, we propose a reaction mechanism in which FMN ox binds to ActVB where it is reduced and the resulting FMN red moves to ActVA-ORF5 where it reacts with O 2 to generate a flavinperoxide intermediate. A direct spectroscopic evidence for the formation of such species within ActVA-ORF5 is reported.

INTRODUCTION
There is a great variety of monooxygenases, catalyzing oxygen activation and oxygen transfer reactions in the living world. The most extensively studied ones are the cytochrome P450-dependent monooxygenases (1)  Therefore, in this two-component system, NAD(P)H oxidation and the hydroxylation reaction are catalyzed by separate polypeptides. For years, the only prototype for this class of monooxygenases was the luciferase enzyme which has been the subject of numerous studies (3,5,6). However, it is only recently that the importance of such systems in oxidative metabolism, in particular within microorganisms, has been realized.
If we except the enzymes involved in the luciferase reaction, the flavin reductases associated with the oxygenase component display strong sequence homologies (4). These flavin reductases have been extensively studied from a biochemical (7-10) and a structural point of view (11,12). On the other hand, much less is known on the structure and the enzymatic mechanisms of the monooxygenase component.
The monooxygenase components can be further divided into two groups on the basis of the type of the flavin used in the reaction. The first group consists in FADH 2 -utilizing monooxygenases which are represented by several well-characterized systems such as 4hydroxyphenylacetate 3-monooxygenase (HpaB) of Escherichia coli (13) and 2,4,5trichlorophenol monooxygenase from Burkholderia cepacia (14,15). 2,4,6-trichlorophenol monooxygenase (TcpA) from Ralstonia eutropha (16), phenol hydroxylase (Phe A1) from Bacillus thermoglucosidasius (9) and styrene monooxygenase (StyA) from Pseudomonas flurescens ST (17,18) belong to that family but have not been characterized yet. The second group consists in FMNH 2 -utilizing monooxygenases which have been much less investigated in terms of their structure, mechanism and substrate specificity. Examples of such FMNdependent systems are those involved in the synthesis of the antibiotic pristinamycin in Streptomyces pristinaespiralis (19,20), biodegradation of polyaminocarboxylates such as ethylenediamine tetracetic acid (EDTA) (21) or nitrilotriacetic acid (NTA) (22) by microorganisms, utilization of sulfur from aliphatic sulfonates (23) and desulfurization of fossil fuels by Rhodococcus species (24).
The results reported here concern the enzyme system we have selected to study the chemistry of FMN-utilizing monooxygenases. The system participates to the last step of the biosynthesis of the antibiotic actinorhodin in Streptomyces coelicolor (Scheme 2) which consists in the hydroxylation of the precursor dihydrokalafungin and the coupling of two molecules of the hydroxylated product (25-27).
Indications that this step was catalyzed by a two-component flavin-dependent monooxygenase came from genetic (25) and biochemical studies (7,10,25). Actinorhodin biosynthesis involves about 20 proteins whose corresponding genes are localized in the same region of the chromosome (27). Inactivation of the actVB gene resulted in the accumulation of dihydrokalafungin without any production of actinorhodin (25). Purification and on the column was achieved with 4 column volumes of Tris-HCl 50 mM pH 7.6, 10 % glycerol and the protein was recovered with 3 column volumes of 500 mM imidazole, 25 mM Tris-HCl pH 7.6, 10% glycerol. As apoActVB was unstable in the presence of imidazole, the protein solution was immediately loaded onto a Nap 10 column and eluted with 25 mM Tris-HCl pH 7.6, glycerol 10%. The yield of the overall process was about 40%.
Analytical Determinations. The native molecular mass of ActVA was determined with an analytical Superdex 200 gel filtration (Amersham Biosciences) equilibrated with 10 mM Tris-HCl pH 7.6 containing 150 mM NaCl. Bovine serum albumin (66 kDa), ovalbumin (45 kDa), chymotrypsin (20.1 kDa) and cytochrome c (12.4 kDa) were used as markers of molecular mass. The void volume was determined with ferritin (450 kDa). Protein concentration was determined using the Bio-Rad protein assay reagent, with bovine serum albumin as a standard.
Flavin Binding to ActVA-ORF5 and ActVB. UV-Visible absorbance spectra were recorded with a Hewlett-Packard 8453 photodiode array instrument at 18°C. K d values of FMN ox for ActVB and ActVA were calculated from the variation of the absorbance values at the selected wavelengths, using eq 1: where A 0 is the initial absorbance, A x the absorbance after a given addition of the protein, A f the absorbance at the end of the titration. Making the hypothesis that ActVA and ActVB contain one binding site per monomer, [FMN bound ] and [protein free ] were calculated from the total concentrations of FMN ([FMN tot ]), and protein ([protein tot ]), using eq 2 and eq 3: ActVA fluorescence emission spectra were recorded at 18 °C with a Jasco FP 6500 spectrofluorimeter, using a 1 cm square cuvette (Helma). The excitation wavelength was set to 295 nm. The measured fluorescence spectra were corrected from inner filter effects due to FMN ox or FMN red absorption during excitation as well as during emission. The corrected fluorescence (F corr ) was calculated according to the method described in (30), using eq 4: where L is the cuvette full path length (1 cm), [FMN] the total concentration of FMN, e ex , the extinction coefficient of FMN at excitation wavelength (295 nm), e em , the extinction coefficient of FMN for each emission wavelength between 300 and 400 nm. Since the inner filter effect is not homogenous in the entire cuvette, we divided it into 100 identical points and defined their position by x and y coordinates making the hypothesis that their surfaces are infinitely small. For each position and each wavelength between 295 and 400 nm, the factors representing the inner filter effect were calculated and were subsequently summed in order to determine an F corrected (F corr ) value. F corr precisely corresponds to the intrinsic protein fluorescence.
K d values were calculated from the variation of the corrected fluorescence intensity at 338 nm (F corr ) using eqs 1, 2 and 3.
Anaerobic experiments. Anaerobic experiments were performed in a Jacomex glove box equipped with a UV-visible cell coupled to a Uvikon XL spectrophotometer by optical fibers (Photonetics system). All the solutions were incubated anaerobically 2 hours before the beginning of each experiment. For FMN reduction, an FMN stock solution (500 mM) was anaerobically and quantitatively photoreduced by 30 min irradiation with a commercial slide projector placed at a distance of 3 cm, in the presence of 2 mM deazaflavin and 10 mM EDTA. For oxidation of reduced flavin, as a standard procedure, aerated water was prepared for each experiment by bubbling a 100 % oxygen gas for 30 min into an air-tight eppendorff tube containing 1 ml of water. FMN red oxidation was performed in the anaerobic glove box by injecting 5 m l of the oxygen saturated water (1 mM) into a 100 mL air-tight spectrophotometric cuvette with an Hamilton syringe.

ActVA-ORF5 as the potential flavin-dependent hydroxylase
Among the six genes present in the actVA region (26), only ORF5 displays sequence homologies with flavin-dependent hydroxylases. This becomes particularly obvious when the sequence of the corresponding protein is compared to the product of the tdsC gene from Paenibacillus spA11-2 (31) and the product of the pheA gene (32) (Supplementary data Figure 1). TdsC is a FMN-utilizing monooxygenase associated with TdsD, a NADH:FMN oxidoreductase, and involved in the oxidation of dibenzothiophene (31). A similar system is found in Rhodococcus erytropolis with DszC acting as a FMN-utilizing monooxygenase and DszD as a NADH:FMN oxidoreductase displaying strong sequence homologies with ActVB (24). Therefore, this analysis suggests that actVA-ORF5 gene product might be the monooxygenase associated with ActVB for the last step of actinorhodin biosynthesis.

Purification of ActVA-ORF5
The details of the methods used to clone, express the actVA-ORF5 gene in E. coli and purify the corresponding protein are given in the experimental section. Briefly, E. coli Bl 21 (DE3) pLysS strain was transformed with the expression vector pActVA-ORF5 derived from pT 7-7 containing the actVA-ORF5 gene. Bacterial soluble extracts were subjected to an anion exchange chromatography and the enriched fractions, as judged by SDS-PAGE analysis, were further purified by gel filtration on a Superdex 200 column. The protein was obtained 95 % pure as determined by SDS-PAGE analysis (data not shown). No proteolysis occurred all through the purification process (data not shown). Determination of its N-terminal sequence (SEDTHT) confirmed that the purified protein was indeed the product of the actVA-ORF5 gene. For the sake of simplicity, in the following, this protein is named ActVA. Analysis by mass spectroscopy showed that the protein, with a molecular mass of 39,715 Da, lacked the N terminal methionine. Chromatography of pure ActVA on a calibrated Superdex 200 gel filtration column showed that the protein was dimeric in solution, with an experimental mass of about 77,000 Da (data not shown). Finally, no evidence for a protein-bound chromophore could be obtained by UV-visible spectroscopy.
At room temperature the protein spontaneously precipitated in the presence of 250 mM NaCl. Thus, the following experiments using pure ActVA were carried out in 50 mM Tris-HCl pH 7.6 buffer, in the absence of salt.

ActVA-ORF5, an active monooxygenase system
The precursor of actinorhodin is dihydrokalafungin. However, this compound is not commercially available and difficult to obtain either from natural sources (25) or by chemical synthesis (33). Thus, we used 1,5-dihydroanthraquinone (DHAQ) as a substrate analog to check whether the ActVA protein was indeed able to catalyze an hydroxylation reaction in combination with ActVB. The assay mixture, under aerobic conditions, contained 200 µM NADH, 5 µM FMN, 50 µM DHAQ, 11 nM ActVB and various amounts of ActVA, in 2 ml of 50 mM Tris-HCl pH 7.6. After one hour at 30 °C, DHAQ and its oxidation products were extracted with MeOH and analyzed by ESI-MS. As shown in Figure 1, a peak at m/z = 255 was observed besides the peak at m/z = 239 corresponding to DHAQ. This proved the formation of a product corresponding to DHAQ plus one oxygen atom (+16), thus deriving from DHAQ by monooxygenation. This peak was absent from the mass spectrum when one of the following components of the system was excluded from the reaction mixture, FMN, NADH, ActVB or DHAQ (data not shown). In addition, a control reaction carried out within an anaerobic glove box did not result in the production of the oxygenated product either (data not shown). In conclusion, taken together, these preliminary experiments reported here clearly indicate that the product of the actVA-ORF5 gene is a flavin-dependent monooxygenase.

Interaction of FMN with ActVA and ActVB: dissociation constants
In order to determine the mechanism of the transfer of reduced flavins between ActVB and ActVA, the dissociation constant (K d ) values for reduced and oxidized FMN with ActVB and ActVA proteins were determined.
As reported previously, a significant proportion (20-30 %) of the purified ActVB polypeptides contained one bound FMN (10). Removal of FMN for the production of ActVB in the pure apoprotein form, named apoActVB, was thus achieved first prior to the binding experiments. The use of the his-tagged form of ActVB bound to a Ni-NTA column allowed a fast and efficient procedure for the preparation of apoActVB, as described in the experimental section. The protein bound to the Ni-NTA column was treated first with a solution containing 2 M urea and 2 M KBr. This was followed by renaturation, elution with imidazole and desalting. The protein obtained from that process retained its ability to bind FMN, as judged from the appearance during incubation of apoActVB with FMN, of an absorption band, centered to 457 nm, characteristic of the ActVB:FMN complex (10). ApoActVB was shown to be active as a flavin reductase in a standard assay using NADH and riboflavin or FMN as substrates, with K m values for flavins (data not shown) comparable to those obtained with the as-isolated enzyme (10). Taken together, these data showed that apoActVB was fully active, well-folded and could thus be used in our following investigations.  (10), but also to ActVA. Therefore, the l max value could be used as an indicator for whether FMN ox is free or bound to ActVA or ActVB.
As shown in Figure 3 Since FMN red does not exhibit marked absorbance bands, K d values for FMN red could not be obtained by the same spectrophotometric titration experiments. For that purpose, we took advantage of specific spectroscopic properties of FMN red as described in the following.
In the case of ActVB, binding of FMN red in its active site in the presence of NAD + results in the formation of a broad absorption band between 550 and 800 nm which is characteristic for a charge-transfer (CT) complex between NAD + and FMN red ((10) and Figure   4). Such a complex is also formed as the result of the oxidation of NADH by FMN ox in the active site of ActVB. As shown in Figure 4, the intensity of the charge transfer band increased upon successive addition of FMN red in the presence of an excess of NAD + . A plot of the bound ActVB versus free FMN red determined from the variation of the absorbance of the charge transfer band at 680 nm gave a K d value of 6.6 ± 0.6 mM for FMN red with regard to the [ActVB-NAD + ] complex (Inset Figure 4).
Finally, in the case of ActVA, the well-established ability of FMN red to quench the fluorescence of the apoproteins (upon excitation of tryptophan residues at 295 nm) allowed us to design another method for the determination of the K d value for FMN red . This approach could be also carried out with FMN ox allowing comparison with the results described above, obtained by UV-visible spectrophotometry. It should be mentioned that such experiments could not be carried out with ActVB since the protein is totally devoid of trytophan residues (27).  from ActVA to ActVB and of FMN red in the reverse direction. In the first one, 10 µM of FMN ox was complexed to an excess of ActVA (140 µM) and, accordingly, the light absorption spectrum of the solution displayed a band centered at 438 nm (data not shown).
Addition of increasing amounts of apoActVB resulted in a shift of that band to lower energies (data not shown). As shown in Figure 6, the l max value increased to reach 457 nm after addition of about one equivalent of apoActVB with regard to ActVA, with no further changes for larger apoActVB/ActVA ratios. The final value of l max was characteristic for the FMN ox -ActVB complex, demonstrating that FMN ox , initially in ActVA, had been transferred to and was bound to apoActVB. Taking into account the concentrations of the different components in the solution, a total transfer of FMN ox from ActVA to ActVB at a [ActVB]/[ActVA] ratio of 1 was consistent with the K d values reported in Table 1 1 . In a second experiment, FMN ox was bound to an excess of ActVB (thus light-absorbing at 457 nm) and was reduced by NADH under anaerobic conditions (data not shown). Reduction could be monitored by visible spectroscopy not only from the disappearance of the 457 nm band, but also from the appearance of the broad absorption between 550 and 800 nm characteristic for the CT complex between NAD + and FMN red discussed above (Figure 4).
Anaerobic addition of one equivalent of ActVA resulted in the instantaneous disappearance of about 90% of the latter absorption (data not shown). This indicated disruption of the chargetransfer complex, most likely resulting from a transfer of FMN red from ActVB to ActVA.

Oxidation of FMNH 2 : evidence for a flavin-peroxide intermediate
As shown above, FMN red is a good ligand for ActVA. In the following, the study of the reaction of the ActVA:FMN red complex with molecular oxygen is reported. In the experiment shown in Figure 7, 16 µM FMN red were anaerobically incubated with 140 µM ActVA, so that the flavin was totally complexed to the protein ( Table 1) Oxidation of FMN red in complex with ActVA is also a very slow process when compared to that of FMN red in complex with ActVB. This was shown in the experiment described below. ActVB-FMN ox complex was treated anaerobically with a slight excess of NADH to generate ActVB-FMN red . Oxidation of the flavin was initiated by the addition of aerated water and monitored by UV-visible spectroscopy. In Figure 8 are shown the initial oxidation rates Vi for the O 2 -dependent oxidation of ActVB-FMN red , in the presence of increasing amounts of ActVA. The observation that Vi decreased with increased ActVA was consistent with the notion that FMN red spontaneously moved from ActVB to ActVA and was oxidized more slowly within the latter than within ActVB. As shown in Figure 8, in the presence of one equivalent of ActVA compared to ActVB, the oxidation rate was similar to that of ActVA-FMN red alone. This is in agreement with the much better affinity of FMN red for ActVA than for ActVB. In addition, formation of FMN ox within ActVB occurred with no observable intermediate species absorbing at 380 nm (data not shown). Therefore a slow oxidation process via a FMN-OOH species is specific for ActVA.
The supplementary data Figure 4 shows one of those experiments of Figure 8 suggests that also in that case, FMN red oxidation occurred in ActVA, as a consequence of a fast transfer of FMN red from ActVB to ActVA. Furthermore it shows that the oxidation reaction is not affected by the presence of ActVB. The only noticeable difference with the experiment in absence of ActVB (Figure 7) was that the final spectrum of the product FMN ox exhibited a maximum at 457 nm showing that at the end of the reaction, FMN ox was bound to ActVB.

Interaction of ActVA with ActVB
In order to verify that ActVA and ActVB might not form a complex in solution, the following experiment has been carried out. When an equimolar mixture of His-tagged ActVB and ActVA was loaded onto the Ni-NTA column, ActVA was totally recovered in the runthrough fraction and could not be detected in the imidazole fraction containing ActVB (data not shown). These data suggest that there is no association between ActVA and ActVB. Our investigation of the final step of actinorhodin synthesis, which consists in the conversion of dihydrokalafungin into actinorhodin (Scheme 2) led us to characterize such a two-component FMN-dependent monooxygenase system. We first reported the purification and the characterization of ActVB, the NADH:FMN oxidoreductase associated with the reaction (10). The oxidation reaction itself is not a trivial reaction since it involves two oxidation steps, a dimerization and an aromatic hydroxylation. Not only is the order of the two steps still unknown but it is also not sure whether there are different genes for each step.
The ActVA region of the act cluster carries several genes involved in actinorhodin biosynthesis (26). Because of the general difficulties in purifying the enzymes of that region coupled with the lack of commercial substrates required for biochemical studies, so far only ORF6 was assigned to a defined function, catalysis of the conversion of 6deoxydihydrokalafungin to dihydrokalafungin (34,35). However, we speculated that the ORF5 product, named ActVA here for sake of simplicity, could be the enzyme partner of ActVB for catalyzing one or the two steps of the oxidation of dihydrokalafungin to actinorhodin for the following reasons. First, a DNA fragment containing the actVA-ORF5 region could cause Streptomyces sp. AM 7161, the producer of medermycin, an analog of actinorhodin which lacks the C-8 hydroxyl group, to make mederrhodin A, which is hydroxylated at C-8 (36). Secondly, as shown here, ORF5 displays significant similarities with genes encoding flavin-dependent monooxygenases (Supplementary material Figure 1).
Our results strongly support this hypothesis. With a pure preparation of ActVA, reported here for the first time, we clearly show that the ActVA-ActVB combination can catalyze an hydroxylation reaction, using 1,5-dihydroanthraquinone as a substrate analog. Oxidation absolutely requires NADH, FMN and molecular oxygen, supporting the notion that ActVB serves for the production of reduced flavins whereas ActVA uses the latter to activate oxygen and oxidize the organic substrate. Accordingly, even though it was purified with no chromophore, we have shown that ActVA is able to bind FMN ox , which displays a specific light absorption spectrum with a low-energy transition at 438 nm. ActVA is also able to bind FMN red , to which it provides protection from oxidation by air. With the true substrate, dihydrokalafungin, when it is available, this in vitro system will allow us to conclude whether dimerization and hydroxylation is catalyzed by the same enzyme and, if not, which step is occurring first.
An interesting issue concerns the transfer of reduced flavin from one protein (ActVB) to another (ActVA), its control and whether specific interactions between ActVB and ActVA are required for that purpose.
The K d values for FMN, both in the reduced and the oxidized form, characterizing their interaction with each component, ActVA and ActVB, have been carefully determined. This is the first time that all these values are given with this class of enzymes. They indicate that the flavin transfer between ActVA and ActVB is under a thermodynamic control, with a preference of FMN ox for ActVB and a much stronger affinity of ActVA for FMN red (K d = 0.39 mM). The data showed that FMN ox can be transferred from ActVA to ActVB ( Figure 6) and also that FMN red generated in ActVB by reduction of FMN ox either by NADH or by photoreduction can be removed from ActVB and transferred to ActVA. This was also in good agreement with the observation that the oxidation of ActVB-FMN red by oxygen is greatly slowed down upon addition of increasing amounts of apoActVA, as a consequence of FMN red transfer to ActVA where oxidation occurs slowly ( Figure 8).
All these results can thus be fitted into a very simple mechanism shown in Scheme 3.
At the very beginning of the reaction, FMN ox binds to ActVB where it is reduced by NADH.
The resulting FMN red then diffuses out of ActVB and tightly binds to ActVA where it reacts with oxygen. In this step a reactive intermediate is formed and used to hydroxylate the substrate thus generating FMN ox as will be discussed below. In the absence of the substrate it is likely that the intermediate decomposes to form hydrogen peroxide. Then, in all cases, FMN ox diffuses out of ActVA and ends up into ActVB ready to start a new cycle. In this mechanism, based on a thermodynamic control of the reaction, we do not find it necessary to invoke a specific interaction between the two proteins and a complex channeling mechanism allowing FMN to travel from one protein to another without equilibrating with FMN free in solution. In fact, we did not get any evidence for a complex between ActVB and ActVA. It is actually the case for most two-component flavin-dependent monooxygenase systems that the flavin reductase component and the monooxygenase component do not form a complex.
Furthermore, in most cases, the oxidoreductase component can be replaced by many flavin reductases, including the non-homologous ones from other microorganisms. Therefore direct protein-protein interactions are in general not essential.
In a few cases it has been reported that the monooxygenase component stabilizes C-4a-flavin hydroperoxides in the absence of the substrate (3). One of the characteristics of these species, which derive from the reaction of reduced flavins with molecular oxygen, is their light absorption spectrum with a broad unique band at around 360-390 nm and the absence of the absorption band at 440-450 nm present in the spectrum of oxidized flavins. In the present study during oxidation of ActVA-FMN red or oxidation of ActVB-FMN red in the presence of ActVA, we have observed that the reaction, in the absence of substrate, occurs in two steps. In the first one, a species absorbing at 380 nm is rapidly generated, and in the second one, the latter is slowly converted to FMN ox . We thus assign the 380 nm-absorbing species to a flavin hydroperoxide, FMN-OOH. The accumulation of the observed peroxide and its slow decomposition in ActVA is consistent with the protective effect of ActVA with regard to flavin oxidation by oxygen as discussed above. It is very likely that the flavin hydroperoxide is the oxidizing agent which attacks the substrate within ActVA. Further studies will address the substrate specificity of this interesting enzyme system as well as the chemistry of the reaction between the flavin hydroperoxide and the substrate.         Effect of ActVA on the initial velocity of ActVB:FMN red complex oxidation. FMN ox (16 µM) in the presence of apoActVB (50 µM) was reduced with one equivalent of NADH, in 10 mM Tris-HCl pH 7.6. Then, the mixture was anaerobically incubated with different amounts of ActVA. O 2 (50 µM final concentration) was rapidly added and the initial velocity of FMN red oxidation was calculated from the increase of the absorbance at 457 nm as a function of time.         Determination of the K d value for FMN red with regard to ActVA. ActVA (1 mM) was incubated in anaerobiosis with increasing amounts of FMN red in 50 mM Tris-HCl pH7.6 at 18°C . Sample was excited at 295 nm and the fluorescence spectra were recorded and the fluorescence intensity at 339 nm was used to determine the amount of FMN red bound to ActVA.