A Regulatory Role of NAD Redox Status on Flavin Cofactor Homeostasis in S. cerevisiae Mitochondria

Flavin adenine dinucleotide (FAD) and nicotinamide adenine dinucleotide (NAD) are two redox cofactors of pivotal importance for mitochondrial functionality and cellular redox balance. Despite their relevance, the mechanism by which intramitochondrial NAD(H) and FAD levels are maintained remains quite unclear in Saccharomyces cerevisiae. We investigated here the ability of isolated mitochondria to degrade externally added FAD and NAD (in both its reduced and oxidized forms). A set of kinetic experiments demonstrated that mitochondrial FAD and NAD(H) destroying enzymes are different from each other and from the already characterized NUDIX hydrolases. We studied here, in some detail, FAD pyrophosphatase (EC 3.6.1.18), which is inhibited by NAD+ and NADH according to a noncompetitive inhibition, with Ki values that differ from each other by an order of magnitude. These findings, together with the ability of mitochondrial FAD pyrophosphatase to metabolize endogenous FAD, presumably deriving from mitochondrial holoflavoproteins destined to degradation, allow for proposing a novel possible role of mitochondrial NAD redox status in regulating FAD homeostasis and/or flavoprotein degradation in S. cerevisiae.


Introduction
Flavin adenine dinucleotide (FAD) and nicotinamide adenine dinucleotide (NAD) are two molecules of pivotal importance for mitochondrial functionality, given their role as redox cofactors of a large number of dehydrogenases, reductases, and oxidases mainly involved in energy production and redox homeostasis [1][2][3][4]. Additional emerging regulatory roles are linked to a number of additional cofactor-dependent events, such as protein folding, apoptosis, gene silencing, transcriptional regulation, DNA repairs and calcium-dependent signaling pathways. In many of these processes NAD and FAD are also involved in nonredox reactions (for recent reviews see [2,5,6]). In particular, NAD homeostasis and NAD-dependent modification of target proteins play a crucial role in calorie-restriction-induced lifespan extension and in age-related metabolic diseases [7][8][9].
Consistently, NAD-and FAD-dependent enzymes deficiency and/or impairment in flavin and NAD homeostasis in humans and experimental animals have been linked to several diseases, such as cancer, cardiovascular diseases, anemia, abnormal fetal development, and different neuromuscular and neurological disorders [10][11][12][13][14].
The relevance of such processes merits further research aimed to better describe NAD and FAD homeostasis and flavoenzyme biogenesis, especially in those organisms that can be simple and suitable model for human diseases. The conserved biological processes with all eukaryotic cells, together with the possibility of simple and quick genetic manipulation, allowed for proposing the budding yeast, Saccharomyces cerevisiae, as the premier model to understand the biochemistry and molecular biology of mammalian cells and to decipher molecular mechanisms underlying human diseases [15][16][17].
In yeast and most other organisms, besides de novo synthesis of NAD, the regeneration of NAD from its nicotinamide degradation products has been described in some detail [8,18,19]. This salvage pathway accompanies NAD +dependent signaling processes which, differently from those in which NAD works as redox cofactor, require constant replenishment of cellular NAD pools. NAD + salvage pathway takes place in the nucleus [20]. Differently from mammals [21,22], NAD + is not synthesized in yeast mitochondria; consistently, two mitochondrial carriers (NDT1/2) seem to be responsible for replenishing mitochondrial NAD + level in yeasts [23,24].
Another way to cleave the pyridine nucleotide molecule is at the level of pyrophosphate bond via hydrolytic enzymes. A diphosphatase (pyrophosphatase), performing an enzymatic activity towards NADH, as preferred substrate, and giving NMNH and AMP as products, has been characterized in yeast as belonging to the NUDIX hydrolase family (EC 3.6.1.-) [27]. This enzyme, namely, Npy1p encoded by YGL067W, is able to catalyze NAD + hydrolysis and also very weakly FAD hydrolysis; it was reported to be located in peroxisomes, active at alkaline pH, and strongly inhibited by F − . The existence of a mitochondrial isoenzyme has not been reported so far [28].
The knowledge of FAD cleavage events in yeast is rather poor, as opposed to its biosynthesis. The first investigation on FAD hydrolysis was carried out on cellular extracts obtained by the flavinogenic yeast Pichia guilliermondii [36,37]. Following the demonstration of the existence of a mitochondrial FAD pyrophosphatase (FADppase, EC 3.6.1.18) and FMN phosphohydrolase (EC 3.1.3.2) in rat liver mitochondria [38], further functional evidence of FAD cleaving enzymes has been obtained in S. cerevisiae mitochondria (SCM) [39,40]. The molecular identification of FADppase is still lacking, while a gene encoding for a specific FADppase so far has been cloned and identified in Arabidopsis and named AtNUDX23. It belongs to the NUDIX hydrolase family and is distributed in plastids [41,42]. The possibility that some members of S. cerevisiae NUDIX hydrolase family were able to hydrolyze FAD remains to be investigated, in the frame of characterization of a putative mitochondrial FADppase [40].
Here we studied the ability of SCM to catalyze NAD and FAD hydrolysis via enzymatic activities which are different from the already characterized NUDIX hydrolases. The differential inhibition by the oxidized and reduced form of NAD, together with the ability of mitochondrial FADppase to metabolize endogenous FAD, presumably deriving from mitochondrial holoflavoproteins destined to degradation, allows for proposing a novel possible role of mitochondrial NAD redox status in regulating FAD homeostasis and, possibly, flavoprotein degradation in S. cerevisiae.

Mitochondria Isolation and
Purification. Crude S. cerevisiae mitochondria (SCM) were isolated according to [31]. For pure SCM preparation, the final mitochondrial pellet was resuspended in the isolation medium, consisting of 0.6 M Mannitol and 20 mM HEPES pH 7.4, to obtain 5 mg mitochondrial protein/mL and subsequently purified from extramitochondrial contaminations using a sucrose gradient basically as described in [43]. The intactness of mitochondrial inner membrane was checked measuring the latency of release of the mitochondrial matrix enzyme fumarase (FUM) as in [34,44] by treating SCM with the nonionic detergent Triton X-100 (TX100, 0.1%) at 0 ∘ C for 1 min. The mitochondrial functionality was assessed by oxygen uptake measurements carried out using a Gilson oxygraph as in [31]. The SCM purity was checked by measuring in SCM and spheroplasts (sphero) the activities of different marker enzymes such as cytosolic pyruvate decarboxylase (PDC), vacuolar alkaline phosphatase (AP), peroxisomal D-aminoacid oxidase (D-AAOX), and mitochondrial citrate synthase (CS) via spectrophotometric assays, essentially as described in [31,45]. Mitochondrial protein concentration was determined according to [46].

FAD Hydrolysis.
FAD externally added or endogenous FAD metabolism in S. cerevisiae was monitored by means of fluorimetric and HPLC measurements, essentially as in [38,47]. In the case of fluorimetric measurements, flavin derivative emission spectra (excitation wavelength at 450 nm) and time drive measurements (excitation and emission wavelengths at 450 nm and 520 nm, resp.) were carried out at 25 ∘ C in 2 mL of a standard medium consisting of 0.6 M Mannitol, 50 mM TRIS-HCl, pH 7.5, and 5 mM MgCl 2 by means of a LS50B Perkin Elmer spectrofluorimeter. Flavin fluorescence emission spectra were corrected as in [48] by adding a few crystals of sodium dithionite to the mitochondrial suspension. When externally added FAD hydrolysis was measured, the endogenous flavin fluorescence was not considered since it was found to be negligible compared to that measured in subcellular suspension.
In each experiment, FAD, FMN, and Rf fluorescence was calibrated individually using standard solutions whose concentrations were calculated by using 450 of 12.2 mM −1 ⋅ cm −1 for FMN and Rf, and 11.3 mM −1 ⋅ cm −1 for FAD. Under the experimental conditions used here, FAD fluorescence constant ( FAD ) proved to be about ten times lower than those of FMN and Rf ( FMN/Rf ) which did not differ significantly from each other [33,49]. Thus, the rate of FAD hydrolysis, that is, the rate of FMN + Rf formation, expressed as nmol FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein, was calculated from the rate of fluorescence increase, measured as tangent to the initial part of the experimental curve by applying the following equation: where Δ is expressed in fluorescence arbitrary units, is expressed in mL, Δ = FMN/RF − FAD is expressed as M −1 , Δ is expressed in min, and is the mass of protein in mg.
In the case of HPLC measurements, from aliquots of the subcellular suspension (100 L), taken at various times, perchloric extracts were obtained and neutralized with KOH [33]. FAD, FMN, and Rf were analyzed by means of HPLC, as in [31,47]. FAD, FMN, and Rf were measured in each experiment, with a quantitative determination carried out by measuring peak areas by means of a calibration curve. It should be noted that, in externally added FAD experiments, the contribution of endogenous flavins to the measured chromatographic peak area was proved to be negligible with respect to that due to flavin derivatives found in the subcellular suspension. Since the rate of FMN/Rf appearance measured fluorimetrically via HPLC showed no significant difference, these experimental approaches were used in this work indifferently.

NADH Oxidation and
Hydrolysis. NADH metabolism in S. cerevisiae was monitored by means of fluorimetric measurements, using a LS50B Perkin Elmer spectrofluorimeter. NADH excitation spectra (emission wavelength at 456 nm) and time drive measurements (excitation and emission wavelength pairs at 260/456 and 340/456 nm, resp.) were carried out at 25 ∘ C in 2 mL of a standard medium. NADH oxidation to give the nonfluorescence NAD + can be revealed (when it is the sole process responsible for NADH disappearance) as a fluorescence decrease at 260/456 nm and 340/456 nm as tangent to the linear part of the experimental curve as where Δ ecc/em is expressed in fluorescence arbitrary units, is expressed in mL, NADH260 and NADH340 are expressed as M −1 , and Δ is expressed in min. Differently from oxidation, NADH hydrolysis results in products which are not fluorescent at 260/456 nm but still fluorescent at 340/456 nm (with a constant value 1.25fold lower). Therefore, when hydrolysis is the sole process responsible for NADH disappearance, it can be revealed as a fluorescence decrease at 260/456 nm and 340/456 nm as tangent to the linear part of the experimental curve as with Δ 260/456 /Δ higher than Δ 340/456 /Δ . Rearranging (2) and (3), it is possible to calculate the two components of the NADH disappearance in a complex system, where oxidation is significantly reduced with respect to hydrolysis (i.e., in the presence of KCN 1 mM).
The rate of NADH hydrolysis in the presence of residual NADH oxidation, expressed as nmol NADH hydrolyzed ⋅ min −1 ⋅ mg −1 protein, can be calculated by applying the following equation: with Δ 260/456 /Δ higher or equal to Δ 340/456 /Δ . The rate of NADH oxidation in the mixed system can be calculated as In each experiment, the NADH260 and NADH340 values were determined by calibrating NADH fluorescence with standard solutions, whose concentrations were spectrophotometrically defined ( 340 was 6.2 mM −1 ⋅ cm −1 ). NMNH fluorescence was calibrated using a standard curve produced by incubating NADH (at concentrations ranging from 1 to 10 M) with excess amounts of commercial nucleotide pyrophosphatase (EC 3.6.1.9) to determine the Δ 340 value.

4
Oxidative Medicine and Cellular Longevity 2.6. NAD + Hydrolysis. NAD + hydrolysis was measured by fluorimetrically monitoring the hydrolysis of nicotinamide 1,N 6 -ethenoadenine dinucleotide ( -NAD + ), essentially as described in [50,51]. Fluorescence measurements were performed using a Perkin Elmer LS5 spectrofluorimeter (excitation and emission wavelengths set at 310 nm and 410 nm) at 25 ∘ C in 2 mL of a standard medium. The fluorescence changes produced by subcellular suspensions were calibrated by using a standard curve produced by incubating -NAD + (at concentrations ranging from 1 to 6 M) with excess amounts of commercial NADase [50]. The concentration of -NAD + was determined by the conversion of -NAD + to -NADH using the commercial alcohol dehydrogenase reaction and assuming a molar extinction coefficient for -NADH of 6.2 mM −1 ⋅ cm −1 at 340 nm.

Kinetic Data Analysis. Data fitting was performed according to the Michaelis-Menten equation:
To fit the experimental data and to obtain estimates of the kinetic parameters use was made of the GraFit software (v. 3.00, 1992, by R. J. Leatherbarrow, Erithacus Software).

Statistical Analysis.
All experiments were repeated at least 3 times with different mitochondrial preparations.
Results are presented as mean ± standard deviation. Statistical significance was evaluated by Student's -test. Values of < 0.05 were considered statistically significant.

Results and Discussion
In order to study NAD(H) mitochondrial metabolism we isolated functionally active SCM, as described before [31,32], and further purified them, basically as in [43]. To control that SCM were significantly depleted by extramitochondrial contaminations; in Figure 1(a) we measured the distribution of the cytosolic marker enzyme (pyruvate decarboxylase, PDC), the vacuolar marker enzyme (alkaline phosphatase, AP), and the peroxisomal marker enzyme (D-aminoacid oxidase, DAAOX), with their specific activities being about tenfold (for the cytosolic and vacuolar markers) and fivefold (for the peroxisomal marker) lower in the mitochondrial fraction than those in spheroplasts (sphero). The purity of mitochondrial preparations was also assessed, by following the enrichment of the mitochondrial matrix marker enzymes citrate synthase (CS) in Figure 1(a) and fumarase (FUM) in Figure 1(b) whose specific activities were about sixfold enriched in the mitochondrial fraction. The intactness of mitochondrial suspension was also checked by following the latency of release of the mitochondrial matrix enzyme FUM, following the rupture of the inner mitochondrial membrane by Triton X-100 (TX100). An integrity of 81 ± 3% was measured in three different mitochondrial preparations ( Figure 1(b)), as described in Section 2. Then, in the frame of assessing mitochondrial functionality, NADH (1 mM) was externally added to purified SCM, to induce respiration, via the two external NADH dehydrogenases (namely, Nde1p and Nde2p), as schematized in Figure 1(c  [52]. To better investigate the fate of externally added NADH, SCM were ruptured by freezing-thawing cycles and added to NADH (5 M) in the absence and in the presence of KCN (1 mM) (Figure 2(a)). Fluorescence excitation spectra of the suspension were registered for 10 min, with the emission wavelength set at 456 nm. As expected, two major emission fluorescence peaks are found (at 260 and 340 nm, resp.) due to the reduced form of the coenzyme, with the corresponding fluorescence intensities decreasing concomitantly in the course of NADH oxidation at both 260 nm and 340 nm.
From the values of fluorescence decrease, a NADH oxidation rate of 520 nmol ⋅ min −1 ⋅ mg −1 protein was calculated as reported in Section 2 (data not shown). KCN (1 mM) addition to ruptured SCM reduced the rate of NADH disappearance to less than 8% of that measured in its absence (Figure 2(a)). Under this experimental conditions, the rate of fluorescence decrease at wavelength pairs 340/456 nm (essentially due to the KCN-insensitive NADH oxidation rate) became lower than that measured at 260/456 nm.
A possible explanation for this asymmetry derived from the hypothesis that oxidation is not the sole process responsible for NADH disappearance in the suspension, with a possible contribution due to putative hydrolytic processes. In this case a NADH hydrolysis rate equal to 14 nmol NADH hydrolyzed ⋅ min −1 ⋅ mg −1 protein was calculated by applying (4) reported in Section 2, while NADH oxidation rate was reduced at 48 nmol ⋅ min −1 ⋅ mg −1 protein. In three independent experiments using different mitochondrial preparations, a mean rate of 14.6 ± 1.1 nmol NADH hydrolyzed ⋅ min −1 ⋅ mg −1 protein was calculated (with NADH oxidation rate of 34 ± 12 nmol ⋅ min −1 ⋅ mg −1 protein). Similar results were obtained when the fluorescence decrease was continuously measured at both the wavelength pairs 260/456 nm and 340/456 nm, respectively. Typical traces are reported in Figure 2(b) (straight lines).
The specific NADH hydrolytic activity measured in the mitochondrial fraction resulted almost equal to that found in whole sphero (14.1 ± 1.6 nmol ⋅ min −1 ⋅ mg −1 protein); this allows us to propose a mitochondrial localization for NADH destroying enzyme, even if not exclusive. The already characterized hydrolase which is known to cleave NADH, namely, Npy1p (see Section 1), seemed to be peroxisomal [28].
Oxidative Medicine and Cellular Longevity   With the aim of identifing the NADH destroying enzyme, we queried a number of protein sorting signal prediction programs (i.e., iPSORT, TargetP, and MITOPROT), thus finding a possible mitochondrial localization for this protein.
To understand whether the mitochondrial NADH hydrolytic activity was due to the NUDIX hydrolase Npy1p, the effects of both NaF (1 mM) and CaCl 2 (1 mM), powerful inhibitors of the peroxisomal enzyme, were tested (Figure 2(b)). Since they clearly did not inhibit the NADH hydrolysis rate (dotted lines), we exclude that the activity we revealed was due to a putative mitochondrial isoform of this enzyme. Independently from an auspicial protein identification, these experiments represent the first evidence in favor of the existence of a novel mitochondrial NAD(H) destroying activity. Nevertheless, the rapid fluorimetric method used prevents a detailed kinetic characterization. A simpler investigation of NAD-destroying activities makes use of a NAD + fluorescent analog, namely, nicotinamide 1,N 6 -ethenoadenine dinucleotide ( -NAD + ) [50,51]. This assay was specifically useful to investigate whether purified SCM can hydrolyze also the oxidized form of the pyridine coenzyme. In Figure 3(a) a typical -NAD + (50 M) emission spectrum is reported, as revealed at pH 7.5, in the 340-500 nm range, with excitation wavelength set a 310 nm. As expected, it reveals a characteristic peak at 410 nm, whose intensity linearly depends on -NAD + concentration. Addition of SCM (previously solubilized with TX100) induced an increase in fluorescence emission at 410 nm, as a linear function of time up to 20 min. Since the -adenine ring specific fluorescence constant is about ten-fold higher than that of the -NAD + , the observed fluorescence increase strongly suggests that -NAD + was cleaved with a rate equal to 0.23 nmol -NAD + cleaved ⋅ min −1 ⋅ mg −1 protein. In Figure 3(b) the fluorescence changes of -NAD + (50 M) suspensions, following the addition of solubilized SCM, were continuously measured, at the fixed excitation and emission wavelengths (set at 310 nm and 410 nm, resp.), corresponding to an initial rate of hydrolysis equal to 0.25 nmol ⋅ min −1 ⋅ mg −1 protein (+TX100). No increase was observed when freshly isolated intact SCM were used (−TX100), thus suggesting an intramitochondrial localization for NAD + -cleaving activity.
The dependence of the rate of cleavage of -NAD + catalyzed by solubilized SCM was determined in the pH range 4.5-10, using 100 mM acetate/acetic acid and 50 mM Tris/HCl buffering mixtures (Figure 4(a)). -NAD + cleavage was found to occur with a bell-shaped profile with the maximum rate measured at pH 5.5.
To gain some insights into the substrate affinity, the dependence of the rate of -NAD + was studied at pH 6 as a function of the substrate concentration (Figure 4(b)). Hyperbolic characteristic was found, with saturation kinetics analyzed by the Michaelis-Menten equation, using the GraFit software (see Section 2). Thus, and max values equal to 56 ± 8.9 M and 1.7 ± 0.1 nmol ⋅ min −1 ⋅ mg −1 of mitochondrial protein, respectively, were calculated from the data in Figure 4(b). In three independent experiments using different mitochondrial preparations, a mean rate of 1.6 ± 0.2 nmol -NAD + hydrolyzed ⋅ min −1 ⋅ mg −1 protein was found, using -NAD + (200 M).
The specific -NAD + hydrolytic activity was also measured in solubilized sphero, under the same experimental conditions. As reported in Figure 4(c), the specific -NAD + hydrolytic activity in sphero is similar to that measured in SCM. This finding, as previously observed for NADH hydrolytic activity, suggests a preferential, but not exclusive, mitochondrial localization for this enzyme.
As previously observed for NADH hydrolyzing activity, the mitochondrial -NAD + hydrolytic activity was not inhibited by both NaF (1 mM) and CaCl 2 (1 mM) (Figure 4(d)), thus again excluding that a NUDIX hydrolase was responsible for this activity. Conversely, the mitochondrial NAD +cleavage activity was totally inhibited by AMP (1 mM) but also significantly inhibited by nicotinamide, with an IC 50 of about 5 mM (Figure 4(d)). This last experiment strongly supports the proposal that a nicotinamide-sensitive pyridine nucleotide-destroying enzyme is enriched in SCM, even if, unfortunately, it did not allow to precisely identify the product of mitochondrial NAD(H) hydrolysis. Nevertheless, the strong inhibition by AMP allows us to postulate that the dinucleotide was broken at the level of pyrophosphate bond, as already demonstrated for certain animal and plants FAD pyrophosphatases (FADppase, EC 3.6.1.18) [54,55].
Particular attention merits the finding that NAD + degradation, as well as FAD degradation (see below), strictly depends on pH, which in yeast is highly dynamic [56]; thus, we expect that local changes in intramitochondrial pH may have profound influence on the availability of intramitochondrial pyridine and flavin cofactors. Under respiratory conditions, an intramitochondrial pH value ranging from 7.7 to 7.3 during the different growth phases was experimentally evaluated [57], presumably accompanied by a very low cofactor degradation, as shown by measurements performed here. Thus, under physiological conditions, respiratory-chaindriven proton pumping prevents cofactor hydrolysis by generating a transmembrane pH gradient, negative inside.
Mild uncoupling [58] or various stress conditions [57] induce a severe decrease in intramitochondrial pH value, which plays a central role, together with vacuoles, in controlling cytosolic pH [56,59,60]. Under these conditions we expected an increase in the rate of NAD + , as well as FAD degradation (see below), which in turn may regulate redox homeostasis, cell growth, and defense responses [61,62].
Since the contribution of mitochondrial NAD(H) degrading enzyme activity on the economy of NAD(H) homeostasis and, in turn, on cell longevity and stress resistance is still unexplored, our future goal will be to identify acidic mitochondrial NAD pyrophosphatase in yeast.
As far as FAD degradation is concerned, the occurrence of AMP-inhibited FAD hydrolysis in isolated intact mitochondria was for the first time demonstrated in rat liver [38] and, more recently, in S. cerevisiae [40]. In order to understand whether FADppase, described in [40], could also be responsible for the mitochondrial NAD(H) hydrolyzing activity observed in our mitochondrial preparation, FADppase was characterized in some detail. Thus, FAD (3 M) was added to solubilized SCM and flavin fluorescence emission spectra (with excitation wavelength set at 450 nm) registered at pH 7.5, as a function of time. As expected, an increase in the 500-600 nm wavelength range with a peak at 520 nm was observed in Figure 5(a), due to the conversion of FAD into FMN/Rf, with a rate of fluorescence increase, measured as described in Section 2, corresponding to 0.28 nmol FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein.
In Figure 5(b), FAD (3 M) fluorescence was continuously measured at the fixed excitation and emission wavelengths (set at 450 nm and 520 nm, resp.). No fluorescence increase was observed after addition of intact and coupled SCM (−TX100). When FAD was added to solubilized SCM an increase was observed with a rate equal to 0.30 nmol FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein (+TX100). In seven experiments performed with different mitochondrial preparations, a mean rate of 0.32 ± 0.05 nmol FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein was measured. FAD hydrolysis could be revealed only when SCM were solubilized by TX100 and in this aspect SCM are different from rat liver mitochondria. In order to characterize the products of FAD hydrolysis catalyzed by solubilized mitochondria, under the same experimental conditions described in Figure 5(a), extracts of the suspension were taken at different incubation times and analyzed via HPLC, as described in Section 2, with measurements made of FAD, FMN, and Rf amount in the suspension.
Solubilized SCM added with FAD (3 M) were incubated either in the absence (panel (A)) or in the presence of AMP (panel (B)). In a parallel assay the effect of NaPPi (panel (C)), which was proved to prevent FMN hydrolysis in SCM [31], was also tested.
As shown in Figure 5(c) (panel (A)), FAD amount was found to decrease with a constant initial rate of 0.29 nmol FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein, in agreement with the fluorimetric measurements, and then it reached a plateau. Concomitantly with FAD disappearance, a significant increase in FMN amount was initially observed with a constant rate of 0.30 nmol ⋅ min −1 ⋅ mg −1 protein in the first 10 min, in a fairly good agreement with a stoichiometric ratio of 1 : 1 with the FAD amount decrease (see inset). At longer incubation time, Rf, that was not initially detectable, appeared in the suspension (see inset) and it increased linearly up to 60 min (with a rate of 0.10 ⋅ min −1 ⋅ mg −1 protein), thus prevailing over FMN (panel (A)) which remained relatively constant (at about 0.06 M). Since FAD hydrolysis rate does not depend on the rate of FMN into Rf conversion in the first 10 min (that is, the step catalyzed by FADppase is the limiting step of mitochondrial FAD degradation pathway), kinetics measurements of FADppase reaction were performed consistently by fluorimetric method.
As expected, the rate of FAD hydrolysis was totally inhibited by AMP (1 mM); neither FMN nor Rf was found to appear during the incubation time (panel (B)). Conversely, in the presence of NaPPi (1 mM), the rate of FAD hydrolysis resulted unchanged, with a notable increase in FMN (with an initial rate of 0.30 nmol ⋅ min −1 ⋅ mg −1 proteins), which was slowly accompanied by Rf appearance with an 85% reduced rate, in comparison to the rate measured in the absence of NaPPi (panel (C)).
When the mitochondrial FAD (3 M) hydrolysis was studied at pH 6 (i.e., the optimum of -NAD + hydrolytic activity), FAD disappearance was found to occur with a 2.5fold higher rate and to be still in a good agreement with a stoichiometric ratio of 1 : 1 with the appearance of FMN, that is the preeminent species at this pH value, whereas the rate of Rf appearance is extremely low (0.02 nmol ⋅ min −1 ⋅ mg −1 mitochondrial proteins).
Consistently, the rate of FAD cleavage catalyzed by solubilized SCM, fluorimetrically determined in the pH range 4.0-10 in Figure 6(a), was found to occur with a bell-shaped profile with the maximum rate measured at pH 6 (being equal to 1.05 nmol ⋅ min −1 ⋅ mg −1 protein, when FAD 10 M was used as substrate). This pH profile is quite different from that described in [40], whose optimum activity was at alkaline pH. Thus, the dependence of FAD hydrolysis rate on substrate concentration was studied at pH 6 ( Figure 6(b)). Hyperbolic characteristic was found with saturation kinetic, analyzed by the Michaelis-Menten equation, using the GraFit software (see Section 2). Thus, and max values equal to 5.9 ± 0.5 M and 1.6 ± 0.1 nmol ⋅ min −1 ⋅ mg −1 of mitochondrial protein, respectively, were calculated from the data in Figure 6(b). A value equal to 10 ± 2 M was determined at pH 7.5 (data not shown).
These results confirm that in SCM, a FADppase exists and it is able to catalyze the cleavage of the diphosphate bound in dinucleotides. This enzyme is inhibited by AMP and stimulated by acidic pH value. A putative mitochondrial FMN phosphohydrolase activity is involved in FMN dephosphorylation; it is specifically inhibited by NaPPi and its activity is significantly reduced at acidic pH value.
As regards inhibition of FAD hydrolysis, the inorganic compounds NaF (1 mM) and CaCl 2 (1 mM), differently from AMP (1 mM), did not reduce FAD hydrolytic activity in SCM (Figure 6(c)), as already observed for -NAD + hydrolysis, thus excluding that FMN is produced by a putative mitochondrial NUDIX hydrolase.
Quite interestingly, differently from -NAD + hydrolysis, nicotinamide (as well as NMN) did not affect FAD hydrolysis rate. Therefore, we may conclude that mitochondrial NAD(H) and FAD destroying activities are different. Another difference between mitochondrial NAD(H) and FAD destroying activities resides in the observation that specific FAD hydrolytic activity catalyzed by solubilized SCM corresponded to about only 27% of that measured in the sphero ( Figure 6(d)). Taking into account the extramitochondrial contaminations present in our mitochondrial preparations (see Figure 1(a)), it represents about 10% of the total activity detectable in the sphero. A similar distribution ratio was also found for other mitochondrial enzymes, that is, RFK and FADS [29,31]. Presumably, other quite active FADppases are localized in the extramitochondrial compartments.
In a first attempt to determine the physiological role for the intramitochondrial FADppase, the possibility that this enzyme can play a major role in the intramitochondrial FAD degradation was studied by both fluorescence and HPLC experiments. A typical fluorimetric experiment is reported in Figure 7(a). At initial time, intact mitochondria showed a typical endogenous flavin emission spectrum with a peak at 520 nm (excitation wavelength set at 450 nm) (−TX100, dotted line). No increase in flavin fluorescence was observed when coupled and intact SCM were used. When SCM were solubilized with TX100 an increase in the 500-600 nm wavelength range was observed (−NAD + ). This strongly suggests that endogenous FAD has been hydrolyzed to FMN/Rf. Endogenous flavin fluorescence was found to increase as a linear function of time up to 30 min with a rate equal to 7 pmol endogenous FAD hydrolyzed ⋅ min −1 ⋅ mg −1 protein (see inset). It should be noted that under our experimental conditions less that 20% of total mitochondrial FAD amount was hydrolyzed, presumably due to an inhibition by FAD hydrolysis product(s) or because the remaining FAD was bound to holoflavoprotein and thus protected by hydrolytic activity.
Since FADppase is located inside mitochondria the stimulatory effect of the detergent could not be easy explained, unless some internal metabolite, whose concentration is high in the matrix of intact organelles and diluted following membrane rupture, is able to inhibit FADppase. In this hypothesis we tested a possible inhibitory effect by NAD + (1 mM), which was found to completely inhibit endogenous FAD to FMN conversion (+NAD + ).
In the same experiment, the hydrolysis of endogenous FAD was confirmed by HPLC measurements of FAD, FMN, and Rf amounts before and after TX100 addition (Figure 7(b)). In the absence of TX100 the mitochondrial amounts of FAD and FMN were about 172 and 26 pmol ⋅ mg −1 protein, respectively, and no Rf was detected (panel (A)). As a result of TX100 addition, FAD and FMN were found to decrease, with significant appearance of Rf (panel (B)). Rf content was found to increase up to 20 min in a fairly good stoichiometric ratio (1 : 1) with FAD decrease (data not shown). The rate of FAD hydrolysis (i.e., Rf appearance) was equal to 7.2 pmol ⋅ min −1 ⋅ mg −1 protein, in good agreement with fluorescence measurements. Rf appearance was strongly prevented by NAD + (panel (C)). The effect of NAD + was confirmed on exogenous FAD hydrolysis by HPLC (data not shown) and the nature of the inhibition studied fluorimetrically (Figure 8(a)). FAD hydrolysis sensitivity to NAD + was consistent with a noncompetitive inhibition, as shown in the Dixon plot with a value equal to 10 M, calculated from the data in Figure 8(a). Adenosine and derived compounds (i.e., ADP-ribose, Ap 5 A, and CoA) and guanosine nucleotides, but not nicotinamide and NMN, are able to reduce FAD hydrolysis rate (data not shown); therefore we feel that the purine moiety is relevant for inhibitor recognition by FADppase. Consistently, NADH is also able to inhibit in a noncompetitive way the rate of FAD hydrolysis, but, interestingly, the differed for an order magnitude, being equal to 100 M for NADH, as calculated from the data reported in Figure 8(b).
These results demonstrate that FAD hydrolyzing enzyme could discriminate between the redox status of pyridine nucleotides, as depicted in Figure 9. It is well known that a high electron pressure, namely, an increase in NADH/NAD + ratio, is a condition favoring ROS generation which can be contrasted by an efficient respiratory chain (which is flavin dependent) and by defense systems (some of which are flavoenzymes) [6]. According to the measurements performed here, reduction of NAD + to NADH is expected to favor hydrolysis of free FAD, since the oxidized form of the pyridine cofactor is more powerful as inhibitor (lower ) than the reduced form (higher ), at least in ruptured SCM. Thus, we can speculate that reducing conditions are accompanied by a more rapid local "recycling" of FAD, presumably deriving from holoflavoprotein turnover. This situation could allow a more rapid intramitochondrial assembly of novel flavoproteins (mainly components of the respiratory chain, like SDH [32,63]) and saving of cytosolic Rf level which is necessary for enzymatic ROS defense.
Whether or not flavin-coenzymes homeostasis is modulated in vivo by pyridine nucleotide redox status inside SCM as postulated and summarized in the scheme in Figure 9, is an interesting interrogative that we would like to address in the next future.

Conclusions
Here we proved the ability of SCM to catalyze NAD(H) and FAD hydrolysis, via enzymatic activities which are different from the already characterized NUDIX hydrolases.
The differential inhibition by the oxidized and reduced form of NAD toward the mitochondrial FADppase, together with the ability of mitochondrial FADppase to metabolize endogenous FAD, presumably deriving from mitochondrial holoflavoproteins destined to degradation, allows for proposing a novel possible role of mitochondrial NAD redox status in regulating FAD homeostasis and, possibly, flavoprotein degradation in S. cerevisiae.