Reversible dissociation of flavin mononucleotide from the mammalian membrane-bound NADH:ubiquinone oxidoreductase (complex I)

Abstract

Conditions for the reversible dissociation of flavin mononucleotide (FMN) from the membrane-bound mitochondrial NADH:ubiquinone oxidoreductase (complex I) are described. The catalytic activities of the enzyme, i.e. rotenone-insensitive NADH:hexaammineruthenium III reductase and rotenone-sensitive NADH:quinone reductase decline when bovine heart submitochondrial particles are incubated with NADH in the presence of rotenone or cyanide at alkaline pH. FMN protects and fully restores the NADH-induced inactivation whereas riboflavin and flavin adenine dinucleotide do not. The data show that the reduction of complex I significantly weakens the binding of FMN to protein thus resulting in its dissociation when the concentration of holoenzyme is comparable with K_d (~10⁻⁸ M at pH 10.0).

1. Introduction

Mitochondrial NADH:ubiquinone oxidoreductase (complex I) is a key enzyme of bioenergetic metabolism. It catalyzes the first step of intramitochondrial NADH oxidation coupled with vectorial translocation of four protons across the inner membrane. The structure, molecular biology and catalytic properties of mitochondrial complex I and its prokaryotic homologues (NDH-1) have been comprehensively reviewed [1–5]. Mammalian complex I is comprised of 45 different subunits [6] and harbors distinct redox centers including FMN [7], iron-sulfur clusters [8] and bound ubiquinone species [9]. Convincing evidence has established that the 51 kDa (Nqo1) subunit contains the NADH-binding site and the FMN and iron-sulfur N3 center [10–12]. The recently solved structure of the Thermus thermophilus NDH-1 hydrophilic fragment at 3.3 Å shows that the FMN is in a solvent accessible cleft of the Nqo1 subunit located ~90Å from the coupling membrane plane [12]. The small water soluble fragment of complex I, (termed FP) containing the 51 kDa subunit was also shown to bind photoactivated NADH analogues [13] and is capable of NADH oxidation by artificial electron acceptors [14, 15]. These and recent structural data [12] strongly suggest that FMN serves as the primary oxidant of NADH during steady-state respiration. The redox properties of the FMN/FMNH₂ couple in isolated complex I [16] and in the FP fragment [17] have been characterized by low temperature EPR and protein film voltammetry, respectively.

When diluted samples of FP are reduced by NADH or dithionite their catalytic activity gradually declines due to dissociation of FMN, a phenomenon operationally called “reductive inactivation” [18]. The NADH-induced loss of enzymatic activity was shown to be irreversible although it was prevented by either NAD⁺ or FMN [18]. Recent data on redox-dependent change of affinity of pyridine nucleotides for the active site of complex I has been interpreted as evidence consistent with the FMN oxidoreduction-dependent conformational change of the enzyme [19]. Thus, we were prompted to apply previously developed approach [18] to study possible “reductive inactivation” of the native membrane-bound complex I. Here we describe experiments showing the reversible dissociation of FMN from membrane-bound complex I.

2. Materials and methods

Bovine heart submitochondrial particles (SMP) were prepared as described [20] and complex I was activated by aerobic incubation with 1 mM NADPH [9]. Activated SMP were precipitated by centrifugation, suspended in 0.25 M sucrose, 50 mM Tris-Cl (pH 8.0), 0.2 mM EDTA and stored in liquid nitrogen.

NADH:Q₁ or NADH:hexaammineruthenium (III) (HAR) reductase activities were assayed at 30°C photometrically as the decrease in NADH concentration ( ${\epsilon }_{\text{mM}}^{340}$ = 6.22 mM⁻¹cm⁻¹ or ${\epsilon }_{\text{mM}}^{380}$ = 1.25 mM⁻¹cm⁻¹) in a standard mixture comprised of 0.25 M sucrose, 25 mM Tris/25 mM CAPS, and 0.2 mM EDTA in the presence of the inhibitors (cyanide or rotenone) added to prevent the NADH oxidase activity. The concentrations of electron acceptors, pH and other additions to the assay mixture are indicated in the Legends to the figures and footnote to the Table. The protein content was determined by the biuret assay using bovine serum albumin as the standard. All fine chemicals were from Sigma.

3. Results

When SMP were preincubated for 1 h at 20°C over a wide range of pH and then assayed their NADH:HAR reductase activity remained quite stable within the pH range of 6.0–10.0 (Fig. 1). Moreover, in contrast to the rotenone-sensitive NADH oxidase which shows only slight pH-dependency [21], NADH:HAR reductase activity was strongly pH-dependent as if deprotonation of a group with a pK_a of about 8.5 is required for catalysis. If NADH (and rotenone) were present in the preincubation medium the catalytic activity initiated by further addition of HAR was only slightly decreased after 1 h preincubation at pH 6.0 and 7.5 whereas, a significant decrease of NADH oxidation was evident at more alkaline pH.

Fig. 1.

NADH-induced inactivation of NADH:HAR reductase activity. Data points and the continuous line show pH-dependence of the activity measured as initial rate of 0.6 mM NADH oxidation by SMP at 20°C in the standard reaction mixture supplemented with rotenone (5 μM) and HAR (0.5 mM). Black and grey bars show the activities measured before and after 1 h preincubation of SMP in the assay cuvette (10 μg of protein per ml), respectively, at pH values indicated in brackets. Open bars: 1 mM NADH was added to the preincubation mixture and the reaction was initiated by the addition of 0.5 mM HAR.

The pH-dependence of NADH:HAR reductase activity was not elaborated in the present study, however, the inhibitory effect of preincubation with NADH at alkaline pH was further investigated. Fig. 2 demonstrates a decline of the NADH:HAR reductase activity during preincubation of SMP in the presence of NADH and rotenone added to prevent respiration. The activity slowly decreased in a first-order decay process (k = 0.034 min⁻¹ at 20°C, pH 10.0). The rate of inactivation was dependent on NADH concentration (apparent K_i was about 10 μM, data not shown). The addition of 1 mM NAD⁺ to the preincubation medium prevented the NADH-induced inactivation thus showing that the reduction of some component poised in redox equilibrium with NADH/NAD⁺ couple was responsible for the inactivation. More importantly, added FMN not only protected but reactivated the enzyme after the NADH-induced inactivation has been completed. Neither FAD, nor riboflavin were able to protect and/or reactivate the enzyme. The data shown in Fig. 2 suggest that dissociation of FMN was the reason for the NADH-induced inactivation.

Fig. 2.

Time course of the NADH-induced inactivation. SMP (10 μg of protein per ml) were preincubated at 20°C in the standard assay mixture (pH 10.0) supplemented with rotenone (5 μM) and the rates of NADH oxidation initiated by the addition of 0.5 mM HAR and NADH (100 μM) were measured. Other additions to the preincubation medium were: none (●); 100 μM NADH (○) or 100 μM NADH and: 1 mM NAD⁺ (■), 10 μM FMN (□), 10 μM riboflavin (▲) or 10 μM FAD (○). Ten μM FMN was added where indicated after the NADH-induced inactivation had been almost completed. One arbitrary unit corresponds to the specific activity of 2.2 ± 0.2 μmol/min per mg of protein.

When a tightly-bound ligand (such as FMNH₂) reversibly dissociates:

$$\text{E}\cdot {\text{FMNH}}_2\stackrel{{\text{K}}_{\text{d}}}{\rightleftarrows }\text{E}+{\text{FMNH}}_2,$$ (1)

the concentration of active holoenzyme in equilibrium is decreased upon dilution (Ostwald’s dilution law) according to Equation (2):

$$[\text{E}\cdot {\text{FMNH}}_2]={[\text{E}]}_{\text{t}}+\frac{{\text{K}}_{\text{d}}-\sqrt{4{[\text{E}]}_{\text{t}}\cdot {\text{K}}_{\text{d}}+{\text{K}}_{\text{d}}^2}}{2},$$ (2)

where [E]_t stands for the sum of holo- and apo-enzyme concentrations and K_d is the dissociation constant in equilibrium (1). Eq. (2) predicts that the residual enzymatic activity after the NADH-induced inactivation has been completed under certain conditions should be dependent on protein concentration ([E]_t). Fig. 3 shows that this was the case and K_d values of 0.4 and 10.0 nM could be approximated for pH 9.0 and 10.0, respectively, by fitting the experimental data to the theoretical function described by Eq. (2). It should be noted that Eqs. (1) and (2) describe the process as the first approximation only: the dissociated free FMNH₂ is presumably oxidized by oxygen and oxidized FMN may rebind to the apoprotein thus resulting in very slow cycling.

Fig. 3.

Dependence of the NADH-induced inactivation on the protein concentration. Different amounts of SMP were preincubated in the presence of rotenone (5 μM) and NADH (100 μM) for 1 h at pH 9.0 (●) or 10.0 (○) and their normalized residual activity initiated by the addition of 0.5 mM HAR was measured. The concentration of complex I shown on abscissa was calculated assuming its content in our preparation as 0.1 nmol per mg of protein [22]. The continuous lines are computer generated dependencies of the residual activity on concentration of complex I according to Eq. (1) with the K_D values of 0.4 nM and 10.0 nM at pH 9.0 and 10.0, respectively. The averaged data of 2 experiments are shown.

The results described above are in qualitative agreement with the data and interpretation previously reported for FP [18]. It was of obvious interest to find out what is the effect of reductive dissociation of FMN on the “complete” rotenone-sensitive NADH:quinone reductase activity. The data presented in Table 1 show that preincubation of SMP at alkaline pH in the presence of NADH under conditions where respiration was prevented resulted in loss of the NADH:Q₁ reductase activity. Most importantly, almost full protection and complete restoration of the rotenone-sensitive activity by FMN was evident. Substantial loss (about 45%) of the activity was observed after prolonged incubation of SMP at 20°C, pH 10.0 in the absence of NADH. This was most likely due to the enzyme de-activation phenomenon (see Refs. [3, 21] for the details). The rate of turnover-dependent activation in the assay was shown to be very slow at alkaline pH [21]. No special attempts to reveal full catalytic capacity of the enzyme after 1 h preincubation at pH 10.0 without NADH were made. Thus, we conclude that reversible dissociation of flavin from its binding site in complex I upon its reduction at alkaline pH is the reason for the loss of activity.

Table 1.

Protection and restoration of the rotenone-sensitive NADH:quinone reductase activity by FMN

Samplea	Activityμmol/min per mg of protein
1. No preincubation	0.35
2. Preincubated, no NADH	0.19 (100%)
3. As (2) + NADH	0.03 (16%)
4. As (3) + FMN	0.16 (84%)
5. As (2), FMN added after preincubation	0.14 (74%)
6. As (2), Riboflavin added after preincubation	0.02 (10%)
7. As (2), FAD added after preincubation	0.02 (10%)

^aSMP (10 μg/ml) were preincubated for 1 h at 20°C in the standard reaction mixture (pH 10.0) supplemented with bovine serum albumin (1 mg/ml), potassium cyanide (1.5 mM), and NADH (50 μM, where indicated) The residual activity was measured with 30 μM Q₁ as electron acceptor. Flavin nucleotides (10 μM) were added where indicated. All the activities were more than 95% inhibited by 5 μM rotenone. Neither riboflavin nor FAD protected NADH-induced inactivation tested as in sample 4.

4. Discussion

Since Theorell’s pioneering resolution and reconstitution of FMN-containing Warburg’s Old Yellow Enzyme [23] spectacular progress have been achieved in understanding of flavin-protein interaction in many flavoproteins (see Ref. [24] for review). Although there is great interest in the catalytic mechanism of complex I and there has been a clear identification of FMN as the prosthetic group needed for NADH oxidation [7] only a few studies have addressed the properties of the enzyme-bound flavin [16–18, 25]. Previous attempts to reconstitute full catalytic activity from the FMN-deficient acid ethanol solubilized “low molecular weight” NADH dehydrogenase, a preparation evidently similar if not identical to FP, have not been successful although these authors did claim approximately 75% restoration of the original activity with cytochrome c as acceptor (about one tenth of that seen with ferricyanide) [7]. The reduction-dependent dissociation of FMN from FP results in irreversible loss of activity thus suggesting that in addition to its electron transferring function FMN plays a structural role in maintaining the native protein conformation [18]. The results presented in this paper are in excellent agreement with those previously reported for FP [18] except for one important difference: the dissociation of FMNH₂ from the membrane-bound complex I is fully reversible (Fig. 2 and Table 1) and both the NADH:artificial acceptor and rotenone-sensitive quinone reductase activities are reconstituted by the externally added FMN.

The critical factors for the dissociation of FMN from its binding site are enzyme reduction and alkalinization. At present we are unable to suggest any particular species for the protonation-deprotonation equilibrium either in the flavin molecule or in the protein environment. The very strong dependence of apparent K_d on pH (almost two-order difference at pH 9.0 and 10.0, Fig. 3) is indicative of a highly cooperative process. Dramatic change of affinity for FMN and FMNH₂ to the flavin binding site which is thermodynamically equivalent to significant negative shift of the midpoint redox potential of bound versus free flavin, is in line with the proposal for the FMN oxidoreduction-induced conformational change of complex I [18, 19].

The conditions required for dissociation of FMN, i.e. very diluted samples and strongly alkaline pH are certainly irrelevant under any conceivable physiological conditions. However, further resolution-reconstitution studies are expected to shed light on the mechanism (spontaneous or specifically catalyzed of the enzyme assembling. Full reversibility of the reduction-induced FMN dissociation opens a number of avenues to be pursued for further studies of complex I such as reconstitution of the activity with modified flavins [24], redox potential titration of FMN and/or its analogues binding and elucidation of the electron transfer sequence in reactions with different electron acceptors.

The major obstacle for these studies is that significant dissociation of FMN from the holoenzyme (membrane-bound complex I) occurs only in a very diluted system (Fig. 3) and a procedure to prepare FMN-free complex I is evidently needed. Search in this direction is currently under way in our laboratories.

Abbreviations

SMP

submitochondrial particles

FMN

flavin mononucleotide

FAD

flavin adenine dinucleotide

CAPS

3-[Cyclohexylamino]-1-propanesulfonic acid

HAR

hexaammineruthenium(III)

Q₁

2,3-Dimethoxy-5-methyl-6-[3-methyl-2-butenyl]-1,4-benzoquinone