Article pubs.acs.org/biochemistry
Key Role of the Adenylate Moiety and Integrity of the AdenylateBinding Site for the NAD+/H Binding to Mitochondrial ApoptosisInducing Factor Luca Sorrentino,† Alessandra Maria Calogero,† Vittorio Pandini,† Maria Antonietta Vanoni,† Irina F. Sevrioukova,‡ and Alessandro Aliverti*,† †
Department of Biosciences, Università degli Studi di Milano, via Celoria 26, 20133 Milano, Italy Department of Molecular Biology and Biochemistry, University of California, Irvine, California 92697-3900, United States
‡
S Supporting Information *
ABSTRACT: Apoptosis-inducing factor (AIF) is a mitochondrial flavoprotein with pro-life and pro-death activities, which plays critical roles in mitochondrial energy metabolism and caspase-independent apoptosis. Defects in AIF structure or expression can cause mitochondrial abnormalities leading to mitochondrial defects and neurodegeneration. The mechanism of AIF-induced apoptosis was extensively investigated, whereas the mitochondrial function of AIF is poorly understood. A unique feature of AIF is the ability to form a tight, air-stable charge-transfer (CT) complex upon reaction with NADH and to undergo a conformational switch leading to dimerization, proposed to be important for its vital and lethal functions. Although some aspects of interaction of AIF with NAD+/H have been analyzed, its precise mechanism is not fully understood. We investigated how the oxidized and photoreduced wild-type and G307A and -E variants of murine AIF associate with NAD+/H and nicotinamide mononucleotide (NMN+/H) to determine the role of the adenylate moiety in the binding process. Our results indicate that (i) the adenylate moiety of NAD+/H is crucial for the association with AIF and for the subsequent structural reorganization of the complex, but not for protein dimerization, (ii) FAD reduction rather than binding of NAD+/H to AIF initiates conformational rearrangement, and (iii) alteration of the adenylate-binding site by the G307E (equivalent to a pathological G308E mutation in human AIF) or G307A replacements decrease the affinity and association rate of NAD+/H, which, in turn, perturbs CT complex formation and protein dimerization but has no influence on the conformational switch in the regulatory peptide.
A
poptosis-inducing factor (AIF)1,2 is a mitochondrial flavoprotein, highly conserved in vertebrates and with homologues in most other eukaryotes,2,3 that belongs to the glutathione reductase (GR) structural superfamily.4,5 It is encoded by a nuclear gene (AIFM1 in humans) located on the X chromosome.6 AIF is synthesized as a precursor apoprotein with an N-terminal mitochondrial targeting peptide, which is removed after import into the organelle.1 After folding and FAD incorporation, the mature protein adopts a three-domain structure7,8 and attaches to the intermembrane side of the inner mitochondrial membrane via the N-terminal segment. Like some other mitochondrial proteins,9 AIF has a role as a mediator in programmed cell death.1 Upon specific stimuli, the AIF membrane-anchoring peptide is cleaved off by cathepsins or calpains, generating the soluble form of AIF,10 which is released from the organelle through the permeabilized external membrane and translocates to the nucleus. Here, after recruiting various nucleases, AIF binds DNA and promotes chromatin condensation and large-scale DNA fragmentation in a caspase-independent apoptotic pathway also known as parthanatos.11,12 © 2015 American Chemical Society
In addition to its well-characterized function in programmed cell death,13 AIF plays a major role in sustaining mitochondrial structural and functional integrity in healthy cells.14−16 Downregulation of AIF gene expression results in profound mitochondrial abnormalities and impairs oxidative phosphorylation (OXPHOS), particularly at the level of complexes I and III,13,14 suggesting that AIF has a function in respiratory chain biogenesis and/or maintenance.2,15 The relation between AIF and mitochondrial morphology and dynamics is indirectly elucidated by the finding that it interacts with OPA1,17 a dynamin-related GTPase colocalized with AIF, which is involved in the formation of cristae and the regulation of the mitochondrial network.18 OPA1 mutations are associated with a type of dominant optic atrophy (DOA), a degenerative optic neuropathy17 with symptoms very similar to those of known pathogenic AIFM1 mutations (described below). Unfortunately, the molecular mechanism by which AIF assists organelle Received: August 13, 2015 Revised: November 3, 2015 Published: November 4, 2015 6996
DOI: 10.1021/acs.biochem.5b00898 Biochemistry 2015, 54, 6996−7009
Article
Biochemistry
rates of AIF reduction by NADH and reoxidation by oxygen.34,38 The AIF-R201Δ variant is also structurally unstable and prone to FAD release.38 In contrast to other known pathological variants, neurodegeneration caused by AIF-E493V seems to be mediated solely by the increase in its apoptogenic activity. Discovery of the second, noncatalytic NAD+/Hbinding site in mammalian AIF23 places the molecular effects of the R201Δ and E493V mutations under a new perspective, because both residues 201 and 493 map within this putative regulatory site. The AIF-G308E variant,36 reported more than three years ago, associates with a disease similar to that caused by the R201Δ deletion. However, no biochemical characterization of the protein variant has been published. G308 is conserved in vertebrate AIFs2,36 and is positioned in a critical region of the catalytic NAD-binding site (Figure 1A), occupying the second
morphology and redox functions is unknown. In the absence of direct evidence, two hypothetical AIF activities have been proposed.2 As a member of the GR family endowed with NADH- or NADPH-dependent reductase activity toward different electron acceptors,19,20 AIF could be a nicotinamide dinucleotide-dependent enzyme acting on ubiquinone21 and/or other endogenous substrates yet to be identified. On the other hand, the observation that its reaction with NADH yields an exceptionally oxygen-stable FADH−−NAD+ charge-transfer (CT) complex and that such reaction is coupled to protein dimerization led to the proposal that AIF could be a redox and/ or NAD+/H sensor, whose quaternary structure changes are part of a novel signal-transduction pathway.20,22 The two possibilities are not incompatible, because the sensing/transducing role of AIF could be mediated by its catalytic action. AIF is unique within the GR family in possessing the 509−559 insertion and the 190−202 β-hairpin.22 Redox-linked reorganization of the aromatic core of the C-terminal domain, which includes residues of the regulatory 509−559 insertion, affects the conformation of the 190−202 β-hairpin and the 438−452 peptide at the molecule surface, promoting monomer−dimer transition.22 It is interesting to mention the recently provided evidence that human AIF contains a second NAD+/H-binding site outside the active site where hydride transfer (HT) to the FAD cofactor takes place.23 The two sites are predicted to be allosterically linked, because occupation of the second NAD+/ H-binding site requires the previously described conformational changes associated with CT complex formation at the catalytically active site.23 Thus, NAD+/H could fulfill the double role of the enzyme substrate and allosteric effector. Moreover, apoptotic and redox functions of AIF could be somehow interrelated, because CT complex formation and protein dimerization are known to weaken its interaction with DNA.20 Very recently, it has been shown that AIF is absolutely required for translation-coupled import into the mitochondrion and function of CHCHD4/MIA40, through a direct interaction with its N-terminal portion.24,25 CHCHD4/MIA40 is a soluble intermembrane-space protein, which participates in the oxidative folding of components of the respiratory complexes,26,27 and the lack of which mimics the effects of AIF deficiency.24 OXPHOS failure is particularly deleterious for the nervous system, being typically associated with neurodegeneration.28,29 Both the harlequin (Hq) mouse, a natural murine strain in which the AIF level is decreased by ∼80%,30 and experimental models of AIF deficiency show an OXPHOS defect, mitochondrial alterations, and neuron loss.2,31−33 To date, six AIF allelic variants that cause severe human mitochondriopathies have been identified, with neurodegeneration as a common feature.34−41 Deletion of R201 (R201Δ)37,38 and the G308E, G262S, and G338E replacements 36,40,41 cause mitochondrial encephalomyopathies with different degrees of life span shortening. The V243L and E493V replacements mainly cause peripheral symptoms, including neuropathy, muscular atrophy, and deafness,34,39 the condition associated with the latter mutation being known as Cowchock syndrome.42,43 Interpretation at the molecular level of the symptoms determined by AIF defects is complicated by the dual, vital and lethal, nature of the protein. The V243L, G262S, and G338E mutations cause a marked decrease in the AIF level, possibly because of the impairment of its folding and/or stability, which in turn lowers complex I and IV activity.39−41 E493V and R201Δ mutations both significantly increase the
Figure 1. Position of G308 of human AIF (equivalent to G307 of mouse AIF) in the three-dimensional structure of the protein. The Rossmann fold of the NAD-binding domain of human AIF in its reduced form complexed with NAD + is shown in ribbon representation. Relevant elements are displayed as wireframes. The carbon atoms of residue 308, FAD, the adenylate moiety, and the nicotinamide mononucleotide portion of NAD+ are colored yellow, gold, white, and green, respectively. The H-bond between the G308 amide and the 2′-hydroxyl of the NAD+ adenylate is shown as a dashed line. (A) Crystal structure of wild-type human AIF (Protein Data Bank entry 4BUR). (B) Modeled structure of the AIF-G308E variant, obtained by replacing the Gly308 residue with Glu in the wildtype molecule. Among the possible side chain rotamers, that generating the lowest degree of clashes with the neighbor groups was chosen. No attempt to optimize the conformation of the Glu308 environment was made. Molecules were rendered using the PyMol Graphic System (Schrödinger, LLC).
position of the canonical Rossmann fold GXGXXG motif.44 Because this region is implicated in the binding of the pyrophosphate moiety of the dinucleotide, the G308E change is expected to greatly perturb the NAD+/H association process and destabilize the CT complex of AIF by both steric hindrance and charge repulsion (Figure 1B). In this study, we investigated the role of the adenylate moiety of the ligand on the AIF− NAD+/H interaction by analyzing how the FAD redox state and structural alterations in the adenylate-binding site of the murine AIF affect the affinity and binding rate of NAD+/H and nicotinamide mononucleotide (NMN+/H). The murine 6997
DOI: 10.1021/acs.biochem.5b00898 Biochemistry 2015, 54, 6996−7009
Article
Biochemistry
performed by gel filtration on a PD10 cartridge (GE Healthcare). The protein concentration was assayed by the biuret method,45 except for purified samples in which Bradford’s method was used.46 The concentration of all AIF forms was determined spectrophotometrically on the basis of an ε452 of 12.8 mM−1 cm−1. Spectrophotometric Titration Studies. A 8453 diodearray spectrophotometer (Agilent) was used for all measurements. The extinction coefficients of the AIF forms were determined on the basis of the amount of FAD released following sodium dodecyl sulfate (SDS) treatment.47 Photoreduction experiments were performed at 25 °C using solutions of 15 μM AIF forms in 50 mM sodium phosphate (pH 7.5) containing 15 mM EDTA, 2 μM 5-deaza-5-carba-riboflavin, and 5 μM methyl viologen, as described elsewhere.48 When present, NAD+ or NMN+ was at a concentration of 20 and 250 μM, respectively. Solutions were made anaerobic within sealed cuvettes by several cycles of vacuum application and N2 flushing and then subjected to successive light irradiation periods using a slide projector lamp. Spectra were recorded after each reduction step, as equilibrium conditions were established under dark conditions. To determine the dissociation constant of the CT complexes (KdCT) of the reduced WT and G307E variant of AIF with NAD+, protein solutions (22 μM) in 50 mM sodium phosphate buffer were made anaerobic. Aliquots of a concentrated NAD+ solution, made anaerobic by N2 bubbling, were added stepwise at 25 °C, until a 2-fold molar excess of the ligand over protein was reached. Absorption spectra were recorded after each titrant addition when no further spectral changes were observed, which indicated that equilibrium had been established. To calculate KdCT and the extinction coefficient of the CT complexes, the titration data points (A750 as a function of added titrant) were corrected for dilution and fitted the theoretical equation for a 1:1 binding equilibrium (eq 1)49 using Grafit 5 (Erithacus Software Ltd.).
homologue was chosen because its sequence is 92% identical with that of human AIF, is better expressed in Escherichia coli, and has been more extensively characterized.2 Two single replacements, G307E (equivalent to the pathological G308E change in human AIF) and G307A, were introduced to alter the adenylate-binding site in the catalytic ligand pocket. Here we report that the adenylate moiety of NAD+/H and integrity of the adenylate-binding site are crucial for ligand association and structural reorganization in AIF but not for the protein dimerization process, and that the redox-linked conformational switch is initiated by FAD reduction rather than by AIF− NAD+/H complex formation. Our results not only provide new mechanistic insight into the AIF−ligand interaction but also explain why the G308E replacement is so deleterious in human AIF.
■
MATERIALS AND METHODS Materials. NAD+, NADH, NADPH, NMN+ and NMNH, 2,6-dichlorophenolindophenol (DCIP), xanthine, xanthine oxidase (XO), and soybean trypsin inhibitor were purchased from Sigma-Aldrich. Bis(sulfosuccinimidyl)suberate (BS3) was obtained from Pierce Biotechnology (Rockford, IL) and L-1tosylamido-2-phenylethyl chloromethyl ketone (TPCK)-treated trypsin from Cooper Biomedical. All other chemicals were of the highest commercially available grade. Protein Production and Purification. The pKK233-3 vector-based construct for the bacterial overproduction of the wild-type (WT) Δ1−101 form of mouse AIF as a fusion with a C-terminal His tag, here named as pKK-AIFΔ101, was described previously.20 To obtain the G307A and G307E variants of the protein, the corresponding codon changes were introduced into the AIF coding region, using the QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene). The WT and mutant forms of His-tagged AIFΔ1−101, namely, AIF-G307A and AIF-G307E, were overproduced in E. coli strain BL21(DE3) and purified according to the protocol previously reported,20 using an Ä KTA-FPLC system (GE Healthcare). In a typical purification, bacterial cells harvested by centrifugation from a 5 L culture were resuspended in 200 mL of 50 mM sodium phosphate buffer (pH 8.0) containing 10 mM imidazole (solvent A), supplemented with 1 mM dithiothreitol and 1 mM phenylmethanesulfonyl fluoride, and disrupted by sonication. The resulting extract was clarified by high-speed centrifugation and microfiltration and applied on a Ni Sepharose HP column (2.6 cm × 3.8 cm, 20 mL, GE Healthcare), equilibrated with solvent A. After extensive washing, initially with solvent A and subsequently with 2% solvent B [50 mM sodium phosphate buffer (pH 8.0) and 500 mM imidazole], elution of the recombinant protein was achieved by applying a 2 to 50% gradient of solvent B over 100 mL at a flow rate of 2 mL/min. After solvent exchange by gel filtration to 50 mM Tris-HCl buffer (pH 8.0) containing 10% glycerol, the protein was loaded on a Q-Sepharose HP column (2.6 cm × 9.5 cm, 50 mL, GE Healthcare Europe) and eluted with a 0 to 0.5 M NaCl gradient in the same buffer over 250 mL at a flow rate of 2 mL/min. This anion-exchange step was required to remove small amounts of contaminating unidentified E. coli oxidoreductase(s), which interfered with subsequent AIF characterization. After concentration by ultrafiltration and desalting by gel filtration, the protein was stored at −20 °C in 50 mM Tris-HCl buffer (pH 7.4) containing 10% glycerol. Prior to each experiment, buffer exchange to meet the required solvent composition was
⎡ ΔA 750 = Δε750⎣[AIF] + [NAD+] + Kd CT −
⎤ ([AIF] + [NAD+] + Kd CT)2 − 4[AIF][NAD+] ⎦ /2 (1)
where Δε750 represents the extinction coefficient at 750 nm of the CT complex minus those of reduced AIF and NAD+, [AIF] and [NAD+] are the total concentrations of AIF and NAD+, respectively, in the mixture after each titrant addition, and KdCT is the dissociation constant of the CT complex. Kinetic Studies. The NADH−DCIP diaphorase reaction catalyzed by various AIF forms was studied under steady-state conditions at 25 °C in 50 mM sodium phosphate buffer (pH 7.5) and in 100 mM sodium phosphate buffer (pH 8.0). To estimate the kinetic parameters, the concentration of NADH was varied between 0.1 and 20 mM, while that of DCIP was kept constant at 30 μM. The pH dependence of the kinetic parameters of the diaphorase reaction catalyzed by AIF and AIF-G307E was studied in a mixed buffer system composed of 50 mM MES, 50 mM HEPES, and 100 mM ethanolamine, adjusted with NaOH at the desired pH and designed to keep the ionic strength constant at 100 mM in the pH range of 6− 10.5.50 The NADPH−DCIP diaphorase activities of the WT and G307E variant of AIF were studied in a similar way, but only in 50 mM sodium phosphate buffer (pH 7.5). All steadystate data, i.e., initial velocity values (v) as a function of the 6998
DOI: 10.1021/acs.biochem.5b00898 Biochemistry 2015, 54, 6996−7009
Article
Biochemistry
To compare the time course of oxidation of the CT complexes of the WT and G307E variant of AIF by O2, 22 μM protein was incubated overnight with 1 mM NADH, after which the excess of the reductant was removed by gel filtration on a PD10 cartridge. The resulting 1:1 complexes were incubated at 25 °C in air-equilibrated 50 mM sodium phosphate buffer (pH 7.5) and their absorption spectra recorded at intervals until full FAD reoxidation was achieved. Protein Dimerization Studies. To compare the quaternary structures of the oxidized and NAD +-bound CT complexes of the WT and G307E forms of AIF, protein solutions (40 μM) in 20 mM Tris-HCl buffer (pH 7.5) containing 200 mM NaCl and 5 mM dithiothreitol were incubated overnight at 0 °C in the absence or presence of 0.5 mM NADH. Protein samples (∼4 nmol) were analyzed on a Superdex 200 HR 10/30 gel filtration column (GE Healthcare) at 8 °C using the same buffer as the mobile phase and a flow rate of 0.5 mL/min. Cross-linking experiments were performed by incubating 4.4 μM AIF forms with 0.44 mM BS3 in 20 mM sodium phosphate buffer (pH 7.0), at 15 °C, in the absence or presence of 0.5 mM NADH. At fixed reaction times, aliquots were withdrawn and the reaction was stopped by adding Tris base (final concentration of 10 mM). The products were analyzed by sodium dodecyl sulfate−polyacrylamide gel electrophoresis (SDS−PAGE) on 10% polyacrylamide gels. The time course of CT complex formation and protein dimerization was compared by mixing 40 μM AIF forms with different concentrations of NADH or NMNH in 50 mM sodium phosphate buffer (pH 7.5) at 10 °C. CT complex formation was monitored spectrophotometrically, while the monomer−dimer transition was followed by measuring dynamic light scattering (DLS) changes with a DynaPro instrument (ProteinSolutions, Wyatt Technology Corp.). To test if AIF dimerization could be coupled to FAD reduction in the absence of nicotinamide nucleotides, 20−40 μM solutions of AIF forms in 50 mM sodium phosphate buffer (pH 7.5) were photoreduced under anaerobic conditions in sealed cuvettes at 10 °C or enzymatically reduced by incubation with a reducing system consisting of 0.25 mM xanthine, 100 milliunits of XO, and 10 μM methyl viologen.52 The progress of FAD reduction and monomer−dimer transition was monitored by spectrophotometric and DLS measurements, respectively. The low temperature for these experiments was chosen to improve the signal-to-noise ratio of DLS measurements. Limited proteolysis was performed by incubating 0.6 mg/mL (10 μM) AIF forms with 15 μg/mL TPCK-treated trypsin (2.5% weight-to-weight ratio) at 37 °C in 20 mM TrisHCl buffer (pH 7.5) containing 15 mM EDTA, 2 μM 5-deaza5-carba-riboflavin, and 5 μM methyl viologen. When required, AIF-bound FAD was anaerobically photoreduced in the absence or presence of 20 μM NAD+ or preincubated with 2 mM NADH. At different reaction times, aliquots were withdrawn, briefly incubated with a 10-fold molar excess of soybean trypsin inhibitor, and analyzed by SDS−PAGE on 12% polyacrylamide gels. Thermal Denaturation. The conformational stability of the AIF forms was investigated by the ThermoFAD technique,53 using a MiniOpticon System (Bio-Rad Laboratories). Solutions of 13 μM AIF forms in 50 mM sodium phosphate buffer (pH 7.5), both in the absence and in the presence of 1 mM NADH, were subjected to thermal denaturation by increasing the temperature from 15 to 99 °C at a constant rate of 3.33 °C/min. The release of endogenous
reductant concentration ([S]), were analyzed using Grafit 5 by nonlinear fitting to either of the following equations: v=
V [S] K + [S]
(2)
V V [S] K V V + K [S]
(3)
and v=
Equation 2 was used to analyze data sets showing a conventional saturation behavior, where both the maximal rate (V) and the Michaelis constant (K) could be estimated with reasonable accuracy. Equation 3 was instead used to analyze data sets in which the ligand concentration was limited to values too much lower than K to allow determination of V, but suitable to accurately estimate the apparent second-order kinetic constant of the process (V/K).51 To obtain the pK values of the ionization processes affecting the kinetic parameters of the diaphorase reaction, data obtained at different pH values were fitted to the equation for a two-ionization equilibrium (eq 4) using Grafit 5. P=
a × 10−2pH + b × 10−(pH − pK1) + c × 10−(pK1+ pK 2) 10−2pH + 10−(pH − pK1) + 10−(pK1+ pK 2) (4)
where P represents either V or V/K, pK1 and pK2 are the pK values of the two apparent ionizable groups, a and c are the limiting P values of the fully protonated and fully deprotonated states of the enzyme, respectively, and b is the limiting P value for the state with the most acidic and less acidic groups deprotonated and protonated, respectively. The kinetics of AIF CT complex formation was investigated under anaerobic conditions by mixing the oxidized or photoreduced protein forms (AIFred) with the complementary redox form of the nucleotide ligands (NADH/NMNH or NAD+/NMN+, respectively) in 50 mM sodium phosphate buffer (pH 7.5). The fastest reactions, i.e., between the oxidized or reduced WT AIF and NADH or NAD+, respectively, and between reduced AIF-G307E and NAD+, were monitored using an SF-61 DX2 diode-array stopped-flow spectrophotometer (Hi-Tech Scientific). The other reactions, considerably slower, were monitored under anaerobic conditions by a conventional 8453 diode-array spectrophotometer, after being initiated by manual mixing in sealed cuvettes. Absorbance traces at 450 and 700 nm were fitted to a single-exponential decay equation using KinetAsyst version 3.0 (Hi-Tech Scientific) to estimate the values of apparent first-order rate constants (kapp) of the processes. The kapp values obtained at different ligand concentrations were fitted using Grafit 5 to equations equivalent to eqs 2 and 3, where v, V, and K were replaced with kapp, its upper limit (klim), and the half-saturating ligand concentration (K0.5), respectively (see Discussion and eq 6). The ligand concentration dependence of kapp failed to obey simple hyperbolic or linear equations over the entire concentration ranges analyzed, possibly because of the occupation at a high ligand concentration of the mentioned allosteric NAD+/H-binding site,23 which could alter the affinity of the catalytic site responsible for the CT interaction. Thus, the klim/K0.5 values were estimated by limiting curve fitting to the three to five points at the lowest ligand concentrations. 6999
DOI: 10.1021/acs.biochem.5b00898 Biochemistry 2015, 54, 6996−7009
Article
Biochemistry
Figure 2. CT complex formation between AIF forms and NAD+ or NMN+, upon anaerobic photoreduction. Anaerobic solutions of 15 μM AIF forms containing 20 μM NAD+ or 250 μM NMN+ were progressively photoreduced and their spectra recorded at different stages of FAD reduction. The spectra of oxidized proteins are colored black. (A) Reduction of WT AIF in the presence of NAD+. (B) Reduction of WT AIF in the presence of NMN+. (C) Reduction of AIF-G307A in the presence of NAD+. (D) Reduction of AIF-G307E in the presence of NAD+. Arrows indicate the direction of spectral changes. Red stars mark the 400 and 490 nm peaks, characteristic of the anionic flavin semiquinone. The four insets show the plots of A370 vs A420, corresponding to expected local maximum and minimum, respectively, of the anionic flavin semiquinone spectrum. In the case of the AIF−NMN+ couple, the concavity of the plot is caused by the accumulation of NMNH in the last phase of the titration.
accumulate when photoreduction was performed in the absence of the ligand. Accumulation of the FAD semiquinone is more evident from the plots of absorbance change at wavelengths characteristic of this species, as shown in the insets of Figure 2. NMN+ behaved like NAD+ in stabilizing a CT complex species with AIFred (Figure 2B). However, it differed from NAD+ in two important details: (i) a much higher concentration of NMN+ was required to saturate AIFred, and (ii) NMN+ was unable to stabilize FAD semiquinone. G307A and G307E replacements did not prevent CT complex formation of the reduced protein with NAD+ (Figure 2C,D). Significantly, while AIF-G307A photoreduction in the presence of NAD+ yielded a series of spectra indistinguishable from that of WT AIF (Figure 2C), AIF-G307E was gradually converted to a CT complex species without detectable accumulation of semiquinone species (Figure 2D). Thus, our results indicate that disruption of the interaction between the adenylate moiety of NAD+ and its binding site, although not preventing CT complex formation, perturbs the electronic interactions between the nicotinamide and flavin rings. To measure the dissociation constant (KdCT) of the CT complexes, we titrated pre-reduced WT and G307E AIF with NAD+ under anaerobic conditions (Figure 3A,B). We found that, in comparison to the procedure involving addition of NADH to oxidized AIF, the use of AIFred greatly shortened the time required to reach equilibrium after each ligand addition. The course of NAD+ titration of WT and G307E variants of AIFred showed that the mutation significantly lowered the stability of the CT complex (Figure 3C). Curve fitting yielded a KdCT of
