|Home | About | Journals | Submit | Contact Us | Français|
The Ndi1 enzyme found in the mitochondrial membrane of Saccharomyces cerevisiae is an NDH-2-type alternative NADH-quinone (Q) oxidoreductase. As Ndi1 is expected to be a possible remedy for complex I defects of mammalian mitochondria, a detailed biochemical characterization of the enzyme is needed. To identify the ubiquinone (UQ) binding site in Ndi1, we carried out photoaffinity labeling using a photoreactive biotinylated UQ mimic (compound 2) synthesized following a concept of least modification of the substituents on the quinone ring possible. Cleavage with CNBr of Ndi1 cross-linked by 2 revealed the UQ-ring of 2 to be specifically cross-linked to the region Phe281-Met410 (130 amino acids). Digestion of the CNBr fragment with V8 protease and lysylendopeptidase (Lys C) gave ~8 kDa and ~4 kDa peptides, respectively. The ~8 kDa V8 digest was identified as Thr329-Glu399 (71 amino acids) by an N-terminal sequence analysis. Although the ~4 kDa Lys C digest could not be identified by N-terminal sequence analysis, the band was thought to cover the region Gly374-Lys405 (32 amino acids). Taken together, the binding site of the Q-ring of 2 must be located in a common region of the V8 and the Lys C digests Gly374-Glu399 (26 amino acids). Superimposition of the Ndi1 sequence onto a 3D-structural model of NDH-2 from Escherichia coli suggested that the C-terminal portion of this region is close to the isoalloxazine ring of FAD.
The alternative NADH-quinone (Q) oxidoreductase (NDH-2) catalyzes electron transfer from NADH via FAD to quinone without proton pumping. The NDH-2 enzymes are found in bacteria and in mitochondria of plants and fungi, but not in that of mammalian mitochondria. Various species accomplishing respiration with NADH as an electron source use either NDH-1 (complex I, proton-pumping NADH-Q oxidoreductase), NDH-2, or both (1, 2). In general, NDH-2 is much more insensitive than NDH-1 to the inhibition by rotenone (1). Plant and fungal mitochondria possess two types of NDH-2 (3); one is directed to the matrix and catalyzes NADH oxidation in the matrix (designated the internal NADH dehydrogenase or Ndi), and the other faces the intermembrane space and oxidizes NADH in the cytoplasmic space (designated the external NADH dehydrogenase or Nde). The Ndi1, an NDH-2-type enzyme, found in the mitochondria of Saccharomyces cerevisiae is composed of a single polypeptide of 53 kDa (630 amino acids) and contains noncovalently bound FAD and no iron-sulfur cluster (4). This enzyme is similar to complex I in terms of the reoxidation of matrix NADH produced in the Krebs cycle (1, 3).
A series of studies by Yagi and colleagues suggest that the S. cerevisiae NDI1 gene encoding Ndi1 may function as a therapeutic agent for mitochondrial diseases caused by complex I deficiencies (5-13). Actually, the Ndi1 expressed in the substantia nigra of the mouse brains has protective effects against Parkinsonian symptoms caused by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine treatment (12, 14). The expressed Ndi1 enzyme may play a dual role in rescuing complex I-deficient cells (15); one role being to restore the NADH oxidase activity, the other being to decrease oxidative damage caused by complex I inhibition. To further test the idea of using Ndi1 to treat complex I defects, it is important to clarify the mechanism of the protective effects. For this purpose, a detailed biochemical characterization of the Ndi1 enzyme is indispensable.
Previously (16), we successfully overexpressed Ndi1 in Escherichia coli membranes using an N-terminal His10 tag fusion system and purified the functional enzyme. The purified recombinant Ndi1 had similar properties to the authentic Ndi1 enzyme isolated from S. cerevisiae mitochondria (4). The Ndi1 enzyme extracted with Triton X-100 contained no UQ8, but the enzyme extracted with dodecyl-β-D-maltoside (DM) contained a substoichiometric amount of UQ8 (~0.2 mol of UQ8/mol of Ndi1). Exogenous UQ, such as UQ6, could be incorporated into the Ndi1 extracted with Triton X-100 to a level where the ratio of bound-UQ/Ndi1 was ~1. From comparative biochemical studies of the UQ-bound and UQ-free Ndi1 enzymes, we suggested that the bound-UQ site is distinct from the UQ catalytic site (16), while a semiquinone radical bound to the enzyme has not been detected yet. The existence of a bound-UQ site has been demonstrated in E. coli membrane-bound glucose dehydrogenase (17, 18) and bo-type ubiquinol oxidase (19-22) with definitive spectroscopic evidence. Identification of the UQ binding site(s) in Ndi1, whose X-ray crystal structure is unavailable at present, is crucial to understanding the molecular mechanism of electron transfer in the enzyme. To this end, we synthesized several photoreactive azido-Qs possessing a biotin at the terminal end of the side chain and carried out photoaffinity labeling. The present study revealed that the binding site of the Q-ring is located in the sequence region Gly374-Lys405 (26 amino acids). On the basis of the structural model for NDH-2 from Escherichia coli, the C-terminal portion of this region may be close to the isoalloxazine ring of FAD.
Ubiquinone-1 (UQ1) and ubiquinoe-2 (UQ2) were generously provided by Eisai (Tokyo, Japan). Ubiquinone-6 (UQ6) and streptavidin-agarose were purchased from Sigma (St. Louis, MO). Protein standards for SDS-PAGE (the Presition plus blue standard and the Kaleidoscope polypeptide standard) were from Bio-Rad (Hercules, CA). Other reagents were of analytical grade.
The expression and purification of S. cerevisiae Ndi1 were carried out using Triton X-100 as described previously (16). The Ndi1 enzyme prepared by Triton X-100 did not contain UQ6 (Q-free Ndi1). The UQ6-bound Ndi1 possessing a stoichiometric UQ6 (UQ6:Ndi1 = ~1:1) was prepared using DM in place of Triton X-100 as reported previously (16). Electron transfer activities of UQ analogues were measured spectrometrically with a Shimadzu UV-3000 (340 nm, ε = 6.2 mM−1cm−1). The reaction medium (2.5 mL) contained 50 mM NaPi buffer (pH 6.0) and 1 mM EDTA, and the protein concentration was 0.066 μg/mL. The reaction was started by adding 100 μM of NADH after the equilibration of the enzyme with the UQ analogues.
The purified Ndi1 (0.1~0.3 mg of protein/mL) was incubated with azido-Q in a buffer containing 50 mM MOPS-KOH (pH 7.0), 0.1 mM EDTA and 10% glycerol for 10 min at room temperature. The molar ratio of azido-Q to Ndi1 was varied in accordance with the purpose of respective experiments and described in the figure legends. The reconstituted Ndi1 was UV-irradiated with a long wavelength UV lamp (Black-lay model B-100A, UVP, Upland, CA) for 10-20 min on ice at a distance of 10 cm from the light source. When suppression of the cross-linking was tested, a competitor was added and incubated for 10 min prior to the treatment with azido-Q. The cross-linking was quenched by adding 4 × Laemmli’s sample buffer or acetone precipitation, and the samples were subjected to further analysis.
SDS-PAGE was performed according to Laemmli (23) and Schägger (24). For the analysis of undigested Ndi1 samples, denatured samples were separated on 10% Laemmli’s gel, and subjected to CBB staining or electroblotting. For the analysis of CNBr-cleaved or enzymatically digested Ndi1 samples, the samples were separated on Schägger’s tricine gel (16.5% T/ 6% T), and subjected to further analysis.
Electrophoresed proteins were transferred onto a PVDF membrane (Immn-blot PVDF membrane, 0.2 μm, Bio-Rad) in a buffer containing 10 mM NaHCO3, 3 mM NaCO3 and 0.025% (w/v) SDS overnight at 35 V (100 mA) in a cold room. The blotted membrane was blocked with 1% gelatin in Tween TBS (10 mM Tris/HCl, pH 7.4, 0.9% NaCl, and 0.05% Tween 20) for an hour at room temperature, followed by incubation for another hour with Streptavidin-AP (Dako Cytomation, Glostrup, Denmark) at room temperature. The treated membrane was washed with Tween TBS (10 min × 3 times), and developed with NBT/BCIP chromogenic substrates (AP color development kit, Bio-Rad).
Ndi1 (0.1-0.3 mg of protein/mL) was photocross-linked by a 2~4-fold molar excess of azido-Q in the buffer described above. The reaction was terminated by acetone precipitation, and the pellet was subjected to cleavage by CNBr or digestion with Trypsin. For CNBr cleavage, the pellet was resuspended in 70% formic acid, and digested by CNBr at a 300-fold excess of the Met residues contained. The digestion was continued overnight at 37°C in darkness. Cleavage was quenched by adding water and vacuum centrifugal evaporation, and the digests were dissolved in 1 × Schägger’s sample buffer, and subjected to Tricine/SDS-PAGE. When the cleaved peptides were further digested by proteases, the CNBr digests were recovered from the tricine gel by electrophoretical elution (electroelution) using a Centrilutor® equipped with a Centricon® YM-10 (Millipore, Billerica, MA).
For secondary enzymatic digestion, the electroeluted peptides were digested by V8 protease (Glu-C, Roche Applied Science, Penzberg, Germany) and lysylendopeptidase (Lys-C, Wako, Osaka, Japan) in 50 mM ammonium bicarbonate buffer and 20 mM Tris/HCl buffer (pH 9.0), respectively. The protease to substrate ratio was set to 1:30 (v/v), and digestion was continued overnight at 37°C.
For in-gel digestion, the CBB-stained Ndi1 bands were digested by trypsin (Promega) or chymotrypsin (Roche) in a buffer containing 50 mM NH4HCO3 and 5 mM CaCl2 for overnight at 37°C. The digests were extracted from the gel once in 0.1% formic acid and twice in 50% acetonitrile/0.1% formic acid. The combined extracts were concentrated using a vacuum-centrifugal evaporator and subjected to the LC-ESI-MS analysis.
The mass spectra of the enzyme digests were recorded on a triple quadrupole mass spectrometer (API-3000, Applied Biosystems, Foster city, CA) in the Q1-positive ion mode. The peptide samples were separated by an “on-line” HPLC system (Agilent technologies, Santa Clara, CA) using a reverse-phase column (Supelco Discovery® BIO Wide Pore C18, 2.1 × 150 mm, Sigma-Aldrich, Bellefonte, PA), and subjected to a mass spectrometric analysis. The mobile phase was comprised of solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile), delivered at a flow rate of 0.2 mL/min, and a linear gradient of solvent B was formed from 5 to 55% (60 min). The spectra were scanned from 500-1500 m/z with a step size of 0.1-0.2 and dwell time of 1.0 ms. BioMultiview software (Sciex) was used for the reconstruction of multiply charged ions, and calculated molecular masses were compared with theoretical digests from the published protein sequences of S. cerevisiae Ndi1 (Swiss-Prot: P32340) using Peptide Mass (http://www.expasy.org/tools/peptide-mass.html).
The CNBr and enzymatic digests of the cross-linked Ndi1 were separated by Tricine/SDS-PAGE and blotted onto a PVDF membrane as described above. The membrane was stained with 0.025% CBB-R250 in 40% MeOH, and destained with 50% MeOH. The identified band corresponding to the cross-linked peptide fragment was excised and the N-terminal sequence was analyzed using the PROCISE 491HT system (Applied Biosystems).
The cross-linked (i.e. biotinylated) Ndi1 was purified using immobilized streptavidin. The Q-free Ndi1 (0.34 mg of protein/mL, total 68 μg) was photo-irradiated with a 20-fold molar excess of azido-Q, and the excess probe was removed by acetone precipitation. The labeled Ndi1 was re-solubilized by incubating in 2% SDS (60 μL) at 40°C for an hour. Then, the Ndi1 was diluted with 500 μL of 1% Triton X-100 in TBS buffer (0.9% NaCl, 10 mM Tris/HCl, pH 7.4, and 0.21% SDS), and incubated with a 50 μL slurry of Streptavidin-Agarose (Sigma) overnight in a cold room. The avidin resin was washed twice with 0.5% Triton X-100 in TBS buffer, and twice with TBS buffer. The Ndi1 captured on the avidin resin was eluted in a Laemmli’s SDS-PAGE buffer at 90°C. These capture/release procedures using avidin resin were monitored by SDS-PAGE and Western blotting.
Peptide mass finger-printing of the labeled Ndi1 using the SELDI (surface enhanced laser desorption ionization)-TOF MS system was performed at ProteNova Co., Ltd. (Takamatsu, Japan). CBB-stained gel bands were treated with 10 mM DTT and 55 mM iodoacetamide, followed by in gel digestion by TPCK-trypsin (Promega, Madison, WI) in a buffer containing 50 mM NH4HCO3 and 0.02% octyl glucoside (pH 8.0) at 35°C for 5 hours. The digested peptides were extracted with 0.6% TFA in 10% acetonitrile, and analyzed by the ProteinChip SELDI system (Bio-Rad) on a NP 20 chip.
Multiple sequence alignment was performed by the Clustal W program (25), using the amino acid sequences of alternative NADH-Q oxidoreductases from various prokaryotic and eukaryotic organisms as follows: Escherichia coli (CAA23586.1), Azotobacter vinelandii (AAK19737.1), Saccharomyces cerevisiae (CAA43787.1), Yarrowia lipolytica (CAA07265.1), Solanum tuberosum (CAB52796.1), Arabidopsis thaliana (AAM61225.1), Trypanosoma brucei (AAM95239.1), Synechocystis sp. PCC 6803 (BAA17787.1), and Thermosynechococcus elongatus BP-1 (BAC08688.1). These sequences, including those of S. serevisiae Ndi1 and E. coli NDH-2, are classified as the group A NDH-2 based on their conserved binding motifs (2).
For the prediction of the UQ-binding site in Ndi1, the in silico 3D-structure of the NDH-2 from E. coli (26) was obtained from the Protein Data Bank (http://www.rcsb.org/pdb/) in the PDB format (PDB: 1OZK). Visualization of the putative UQ-binding site was carried out using Mac Pymol software (version 0.99, DeLano Scientific LLC, Palo Alto, CA, downloaded from http://pymol.org/).
We synthesized three photoreactive biotinylated azido-Qs (1-3, Figure 1) following a concept of least modification of the substituents on the quinone ring possible. 3H-labeled azido-Qs have been powerful chemical probes for determining the UQ-binding site in various respiratory enzymes (27-31). The substitution pattern on the Q ring used in earlier studies was limited to 2-methyl-3-azido-5-methoxy-(6-alkyl tail)-1,4-benzoquinone (27-31), while the substitution pattern of natural UQ is 2,3-dimethoxy-5-methyl-6-(alkyl tail)-1,4-benzoquinone. Any photoaffinity labeling probe should ideally be as similar in structure as possible to the original biologically active chemical. Therefore we here synthesized 2-methoxy-3-azido-5-methyl-6-(alkyl tail)-1,4-benzoquinone according to our previous procedure (32). This Q-ring structure was shown to react as an efficient photoreactive Q mimic with the cytochrome bd complex in E. coli (33).
Concerning the hydrophobic isoprene tail moiety, it is noteworthy that, although the tail moiety has been generally believed to merely enhance the hydrophobicity of the UQ molecule, the isoprene moiety in the vicinity of the Q-ring (i.e. first and/or second isoprene) plays a specific role as a substrate for respiratory enzymes such as mitochondrial complex I and cytochrome bo complex in E. coli (34). Therefore, to ensure inherent binding of the Q-ring to the enzyme, two isoprenyl groups were directly attached to the Q-ring in compounds 1-3.
Moreover, the incorporation of a radioisotope into the probe molecule is useful as a tag for monitoring the cross-linked protein in photoaffinity labeling experiments; however, this is sometimes accompanied by inconvenience in experimental handling. We therefore attached a biotin tag at the terminal end of the tail through a linker, such as amide (1 and 2) and diethyl glycol (3), to enable several applications in the detection and isolation procedures. In addition, the linker’s structure was varied since this moiety may affect the reactivity of the probe molecule with the enzyme and/or avidin.
To verify the electron-accepting ability of the biotinylated azido-Qs, we determined kinetic parameters with UQ-free Ndi1 using UQ1 and UQ2 as references (Table 1). Judging from the intrinsic electron-accepting efficiency in terms of the Vmax/Km value, we found that compounds 2 and 3 work well as an electron-acceptor, indicating that these compounds are suitable for photoaffinity labeling as UQ mimics. We also confirmed that a specific inhibitor, aurachin analogue AC0-10 (16), completely blocks the reduction of the three azido-Qs.
To find out the best azido-Q for the cross-linking reaction with Ndi1, we compared the extent of cross-linking among the three compounds. The UQ-free Ndi1 was incubated with each azido-Q for 10 min before photolysis with the azido-Q/Ndi1 ratio of 1. The reconstituted azido-Q was UV-irradiated to generate the nitrene, which reacts to an amino acid residue in close proximity to the Q-ring. The cross-linking was detected by Western blotting using AP-conjugated streptavidin (Figure 2). The signal quantity indicated that the extent of cross-linking by 2 is the best among the three. It is however difficult to establish whether this is attributable to the affinity of 2 for the enzyme or to reactivity against AP-conjugated streptavidin or both. Although 3 is the best candidate with regard to the electron-accepting ability, given the extent of the cross-linking, we decided to use 2 as a photoaffinity probe in the subsequent experiments.
To specify the cross-linking by 2, we examined whether other short-chain UQ analogues (UQ2 and azido-Q2) suppress the cross-linking. As shown in Figure 3, both UQ analogues suppressed the cross-linking in a concentration dependent manner, verifying a specific binding of 2.
To identify the UQ-binding site, the Ndi1 photoirradiated with 2 was cleaved by CNBr into several fragments (Figure 4). The cleavage yield was over 90%, as judged by gel electrophoresis. The cross-linking was predominantly found in the ~18 kDa fragment regardless of the 2/Ndi1 ratio (1~20). The specificity of the cross-linking was also confirmed by digestion with V8 and Lys-C as described later, indicating that random cross-linking during photolysis is negligible. On the basis of theoretical sites of cleavage by CNBr (summarized in Figure 6), this fragment was presumed to be Phe281-Met410 (130 amino acids, calculated peptide mass; 14439.5). To verify this, we analyzed the N-terminal sequence of the fragment on the PVDF membrane, which yielded H2N-FEKKLSSYAQSHLENTSIKV-, which corresponds to the amino acid residues Phe281-V300.
The CNBr fragment was isolated by electroelution from the tricine gel and further cleaved into smaller peptides by the treatment with V8 protease or Lys-C. After the V8 cleavage, major cross-linking was found in the ~8 kDa fragment (Figure 5), and the N-terminal sequence analysis of the CBB-stained band on the membrane provided H2N-T329IPYG-. Taking into consideration the theoretical cleavage site of V8 and apparent molecular mass of the band, we identified this fragment as Thr329-Glu399 (71 amino acids, calculated peptide mass; 7648.1, Figure 6). On the other hand, after the Lys-C cleavage, the cross-linking was found solely in the ~4 kDa fragment. On the basis of suspected sites of cleavage by Lys-C, this fragment was thought to be Gly374-Lys405 (32 amino acids, calculated peptide mass; 3314.6, Figure 6). We could not identify the CBB-stained band corresponding to the western band on the PVDF membrane.
To confirm the occurrence of the two fragments, we checked the digestion pattern of the non-labeled ~18 kDa fragment by LC-ESI-MS (Table S1). Actually, both fragments produced by V8 (Thr329-Glu399, 7652.6 Da, 0.06% error) and Lys-C (Gly374-Lys405, 3318.4 Da, 0.1% error) were determined. Taken together, we concluded that the binding site of Q-ring of 2 must be located in a common region of the V8 and Lys C digests (i.e. Gly374-Glu399, 26 amino acids).
A previous study suggested that Ndi1 bears two distinct UQ binding sites: one for bound UQ and the other for catalytic UQ (16). To test this hypothesis and examine the effect of bound-UQ on the photocross-linking, we compared the extent of cross-linking between the UQ6-bound and the UQ-free Ndi1 at a 2:Ndi1 ratio of 1~4. The UQ6-bound Ndi1 containing a stoichiometric UQ6 (~1 mol/mol) was prepared as described previously (16). As shown in Figure S1, there was no difference in the extent of cross-linking between the two enzyme preparations. We also compared the position of the CNBr-cleaved peptides, and the V8 and Lys C digests between the UQ6-bound and the UQ-free Ndi1. Irrespective of the presence of bound-UQ6, the photocross-linking occurred solely at the ~18 kDa band (Figure S1), and the migration pattern of the other digests was also identical to that of UQ-free Ndi1 (Figure S2). If 2 also binds to the site for bound UQ, which may have a higher affinity for UQ than the site for catalytic UQ, the extent of cross-linking should be suppressed in the presence of bound UQ. However, this was not the case. Taken together, the present study did not provide evidence for the occurrence of the bound UQ site in Ndi1. We will further discuss this point later.
In order to pinpoint the amino acid cross-linked by 2, we carried out a MALDI-TOF MS analysis of the triptic digests of the CNBr fragment prepared from Ndi1 photoirradiated with or without 2. It should be noted that the Ndi1 photoirradiated with 2 is a mixture of cross-linked and non-cross-linked enzymes because the reaction yield of the cross-linking is far less than 100%. Although the peptide mass fingerprinting by MALDI-TOF MS covered 87% of the CNBr fragment, no significant difference was observed between the two Ndi1 preparations (data not shown).
We therefore tried to purify the cross-linked (i.e. biotinylated) enzyme using immobilized-streptavidin, as described in the Experimental Procedures. As confirmed by SDS-PAGE and Western analysis at each purification step (Figure 7), the Ndi1 cross-linked by 2 was specifically captured by the avidin resin and released by treatment with Laemmli’s buffer containing 4% SDS. Judging from the protein bands on the CBB-stained gel, we figured that the recovery of the cross-linked Ndi1 from a mixture of cross-linked/non-cross-linked Ndi1 is about 5%.
With the purified cross-linked enzyme in hand, we compared peptide mass fingerprinting of the triptic digests of the cross-linked and non-cross-linked Ndi1 with the SELDI-TOF MS system, which is superior in sensitivity, quantification performance and reproducibility to MALDI-TOF MS (35). While the total sequence coverage was 82%, no significant difference was observed between the cross-linked and non-cross-linked enzymes (e.g., Figure S3). In addition, we carried out LC-ESI-MS analysis of the tryptic, Lys-C, and chymotryptic digests of the cross-linked and no cross-linked Ndi1. However we could not detect the peaks corresponding to the peptides cross-linked by the quinone probe, as shown in Figure S4 taking the triptic digests as an example.
We synthesized photoreactive Q probes bearing a biotin and carried out photoaffinity labeling to identify the UQ binding site in Ndi1. Our results strongly suggest that the sequence Gly374-Glu399 (26 amino acids) constructs the binding domain of the Q-ring. In agreement with this, the region contains Asp383, a residue invariantly conserved among several NDH-2 enzymes (Figure S4) and known to be essential for the enzyme activity.1) To further pinpoint the binding site, we purified the cross-linked Ndi1 and carried out peptide mass fingerprinting by MALDI-TOF MS and SELDI-TOF MS, and also LC-ESI-MS analysis. Unfortunately, we could not detect the peptide fragment cross-linked by 2 using these techniques. This is probably because compound 2 cross-linked to the enzyme may be highly unstable under the protease treatment conditions, and left the peptide. It should be noted that the lifetimes of nitrene and nitrenium ion (a protonated form of nitrene) are rather long, being microseconds order (36). Therefore we can’t exclude the possibility that the cross-linking may have occurred with multiple residues within the region Gly374-Glu399.
Schmid and Gerloff (26) created a 3D-structural model for the alternative NADH-Q oxidoreductase (NDH-2) from E. coli through comparative modeling onto a template from SCOP-family “FAD/NAD-linked reductases” (Figure 8). In the model, they suggested that the redox active parts of the cofactors (i.e. the nicotinamide ring of NADH and the isoalloxazine ring of FAD) are orientated in parallel, in close proximity to each other (< 3 angstrom), to facilitate electron transfer. In contrast to the FAD and NADH binding sites, details of the UQ binding site, such as residues involved in hydrogen bonding, cannot be derived from the structural mode directly. However, they suggested that the UQ binding site would be close to FAD for achieving efficient electron transfer. On the basis of an amino acid sequence alignment of Ndi1 from S. cerevisiae and NDH-2 from E. coli (see Figure S4), we superimposed the region identified in the present study (Gly374-Glu399) into the NDH-2 structure; colored in red in Figure 8. The C-terminal portion of this region is close to the isoalloxazin ring of FAD, enabling facilitative electron transfer between the isoalloxazin ring and UQ-ring. It is likely that NDH2, in the orientation in which it is displayed in Figure 8, tops the membrane (26). Therefore if the UQ-ring is located in this region, the hydrophobic isoprene tail may favorably anchor into the membrane.
The Ndi1, overexpressed in E. coli membranes, extracted with Triton X-100 contains no UQ8, but the enzyme extracted by DM contains a substoichiometric amount of UQ8 (16). The Ndi1 extracted with Triton X-100 can accommodate exogenous UQ, such as UQ6, at a molar ratio of ~1:1 (16). From these results along with other biochemical studies of the UQ-bound and UQ-free Ndi1 enzymes, we suggested that Ndi1 bears two distinct UQ binding sites: one for bound UQ and the other for catalytic UQ (16), with analogy to E. coli membrane-bound glucose dehydrogenase (17, 18) and bo-type ubiquinol oxidase (19-22). However, the present photoaffinity labeling study did not support the occurrence of a bound UQ site in Ndi1; neither the extent nor the pattern of photocross-linking was affected irrespective of the presence of bound (i.e. reconstituted) UQ6. It may be speculated that the bound UQ site proposed previously is identical to the catalytic UQ site. In this case, the reconstituted UQ6 may occupy the catalytic site of Ndi1. Given the observation that an excess amount of UQ2 or 3-azido-Q2 was needed to prevent the cross-linking by 2 (Figure 3), the total amount of UQ6 in the UQ6-reconstitued Ndi1 may be too small to efficiently compete with 2. We actually confirmed that the cross-linking by 2 is suppressed by UQ6 in a concentration dependent way under the experimental conditions identical to the competition test by UQ2 and azido-Q2, (data not shown).
In conclusion, to identify the UQ binding site in Ndi1, we carried out photoaffinity labeling using a newly synthesized photoreactive UQ mimic, 2. Our results reveal the binding site of the Q-ring of 2 to be located in the sequence region Gly374-Glu399 (26 amino acids), the C-terminal portion of which may be close to the isoalloxazin ring of FAD. It is likely that the bound UQ site proposed previously (16) is identical to the catalytic UQ site.
We thank Dr. Eiji Mjima (ProteNova Co., Ltd., Takamatsu, Japan) for kind advice on SELDI-TOF MS analysis.
†This work was supported in part by a Grant-in-Aid for Scientific Research (Grant 20380068 to H. M.) and for Young Scientists (Grant 21880024 to M. M.) from the Japan Society for the Promotion of Science and NIH Grant R01GM033712 (to T. Y.).
1)From site-specific mutations of Ndi1, Yamashita et al. found that D383A/N/Q mutants almost completely lose the NADH-UQ1 oxidoreductase activity (<1%) (Yamashita, T., Nakamaru-Ogiso, E., Yagi, A. and Yagi, T., unpublished data).
SUPPORTING INFORMATION AVAILABLE Synthesis of compounds 1-3, Table S1, and Figures S1-S4. This material is available free of charge via the Internet at http://pubs.acs.org.