| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 282, Issue 20, 14708-14718, May 18, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1
From the
Medical Research Council Dunn Human Nutrition Unit, Wellcome Trust/MRC Building, Hills Road, Cambridge CB2 2XY, United Kingdom, the ¶Departments of Pediatrics and Pharmaceutical Sciences, University of Colorado Health Sciences Center, Denver, Colorado 80262, and the Department of Chemistry, University of Otago, P. O. Box 56, Dunedin 9001, New Zealand
Received for publication, December 14, 2006 , and in revised form, February 27, 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Whereas the greater hydrophobicity of the longer MitoQ analogs would have increased their partitioning into the lipid phase where they would be most available for reduction by dehydrogenases and protective against lipid peroxidation (2, 3), differences in how MitoQ reacted with individual ubiquinone oxidoreductases suggested two additional factors arising from the size and the charge of its TPP+ moiety (3). The TPP+ group of MitoQ is somewhat bulkier than the ubiquinone moiety raising the possibility of sterically hindered access to the active sites of some mitochondrial ubiquinone oxidoreductases. Using a sterically similar analog of MitoQ10 in which the positively charged phosphonium is replaced with a neutral carbon (tritylQ10; Fig. 1A), we show that bulkiness of the TPP+ moiety is likely to diminish MitoQ10 reduction by complex I.
|
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Fluorescence Quenching—To prepare small unilamellar vesicles (SUV), 4.8 µl of 10 mM 1-pyrene carboxylic acid in Me2SO and 480 µl of either 25 mg·ml–1 turkey egg yolk L-
-phosphatidyl choline (PC; 
60% PC, Type XII-E, Sigma) or 25 mg·ml–1 soybean asolectin (55% PC, Type IV-S, Sigma) in chloroform were evaporated to dryness under a stream of nitrogen in a 15-ml glass Kimax tube. Residual chloroform was removed under vacuum before 12 ml of KPi buffer (50 mM KPi-KOH (pH 7.8), 100 µM EDTA, and 100 µM diethylenetriaminepentaacetic acid) was added followed by incubation for 1 h at room temperature. The tube was then vortexed vigorously and placed in a Decon F5 Minor sonicating water bath for 30 min at room temperature. To assess the size distribution of the egg yolk SUV preparation, 2 µl was adsorbed to a Cu/Rh grid for 2 min before blotting. Uranyl acetate (1%) was immediately applied for 8 s before thorough blotting. The grids were visualized using a Tecnai 12 electron microscope at x26,000 magnification. The SUVs had a mean external spherical diameter of 86 nm with a S.D. of 26 nm (n = 35).
Pyrene quenching was assayed in 2.5 ml of the SUV suspension (1 mg·ml–1) in a stirred cuvette with a Shimadzu RF 5301-PC fluorimeter (
ex 343–346 ± 1.5 nm, em 377 ± 0.75 nm) at 30 °C. Five 2-µl additions from 10 mM stock solutions of MitoQ10, decylQ, idebenone, or decylTPP in ethanol were made at 30–60 s intervals, and the fluorescence measured before (I0) and after (I) each addition. Fluorescence quenching was plotted as I0/I – 1 versus the concentration (mM) of MitoQ10, decylQ, idebenone, or decylTPP, with the slope of the line being the Stern-Volmer (SV) constant (14). The loss of Pyr2 fluorescence upon ubiquinone addition in bulk phase ethanol was assumed to result from an inner filter effect, and the values for I0 and I were corrected accordingly. Correction for the inner filter effect decreased the SV constants in SUVs and mitochondrial membranes when a ubiquinone was the quencher by 1 mM–1. Experiments with decylTPP showed there was no inner filter effect due to the TPP+ moiety.
Bovine heart mitochondria were isolated, and mitochondrial membranes were prepared from these as described previously (16, 17). For fluorescence quenching of 1-pyrene carboxylic acids, bovine heart mitochondrial membranes (800 µg protein·ml–1) were added to a stirred cuvette containing 2.5 ml of KPi buffer in a Shimadzu RF 5301-PC fluorimeter (ex 343–346 ± 1.5 nm, em 377 ± 0.75 nm) at 37 °C. To this was added 1 µl of 10 mM 1-pyrene carboxylic acid in Me2SO and once the fluorescence had stabilized, fluorescence quenching was determined as above. Binding of 1-pyrene carboxylic acids to bovine heart mitochondrial membranes (800 µg of protein) was measured by incubating them in 1 ml of KPi buffer containing 20 µM 1-pyrene carboxylic acid for 10 min at 37 °C. Membranes were pelleted by centrifugation (30 min at 16,000 x g, 37 °C), after which the supernatant was removed and extracted 1–3 times with 1 ml of octan-1-ol. The fraction of pyrene carboxylic acid in the aqueous phase was determined by measuring the A345 of the combined octan-1-ol extracts and comparing it to the A345 of the original 1-pyrene carboxylic acid solution in octan-1-ol. The concentration of Pyr16 in the aqueous phase was too low to be measured accurately and thus all Pyr16 was assumed to be membrane associated.
Solvent Solubility—Serial dilutions of a 100 mM ethanol stock of the relevant ubiquinone were evaporated to dryness under vacuum in a 1.5-ml Eppendorf tube. The ubiquinone (3–1000 nmol) was then resuspended in 1 ml of octan-1-ol or cyclohexane with vigorous vortexing. As MitoQ10 was minimally soluble in cyclohexane and formed a separate orange phase, attempts were made to solubilize it further by placing it for 3 h in a shaking water bath at 37 °C or for 30 min in a Decon F5 Minor sonicating water bath. MitoQ10 was solubilized to the same extent in cyclohexane after either treatment. After centrifugation (30 s at 13,000 x g) the A275 of the supernatant was measured to calculate the ubiquinone concentration.
Protein Purification—Porcine ETF-QOR was purified from porcine liver submitochondrial particles as described previously (18). Human ETF and human medium chain acyl-CoA dehydrogenase (MCAD) were expressed from pET vectors in Escherichia coli and purified (19, 20). Complex I was purified from bovine heart mitochondria (21). Complex II was partially purified from bovine heart mitochondria. Briefly, mitochondria were solubilized in 20 mM Tris-HCl (pH 8), 1% (w/v) dodecylmaltoside, 500 µM EDTA. After centrifugation, the supernatant was applied to a Q-Sepharose column in 20 mM Tris-HCl, pH 8, 0.1% (w/v) dodecylmaltoside, 50 mM sucrose, 2 mM MgSO4, 1 mM EDTA, 10% (v/v) glycerol. Protein was eluted using a linear gradient of 0–500 mM NaCl with complex II eluting at 200 mM NaCl. Fractions containing complex II activity were pooled and concentrated using a 100-kDa cutoff spin column (Vivascience) then applied to a 1.6 x 60 cm S-300 column equilibrated with 20 mM Tris-HCl (pH 8), 0.1% (w/v) dodecylmaltoside, 50 mM sucrose, 2 mM MgSO4, 1 mM EDTA, 10% (v/v) glycerol, 200 mM NaCl. Fractions containing complex II were pooled and concentrated using a 100-kDa cutoff spin column.
Isolated Enzyme Assays—ETF-QOR was assayed in a reaction mixture containing 10 mM Hepes-KOH (pH 7.4), 1 µM MCAD, 1 µM ETF, 100 µM octanoyl-CoA, 6 mM CHAPS, and 60 µM of either CoQ2, decylQ, idebenone, tridecylQ, MitoQ10, or MitoQ15 at 25 °C. The reaction was initiated by the addition of ETF-QOR and monitored as the decrease in absorbance at 275 nm (
ox-red = 7.4 mM–1·cm–1; this extinction coefficient accounts for the contribution of octenoyl-CoA at 275 nm (18, 22)).
Ubiquinone reduction by isolated complex I was assayed in a stirred cuvette containing 20 mM Hepes-KOH (pH 7.5), 50 µM NAD+, 500 µg·ml–1 asolectin, 0.05% (w/v) CHAPS, 5 mM lactate, 5 units·ml–1 lactate dehydrogenase, and 10 µM of either decylQ or MitoQ10 at 32 °C. After a 2-min preincubation, the reaction was initiated by addition of 2 µg·ml–1 isolated complex I and monitored by measuring the decrease in A275. Rotenone (8 µg·ml–1) was added as indicated. Asolectin (Fluka) was partially purified by several rounds of precipitation with acetone (23), stored under nitrogen at –20 °C, and added as a 10 mg·ml–1 stock in 1% (w/v) CHAPS, 20 mM Hepes-KOH (pH 7.5). NADH oxidation by isolated complex I was assayed in a reaction mixture containing 20 mM Hepes-KOH (pH 7.5), 100 µM NADH, 500 µg·ml–1 asolectin, 0.05% (w/v) CHAPS, and 200 µM of either decylQ or MitoQ10 at 32 °C. The reaction was initiated by addition of 2 µg·ml–1 isolated complex I and measured as a decrease in A340–380. Some assays also contained 8 µg·ml–1 rotenone.
Ubiquinone reduction by isolated complex II was assayed in a stirred cuvette containing 20 mM Hepes-KOH (pH 7.5), 5 mM succinate, 500 µg·ml–1 asolectin, 0.05% (w/v) CHAPS, and 50 µM of either decylQ, idebenone, or MitoQ10 at 32 °C. The reaction was initiated by addition of 10 µg·ml–1 isolated complex II and measured as a decrease in A275 (ox-red = 12.5 mM–1·cm–1).
Enzyme Assays in Bovine Heart Mitochondrial Membranes—Although tritylQ10 is soluble in octan-1-ol and cyclohexane up to 1 mM (data not shown), it is significantly more hydrophobic than MitoQ10. Direct addition to lipid containing buffer results in slow or uneven incorporation into bilayers as judged by fluorescence quenching of Pyr12 (data not shown). As this effect was also observed with CoQ4 it was used as a control. Incorporation of tritylQ10 or CoQ4 into phospholipids could be achieved by either sonication or reconstitution from chloroform. However, as conditions could not be easily achieved where either CoQ4 or tritylQ10 were substrates for isolated complex I or complex II, we used bovine heart mitochondrial membranes sonicated in the presence of tritylQ10 or CoQ4.
Ubiquinone reduction by NADH in bovine heart mitochondrial membranes (100 µg·ml–1) was assayed in 20 mM Hepes-KOH (pH 7.5), 50 µM NAD+, 5 mM lactate, 200 µM KCN, and 50 µM of either CoQ4, tritylQ10, decylQ, or MitoQ10 at 32 °C. After microtip sonication (Misonix 3000; 6 x 5 s, 0 °C) in a glass vial the reaction mix was transferred to a stirred cuvette. Following preincubation, the reaction was initiated by addition of lactate dehydrogenase (5 units·ml–1) and monitored by measuring the decrease in A275. No decrease in A275 occurred in the absence of ubiquinone or presence of 8 µg·ml–1 rotenone.
Ubiquinone reduction by succinate in bovine heart mitochondrial membranes (100 µg·ml–1) was assayed in 20 mM Hepes-KOH (pH 7.5), 200 µM KCN, 8 µg·ml–1 rotenone, and 50 µM of either CoQ4, tritylQ10, decylQ, or MitoQ10 at 32 °C. After microtip sonication (6 x 5 s, 0 °C) in a glass vial the reaction mix was transferred to a stirred cuvette. Following a 2-min preincubation, the reaction was initiated by addition of 5 mM succinate and monitored by measuring the decrease in A275. No decrease in A275 occurred in the presence of 20 mM malonate.
Ubiquinol oxidation by bovine heart mitochondrial membranes (50 µg·ml–1) was assayed in 20 mM Hepes-KOH (pH 7.5), 8 µg·ml–1 rotenone, 200 µM KCN, and 50 µM of the ubiquinol form of either CoQ4, tritylQ10, decylQ, or MitoQ10 at 32 °C. After microtip sonication (6 x 5 s, 0 °C) in a glass vial the reaction mix was transferred to a stirred cuvette. Following a 2-min preincubation, the reaction was initiated by addition of 50 µM cytochrome c and monitored by measuring the increase in A550 (red-ox = 21 mM–1·cm–1). The increase in A550 in the presence of 400 nM myxothiazol was subtracted. Ubiquinone was reduced by NaBH4, extracted and stored in ethanol (pH 2) as described previously (3).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-oxidation, via electron-transferring flavoprotein:quinone oxidoreductase (ETF-QOR) could reduce MitoQ10. To investigate the interaction of MitoQ10 with the major sites of ubiquinone reduction we isolated complex I, complex II, and ETF-QOR and assayed their ability to reduce MitoQ10.
When isolated complex II (succinate dehydrogenase) was incubated with MitoQ10 the rate of ubiquinone reduction was identical to that for idebenone and decylQ (Fig. 1B). This confirms that MitoQ10 is a good substrate for this enzyme and is consistent with previous observations in bovine heart mitochondrial membranes (3). Complex I (NADH:ubiquinone oxidoreductase) is the route by which electrons enter the CoQ pool from mitochondrial NADH. When we incubated MitoQ10 or MitoQ15 with isolated complex I in the presence of supplementary phospholipids no rotenone-sensitive ubiquinol generation (supplemental Fig. S2) or NADH oxidation (Fig. 1C), was observed. These results contrast with those when decylQ was used as an electron acceptor (Fig. 1C and supplemental Fig. S2) and demonstrate that MitoQ10 and MitoQ15 are not substrates for isolated complex I, consistent with previous observations in mitochondrial membranes (3). During -oxidation ETF-QOR accepts electrons from the soluble protein ETF, and in turn donates these to the CoQ pool. When isolated ETF-QOR was incubated with MitoQ10 or MitoQ15 the rates of reduction to the ubiquinol form were slow (Fig. 1D). As decylQ, idebenone and a range of other ubiquinone derivatives are readily reduced by ETF-QOR (Fig. 1D) (15), it is concluded that the TPP+ moiety inhibits access of MitoQ10 and MitoQ15 to the ubiquinone reduction site in ETF-QOR.
In summary, MitoQ10 is a good substrate for isolated complex II, but a poor substrate for isolated ETF-QOR and complex I. Therefore complex II appears primarily responsible for reducing MitoQ10 in vivo. We note that while glycerol-3-phosphate dehydrogenase can reduce MitoQ10 (3), its location on the outer surface of the mitochondrial inner membrane and low expression in many tissues may limit its importance in vivo.
The Bulkiness of Its TPP+ Moiety Contributes to the Poor Reactivity of MitoQ10 with Complex I and ETF-QOR, but Not Complex III—Having determined the likely site of MitoQ10 reduction within mitochondria, our aim was to understand why MitoQ10 was not reduced by ETF-QOR or complex I (Fig. 1), or oxidized by complex III (3). Three possible explanations exist for the poor reactivity of MitoQ10 with ETF-QOR, complex I and complex III; either the ubiquinone binding site is in a hydrophobic environment to which the positive charge of the TPP+ moiety prevents access, there is a long narrow access channel that the bulky TPP+ moiety cannot enter or they are inhibited by MitoQ10. Recently the structure of ETF-QOR was determined and shown to have a long narrow ubiquinone binding channel (24). Overlaying MitoQ10 with the ubiquinone moiety and five isoprenoid units of CoQ10 visible in the ETF-QOR structure clearly indicates that the TPP+ moiety of MitoQ10 prevents its ubiquinone group from reaching the ubiquinone binding site of ETF-QOR (data not shown). Thus, steric hindrance caused by the bulkiness of its TPP+ moiety is sufficient to explain the lack of reactivity of MitoQ10 with ETF-QOR. In contrast, molecular modeling of MitoQ10 with complex III shows that the TPP+ moiety does not sterically prevent access of the ubiquinone group to active sites within complex III (3), suggesting an alternative explanation (see below).
|
Inhibition of Complexes I and III Is Not Responsible for Their Poor Reactivity with MitoQ10—One potential explanation for the poor reactivity of MitoQ10 with complexes I and III is that high concentrations of lipophilic cations can inhibit mitochondrial function. Consistent with this in bovine heart mitochondrial membranes there is some evidence of inactivation of complexes I and III in the presence of 50 µM decylTPP (data not shown). To confirm that the lack of MitoQ10 reactivity with complex I and III (Fig. 2) does not solely arise from inhibition or inactivation, we measured the rate of decylQ reactivity with these complexes in the presence of lower concentrations of both redox forms of MitoQ10 or decylTPP. The rate of decylQ reduction by Complex I was not inhibited by either the reduced or the oxidized forms of MitoQ10, or by decylTPP (supplemental Fig. S3A). In contrast, MitoQ10 is largely unreactive with complex I when compared with decylQ in the presence of an equivalent amount of lipophilic cation (supplemental Fig. S3A). Similarly the rate of decylQ oxidation by complex III was not inhibited by the presence of either the reduced or the oxidized forms of MitoQ10, or by decylTPP (supplemental Fig. S3B). Like complex I, MitoQ10 is largely unreactive with complex III when compared with decylQ in the presence of an equivalent amount of lipophilic cation (supplemental Fig. S3B). While in both cases, there is some redox exchange between the two hydrophilic ubiquinones, decylQ and MitoQ10, it is clear that complexes I and III are still active in the presence of both redox forms of MitoQ10.
|
The TPP+ Moiety of MitoQ10 Is Largely Excluded from the Hydrophobic Core of Phospholipid Bilayers While the Ubiquinone Moiety Is Not—Although lipophilic cations such as MitoQ10 are membrane permeable, their charged nature could mean that their steady-state concentration deep within phospholipid bilayers is minimal. This could explain the inability of MitoQ10 to interact effectively with complex III. To determine whether the local distribution of the TPP+ and ubiquinone moieties of MitoQ10 in different regions of the phospholipid bilayer varies and hence to gauge its overall molecular orientation, we measured its ability to quench the fluorescence of a pyrene moiety separated from a carboxylic acid by a carbon chain of n-1 methylene groups (Pyrn, n = 2, 4, 6, 10, 12, 16). This probe was chosen because at neutral pH the depth of the fluorescent pyrene label within the membrane will depend on the length of the chain between it and the carboxylate (14, 26). As a result, the chain length-dependence of pyrene fluorescence quenching gives an indication of the relative position of the quencher within a membrane, as illustrated in Fig. 3A. To determine the position of a quenching moiety within a membrane, we prepared small unilamellar vesicles (SUVs) from egg yolk phospholipids containing a range of these 1-pyrene carboxylic acids. We then measured fluorescence quenching of the incorporated pyrene moiety upon addition of increasing amounts of various quenching molecules, the structures of which are shown in Fig. 1A. The fluorescence before (I0) and after (I) each addition of quencher was plotted versus the quencher concentration, the slope being the Stern-Volmer (SV) constant. That the plot of I0/I – 1 versus quencher concentration is linear indicates that quenching of pyrene fluorescence by TPP+ and ubiquinone in SUVs is collisional (Fig. 3B) (14). MitoQ10 strongly quenched Pyr10–16 which will lie deep in the hydrophobic core of the membrane, but only weakly quenched Pyr2 and Pyr4, which will lie closer to the bilayer surface (Fig. 3, C and D). As MitoQ10 is a composite of a ubiquinone group and a TPP+ moiety (Fig. 1A), both of which might quench pyrenes, we next compared the fluorescence quenching of MitoQ10 with decylTPP and idebenone (Fig. 3C). These compounds were chosen as they are structurally related to MitoQ10, but lack the ubiquinone or TPP+ moieties, respectively. To minimize discrepancies due to differences in quencher concentration in the lipid phase and/or quenching efficiency, and to facilitate comparison of the depth within the membrane of each quenching moiety, these data were replotted as a percentage of the SV constant of the pyrene carboxylic acid that quenched most strongly (Fig. 3D). Like MitoQ10, the ubiquinone moiety of idebenone maximally quenches the deeper lying Pyr10–16 with significantly less quenching of the shallower Pyr2–6 (Fig. 3, C and D). In contrast, decylTPP exhibits maximal quenching with Pyr4 and none at all with Pyr10-Pyr16 (Fig. 3, C and D), suggesting that the TPP+ moiety is concentrated nearer the surface and does not have access to the core of the phospholipid bilayer. While the quenching profiles of MitoQ10 and idebenone are similar, close inspection indicates that there are subtle differences. Pyr4 is quenched significantly more by MitoQ10 than idebenone (p = 0.004) and a relative increase in quenching by MitoQ10 over idebenone is also observed with Pyr2 and Pyr6. In fact the difference between the quenching profiles of idebenone and MitoQ10 is very similar to the profile for decylTPP (Fig. 3E) suggesting that quenching by the two moieties is additive. This implies that the TPP+ moiety of MitoQ10, like that of decylTPP, is found nearer the phospholipid bilayer surface while the ubiquinone group is buried within the bilayer.
The position of the carboxyl carbon of pyrene carboxylic acids is expected to be similar to that of anthroloxy-labeled fatty acids where the ionized carboxyl carbon resides 18.6 Å from the bilayer midplane, with the protonated uncharged form lying slightly deeper at 16 Å (26). This places the carboxyl carbon close to the depth of the ester carbonyl groups on the acyl chains of phospholipids (27). At pH 7.8, pyrene carboxylic acids are largely ionized while at pH 5 they will be 50% neutral and thus will on average lie deeper within the membrane. In contrast, the net charges on the TPP+ and ubiquinone moieties are pH-independent. Therefore lowering the pH to 5.0 should result in the pyrene moiety moving deeper into the membrane relative to the quenching groups. Consistent with this scenario, MitoQ10 and idebenone were much more effective at quenching Pyr2–6 at pH 5.0 than at pH 7.8 while decylTPP was largely ineffective at quenching any of the pyrenes at pH 5.0 (Fig. 3C). Thus this experimental system gives an indication of the position of the TPP+ and ubiquinone moieties within the phospholipid bilayer.
DecylTPP, idebenone, and MitoQ10 were used because they have similar hydrophobicities (log(P) = 3.7, 4.2, 3.4, respectively in octan-1-ol:PBS) (3) and thus, all should partition predominantly (>90%) into SUVs under our conditions, assuming a phospholipid volume of 1 µl·mg–1 (28). This is an important consideration as the observed quenching is a function of the concentrations of quencher and fluorophor actually in the lipid phase, and is thus dependent on their hydrophobicities. To demonstrate that the quenchers were predominantly in the lipid phase decylQ, a ubiquinone analog 10-fold more hydrophobic than idebenone (logP = 5.5 in octan-1-ol:PBS) (3), was used and showed a similar quenching profile to both idebenone and MitoQ10 (Fig. 3D). Thus all the quenchers used were predominantly in the lipid phase, and the results are not compromised by small differences in quencher hydrophobicity. However, differential quencher partitioning became problematic for MitoQ derivatives with shorter chain lengths. When we compared MitoQ3 (logP = 0.5 in octan-1-ol:PBS) with propylTPP and a 3-carbon analog of the 10-carbon linked idebenone, the level of quenching was weak, indicating that most of the quencher was present in the aqueous phase, and that the results were susceptible to small differences in quencher hydrophobicity (data not shown). This caveat is also true for the shorter pyrene carboxylic acids, a proportion of which may not be in the lipid phase.
To evaluate and address this latter issue we changed to a mitochondrial membrane system where we could separate the lipid and aqueous phases and thus correct for the proportion of pyrene carboxylic acid in the aqueous phase. Additionally, while the above work demonstrates that the TPP+ moiety of MitoQ10 is excluded from the hydrophobic core of a phospholipid bilayer, this may not be physiologically relevant as mitochondrial membranes contain a high proportion of protein and have a different lipid composition. To test whether these factors would facilitate penetration of the TPP+ moiety of MitoQ10 into the hydrophobic core, and to correct for the proportion of pyrene carboxylic acid in the aqueous phase, we measured fluorescence quenching in membranes isolated from bovine heart mitochondria. When compared with egg yolk SUVs, experiments with decylTPP gave stronger fluorescence quenching of Pyr6–12 in mitochondrial membranes (Fig. 4A). While this suggests the TPP+ moiety of decylTPP can assume a slightly deeper position within mitochondrial membranes, it was ineffective at quenching Pyr16, the deepest penetrating pyrene carboxylic acid. As heart mitochondrial membranes differ from egg yolk phospholipids in several potentially important ways, e.g. they contain large amounts of protein, a higher proportion of unsaturated fatty acids, cardiolipin and little cholesterol (29), it is difficult to assign a single reason for this increased penetration. However, to partially address this issue we tested the penetration of decylTPP+ into soybean asolectin, which also has a high degree of unsaturation, but is free of cholesterol and has a significant proportion of negatively charged phospholipids. In asolectin there was a similar increase in quenching of Pyr10–16 by decylTPP (Fig. 4A), suggesting the increased penetration of decylTPP into mitochondrial membranes may result from different lipid composition. In contrast to decylTPP, fluorescence quenching in mitochondrial membranes by MitoQ10 was largely unaffected by differences between egg yolk SUVs and mitochondrial membranes (Fig. 4A).
To determine the proportion of each pyrene carboxylic acid that was membrane-associated, we measured the fraction of each pyrene carboxylic acid in the aqueous phase. The percentages of Pyr2, Pyr4, Pyr6, Pyr10, and Pyr12 in the aqueous phase were 76 ± 5, 39 ± 1, 7 ± 1, 3, and 2, respectively, while the percentage of Pyr16 in the aqueous phase was negligible. When the data in Fig. 4A are corrected for the concentration of membrane-associated pyrene carboxylic acid, the true position of the TPP+ moiety of decylTPP is evaluated as being closer to the membrane surface as indicated by enhanced quenching of Pyr2 and Pyr4 and the decreased relative quenching of Pyr6–12 (Fig. 4B). Correction for pyrene carboxylic acid binding to mitochondrial membranes confirms that the ubiquinone moieties of both MitoQ10 and idebenone are inserted into the membrane as they maximally quench Pyr10–16. However, correction enhances the relative quenching of Pyr2 and Pyr4 (Fig. 4B), indicating the steady-state concentration of their ubiquinone moieties close to the membrane surface is higher in reality.
|
The TPP+ Moiety of MitoQ10 Greatly Limits Its Solubility in Cyclohexane—The quenching data indicate that the TPP+ moiety of MitoQ10 is predominantly localized to near the membrane surface while the ubiquinone moiety is present in the membrane core. That the ubiquinone moiety of MitoQ10 can quench buried pyrenes as effectively as lipid soluble ubiquinones such as idebenone indicates that the concentration of the ubiquinone moiety of MitoQ10 within the membrane core is high. However, a notable difference between idebenone and MitoQ10 is that the TPP+ moiety constrains the orientation of MitoQ10 and the free movement of its attached ubiquinone moiety within the hydrophobic core. While this should not affect interactions between the ubiquinone moiety of MitoQ10 and diffusible pyrenes or lipid radicals, access to a deeply buried ubiquinone binding site of a ubiquinone oxidoreductase might not be possible unless the TPP+ moiety moves into the hydrophobic core of the membrane. In this case the concentration of the entire MitoQ10 molecule in the hydrophobic core of the phospholipid lipid bilayer will become the important variable and for efficient reduction it would need to be of the order of the Km for CoQ10 of mitochondrial ubiquinone oxidoreductases. To estimate the steady-state concentration of MitoQ10 in the hydrophobic core of the membrane, we measured the solubility of MitoQ10 in cyclohexane (relative dielectric constant (r) = 2) a solvent mimicking the hydrocarbon tail region of phospholipids in bilayers (30). MitoQ10 was largely insoluble in cyclohexane with a maximum concentration of 1–2 µM in solution (Fig. 5). In contrast, MitoQ10 was very soluble in octan-1-ol (r = 10.3), a more polar solvent that can form hydrogen bonds and mimics the ester region below the phospholipid head groups in membranes (Fig. 5). The insolubility of MitoQ10 in cyclohexane was caused by the TPP+ moiety as idebenone, a ubiquinone of similar structure that lacks the TPP+ moiety, was freely soluble in both cyclohexane and octan-1-ol up to at least 1 mM (Fig. 5).
The relative insolubility of MitoQ10 in cyclohexane could simply reflect insolubility induced by the counter-ion, methanesulfonate (
), in cyclohexane. In vivo the dominant anion would be Cl– so to mimic this environment we incubated 300 nmol of MitoQ10 with 1 ml of cyclohexane:100 µl of phosphate-buffered saline; however there was no increase in the solubility of MitoQ10 in cyclohexane (data not shown). More hydrophobic anions may be available in vivo for ion pair formation thereby facilitating the solubilization of MitoQ10 in the membrane. However, it was found that an equimolar amount (300 nmol) of sodium tetraphenylborate (TPB–), a lipophilic anion that facilitates the membrane permeability of lipophilic cations (31, 32), failed to solubilize MitoQ10 in cyclohexane (data not shown). As a result, we conclude that irrespective of the counter-ions available to it in vivo, MitoQ10 will have low solubility in the non-polar hydrocarbon environment of the membrane core.
In summary, although MitoQ10 is very soluble in octan-1-ol and will thus easily access the periphery of the bilayer, the solubility of MitoQ10 in cyclohexane is 2 µM. This indicates that the steady-state concentration of the entire MitoQ10 molecule within the hydrophobic core of the phospholipid bilayer is likely to be significantly lower than the membrane concentration of endogenous CoQ (1–10 mM) in mitochondria from a range of tissues and species (33, 34). As estimates of the Km for CoQ10 of various mitochondrial ubiquinone oxidoreductases are of that order (34–36), we conclude that the poor reactivity of MitoQ10 with mitochondrial complex III could easily result from the low steady-state concentration of MitoQ10 in non-polar environments, such as the membrane core.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Even though incubation with MitoQ10 led to similar concentrations of the ubiquinone moiety within the membrane as were found on incubation with short-chain ubiquinone analogs (Figs. 3 and 4), their interactions with ubiquinone oxidoreductases were dramatically different (Figs. 1 and 2). The concentration of the ubiquinone moiety, not the whole MitoQ10 molecule, in the membrane will be the important determinant in the quenching reaction with 1-pyrene carboxylic acids as there is no requirement for the TPP+ moiety to move into the hydrophobic core for an interaction between pyrene and ubiquinone to take place. In contrast, the ubiquinone binding sites of ubiquinone oxidoreductases are of a fixed depth and access from the direction of the membrane surface may be sterically hindered. If the TPP+ moiety of MitoQ10 had to move into a hydrophobic environment for its ubiquinone moiety to reach the active site, this would be thermodynamically unfavorable; in this case the concentration of the entire MitoQ10 molecule, not its ubiquinone moiety, in the membrane core would become the relevant variable. As shown in Fig. 5 the concentration of the entire MitoQ10 molecule in the membrane core is much lower than uncharged ubiquinones.
Reduction of MitoQ10 to a ubiquinol by mitochondrial ubiquinone reductases is essential for its antioxidant function in vivo. Additionally, subsequent oxidation of MitoQ10 by complex III would be required for it to function as an effective electron carrier in oxidative phosphorylation. Here we show that the interaction of MitoQ10 with mitochondrial ubiquinone oxidoreductases is significantly different from that of artificial short-chain ubiquinone analogs, such as decylQ or idebenone. These findings indicate that MitoQ10 is not effectively reduced to its antioxidant form by complex I or ETF-QOR, and that the principal mitochondrial enzyme responsible for reducing MitoQ10 to its ubiquinol form in vivo is complex II. This is supported by further measurements in heart mitochondrial membranes where the physiological ratios of respiratory complexes are conserved; the absolute rate of MitoQ10 reduction was 138 nmol·min–1·mg protein–1 for succinate compared with 4 nmol·min–1·mg protein–1 for NADH and 3.7 nmol·min–1·mg protein–1 for glycerol-3-phosphate (3). This level of MitoQ10 reduction by complex II is clearly sufficient for it to function as a highly effective antioxidant in preventing mitochondrial oxidative damage in vivo (1–3, 5). In contrast, MitoQ10 is poorly oxidized by complex III suggesting it would be an ineffective replacement for CoQ10 in oxidative phosphorylation.
Can the reactivity of MitoQ10 be rationalized with what is known about the structures of the various mitochondrial ubiquinone oxidoreductases? Complex II is a transmembrane protein with a ubiquinone binding site near the matrix surface of the membrane (37). The ubiquinone binding site of complex II is accessed via a short channel which the ubiquinone moiety of MitoQ10 can access while the TPP+ moiety remains on the membrane surface (3). Consistent with this, the rate of reduction of MitoQ10 and short-chain ubiquinones by complex II was identical (Figs. 1 and 4), indicating that all compounds could easily access its ubiquinone binding site. While the access channel is too short to sterically hinder MitoQ10, molecular modeling indicates steric hindrance of MitoQ3, and MitoQ3 is a less effective substrate for complex II (3).
In contrast, MitoQ10 is not oxidized by complex III (Fig. 2) and is not reduced by ETF-QOR or complex I, even though the other short-chain ubiquinone analogs were (Figs. 1 and 2). This occurs despite similar concentrations of the ubiquinone moieties of MitoQ10, idebenone and decylQ within the membrane (Fig. 3C). One possible explanation for the lack of reactivity is that the ubiquinone binding sites of complex I, complex III and ETF-QOR have long, narrow access channels to which entry of MitoQ10 is sterically hindered by its TPP+ moiety. A second potential reason is that the ubiquinone binding site is not sterically hindered, but is buried within the hydrophobic core of the membrane where the concentration of MitoQ10 is low. A further consideration for the latter explanation is that the long hydrophobic tail of CoQ10 resides in the midplane of phospholipid bilayers (34). Such a location for CoQ10 requires that ubiquinone oxidoreductases have ubiquinone-binding sites that can be entered from the core of the membrane, often at the expense of access from the direction of the membrane surface. Entry of the ubiquinone moiety of MitoQ10 to ubiquinone binding sites facing the midplane could thus be hindered as it would be thermodynamically unfavourable for MitoQ10 to take on the required orientation, i.e. with its TPP+ moiety in the hydrophobic core of the membrane.
ETF-QOR is a monotopic membrane protein with the molecular structure depicting five isoprenoids from CoQ10 entering the ubiquinone reduction site from the lipid phase via a channel at the base of the enzyme (24). When MitoQ10 is modeled into this active site the bulky TPP+ moiety sterically hinders access of the ubiquinone group and this would appear sufficient to prevent MitoQ10 reacting with ETF-QOR (data not shown). However, as CoQ10 enters from the direction of the bilayer midplane and the TPP+ of MitoQ10 would need to enter the lipid phase for its ubiquinone to reach the active site, both mechanisms could in principle contribute to the lack of reactivity of MitoQ10 with ETF-QOR. There is no molecular structure for the ubiquinone binding site(s) of complex I (25). Here we showed that while tritylQ10 can be reduced by complex I, this rate is slower than CoQ4, but faster than MitoQ10 (Fig. 2A). This indicates that steric hindrance is a substantial component of the diminished reactivity of MitoQ10 with complex I. This may be amplified by the low concentration of MitoQ10 in the membrane core. Complex III contains both a ubiquinone reduction site and a ubiquinol oxidation site, each of which is nearer to and accessed from the direction of the bilayer midplane. The structure of complex III indicates that MitoQ10 can enter both ubiquinone binding sites of complex III without steric hindrance (3, 38). Here we confirm this by showing that a neutral, sterically similar version of MitoQ10, tritylQ10, can be oxidized by complex III (Fig. 2C). As steric hindrance is not a factor and the ubiquinone binding sites of complex III are buried within the hydrophobic core, we conclude that MitoQ10 is weakly oxidized by complex III primarily because the positive charge of the TPP+ tethers the ubiquinone group to the membrane surface.
That the effective interaction of MitoQ10 with enzyme active sites could be limited by tethering of its ubiquinone group to a TPP+ moiety constrained to near the membrane surface, is further supported by the low µM steady-state concentration of the entire MitoQ10 molecule in a solvent mimicking the bilayer core (Fig. 5). This concentration is far lower than that of CoQ10 in mitochondrial membranes (5 mM) (33, 34). Furthermore, estimates of the Km for endogenous CoQ10 for NADH, succinate and glycerol-3-phosphate oxidation in the lipid phase are 5 mM, 500 µM, and 700 µM (34–36) with the Km of complex I for artificial decylQ higher still (140 mM) (34). Therefore, the lack of oxidation of MitoQ10 by complex III (Fig. 2) is consistent with the TPP+ moiety inhibiting penetration of the ubiquinone group into the hydrophobic environment surrounding the ubiquinone-binding sites of this enzyme.
In summary, here we have shown that the favored orientation of MitoQ10 is with the TPP+ moiety near the membrane surface and the ubiquinone penetrating into the membrane core (Figs. 3 and 4). This orientation enables the ubiquinone moiety to access the membrane core to act as a chain breaking antioxidant and allows recycling of MitoQ10 to its ubiquinol form via reduction by complex II. In contrast MitoQ10 cannot be oxidized by complex III explaining why it does not function as an electron carrier in mitochondrial respiration (3). These findings have significant implications for our understanding of the mode of action of mitochondria-targeted ubiquinones and will allow the rational optimization of other mitochondria-targeted molecules.
|
FOOTNOTES |
|---|
The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1–S3.
1 To whom correspondence should be addressed: Medical Research Council Dunn Human Nutrition Unit, Wellcome Trust/MRC Bldg., Hills Road, Cambridge CB2 2XY, UK. Tel.: 44-1223-252900; Fax: 44-1223-252905; E-mail: mpm{at}mrc-dunn.cam.ac.uk.
2 The abbreviations used are: TPP+, triphenylphosphonium cation; BHM, bovine heart mitochondrial membranes; CHAPS, 3-[(3-cholamidoproyl) dimethylammonio]-1-propane-sulfonate; CoQ1–10, ubiquinone with a tail of 1–10 isoprenoid units (2,3-dimethoxy-5-methyl-6-polyprenyl-1,4-benzoquinone); cyt c, cytochrome c; ETF, electron-transferring flavoprotein; ETF-QOR, electron-transferring flavoprotein:quinone oxidoreductase; MCAD, medium-chain acyl-CoA dehydrogenase; MitoQ, ubiquinone linked to a triphenylphosphonium cation by an alkyl chain of unspecified length; MitoQn, ubiquinone linked to a triphenylphosphonium cation by an alkyl chain of n methylene groups; PC, phosphatidylcholine; Pyrn, pyrene linked to a carboxylic acid by an alkyl chain of n-1 methylene groups; SUV, small unilamellar vesicle; SV, Stern-Volmer constant; TPB–, tetraphenylborate anion; TPMP, methyltriphenylphosphonium cation; r, relative dielectric constant.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
), Pyr4 (
), Pyr6 (
), Pyr10 (x), Pyr12 (+), or Pyr16 (
). SV constants are the slope of I0/I – 1 plotted against the MitoQ10 or decylTPP concentration. C, fluorescence quenching of 1-pyrene carboxylic acids by decylTPP (
) and MitoQ10 (
) at pH 5.0 as a function of 1-pyrene carboxylic acid chain length. D, alkyl length at which decylTPP (
