Báo cáo khoa học: Metabolic control of mitochondrial properties by adenine nucleotide translocator determines palmitoyl-CoA effects Implications for a mechanism linking obesity and type 2 diabetes

Metabolic control of mitochondrial properties by adenine nucleotide translocator determines palmitoyl-CoA effects

Implications for a mechanism linking obesity and type 2 diabetes Jolita Ciapaite1,5, Stephan J. L. Bakker2, Michaela Diamant3, Gerco van Eikenhorst1, Robert J. Heine3, Hans V. Westerhoff1,4 and Klaas Krab1

1 Department of Molecular Cell Physiology, Institute for Molecular Cell Biology, Faculty of Earth and Life Sciences, VU University,

Amsterdam, the Netherlands

2 Department of Internal Medicine, University of Groningen and University Medical Center Groningen, the Netherlands 3 Department of Endocrinology, Institute for Cardiovascular Research, VU University Medical Center, Amsterdam, the Netherlands 4 Manchester Centre for Integrative Systems Biology, MIB, University of Manchester, UK 5 Centre of Environmental Research, Faculty of Nature Sciences, Vytautas Magnus University, Kaunas, Lithuania

Keywords metabolic control analysis; oxidative phosphorylation; palmitoyl-CoA; reactive oxygen species; type 2 diabetes

Correspondence J. Ciapaite, Centre of Environmental Research, Faculty of Nature Sciences, Vytautas Magnus University, Kaunas, Vileikos 8, LT-44404, Lithuania Fax: +370 37 327904 Tel: +370 37 327905 E-mail: jolita.ciapaite@falw.vu.nl

(Received 21 June 2006, revised 19 August 2006, accepted 4 October 2006)

doi:10.1111/j.1742-4658.2006.05523.x

Inhibition of the mitochondrial adenine nucleotide translocator (ANT) by long-chain acyl-CoA esters has been proposed to contribute to cellular dys- function in obesity and type 2 diabetes by increasing formation of reactive oxygen species and adenosine via effects on the coenzyme Q redox state, mitochondrial membrane potential (Dw) and cytosolic ATP concentrations. We here show that 5 lm palmitoyl-CoA increases the ratio of reduced to oxidized coenzyme Q (QH2 ⁄ Q) by 42 ± 9%, Dw by 13 ± 1 mV (9%), and the intramitochondrial ATP ⁄ ADP ratio by 352 ± 34%, and decreases the extramitochondrial ATP ⁄ ADP ratio by 63 ± 4% in actively phosphorylat- ing mitochondria. The latter reduction is expected to translate into a 24% higher extramitochondrial AMP concentration. Furthermore, palmitoyl- CoA induced concentration-dependent H2O2 formation, which can only partly be explained by its effect on Dw. Although all measured fluxes and intermediate concentrations were affected by palmitoyl-CoA, modular kin- etic analysis revealed that this resulted mainly from inhibition of the ANT. Through Metabolic Control Analysis, we then determined to what extent the ANT controls the investigated mitochondrial properties. Under steady- state conditions, the ANT moderately controlled oxygen uptake (control coefficient C ¼ 0.13) and phosphorylation (C ¼ 0.14) flux. It controlled intramitochondrial (C ¼ )0.70) and extramitochondrial ATP ⁄ ADP ratios (C ¼ 0.23) more strongly, whereas the control exerted over the QH2 ⁄ Q ratio (C ¼ )0.04) and Dw (C ¼ )0.01) was small. Quantitative assessment of the effects of palmitoyl-CoA showed that the mitochondrial properties that were most strongly controlled by the ANT were affected the most. Our observations suggest that long-chain acyl-CoA esters may contribute to cellular dysfunction in obesity and type 2 diabetes through effects on cellular energy metabolism and production of reactive oxygen species.

, flux

, concentration control coefficient, quantifying control of intermediate Xm by module i; C J k i

i

h , proton leak flux; Jo, oxygen uptake flux; Jp, phosphorylation flux; LCAC, long chain acyl-CoA ester; QH2 ⁄ Q ratio, ratio of

Abbreviations [AMP]out, concentration of extramitochondrial AMP; AMPK, AMP-activated protein kinase; ANT, adenine nucleotide translocator; Ap5A, P1,P5-di(adenosine-5¢)-pentaphosphate; ATPin ⁄ ADPin ratio, ATP to ADP ratio in the mitochondrial matrix; ATPout ⁄ ADPout ratio, extramitochondrial ATP to ADP ratio; C X m control coefficient, quantifying control of flux Jk by module i; Dw, membrane potential, i.e. electrical potential across the inner mitochondrial membrane; J1 reduced to oxidized coenzyme Q; ROS, reactive oxygen species; S-13, 5-chloro-3-t-butyl-2¢-chloro-4¢-nitrosalicylanilide.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5288

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

roles.

Impaired mitochondrial

changes

in nonadipose

triacylglycerols

tissues

the majority In mammals, mitochondria produce of ATP required to drive energy-dependent cellular processes. However, mitochondria also play more indirect function is emerging as an important factor in insulin-resistant states: less efficient mitochondrial oxidative phos- phorylation has been demonstrated in both the elderly and insulin-resistant offspring of patients with type 2 diabetes compared with young, healthy con- trols [1,2]. Although under physiological conditions nonesterified fatty acids are an important source of fuel for many tissues because they can yield relatively large quantities of ATP, obesity-related persistent oversupply of nonesterified fatty acids and accumula- tion of is thought to contribute to the molecular mechanisms underlying both insulin resistance and b-cell dysfunc- tion in type 2 diabetes [3,4]. Nonesterified and esteri- fied fatty acids interfere with mitochondrial oxidative phosphorylation in vitro [5,6]. Furthermore, an imbal- ance in fatty acid metabolism resulting in activation of nonoxidative rather than oxidative pathways and accumulation of biologically active molecules [e.g. long-chain acyl-CoAs (LCACs), ceramide, diacylglyc- erol] could adversely affect cellular function by direct effects on a variety of enzymes and induction of apoptosis [4].

Fatty acid-induced insulin resistance in liver is one of the main causes of hyperglycemia in type 2 diabetes [15], and the role of mitochondria in this dysfunction is not fully elucidated. We have shown that, in isolated rat liver mitochondria oxidizing succinate, palmitoyl- CoA inhibited the ANT and induced working-condi- tion-dependent in intramitochondrial and extramitochondrial ATP concentrations and Dw [16]. The relative contribution of a particular enzyme to the control of metabolic pathway flux and concentrations of reaction intermediates determines to what extent inhibition of that enzyme would affect pathway flux and intermediate concentrations. The control can be quantitatively assessed using Metabolic Control Analy- sis [17–19]. The control of fluxes and intermediates is a system property, i.e. it is determined by all enzymes constituting the pathway. For this reason here we quantitatively assessed the control of fluxes and inter- mediates of oxidative phosphorylation not only by the ANT but also by other components of oxidative phos- phorylation. Furthermore, we tested parts of the above hypothesis by determining the effects of palmitoyl- CoA on actively phosphorylating (state 3) mitochon- dria oxidizing a more physiological NADH-delivering substrate, i.e. glutamate plus malate. To investigate which mitochondrial enzymes are involved in the multiple effects that we encountered, we implemented modular kinetic analysis. We found that palmitoyl- CoA acts directly on the ANT, and then indirectly induces ROS production and a concomitant reduction in the extramitochondrial ATP ⁄ ADP ratio. The extent to which palmitoyl-CoA affected different mitochond- rial properties can largely be explained by the magni- tude of the control exerted by the ANT over these properties.

Results

Palmitoyl-CoA effects on steady-state fluxes and intermediate concentrations

reactive oxygen species

formation of

caused

kinase

the

by

Tight regulation of intracellular concentrations of free LCACs by acyl-CoA-binding protein can be impaired under pathological conditions with excess lipid supply (e.g. obesity) because of inadequate the latter [7]. LCACs modulate the expression of activity of the mitochondrial adenine nucleotide translocator (ANT) from both the outer and matrix sides of the inner mitochondrial membrane by com- petitive displacement of the nucleotide from its bind- ing site on the protein [8]. It has been hypothesized that increased concentrations of free LCACs interfere with mitochondrial function through inhibition of the ANT, leading to lower cytosolic ATP and matrix ADP availability, increased mitochondrial membrane potential (Dw), and reduction level of coenzyme Q [9]. The two latter events are expected to promote the (ROS) [10,11], resulting in impaired cellular functions and cell death. Moreover, increased AMP production by low cytosolic adenylate ATP ⁄ ADP ratio and further breakdown of AMP to adenosine is expected to result in an increase in extracellular adenosine concentration [12]. The latter [13], which can promote is a potent vasodilator sodium retention in the kidney and stimulate sympa- thetic nervous system activity [14].

Table 1 summarizes the effects of 5 lm palmitoyl-CoA on the steady-state fluxes and intermediate concentra- tions in isolated rat liver mitochondria respiring on glutamate plus malate. Palmitoyl-CoA decreased oxygen uptake flux (Jo) by 56 ± 3% and phosphoryla- tion flux (Jp) by 58 ± 7%, and increased proton-leak flux ( J1 h) by 37 ± 6%. The opposite effect was found on the extramitochondrial and matrix ATP ⁄ ADP ratios: the former decreased by 63 ± 4%, and the lat- ter increased by 352 ± 34%. The reduced to oxidized coenzyme Q ratio (QH2 ⁄ Q) increased by 42 ± 9%, and Dw increased by 13 ± 1 mV (9%). The QH2 ⁄ Q

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5289

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Table 1. Steady-state values of fluxes and intermediates, as affec- ted by palmitoyl CoA. Values are mean ± SEM from four experi- ments.

No Palmitoyl-CoA

+ 5 lM Palmitoyl-CoA

53 ± 3 375 ± 26 3.3 ± 0.3 150 ± 2 0.67 ± 0.04 0.16 ± 0.01 6.7 ± 1.0

23 ± 2** 160 ± 15** 4.5 ± 0.4** 163 ± 3** 3.03 ± 0.04** 0.06 ± 0.01* 9.4 ± 1.3*

notion that the palmitoyl-CoA-induced increase in Dw would stimulate ROS production [10], we determined the effect of palmitoyl-CoA on H2O2 production in mitochondria respiring on succinate. Figure 1A shows that palmitoyl-CoA induced H2O2 production in state 3 in a concentration-dependent manner. The palmitoyl-CoA-induced H2O2 production was partially sensitive to protonophore S-13, suggesting dependence inhibition of the on Dw (Fig. 1B). In line with this, ANT with atractyloside or carboxyatractyloside and ATP synthase with oligomycin also induced H2O2 formation, although to a lower extent (Fig. 1B).

70.02

86.68

Jo [nmol O2Æmin)1Æ(mg protein))1] Jp [nmol ADPÆmin)1Æ(mg protein))1] h (nmol O2Æmin)1Æ(mg protein))1] J1 Dw (mV) ATPin ⁄ ADPin ATPout ⁄ ADPout QH2 ⁄ Q AMPout (calculated) (lM)

*P < 0.05 and **P < 0.01 versus no palmitoyl-CoA.

ratio in mitochondria respiring on succinate was 4.9 ± 0.3 and 5.4 ± 0.2 in the absence and presence of 5 lm palmitoyl-CoA, respectively. We conclude that palmitoyl-CoA affects virtually all steady-state proper- ties of these mitochondria, albeit to various extents.

Palmitoyl-CoA effects on mitochondrial H2O2 production

We have shown that 5 lm palmitoyl-CoA caused a significant increase in Dw in actively phosphorylating mitochondria (state 3) respiring on succinate [16] and the NADH-delivering substrate (Table 1). To test

To test whether palmitoyl-CoA metabolism via b-oxidation contributes to increased H2O2 production, we determined the effect of palmitoyl-l-carnitine (sub- strate for b-oxidation) and malonyl-CoA (inhibitor of palmitoyl-carnitine transferase 1, part of the mito- chondrial acyl-CoA transport system). Palmitoyl- l-carnitine (5 lm) alone and in combination with atractyloside (to test whether the effect of palmitoyl- CoA requires both its oxidation and its inhibition of the ANT) stimulated H2O2 production rate less than 5 lm palmitoyl-CoA (Fig. 1B), suggesting that b-oxi- dation was not involved. Furthermore, rotenone, an inhibitor of respiratory chain complex I, had no significant effect on palmitoyl-CoA-induced H2O2 pro- duction. However, partial inhibition of palmitoyl-CoA- induced H2O2 production with malonyl-CoA (Fig. 1B) suggests that palmitoyl-CoA partially exerts its effect from the matrix side.

B

A

Fig. 1. Effect of palmitoyl-CoA on H2O2 production in isolated mitochondria respiring on succinate. (A) Dependence of H2O2 production on palmitoyl-CoA concentration. (B) Comparison of the effects of various inhibitors on H2O2 production. St 3, State 3; p-CoA, palmitoyl-CoA (5 lM), protonophore S-13 (0.2 lM); AT, atractyloside (1.5 lM); CAT, carboxyatractyloside (0.1 lM); Oligo, oligomycin (0.5 lM); Ro, rotenone (2 lM); M-CoA, malonyl-CoA (0.1 mM); PC, palmitoyl-L-carnitine (5 lM). All inhibitors were added in state 3. Values are mean ± SEM from four experiments. *P < 0.001 versus state 3; #P < 0.02 and $P < 0.002 versus 5 lM palmitoyl-CoA.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5290

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Palmitoyl-CoA effects on extramitochondrial AMP concentration

ratios. As

are regulated by the adenylate kinase equilibrium. In the range of relatively low ATPout ⁄ ADPout ratios, a small decrease leads to a large increase in [AMP]out, whereas [AMP]out changes relatively little in the range indicated in of high ATPout ⁄ ADPout Fig. 2A, when experimentally obtained values of ATPout ⁄ ADPout ratios (Table 1 and [16]) are used in the calculation, inhibition with palmitoyl-CoA would cause an increase in [AMP]out of 17% and 24% with succinate and glutamate plus malate as substrates, respectively.

We have shown that 5 lm palmitoyl-CoA caused a sig- nificant decrease in the extramitochondrial ATP ⁄ ADP ratio (ATPout ⁄ ADPout) (Table 1). However, in this par- ticular experiment it was not possible to determine the effect of decreased extramitochondrial ATP availability on extramitochondrial AMP formation experimentally, used P1,P5-di(adenosine-5¢)-pentaphosphate as we (Ap5A) as inhibitor of adenylate kinase to prevent depletion of available ATP and ADP and to maintain steady-state respiration. Instead, we did a theoretical calculation of the extramitochondrial AMP concentra- tion ([AMP]out) expected at different ATPout ⁄ ADPout ratios. This calculation assumes that the adenylate kin- ase reaction is at equilibrium, which is a safe assump- tion because there is not much net flux expected through this enzyme under the conditions investigated. Figure 2A shows [AMP]out predicted to be present at different steady-state ATPout ⁄ ADPout ratios when the total adenylate concentration is 0.1 mm, with the assumption that the proportions of adenine nucleotides

A

B

concentration-dependent

decrease

in

However, such a low ATPout ⁄ ADPout ratio (< 0.2) obtained using excess hexokinase and low concentra- tion of total adenylates (i.e. 100 lm) is not likely to be relevant under physiological conditions. For this reason we performed an experiment without adenylate kinase inhibitor and with higher and more physiologi- cally relevant total adenylate concentration (2 mm). We determined how palmitoyl-CoA (5 and 10 lm) affects the ATPtotal ⁄ ADPtotal ratio and [AMP]total in actively phosphorylating (state 3) mitochondria respir- ing on succinate and compared the experimental and calculated values (Fig. 2B). Because the total adeny- late concentration in the medium was high (2 mm) and the contribution of the matrix adenylates was relatively low ((cid:2) 10 lm), we assumed that changes in the ATPtotal ⁄ ADPtotal ratio reflect changes in the ATPout ⁄ ADPout ratio. Palmitoyl-CoA caused a signi- ficant the ATPtotal ⁄ ADPtotal ratio and increase in [AMP]total, which corresponded quite well to the correlation of [AMP]out and the ATPout ⁄ ADPout ratio predicted by the calculation.

Palmitoyl-CoA specifically affects the ANT

Fig. 2. Dependence of AMP concentration on the ATP ⁄ ADP ratio. (A) Dependence of AMP concentration on the ATP ⁄ ADP ratio when the total concentration of adenylates is 100 lM. [AMP] was calcula- ted as described in Experimental procedures using an equilibrium constant for adenylate kinase equal to 0.442 [42]. The points on the curve show [AMP] expected to be present at the experimentally obtained mean values of ATPout ⁄ ADPout for succinate [16] and glutamate plus malate (Table 1), respectively, if adenylate kinase was not inhibited. (B) Dependence of AMP concentration on the ATP ⁄ ADP ratio when the total concentration of adenylates is 2 mM. The points show experimentally determined dependence of [AMP]total on the ATPtotal ⁄ ADPtotal ratio in actively phosphorylating (state 3) mitochondria respiring on succinate with no adenylate kin- ase inhibitor added. The points correspond to conditions with 0, 5 or 10 lM palmitoyl-CoA added, and are mean ± SEM from three independent experiments. *P < 0.05 versus no palmitoyl-CoA. Succ, Succinate; g + m, glutamate plus malate; p-CoA, palmitoyl- CoA. Open symbols, no palmitoyl-CoA; closed symbols, + palmi- toyl-CoA.

To identify the sites of oxidative phosphorylation directly affected by palmitoyl-CoA, we applied modu- lar kinetic analysis in two different ways: with either Dw or matrix ATP ⁄ ADP ratio (ATPin ⁄ ADPin) as an intermediate. Modular kinetic analysis with Dw as con- necting intermediate revealed that palmitoyl-CoA inhibits the phosphorylating module (Fig. 3A), as the flux through the module (Jp) was significantly lower in the presence of palmitoyl-CoA than in its absence, when both conditions were compared for the same lev- els of Dw. The flux through the substrate-oxidation module was slightly, although not significantly, higher in the presence of palmitoyl-CoA (Fig. 3B), indicating a tendency of palmitoyl-CoA to stimulate the activity of this module, possibly via its effect on b-oxidation. The proton-leak module was not affected directly by palmitoyl-CoA (Fig. 3C).

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5291

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

A

B

C

Fig. 3. Effect of palmitoyl-CoA on the kinetics of the oxidative phosphorylation modules around Dw. (A) Kinetics of the phosphorylation mod- ule as determined by titration of the substrate oxidation module with 0–25 nM myxothiazol. (B) Kinetics of the substrate oxidation module determined by titrating the phosphorylation module with 0–0.3 lM oligomycin. (C) Kinetics of the proton leak module as determined by titra- tion of the substrate oxidation module with 0–55 nM rotenone when the phosphorylation module was blocked with 0.3 lM oligomycin. Jp was calculated as: Jp ¼ Jo ) Jh at the same value of Dw; J1 h was measured as Jo in the absence of ADP phosphorylation [46]. Values are mean ± SEM from four experiments. Open symbols, no palmitoyl-CoA; closed symbols, +5 lM palmitoyl-CoA.

mitochondria respiring on the NADH-delivering sub- strate, ANT is the only component of oxidative phos- phorylation affected by palmitoyl-CoA, although a stimulatory effect on substrate oxidation cannot be excluded. Thus the multitude of effects on steady-state fluxes and intermediate concentrations exerted by palmitoyl-CoA is achieved through inhibition of the ANT.

Further analysis with the ATPin ⁄ ADPin ratio as an intermediate showed that the ATP-consuming module (comprising the ANT and hexokinase) was inhibited by palmitoyl-CoA (Fig. 4A), as concluded from lower flux through the module in the presence of palmitoyl- CoA, while the ATP-producing module was not affected (Fig. 4B). We have shown previously that hexokinase is not inhibited by palmitoyl-CoA [16]. Therefore our current data indicate that, also in

Metabolic control of mitochondrial properties

A

B

To determine whether palmitoyl-CoA affected the properties it would be expected to affect for its direct action on the ANT, and to see if we could account for the observation that some properties were affected more than others, we used the systems biology method of Metabolic Control Analysis. To assess the control of fluxes and intermediates, we took a modular approach (Fig. 5).

Metabolic control of fluxes

Fig. 4. Effect of palmitoyl-CoA on the kinetics of the modules of oxidative phosphorylation around the intramitochondrial ATP ⁄ ADP ratio. (A) Kinetics of the ATP-consuming module as determined by titration of the ATP-producing module with 0–20 nM myxothiazol. (B) Kinetics of the ATP-producing module as determined by titration of the ATP-consuming module with 0–0.75 lM atractyloside. Jp was calculated as: Jp ¼ Jo ) Jh at the same value of Dw and multi- (ADP ⁄ O ¼ 2.7 ± 0.1). Values are plied by the ADP ⁄ O ratio [16] mean ± SEM from four experiments. Open symbols, no palmitoyl- CoA; closed symbols, +5 lM palmitoyl-CoA.

Control coefficients of the six modules of oxidative phosphorylation over the oxygen uptake (Jo) and phosphorylation flux (Jp) for both respiratory sub- strates are summarized in Table 2. The distribution pattern of the control over Jp among the modules was similar to that of Jo for both substrates used except for the negative control exerted by the proton leak (because it dissipates Dw which is needed to drive ADP phosphorylation and adenine nucleotide trans- location). In all conditions, the control distribution

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5292

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Fig. 5. Division of the oxidative phosphorylation into modules. The modules: 1, Q-reducing module, comprising dicarboxylate carrier and sub- strate dehydrogenases (malate and NADH dehydrogenases in the case of glutamate plus malate oxidation, or succinate dehydrogenase in the case of succinate oxidation); 2, QH2-oxidizing module, comprising cytochrome bc1 and cytochrome c oxidase; 3, proton leak module, comprising passive membrane permeability to protons and cation cycling; 4, ATP synthesis, comprising ATP synthase and phosphate carrier; 5, adenine nucleotide translocator; 6, hexokinase. The intermediates: a, QH2 ⁄ Q ratio; b, membrane potential (Dw); d, matrix ATP ⁄ ADP ratio (ATPin ⁄ ADPin); c, extramitochondrial ATP ⁄ ADP ratio (ATPout ⁄ ADPout). Arrows marked e, h1 and p indicate electron flux, transmembrane pro- ton flux, and ATP flux, respectively. The dashed arrow h1 going from Q-reducing module to Dw is valid only when glutamate + malate is used as a substrate.

of the ANT to the control of Jo and Jp was moderate and similar with both respiratory substrates.

four

Table 2. Metabolic control of fluxes. The control coefficients were calculated from elasticity coefficients (Supplementary material, Table S2) and steady-state fluxes (Table 1 and [16] for glutamate plus malate and succinate, respectively). Values are mean ± SEM from three (succinate) or (glutamate plus malate) experi- ments (indicated as subscript). Q red, Q-reducing module; QH2 ox, QH2-oxidizing module; Leak, proton-leak module; ATP synth, ATP- synthesis module; ANT, adenine nucleotide translocator; Hk, hexo- kinase; p-CoA, palmitoyl-CoA.

When the two substrates are compared, using glu- tamate plus malate instead of succinate, control of the fluxes shifts from the respiratory chain to ATP synthe- sis. Furthermore, the distribution of the control within the respiratory chain shifts from the part downstream of coenzyme Q with succinate to the part upstream of coenzyme Q with glutamate plus malate.

C J o i

C J p i

Module, i

No p-CoA + 5 lM p-CoA No p-CoA + 5 lM p-CoA

0.140.02 0.460.01

0.130.02 0.450.01 0.020.00 0.060.00 0.120.03 0.220.01

0.090.02** 0.310.04* 0.060.01** 0.120.04* 0.200.02* 0.210.01

0.090.02** 0.300.03* ) 0.030.01 ) 0.040.01 0.150.05* 0.240.02** 0.260.02

0.060.01 0.130.03 0.240.01

0.270.04 0.150.02

0.160.02* 0.160.02

) 0.030.00 ) 0.060.01**

Succinate Q red QH2ox Leak ATP synth ANT Hk Glutamate plus malate Q red QH2 ox Leak ATP synth ANT Hk

0.260.04 0.140.02 0.040.00 0.280.04 0.130.01 0.150.03

0.140.02* 0.140.01 0.140.01** 0.220.02 0.200.01* 0.150.01

0.310.04 0.140.01 0.160.04

0.280.03 0.260.01** 0.200.01

*P < 0.05 and **P < 0.01 versus no palmitoyl-CoA.

was as expected for state 3: the bulk of flux control was shared between the respiratory chain and the mod- ules involved in the production of extramitochondrial ATP (actually glucose 6-phosphate), with hardly any control by the proton-leak module. The contribution

In agreement with the fact that the ANT is the only target of palmitoyl-CoA in the system of oxidative phosphorylation under these experimental conditions, we found that, with both respiratory substrates, the control exerted by the ANT over Jo and Jp signifi- cantly increased upon inhibition with palmitoyl-CoA. The control of Jo increased by 67% and 55% with suc- cinate and glutamate plus malate, respectively, whereas the control of Jp was affected more strongly: it increased by 87% and 83% with succinate and glutam- ate plus malate, respectively. Owing to the summation property of flux control coefficients [17,18], an increase in the control strength of one component of the system to a decrease in the control automatically leads strength of other component(s). In our case, an increase in the control of fluxes by the ANT was mainly compensated for by decreased control by the respiratory chain modules (Table 2). Furthermore, the control by the proton-leak module slightly but signifi- cantly increased because palmitoyl-CoA increases Dw, moving the system to a new steady state that is closer to state 4, where control by proton leak is high.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5293

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Control of the QH2/Q ratio and Dw

the ANT stimulates

substrates, the ANT exerted negative control over the QH2 ⁄ Q ratio and Dw (Table 3). This is because activa- tion of the phosphorylation branch of oxidative phosphorylation, which consumes Dw. The negative control over the QH2 ⁄ Q ratio is explained similarly.

Palmitoyl-CoA had hardly any effect on the control of the QH2 ⁄ Q ratio when glutamate plus malate was used as a substrate. With succinate, palmitoyl-CoA mainly affected the control of the QH2 ⁄ Q ratio by respiratory-chain modules: control by both coen- zyme Q-reducing and coenzyme QH2-oxidizing mod- ules has decreased. Furthermore, palmitoyl-CoA had little effect on the control of Dw except that the con- trol by the coenzyme Q-reducing and ATP-synthesis module significantly decreased with glutamate plus malate as substrate, and for both substrates the negat- ive control exerted by the proton leak slightly increased because of the effect of palmitoyl-CoA on Dw.

Control of matrix and extramitochondrial ATP/ADP ratios

The control of the ATPin ⁄ ADPin ratio and ATPout ⁄ ADPout ratio is summarized in Table 3. For both sub- strates used, control of the ATPin ⁄ ADPin ratio was shared among all modules of oxidative phosphoryla- tion, with a slight negative control exerted by the proton leak. The ANT exerted a large negative control

Coenzyme Q reduction level and Dw are among the factors that determine ROS production by the mitoch- ondrial respiratory chain [11,12]. The control of these intermediates by the six modules of oxidative phos- phorylation is summarized in Table 3. The sum of all concentration (also Dw) control coefficients in a path- way is zero, by definition [17,18]. Accordingly, the val- ues of the coefficients can be positive or negative depending on whether an enzyme is involved in the production or the consumption of an intermediate, the QH2 ⁄ Q respectively. For both substrates used, ratio was almost solely controlled by the respiratory chain enzymes, with the coenzyme Q reducing module exerting a positive control and the coenzyme QH2 oxidizing module exerting a negative control, while the control of Dw was shared equally between the Dw-gen- erating (positive control) and Dw-consuming or Dw- consumption-stimulating processes (negative control) (Table 3). In the case of succinate oxidation, most of the control of Dw within the respiratory chain resided in the part downstream of coenzyme Q (cytochrome bc1 complex and cytochrome c oxidase), whereas, in the case of glutamate plus malate, the part upstream of coenzyme Q (dicarboxylate carrier and substrate dehydrogenases) had slightly more control of Dw, poss- ibly because NADH dehydrogenase, a proton-pumping active. With both respiratory enzyme, becomes

Table 3. Metabolic control of intermediates. The control coefficients were calculated from elasticity coefficients (Supplementary material, Table S2) and steady-state fluxes (Table 1 and [16] for glutamate plus malate and succinate, respectively). Values are mean ± SEM from three (succinate) or four (glutamate plus malate) experiments (indicated as subscript). Q red, Q-reducing module; QH2 ox, QH2-oxidizing module; Leak, proton-leak module; ATP synth, ATP-synthesis module; ANT, adenine nucleotide translocator; Hk, hexokinase; p-CoA, palmi- toyl-CoA.

C ATPin =ADPin

C ATPout =ADPout

C QH2=Q

i

C Dw i

i

i

No p-CoA

No p-CoA

No p-CoA

No p-CoA

Module, i

+ 5 lM p-CoA

+ 5 lM p-CoA

+ 5 lM p-CoA

+ 5 lM p-CoA

0.570.11 ) 0.300.06 ) 0.010.00 ) 0.040.01 ) 0.070.01 ) 0.150.03

0.380.06 ) 0.140.04 ) 0.020.00 ) 0.050.01 ) 0.080.01 ) 0.090.02

0.020.00 0.060.00 0.000.00 ) 0.010.00 ) 0.020.00 ) 0.040.01

0.010.00 0.060.01 ) 0.010.00 ) 0.010.00 ) 0.020.00 ) 0.030.01

0.200.05 0.780.12 ) 0.040.01 0.260.02 ) 0.390.02 ) 0.810.19

0.180.10 0.790.31 ) 0.110.05 0.790.13* ) 0.780.11* ) 0.880.20

0.250.05 0.850.03 ) 0.050.01 0.110.01 0.240.05 ) 1.410.04

0.140.04* 0.460.06* ) 0.060.01 0.230.07* 0.370.03* ) 1.130.01*

0.280.03* 0.300.04 ) 0.110.02** 0.530.07 0.480.03**

Succinate Q red QH2 ox Leak ATP synth ANT Hk Glutamate plus malate Q red QH2 ox Leak ATP synth ANT Hk

0.560.14 ) 0.400.13 ) 0.010.00 ) 0.080.02 ) 0.040.00 ) 0.040.01

0.600.11 ) 0.420.08 ) 0.030.00* ) 0.050.01 ) 0.060.02 ) 0.040.01

0.030.00 0.020.00 0.000.00 ) 0.020.00 ) 0.010.00 ) 0.010.00

0.020.00* 0.020.00 ) 0.010.00* ) 0.010.00* ) 0.010.00 ) 0.010.00

0.550.14 0.230.08 ) 0.050.01 0.700.11 ) 0.700.14 ) 0.730.11

0.460.18 0.590.21 ) 0.200.08 1.300.40 ) 1.280.36* ) 0.880.25

0.410.06 0.230.04 ) 0.040.01 0.480.06 0.230.02 ) 1.310.03

) 1.470.09

*P < 0.05 and **P < 0.01 versus no palmitoyl-CoA.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5294

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

to the

intermembrane

succinate. The control of the ATPout ⁄ ADPout ratio by the ANT increased by 56% and 113% with succinate and glutamate plus malate as substrate, respectively.

for succinate,

Partial integrated responses to palmitoyl-CoA

over the ATPin ⁄ ADPin ratio, because it functions as a ‘consumer’ of matrix ATP by transporting it from mitochondria space. The distribution of control within the respiratory chain the depended on the substrate used: QH2-oxidizing module exerted more control than Q-reducing module, whereas, with glutamate plus ma- late as substrate, it was the opposite. Palmitoyl-CoA tended to increase the positive control of the ATPin ⁄ ADPin ratio by ATP synthesis and the negative control by the proton leak. The negative control of the ATPin ⁄ ADPin ratio by the ANT increased by 100% and 82% with succinate and glutamate plus malate as substrate, respectively.

on

control

the ATPout ⁄ ADPout

For both substrates, hexokinase exerted the highest ratio negative (Table 3). The remainder of the control was distributed among the respiratory-chain modules, ATP synthesis, and the ANT, with negligible negative control exerted by the proton leak. Similarly to the control of the ATPin ⁄ ADPin ratio, the distribution of the control of the ATPout ⁄ ADPout ratio within the respiratory chain depended on the substrate used. Comparing the two substrates, ATP synthesis exerted less control over the ATPout ⁄ ADPout ratio in the case of succinate oxida- the tion. Palmitoyl-CoA increased the control of ATPout ⁄ ADPout ratio by ATP synthesis and proton leak, and decreased the control by the Q-reducing module with both respiratory substrates, and the con- trol by the QH2-oxidizing module and hexokinase with

Table 4 summarizes integrated elasticities to palmitoyl- CoA and partial integrated responses of system fluxes and intermediates to palmitoyl-CoA mediated through each module of oxidative phosphorylation. With both respiratory substrates, the ANT had the largest elasti- city to palmitoyl-CoA, in agreement with the finding that, under our experimental conditions, the ANT is the main target of palmitoyl-CoA in oxidative phos- phorylation. As a consequence, the response mediated through the ANT contributed most to the overall response of the system fluxes and intermediates to palmitoyl-CoA, i.e. the response through the ANT was responsible for 68% of the decrease in Jo, 68% of the decrease in Jp, 56% of the increase in the QH2 ⁄ Q ratio, 70% of the increase in Dw, 72% of the increase in the ATPin ⁄ ADPin ratio, and 59% of the decrease in the ATPout ⁄ ADPout ratio with succinate as substrate. Similar results were obtained when glutamate plus malate was used as substrate: the response through the ANT contributed 75% of the reduction in Jo, 76% of the reduction in Jp, 68% of the increase in Dw, 88% of the increase in the ATPin ⁄ ADPin ratio, and 69% of the reduction in the ATPout ⁄ ADPout ratio. The exception was the QH2 ⁄ Q ratio, where the response through the

Table 4. Contribution of individual modules of oxidative phosphorylation to the overall response of system variables to palmitoyl-CoA. The partial integrated responses (IR) of each module to 5 lM palmitoyl-CoA were calculated using control coefficients (Tables 2 and 3) and integ- rated elasticity coefficients (Ie) of modules to palmitoyl-CoA as described in [21]. Values are mean ± SEM from three (succinate) or four (glu- tamate plus malate) experiments (indicated as subscript). Modules: 1, Q reducing; 2, QH2 oxidizing; 3, proton leak; 4, ATP synthesis; 5, ANT; 6, hexokinase. p-CoA, Palmitoyl-CoA; OR, overall response.

i

i IRJo

i IRJp

i IRDw

Iei

p(cid:3)CoA

i IRQH2=Q p(cid:3)CoA

p(cid:3)CoA

p(cid:3)CoA

p(cid:3)CoA

i IRATPin =ADPin p(cid:3)CoA

i IRATPout =ADPout p(cid:3)CoA

) 0.140.04 0.150.03 ) 0.070.10 ) 0.830.29 ) 3.320.43 ) 0.180.03

) 0.020.00 0.070.01 0.000.00 ) 0.050.02 ) 0.400.04 ) 0.040.00 ) 0.440.01

) 0.080.03 ) 0.050.02 0.000.00 0.040.02 0.240.04 0.030.00 0.180.04

0.000.00 0.010.00 0.000.00 0.010.00 0.070.00 0.010.00 0.090.00

) 0.020.00 0.110.01 0.000.01 ) 0.220.08 1.270.10 0.140.01 1.280.03

) 0.030.01 0.130.03 0.010.01 ) 0.100.04 ) 0.750.08 0.260.04 ) 0.490.04

0.620.17 0.250.44 0.170.08 0.020.38 ) 5.000.89 ) 0.120.01

Succinate ) 0.020.00 1 0.070.01 2 0.000.00 3 ) 0.050.02 4 ) 0.370.04 5 ) 0.040.00 6 ) 0.410.02 OR Glutamate plus malate 0.160.04 1 0.010.05 2 0.010.00 3 4 0.030.10 ) 0.660.12 5 ) 0.020.00 6 ) 0.470.08 OR

0.160.05 0.010.05 0.000.00 0.030.11 ) 0.710.13 ) 0.020.00 ) 0.530.08

0.330.08 ) 0.100.13 0.000.00 ) 0.010.03 0.180.06 0.000.00 0.400.10

0.020.01 0.000.00 0.000.00 0.000.01 0.060.01 0.000.00 0.080.01

0.330.11 ) 0.010.03 ) 0.010.00 0.090.27 3.791.44 0.090.01 4.281.68

0.250.07 0.020.08 ) 0.010.00 0.050.18 ) 1.100.18 0.160.01 ) 0.620.11

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5295

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

[12]. The primary mechanism of

ANT contributed only 29% of the overall increase in the QH2 ⁄ Q ratio, most of the rest of the increase stem- ming from stimulation of the Q-reducing module (52%).

characterized by

conditions

The response of a system variable to an external effector mediated through a specific module is deter- mined by the control exerted by that module over a system variable and the elasticity of that module to the effector [20,21]. Table 4 shows that the overall effects of palmitoyl-CoA on system fluxes and intermediates were mainly mediated through the ANT and that the properties that were controlled most strongly by the ANT were affected the most.

Discussion

AMP to adenosine and extracellular release of the lat- ter intracellular adenosine production is hydrolysis of AMP by a cyto- solic 5¢-nucleotidase [26]. Increased concentrations of free ADP and AMP in the cytosol are major determi- nants of adenosine production, with extracellular adenosine release correlating linearly with free cyto- solic AMP concentration [27]. Exogenous adenosine is a potent vasodilator (EC50 @ 0.1 lm), and, under phy- siological conditions, it facilitates tissue recovery after intensive workload by increasing blood flow and sup- ply of oxygen and metabolic substrates. Under patho- inappropriate logical intracellular triacylglycerol accumulation, a low cyto- solic ATP ⁄ ADP ratio may persist because of constant inhibition of the ANT leading to a sustained increase in extracellular adenosine concentrations, resulting in hyperperfusion, hypertension, increased urate produc- tion, and other abnormalities common to insulin-resist- ant states [12].

We have shown that palmitoyl-CoA induces ROS pro- duction in actively phosphorylating isolated rat liver mitochondria. Furthermore, it influences the ATPout ⁄ ADPout ratio in such a way that these changes result in increased [AMP]out. This is in line with a mechanism we proposed to underlie the association between obes- ity and type 2 diabetes [9,12]. However, owing to mul- tiple interactions in living systems, it can be difficult to differentiate whether all the effects relate to one or many primary effects. To acknowledge this, using modular kinetic analysis, we established that the pri- mary cause of the effects of palmitoyl-CoA was inhibi- tion of the ANT. Assessment of the metabolic control of fluxes and intermediate concentrations by the ANT then revealed that this enzyme partially controls many fluxes, concentrations and potentials. This then con- firmed that an increase in AMP concentration and, at least partly, stimulation of ROS production are effects of ANT inhibition. This study thereby shows how sys- tems biology methodologies might help in dissecting the convoluted cause–effect chains in multifactorial diseases such as obesity and type 2 diabetes.

range. However, even at

in perfused rat

We have shown that inhibition of the ANT with palmitoyl-CoA results in a significantly lower ATPout ⁄ ADPout ratio. With respect to the interrelation between ATPout ⁄ ADPout ratios and [AMP]out, we have shown here how [AMP]out increases with decreasing ATPout ⁄ ADPout ratio, with larger increases observed at low ratios and smaller changes at high ratios. This indi- cates that the effect of LCACs on AMP production will vary depending on the energy state of the cell. The theoretical assessment of the correlation was supported by experimental findings showing that inhibition of the ANT with palmitoyl-CoA leads to a significant palmi- toyl-CoA concentration-dependent decrease in the ATPtotal ⁄ ADPtotal ratio and a concomitant increase in [AMP]total. On the basis of our findings, we expect that, in intact cells, the absolute cytosolic AMP con- centration will increase moderately in response to a decrease in the cytosolic ATP ⁄ ADP ratio in the phy- siologically relevant low concentrations of AMP, the relative increase in con- centration would still be substantial and so would the relative effect on the rate of production of adenosine; 5¢-nucleotidase operates in vivo at substrate concentra- tions three orders of magnitude below its Km of 1.2 mm [28].

Both starvation and incubation with fatty acids have been shown to cause a concomitant decrease in cyto- solic ATP ⁄ ADP ratios and an increase in total LCAC concentrations liver and isolated hepatocytes, indicating that inhibition of the ANT by LCACs may be relevant in vivo [22–24]. It has been suggested that modulation of ANT activity by LCACs might be physiologically significant in the regulation of gluconeogenesis by fatty acids through effects on the intramitochondrial ATP ⁄ ADP ratio [23].

Inhibition ⁄ deinhibition of the ANT depending on LCAC concentration may be relevant in the regulation of cellular metabolism in vivo via effects on AMP-acti- vated protein kinase (AMPK). Activation of AMPK acts as a switch from anabolic to catabolic metabolism which generates ATP (e.g. stimulation of b-oxidation) [25]. Thus activation of AMPK would seem to be a desirable effect in obesity, as it would promote the

A decrease in the extramitochondrial concentration of ATP may result in increased formation of AMP by adenylate kinase. This may subsequently stimulate a cellular response to stress through activation of AMP- dependent processes [25] or lead to the breakdown of

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5296

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

top of the electron-transfer chain might cause extra ROS production. Our observation that atractyloside, a direct inhibitor of the ANT, caused ROS production that could only partly account for palmitoyl-CoA- induced ROS production indicates that the ANT is only partly involved in this process. It is possible that palmitoyl-CoA decreased mitochondrial antioxidant capacity by inhibiting nicotinamide nucleotide trans- hydrogenase [32], an enzyme that provides NADPH for regeneration of two important antioxidant com- pounds, glutathione and thioredoxin, in the mitochon- dria, and in this way contributed to increased ROS production.

contributing

carboxylase,

pyruvate

consumption of excess fat. However, the combination of persistent ANT inhibition by LCACs with constant activation of AMPK may have some adverse effects, because stimulation of b-oxidation in response to acti- vation of AMPK cannot lead to production of ATP because of lack of mitochondrial ADP. As AMPK sti- mulates cellular fatty acid uptake [29] and the availab- ility of circulating fatty acids is increased in obesity, this may lead to accumulation of intracellular triacyl- glycerols. Furthermore, owing to the decreased flux through the tricarboxylic acid cycle in the case of ANT inhibition, b-oxidation-derived acetyl-CoA may to stimulate increased rates of gluconeogenesis, or may be used for ketone body synthesis in the liver. Indeed, short-term overexpression of AMPK in mouse liver has been shown to induce fatty liver and increase ketogenesis [30].

in particular,

that

shows

the process

substantiates

Our data show that the ANT controls many steady- state concentrations, potentials and fluxes. In agree- ment with this, the specific effect of palmitoyl-CoA on the ANT appears to be consistent with its ability to affect many fluxes, concentrations and potentials. Table 1 shows that palmitoyl-CoA affects different mitochondrial properties to different extents. In the light of these observations, we asked whether Meta- bolic Control Analysis could have served to predict the palmitoyl-CoA effects. We showed that the ANT con- trolled Dw and the QH2 ⁄ Q ratio to the least extent, and indeed it was the least affected by palmitoyl-CoA. Jo, Jp, ATPin ⁄ ADPin and ATPout ⁄ ADPout ratios were more strongly controlled by the ANT, and again this corresponded to a stronger effect of palmitoyl-CoA. We conclude that the specific effect of palmitoyl-CoA on the ANT and the varying extent to which the ANT controls various mitochondrial properties at steady- state can largely explain the observed palmitoyl-CoA effects.

for

The relatively weak control of the QH2 ⁄ Q ratio and Dw by the ANT is in agreement with the finding that ANT inhibition by palmitoyl-CoA and the resulting increase in Dw can only partly account the observed increase in ROS production. We found that, for Dw and the QH2 ⁄ Q ratio, their immediate produc- ers and consumers, i.e. the respiratory-chain compo- nents, exerted the strongest control. This indicates if an increase in ROS production is brought that, about by alterations in Dw and the QH2 ⁄ Q ratio, inter- ference with respiratory-chain function will contribute more than interference with any other component of oxidative phosphorylation.

Stimulation of ROS production is thought to con- tribute to dysfunction of many different cell types, but, to b-cell dysfunction in insulin- resistant states through low expression of antioxidant enzymes in these cells [31]. Mitochondrial Dw and the redox state of coenzyme Q are known to affect ROS formation [10,11]. We have shown that 5 lm palmi- toyl-CoA caused a substantial increase in Dw (13 mV with glutamate plus malate and 15 mV with succinate [16] as substrate, compared with a total state 3–state 4 difference of (cid:2) 25 mV) and induced H2O2 production in mitochondria respiring on succinate. The sensitivity the palmitoyl-CoA-induced H2O2 production to of is partly protonophore Dw-dependent. This the part of our hypothesis suggesting that LCACs bring about ROS production through an increase in Dw [9]. The effect of palmitoyl-CoA on the QH2 ⁄ Q ratio with both res- piratory substrates was less pronounced, casting doubt on the alternative route by which palmitoyl-CoA may affect ROS production by the respiratory chain. Effects through the more elusive local ubiquinone rad- ical remain an option. Our results indicate that the palmitoyl-CoA effect on H2O2 production might be partly exerted from the matrix side, but the effect is b-oxidation-independent as the substrate of b-oxida- tion, palmitoyl-l-carnitine, stimulated H2O2 produc- tion less than did equal amounts of palmitoyl-CoA. Moreover, palmitoyl-CoA did not enhance respiration directly, as measured by modular kinetic analysis. A possibility remains that palmitoyl-CoA increased the redox level of intramitochondrial NADH and flavo- proteins, but it was not able to further stimulate res- piration because it was already operating at Vmax. In such a case, the most reduced redox potential at the

It has been shown that the control of Jo by the ANT is comparable in isolated liver mitochondria [33] and isolated hepatocytes [34], indicating that, at least to a certain extent, results obtained in isolated mito- chondria can be extrapolated to the intact cell. Our results reconfirmed the observation that, in isolated rat liver mitochondria, the ANT has limited control over

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5297

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Determination of adenine nucleotide concentrations

Jo [19,33,35,36] and that it controls Jp more strongly than Jo. However, it has been demonstrated that ANT control of Jo changes depending on intramitochond- rial and extramitochondrial ATP utilization [37,38]. Accordingly, the effect of LCACs on ANT control of Jo must depend on the ATP elasticity of the ATP-util- izing processes active at the moment of inhibition, e.g. inhibition of the ANT in rat liver cells with the specific inhibitor, atractyloside, decreased glucose synthesis to a greater extent than urea synthesis, even though both processes require ATP [24].

kit

Measurement of coenzyme Q reduction

In conclusion, we have shown that the ANT con- trolled all investigated properties of the mitochondrial oxidative phosphorylation to different extents, with the largest control exerted over the ATPin ⁄ ADPin and ATPout ⁄ ADPout ratios. The effects of palmitoyl-CoA largely corresponded to this. Our results suggest that inhibition of the ANT by LCACs may be important in the control of cellular energy metabolism, but it accounts only partly for stimulation of ROS produc- tion.

Adenine nucleotides were extracted with phenol as des- cribed [41]. Concentrations were measured using a lucifer- (BioOrbit, Turku, in–luciferase ATP-monitoring Finland). ATP concentrations in the medium and the mitochondrial matrix were determined from yeast hexo- kinase kinetics as described [16]. As hexokinase kinetics were determined in medium containing creatine and creat- ine phosphate, this medium was used in all experiments with mitochondria. AMP concentration was determined spectrophotometrically using a standard enzymatic assay [42].

Experimental procedures

Materials

Coenzyme Q reduction levels were determined in a thermo- statically controlled (25 (cid:2)C) vessel equipped with platinum and oxygen electrodes, by polarographically measuring the redox state of exogenous coenzyme Q1 (2 lm) [43]. To cal- ibrate the platinum electrode traces, samples were taken from incubations of mitochondria in standard assay med- ium without further additions (state 1) and mitochondria incubated with substrate (10 mm succinate plus 2 lm rote- none, or 5 mm glutamate plus 5 mm malate, state 2). Then 1 mL of sample was quenched with 3 mL 0.2 m HClO4 in methanol (0 (cid:2)C), coenzyme Q was extracted with 3 mL pet- roleum ether (40–60 (cid:2)C), and reduced and oxidized coen- zyme Q in the samples was determined by HPLC as described [44].

Measurement of H2O2 production

Isolation of mitochondria

Yeast hexokinase was from Roche (Mannheim, Germany). Horseradish peroxidase, superoxide dismutase, oligomycin, myxothiazol, atractyloside, carboxyatractyloside, rotenone, palmitoyl-CoA, malonyl-CoA, Ap5A, p-hydroxyphenyl- acetic acid and coenzyme Q1 were from Sigma-Aldrich (Zwijndrecht, the Netherlands).

Measurement of oxygen uptake and Dw

Liver mitochondria were isolated from male Wistar rats (250–300 g) as in [39]. Protein content was determined by the method of Bradford [40], with BSA as a standard.

Calculation of extramitochondrial AMP concentrations

The rate of H2O2 production was estimated from the rate of p-hydroxyphenylacetic acid oxidation (excitation and emission wavelengths 317 nm and 414 nm, respectively) as described [45]. Briefly, mitochondria were incubated at 25 (cid:2)C in 2 mL standard assay medium containing 1 mm diethylenetriaminepenta-acetic acid, 0.2 mm p-hydroxyphe- nylacetic acid, 10 UÆmL)1 horseradish peroxidase and 30 UÆmL)1 superoxide dismutase under the following condi- tions: state 3 and state 3 plus inhibitors [palmitoyl-CoA (1, 2.5 and 5 lm), 5-chloro-3-t-butyl-2¢-chloro-4¢-nitrosalicylan- ilide (S-13, 0.2 lm), atractyloside (1.5 lm), carboxyatrac- tyloside (0.1 lm), oligomycin (0.5 lm), rotenone (2 lm), malonyl-CoA (0.1 mm)] or palmitoyl-l-carnitine (5 lm). Fluorescence signal was quantified using H2O2 as standard.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5298

AMP concentration was calculated as (for derivation see supplementary data, Appendix S1): Mitochondria were incubated at 25 (cid:2)C in a closed, stirred and thermostatically controlled glass vessel equipped with Clark-type oxygen electrode and tetraphenylphosphonium ion (TPP+)-sensitive electrode as described [16]. The assay medium contained 25 mm creatine, 25 mm creatine phos- phate, 75 mm KCl, 20 mm Tris, 2.3 mm MgCl2, 5 mm glutamate plus 5 mm malate, 50 lm Ap5A, pH 7.3. An ADP-regenerating system consisting of excess hexokinase (5.78 UÆmL)1), glucose (12.5 mm) and KH2PO4 (5 mm) was used to maintain steady-state respiration rates. ATP at a concentration of 100 lm was added to initiate state 3 respiration.

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

intermediates constant

intermediates constant

Calculation of control coefficients

(cid:3) (cid:2) ½AMP(cid:4) ¼ ð1Þ ¼ ei Xm aKeq r2 þ r þ Keq (cid:2) . @Xm Xm (cid:3) ¼ ð4Þ @vi vi @lnvi @lnXm

Modular kinetic analysis

where r is the ATP ⁄ ADP ratio (equal to our experimental extramitochondrial ATP ⁄ ADP ratio only when formation of AMP is blocked by Ap5A), a is the total amount of adenylate (100 lm), and Keq is the equilibrium constant of adenylate kinase (Keq ¼ 0.442 [42]).

In the calculation, we assumed that

Metabolic Control Analysis

Definitions

To localize the sites of action of palmitoyl-CoA, the system of oxidative phosphorylation was conceptually subdivided into a small number of functional modules interacting via a limited number of intermediates. The kinetic response of flux through the modules to changes in the concentration of the connecting intermediate was determined in the pres- ence and absence of palmitoyl-CoA by titrating with speci- fic inhibitors as described [16,46]. In the first application, the system was divided into a substrate-oxidation module, a phosphorylation module, and a proton-leak module, with Dw as the connecting intermediate [46]. In the second appli- cation, it was divided into an ATP-producing module and an ATP-consuming module, with the matrix ATP ⁄ ADP ratio as the connecting intermediate [16].

For analysis of metabolic control, we conceptually subdivi- ded the system of oxidative phosphorylation into six (coenzyme Q-reducing module, coenzyme QH2- modules oxidizing module, proton-leak module, ATP-synthesis module, ANT, and hexokinase) connected by four interme- diates: QH2 ⁄ Q ratio; Dw; ATPin ⁄ ADPin; ATPout ⁄ ADPout (Fig. 5). The control coefficients of the modules for the oxygen-uptake and phosphorylation fluxes and concentra- tions of four intermediates were calculated from the system fluxes and elasticity coefficients (i.e. coefficients quantifying sensitivity of flux through the module to changes in concen- tration of an intermediate [17]) using the matrix method [47]. the coen- zyme Q-reducing and coenzyme QH2-oxidizing modules are insensitive to changes in the ATPin ⁄ ADPin and ATPout ⁄ ADPout ratios (i.e. elasticity coefficients are zero) [46]; in the case of succinate oxidation, the coenzyme Q-reducing module is insensitive to Dw, ATP synthesis is sensitive only to Dw and the ATPin ⁄ ADPin ratio, the hexokinase rate is sensitive only to the ATPout ⁄ ADPout ratio, whereas proton leak is sensitive only to Dw [46]; ANT is sensitive to all four intermediates including the QH2 ⁄ Q ratio [48].

i k ¼

ss

ss

The flux control coefficient is defined as the fractional change in the system flux (Jk) at steady-state in response to an infinitesimal change in the rate of an enzyme (module) i (vi) [18]: (cid:2) (cid:3) (cid:2) (cid:3) CJ ¼ ð2Þ @Jk Jk . @vi vi @lnJk @lnvi

The subscript ss refers to the steady-state condition and is hereafter omitted, as are the parentheses. To obtain the elasticity coefficients that were assumed to have a nonzero value, we used a multiple modulation method [49], i.e. each module was titrated with a specific inhibitor (Table 5) and the co-response of the flux and intermediate concentration was measured. The co-response coefficients quantifying the ratio of responses of intermedi- ate Xm and flux Jk after perturbation of module i [50] were determined from the slopes of inhibitor titration curves at steady-state as:

iOXm Jk

¼ ¼ ð5Þ @lnXm @lnvi . @lnJk @lnvi @lnXm @lnJk CXm i CJk i

The concentration control coefficient is defined as the fractional change in the steady-state concentration of inter- mediate (or ratio of concentrations, Dw) (Xm) in response to an infinitesimal direct perturbation of the enzyme (mod- ule) i rate (vi) [17,18]:

i ¼

Table 5. Modulations used to determine the co-response coeffi- cients. Mal, Malonate (0–0.625 mM); Oligo, oligomycin (0–0.3 lM); Atr, atractyloside (0–1.5 lM); Rot, rotenone (0–30 nM); Myx, myxo- thiazol (0–25 nM); Hk, hexokinase (0–5.78 UÆmL)1).

Module

Succinate

Glutamate + malate

CXm ¼ ð3Þ @Xm Xm . @vi vi @lnXm @lnvi

Q reducing QH2 oxidizing Proton leak ATP synthesis ANT Hk

Myx, Oligo, Atr Mal, Oligo, Atr Mal, Myx Mal, Myx, Atr Mal, Myx, Oligo, Hk Mal, Myx, Oligo, Atr

Myx, Oligo, Atr Rot, Oligo, Atr Rot, Myx Rot, Myx, Atr Rot, Myx, Oligo, Hk Rot, Myx, Oligo, Atr

Effectively, the value of the control coefficient of an enzyme indicates the percentage reduction in a system flux (for flux control coefficients) or in an intermediate concentration (for concentration control coefficients) in response to 1% inhibition of the reaction rate of that enzyme. The elasticity coefficient

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5299

is defined as the fractional change in rate v through enzyme (module) i, caused by the fractional change in the concentration of intermediate Xm, when concentrations of other intermediates are held con- stant [17,18]:

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

Acknowledgements

This research was funded by the Dutch Diabetes Foundation (grant no. 1999.007), and supported by various other grants [e.g. BioSim and NucSys (FP6 EU)]. We thank G. Wardeh for providing rat livers.

i ¼

The response of the flux Jk to the change in enzyme (module) i (i „ k) is transmitted via changes in the inter- mediates Xm. The response can be approximated by the expression [18,51]: X CJk ð6Þ (cid:5) ek Xm CXm i intermediates

Dividing by the flux control coefficient yields:

References

iOXm Jk

intermediates

X 1 ¼ ð7Þ (cid:5) ek Xm

1 Petersen KF, Befroy D, Dufour S, Dziura J, Ariyan C, Rothman DL, DiPietro L, Cline GW & Shulman GI (2003) Mitochondrial dysfunction in the elderly: possi- ble role in insulin resistance. Science 300, 1140–1142. 2 Petersen KF, Dufour S, Befroy D, Garcia R &

Shulman GI (2004) Impaired mitochondrial activity in the insulin-resistant offspring of patients with type 2 diabetes. N Engl J Med 350, 664–671.

1 ¼ eQred

þ eQred

ð8Þ

Jo

Dw (cid:5) MyxODw Jo

QH2=Q (cid:5) MyxOQH2Q

1 ¼ eQred

ð9Þ

þ eQred

Jo

Dw (cid:5) OligoODw Jo

QH2=Q (cid:5) OligoOQH2Q

1 ¼ eQred

ð10Þ

þ eQred

Jo

Dw (cid:5) AtrODw Jo

QH2=Q (cid:5) AtrOQH2Q

Thus when the co-response coefficients are known, the elas- ticity coefficients of each module to each intermediate can be calculated from sets of equations such as Eqn 7. For example, the elasticity coefficients of the coenzyme Q-redu- cing module for the QH2 ⁄ Q ratio and Dw in the case of glutamate plus malate oxidation was calculated from the following set: 3 Felber JP & Golay A (2002) Pathway from obesity to diabetes. Int J Obes Relat Metab Disord 26, S39–S45. 4 Unger RH (2002) Lipotoxic diseases. Annu Rev Med 53, 319–336. 5 Schonfeld P, Wieckowski MR & Wojtczak L (2000)

Long-chain fatty acid-promoted swelling of mitochon- dria: further evidence for the protonophoric effect of fatty acids in the inner mitochondrial membrane. FEBS Lett 471, 108–112.

and elasticity (Table S2), coefficients 6 Chua BH & Shrago E (1977) Reversible inhibition of adenine nucleotide translocation by long chain acyl- CoA esters in bovine heart mitochondria and inverted submitochondrial particles. J Biol Chem 252, 6711– 6714.

Calculation of partial integrated response

The calculations for succinate as a respiratory substrate were performed using data obtained previously [16]. The inhibitor titration curves (Figs S1 and S2), the co-response (Table S1) and detailed calculation of control coefficients (Eqn S6) are given in Supplementary material, Appendix S2.

7 Franch J, Knudsen J, Ellis BA, Pedersen PK, Cooney GJ & Jensen J (2002) Acyl-CoA binding protein expression is fiber type- specific and elevated in muscles from the obese insulin-resistant Zucker rat. Diabetes 51, 449–454. 8 Woldegiorgis G, Yousufzai SY & Shrago E (1982) Stu- dies on the interaction of palmitoyl coenzyme A with the adenine nucleotide translocase. J Biol Chem 257, 14783–14787.

9 Bakker SJL, IJzerman RG, Teerlink T, Westerhoff HV, Gans ROB & Heine RJ (2000) Cytosolic triglycerides and oxidative stress in central obesity: the missing link between excessive atherosclerosis, endothelial dysfunc- tion, and b-cell failure? Atherosclerosis 148, 17–21. 10 Korshunov SS, Skulachev VP & Starkov AA (1997) Partial integrated responses quantify the response of a sys- tem variable a (flux J or intermediate X), mediated through a pathway enzyme (module) i, to a singe-step change in concentration of external effector q, and were calculated using control coefficients and integrated elasticity coeffi- cients as described [21]. The overall response of a system variable is the sum of all partial responses, i.e. the partial response indicates how much of the effector-induced change in a system variable is caused by changes in activity of each pathway enzyme (module).

Data presentation

High protonic potential actuates a mechanism of pro- duction of reactive oxygen species in mitochondria. FEBS Lett 416, 15–18.

11 Gille L & Nohl H (2001) The ubiquinol ⁄ bc1 redox cou- ple regulates mitochondrial oxygen radical formation. Arch Biochem Biophys 388, 34–38. 12 Bakker SJL, Gans ROB, ter Maaten JC, Teerlink T, test. A P < 0.05 that

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5300

Westerhoff HV & Heine RJ (2001) The potential role of adenosine in the pathophysiology of the insulin resis- tance syndrome. Atherosclerosis 155, 283–290. Data are expressed as mean ± SEM from n independent mitochondrial preparations. The statistical significance of palmitoyl-CoA effects was determined using paired Stu- dent’s t the difference arose by chance was considered to make the difference statistically significant.

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

13 Stepp DW, Van Bibber R, Kroll K & Feigl EO (1996) Quantitative relation between interstitial adenosine concentration and coronary blood flow. Circ Res 79, 601–610. 28 Truong VL, Collinson AR & Lowenstein JM (1988) 5¢-nucleotidases in rat heart. Evidence for the occur- rence of two soluble enzymes with different substrate specificities. Biochem J 253, 117–121. 14 Costa F & Biaggioni I (1994) Role of adenosine in the

sympathetic activation produced by isometric exercise in humans. J Clin Invest 93, 1654–1660.

29 Luiken JJ, Coort SL, Willems J, Coumans WA, Bonen A, van der Vusse GJ & Glatz JF (2003) Contraction- induced fatty acid translocase ⁄ CD36 translocation in rat cardiac myocytes is mediated through AMP-activated protein kinase signaling. Diabetes 52, 1627–1634. 15 Krebs M & Roden M (2005) Molecular mechanisms of lipid-induced insulin resistance in muscle, liver and vas- culature. Diabetes Obes Metab 7, 621–632.

30 Foretz M, Ancellin N, Andreelli F, Saintillan Y, Gron- din P, Kahn A, Thorens B, Vaulont S & Viollet B (2005) Short-term overexpression of a constitutively active form of AMP-activated protein kinase in the liver leads to mild hypoglycemia and fatty liver. Diabetes 54, 1331–1339. 16 Ciapaite J, Van Eikenhorst G, Bakker SJ, Diamant M, Heine RJ, Wagner MJ, Westerhoff HV & Krab K (2005) Modular kinetic analysis of the adenine nucleo- tide translocator-mediated effects of palmitoyl-CoA on the oxidative phosphorylation in isolated rat liver mito- chondria. Diabetes 54, 944–951. 17 Kacser H & Burns JA (1973) The control of flux. Symp Soc Exp Biol 27, 65–104.

31 Tiedge M, Lortz S, Munday R & Lenzen S (1998) Com- plementary action of antioxidant enzymes in the protec- tion of bioengineered insulin-producing RINm5F cells against the toxicity of reactive oxygen species. Diabetes 47, 1578–1585.

32 Rydstrom J (1972) Site-specific inhibitors of mitochond- rial nicotinamide-nucleotide transhydrogenase. Eur J Biochem 31, 496–504.

18 Heinrich R & Rapoport TA (1974) A linear steady-state treatment of enzymatic chains. General properties, con- trol and effector strength. Eur J Biochem 42, 89–95. 19 Groen AK, Wanders RJ, Westerhoff HV, van der Meer R & Tager JM (1982) Quantification of the contribution of various steps to the control of mitochondrial respira- tion. J Biol Chem 257, 2754–2757. 33 Tager JM, Groen AK, Wanders RJ, Duszynski J, West- erhoff HV & Vervoorn RC (1983) Control of mitochon- drial respiration. Biochem Soc Trans 11, 40–43.

20 Kholodenko BN (1988) How do external parameters control fluxes and concentrations of metabolites? An additional relationship in the theory of metabolic con- trol. FEBS Lett 232, 383–386.

34 Duszynski J, Groen AK, Wanders RAJ, Vervoorn RC & Tager JM (1982) Quantification of the role of the adenine nucleotide translocator in the control of mito- chondrial respiration in isolated rat-liver cells. FEBS Lett 146, 262–266. 21 Ainscow EK & Brand MD (1999) Quantifying elasticity analysis: how external effectors cause changes to meta- bolic systems. Biosystems 49, 151–159. 22 Schwenke WD, Soboll S, Seitz HJ & Sies H (1981)

Mitochondrial and cytosolic ATP ⁄ ADP ratios in rat liver in vivo. Biochem J 200, 405–408. 35 Bohnensack R, Kuster U & Letko G (1982) Rate-con- trolling steps of oxidative phosphorylation in rat liver mitochondria. A synoptic approach of model and experiment. Biochim Biophys Acta 680, 271–280. 36 Westerhoff HV, Plomp PJ, Groen AK, Wanders RJ,

23 Soboll S, Seitz HJ, Sies H, Ziegler B & Scholz R (1984) Effect of long-chain fatty acyl-CoA on mitochondrial and cytosolic ATP ⁄ ADP ratios in the intact liver cell. Biochem J 220, 371–376. Bode JA & van Dam K (1987) On the origin of the lim- ited control of mitochondrial respiration by the adenine nucleotide translocator. Arch Biochem Biophys 257, 154–169. 24 Akerboom TP, Bookelman H & Tager JM (1977)

Control of ATP transport across the mitochondrial membrane in isolated rat-liver cells. FEBS Lett 74, 50–54.

37 Gellerich FN, Bohnensack R & Kunz W (1983) Control of mitochondrial respiration. The contribution of the adenine nucleotide translocator depends on the ATP- and ADP-consuming enzymes. Biochim Biophys Acta 722, 381–391. 38 Wanders RJA, Groen AK, van Roermund CWT & 25 Hardie DG (2004) The AMP-activated protein kinase pathway – new players upstream and downstream. J Cell Sci 117, 5479–5487. 26 van den Berghe G, Bontemps F & Vincent MF

Tager JM (1984) Factors determining the relative contri- bution of the adenine nucleotide translocator and the ADP-regenerating system to the control of oxidative phosphorylation in isolated rat-liver mitochondria. Eur J Biochem 142, 417–424. (1988) Cytosolic purine 5¢-nucleotidases of rat liver and human red blood cells: regulatory properties and role in AMP dephosphorylation. Adv Enzyme Regul 27, 297–311. 39 Mildaziene V, Nauciene Z, Baniene R & Grigiene J 27 Bunger R & Soboll S (1986) Cytosolic adenylates and

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5301

(2002) Multiple effects of 2,2,5,5-tetrachlorbiphenyl on oxidative phosphorylation in rat liver mitochondria. Toxicol Sci 65, 220–227. adenosine release in perfused working heart. Comparison of whole tissue with cytosolic non-aqueous fractionation analyses. Eur J Biochem 159, 203–213.

J. Ciapaite et al.

Palmitoyl-CoA and control of mitochondrial function

fluxes in biotechnological processes. Biotechnol Bioeng 30, 101–107. 48 Echtay KS, Winkler E & Klingenberg M (2000) Coen- 40 Bradford MM (1976) A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72, 248–254. 41 Jensen PR, Westerhoff HV & Michelsen O (1993) zyme Q is an obligatory cofactor for uncoupling protein function. Nature 408, 609–613.

Excess capacity of H+-ATPase and inverse respiratory control in Escherichia coli. EMBO J 12, 1277–1282. 42 Bergmeyer HU (1974) Methods of Enzymatic Analysis. 49 Giersch C (1994) Determining elasticities from multiple measurements of steady-state flux rates and metabolite concentrations: theory. J Theor Biol 169, 89–99. Academic Press, New York, NY.

50 Hofmeyr JH, Cornish-Bowden A & Rohwer JM (1993) Taking enzyme kinetics out of control; putting control into regulation. Eur J Biochem 212, 833–837. 51 Westerhoff HV & Chen YD (1984) How do enzyme 43 Moore A, Dry IB & Wiskich JT (1988) Measurement of the redox state of the ubiquinone pool in plant mito- chondria. FEBS Lett 235, 76–80.

activities control metabolite concentrations? An addi- tional theorem in the theory of metabolic control. Eur J Biochem 142, 425–430.

44 Van den Bergen CW, Wagner AM, Krab K & Moore AL (1994) The relationship between electron flux and the redox poise of the quinone pool in plant mitochondria. Interplay between quinol-oxidizing and quinone-redu- cing pathways. Eur J Biochem 226, 1071–1078.

Supplementary material

is available

The following supplementary material online: Appendix S1. Equations S1 to S5, used for calculation of extramitochondrial AMP concentration. Appendix S2. Calculation of control coefficients.

This material is available as part of the online article

from http://www.blackwell-synergy.com

45 Liu Y, Fiskum G & Schubert D (2002) Generation of reactive oxygen species by the mitochondrial electron transport chain. J Neurochem 80, 780–787.

46 Hafner RP, Brown GC & Brand MD (1990) Analysis of the control of respiration rate, phosphorylation rate, proton leak rate and protonmotive force in isolated mitochondria using the ‘top-down’ approach of metabolic control theory. Eur J Biochem 188, 313–319.

FEBS Journal 273 (2006) 5288–5302 ª 2006 The Authors Journal compilation ª 2006 FEBS

5302

47 Westerhoff HV & Kell DB (1987) Matrix method for determining the steps most rate- limiting to metabolic