IF1 : Setting the Pace of the F1 Fo -ATP Synthase

Review

IF1: setting the pace of the F1Fo-ATP synthase

Michelangelo Campanella1,2, Nadeene Parker1, Choon Hong Tan1, Andrew M. Hall1 and Michael R. Duchen1

1 Department of Cell and Developmental Biology, Mitochondrial Biology Group, University College London, London, WC1E 6BT, UK 2 Royal Veterinary College, University of London, London, NW1 0TU, UK

When mitochondrial function is compromised and the establishes a proton-motive force, which has two mitochondrial membrane potential (Dcm) falls below components: a pH differential and an electrical membrane a threshold, the F1Fo-ATP synthase can reverse, potential (Dcm). The pH component of mitochondrial hydrolysing ATP to pump protons out of the mitochon- proton motive force is small relative to the membrane drial matrix. Although this activity can deplete ATP and potential and, hence, it is the latter that provides the precipitate cell death, it is limited by the mitochondrial predominant driving force for mitochondrial transport, 2+ protein IF1, an endogenous F1Fo-ATPase inhibitor. IF1, ADP phosphorylation, Ca accumulation and the import therefore, preserves ATP at the expense of Dcm. of nuclear-encoded, mitochondrial-localized proteins. Despite a wealth of detailed knowledge on the bio- Impaired mitochondrial function resulting from a variety chemistry of the interaction of IF1 and the F1Fo-ATPase, of different mechanisms will usually cause a decrease in little is known about its physiological activity. Emer- Dcm as a common endpoint. These include limited sub- ging research suggests that IF1 has a wider ranging strate or oxygen availability such as occurs in ischaemia impact on mitochondrial structure and function than (e.g. in a stroke or heart attack); genetic or acquired defects previously thought. in respiratory chain activity (due to mutations or to oxidative or nitrosative damage to mitochondrial respiratory Mitochondria as ATP consumers proteins); or a leak of protons back into the mitochondrial In most Biochemistry text books, mitochondria are matrix (through opening of the mitochondrial permeability described as ‘the powerhouse of the cell’. The bulk of transition pore [mPTP] [3,6] or the activation of uncoupling ATP in most mammalian cells comes from mitochondrial proteins; see Ref. [7] for a review). oxidative phosphorylation [1]. The reliable supply of ATP is The F1Fo-ATP synthase is the enzyme complex respon- fundamental to cell function; ATP is required for all the sible for ATP synthesis driven by oxidative phosphoryl- work that cells do: for muscle contraction, cell migration, ation [8,9]. The complex is an ancestral proton- secretion, and the maintenance of the ion gradients that translocating ATPase, a molecular motor that normally underlie membrane excitability. Although ATP availabil- operates as an ATP synthase in the mitochondrial inner ity is a basic cellular requirement, the role of mitochondria membrane in which ADP phosphorylation is driven by the in cell physiology extends far beyond even this. Mitochon- movement of protons down the electrochemical potential 2+ dria have essential roles in calcium (Ca ) homeostasis (for gradient established by respiration (Figure 1a[i]; a recent review see Ref. [2]), in free radical signalling [3], animations at: www.mrc-mbu.cam.ac.uk/research/atp- and they act as gatekeepers of cell death by harbouring synthase). The directionality of the enzyme is dictated both pro- and anti-apoptotic proteins [4]. When a system is by the balance between the bioenergetic parameters of free so fundamental to cell life and death, it follows inevitably energy available from the phosphorylation potential and that, when it is compromised, resultant impairment of cell from Dcm. In normally respiring mitochondria, the and tissue function will manifest as disease. Indeed, dis- removal of ATP by the adenine nucleotide translocase ordered mitochondrial function has been implicated in the (ANT) ensures that the intramitochondrial phosphoryl- pathogenesis of an array of major human diseases [5].A ation potential is held relatively low while Dcm is high, comprehensive understanding of the mechanisms that (estimated at between 150 and 180 mV negative to the govern mitochondrial homeostasis and that dictate cytosol), favouring ADP phosphorylation (i.e. ATP syn- responses to altered mitochondrial homeostasis is clearly thesis). However, when mitochondrial homeostasis is com- essential if we are to develop rational approaches to mana- promised, the situation can reverse. A decrease in Dcm ging these diseases. accompanied by an increase in the phosphorylation poten- Central to mitochondrial function is an electrochemical tial as glycolysis is upregulated (termed the Pasteur effect proton gradient across the mitochondrial inner membrane [10];seealsoRef.[11]) together with reversal of the ANT, that is established by the proton pumping activity of which imports glycolytic ATP, will favour ATP hydrolysis. the respiratory chain (Figure 1). The proton gradient Therefore, during mitochondrial dysfunction, the F1Fo- ATPase can run ‘backwards’, acting as an ATP-consuming Corresponding author: Duchen, M.R. ([email protected]) proton pump (Figure 1aii).

Figure 1. Role of F1Fo-ATP synthase in mitochondrial bio-energetics. F1Fo-ATP synthase activity and co-localisation with IF1 are depicted. (a) Cartoons illustrating the operations of the F1Fo-ATP synthase in (i) respiring mitochondria and (ii) mitochondria acting as ATP consumers. (i) A substrate (e.g. through the tricarboxylic acid cycle [TCA]) supplies reducing equivalents as NADH and FADH2 to the respiratory chain. The respiratory complexes transfer electrons from NADH to cytochrome c (cyt c)and hence to molecular oxygen. In the process, the respiratory complexes pump protons across the inner membrane, establishing a proton gradient expressed predominantly as an electrochemical potential (Dcm)of150 to 180 mV negative to the cytosol. This Dcm provides the energy to drive proton flux through the F1Fo-ATPsynthase (grey), promoting ATP synthesis. ADP is supplied through the adenine nucleotide translocase (ANT) which exports ATP to the cytosol. (ii) If Dcm is not maintained, then the F1Fo- ATP synthase can reverse, consuming ATP and acting as a proton translocating ATPase. The ANT reverses as well, transporting glycolytic ATP into the matrix (modified from Ref. [30] with permission). Panels (b) and (c) illustrate the mitochondrial co-localisation of IF1 (red) with the b-subunit of the F1Fo-ATP synthase (green) using immunofluorescence of the HeLa cell line (b) and intact rat kidney (c). DAPI staining of the nuclei is shown in blue.

The ability of the F1Fo-ATP synthase to reverse and act represent an important pathogenic mechanism in dis- as an ATPase during conditions of reduced membrane eases in which mitochondrial respiration is compromised. potential has been appreciated for many years [11].The These diseases include the lack of oxygen or substrate (as proton-pumping capacity of the F1Fo-ATPase has two in a stroke or heart attack), but also include many diseases functional consequences. First, when respiration is in which more subtle defects in mitochondria are impli- impaired, it can sustain Dcm, or at least slow its rate of cated, including Parkinson’s, Alzheimer’s and motor dissipation by counteracting proton leaks. Alternatively, neuron diseases and a large number of rare but debilitat- underconditionsinwhichthelossofDcm is extreme (e.g. if ing diseases that involve genetic defects affecting mito- the mPTP opens), when no amount of proton pumping chondrial proteins. wouldbeabletorestorethepotential,theenzymecanhave Both of these mechanisms have been described in cel- sufﬁcient activity to drive depletion of cellular ATP. lular models of pathology. Thus, in cells derived from a Indeed, ATP consumption by the F1Fo-ATPase could patient with a mitochondrial genetic defect (with clinical

344 Review Trends in Biochemical Sciences Vol.34 No.7

mitochondrial encephalopathy and lactic acidosis) carrying gene ATPIF1.IF1 binds to and inhibits the F1Fo-ATPase a mitochondrial DNA mutation affecting proton pumping activity under conditions of both matrix acidification [18] at complex I, Dcm collapsed when cells were treated with and ATP hydrolysis [19]. The mammalian ATPIF1 gene oligomycin, an inhibitor of the F1Fo complex [12]. This product contains 106–109 amino acids (depending on the simple experiment suggests that proton pumping by the species of origin). The first 25 amino acids (residues -25 to - F1Fo-ATPase maintains Dcm (albeit reduced) in these cells 1) represent a mitochondrial targeting presequence that is at the expense of glycolytic ATP, probably compounding cleaved within the mitochondria to form the mature IF1 the lactic acidosis observed in the patient. protein of 84 amino acids [9].IF1 is highly conserved The power of the F1Fo-ATPase to deplete ATP has been throughout evolution, with homologues found in birds, demonstrated most dramatically in isolated rat cardiomyo- nematodes, yeasts and plants (although yeast and plant cytes after exposure to a mitochondrial uncoupler or in IF1 show significant functional differences from mamma- response to mPTP opening. The cardiomyocytes shortened lian IF1). This degree of conservation suggests that IF1 is a to a rigor contracture within minutes of Dcm collapse, protein of major functional importance. Indeed, structure indicating ATP depletion. The rigor contracture was pre- is sufficiently conserved such that IF1 from one species can vented by oligomycin exposure, emphasizing the power of inhibit F1Fo-ATPase from another, albeit with varying the F1Fo-ATPase as an ATP consumer in a cell type in degrees of efficacy [20]. It was recently suggested that which mitochondria make up 40% of the cell volume IF1 might also localize to the plasma membrane where [3,12–14]. Grover et al. [15] demonstrated the clinical it is presumed to associate with an F1Fo-ATPase that has potential of inhibiting this ATPase activity; treatment with also been localized to the plasma membrane [21]. A calmo- a compound, known as BMS-199264, that inhibits ATPase dulin consensus binding motif is present in the middle of activity without impairing ATP synthase activity, substan- the IF1 protein [22], and it has been suggested that this tially decreased infarct size after coronary artery occlusion motif might dictate its plasma membrane localisation in in an isolated perfused rat heart preparation [15]. hepatocytes [22]. However, there remain many questions about a role (and even the presence) of the F1Fo-ATPase at IF1, the endogenous mitochondrial ATP hydrolase the plasma membrane and, therefore, this theme will not inhibitor be explored further in this review. Although the role of mitochondria in triggering cell death by IF1-mediated F1Fo-ATPase inhibition is optimal at a pH initiating the complex signalling pathways of apoptosis is of 6.7 [18], a condition achieved in the mitochondrial well defined [4,16], it is becoming clear that mitochondria matrix during severe ischaemia [23]. The action of IF1 could also accelerate progression towards necrotic cell death on F1Fo-ATP synthase activity, however, remains poorly through the simple mechanism of ATP depletion due to documented in the literature. Given the requirement for an ATPase activity during mitochondrial dysfunction. This electrochemical potential, this partly reflects the technical process is limited by an endogenous inhibitor protein known difficulties involved in the study of F1Fo-ATP synthase as IF1 – the inhibitory factor of the F1Fo-ATPase. Since its activity in contrast with relatively straightforward discovery [17], a wealth of information has been gathered measurements of ATP hydrolysis. Although some evidence about the biochemical and molecular structure of this small suggests that IF1 can inhibit synthase activity [24], the protein, and yet it seems to have been remarkably neglected importance of these observations remains unclear. in more physiological studies. The protein inhibits ATPase The detailed crystal structure of IF1-inhibited F1- activity in response to acidification of the mitochondrial ATPase has been solved [19,25], revealing two main fea- matrix [18], which will usually accompany inhibition of tures [19,25]. First, IF1 acts as a homodimer, simul- mitochondrial respiration (i.e. during hypoxia/ischaemia) taneously inhibiting two F1-ATPase units [25] and, and in response to the reversal of F1Fo-ATP synthase second, residues in the two protein complexes form numer- activity to act as an ATPase [19]. Remarkably, we know ous associations that involve several F1-ATPase subunits almost nothing about the relative expression levels of the [19]. Full association of IF1 with the F1Fo-ATPase is protein in different tissues or cell types, or about the phys- thought to occur only during ATP hydrolysis. Gledhill iological impact of varied IF1 expression levels, or the mech- et al. [19] suggest that low-affinity binding of IF1 to the anisms that regulate its expression. Currently, there are no surface of the F1 domain might occur [19] even in the animal models in which the gene is either overexpressed or ground state; however, the functional consequence of this knocked out; thus, the consequences of altered IF1 expres- activity is not clear. sion levels for cell or tissue function remain unknown. Dimerization of IF1 promotes dimerization of the F1- Therefore, some recent investigations have set out to explore ATPase during ATP hydrolysis as demonstrated either these issues by looking at the functional consequences of using blue native gel electrophoresis [26] or in the IF1- varying IF1 expression levels in cell lines. These data inhibited F1-ATPase crystal structure [25]. It has been suggest that IF1 not only inhibits ATPase activity, and so suggested that dimerization of the IF1-inhibited ATPase protects cells from ATP depletion in response to hypoxia, bus structure could help to stabilize the complex against the also IF1 appears to have a role in defining the conformation torque induced by ATP hydrolysis in the F1 domain, or it of the F1Fo-ATP synthase and mitochondrial cristae struc- might bring the F1-ATPase domains sufficiently close ture, as well as in regulating oxidative phosphorylation together (100 A˚ ) that they hinder each others’ rotation under normal physiological conditions. [27]. Dimerization of yeast F1Fo-ATP synthase occurs In 1963, Pullman and Monroy [17] discovered the mito- independently of IF1 [28]. Buzhynskyy et al. [27] used chondrial protein IF1 (inhibitor factor 1), encoded by the high-resolution atomic force microscopy topography to

345 Review Trends in Biochemical Sciences Vol.34 No.7

study F1Fo protein complex dimerization in native yeast cardiomyocytes against contracture, this experiment mitochondrial membranes. They described two popu- demonstrates that inhibition of F1Fo-ATPase activity is lations of F1Fo protein complex dimers, with the gamma protective to cells. This is consistent with the finding that ˚ ˚ subunits separated by 150 A and 100 A. This was inter- IF1 overexpression significantly reduced cell death in preted as representing active F1Fo-ATP synthase dimers response to oxygen and glucose deprivation [30]. and the IF1-inhibited F1Fo-ATPase dimers, respectively. In a schematic cartoon (Figure 2b) we have illustrated IF1-dependent dimerization of the F1Fo-ATP synthase the predicted impact of IF1 on changes in [ATP] and Dcm during normal respiration remains controversial [29].In during the progression of a period of ischaemia. In cells HeLa cells overexpressing IF1, blue native gel analysis with levels of IF1 sufficient to completely inhibit the revealed that the ratio of dimeric to monomeric forms of the ATPase activity (equivalent to the action of oligomycin), F1Fo protein complex was increased [30]. These findings Dcm collapses rapidly as soon as the respiratory chain is suggest that increased IF1 expression might promote F1Fo- inhibited, whereas [ATP] can be preserved for a consider- ATP synthase dimerization in mammalian cells. able period of time until other ATP consumers drive ATP depletion. This timing will vary depending on the glycolytic IF1 cell biology: functional consequences of altered capacity of the cell type and the activity of the ATP con- expression sumers. In marked contrast, in cells in which IF1 is absent, The functional consequences of genetic manipulation of IF1 Dcm can be maintained at a new steady state for prolonged protein levels were recently explored in cell lines (HeLa periods of time, but this occurs at the expense of cellular cells and muscle-derived C2C12 cells) [30], in which IF1 ATP, which becomes depleted more quickly. Once ATP is was either overexpressed by transient transfection or depleted, Dcm will collapse as there is no ATP left as a knocked down using small interfering RNA (siRNA). substrate for the F1Fo- ATPase. Two approaches were used to assay IF1-mediated inhi- These principles are readily established experimentally. bition of F1Fo-ATPase activity: (i) measurements of the Thus, CN inhibits electron transfer along the electron rate of ATP depletion after inhibition of oxidative phos- transport chain and hence its proton-pumping capacity, phorylation and glycolysis (halting all ATP synthesis) to leading to depolarisation of mitochondria. Recently, the assess ATP hydrolysis by the F1Fo-ATPase (Figure 2a); and extent of mitochondrial depolarisation in HeLa cells over- (ii) measurements of Dcm in the face of inhibition of oxi- expressing IF1 was shown to be greater than that seen in dative phosphorylation to assess the proton pumping wild-type cells. Conversely, when IF1 expression was sup- activity of the F1Fo-ATPase (Figures 2b and 3). pressed, Dcm was maintained at a new steady state for In the first of these assays, cellular ATP synthesis was prolonged periods of time [30]. These experiments are more completely inhibited using either (i) iodoacetic acid (IAA) to revealing than they might initially seem as they show that inhibit glycolysis together with sodium cyanide (CN )to endogenous IF1 protein is not necessarily expressed with a inhibit mitochondrial respiration, or (ii) IAA to inhibit fixed stoichiometry in relation to the F1Fo-ATP synthase. glycolysis and oligomycin to inhibit oxidative phosphoryl- That activity can be increased or decreased by genetic ation. After inhibition of all cellular ATP synthesis, the manipulations argues that there is room for regulation of rate of ATP depletion reflects the activity of all active ATP- expression of this protein in relation to that of the F1Fo consuming processes in the cell. When oxidative phos- protein complex. This idea is strengthened by several obser- phorylation is inhibited with CN , the F1Fo-ATPase vations. In central nervous system cultures containing both activity contributes to the global rate of ATP consumption. astrocytes and neurons, immunofluorescence measure- By contrast, when oxidative phosphorylation is inhibited ments showed that ratios between the expression levels of with oligomycin, the F1Fo-ATPase activity cannot contrib- the F1Fo-ATP synthase b-subunit and IF1 protein levels ute to ATP consumption. The difference between the two varied dramatically between the two cell types. Neurons, rates therefore gives a measure of the specific contribution with a relatively low b-subunit:IF1 ratio, exhibited rapid of the F1Fo-ATPase as an ATP consumer. The concen- loss of Dcm in response to CN , whereas astrocytes, with 2+ tration of free intracellular magnesium ([Mg ]c) increases high b-subunit:IF1 ratio, maintained Dcm, albeit at a 2+ as an index of ATP consumption as Mg is released upon reduced level. F1Fo-ATPase activity was clearly required ATP hydrolysis (Figure 2). This provides a useful assay at for the maintenance of Dcm because treatment of astrocytes the level of single cells, where direct measurements of with oligomycin caused rapid mitochondrial depolarisation [ATP] are not really practical. After complete ATP (Figure 3b[ii]). Similarly, in a series of experiments using depletion, cells underwent lysis; this occurred between multiphoton imaging of rat kidney slices, the mitochondria 45 and 75 min in HeLa cells. In cells lacking IF1, the initial in the proximal tubules showed a relatively low ratio of the rate of ATP consumption was significantly faster in the F1Fo-ATP synthase b-subunit:IF1 and the Dcm depolarised presence of CN and IAA, and the time to lysis was rapidly in response to respiratory inhibition; by contrast, in significantly shorter compared with wild-type cells (grey the more glycolytic neighbouring distal tubules, which versus black; Figure 2). In the presence of oligomycin and showed a high b-subunit:IF1 ratio, Dcm was maintained IAA, the rate of ATP consumption was slower and the time for long periods of time despite respiratory inhibition [31] to lysis in wild-type cells was substantially delayed (red (Figure 3). It is interesting to note that both neurons and the trace; Figure 2), indicating the overall ATP-consuming proximal tubules are highly oxidative [32,33], and as a result activity of the F1Fo-ATPase. Together, these experiments the kidney proximal tubules, for example, are clinically show that endogenous IF1 functionally inhibits ATPase more vulnerable to mitochondrial dysfunction than activity in intact cells. Similar to the protection of rat distal tubules [34]. It could be advantageous to have

346 Review Trends in Biochemical Sciences Vol.34 No.7

Figure 2. Impact of IF1 on ATP consumption and Dcm.IF1 inhibits cellular ATP consumption by F1Fo-ATPase and promotes Dcm collapse. (a) Experimental traces illustrate 2+ the impact of IF1 on the rate of ATP depletion in HeLa cells after inhibition of ATP synthesis. The fluorescent dye MagFura was used to measure free intracellular [Mg ]c, 2+ 2+ which increases as ATP is hydrolysed. Most intracellular Mg is bound to ATP (with an affinity for ADP about ten times lower than that for ATP). Free [Mg ]c rises until cells undergo lysis (arrows). In cells exposed to IAA and CN (grey line), blocking both glycolysis and respiration, the rate of ATP depletion reflects the summed activity of all ATP consumers. In cells exposed to IAA and oligomycin (oligo; red line), blocking both glycolysis and oxidative phosphorylation, the rate of ATP depletion reflects the summed activity of all ATP consumers now excluding the F1Fo-ATPase. The differences between the two responses are attributable to ATP depletion driven by the F1Fo-ATPase. 2+ Representative traces of relative [Mg ]c with time are shown for HeLa cells treated with CN (black line in WT cells, grey line in cells lacking IF1) or oligomycin (red line), in wild-type (WT; black line) or IF1 suppressed (-IF1; grey and red lines) cells. The blue-shaded area indicates the effective inhibition of ATPase activity by endogenous IF1, whereas the grey-shaded area shows that the total difference in time is due to uninhibited F1Fo-ATPase activity in the absence of IF1 (modified from Ref. [30] with permission). (b) Schematic to show the changes in [ATP] (solid lines) and Dcm (dashed lines) during a period of ischaemia in cells containing either maximally inhibitory levels of IF1 (red) or cells in which IF1 is absent (black). In the latter, Dcm is maintained at a new steady state by the proton pumping of the F1Fo-ATPase. ATP falls progressively, driven partly by the F1Fo-ATPase itself. Once ATP is depleted, the Dcm collapses as it can no longer be maintained by the ATPase. By contrast, in cells expressing IF1, Dcm falls quickly whereas ATP depletion is significantly delayed.

alowb-subunit:IF1 ratio to prevent rapid ATP consumption levels [37]. It should be emphasized that measurements of during respiratory inhibition. IF1 protein levels alone are almost meaningless without Although it seems clear that IF1 expression levels equivalent measurements of F1Fo-ATP synthase protein relative to the F1Fo-ATP synthase vary between tissues concentration or activity. It is the expression of IF1 relative and even between cell types within given tissues [30],itis to that of the F1Fo-ATP synthase that dictates the func- not clear how its expression is modulated. Some evidence tional outcome. It seems probable that IF1 expression is suggests that IF1 might be downregulated in septic shock regulated independently from that of the F1Fo-ATP syndrome [35]. The same authors have suggested that synthase and it seems important to establish the mechan- hypoxia inducible factor 1-a (HIF1-a) might regulate IF1 ism by which this is achieved. expression levels [36]. More recently, immediate early response gene X-1 (IEX-1) was implicated in targeting Unexpected roles for IF1 as a determinant of IF1 for degradation. IEX-1 deletion stabilized IF1, thereby mitochondrial structure and function reducing ATPase activity and suggesting a mechanism for All the data that we have discussed suggest that the post-translational regulation of steady state IF1 protein canonical actions of IF1 described in isolated mitochondria,

347 Review Trends in Biochemical Sciences Vol.34 No.7

Figure 3. The ratio of expression of F1Fo-ATP synthase and IF1 varies between different cell types. F1Fo-ATP synthase and IF1 expression levels are shown for co-cultures of neurons and astrocytes and in slices of rat kidney. (a) Representative immunofluorescence images of rat primary neurons (n) and astrocytes (a) in coculture, labelled with an antibody against the b-subunit of the F1Fo-ATP synthase (green) or IF1 (red). A colour-coded image of the ratio between F1Fo-ATP synthase b-subunit:IF1 is shown. (b)(i) Quantification (mean SD) of the F1Fo-ATP synthase b-subunit:IF1 ratio (*p <0.005). (ii) Relative Dcm was measured in cells treated with 1 mM CN , with or without 2.5 mg/ ml oligomycin, by the retention of tetramethyl rhodamine methyl ester (TMRM) [30] and normalized to maximally polarized values (modified from Ref. [30] with permission). (c) In sections of rat kidney, immunostaining of the b-subunit of the F1Fo-ATP synthase (green) and IF1 (red) are shown along with a colour-coded image of the ratio between the two images showing the b-subunit:IF1 ratio. Quantification (mean SD) of b-subunit:IF1 ratio from 15–18 tubules is shown (**p <0.001) (d)(i). (d)(ii) Changes in Dcm in response to CN and hypoxia were measured using TMRM with multiphoton imaging of rat kidney slices (modified from Ref. [31] with permission). Abbreviations: CD, collecting duct; DT, distal tubule; FCCP, carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone; PT, proximal tubule. submitochondrial particles or isolated enzyme prep- after suppression of IF1 [30]. That these effects were arations [18,38] are maintained in cell culture models; normalised in the presence of oligomycin implicated IF1- IF1 thus behaves as expected from experiments with dependent modulation of F1Fo-ATPase activity. Mitochon- isolated enzyme in a test tube. Recent work also suggests drial NADH was more oxidised in the IF1-overexpressing that IF1 might have unexpected roles in regulating cells, consistent with an increased rate of respiration, mitochondrial function. In normally respiring cells, IF1 whereas the IF1 siRNA treated cells showed the opposite overexpression reduced Dcm, whereas Dcm was increased effect [30]. Again, these differences were normalised by

348 Review Trends in Biochemical Sciences Vol.34 No.7

Figure 4.IF1 promotes F1Fo-ATP synthase dimerization. IF1-mediated F1Fo-ATP synthase dimerization alters membrane surface-charge distribution. (a) Model of the dimeric F1Fo-ATP synthase complex. It is suggested that IF1 (as labelled, yellow) might promote an association between F1 subunits as shown (adapted from Ref. [42] with permission). The components representing the F1 ‘motor’ of the synthase are shown in blue and the Fo component as green. Subunits are indicated. OSCP, oligomycin sensitivity conferring protein. (b) 3D reconstruction of the dimeric ATPase and its membrane derived from cryo electron tomography data (adapted from Ref. [39] with permission). The protruding ATP synthase components are shown as yellow and the crista membrane as grey. (c) A model proposed by Strauss et al. [39] showing accumulated surface charge forced by the high membrane curvature, which is, in turn, generated by the F1Fo-ATP synthase dimeric structure (shown as yellow; adapted from Ref. [40] with permission).

Figure 5.IF1 expression level modulates mitochondrial cristae structure. IF1 expression alters the structure and overall number of mitochondrial cristae. (a) Representative mitochondrial profiles created from electron micrographs of HeLa cells, showing mitochondria from control cells, and cells with IF1 overexpressed (+IF1) or with IF1 suppressed using siRNA (IF1). (b) The number of cristae per mitochondrion has been quantified (mean SD, n = 5) and plotted as a histogram (modified from Ref. [30] with permission); *p <0.005. oligomycin. At the time, the available literature could not an increase in electrostatic charge at the apex of the account for these observations, with the exception that IF1 cristae formed by F1Fo-ATP synthase dimers (Figure 4c) has been implicated in the formation of F1Fo protein com- could increase proton flux through the F1Fo-ATP synthase plex dimers (Figure 4a) [26,30] and that it has been and so increase the rate of ADP phosphorylation, although suggested that F1Fo-ATP synthase dimerization could it is not yet clear how or whether this could be sustained at improve the efficiency of ATP synthesis [39]. a steady state. Consistent with this model, the EM ultra- How or why might dimerization influence F1Fo-ATP structure was transformed by alterations in IF1 expres- synthase activity? A recent study using high-resolution sion levels; cells overexpressing IF1 showed an increase in electron microscopy (EM) showed that F1Fo-ATP synthase the number and density of cristae per mitochondrion, dimers tend to align ‘in ribbons’ along mitochondrial whereas mitochondria in cells with reduced IF1 expression membranes with a high curvature [39] (Figure 4b). The had few cristae [30] (Figure 5). authors suggested that the insertion of the dimeric enzyme complex into the membrane forces the high cur- Concluding remarks and future perspectives vature of the membrane. The idea that the F1Fo-ATP The data that we have discussed suggest that IF1 might synthase might define cristae structure has been act as a ‘coupling factor’, increasing the relative efficiency suggested for some years [40,41]. Indeed, Minauro-San- of mitochondrial oxidative phosphorylation. Although this miguel et al. [42] have already suggested that IF1 might model remains very speculative, it fits with the available shape cristae structure by promoting F1Fo-ATP synthase expression data. In the cells examined thus far, highly dimerization. A mathematical model [39] showed that the oxidative cell types have lower b-subunit:IF1 ratios high curvature of the mitochondrial inner membrane compared with their glycolytic neighbours (neurons ver- creates areas of concentrated surface charge equivalent sus astrocytes; proximal versus distal renal tubules; to a pH gradient of 0.5 units. The authors suggested that Figure 3) [30,31]. More work will be required to establish

349 Review Trends in Biochemical Sciences Vol.34 No.7 the generality of this observation and to explore the 17 Pullman, M.E. and Monroy, G.C. (1963) A naturally occurring inhibitor regulation of expression and the functional properties of of mitochondrial adenosine triphosphatase. J. Biol. Chem. 238, 3762– 3769 IF1. We need to understand its role in intact tissues and 18 Rouslin, W. and Pullman, M.E. (1987) Protonic inhibition of the how it impacts on normal tissue physiology and patho- mitochondrial adenosine 50-triphosphatase in ischemic cardiac physiology. Obvious next steps will include explorations of muscle. Reversible binding of the ATPase inhibitor protein to the mitochondrial ATPase during ischemia. J. Mol. Cell. Cardiol. 19, the impact of altered IF1 expression levels in transgenic animal models. What happens in the heart when IF is 661–668 1 19 Gledhill, J.R. et al. (2007) How the regulatory protein, IF1, inhibits F1- knocked out or overexpressed, both in terms of normal ATPase from bovine mitochondria. Proc. Natl. Acad. Sci. U. S. A. 104, physiology and in response to cellular injury? What are the 15671–15676 physiological roles of IF1 in other tissues? To what extent 20 Rouslin, W. et al. (1995) ATPase activity, IF1 content, and proton conductivity of ESMP from control and ischemic slow and fast heart- does IF1 expression dictate the balance between oxidative metabolism and glycolysis? What are the factors that rate hearts. J. Bioenerg. Biomembr. 27, 459–466 21 Contessi, S. et al. (2007) IF distribution in HepG2 cells in relation to dictate the expression levels of IF in different cell types 1 1 ecto-F0F1ATPsynthase and calmodulin. J. Bioenerg. Biomembr. 39, or tissues? These questions await an answer. It seems that 291–300 the role of IF1 extends beyond that envisaged in the 22 Contessi, S. et al. (2005) Identiﬁcation of a conserved calmodulin- existing literature and that we still have a great deal to binding motif in the sequence of F0F1 ATP synthase inhibitor protein. J. Bioenerg. Biomembr. 37, 317–326 learn about this fascinating little protein. 23 Rouslin, W. (1983) Protonic inhibition of the mitochondrial oligomycin- sensitive adenosine 50-triphosphatase in ischemic and autolyzing cardiac Acknowledgements muscle. Possible mechanism for the mitigation of ATP hydrolysis under We thank the Wellcome Trust (www.wellcome.ac.uk), the Medical nonenergizing conditions. J. Biol. Chem. 258, 9657–9661 Research Council (www.mrc.ac.uk), Kidney Research UK 24 Tuena Gomez-Puyou, M.T. et al. (1988) Synthesis and hydrolysis of ATP (www.kidneyresearchuk.org) and Royal Veterinary College internal by the mitochondrial ATP synthase. Biochem. Cell Biol. 66, 677–682 funds for supporting the work of our laboratory. We also thank Andrey 25 Cabezon, E. et al. (2003) The structure of bovine F1-ATPase in complex Abramov and Andrew Tinker for their invaluable contributions to the with its regulatory protein IF1. Nat. Struct. Biol. 10, 744–750 work and for useful discussion and Werner Ku¨ hlbrandt and Jose´.J. 26 Garcia, J.J. et al. (2006) The inhibitor protein (IF1) promotes

Garcia for allowing us to use their material to illustrate this review. dimerization of the mitochondrial F1F0-ATP synthase. Biochemistry 45, 12695–12703 References 27 Buzhynskyy, N. et al. (2007) Rows of ATP synthase dimers in native 1 Rich, P. (2003) Chemiosmotic coupling: The cost of living. Nature 421, mitochondrial inner membranes. Biophys. J. 93, 2870–2876 583 28 Wittig, I. et al. (2008) Characterization of domain interfaces in 2 Szabadkai, G. and Duchen, M.R. (2008) Mitochondria: the hub of monomeric and dimeric ATP synthase. Mol. Cell. Proteomics 7, 995– cellular Ca2+ signalling. Physiology (Bethesda) 23, 84–94 1004 2+ 3 Duchen, M.R. (2000) Mitochondria and Ca in cell physiology and 29 Tomasetig, L. et al. (2002) Dimerization of F0F1ATP synthase from pathophysiology. Cell Calcium 28, 339–348 bovine heart is independent from the binding of the inhibitor protein 4 Kroemer, G. et al. (1997) Mitochondrial control of apoptosis. Immunol. IF1. Biochim. Biophys. Acta 1556, 133–141 Today 18, 44–51 30 Campanella, M. et al. (2008) Regulation of mitochondrial structure and 5 Duchen, M.R. (2004) Mitochondria in health and disease: perspectives function by the F1Fo-ATPase inhibitor protein, IF1. Cell Metab. 8, 13–25 on a new mitochondrial biology. Mol. Aspects Med. 25, 365–451 31 Hall, A.M. et al. (2009) Multiphoton imaging reveals differences in 6 Irwin, W.A. et al. (2003) Mitochondrial dysfunction and apoptosis mitochondrial function between nephron segments. J. Am. Soc. in myopathic mice with collagen VI deﬁciency. Nat. Genet. 35, Nephrol. 20, 1293–1302 367–371 32 Bagnasco, S. et al. (1985) Lactate production in isolated segments of the 7 Affourtit, C. et al. (2007) Novel uncoupling proteins. Novartis Found. rat nephron. Am. J. Physiol. 248, F522–F526 Symp. 287, 70–80 33 Magistretti, P.J. (2006) Neuron-glia metabolic coupling and plasticity. 8 Boyer, P.D. (1997) The ATP synthase–a splendid molecular machine. J. Exp. Biol. 209, 2304–2311 Annu. Rev. Biochem. 66, 717–749 34 Niaudet, P. and Rotig, A. (1997) The kidney in mitochondrial 9 Walker, J.E. et al. (1987) ATP synthase from bovine mitochondria: cytopathies. Kidney Int. 51, 1000–1007 sequences of imported precursors of oligomycin sensitivity conferral 35 Huang, L.J. et al. (2007) Suppression of mitochondrial ATPase protein, factor 6, and adenosinetriphosphatase inhibitor protein. inhibitor protein (IF1) in the liver of late septic rats. Biochim. Biochemistry 26, 8613–8619 Biophys. Acta 1767, 888–896 10 Krebs, H.A. (1972) The Pasteur effect and the relations between 36 Huang, L-J. and Yang, R-C. (2008) Hypoxia-inducible factor-1a respiration and fermentation. Essays Biochem. 8, 1–34 regulates the hepatic mitochondrial ATPase inhibitory protein (IF1) 11 David, G. and Nicholls, S.J.F. (2002) Bioenergetics. Academic Press [abstract]. FASEB J. 22, 1190.13 12 McKenzie, M. et al. (2007) Mitochondrial ND5 gene variation 37 Shen, L. et al. (2008) IEX-1 targets mitochondrial F1Fo-ATPase inhibitor associated with encephalomyopathy and mitochondrial ATP for degradation. Cell Death. Differ. DOI:10.1038/ cdd.2008.184. consumption. J. Biol. Chem. 282, 36845–36852 38 Green, D.W. and Grover, G.J. (2000) The IF1 inhibitor protein of the 13 Allue, I. et al. (1996) Evidence for rapid consumption of millimolar mitochondrial F1F0-ATPase. Biochim. Biophys. Acta 1458, 343–355 concentrations of cytoplasmic ATP during rigor-contracture of 39 Strauss, M. et al. (2008) Dimer ribbons of ATP synthase shape the inner metabolically compromised single cardiomyocytes. Biochem. J. 319, mitochondrial membrane. EMBO J. 27, 1154–1160 463–469 40 Allen, R.D. et al. (1989) An investigation of mitochondrial inner 14 Leyssens, A. et al. (1996) The relationship between mitochondrial membranes by rapid-freeze deep-etch techniques. J. Cell Biol. 108, state, ATP hydrolysis, [Mg2+]i and [Ca2+]i studied in isolated rat 2233–2240 cardiomyocytes. J. Physiol. 496, 111–128 41 Giraud, M.F. et al. (2002) Is there a relationship between the 15 Grover, G.J. et al. (2004) Excessive ATP hydrolysis in ischemic supramolecular organization of the mitochondrial ATP synthase and

myocardium by mitochondrial F1F0-ATPase: effect of selective the formation of cristae? Biochim. Biophys. Acta 1555, 174–180 pharmacological inhibition of mitochondrial ATPase hydrolase 42 Minauro-Sanmiguel, F. et al. (2005) Structure of dimeric mitochondrial activity. Am. J. Physiol. Heart Circ. Physiol. 287, H1747–H1755 ATP synthase: novel F0 bridging features and the structural basis of 16 Jacobson, M.D. and McCarthy, N., eds (2002) Apoptosis: The mitochondrial cristae biogenesis. Proc. Natl. Acad. Sci. U. S. A. 102, Molecular Biology of Programmed Cell Death, Oxford University Press 12356–12358

350