Targeted polyphosphatase expression alters mitochondrial ...

Targeted polyphosphatase expression alters mitochondrial metabolism and inhibits calcium-dependent cell death Andrey Y. Abramov*, Cresson Fraley†‡, Catherine T. Diao§, Robert Winkfein§, Michael A. Colicos§, Michael R. Duchen*, Robert J. French§, and Evgeny Pavlov§¶ *Department of Physiology and Mitochondrial Biology Group, University College London, Gower Street, London WC1E 6BT, United Kingdom; †Department of Biochemistry, Stanford University School of Medicine, 279 West Campus Drive, Stanford, CA 94305-5307; ‡Metis Biosciences, Inc., 6969 Bonny Doon Road, Santa Cruz, CA 95060-9754; and §Department of Physiology and Biophysics, University of Calgary, 3330 Hospital Drive N.W., Calgary, AB, Canada T2N 4N1 Communicated by Arthur Kornberg, Stanford University School of Medicine, Stanford, CA, September 20, 2007 (received for review June 7, 2007)

mitochondria 兩 permeability transition 兩 polyphosphate 兩 ␤-amyloid peptide 兩 necrosis

T

he chemical and physical properties of polyphosphate (polyP), including its high negative charge and its ability to form complexes with Ca2⫹ and to form high energy bonds, underlie its potential to play an important role in cell metabolism. Significant amounts of polyP have been found in bacteria and in lower eukaryotes. In those organisms, it provides energy storage and a reserve pool of inorganic phosphate, participates in regulation of gene expression, protects cells from the toxicity of heavy metals by forming complexes with them, and participates in channel formation through assembly into complexes with Ca2⫹ and polyhydroxybutyrate (PHB) (polyP/Ca2⫹/PHB complex) (1, 2) and possibly through interaction with channelforming proteins (3). PolyP has also been found in all higher eukaryotic organisms tested, where it is localized in various subcellular compartments, including mitochondria (4). Furthermore, mitochondrial polyP can form polyP/Ca2⫹/PHB complexes (5) with ion-conducting properties similar to those of native mitochondrial permeability transition pore (mPTP) (6). mPTP opening or formation in the mitochondrial inner membrane is believed to underlie the Ca2⫹induced permeability transition (PT), a phenomenon that causes inner membrane depolarization and disruption of ATP synthesis and plays a central role during various types of necrotic and apoptotic cell death (7). The molecular composition of the conducting pathway of mPTP is currently not well defined. Recently, we have raised the possibility that, in vivo, the polyP/ Ca2⫹/PHB complex might comprise the ion-conducting part of the mPTP complex (6). If so, mitochondrial polyP should be essential for mPTP opening/formation. Here, we examine the involvement of polyP in normal mitochondrial function and in PT development during stress. To this end, we specifically reduced levels of mitowww.pnas.org兾cgi兾doi兾10.1073兾pnas.0708959104

chondrial polyP by targeted expression of yeast exopolyphosphatase scPPX1 (8). We found that polyP affects both mitochondrial metabolism and mitochondrial Ca2⫹ accumulation. Furthermore, reduction of mitochondrial polyP levels was profoundly protective, dramatically reducing the probability of Ca2⫹-induced PT. Associated with this PT inhibition, cultured cell lines, as well as primary cocultures of neurons and astrocytes, were protected from stressinduced death. Results To test the generality of effects of PPX expression, we examined several cell lines, including HepG2 (hepatic carcinoma cells), HEK293 (human embryonic kidney), and C2C12 (undifferentiated mouse myoblasts), as well as primary cultures of astrocytes and neurons from rat brain. In most cases, illustrations are presented for HepG2 cells, but data did not differ significantly among cell types. Expression of a Polyphosphatase (PPX) in Mitochondria. Targeting and functional activity of the heterologously expressed enzyme. To decrease

the amount of polyP specifically in mitochondria, we constructed DNA encoding a fusion protein, MGP, composed of mitochondrially targeted GFP protein (MTS-GFP), and an exopolyphosphatase enzyme from yeast (scPPX1) that specifically hydrolyzes polyP into inorganic phosphate (8). Expression of MGP in mitochondria of transiently transfected cells was confirmed by confocal microscopy. Fig. 1a shows colocalization of the green fluorescent signal from MGP protein with the red mitochondrial TMRM signal. We confirmed the activity of the expressed enzyme by measuring polyphosphatase (PPX) activity of mitochondrial lysate from cells transfected with MGP. PPX activity, estimated from the rate of hydrolysis of synthetic poly 32P into orthophosphate, was ⬇500,000 pmol of Pi released per min/mg of mitochondria lysate protein. No detectable endogenous PPX activity was seen in lysate from mitochondria of nontransfected cells (Fig. 1b), whereas in the presence of the lysate containing MGP protein, 95% of polyP was converted into Pi within 20 min (Fig. 1c). Author contributions: A.Y.A., C.F., C.T.D., R.W., M.A.C., M.R.D., R.J.F., and E.P. designed research; A.Y.A., C.F., C.T.D., R.W., M.A.C., and E.P. performed research; A.Y.A., C.F., M.R.D., R.J.F., and E.P. analyzed data; and A.Y.A., C.F., M.R.D., R.J.F., and E.P. wrote the paper. The authors declare no conflict of interest. Abbreviations: PolyP, polyphosphate; mPTP, mitochondrial permeability transition pore; PT, permeability transition; MTS-GFP, mitochondrially targeted GFP protein; MGP, mitochondrially targeted GFP linked to exopolyphosphatase enzyme; ␤A, ␤-amyloid peptide; PPX, polyphosphatase. ¶To

whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0708959104/DC1. © 2007 by The National Academy of Sciences of the USA

PNAS 兩 November 13, 2007 兩 vol. 104 兩 no. 46 兩 18091–18096

CELL BIOLOGY

Polyphosphate (polyP) consists of tens to hundreds of phosphates, linked by ATP-like high-energy bonds. Although polyP is present in mammalian mitochondria, its physiological roles there are obscure. Here, we examine the involvement of polyP in mitochondrial energy metabolism and ion transport. We constructed a vector to express a mitochondrially targeted polyphosphatase, along with a GFP fluorescent tag. Specific reduction of mitochondrial polyP, by polyphosphatase expression, significantly modulates mitochondrial bioenergetics, as indicated by the reduction of inner membrane potential and increased NADH levels. Furthermore, reduction of polyP levels increases mitochondrial capacity to accumulate calcium and reduces the likelihood of the calcium-induced mitochondrial permeability transition, a central event in many types of necrotic cell death. This confers protection against cell death, including that induced by ␤-amyloid peptide, a pathogenic agent in Alzheimer’s disease. These results demonstrate a crucial role played by polyP in mitochondrial function of mammalian cells.

Fig. 1. Mitochondrial localization and functional activity of the MTS-GFP-PPX polyphosphatase construct (MGP). (a) Confocal images showing green signal from MGP (Upper Left), red signal from TMRM located in mitochondria (Upper Right), superposed MGP and TMRM images (Lower Left), and intensity profiles for MGP and TMRM (Lower Right), which were collected along the line of the red arrow (Lower Left). (b and c) Thin-layer chromatograms of mitochondrial lysates, from nontransfected controls and transfected cells, incubated in the presence of poly 32P. Activity is indicated by the conversion of polyP to Pi during the course of the incubation (uppermost spots in each lane). (b) Lack of endogenous PPX activity in lysates from nontransfected cells. Chromatograms showed no detectable hydrolysis of polyP until the addition of purified wild-type PPX after 45 min of incubation (last lane). (c) Polyphosphatase (PPX) activity of the MGP construct.

Reduction of polyphosphate levels in mitochondria of living cells. Be-

cause of relatively low amounts of material available, we were not able to estimate the decrease of polyP using enzymatic assay on isolated mitochondria. Thus, we compared levels of polyP in control and MGP-expressing cells by measuring fluorescence intensity of DAPI (Fig. 2). Fluorescence of the DAPI–polyP complex has a broad emission spectrum with a peak at ⬇525 nm and can be distinguished from fluorescence of free DAPI or DAPI-DNA, which have emission maxima at ⬇460 nm (Fig. 2b), (9, 10). Using confocal microscopy, we measured DAPI-polyP fluorescence (at ⬎580 nm) in control and PPX-expressing cells, identified by green GFP fluorescence. The DAPI-polyP signal varied significantly within a single population of cells. In some cells, the maximal signal appeared in mitochondria, whereas in others, the signal was stronger in the nuclei (not shown). For the following experiments, we selected cells with a low nuclear DAPI-polyP signal for analysis. Fig. 2a shows that the DAPIpolyP signal (in blue) is mostly colocalized with the TMRM signal from mitochondria. From this figure, we estimate that

⬎70% of DAPI-polyP signal merges with that of TMRM, indicating that it originates from mitochondria. Mitochondrial expression of the MGP enzyme in HepG2, C2C12 or HEK293 cells resulted in a significantly reduced DAPI fluorescence (P ⬍ 0.001 for all three cell types). Fluorescence intensities, in arbitrary units, were: for HepG2 cells, 3,300 ⫾ 200 (controls, n ⫽ 53 cells), 990 ⫾ 80 (MGP, n ⫽ 47) (see Fig. 2c); for C2C12 cells, 2,900 ⫾ 200 (controls, n ⫽ 62), 1,130 ⫾ 50 (MGP, n ⫽ 48); and for HEK293 cells, 3,000 ⫾ 200 (controls, n ⫽ 68), 1,240 ⫾ 70 (MGP, n ⫽ 61). DAPI-polyP fluorescence in cells expressing only MTS-GFP, was indistinguishable from that of control levels (Fig. 2c). Thus, recombinant MGP is specifically expressed in mitochondria, and it reduces levels of polyP in living cells, as indicated by DAPI fluorescence. Functional Changes in Mitochondria of Cells Expressing PPX. As the transfection efficiency of the MGP was low, we were unable to assess respiratory activity in cell populations using conventional respirometry. To define the effects of a reduction in mitochondrial polyP, we measured the following indicators of mitochondrial

Fig. 2. Mitochondrially targeted expression of polyphosphatase (MGP) reduces levels of polyP (mouse C2C12 cells; similar results were obtained with HepG2 cells). (a) In each set of four images, three separate color channels plus their superposition are shown: green, GFP fluorescence indicating MGP expression (I); red, TMRM signal reflecting inner membrane potential (II); blue, DAPI-polyP signal (III); and superposition of the three individual channels (IV). Data are shown from an MGP-expressing cell (Left) and a nonexpressing control (Right). Below the images are the intensity profile traces (green, MGP; red, TMRM; blue, DAPI) showing fluorescence intensity along the red arrows in image IV for each cell. Note the reduced DAPI emission associated with MGP expression, reflecting a reduction in polyP. (b) Differing DAPI fluorescence emission spectra recorded from nuclear (solid line) and mitochondrial (dotted line) regions, excitation wavelength, 405 nm. These spectra resemble the distinct emissions previously reported for pure DNA and for polyP, respectively, in test-tube experiments (see Results and SI Methods). (c) Relative intensity of DAPI-polyP fluorescence (integrated, ⬎580 nm), expressed as a percentage of the signal from nontransfected control cells (100%). Cells expressing MGP show a reduction to ⬇30%, whereas the signal from cells expressing only MTS-GFP is indistinguishable from nontransfected control cells. **, P ⬍ 0.001. 18092 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0708959104

Abramov et al.

function in control and MGP-expressing cells: mitochondrial membrane potential, the redox state of NADH, and the effects of interventions focused on specific complexes in the electron transport chain. Reduction of mitochondrial membrane potential. The inner mitochondrial membrane potential (⌬⌿m) is reflected by the intensity of the fluorescent signal from TMRM. When TMRM (20 nM) is added to the bath, it concentrates in the mitochondrial matrix, relative to the cytoplasm, according to the Nernst equation. In cells expressing MGP, ⌬⌿m was significantly reduced. For HepG2 cells, the intensity of the TMRM signal was reduced by 51 ⫾ 4% from control levels (n ⫽ 41 control cells; n ⫽ 39 MGP cells; P ⬍ 0.001; see Fig. 3a). For HEK293 cells, the reduction from controls was 43 ⫾ 4% (n ⫽ 36 control cells; n ⫽ 44 MGP cells; P ⬍ 0.001). Redox state of NADH. A decrease in mitochondrial membrane potential may result from an increased leak, decreased substrate supply or decreased respiratory activity. It is possible to differentiate among different causes by measuring NADH redox state. For example, increased leak would be associated with increased oxidation of NADH (decreased fluorescence), whereas impaired respiration would be associated with a more reduced state (increased fluorescence). We observed that reduction of polyP levels by MGP expression was associated with an increase in NADH autofluorescence, suggesting an increase in the reduced state of NADH, the main substrate of the respiratory chain (Fig. 3b). The resting level of NADH autofluorescence in the cells was expressed as a ‘‘redox index,’’ a function of the maximally oxidized and maximally reduced signals. These limits were estimated from the response to 1 ␮M FCCP (stimulates maximal respiration, completely oxidizing the mitochondrial NADH pool), taken as 0%, and the response to 1 mM NaCN (inhibits respiration and promotes maximal NADH reduction), taken as 100%. In response to polyP depletion by MGP, the redox index in HepG2 cells increased from 48 ⫾ 4% in control (n ⫽ 31 cells) to 70 ⫾ 6% (n ⫽ 34 cells, P ⬍ 0.05). Thus, polyP depletion most likely decreases potential by inhibition of mitochondrial respiration, suggesting that polyP plays a role in the maintenance of normal mitochondrial respiration. PolyP depletion impairs function of the mitochondrial respiratory chain.

Systematic provision of substrates to complex I, II, and IV was used to determine where, in the chain, activity might be impaired. Supporting information (SI) Fig. 6a demonstrates changes in NADH autofluorescence in response to substrates for mitochondrial complex I, pyruvate (5 mM), and glutamate (5 mM) in control (n ⫽ 39) and MGP (n ⫽ 42) cells. The two substrates induced Abramov et al.

PolyP in Mitochondrial Ca2ⴙ Transport and the Permeability Transition (PT). Ca2ⴙ accumulation and PT in permeabilized cells. Mitochondrial

calcium uptake is a key regulator of mitochondrial function and of cell death. Furthermore, polyP might be expected to provide binding sites for calcium and thus contribute to calcium buffering within mitochondria. We therefore designed experiments to explore the impact of MGP expression on mitochondrial calcium uptake. To have complete control of the calcium concentrations bathing the mitochondria, we chose to preempt cytoplasmic calcium signals and calcium buffering as variables. Cells were exposed to low concentrations of digitonin, which selectively permeabilizes the plasmalemma, but not the inner mitochondrial membrane (11). Cells were loaded with the high-affinity Ca2⫹ indicator X-Rhod-1 (Fig. 4 a and b) and then permeabilized by using 20 ␮M digitonin in a cytosol-like Ca2⫹ free saline (see Methods). X-Rhod-1-AM is accumulated in the mitochondrial matrix by virtue of a matrix-negative potential but is also present in the cytosol (Fig. 4a Upper Left, before permeabilization). Upon addition of digitonin, X-Rhod-1 leaves the cytoplasm but is retained by mitochondria (Fig. 4a Upper Right, 30 s), allowing clear images of the mitochondrial fluorescence to be recorded. Mitochondrial accumulation or loss of Ca2⫹ can be then estimated from X-Rhod-1 fluorescence from mitochondrial regions of the permeabilized cells (Fig. 4a Lower, 210 and 330 s). We detected only small changes in [Ca2⫹]m after addition of 1 ␮M of Ca2⫹ to either control (n ⫽ 26) or MGP-transfected HepG2 cells (n ⫽ 21), (Fig. 4b). Addition of 20 ␮M Ca2⫹ clearly increased [Ca2⫹]m in MGP-expressing mitochondria (X-Rhod-1 fluorescence increased by 870 ⫾ 60 arbitrary units), but in control cells, the X-Rhod-1 fluorescence disappeared, suggesting loss of Ca2⫹ (and/or dye). This loss of fluorescence was prevented by preincubation of the cells with 0.5 ␮M cyclosporin A (CsA) (n ⫽ 20, data not shown), confirming that it reflected the occurrence of PT. Exposure to even higher Ca2⫹ concentrations did not change the signal of the high-affinity Ca2⫹ indicator, X-Rhod-1, in MGP mitochondria, likely because of saturation of the dye. To test this, we carried out similar experiments using the low affinity Ca2⫹ indicator, rhod-5N (Fig. 4c). Addition of 1.2 mM Ca2⫹ induced a fast loss of the rhod-5n fluorescence from control mitochondria (n ⫽ 31 cells). In contrast, in MGP-expressing mitochondria, a further increase of fluorescence was observed (n ⫽ 23), suggesting that MGP PNAS 兩 November 13, 2007 兩 vol. 104 兩 no. 46 兩 18093

CELL BIOLOGY

Fig. 3. Influence of polyP on mitochondrial function. Cells with no identifiable green fluorescence were used as controls. (a) Reduction of mitochondrial inner membrane potential associated with reduced polyP levels. HepG2 expressing MGP showed lower TMRM fluorescence than controls. **, p ⬍ 0.001. (b) Increase of the basal level of NADH autofluorescence in cells expressing MGP. Values of fluorescence in the presence of FCCP (1 ␮M), or NaCN (1 mM), were taken as 0% and 100%, respectively.

similar changes in NADH levels. Measurements of TMRM fluorescence in similar experiments (SI Fig. 6b) indicate that both glutamate and pyruvate increase mitochondrial membrane potential in control (n ⫽ 44) and MGP (n ⫽ 36 cells) HepG2 cells, but the increase was 27.3 ⫾ 1.9% less for MGP cells. This strongly suggests that complex I in MGP mitochondria is expressed and can be stimulated by external substrates but that the activity of the complex I in MGP cells is lower than in controls. The membranepermeable analogue of a complex II substrate, methyl succinate (5 mM), restored mitochondrial membrane potential in rotenonetreated control (n ⫽ 37) and MGP (n ⫽ 46) HepG2 cells (SI Fig. 6c), confirming the activity of complex II after polyP depletion. Again, however, the effect of methyl succinate was 36.8 ⫾ 3.1% less in MGD HepG2 cells than in controls. TMPD (200 ␮M)/ascorbate (5 mM) increased TMRM fluorescence, i.e., increased potential, in both control (n ⫽ 44) and MGP (n ⫽ 33) HepG2 cells (SI Fig. 6d). TMPD/Asc normally acts via complex IV to maximize mitochondrial respiration and establish maximal potential. Here, it was not able to generate a maximal potential in the MGP-transfected cells. After application of TMPD/Asc, the TMRM fluorescence in MGP cells was 68.7 ⫾ 4.9% of the value seen in control cells. Conservatively speaking, the fact that saturating levels of substrate for none of complexes I, II, or IV could drive TMRM fluorescence in MGP cells to maximal control levels suggests that, at least, complex IV is inhibited by MGP expression.

Fig. 4. Role of polyP in mitochondrial Ca2⫹ uptake and PT in permeabilized HepG2 cells. Cells were permeabilized by including 20 ␮M digitonin in the external medium, which also contained 3 mM EGTA. Calcium was added to give the free concentrations indicated. (a) Confocal image of cell loaded with the high-affinity mitochondrial calcium dye XRhod-1, showing cells before (Upper Left) and at 30, 210, and 330 s after permeabilization. The green arrows indicate a cell with MGP overexpression. Cells with no identifiable green fluorescence were used as controls. (b) Time courses of XRhod-1 fluorescence in a group of control and MGP-expressing cells. The controls, but not MGP-expressing cells, show an abrupt calcium release. (c) Fluorescence intensity from mitochondria loaded with Rhod-5n, a low-affinity dye. MGP-expressing cells were able to accumulate calcium until the uncoupler FCCP (1 ␮M) was added.

allowed mitochondria to accumulate and store substantial additional Ca2⫹ without PT opening. Mitochondria in these cells showed no PT opening even in response to 1.2 mM Ca2⫹ but immediately released Ca2⫹ in response to addition of the mitochondrial uncoupler, FCCP (Fig. 4c). Thus, despite reducing mitochondrial membrane potential, depletion of polyP increases the capacity of mitochondria to accumulate and retain calcium and decreases the probability of PTP opening Depolarization of mitochondria by increased Ca2ⴙ in permeabilized cells.

We also assayed PT opening in response to Ca2⫹ additions by monitoring changes in mitochondrial membrane potential. Application of Ca2⫹ at concentrations likely to occur during normal signaling (850 nM free Ca2⫹) induced a modest mitochondrial depolarization in both control (17 ⫾ 1% decrease in TMRM signal, n ⫽ 19) and MGP-transfected cells (25 ⫾ 2% decrease, n ⫽ 15; SI Fig. 7a). Progressive increases in [Ca2⫹] above 10 ␮M induced loss of the mitochondrial potential (signaled by the rapid and almost complete loss of TMRM fluorescence) in control cells (SI Fig. 7b, gray symbols, n ⫽ 36 cells). In contrast, mitochondria from MGP-transfected cells (n ⫽ 35) showed no abrupt depolarization until addition of the uncoupler, FCCP (SI Fig. 7b, green symbols). These observations are consistent with the findings described above and suggest that mitochondria in MGP-expressing cells maintained a substantial inner membrane potential in the face of increasing [Ca2⫹] up to 1.2 mM. Preincubation of control cells with CsA, an inhibitor of PT (0.5 ␮M, 20 min, n ⫽ 22, SI Fig. 7c), prevented Ca2⫹-induced depolarization. Thus, MGP-induced reduction of mitochondrial polyP levels protected mitochondria from Ca2⫹-induced PT. PT induced by reactive oxygen species (ROS). ROS are also major inducers of PT. Although ROS-induced PT is also sensitive to CsA, it is not known whether the same signaling pathway(s) are involved as for Ca2⫹-induced PT. We investigated ROS-induced PT using high-power laser illumination of TMRM. This generates superoxide ROS production and consequently triggers PT (12, 13). Illumination of the cells induced PT in both control (n ⫽ 26) and MGP-transfected (n ⫽ 22) HepG2 cells, within 5–6 min (SI Fig. 8). Preincubation of the HepG2 cells with CsA (0.5 ␮M for 10 min) prevented, or delayed, the rapid decrease in TMRM fluorescence in the majority of the nontransfected HepG2 cells (27 of 33, data not shown). Thus, lowering the level of polyP did not prevent ROS-induced PT opening, in contrast to its effect on Ca2⫹-induced PT. PT-Dependent Cell Death and PolyP. PPX expression protects against ionomycin-induced cell death. Ionomycin is a Ca2⫹ ionophore, which,

when added to cells in culture, severely perturbs Ca2⫹ homeosta18094 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0708959104

sis and causes necrotic cell death. It was demonstrated recently that cells lacking cyclophilin D, a protein required for normal PT opening, are protected from death induced by application of Ca2⫹ and ionomycin (14). Thus, at least in part, the toxic effect of ionomycin involves activation of the PT. We tested whether expression of MGP could protect from ionomycin-induced necrosis. Cells transiently transfected with MGP, or MG (control cells expressing mitochondrially targeted GFP protein, but lacking PPX activity) were treated for 2 h with different concentrations of ionomycin. The percentage of dead cells was determined by using propidium iodide (PI) (Fig. 5a). Cell death was significantly reduced in the MGP-expressing group, compared with the cells expressing MG (Fig. 5b), suggesting that polyP participates in Ca2⫹-induced cell death. Reduction of polyP levels decreases ␤-amyloid toxicity. Alzheimer’s disease is a neurodegenerative disorder characterized by the presence in the brain of senile plaques, containing an amyloid core made of ␤-amyloid peptide (␤A). ␤A is a neurotoxic polypeptide of 39–43 aa that can induce mPTP opening in isolated brain mitochondria (15–17). Previously, we have shown that addition of ␤A promotes fluctuations of intracellular Ca2⫹ concentration, and Ca2⫹- and CsA-dependent fast, transient mitochondrial depolarizations in astrocytes, indicating activation of PT (18, 19). Here, we studied how decrease in polyP affects the action of ␤A. MGP expression decreased DAPI-polyP fluorescence in both neurons and astrocytes, (SI Fig. 9). Control and MGP-transfected mixed cultures of glia and neurons from rat hippocampus were incubated with Rhodamine123 (10 ␮M) for 15 min, followed by washing. This protocol gave an increase of Rhodamine123 fluorescence upon membrane depolarization. ␤A did not cause any change in mitochondrial potential in neurons (n ⫽ 42) over ⬇50 min but did induce profound potential changes in astrocytes (Fig. 5c, n ⫽ 69 astrocytes), with a slow, modest, mitochondrial depolarization, on which were superimposed sporadic large, transient depolarizations. As we showed previously (19), the transient depolarizations depend on Ca2⫹ and ROS (PT inducers) and are inhibited by CsA and antioxidants, known inhibitors of the PT. Transfection of hippocampal cultures with MGP completely prevented the sporadic, fast transient mitochondrial depolarizations induced by ␤A (Fig. 5d, n ⫽ 87). Finally, we explored the protective effect of MGP against toxicity caused by 24-h incubation of cultured cells with 5 ␮M ␤A 1–42. PT in astrocytes is an important step toward cell death caused by ␤A (20). For hippocampal astrocytes expressing MGP, the proportion of dead cells decreased from 23.7 ⫾ 2.2% to Abramov et al.

Fig. 5. Protection against cell death by MGP. (a) Fluorescent images of GFP-positive HepG2 cells treated with 10 ␮M ionomycin for 2 h and loaded with PI. Nuclei of dead cells were stained by PI; only GFP-positive cells were counted. (b) Control cells, transfected with MTS-GFP (no PPX expression), were more vulnerable to ionomycin-induced death than cells expressing MTS-GFP-PPX (MGP). (c) ␤A caused mitochondrial membrane depolarization in astrocytes. (d) MGP-expressing astrocytes were protected from both fast, transient depolarizations and from the profound baseline membrane depolarization occurring in some controls, which precluded additional depolarization after FCCP application. (e and f ) Cell death in, respectively, control and MGP-expressing cocultures of hippocampal neurons and astrocytes. Staining and cell counts were performed after 24-h incubation with ␤A. MGP expression was associated with protection of both neurons and astrocytes from ␤A-induced cell death. Error bars represent SEM from four independent experiments.

Discussion We present experimental evidence that polyP is an integral participant in mitochondrial function in higher eukaryotes. Mitochondrial polyP appears to play important roles in cell metabolism and is a major determinant of the cellular response to stress by regulation of the PT. Involvement of PolyP in Regulation of Normal Mitochondrial Function.

One of our most intriguing findings is the demonstration that levels of polyP significantly affect mitochondrial respiration under normal physiological conditions. Given the physical and chemical properties of polyP, levels of polyP inside mitochondria might regulate their function in several ways, including: (i) PolyP, as a high energy polymer, might serve as a membrane potential independent source of ATP production and thus affect activity of the ATP synthase and the respiratory chain; (ii) levels of polyP, a strong chelator of divalent cations, might affect mitochondrial Ca2⫹-buffering capacity and thus alter the Ca2⫹dependent regulation of multiple mitochondrial enzymes; (iii) polyP, a highly charged polyanion, might directly bind to mitochondrial proteins and thus regulate their activities. Our data suggest that polyP may be required for normal function of the respiratory chain, perhaps most importantly at Complex IV, but they do not exclude other potential sites of action. PolyP: A Component of the Permeability Transition Pore? We have also shown that MGP expression, and hence polyP depletion, increases the calcium accumulation capacity of mitochondria and reduces the probability of Ca2⫹-dependent PT opening. This runs counter to expectations because: (i) polyP depletion reduces mitochondrial membrane potential, and calcium uptake is potential-dependent, and (ii) polyP might be expected to act as a calcium buffer, given its large number of potential calciumbinding sites. We have considered the possibility that polyP may Abramov et al.

represent a pore component that contributes to the development of the PT. Development of PT is a complicated process involving participation of a large number of proteins and regulation by multiple agents. The specific components that contribute to PT are highly controversial (21). This complexity, together with involvement of polyP in several physiological functions, makes it difficult to conclude unambiguously how polyP is involved in PT. However, multiple correlations between conditions favoring development of PT, and the presence/opening of the polyP/ Ca2⫹/PHB channel, suggest that polyP could participate directly in formation of the PT pore by forming a polyP/Ca2⫹/PHB complex. This idea is consistent with several observations: (i) ion-conducting properties of the mitochondrial polyP/Ca2⫹/PHB channel mimic properties of PT pore seen by patch-clamp of native membranes and in experiments on intact mitochondria (see discussion in ref. 6; (ii) polyP/Ca2⫹/PHB complex formation requires elevated concentrations of Ca2⫹ (22), a condition characteristic of Ca2⫹-induced PT; (iii) PT development is favored by several membrane proteins (7), as is formation of polyP/Ca2⫹/PHB channels in bacteria (23); (iv) both PT in mitochondria and polyP/Ca2⫹/PHB complex formation in bacteria are associated with passage of DNA across membranes (24, 25); and finally, (v) our inability to detect PT in polyP-deficient cells, even under strong Ca2⫹ overload, suggests that polyP probably plays a direct, rather than a regulatory, role. Possible Mechanisms of Protection Against Cell Death. A large body

of experimental data suggests that prevention of PT is a powerful way to prevent cell death under pathophysiological conditions (7). The fact that depletion of polyP protects cells from Ca2⫹dependent death supports the idea that it is involved in PT development in vivo. In our ␤A experiments, polyP depletion protected both neurons and astrocytes. We have demonstrated that ␤A acts directly on astrocytes, inducing increases in cytoplasmic Ca2⫹ and consequent PT, whereas neuronal cell death is caused by oxidative stress transmitted from astrocytes (20). Conclusions Our findings show that polyP can strongly influence mitochondrial metabolism and Ca2⫹ accumulation and induction of the PNAS 兩 November 13, 2007 兩 vol. 104 兩 no. 46 兩 18095

CELL BIOLOGY

12.6 ⫾ 1%; P ⬍ 0.001, and, for neurons, it decreased from 51.9 ⫾ 4.6% to 17.9 ⫾ 1.4%, P ⬍ 0.001. (Fig. 5 e and f, n ⫽ 4 experiments). Thus, MGP expression protected both neurons and astrocytes against ␤A.

Ca2⫹-dependent PT. Depletion of polyP exerts a remarkably potent action to protect cells from Ca2⫹-dependent cell death in a pathophysiological model of ␤A toxicity. These observations raise major questions about the normal physiological pathway(s) for regulation of polyP in mammalian cells, and suggest possible new strategies to intercept disease processes which perturb the normal pathways controlling cell death. Methods Generation of MTS-GFP-PPX Construct and Estimation of PPX Activity.

pAcGFP1-Mito cDNA (MTS-GFP), which encodes a fusion of a mitochondrial targeting sequence derived from the precursor of subunit VIII of human cytochrome c oxidase and the GFP from Aequorea coerulescens, was purchased (Clontech, Mountain View, CA) and amplified by PCR with the addition of flanking NheI (5⬘) and EcoRI (3⬘) sites. The product was then subcloned into the vector, pSWF1 (kindly provided by Arthur Kornberg), containing the entire coding region of the exopolyphosphatase, scPPX1, for mammalian cell expression. Sequencing verified that the full-length code of MTS-GFP-scPPX1 exactly fills the ORF. Cultured cells were transfected with the DNA by using PolyFect Transfection Reagent (Qiagen, Valencia, CA). Activity of the MTS-GFP-scPPX1 construct in cell lysates was assayed by using a standard protocol (26) to measure the hydrolysis of radioactive 32P-polyP, as detailed in SI Text.

Fluorescent Microscopy. Epifluorescence measurements were obtained by using an inverted microscope with a ⫻20 fluorite objective. For Rhodamine123 measurements, excitation light from a xenon arc lamp was selected by using a 10-nm bandpass filter centered at 490 nm, housed in a computer-controlled filter wheel (Cairn Research, Faversham, U.K.). Cell Death Experiments. Dead cells were counted after staining

with propidium iodide (PI). For ionomycin experiments, HepG2 cells were transfected with MTS-GFP-PPX (abbreviated as MGP) or MTS-GFP (control) constructs. For ␤A toxicity assays, mixed cultures of hippocampal neurons and glial cells were prepared as described (19) from Sprague–Dawley rat pups 2–4 days postpartum (University College London breeding colony). Transfection with MGP was done 7 days after culture preparation. Cell count was done by using PI and Hoechst 33342. Further details of the experiments described in SI Text. Statistical Analysis. Statistical analysis was performed by using Origin 7 (Microcal Software, Northampton, MA) software. Results are expressed as means ⫾ SEM).

510 CLSM (Zeiss, Thornwood, NY) equipped with a META detection system and a ⫻40 oil-immersion objective. Dyes used and protocol details described in SI Text.

We thank Dr. Arthur Kornberg for scPPX1 DNA and the use of his laboratory to perform some experiments. Drs. George Chaconas, Wayne Giles, Kathleen Kinnally, Catherine Morris, Michael Walsh, and Gerald Zamponi provided critical comments on drafts of the manuscript. A.Y.A. and M.R.D. are supported by the Wellcome Trust, E.P. and R.J.F. are supported by the Canadian Institutes of Health Research, and R.J.F. is an Alberta Heritage Foundation for Medical Research Medical Scientist.

Kulaev I, Vagabov V, Kulakovskaya T (1999) J Biosci Bioeng 88:111–129. Kornberg A, Rao NN, ult-Riche D (1999) Annu Rev Biochem 68:89–125. Negoda A, Xian M, Reusch RN (2007) Proc Natl Acad Sci USA 104:4342–4346. Kumble KD, Kornberg A (1995) J Biol Chem 270:5818–5822. Reusch RN (1989) Proc Soc Exp Biol Med 191:377–381. Pavlov E, Zakharian E, Bladen C, Diao CT, Grimbly C, Reusch RN, French RJ (2005) Biophys J 88:2614–2625. Bernardi P, Krauskopf A, Basso E, Petronilli V, Blachly-Dyson E, Di LF, Forte MA (2006) FEBS J 273:2077–2099. Wurst H, Kornberg A (1994) J Biol Chem 269:10996–11001. Tijssen JP, Beekes HW, Van SJ (1982) Biochim Biophys Acta 721:394–398. Rodrigues CO, Ruiz FA, Rohloff P, Scott DA, Moreno SN (2002) J Biol Chem 277:48650–48656. Gukovskaya AS, Zinchenko VP, Petrunyaka VV, Khodorov BI, Evtodienko YV (1986) Eur J Biochem 161:249–256. Duchen MR (2000) Cell Calcium 28:339–348. Huser J, Rechenmacher CE, Blatter LA (1998) Biophys J 74:2129–2137. Baines CP, Kaiser RA, Purcell NH, Blair NS, Osinska H, Hambleton MA, Brunskill EW, Sayen MR, Gottlieb RA, Dorn GW, et al. (2005) Nature 434:658–662.

15. Parks JK, Smith TS, Trimmer PA, Bennett JP, Jr, Parker WD, Jr (2001) J Neurochem 76:1050–1056. 16. Shevtzova EF, Kireeva EG, Bachurin SO (2001) Bull Exp Biol Med 132:1173– 1176. 17. Moreira PI, Santos MS, Moreno A, Oliveira C (2001) Biosci Rep 21:789–800. 18. Abramov AY, Canevari L, Duchen MR (2003) J Neurosci 23:5088–5095. 19. Abramov AY, Canevari L, Duchen MR (2004) J Neurosci 24:565–575. 20. Abramov AY, Duchen MR (2005) Philos Trans R Soc London Ser B 360:2309– 2314. 21. Kokoszka JE, Waymire KG, Levy SE, Sligh JE, Cai J, Jones DP, MacGregor GR, Wallace DC (2004) Nature 427:461–465. 22. Reusch RN, Sadoff HL (1988) Proc Natl Acad Sci USA 85:4176–4180. 23. Jones HE, Holland IB, Jacq A, Wall T, Campbell AK (2003) Biochim Biophys Acta 1612:90–97. 24. Patrushev M, Kasymov V, Patrusheva V, Ushakova T, Gogvadze V, Gaziev A (2004) Cell Mol Life Sci 61:3100–3103. 25. Castuma CE, Huang R, Kornberg A, Reusch RN (1995) J Biol Chem 270:12980–12983. 26. Wang L, Fraley CD, Faridi J, Kornberg A, Roth RA (2003) Proc Natl Acad Sci USA 100:11249–11254.

Confocal Experiments. Confocal images were obtained by using a

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

18096 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0708959104

Abramov et al.