JBIC Journal of Biological Inorganic Chemistry

, Volume 12, Issue 7, pp 1029–1053

Electron paramagnetic resonance and Mössbauer spectroscopy of intact mitochondria from respiring Saccharomyces cerevisiae

  • Brandon N. Hudder
  • Jessica Garber Morales
  • Audria Stubna
  • Eckard Münck
  • Michael P. Hendrich
  • Paul A. Lindahl
Original Paper

DOI: 10.1007/s00775-007-0275-1

Cite this article as:
Hudder, B.N., Morales, J.G., Stubna, A. et al. J Biol Inorg Chem (2007) 12: 1029. doi:10.1007/s00775-007-0275-1

Abstract

Mitochondria from respiring cells were isolated under anaerobic conditions. Microscopic images were largely devoid of contaminants, and samples consumed O2 in an NADH-dependent manner. Protein and metal concentrations of packed mitochondria were determined, as was the percentage of external void volume. Samples were similarly packed into electron paramagnetic resonance tubes, either in the as-isolated state or after exposure to various reagents. Analyses revealed two signals originating from species that could be removed by chelation, including rhombic Fe3+ (g = 4.3) and aqueous Mn2+ ions (g = 2.00 with Mn-based hyperfine). Three S = 5/2 signals from Fe3+ hemes were observed, probably arising from cytochrome c peroxidase and the a3:Cub site of cytochrome c oxidase. Three Fe/S-based signals were observed, with averaged g values of 1.94, 1.90 and 2.01. These probably arise, respectively, from the [Fe2S2]+ cluster of succinate dehydrogenase, the [Fe2S2]+ cluster of the Rieske protein of cytochrome bc1, and the [Fe3S4]+ cluster of aconitase, homoaconitase or succinate dehydrogenase. Also observed was a low-intensity isotropic g = 2.00 signal arising from organic-based radicals, and a broad signal with gave = 2.02. Mössbauer spectra of intact mitochondria were dominated by signals from Fe4S4 clusters (60–85% of Fe). The major feature in as-isolated samples, and in samples treated with ethylenebis(oxyethylenenitrilo)tetraacetic acid, dithionite or O2, was a quadrupole doublet with ΔEQ = 1.15 mm/s and δ = 0.45 mm/s, assigned to [Fe4S4]2+ clusters. Substantial high-spin non-heme Fe2+ (up to 20%) and Fe3+ (up to 15%) species were observed. The distribution of Fe was qualitatively similar to that suggested by the mitochondrial proteome.

Keywords

Iron Sulfur Cluster assembly Heme biosynthesis Non-heme 

Abbreviations

CoQ

Coenzyme Q

DTT

Dithiothreitol

EDTA

Ethylenediaminetetraacetic acid

EGTA

Ethylenebis(oxyethylenenitrilo)tetraacetic acid

EPR

Electron paramagnetic resonance

ETF

Electron transfer flavoprotein

HEPES

N-(2-Hydroxyethyl)piperazine-N′-ethanesulfonic acid

IM

Inner membrane

IMS

Intermembrane space

NHE

Normal hydrogen electrode

OM

Outer membrane

SH buffer

0.6 M sorbitol/20 mM N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid buffer pH 7.4

SP buffer

1.2 M sorbitol/20 mM potassium phosphate buffer pH 7.4

Introduction

Mitochondria are the cellular organelles in which oxidative phosphorylation and a myriad of related processes involving iron, copper and manganese occur. These branched tubular structures have an outer membrane (OM), an aqueous intermembrane space (IMS), an inner membrane (IM) and an aqueous matrix region. The IM is highly invaginated, with cristae protruding into the aqueous matrix region. Imported iron ions are used in heme and iron–sulfur (Fe/S) cluster biosynthesis. A portion of these nascent prosthetic groups are incorporated into mitochondrial apoproteins, while the remainder are exported to the cytosol. Imported copper and manganese ions are installed into cytochrome c oxidase and manganese superoxide dismutase, respectively.

The proteins involved in these processes can be categorized in terms of the metal centers they contain. Proteins containing Fe2S2, Fe3S4 and/or Fe4S4 clusters include succinate dehydrogenase [1, 2, 3], the Rieske protein [4], aconitase and homoaconitase [5, 6], ferredoxin/adrenodoxin [7, 8, 9, 10], biotin synthase [11, 12, 13, 14, 15] and lipoic acid synthase [16, 17, 18]. Dihydroxyacid dehydratase catalyzes the dehydration of an intermediate in the biosynthesis pathway of branched-chain amino acids [19]. Although the metal center in this enzyme has not been well studied, the homologous enzyme from Escherichia coli contains an Fe4S4 cluster [20]. Such clusters are also found in scaffold proteins which are used in the synthesis of Fe/S clusters, including Isu1p, Isu2p, Isa1p and Nfu1p [21, 22, 23, 24]. A BLAST search suggests that the open reading frame YOR356W encodes the flavin adenine dinucleotide-containing and Fe4S4-containing electron transfer flavoprotein (ETF) dehydrogenase [25, 26, 27].

Other mitochondrial proteins contain heme groups. Heme b is found in cytochrome bc1 [28], cytochrome c peroxidase [29], succinate dehydrogenase and flavocytochrome b2 [30]. Heme a is found in cytochrome c oxidase [31], while heme c is contained in cytochrome c1 and in both isoforms of cytochrome c [32]. Heme monooxygenase catalyzes the conversion of heme b to heme a within the heme biosynthetic pathway [33]. The homolog from E. coli contains heme b and heme a prosthetic groups [34, 35, 36]. When yeast cells are grown under respiratory conditions, the heme-b-containing catalase A (Cta1p) is targeted to the mitochondrial matrix [37].

Another group of proteins are involved in iron trafficking. Isa2p [38] and Yfh1p [39] help import Fe2+ ions into the matrix and insert iron into ferrochelatase (Hem15p) for heme biosynthesis [40, 41, 42] and into scaffold proteins for Fe/S synthesis. IM proteins Mrs3p, Mrs4p, Mmt1p and Mmt2p carry iron into mitochondria [43, 44]. Heme O synthase (Cox10p) and heme A synthase (Cox15p) bind intermediate states of hemes [33, 35, 45]. Cytochrome c heme lyase (Cyc3p) and cytochrome c1 heme lyase (Cyt2p) install heme c into cytochromes c and c1, respectively [46]. Mdl1p exports heme groups [47], while Atm1p and Erv1p export Fe/S clusters [48, 49]. Coq7p is a yeast mitochondrial protein that contains a diiron center and serves as a monooxygenase/hydroxylase in coenzyme Q (CoQ) biosynthesis [50, 51].

Cytochrome c oxidase is the best-known copper-containing mitochondrial protein. The Cox1p subunit of this complex contains one copper ion (CuB) adjacent to heme a3 in its active site, while the electron-transfer CuA site in Cox2p contains two copper ions [52]. Cox23p, Cox17p, Sco1p and Cox11p are chaperones that import Cu ions into mitochondria and insert them into Cox1p and Cox2p during their assembly [53, 54, 55]. Copper ions in these chaperones are in the diamagnetic Cu+ oxidation state. A small amount of the cytosolic copper-containing (Cu–Zn) superoxide dismutase (Sod1p) appears to localize in the IMS of mitochondria [56]. Approximately 90% of mitochondrial copper is found in the matrix as a nonproteinaceously bound pool of Cu+ ions [57].

Manganese superoxide dismutase (Sod2p) appears to be the only manganese-containing enzyme in Saccharomyces cerevisiae mitochondria. The mitochondrial manganese chaperone protein (Mtm1p) helps to import manganese ions and to install one of these ions into matrix-localized apo-Sod2p [58].

Flavins and ubiquinone can be stabilized in an S = 1/2 semiquinone state that affords electron paramagnetic resonance (EPR) signals at the free-electron g value, 2.00. Flavin-containing mitochondrial proteins include α-ketoglutarate dehydrogenase [59], d-lactate cytochrome c oxidoreductases [60], glutathione reductase [61], thioredoxin reductase [62], glycerol-3-phosphate dehydrogenase [63], d-arabinono-1,4-lactone oxidase [64], acetolactate synthase [65], methylene tetrahydrofolate reductase [66], succinate dehydrogenase [67], Coq6p [68], Mmf1p [69] and ETF dehydrogenase [25, 26, 27].

The most important spectroscopic technique that has been applied to intact mitochondria is EPR, dating from the pioneering work of Beinert [70], who initially described high-spin heme signals from cytochrome c oxidase [71]. The gave = 2.01 EPR signal from the inactivated [Fe3S4]+ form of aconitase was observed in crude intact rat-heart mitochondria exposed to H2O2 [72]. EPR signals from the Rieske cluster of cytochrome bc1 and the [Fe2S2]+ cluster of succinate dehydrogenase have also been observed in intact mitochondria [73, 74, 75, 76, 77, 78]. EPR spectra of intact mitochondria were examined to determine the effect of abolishing heme biosynthesis on succinate dehydrogenase and the Rieske protein [75] and to determine the effects of Ca2+ and Mn2+ ions [79, 80]. Adrenodoxin levels in intact human placental mitochondria were examined by EPR [81]. Respiratory complexes in submitochondrial fractions have also been examined [82, 83, 84]. In contrast, there has been just one report of a Mössbauer spectrum of intact mitochondria, specifically of a strain in which yfh1 was deleted [39]. This genetic modification causes iron to accumulate in the matrix, and the observed Mössbauer spectral intensity exclusively reflected the accumulated iron. The “control” Mössbauer spectrum of wild-type mitochondria was devoid of any signals.

This overview highlights the complexity of transition metal metabolism occurring within these organelles. We report on our efforts to establish a few simple yet unestablished aspects of iron metabolism in yeast mitochondria, namely, the absolute concentration of iron and of overall protein contained therein, and the proportion of that iron present in various types of centers (e.g., hemes, Fe/S clusters, etc.). Our approach was to investigate mitochondria from S. cerevisiae using EPR and Mössbauer spectroscopy along with various bioanalytical characterizations. For the first time using whole mitochondria, the absolute spin concentrations of detectable metal protein species have been quantified from EPR spectra. We investigated intact mitochondria prepared under different redox and/or isolation conditions. We determined the proportion of excluded buffer in these packed samples, which, when combined with metal and protein determinations of the packed samples, allowed us to estimate the absolute iron concentration contained within these organelles. This information, when combined with our spectroscopic results, allowed us to estimate, albeit in broad terms, how iron is distributed within the organelle. This distribution was then compared with that calculated from the iron-containing proteins known to be present in the mitochondrial proteome.

Materials and methods

Cell growth and isolation of mitochrondia

S. cerevisiae cells (strain D273-10B) were grown on SSlac medium (0.3% glucose, 1.7% lactate) [72 g yeast extract, 25 g ammonium chloride, 25 g potassium hydrogen phosphate, 12.5 g NaCl, 12.5 g CaCl2, 14.4 g MgCl2, 12.5 g glucose and 0.7 L of 60% sodium lacate syrup (Fisher) in 25-L solution] in a custom-built thermostatically controlled autoclavable 25-L glass fermenter in which cultures were stirred and bubbled with pure O2 at a rate of approximately 3 L/min, dispersed through a fine glass frit with a diameter of 5 cm. Under these growth conditions, cells ferment on glucose at early stages of growth and then switch to respiration on lactate once the glucose has been consumed. Harvesting commenced when the optical density of a 1-cm solution at 600 nm reached 1.2–1.4. The culture was chilled to 278 K and harvested at 5,000 rpm using a Sorvall SLC-6000 rotor and a Sorvall Evolution centrifuge.

Immediately after harvesting and without freezing the pelleted cells, mitochondria were isolated essentially as described in [85] except that all steps were performed anaerobically. Cell paste (100–150 g) was transferred into a refrigerated argon-atmosphere glove box (M. Braun) maintained at approximately 278 K and approximately 1 ppm O2 as monitored continuously using a model 310 Teledyne analyzer. Buffers used in the isolation were degassed on a Schlenk line. For some preparations, all isolation buffers were supplemented with ethylenediaminetetraacetic acid (EDTA) or ethylenebis(oxyethylenenitrilo)tetraacetic acid (EGTA) (Acros) at final concentrations of 1 or 10 mM. In other preparations, no metal chelators were included. Cell paste was suspended in a 100 mM tris(hydroxymethyl)aminomethane sulfate/10 mM dithiothreitol (DTT) buffer (500 mL) and then spun at 5,000 rpm for 5 min in the SLC-6000 rotor. Subsequent centrifugations were performed under these conditions unless otherwise stated. The resulting pellet was suspended in 1.2 M sorbitol/20 mM potassium phosphate buffer, pH 7.4 (500 mL), hereafter referred to as SP buffer, using a rubber policeman. The resulting suspension was centrifuged, resuspended in SP buffer (500 mL), centrifuged again, and resuspended again in the same buffer. Cell walls were disrupted by adding 3 mg of 100 units/mg yeast lytic enzyme (Sigma) per gram of cell paste. The resulting spheroplasts were centrifuged, suspended in SP buffer (500 mL) and then centrifuged. The pellet was resuspended in 250 mL of 1.2 M sorbitol/40 mM N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid (HEPES) pH 7.4 and 250 mL of 1 mM phenylmethylsulfonyl fluoride in double-distilled H2O. The mixture was homogenized using 25 strokes of a 40-mL Dounce homogenizer (Fisher Scientific) during a period of 2–4 min. The suspension was centrifuged at 2,500 rpm for 5 min, and the supernatant was transferred to a fresh centrifuge bottle and centrifuged again under the same conditions. The supernatant, which consisted of crude mitochondria, was then centrifuged at 10,000 rpm in a Sorvall SLA-1500 rotor for 10 min. The resulting pellet was resuspended in 200 mL of a 0.6 M sorbitol/20 mM HEPES buffer pH 7.4, hereafter referred to as SH buffer. The resulting solution was centrifuged three more times, in the manner described in the previous three sentences, and the final pellet of crude mitochondria was resuspended in 20 mL SH buffer. This solution was loaded onto a discontinuous gradient solution composed of 10 mL of 15% and 10 mL of 20% (w/v) Histodenz® (Sigma) prepared in SH buffer and contained in Beckman Ultra Clear™ centrifuge tubes. The tubes were placed in the buckets of an SW-32Ti rotor (Beckman Coulter). The buckets were sealed, removed from the box and spun at 9,000 rpm in an SW-32Ti rotor (Beckman Coulter) for 1.5 h using a Beckman L7 ultracentrifuge. The buckets were returned to the box, and the tubes were placed in a support which allowed the pure mitochondrial band at the interface of the gradient to be collected after first removing the layer above the band. From 150 g cell paste, a total of 5–15 mL of mitochondrial solution in the “as-isolated” state was obtained using three to six buckets depending on the yield. The only reductant used during the procedure was DTT and then only at an early step of the isolation procedure before cell walls were disrupted. E°′ for the disulfide/DTT half-cell is −330 mV versus the normal hydrogen electrode (NHE) [86]. Anaerobically prepared isolation buffers undoubtedly contained a trace of oxidizing ability [87]. Both factors considered, the resulting solution potential of mitochondria in the non-redox-buffered “as-isolated” state was estimated to be between −0.1 and 0 mV versus NHE. Prior to freezing, some samples were exposed to air (typically for 1 day at 277 K), sodium dithionite (10 mM at pH 7.5 or 8.5), potassium ferricyanide (1 mM) or nitric oxide (1 atm).

For Mössbauer spectroscopy studies, S. cerevisiae cells were grown similarly except that the medium was supplemented with 20 μM 57FeCl3. With use of a custom-made Delrin™ insert that fit in the buckets of the SW-32Ti rotor, isolated mitochondria were packed tightly into Mössbauer cuvettes by centrifugation, typically at 9,000 rpm for 2 h. Samples were then frozen inside the glove box by contact with a liquid-nitrogen-cooled aluminum block. There was some variation in speed and duration used in packing, resulting in some differences in terms of observed 57Fe concentrations. Each spectrum presented here was recorded with an approximately 40 mCi 57Co source.

Electron and fluorescence microscopy

One milliliter of as-isolated mitochondrial solution was microcentrifuged (Fisher Scientific) at 6,400 rpm for 5 min in a 1.5-mL Eppendorf tube. The pellet was resuspended in SH buffer and glutaraldehyde (2.0% v/v final concentration) was added. The solution was recentrifuged and the pellet was resuspended in 1% osmium tetroxide and 0.5% potassium ferrocyanide (w/v) in SH buffer. This was followed by en bloc staining using 1% uranyl acetate in SH buffer. Samples were dehydrated by incubation in increasingly concentrated ethanol solutions and then embedded using epoxy-based resin. Thin-sectioning was performed using a glass knife/water trough on a microtome, followed by retrieval of the thin sections using 200 mesh grids. Positive staining of these sections was performed using lead acetate/sodium hydroxide [88]. Images were obtained using a JEOL 1200 EX transmission electron microscope.

For fluorescence images, equivalent mitochondrial solutions were incubated in SH buffer, containing 500 nM MitoTracker® (Molecular Probes) or, in another experiment, 1 μM ERTracker® at 310 K for 45 min. The solution was centrifuged, and the pellet was resuspended in SH buffer. Images were obtained using a Bio-Rad Radiance 2000 MP instrument equipped with a ×63 (water-immersion) objective.

Oxygen consumption measurements

A sample of non-chelator-treated intact mitochondria was suspended in SH buffer. A 5-mL portion was assayed for protein concentration using the biuret method [89] as described in “Protein and metal ion concentrations.” Another portion of the intact mitochondrial solution was divided into three samples. One sample was incubated anaerobically for 4–5 h with 10 mM EDTA, another was incubated similarly with 10 mM EGTA, and the remaining sample was not treated. Each sample (1.2 mL) was injected into 29 ± 1 mL of air-saturated SH buffer containing 1.5 mM NADH, 0.2 mM ADP, 2 mM MgCl2, 20 mM phosphates pH 7.4, 250 mM sucrose and 10 mM KCl, essentially as described in [90]. The solution was maintained at 298 K in a water-jacketed glass vessel which contained no gas head space. Included in this vessel was a Clark oxygen electrode (YSI Bioanalytical Products). The final protein concentration was 0.10 mg/mL.

Electron paramagnetic resonance

Custom-built Delrin™ inserts were designed to fit within the buckets of the SW-32Ti rotor. Holes were drilled into the center of these inserts, with a diameter just sufficient to fit a modified EPR tube (4.96-mm outer diameter; 3.39-mm inner diameter; 80-mm long; Wilmad/Lab Glass, Buena, NJ, USA). A 2-mm-long cylinder of silicone rubber was inserted at the bottom of the hole. The brown mitochondrial solution obtained from the gradient step described in “Cell growth and isolation of mitochondria” was diluted with an equal volume of SH buffer. Tubes were filled with this solution and the entire assembly was sealed, removed from the box and spun by centrifugation at 9,000 rpm for 1 h. Samples were returned to the box, and the supernatant was replaced with additional mitochondrial solution. This process was repeated until the volume of tightly packed mitochondria at the bottom of the tube reached approximately 400 μL. EPR tubes were removed from the inserts and frozen in less than 1 min using liquid N2. Two to four EPR samples were prepared from a solution of gradient-purified mitochondria isolated from 25 L of culture.

One end of a stainless steel wire (20 cm × 0.5-mm diameter) was attached to one end of a stainless steel rod (20 cm × 4.8-mm diameter), with the wire extended coaxially with the rod. Approximately 5 cm beyond the point of attachment, the wire was bent back towards the rod (like a hairpin) and coiled around itself up towards the rod. The outer diameter of the coil at the base of the hairpin was slightly less than the inner diameter of the modified EPR tube, while the outer diameter of the remainder of the coil was slightly greater than the inner diameter of the EPR tube. In this way, the wire coil fit snugly into the upper region of the EPR tubes. The entire assembly was just sufficiently robust to be inserted into and removed from the EPR cavity. Spectra were obtained with a Bruker EMX X-band EPR spectrometer operating in perpendicular mode with an Oxford Instruments EM910 cryostat. Signals were simulated with SpinCount written by one of the authors (M.P.H.). Signal intensities were quantified relative to a CuEDTA spin standard using the same software.

Protein and metal ion concentrations

A line was drawn on the exterior of the EPR tubes to indicate the height of the packed mitochondria. The organelles were thawed and quantitatively transferred to plastic screw-top vials using a slightly twisted quartz rod and a minimal volume of SH buffer. The volume of packed organelles was determined by weighing the tubes before and after filling them with an equivalent volume of water, and then dividing the difference by the density of water. The final volume of the solution in the screw-top vial, typically 5 mL, was similarly determined. The ratio of these two volumes constituted the dilution factor by which measured protein and metal concentrations, obtained using the solution in the vial, were multiplied to yield the respective concentrations in packed mitochondria.

Samples contained in the vial were sonicated using a Branson Sonifier 450 operating for 5–10 min at 60% capacity. Protein analyses were performed in either of two ways, namely, by quantitative amino acid analysis, which is the most accurate method available [91], or by the biuret colorimetric method. Relative to amino acid analysis, the results obtained using the biuret method were similar within the uncertainty of the measurements. For quantitative amino acid analysis, aliquots were hydrolyzed in 6 M HCl/2% phenol at 383 K and analyzed using a Hewlett-Packard AminoQuant system. Amino acid percentages were similar among preparations. Primary and secondary amino acids present in the samples were derivatized using o-phthalaldehyde and 9-fluoromethylchloroformate, respectively. Metal concentrations were determined by atomic absorption spectrometry (PE AAnalyst 700 operating in furnace mode) and by inductively coupled plasma mass spectrometry (PerkinElmer). Sonicated samples (250–400 μL) were digested using an equal volume of 15.8 M trace-metal-grade HNO3 (Fischer Scientific) in a sealed plastic tube that was then incubated overnight at 353 K. The resulting solution was diluted with deionized and distilled H2O to a final HNO3 concentration of 0.2 M.

Percentage of external solution in packed samples

Custom-built Lexan “graduated cylinders” were constructed within inserts that fit within buckets of the SW-32Ti rotor (Fig. 1). These inserts were used to accurately measure the volume of a packed mitochondria sample, obtained by loading a solution of isolated mitochondria and spinning the sample for 1 h at 9,000 rpm (10,000g). The supernatant was decanted and the volume of the packed sample was measured using this apparatus. This volume (Vpel) was assumed to be composed of the volume of the mitochondria plus the volume of excluded water: \( {\text{(}}V_{{{\text{pel}}}} {\text{ = }}V_{{{\text{mito}}}} {\text{ + }}V_{{{\text{H}}_{{\text{2}}} {\text{O}}}} {\text{)}} \). To determine the ratio \( V_{{{\text{H}}_{{\text{2}}} {\text{O}}}} {\text{/}}V_{{{\text{pel}}}} \), a 1.00-mL stock solution of radioactively labeled sucrose (American Radiolabeled Chemicals, 625 mCi/mmol), prepared in SH buffer (with Cstock* in counts per minute per milliliter given in Table 1 for each experiment), was added to the pellet and the pellet was resuspended. The inserts were spun by centrifugation as described above, the supernatant was removed, the volume (Vsup1) was determined using a gastight syringe (Hamilton), and the concentration of radioactivity (Cstock*) was determined by scintillation counting (Beckman 5000SL). Assuming that none of the sucrose entered the mitochondria, the excluded water will also have a concentration of radioactivity given by Cstock*. The conservation of matter suggests that
$$ C_{{{\text{stock}}}} ^{*} V_{{{\text{stock}}}} = C_{{{\text{sup1}}}} ^{*} V_{{{\text{sup1}}}} + C_{{{\text{sup1}}}} ^{*} V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{1}}} }} . $$
This equation was solved for \( V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{1}}} }} \). The resulting pellet was found to have essentially the same volume as the original pellet. This pellet, containing radioactively labeled sucrose in the external volume, was resuspended with a solution of nonradioactively labeled sucrose, and the other steps of the same process were repeated. In this case, the resulting concentration of radioactivity in the supernatant fraction was called Csup2* and the corresponding conservation of matter relationship becomes
$$ C_{{{\text{sup1}}}} ^{*} \cdot V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{2}}} }} = C_{{{\text{sup2}}}} ^{*} \cdot V_{{{\text{sup2}}}} + C_{{{\text{sup2}}}} ^{*} \cdot V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{2}}} }} . $$
This equation was solved for \( V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{2}}} }}\). The average of the two values for \( V_{{{\text{H}}_{{\text{2}}} {\text{O}}}} \) was divided by Vpel, affording the fraction of the pellet volume due to excluded water.
Fig. 1

Graduated cylinder used to measure the volume of packed mitochondria samples

Table 1

Determination of excluded buffer in packed mitochondria samples

Cstock* (cpm/mL)

Vstock (mL)

Csup1* (cpm/mL)

Vsup1 (mL)

\( V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{1}}} }} \) (mL)

Csup2* (cpm/mL)

Vsup2 (mL)

\( V_{{{\text{H}}_{{\text{2}}} {\text{O}}_{{\text{2}}} }} \) (mL)

Vpel (mL)

Average % H2O In Vpel

23,830

1.00

20,960

1.00

0.14

2,091

0.98

0.11

0.71

17

37,750

1.00

35,940

1.00

0.05

3,210

0.98

0.10

0.40

18

49,440

1.00

45,150

0.99

0.11

8,220

1.00

0.22

0.82

20

251,260

1.00

238,640

0.98

0.07

52,510

1.00

0.28

0.52

34

58,880

1.00

51,010

1.01

0.14

7,920

0.99

0.18

0.63

26

58,880

1.00

44,580

1.12

0.20

4,120

0.99

0.10

0.92

16

58,880

1.50

47,660

1.49

0.37

14,800

0.97

0.44

1.40

29

Results

Characterization of intact mitochondria

Intact yeast mitochondria were isolated as described in “Materials and methods.” Some preparations were isolated without adding a metal chelator to the isolation buffers, while others were isolated in the presence of either EDTA or EGTA. These chelators were added to remove adventitious metal ions associated with mitochondria. EGTA is unable to penetrate biological membranes [92], while this property is uncertain with respect to EDTA. However, EDTA has been used in isolating mitochondria [93] and as far as we are aware, there have been no reports of EDTA stripping essential metal ions from these organelles.

We assayed a number of preparations for purity and membrane integrity using electron microscopy. Although significant size dispersion was typically evident (Fig. 2, top), there was no obvious evidence of impurities (bacteria or Golgi apparatus) or disrupted membrane structures. Sample morphology was independent of the method of isolation (as-isolated, EDTA-treated or EGTA-treated). Our images are similar to those obtained in the classical studies of Hackenbrock [94] and more recently [95]. Dispersion probably results from the dynamic fission and fusion processes that occur in yeast mitochondria [96]. Confocal microscopic images reveal that mitochondria form extensive tubelike networks extending throughout the cell [97]. These dynamic changes in size and shape would appear to render the concept of the number of mitochondria per cell rather meaningless. A more quantifiable parameter is the volume occupied by these organelles, and we will use this parameter throughout this paper.
Fig. 2

Electron microscopy (top) and fluorescence microscopy (bottom) images of whole mitochondria isolated from Saccharomyces cerevisiae

Fluorescence microscopy was also used to assess purity. One sample was stained for fluorescence with MitoTracker®, while another was stained with ERTracker®. The former dye associates with mitochondria, while the latter associates with the endoplasmic reticulum. As shown in Fig. 2, the vast majority of objects in our samples assimilated the MitoTracker® stain. There was no obvious sign of endoplasmic reticulum contamination using ERTracker® (data not shown). Both results suggest that the samples examined here were relatively pure and intact.

We assayed a number of preparations for their ability to consume O2. As shown in Fig. 3, preparations incubated in the absence of chelator or in the presence of EDTA or EGTA consumed 240, 160 and 200 nmol O2 per minute per milligram of protein, respectively (estimated relative error of ±20%) when incubated in buffer containing NADH. Control samples assayed in the absence of NADH consumed little O2. These rates are similar to those reported previously [98, 99, 100]. We also evaluated the coupling ratio of our preparations, defined as the rate of O2 consumption with ADP added to the assay solution divided by the rate of consumption when ADP was absent. In our fresh samples, this ratio was approximately 2, similar to previously reported ratios [99, 100], whereas it approached 1 for mitochondria stored anaerobically at 278 K for approximately 3 days. Preparations used for EPR and Mössbauer analyses were frozen between 6 h and 3 days after they were isolated. We have not yet been able to correlate the age of the mitochondria to specific spectral changes, but we suspect that spectral features might become slightly broader with age.
Fig. 3

Oxygen consumption by isolated intact mitochondria. No chelator (squares), ethylenediaminetetraacetic acid (EDTA; triangles), ethylenebis(oxyethylenenitrilo)tetraacetic acid (EGTA; circles). The experiment was performed as described in “Materials and methods

We determined protein and metal concentrations in our packed samples. The mean protein concentration (n = 15) was 55 ± 13 mg/mL, independent of whether chelators were or were not included in buffers during mitochondria isolations. In the absence of chelators, mean Fe, Cu, Mn and Zn concentrations in our packed mitochondria were 860 ± 480 μM (n = 5), 240 ± 150 μM (n = 5), 40 ± 30 μM (n = 5) and 1,000 ± 200 μM (n = 2), respectively. Corresponding metal concentrations for packed mitochondria samples isolated in the presence of chelators were 570 ± 100 μM Fe (n = 11), 220 ± 150 μM Cu (n = 11),  20 ± 10 μM Mn (n = 11) and 330 ± 170 μM Zn (n = 5). The scatter in the Cu, Mn and Zn data precludes us from drawing strong conclusions regarding the concentration of these ions in mitochondria. However, the modest scatter for the protein and Fe concentrations measured in samples isolated in the presence of chelators indicates that these concentrations (and their ratio, approximately 10 nmol Fe/mg protein) are reliable within a relative uncertainty of 25%.

Next, we determined the proportion of the packed mitochondria samples due to the mitochondria themselves (rather than to excluded solution). Using the procedure described in “Materials and methods,” we found the percentage of mitochondria in our packed samples to be 77 ± 7 (n = 7), as shown in Table 1. The absolute concentrations of protein and Fe concentrations contained in “neat” mitochondria (devoid of solvent) could then be calculated by dividing the measured concentrations for the packed samples by 0.77, affording a protein concentration of approximately 70 mg/mL and Fe concentrations of 0.74 and 1.1 mM for samples isolated in the presence and absence of chelators, respectively. Given the uncertainty as to whether the Fe removed by chelators had any functional relevance, and on the basis of 57Fe concentration estimates based on Mössbauer intensities (see “Mössbauer spectra of mitochondria”), we conclude that the concentration of Fe in respiring yeast mitochondria is 800 ± 200 μM.

EPR of mitochondria

Mitochondria prepared in three different redox states, including as-isolated, oxidized and reduced, were packed tightly into custom-designed EPR tubes so as to expel external buffer and maximize the intensity of mitochondrial EPR signals. As-isolated samples are defined as those prepared anerobically in the absence of either oxidant or reductant. Oxidized samples were treated with either O2 or ferricyanide. Reduced samples were treated with sodium dithionite. Some samples were prepared in the presence of the metal chelators EDTA and EGTA, while others were prepared in the absence of such chelators. This was done in an attempt to distinguish EPR signals originating from functional species within mitochondria from species that were adventitiously bound to the organelle. Owing to concern that membrane integrity would be compromised by freeze/thaw cycles, samples were never used twice (i.e., they were not thawed, treated in some manner, refrozen and reanalyzed spectroscopically). Once thawed, samples were used for protein and metal analyses and then discarded. This procedure produced reasonable but not perfect correlation between the redox state in which the sample was prepared and the types and intensities of EPR signals observed.

EPR signals observed during this study are shown in Figs. 4 and 5. The principal g values for these signals and associated spin concentrations are compiled in Table 2. EPR signals of high-spin (S = 5/2) Fe3+ species were analyzed with the conventional spin Hamiltonian
$$ H = D[S^{2}_{z} - 35/12 + E/D(S^{2}_{x} - S^{2}_{y} )] + g_{0} \beta{\mathbf{S}}\cdot{\mathbf{B}}. $$
Fig. 4

Low-field X-band electron paramagnetic resonance (EPR) spectra of intact mitochondria. A Non-chelator-treated, as-isolated, B EGTA-treated, O2-exposed, C EGTA-treated, as-isolated, D EGTA-treated, reduced with 10 mM dithionite pH 7.4, E same as D but at pH 8.5. EPR conditions as follows: average microwave frequency, 9.45 GHz; microwave power, 20 mW; modulation amplitude, 10 G; receiver gain 1 × 104 for AC, 5.02 × 104 for D and E. Temperature, 10 K. The intensities of D and E have been multiplied by 5 and 2, respectively

Fig. 5

High-field X-band EPR spectra of intact mitochondria. A Non-chelator-treated, as-isolated, B a more recent preparation of non-chelator-treated, as-isolated (200 μW), C EDTA treated, NO-exposed (9.458 GHz, 200 μW, gain 5.02 × 104), D EGTA-treated, O2-exposed, E EGTA-treated, oxidized with 1 mM ferricyanide. Other conditions were as for Fig. 4 except that the average microwave frequency was 9.43 GHz. The intensities of BE have been multiplied by 5, 5, 2, and 5, respectively. Microwave power in A, D, and E was 200 μW

Table 2

Electron paramagnetic resonance (EPR) signals observed from whole mitochondria from Saccharomyces cerevisiae

Signal

Spin state parameters

g values (g1, g2, g3)

Concentration range (μM)

Tentative assignment

High-spin Fe3+ heme 1

S = 5/2, E/D = 0.041

6.9, 5.0

0–3

Cytochrome c peroxidase (Ccp1p)

High-spin Fe3+ heme 2

S = 5/2, E/D = 0.021

6.4, 5.4

0–2

Cytochrome c oxidase, heme a3:Cub

High-spin Fe3+ heme 3

S = 5/2, E/D = 0

6.0

0–1

Cytochrome c oxidase, heme a3:Cub

g = 4.3

S = 5/2, E/D = 0.33

4.27

Minor to 40

Adventitious Fe3+

gave = 2.02

S = 1/2 or spin-coupled system

2.085, 1.989, 1.985

1–20

Unassigned; possibly spin-interacting Fe/S clusters of ETF dehydrogenase

gave = 2.01

S = 1/2

2.026, 2.022, 2.003

0–5

[Fe3S4]+ probably from aconitase or succinate dehydrogenase

g = 2.00 (hyperfine)

S = 5/2; I = 5/2; a = 90 G

2.000, 2.000, 2.000

0–20

Adventitious Mn2+

g = 2.00 (isotropic)

S = 1/2

2.000, 2.000, 2.000

<2

C- or O-based organic radical

gave = 1.94

S = 1/2

2.026, 1.934, 1.913

0–23

Succinate dehydrogenase [Fe2S2]+ (Sdh2p)

gave = 1.90

S = 1/2

2.025, 1.897, 1.784

0–45

Rieske [Fe2S2]+ cluster (Rip1p)

ETF electron transfer flavoprotein

For βB ≪ |D| it is customary to describe the magnetic properties of the three Kramers doublets by effective g values, which are dependent on the rhombicity parameter E/D and the intrinsic g value g0 ≈ 2.0 [101].

Mitochondria prepared in the absence of metal chelators sometimes exhibited a signal at g = 4.3, as shown in Fig. 4, spectrum A. This signal is typical of non-heme Fe3+ species with E/D ≈ 0.27–0.33. Such spectra also typically include features at g = 6.9 and 5.0, indicative of a high-spin Fe3+ heme with E/D = 0.041. The minor signal at g = 6.0 indicates a second high-spin Fe3+ heme with E/D ≈ 0 (discussed later). In such samples, the g = 2 region (Fig. 5, spectrum A) is often dominated by a signal with a six-line hyperfine pattern (magnetic hyperfine coupling constant, a = 90 G) typical of an S = 5/2 Mn(II) species. A feature at g ∼ 1.94 is also evident but is obscured by overlap with the Mn(II) signal. In all samples, the spectral region between g = 4.3 and 2.2 was devoid of signals. A more recently prepared sample, as-isolated in the absence of chelators, exhibited a g = 4.3 signal with significantly lower intensities than that in Fig. 4, spectrum A and did not show the Mn(II) signal of Fig. 5, spectrum A; rather it exhibited the spectrum shown in Fig. 5, spectrum B (discussed later). The g ≈ 4.3 and Mn(II) signals were also either absent or present at low intensity in spectra of samples as-isolated with chelators included in all isolation buffers. This suggests that all or most of the species yielding these signals arise from adventitious Mn2+ and Fe3+ ions that can be chelated by EDTA and EGTA.

The same two high-spin heme species described above were also observed in spectra of chelator-treated as-isolated preparations, but an additional high-spin heme signal was also observed, with g = (6.4, 5.4) and E/D = 0.021. This signal was observed either alone (Fig. 4, spectrum B) or overlapped with the g = 6.0 feature (Fig. 4, spectrum C). The preparation affording the strong g = (6.4, 5.4) signal of Fig. 4, spectrum B had been exposed for approximately 20 min to O2 prior to centrifugation and freezing under standard anaerobic conditions. In chelator-treated samples, the region between g = 4.3 and 2.2 was also devoid of signals.

In general, the dominant signal in the g = 2 region from as-isolated chelator-treated samples had g = 2.026, 1.934 and 1.913 (gave = 1.94), as in Fig. 5, spectrum B. The microwave power which caused the gave = 1.94 signal intensity divided by the square root of the power to reach half maximum was P1/2 = 57 mW at 10 K. On closer inspection, a second signal, with g = 2.02, 1.90 and 1.75 (gave = 1.90) is also evident. The g values of the gave = 1.94 and 1.90 signals strongly suggest that they arise from Fe/S proteins.

An isotropic giso = 2.00 signal was observed in most preparations. The signal was broader for some preparations (Fig. 5, spectrum B) and sharper in others (Fig. 5, spectrum D). A microwave power study at 10 K indicates that the sharp giso = 2.00 signal is easily saturated at less than 1 μW. The other signals in the spectrum begin to saturate at powers greater than 80 μW.

A fourth signal with one principal g value near 2.08 can also be observed in many preparations; however, the other associated g values are poorly resolved at 10 K. Spectral overlap became less problematic at 130 K, as this signal remains slow-relaxing, while the gave = 1.94 and 1.90 signals are broadened at that temperature; spectra collected at that temperature suggest that the other features associated with the g = 2.08 resonance are near g = 1.99, affording gave = 2.02 for the signal. This was confirmed by spectral simulation and decomposition. The 10 K and 200 μW spectrum of an EGTA-treated sample was decomposed (Fig. 6, spectrum A, solid line) by simulating the gave = 1.90 (Fig. 6, spectrum B), 1.94 (Fig. 6, spectrum C), 2.00 (Fig. 6, spectrum D) and 2.02 (Fig. 6, spectrum E) signals, using g values listed in Table 2. The intensity of each simulation was adjusted to produce a sum of the four simulations (Fig. 6, spectrum A, dashed line) that best matched the experimental spectrum. Experimental spectra from other preparations gave similar deconvolutions.
Fig. 6

Simulations of the EPR spectrum of Fig. 5, spectrum B for as-isolated mitochondria without chelator. A Data (solid line) and sum (dashed line) of simulations (BE). The g values of the simulations are stated for species with gave and concentrations of B 1.90, 44 μM; C 1.94, 23 μM; D 2.00, 2 μM; E 2.02, 17 μM. Experimental conditions are the same as for Fig. 5, spectrum B

EPR of intact mitochondria treated with various redox agents

Earlier mitochondrial preparations that had been exposed to air for 1–2 days exhibited low-field regions essentially devoid of heme-containing signals. More recent preparations, exposed to O2 for 6 h, exhibited the high-spin heme signal at g = (6.4, 5.4) at high concentration (Fig. 4, spectrum B). In these samples, the g = 2 region generally consisted of intense signals with gave = 2.01 and giso = 2.00, and were largely devoid of gave = 1.94 and 1.90 signals (e.g., Fig. 5, spectrum D). A similar set of signals were observed in a sample oxidized with ferricyanide. In this case, the low-field region displayed a mixture of the g = 6.0 and g = (6.4, 5.3) high-spin heme signals, while the high-field region revealed an intense gave = 2.01 signal (spin concentration approximately 5 μM) (Fig. 5, spectrum E) along with a sizable giso = 2.00 signal (1 μM) and low-intensity gave = 1.94 and 1.90 signals.

The gave = 2.01 signal was not observed in spectra of samples treated with dithionite or in spectra of most as-isolated preparations, indicating that the species exhibiting this signal is EPR-silent when reduced. The low-field region of spectra from samples treated with dithionite was largely devoid of signals, as expected from the thermodynamic ability of dithionite to reduce Fe3+ hemes. The gave = 1.94 and 1.90 signals were present, as expected, but with concentrations similar to that observed in as-isolated samples. The reduction ability of dithionite declines as pH is lowered [102], and we anticipated that the intensity of these signals might increase significantly at pH 8.5 relative to the intensity at pH 7.4. This expectation was not fulfilled. However, an unresolved absorption-like feature at g ∼ 6.4 was observed in spectra of a sample reduced with dithionite at pH 8.5 (Fig. 4, spectrum E) but not at pH 7.4 (Fig. 4, spectrum D). This may be associated with S = 3/2 [Fe4S4]+ clusters [103].

Another preparation was exposed to 1 atm NO. This afforded a signal at g = 2.07 and a g|| = 2.01 resonance that exhibited a 14N hyperfine splitting of a = 14 G (Fig. 5, spectrum C). This signal is characteristic of a pentacoordinate heme–nitrosyl complex [101]. The spin concentration associated with this signal (20 μM) was quite high, and it may reflect the overall concentration of pentacoordinate Fe2+ heme species present, as such species are known to bind to NO to yield similar signals.

Mössbauer spectra of mitochondria

For readers not familiar with details of Mössbauer spectroscopy we have given a short tutorial-type section in the supplementary material. For the present study we have collected Mössbauer spectra from numerous samples of intact mitochondria. A spectrum from an as-isolated sample not exposed to metal chelators, shown in Fig. 7, spectrum A, exhibits three distinct spectral features (similar spectra were observed for preparations treated with metal chelators, EGTA and EDTA). Approximately 15–20% of the iron belongs to a doublet with quadrupole splitting ΔEQ ≈ 3.3 mm/s and isomer shift δ ≈ 1.3 mm/s; this doublet is outlined in the experimental spectrum. The quoted values are typical of high-spin mononuclear Fe2+ ions in penta- or hexacoordinate nitrogen/oxygen environments: FeII(H2O)6 complexes, the iron sites in reduced dioxygenases and iron superoxide dismutase, and of fully reduced binuclear iron-oxo centers at low applied field. High-spin Fe2+ hemes have distinctly smaller δ values (approximately 0.83–0.93 mm/s); however, such species would be difficult to resolve if they were to account for less than 5% of the Fe in the present samples.
Fig. 7

Low-temperature (4.2 K) Mössbauer spectra of intact as-isolated mitochondria. A Spectrum of a preparation not treated with chelators, recorded in a 45 mT applied magnetic field. The spectrum of high-spin Fe2+ components is outlined above the experimental data. B EGTA-treated mitochondria with magnetic field as for A. The dashed line indicates the ΔEQ = 1.15 mm/s doublet, a component comprising predominantly [Fe4S4]2+ clusters. The solid line represents the sum of the Fe2+ and [Fe4S4]2+ components. C Same as for B but in the presence of a parallel applied magnetic field of 8.0 T. The dashed line is a spectral simulation generated under the assumption that the ΔEQ = 1.15 mm/s component is diamagnetic. The solid line above the data is a simulation for a high-spin Fe3 component, and the solid line drawn through the data is the sum of the Fe3+ and [Fe4S4]2+ species. The Fe2+ component was not simulated because this would require use of many unknown parameters. D The 45-mT spectrum of mitochondria treated with O2/antimycin. The solid line outlines the contributions of the ΔEQ = 1.15 mm/s doublet (52%) and the Fe2+ component (12%)

A second doublet in Fig. 7, spectrum A (outlined as the dashed line in Fig. 7, spectrum B), accounting for 55–65% of the iron, has ΔEQ ≈ 1.15 mm/s and δ ≈ 0.46 mm/s.1 This doublet most probably represents Fe4S4 clusters in the 2+ core oxidation state. In this state [Fe4S4]2+ clusters have a ground state with cluster spin S = 0. Low-spin Fe2+ hemes such as cytochromes b and c have very similar ΔEQ and δ values and thus their contributions would be difficult to separate from those of [Fe4S4]2+ clusters. In principle, the cytochromes should be oxidizable and thus detectable by EPR; however, no such signals were identified in the analogous samples examined by EPR. Thus, we suspect that low-spin Fe2+ hemes do not contribute substantially to this doublet. We comment further on the ΔEQ = 1.15 mm/s component when we discuss the spectrum of Fig. 7, spectrum C.

A third component present in Fig. 7, spectrum A exhibits broad absorption extending over a velocity range of roughly 10 mm/s; this feature reflects unresolved magnetic hyperfine structure of (mostly) high-spin Fe3+ ions as well as other unidentified low-spin magnetic species. Finally, as much as 12% of the total iron may belong to S = 1/2 [Fe4S4]+ clusters (discussed below).

In principle, Mössbauer spectroscopy can be used, with some effort and proper calibration, to determine the absolute 57Fe concentration of a sample. We have done this for many years, mainly for keeping track of 57Fe enrichment in proteins, which we have calibrated with ferredoxins and dioxygenases. Thus, with our equipment, a 5-mm-thick frozen aqueous solution sample containing 1 mM 57Fe exhibiting a quadrupole doublet of 0.30 mm/s full width at half maximum yields 5% resonance absorption. Using this empirical rule (comparing the total absorption area to that under the “standard” doublet), the sample of Fig. 7, spectrum A has a 57Fe concentration of approximately 0.5 mM. A similarly prepared sample (Fig. 7, spectrum B), but treated with the chelator EGTA, has an 57Fe concentration of approximately 0.3 mM (both calculations taking into effect the solvent void volume). Although unsupplemented with Fe, the media in which the yeast were grown certainly contained some natural-abundance Fe; thus, these Mössbauer spectroscopy-based estimates may be somewhat lower than those obtained by chemical analysis since Mössbauer spectroscopy only detects 57Fe in our samples.

The EGTA-treated sample (Fig. 7, spectrum B) contains essentially the same spectral components as the sample that was not treated with chelators (Fig. 7, spectrum A); however, the proportions were somewhat different, with 40–50% in the ΔEQ = 1.15 mm/s doublet, approximately 20% high-spin Fe2+, 15% high-spin Fe3+ ions and some as yet unidentified iron. These percentages for the samples discussed in this study are summarized in Table 3. Figure 7, spectrum C was recorded at 4.2 K in the presence of an external magnetic field of 8.0 T applied parallel to the γ beam. The central feature, outlined separately by the dashed line, belongs to the feature assigned to [Fe4S4]2+ clusters. This simulation was generated with the assumption that the ΔEQ = 1.15 mm/s doublet represents iron residing in a diamagnetic (S = 0) environment, in good agreement with the data. The two absorption bands at +8 and −8 mm/s Doppler velocity belong to high-spin Fe3+ species with N/O octahedral coordination; a spectral simulation (solid line) is shown above the data. This component, representing approximately 15% of the 57Fe, is probably a collection of various mononuclear Fe3+ species with octahedral N/O ligation; such species typically have zero-field splitting parameters |D| < 2/cm and isotropic magnetic hyperfine coupling constants A0 ≈ −(27–29) MHz.2
Table 3

Summary of Mössbauer and EPR results

Fe center

O2/antimycin

As-isolated, no chelator

As-isolated, EGTA

Dithionite-treated

EPR

S = 0 [Fe4S4]2+ + low-spin Fe2+

52–57%

55–65%

40–50%

45%

 

S  = 1/2, 3/2 [Fe4S4]+

<8%

<12%

 

40% to minor

6% (gave = 2.02)

S = 2, 1/2 [Fe3S4]0/+

<5%

<5%

<5%

<5%

3% (gave = 2.01)

S = 0 [Fe2S2]2+

<5%

ND

ND

ND

 

S = 1/2 [Fe2S2]+

<12%

  

<12%

10%

High-spin Fe3+, octahedral, N/O ligands

5%

ND

15%

 

1% + adventitious

High-spin Fe2+ 5/6-CN, O/N ligands

12%

15–20%

20%

20%

 

Low-spin Fe3+

ND

   

ND

EGTA ethylenebis(oxyethylenenitrilo)tetraacetic acid, ND not determined

Figure 8 shows 4.2 K Mössbauer spectra of mitochondria treated with 10 mM dithionite at pH 8.5; the spectra were recorded in parallel applied fields of 50 mT (Fig. 8, spectrum A) and 8.0 T (Fig. 8, spectrum B). For this sample approximately 45% of the 57Fe is found to be associated with the ΔEQ = 1.15 mm/s doublet. Compared with the Fig. 7, spectrum A, there is increased absorption (from various paramagnetic species) around −1.8 and +2.2 mm/s Doppler velocity (arrows). These features most probably belong to S = 1/2 [Fe4S4]+ clusters. The solid lines overlapping the data in Fig. 8 are simulations typical of S = 0 [Fe4S4]2+ clusters (drawn as the dashed lines to represent 45% of total Fe) and from S = 1/2 [Fe4S4]+ (40%) clusters; the latter are, somewhat arbitrarily, represented by two cluster forms using the parameters of reduced aconitase (20%) and a (generic) set of parameters similar to those of the [Fe4S4]+ cluster E. coli sulfite reductase (20%). These two spectral components which have been used to represent the [Fe4S4]+ state could also be drawn into Fig. 7, spectrum A, each representing 10% of the 57Fe in that spectrum. Interestingly, with the present decomposition, approximately 85% of the 57Fe would belong to Fe4S4 clusters in the dithionite-reduced sample. This estimate is probably a bit high, as some of the absorption attributed to [Fe4S4]+ clusters may result from [Fe2S2]+ clusters (but not more than 12%; see next paragraph). Additional absorption attributed to [Fe4S4]2+ clusters may also arise from low-spin Fe2+ cytochromes. In a low external field, oxidized Fe3S4 clusters would be present as an S = 1/2 absorption extending beyond the central [Fe4S4]2+ doublet. While we see no direct evidence for the presence of such species, they could be present at concentrations below 5% of the total 57Fe.
Fig. 8

Mössbauer spectra at 4.2 K of dithionite-treated mitochondria at pH 8.5. Spectra were recorded in 50 mT (A) and 8.0 T (B) parallel applied fields. The dashed line outlines the contribution (45%) of species contained in the ΔEQ = 1.15 mm/s doublet. The solid lines drawn above the data are spectral simulations of S = 1/2 [Fe4S4]+ cluster spectra using parameters of reduced aconitase (a 20%) and parameters similar to those of Escherichia coli sulfite reductase (b 20%). The solid lines drawn through the data are the sum of the three species. The rightmost feature in A is the high-energy line of a high-spin Fe2+ component. Its contribution, at 20%, is not taken into account in the simulation

We attempted to oxidize anaerobically isolated mitochondrial samples by exposing them to air for 1–2 days. Such treatment had little effect on Mössbauer spectral features, and we suspected that this redox-buffering ability was related to the functioning of the respiratory electron transport chain. In an attempt to block this chain and thus prevent cytochrome oxidase from reducing O2, we treated a sample with antimycin A, a potent inhibitor of cytochrome bc1 [105], and then exposed it to O2. The spectrum of this sample (Fig. 7. spectrum D) exhibited the ΔEQ = 1.15 mm/s doublet with an intensity representing 52–57% of total 57Fe in the sample. The major difference between Fig. 7, spectra A and D is a decline from approximately 20% in Fig. 7, spectrum A to 12% in Fig. 7, spectrum D of the high-spin Fe2+ species. An 8.0-T spectrum (not shown) revealed that 5% of the total iron of the sample giving Fig. 7, spectrum D is high-spin Fe3+.

Because the central region of the spectrum of the O2/antimycin A treated sample is comparatively clean, we studied this sample in an expanded velocity scale. This allowed us to search for the presence of [Fe2S2]2+ and [Fe2S2]+ clusters. Because of spectral crowding and the low signal amplitudes, we do not have unequivocal evidence for the presence of this cluster type, but we can give some upper limits. We are quite confident that less than 5% of the total iron belongs to (diamagnetic) [Fe2S2]2+ clusters. Up to 12% of the iron may belong to conventional and Rieske-type [Fe2S2]+ clusters, indicating that no more than approximately 17% of the total iron in these samples arises from Fe2S2 clusters. Finally, only 8% of the total absorption may belong to [Fe4S4]+ clusters. These estimates could be improved by studying matched samples with both EPR and Mössbauer spectroscopy, and we plan to do this in the future.

Discussion

The objectives of this study were to estimate (1) the concentration of Fe in mitochondria; (2) the distribution of that Fe into various structural groups (heme, Fe/S clusters, etc.) and (3) the degree to which the redox state of these groups could be altered by treating intact mitochondria with redox agents. Our approach was to obtain EPR and Mössbauer spectra of whole intact mitochondria. We consider the study to be exploratory and more qualitative than is normally the case for studies from our laboratories, for two essential reasons. First, it is impossible to determine metal concentrations of an organelle with anywhere near the precision typical of a purified metalloprotein. Second, it is impossible to deconvolute spectra into individual and assignable protein components. We explored whether obtaining even approximate and qualitative information relevant to these objectives would provide new insights into how Fe is metabolized within these organelles.

Metal and protein concentrations

In order to determine the absolute concentration of proteins and Fe in whole intact mitochondria, we determined their concentration in packed mitochondria samples as well as the fraction of the volume due to the mitochondria themselves. Our calculations assumed that samples were devoid of impurities and that none of the radioactive sucrose used in the experiment moved into the mitochondria. Sucrose is commonly used to match the osmotic pressure of mitochondrial buffers to that within the organelles, a property that implies the inability of sucrose to penetrate mitochondrial membranes. Electron micrographs and fluorescence images of our preparations did not reveal significant contamination. The values obtained (approximately 800 μM Fe by chemical analysis and approximately 500 μM 57Fe by Mössbauer spectroscopy, and approximately 70 mg/mL protein) have relative uncertainties of about ±25%, as assessed from repeated measurements.

As far as we are aware, all previously reported metal ion contents of mitochondria have been in terms of ratios of Fe concentrations to protein concentrations, typically given in units of nanomoles of Fe per milligram of mitochondrial protein (which in our case is approximately 10 nmol Fe/mg protein). Given the complexity of the organelle, it may not be possible to fully rationalize the ratio we obtained with other reported ratios. Cobine et al. [57] measured 2.3 nmol Fe/mg for mitochondria biosynthesized under respiratory conditions when the growth medium was not supplemented with Fe, and 13 nmol Fe/mg when the medium was supplemented with Fe (our medium was not supplemented). It does not appear that metal chelators were added during isolation. The nearly sixfold observed difference suggests that the Fe-to-protein concentration ratio depends sensitively on the Fe concentration of the growth medium, and perhaps on other factors. Because these are ratios, differences may be due to changes in either Fe concentration and/or protein concentration. Other reported values range between 2.5 and 5 nmol Fe/mg mitochondrial protein [106, 107]. Tangaras et al. [108] reported 4.3 nmol Fe/mg, divided roughly into 20% heme, 50% Fe/S clusters and 30% “non-heme non-FeS”. Wallace et al. [6] reported 1.2 nmol Fe/mg, while Kispal et al. [109] reported 2 nmol/mg“free” Fe (i.e., non-heme non-Fe/S).

With use of in vivo fluorescence, the absolute concentration of chelatable Fe within mitochondria from rat hepatocyte and endothelial cells has been estimated to be between 4.8 and approximately 12 μM [110, 111]. This should represent Fe on the inside of mitochondria but not tightly associated with proteins. Mitochondria from human fibroblasts and lymphoblasts contain 1–2 μM of such chelatable Fe [112]. If similar concentrations of such Fe were present in our samples (approximately 8 μM), this would represent only approximately 1% of the total Fe in these organelles; thus, we suspect that this is an underestimate since we observe approximately 20% due to mononuclear high-spin Fe. We also caution that this refers to chelatable Fe within the organelle, and is not related to the chelatable Fe which is responsible for the difference in Fe concentrations observed with/without added chelators.

Anaerobic isolation

We isolated yeast mitochondria under anaerobic conditions to afford better control of the redox status of these organelles. Anaerobic isolation was also a precautionary measure, because a number of mitochondrial proteins are inactivated by exposure to excess O2 or by the effects of oxidative stress. For example, iron is imported into the matrix and delivered to the scaffold proteins in the reduced Fe2+ state [113]. Similarly, copper ions appear to be imported in the reduced cuprous state. In vitro Fe/S biosynthesis requires anaerobic conditions [114], biotin synthase is inactivated by O2 [18] and maximal ferrochelatase activity is observed under anaerobic conditions [115]. Under oxidizing conditions, the labile iron in the Fe4S4 cluster of aconitase dissociates into an Fe3S4 cluster, thereby inactivating the enzyme [116]. With the exception of the IM ferrochelatase, these O2-sensitive enzymes are located in the matrix (however, ferrochelatase receives Fe2+ ions from the matrix). Henze and Martin [117] and Mühlenhoff and Lill [118] suggest that the matrix is the most anaerobic compartment in O2-respiring cells. Given the predominant role of O2 in reactions occurring within mitochondria, this might seem counterintuitive. However, the O2-consuming reactions occur at the IM which encapsulates the matrix, and these reactions may occur fast enough to effectively remove any O2 that diffuses into the matrix.

In the oxidized inactivated state, the [Fe3S4]+ cluster of aconitase affords an EPR signal with gave = 2.01 [5, 119], similar to that observed here. The Fe3S4 cluster of succcinate dehydrogenase exhibits a similar signal in the oxidized state [2]. This signal was absent in all but one of the as-isolated samples, which may have been slightly oxidized relative to other as-isolated samples. The more intense Fe3+ heme signals exhibited by this particular batch relative to the other as-isolated samples are congruent with this possibility. The absence of the gave = 2.01 signal in spectra of our anaerobically prepared samples (and in spectra of dithionite-reduced mitochondria) as well as the presence of this signal under oxidizing conditions indicate that this signal probably arises from an oxidized [Fe3S4]+ cluster, either from inactivated aconitase, homoaconitase or the [Fe3S4]+ cluster of succinate dehydrogenase.

Adventitious Fe and Mn

The hyperfine-split signal observed in the g = 2 region of various as-isolated samples is typical of adventitious or weakly bound Mn2+ ions and we assign it as such. No additional Mn-based signals were observed in any of our samples, which suggests that the two known Mn-containing proteins in yeast mitochondria, Sod2p and Mtm1p, are present at concentrations below our EPR detection limit. Sod2p is a matrix-localized manganese superoxide dismutase while Mtm1p is an IM manganese chaperone [58]. Based on S = 5/2 spin Hamiltonian simulations using the known parameters for Sod2p, namely, D = 0.348/cm and E/D = 0.026 [119, 121], our calculations indicate that we could detect a minimum concentration of 10 μM Mn-Sod2p. Thus, we suspect that Mn-Sod2p is present at a concentration less than this.

The g = 4.3 signal probably arises from Fe3+ ions that are adventitiously bound to mitochondrial proteins or membrane phospholipids. Our studies show that this type of iron can be removed by EDTA and EGTA, suggesting that it resides in a region that can be accessed by these chelators, such as the outer face of the OM or perhaps the IMS. In more recent preparations, the amount of this adventitious Fe3+ appears to be minimal, even in samples prepared in the absence of chelators. The presence of aqueous Fe3+ ions in the OM or the IMS suggests that this region has an electrochemical potential sufficiently high to support this state. Previous reports have used the g = 4.3 signal as a quantitative diagnostic for “free” Fe generated by cellular damage [121], but our study suggests that caution should be applied in these interpretations in that this signal may also arise from Fe3+ peripherally associated with these organelles.

Species containing Fe/S clusters

The gave = 1.94 signal was a reproducible spectral feature of all as-isolated and dithionite-reduced samples. We tentatively assign this signal to the [Fe2S2]+ cluster of succinate dehydrogenase. This cluster, when reduced, exhibits an EPR signal with g values nearly identical to those observed here [123]. A similar signal in other reported mitochondrial preparations has been observed and assigned similarly [77]. However, the assignment should be cautiously accepted because gave = 1.94 signals are characteristic of both [Fe4S4]+ and [Fe2S2]+ containing proteins and we cannot exclude the possibility that the observed signal arises from one or more [Fe4S4]+ clusters rather than, or in addition to, the [Fe2S2]+ cluster in Sdh2p.

Further support for assigning the gave = 1.94 signal to the [Fe2S2]+ cluster of succinate dehydrogenase comes from the saturation properties of the gave = 1.94 signal (P1/2 = 57 mW at 10 K), which are similar to those reported for the succinate dehydrogenase [Fe2S2]+ cluster under conditions where the Fe4S4 cluster in the same enzyme is reduced to the paramagnetic 1+ state [3]. When the Fe4S4 cluster is oxidized to the diamagnetic 2+ state, the saturation behavior of the [Fe2S2]+ cluster differs substantially (P1/2 = 0.63 mW). The observed saturation behavior suggests that the Fe4S4 cluster is reduced to the 1+ core oxidation state in our samples, and spectral features due to this species should be a component of the Mössbauer spectra shown in Fig. 7. Our inability to observe the broad EPR features reported for the [Fe4S4]+ cluster of Sdh2 is not surprising as these low-intensity features are easily missed [3]. Since E°′ for the [Fe4S4]2+/+ couple is −270 mV [124, 125], this implies that the potential of the solution for which this cluster is in redox-equilibrium is at or below this value.

The gave = 1.90 signal is similar to that exhibited by the isolated Rieske [Fe2S2]+ protein [4], and we assign it as such. It should contribute to the magnetic components of Fig. 7. E°′ for the 2+/1+ redox couple of this cluster is +280 mV [126]. This protein is part of the cytochrome bc1 complex, which is located in the IM, but the Rieske protein itself is tethered to the rest of the complex and extends into the IMS. This cluster transfers electrons to the IMS protein cytochrome c, suggesting that it is in redox-equilibrium with the IMS. Assuming this, the presence of this signal suggests that the potential of the IMS is less than +280 mV in our samples.

Species containing hemes

Two of the observed signals which typify S = 5/2 Fe3+ hemes, including the g = (6.4, 5.4) signal with rhombic symmetry and the g = 6.0 signal with axial symmetry (E/D ∼ 0) most likely arise from cytochrome c oxidase. Signals with identical g values have been reported to arise from the heme a3:Cub active site in an intermediate redox state in which Cub is reduced to the 1+ state, while heme a3 is high-spin Fe3+ [70, 127, 128]. These signals are observed for samples that were oxidized by O2 or ferricyanide (and in spectra from the single “as-isolated” batch that exhibited the gave = 2.01 signal and was slightly more oxidized than the others). The occurrence of this signal probably requires mildly oxidizing conditions, in that exposure to O2 in our protocol was followed by a relatively slow anaerobic packing procedure during which time some re-reduction could have occurred. The absence of these high-spin Fe3+ signals in the dithionite-treated samples is consistent with the ability of dithionite to reduce Fe3+ heme a3. This behavior is also consistent with a reduction potential for the Fe3+/Fe2+ heme a3 site of approximately +350 mV [129].

In EPR studies of isolated cytochrome c oxidase, the combined quantified intensity of these signals corresponded to 23–50% of the cytochrome c oxidase concentration [70, 127, 128]. Since the maximum combined spin concentration observed here for these signals was approximately 3 μM, these percentages suggest a minimum cytochrome c oxidase concentration in mitochondria of 6–13 μM. The combined heme a plus heme a3 concentration in mitochondria has been estimated at 0.15–0.3 μmol/g mitochondrial protein [130], which suggests a cytochrome c oxidase concentration (assuming 70 mg/mL protein concentration) of 5–10 μM, close to what we observe by spin quantification. The intense signal that developed upon treating mitochondria with nitric oxide undoubtedly arose from pentacoordinate heme–nitrosyl groups [104] and it indicates a minimum heme concentration in our samples of 20 μM. A significant contribution to this signal is likely from NO binding to the Fe2+ heme a3 of cytochrome c oxidase. Considered collectively, we suspect that the concentration of cytochrome c oxidase in mitochondria from respiring yeast is between 6 and 20 μM.

The third high-spin Fe3+ heme signal (g = 6.8, 5.0; E/D = 0.042) probably originates from heme b in cytochrome c peroxidase (Ccp1p), as similar g values have been reported [131, 132]. The particular degree of rhombic distortion in this heme depends on pH and subtle structural changes. The redox potential for the Fe3+/Fe2+ couple of this IMS protein is −182 mV at pH 7 [133]. If this signal arises from Ccp1p, the potential of the IMS region in our samples would appear to be greater than this value. Alternatively, this Fe3+ heme signal may originate from matrix-localized catalase A (Cta1p) or flavohemoprotein (Yhb1p) as the Fe3+ states of these proteins exhibit similar g values [134, 135]. A similar signal from mitochondria from Spodeptera littoralis was assigned to catalase [77]. However, we expect that the potential of the matrix would be sufficiently low to reduce these centers fully.

EPR signals from low-spin Fe3+ hemes are typically found at gz = 3.7–2.4 and gy = 2.5–2.1 [136], but no recognizable signals were found in this region. Signals from low-spin Fe3+ hemes were probably not observed either because such groups were in the Fe2+ state or because they are highly anisotropic with very broad signals. There are a number of such groups in mitochondria (e.g., cytochromes b, b2, c, c1, heme a), many of which should be in redox-equilibrium with the IMS. Previous studies reported concentrations of 0.2–0.4 μmol/g protein for cytochrome c and 0.07–0.25 μmol/g protein for cytochrome c1 [130], both of which correspond to easily detectable concentrations in our samples with 70 mg/mL protein. With an estimate of 0.8 μM for the concentration of flavocytochrome b2 (Table 4), this protein would also be detectable. Since E°′ = +290 mV for cytochrome c [137], +230 mV for cytochrome c1 [138], −3 mV for flavocytochrome b2 [139] and +255 mV for heme a [140], it seems likely that the potential of the IMS in our samples was below approximately 0 mV, rendering these centers ferrous and EPR-silent.
Table 4

Iron-containing proteins in mitochondria from S. cerevisiae

Protein

Gene

Copies per cell

Concentration (μM)

Location

Prosthetic group

E°′ (mV, NHE)

Electronic and magnetic properties

Cytochrome c isoform I

cyc1

7,730

1.2

IMS [46]

Heme c

+290 [137]

S = 1/2 Fe3+ (g = 3.06, 2.26, 1.25) [32] and S = 0 Fe2+

Cytochrome c isoform II

cyc7

1,310

0.2

IMS [46]

Heme c

+286 [156]

S = 1/2 Fe3+ (g = 3.2, 2.05, 1.39) [32] and S = 0 Fe2+

Cytochrome c peroxidase

ccp1

6,730

1.1

IMS [157]

Heme b

−182 [133]

S = 5/2 Fe3+ (g = 6.60, 5.23, 5-CN; and g = 6.13, 5.81, 6-CN) [158] and S = 2 Fe2+

Flavocytochrome b2

cyb2

4,590

0.8

IMS [157]

Heme b2

−3 [139]

S = 1/2 Fe3+ (g = 2.99, 2.22, 1.47) [159] and S = 0 Fe2+

Cytochrome bc1

rip1

IM (facing IMS) [28, 155]

Fe2S2 (Rieske)

+285 [126, 160]

S = 0 [Fe2S2]2+ and S = 1/2 [Fe2S2]+ (g = 2.02, 1.90, 1.80) [4]

Cytochrome bc1

cyt1

39,900

6.6

IM (redox with rip1) [28, 155]

Heme c1

+230 (est) [138]

S = 1/2 Fe3+ (g = 3.33 or 3.35) [161, 162] and low-spin Fe2+

Cytochrome c oxidase

cox1

IM [163]

Heme a

+320 [31]

S = 1/2 Fe3+ (g = 3.03, 2.21, 1.45) [127, 164] and low-spin Fe2+

Cytochrome c oxidase

cox1

IM [163]

Heme a3:Cub

+350 [129]

Fully oxidized: EPR-silent Fe3+ spin-coupled to Cu2+ with J ∼ 1/cm. Intermediate: S = 5/2 Fe3+ (g = 6.4, 5.3) [127] mixed with (g = 6.0) when Cu+. Fully reduced: high-spin Fe2+:Cu+ [164, 165]

Succinate dehydrogenase

sdh3:sdh4

238:7,920

0.04:1.3

IM [1]

Heme b

+60 [166] (but this is for non-Sc enzyme which has novel Cys)

S = 1/2 Fe3+ (g = 3.63) [167] and S = 0 Fe2+

Cytochrome bc1

cob1

IM [28, 155]

Heme bH

−45 (−35 to +25) [155]

S = 1/2 Fe3+ (g = 3.45) [162] and S = 0 Fe2+

Cytochrome bc1

cob1

IM [28, 155]

Heme bL

−150 (−95) [155]

S = 1/2 Fe3+ (g = 3.78) [162] and S = 0 Fe2+

Ferrochelatase

hem15

22,700

3.8

IM (facing M) [42]

Mononuclear Fe

S = 2 Fe2+ (δ = 1.36 mm/s; ΔEQ = 3.04 mm/s) [115]

Succinate dehydrogenase

Sdh2

9,540

1.6

IM (facing M) [1]

Fe2S2

0 [3]

S = 0 [Fe2S2]2+ and S = 1/2 [Fe2S2]+ (g = 2.026, 1.935, 1.912) [3, 166]

Succinate dehydrogenase

Sdh2

9,540

1.6

IM (facing M) [1]

Fe3S4

+60 [3]

S = 1/2 [Fe3S4]+ (g = 2.01) and S = 2 [Fe3S4]0 [166]

Succinate dehydrogenase

Sdh2

9,540

1.6

IM (facing M) [1]

Fe4S4

−260 [3]

S = 0 [Fe4S4]2+ and S = 1/2 [Fe4S4]+ (g = 2.064, 1.992, 1.847 and magnetic interactions affording features at 2.27 and 1.63) [3]

Heme monooxygenase

cox15

IM [35]

Heme a [34, 168]

+242 [168]

S = 1/2 Fe3+ (g = 3.5) and S = 0 Fe2+, [168]

Heme monooxygenase

cox15

IM [35]

Heme b [34, 168]

+85 [168]

S = 1/2 Fe3+ (g = 3.7) and S = 0 Fe2+, [168]

Carboxylate monoxygenase

Coq7

IM [169]

Fe–O–Fe [48, 170]

+48 and −135 [171]

(Putative) S = 0 [Fe2+ Fe2+], S = 1/2 [Fe3+ Fe2+] (g = 1.95, 1.86, 1.77) and S = 4 [Fe2+ Fe2+] [172]

ETF dehydrogenase

YOR356W (putative)

3,320

0.6

IM

Fe4S4 [26]

+47 [27]

S = 0 [Fe4S4]2+ and S = 1 /2 [Fe4S4]+ (g = 2.086, 1.939, 1.886) [25]

Aconitase

Aco1

96,700

16

M [173]

Fe4S4 and Fe3S4

−450, −268, +100 [174]

S = 0 [Fe4S4]2+ and S = 1/2 [Fe4S4]+ (g = 2.06, 1.93, 1.86) [5] S = 1/2 [Fe3S4]+ (g = 2.024, 2.016, 2.004) and S = 2 [Fe3S4]0S = 1/2 [Fe4S4]3+ and S = 0 [Fe4S4]2+

Homoaconitase

Lys4

7,350

1.2

M [6]

Fe4S4 and Fe3S4 (putative)

Similar to aconitase [6]

Similar to aconitase [6]

Ferredoxin

Yah1

14,800

2.4

M [7]

Fe2S2

−353 [8]

S = 0 [Fe2S2]2+ and S = 1/2 [Fe2S2]+ (g = 2.024, 1.937, 1.937) [8]

Fe/S scaffold protein

Isu1

10,800

1.8

M [22]

Fe2S2

(Probably low)

S = 0 [Fe2S2]2+ [175]

Fe/S scaffold protein

Isu2

3,420

0.6

M [22]

Fe2S2

(Probably low)

S = 0 [Fe2S2]2+ [175]

Fe/S scaffold protein

Isa1

125

0.02

M [23]

Fe2S2

(Probably low)

S = 0 [Fe2S2]2+ [175]

Fe/S scaffold protein

Isa2

1,560

0.3

M [174] or IMS [23]

Fe2S2 (putative) [176]

(Probably low)

S = 0 [Fe2S2]2+ [175]

Fe/S scaffold protein

Nfu1

11,300

1.9

M [22]

Fe2S2

(Probably low)

S = 0 [Fe2S2]2+ [175]

Fe/S scaffold protein

Isu1

10,800

1.8

M [22]

Fe4S4

(Probably low)

S = 0 [Fe4S4]2+ [175]

Fe/S scaffold protein

Isu2

3,420

0.6

M [22]

Fe4S4

(Probably low)

S = 0 [Fe4S4]2+ [175]

Fe/S scaffold protein

Isa1

125

0.02

M [23]

Fe4S4

(Probably low)

S = 0 [Fe4S4]2+ [175]

Fe/S scaffold protein

Isa2

1,560

0.3

M [176] or IMS [23]

Fe4S4 (putative) [176]

(Probably low)

S = 0 [Fe4S4]2+ [175]

Fe/S scaffold protein

Nfu1

11,300

1.9

M [22]

Fe4S4

(Probably low)

S = 0 [Fe4S4]2+ [175]

Biotin synthase

Bio2

504

0.08

M [177]

Fe4S4

−440 [178]

S = 0 [Fe4S4]2+ and S = 1/2 [Fe4S4]+ (g = 2.042, 1.937, 1.937) [18] or (g = 2.035, 1.937, 1.937) [14] or (g = 2.044, 1.944, 1.914 and S = 3/2) [179]

Lipoic acid synthase

Lip5

1,630

0.3

M [180]

Fe4S4

−505 [181]

S = 0 [Fe4S4]2+ and S = 1/2 [Fe4S4]+ (g = 2.039, 1.937, 1.937) [18]

Biotin synthase

Bio2

504

0.08

M [177]

Fe2S2

−140 [178]

S = 0 [Fe2S2]2+ and S = 1/2 [Fe2S2]+ (g = 2.01, 1.96, 1.88 and g = 2.00, 1.94, 1.85) [14, 182]

Lipoic acid synthase

Lip5

1,630

0.3

M [180]

Fe2S2

−430 [181]

S = 0 [Fe2S2]2+ and S = 1/2 [Fe2S2]+

Dihydroxyacid dehydratase

Ilv3

171,000

28

M (putative) [19]

Fe4S4 (putative)

(Dithionite-reducible) [20]

S = 0 [Fe4S4]2+ and S = 3/2 [Fe4S4]+ (g = 5.2, 4.7) [19, 20]

Frataxin homolog

Yfh1

1,560

0.3

M [183]

2 mononuclear Fe’s [184]

(Probably high)

S = 5/2 Fe3+ and Fe2+ [184, 185]

Catalase A

Cta1

623

0.1

M [37]

Heme b

−226 (est [134])

S = 5/2 Fe3+ (g = 6.48, 5.10) [186]

Flavohemoglobin

Yhb1

13,000

2.2

M [135] (and cytosol)

Heme b

−230 to −320 (est [187])

S = 5/2 Fe3+ (g = 5.75, 6.47, 5.22) [188]

NHE normal hydrogen electrode, IMS intermembrane space, IM inner membrane, M matrix

Organic radical species

The isotropic giso = 2.00 signal has g values and saturation properties typical of organic radicals, and we assign this signal to the population of such radicals in our samples. Possible sources include the semiquinone states of flavins and ubiquinone, and conceivably reactive oxygen species. Given the preponderance of ubiquinone (0.6–4.0 μmol/g mitochondrial protein) [141] and flavin-containing proteins (see “Introduction”), we were surprised that the spin concentrations of the giso = 2.00 signal were so low. This circumstance may have arisen because our mitochondria were isolated anaerobically such that any radical species generated during cell growth could have decayed during the lengthy isolation period. The increased intensity of the giso = 2.00 signal for samples prepared under oxidizing conditions supports this possibility and highlights the importance of preparing these organelles anaerobically.

An unassigned mitochondrial EPR signal

Our samples exhibited an EPR signal with a distinct resonance at g = 2.08 and having gave = 2.02. Positive features near g = 2.08 are usually associated with Fe4S4 clusters, however our signal does not show the corresponding higher-field features typical of such clusters; rather, the partners for this species appear to be in the 2.00 region. Broader signals at or near g = 2.08 have been assigned to a spin-coupled cluster involving the reduced S2 cluster of complex II (succinate CoQ oxidoreductase) [3, 142] and we have considered assigning it as such. We have also considered assigning it to ETF dehydrogenase [25, 26, 27]. Given the uncertainties, we leave this signal unassigned pending further study.

Absence of Cu2+-based signals

We did not observe signals characteristic of Cu2+ ions, even though, by chemical analysis, our samples contained copper at detectably high concentrations. The lack of such signals suggests that the vast majority of Cu in our samples is in the diamagnetic Cu+ state. The most well-known Cu centers in mitochondria are the CuA and CuB sites in cytochrome c oxidase. Oxidized CuA exhibits an EPR signal with g = 2.17 [143], but no such signal was obviously present. E°′ for CuA is +240 mV [144]. This center should be in redox-equilibrium with the IMS, as it functions by accepting electrons from cytochrome c. Since there is a detectable concentration of cytochrome c oxidase in our samples, this implies that the absence of a signal arises because the potential of the IMS is less than approximately +230 mV. Mitochondria also contain a number of Cu chaperones, but such centers were probably in the diamagnetic Cu+ state [145, 146]. The majority of Cu in yeast mitochondria appears unassociated with proteins and located in the matrix in a Cu+ form [57]. Our results are consistent with this, both in terms of the concentration of Cu observed and the absence of Cu2+-based EPR signals. However, we remain puzzled why no Cu2+ signals were observed under oxidizing conditions.

Electrochemical potentials of mitochondrial compartments

It would be difficult to interpret our EPR results by assuming that all redox centers in our mitochondrial samples sensed the same electrochemical potential. Given the likelihood that the [Fe2S2]+ cluster of succinate dehydrogenase was observed, with saturation properties indicating that the Fe4S4 cluster of this enzyme was also reduced, the potential of the solution with which this cluster is in redox-equilibrium should be less than approximately −270 mV. These clusters of succinate dehydrogenase should be in redox-equilibrium with the mitochondrial matrix, which implies a matrix potential in our samples of less than −270 mV. We observe an oxidized Fe3+ heme which is most likely from the IMS cytochrome c peroxidase, and the lack of a signal from flavocytochrome b2 suggests that it is in the reduced state. This implies an IMS solution potential in the range from −200 to 0 mV in our samples. If we assume a potential difference between the IMS and the matrix of 180 mV [147], an IMS potential range of −100 to −200 mV would predict a matrix potential range of −300 to −400 mV, which are both compatible with our observations.

Mössbauer spectroscopy

To date, only one Mössbauer spectrum of wild-type mitochondria has been published and it was devoid of any signals [39]. In contrast, a sample of the mutant Δyfh1 displayed a quadrupole doublet with ΔEQ = 0.67 mm/s and δ = 0.52 mm/s which was assigned to amorphous nanoparticles of iron(III) phosphate; other Fe-containing components were not observed. We have recorded Mössbauer spectra of more than 25 preparations of intact mitochondria, over the course of 2 years and involving various group members preparing these samples. Also, the Mössbauer spectroscopy and EPR studies developed independently, and we have therefore not studied aliquots of the matched samples with both techniques. Isolation procedures were adjusted based on feedback from our Mössbauer spectroscopy and EPR results, so it is not surprising that we observed some variation in the concentration of the various spectroscopic components.

For studies of mitochondria, EPR and Mössbauer spectroscopy are complementary. EPR detects with high sensitivity species with half-integral spin (Kramers systems), while Mössbauer spectroscopy, in this first study, is mainly useful in the exploration of components with integer or zero electronic spin, i.e., components either not accessible (diamagnetic complexes) or only difficult to access by EPR. We consider our present Mössbauer spectroscopy results to be preliminary, but we believe that the proven power of the technique can be exploited in the future, once the system is dissected by metabolic and/or genetic manipulations, e.g., overexpression or deletion of particular mitochondrial proteins. Our present studies suggest that one should conduct the Mössbauer spectroscopy studies at 4.2 K and in weak as well as in strong applied fields. A strong-field capability is essential as it allows one to identify S = 0 species such as [Fe4S4]2+ and [Fe2S2]2+ clusters. Spectra recorded in strong applied fields distinguish also between monomeric high-spin Fe3+ and nanoparticles containing high-spin Fe3+.

Approximate iron distribution in mitochondria from respiring yeast

Our Mössbauer spectroscopy and EPR results are insufficient to establish precisely how Fe ions in mitochondria are distributed, but they are sufficient to allow us to draw some approximate and preliminary conclusions, which are summarized by the pie chart shown in Fig. 9. The majority of Fe in mitochondria from respiring yeast is present as Fe4S4 clusters; in Fig. 9 we estimate this to be approximately 60%, but values as low as 50% and as high as 85% are possible. In the as-isolated state, most of these Fe4S4 clusters are in the 2+ state, but a substantial fraction can be reduced to the 1+ state by incubation of mitochondria with dithionite at pH 8.5. Thus, in Fig. 9 we distinguish dithionite-reducible [Fe4S4]2+/+ clusters from irreducible [Fe4S4]2+ clusters. The next most abundant class of Fe-containing species in mitochondria, representing approximately 20% of the Fe in Fig. 9 (but with an acceptable range of 15–35%), is high-spin non-heme Fe. In the as-isolated state, most of these ions are high-spin Fe2+, with smaller proportions of high-spin Fe3+. We have not observed low-spin Fe3+ ions in any of our samples. Our EPR results suggest that approximately 9% (but with a range of 5–11%, as limited by our Mössbauer spectroscopy analysis) of the Fe is present as [Fe2S2]+ clusters, whereas no such clusters in the 2+ state were observed. The ΔEQ = 1.15 mm/s doublet, attributed essentially to [Fe4S4]2+ clusters, may contain contributions from low-spin ferrous hemes. The remaining few percent of mitochondrial Fe is present as high-spin heme and perhaps Fe3S4 centers.
Fig. 9

Comparison of observed and calculated percentile Fe distribution. Percentages used in the pie chart for the observed distribution were as follows: 37% ([Fe4S4]2+ + low-spin Fe2+ hemes), 25% [Fe4S4]2+/+, 22% high-spin (Fe3+ + Fe2+) non-heme mononuclear, 9% [Fe2S2]2+/+, 4% high-spin hemes (Fe3+ + Fe2+) and 3% [Fe3S4]+/0 clusters. These values are based on our results, collectively considered, but should be viewed as a hypothesis, with substantial latitude in our estimates for each category. Calculated percentages were 78% [Fe4S4]2+/+, 5% low-spin hemes (Fe2+ only), 5% high-spin Fe3+/2+ non-heme mononuclear, 8% [Fe2S2]2+/+, 2% high-spin hemes (Fe3+/2+) and 2% [Fe3S4]+/0. These values are taken from Table S1, assuming 8 μM nonproteinatious high-spin non-heme Fe2+ ions. The concentration of Fe associated with each Fe-containing mitochondrial protein was calculated and percentages were obtained by dividing each individual Fe concentration by the sum of all such values and multiplying by 100. HS low spin, LS low spin

The spectral simulations of Fig. 8 suggest that the main part of the magnetic features represent S = 1/2 [Fe4S4]+ clusters. While we are reasonably certain that the magnetic features reflect paramagnetic Fe/S clusters, we do not wish to suggest that the entire magnetic component belongs to the S = 1/2 forms (the gave = 1.94 species) of [Fe4S4]+ clusters; we suspect that S = 3/2 [Fe4S4]+ as well as [Fe2S2]+ clusters, albeit to a lesser extent, also contribute. By EPR, we see signals assigned to these latter species (S = 3/2 [Fe4S4]+ and S = 1/2 [Fe2S2]+ clusters). By studying the Mössbauer and EPR spectra of aliquots of the same dithionite-treated sample, one should be able to assess the cluster type and concentration of the reduced Fe/S species better.

Regarding high-spin ferrous species, many hexa-and pentacoordinated complexes with N/O ligation contribute doublets with ΔEQ = 3.0–3.5 mm/s and δ = 1.3–1.4 mm/s. With a few exceptions the high-field spectra of these complexes are broad and difficult to analyze.

Included in the [Fe4S4]2+ cluster portion are clusters that convert into Fe3S4 clusters upon oxidation (e.g., aconitase); this fraction could represent as much as approximately 25 μM Fe (3% of the total). Treatment with dithionite at pH 8.5 causes approximately half of the [Fe4S4]2+ portion to become reduced to the [Fe4S4]+ state.

Mössbauer spectra of as-isolated mitochondria indicate that approximately 15% of the Fe is paramagnetic and in half-integer spin states. According to our EPR results, this would include approximately 3 μM due to the high-spin heme signal from cytochrome c peroxidase/catalase, approximately 3 μM due to the high-spin heme signals from cytochrome c oxidase, approximately 20 μM due to g = 4.3 high-spin Fe3+, approximately 40 μM due to the [Fe2S2]+ cluster of the Rieske protein and approximately 20 μM due to the [Fe2S2]+ cluster from succinate dehydrogenase. The [Fe4S4]+ cluster of succinate dehydrogenase might also contribute to the paramagnetic component. Summing these EPR contributions affords approximately 90 μM Fe, translating into approximately 11% of the total Fe, in qualitative agreement with what is observed by Mössbauer spectroscopy.

Comparison with known Fe-containing proteins in mitochondria

We have organized known mitochondrial proteins starting with the results of three proteomic studies [148, 149, 150], the information provided by the Saccharomyces Genomics Database and the reconstructed metabolic network of Forster et al. [151]. The integration of this information led to the identification of approximately 600 candidate mitochondrial proteins, a number comparable to the approximately 800 proteins estimated to constitute the complete yeast mitochondrial proteome [150]. Primary research literature describing properties of each of these proteins was accessed using the Web of Science (http://www.isi10.isiknowledge.com) and information specifically regarding Fe content and suborganellar localization was sought.

The result of this analysis afforded the proteins and protein complexes included in Table 4. It is difficult to establish that this or any such list is complete, and we suspect that there are Fe-containing mitochondrially localized proteins that are not included. Some such proteins might be unidentified currently, or the presence of Fe in them might be uncertain. Also not included in this list are proteins that are known to interact with Fe (e.g., transporters) but for which no Fe-bound state has been characterized.

The concentrations of many of these proteins within the mitochondria have been estimated. Ghaemmaghami et al. [153] created a comprehensive fusion library of S cerevisiae cells in which each member had a different open reading frame tagged with the same epitope. Natural expression levels of the corresponding fusion proteins were quantified to afford copy numbers per cell (Table 4). The volume of an S. cerevisiae cell is approximately 1 × 10−13 L, and mitochondria occupy approximately 10% of this [154]. Thus, one copy of a mitochondrial component per cell reflects a concentration of 170 pM in the organelle. Such information can be useful in interpreting Mössbauer spectra.

It is interesting to compare this list with the results observed in this study. We calculated the overall Fe concentration in mitochondria implied by the proteins and concentrations given in this table, by summing the products of the concentration of each known Fe-containing mitochondrial protein (Ci) and the number of Fe’s per protein (mi); i.e., \( {\text{[Fe]}}_{{{\text{overall}}}} = {\sum\nolimits_{i = 1}^n {m_{i} C_{i} } } \) (if the number of copies per cell was not reported, a concentration of 1 μM was assumed). The calculated overall concentration of Fe in mitochondria was 265 μM (Table S1), corresponding to only one third of our experimentally determined value. If correct, this suggests that two thirds of the detected Fe in yeast mitochondria is not accounted for by Table 4, assuming the protein concentrations in that table. On the other hand, systematically low estimates of protein concentration in the literature might also be responsible for this discrepancy. Tempered by this caveat, some interesting trends are apparent, namely:
  • A large proportion of Fe in mitochondria appears to be associated with the Fe4S4 cluster from a single protein, namely, dihydroxyacid dehydratase.

  • The OM is devoid of Fe-containing proteins.

  • The matrix contains few heme-containing proteins (only catalase and flavohemoglobin).

  • The matrix is dominated by Fe/S centers, especially [Fe4S4]2+ clusters.

  • Only one mitochondrial protein with an Fe–O–Fe center is known (Coq7p).

In order to compare our experimental EPR and Mössbauer spectroscopy results with predictions made from these calculations, two additional pieces of information for each site are required, namely, the region of the mitochondria with which the site is in redox-equilibrium and the solution potential of that region under the conditions when our samples were frozen. For some redox centers (e.g., cytochrome c), there is no doubt as to the region with which they are in redox-equilibrium (e.g., the IMS), but such information is not certain for all entries in Table 4. Nor is the solution potential for each region of the mitochondria (under the specific conditions for which our samples were prepared) known. For proteins located in either aqueous region (IMS or matrix), the region with which they were assumed to be in redox-equilibration was the region where the proteins were located. For proteins located in the IM, there were a number of possibilities. Some sites extend into either the IMS or the matrix, and if such sites are known to accept/donate electrons with donors/acceptors in that aqueous region, such centers were deemed to be in redox-equilibrium with that aqueous region rather than with the IM. Other sites contained within IM-bound respiratory complexes might be along a known electron pathway which leads to either aqueous region, and such sites were deemed to be in redox-equilibrium with that aqueous region. In a few cases, redox-dependent protein conformation changes occur such that sites contained within those proteins might be in redox-equilibrium with more than one region, depending on the protein conformation at the time our samples were frozen, and so no definitive assignment could be made. Finally, the active site for cytochrome c oxidase (heme a3:Cub) might be unique in being in redox-equilibrium with the O2/H2O couple (E° ∼ +800 mV). For our samples prepared under anaerobic conditions, the appropriate non-standard-state reduction potential to use would be much less than +800 mV. Our set of tentative assignments is given in Table S1.

Next, we estimated the solution potential of the IMS and matrix to be EIMS ≈ −0.1 and Ematrix ≈ −0.3 V, respectively (see discussion above). EIM probably lies somewhere between these two values and may be controlled by the E° value for the CoQox/CoQred couple, namely, +60 mV in the IM [155]. What must separate the redox potential of one region from another and maintain regions in redox isolation are the redox-dependent conformational changes known for the IM-bound respiratory complexes.

Then, we grouped all species of Table 4 that would give rise to equivalent Mössbauer spectral features, and calculated the concentration of Fe associated with each group. This procedure resulted in the nine groups listed in Table S1. The calculated percentages of mitochondrial Fe in various forms are also shown in Fig. 9. Comparison with what we have observed in this study indicates overall qualitative agreement. The greatest apparent discrepancy is the greater percentage of high-spin non-heme Fe observed in our samples relative to that predicted by the calculations. The qualitative similarities in the calculated versus observed distribution of Fe in mitochondria indicates a general agreement between our results and the calculated contents of these organelles. Calculations predict that there should be fewer high-spin Fe2+ ions and more ([Fe4S4]2+ clusters + low-spin Fe2+ hemes) than we observe. Some of the observed high-spin Fe2+ ions may be adventitiously bound, despite our attempt to remove such ions by chelation. Alternatively, some Fe-containing mitochondrial proteins have not been included in the calculations, or a portion of these ions might represent a transient and chelatable ferrous ion pool, as has been reported previously [111]. In this latter case, the concentration of ions estimated for this pool (2–12 μM Fe) would be substantially less than we observe (approximately 180 μM).

Conclusions

The major contributions of this study are:
  1. 1.

    Protein and metal concentrations. We determined the absolute concentrations of protein and Fe in “neat” (solvent-free) yeast mitochondria, synthesized by respiring cells and isolated in the presence of metal chelators to be approximately 70 mg/mL and 800 ± 200 μM, respectively.

     
  2. 2.

    EPR signals from proteins containing Fe/S cluster and heme prosthetic groups. Signals were observed that have been tentatively assigned to succinate dehydrogenase, the Rieske Fe/S protein, aconitase, cytochrome c oxidase and cytochrome c peroxidase. An intense signal with gave = 2.02 was observed. Although unassigned, this signal probably originates from an Fe/S cluster.

     
  3. 3.

    Species not observed. No signals from Cu2+ ions, Mn2+ superoxide dismutase or low-spin Fe3+ hemes were observed. These species may be present at undetectably low concentrations or in EPR-silent states. The collective concentration of organic-based radical species was unexpectedly low.

     
  4. 4.

    Electrochemical compartmentalization. In the as-isolated state of our samples, the ranges of the potentials for the IMS and the matrix are −0.2 < EIMS < −0.1 V and −0.4 < Ematrix < −0.3 V (vs NHE), respectively.

     
  5. 5.

    Dominance of Fe4S4 clusters. The majority of Fe (50–85%) in mitochondria isolated from respiring cells is present in this form.

     
  6. 6.

    High-spin ferrous ions. Approximately 20% of the Fe in mitochondria is present as high-spin non-heme Fe2+ ions with five to six O/N ligands coordinating. A portion of this is probably adventitiously bound to the mitochondria, while the remainder is associated with mitochondrial proteins and/or conceivably a “pool” of such ions, perhaps in the mitochondrial matrix.

     
  7. 7.

    Fe2S2 clusters. A modest proportion of the Fe (approximately 5–10%) in mitochondria is present as these types of clusters, including those from succinate dehydrogenase and the Rieske Fe/S protein.

     
  8. 8.

    Hemes. A similarly modest proportion of Fe in mitochondria is present as heme prosthetic groups.

     
  9. 9.

    Distribution of Fe in mitochondrial regions. The OM is largely devoid of Fe-containing proteins, while the matrix is dominated by Fe4S4 clusters.

     
Footnotes
1

For a purified Fe4S4 ferredoxin the area under the doublet can be quantified to within 1–2%. Here, the uncertainties are considerably larger, primarily because more than one cluster contributes. The primary contributors to the doublet may be aconitase and dihydroxyacid dehydratase. Because species with slightly different but unresolved parameters contribute, lineshapes are heterogeneously broadened Lorentzians. We used both the Lorentzian and the Voight lineshape options of WMOSS. As Voight shapes are narrower at the base, this option yields, upon visual inspection, a lower estimate for the concentration.

 
2

In weak applied fields, the lowest three Kramers doublets of the spin sextet are generally populated at 4.2 K, yielding three Mössbauer spectra per site. Moreover, under these conditions the magnetic splittings, like the effective g values observed by EPR, are very sensitive to the rhombicity parameter E/D. Consequently, the high-spin Fe3+ ions in our sample produce broad and barely discernible features in weak fields. However, the 8.0-T spectra are fairly insensitive to D and E/D, because the large Zeeman splitting puts essentially all Fe3+ ions into the MS = −5/2 state, facilitating detection and quantification.

 

Acknowledgements

We thank the following people: Art Johnson and Holly Cargill (Department of Biochemistry and Biophysics, Texas A&M University) for instructions on isolating mitochondria; Rola Barhoumi (Image Analysis Laboratory, Texas A&M University) and Anne Ellis (Microscopy and Imaging Center, Texas A&M University) for collecting microscopic images; Jinny Johnson (Protein Chemistry Laboratory, Texas A&M University) for performing amino acid analyses; David P. Giedroc (Department of Biochemistry and Biophysics, Texas A&M University) for access to his atomic absorption spectrophotometer; William James (Department of Chemistry, Texas A&M University) for training on and assistance with the inductively coupled plasma mass spectrometer; Shelly Henderson Possi for help in isolating some batches and in measuring O2 consumption; Tanner Freeman for preparing one of the EPR samples; and Roland Lill for helpful discussion.

This study was supported by the Robert A. Welch Foundation (A1170) and The National Institutes of Health [GM077387 (M.P.H.), EB001475 (E.M.) and The Chemistry Biology Interface training program (B.N.H. and J.G)].

Supplementary material

Copyright information

© SBIC 2007

Authors and Affiliations

  • Brandon N. Hudder
    • 1
  • Jessica Garber Morales
    • 1
  • Audria Stubna
    • 2
  • Eckard Münck
    • 2
  • Michael P. Hendrich
    • 2
  • Paul A. Lindahl
    • 1
    • 3
  1. 1.Department of ChemistryTexas A&M UniversityCollege StationUSA
  2. 2.Department of ChemistryCarnegie Mellon UniversityPittsburghUSA
  3. 3.Department of Biochemistry and BiophysicsTexas A&M UniversityCollege StationUSA

Personalised recommendations