Mechanism of Iron Transport to the Site of Heme Synthesis inside Yeast Mitochondria*

The import of metals, iron in particular, into mitochondria is poorly understood. Iron in mitochondria is required for the biosynthesis of heme and various iron-sulfur proteins. We have developed an in vitro assay to follow the uptake of iron into isolated yeast mitochondria. By measuring the incorporation of iron into porphyrin by ferrochelatase in the matrix, we were able to define the mechanism of iron import. Iron uptake is driven energetically by a membrane potential across the inner membrane but does not require ATP. Only reduced iron is functional in generating heme. Iron cannot be preloaded in the mitochondrial matrix but rather has to be transported across the inner membrane simultaneously with the synthesis of heme, suggesting that ferrochelatase receives iron directly from the inner membrane. Transport of iron is inhibited by manganese but not by zinc, nickel, and copper ions, explaining whyin vivo these ions are not incorporated into porphyrin. The inner membrane proteins Mmt1p and Mmt2p proposed to be involved in mitochondrial iron movement are not required for the supply of ferrochelatase with iron. Iron transport can be reconstituted efficiently in a membrane potential-dependent fashion in proteoliposomes that were formed from a detergent extract of mitochondria. Our biochemical analysis of iron import into yeast mitochondria provides the basis for the identification of components involved in transport.

The import of metals, iron in particular, into mitochondria is poorly understood. Iron in mitochondria is required for the biosynthesis of heme and various ironsulfur proteins. We have developed an in vitro assay to follow the uptake of iron into isolated yeast mitochondria. By measuring the incorporation of iron into porphyrin by ferrochelatase in the matrix, we were able to define the mechanism of iron import. Iron uptake is driven energetically by a membrane potential across the inner membrane but does not require ATP. Only reduced iron is functional in generating heme. Iron cannot be preloaded in the mitochondrial matrix but rather has to be transported across the inner membrane simultaneously with the synthesis of heme, suggesting that ferrochelatase receives iron directly from the inner membrane. Transport of iron is inhibited by manganese but not by zinc, nickel, and copper ions, explaining why in vivo these ions are not incorporated into porphyrin. The inner membrane proteins Mmt1p and Mmt2p proposed to be involved in mitochondrial iron movement are not required for the supply of ferrochelatase with iron. Iron transport can be reconstituted efficiently in a membrane potential-dependent fashion in proteoliposomes that were formed from a detergent extract of mitochondria. Our biochemical analysis of iron import into yeast mitochondria provides the basis for the identification of components involved in transport.
The acquisition of iron by eukaryotic cells involves various transport systems that ensure the regulated uptake of this essential metal (for review, see Refs. [1][2][3][4][5][6]. Most of the cellular iron is utilized within mitochondria, where it is required for the generation of heme by ferrochelatase in the matrix (7) and for several iron-sulfur cluster-containing proteins (Fe/S proteins) 1 of both the matrix (e.g. aconitase and homoaconitate hydratase) and the inner membrane (e.g. Rieske Fe/S protein; see Refs. 8 and 9). The import of iron into mitochondria must be tightly regulated, as hardly any accumulation of iron is observed in mutants defective in heme biogenesis (10). Limiting the iron content would impair the metabolic and respiratory activity of the organelle, whereas excess iron may exert toxic effects by generation of an oxidative stress through radical formation (11). Free iron ions might be particularly harmful in mitochondria, where free reactive oxygen species are generated as a side reaction of the electron transport (see, e.g. Ref. 12). Three proteins have recently been described as being important for iron homeostasis within mitochondria: the yeast ABC transporter Atm1p of the inner membrane (13,14), and the matrix proteins frataxin (yeast Yfh1p; 15,16) and Ssq1p (17). Defects in any of these proteins cause large increases in the mitochondrial iron content. Mutations in the human homolog of Atm1p, hABC7, may cause a form of sideroblastic anemia (14,18). Frataxin is mutated in patients with Friedreich's ataxia (19). Ssq1p, a member of the 70-kDa heat shock protein family, was proposed to be required for correct processing of frataxin (17) even though recent findings indicate that this relation may be more complex (20). In any case, the distinct function of these proteins in iron homeostasis within mitochondria remains to be determined.
Despite the central importance of iron for mitochondrial biogenesis, the molecular basis of its import into the organelles is not understood. Earlier studies have attempted to investigate iron uptake into mammalian mitochondria by assessing the iron associated with the organelles (summarized in Refs. 11 and 21). Contradicting results have been achieved concerning the energy dependence of the import reaction. Although some reports found no obvious requirement for a potential across the mitochondrial inner membrane (22,23), evidence has been presentedforironuptakeoccurringinbothamembranepotentialdependent and an independent fashion (24). No conclusive evidence is available in what form (free, chelated, reduced, or oxidized) iron must be presented to mitochondria to be competent for import (for review, see Ref. 21). A serious shortcoming of previous reports has been the failure to localize the iron associated with mitochondria to one of the organellar compartments; i.e. it was not possible to differentiate truly imported iron from iron unspecifically associated with mitochondrial membranes. Using isolated yeast mitochondria we also failed to distinguish between imported and unspecifically bound iron. The reason for these obstacles is the avid association of iron with biological membranes.
We have developed an in vitro assay for the import of iron into the matrix of yeast mitochondria to study the biochemical mechanism of iron transport. The assay circumvents previous technical difficulties by following the uptake of iron through the generation of radioactively labeled heme from [ 55 Fe]iron and a porphyrin substrate. This reaction is catalyzed by ferrochelatase, a protein of the mitochondrial matrix peripherally associated with the inner membrane (7,25). Because this procedure detects only biochemically competent iron that has crossed the inner membrane, well known problems with pre-cipitation and unspecific membrane attachment of iron are avoided. Our strategy also makes it unnecessary to localize the iron associated with mitochondria. Using this assay, we were able to unravel mechanistic details of iron transport to ferrochelatase in the mitochondrial matrix. The energy for iron transport is provided by the potential across the mitochondrial inner membrane but not by hydrolysis of ATP. Our data suggest that iron must be delivered to ferrochelatase in its reduced form and reaches the enzyme directly from the inner membrane without transient passage through the matrix. The transport reaction is highly specific for iron ions, i.e. the transporter precludes other metals such as zinc and copper from access to ferrochelatase. Furthermore, the generation of heme can be reconstituted efficiently using proteoliposomes formed from detergent extracts of mitochondria. Reconstitution of iron transport across the mitochondrial inner membrane will now facilitate identification of components involved in this reaction and their biochemical characterization. Similar developments were opened by detergent solubilization and reconstitution of other central biochemical processes such as preprotein translocation and vesicular transport (26 -28).
Assay of Iron Import into Isolated Mitochondria-In the standard experiment isolated yeast mitochondria (100 g) were incubated with 2 mM NADH for 3 min at 25°C in 500 l of buffer A (0.6 M sorbitol, 40 mM Hepes-KOH, pH 7.4, 50 mM KCl, 1 mM MgSO 4 ). Iron import and heme synthesis were initiated by the addition of 1 mM ascorbate, 0.2 M 55 FeCl 3 (corresponding to approximately 10 6 cpm), and 1 M deuteroporphyrin (DP; final concentrations). After 10 min at 25°C, the reaction was stopped by the addition of 25 l of 10 mM FeCl 3 in 5 M HCl. Formed [ 55 Fe]deuteroheme was extracted by the addition of 500 l of n-butyl acetate and vortexing for 45 s. After phase separation by brief centrifugation, 200 l of the organic phase was added to 1 ml of liquid scintillation mixture (Ultima Gold TM , Packard) and counted in a Philips Tw 4700 liquid scintillation counter. A DP stock solution was prepared by dissolving 8 mg of DP in 10 l of 1 M KOH. This solution was diluted with 1,360 l of 100 mM Tris-HCl, pH 7.5. The resulting stock solution (10 mM) was stored at Ϫ20°C and diluted 200-fold into buffer A before its addition to the samples. Experiments under anaerobic conditions were performed using an anaerobic chamber.
Small Scale Isolation of Crude Mitochondria-After growth in the desired media yeast cells were collected by centrifugation, washed once in H 2 O, and resuspended in ice-cold SoH buffer (0.6 M sorbitol, 20 mM Hepes-KOH, pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride at a density of 0.5 g of cells/ml. Glass beads (0.4 -0.5-mm diameter) were added to occupy two-thirds of the final volume. Cells were broken by vortexing for 15 s at maximum speed. After 15 s of chilling on ice, vortexing was repeated. Samples were subjected to differential centrifugation to obtain a crude mitochondrial fraction. First, cells were centrifuged for 5 min at 2,500 rpm in a Beckman JA18.1 rotor at 2°C. The supernatant was removed and centrifuged for 12 min at 9,000 rpm at 2°C. The resulting pellet fraction containing mitochondria was resuspended in SoH buffer with 1 mM phenylmethylsulfonyl fluoride, and the protein concentration (typically adjusted to 5-10 mg/ml) was determined by the Coomassie color reagent.
Miscellaneous Methods-Standard methods for the manipulation of DNA and for polymerase chain reaction were used (31). Other methods were used as published: transformation of yeast cells (32), isolation of yeast mitochondria (33), whole cell lysates by breaking the cells with glass beads (34), gel electrophoresis and immunostaining of mitochondrial proteins (35,36), measurement of mitochondrial integrity by protease protection assays of matrix, and intermembrane space-exposed proteins (37). Purified ferrochelatase was a kind gift of Dr. R. Labbe-Bois.

An Assay to Study Iron Import into Isolated Mitochondria-
Because of the avid and unspecific association of iron with membranes it was not possible to use iron binding to mitochondria as a measure for import. We therefore developed an assay for iron import by following the generation of 55 Fe-radiolabeled heme catalyzed by ferrochelatase in the mitochondrial matrix. Isolated mitochondria were incubated with [ 55 Fe]iron and the protoporphyrin IX analog DP (Fig. 1A). Synthesis was performed at 25°C with energized organelles (by the addition of NADH) in the presence of the reducing agent ascorbate and was stopped by the addition of excess iron chloride. The formed radiolabeled [ 55 Fe]deuteroheme (for simplicity denoted as [ 55 Fe]heme hereafter) was extracted into an organic phase (n-butyl acetate) and quantitated by liquid scintillation counting. Mitochondria isolated from wild-type yeast cells generated 25 pmol of [ 55 Fe]heme/min/mg of mitochondrial protein, i.e. 25% of added [ 55 Fe]iron was incorporated into DP in a 10-min reaction. In contrast, hardly any heme synthesis was observed in the absence of added DP or upon incubation at 0°C (Fig. 1B).
No [ 55 Fe]heme was generated in the absence of mitochondria or with mitochondria isolated from a yeast strain (⌬hem15) in which the gene of ferrochelatase has been deleted (29). This suggests that the incorporation of iron into DP did not occur spontaneously under our experimental conditions but rather was catalyzed by ferrochelatase. In keeping with this, heme formation was inhibited strongly by the addition of zinc ions, known to serve as a competitive inhibitor of the ferrochelatase enzyme activity (38,39; see below). The integrity of the isolated mitochondria was estimated by both the activity of citrate synthase (40) and the protease accessibility of inner membrane and matrix proteins (37) and was higher than 98% (not shown). This indicates that the generation of heme in intact isolated mitochondria can be taken as a measure of iron import into mitochondria. The assay system was used to analyze the biochemical mechanism of this process.
Iron Import into Mitochondria Requires a Membrane Potential-First, we tested whether the uptake into mitochondria of the substrates of ferrochelatase requires the presence of a membrane potential. When isolated mitochondria were energized by the addition of NADH or ethanol, efficient synthesis of 55 Fe-radiolabeled heme was observed ( Fig. 2A). Comparatively slow synthesis of heme occurred in the absence of these compounds, which feed electrons into the respiratory chain. Hardly any heme was formed upon depleting the membrane potential by the addition of the uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP; Fig. 2A). Similar results were obtained by inhibiting the formation of the membrane potential by the addition of antimycin A (inhibitor of complex III), the ionophore valinomycin, and oligomycin (inhibitor of the F 1 F 0 -ATPase) to the reaction (not shown; see below). Thus, the transport of iron and/or DP to matrix-localized ferrochelatase requires a membrane potential in isolated mitochondria.
To define which of the substrates of ferrochelatase requires a membrane potential for import, we investigated the transport of the hydrophobic DP. To this end, we incubated mitochondria with or without DP in the presence or absence of antimycin A, valinomycin, and oligomycin. Under all conditions, DP avidly associated with mitochondria, even at 0°C (not shown). After removal of free DP by reisolation of the mitochondria, the soluble matrix fraction was isolated and used as the source of DP in a heme synthesis assay employing purified ferrochelatase. The matrix fraction obtained from mitochondria that were incubated with DP supported 5-fold higher heme synthesis than the matrix fraction derived from mock-treated mitochondria (Fig. 2B). The stimulation of heme formation by the matrix extract was independent of the presence of a membrane potential during the import of DP into mitochondria. It appears from these data that DP can be imported into the mitochondrial matrix independently of a membrane potential. Together with data presented below, we conclude that DP can permeate the inner membrane in the absence of a potential under our experimental conditions. This suggests that the requirement of a membrane potential for in organello heme synthesis is related to the uptake of iron rather than porphyrin.
To analyze whether iron import into mitochondria requires a membrane potential in vivo, we added radiolabeled [ 55 Fe]iron to wild-type yeast cells in the presence or absence of the uncoupler CCCP (41). After labeling for 1 h, cells were washed with citrate and EDTA to remove iron that had not been incorporated by the cells. A cell lysate was prepared by breaking yeast cells with glass beads, and a postmitochondrial supernatant and a crude mitochondrial fraction were obtained. The [ 55 Fe]heme generated during the labeling period was extracted by n-butyl acetate and quantitated. In the presence of CCCP, only 25% of heme was formed compared with untreated cells (Fig. 3A). This strong impairment of cellular heme synthesis was not the result of defective uptake of iron into CCCP-treated cells because compared with wild-type cells even slightly higher amounts of total [ 55 Fe]iron were observed in the cell lysate and the postmitochondrial supernatant (Fig. 3B). We also noticed in this in vivo experiment that similar amounts of radiolabeled iron associated with mitochondria independently of the existence of a membrane potential. This strengthens our view that iron association is not a reliable measure for mitochondrial import even under in vivo conditions (see above). Together, these data demonstrate a membrane potential dependence of heme synthesis in vivo, suggesting that the import of iron into mitochondria requires energized organelles.
Iron Import into Mitochondria Occurs Independently of ATP-The requirement for nucleotides, in particular for ATP, for iron import was tested with isolated mitochondria in the presence of ascorbate. No stimulatory effect on the generation of heme was observed by inclusion of 100 M adenine nucleotides in the assay (Fig. 4A). Higher amounts inhibited the FIG. 2. Iron import into isolated mitochondria requires a membrane potential across the inner membrane. Panel A, the synthesis of 55 Fe-radiolabeled heme was performed as described in Fig.  1B at 25°C in the presence of NADH unless indicated otherwise (No addition). In one set of samples NADH was replaced by 0.5% EtOH to generate a membrane potential; to another set 40 M CCCP was added. One series of reactions was carried out at 0°C. After incubation for various time periods the reaction was terminated, and radiolabeled heme was quantitated as described in Fig. 1B. Panel B, accumulation of DP in isolated mitochondria does not require a membrane potential. Mitochondria (100 g) in buffer A were incubated for 10 min at 25°C in the presence or absence of 2 M DP, 8 M antimycin A (A), 0.5 M valinomycin (V), and 20 M oligomycin (O) as indicated. Organelles were reisolated, washed once with buffer A, and briefly sonicated (10 s, 40% duty, 70 watts). Membranes were pelleted (1 h, 100,000 ϫ g), and the supernatant was added to 25 g of purified ferrochelatase. Heme synthesis was started by adding radiolabeled [ 55 Fe]iron. After incubation for 10 min at 25°C the [ 55 Fe]iron that formed was quantitated as described in Fig. 1B. 3-fold higher heme synthesis was observed when the assay was performed in the presence of additional 2 M DP. reaction (Fig. 4B). Presumably, this inhibition is caused by chelation of iron by nucleotides because similar effects were seen upon the addition of citrate, an efficient chelating agent, or of branched chain ␣-keto acids, which serve as siderophores in certain bacteria ( Fig. 4A and Ref. 42). To test the requirement for intramitochondrial ATP, mitochondria were treated with apyrase before assaying for the synthesis of radiolabeled heme (43). Depletion of ATP by apyrase had no effect on heme synthesis, also when new synthesis of ATP by the F 1 F 0 -ATP synthase was precluded by adding the inhibitor oligomycin (Fig. 4B). Thus, neither extramitochondrial nor internal ATP is required for the uptake of iron into mitochondria. Likewise, the addition of ADP and AMP does not appear to stimulate the transport of iron to ferrochelatase, as this has been reported earlier for the association of iron with mammalian mitochondria (22,44).
Iron Must Be Reduced before Transport to Ferrochelatase-In our standard assay using isolated mitochondria energized with NADH, iron is maintained in a reduced state by the addition of ascorbate. When this reducing agent was omitted, hardly any radiolabeled heme was generated by isolated mitochondria (Fig. 5A). Efficient synthesis of heme could be restored when iron was reduced by inclusion of dithiothreitol or reduced glutathione in the assay. Both reagents were capable of reducing iron in vitro (not shown). The requirement for reduction of iron could not be replaced by the inclusion of various chelating compounds such as adenine nucleotides, ␣-keto acids, or citrate, which prevent the precipitation of ferric iron out of solu-tion (Fig. 4A, ϪAscorbate).
In the absence of NADH as a substrate of the electron transport chain, isolated mitochondria synthesized about 5-fold less heme compared with the presence of NADH, presumably because of a residual membrane potential ( Fig. 5B; see also Fig.  2A). Similar to the reaction in fully energized mitochondria, synthesis of heme required the reduction of iron by the addition of ascorbate, dithiothreitol, or reduced glutathione (Fig. 5B). The reduction of iron could not be performed by isolated mitochondria themselves even when they were energized with the reducing compound NADH. A comparable requirement for reduction of iron by ascorbate, dithiothreitol, reduced glutathione, or sodium dithionite was observed under anaerobic conditions using mitochondria that were energized by the addition of 0.1 mM ATP (not shown). In support of these findings, we found relatively weak inhibition of heme formation by adding the chelator desferrioximine, which binds ferric iron with an affinity constant of 10 30 (Fig. 5C). This indicates that it is not ferric iron that is transported across the mitochondrial inner membrane. We conclude from these observations that iron has to be in a reduced state before its uptake into mitochondria to be  4. Import of iron does not require intra-or extramitochondrial ATP. Panel A, 55 FeCl 3 was incubated for 5 min at 4°C with ATP, ADP, AMP, citrate (10 mM each), or a mixture of ␣-ketoisovalerate, ␣-ketoisocaproate, and ␣-ketomethylvalerate (10 mM each; ␣-Ketoacids) in the presence or absence of 0.1 M ascorbate. Samples were diluted 100-fold into a mixture containing 100 g of mitochondria that had been preincubated for 2 min with 2 mM NADH and 1 M DP in a final volume of 500 l of buffer A. Heme synthesis was allowed for 5 min at 25°C and quantitated as described in Fig. 1B. The sample containing ascorbate without further additions (None) was set as 100%. Panel B, mitochondria were treated with or without 10 unit/ml apyrase and 20 M oligomycin as indicated for 5 min at 25°C and were reisolated and resuspended in buffer A. Various concentrations of ATP were added, and the synthesis of [ 55 Fe]heme was measured as described in Fig. 1B. Data are normalized to the amount of [ 55 Fe]heme formed in the absence of added ATP, apyrase, and oligomycin. competent in the ferrochelatase reaction.
Iron Is Supplied to Ferrochelatase from the Inner Membrane-We tested whether iron can be preloaded into isolated mitochondria before starting the ferrochelatase reaction by the addition of DP. We took advantage of the fact that the import of iron can be inhibited efficiently by the addition of EDTA, a chelator that is not capable of entering mitochondria (Fig. 6A,  left). [ 55 Fe]Iron was incubated with energized mitochondria for 4 min, before further import of iron was inhibited by chelating with EDTA those iron ions that were still outside of the or-ganelles. When the reaction was supplemented with DP, and samples were incubated further, no significant heme formation was observed compared with the simultaneous addition of iron and DP (Fig. 6A, right). Thus, iron cannot be imported into mitochondria before the addition of DP. We conclude from these data that iron present in the mitochondrial matrix does not form a pool that can be used by ferrochelatase. Rather, iron is supplied to ferrochelatase directly from the inner membrane without transient passage across the matrix.
When DP was preincubated with mitochondria in the absence of iron, and excess DP was removed by reisolation of the organelles before the addition of [ 55 Fe]iron, the amounts of radiolabeled heme generated were similar to those in the onestep incubation (Fig. 6B). The initial association of DP with mitochondria did not require the presence of a membrane potential across the inner membrane ( Fig. 6B; see above). This supports our view that DP can associate with mitochondria in a manner that is competent for the generation of [ 55 Fe]heme upon the subsequent addition of radiolabeled iron.
The Putative Mitochondrial Metal Transporters Mmt1p and Mmt2p Are Not Required for Delivery of Iron to Ferrochelatase-Transport of iron into yeast mitochondria has been suggested to involve two highly homologous proteins of the inner membrane, Mmt1p and Mmt2p (45). To analyze whether these putative metal transporters mediate the delivery of iron to ferrochelatase, we constructed a yeast mutant in which both the MMT1 and MMT2 genes were deleted. These mutant cells (strain ⌬mmt1⌬mmt2) displayed hardly any growth defect on standard media (45; not shown). Mitochondria isolated from the mutant cells contained similar amounts of cytochromes and harbored the same activity of the mitochondrial Fe/S protein aconitase as did wild-type cells (not shown). Mutant mitochondria showed wild-type activity in the generation of 55 Fe-radiolabeled heme (Fig. 7). We conclude that Mmt1p and Mmt2p are not required for the transport of iron to ferrochelatase.
Transport of Iron to Ferrochelatase Can Be Reconstituted in Proteoliposomes-We attempted to dissolve the mitochondrial membranes in detergent and reconstitute proteoliposomes that faithfully harbor the activity to generate heme. To this end, mitochondria were lysed in dodecyl-maltoside, and aggregated material was removed by ultracentrifugation. The clarified extract was supplemented with a phospholipid mixture, diluted below the critical micellar concentration, and the proteoliposomes formed were reisolated. They were capable of synthesizing ATP upon the generation of a membrane potential by NADH (not shown), i.e. the components of oxidative phosphorylation were at least partially reintegrated into the lipid bilayer in a functional fashion. When the proteoliposomes were tested for their ability to generate heme from added [ 55 Fe]iron and DP, they synthesized heme at almost the same efficiency as intact mitochondria (Fig. 8). The generation of radiolabeled heme was dependent upon the addition of NADH (not shown) and was inhibited by antimycin A or CCCP (Fig. 8), i.e. iron import required a membrane potential in a manner similar to that observed with intact mitochondria. In contrast, by using the supernatant of the reconstitution procedure (Fig. 8, middle panel) or purified ferrochelatase (not shown), no effects of antimycin A or CCCP were detectable for the synthesis of heme. The activity of this extract containing free ferrochelatase was almost 5-fold higher than that of intact mitochondria. The simplest explanation for this observation is that the transport of iron to ferrochelatase across the mitochondrial inner membrane is rate-limiting for heme synthesis.
The Specificity of Iron Transport Precludes Access of Other Metals to Ferrochelatase-We investigated the metal specificity of the iron transport reaction by comparing the metal ion FIG. 5. Iron must be reduced prior to transport to ferrochelatase. Panel A, synthesis of 55 Fe-radiolabeled heme was performed as described in Fig. 1B using mitochondria that were energized with 2 mM NADH. Where indicated, ascorbate was omitted from the reaction or replaced by 1 mM dithiothreitol (DTT) or 1 mM glutathione (GSH) as a reducing agent. Further treatment and determination of [ 55 Fe]heme were as described in Fig. 1B. Panel B, synthesis and analysis of radiolabeled heme was performed as in panel A without the addition of NADH unless indicated otherwise. All data are given relative to the amount of [ 55 Fe]heme formed in the standard reaction in panel A (containing ascorbate and NADH). Panel C, the experiments were performed as described in Fig. 1B in the presence of the indicated concentrations of chelators. [ 55 Fe]Heme formed in the absence of chelators was set as 100%.
inhibition of heme synthesis by intact mitochondria and by free ferrochelatase. Submicromolar concentrations of zinc ions strongly inhibited heme synthesis by free or purified ferrochelatase (see Ref. 38) but hardly affected the reaction occurring in intact mitochondria (Fig. 9A). Only at higher concentrations of zinc was the reaction in intact mitochondria inhibited (see Fig.  1B). Thus, below micromolar concentrations zinc ions do not interfere with the transport of iron across the mitochondrial inner membrane. Comparable results were obtained with copper and nickel ions (not shown). In conclusion, zinc, copper, and nickel ions are efficient substrates or inhibitors of ferrochelatase, yet in intact mitochondria their access to the enzyme is precluded by the specificity of iron transport across the inner membrane.
The opposite result was obtained with manganese ions. They impaired heme synthesis mediated by intact organelles but did not block free or purified ferrochelatase below a 500 M concentration (Fig. 9B). The effect of manganese ions on heme In other samples (Two Steps) mitochondria were energized with 2 mM NADH in 500 l of buffer A for 2 min at 25°C. 1 mM ascorbate (Asc) and 0.2 M 55 FeCl 3 were added, and samples were incubated for 4 min at 25°C. Further import of iron was prevented by chelation with 1 mM EDTA followed by incubation for 1 min. Synthesis of heme was started by the addition of 1 M DP. Samples were incubated for 5 min at 25°C, the reaction was stopped, and heme was quantitated as described in Fig. 1B. Control samples contained 1 mM EDTA from the beginning of the reaction. Heme formed in the one-step reaction in the absence of EDTA was set as 100%. Panel B, heme formation was performed as in panel A (One Step). For the two-step reaction, 100 g of mitochondria was first incubated with 1 M DP in 500 l of buffer A for 5 min at 25°C. Mitochondria were reisolated by centrifugation (10 min, 15,000 ϫ g, 2°C), washed once with buffer A, and resuspended in 500 l of buffer A containing 2 mM NADH. After 2 min, the iron import was started by the addition of 1 mM ascorbate and 0.2 M 55 FeCl 3 . Samples were incubated for 5 min at 25°C. Further treatment was as in panel A. Heme formed in the one-step reaction was set as 100%.
synthesis was comparable using intact mitochondria (Fig. 9) or reconstituted proteoliposomes (not shown), demonstrating further that we have faithfully reconstituted the process of iron import into mitochondria. Apparently, manganese ions specifically inhibit the iron transport step. However, the concentrations needed for efficient inhibition are rather high, and therefore manganese may not interfere significantly with iron transport to ferrochelatase in vivo.

DISCUSSION
The goal of the present communication is a mechanistic characterization of iron import into yeast mitochondria. We have used the in organello synthesis of heme to monitor the transport of this metal across the inner membrane thereby avoiding well known problems resulting from the unspecific and membrane potential-independent association of iron with mitochondria. Despite the increased complexity of the assay involving the transport of both iron and porphyrin, it was possible to define the requirements for iron transport. This was due to the observation that porphyrin is transported across the mitochondrial membranes in a spontaneous fashion. Porphyrin accumulated within mitochondria at 0°C in the absence of a membrane potential and could be provided for heme synthesis before iron was added. Apparently, under our in vitro conditions this compound is able to permeate the membranes by virtue of its hydrophobicity.
Iron import into isolated mitochondria was dependent on a membrane potential as an energy source. A similar requirement for energized mitochondria was observed in vivo. This appears to be the only driving force, because iron import occurred independently of adenine nucleotides both inside and outside of mitochondria. It is therefore unlikely that the transport of iron is mediated by a transport ATPase. Further studies are needed to define whether the electrochemical or the pH gradient at the inner membrane drives import of iron.
Only the reduced (ferrous) form of iron was found to be competent for transport to ferrochelatase. Iron could not be reduced by isolated mitochondria even when the organelles were supplied with NADH and thus exhibited a high reducing capacity. Transport of reduced iron to ferrochelatase across the inner membrane might provide a mechanistic advantage because ferrochelatase is known to utilize specifically ferrous iron for incorporation into the porphyrin ring (38). Thus, ferrochelatase appears to be supplied with the proper iron substrate.
Iron preloaded into mitochondria does not serve as a pool for ferrochelatase activity. Rather, iron had to be delivered to ferrochelatase directly from the inner membrane to be competent in heme synthesis. It seems unlikely that iron, upon entry into the matrix, becomes rapidly incompetent, e.g. by oxidation, because the mitochondrial matrix represents a reducing environment. We therefore propose that ferrous iron reaches the active site of ferrochelatase directly after its passage across the inner membrane without transient movement through the matrix. This raises the interesting question of whether the transporter supplying ferrochelatase with iron is also involved in importing iron into the mitochondrial matrix for biogenesis of Fe/S proteins or whether a separate transporter performs this reaction.
Heme synthesis by intact mitochondria and by ferrochelatase in solution displays differential sensitivities to various metal ions. Zinc, copper, and nickel do not appreciably inhibit ferrochelatase function in intact organelles at concentrations that block free ferrochelatase almost completely. Apparently, the iron transporter of the mitochondrial inner membrane is highly selective and efficiently excludes zinc, copper, and nickel ions from being transferred to ferrochelatase. Zinc has long been known to represent a strong competitive inhibitor of free ferrochelatase because it becomes incorporated into the porphyrin ring instead of iron (38,39). An unsolved question therefore was how the formation of zinc protoporphyrin IX is precluded in vivo. The observed selectivity of iron transport across the mitochondrial inner membrane provides a clear explanation for this problem. In contrast to what is found for these metals, manganese ions efficiently inhibit heme synthesis in organello at concentrations that do not impair soluble ferrochelatase. Thus, manganese ions interfere directly with iron transport. However, inhibition occurs only at rather high concentrations that may not elicit any significant inhibition of iron transport to ferrochelatase in vivo.
The rate of heme synthesis in our in vitro experiments is about 1.5 nmol of heme/mg of mitochondrial protein/h. Yeast cells contain about 100 nmol of total heme/g of dry weight (7,46). Assuming that 30% of dry weight corresponds to protein and 1/10 of that represents mitochondrial protein, a total of 3.3 nmol of heme/mg of mitochondrial protein has to be synthesized per generation time. With an average doubling time of 2 h, the estimated in vivo heme requirement of a growing yeast cell (about 1.7 nmol of heme/mg of mitochondrial protein/h) nicely matches the amounts of heme produced in our in vitro experiments. Clearly, our experimental system using isolated mitochondria faithfully and efficiently reproduces the in vivo reaction. Previous studies have suggested that ferrochelatase is present in yeast cells in amounts exceeding the needs of a living cell (7,46). Our studies on detergent solubilization and reconstitution of heme synthesis confirm these observations and provide an explanation for the previous measurements. A 5-fold higher enzyme activity was observed using solubilized (free) ferrochelatase compared with the reaction in intact organelles. Apparently, direct access of iron to ferrochelatase results in significantly higher activity in heme synthesis. These results suggest that iron import is rate-limiting for heme formation in intact mitochondria.
Two proteins of the inner membrane, Mmt1p and Mmt2p, have been claimed to function as mitochondrial iron transporters (45). However, transport has not been measured directly. Our quantitative analysis of heme synthesis in mitochondria isolated from ⌬mmt1⌬mmt2 cells shows that these proteins are not required for the transport of iron to ferrochelatase. These data fit well the observation of unchanged amounts of cytochromes in these cells relative to wild-type cells. Our study does not experimentally address the question of whether these two proteins might be involved in the transport of iron into the matrix to provide iron for the biogenesis of Fe/S proteins such as aconitase. This seems unlikely though because the activity of mitochondrial aconitase was unchanged upon simultaneous deletion of the MMT1 and MMT2 genes (not shown). Together, these data demonstrate that Mmt1p and Mmt2p are not essential for iron import into mitochondria.
Iron transport into mitochondria can be reconstituted efficiently in proteoliposomes. As found with intact mitochondria, the reaction is dependent on a membrane potential and requires reduced iron. The possibility of dissolving the mitochondrial membranes in detergent and forming proteoliposomes that retain the activity of transporting iron across the lipid bilayer breaks the grounds for isolation and characterization of the components involved in this process. Our biochemical characterization of the mechanism of iron transport into mitochondria represents a crucial step toward better understanding of mitochondrial iron homeostasis. Our study will also facilitate the elucidation of the function of the ABC transporter Atm1p and of frataxin in maintaining proper mitochondrial iron levels.