The Structures of Frataxin Oligomers Reveal the Mechanism for the ...

Structure 14, 1535–1546, October 2006 ª2006 Elsevier Ltd All rights reserved

DOI 10.1016/j.str.2006.08.010

The Structures of Frataxin Oligomers Reveal the Mechanism for the Delivery and Detoxification of Iron Tobias Karlberg,1 Ulrika Schagerlo¨f,1 Oleksandr Gakh,3 Sungjo Park,3 Ulf Ryde,2 Martin Lindahl,1 Kirstin Leath,4 Elspeth Garman,4 Grazia Isaya,3,* and Salam Al-Karadaghi1,* 1 Department of Molecular Biophysics 2 Department of Theoretical Chemistry Center for Chemistry and Chemical Engineering Lund University P.O. Box 124 SE-221 00 Lund Sweden 3 Departments of Pediatric & Adolescent Medicine and Biochemistry & Molecular Biology Mayo Clinic College of Medicine Rochester, Minnesota 55905 4 Department of Biochemistry University of Oxford South Parks Road Oxford, OX1 3QU United Kingdom

Summary Defects in the mitochondrial protein frataxin are responsible for Friedreich ataxia, a neurodegenerative and cardiac disease that affects 1:40,000 children. Here, we present the crystal structures of the ironfree and iron-loaded frataxin trimers, and a singleparticle electron microscopy reconstruction of a 24 subunit oligomer. The structures reveal fundamental aspects of the frataxin mechanism. The trimer has a central channel in which one atom of iron binds. Two conformations of the channel with different metalbinding affinities suggest that a gating mechanism controls whether the bound iron is delivered to other proteins or transferred to detoxification sites. The trimer constitutes the basic structural unit of the 24 subunit oligomer. The architecture of this oligomer and several features of the trimer structure demonstrate striking similarities to the iron-storage protein ferritin. The data reveal how stepwise assembly provides frataxin with the structural flexibility to perform two functions: metal delivery and detoxification. Introduction The ability to incorporate iron into prosthetic groups and proteins is essential for the biogenesis of many vital enzymes in all organisms (Al-Karadaghi et al., 2006; Lill et al., 1999). However, in the presence of atmospheric oxygen at neutral pH, ferrous iron (Fe2+) is readily oxidized to the ferric form (Fe3+) with a decrease in solubility from %10210 M to %10218 M (Williams, 1982). In addition, Fe2+ can interact with hydrogen peroxide, a byproduct of oxygen metabolism, and catalyze the

*Correspondence: [email protected] (G.I.), salam.al-karadaghi@ mbfys.lu.se (S.A-K.)

generation of highly toxic hydroxyl radicals, which ultimately results in oxidative damage to proteins, membranes, and DNA (Halliwell, 1978). Thus, to maintain an adequate supply of iron, cells require molecular mechanisms to overcome both the limited bioavailability and potential toxicity of this transition metal. A well-characterized mechanism is the one utilized by the ferritins, a conserved family of proteins that assemble into hollow particles able to accommodate large amounts of iron, which play a critical role in iron detoxification and storage in the cytoplasm and other cellular compartments (Chasteen and Harrison, 1999). In mitochondria of most tissues, another highly conserved protein, frataxin, has been recently shown to perform the dual function of making Fe2+ bioavailable and detoxifying surplus iron (reviewed in Al-Karadaghi et al., 2006). The function of frataxin has been the subject of intense investigation ever since defects in this protein were linked to Friedreich ataxia, a progressive disease of children and adolescents characterized by neurological impairment, cardiomyopathy, and diabetes mellitus (Campuzano et al., 1996). In vivo frataxin promotes the biosynthesis of heme (Schoenfeld et al., 2005; Zhang et al., 2005) as well as the assembly and repair of ironsulfur clusters (Bulteau et al., 2004; Duby et al., 2002; Gerber et al., 2003). This correlates well with the ability of frataxin to deliver Fe2+ to ferrochelatase, the enzyme that catalyzes the last step of heme biosynthesis (O’Neill et al., 2005a; Park et al., 2003; Yoon and Cowan, 2004); to the iron-sulfur cluster scaffold protein IscU, which initiates the assembly of iron-sulfur clusters (Layer et al., 2006; Yoon and Cowan, 2003); and to the [3Fe-4S]+ form of mitochondrial aconitase, which yields the active [4Fe-4S]2+ enzyme (Bulteau et al., 2004). In vivo frataxin also plays a primary role in the protection against oxidative stress (Gakh et al., 2006; Schultz et al., 2000; Thierbach et al., 2005; Vazquez-Manrique et al., 2006). This is consistent with the ability of frataxin to catalyze the oxidation of Fe2+ to Fe3+, by using O2 or H2O2 as the oxidant, and to promote the conversion of Fe3+ to a stable protein-bound mineral (Bou-Abdallah et al., 2004; Nichol et al., 2003; O’Neill et al., 2005a; Park et al., 2002). Biochemical studies have shown that self-assembly is a central part of the mechanism of iron detoxification by frataxin. Thus, the bacterial frataxin homolog, CyaY, forms a tetramer when Fe2+ is added anaerobically, while larger oligomers are formed in the presence of Fe2+ and atmospheric O2 (Bou-Abdallah et al., 2004; Layer et al., 2006). Using this mechanism, CyaY sequesters w25 atoms of iron per subunit in a polynuclear Fe3+ hydroxo(oxo) mineral. Assembly of yeast frataxin also depends on the presence of iron, and it proceeds according to the progression a / a3 / a6 / a12 / a24 / a48 (Adamec et al., 2000; Gakh et al., 2002). The 48 subunit oligomer can store w50–75 iron atoms per subunit in 1–2 nm cores, which are structurally similar to ferrihydrite, the main biomineral formed by vertebrate ferritins (Park et al., 2003; Nichol et al., 2003). Interactions leading to stepwise assembly of yeast frataxin

Structure 1536

oligomers have been suggested to be mediated by alignment and complexation of ferrihydrite crystallites, formed at separate mineralization sites (Park et al., 2003). Indeed, yeast frataxin oligomers readily disassemble into monomers upon reduction of their ferric iron cores (Park et al., 2003). Unlike CyaY and yeast frataxin, human frataxin assembles in an iron-independent manner via stable subunit-subunit interactions mediated by the nonconserved N-terminal region of the protein (O’Neill et al., 2005b). Like yeast frataxin, the human protein has ferroxidase activity and forms iron cores structurally similar to ferrihydrite (Nichol et al., 2003). Although the three-dimensional structures of the monomers of CyaY, yeast, and human frataxin have been determined (Adinolfi et al., 2002; Cho et al., 2000; Cook et al., 2006; Dhe-Paganon et al., 2000; He et al., 2004), the molecular mechanism that enables frataxin to function in both iron delivery and detoxification remains elusive. In light of the high degree of amino acid sequence conservation among frataxins from different organisms and the fact that self-assembly is a conserved property of the protein, we screened for point mutations that would enable yeast frataxin to form stable oligomers in an iron-independent manner, similar to human frataxin. Here, iron-free and iron-loaded trimers and a 24 subunit oligomer formed by an Y73A variant of yeast frataxin were characterized by X-ray crystallography and single-particle electron microscopy (EM) reconstruction. The data suggest a conserved mechanism in which the trimer is the basic functional unit of frataxin responsible for Fe2+ binding and delivery to other proteins, while higher-order oligomers provide a basic structure suitable for iron detoxification and storage.

Results and Discussion Preparation of Stable Oligomers of Yeast Frataxin The yeast frataxin monomer assembles stepwise in an iron-dependent manner (Adamec et al., 2000). Trimer was suggested to represent the functional unit of yeast frataxin, but the instability of this species hampered structural studies (Park et al., 2002; unpublished data). In contrast, the human frataxin monomer assembles in an iron-independent manner via stable protein-protein interactions (O’Neill et al., 2005b). However, human frataxin particles are not suitable for structural studies due to their tendency to polymerize (Cavadini et al., 2002). We therefore screened for point mutations that would enable the yeast frataxin monomer to oligomerize in an iron-independent manner. The mature form of the protein (i.e., lacking the mitochondrial matrix-targeting peptide) (Branda et al., 1999) was subjected to site-directed mutagenesis whereby residues conserved between yeast and human frataxin were replaced by alanine residues; then, mutant proteins were expressed in E. coli and bacterial cell extracts analyzed by size-exclusion chromatography as described (Gakh et al., 2006; O’Neill et al., 2005b). Mutations in two regions (N terminus and strands b2–b3) resulted in proteins that oligomerized during expression in E. coli, yielding trimer and lower levels of larger oligomers. The trimer and a 24 subunit oligomer formed by one of these proteins (Y73A) were further characterized, as described below.

Table 1. Crystallographic Data Collection and Refinement Statistics Native

Fe(II)-Loaded

I911-2 I213 1.0516 121.22, 121.22, 121.22 20–3.0 (3.2–3.0) 98.3 (94.7) 86.3 (61.1) 11.6 (11.9) 0.051 (0.22) 77.2

ID23-1 I213 0.97625 123.80, 123.80, 123.80 25–3.6 (3.8–3.6) 99.6 (99.4) 95.6 (83.2) 22.8 (22.4) 0.036 (0.11) 78.6

Data Collection Beamline Space group Wavelength (A˚) Cell dimensions: a, b, c (A˚) Resolution range (A˚) Completeness (%) I/s(I) > 3 (%) Multiplicity Rmergea Solvent content (%) Refinement Rcryst (Rfree)b Rmsd bond distances (A˚) Rmsd bond angles ( )

0.230 (0.276) 0.011 1.9

a

Rmerge = SjIi 2 j/SI, where Ii is an individual intensity measurement and is the average intensity for this reflection. b Rcryst = SjFo 2 Fcj/SFo, where Fo and Fc are the observed and calculated structure factor amplitudes, respectively. Rfree is the same as Rcryst, but it is calculated on 5% of the data excluded from refinement.

Crystal Structure of the Trimer The Y73A yeast frataxin crystallized in space group I213. The structure was solved by molecular replacement by using the structures of CyaY (Cho et al., 2000) and human frataxin (Dhe-Paganon et al., 2000) monomers as search models. The NMR structure of the yeast frataxin monomer available at that time (He et al., 2004) did not give a satisfactory molecular-replacement solution. The trimer structure was refined to the crystallographic R factor of 0.23 (Rfree = 0.28). The mature form of the protein used in this study consisted of 123 amino acid residues (52–174). Of these, the first 9 and the last 2 were disordered in the electron density map and could not be modeled. The final refinement statistics are presented in Table 1. The overall arrangement of monomers in the trimer structure is almost flat, with dimensions of 75 A˚ 3 75 A˚ 3 30 A˚ (Figures 1A and 1B). Each subunit is folded into an a/b sandwich with two a helices (a2, a3) packed against a five-stranded (b1–b5), antiparallel, twisted b sheet. Two additional b strands (b6, b7) build up a b hairpin almost perpendicular to strand b5. A short helix at the protein N terminus (a1), together with a 7 residue extended coil region connected to helix a2, build up an N-terminal extension that interacts with the neighboring monomer within the trimer (Figures 1A and 1B). The overall fold of each subunit is similar to that reported earlier for different monomeric forms of the protein, although with some important deviations. Helix a2 is about two turns shorter in the trimer structure, suggesting that oligomerization involves partial unfolding of this element. Moreover, the N-terminal extension, which is highly flexible in the yeast frataxin monomer (Cook et al., 2006), is stabilized within the trimer by the interactions between neighboring subunits (Figure 1B).

The Structures of Frataxin Oligomers 1537

The Role of the N-Terminal Region The N-terminal region of each subunit appears to play a crucial role in the stabilization of the trimer. In each monomer, the N-terminal region is anchored to the core structure at the base (Figure 2A) and through helix a1, which interacts with the b sheet of another subunit (Figure 2B). Figure 2C shows a 3Fo 2 2Fc electron density map superimposed on the N-terminal region. At the base, the conformation of the N-terminal extension is stabilized by an extensive network of interactions between the main chain carbonyls of A72, H74, and E75 and the amide groups of H74, E76, and D78, respectively, and between the side chains of H74, E76, and D78 and the side chains of D79, R141, and K123, respectively (Figure 2A). Interactions of helix a1 with the b sheet of another subunit include residues P62, E64, V65, and L68 from helix a1 and residues T110, E112, T118, and V120 from the b sheet (Figure 2B). Hydrophobic residues P62, V65, and L68 in the N-terminal extension are directly packed against residues T110 and T118 (Figure 2B). Only the side chain of T118 and the carbonyl oxygen of E64 are at hydrogen-bonding distance from each other. Interestingly, the N-terminal extension also plays a major role in the packing of trimers within the crystals, where two monomers from neighboring trimers pack with their b sheets facing each other. The N-terminal extension is responsible for the interactions within the crystal lattice and is sandwiched between the structural elements (not shown). Similar interactions may be responsible for the packing of trimers within the 24 subunit oligomer (see below).

Figure 1. Overall Architecture of the Yeast Frataxin Trimer in Ribbon Representation (A) Suggested outer surface of the trimer with the exposed helices a2 and a3 (shown in yellow). Loops and b strands are shown in brown or red. (B) Suggested inner surface of the trimer. The N-terminal extension, which contributes to the stabilization of the trimer, is shown in green.

The plane of the b sheet in each subunit is inclined with an angle of about 30 relative to the plane of the trimer. Solvent-accessible parts of helix a3 and strand b7, the hairpin loop between strands b6 and b7, and the loop between strands b5 and b6 create one surface of the trimer (Figure 1A). The opposite surface is created by elements from two different subunits, with the N-terminal extension of one subunit packed against the b sheet surface of the next, which, in turn, contributes helix a2, strands b1–b3, and the loop between b3 and b4 to the surface (Figure 1B).

The Channel at the 3-Fold Axis of the Trimer The trimer is further stabilized by interactions between the loops around a central channel at the 3-fold axis (Figures 1A and 1B). On one surface of the trimer, the b hairpin loop between strands b6 and b7 (residues S151–G155), the C terminus of strand b5, and the loop between strands b5 and b6 (residues L144–W149) build up a unique helical structural element (shown in red in Figures 1A and 3A). Three such elements are arranged around the 3-fold axis of the trimer, with the loop between strands b5 and b6 of one subunit packed against the hairpin loop of the next subunit (Figure 3A). Two putative hydrogen bonds between subunits, one between the carbonyl oxygen of L152 and the amide group of G147 and the other between the side chain of N146 and the carbonyl oxygen of Q154, contribute to the stabilization of the loops (Figure 3A). The side chains of L145, V150, and L152 from the three subunits are exposed to the solvent and together create a hydrophobic edge around the entrance to the channel (Figure 3A). In the sequence of human frataxin, L145, which is closer to the surface, is replaced by a threonine, L152 is replaced by a serine, while V150 is conserved (Figure 4). Each subunit contributes an invariant aspartate residue (D143, at the end of strand b5); the three D143 residues are located approximately in the middle of the channel, with their side chains directed toward the opening at the opposite end (Figure 3A). On the other surface of the trimer, the channel opening is built up by residues between G117 and D143, which belong to strands b3, b4, and b5 (Figures 1B and 3B).

Structure 1538

Figure 2. Stabilization of the Trimer by the N-Terminal Extension (A) Interactions at the base of the N-terminal extension. (B) Interactions of helix a1 of one subunit of the trimer with the b sheet of another subunit. (C) Stereoview showing a 3Fo 2 2Fc density map (blue) contoured at the 1.3s level superimposed on helix a1.

This is the most highly conserved stretch of amino acids within eukaryotic frataxin sequences (Figure 4). Residues K123–I130 build up part of strands b3 and b4 and the loop between them. The strands are sharply bent by w80 relative to the plane of the central b sheet, while the loop between them (residues P125–K128) extends into the putative exit of the channel. Several interactions may contribute to the stabilization of this loop. Among them is a putative hydrogen bond between the side chain of D143 and the amide of Q129 of the same subunit. Additional interactions involve putative hydrogen bonds between the side chain of Q129 and the carbonyl oxygen of P126 of the next subunit, and between the side chains of N127 of one subunit and E75 of another (Figure 3B). Residues P126 and Q129 are invariant in all frataxin sequences (Figure 4). Difference electron density maps show that, within the same crystallographic structure, loop P125–K128 may exist in two alternative conformations, which approximately have the same occupancy (w0.5) (Figure 3C). In the second conformation, the tip of the loop is displaced by about 8 A˚. This brings residues N127 and Q129 in contact with the opposite subunit (Figure 3C). New interactions are formed between N127 and Q129, and between D143 and the side chains of N127 and Q129 of the same subunit. The interactions between N127 and E75 and between Q129 and P126 are not present in

this conformation. The peptide unit of P126 is flipped, resulting in the carbonyl oxygen pointing toward the interior of the channel, a direction opposite to that observed in the first conformation (Figure 3C). A shift of the side chain of K128 creates a positively charged lid at the channel opening (Figure 3C). The channel is funnel shaped, with openings of w15 A˚ and w4 A˚ at the two sides of the trimer, and with increasing acidity toward the latter (Figures 5A–5D). In analogy with the channels at the 3-fold axis of ferritin shells (Grant et al., 1998; Hempstead et al., 1994; Stillman et al., 2001), the larger and smaller openings may serve as an entrance and exit, respectively, for iron ions. However, while the entrance to the ferritin channel is lined by basic residues, the putative entrance to the frataxin channel is lined by hydrophobic residues (Figures 3A and 5C), which may serve not only to guide ions into the channel, but also to provide a docking surface for the proteins that interact with frataxin (Bulteau et al., 2004; Park et al., 2003; Yoon and Cowan, 2003). The Crystal Structure of the Iron-Loaded Trimer Iron loading of the frataxin trimer was achieved by aerobic incubation of the purified oligomer with Fe2+ at a molar ratio of two iron atoms per monomer, followed by isolation of the iron-loaded trimer by size-exclusion chromatography. A stoichiometric ratio of 0.9 (60.1)


Figure 3. Stereoview of the Structure of the Central Channel (A) Putative channel entrance. The helical structural motif from each monomer is shown in red. (B) The first conformation of the proposed channel exit. The residues that contribute to the structure and stabilization of the trimer are shown as sticks. In (A) and (B), the electron density contoured at the 1.3s level is shown superimposed on the model. (C) The second conformation of the proposed channel exit. The teal ribbon model with amino acid side chains (shown as sticks) represents the second conformation of the channel exit. The structure of the first conformation shown as a brown ribbon model is superimposed to demonstrate the two positions of the P125–K128 loop.

iron atoms per iron-loaded trimer was measured by using particle-induced X-ray emission (microPIXE) (Garman and Grime, 2005), whereas there was no measurable iron in the unloaded control. The iron-loaded trimer was crystallized at conditions similar to those of the iron-free complex, and diffraction data were collected to a resolution of 3.6 A˚. Despite the low resolution of the data, the difference electron density clearly shows a high sigma peak (electron density is visible up to the level of 5s; Figure 5C) at the center of the channel at the 3-fold axis of the trimer. The presence of one iron atom per trimer is in agreement with the microPIXE results. Three additional peaks, which most probably correspond to solvent molecules, are positioned at w2 A˚ from the center of the density (Figures 5C and 5D). This distance is within the range expected for metal-solvent bond distances. Interestingly, in the presence of iron, the P125–K128 loop is present only in the first of the two conformations described above. The metal binds close to the suggested entrance at w4 A˚ from the side chains of the three D143 residues (Figures 5C and 5D). It cannot be excluded that additional solvent molecules, not visible at the present resolution of the

data, bridge the interactions between the iron and protein groups. Owing to the hydrophobic lid formed by V150, L152, and especially L145, the iron atom is fully enclosed within the channel (Figure 5C). Calculations of the electrostatic potential and corresponding metal-binding energies in the central pore of the frataxin trimer indicate that the change in the conformation of the P125–K128 loop may affect the ironbinding properties of the channel, with the second conformation demonstrating higher metal-binding affinity (Figure 6; also see Experimental Procedures). This is in apparent contrast with the fact that in the structure of the iron-loaded trimer the P125–K128 loop is present in the first, low-affinity binding conformation. A possible explanation is that upon iron binding the interaction between D143 and the metal prevents D143 from contributing to the stabilization of the second, high-affinity binding conformation, favoring a shift to the low-affinity conformation. This may facilitate iron delivery to proteins docked at the channel entrance, while conditions that stabilize the second conformation, which remain to be identified, may promote iron passage through the channel. The suggested channel exit appears to be

Structure 1540

Figure 4. Alignment of Frataxin Sequences The amino acid sequences of frataxin from yeast, human, mouse, and Drosophila are aligned with the bacterial CyaY frataxin homologs from S. typhimurium, E. coli, and P. aeruginosa. Residues conserved only in eukaryotic or bacterial sequences are marked in green and blue, respectively; invariant residues are marked in red. The positions of secondary structural elements are shown along the sequence. The overall sequence identity is 40% between the yeast and human sequences, and it is 28% between the 106 overlapping residues (68–174) of the yeast and E. coli sequences. This figure was generated with Alscript (Barton, 1993). The sequences are identified by their Swiss-Prot entry names.

connected to negatively charged patches on the side of each subunit (Figure 5B). The residues forming the patches, D78, D79, D82, D86, E89, E90, and E93 from helix a2, and D101 from strand b1, are conserved in eukaryotic sequences (Figure 4). NMR iron titrations of monomeric frataxins have implicated these residues in iron binding (Cook et al., 2006; Nair et al., 2004). In addition, site-directed mutagenesis has shown that they comprise functionally distinct sites that are involved in iron oxidation and mineralization, and that they are important for iron detoxification in vivo (Gakh et al., 2006). Each patch is lined by hydrophobic and positively charged residues (H74, H106, K123), which together with the side chain of K128 (in the high-affinity binding conformation of the P125–K128 loop), may serve in leading metal ions from the channel exit directly to the detoxification sites. The putative ferroxidation site is in close vicinity of the channel exit and entails an arrangement of residues (H74, D78, D79, D82, and H83) reminiscent of the di-iron site of ribonucleotide reductase (Nordlund and Eklund, 1995) (Figure 5E). The residues constituting this site have a different arrangement in the trimer compared to the monomer structure of yeast frataxin (Cook et al., 2006), due to the partial unfolding of helix a2, implying that trimer formation is needed for the formation of a functional ferroxidation site. Position of Pathogenic Frataxin Point Mutations Defects in frataxin are linked to the neurodegenerative disease Friedreich ataxia (Campuzano et al., 1996). Most patients carry large GAA trinucleotide repeat expansions in the first intron of the frataxin gene that hamper transcriptional elongation, causing a severe reduction in the levels of frataxin. About 4% of the patients carry an expansion in one allele and a point mutation in the other allele (Campuzano et al., 1996). While the

majority of the pathogenic point mutations affect the hydrophobic core of the structure (L84S, I130F, L132P, W149G, L158H, and L158F, yeast numbering), four mutations are likely to destabilize formation of the trimeric complex (Figure 7). These include G107V, W131R, R141C (G130V, W155R, and R165C, human numbering), and D122Y at a position that corresponds to P100 in yeast. P100 is located in the loop between helix a2 and strand b1, at the tip of the subunit close to the surface against which helix a1 is packed. G107 is located in a loop between strands b1 and b2. Its carbonyl oxygen makes a putative hydrogen bond with the amide group of K123, thus contributing to the stabilization of the conformation of the regions between K123 and I130. K123 is also involved in interactions with the N-terminal extension through a putative hydrogen bond with D78. Moreover, W131, which is exposed to the surface of the central b sheet, is involved in interactions with the N-terminal extension. Another mutation, R141C, may disrupt a putative hydrogen bond to the carbonyl of P125, which may affect the conformation of the loop at the channel exit. Solution studies of the effect of these mutations, in combination with a rational design of compounds that stabilize the frataxin trimer, might offer a new opportunity for the treatment of Friedreich ataxia. EM Single-Particle Reconstruction of the 24 Subunit Oligomer The 24 subunit oligomer formed by yeast frataxin Y73A was reconstructed from 4,000 particles at a resolution of 19 A˚ (Figures 8A and 8B). The reconstruction shows a cubic shape in which the trimer represents the basic structural unit, with the center of each trimer positioned at the 3-fold axis of the cube (Figure 8C). Docking of the trimer into the reconstruction (Figure 8D) shows that the particle is primarily stabilized by interactions between


Figure 5. Potential Functional Sites and Metal Binding to the Central Channel of the Trimer (A and B) Electrostatic surface-potential distribution for the suggested outer (A) and inner (B) surfaces of the frataxin trimer. Red and blue denote negative and positive potentials, respectively. The residues constituting one of the three proposed ferroxidation sites are shown as sticks in (B). (C) Top view of the channel with bound iron. The Fo 2 Fc difference electron density map contoured at the 4s level (magenta) is shown superimposed on the 3Fo 2 2Fc density map (blue). The bound iron (shown as a blue sphere) is located w6 A˚ from the entrance to the channel, as defined by the side chain of L145. The red spheres correspond to solvent molecules. (D) Cross-section through the central channel showing the electrostatic-potential distribution within the channel. The bound iron and surrounding solvent molecules are shown as spheres. (E) The proposed ferroxidation site of frataxin viewed with an electrostatic-potential surface. Distances (A˚) between the side chains are indicated in the figure.

the N termini of subunits from neighboring trimers (Figure 8E). The best fit is obtained with the N termini pointing toward the interior of the complex. This location of the N termini places the suggested entrance to the channel on the outside of the particle (Figure 8F), while the exit and the oxidation and mineralization sites are located in the interior. Ongoing reconstruction of ironloaded oligomers shows that the packing of trimers delimits a cavity for iron deposition (unpublished data). This architecture is in agreement with the iron-storage function of frataxin and is remarkably reminiscent of the iron-storage protein ferritin (Grant et al., 1998; Hempstead et al., 1994; Stillman et al., 2001) (Figure 8F). Concluding Remarks A reduction in the levels of frataxin is responsible for Friedreich ataxia, a fatal neurodegenerative and cardiac disease (Campuzano et al., 1996). Frataxin is an iron-

binding protein required for the maintenance of mitochondrial iron balance in humans and other eukaryotes, owing to its roles in the delivery of iron to the heme and iron-sulfur cluster biosynthetic pathways and the detoxification of surplus iron (reviewed in Al-Karadaghi et al., 2006). The mechanism that enables frataxin to perform such essential functions has been eagerly pursued but has remained elusive even after different structures of monomeric frataxins were solved. The structures of oligomeric frataxin species described here suggest that self-assembly provides the protein with the means to bind iron and either transfer it to other proteins or detoxify and store it. Whereas the assembly of yeast frataxin and CyaY is driven by the binding and oxidation of iron (Layer et al., 2006; Park et al., 2003), assembly of human frataxin occurs in an iron-independent manner via subunit-subunit interactions mediated by the protein N

Structure 1542

Figure 6. Calculated Metal Binding Energy along the 3-Fold Symmetry Axis of the Trimer Electrostatic interaction energies with a +2-charged ion were calculated for each of the two conformations of the P125–K128 loop as described in Experimental Procedures. The red and blue plots show the metal-binding energies for the first and second configurations, respectively, along the 3-fold-symmetry axis of the trimer. The position of the two plots along the 3-fold axis is defined by the MolScript y coordinate (y coord, in angstroms from the unit cell origin). The position of the metal-binding energy in the first configuration of the channel (red plot) is in agreement with the position of the metal ion in the iron-loaded structure (w6 A˚ from the side chain of L145).

terminus (O’Neill et al., 2005b). The Y73A variant used in the present work was identified by screening for point mutations that would enable yeast frataxin to form stable oligomers in an iron-independent manner like human frataxin. Similar variants found in the screen included single substitutions at the base of the N-terminal extension, H74A and D78A, and double substitutions in the b sheet, V108A/T110A and T118A/V120A. These mutations probably induce conformational changes that favor the interaction of the N-terminal extension of one subunit with the b sheet of another, thereby facilitating oligomerization. The N-terminal region of human frataxin is 18 residues longer than that of the yeast protein and has a significantly different amino acid sequence, whereas CyaY only has a short stretch of amino acids upstream of helix a2 (Figure 4). Thus, we speculate that different N-terminal regions reflect different modes by which frataxin assembly can be induced and regulated in different organisms. We have shown that one atom of iron binds in the channel at the 3-fold axis of the trimer. The two possible conformations of the channel suggest that a gated mechanism controls whether the iron stays in the channel or is transferred to the detoxification sites at the channel exit. Because a single iron atom is delivered to ferrochelatase during heme synthesis, IscU during iron-sulfur cluster assembly, and aconitase during [3Fe-4S]+ cluster repair, the trimer could be a suitable source of metal in all of these processes in which frataxin has been implicated as the iron donor. In addition, the relatively large hydrophobic surface that lines the entrance to the channel may enable the trimer to interact with structurally

Figure 7. Position of Pathogenic FRDA Mutations Residues that are mutated in some Friedreich ataxia patients are mapped on one subunit of the trimer. See Results for details.

different proteins. The reconstruction of a 24 subunit oligomer further shows that the trimer is the building block of larger frataxin complexes. It also reveals striking similarities in the structural organization of frataxin oligomers and the iron-storage protein ferritin, despite the lack of any apparent evolutionary relationship between these two protein families. It can also be expected that the organization of the iron core of frataxin will be reminiscent of the iron core of ferritin (Nichol et al., 2003). Together these data suggest that self-assembly provides frataxin with the structural features required to perform both iron delivery and iron detoxification. The insights gained from this work provide a framework to identify natural factors or develop compounds that enhance frataxin activity by modulating its oligomerization behavior. Experimental Procedures Protein Preparation and Crystallization The mutant Y73A frataxin from S. cerevisiae (residues 52–174 of the yeast frataxin sequence, corresponding to the mature form of the protein) (Branda et al., 1999) was recombinantly expressed in E. coli and purified with a modification of a previously described procedure (Cavadini et al., 2002). Briefly, bacterial lysate (w25 ml at 26 mg/ml total protein) was applied to a Macro-Prep DEAE column (16 mm 3 50 cm) (Bio-Rad), and protein was eluted with a 1 l linear gradient, from 50 to 525 mM NaCl, in 20 mM Tris-HCl (pH 8.0) at a flow rate of 10 ml/min. Most Y73A frataxin was eluted in two separate pools, a low-salt pool (300–380 mM NaCl) containing a3, and a high-salt pool (440–510 mM) containing a24. Each DEAE pool was diluted to 600 ml and loaded onto a Macro-Prep High Q column (Bioscale MT20) (Bio-Rad), and the protein was eluted with a 500 ml linear NaCl gradient as described above at a flow rate of 5 ml/min. Two High Q pools were obtained, a24 (w410–510 mM NaCl) and a3 (w200–290 mM NaCl). Each pool was concentrated to 1 ml, loaded onto a Sephacryl 300 column (16 mm 3 60 cm) (Amersham-Biosciences), and eluted with 120 ml of 10 mM HEPES-KOH (pH 7.3), 100 mM NaCl at a flow rate of 0.4 ml/min. Fractions containing trimer (a3) or a 24 subunit oligomer (a24) were pooled and applied to a Mono Q HR5/5 column (Amersham-Biosciences) and eluted with a linear gradient, from 100 to 450 mM NaCl, at flow rate of 1ml/min. Fractions


Figure 8. Electron Microscopic Model of a 24 Subunit Frataxin Oligomer (A) EM images of the 24 subunit oligomer in 1% uranyl acetate negative stain on the recorded micrograph. (B) A comparison between projection images of the reconstructed model (to the left) and the class averages (to the right). See Experimental Procedures for details. (C) View approximately along the 2-fold axis, which relates two trimers. A total of eight trimers make the oligomer. (D) Packing of trimers docked into the EM reconstruction. The trimers (green and purple) are shown in surface representation. (E) A stereoview showing interactions between the N termini of two subunits from neighboring trimers packed within the 24 subunit particle. (F) Surface representation of frataxin (left) and horse-spleen ferritin (right) 24-meric oligomers viewed along the 3-fold axis, represented with the same scale. The width of the particles is 115 A˚ and 130 A˚, respectively. Trimers are colored according to their electrostatic surface potentials.

containing a3 (w200–250 mM NaCl) or a24 (w280–310 mM NaCl) were concentrated and buffer exchanged to 10 mM HEPES-KOH (pH 7.3). Trimer crystals were grown by vapor diffusion in a hanging drop at 15 C. Crystallization drops were made by mixing 3 ml of a protein solution at a concentration of 7 mg/ml in 10 mM KOH-HEPES (pH 7.3) with 3 ml of a reservoir solution containing 0.1 M Bis-Tris (pH 5.5), 2.0 M ammonium sulfate, and 4% (v/v) g-butyrolactone. Crystals appeared after 5 days. The best crystals had dimensions of w0.15 mm 3 0.15 mm 3 0.15 mm. The crystals belonged to space

group I213. Iron loading of frataxin trimer was achieved by incubating the purified oligomer with ferrous iron at a molar ratio of two iron atoms per monomer for 2 hr at 30 C, followed by isolation of the iron-loaded trimer by size-exclusion chromatography. Crystals in space group I213 were obtained in the same manner as for the metal-free trimer. Data Collection and Structure Determination For data collection, 25% glycerol was used as cryoprotectant, and crystals were mounted on a rayon loop and flash frozen directly in

Structure 1544

a stream of boiled-off nitrogen gas held at 100 K. A native data set to 3.0 A˚ resolution (I213) was collected with a MAR Research CCD detector at MAX II synchrotron laboratory stations I711 and I911-2 in Lund, Sweden (Cerenius et al., 2000; Mammen et al., 2004). The resolution of the data could not be improved at the higher-energy Swiss Light Source synchrotron. A data set of the iron-loaded trimer was collected with an ADSC CCD detector at beamline ID23-1, ESRF synchrotron facility in Grenoble, France, to a resolution of 3.6 A˚ (I213). All data sets were processed with the XDS package (Kabsch, 1993). The structure was solved by molecular replacement with the program Phaser (McCoy et al., 2005) by using a combination of the solved crystal structures of human frataxin and the E. coli homolog, CyaY (Cho et al., 2000; Dhe-Paganon et al., 2000), as a search probe. The model was built by using the graphical program O (Jones et al., 1991) and was refined with CNS (Brunger et al., 1998). The progress of refinement was followed by decreasing R and Rfree values. Data collection and refinement statistics are shown in Table 1. Particle-Induced X-Ray Emission Measurements The particle-induced X-ray emission (microPIXE) measurements were carried out at the National Ion Beam Centre, University of Surrey, UK on a beamline arranged as described by Grime et al. (1991). Small volumes (0.2 ml) of the unloaded and iron-loaded protein concentrated to 20 mg/ml in Tris-HCl buffer (pH 7.3) were each pipetted onto two separate 2 mm thick mylar films stretched over three aluminum target holders and dried, as described (Garman, 1999). A 2.5 MeV proton beam of 3 mm diameter was used to induce characteristic X-ray emission from the dried liquid droplet under vacuum. The X-rays were detected in a solid-state lithium-drifted silicon detector with high energy resolution. The proton beam was then scanned spatially in the x and y dimensions. Using the data collection software, spatial maps of all elements heavier than sodium that were present in the sample were obtained. Quantitative information was obtained by collecting spectra at four separate selected points on the drop and also at a point on the backing foil. These spectra were analyzed with GUPIX (Johansson et al., 1995) to extract the areal density of each element of interest in the sample with respect to the sulfur peak from the methionines and cysteines in the protein. The sulfur signal provides a very convenient internal calibration, allowing the number of iron atoms per protein monomer, and hence per trimer, to be obtained. Calculation of Metal Interaction Energies in the Channel along the 3-Fold Axis Calculations of the electrostatic potential in the central pore of the frataxin trimer were performed by solving the Poisson-Boltzmann equation with the program MEAD (Macroscopic Electrostatics with Atomic Detail) (Bashford, 1997; Bashford and Gerwert, 1992). The electrostatic potential was calculated for each of the two different conformations of the P125–K128 loop present in the iron-free structure of the trimer. The protein was described as a volume with a dielectric constant of 4, dissolved in water with a dielectric constant of 80. All atoms in the protein were assigned a partial charge, taken from the AMBER 2003 force field (Case et al., 2004; Cornell et al., 1995). Aspartate and glutamic acid residues were assumed to be negatively charged, and lysine and arginine residues were assumed to be positively charged. Based on the presumed hydrogen bond pattern, the surroundings, and the solvent exposure, we let H74, H83, and H106 be doubly protonated and therefore positively charged, whereas H95 was protonated on the N32 atom. Several additional calculations were made with different assignments of the histidine residues, e.g., all residues protonated on the N32 atom or all residues doubly protonated, as well as different ionic strength values. These variations had some influence on the absolute value of the potential, but the shapes and positions of the two curves shown in Figure 6 were not affected. Hydrogen atoms were added and optimized by AMBER 8 (Case et al., 2004) before the MEAD calculations. All atoms were assigned PARSE radii (Sitkoff et al., 1994), and the solvent probe radius was 1.4 A˚ (water). The electrostatic potential was calculated along the 3-fold symmetry axis by using a 0.5 A˚ grid with at least twice the size of the protein (2513 grid points). The calculations were run at 300 K. The electrostatic potential was used to calculate the electrostatic interaction energies with a +2 ion along the 3-fold symmetry axis of the trimer, which were

obtained by multiplying the electrostatic potential at any given position along the 3-fold axis by the charge of the metal ion. The energies in Figure 6 exclude the solvation energy of the unbound metal ion, which will reduce the binding energies by a constant factor for both curves. In Figure 6, the position of the two curves along the 3-fold axis of the trimer is defined by the y coordinate of the PDB files used to calculate the electrostatic potential.

EM Data Collection and Image Processing Protein solution (2 ml of 0.2 mg/ml) was applied to carbon-coated, glow-discharged copper grids (400-mesh, Electron Microscopy Sciences, USA) and was allowed to adsorb for 1 min. Excess solution was blotted, and the grids were subsequently stained with either 1% uranyl acetate in water (2 ml for 10 s) or 1% phosphor tungsten acid in 20 mM phosphate buffer (pH 6.9) (2 ml for 1 min). Images were acquired with a Philips CM120 equipped with a Gatan GIF 100 energy filter and Gatan 791 CCD camera (1024 3 1024 pixels) at a magnification of 52,400-fold (sampling distance 4.67 A˚). The image processing was performed with the Eman software package (Ludtke et al., 1999), applying octahedral point group symmetry. The iterative classification procedure converged after 8 cycles of reclassifications, and 4000 particles out of the 5400 were used for calculating the final reconstruction. The resolution of the frataxin oligomer was determined to 19 A˚ according to the 0.5 Fourier shell correlation criteria. Strict damping of the FSC curve in highresolution shells was observed, which, together with the comparison between class averages and projections, is an indication that the envelope of the classification process is not limiting. The docking of the X-ray structure into the EM reconstruction was done in Chimera (Pettersen et al., 2004). The modification of the trimeric X-ray structure was done in O (Jones et al., 1991). The final reconstruction of the 24 subunit oligomer was evaluated by comparing class averages with projection images in the same angular orientation of the interpolated model (Figure 8B). The image pairs revealed the same structural features, an indication of the reconstruction being in a global minimum, rather than trapped in a local one. An initial reconstruction was performed without any symmetry. The result reveals a cubic shape of the molecule, vindicating our use of octahedral symmetry (result not shown). Several independent reconstructions aligning on either point group symmetry or varying reference models gave similar results. To induce a better fit to the EM model, the monomer had to be tilted by an angle of about 12 compared to its position in the X-ray model. The tilt did not introduce any notable changes in the internal interactions within the trimeric structure, and the size of the channel was conserved.

Acknowledgments We acknowledge MAX-lab and the European Synchrotron Radiation Facility for provision of synchrotron radiation facilities, and we thank the National Center for High-Resolution Electron Microscopy electron biomicroscopy unit. This work was supported by grants from the Swedish Research Council to S.A.-K. and from the National Institutes of Health/National Institute on Aging (AG15709) and the Friedreich Ataxia Research Alliance (FARA) to G.I. Received: March 21, 2006 Revised: August 21, 2006 Accepted: August 28, 2006 Published: October 10, 2006

References Adamec, J., Rusnak, F., Owen, W.G., Naylor, S., Benson, L.M., Gacy, A.M., and Isaya, G. (2000). Iron-dependent self-assembly of recombinant yeast frataxin: implications for Friedreich ataxia. Am. J. Hum. Genet. 67, 549–562. Adinolfi, S., Trifuoggi, M., Politou, A.S., Martin, S., and Pastore, A. (2002). A structural approach to understanding the iron-binding properties of phylogenetically different frataxins. Hum. Mol. Genet. 11, 1865–1877.


Al-Karadaghi, S., Franco, R., Hansson, M., Shelnutt, J.A., Isaya, G., and Ferreira, G.C. (2006). Chelatases: distort to select? Trends Biochem. Sci. 31, 135–142. Barton, G.J. (1993). An efficient algorithm to locate all locally optimal alignments between two sequences allowing for gaps. Comput. Appl. Biosci. 9, 729–734. Bashford, D. (1997). An object-oriented programming suite for electrostatic effects in biological molecules. In Scientific Computing in Object-Oriented Parallel Environments, Y. Ishikawa, R.R. Oldehoeft, J.V.W. Reynders, and M. Tholburn, eds. (Berlin: Springer), pp. 233– 240. Bashford, D., and Gerwert, K. (1992). Electrostatic calculations of the pKa values of ionizable groups in bacteriorhodopsin. J. Mol. Biol. 224, 473–486. Bou-Abdallah, F., Adinolfi, S., Pastore, A., Laue, T.M., and Chasteen, N.D. (2004). Iron binding and oxidation kinetics in frataxin CyaY of Escherichia coli. J. Mol. Biol. 341, 605–615. Branda, S.S., Cavadini, P., Adamec, J., Kalousek, F., Taroni, F., and Isaya, G. (1999). Yeast and human frataxin are processed to mature form in two sequential steps by the mitochondrial processing peptidase. J. Biol. Chem. 274, 22763–22769. Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54, 905–921. Bulteau, A.L., O’Neill, H.A., Kennedy, M.C., Ikeda-Saito, M., Isaya, G., and Szweda, L.I. (2004). Frataxin acts as an iron chaperone protein to modulate mitochondrial aconitase activity. Science 305, 242– 245. Campuzano, V., Montermini, L., Molto, M.D., Pianese, L., Cossee, M., Cavalcanti, F., Monros, E., Rodius, F., Duclos, F., Monticelli, A., et al. (1996). Friedreich’s ataxia: autosomal recessive disease caused by an intronic GAA triplet repeat expansion. Science 271, 1423–1427. Case, D.A., Darden, T.A., Cheatham, T.E., Simmerling, C.L., Wang, J., Duke, R.E., Luo, R., Merz, K.M., Wang, B., Pearlman, D.A., et al. (2004). AMBER 8 (www.amber.scripps.edu). Cavadini, P., O’Neill, H.A., Benada, O., and Isaya, G. (2002). Assembly and iron binding properties of human frataxin, the protein deficient in Friedreich ataxia. Hum. Mol. Genet. 33, 217–227. Cerenius, Y., Sta˚hl, K., Svensson, L.A., Ursby, T., Oskarsson, A˚., Albertsson, J., and Liljas, A. (2000). The crystallography beamline I711 at MAX II. J. Synchrotron Radiat. 7, 203–208. Chasteen, N.D., and Harrison, P.M. (1999). Mineralization in ferritin: an efficient means of iron storage. J. Struct. Biol. 126, 182–194. Cho, S.J., Lee, M.G., Yang, J.K., Lee, J.Y., Song, H.K., and Suh, S.W. (2000). Crystal structure of Escherichia coli CyaY protein reveals a previously unidentified fold for the evolutionarily conserved frataxin family. Proc. Natl. Acad. Sci. USA 97, 8932–8937. Cook, J.D., Bencze, K.Z., Jankovic, A.D., Crater, A.K., Busch, C.N., Bradley, P.B., Stemmler, A.J., Spaller, M.R., and Stemmler, T.L. (2006). Monomeric yeast frataxin is an iron-binding protein. Biochemistry 45, 7767–7777. Cornell, W.D., Cieplak, P., Bayly, C.I., Gould, I.R., Merz, K.M., Ferguson, D.M., Spellmeyer, D.C., Fox, T., Caldwell, J.W., and Kollman, P.A. (1995). A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 117, 5179–5197.

function of frataxin that limits oxidative damage and preserves cell longevity. Hum. Mol. Genet. 15, 467–479. Garman, E. (1999). Leaving no element of doubt: analysis of proteins using microPIXE. Structure 7, R291–R299. Garman, E.F., and Grime, G.W. (2005). Elemental analysis of proteins by microPIXE. Prog. Biophys. Mol. Biol. 89, 173–205. Gerber, J., Muhlenhoff, U., and Lill, R. (2003). An interaction between frataxin and Isu1/Nfs1 that is crucial for Fe/S cluster synthesis on Isu1. EMBO Rep. 4, 906–911. Grant, R.A., Filman, D.J., Finkel, S.E., Kolter, R., and Hogle, J.M. (1998). The crystal structure of Dps, a ferritin homolog that binds and protects DNA. Nat. Struct. Biol. 5, 294–303. Grime, G.W., Dawson, M., Marsh, M., McArthur, I.C., and Watt, F. (1991). The Oxford submicron nuclear facility. Nucleic Instrum. Meth. B 54, 52–53. Halliwell, B. (1978). Superoxide-dependent formation of hydroxyl radicals in the presence of iron salts. Its role in degradation of hyaluronic acid by a superoxide-generating system. FEBS Lett. 96, 238– 242. He, Y., Alam, S.L., Proteasa, S.V., Zhang, Y., Lesuisse, E., Dancis, A., and Stemmler, T.L. (2004). Yeast frataxin solution structure, iron binding, and ferrochelatase interaction. Biochemistry 43, 16254– 16262. Hempstead, P.D., Hudson, A.J., Artymiuk, P.J., Andrews, S.C., Banfield, M.J., Guest, J.R., and Harrison, P.M. (1994). Direct observation of the iron binding sites in a ferritin. FEBS Lett. 350, 258–262. Johansson, S.A.E., Campbell, J.L., and Malqvist, K.G. (1995). Particle induced X-ray emission spectrometry (PIXE). In Chemical Analysis: A Series of Monographs on Analytical Chemistry and Its Applications, S.A.E. Johansson, J.L. Campbell, and K.G. Malmqvist, eds. (New York: John Wiley and Sons, Inc.), p. 167. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47(Pt. 2), 110–119. Kabsch, W. (1993). Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800. Layer, G., Ollagnier-de Choudens, S., Sanakis, Y., and Fontecave, M. (2006). Iron-sulfur cluster biosynthesis: characterization of Escherichia coli CyaY as an iron donor for the assembly of [2Fe-2S] clusters in the scaffold IscU. J. Biol. Chem. 281, 16256–16263. Lill, R., Diekert, K., Kaut, A., Lange, H., Pelzer, W., Prohl, C., and Kispal, G. (1999). The essential role of mitochondria in the biogenesis of cellular iron-sulfur proteins. Biol. Chem. 380, 1157–1166. Ludtke, S.J., Baldwin, P.R., and Chiu, W. (1999). EMAN: semiautomated software for high-resolution single-particle reconstructions. J. Struct. Biol. 128, 82–97. Mammen, C.B., Ursby, T., Thunnissen, M., and Als-Nielsen, J. (2004). Bent Diamond Crystals and Multilayer Based Optics at the new 5-Station Protein Crystallography Beamline ‘Cassiopeia’ at MAX-lab. AIP Conference Proceedings 705, 808–811. McCoy, A.J., Grosse-Kunstleve, R.W., Storoni, L.C., and Read, R.J. (2005). Likelihood-enhanced fast translation functions. Acta Crystallogr. D Biol. Crystallogr. 61, 458–464. Nair, M., Adinolfi, S., Pastore, C., Kelly, G., Temussi, P., and Pastore, A. (2004). Solution structure of the bacterial frataxin ortholog, CyaY: mapping the iron binding sites. Structure 12, 2037–2048.

Dhe-Paganon, S., Shigeta, R., Chi, Y.I., Ristow, M., and Shoelson, S.E. (2000). Crystal structure of human frataxin. J. Biol. Chem. 275, 30753–30756.

Nichol, H., Gakh, O., O’Neill, H.A., Pickering, I.J., Isaya, G., and George, G.N. (2003). Structure of frataxin iron cores: an X-ray absorption spectroscopic study. Biochemistry 42, 5971–5976.

Duby, G., Foury, F., Ramazzotti, A., Herrmann, J., and Lutz, T. (2002). A non-essential function for yeast frataxin in iron-sulfur cluster assembly. Hum. Mol. Genet. 11, 2635–2643.

Nordlund, P., and Eklund, H. (1995). Di-iron-carboxylate proteins. Curr. Opin. Struct. Biol. 5, 758–766.

Gakh, O., Adamec, J., Gacy, M.A., Twesten, R.D., Owen, W.G., and Isaya, G. (2002). Physical evidence that yeast frataxin is an iron storage protein. Biochemistry 41, 6798–6804.

O’Neill, H.A., Gakh, O., Park, S., Cui, J., Mooney, S.M., Sampson, M., Ferreira, G.C., and Isaya, G. (2005a). Assembly of human frataxin is a mechanism to detoxify redox-active iron. Biochemistry 44, 537– 545.

Gakh, O., Park, S., Liu, G., Macomber, L., Imlay, J.A., Ferreira, G.C., and Isaya, G. (2006). Mitochondrial iron detoxification is a primary

O’Neill, H.A., Gakh, O., and Isaya, G. (2005b). Supramolecular assemblies of human frataxin are formed via subunit-subunit

Structure 1546

interactions mediated by a non-conserved amino-terminal region. J. Mol. Biol. 345, 433–439. Park, S., Gakh, O., Mooney, S.M., and Isaya, G. (2002). The ferroxidase activity of yeast frataxin. J. Biol. Chem. 277, 38589–38595. Park, S., Gakh, O., O’Neill, H.A., Mangravita, A., Nichol, H., Ferreira, G.C., and Isaya, G. (2003). Yeast frataxin sequentially chaperones and stores iron by coupling protein assembly with iron oxidation. J. Biol. Chem. 278, 31340–31351. Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612. Schoenfeld, R.A., Napoli, E., Wong, A., Zhan, S., Morin, D., Buckpitt, A.R., Taroni, F., Lonnerdal, B., Ristow, M., Puccio, H., and Cortopassi, G.A. (2005). Frataxin deficiency alters heme pathway transcripts and decreases mitochondrial heme metabolites in mammalian cells. Hum. Mol. Genet. 14, 3787–3799. Schultz, J.B., Dehmer, T., Schols, L., Mende, H., Hardt, C., Vorgerd, M., Burk, K., Matson, W., Dichgans, J., Beal, M.F., and Bogdanov, M.B. (2000). Oxidative stress in patients with Friedreich’s ataxia. Neurology 55, 1719–1721. Sitkoff, D., Sharp, K.A., and Honig, B. (1994). Accurate calculation of hydration free-energies using macroscopic solvent models. J. Phys. Chem. 98, 1978–1988. Stillman, T.J., Hempstead, P.D., Artymiuk, P.J., Andrews, S.C., Hudson, A.J., Treffry, A., Guest, J.R., and Harrison, P.M. (2001). The highresolution X-ray crystallographic structure of the ferritin (EcFtnA) of Escherichia coli; comparison with human H ferritin (HuHF) and the structures of the Fe(3+) and Zn(2+) derivatives. J. Mol. Biol. 307, 587–603. Thierbach, R., Schulz, T.J., Isken, F., Voigt, A., Mietzner, B., Drewes, G., von Kleist-Retzow, J.C., Wiesner, R.J., Magnuson, M.A., Puccio, H., et al. (2005). Targeted disruption of hepatic frataxin expression causes impaired mitochondrial function, decreased life span and tumor growth in mice. Hum. Mol. Genet. 14, 3857–3864. Vazquez-Manrique, R.P., Gonzalez-Cabo, P., Ros, S., Aziz, H., Baylis, H.A., and Palau, F. (2006). Reduction of Caenorhabditis elegans frataxin increases sensitivity to oxidative stress, reduces lifespan, and causes lethality in a mitochondrial complex II mutant. FASEB J. 20, 172–174. Williams, R.J. (1982). Free manganese (II) and iron (II) cations can act as intracellular cell controls. FEBS Lett. 140, 3–10. Yoon, T., and Cowan, J.A. (2003). Iron-sulfur cluster biosynthesis. Characterization of frataxin as an iron donor for assembly of [2Fe2S] clusters in ISU-type proteins. J. Am. Chem. Soc. 125, 6078–6084. Yoon, T., and Cowan, J.A. (2004). Frataxin-mediated iron delivery to Ferrochelatase in the final step of Heme biosynthesis. J. Biol. Chem. 279, 25943–25946. Zhang, Y., Lyver, E.R., Knight, S.A., Lesuisse, E., and Dancis, A. (2005). Frataxin and mitochondrial carrier proteins, Mrs3p and Mrs4p, cooperate in providing iron for heme synthesis. J. Biol. Chem. 280, 19794–19807.

Accession Numbers Atomic coordinates for the reported crystal structures have been deposited in the Protein Data Bank with accession number 2FQL.