ISCA1 is essential for mitochondrial Fe4S4 biogenesis in vivo

Abstract

Mammalian A-type proteins, ISCA1 and ISCA2, are evolutionarily conserved proteins involved in iron-sulfur cluster (Fe-S) biogenesis. Recently, it was shown that ISCA1 and ISCA2 form a heterocomplex that is implicated in the maturation of mitochondrial Fe₄ S₄ proteins. Here we report that mouse ISCA1 and ISCA2 are Fe₂ S₂ -containing proteins that combine all features of Fe-S carrier proteins. We use biochemical, spectroscopic and in vivo approaches to demonstrate that despite forming a complex, ISCA1 and ISCA2 establish discrete interactions with components of the late Fe-S machinery. Surprisingly, knockdown experiments in mouse skeletal muscle and in primary cultures of neurons suggest that ISCA1, but not ISCA2, is required for mitochondrial Fe₄ S₄ proteins biogenesis. Collectively, our data suggest that cellular processes with different requirements for ISCA1, ISCA2 and ISCA1-ISCA2 complex seem to exist.

ARTICLE

Received 19 May 2016 | Accepted 2 Mar 2017 | Published 11 May 2017

DOI: 10.1038/ncomms15124 OPEN

ISCA1 is essential for mitochondrial Fe4S4 biogenesis in vivo

Lena Kristina Beilschmidt1,2,3,4,5, Sandrine Ollagnier de Choudens6,7,8, Marjorie Fournier1,2,3,4,w, Ioannis Sanakis9,

Marc-Andr Hograindleur6,7,8,10, Martin Clmancey7,8,10, Genevive Blondin7,8,10, Stphane Schmucker1,2,3,4,5,

Aurlie Eisenmann1,2,3,4,5, Amlie Weiss1,2,3,4, Pascale Koebel1,2,3,4, Nadia Messaddeq1,2,3,4,

Hlne Puccio1,2,3,4,5 & Alain Martelli1,2,3,4,5,w

Mammalian A-type proteins, ISCA1 and ISCA2, are evolutionarily conserved proteins involved in ironsulfur cluster (FeS) biogenesis. Recently, it was shown that ISCA1 and ISCA2 form a heterocomplex that is implicated in the maturation of mitochondrial Fe4S4 proteins. Here we report that mouse ISCA1 and ISCA2 are Fe2S2-containing proteins that combine all features of FeS carrier proteins. We use biochemical, spectroscopic and in vivo approaches to demonstrate that despite forming a complex, ISCA1 and ISCA2 establish discrete interactions with components of the late FeS machinery. Surprisingly, knockdown experiments in mouse skeletal muscle and in primary cultures of neurons suggest that ISCA1, but not ISCA2, is required for mitochondrial Fe4S4 proteins biogenesis. Collectively, our data suggest that cellular processes with different requirements for ISCA1, ISCA2 and ISCA1ISCA2 complex seem to exist.

1 IGBMC (Institut de Gntique et de Biologie Molculaire et Cellulaire) Translational Medicine and Neurogenetics Department, 67404 Illkirch, France.

2 Inserm U596, 67404 Illkirch, France. 3 CNRS, UMR7104, 67404 Illkirch, France. 4 Universit de Strasbourg, 67000 Strasbourg, France. 5 Collge de France, Chaire de Gntique Humaine, 67404 Illkirch, France. 6 CEA/DRF/BIG/CBM/BioCat, 38054 Grenoble, France. 7 CNRS UMR 5249, LCBM, 38054 Grenoble, France. 8 Universit Grenoble Alpes, LCBM, 38054 Grenoble, France. 9 NCSR, Demokritos, Institut of Materials Science, Attiki, Greece. 10 CEA/DRF/BIG/ CBM/pmb, 38054 Grenoble, France. w Present addresses: Sir William Dunn School of Pathology, University of Oxford, Oxford OX1 3RE, UK (M.F.); Rare

Disease Research Unit, Pzer Inc., Cambridge, Massachusetts 02139, USA (A.M.). Correspondence and requests for materials should be addressed to H.P. (email: mailto:[email protected]

Web End [email protected] ) or to A.M. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

Ironsulfur clusters (FeS) are ancient and essential cofactors that participate in a number of cellular processes ranging from mitochondrial respiration to DNA metabolism1,2. In

eukaryotes, de novo FeS biogenesis takes place within mitochondria and relies on proteins that are highly conserved, with homologues found in bacteria and yeast. During de novo FeS biogenesis, a cysteine desulfurase complex, consisting of NFS1 and ISD11, provides the inorganic sulfur and binds the scaffold protein ISCU on which the cluster is assembled. Frataxin (FXN), the protein decient in Friedreich ataxia, interacts with the ternary ISCUNFS1ISD11 complex and controls both iron entry and sulfur transfer to ISCU by regulating the cysteine desulfurase activity36. Electrons that are required for FeS synthesis are most likely provided by the ferredoxin FDX2 (ref. 7). After de novo assembly, the cluster on ISCU is transferred to mitochondrial acceptor proteins via a late-acting FeS machinery involving poorly characterized accessory proteins1,2. In parallel, a sulfur compound, most likely derived from glutathione, is provided by the mitochondrial FeS machinery and is exported to the cytosol by the ABCB7 transporter where it is used as substrate to elaborate FeS for extra-mitochondrial acceptor proteins8.

A-type proteins are evolutionarily conserved proteins involved in FeS biogenesis. In bacteria, several A-type proteins can co-exist. Escherichia coli encodes four A-type proteins: IscA encoded by the isc operon9, SufA encoded by the suf operon10, ErpA11 and NfuA12,13. In yeast Saccharomyces cerevisiae and in mammals, two A-type proteins, ISCA1 (yeast Isa1) and ISCA2 (yeast Isa2), are nuclear-encoded and addressed to mitochondria14,15. Despite their presence throughout phyla, the role of A-type proteins remains unclear.

Bacterial A-type proteins were reported to bind either iron or FeS, and accordingly to display a role either as iron donor16,17, in particular during FeS cluster assembly on IscU, or as FeS carrier/scaffold proteins that can provide FeS to acceptor proteins10,12,13,1820. Furthermore, under aerobic condition in E. coli, only the double iscA sufA deletion is lethal, suggesting that bacterial A-type proteins are functionally redundant2123.

In S. cerevisiae, Isa1 and Isa2 single or double deletion strains are viable and were initially reported to present both mitochondrial and cytosolic FeS defects14,24,25. However, Isa1 and Isa2, together with their binding partner Iba57, were more recently shown to be specically required for the maturation of mitochondrial Fe4S4 proteins26,27. Interestingly, only DIsa1 yeast strain was shown to be complemented by E. coli A-type proteins (IscA, SufA or ErpA)27, thus supporting the non-redundant function of eukaryotic A-type proteins. In mammals, human ISCA1 and ISCA2 knockdown experiments in HeLa cells suggest that both proteins may function together during the late biogenesis of mitochondrial Fe4S4 (ref. 15). Recent in vitro biochemical studies using recombinant human ISCA1 and ISCA2 demonstrated a cluster transfer and a proteinprotein interaction between human glutaredoxin GLRX5 and ISCA1 or ISCA2 (ref. 28). Furthermore, a human ISCA1ISCA2 heterodimeric complex was shown to act as a platform to assemble a Fe4S4

cluster from two Fe2S2 clusters present on GLRX5 dimer in vitro29, thus providing a potential mechanism by which ISCA1 and ISCA2 could supply Fe4S4 in vivo.

To get further insights into the function of eukaryotic ISCA1 and ISCA2, we use the murine proteins and performed multiple complementary in vitro and in vivo characterizations. Although both proteins display key characteristics of FeS carriers in vitro, in vivo data provide evidence that only ISCA1 is essential for mitochondrial Fe4S4 proteins under dened physiological conditions in skeletal muscle or primary neuronal cells, while ISCA2 is

dispensable. Together, our results suggest that ISCA1 and ISCA2 do not necessarily act as a complex to provide mitochondrial Fe4S4, and that a dynamic network probably exists in vivo.

ResultsMouse ISCA1 and ISCA2 are Fe2S2-containing proteins. Mouse

Isca1 and Isca2 were cloned to produce the respective recombinant proteins in bacteria. Whereas pure His-ISCA2 was obtained under anaerobic condition, His-ISCA1 could not be puried under this condition due to precipitation of the recombinant protein during purication, and could only be puried aerobically. On analytical Superdex-75 column, His-ISCA2 was eluted mainly as a single symmetric peak corresponding to a dimer (theoretical mass of the monomer: 13,554 Da) while His-ISCA1 behaved as a mixture of dimeric (main) and tetrameric (minor) forms (theoretical mass of the monomer: 13,898.8 Da) (Supplementary Fig. 1A). After purication both proteins are coloured. Chemical analysis revealed roughly stoichiometric amounts of iron and sulde for both proteins with an average of0.9 iron and 0.9 sulfur atom/monomer for His-ISCA2 and 0.9 iron and 0.8 sulfur atom/monomer for His-ISCA1, suggesting the presence of an ironsulfur cluster. The UV-visible spectra of both proteins are in agreement with this hypothesis with absorption bands at 420 and 320 nm for both proteins and additional shoulders at 460 and 550 nm for His-ISCA2 (Supplementary Fig. 1B), suggesting the presence of a Fe2S2 cluster. The 4.2 K

Mssbauer spectra of a 57Fe-labelled His-ISCA2 and 57Fe-labelled His-ISCA1 consisted of one symmetric quadrupole doublet, representing almost 100% of the iron content, whose parameters (d 0.27(0.01) mm s 1 and DEQ 0.53(0.03) mm s 1 for

ISCA2; d 0.28(1) mm s 1 and DEQ 0.50(2) mm s 1 for His-

ISCA1), unambiguously demonstrated the presence of a [Fe2S2]2 (S 0) cluster (Fig. 1a). We further analysed both

ISCA1 an ISCA2 clusters by EPR spectroscopy. After reduction with dithionite, the initial EPR-silent ISCA2 was converted to a S 1/2 species, characterized by a rhombic EPR signal with

g values at g1 1.99, g2 1.97 and g3 1.91 (Fig. 1b).

Temperature dependence and microwave power saturation properties of the signal were in agreement with a [Fe2S2] . The signal integrated for 20% of total iron. A similar EPR signal, integrating for 15% of total iron, was obtained using dithiothreitol (DTT), showing that the low yield of the FeS under the

1 oxidation state was not due to a strong reducing treatment that can partially destroy the cluster18. Similar results were obtained for His-ISCA1 (Fig. 1b). Collectively, our data indicate that both as-puried His-ISCA1 and His-ISCA2 bind an oxidized Fe2S2 under a 2 oxidation state.

To ensure that the Fe2S2 cluster on ISCAs do not result from degradation of a Fe4S4 cluster during purication steps, we performed in cellulo Mssbauer experiments by comparing FeS content of bacteria overexpressing ISCA1 or ISCA2 with control bacteria cultured under the same conditions. Despite some variations, all control spectra (Fig. 1c, upper panels) were comparable to those previously published in the literature3034. The main difference between the spectrum recorded on bacteria overexpressing ISCA2 proteins (Fig. 1c, lower right panel) and that of the control (Fig. 1c, upper right panel) is the additional line clearly observed at about 0.5 mm s 1 (Fig. 1c, dashed line), suggesting the contribution of [Fe2S2]2 clusters. Satisfying simulations were obtained for both spectra (see Methods Section,

Fig. 1c, Table 1). Twenty per cent of the total iron content were observed as [Fe2S2]2 clusters for ISCA2 samples. Similar results were obtained on four different sample preparations for ISCA2.

Furthermore, using a more efcient expression system (see Methods section), we were able to increase the spectral

2 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124 ARTICLE

a

ISCA1

ISCA2

0.1%

0.4%

2

1

0 1 2

2

1

0 1 2

Velocity (mm s1)

Velocity (mm s1)

b

1.99

1.99

1.97

1.91

ISCA1

ISCA2

1.958

1.92

3,200 100 300 500 700 900

3,400 3,600 3,800

Magnetic field (Gauss) Magnetic field (Gauss)

c

ISCA1 ISCA2

Control cells Overexpressing cells

0.5 %

0.5 %

0.5 %

0.5 %

2 0 2 4

2 0 2 4

Velocity (mm s1) Velocity (mm s1)

Figure 1 | Spectroscopic characterization of ISCA1 and ISCA2. (a) Mssbauer spectra of the puried 57Fe/S-ISCA1 (1.1 mM, 1 Fe and 0.9 S/protein) and

57Fe/S-ISCA2 (380 mM, 0.7 Fe and 0.65 S/protein). Mssbauer spectra were recorded at 4.2 K in a magnetic eld of 83 and 60 mT applied perpendicular to the g-beam for ISCA1 and ISCA2, respectively. The solid lines correspond to simulation of the experimental spectra. (b) X-band EPR spectra of as-isolated ISCA1 (250 mM) and ISCA2 (500 mM) after reduction with 1 mM dithionite for 10 min. (c) Mssbauer spectra recorded on whole control cells (upper panels) and ISCA1 or ISCA2-overexpressing cells (lower panels) at 5.5 K using a 60 mT external magnetic eld applied parallel to the g-beam.

Experimental spectra are shown with hatched marks and simulations are overlaid as solid black lines. Five components were used for simulation: HS FeII

(light and dark green), Fe4S4 clusters and LS ferrous haems (light blue), FeIII NP (dark blue) and [Fe2S2]2 (red) (see Table 1 for parameter values).

intensity signal, and 30% of the total iron content was observed as [Fe2S2]2 clusters for ISCA2 (Supplementary Fig. 1C and

Supplementary Table 1). Comparable results were obtained by overexpressing ISCA1. However, due to the low solubility of the ISCA1 protein in these conditions, the signal was more variable and a signicant signal for [Fe2S2]2 clusters could only be observed in three performed experiments with the best one

showing up to 24% contribution of the [Fe2S2]2 clusters (Fig. 1c, lower left panel, Table 1). For both induced ISCA cell samples, the amount of additional FeS cluster is larger than the experimental uncertainty and comparable to that measured in other studies30,31. A perusal of Table 1 reveals that the total contribution of high-spin (HS) FeII and Fe4S4 components did not vary signicantly between control and induced ISCA cell

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

Table 1 | Values of the nuclear parameters and contributions for the ve components issued from simulations of the spectra*.

Species d(mm s 1) DEQ(mm s 1) C(mm s 1)w % in control cell samples % in induced ISCA cells ISCA1HS FeII 1.36 3.13 0.43 25 39 30 43HS FeII 1.18 2.97 0.43 14 13[Fe4S4]2 and LS FeII haem 0.45 1.15 0.40 15 61 12 34

FeIII NP 0.48 0.61 0.57 45 22[Fe2S2]2 z 0.27 0.50 0.28 o2 o4 12 240.28 0.68 0.28 o2 12

ISCA2HS FeII 1.36 3.13 0.52 36 57 41 54 HS FeII 1.18 2.97 0.52 21 13[Fe4S4]2 and LS FeII haem 0.45 1.15 0.41 13 43 10 26

FeIII NP 0.48 0.61 0.60 30 16[Fe2S2]2 z 0.27 0.50 0.32 o2 o4 10 200.28 0.68 0.32 o2 10

*Shown in Fig. 1c Italic numbers indicate the overall contributions of HS FeII species, of the two components with 0.40.5 mm s 1 isomer shift values (Fe S clusters, LS ferrous haems and FeIII NP), and of the two sites of the Fe S clusters. Uncertainties are 0.02mms 1 on d, DE and G and 3 on the percentage of each site.wThe linewidth for HS FeII components were constrained to be equal. The same holds for the two sites of the [Fe S ]2 clusters.

zIsomer shift and quadrupole splitting values are those published for the [Fe S ]2 cluster of NifIscA (ref. 53) and were xed. The two sites were assumed to contribute in a 1:1 ratio.

samples. Conversely, the contributions of the component corresponding to FeIII nanoparticles (FeIII NP) exhibited an important decrease (23% for ISCA1 and 14% for ISCA2), suggesting that the oxidized Fe2S2 detected in the ISCA-overexpressing cells mainly originated from this pool of iron. Since oxygen-sensitive Fe4S4 clusters of proteins were shown to be detectable by in cellulo Mssbauer spectroscopy using bacteria overexpressing the corresponding protein31,3537, our results indicate that mouse recombinant ISCA1 and ISCA2 are produced as Fe2S2 proteins in bacteria.

Mouse ISCA1 and ISCA2 behave as FeS carriers in vitro. An FeS carrier protein is dened by its capacity (i) to accept FeS from a scaffold protein and (ii) to transfer its FeS to target proteins. The FeS transfer between the ISCU scaffold and ISCAs was rst investigated using the His-tagged mouse ISCA and ISCU proteins. Fe2S2-ISCA1 or Fe2S2-ISCA2 was incubated, under anaerobic condition, with one equivalent of apo-ISCU. After desalting and separation, ISCA2 (initially with 0.6 Fe and 0.6 S per monomer) was still containing 0.4 Fe and 0.45 S/monomer, and the UV-visible spectrum (absorption bands at 320, 420 and 460 nm) was very similar to the one before incubation, showing that no cluster transfer occurred (Fig. 2a, upper left panel). Accordingly, no iron or sulfur could be detected on ISCU and the UV-visible spectra of ISCU were similar before and after incubation (Fig. 2a, lower left panel). A similar result was obtained using Fe2S2-ISCA1 as FeS donor (Supplementary Fig. 1D).

However, when holo-ISCU was incubated with apo-ISCA2, cluster transfer was observed, with changes in the UV-visible spectra (Fig. 2a, right panels), a loss of 98% of the iron on ISCU, and 0.7 Fe/monomer and 0.6 S/monomer on ISCA2 after separation. Owing to the instability of apo-ISCA1, the transfer between Fe2S2-ISCU and apo-ISCA1 could not be conclusively assessed.

The capacity of ISCA proteins to transfer their cluster to Fe2S2

and Fe4S4 target proteins was then investigated using E. coli apo ferredoxin (Fdx) and mouse mitochondrial aconitase (ACO2), respectively. After 30 min incubation of apo-Fdx with an excess of Fe2S2-ISCA2, the characteristic UV-visible spectrum of Fe2S2-

Fdx, with lmax at 415 and 460 nm, was obtained (Fig. 2b).

Reduction of the ISCA2-Fdx mixture with dithionite and further analysis by EPR unambiguously demonstrated the formation of

holo-Fdx, since a S 1/2 EPR signal characteristic of the reduced

[Fe2S2] cluster of Fdx was observed (Supplementary Fig. 1E)38. Similar results were obtained when Fe2S2-ISCA1 was used as a FeS donor protein (Supplementary Fig. 1F). Apo-ACO2 was incubated anaerobically with a 3.5-fold molar excess of either native ISCA1 or ISCA2, and transfer was monitored by measuring ACO2 enzyme activity. Upon incubation with Fe2S2-ISCA1 or Fe2S2-ISCA2, ACO2 activity was progressively increasing and reached more than 70% and 80% of the activity of a chemically reconstituted ACO2 after 30 min incubation, respectively (Fig. 2c). FeS transfer was observed only in the presence of a reducing agent such as DTT in agreement with the requirement of electrons to generate Fe4S4 from Fe2S2 clusters18.

These data clearly show that native ISCA proteins can transfer their FeS to both Fe2S2 and Fe4S4 apo-proteins.

Together, our results show that both ISCA1 and ISCA2 have key characteristics of FeS carrier proteins in vitro.

ISCA1 and ISCA2 have distinct interacting partners. To further dene the process in which ISCA proteins are involved, we sought interacting partners by performing immunoprecipitation (IP) experiments coupled to multidimensional protein identication technology (MudPIT). IP were carried out using HeLa mitochondrial extracts expressing mouse ISCA1, ISCA2, IBA57, FDX2 or GLRX5 with a C-terminal Flag epitope, respectively. MudPIT analysis was then performed on crude IP elutions. Specic interacting partners were ltered with high stringency criteria, to be enriched ten times minimum over both control-IP and mitochondrial-enriched extracts. Their relative abundance was estimated by label-free quantication analysis, based on normalized spectral abundance factor (NSAF). The full data set can be found in Supplementary data le. To get insights on the role of ISCA proteins in FeS biogenesis, we restricted our current analysis to the identication of proteins known to be involved in the FeS pathway (Fig. 3a, Supplementary Data 1). As previously described27,29, a reciprocal interaction between ISCA1 and ISCA2 was found (Fig. 3a). The interaction was conrmed in an independent ISCA1-Flag IP in N2a cells, followed by western blot analysis with an ISCA2 antibody (Fig. 3b). Interestingly, ISCA2 and IBA57 reciprocally co-puried, whereas no interaction was detected between IBA57 and ISCA1 (Fig. 3a). The specic interaction between ISCA2 and IBA57

4 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124 ARTICLE

a

Holo-ISCA2 + apo-ISCU

2

Apo-ISCA2 + holo-ISCU

0.6

ISCA2 before

ISCA2 after

ISCA2 before

ISCA2 after

1.5

0.4

Absorbance

Absorbance

1

0.2

0.5

0

0

0.8

2.5

ISCU before

ISCU after

0.6

Absorbance

0.4

Absorbance

1.5

0.2

0.5

0 300 500 600 700Wavelength (nm)

400

400

300 500 600 700Wavelength (nm)

b c

100

2

Fe2S2-Fdx

ISCA1

ISCA2

0.2

80

t=0

Absorbance

ACO2 activity (%)

1.5

0.1

60

1

0 300

400 500 600 700

40

20

0.5

t=0

t=15 min

t=30 min

400

0 5 10 15 20 30

0 300 500 600 700Wavelength (nm)

Time (min)

Figure 2 | FeS transfer experiments using ISCA1 and ISCA2 recombinant proteins. (a) UV-visible spectra obtained by mixing Fe2S2-ISCA2 with apo-ISCU (left panels) and apo-ISCA2 and Fe2S2-ISCU (right panels). (b) Monitoring of the apo-ferredoxin (thin line, 100 mM) and as-isolated Fe2S2-ISCA2 (bold line, 200 mM) mixture after 15 min (dotted line) and 30 min (dashed line) incubation. The inset shows the UV-visible spectrum of the

Fe2S2-ferredoxin (20 mM) obtained after separation onto a NiNTA column. (c) ACO2 (0.2 nmol) activity after incubation with tenfold excess of as-isolated ISCA1 (0.9 Fe/monomer) or ISCA2 (0.6 Fe/monomer) at 5, 10, 15, 20 and 30 min. Reconstituted ACO2 (0.2 nmol, 3.8 iron/protein) was used as positive control to set the 100% activity. Data are represented as the mean of three measurementss.d.

could be conrmed in IP experiments performed with N2a cells (Fig. 3b). Similarly, GLRX5 and ISCA2 reciprocally co-puried, but with rather low NSAF values in the MudPIT analysis (Fig. 3a). In agreement, a low GLRX5 signal was detected in ISCA2-Flag IP experiments with N2a cells (Fig. 3b). Finally, whereas a faint interaction (low NSAF) with FDX2 was observed in the ISCA2-Flag IP (Fig. 3a), no peptide was detected in the reverse IP, suggesting that FDX2 may rather be an indirect or weak binding partner of ISCA2. In contrast, the de novo FeS biosynthesis complex components, NFS1, ISCU and ISD11 were found in high abundance in the FDX2-Flag IP (Fig. 3a), strongly supporting the current model of FDX2 as an electron donor for de novo FeS assembly7. In addition, NFU1 was detected in the ISCA1-Flag IP by MudPIT (Fig. 3a), and was conrmed in the ISCA1-Flag independent IP using N2a cells (Fig. 3b). The interaction appeared specic to ISCA1 since NFU1 was neither co-immunoprecipitated with ISCA2-Flag nor IBA57-Flag (Fig. 3a,b).

Collectively, these results demonstrate a direct interaction between ISCA1 and ISCA2, as well as between ISCA2, IBA57 and GLRX5, whereas NFU1 is a partner of ISCA1 (Fig. 3c). Our results show that despite the existence of a prominent ISCA1 ISCA2 complex, distinct complexes containing either ISCA1 or ISCA2 may exist in vivo.

ISCA1 is required for the maturation of mitochondrial Fe4S4.

To determine the in vivo functions of ISCA1 and ISCA2, we

carried out knockdown experiments by injecting recombinant adeno-associated virus (AAVs) encoding for specic shRNAs into the tibialis anterior (TA) of 4-week-old mice. A scrambled shRNA (CTL) and a shRNA directed against the expression of the ISCU scaffold protein were used as controls. Strong knockdowns were conrmed 3 weeks post injection (w.p.i.) by qRTPCR for ISCA1 and by western blot for ISCA2 and ISCU (Fig. 4a and Supplementary Fig. 2A). At dissection, no gross muscle defect due to the viral infection was observed and histological analysis using haematoxylin-eosin staining conrmed an overall preserved muscle organization (Fig. 4b, left panels). By electron microscopy, muscle bres were completely normal, and in particular, no mitochondrial disorganization was observed (Supplementary Fig. 2B). We next analysed FeS-dependent proteins in the TA muscle. The activity of succinate dehydrogenase (SDH), containing both Fe2S2 and Fe4S4 clusters, was strongly decreased in skeletal muscle after Isca1 and Iscu knockdowns (Fig. 4b, right panels and Supplementary Fig. 2C). A substantial decrease of mitochondrial Fe4S4 containing proteins, including aconitase (ACO2), complex II (SDH B) and two subunits of complex I (NDUFS3 and NDUFS5) was observed by western blot after Isca1 and Iscu knockdowns (Fig. 4a), in line with previously reported instability of the respective apo-proteins or complexes in the absence of FeS in eukaryotic cells15,39. In addition, a decrease in lipoic acid (LA) bound to pyruvate dehydrogenase (PDH) and a-ketogluterate dehydrogenase (KGDH) complexes was observed after Isca1 and Iscu knockdowns (Fig. 4a). Since the levels of the PDH-E2 subunit, on which LA is bound, were unchanged

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

a

Flag IP b

CTL ISCA1-Flag

IP IP

Input Input

GLRX5

M

ISCA1

ISCA2

IBA57

FDX2

FLAG

GLRX5

17

ISCA2

NFU1

FLAG

GLRX5

IBA57

NFU1

FLAG

GLRX5

17

ISCA1

ISCA2

IBA57

FDX2

GLRX5

NFU1

ISCU

ISD11

NFS1

HSCB

Relative protein abundance

NSAF (*100)

36

28

17

IBA57

Immunoprecipitated proteins

CTL ISCA2-Flag

IP IP

Input

Input

M

17

36

17

28

CTL IBA57-Flag

IP IP

Input

M

c

Input

36

17

28

ISCA2

NFU1

FDX2

ISCA1

17

ISCU

ISD11

NFS1

IBA57

NFU1

GLRX5

Figure 3 | ISCA1 and ISCA2 interaction network. (a) FeS machinery-related proteins and their relative abundance (NSAF) determined by MudPIT analysis after immunoprecipitation of the indicated Flag-tagged baits. Data represent proteins enriched 10 over controls. The colour intensity reects the

NSAF values multiplied by 100. (b) Representative western blot analysis using different available antibodies (as indicated) of ISCA1-Flag, ISCA2-Flag and IBA57-Flag IPs versus control IP (CTL) obtained after transfection of N2a cells with the respective constructs (nZ3). M indicates size markers.

(c) Interaction network within the FeS machinery based on western blot and MudPIT analysis.

(Fig. 4a), the observed decrease in LA reects a defect in the Fe4S4

enzyme lipoic acid synthase (LIAS). Conversely, no decrease in the tested mitochondrial Fe2S2 enzymes (FECH and RIESKE) or in the non-FeS containing enzyme COX IV was observed for the Isca1 knockdown. A decrease of Fe2S2 enzymes, in addition to

Fe4S4 enzymes, was however seen with the depletion of ISCU in accordance with the central role of the scaffold protein for all FeS (Fig. 4a). ISCA1 depletion also had no effect on extra-mitochondrial FeS protein levels (ABCE1 and GPAT) (Fig. 4a). A similar pattern for the ISCA1 knockdown was observed at 6 w.p.i. (Supplementary Fig. 2D).

The specicity of the observed effects was further controlled through the expression of a shRNA-resistant version of ISCA1 (ISCA1R), that is, containing a silent mutation in the region targeted by the shRNA. The combination of ISCA1 knockdown with overexpression of the ISCA1R was validated by qRTPCR using specic primer pairs (Supplementary Fig. 2F). All alterations in maturation of mitochondrial Fe4S4 enzymes were fully prevented by the expression of ISCA1R (Fig. 4c, left panel),

demonstrating the specic implication of ISCA1 in mitochondrial Fe4S4 protein maturation.

Quite surprisingly, knockdown of ISCA2 had no effect on any of the tested FeS proteins in muscle 3 w.p.i. (Fig. 4a) and 6 w.p.i. (Supplementary Fig. 2D). Accordingly, the SDH activity was normal on muscle sections (Fig. 4b, right panels and Supplementary Fig. 2C). Furthermore, IBA57 knockdown also did not impair any of the tested FeS proteins (Supplementary Fig. 2E).

To fully evaluate the potential functional redundancy between ISCA1 and ISCA2, we combined knockdown of endogenous Isca1 with overexpression of ISCA2. Knockdown of Isca1 and over-expression of ISCA2 were validated by qRTPCR and western blot, respectively (Fig. 4c, right panel and Supplementary Fig. 2F). In contrast to the full prevention that was obtained with ISCA1R expression (Fig. 4c, left panel), ISCA2 overexpression had no effect on the maturation of mitochondrial Fe4S4 proteins after

ISCA1 knockdown (Fig. 4c, right panel), hence demonstrating that ISCA2 cannot compensate for the loss of ISCA1 function

6 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124 ARTICLE

a

sh Isca1

CTL

sh Isca1

CTL

sh Isca2

CTL

sh Isca2

CTL

sh Iscu

CTL

sh Iscu

CTL

M

M

M

90

90

90

ACO2

SDH B

NDUFS3

NDUFS5

PDH-E2

28

28 17

28

17

72

55

36

55

36 17

28

17

28

17

72

55

36

55

36 17

28

28 17

28

17

28

FECH

RIESKE

GPAT

Mitochondrial

36

36

36

COX IV

LA PDH

KGDH

55

55

55

55

55

55

Cytosolic

72

55

36

55

36 17

17 ISCU

ABCE1

GAPDH

bTUB

IBA57

ISCA2

b SDH

H&E

c

CTL

sh Isca1

CTL

sh Isca1

CTL

+

Isca1R

ACO2

SDH B

GAPDH

ISCA2

+

+

Isca2

+

M

28

28

11

90

sh Isca1

sh Isca2

NDUFS3

LA PDH

KGDH

55

36

sh Iscu

Figure 4 | ISCA1 and ISCA2 knockdowns in skeletal muscle. (a) Representative western blots of two independent samples for the indicated proteins using extracts from TA muscle at 3 w.p.i., injected with rAAV-scrambled shRNA (CTL), rAAV-shIsca1, rAAV-shIsca2 or rAAV-shIscu (nZ5). GAPDH and beta-tubuline (bTUB) were used as loading controls. M indicates size markers. (b) Histological analysis after haematoxylin-eosin (H&E) or SDH activity staining on cryosections from TA muscle at 3 w.p.i., injected with rAAV-scrambled shRNA (CTL), rAAV-shIsca1, rAAV-shIsca2 or rAAV-shIscu (n 5). Scale bar,

50 mM (H&E) or 100 mM (SDH). (c) Representative western blots for the indicated proteins using extracts from TA muscle at 3 w.p.i., co-injected with rAAV-shISCA1 or rAAV-scramble shRNA (CTL) and rAAV-ISCA1R or rAAV-ISCA2 (n 4). GAPDH was used as loading control. M indicates size markers.

in vivo and that ISCA1 and ISCA2 are not functionally redundant.

Together, these results provide clear evidence that ISCA1 is required for the maturation of mitochondrial Fe4S4 proteins in mature skeletal muscle, whereas ISCA2 is not essential in this tissue, under standard physiological conditions.

To evaluate the cellular function of ISCA1 and ISCA2 in another cell type, we performed knockdown experiments on primary cultures of mouse sensory neurons. Primary cultures

were efciently transduced using AAV2/9 recombinant vectors coding for the shRNAs against Isca1 and Isca2, as indicated by the GFP signal observed in transduced cells (Fig. 5a). Fifteen days post-infection, a strong acidication of the medium (yellow colour) was observed in the culture infected with Isca1 shRNA (Fig. 5b), suggesting that ISCA1-depleted neurons are highly glycolytic and produce high levels of lactic acid due to mitochondrial dysfunction. Accordingly, the evaluation of LA levels by western blot showed a decrease of LA on a-ketogluterate

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

a

CTL

shlsca1 shlsca2

b

CTL shlsca1 shlsca2

SDH B

BF

GFP

M

70

d

shlsca1

shlsca2

c

CTL

shlsca1

shlsca2

PDH

KGDH

PDH-E2

GAPDH

NDUFS3

RIESKE

FECH

GAPDH

COX4

ISCA2

LA

55

70

35

600

500

OCR (pmol min1 )

M 25

25 25

55

35

15

15

e

CTL

shlsca1 shlsca2

FCCP

Antimycin

+ rotenone

CTL shlsca1 shlsca2

CTL

120

Basal respiration

Spare capacity

Oligomycin

400

300

200

100

0 0 10 20 30 40Time (min)

OCR (pmol min1 )

100

80

60

40

OCR (pmol min1 )

400

300

200

100

0

20

0

50 60 70 80

Figure 5 | ISCA1 and ISCA2 knockdowns in primary sensory neurons. (a) Brighteld and immunouorescence (GFP) analysis of neuronal cultures 4 days after infection with rAAV-scrambled shRNA (CTL), rAAV-shIsca1 or rAAV-shIsca2. Scale bar, 100 mm. (b) Representative picture of the neuronal cultures 15 days after infection with rAAV-scrambled shRNA (CTL), rAAV-shIsca1 or rAAV-shIsca2 (n 6). (c) Representative western blots of two independent

samples for lipoic acid (LA) and the E2 subunit of PDH (PDH-E2) using neuronal extracts 15 days after infection with rAAV-scrambled shRNA (CTL), rAAV-shIsca1 or rAAV-shIsca2. GAPDH was used as loading control (n 6). M indicates size markers. (d) Representative western blots of two

independent samples for the indicated proteins using neuronal extracts 15 days after infection with rAAV-scrambled shRNA (CTL), rAAV-shIsca1 or rAAV-shIsca2. GAPDH was used as loading control (n 6). M indicates size markers. (e) Representative Seahorse analysis using XF Cell Mito Stress Test

Kit obtained with neurons 8 days after infection with rAAV-scrambled shRNA (CTL), rAAV-shIsca1 or rAAV-shIsca2 (nZ5). Basal respirations and spare capacities are represented as the means.d. ***po0.001. OCR: oxygen consumption rate.

dehydrogenase and PDH complexes in Isca1 knockdowns, whereas no effect was observed in control or ISCA2-decient neurons (Fig. 5c). Similarly, the levels of mitochondrial Fe4S4-containing proteins (SDHB and NDUFS3) were specically reduced in Isca1 knockdowns (Fig. 5d).

The impact of both ISCA1 and ISCA2 depletions on mitochondrial respiration was investigated on primary neurons in culture using Seahorse and the XF Cell Mito Stress Test Kit. Four days after infection with AAVs, most cells expressed GFP (Supplementary Fig. 3A). Seahorse measurement was performed 8 days post-infection. A clear and drastic reduction of both the basal respiration and the spare respiratory capacity was observed in ISCA1-decient neurons (Fig. 5e), whereas no signicant difference was detected between control and ISCA2-depleted neurons despite a signicant decrease of ISCA2 protein level (Fig. 5e and Supplementary Fig. 3B). Hence, these results further demonstrate the specic role of ISCA1 in the maturation of

mitochondrial Fe4S4 that are essential for mitochondrial respiration and function.

DiscussionAmong components of the FeS protein assembly, the A-type ISC assembly factors are not fully characterized. In the present manuscript, we further validate the central role of ISCA1 for the maturation of mitochondrial Fe4S4 proteins while demonstrating that ISCA2 and IBA57 are not required under standard physiological conditions in two post-mitotic cell types.

Yeast Isa1 was the rst eukaryotic A-type protein to be puried, however obtained as an apo-protein that could bind an Fe2S2 only after chemical reconstitution40. More recently, Bianci et al. isolated anaerobically human ISCA2 in an Fe2S2 form, but again ISCA1 was shown to contain an Fe2S2 only after chemical reconstitution28. Our results demonstrate for the rst time that

8 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124 ARTICLE

both ISCA1 and ISCA2, as isolated (without any chemical treatment), are Fe2S2 proteins, and that the Fe2S2 clusters do not result from Fe4S4 degradation during purication based on our in cellulo experiments. Furthermore, our results are in agreement with the observation that ISCA proteins can function as FeS carriers. Accordingly, in vivo, we showed specic interactions of ISCA1 and ISCA2 with proteins of the late mitochondrial machinery, namely GLRX5, IBA57 and NFU1 proteins. Although a strong interaction of ISCA1 with ISCA2 was observed, consistent with previously reported data obtained in yeast or using the recombinant human proteins27,29, the functional relevance of this interaction in cells remains to be determined. Interestingly, the distinct respective interaction between ISCA2 and IBA57 contrasts with results published in yeast where both Isa1 and Isa2 interact with Iba57 protein by co-IP26. However, this specic ISCA2IBA57 interaction correlates with our in vivo data since neither ISCA2 nor IBA57 knockdown led to a phenotype associated with FeS metabolism. These results point to the complexity of the late FeS cluster assembly machinery, and that dynamic networks probably exist. Of note NFU1 was reported to bind Fe4S4 in vitro41, and recently identied mutation in NFU1 lead to multiple mitochondrial dysfunction syndrome42,43 with respiratory complex and LA deciencies that are relying on mitochondrial Fe4S4 biogenesis. Through the distinct interaction of ISCA1 with NFU1, we can therefore speculate that NFU1 might be involved with ISCA1 in Fe4S4

assembly and/or delivery to mitochondrial proteins.

Finally, the different phenotypes highlighted between ISCA1 and ISCA2 under our conditions contrast with data obtained in HeLa cells in which similar phenotypes were observed for ISCA1 and ISCA2 knockdowns15. However, the use of multiple transfections with specic siRNAs in HeLa cells that were required to observe a phenotype may have hampered the identication of primary specic effects. The role of ISCA2 in vivo remains difcult to dene as no phenotype was revealed under our conditions. The fact that a phenotype was found using HeLa cells, which are highly proliferative cells, may suggest that ISCA2 is required in FeS biogenesis during cell division, which may be in agreement with its implication during development4446 or depending on the physiological state of the cell (that is, proliferative, post-mitotic, oxidative stress, hypoxia, and so on). Interestingly, recently identied mutations in ISCA2 and IBA57 in human lead to severe genetic disorders in young children that may reect a specic implication of these two proteins during development4446. Differences between ISCA1 and ISCA2 are also highlighted by the fact that holo-ISCA proteins were obtained under different conditions: ISCA2 under anaerobiosis, whereas ISCA1 should be puried under aerobiosis. This may suggest that ISCA clusters have a different response to redox conditions. Whether such properties are related to specic physiological functions for each protein remains to be determined. Collectively, our results with the results in the literature strongly suggest that late mitochondrial FeS biogenesis is a complex and dynamic system that may have tissue and temporal specicity.

Methods

Cloning and plasmid constructs. Full length Isca1, Isca2, Glrx5, Fdx2 and Iba57 (C1orf69 homolog) cDNAs were PCR amplied from a bank of mouse heart cDNAs using specic primers and cloned in pCDNA3.1 (zeo) for the mamma

lian expression of the Flag-tagged proteins. Mature forms of Isca1 and Isca2 were cloned into pET-11a (Novagen) for the bacterial expression of the N-terminal His-tagged recombinant proteins or into pACYC-DUET-1 (Novagen) for the bacterial expression of non-tagged recombinant proteins. Specic shRNAs against mouse Isca1, Isca2, Iba57 or Iscu and scrambled shRNAs were cloned into a modied pENTR1A (Invitrogen Life Technologies) vector containing the U6 promoter. ShRNAs were subcloned into pAAV vector containing a CMV-eGFP cassette and

AAV2-specic ITR sequences using Gateway technology. Mouse ISCA1 encoded by a cDNA resistant to the shRNA (ISCA1R) was obtained by directed mutagenesis and cloned into pAAV-CMV (Stratagene). Site-directed mutagenesis was carried out using specic primers and the Phusion high-delity DNA polymerase (Finnzymes) with the GC-Buffer and following the manufacturers recommendations. All constructs were veried by Sanger sequencing. Oligonucleotide sequences are listed in Supplementary Table 2.

Production and purication of recombinant proteins. ISCA1 and ISCA2 were obtained by growing E. coli BL21(DE3)/pET-ISCA1 or pET-ISCA2 in LB medium containing 100 mg ml 1 ampicillin, at 37 C. ISCA2 expression was induced for 4 h by adding 0.5 mM isopropyl b-D-thiogalactoside (IPTG) to an exponentially growing culture. ISCA1 protein induction was performed at 20 C overnight. Bacterial pellets were resuspended in buffer A (50 mM Tris-HCl pH 8, 50 mM NaCl, 0.1% TritonX100, 5 mM DTT, 1 mM PMSF) and sonicated before ultra-centrifugation (90 min, 144,000g, 4 C). Supernatants were treated with DNAse I and 10 mM MgCl2 at 4 C for 45 min, and after centrifugation (30 min, 13,000g, 4 C), soluble proteins were loaded onto a 5 ml Ni-NTA afnity column (Qiagen) equilibrated with buffer B (50 mM Tris-HCl pH 8, 50 mM NaCl, 0.1% TritonX100). Fractions containing ISCA proteins were pooled after elution with buffer B containing 0.2 M imidazole. This protocol allowed the purication of approximately 5 and 40 mg of ISCA1 and ISCA2, respectively. For Mssbauer experiments, bacteria expressing ISCA1 or ISCA2 were grown aerobically in minimal M9

57Fe-enriched medium (35 mM) supplemented with 2 mM MgSO4, 0.4% glucose, 2 mg ml 1 thiamine, 1 mM CaCl2. 57Fe (Cambridge Isotopes) was prepared by resuspending 57Fe powder in concentrated HCl:HNO3 (1:1 ratio) in order to get a 500 mM solution. Proteins expressions as well as purications were performed as described above.

Mouse aconitase (ACO2) was puried from an E. coli BL21(DE3) strain containing pRARE2 plasmid (encoding rare tRNAs) and pGEX-ACO2 that were grown in LB at 37 C for 3 h after addition of 0.5 mM IPTG. The bacterial pellet was resuspended in buffer C (PBS supplemented with 0.1% TritonX100 and 1 mM PMSF) and sonicated before ultracentrifugation (90 min, 144,000g, 4 C). The supernatant was treated with DNAse I and 10 mM MgCl2 at 4 C for 45 min, and after centrifugation (30 min, 13,000g, 4 C), soluble proteins were loaded onto a5 ml GST-trap afnity column (Qiagen) equilibrated with buffer C without PMSF. The GST-fused ACO2 protein was eluted with buffer C containing 30 mM glutathione and pooled fractions were stored at 80 C. His-ISCU was obtained

from purication of the NFS1ISD11ISCU complex47 and a further gel ltration step onto a Superdex-75 (Pharmacia) equilibrated with buffer D (100 mM Tris-HCl pH 7.5, 100 mM NaCl). E. coli apo-ferredoxin (Fdx) was expressed in Terric broth for 18 h at 28 C and puried in three steps using DEAE-cellulose, anion and hydrophobic columns. Apoferredoxin was prepared by boiling puried holoferredoxin in the presence of 100 mM EDTA and 500 mM DTT and desalting in buffer D. Protein concentrations were measured by Bradford assay using bovine serum albumin as a standard. FPLC gel ltration with an analytical Superdex-75 (Pharmacia Amersham Biotech) at a ow rate of 0.5 ml min 1 equilibrated with buffer E (100 mM Tris-HCl pH 7.5, 100 mM NaCl, 5 mM DTT) was used for size determination of oligomerization state of the ISCA proteins. A gel ltration calibration kit (calibration protein II, Boehringer Inc) was used as molecular weight standards.

Whole cell Mssbauer spectroscopy. BL21(DE3) E. coli strain transformed with pET-ISCA1 or pET-ISCA2 or pACYC-ISCA2 plasmids were cultivated aerobically in 57Fe-enriched MOPS medium prepared as described48. Briey, transformed bacteria were cultured overnight at 37 C in LB medium supplemented with100 mg ml 1 ampicillin. Three ml were used to inoculate 2 1 l of MOPS

supplemented with 100 mg ml 1 ampicillin and 20 mM 57FeCl3. Cells were incubated at 37 C with agitation (200 rpm) until reaching an OD600 of about

0.50.6. Protein expression was then induced by adding 0.5 mM IPTG during 3 h at 37 C for ISCA2 or overnight at 20 C for ISCA1. Under these conditions both proteins were expressed under a soluble form, although the amount varied for ISCA1 from one experiment to another. Control cultures were obtained without IPTG induction. Cells were harvested by centrifugation (4,000g, 20 min, 4 C), washed twice with MOPS buffer containing 50 mM NaCl and 22 mM glucose and centrifuged again for 10 min at 5,000g, 4 C. Cells were then packed in a Mssbauer cup (0.60.7 g of cells per Mssbauer cup) and frozen in liquid nitrogen before analysis. Mssbauer experiments were successfully repeated three times for ISCA1, four times for ISCA2, and twice for pACYC-ISCA2 in order to ensure reliable data and the presented spectra reect the best experiment in terms of intensity and simulation. In cellulo Mssbauer spectra were recorded using a horizontal transmission 4 K closed cycle refrigerator system from Janis and SHI and a 100mCi source of 57Co(Rh) (ref. 49). All velocity scales and isomer shifts were referred to the metallic iron standard at room temperature. Analyses of the data were performed with the software WMOSS4 Mssbauer Spectral Analysis Software (http://www.wmoss.org

Web End =www.wmoss.org, 20092015).

Simulations were rst performed on the two control cell spectra. Four contributions were necessary to fully reproduce the experimental data. Introduction of a fth contribution due to [Fe2S2]2 unavoidably led to low percentages (o4% for the whole cluster). The quality of the simulations is

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

signicantly lost for Fe2S2 contributions larger than 4%. We constrained the nuclear parameters to be identical for the two calculations; only the linewidths and the relative area were allowed to differ. From the isomer shift value, two components were associated to HS FeII species. Nuclear parametersd 0.45 mm s 1 and DEQ 1.15 mm s 1 were typical of S 0 [Fe4S4]2 clusters

and low-spin (LS) ferrous haem centres. A broad doublet with parameters d 0.48 mm s 1 and DEQ 0.61 mm s 1 is typical of FeIII (phosphate)

oxyhydroxide nanoparticles (denoted FeIII NP)5052. Simulations were then run to reproduce the two experimental spectra recorded on cells with overexpressed ISCA1 or ISCA2. The nuclear parameters (isomer shift and quadrupole splitting) for the four components identied in the control cell spectra were xed to their determined values. Only the relative areas were allowed to vary. A fth component was introduced owing to the presence of the line atE0.5 mm s 1 (see main text).

Since we suspect the presence of [Fe2S2]2 clusters, we xed the nuclear parameter values of the two Fe sites in a 1:1 ratio to that determined for NifIscA (ref. 53).

Spectroscopic analyses on puried proteins. UV-visible absorption spectra were recorded with an Uvikon XL spectrophotometer (BioTek Instruments).

57Fe-Mssbauer spectra on puried proteins were recorded using 400 ml cuvettes containing 0.381.1 mM protein. Spectra were recorded on a spectrometer operating in constant acceleration mode using an Oxford cryostat that allowed temperatures from 1.5 to 300 K and a 57Co source in rhodium. Isomer shifts are reported relative to metallic iron at room temperature. EPR spectra of puried proteins (250500 mM) were obtained after reduction for 10 min with 1 mM dithionite prepared in buffer F (0.1M Tris-HCl pH 8) or for 40 min with 50 mM DTT prepared in buffer F. Spectra were recorded on a Bruker EMX (9.5 GHz) or ER200D EPR spectrometers equipped with an ESR 900 helium ow cryostat (Oxford Instruments). Double integrals of the EPR signals and spin concentration were obtained through the Win-EPR software using the spectrum of a 200 mM

Cu(EDTA) standard recorded under non-saturating conditions.

In vitro FeS cluster transfer. For FeS transfer between ISCA and ISCU, FeS-ISCA (ISCA1 and ISCA2, 60 nmol) were incubated in buffer G (0.1 M Tris-HCL pH 8, 50 mM NaCl) under anaerobic conditions with one equivalent of apo-ISCU for 1 h. After separation onto an analytical Superdex-75 column, equilibrated in buffer G, each protein was analysed for its iron and sulfur content and its UV-vis. spectrum. The experiment was also performed the other way around, using FeS-ISCU and apo ISCA (ISCA1 and ISCA2) under the same conditions (incubation time, protein equivalent, and so on). ISCA apo-proteins were prepared by incubating as-puried proteins with 20 mM EDTA and 2 mM dithionite under anaerobic conditions for one night and desalting onto a NAP25 column equilibrated with buffer G. Iron content was measured to check that negligible iron is still bound to proteins.

For FeS transfer between ISCA and Apo-Fdx, puried Apo-Fdx (180 mM) was incubated anaerobically in 200 ml buffer H (0.1 M Tris-HCl pH: 8; 50 mM KCl) with native ISCA1 (0.9 iron and 0.9 sulfur atom/monomer) or ISCA2 (1 iron and0.9 sulfur atom/monomer) in order to provide 2 iron and sulfur atoms/ferredoxin. FeS transfer was followed by monitoring the UV-visible absorption in the 300600 nm region and by EPR analysis of the ISCA-Fdx protein mixture after reduction with 2 mM dithionite.

To assess mouse aconitase activation, apo-ACO2 (0.2 nmol) was pretreated with DTT and then incubated anaerobically in a glove box at 18 C in buffer I (50 mM Tris-HCl pH: 7.6) with a tenfold molar excess of the native ISCA1 (0.9 iron and sulfur atom/monomer) or ISCA2 (0.6 iron and 0.7 sulfur atom/monomer) in order to provide 4 iron and 4 sulfur atoms per ACO2. After 5, 10, 15, 20 and 30 min incubation, aconitase activity was assessed as described54. Briey, ACO2-ISCA mixtures were added to 0.6 mM MnCl2, 25 mM citrate, 0.5 U isocitric dehydrogenase, 0.25 mM NADP , 50 mM Tris-HCl, pH 7.6, in a 100 ml nal volume and NADPH formation was monitored at 340 nm. The 100% activity corresponds to the activity of the chemically reconstituted ACO2 prepared by incubating apo-ACO2 with 4 molar excess of ferrous iron and sulfur for 30 min in the presence of 5 mM DTT (4 mmol min 1 mg 1).

Cell cultures. HeLa cells (ATCC-CCL-2) and Neuro-2a (N2a)(ATCC-CCL-131) cells were cultured in DMEM medium supplemented with 1 g l 1 glucose, 5% fetal calf serum and 40 mg ml 1 gentamycine at 37% with 5% CO2. Cell lines tested negative for mycoplasma contamination. HeLa cells and N2a cells were transfected with the different pcDNA3.1 constructs (Isca1, Isca2, Iba57, Glrx5, Fdx2) using Fugene 6 transfection reagent (Roche) and Lipofectamine transfection reagent (Thermosher), respectively, following the manufacturers recommendations. Transfected cells were harvested 2448 h after transfection.

Primary sensory neurons were obtained from the dissection of dorsal root ganglia (DRGs) of E13.5 mouse embryos55. Briey, DRGs were collected and digested in 1 ml of 2.5% trypsin for 45 min at 37 C. Digestion was stopped by adding two volumes of C-medium (MEM, glucose 4 g l 1, 10% fetal bovine serum, glutamine 2 mM, NGF 50 ng ml 1) supplemented with penicillin and streptomycin (P/S), and 1 volume of fetal bovine serum. Cells were centrifuged at 800g, 10 min at RT and mechanically dissociated in C-medium P/S using a 1 ml

micropipette. For western blot analysis, cells were plated on poly-lysine/collagen-

coated 35 mm plates in C-medium P/S (at a concentration of about 30 DRGs per

plate). For Seahorse analysis, the cell suspension was complemented with matrigel (Corning) in a 1:19 ratio and with 10 mg ml 1 of laminin (Sigma) to a nal concentration of approximately 0.03 DRG per ml. Eighty ml per well were then plated on poly-lysine/laminin-coated XF96 cell culture microplates (Seahorse bioscience). The day after dissection medium was changed to C-medium supplemented with 10 mM uorodeoxyuridine 10 mM uridine (FUdR) to get rid

of dividing cells (treatment was repeated the day after transduction with rAAV for cultures on 35 mm plates). The culture medium was then changed every 2 days with NB medium (Neurobasal, glucose 4 g l 1, B27, glutamine 2 mM, NGF50 ng ml 1). Cultures were maintained at 37% with 5% CO2. Transduction of neurons with the different rAAV2/9 was carried out overnight using a concentration of 9 109 vg ml 1 in C-medium (using 1 ml for 35 mm plates and

80 ml per well for the XF96 microplate). For 35 mm plates, infection was performed 4 days after dissection, whereas infection was performed the night after dissection for XF96 microplates. Cells were collected 15 days after infection and pellets were kept at 80 C.

Immunoprecipitation. IPs were carried out using mitochondrial-enriched protein extracts from HeLa or N2a cells overexpressing FLAG-tagged proteins56. Cells were collected in PBS and mitochondrial-enriched fractions were obtained by incubating cell pellets in Tris-HCl 100 mM, pH 7.5, 10% glycerol supplemented with 0.014% digitonin at 4 C for 10 min, and by centrifuging the suspension 10,000g 10 min at 4 C. The resulting pellet was resuspended in Tris-HCl 10 mM pH 7.4, 10% glycerol, 100 mM NaCl, 50 mM KCl, Complete protease inhibitor cocktail (Roche),0.2% Triton X-100, incubated on ice 20 min and centrifuged 10,000g 10 min at4 C. The resulting supernatant corresponding to the mitochondria-enriched protein extracts (0.38 mg) were incubated with 25100 ml FLAG M2 coupled resin (SIGMA) (50% slurry, prepared as recommended by the manufacturer) in the presence of Complete protease inhibitor (Roche). Suspensions were incubated 2 h at 4 C, centrifuged 10 min at 12,000g and beads were washed twice with 1 ml PBS. Elution was performed by incubating the resin 5 min with glycine buffer(0.1 M glycine, pH 2.8) and centrifugation for 10 min at 12,000g.

Mass spectrometry analysis. Immunoprecipitated samples were analysed by MudPIT as follows (three biological replicates for each individual protein). Flag IP eluates were TCA-precipitated, urea-denatured, reduced, alkylated and digested with endoproteinase Lys-C (Roche), followed by modied trypsin digestion (Promega). Peptide mixtures were loaded onto a triphasic 100 mm inner diameter fused silica microcapillary column, packed with C18 reverse phase (Aqua, Phenomenex) and strong cation exchange (Partisphere SCX, Whatman) particles. Loaded columns were placed in-line with a Dionex Ultimate 3,000 nano-LC (Thermo Fisher Scientic) and a LTQ Velos linear ion trap mass spectrometer equipped with a nano-LC electrospray ionization source (Thermo Fisher Scientic). A fully automated 12-step MudPIT run was performed as previously described57, during which each full MS scan (from 300 to 1,700 m/z range) was followed by 20 MS/MS events using data-dependent acquisition. Proteins were identied by database searching using SEQUEST HT within Proteome Discoverer1.4 software (Thermo Fisher Scientic) against a Homo sapiens database containing 20,206 protein sequences (Swissprot release from 2012-10-05) in which protein sequences of ISCA1, ISCA2, IBA57, FDX2 or GLRX5 have been replaced by the corresponding mouse protein sequence fused to Flag tag. Cysteine residues were considered to be fully carbamidomethylated ( 57 Da statically added) and

methionine considered to be oxidized ( 16 Da dynamically added). Peptides were

ltered with XCorr values equal to 1.5 ( 1), 2.5 ( 2), 3.0 ( 3) and 3.2 (4 3).

Only peptides with a minimum length of seven residues and maximum DCn values of 0.1 were retained. Specic interacting partners were considered to be enriched 10 over negative controls and mitochondrial input samples. IP normalization

was performed using at least one control IP and two mitochondrial input samples. Remaining proteins were selected for mitochondrial localization using SWISSprot library. SAF (PSM/protein length) and NSAF (SAF(protein)/sum SAF(all proteins)) values were calculated from sum PSM values of replicate IP experiments57. NSAF 100 values were used for representation.

Mice. All mice were in 100% C57BL/6J background. Mice were maintained in a temperature and humidity-controlled animal facility, with a 12-h lightdark cycle and free access to water and a standard rodent chow (D03, SAFE). Both male and female mice were used in all experiments. All animal procedures and experiments were approved by the local ethical committee (Comit dEthique en Exprimentation Animale IGBMC-ICS) and the Ministre de lEducation Nationale et de lEnseignement Suprieur et de la Recherche (project #02406.03). For knockdown experiments, 4-week-old C57BL/6J wild-type mice were anaesthetized by intraperitoneal injection of ketamine-xylazine (75 and 10 mg per kg body weight, respectively) and a dose of 2.5 1010 vg in a total volume of 25 ml of rAAV2/1 was

administered in each TA. Recombinant AAV2/1 expressing the specic shRNAs and the rAAV2/1 with the scrambled shRNA were injected at contralateral sides, respectively. Three or 6 weeks post-infection, mice were anaesthetized by intraperitoneal injection of ketamine-xylazine, dissected and killed. All experiments have biological replicates (n 510) for 3 w.p.i. and n 3 for 6 w.p.i. The sample

10 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124 ARTICLE

Statistical analysis. Differences between mean values were evaluated using the bilateral Students t-test. po0.05 was considered signicant.

Data availability. Data supporting the ndings of this study are available within the article and its supplementary information les and from the corresponding author on reasonable request. The genes with Ensembl accession codes ENSMUSG00000044792 (Isca11), ENSMUSG00000021241 (Isca2), ENSMUSG00000021102 (Glrx5), ENSMUST00000215338 (Fdx2) and ENSMUSG00000049287 (Iba57) were used in this work.

References

1. Beilschmidt, L. K. & Puccio, H. M. Mammalian Fe-S cluster biogenesis and its implication in disease. Biochimie 100, 4860 (2014).

2. Lill, R. Function and biogenesis of iron-sulphur proteins. Nature 460, 831838 (2009).

3. Colin, F. et al. Mammalian frataxin controls sulfur production and iron entry during de novo Fe4S4 cluster assembly. J. Am. Chem. Soc. 135, 733740 (2013).

4. Tsai, C. L. & Barondeau, D. P. Human frataxin is an allosteric switch that activates the Fe-S cluster biosynthetic complex. Biochemistry 49, 91329139 (2010).

5. Bridwell-Rabb, J., Fox, N. G., Tsai, C. L., Winn, A. M. & Barondeau, D. P. Human frataxin activates Fe-S cluster biosynthesis by facilitating sulfur transfer chemistry. Biochemistry 53, 49044913 (2014).

6. Parent, A. et al. Mammalian frataxin directly enhances sulfur transfer of NFS1 persulde to both ISCU and free thiols. Nat. Commun. 6, 5686 (2015).7. Kim, J. H., Frederick, R. O., Reinen, N. M., Troupis, A. T. & Markley, J. L. [2Fe-2S]-ferredoxin binds directly to cysteine desulfurase and supplies an electron for iron-sulfur cluster assembly but is displaced by the scaffold protein or bacterial frataxin. J. Am. Chem. Soc. 135, 81178120 (2013).

8. Schaedler, T. A. et al. A conserved mitochondrial ATP-binding cassette transporter exports glutathione polysulde for cytosolic metal cofactor assembly. J. Biol. Chem. 289, 2326423274 (2014).

9. Zheng, L., Cash, V. L., Flint, D. H. & Dean, D. R. Assembly of iron-sulfur clusters. Identication of an iscSUA-hscBA-fdx gene cluster from Azotobacter vinelandii. J. Biol. Chem. 273, 1326413272 (1998).

10. Ollagnier-de-Choudens, S., Sanakis, Y. & Fontecave, M. SufA/IscA: reactivity studies of a class of scaffold proteins involved in [Fe-S] cluster assembly. J. Biol. Inorg. Chem. 9, 828838 (2004).

11. Loiseau, L. et al. ErpA, an iron sulfur (Fe S) protein of the A-type essential for respiratory metabolism in Escherichia coli. Proc. Natl Acad. Sci. USA 104, 1362613631 (2007).

12. Angelini, S. et al. NfuA, a new factor required for maturing Fe/S proteins in Escherichia coli under oxidative stress and iron starvation conditions. J. Biol. Chem. 283, 1408414091 (2008).

13. Bandyopadhyay, S. et al. A proposed role for the Azotobacter vinelandii NfuA protein as an intermediate iron-sulfur cluster carrier. J. Biol. Chem. 283, 1409214099 (2008).

14. Jensen, L. T. & Culotta, V. C. Role of Saccharomyces cerevisiae ISA1 and ISA2 in iron homeostasis. Mol. Cell Biol. 20, 39183927 (2000).

15. Sheftel, A. D. et al. The human mitochondrial ISCA1, ISCA2, and IBA57 proteins are required for [4Fe-4S] protein maturation. Mol. biol. cell 23, 11571166 (2012).

16. Ding, H. & Clark, R. J. Characterization of iron binding in IscA, an ancient iron-sulphur cluster assembly protein. Biochem. J. 379, 433440 (2004).

17. Ding, H., Clark, R. J. & Ding, B. IscA mediates iron delivery for assembly of iron-sulfur clusters in IscU under the limited accessible free iron conditions.J. Biol. Chem. 279, 3749937504 (2004).18. Gupta, V. et al. Native Escherichia coli SufA, coexpressed with SufBCDSE, puries as a [2Fe-2S] protein and acts as an Fe-S transporter to Fe-S target enzymes. J. Am. Chem. Soc. 131, 61496153 (2009).

19. Krebs, C. et al. IscA, an alternate scaffold for Fe-S cluster biosynthesis. Biochemistry 40, 1406914080 (2001).

20. Zeng, J. et al. The IscA from Acidithiobacillus ferrooxidans is an iron-sulfur protein which assemble the [Fe4S4] cluster with intracellular iron and sulfur. Arch. Biochem. Biophys. 463, 237244 (2007).

21. Lu, J., Yang, J., Tan, G. & Ding, H. Complementary roles of SufA and IscA in the biogenesis of iron-sulfur clusters in Escherichia coli. Biochem. J. 409, 535543 (2008).

22. Tan, G., Lu, J., Bitoun, J. P., Huang, H. & Ding, H. IscA/SufA paralogues are required for the [4Fe-4S] cluster assembly in enzymes of multiple physiological pathways in Escherichia coli under aerobic growth conditions. Biochem. J. 420, 463472 (2009).

23. Vinella, D., Brochier-Armanet, C., Loiseau, L., Talla, E. & Barras, F. Iron-sulfur (Fe/S) protein biogenesis: phylogenomic and genetic studies of A-type carriers. PLoS Genet. 5, e1000497 (2009).

24. Kaut, A., Lange, H., Diekert, K., Kispal, G. & Lill, R. Isa1p is a component of the mitochondrial machinery for maturation of cellular iron-sulfur proteins and

size of a minimum of three biological replicates was estimated based on the inter-individual biological variability and the variability of injection. The samples were not randomized since the contralateral side injected with the scrambled shRNA served as the control for each experiment. For molecular analysis, tissue samples were directly snap-frozen in liquid nitrogen and kept at 80 C. For histology,

skeletal muscle were embedded in OCT Tissue Tek (Sakura Finetechnical, Torrance, CA, USA) and snap-frozen in isopentane chilled in liquid nitrogen. For electron microscopy, tissue samples were xed in 2.5% paraformaldehyde and 2.5% glutaraldehyde in cacodylate buffer (0.1 M, pH 7.2), post-xed in 0.1 M cacodylate buffer supplemented with 1% osmium tetroxide for 1 h at 4 C, dehydrated and embedded in Epon.

Western blot and enzymatic measurements. Tissues were homogenized using the ULTRA-TURRAX T25 basic (IKA-WERKE) in SDS-buffer (10 ml per 1 mg tissue; 280 mM Tris, 43% glycerol, 10% SDS, pH 6.8). Primary neurons were processed using TGEK buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA,10% glycerol, 50 mM KCl) supplemented with 0.2% Triton X100 and Complete inhibitor cocktail (Roche). After homogenization, samples were centrifuged at 12,000g, 10 min, 4 C and the protein concentration of supernatants was determined using Bradford assay (Bio-Rad). Extracts were further processed using the SDS buffer to resuspend membrane proteins. Unless otherwise indicated, protein extracts were further denaturated in loading buffer and loaded on Tris-glycine polyacrylamide gel58. Antibodies were incubated overnight at 4 C in PBS-Tween or TBS-Tween (0.05%) using antibodies listed in Supplementary Table 3. Western blot was revealed using Amersham Hyperlm ECL or Amersham Imager 600 (GE Healthcare). All original western blots can be found in Supplementary Fig. 4. SDH activities were measured spectrophotometrically by following the reduction of dichlorophenol indophenol (DCPIP) at 600 nm (ref. 59).

Quantitative real-time PCR. Total RNA was obtained from mouse tissues using Tri Reagent (Molecular Research Centre, Inc.) according to the manufacturers protocol. Reverse transcription was carried out using Superscript II kit (Invitrogen). Quantitative RTPCR was performed on a LightCycler 480 apparatus (Roche) using specic oligonucleotides, listed in Supplementary Table 4.

Hprt expression was used as control.

Histology and electron microscopy. For histological analysis, 10 mm cryostat sections were stained with haematoxylin and eosin, or for the activity of SDH59. An incubation time of 10 min for SDH at 37 C was required for the appearance of the nitroblue-diformozan precipitate before mounting. For electron microscopy, ultrathin sections (70 nm) of skeletal muscle were contrasted with uranyl acetate and lead citrate and examined with a Morgagni 268D electron microscope59.

AAV production and purication. AAV2/1 and AAV2/9 vectors were generated by a triple transfection of AAV-293 cell line with the different pAAV constructs, pHelper and pXR1 or pAAV2_9 PENN P0008 (PENN Vector Core) for AAV serotype 1 or AAV serotype 9, respectively. Cell lysates were subjected to three freeze/thaw cycles, then treated with 50 U ml 1 of Benzonase (Sigma) for 30 min at 37 C, and claried by centrifugation. Viral vectors were puried by Iodixanol gradient ultracentrifugation followed by dialysis and concentration against Dulbeccos Phosphate Buffered Saline using centrifugal lters (Amicon Ultra-15 Centrifugal Filter Devices 30 K, Millipore, Bedford, MA, USA). Physical particles were quantied by real-time PCR using a plasmid standard pAAV-eGFP, and titers were expressed as viral genomes per milliliter (vg ml 1). All rAAVs were obtained with titers over 1012 vg ml 1.

Seahorse analysis. Seahorse analyses were carried on a XFe96 extracellular ux analyser using the XF Cell Mito Stress Test Kit (Seahorse bioscience), following the manufacturers recommendations and protocol. Experiments were performed in XF Base medium supplemented with 10 mM glucose, 2 mM pyruvate and 2 mM glutamine, pH 7.4, using 2 mM oligomycin, 0.5 mM FCCP and 0.5 mM of rotenone and antimycin. Data were analysed using the XF Cell Mito Stress Test Report Generator V2 (http://www.seahorsebio.com/support/software/stress-test-generator.php

Web End =http://www.seahorsebio.com/support/software/stress-test-gen http://www.seahorsebio.com/support/software/stress-test-generator.php

Web End =erator.php ) and by calculating the basal respiration and spare respiratory capacity for each well.

Antibody production and purication. Polyclonal rabbit antibodies against mouse ISCA2 or IBA57 protein were produced using the GST-ISCA2 recombinant protein or the CLGDLQDYHKYRYQQG peptide, respectively. The peptide was coupled to activated KLH using Imject Maleimide Activated Carrier Protein Spin Kits (Thermo Scientic) in Tris-EDTA buffer (50 mM Tris, pH 8.5, 5 mM EDTA). Rabbits were injected with a 1:1 solution of the respective antigen and Freunds Complet adjuvant. Sera were puried using SulfoLink Coupling Resin (Thermo Scientic), according to the manufacturers protocol, using His-ISCA2 or the respective epitope-peptide. Antibodies were eluted in acidic conditions (0.1 M glycine, pH 2.8). Antibody-containing fractions were then dialysed against PBS and stored in glycerol 29% with NaN3 at 20 C.

NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15124

requires conserved cysteine residues for function. J. Biol. Chem. 275, 1595515961 (2000).25. Pelzer, W. et al. Mitochondrial Isa2p plays a crucial role in the maturation of cellular iron-sulfur proteins. FEBS Lett. 476, 134139 (2000).

26. Gelling, C., Dawes, I. W., Richhardt, N., Lill, R. & Muhlenhoff, U. Mitochondrial Iba57p is required for Fe/S cluster formation on aconitase and activation of radical SAM enzymes. Mol. Cell Biol. 28, 18511861 (2008).

27. Muhlenhoff, U., Richter, N., Pines, O., Pierik, A. J. & Lill, R. Specialized function of yeast Isa1 and Isa2 proteins in the maturation of mitochondrial [4Fe-4S] proteins. J. Biol. Chem. 286, 4120541216 (2011).

28. Banci, L. et al. [2Fe-2S] cluster transfer in iron-sulfur protein biogenesis. Proc. Natl Acad. Sci. USA 111, 62036208 (2014).

29. Brancaccio, D. et al. Formation of [4Fe-4S] clusters in the mitochondrial iron-sulfur cluster assembly machinery. J. Am. Chem. Soc. 136, 1624016250 (2014).

30. Fleischhacker, A. S. et al. Characterization of the [2Fe-2S] cluster of Escherichia coli transcription factor IscR. Biochemistry 51, 44534462 (2012).

31. Yang, J. et al. The iron-sulfur cluster of pyruvate formate-lyase activating enzyme in whole cells: cluster interconversion and a valence-localized [4Fe-4S]2 state. Biochemistry 48, 92349241 (2009).

32. Cosper, M. M., Jameson, G. N. L., Eidsness, M. K., Huynh, B. H. & Jonhson, M. K. Recombinant Escherichia coli biotin synthase is a [2Fe2S]2 protein in whole cells. FEBS Lett. 529, 332336 (2002).

33. Hristova, D., Wu, C.-H., Jiang, W., Krebs, C. & Stubbe, J. Importance of the maintenance pathway in the regulation of the activity of Escherichia coli ribonucleotide reductase. Biochemistry 47, 39893999 (2008).

34. Matzanke, B. F., Mller, G. I., Bill, E. & Trautwein, A. X. Iron metabolism of Escherichia coli studied by Mssbauer spectroscopy and biochemical methods. Eur. J. Biochem. 183, 371379 (1989).

35. Popescu, C. V., Bates, D. M., Beinert, H., Munck, E. & Kiley, P. J. Mossbauer spectroscopy as a tool for the study of activation/inactivation of the transcription regulator FNR in whole cells of Escherichia coli. Proc. Natl Acad. Sci. USA 95, 1343113435 (1998).

36. Benda, R. et al. Iron-sulfur clusters of biotin synthase in vivo: a Mossbauer study. Biochemistry 41, 1500015006 (2002).

37. Seemann, M. et al. Isoprenoid biosynthesis via the MEP pathway: in vivo Mossbauer spectroscopy identies a [4Fe-4S]2 center with unusual coordination

sphere in the LytB protein. J. Am. Chem. Soc. 131, 1318413185 (2009).38. Ta, D. T. & Vickery, L. E. Cloning, sequencing, and overexpression of a [2Fe-2S] ferredoxin gene from Escherichia coli. J. Biol. Chem. 267, 1112011125 (1992).

39. Guillon, B. et al. Frataxin deciency causes upregulation of mitochondrial Lon and ClpP proteases and severe loss of mitochondrial Fe-S proteins. Febs J. 276, 10361047 (2009).

40. Wu, G. et al. Iron-sulfur cluster biosynthesis: characterization of Schizosaccharomyces pombe Isa1. J. Biol. Inorg. Chem. 7, 526532 (2002).

41. Tong, W. H., Jameson, G. N., Huynh, B. H. & Rouault, T. A. Subcellular compartmentalization of human Nfu, an iron-sulfur cluster scaffold protein, and its ability to assemble a [4Fe-4S] cluster. Proc. Natl Acad. Sci. USA 100, 97629767 (2003).

42. Cameron, J. M. et al. Mutations in iron-sulfur cluster scaffold genes NFU1 and BOLA3 cause a fatal deciency of multiple respiratory chain and 2-oxoacid dehydrogenase enzymes. Am. J. Hum. Genet. 89, 486495 (2011).

43. Navarro-Sastre, A. et al. A fatal mitochondrial disease is associated with defective NFU1 function in the maturation of a subset of mitochondrial Fe-S proteins. Am. J. Hum. Genet. 89, 656667 (2011).

44. Ajit Bolar, N. et al. Mutation of the iron-sulfur cluster assembly gene IBA57 causes severe myopathy and encephalopathy. Hum. Mol. Genet. 22, 25902602 (2013).

45. Debray, F. G. et al. Mutation of the iron-sulfur cluster assembly gene IBA57 causes fatal infantile leukodystrophy. J. Inherit. Metab. Dis. 38, 11471153 (2015).

46. Al-Hassnan, Z. N. et al. ISCA2 mutation causes infantile neurodegenerative mitochondrial disorder. J. Med. Genet. 52, 186194 (2015).

47. Schmucker, S. et al. Mammalian frataxin: an essential function for cellular viability through an interaction with a preformed ISCU/NFS1/ISD11 iron-sulfur assembly complex. PLoS ONE 6, e16199 (2011).

48. Garboczi, D. N., Hullihen, J. H. & Pedersen, P. L. Mitochondrial ATP synthase. Overexpression in Escherichia coli of a rat liver beta subunit peptide and its interaction with adenine nucleotides. J. Biol. Chem. 263, 1569415698 (1988).

49. Gour, E. et al. Reversible (de)protonation-induced valence inversion in mixed-valent diiron(II,III) complexes. Inorg. Chem. 50, 64086410 (2011).

50. Jhurry, N. D., Chakrabarti, M., McCormick, S. P., Holmes-Hampton, G. P. & Lindahl, P. A. Biophysical investigation of the ironome of human jurkat cells and mitochondria. Biochemistry 51, 52765284 (2012).

51. Miao, R. et al. EPR and Mssbauer spectroscopy of intact mitochondria isolated from Yah1p-depleted Saccharomyces cerevisiae. Biochemistry 47, 98889899 (2008).

52. Park, J., McCormick, S. P., Cockrell, A. L., Chakrabarti, M. & Lindahl, P. A. High-spin ferric ions in Saccharomyces cerevisiae vacuoles are reduced to the ferrous state during adenine-precursor detoxication. Biochemistry 53, 39403951 (2014).

53. Mapolelo, D. T., Zhang, B., Naik, S. G., Huynh, B. H. & Johnson, M. K. Spectroscopic and functional characterization of iron-sulfur cluster-bound forms of azotobacter vinelandii(Nif)IscA. Biochemistry 51, 80718084 (2012).

54. Gardner, P. R. & Fridovich, I. Inactivation-reactivation of aconitase in Escherichia coli. A sensitive measure of superoxide radical. J. Biol. Chem. 267, 87578763 (1992).

55. Noseda, R. et al. DDIT4/REDD1/RTP801 is a novel negative regulator of Schwann cell myelination. J. Neurosci. 33, 1529515305 (2013).

56. Schmucker, S., Argentini, M., Carelle-Calmels, N., Martelli, A. & Puccio, H. The in vivo mitochondrial two-step maturation of human frataxin. Hum. Mol. Genet. 17, 35213531 (2008).

57. Florens, L. & Washburn, M. P. Proteomic analysis by multidimensional protein identication technology. Methods Mol. Biol. 328, 159175 (2006).

58. Martelli, A. et al. Frataxin is essential for extramitochondrial Fe-S cluster proteins in mammalian tissues. Hum. Mol. Genet. 16, 26512658 (2007).

59. Puccio, H. et al. Mouse models for Friedreich ataxia exhibit cardiomyopathy, sensory nerve defect and Fe-S enzyme deciency followed by intramitochondrial iron deposits. Nat. Genet. 27, 181186 (2001).

Acknowledgements

The authors thank Jean-Marc Latour for helpful discussions. We thank Franoise Piguet and Olivier Griso for contributions to primary neuronal cultures, Laurence Reutenauer for technical help, Belinda Cowling and members of Laportes team for the set up of AAV intramuscular injection and helpful discussions. We also acknowledge the Ingestem platform and all the IGBMC platforms for their contributions and help. This work was supported by grants from the Friedreich Ataxia Research Alliance to A.M.; The Agence Nationale pour la Recherche with the ANR FRATISCA (ANR-14-CE09-0026) to S.O.d.C., G.B. and H.P.; the ANR FeStreS (ANR-11-BSV3-022-02) to S.O.d.C.; the Labex ARCANE (ANR-11-LABX-0003-01) to S.O.d.C. and G.B.; the Labex (ANR-10-LABX-0030-INRT) and Idex (ANR-10-IDEX-0002-02), as well as the European Community under the European Research Council (206634/ISCATAXIA) to H.P.

Author contributions

L.K.B., S.O.d.C., I.S., M.-A.H., M.C., S.S., A.E., A.W., P.K., N.M. and A.M. performed experiments, M.F., S.O.d.C., G.B., H.P. and A.M. designed experiments, L.K.B., S.O.d.C., M.F., I.S., M.C., G.B., H.P. and A.M. analysed data, L.K.B., S.O.d.C., G.B., H.P. and A.M. wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Beilschmidt, L. K. et al. ISCA1 is essential for mitochondrial Fe4S4 biogenesis in vivo. Nat. Commun. 8, 15124 doi: 10.1038/ncomms15124 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

12 NATURE COMMUNICATIONS | 8:15124 | DOI: 10.1038/ncomms15124 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications