Supporting information

Banci et al. 10.1073/pnas.1302378110 SI Materials and Methods Electrospray Ionization-MS, Circular Dichroism, and EPR. Electrospray Cloning and Protein Production of the FMN-Binding Domain of NADPH- ionization (ESI)-MS spectra were recorded by direct introduction dependent diflavin oxidoreductase 1. The FMN-binding domain of of the samples at 5 μL/min flow rate in an LTQ-Orbitrap high- NADPH-dependent diflavin oxidoreductase 1 (Ndor1), encoding resolution mass spectrometer (Thermo), equipped with a con- residues 1–161 or 1–174, was PCR amplified from a commercially ventional ESI source. For acquisition, Xcalibur 2.0. software acquired plasmid (IMAGE Full Length cDNA clone IRAT- (Thermo) was used and monoisotopic and average deconvo- p970E05127D) encoding human Ndor1 and inserted in Gateway luted masses were obtained by using the integrated Xtract tool. pEntr-TEV-d-Topo vector. The protein constructs were expressed For spectra acquisition a nominal resolution (at m/z 400) of with an N-terminal protein G β1 domain (GB1)-His-tag in BL21 100,000 was used. – (DE3)pLysS cells. The cultures were grown at 37 °C overnight in Circular dichroism spectra (290 340 nm) were run at 298 K with LB or minimal medium containing 100 μg/mL ampicillin, 34 μg/mL a JASCO Jasco-810 spectropolarimeter, using a 1-cm cell and chloramphenicol, and 100 μMriboflavin and grown further at 37 °C bandwidth of 1 nm. Spectra were accumulated 10 times and the values were corrected for buffer contributions. The CDPro soft- until OD600 reached 0.6–0.8. The culture was transferred to 17 °C and protein expression was induced by addition of isopropyl-β-D- ware package was used to determine the secondary structure thiogalactopyranosid (IPTG) to a final concentration of 0.5 mM content, considering reference dataset 1 (http://lamar.colostate. and the incubation was continued for 16 h. The cell pellets were edu/~sreeram/CDPro/main.html) (9, 10). The circular dichroism resuspended in lysis buffer (20 mM Tris·HCl, 500 mM NaCl, 5 mM spectrum of [2Fe-2S]-cytokine-induced apoptosis inhibitor 1 imidazole, pH 8) and subsequently disrupted on ice by sonication. (CIAPIN1)-single shows a negative band at 205 nm characteristic The resulting cell lysate was centrifuged and the supernatant was of a polyproline II conformation, which is typically found in un- folded proteins (11–14), and a shoulder at 222 nm characteristic of applied to a nickel-charged HiTrap chelating HP column. The α FMN-binding domain was eluted with lysis buffer containing 500 an -helical conformation (estimated at 15%). mM imidazole and 3 mM DTT. The GB1-His-tag was cleaved from EPR spectra of reduced [2Fe-2S]-anamorsin and [2Fe-2S]- CIAPIN1-single in 50 mM Tris·HCl (pH 8.0), 500 mM NaCl or 50 the recombinant protein by incubation with tobacco etch virus mM phosphate buffer (pH 7.0), and 10% glycerol were performed (TEV) protease (5 μL TEV/1 mg protein) and removed by reverse on a Bruker Elexsys E500 spectrometer equipped with a X-band Ni(II) affinity chromatography. Size exclusion chromatography microwave bridge (microwave frequency, 9.45 GHz) and a unit was performed as the final purification step, using a HiLoad 16/60 for temperature control (ER 4131 VT). EPR parameters were: Superdex 75-pg column and degassed 50 mM Tris·HCl, 500 mM sample temperature, 45 K; microwave frequency, 9.45 GHz; mi- NaCl, and 2 mM DTT, pH 8, as a running buffer. crowave power, 5 mW; modulation frequency, 9,387,691 GHz; X-Ray Data Analysis. The crystal diffracted to 1.8 Å resolution. The modulation amplitude, 2,500 G; and time constant, 167 ms. To reduce the cluster, 1 mM dithionite was added under anaerobic crystal belongs to spacegroup P43212 with two molecules in the asymmetric unit, a solvent content of about 50%, and a mosaicity conditions and the sample was immediately frozen. of 0.3°–0.4°. The data were processed and scaled using the XDS 15N NMR Relaxation Data. 15Nlongitudinal(R ) and transverse (R ) package (1). The structure was solved through the molecular re- 1 2 relaxation rates and steady-state heteronuclear NOE measure- placement technique, using the Protein Data Bank (PDB) entry ments were performed at 11.7 T (500 MHz) for FMN-Ndor1 and at 1B1C as the template, which corresponds to the FMN-binding fl 14.1 T (600 MHz) for [2Fe-2S]-CIAPIN1-single, using the standard domain of the di avin reductase NADPH-cytochrome P450 re- pulse sequences on 15N-labeled samples. The overall rotational ductase; water molecules and FMN were omitted from the starting correlation times of FMN-Ndor1 were estimated from the R2/R1 model. The selected template model was the PDB entry having the ratio, using the program QUADRATIC_DIFFUSION, excluding highest sequence homology with that of the FMN-binding domain relaxation data of NHs having an exchange contribution to the R2 of Ndor1. The correct orientation and translation of the molecule value or exhibiting large-amplitude internal motions. The second- within the crystallographic unit cell were determined with standard ary structure elements of FMN-Ndor1, as obtained from chemical Patterson search techniques (2, 3) as implemented in the program shift analysis through the TALOS+ program (15), are 5–12 (strand fi MOLREP (4). The re nement was carried out using REFMAC5 β1), 16–30 (helix α1), 34–38 (strand β2), 45–49 (helix α2), 53–59 (5), using local NCS and default TLS restraints. In between the (strand β3), 71–77 (helix α3), 90–97 (strand β4), 107–118 (helix α4), fi re nement cycles the structure was subjected to manual rebuilding 123–125, 126–129 (strand β5andβ5′), and 139–154 (helix α5). The by using XtalView (6). Water molecules and FMN were added heteronuclear relaxation NMR data of FMN-Ndor1 indicate an using the standard procedures within the ARP/WARP suite (7). essentially rigid protein, apart from the last 15 residues of the C The crystal structure of the FMN-binding domain of Ndor1 reveals terminus, which are unstructured and flexible, with a molecular the presence of two molecules in the asymmetric unit related by tumbling correlation time of 15.3 ± 1.3 ns at 298 K, indicating that noncrystallographic symmetry. The total surface of both molecules in solution the protein is in a monomeric state. is about 14,000 A2, whereas the buried surface accounts for 1,300 2 A , indicating that the presence of the apparent dimer is a crystal Paramagnetic NMR Experiments. 1H-detected HNCA, HNCO, packing artifact. In fact, the comparison of the thermal factor CBCACONH, and 13C-detected CON, CC-COSY, CACO, IR- values of both chains clearly shows that these values are signifi- CACO-AP, and CBCACO were modified to be tailored for very cantly and systematically higher for chain B, indicating a re- fast relaxing resonances. Experiments were acquired with (i) op- markably higher mobility. The stereochemical quality of the timized transfer delays during coherence transfers; (ii) acquisition refined model was assessed using the program Procheck (8), which and recycle delays tailored to relaxation properties of target spins; shows that residues in the most favored and additional allowed (iii) removal of time-consuming building blocks such as selective regions are, respectively, 91.6% and 8.4%. Table S1 shows the data pulses during watergate, sensitivity improvement, and echo and collection and refinement statistics. antiecho selection; (iv) use of relaxation-based filters implemented

Banci et al. www.pnas.org/cgi/content/short/1302378110 1of9 as building blocks into the standard sequences such as inversion Structure Calculations and Molecular Dynamics Simulation of [2Fe- recovery (R1-based filters) and broadband saturation recovery 2S]-CIAPIN1-Single. A structural model of [2Fe-2S]-CIAPIN1- (R2-based filters); and (v) removal of scalar coupling evolution single was obtained by performing CYANA calculations with 51 from antiphase to in phase of the observed spin (either 1Hor13C) dihedral angle restraints derived from TALOS+ software, 17 1 13 α with acquisition of the antiphase component, detected as a doublet R1-based distance restraints (10 HN, 7 C ), and 20 distance fi in negative dispersion. restraints used to de ne the [2Fe-2S] cluster binding to the 13 protein. The paramagnetic longitudinal relaxation rates of 1H A C, antiphase observed CACO experiment (CACO-SQ-AP) 13 was recorded at 176.08 MHz, using a 1,024 × 64 data point matrix and C nuclei were translated in upper and lower distance α collected over 80 (F2, C′) × 50 (F1, C ) ppm. A total of 1,024 scans limits by applying a methodology already applied on highly – were acquired per fid, using Cα-C′ inept transfer and recycle de- paramagnetic systems (16 19). Unrestrained molecular dynam- lays of 2.78 ms and 366 ms, respectively. To filter signal intensity ics (MD) simulations of [2Fe-2S]-CIAPIN1-single (using only α – according to longitudinal relaxation rates of C spins, an inversion residues 234 299 that comprise the more structured, compact segment) were performed through the Amber 12 software recovery version of the experiment was also performed. The package (20, 21), using the Amber FF03 force field library and CACO-SQ-AP sequence was preceded by a nonselective 180° 13C the structural model obtained from CYANA calculations, as pulse in the region of aliphatic signals. Inversion recovery and a starting point of the simulation. In the MD, the atomic charges recycle delays were 100 ms and 260 ms, respectively. A CBCACO +2 13 ′ and geometries parameters of the [2Fe-2S] cluster were de- experiment was also recorded in a C antiphase observed mode rived from previous density-functional theory (DFT) works (22, fi (CBCACO-AP). The conventional pulse sequence was modi ed 23) and from the crystal structure of bovine adrenodoxin (PDB fi by removing the nal in phase antiphase (IPAP) building block ID 1AYF), respectively. The system was solvated with water in α ′ used for virtual decoupling of C -C homonuclear couplings. a periodic TIP4P/2005 box, large enough to contain the system CBCACO-AP was recorded at 176.08 MHz, using a 1,024 × 128 α and 1 nm of solvent on all sides. During the productive phase of data point matrix collected over 80 (F2, C′) × 80 (F1, C ) ppm. A the simulation, the temperature and pressure were kept constant total of 1,024 scans were acquired per fid, using Cβ-Cα,Cα-C′ at 300 K and 1 bar, respectively. In the initial steps of the dy- inept transfer delays of 4.0 and 2.78 ms and a recycle delay of 300 namic calculation, water molecules were allowed to adjust with ms. A CON experiment was recorded, at 176.08 MHz, using the position restraints on the protein. Then, the temperature was routine pulse sequence. A 512 × 128 data point matrix was ac- raised gradually from 0 K to 300 K, the system was equilibrated quired over 36.1 (F2, C′) × 50 (F1, 15N) ppm. A total of 4,096 scans for 500 ps, and a 100-ns trajectory was produced. α were acquired per fid, using C -C′ refocusing and C′-N transfer Redox Chemistry. FMN-Ndor1 (200 μM) was stepwise reduced to delays of 4.5 and 8 ms, respectively. Recycle delay was 270 • ms.13C-13C COSY experiments were recorded, at 176.08 MHz, form the neutral blue semiquinone (FMNH ) or the fully reduced using a 220 × 220-ppm spectral region, using a 4,096 × 900 data (FMNH2) state by addition of 0.5 and 1 eq of dithionite, respec- point matrix, 400 scans per fid, and 412-ms recycle delays. Before tively. Dithionite solution was prepared in an anaerobic chamber Fourier transformation, the data point matrix was reduced to se- with degassed buffer. Dithionite concentration was determined by 1 measuring UV absorbance at 315 nm, using a molar absorptivity of lect suitable values of t max and t max. HR measurements of − − 2 1 1 6,900 M 1·cm 1. Protein concentrations of FMN-Ndor1, [2Fe-2S]- paramagnetic backbone NHs were obtained via inversion re- 1 15 anamorsin, and [2Fe-2S]-CIAPIN1 were measured by UV spec- covery, edited by H- N heteronuclear single quantum coherence − − troscopy at 280 nm, using molar absorptivities of 21,095 M 1·cm 1, − − − − (HSQC) experiments. Nineteen experiments were collected using 14,293 M 1·cm 1,and3,230M1·cm 1, respectively, and/or by recovery delays spanning from 250 ms to 1 ms. Longitudinal re- • 13 α Bradford protein assay. The FMNH2 or FMNH form was then laxation rates of C spins were obtained via an inversion re- mixed with [2Fe-2S]-anamorsin or [2Fe-2S]-CIAPIN1-single up to covery-CACO-SQ-AP experiment, recorded at 176.08 MHz. A molar ratios of 2 (FMNH2):1 ([2Fe-2S]-anamorsin or [2Fe-2S]- × fi • total of 1,024 80 data point matrices, 256 scans per d, were CIAPIN1-single) and 1 (FMNH ):2 ([2Fe-2S]-anamorsin or [2Fe- × collected, over a 80 56-ppm spectral region, using a relaxation 2S]-CIAPIN1-single). UV/visible spectra of FMN-Ndor1 with • delay of 1 s. Eight experiments were recorded using recovery de- FMN in its oxidized and reduced FMNH and FMNH2 states be- lays of 5 ms, 50 ms, 100 ms, 200 ms, 300 ms, 400 ms, 600 ms, and 1 s, fore and after the additions of [2Fe-2S]-anamorsin or [2Fe-2S]- respectively. To identify signals affected by paramagnetic re- CIAPIN1-single were recorded in anaerobic conditions on a Cary laxation and to obtain an estimate of diamagnetic contributions to 50 Eclipse spectrophotometer in degassed 300 mM NaCl, 1 mM R1 relaxation, another series was recorded using relaxation delay DTT, 50 mM phosphate buffer, pH 7.0. Absorbance changes at 432 of 2.5 s and recovery delays ranging from 5 ms to 2.5 s. nm were monitored for the indicated time period.

1. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66(Pt 2):125–132. 11. Shi Z, Woody RW, Kallenbach NR (2002) Is polyproline II a major backbone 2. Rossmann MG, Blow DM (1962) The detection of sub-units within the crystallographic conformation in unfolded proteins? Adv Protein Chem 62:163–240. asymmetric unit. Acta Crystallogr D Biol Crystallogr 15(1):24–31. 12. Rucker AL, Creamer TP (2002) Polyproline II helical structure in protein unfolded 3. Crowther RA (1972) The Molecular Replacement Method, ed Rossmann MG (Gordon states: Lysine peptides revisited. Protein Sci 11(4):980–985. & Breach, New York), pp 173–178. 13. Dukor RK, Keiderling TA (1991) Reassessment of the random coil conformation: 4. Vagin A, Teplyakov A (1997) MOLREP: An automated program for molecular Vibrational CD study of proline oligopeptides and related polypeptides. Biopolymers replacement. J Appl Cryst 30(6):1022–1025. 31(14):1747–1761. 5. Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular structures 14. Woody RW (1992) Circular dichroism of unordered polypeptides. Adv Biophys Chem 2:37–79. by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53(Pt 3):240–255. 15. Shen Y, Delaglio F, Cornilescu G, Bax A (2009) TALOS+: A hybrid method for predicting 6. McRee DE (1992) XtalView: A visual protein crystallographic software system for protein backbone torsion angles from NMR chemical shifts. JBiomolNMR44(4):213–223. XII/XView. J Mol Graph 10(1):44–47. 16. Bertini I, et al. (2001) Paramagnetism-based versus classical constraints: An analysis of

7. Lamzin VS, Wilson KS (1993) Automated refinement of protein models. Acta the solution structure of Ca Ln calbindin D9k. J Biomol NMR 21(2):85–98. Crystallogr D Biol Crystallogr 49(Pt 1):129–147. 17. Arnesano F, et al. (2003) A strategy for the NMR characterization of type II copper(II) 8. Laskowski RA, MacArthur MW, Moss DS, Thornton JM (1993) PROCHECK: A program proteins: The case of the copper trafficking protein CopC from Pseudomonas to check the stereochemical quality of protein structures. J Appl Cryst 26(2):283–291. Syringae. J Am Chem Soc 125(24):7200–7208. 9. Sreerama N, Woody RW (2004) Computation and analysis of protein circular dichroism 18. Im S-C, Liu G, Luchinat C, Sykes AG, Bertini I (1998) The solution structure of parsley spectra. Methods Enzymol 383:318–351. [2Fe-2S]ferredoxin. Eur J Biochem 258(2):465–477. 10. Sreerama N, Woody RW (2000) Estimation of protein secondary structure from 19. Bertini I, et al. (1996) The solution structure refinement of the paramagnetic reduced circular dichroism spectra: Comparison of CONTIN, SELCON, and CDSSTR methods HiPIP I from Ectothiorhodospira halophila by using stable isotope labeling and with an expanded reference set. Anal Biochem 287(2):252–260. nuclear relaxation. Eur J Biochem 241(2):440–452.

Banci et al. www.pnas.org/cgi/content/short/1302378110 2of9 20. Case DA, et al. (2012) AMBER 12 (University of California, San Francisco). multireference second-order Møller-Plesset perturbation theory (MRMP) approach. 21. Bertini I, Case DA, Ferella L, Giachetti A, Rosato A (2011) A Grid-enabled web portal J Phys Chem A 109(43):9867–9874. for NMR structure refinement with AMBER. Bioinformatics 27(17):2384–2390. 23. Li J, et al. (1998) Incorporating protein environments in density functional theory? A 22. Higashi M, Kato S (2005) Theoretical study on electronic and spin structures of [Fe2S2] self-consistent reaction field calculation of redox potentials of [2Fe2S] clusters in (2+,+) cluster: Reference interaction site model self-consistent field (RISM-SCF) and ferredoxin and phthalate dioxygenase reductase. J Phys Chem A 102(31):6311–6324.

1.5 A 1.0 0.5 0.0 (ppm) α -0.5 C Δ -1.0 -1.5 175 180 185 190 195 200 205 210 215 220 1.5 B 1.0

0.5

0.0 C' (ppm) C'

Δ -0.5

-1.0 175 180 185 190 195 200 205 210 215 220 1.5 C 1.0

(ppm) 0.5 )

β 0.0 -C

α -0.5 C

( -1.0 Δ -1.5 175 180 185 190 195 200 205 210 215 220 Residue number

Fig. S1. Differences (in ppm) of the observed chemical shifts of [2Fe2S]-anamorsin with respect to reference values characteristic of the random-coil con- formation of small peptides. The reference shifts were properly corrected to account for the contributions by sequential neighbors and by temperature and pH effects. (A) ΔCα,(B) ΔC′, and (C) Δ(Cα-Cβ) of the linker region of [2Fe2S]-anamorsin are shown. Residues 192–195 have positive ΔCα, Δ(Cα-Cβ), and ΔC′ values, indicating an α-helical secondary structure propensity.

Banci et al. www.pnas.org/cgi/content/short/1302378110 3of9 A B 2500 ) -1

cm 2000 -1

1500

1000 sorbance (M b a 500

molar 0 250 450 650 3000 3200 3400 3600 3800 wavelength (nm) magnec field (Gauss)

C [Fe S ]-CIAPIN1 2 2 D [Fe S ]-CIAPIN1 12302.44 2 2 100 12302.44 100 (%) 80 80 ndance (%) ndance u 60 u 60 Ab 40 40 apo-CIAPIN1 12126.62 Relave Ab Relave Relave 20 apo-CIAPIN1 20 12126.62 0 0 12100 12200 12300 12400 12100 12200 12300 12400 m/z m/z

Fig. S2. CIAPIN1 domain of anamorsin binds a [2Fe-2S] cluster as found in the full-length protein. (A) Comparison of UV-visible spectra (50 mM Tris, pH 8, 500 mM NaCl, 5% glycerol) of [2Fe-2S]-anamorsin (solid line) and [2Fe-2S]-CIAPIN1-single (dashed line) purified from Escherichia coli cells. For both spectra, the molar absorbance is relative to the overall protein concentration estimated from the absorbance at 280 nm. After purification the cluster-bound form is ∼40% for the CIAPIN1 domain and ∼65% for the full-length protein. After chemical reconstitution, the level of bound cluster reaches ∼100%, corresponding to one [2Fe-2S] cluster per protein molecule, for the full-length protein (1), and it is ∼90% in the CIAPIN1 domain. (B) EPR spectra at 45 K (50 mM Tris, pH 8, 500 mM NaCl, 10% glycerol) of [2Fe-2S]-anamorsin (solid line) and [2Fe-2S]-CIAPIN1-single (dashed line). (C and D) Electrospray ionization mass spectrometry spectrum of [2Fe-2S]-CIAPIN1-single before (C) and after (D) addition of 0.02% of formic acid. The molecular weight difference between [2Fe-2S]-CIAPIN1-single and apo-CIAPIN1-single is 175.8, corresponding to the mass of 2Fe and 2S atoms. The apo form of CIAPIN1-single (molecular weight 12,302.44) forms by acidifi- cation. The relative abundance was scaled with respect to the peak corresponding to the molecular weight of the [2Fe-2S]-CIAPIN1-single.

1. Banci L, et al. (2011) Anamorsin is a [2Fe-2S] cluster-containing substrate of the Mia40-dependent mitochondrial protein trapping machinery. Chem Biol 18(6):794–804.

Banci et al. www.pnas.org/cgi/content/short/1302378110 4of9 A 5 B 15N

) 0 -1 ol

dm -5 2

-10 deg cm ( -3 0 -15 x 1 ] [

-20

190 200 210 220 230 240 250 260 1 Wavelength (nm) H C 1.0 D 4 050.5

2 0.0 (ppm) )

H}-NOEs -0.5 0 -C 1 { N C ( 15 -1.0 -2 -1.5

-4 -2.0 220 240 260 280 300 220 240 260 280 300 Residue number Residue number

Fig. S3. [2Fe-2S]-CIAPIN1-single of anamorsin is largely unstructured. (A) Far-UV CD spectrum (10-μM protein concentration in 5 mM phosphate buffer, pH 7.0) and (B) 1H-15N HSQC spectrum (0.5-mM protein concentration in 50 mM phosphate buffer, pH 7.0, and 1 mM DTT) of [2Fe-2S]-CIAPIN1-single. (C) 15N{1H}-NOE values per residue for the [2Fe-2S]-CIAPIN1-single (shaded). The position of proline and cysteine residues is indicated, respectively, as solid and open bars, to which, only for graphical representation, a 15N{1H}-NOE value of 1 was arbitrarily assigned. (D) Differences (in ppm) of the observed 13C chemical shifts of [2Fe- 2S]-CIAPIN1-single with respect to reference values characteristic of the random-coil conformation of small peptides. The reference shifts were properly α corrected to account for the contributions by sequential neighbors and by temperature and pH effects. The bars represent the difference between the 13C β chemical shift (experimental shift minus random coil shift) and the 13C chemical shift (experimental shift minus random coil shift).

Fig. S4. Protein–protein interaction between 15N-labeled FMN-binding domain of Ndor1 and unlabeled [2Fe-2S]-anamorsin as characterized by NMR. (A) Overlay of 1H-15N HSQC spectra at 298 K acquired at 900 MHz, of oxidized, 15N-labeled FMN-binding domain (construct 1–174 aa) before (red) and after (black) the addition of 1 eq of oxidized, unlabeled [2Fe-2S]-anamorsin. (B) Overlay of 1H-15N HSQC spectra at 298 K, acquired at 900 MHz, of oxidized, 15N-labeled FMN-binding domain (construct 1–174 aa) before (black) and after addition of 1 eq of unlabeled N-terminal domain of anamorsin (red).

Banci et al. www.pnas.org/cgi/content/short/1302378110 5of9 Fig. S5. Protein–protein interaction between 13C,15N- or 15N-labeled [2Fe-2S]-anamorsin and unlabeled FMN-binding domain of Ndor1 as characterized by NMR. (A) Overlay of 1H-15N HSQC spectra at 298 K, acquired at 900 MHz, of oxidized,15N-labeled [2Fe-2S]-anamorsin before (black) and after addition of 1 eq of oxidized, unlabeled FMN-binding domain (construct 1–174 aa) (red). (B) Superposition of the CON spectra of oxidized, 13C,15N-labeled [2Fe-2S]-anamorsin in the presence of 0 (black) and 1 eq (red) of unlabeled FMN-binding domain of Ndor1. (C) Overlay of paramagnetic 1H-15N IR-HSQC-AP spectra at 298 K and 700 MHz of oxidized, 15N-labeled [2Fe-2S]-anamorsin before (black) and after (red) addition of 1 eq of oxidized, unlabeled FMN-binding domain.

Banci et al. www.pnas.org/cgi/content/short/1302378110 6of9 Fig. S6. Schematic representation of the region of [2Fe-2S]-anamorsin experiencing spectral changes upon interaction with the FMN-binding domain of Ndor1. Two main interacting areas are identified: a highly hydrophobic area (residues 188–202) and a negatively charged area (residues 204–223). Positively charged areas possibly involved in protein–protein recognition nearby the [2Fe-2S]-cluster are also shown. Positively charged residues are colored in blue, negatively charged residues in red, hydrophobic residues in green, and [2Fe-2S] cysteine ligands are shown in yellow.

Banci et al. www.pnas.org/cgi/content/short/1302378110 7of9 A 0.9 080.8

0.7

0.6

0.5 nce [a.u.] a 0.4

Absorb 0.3

0.2

0.1

0 300 350 400 450 500 550 600 650 700 750 Wavelength [nm]

B -0.01

-0.03

au] -0050.05 [

-0.07

-0.09 ∆Absorbance

-0.11

-0.13 0204060 Time [min]

Fig. S7. Electron transfer between the FMN-binding domain of Ndor1 and [2Fe-2S]-anamorsin. (A) Overlay of UV-vis spectra of the FMN-binding domain of Ndor1 and [2Fe-2S]-anamorsin in their various redox states. The FMN-binding domain in its FMN oxidized state (green) was stoichiometrically reduced with dithionite to

FMNH• (brown) and to FMNH2 (dark green). The black arrow indicates the isosbestic point between FMNH• and FMNH2 (at 432 nm). FMNH2 was then mixed with [2Fe-2S]-anamorsin (blue) in 1:0.5 molar ratio (the protein mixture is in violet). Dithionite-reduced [2Fe-2S]-anamorsin is in orange. (B) Changes of the absorbance at

432 nm upon incubation of the reduced FMNH2 domain with [2Fe-2S]-anamorsin (green line) or [2Fe-2S]-CIAPIN1 domain (black line). The FMN-binding domain was first stoichiometrically reduced with dithionite to FMNH2 and then mixed with oxidized [2Fe-2S]-anamorsin or oxidized [2Fe-2S]-CIAPIN1 domain.

Fig. S8. Redox state of the FMN moiety and the presence or not of the [2Fe-2S] cluster do not affect protein–protein recognition in the anamorsin–FMN domain protein complex. Shown is overlay of TROSY 1H-15N HSQC spectra at 308 K, acquired at 900 MHz, of the 1:1 mixture of 15N-labeled oxidized FMN domain 15 (construct 1–174 aa) and unlabeled oxidized [2Fe-2S]-anamorsin (black), of the 1:1 mixture of N-labeled FMNH2 domain of Ndor1 (construct 1–174 aa) and unlabeled oxidized [2Fe-2S]-anamorsin (blue) and of the 1:1 mixture of 15N-labeled oxidized FMN-binding domain of Ndor1 (construct 1–174 aa) and unlabeled apo-anamorsin (red). The 1H-15N HSQC maps of all final mixtures are completely superimposable, indicating the formation of the same protein adduct.

Banci et al. www.pnas.org/cgi/content/short/1302378110 8of9 Table S1. Data collection and refinement statistics of the FMN- binding domain of Ndor1 Data collection and refinement Statistics

Data collection

Space group P43212 Cell dimensions a, b, c, Å a = b = 80.446, c = 101.483 α, β, γ, ° 90.0, 90.0, 90.0 Resolution, Å 20.10–1.80 (1.85–1.80)*

Rsym or Rmerge 0.083 (0.52) I/σI 16.7 (2.1) Completeness, % 99.7 (98.7) Redundancy 10.6 (7.8) Refinement Resolution, Å 20.10–1.80 (1.85–1.80) No. reflections 31,433 (4,943)

Rwork/Rfree 0.21/0.23 No. atoms 2,680 Protein 2,469 Ligand/ion 62 Water 149 B-factors Protein, chain A/B 19.7/26.7 Ligand/ion, chain A/B 15.6/26.2 Water 29.4 rms deviations Bond lengths, Å 0.007 Bond angles, ° 1.28

*Values in parentheses are for highest-resolution shell.

Banci et al. www.pnas.org/cgi/content/short/1302378110 9of9