Characterization of poplar GrxS14 in different structural forms

Glutaredoxins (Grxs) are glutathione-dependent thiol disul-fide oxidoreductases of the thioredoxin family present in all organisms from bacteria to human (Noguera et al., 2005). Depending on their active site sequence, Grxs are essentially classified into three families: the dithiol Grxs, the monothiol Grxs and the CC type restricted to plants (Rouhier et al., 2008). Grxs play important biological functions in plants, such as oxidative stress responses, iron-sulfur (FeS) cluster assembly, and cell signaling, etc. It was found that Arabidopsis GrxS14 is a new class of signaling molecules in plants that can regulate the Ca 2+ transport activity of CAX1 (cation exchangers) by interacting with the N-terminal region of CAX1 (Cheng and Hirschi, 2003). It was suggested that Arabidopsis GrxS14 functions to protecting cells against protein oxidative damage (Cheng et al., 2006). Both Arabidopsis and poplar GrxS14 are monothiol Grxs located in the chloroplasts, which exist as an apo form and a holo form bridged by a [2Fe-2S] cluster with two external glutathione (GSH) ligands, and they can complement a yeast grx5 mutant defective in FeS cluster assembly in vivo (Bandyopadhyay et al., 2008). It was proposed that Arabidopsis and poplar GrxS14 may function as scaffold protein for the assembly of [2Fe-2S] cluster, as GrxS14 can transfer intact cluster to physiologically relevant acceptor proteins which is regulated by GSH (Bandyopad-hyay et al., 2008; Wang et al., 2012; Liu et al., 2013). Here we report the solution structure of reduced poplar GrxS14 and structure models for the non-covalent apo GrxS14 dimer and GrxS14/GSH complex, as well as the NMR characterization of holo GrxS14. The quality of the 2D 1 H-15 N HSQC spectrum of apo GrxS14 at 1 mmol/L concentration was very poor (Fig. S1), and very few signals could be observed in 3D triple-resonance NMR spectra. Dilution of the sample did not improve the quality of NMR spectra very significantly. Interestingly, much better NMR spectra were obtained with the addition of GSH (Fig. S1). Although apo GrxS14 appeared to be a monomer on the gel filtration column, analytical ultracentri-fugation analysis showed two peaks (with molecular weight about 24 kDa and 12 kDa) for apo GrxS14 without GSH, while there was only one peak at ∼12 kDa for apo GrxS14 with GSH (Fig. S2A and S2B). All these suggest that apo GrxS14 should be in a monomer-to-dimer equilibrium, and the dimerization can be inhibited by GSH. Thus, there exists another type of GrxS14 …


NMR sample preparation
The apo and holo protein were purified according to the procedure described previously (Wang et al., 2011). The NMR samples contain 0.2 mM (apo form), 0.5 mM (holo form) uniformly 15 N-and/or 13 C-labeled protein in 30 mM Tris-HCl buffer with 50 mM NaCl and 95% H 2 O/5% D 2 O at pH 7.5, along with 0.01% NaN 3 and 0.01% DSS. All NMR samples had 20 mM GSH for both apo and holo proteins, and oxygen was removed from the samples in order to keep a reducing condition for stabilizing the samples.

Analytical ultracentrifuge experiment
The Sedimentation velocity experiments were carried out with a Beckman Coulter ProteomeLab TM XL-I instrument. All AUC runs were carried out at the rotation speed of 60,000 rpm at 16 °C. The sample volume was 400 μL and the protein concentration was ~ 0.075 mM. A wavelength of 280 nm was used to record the UV absorption of the cells which scanned every minute for 5 h. The AUC data were analyzed using SEDFIT program (Schuck, 2000).

NMR spectroscopy
NMR experiments were performed at 25 o C on a Bruker Avance 500 MHz or an 800 MHz spectrometer with a cryoprobe. Proton chemical shifts were referenced to internal DSS. 15 N and 13 C chemical shifts were referenced indirectly to DSS (Markley et al., 1998). Three-dimensional 15 N-edited NOESY-HSQC and 13 C-edited aliphatic and aromatic NOESY-HSQC spectra were collected with mixing times of 120 and 90 ms for the apo protein, respectively (Ferentz and Wagner, 2000). All NMR spectra were processed using NMRPipe (Delaglio et al., 1995) and analyzed with NMRView (Johnson and Blevins, 1994).

Structure calculations
The distance restraints were derived from NOESY series spectra and dihedral angle (φ, ψ) restraints were derived from chemical shifts using TALOS (Cornilescu et al., 1999). χ1 angles are determined by comparing cross peak volumes of intra-residual HN-HB and HA-HB NOEs in NOESY spectra with short mixing time.
Hydrogen bond restraints were added based on the secondary structure prediction and NOE restraints. The CYANA gave a bad initial structure using restraints from the CANDID module (Herrmann et al., 2002). Therefore we used a model as the initial structure which obtained from homology modeling based on the structure of E. coli Grx4 (PDB code 1YKA). Then the initial structure was used as filter model for SANE (Duggan et al., 2001) to obtain the assignments and generate new distance restraints.
A new round of CYANA (Guntert et al., 1997) calculation was then run using the new distance restraints. Until the distance restraint violations were smaller than 0.3 Å, 100 structures were selected as the initial structures for refinement using AMBER9 (Pearlman et al., 1995). Finally the 20 lowest energy structures were selected for final analysis. The quality of the final structures was analyzed using MOLMOL (Koradi et al., 1996), andPROCHECK_NMR (Laskowski et al., 1996). The mean structure was generated using MOLMOL and was energy minimized in AMBER9.
The solution structure of apo GrxS14 has been submitted to PDB with the code 2LKU.

Titration of apo GrxS14 with GSH and K d determination
A concentration of 0.5 M stock solution of GSH was prepared (pH 7.5). The concentration of apo GrxS14 protein was 0.1 mM, and the NMR sample contained 10 mM DTT in order to keep the reducing condition during the titration.
where δ is the observed change in chemical shift, δ 0 is the total change in chemical shift at saturation condition, [L] 0 is the total molar concentrations of the apo GrxS14 protein and [P] 0 is the total molar concentrations of GSH. Only the residues with significant chemical shift changes were involved in fitting to equation (2) given by Morton et al (Morton et al., 1996) from the titration spectra.

HADDDOCK protocol
The docking was performed with HADDOCK 2.0 (Dominguez et al., 2003) software using NMR data as restraint files. The docking model was calculated based on chemical shift changes between different 2D 1 H-15 N HSQC spectra. First, the combined chemical shift differences were calculated using equation (1) Then, the average solvent accessible area for each of those residues was calculated in MOLMOL. Only the residues exhibiting over 50% solvent accessibility were used as active residues. The passive residues were defined as the residues closing to the active residues and showing large (>50%) solvent accessibility. Then AIR restraint files were generated in HADDOCK homepage (http://www.nmr.chem.uu.nl/haddock/) for docking calculation. A total of 1000 rigid-body docking structures were generated with HADDOCK software. The 200 best structures were used for the semi-flexible simulated annealing and refinement in water. Finally, the reasonable docking structures were extracted through cluster analysis on the final refined structures.

Structure models of apo GrxS14 dimer
Comparison of 2D 1 H-15 N HSQC spectra of apo GrxS14 without GSH for samples at 0.1 and 0.4 mM, it is found that shifted NH peaks are mainly from residues F35, Q37, K66, W71, G86, D88, I89, V91, E92 and S96 (Fig. S4), which should be involved in the dimerization. For calculating structure models of apo GrxS14 dimer using HADDOCK 2.0, residues F35, Q37, K66, W71, G86, D88, I89, V91, E92 and S96 were defined as active, and residues C33, G34, T38, Q41, Q63, E67, S70, P72, T73, F83, K95, G97 and E98 were defined as passive. The semi-flexible regions were defined by considering all residues within 5Å of another molecule in the top 200 initial rigid-body docking structures. The five pairs of segments C33-Q37, K66-E67, S70-T73, G86-E92 and S96-E98 located at the interface were defined as NCS restraints for keeping the dimer symmetric. Based on the pairwise backbone root-mean-square deviation value, ten best docking structures were chosen from the lowest intermolecular energy cluster of the final round structures for analysis. The intermolecular energy of the lowest cluster was -271.40 ± 66.44 kcal/mol.

Structure models of apo GrxS14/GSH complex
Comparing 2D 1 H-15 N HSQC spectra of apo GrxS14 without GSH and with 230-fold GSH, it was found that residues with significant combined NH chemical shift changes (> 0.05 ppm) are distributed on all secondary structure regions.
The residues with ratio value 3 denote residues not observed in the HSQC spectrum of holo GrxS14. (C) 13 C α chemical shift differences between holo and apo GrxS14 vs residue number.