Disordered form of the scaffold protein IscU is the substrate for iron-sulfur cluster assembly on cysteine desulfurase

Two Conformational States of Apo-IscU.

NMR spectra of apo-IscU (Fig. S1A) display two sets of signals reporting on two conformational states, a structured state and a dynamically disordered state. For a given residue, the signals from the two states can be linked on the basis of an exchange NMR spectrum, which show that the lifetimes of each state are on the order of 1 s (7). Our extensive assignments of the and NMR signals from the S and D states of apo-IscU have enabled us to determine the secondary structural features of the two states (Fig. S1B). The D state exhibits little evidence of secondary structure, whereas the secondary structural elements of the S state correspond to those found in an NMR structure of the Zn²⁺ complex of IscU (11) and in the X-ray structure of the IscU:IscS complex (8). The assigned chemical shifts for the S (17837) and D (17836) forms of apo-IscU have been deposited at the BioMagResBank under the accession codes provided above.

Single Amino Acid Substitutions in IscU Alter the Equilibrium Between the S and D States.

The D39A substitution (E. coli numbering system; Fig. 1A) of IscU is known to increase the stability of holo-IscU (12–14). In fact, the protein used in determining the X-ray structure of holo-IscU incorporated this stabilizing substitution (15). Our laboratory showed that E. coli apo-IscU(D39A) exists in solution primarily in the S state and adopts the same fold as the structured component of WT apo-IscU (7). This finding prompted us to search for other conserved residues in E. coli IscU (Fig. 1A) whose substitution perturbs the equilibrium between the S and D states. We report here three additional single-site substitutions that strongly favor the S state—N90A, S107A, and E111A (Fig. 1B)—and two that strongly favor the D state—K89A and N90D (Fig. 1C). Unlike D39A, these substitutions do not remove a potential cluster ligand. As expected, the substitutions that favor the S state increase the temperature at which the variants unfold (Fig. 1D). The structural differences among the variants were confirmed by circular dichroism (CD) spectroscopy (Fig. S2).

IscS Binds Preferentially to the D State of apo-IscU.

Given that apo-IscU exists in equilibrium between two states, S and D, it was of interest to determine whether one of these states interacts preferentially with IscS. We used three approaches to investigate this question: (i) We measured the effect of adding a substoichiometric quantity of IscS on the apo-IscU S → D and D → S transition rates as determined by two-dimensional exchange spectroscopy, (ii) we investigated the effect of added IscS on deuterium exchange at protected sites in the S state of apo-IscU, and (iii) we determined the conformational state of apo-IscU in the complex with IscS.

The S → D and D → S rates were determined by two-dimensional exchange spectroscopy by measuring the intensity of the exchange cross peak as a function of the exchange mixing time at short values of this delay before relaxation effects become dominant (16). We found that the addition of IscS increased the S → D rate and decreased the D → S rate for apo-IscU (Fig. 2A).

Hydrogen exchange of protected amide groups in proteins can be measured by transferring ¹⁵N-labeled protein from H₂O into D₂O and following the decrease in intensity of two-dimensional (2D) NMR ¹H-¹⁵N cross peaks as a function of time. The band-selective optimized flip-angle short transient NMR approach (17) enabled rapid recording of 2D spectra immediately after the solvent transfer. We found that the exchange rates of signals from WT apo-IscU increased in the presence of a small amount of added IscS; however, because protection factors for apo-IscU alone were low, the effects were difficult to quantify. By using N90A, a variant that exists primarily in the S state, the effects could be seen more clearly. In the absence of IscS, the exchange half-times for several residues were on the order of 1,000–2,000 s; however, in the presence of 0.11 equivalent of IscS, exchange was complete by the time (approximately 250 s) the first spectrum was collected (Fig. S3).

When we added unlabeled E. coli IscS to [U-¹⁵N]-apo-IscU, the ¹H-¹⁵N heteronuclear single-quantum correlation (HSQC) spectrum changed to one indicative of the disordered state (Fig. 2B). By repeating this experiment with variants with differing stabilities of the D and S states, we found that higher concentrations of IscS were required to convert apo-IscU variants favoring the S state than those favoring the D state (Fig. S4), but with sufficient added IscS, conversion to the D state became nearly complete (Fig. S5).

Taken together, these results indicate that IscS binds preferentially to the D state of apo-IscU. Mass action serves to convert the remaining S state to a D-like state in the complex. The effect is reversible as shown by an experiment in which NMR spectra were taken of the complex formed between ¹⁵N-labeled apo-IscU and IscS before and after the addition of unlabeled apo-IscU; the spectrum of the initial complex showed signals only from disordered residues, whereas addition of unlabeled apo-IscU partially displaced ¹⁵N-apo-IscU from the complex, resulting in a spectrum showing signals from both the S and D states (Fig. S6).

Conformation of apo-IscU when Complexed with IscS.

We used 2D and 3D NMR methods to assign most of the observable backbone ¹H-¹⁵N signals from WT apo-IscU in the complex with IscS (deposited at BioMagResBank under accession code 17844). The ¹H-¹⁵N chemical shifts of apo-IscU in the IscU:IscS complex closely matched those of the D state of free apo-IscU but not those of the S state (Fig. 2C). These results identified which IscU residues are disordered in the complex in solution (residues colored green in the X-ray structure of the complex, Fig. 2D).

We used analytical gel filtration (Fig. S7) to estimate the mass of the complex in solution. The size determined (102 kDa) is in agreement with the (IscU)₂∶(IscS)₂ stoichiometry of the complex and the X-ray structure, which shows one IscU bound to each IscS subunit (8). Given the size of the complex, backbone signals from IscU residues in the binding site with IscS are expected to be exceedingly broad and not visible in the ¹H-¹⁵N spectrum. Signals from residues (K59-G62, V73, and A117) in the D state of free apo-IscU broadened and disappeared upon addition of IscS (these residues are indicated by a red “X” in Fig 2C); they correspond to potential candidate residues in the interaction site with IscS (IscU residues colored red in the X-ray structure of the complex, Fig. 2D). IscU residues colored black in the X-ray structure of the complex (Fig. 2D) correspond to ones whose NMR signals were not observed in either free IscU or the complex.

Influence of Metal Binding on the Order-Disorder Processes.

It has been proposed that Zn²⁺ in the growth medium is required for the production of properly folded apo-IscU (9). To examine this claim, we produced WT [U-¹⁵N]-IscU in two M9 media, one supplemented with 8.4 μM ZnSO₄ and the other with no added Zn²⁺; in both cases, we produced apo-IscU by EDTA treatment to remove metal ions followed by gel filtration to remove EDTA. The 2D ¹⁵N-HSQC spectra of apo-IscU produced from the two media showed the protein to be present as an identical mixture of S and D states (Fig. S8 A and B). We further observed that IscU adopts a single structured state upon the addition of either Zn²⁺ (Fig. S8C) or Fe²⁺ (Fig. S8D), with zinc forming a complex with higher thermal stability than iron (Fig. S8E). Thus, it is clear that the presence of metal ions can mask the interesting order-disorder equilibrium in apo-IscU by stabilizing an S-like state. We also examined whether IscU has similar properties under more physiological conditions of ionic composition or macromolecular crowding. The 2D ¹⁵N-HSQC spectrum of [U-¹⁵N]apo-IscU exhibited peaks from the S and D states under the following three conditions: (i) 150 mM KCl and 10 mM MgCl₂ (Fig. S8F), (ii) 100 mg/mL Ficoll® PM 70 (Fig. S8G), and (iii) 100 mg/mL BSA (Fig. S8H). Furthermore, the addition of IscS to each of these solutions led to the appearance of disordered IscU peaks characteristic for the IscU:IscS complex (Fig. S8 I–M).

In Vitro Cluster Assembly.

To examine the functional relevance of these findings, we carried out in vitro cluster assembly reactions (18) with a panel of IscU variants. We added each IscU variant anaerobically to the cluster assembly mixture (fivefold excess ammonium ferrous sulfate, fivefold excess l-cysteine, and 1/50-fold IscS) and monitored Fe-S cluster assembly by following the absorbance at 456 nm (Fig. 3). Under these conditions, IscU assembles a mixture of [2Fe-2S] and [4Fe-4S] clusters (18). We found that WT IscU assembled clusters most efficiently. The fully structured variants (N90A, S107A, and E111A) exhibited a biphasic reaction with an initial slow step and assembled clusters at an overall rate about seven times slower than WT IscU. By contrast, the partially disordered variants (K89A and N90D) exhibited no slow initial phase and assembled clusters about three times more slowly than WT IscU. The duration of the observed initial slow step in cluster assembly appears to correlate with the level of residual structured IscU in the mixture, as estimated from the relative intensities signals from the ¹⁵N-labeled side chain of tryptophan-76 of the apo-IscU variants, alone and in 1∶1 IscU:IscS mixtures (Fig. S4). Trp-76 is located in a mobile region in the complex, and its side-chain signal was observable in the complex.

In agreement with this model, we found that Fe-S cluster assembly in the presence of 250 μM ZnSO₄, which stabilizes an S-like state of IscU, proceeded slowly with an initial lag (Fig. S9, scarlet). The rate of cluster assembly on WT IscU in the presence of Zn(II) was comparable to that of the most structured variant N90A in the absence of Zn(II), but the level of clusters assembled was lower. As expected, the reaction mixture without IscS did not show any cluster assembly (Fig. S9, orange).

Production and Purification of Proteins.

[U-¹⁵N]-labeled samples of IscU variants were produced and purified in the apo form as described previously (35). The expression plasmids of IscU variants were created by applying the QuikChange II Site-Directed Mutagenesis Kit (Stratagene) to the pTrcIscU expression vector (35). IscS was prepared as described in SI Methods. To check whether the presence of ZnSO₄ in the expression media served to remove the conformational heterogeneity of apo-IscU as previously proposed (9), two separate [U-¹⁵N]-apo-IscU samples were prepared; one of them was expressed in a regular M9 medium without a zinc supplement, whereas the other was expressed in a M9 medium containing 8.4 μM ZnSO₄. The two protein samples were purified by the same procedure as described previously (35).

NMR Spectroscopy.

All NMR spectra were acquired at 25 °C with 600- or 800-MHz Varian NMR System spectrometers equipped with a z-gradient cryogenic probe, and all samples contained 7% D₂O, 0.7 mM 2,2-dimethyl-2-silapentane-5-sulfonate, and 0.02% sodium azide. NMRPipe (36) was used to process the raw NMR data, and Sparky (University of California, San Francisco) was used for data analysis. The samples for the signal assignments of the S and D states of WT apo-IscU and the exchange rate determination between the S and D states of WT apo-IscU were prepared in 50 mM Tris·HCl buffer at pH 7.5 containing 0.5 mM EDTA and 5 mM DTT. The samples for the hydrogen exchange rate measurements of apo-IscU(N90A) were prepared in 50 mM Tris·HCl buffer at pH 7.5 containing 150 mM NaCl, 0.5 mM EDTA, and 5 mM DTT. Salt was added to the samples used for the H/D exchange measurements because it stabilizes the S state of apo-IscU and made it easier to measure the exchange rates. We used fast maximum-likelihood reconstruction as implemented in the Newton software package (37) in analyzing the hydrogen exchange and 2D NMR exchange data. More experimental details are provided in SI Methods.

CD Spectroscopy.

The buffer used for CD spectroscopy contained 20 mM Tris·H₂SO₄ (pH 8.0) and 0.5 mM Tris(2-carboxyethyl)phosphine. The 1-mm path-length cuvettes were sealed with a rubber septum as needed. The concentration of apo-IscU was 10–20 μM. To prepare the metal complexes, IscU∶Zn²⁺ and IscU∶Fe²⁺, ZnSO₄, or (NH₄)₂Fe(SO₄)₂ were added to a protein sample to achieve a metal ion concentration of 100 μM. Far-UV CD spectra of IscU variants and complexes were recorded with an Aviv 202SF CD spectrophotometer (Aviv Biomedical) at 25 °C. Thermal denaturation curves of IscU variants were obtained by following CD signals at 222 nm while increasing the temperature in 3 °C increments from 9 to 69 °C. The protein samples were incubated for 5 min at each temperature. Because of the variable secondary structure content of IscU variants, the final thermal denaturation curves were normalized so that “0” represented the structured state whereas “1” represented the unfolded state (Fig. 1D).

Reconstitution of [Fe-S]-IscU.

The approach used was adapted from a published protocol (18). All the experiments involving [Fe-S]-IscU were conducted in an anaerobic chamber (Coy Laboratory) filled with 90% N₂ and 10% H₂ at room temperature. The O₂ level was maintained at less than 5 ppm. The buffer used to reconstitute [Fe-S]-IscU was 0.1 M Tris∙HCl (pH 7.5) and 5 mM DTT. The reconstitution mixture in this buffer contained 50 μM apo-IscU, 250 μM ferrous ammonium sulfate, and 1 μM IscS. The reaction was initiated by adding l-cysteine to achieve a concentration of 250 μM. We used 1-cm path-length cuvettes sealed with a rubber septum to measure the absorbance at 456 nm, which positively correlates with the increasing amount of the reconstituted [Fe-S]-IscU. A UV-1700 UV-visible spectrophotometer (Shimadzu) equipped with a temperature-controlled cell positioner was used for the absorbance measurements, and the raw data were processed with UVProbe 2.21 software (Shimadzu). All reconstitution experiments were repeated at least twice to confirm reproducibility.