A role for two conserved arginine residues in protected persulfide transfer by SufE-dependent SufS cysteine desulfurases

Abstract

Under stress conditions, iron-sulfur cluster biogenesis in E. coli is initiated by the cysteine desulfurase, SufS, via the SUF pathway. SufS is a type II cysteine desulfurase that catalyzes the PLP-dependent breakage of an L-cysteine C-S bond to generate L-alanine and a covalent active site persulfide. The cysteine desulfurase activity of SufS is activated by SufE, which accepts the covalent persulfide from SufS to regenerate the active site. Based on analysis of the SufS/SufE structure, it was hypothesized that two conserved arginine residues in the SufS active site, R56 and R359, could be important for persulfide transfer from SufS to SufE by regulating the positioning of the α3-α4 loop on SufS. To investigate this hypothesis, site-directed mutagenesis was used to obtain R56A/K and R359A/K SufS variants. Alanine substitution at either position caused defects to SufE-dependent SufS activity, with more conservative lysine substitutions resulting in varying levels of rescued activity. Fluorescence polarization binding assays showed that the loss of SufS activity was not due to a defect in forming the SufS/SufE complex. Surprisingly, the R359A substitution resulted in a 10-fold improvement in the K_D value for complex formation. The structure of R359A SufS explains this result as it exhibits a conformational change in the α3-α4 loop allowing SufE better access to the SufS active site. Taken together, the kinetic, binding, and structural data support a mechanism where R359 plays a role in linking SufS catalysis with modulation of the α3-α4 loop to promote a close-approach interaction of SufS and SufE conducive to persulfide transfer.

Graphical Abstract

Introduction

Sulfur is an essential element for life and is involved in the formation of a variety of biomolecules such as amino acids, coenzymes, modified tRNA bases, and iron-sulfur (Fe-S) clusters. Mobilization of sulfur often involves cysteine desulfurases that use a pyridoxal 5’-phosphate (PLP) cofactor to catalyze the breakage of the L-cysteine C-S bond yielding L-alanine and covalent persulfide on an active site cysteine (Cys-S-S⁻) as products.^{1, 2} The model organism Escherichia coli contains three cysteine desulfurases in its genome, two of which participate in bioassembly pathways for Fe-S cluster cofactors. IscS catalyzes the cysteine desulfurase step in the housekeeping ISC pathway passing the nascent persulfide intermediate directly to IscU for cluster formation under normal coditions.³ IscS is capable of interacting with multiple persulfide acceptors and also supports formation of additional thio-containing metabolites.⁴ Under conditions of oxidative stress or iron-limitation, E. coli utilizes the SUF pathway as an alternative to the ISC pathway for Fe-S cluster formation (Figure 1A).^{5, 6} Intermediates in the SUF pathway have been shown to be resistant to external oxidants and reductants, consistent with its activation under harsh conditions.^{7, 8} The sulfur mobilization step in the SUF pathway is catalyzed by the cysteine desulfurase SufS. SufS requires a specific transpersulfurase partner, SufE, to act as an intermediary and transfer the persulfide to the SufBC₂D scaffold for assembly into iron-sulfur clusters.^{9, 10}

Figure 1. The location of conserved Arg residues in the SufS active site.

(A) Sulfur liberated from Cys by SufS is utilized by the Suf pathway for iron-sulfur cluster biogenesis. (B) SufS catalyzes a desulfurase reaction to generate a covalently bound persulfide which is then transferred to SufE during the transpersulfuration reaction. Several covalent intermediates form during the desulfurase reaction. (C) . A surface rendering of the SufS dimer highlights C364 in the active site, regions involved in SufE interactions, and their proximity to the α3-α4 loop. (D) SufS crystal structures bearing the PLP cys-aldimine reaction intermediate reveal that R56 and R359 undergo conformational changes to form interactions with the substrate at this step. The green arc represents the region of SufE approach to the active site. The structures shown are based on PDB codes 6mr2, 6o11. PDB code 6o11 contains a C364A active site substitution that blocks chemistry but is rendered here as wild-type to mimic the natural state.

Both SufS and IscS belong to the family of homodimeric aminotransferase type V-fold PLP-dependent enzymes; however, there are several key structural differences that separate the two enzymes. IscS is an example of a type I cysteine desulfurase as its active site cysteine residue is located on a long, flexible loop.¹¹ The flexible loop structure allows IscS to directly transfer the persulfide product to the IscU scaffold for cluster assembly.¹² In the absence of IscU, IscS can be activated by external reductants that reduce the active site persulfide to SH₂ and promote turnover. SufS is a type II cysteine desulfurase with its active site cysteine located on a short, rigid loop buried ~10 Å below the active site entrance and a sequence insertion relative to the type I enzymes that forms a β-hairpin that partially covers the active site of the adjacent monomer.¹³ Due in part to these unique sturctural features, SufS exhibits minimal activity with exogenous reductants and requires SufE as a persulfide acceptor for optimal activity.

An abbreviated chemical mechanism for SufS is shown in Figure 1B. SufS is isolated with a PLP cofactor attached to K226 (E. coli numbering) as an internal aldimine. In the presence of L-cysteine, a transamination reaction follows to give an external PLP-Cys aldimine. A 1,3-proton transfer results in a PLP-Cys ketimine intermediate, preceeding breakage of the C-S bond to generate a PLP-Ala ketimine and the covalent persulfide product on C364. The PLP internal aldimine is regenerated by proton rearrangement of the PLP-Ala ketimine to form the PLP-Ala aldimine followed by transamination to produce the L-alanine product. In the absence of SufE, SufS exhibits low steady-state activity with a pre-steady state burst for both C-S bond cleavage and alanine formation, suggesting a rate-determining step in the transpersulfurase half-reaction.¹⁴ The presence of SufE activates this rate-determining step ~10-fold, consistent with its role as a preferred persulfide acceptor for SufS.

Our understanding of the kinetic activation of SufS by SufE is complemented by improvements in a structural description of SufS/SufE complex formation. A two-step binding process has been proposed based on a 10-fold decrease in K_D value from ~5 μM for the intial complex to ~0.5 μM for a “close-approach” complex formation measured when either SufE or SufS contains a persulfide or persulfide analog.^{15, 16} Using results from structural, kinetic, and binding assays, separate SufS interfaces were identified for these two forms of the complex with residues on the α16 helix of SufS responsible for forming the initial complex with SufE and a conserved β-latch motif, which includes the type II desulfurase β-hairpin structure, required for formation of the “close-approach” complex (Figure 1C).^{15, 17}

A recent crystal structure of the SufS/SufE complex (PDB: 8vbs) showed that the α3-α4 loop of SufS (residues 50–59, Figure 1C) may be an additional conserved structural interface involved in formation of the “close-approach” complex between SufS and SufE.¹⁷ In this structure, the SufS and SufE active site cysteines are not close enough for persulfide transfer, and the α3-α4 loop of SufS is positioned to block a closer approach of SufE. The position of the loop in the complex structure is identical to the conformation in the “resting state” SufS-alone structure (PDB: 6mr2) suggesting a conformational change is required prior to persulfide transfer.

The conformation of the α3-α4 loop is flexible as an alternate orientation was observed in the structure of the C364A variant of SufS crystallized with a trapped PLP-Cys aldimine catalytic intermediate (PDB: 6o11).¹⁸ A comparison of the trapped intermediate structure with the “resting state” structure identifies three SufS active site residues that change conformation: H55 and R56, located on the α3-α4 loop, and R359, all of which are strongly conserved in type II cysteine desulfurases (Figure S1). In the resting state structure, the conformation of the α3-α4 loop adopts an orientation placing H55 into the active site and R56 out of the active site. R359 lies on a β-strand adjacent to the active site but faces away from the PLP cofactor and interacts with residues on α3-α4 loop (Figure 1D, left panel). We define the positions of R56 and R359 as “out” in this structure. These three residues shift conformation in the 6o11 structure containing the PLP-Cys aldimine intermediate, with R56 and R359 moving into the active site to make interactions with the sulfur atom on the aldimine, while H55 shifts out of the active site (Figure 1D, right panel). In this “in” position, R359 no longer interacts with the α3-α4 loop and is poised to interact with both the PLP-Cys external aldimine sulfur and the C364 sidechain suggesting it may play a key role in chemistry. Movement of R359 into the active site allows the α3-α4 loop to adopt a new conformation with the terminal group of R56 swinging ~ 10 Å into the active site (“in” position) where it also makes contact with the sulfur of the PLP-Cys aldimine. This conformation of the α3-α4 loop would relieve conflict between SufS and SufE and allow for formation of the “close-approach” complex.

We hypothesize that coordinated motions of R359 and the α3-α4 loop work together to link the catalytic mechanism of SufS with protected persulfide transfer to SufE. To investigate this hypothesis, site-directed mutants of conserved R359 and R56 residues in SufS were generated and characterized by cysteine desulfurase kinetics, SufS/SufE binding studies, and X-ray crystallography. The H55A SufS variant has been previously characterized and shown to have minimal effects on catalysis and structure.¹⁸ In contrast, substitution of either the R56 and R359 residue with alanine results in catalytically impaired SufS variants, with the partial recovery of enzyme activity upon substitution with lysine. Fluorescence polarization binding assays show catalytic defects in the alanine variants are not due to disruption of the initial SufS/SufE complex. X-ray crystal structures of R56A and R359A SufS variants show structural changes in the active site focused on the conformation of the α3-α4 loop. Importantly, characterization of the R359A SufS variant shows a correlation between a sub-micromolar binding constant for SufE and an α3-α4 loop conformation mimicking the PLP-Cys aldimine structure. Overall, the data suggest an orchestrated mechanism followed by R56 and R359 SufS in response to the changes in the active site, where the positioning of the R359 SufS residue governs the movement of the α3-α4 mobile loop and in turn interactions with SufE.

Experimental Procedures

Generation of site-directed SufS variants

Mutations to generate the SufS variants R56A, R56K, R359A, R359K were introduced into the wildtype pET-21a-sufS vector via Quick Change Site-Directed Mutagenesis Kit (Agilent) using forward and reverse primers with desired mutations (Table S1). The resulting plasmids were transformed into XL-10 Gold E. coli cells, purified, and sequenced to confirm the desired mutations (Eurofin Genomics). Recombinant protein expression for SufS, SufE, and SufS variants was done by transforming pET-21a-sufS plasmids carrying the desired mutations into electrocompetent BL21(DE3)Δsuf E. coli cells, which lack the genomic suf operon. Next, cells were grown overnight in Luria Broth (LB) with 100 μg/ml ampicillin and 50 μg/ml kanamycin at 37 °C. The next morning the overnight cultures were diluted 100-fold in fresh LB containing 100 μg/ml ampicillin and 50 μg/ml kanamycin. The cells were incubated at 37 °C with shaking at 250 rpm until an A₆₀₀ of 0.4–0.6 was reached. At this point, protein expression was induced with 500 μM IPTG. Cells were grown for 3 h post induction and harvested by centrifugation. Cell pellets were stored at −80 °C until purification. For SufS purification, cells were thawed and resuspended in lysis buffer consisting of 25 mM Tris pH 7.5, 100 mM NaCl, 5 mM DTT, 0.02 mg/mL DNase, 1 mM PMSF, and 1 mM MgCl₂. Resuspended cells were lysed via sonication for 10 min (30 sec on and 30 sec off) at output control of 4 and duty cycle of 50% on a Branson sonifier. The lysate was centrifuged at 20,000g for 30 min at 4 °C to pellet cell debris. SufS and SufE proteins were purified using methods described previously.¹⁹ Briefly, WT SufS and SufS mutants were purified using three sets of columns in a sequence of Q-XL anion-exchange, phenyl-HP hydrophobic interactions, and Superdex-200 size exclusion chromatography (Cytvia). Lysate containing wildtype SufE was purified using two sets of columns in a sequence of Q-XL anion exchange and Superdex-200 size exclusion chromatography. Fractions containing desired proteins (SufS or SufE) were identified using SDS-PAGE, pooled, concentrated, and were stored in 10% glycerol at −80 °C for further use. Concentrations for SufS were based on PLP quantification at 388 nm in 0.1 M NaOH (ε₃₈₈ = 6600 M⁻¹ cm⁻¹).²⁰ PLP occupancy was determined by comparing the concentration of PLP with the concentation of SufS determined by absorbance at 280 nm (calculated ε₂₈₀ = 49,850 M⁻¹ cm⁻¹) and was greater than 90% for all enzymes used in this study. The concentration of SufE was determined by absorbance at 280 nm (calculated ε₂₈₀ = 20,970 M⁻¹ cm⁻¹).

SufS alanine-NDA assay

SufS cysteine desulfurase activity was measured by monitoring the production of L-alanine using naphthalene-2,3-dicarboxaldehyde (NDA) derivatization through fluorescence detection as developed by Dos Santos et al.²¹ Briefly, components of the assay mixture contained 0.25–1 μM SufS (based on PLP absorbance), 50 mM MOPS, pH 8.0, 150 mM NaCl, 2 mM tris(2-carboxyethyl)phosphine) (TCEP) and varying amounts of L-cysteine (0–500 μM) and SufE (0–20 μM). Reactions (50 μL volume) were initiated by the addition of SufS and quenched using 10 μl of 10% trichloroacetic acid (TCA) at various time points ranging from 30 s to 30 min. The quenched reaction mixtures were centrifuged to remove any precipitated protein. Next, 500 μL of freshly prepared NDA-mix containing 100 mM borate, pH 9.0, 2 mM KCN, and 0.2 mM NDA was added to the quenched reaction mixtures and were incubated in the dark for 60 mins. Following incubation, 10 μL of each sample was run on a Zorbax C18 column (Agilent) using a Shimadzu Prominence HPLC with an isocratic gradient of 50% ammonium acetate (pH 6.0) and 50% methanol at the flow rate of 0.4 mL/min. A fluorescence detector was used to detect the Ala-NDA fluorescence adduct (390 nm excitation/440 nm emission). Peak areas were integrated for the Ala-NDA adduct, and the peak area was converted to nanomoles of alanine using an alanine standard curve, made under the same conditions. Initial velocities for alanine formation were determined from at least three time points.

Fluorescence polarization SufS-SufE binding assay

A fluorescently labeled variant of SufE (C51A/E107C SufE) was obtained as previously discussed.¹⁵ The pET21 plasmid carrying the C51A/E107C SufE gene was transformed into BL21(DE3)Δsuf E. coli cells and expressed and purified as described for SufE above with a change in purification buffer pH from 7.5 to 8.0. Following purification and concentration, the protein was dialyzed into 50 mM MOPS (pH 8) and 150 mM NaCl to remove DTT. Fluorescently labeled samples of C51A/E107C SufE were created by incubating 50 μM protein with 500 μM BODIPY FL maleimide (Invitrogen) for 4 h at 25 °C. After the incubation period, unreacted dye was removed using a PD-10 column. Labeling efficiency was determined by comparing the ratio of label concentration (ε₅₀₅ = 80,000 M⁻¹cm⁻¹) to protein concentration (ε₂₈₀ = 20,970 M⁻¹cm⁻¹) and was determined to be between 60–80% using this method. SufS-SufE binding titrations were carried out in black 96-well plates in 50 mM MOPS (pH 8), 150 mM NaCl, and 0.1 mg/mL bovine serum albumin. SufS (0.2–80 μM) was mixed with 100 nM labeled C51A/E107C SufE and allowed to equilibrate at room temperature for 30 min. After equilibration, the fluorescence polarization (480 nm excitation, 520 nm emission) was measured using a BioTek Synergy2 multiwell plate reader.

X-ray structure determination

Crystallization and structure solution of SufS site-directed mutants was performed essentially as previously reported.¹⁵ Briefly, 11–16 mg/mL protein was mixed 1:2 (vol/vol) with crystallization solution, 4–4.5 M NaCl, and 100 mM MES pH 6.5, and subjected to sitting drop vapor diffusion at 20 °C. Crystals appeared after several days. Crystals were cryoprotected by soaking for several minutes in a solution consisting of mother liquor with 50% glycerol. Crystals were plunge-frozen in liquid nitrogen and X-ray data collection proceeded at 100 K with 1.54 Å X-rays generated by a Rigaku Synergy DW equipped with a HyPix6000-HE detector. Data reduction was performed with XDS and merged with XSCALE.²² The structure given by PDB code 6mr2 (with heteroatoms removed) provided phases for structure solution. Model building and refinement were performed iteratively using Coot and Phenix.^{23, 24} X-ray data statistics and model quality data are reported in Table 3. Final coordinates were submitted to the PDB as PDB codes 7rrn (R56A SufS) and 9d2d (R359A SufS).

Table 3.

X-ray data statistics.

	SufS R56A	SufS R359A
	7RRN	9D2D
Data collection
Space group	P43212	P43212
Cell dimensions
a, b, c (Å)	126.14, 126.14, 133.27	126.14, 126.14, 133.45
α, β, γ (°)	90, 90, 90	90, 90, 90
Wavelength (Å)	1.54	1.54
Resolution (Å)a	91.99–2.30 (2.36–2.30)	45.83–2.70 (2.77–2.70)
Rmeas (%)	7.1 (83.8)	13.3 (85.2)
I/σI	25.93 (1.40)	19.31 (2.78)
Completeness (%)	98.5 (91.5)	90.9 (80.1)
Redundancy	10.95 (2.78)	10.59 (8.50)
CC1/2 (%)	99.9 (63.1)	99.7 (84.2)
Refinement
Resolution (Å)	31.79–2.30 (2.36–2.30)	42.30–2.70 (2.77–2.70)
No. reflections	48316	27426
Rwork/Rfree (%)	20.63 (28.21)/ 23.07(30.45)	21.03 (34.12)/ 24.47 (37.06)
No. atoms	3165	3158
Protein	3104	3113
Ligand	15	15
B factors	45.27	36.88
Protein	45.30	36.90
Ligand	43.52	38.92
R.m.s deviations
Bond lengths (Å)	0.008	0.008
Bond angles (°)	0.957	0.954

Values in parentheses are for the highest-resolution shell.

Data analysis

Initial velocity data for alanine formation was fit to the Michaelis-Menten equation (Eq 1) to determine the kinetic parameters for each enzyme. Here, $v$ is the initial velocity, $E_t$ is the total enzyme concentration, $k_{\mathrm{cat}}$ is the maximal turnover number, $S$ is the substrate concentration, $K_M$ is the Michaelis constant. Equation 2 describes substrate inhibtion kinetics with an added inhibition constant ($K_i$). Fluorescence anisotropy data used to measure protein-protein interactions were fit to equation 3 where $A_o$ is the anisotropy in the absence of the ligand, ΔA is the total change in anisotropy, and $K_D$ is the dissociation constant.

$$\frac{v}{E_t}=\frac{k_{cat}\left[S\right]}{K_M+\left[S\right]}$$ (1)

$$\frac{v}{E_t}=\frac{k_{cat}[S]}{K_M+[S](1+\frac{[S]}{K_i})}$$ (2)

$$Flpolarization=A_0+\Delta A\times \left(\frac{\left(\left[SufE\right]+\left[SufS\right]+K_D\right)-\sqrt{{\left(\left[SufE\right]+\left[SufS\right]+K_D\right)}^2-(4\times [SufE]\times \left[SufS\right])}}{2\times [SufE]}\right)$$ (3)

Results

Steady-state kinetics suggests that R56 and R359 are required for optimum SufS activity.

R56 and R359 SufS residues were separately substituted with either alanine or lysine to generate four SufS variants. The R56 and R359 SufS variant proteins along with WT SufS were expressed and purified to homogeneity as determined by SDS-PAGE (Figure S2). Steady-state kinetic parameters of R56 and R359 SufS variants were evaluated using an HPLC-based fluorescence alanine detection assay.²¹ The rate of alanine formation for the SufS enzyme was quantified via NDA-derivatization and HPLC coupled with a fluorescence detector. Kinetic parameters were obtained using the Michaelis-Menten equation. Optimal steady-state activity of SufS requires the presence of both SufE, as a preferred persulfide acceptor, and TCEP, as a preferred reductant for the SufE persulfide. In the absence of SufE, TCEP reduction of the SufS persulfide can support cysteine desulfurase activity, but at a reduced rate limited by the reduction step.^{7, 15} Thus, WT SufS has a low SufE-independent activity with a $k_{\mathit{cat}}$ value of 1.9 ± 0.1 min⁻¹ and $K_m$ value of 28 ± 1 μM for cysteine (Table 1, SufE-independent activity, and Figure 2A). When assayed in the absence of SufE, all four SufS variants exhibit similar $K_{\mathrm{M}}$ values for cysteine (25–60 μM) suggesting the initial interaction with cysteine is not perturbed. Substitutions with alanine at both positions results in decreases in $k_{\mathrm{cat}}$ values of 5- and 10-fold for the R56A and R359A variants, respectively. The more conservative R56K substitution is able to fully restore the SufE-independent WT SufS activity, while the R359K substitution results in partial rescue.

Table 1.

Kinetic parameters for R56 and R359 SufS variants.^a

	SufE-independent activity b		SufE-dependent activity c
Enzyme	$k_{\mathrm{cat}}$ (min−1)	$K_{\mathrm{cys}}$ (μM)	$K_{\mathrm{cat}(\mathrm{SufE})}$ (min−1)	$K_{\mathrm{SufE}}$ (μM)
WT SufS	1.9 ± 0.01	28 ± 1	22 ± 2d	0.7 ± 0.2d
R56A SufS	0.4 ± 0.01	60 ± 7	not activatede	-
R56K SufS	1.9 ± 0.10	24 ± 4	13 ± 1	0.8 ± 0.3
R359A SufS	0.2 ± 0.01	27 ± 4	not activatede	-
R359K SufS	0.9 ± 0.03	33 ± 5	2 ± 0.4	0.5 ± 1.0

^aExperiments performed as described in Methods section.

^bMichaelis-Menten kinetic parameters under varying cysteine concentration in presence of 2 mM TCEP.

^cMichaelis-Menten kinetic parameters under varying cysteine concentration in the presence of SufE and 2 mM TCEP.

^dThe WT SufS kinetics displays substrate inhibition by SufE with a $K_{\text{i}}$ value of 20 ± 6μM. Data from reference 15.

^eSufS activity is not improved in the presence of SufE.

Figure 2. Kinetics of alanine production by WT, R56A, and R359A SufS variants under steady state conditions.

Alanine production was measured by HPLC-based assay. (A) Formation of alanine in the absence of SufE. (B) Formation of alanine in the presence of SufE and 500 μM cysteine. Constant reaction conditions for both experiments were 50 mM MOPS pH 8.0, 150 mM NaCl, 2 mM TCEP, and 0.1 to 1.0 μM SufS variant. When SufE was present, SufS concentration was at least 5-fold lower than the SufE concentration in each assay. Solid lines are from fits of the data to equations describing Michaelis-Menten kinetics (equation 1) or substrate inhibition kinetics (equation 2). Duplicate data points are shown but may be obscured by closely overlapping symbols.

In the presence of SufE, SufS turnover is activated ~10-fold due to the increased rate of persulfide transfer to SufE.^{7, 14} Using SufE as a varied substrate under saturating concentrations of cysteine results in substrate inhibition kinetics for WT SufS (Table 1, SufE-dependent activity, and Figure 2B).¹⁵ This result has been reported previously, albeit at lower cysteine concentrations⁸, and likely involves SufE binding to SufS in a non-productive manner at higher concentrations. The rate of alanine production by the two alanine substituted variants, R56A and R359A SufS, was not stimulated by the addition of SufE up to 20 μM and remained at the level of the SufE-independent activity (Figure 2B). In contrast, the more conservative lysine substitutions do allow for activation of SufS by SufE and do not appear to be subject to significant substrate inhibition in the experimental SufE concentration range. Similar to the results above, R56K SufS rescues SufE-dependent activity, while the R359K substitution only partially rescued activity. Overall, the conserved arginine residues are important for SufS catalysis.

Substitution of R56 or R359 does not negatively perturb SufS/SufE interactions.

One explanation for the failure of SufE to activate several of the SufS variants is that the substitution disrupts formation of the SufS/SufE complex. To investigate this possibility, $K_{\text{D}}$ values for the SufS/SufE complex formation were determined via a fluorescence polarization binding assay.¹⁵ As previously described, 0.05 μM C51A/E107C SufE labeled with BODIPY-FL-maleimide was titrated with the SufS variants (0–20 μM) (Figure 3). Changes in fluorescence polarization were fit to equation 3 to determine a $K_{\text{D}}$ value (Table 2). Previous studies on WT SufS have shown the presence of cysteine promotes tighter assocation between SufS and SufE with a 10-fold improvement in $K_{\text{D}}$ value (from ~5 μM to ~ 0.5 μM)¹⁵, so binding assays were determined in the absence and presence of 500 μM cysteine. In the absence of cysteine, the R56A, R56K, and R359K variants show $K_{\text{D}}$ values similar to that determined with WT SufS, suggesting that binding at the α16 helix of SufS has not been perturbed. In contrast, R359A SufS exhibits a reduced $K_{\text{D}}$ value in the absence of cysteine suggesting this variant has improved the SufS/SufE interaction. The addition of cysteine causes small decreases in $K_{\text{D}}$ values for complex formation with the value for R359A remaining lower than the values determined with the other variants. With reasonable $K_{\text{D}}$ values determined for all SufS variants, the kinetic defects are unlikely to be due to loss of SufE binding.

Figure 3. SufS-SufE binding measured by fluorescence polarization.

The data show changes in fluorescence polarization due to titration of BODIPY-FL labeled C51A/E107C SufE (0.05 μM) by WT, R56 and R359 SufS variants (0.05 – 20 μM) in the (A) absence and (B) presence of 500 μM cysteine. Error bars are the results of standard deviation from triplicate data. Solid lines in both panels are from a fit of the data to Equation 3. $K_{\text{D}}$ values determined from the fits are shown in Table 2.

Table 2.

Dissociation constants for formation of the SufS/SufE complex determined via fluorescence polarization.^a

Enzyme	$K_{\text{D}}$ w/o cysteine (μM)	$K_{\text{D}}$ w/ cysteine (μM)
WT SufS	2.5 ± 0.3	0.2 ± 0.1
R56A SufS	5.8 ± 0.9	4.0 ± 0.4
R56K SufS	8.5 ± 1.1	2.8 ± 0.3
R359A SufS	0.9 ± 0.2	0.3 ± 0.1
R359K SufS	4.6 ± 0.5	2.8 ± 0.3

^aExperiments performed as described in Methods section.

The shift in $K_{\text{D}}$ values in the presence of cysteine is thought to occur due to formation of a trapped SufS persulfide intermediate that cannot be transferred to the labeled SufE as it lacks the C51 active site residue.¹⁵ This high-affinity SufS/SufE complex has been termed the “close-approach” complex as it likely mimics the conformation leading to persulfide transfer. Thus, a decrease in $K_{\text{D}}$ value for the SufS/SufE complex determined in the presence of cysteine reports on both the catalytic fitness of SufS and its ability to form the “close-approach” conformation. Among the R56A, R56K, and R359K variants, none show a decrease in $K_{\text{D}}$ value on the order of that seen with the WT SufS enzyme. Of these three variants, only R56K SufS exhibits a catalytic ability close to that of the WT enzyme and a measurable shift in $K_{\text{D}}$ value suggesting that the conservative substitution may lead to an altered “close-approach” complex that is still able to support catalysis. In contrast, the lack of change in R56A and R359K SufS may derive from low catalytic activity. R359A SufS displayed an unexpected behavior, but one that was rationalized by its crystal structure (below): it shows high-affinity for SufE in the presence and absence of cysteine. As this variant is catalytically impaired, these results suggest that the substitution allows SufS to form a version of the “close-approach” complex in the absence of any catalytic intermediates.

Both the R56A and R359A substitutions result in conformational changes in the α3-α4 loop on SufS.

To better understand the structural reasons for the kinetic defects and altered SufE binding effects of the SufS variants, X-ray crystal structures were determined for the R56A and R359A variants (Table 3). In the R56A SufS structure (PDB ID: 7rrn), residues 56 and 57 are not ordered in the α3-α4 loop (residues 50–59) (Figure 4A). Compared to a high-resolution structure of WT SufS (PDB ID: 6mr2), the adjacent residues of H55 and I58 are displaced slightly from their WT positions. The overall disorder of the loop appears to have minimial effects on positions of the active site residues, and does not provide an immediate explanation to the lack of catalytic activity for the transpersulfurase reaction seen with the R56A variant. Important to discussion later in the text, R359 is found in the out position interacting with the α3-α4 loop.

Figure 4. Active site structure of R56A and R359A SufS.

(A) The X-ray structure SufS R56A reveals no electron density is present for loop residues R56A and G57, indicating this two amino acid segment has become dynamic in the mutant. R359 is located in the “out” position and interacts with the α3-α4 loop. A 2F_o-F_c electron density map contoured at 1.0σ is shown as blue mesh. (B) A comparison of the X-ray structures of wild type SufS (PDB code 6mr2) versus the R359A mutant reveals that loss of the R359 sidechain promotes a dramatic conformational change in H55, R56 and the α3-α4 loop that now has an inward pointing R56. The conformation with R56 “in” and H55 “out” was previously observed in the PLP-Cys aldimine structure of SufS (PDB cod 6o11). A semi-transparent R56 residue from the wild type structure is shown superpositioned against the R359A mutant to reveal the magnitude (arrow) of the conformational change.

The X-ray crystal structure of R359A SufS also exhibits changes in the α3-α4 loop conformation (Figure 4B). In this structure, H55 and R56 have swapped locations with H55 pointing out of the active site and R56 in the “in” position protruding into the active site. In the inward position, R56 interacts with the sidechain of T278 and is located ~ 4Å from C364. Superposition of the R359A structure with a C364-persulfide containing structure (PDB ID: 6mr2) shows the R56 sidechain within 3.3 Å of the terminal sulfur atom. In the new position for H55, it interacts with the side chain of nearby E70, located in the adjacent α4 helix. The positioning of the R56 and the α3-α4 loop are in a near identical pose to that seen in the C364A SufS structure that contains a PLP-Cys aldminine intermediate, which originally initiated the interest in these positions. This structural result provides an explanation for the low $K_{\text{D}}$ values seen with the R359A SufS variant for the SufS/SufE complex. A recent co-crystal of the SufS/SufE complex from E. coli shows that positioning of the R56 residue on the α3-α4 loop can physically block the approach of SufE.¹⁷ The R359A-induced “in” conformation of the α3-α4 loop would be expected to result in formation of the ”close-approach” SufS/SufE complex in the absence of cysteine, consistent with the binding assay results for this variant.

Discussion

The SufS cysteine desulfurase from E. coli is among the best characterized type II systems. With its dedicated role in Fe-S cluster bioassembly, it is also a model system for the study of protected persulfide transfer in living systems. Recent structures of C364A SufS with a trapped PLP-Cys aldimine and for H123A SufS with a trapped PLP-Cys ketimine intermediate allowed for identification of acid/base roles in catalysis including K226 for removal of the α-proton and H123 for deprotonation of C364 prior to nucleophilic attack on the PLP-Cys ketimine intermediate.¹⁸ The C364A SufS structure with a trapped Cys-aldimine intermediate also showed movements in the positioning of H55 and R56 on the α3-α4 loop and residue R359 with both R56 and R359 forming interactions with the sulfur atom of the PLP-Cys aldimine structure. These results took on a new importance when a recent structure of the SufS/SufE complex from E. coli suggested that the conformation of the α3-α4 loop may play a role in regulating persulfide transfer between SufS and SufE. To probe the significance of these residues in the SufS desulfurase reaction, site-directed variants of R56 and R359 were generated in SufS and the resulting enzymes were analyzed for effects on catalysis, affinity for SufE, and structure.

R56 is a key residue on theα3-α4 loop for SufS function.

H55 and R56 reside on the mobile α3-α4 loop, which undergoes a change in conformation upon formation of the PLP-Cys aldimine intermediate with H55 moving from “in” to “out” and R56 moving from “out” to “in”. The SufS H55A variant was previously characterized and showed no detrimental effects on the activity of SufS or the structure.¹⁸ In contrast, the R56A substitution results in loss of catalytic activity. The more conservative substitution of R56K is able to partially restore activity activity suggesting the positive charge of R56 is required. On the basis of $K_{\text{D}}$ values determined for formation of the SufS/SufE complex, loss of R56 is not detrimental to formation of the initial complex at the SufS α16 helix. The differential response of the H55A and R56A substitutions suggest that although both residues change positions, it is the “in” position of R56 that is of catalytic importance.

R359 is essential for catalysis and regulating formation of the “close-approach” complex.

R359 appears to play a more specific role in catalysis compared to R56, based on the result that the charge conservation of a R359K substitution still exhibits very poor kinetic parameters. The contribution to catalysis is perhaps not surprising given the location of R359 when it is in the “in” position. The PLP-cys aldimine structure (PDB ID: 6o11) highlights the 3.3 Å proximity of the R359 sidechain to the sulfur atom of the PLP-cys. Superposition of the PLP-cys aldimine SufS structure with a structure containing a C364 persulfide intermediate (PDB ID: 6mr2) shows that R359 in the “in” position would also be placed with 2.5 Å of the SufS persulfide intermediate. Finally, the R359 residue is conserved in other SufS homologs including SufS from Bacillus subtillis and Mycobacterium tuberculosis. In these organisms, the SufE persulfide acceptor is functionally replaced with SufU. Analysis of the B. subtillis and M. tuberculosis SufS/SufU structures captured near the point of persulfide transfer shows the equivalent R359 arginine residue (BsSufS R356, MtSufS R368) within 3–4 Å from the incoming SufU cysteine that will act as a persulfide acceptor (Figure S3).^{25, 26} In total, structural evidence places the “in” conformation R359 in SufS enzymes at a location where it is able to interact with all facets of persulfide transfer within the system.

One of the more surprising results from this set of experiments is the finding that the R359A SufS variant binds to SufE with high-affinity in the absence of cysteine. Inspection of the structure shows that conformational changes in response to the R359A substitution are limited to changes in the α3-α4 mobile loop. As described above, the new conformation moves R56 from the “out” to the “in” position. A recent structure of the E. coli SufS/SufE complex identified R56 as a residue that could physically block SufE from forming a “close-approach” complex.¹⁷ The combined results from the R359A substitution support this hypothesis as the R56 “in” structure correlates with tight binding between R359A SufS and SufE.

A proposed mechanism of coordinated motion for R359 and R56 in the SufS mechanism.

Considering all the data, we propose a new mechanism explaining the coordinated function of R56 and R359 during the SufS catalytic cycle. In the resting state of SufS, R359 is positioned “out” of the active site and interacts with residues and backbone atoms in the α3-α4 mobile loop (Figure 5, left image). This interaction prevents the loop from changing conformation and maintains the “out” position for R56 and the α3-α4 mobile loop. This conformation may allow for interactions between SufS and SufE at the α16 helix, but prevents formation of a “close-approach” complex in the absence of a catalytic intermediate in the SufS active site as seen in the recent SufS-SufE structure. Upon formation of catatlyically active intermediates, R359 moves to the “in” position and coordinates persulfide transfer steps (Figure 5, right image). With R359 in the “in” position, the α3-α4 mobile loop is no longer sterically constrained and is allowed to undergo a conformational change to bring R56 to the “in” position and promotes formation of the high-affinity, close-approach complex with SufE. It should be noted that while movement of R359 to the “in” position is required for movement of the α3-α4 mobile loop, it may not be sufficient as the recent SufS/SufE complex structure captures R359 facing “in” with R56 still facing “out”. The identities of catalytic intermediates that trigger the conformational change are not currently clear. The transition may also rely on a conformational change of the β-latch motif, which is also required for formation of the close-approach complex.

Figure 5. Regulatory model for SufS/SufE complex formation based on movement of the α3- α4 loop.

The catalytic cycle of SufS is shown in the center reaction scheme. Models for the resting state (left) and persulfide transfer state (right) of SufS are shown that highlight the coordinated motions of R359 and the α3-α4 loop containing R56.

Conclusions

In E. coli, a PLP-based cysteine desulfurase, SufS, mobilizes sulfur for biosynthesis of iron-sulfur clusters via the SUF pathway. Recent structural, kinetic, and bioinformatic studies of SufS proposed the involvement of R56, on the α3-α4 mobile loop, and R359, in the active site, in stabilizing PLP-intermediates during the SufS catalytic cycle and regulating SufE access to the active site. Here, we have shown that both residues are necessary for optimal SufS catalysis. Despite the diminished catalytic activity, loss of R359 results in tight binding of SufE, and the structure of the R359A variant indicated this result could be linked to displacement of R56 and α3-α4 mobile loop. Taken together, the data support a coordinated mechanism to link SufS catalytic activity and SufE active site access through conformational changes in R359 and the α3-α4 mobile loop. This molecular mechanism of protected persulfide transfer may be broadly applicable to other SufE-dependent cysteine desulfurase systems.

Supplementary Material

Supporting Information

Supporting information contains one table and two supplementary figures. Table S1 describes the oligonucleotide primers used for mutagenesis. Figure S1 is a Hidden Markov Model logo for the regions surround R56 and R359 in SufS. Figure S2 is an SDS-PAGE image showing the quality of purified proteins used in this study. Figure S3 is a comparison of active site structures of SufS from Mycobacterium tuberculosis and Bacillus subtilis.

Abbreviations

Ala

alanine

Cys

cysteine

DTT

dithiothreitol

K_D

dissociation constant

MOPS

3-morpholinopropane-1-sulfonic acid

NDA

naphthalene, 2,3-dicarboxaldehyde

PLP

pyridoxal 5’-phosphate

TCEP

tris (2-carboxyethyl)phosphine