Reaction Mechanism and Molecular Basis for Selenium/Sulfur Discrimination of Selenocysteine Lyase*

Selenocysteine lyase (SCL) catalyzes the pyridoxal 5′-phosphate-dependent removal of selenium from l-selenocysteine to yield l-alanine. The enzyme is proposed to function in the recycling of the micronutrient selenium from degraded selenoproteins containing selenocysteine residue as an essential component. The enzyme exhibits strict substrate specificity toward l-selenocysteine and no activity to its cognate l-cysteine. However, it remains unclear how the enzyme distinguishes between selenocysteine and cysteine. Here, we present mechanistic studies of selenocysteine lyase from rat. ESI-MS analysis of wild-type and C375A mutant SCL revealed that the catalytic reaction proceeds via the formation of an enzyme-bound selenopersulfide intermediate on the catalytically essential Cys-375 residue. UV-visible spectrum analysis and the crystal structure of SCL complexed with l-cysteine demonstrated that the enzyme reversibly forms a nonproductive adduct with l-cysteine. Cys-375 on the flexible loop directed l-selenocysteine, but not l-cysteine, to the correct position and orientation in the active site to initiate the catalytic reaction. These findings provide, for the first time, the basis for understanding how trace amounts of a selenium-containing substrate is distinguished from excessive amounts of its cognate sulfur-containing compound in a biological system.

Selenium is an essential trace element required for the normal function of mammals (1). It is similar to sulfur in its chemical properties and needs to be discriminated biologically from abundant sulfur during its metabolism. During the synthesis of selenoproteins, selenium is specifically incorporated into selenocysteyl-tRNA Sec and eventually into the specific position of the nascent polypeptide chains as directed by the UGA codon (2). The specific synthesis of selenocysteyl-tRNA Sec from seryl-tRNA Sec and selenide involves coordinated sequential reactions catalyzed by selenophosphate synthetase (3,4), seryl-tRNA kinase (5), and selenocysteine synthase (6,7). However, the mechanism underlying selenium-specific substrate recognition in these enzymes remains unknown.
The most common selenium source in normal foods is selenocysteine and selenomethionine in proteins (8). Previous studies suggest that selenomethionine is transformed to selenocysteine through the trans-selenation pathway (9,10). Specific cleavage of selenocysteine by selenocysteine lyase (SCL) 5 is thought to be the critical metabolic step for the recycling of selenocysteine, which is formed by proteolytic degradation of selenoproteins.
SCL is a pyridoxal 5Ј-phosphate (PLP) enzyme that catalyzes the removal of selenium from L-selenocysteine to yield L-alanine. In previous studies, we identified and characterized SCL from pig liver (11) and Citrobacter freundii (12) as the first enzyme that discriminately acts on a selenium-containing compound but not on its cognate sulfur analog. The cDNA cloning of mouse SCL (13) revealed that SCL belongs to subgroup V in fold type I of aminotransferases (14). Among them, overall sequence similarity (ϳ30%) is found between mammalian SCL and cysteine desulfurases (13).
SCL and cysteine desulfurase catalyze the same type of reaction, namely the removal of selenium or sulfur from selenocysteine or cysteine to yield alanine. Although bacterial cysteine desulfurases exhibit activity to both cysteine and selenocysteine (15), SCL shows significant specificity to selenocysteine. Mechanistic and crystal structure studies showed that the catalytic reaction of cysteine desulfurase involves the formation of sulfane sulfur in the form of a cysteine persulfide in the active site of the enzyme (16 -18). Sulfur transfer from the persulfide to a cysteine residue of its partner proteins, such as IscU (19), ThiI (20), TusA (21), and SufE (22,23), is associated with direct protein-protein interaction, thereby ensuring the specific delivery of the persulfide sulfur to eventual target molecules, such as an iron-sulfur cluster, thiamine, and thiolated nucleosides, without nonspecific spreading of reactive sulfur in the cell (24). Although SCL has a conserved cysteine residue at the position corresponding to the persulfide-forming cysteine residue of cysteine desulfurase, its role in the catalytic function of the enzyme has not been established.
Here, we investigate the mechanism of rat SCL. Site-directed mutagenesis and ESI-MS studies using wild-type and C375A mutant of SCL show that Cys-375 is essential for the activity of SCL as the site for cysteine selenopersulfide formation. UVvisible spectrum analysis and crystal structure of SCL complexed with L-cysteine in a non-productive adduct form shows the structural basis for discrimination between selenocysteine and cysteine by the enzyme. UV-visible spectrum analysis also shows that loss of Cys-375 resulted in the formation of a nonproductive adduct with selenocysteine, suggesting that Cys-375 serves as a guide to direct the substrate in an appropriate orientation for the catalytic reaction.

EXPERIMENTAL PROCEDURES
Materials-L-Selenocystine and L-cysteine were purchased from Sigma Aldrich. Selenopropionate was synthesized as described in supplemental Experimental Procedures. Oligonucleotides were synthesized by Hokkaido System Science (Sapporo, Japan). Restriction and DNA modification enzymes were purchased from New England Biolabs (Beverly, MA), Takara Shuzo (Kyoto, Japan) and Toyobo (Osaka, Japan). All other chemicals were analytical grade reagents.
Gene Expression and Purification of SCL-cDNA for the fulllength SCL protein of Rattus norvegicus (GenBank TM /EMBL accession number NM_001007755) was cloned by PCR using the primers 5Ј-AGCGATCAGGCATATGGACGTGGCGCG-GAATGGC-3Ј and 5Ј-CCGGAATTCCTAGACCGGGCTTC-CAGCTGGTTC-3Ј. PCR products were cloned into the NdeI and EcoRI sites of pET21a(ϩ) for heterologous expression in Escherichia coli Rosetta(DE3). The E. coli cells harboring the expression plasmid were grown in an LB medium containing 50 g/ml ampicillin at 28°C for 16 h and were harvested by centrifugation. A Tris-HCl buffer (10 mM) (pH 7.4) containing 0.1 mM phenylmethylsulfonyl fluoride and 1 mM EDTA was used as the standard buffer. The cells were disrupted by sonication. The cell debris was removed by centrifugation, and the supernatant solution was applied to a Q-Sepharose column (5 ϫ 5 cm). After the column was washed, the enzyme was eluted with a 1.0-liter linear gradient of 0 -0.5 M NaCl in the buffer. The fractions containing SCL were pooled, mixed with an equal volume of standard buffer containing 2.0 M ammonium sulfate, and applied to a Butyl-Toyopearl column (3 ϫ 15 cm). After the column was washed, the enzyme was eluted with a 1.0-liter linear gradient of 1.0 -0.6 M ammonium sulfate in the buffer. The enzyme fractions were concentrated with Centriprep YM-30 (Millipore, Bedford, MA) and stored at Ϫ80°C until use without significant inactivation. Purification of the C375A mutant of SCL was performed in the same manner as described for the wild-type enzyme.
Enzyme Assay and Kinetic Analysis-SCL activity was assayed by the determination of selenide with lead acetate as described previously (25,26). Protein was determined using the CBB protein assay (Nacalai Tesque, Kyoto, Japan) with bovine serum albumin as a standard.
ESI-MS Sample Preparation and Measurements-Samples of wild-type SCL and the C375A mutant for ESI-MS analysis were prepared under a nitrogen atmosphere in a glove box (Miwa, Osaka, Japan) at room temperature. SCL (230 M) was incubated with L-selenocysteine (70 M) in a 50 mM Tris-HCl buffer (pH 7.0) and immediately separated with a MicroBioSpin 6 column (Bio-Rad). The desalted enzyme was introduced at a flow rate of 18 l/min into a PE Sciex API 2000 single quadrupole mass spectrometer equipped with an electrospray ionization source (PE-Sciex, Thornhill, ON, Canada). The mass spectrometer scanned from m/z ϭ 800 to m/z ϭ 1800 with a 2.0 ms dwell time and a 0.2 m/z step size.
Crystallization and Data Collection-The purified enzyme was dialyzed against 10 mM HEPES-NaOH (pH 7.5). SCL crystals were grown by the hanging-drop vapor-diffusion method at 297 K. A 2 l droplet of a 35 mg⅐ml Ϫ1 protein solution mixed with the same amount of reservoir solution was equilibrated at 293 K against a 500 l reservoir solution (100 mM citrate buffer pH 4.7 and 1.5 M ammonium dihydrogen phosphate). The native crystals appeared within 2 weeks of incubation and grew to maximum dimensions of 0.3 ϫ 0.3 ϫ 0.1 mm. The crystals (pH 4.7) were transferred to the reservoir solution adjusted to pH 8.0 using 1.5 M di-ammonium hydrogen phosphate instead of ammonium dihydrogen phosphate, to produce the native crystals at pH 8.0. Crystals of SCL⅐L-cysteine and SCL⅐selenopropionate were obtained at 293 K by soaking the native crystals in the reservoir solution adjusted to pH 8.0. These crystals are isomorphous with the orthorhombic space group P2 1 2 1 2 1 , average unit cell parameters a ϭ 55.0, b ϭ 101.9, c ϭ 197.3 Å. Assuming two monomers in the asymmetric unit, the Matthews coefficient (V M ) value was calculated to be 2.96 Å 3 Da Ϫ1 , indicating an estimated solvent content of 58.3% in the unit cell (27).
Data collection was performed at 100 K using a wavelength of 1.00 Å from the synchrotron-radiation source and an ADSC Quantum CCD Camera. X-ray diffraction data sets for the unliganded SCL (pH 8.0) were collected to 2.0 Å resolution on the BL5A station at the Photon Factory, KEK (Tsukuba, Japan). X-ray diffraction data sets for the SCL⅐L-cysteine crystal and the SCL⅐selenopropionate crystal were collected to 1.85 and 1.55 Å resolution on the BL41XU and the BL38B2 station, respectively, at SPring8 (Hyogo, Japan). Data were processed using DENZO/SCALEPACK (28). Data collection statistics are presented in Table 1.
Structure Determination and Refinement-The structure of the unliganded SCL (pH 8.0) was solved by the molecular replacement method with the program AMoRe (29) using the IscS structure (PDB ID: 1P3W) as the search model. The modeling of the polypeptide chain was performed using program O (30), and the structure was refined with program CNS (31). Water molecules were picked up on the basis of the peak heights (1.5 ) and distance criteria (4.0 Å from protein and solvent) from the sigma-weighted 2F o Ϫ F c map.
The water molecules whose thermal factors were above the maximum thermal factor of the main chain after refinement were removed from the list. Model building and refinement cycles resulted in an R factor of 0.197 and an R free of 0.236% at 2.0 Å resolution, calculated for 69817 reflections. The same refinement procedure was applied to SCL⅐L-cysteine and SCL⅐selenopropionate but using the coordinates of the native enzyme as the initial model. Refinement cycles resulted in an R factor of 0.187 and 0.211 and R free of 0.187 and 0.209 for SCL⅐Lcysteine and SCL⅐selenopropionate, respectively. Structure diagrams were drawn using MOLSCRIPT (32), POVSCRIPTϩ (33), and PyMol (34).

RESULTS
Purification and Properties of Rat SCL-Because our attempt to obtain crystals of SCL cloned from mouse liver (13) was unsuccessful, we constructed an expression system for rat SCL. The recombinant SCL was purified to homogeneity. Steadystate kinetic analysis revealed that V max and K m values for L-selenocysteine were 26 mol⅐min Ϫ1 ⅐mg Ϫ1 and 5.5 mM, respectively (supplemental Fig. S1). The enzyme showed no detectable activity toward L-cysteine (examined at 5-50 mM). An inhibition study showed that L-cysteine is a competitive inhibitor of the enzyme with K i ϭ 9.6 mM (supplemental Fig. S1). Thus, the fundamental properties SCL are similar to those reported for the enzymes from pig (11) and mouse (13).
Role of Cys-375-SCL has a conserved cysteine residue that corresponds to the persulfide-forming cysteine residue in cysteine desulfurases (supplemental Fig. S2) (16,35). To investigate the role of the conserved cysteine residue in SCL, we prepared a mutant SCL protein (C375A) in which Cys-375 was replaced by an alanine residue. The purified C375A mutant exhibited no activity to L-selenocysteine, suggesting that Cys-375 is a catalytically essential residue of the enzyme similar to the conserved catalytic cysteine residue of cysteine desulfurases (35).  (Fig. 1, A and  D). Incubation of the wild-type SCL with L-selenocysteine resulted in the formation of a new species with a mass of [M ϩ 80] (Fig. 1C). In contrast, incubation with L-cysteine resulted in no formation of such species (Fig. 1B). The data suggest   UV-visible Spectrum Analysis-SCL shows an absorption maximum at 416 nm, which is characteristic of the internal aldimine form of PLP enzymes. Incubation of SCL with L-selenocysteine resulted in a broad increase in the absorbance at ϳ420 nm ( Fig. 2A). In contrast, the addition of L-cysteine to the enzyme caused a significant decrease in the absorption peak at 416 nm, concomitantly with the increase in that at 350 nm (Fig.  2B). The incubation with D-cysteine, D-selenocysteine, or 2-mercaptoethanol also resulted in an absorption peak shift from 416 nm to 330 -340 nm in the same manner as observed in the enzyme incubated with L-cysteine (data not shown). These results indicate that L-cysteine, D-cysteine, D-selenocysteine, and 2-mercaptoethanol can enter the active site of the enzyme to affect the absorption properties of the PLP moiety even though they do not serve as a substrate of the enzyme. Interestingly, the C375A mutant enzyme incubated with L-selenocysteine or L-cysteine also showed a peak shift from 416 nm to 350 nm (Fig. 2, C and D). These results indicate that Cys-375 is essential for the proper capture of the substrate L-selenocysteine.

ESI-MS Analysis of Enzymebound Selenopersulfide
Overall Structure-The overall structure of SCL is shown in Fig. 3A. The polypeptide chain is folded into a dimeric form with each subunit consisting of the small (from N terminus to Leu-30 and from Thr-306 to the C terminus) and large (from Glu-31 to Glu-305) domains. The small domain has an ␣/␤ structure which consists of a two-stranded parallel ␤-sheet, three-stranded antiparallel ␤-sheet, and six helices (supplemental Fig. S3B). The large domain exhibits a typical ␣/␤ structure in which seven parallel ␤-strands (S2-S8), except for S8 form an open twisted ␤-sheet as a core surrounded by six ␣-helices from the solvent side and two ␣-helices from the protein side. The sevenstranded ␤-sheet forms the base of the active site that binds the coenzyme PLP, and is characteristic of PLP-dependent fold type I enzyme (14,36). The molecule has two active site cavities around a molecular 2-fold axis. Each cavity is located at the domain interface of one subunit and at the subunit interface.
The program DALI (37) was used to detect the enzymes possessing three-dimensional structures most similar to that of the unliganded SCL from among PLP-dependent fold type I enzymes in the Protein Data Bank data base. The highest Z scores were calculated to be 62.4 for human SCL (an as yet unpublished structure, PDB 2HDY), 49 (42). This result shows that the structure of SCL is quite similar to that of human SCL, and more similar to those of group I cysteine desulfurases (IscS and NifS-like protein) than those of group II cysteine desulfurases (SufS and CsdB) (35, 43) (Fig. 3B).
SCL changes its overall conformation from the open form to the closed form by a 3.8°rotation of the small domain as a rigid body upon binding of cysteine to the active site crevice to form the cofactor-cysteine complex (Fig. 3A). In addition to the rigid body rotation of the small domain, the cysteine binding induces the local conformational change of the extended lobe (the small domain residues from Ser-374 to Ile-392). The extended lobe of SCL in the unliganded open form was disordered and was not located on the electron density map. In contrast, the lobe in the liganded closed form was ordered to form the helix 11 (3 10 helix)-loop-␣-helix 12-loop structure which covers the active site as a lid. In contrast to SCL, the open-closed conformational change on ligand binding was not observed in cysteine desulfurases (17,18,38,39,44) (see Open-Closed Conformational Change in "Discussion").
Active Site Structure of SCL⅐L-Cysteine Complex-PLP is located at the bottom of the active site pocket formed at the subunit interface and the domain interface (Fig. 4A). The side chain S␥ atom of the bound cysteine, but not the ␣-amino group, makes a covalent bond with the C4Ј atom of PLP. The residual electron density of the S␥ atom of the bound cysteine was so close to C4Ј of PLP that the S␥ atom made a covalent bond with C4Ј (supplemental Fig. S3C). When the C4Ј-N bond between PLP and Lys-247 was refined as a double bond (Schiff base), C4Ј-N linkage deviated from the corresponding electron density compared with the Schiff base in the unliganded form. The C4Ј-N and C4Ј-S␥ were refined as single bonds for a good fit with the electron density, with distances of 1.51 and 1.86 Å, respectively. The bond angles around the C4Ј atom are 106, 111, and 112°, confirming the tetrahedral geometry of the C4Ј atom. The peak shift in the absorption spectrum from 416 to 350 nm observed in the SCL⅐L-cysteine complex crystal (supplemental Fig. S3A) and that in solution (Fig. 2) are thus due to the formation of the tetrahedral adduct of cysteine-PLP-Lys247 (supplemental Fig. S3D).
Upon binding of L-cysteine, the small domain moves toward the active site and the extended lobe (small domain residues Ser-374 to Ile-392), which is disordered in the unliganded form, shows its ordered structure, encapsulating L-cysteine within the active site cavity (Figs. 3 and 4A). The carboxylate of the bound cysteine makes a salt bridge with the guanidino group of Arg-402 and is further fixed by hydrogen bond interactions with Ser-374 (extended lobe) and Asn-186. The protonated ␣-amino group of the bound cysteine, which is supposed to form a Schiff base (external aldimine) with PLP in place of Lys-247 in cysteine desulfurases (15,16), is hydrogen-bonded to the side chain OH group of Ser-374, the main chain CϭO group of Ala-26, and water molecule W1.
Structure of SCL⅐Selenopropionate Complex-We solved the crystal structure of an SCL⅐selenopropionate complex. Selenopropionate is the substrate analog in which the ␣-amino group of the substrate L-selenocysteine is replaced by a hydrogen atom. There is no significant difference in the overall and active site fold between selenopropionate and the L-cysteine complex. The residual electron density showed that the bound selenopropionate is disordered over two sites such that the structure is comprised of two conformers, A and B (Fig. 4B). Conformer A is just like the bound L-cysteine in SCL⅐L-cysteine in that the Se␥ atom of selenopropionate forms a covalent bond with the C4Ј atom of the cofactor, although the distance of 2.28 Å is slightly longer than the normal Se-C bond length (1.98 -2.00 Å) (45). The geometry around the C4Ј atom is tetrahedral with bond angles of 109, 110, and 113°. On the other hand, conformer B directs its Se␥ atom toward the extended lobe and forms a hydrogen bond with the thiol group of the lobe residue Cys-375 (Se␥-S␥ ϭ 3.29 Å) because the selenol group (pK a of ϳ5.2) is deprotonated resulting in the selenolato-thiol interaction (46). The Se␥ atom further interacts with His-133 NE2 with an interatomic distance of 3.5 Å. The spectrum of SCL measured in solution in the presence of selenopropionate showed peaks at 350 and 416 nm, suggesting that selenopropionate is bound to the active site in the two conformers as observed in the SCL⅐selenopropionate crystal (supplemental Fig. S4).

DISCUSSION
Catalytic Mechanism-The C375A mutant of SCL does not catalyze the removal of selenium from L-selenocysteine. C375A⅐L-selenocysteine shows an absorption spectrum that is quite similar to that of the SCL⅐L-cysteine complex (Fig. 2), which suggests that L-selenocysteine forms the tetrahedral adduct with a mutant enzyme, such as L-cysteine bound to the wild-type enzyme. With consideration of the foregoing observations and this fact, we propose that the interaction of the thiol group of Cys-375 with the selenolate of the substrate selenocysteine brings the substrate L-selenocysteine into a favorable arrangement to form a Schiff base with PLP, as was observed in the conformer B of SCL⅐selenopropionate. SCL is designed to form the productive Michaelis complex by the thiol-selenolate interaction as the guide, although the nonproductive form might be in equilibrium with the productive form.
The productive L-selenocysteine is encapsulated in the active site by a small domain rotation and the ordering of the extended lobe with its Se␥ atom interacting with the thiol group of Cys-375. The deprotonated amino group of the substrate makes a nucleophilic attack on the C4Ј of PLP to produce the external aldimine and release the neutral side chain of Lys-247. The ␣-proton of the substrate is eliminated by Lys-247 to yield a quinonoid intermediate. Lys-247 adds a proton to the C4Ј of the quinonoid intermediate, producing the substrate-ketimine intermediate. The substrate-ketimine then transfers selenium to Cys-375 to form cysteine selenopersulfide (Cys375-S-Se(H)) and the enamine. Cys-375, which approaches and interacts with the substrate Se atom, plays a critical role in the formation of the selenopersulfide as well as the productive form.
It remains unclear whether selenium is directly released from Cys375-S-Se(H) or trapped by a seleniumtransferring protein and subsequently released by a reductant in the reaction system. If selenopersulfide selenium, which is more sensitive to oxygen than persulfide sulfur, is released directly from an intermediate and diluted by the bulk solvent, then delivery to a specific acceptor molecule would be inefficient. Plausibly, selenium is shipped to the target protein through interaction with the selenium-transferring protein in a manner similar to the reaction of cysteine desulfurases (16,35).
Open-Closed Conformational Change-Cysteine desulfurases thus far determined by x-ray methods exhibit no marked change in their overall or local conformation because the liganded forms have the same structures as those of the unliganded forms (17, 18, 38 -40). In contrast, upon binding of L-cysteine or selenocysteine, the disordered extended lobe of SCL in the unliganded form shows its ordered structure composed of the helix 11 (3 10 helix)-loop-␣-helix 12-loop to close the active site in concert with a 3.8°rotation of the small domain as a rigid body (Figs. 3 and 4). The accessible surface area of the bound L-cysteine was calculated to be 0 Å 2 , indicating that the open-closed conformational change completely shields bound ligands from the solvent region. The disorder-order transition of the extended lobe is deeply involved in the catalytic event in that the lobe residue Cys-375 is oriented favorably for interaction with the selenol group of the substrate and the oxygen-sensitive selenopersulfide is synthesized in the cavity shielded from the solvent region.
In contrast to SCL, the extended lobe of E. coli IscS (38) or the T. maritima NifS-like protein (39) (group I cysteine desulfurase) in the unliganded form is only one residue shorter than that of SCL, and is disordered in the region bearing the catalytic cysteine residue (Fig. 3B). Upon binding of cysteine to the T. maritima NifSlike protein, the lobe remains disordered and does not change its conformation. The accessible surface area of the bound cysteine is 49.3 Å 2 . E. coli CsdB (group II) has a shorter extended lobe than the group I enzymes. The lobe is ordered and maintains the same conformation regardless of whether it has the unliganded or liganded form. The accessible surface area of the bound L-propargylglycine is 10.2 Å 2 . These results suggest that the ligand bound to group I or II cysteine desulfurase is solvent accessible.
Discrimination between Selenium and Sulfur-Cysteine desulfurases, such as A. vinelandii NifS, E. coli IscS, and E. coli CsdB catalyze the elimination of the Se atom from L-selenocysteine as well as the S atom from L-cysteine (15,35,43,47,48). On the other hand, SCL is highly specific for L-selenocysteine, as described above and elsewhere (11,13). UV-visible spectrum analysis and the detailed active site structures of SCL⅐L-cysteine and SCL⅐selenopropionate uncovered the molecular basis of the specificity of SCL for L-selenocysteine.
The substrate L-selenocysteine binds to the active site in the productive form with the aid of the interaction of the substrate selenolato with Cys-375 thiol. This interaction allows the substrate to form the Schiff base with PLP resulting in the production of the external aldimine. On the other hand, the bound L-cysteine is not involved in the interaction with the thiol group of Cys-375, but the S␥ atom of the bound L-cysteine makes a covalent bond with the C4Ј of PLP to form a stable tetrahedral adduct with the ␣amino group far away from the C4Ј atom and the C␣-proton directed to the solvent side ( Fig. 4A and supplemental Fig. S3). Therefore, SCL does not catalyze L-cysteine desulfuration.
SCL acting on selenium but not on its sulfur counterpart is significant among the enzymes participating in selenoprotein biosynthesis: neither selenophosphate synthetase nor selenocysteine synthase shows strict specificity toward selenium compounds (49). Thus, SCL acts as a sorter for selenium from the mixture of selenium and sulfur in a biological system.