Structural Evidence of Substrate Specificity in Mammalian Peroxidases

The crystal structure of the complex of lactoperoxidase (LPO) with its physiological substrate thiocyanate (SCN–) has been determined at 2.4Å resolution. It revealed that the SCN– ion is bound to LPO in the distal heme cavity. The observed orientation of the SCN– ion shows that the sulfur atom is closer to the heme iron than the nitrogen atom. The nitrogen atom of SCN– forms a hydrogen bond with a water (Wat) molecule at position 6′. This water molecule is stabilized by two hydrogen bonds with Gln423 Nϵ2 and Phe422 oxygen. In contrast, the placement of the SCN– ion in the structure of myeloperoxidase (MPO) occurs with an opposite orientation, in which the nitrogen atom is closer to the heme iron than the sulfur atom. The site corresponding to the positions of Gln423, Phe422 oxygen, and Wat6′ in LPO is occupied primarily by the side chain of Phe407 in MPO due to an entirely different conformation of the loop corresponding to the segment Arg418–Phe431 of LPO. This arrangement in MPO does not favor a similar orientation of the SCN– ion. The orientation of the catalytic product OSCN– as reported in the structure of LPO·OSCN– is similar to the orientation of SCN– in the structure of LPO·SCN–. Similarly, in the structure of LPO·SCN–·CN–, in which CN– binds at Wat1, the position and orientation of the SCN– ion are also identical to that observed in the structure of LPO·SCN.

eosinophil peroxidase (3), thyroid peroxidase (4), and myeloperoxidase (MPO) (5). LPO, eosinophil peroxidase, and MPO are responsible for antimicrobial function and innate immune responses (6 -8), whereas thyroid peroxidase plays a key role in thyroid hormone biosynthesis (9). These peroxidases are different from plant and fungal peroxidases because unlike plant and fungal enzymes, the prosthetic heme group in mammalian peroxidases is covalently linked to the protein (10). There are also several striking structural and functional differences among the mammalian peroxidases (11). The heme group in MPO is attached to the protein via three covalent linkages (12), whereas LPO (12,13), eosinophil peroxidase (12), and thyroid peroxidase (12) contain only two ester linkages. These covalent and various non-covalent linkages contribute differentially to the high stability of the heme core as well as for the peculiar values of their redox potentials (2,14). Furthermore, MPO consists of two disulfide-linked protein chains, whereas LPO, eosinophil peroxidase, and thyroid peroxidase are single chain proteins, although their chain lengths differ greatly. In addition, their sequences contain several critical amino acid differences that may also contribute to the variations in the stereochemical environments of the substratebinding sites. As a consequence of these differences, the mammalian enzymes oxidize various inorganic ions such as SCN Ϫ , Br Ϫ , Cl Ϫ , and I Ϫ with differing specificities and potencies. Biochemical studies have shown that LPO catalyzes preferentially the conversion of SCN Ϫ to OSCN Ϫ (15,16), whereas MPO uses halides (17,18) with a preference for chloride ion as the substrate. The preferences of eosinophil peroxidase and thyroid peroxidase are bromide and iodide, respectively. However, the stereochemical basis of the reported preferences for the substrates by mammalian heme peroxidases is still unclear. So far, the structures of only two mammalian enzymes, MPO and LPO, have been determined (12,13). It is of considerable importance to identify the structural parameters that are responsible for the subtle specificities. In the present work, we have attempted to address this question through the new crystal structures of LPO complexes with SCN Ϫ ions using goat, bovine, and buffalo lactoperoxidases. Because the overall structures of complexes of SCN Ϫ with LPO from all three species were found to be identical, the structure of the complex of buffalo LPO with SCN Ϫ and the ternary complex with SCN Ϫ and CN Ϫ will be discussed here, and buffalo LPO will be termed hereafter as LPO. To highlight the factors pertaining to binding specificity of SCN Ϫ , a comparison of the structures of LPO⅐SCN Ϫ and MPO⅐SCN Ϫ has also been made, revealing many valuable differences pertaining to the observed orientations of the common substrate, SCN Ϫ ion, when bound at the substrate-binding site in the distal heme cavity of the two structures. The structures of LPO⅐SCN Ϫ and MPO⅐SCN Ϫ clearly show that the bound SCN Ϫ ions are present in the distal heme cavity of two enzymes with opposite orientations. In the structure of LPO⅐SCN Ϫ , the sulfur atom is closer to the heme iron than the nitrogen atom, whereas in that of MPO⅐SCN Ϫ , the nitrogen atom is closer to the heme iron than the sulfur atom. As a result of this, the interactions of the SCN Ϫ ion in the distal site of two proteins differ drastically. Gln 423 , a conserved water (Wat) molecule at position 6Ј, and a well aligned carbonyl oxygen of Phe 422 in the proximity of the substrate-binding site in LPO against a protruding Phe 407 in MPO seem to play the key roles in inducing the observed orientations of SCN Ϫ ions in LPO and MPO. The structure of LPO⅐SCN Ϫ has also been compared with the structure of its ternary complex with SCN Ϫ and CN Ϫ ions.

EXPERIMENTAL PROCEDURES
Purification of the Protein-Lactoperoxidase-catalyzed oxidation of SCN Ϫ in milk and saliva contributes to the antimicrobial activity of these fluids (19). Lactoperoxidase was isolated from milk samples collected from Murrah buffaloes (Bubalus bubalis) available at the Indian Veterinary Research Institute (Izatnagar, India). The fat was separated from the fresh milk by skimming. Buffer containing 50 mM Tris-HCl, pH 8.0, and 2 mM CaCl 2 was added to the skimmed milk. The cation exchanger CM-Sephadex C-50 (7 g liter Ϫ1 ) (GE Healthcare, Uppsala, Sweden) equilibrated in 50 mM Tris-HCl, pH 7.8, was added and stirred slowly with a glass rod for ϳ1 h. The gel was kept overnight undisturbed at 280 K. It was decanted on the next day. The unbound proteins were removed by washing the gel with an excess of 50 mM Tris-HCl, pH 8.0. The washed gel was loaded on a CM-Sephadex C-50 column (10 ϫ 2.5 cm) and equilibrated with 50 mM Tris-HCl, pH 8.0. The protein samples were eluted using a linear gradient of 0.0 -0.5 M NaCl with the same buffer. The protein fractions eluted at 0.2 M NaCl were pooled, desalted, and concentrated using an Amicon ultrafiltration cell. This was loaded on a Sephadex G-100 column (100 ϫ 2 cm) using 50 mM Tris-HCl buffer, pH 8.0. This was eluted using the same buffer at a flow rate of 6.0 ml/h. The various protein fractions were collected and pooled separately. These protein samples were examined on SDS-PAGE. The fractions corresponding to an approximate molecular mass of 68 kDa were pooled, lyophilized, and stored at Ϫ20°C for further analysis. The N-terminal sequence of the first 20 amino acid residues was also determined using Edman degradation with Protein Sequencer PPSQ-21A (Shimadzu, Japan).
LPO Activity Measurements-The activity assay was carried out following the procedure of Shindler and Bardsley (20) with some modifications to suit certain requirements. 3.0 ml of 1 mM 2,2-azinobis(3-ethylbenzthiazolinesulfonic acid) (ABTS) in 0.1 M phosphate buffer, pH 6.0, was mixed with 0.1 ml of sample in 0.1 M phosphate buffer, pH 7.0, containing 0.1% gelatin to initialize the spectrophotometer (PerkinElmer Life Sciences).
3.0 ml of 1 mM ABTS solution was mixed with 0.1 ml of protein sample and 0.1 ml of 3.2 mM hydrogen peroxide solution. The absorbance was measured at 412 nm as a function of time for 2 to 3 min. The rate of change of absorbance was constant for at least 2 min. 1 unit of activity is defined as the amount of enzyme catalyzing the oxidation of 1 mol of ABTS min Ϫ1 at 293 K (molar absorption coefficient 32,400 M Ϫ1 cm Ϫ1 ). The activity of lactoperoxidase was found to be 5.3 units ml Ϫ1 . The purity of LPO was also determined by absorbance ratio A 412 /A 280 (RZ value). The RZ value for the purified LPO was found to be 0.932. In addition to this, two separate binding experiments were also carried out using buffers containing 50% methanol as the first condition and using 3 M NaCl as the second condition.
Crystallization of LPO and Its Complexes-The purified samples of LPO were dissolved in 0.01 M phosphate buffer, pH 6.0, containing 2 mM CaCl 2 to a concentration of 25 mg/ml. A reservoir solution consisting of 0.2 M ammonium iodide and 20% (w/v) polyethylene glycol 3350 was prepared. 3 l of protein solution was mixed with 3 l of reservoir solution to prepare 6 l of drops for hanging drop vapor diffusion method. The rectangular dark brown-colored crystals of LPO measuring up to 0.3 ϫ 0.3 ϫ 0.2 mm 3 were obtained after 1 week. The crystals of the complex of LPO with SCN Ϫ were prepared by soaking the crystals of native LPO in the mother liquor composed of 20% (w/v) polyethylene glycol 3350, 0.01 M sodium phosphate, 2 mM CaCl 2 , 50 mM NaSCN at 24°C at pH 6.0 for 48 h. The crystals of the complex of LPO with the SCN Ϫ ion were soaked in the mother liquor containing 20% (w/v) polyethylene glycol 3350, 0.01 M sodium phosphate, 2 mM CaCl 2 , 50 mM sodium cyanate (NaCN) at 25°C at pH 6.0 for 48 h.
Detection of SCN Ϫ Ions in Crystals-To confirm the presence of the SCN Ϫ ion in the soaked crystals of LPO, the crystals were washed thoroughly with distilled water. The washed crystals were crushed. This solution was incubated with 1 M NaCl for 1 h and then ultrafiltered using a membrane with a molecular mass cutoff of 1 kDa. The presence of SCN Ϫ ions in the filtrate was detected by colorimetric determination of SCN Ϫ ions in the sample (21). To make ferric thiocyanate, the ferric nitrate reagent was prepared by dissolving 1 g of ferric nitrate crystals in 10 ml of Milli-Q water, to which 10 ml of concentrated nitric acid was added for making the final volume to 200 ml. 2 ml of the filtrate was mixed with equal volume of the ferric nitrate reagent. The resulting appearance of red color confirms the presence of SCN Ϫ ions in the crystal filtrate.
Spectroscopic Analysis of LPO⅐SCN Ϫ Crystal-The spectral changes in the heme absorption at 412 nm were also recorded spectrophotometrically (PerkinElmer Life Sciences spectrophotometer). To determine the presence of SCN Ϫ ions in the crystals, three different solutions were made. Solution 1 contained the dissolved protein crystals. Solution 2 was obtained after removing the small molecular contents from solution 1 after incubating it with 1 M NaCl and ultrafiltrating it using a membrane with a molecular mass cutoff of 1 kDa. Solution 3 was prepared by mixing purified protein and potassium thiocyanate. The absorption spectra were recorded on all the three samples using a wavelength of 412 nm.
X-ray Intensity Data Collection-The x-ray intensity data were collected at 285 K using a 345-mm diameter MAR Research dtb imaging plate scanner mounted on a Rigaku RU-300 rotating anode x-ray generator operating at 100 mA and 50 kV. The CuK ␣ radiation was obtained using Osmic Blue confocal optics. The intensities were processed using programs DENZO and SCALEPACK (22). The space group was found to be monoclinic P2 1 with approximate cell dimensions of a ϭ 54.2, b ϭ 80.5, and c ϭ 77.3 Å and ␤ ϭ 102.7°. The final data set shows an overall completeness of 98.6% for resolution to 2.4 Å. The summary of final data collection statistics are given in Table 1.
Structure Determination and Refinement-The crystal structures of LPO⅐SCN Ϫ (bovine), LPO⅐SCN Ϫ (buffalo) and LPO⅐SCN Ϫ ⅐CN Ϫ (buffalo) were determined with molecular replacement method using the principle of maximum likelihood in PHASER (23). The native structure of LPO (Protein Data Bank Code 2R5L) was used as the search model (13). The rotation and translation search functions were computed using reflections in the resolution range of 12.0 -4.0 Å. This yielded clear solutions with distinct peaks in each case. The molecular packing in the unit cell calculated using coordinates from these solutions did not produce unfavorable short contacts. The coordinates were transferred using PHASER (23) and were subjected to 25 cycles of rigid body refinement with REFMAC (24) from the CCP4i Version 4.2 program package (25). After the first rounds of refinement, the R work and R free factors reduced to the range between 0.30 and 0.33 and between 0.40 and 0.41, respectively. (5% of the reflections were used for the calculation of R free ; the reflections were not included in the refinement.) The initial maps showed good electron densities for the heme groups. The coordinates of the heme groups were included in further rounds of refinement that were carried out with intermittent manual model building of the protein using Fourier ͉2F o Ϫ F c ͉ and difference Fourier ͉F o Ϫ F c ͉ maps with graphics program O (26) and Coot (27) on a Silicon Graphics O 2 workstation. In the refinement calculations, the ligand-Fe restraints were not used. At the end of these refinements, the R work and R free factors converged to ϳ0.248 and 0.278, respectively, for the structures of LPO⅐SCN Ϫ (bovine), LPO⅐SCN Ϫ (buffalo), and LPO⅐SCN Ϫ ⅐CN Ϫ (buffalo). The difference Fourier ͉F o Ϫ F c ͉ maps calculated at this stage revealed interpretable electron densities for glycan chains at four sites with 2 GlcNAc residues and 1 Man at Asn 95 , 2 GlcNAc residues at Asn 205 , 2 GlcNAc residues ϩ 1 Man residue at Asn 241 , and 2 GlcNAc residues at Asn 332 . The pear-shaped electron densities for SCN Ϫ in both complexes of LPO⅐SCN Ϫ and an additional linear electron density in the structure of LPO⅐SCN Ϫ ⅐CN Ϫ were observed in the distal cavity (Figs. 1 and 2). An identical pear-shaped electron density was observed in the structure of LPO⅐SCN Ϫ (bovine) (supplemental Fig. S1). The electron densities for one calcium ion in each structure were also observed. The difference electron density maps also revealed additional electron densities for seven iodide ions in each structure. All of these were included in the subsequent rounds of refinements. The difference Fourier  Table 1. The refined atomic coordinates have been deposited in the Protein Data Bank with codes 3ERI, 3ERH, and 3FAQ, respectively.   well for the samples of crystal and protein solution with SCN Ϫ ions. However, the measurements carried out on the SCN Ϫ ion-free samples showed different values, and the absorption curve did not superimpose on spectral curves obtained for the sample containing SCN Ϫ ions. These experiments clearly indicated that the SCN Ϫ ions are present in the crystals of the LPO enzyme.

SCN
Overall Protein Structure-The crystallographic parameters and refinement statistics for the crystal structure of the complex of lactoperoxidase with thiocyanate (LPO⅐SCN Ϫ ) are given in Table 1. The structure determination clearly revealed that a thiocyanate ion is present in the distal heme cavity with the sulfur atom of the SCN Ϫ ion being nearer to the heme iron atom than its nitrogen atom. The overall structural organization of the complex of LPO⅐SCN Ϫ is shown in Fig. 3. One of the most remarkable features of the structure of the LPO⅐SCN Ϫ complex is the placement of the SCN Ϫ ion, which is supported by favorable conformation of the loop Arg 418 -Phe 431 . The structure of the corresponding loop is strikingly different in MPO (Fig. 4). This loop constitutes one of the walls of the substrate-binding site. The structure of this rigid loop is stabilized by a salt bridge between His 426 and Glu 130 from helix H 2 a. The helix H 2 a is a unique feature of LPO. The iron atom is displaced approximately by 0.1 Å toward proximal His 351 from the plane made by four heme nitrogen atoms. The four pyrrole rings of the heme group are planar, whereas the heme group as a whole  is slightly non-planar. The bond angles at the iron atom, 208 N1A-Fe-N1C and 208 N1B-Fe-N1D, are 165.8°and 166.2°, respectively, indicating that heme iron atom is equally of the lines of N1A-N1C and N1B-N1D.
Binding of SCN Ϫ -The LPO enzyme catalyzes the peroxidation of thiocyanate in the presence of H 2 O 2 . In order for the reaction to occur, the SCN Ϫ ion must bind to LPO at the sub-strate-binding site in the distal heme cavity with a favorable orientation. The most important question arises here as to how the stereochemical environment around the heme moiety in LPO determines the preference for the SCN Ϫ ion so as to make it a preferred substrate. The structure determination has shown an excellent pear-shaped electron density at the distal site, indicating optimum conditions for binding. The observed density at one end corresponds to nearly 9, whereas at the other end, it is equivalent to the 2 cutoff (Fig. 1). The heme environment in the native LPO is characterized by a unique water structure at the distal heme cavity. The binding of different ligands occurs by displacing appropriate water molecules for the ligands to act either as a substrate or as an inhibitor. The positions of conserved water molecules Wat 1 , Wat 2Ј , Wat 3Ј , Wat 4Ј , Wat 5Ј , and Wat 6Ј (Fig. 5a)    that such an array of water molecules may function as a proton transfer chain (28 -31). It is expected that ligand binding in the distal site either displaces or deletes one or more conserved water molecules. In the present complex, the binding of thiocyanate in the distal site results in the deletion of two water molecules, Wat 4Ј and Wat 5Ј , whereas Wat 2Ј and Wat 3Ј are shifted away considerably. The observed orientation of the SCN Ϫ ion in the distal site shows that the sulfur atom is placed toward the heme iron at a distance of 3.2 Å from Wat 1 and 4.2 Å from His 109 N ⑀2 , whereas the nitrogen atom is located away from Wat 1 and forms a hydrogen bond with Gln 423 N ⑀2 through a conserved, well stabilized water molecule, Wat 6Ј (Fig. 5b). In addition to interactions with Gln 423 N ⑀2 and the nitrogen atom of SCN Ϫ , Wat 6Ј is also hydrogen-bonded to Phe 422 oxygen. In another recent structure of the lactoperoxidase complex with its product OSCN Ϫ (LPO⅐OSCN Ϫ ), the orientation of the OSCN Ϫ ion was also found similar to that of the SCN Ϫ ion, as observed in the present LPO⅐SCN Ϫ complex where the sulfur atom is closer to heme iron than the nitrogen atom ( Fig. 5c) (32). A similar orientation of the SCN Ϫ ion has also been reported in the structure of octaheme tetrathionate reductase (33). An identical arrangement for the SCN Ϫ ion in the distal cavity was also observed in the structure of the complex of LPO⅐SCN Ϫ (bovine) (supplemental Fig. S2). The superimposition of distal heme cavities of LPO⅐SCN Ϫ (buffalo) and LPO⅐SCN Ϫ (bovine) shows a root mean square shift of 0.05 Å for 124 atoms (supplemental Fig. S3). These results are in striking contrast to the observation made in the MPO⅐SCN Ϫ complex where the nitrogen atom was reported to be closer to the heme iron, whereas the sulfur atom was placed farther away (Fig. 5d). It also should be mentioned that NMR relaxation methods (34 -36) indicate that in the LPO⅐SCN Ϫ complex, the nitrogen atom is closer to the heme iron than the carbon. However, the reported distances in the two studies were unusually different for drawing a definite conclusion. It is also noteworthy here that the plane of the SCN Ϫ ion is almost parallel to the heme plane in the LPO⅐SCN Ϫ complex with a 9.5°angle between the plane of the heme and the SCN Ϫ ion. It is pertinent to note that the substrate-binding pocket in the distal heme cavity is surrounded by the heme moiety on one side, whereas the side chain of Arg 255 with C ␤ and C ␥ atoms forms the opposite hydrophobic side. The third side consists of a conserved water molecule (Wat 6Ј ), Gln 423 , Phe 422 oxygen, Pro 424 , and Phe 381 , and the fourth side is made by Gln 105 and His 109 with other conserved water molecules of the distal site, including Wat 1 . The observed position of the SCN Ϫ ion is delicately balanced by van der Waals forces from the nearest heme moiety atoms on one side and C ␤ and C ␥ atoms of Arg 255 on the other. The plane of heme moiety is nearly planar, and it is nearly parallel to the C ␤ -C ␥ bond of Arg 255 . The conserved water molecule stabilized by interactions with Gln 423 N ⑀2 , and Phe 422 oxygen locks the position of the nitrogen atom of the SCN Ϫ ion. This is also supported by the side chains of Pro 424 and Phe 381 . This shows that the stereochemistry in LPO appears to be specific for the binding of the SCN Ϫ ion in the distal heme cavity. In contrast, a similar environment is not available in MPO. The most notable structural element in LPO seems to be the conformation of the Arg 418 -Phe 431 loop in LPO, which is strikingly different from that of the corresponding loop in MPO. This loop in MPO adopts a conformation that places Phe 407 at the site occupied by the side chain of Gln 423 and the conserved water molecule Wat 6Ј . Therefore, the orientation of the SCN Ϫ ion in MPO is bound to differ from that of LPO. This seems to have evolved to attain substrate specificity in mammalian peroxidases. It is also noteworthy that the SCN Ϫ plane is inclined at a much larger angle (70.4°in molecule A and 35.2°in molecule B in the structure of the MPO dimer) in the MPO⅐SCN Ϫ complex (37). The fact that the SCN Ϫ ion is inclined at different angles in the two molecules of the MPO dimer is itself an indication of low grade binding specificity for the SCN Ϫ ion in MPO. The corresponding angle between the planes of OSCN Ϫ and the heme moiety in the complex of LPO⅐OSCN Ϫ was reported to be 21.5° (32). The plane of the SCN Ϫ ion in the LPO⅐SCN Ϫ complex is shifted away from the line of His 351 -Fe-His 109 by ϳ3.8 Å, whereas in the complex of MPO⅐SCN Ϫ , it is at a distance of 3.3 Å.
Comparison with the Structure of LPO⅐SCN Ϫ ⅐CN Ϫ -The crystals of LPO⅐SCN Ϫ were soaked in the mother liquor containing 100 mM NaCN. The presence of CN Ϫ in the crystals was verified by spectroscopic and chemical methods. 5 The x-ray intensity data on the crystals of LPO⅐SCN Ϫ ⅐CN Ϫ were collected to 2.7 Å resolution, and the structure was refined to R work and R free factors of 0.189 and 0.238, respectively (Protein Data Bank Code 3FAQ). The major features of the difference Fourier maps were examined in the vicinity of heme moiety. An elongated positive peak at the 3.0 cutoff on the distal side of the heme iron at the position of the conserved water molecule Wat 1 was interpreted as cyanide binding to the heme iron (Fig. 2). It also showed a pear-shaped electron density for the SCN Ϫ ion (Fig.  2). The refined structure shows a heme iron-to-cyanide carbon distance of 1.9 Å. It is considerably shorter than the distance of the heme iron to the water oxygen atom of 2.6 Å in the LPO⅐SCN Ϫ complex. The C-N bond in the cyanide ion is almost perpendicular to the heme plane with a Fe-C-N bond angle of 161.0°. The position of the SCN Ϫ ion in the present structure is identical to that observed in the structure of LPO⅐SCN Ϫ . However, the plane of the SCN Ϫ ion is not as parallel (inclination, 51°) as that found in the complex of LPO⅐SCN Ϫ (inclination, 10°), indicating a slight adjustment upon binding of the cyanide ion. This is in striking contrast to the binding of the SCN Ϫ ion in MPO, where the planes of the heme moiety and SCN Ϫ ion are considerably more inclined in the MPO⅐SCN Ϫ complex (Protein Data Bank Code 1DNU) (inclination, 70°) than in the ternary complex of MPO⅐SCN Ϫ -CN Ϫ (inclination, 15°) (Protein Data Bank Code 1DNW). The orientations of the SCN Ϫ ion, with the sulfur atom of SCN Ϫ being closer to the heme iron than its nitrogen atom in both structures of LPO⅐SCN Ϫ and LPO⅐SCN Ϫ ⅐CN Ϫ , are found to be identical. It may also be mentioned here that the OSCN Ϫ in the complex of LPO⅐OSCN Ϫ (32) has a similar orientation as in the present structure. bound SCN Ϫ ion at the distal site may restrict the accessibility of the binding site to H 2 O 2 . Additionally, the SCN Ϫ ion interacts with heme water molecule Wat 1 , thus enhancing the binding affinity of water. As a result, it would be more difficult for H 2 O 2 to displace Wat 1 . The above two factors will have adverse effects for the affinity of H 2 O 2 , resulting in reducing the rate of reaction. However, the stereochemical characteristics of the substrate-binding site in the resting LPO allow SCN Ϫ to bind with favorable orientation in LPO, whereas in MPO, SCN Ϫ is held with unfavorable orientation. Therefore, this will also adversely affect the efficiency of the catalytic function of MPO for the substrate SCN Ϫ . In summary, the results of these investigations clearly establish the structural preference of the SCN Ϫ ion as a good substrate for LPO and mediocre substrate for MPO.