INTRODUCTION

Nitrite reductase (NIR)¹ from Alcaligenes faecalis S-6 is a copper-containing enzyme that catalyzes the reduction of NO₂ to NO in the denitrifying pathway in this bacterium (1, 2). Pseudoazurin, a 14-kDa protein containing a type 1 (blue) copper atom, has been isolated as a direct electron donor to NIR in vivo (3, 4). We have cloned genes of both pseudoazurin and NIR from A. faecalis S-6 and developed expression systems in Escherichia coli for site-directed mutagenesis of these proteins (5, 6). X-ray crystallographic analysis of pseudoazurin revealed its typical -barrel structure followed by -helices at the C terminus, binding a single type 1 copper atom at the top of the -barrel (7, 8). Our recent site-directed mutagenesis studies suggested that a ring of lysine residues close to the type 1 copper site of pseudoazurin is involved in the interaction with its redox partner, NIR (9). To further clarify the electron transfer between pseudoazurin and NIR, identification of amino acid residues of NIR that are involved in the interaction with pseudoazurin is now required.

Three-dimensional structures of NIRs from two sources, Achromobacter cycloclastes and A. faecalis S-6, were determined by x-ray crystallographic analyses (10, 11). The analyses revealed that both NIRs have the same trimeric structure with a type 1 copper atom and a type 2 copper atom in each subunit. The type 1 copper is bound inside domain 1 of the two -barrel subunits, and the type 2 copper is bound between two adjacent subunits. Using site-directed mutagenesis of selected ligands to these two types of copper atoms, we have shown that the type 1 copper plays a role as an efficient electron acceptor from pseudoazurin and the type 2 copper is involved in a catalytic mechanism of this enzyme (11). We therefore expected that the surface near the type 1 copper was involved in interaction with the lysine residues of pseudoazurin. In the present study, 10 negatively charged residues on the surface of NIR were chosen as candidates for site-directed mutagenesis and were replaced independently by alanine or serine. Glu-46, Glu-58, Glu-89, Glu-113, Glu-118, Glu-197, Asp-201, Glu-204, and Asp-205 are located on the surface close to the type 1 copper site, and Glu-160 is located on the opposite side of the molecule (Fig. 1). We report here the involvement of electrostatic interactions in forming an electron transfer complex composed of NIR and pseudoazurin and the identification of pairs of the amino acid residues responsible for the electrostatic interaction between these two copper-containing proteins.

EXPERIMENTAL PROCEDURES

E. coli strain JM105 ((lac pro) thi strA endA sbcB15 hsdR4 F traD36 proAB lacI^q lacZM15) was used as a host for the production of wild-type and altered NIRs. pNIR701 is a plasmid that was designed to give efficient production of NIR in the periplasmic space of E. coli as described previously (11).

Site-directed mutagenesis of Glu-46 to Ser, Glu-58 to Ala, Glu-89 to Ser, Glu-113 to Ala, Glu-118 to Ala, Glu-160 to Ala, Glu-197 to Ala, Asp-201 to Ala, Glu-204 to Ala, and Asp-205 to Ala of NIR was carried out by the Kunkel method (12) using oligonucleotides 5-ATGGTGATCAGCGAAAAGAAG-3, 5-GATGCGGGTACCGCAGTTCACGCC-3, 5-CTCATCAACCCGTCGACCAACACG-3, 5-CGGCGGGTTAACCGCAATCAATCCG-3, 5-ATCAATCCCGGGGCAAAGACCATC-3, 5-GGTGCTGCCCCGGGCGGGTCTGCAT-3, 5-AAATACGCGGCGCCCGGGGACGCTTAT-3, 5-GCGCCCGGGGCCGCTTATGAA-3, 5-CGGCGACGCGTATGCAGACACCGTC-3, and 5-CGGCGACGCGTATGAAGCCACCGTCCAA-3, respectively. Hereafter, these altered NIRs will be abbreviated as follows: E46S, E58A, E89S, E113A, E118A, E160A, E197A, D201A, E204A, and D205A, respectively. We also constructed another NIR, E118A/E197A/D201A, in which Glu-118, Glu-197, and Asp-201 were simultaneously replaced by Ala. DNA sequencing of all the mutated genes was carried out by the M13 dideoxy chain termination method (13, 14) to confirm the mutations introduced by the mutagenesis. Each of the mutated genes was inserted back into the corresponding portion of pNIR701 and then introduced in E. coli according to the method of Norgard et al. (15).

Wild-type and altered NIRs were purified from the periplasmic space of E. coli harboring pNIR701 as described previously (11). Most of the altered NIRs (E46S, E58A, E89S, E118A, E197A, D201A, E204A, D205A, and E118A/E197A/D201A) showed lower affinity to DEAE Toyopearl (TOSOH, Tokyo) and Mono Q FPLC (Pharmacia Biotech Inc.) columns compared with wild-type NIR. A. faecalis S-6 pseudoazurin and three altered pseudoazurins with replacement of Lys-10 by Ala, Lys-57 by Ala, and Lys-77 by Ala, abbreviated as K10A, K57A, and K77A, respectively, were purified as described previously (9, 16). Purity of each protein was analyzed by SDS-polyacrylamide gel electrophoresis (17).

Absorption spectra of altered NIRs were measured with a Spectronic 3000 Array (Milton Roy Co.) spectrophotometer. CD spectra of altered NIRs were measured with a Jasco J-720 spectrophotometer.

Nitrite reductase activities of altered NIRs were measured according to the method of Kakutani et al. (2) with dithionite and methyl viologen as an electron donor and a mediator, respectively. One unit of the enzyme activity was defined as the amount of enzyme that catalyzes the reduction of 1 µmol NO₂/min. The deviations in Table I were calculated from more than five observed values.

Table I. Enzyme activities of altered NIRs The enzyme activities were determined by the method of Kakutani et al. (2) as described under ``Experimental Procedures.'' NIRs Specific activities Relative activities units/mg protein Wild type 372  ± 10 100 E46S 299  ± 6 80 E58A 217  ± 10 58 E89S 206  ± 14 55 E113A 46  ± 2 12 E118A 95  ± 6 26 E160A 123  ± 1 33 E197A 174  ± 8 47 D201A 176  ± 10 47 E204A 116  ± 1 31 D205A 139  ± 2 37 E118A/E197A/D201A 63  ± 8 17

Kinetic analyses of wild-type or of altered NIRs with wild-type or altered pseudoazurins were carried out as described previously (11). Each reaction mixture contained 2 mM KNO₂ and reduced pseudoazurin with concentrations of 20-270 µM. An appropriate amount of NIR was then added to the reaction mixture, and the increase in absorbance at 593 nm due to the oxidation of pseudoazurin was monitored. The molar extinction coefficients at 593 nm of wild-type, K10A, K57A, and K77A pseudoazurins (2.9, 2.9, 2.9, and 2.8 mM¹ cm¹, respectively (9)) were used to calculate the kinetic parameters. Data were analyzed by Cleland's initial velocity program (18). Protein concentrations were determined by the method of Bradford (19).

RESULTS

All the altered NIRs were expressed in E. coli and purified to homogeneity on SDS-polyacrylamide gel electrophoresis (data not shown). In the oxidized state, wild-type NIR shows two intense absorption peaks at 462 and 583 nm along with a broad one at 690 nm in the optical spectrum, which is characteristic of the type 1 copper of this enzyme (11). When optical spectra of the altered NIRs were analyzed, no significant changes in the spectra were observed. This indicated that the environment of the type 1 copper site of each altered NIR was similar to that of wild-type NIR. To examine the structure of altered NIRs, we next measured CD spectra of each of the altered NIRs. All the altered NIRs showed CD spectra identical to that of wild-type NIR (data not shown), suggesting that overall structure was not distorted by the replacements.

Specific activities of the altered NIRs were determined using dithionite/methyl viologen as an electron donor to NIR. As described above, spectral analysis of these altered NIRs suggests that no gross change in both overall structure and environment of the type 1 copper site was caused by the replacements. However, most of the altered NIRs had reduced activities (Table I). E46S showed about 80% activity compared with wild-type NIR. E58A, E89S, E197A, and D201A possessed 50-60% activities, while E118A, E160A, E204A, and D205A showed 30-40% activities. The largest decrease in the enzyme activity was observed with E113A and E118A/E197A/D201A, which showed only 10-20% enzyme activities. In addition, E113A seemed to be unstable since the activity was further reduced during long storage of the enzyme at 4 °C (data not shown).

We determined next the apparent kinetic parameters of electron transfer between pseudoazurin and wild-type or altered NIRs by steady-state kinetics according to the method described under ``Experimental Procedures.'' As shown in Table II, the K_m values of several altered NIRs for pseudoazurin were significantly increased, suggesting decreased affinity in these altered NIRs; E118A, E197A, D201A, E204A, and D205A showed a 2.0-, 2.2-, 2.1-, 2.1-, and 2.3-fold increase in the K_m value, respectively. E113A showed a small increase in the K_m value, while no change in the K_m value was observed in the other altered NIRs (E160A, E46S, E58A, and E89S). Furthermore, multiple replacement of the three residues, Glu-118, Glu-197, and Asp-201, in the single subunit caused a further increase in the K_m value (13.9-fold). These results suggested that the negatively charged residues around the type 1 copper site, Glu-118, Glu-197, Asp-201, Glu-204, and Asp-205, are involved in the interaction with pseudoazurin. As for E113A, although we could not determine the exact values of the kinetic parameters due to low yield of this protein, the observed increase in the K_m value suggested the possibility that Glu-113 may also be involved in interaction with pseudoazurin.

Table II. Kinetic parameters of altered NIRs for wild-type pseudoazurin in the electron transfer NIRs kcat Km s1 µM Wild type 387  ± 8 50  ± 3 E160A 121  ± 2 56  ± 3 E46S 221  ± 7 41  ± 3 E58A 223  ± 10 43  ± 5 E89S 93  ± 2 49  ± 2 E113A 26  ± 3 73  ± 22 E204A 163  ± 7 106  ± 9 D205A 195  ± 12 115  ± 14 E118A 111  ± 3 101  ± 5 E197A 238  ± 16 108  ± 14 D201A 217  ± 19 104  ± 18 E118A/E197A/D201A 29  ± 5 692  ± 140

As described above, most of the replacements caused a decrease in NO₂ reducing activity, suggesting that the replacements had some effect on the type 2 copper site where NO₂ reduction occurs. Since the NO₂ reduction is the rate-limiting step in the steady-state assay system monitoring the overall reaction (20), a decrease in the k_cat value in the altered NIRs was anticipated in this study. As shown in Table II, the k_cat values of the altered NIRs with pseudoazurin were reduced by the replacements, similar to those in the assay measuring direct nitrite reduction as described above.

Our previous study revealed that several lysine residues surrounding the type 1 copper site of pseudoazurin are involved in the interaction with NIR (9). Furthermore, the involvement of glutamic and aspartic acid residues of NIR in the interaction with pseudoazurin has been shown in the present study. These observations suggest an electrostatic interaction between NIR and pseudoazurin. To identify specific electrostatic pairs between the two proteins, we next carried out the same kinetic analysis using the altered pseudoazurins, K10A, K57A, or K77A, which were already shown to possess increased K_m values compared with wild-type pseudoazurin (9). Although replacement of Lys-38 of pseudoazurin was also shown to cause an increase in the K_m value (9), we could not use K38A pseudoazurin for the kinetic analysis due to the low yield of this protein. Among the altered NIRs possessing the replacements that caused increases in the K_m values for pseudoazurin, three NIRs, E118A, E197A, and D201A, were chosen; and electron transfer between each of the altered pseudoazurins and the altered NIRs was analyzed in the same way as described above (Table III). In this assay system, it is possible to show putative pairs of charged residues that are involved in the electrostatic interaction by comparing the K_m values. When a replacement of two residues that do not interact with each other is introduced, a cumulative effect of the replacements is expected. On the other hand, simultaneous replacement of a pair of residues interacting with each other would cause no such additive effect on the apparent affinity. For example, the K_m value of NIR for pseudoazurin was increased 2.2-fold by the replacement of K10A pseudoazurin and was increased 2.0-fold by the introduction of E118A to NIR. When the kinetic analysis was carried out with E118A NIR and K10A pseudoazurin, a 3.9-fold increase in the K_m value was observed. This suggested no interaction between the two residues. On the other hand, the replacement of Glu-197 of NIR by Ala did not cause a further increase in the K_m value for K10A pseudoazurin, and the K_m value of E197A NIR for K10A pseudoazurin was almost the same as that of wild-type NIR for K10A pseudoazurin (2.0- and 2.2-fold increases, respectively). This suggests that Glu-197 of NIR could interact with Lys-10 of pseudoazurin in the electron transfer complex. By similar analysis, a possible interaction of Glu-118 of NIR with Lys-77 and/or Lys-57 is suggested.

Table III. Kinetic parameters of altered NIRs for pseudoazurins Pseudoazurins NIRs kcat Km s1 µM Wild type Wild type 387  ± 8 50  ± 3 E118A 111  ± 3 101  ± 5 E197A 238  ± 16 108  ± 14 D201A 217  ± 19 105  ± 18 K10A Wild type 414  ± 16 110  ± 7 E118A 49  ± 5 196  ± 33 E197A 131  ± 13 101  ± 11 D201A 135  ± 6 156  ± 13 K57A Wild type 440  ± 21 98  ± 11 E118A 65  ± 2 122  ± 7 E197A 106  ± 5 149  ± 13 D201A 99  ± 5 221  ± 17 K77A Wild type 552  ± 21 143  ± 9 E118A 56  ± 2 140  ± 9 E197A 128  ± 5 183  ± 12 D201A 106  ± 8 268  ± 33

DISCUSSION

In a previous study, we showed that replacement of four lysine residues (Lys-10, Lys-38, Lys-57, and Lys-77) that form a ring surrounding the type 1 copper site of pseudoazurin increased the K_m value of NIR, suggesting that these residues of pseudoazurin are necessary in docking to NIR and subsequent efficient electron transfer. These results suggest the involvement of electrostatic interaction between pseudoazurin and NIR. We previously showed that the type 1 copper atom of NIR served as a recipient of electrons from pseudoazurin (11). Therefore, residues around the type 1 copper site of NIR were deemed good candidates for mutagenesis. We chose 10 negatively charged residues (8 glutamic acids and 2 aspartic acids) located mainly on the surface close to the type 1 copper site of NIR, replaced each independently by a non-charged residue (alanine or serine), and analyzed their apparent kinetic parameters for pseudoazurin. When the effects of the replacement of the negatively charged residues were examined, apparent increases in the K_m value for pseudoazurin were observed for replacement by alanine of Glu-118, Glu-197, Asp-201, Glu-204, or Asp-205, which are close to the type 1 copper site of NIR. These results confirm that both copper proteins interact with each other through the regions close to their type 1 copper sites and strongly suggest electrostatic interactions in an electron transfer complex of the two copper-containing proteins.

When the amino acid sequences of copper-containing NIRs from A. cycloclastes (21) and A. faecalis S-6 (6) are compared, they show that Glu-118, Glu-197, Glu-204, and Asp-205 are conserved, and the other position, 201, is occupied by an acidic residue, Glu or Asp. This suggests that the recognition of positively charged residues of pseudoazurin by negatively charged ones of NIR may be a common feature for the nitrite-reducing complex. In Pseudomonas aureofaciens NIR, however, only Asp-205 is conserved, and the other residues are occupied by non-polar residues (22). It is notable that the NIRs from A. cycloclastes (23) and A. faecalis S-6 (2) are green in color, whereas the NIRs from P. aureofaciens (24) and Alcaligenes xylosoxidans (25) are blue. For the former NIRs, pseudoazurins were found as the redox partners (3, 26), while azurin serves as a redox partner for the latter NIRs (24, 27). Both pseudoazurins from A. faecalis S-6 and A. cycloclastes have a lysine ring around their type 1 copper sites, and all the lysine residues that were shown to be involved in the interaction with NIR are conserved in both pseudoazurins (8), whereas azurin has no such ring on the surface (28). Although both NIR with pseudoazurin and NIR with azurin pairs function in comparable dissimilatory nitrite reduction systems, the interaction of blue NIR with its redox partner may be different from the interaction of green NIR with pseudoazurin.

Each replacement of 10 negatively charged residues on the surface of the molecule caused reduced specific activity for nitrite reduction, although no gross change in the overall structure was observed by spectral analyses. Since it is found that depletion of the type 2 copper of NIR from A. cycloclastes causes loss of the enzyme activity (29), one might suppose that the low activities observed in the altered NIRs could be due to the lower type 2 copper contents of these enzymes. However, when we measured copper content by plasma emission spectroscopy, all of the altered NIRs were found to possess 1.5-2.0 coppers/monomer, similar to wild-type NIR (data not shown). The low activities in the altered NIRs, therefore, could not be attributed to low occupancy of copper in the type 2 site. We now speculate that a slight conformational change introduced by the replacements on the surface of the molecule would have a long distance effect on the environment of the active site, which is located in the subunit-subunit interface. In order to understand the effect of the replacements on the subunit interaction, determination of three-dimensional structures of those altered NIRs will be required.

Kinetic analysis of altered NIR with several altered pseudoazurins in various combinations suggested that Glu-197 of NIR interacts with Lys-10 of pseudoazurin, and Glu-118 of NIR interacts with Lys-77 and/or Lys-57. We propose here a model for an electron transfer complex between pseudoazurin and NIR (Fig. 2A). Fig. 2B illustrates the NIR and pseudoazurin interaction surface. As shown in the figure, when Glu-197 interacts with Lys-10, Glu-118 can be placed at the position close enough for the electrostatic interaction to Lys-77 and/or Lys-57. In this orientation, Lys-38 of pseudoazurin, previously shown to have a role in the interaction to NIR, lies in a position capable of interacting with Glu-204 and/or Asp-205. The suggested orientation of pseudoazurin relative to NIR positions the molecule so that the type 1 coppers in each would be 14-15 Å apart. However, the northern histidines of each are not directly apposed as we once thought. Intriguingly, His-81 (pseudoazurin) is within hydrogen-bonding distance of the carbonyl oxygen of residue Thr-92 (NIR), three residues away from the copper ligand His-95. This is similar to a pathway seen in the cupredoxin-like subunit II of Paracoccus cytochrome oxidase, the initial recipient of electrons from cytochrome c. In this subunit, His-224, a ligand to the binuclear Cu_A site, lies within hydrogen-bonding distance of a carbonyl oxygen (Arg-473) of subunit I. Arg-473 is in turn hydrogen-bonded to the propionate of heme a₃, and the peptide containing the carbonyl is hydrogen-bonded to a propionate of heme a, thus providing electron transfer pathways to both of the other redox centers in this protein (30). Both our proposed model and the cytochrome oxidase subunit II structure are consistent with a recent study providing direct evidence that hydrogen bonds are more important than previously believed in electron transfer pathways (31). Clearly, a structure of a complex of NIR and pseudoazurin would clarify our model. Site-directed mutagenesis in this study revealed involvement of electrostatic interaction between these two copper-containing proteins for the optimal electron transfer. The complex presented here will be a good guide, for the next site-directed mutagenesis, to understand the mechanism of intermolecular electron transfer.