The Nitric-oxide Reductase from Paracoccus denitrificans Uses a Single Specific Proton Pathway*

Background: NO reductase (NOR) takes up protons from the opposite side of the membrane compared with other heme-copper oxidases. Results: NOR is sensitive to mutations along the suggested proton pathway 1 but not the others. Conclusion: Only pathway 1 is used for proton transfer. Significance: Although no energy is conserved, proton transfer still occurs through a specific pathway. The NO reductase from Paracoccus denitrificans reduces NO to N2O (2NO + 2H+ + 2e− → N2O + H2O) with electrons donated by periplasmic cytochrome c (cytochrome c-dependent NO reductase; cNOR). cNORs are members of the heme-copper oxidase superfamily of integral membrane proteins, comprising the O2-reducing, proton-pumping respiratory enzymes. In contrast, although NO reduction is as exergonic as O2 reduction, there are no protons pumped in cNOR, and in addition, protons needed for NO reduction are derived from the periplasmic solution (no contribution to the electrochemical gradient is made). cNOR thus only needs to transport protons from the periplasm into the active site without the requirement to control the timing of opening and closing (gating) of proton pathways as is needed in a proton pump. Based on the crystal structure of a closely related cNOR and molecular dynamics simulations, several proton transfer pathways were suggested, and in principle, these could all be functional. In this work, we show that residues in one of the suggested pathways (denoted pathway 1) are sensitive to site-directed mutation, whereas residues in the other proposed pathways (pathways 2 and 3) could be exchanged without severe effects on turnover activity with either NO or O2. We further show that electron transfer during single-turnover reduction of O2 is limited by proton transfer and can thus be used to study alterations in proton transfer rates. The exchange of residues along pathway 1 showed specific slowing of this proton-coupled electron transfer as well as changes in its pH dependence. Our results indicate that only pathway 1 is used to transfer protons in cNOR.

Cytochrome c-dependent NO reductase (cNOR) 5 from Paracoccus denitrificans is an integral membrane protein complex that reduces nitric oxide to nitrous oxide (2NO ϩ 2H ϩ ϩ 2e Ϫ 3 N 2 O ϩ H 2 O). Nitric oxide reductases (NORs) are mostly found in denitrifying bacteria that stepwise reduce NO 3 Ϫ to N 2 gas (via NO 2 Ϫ , NO and N 2 O). NORs are also expressed in various pathogenic non-denitrifying bacteria in order to inactivate the toxic NO that is produced by the immune system of the host. NORs are members of the heme-copper oxidase (HCuO) superfamily. Most members of this superfamily reduce O 2 to H 2 O and conserve the liberated free energy by pumping protons across the membrane, thus maintaining an electrochemical gradient. In addition, electrons are donated from the (positive, lower pH) outside, and protons needed for water formation are taken up strictly from the (negative, higher pH) inside. For this purpose, the O 2 -reducing HCuOs use defined proton transfer pathways (one or two) from the cytoplasm into the active site and a route for pumped protons to the outside that is less well defined (for recent reviews on cNOR and proton transfer pathways in the HCuO superfamily, see Refs. 1 and 2). cNOR, on the other hand, has been shown not to contribute to the electrochemical proton gradient, although the amount of free energy available from NO reduction is similar to that for O 2 reduction (3)(4)(5). Electrons are donated by soluble carriers in the periplasm, so for the overall reaction to be non-electrogenic, protons used for NO reduction must also originate in the periplasmic (outside) solution.
cNOR is isolated as a complex of two subunits, NorB and NorC (see Fig. 1A). NO reduction takes place in the NorB subunit that contains three redox centers: two b-type hemes (hemes b and b 3 ) and a non-heme iron (Fe B ). Heme b 3 and the non-heme iron form the binuclear center, where nitric oxide is bound and reduced. NorC is a membrane-anchored cytochrome c with one c-type heme, which presumably forms the site of electron entry. In addition to NO reduction, cNOR can also catalyze O 2 reduction (6 -8). Site-directed mutations that affect NO reduction affect O 2 reduction in a similar manner (7,9,10); thus, presumably the same catalytic components are used for both reactions. Furthermore, proton transfer occurs from the same side with similar rates and amplitudes during single-turnover reactions between the fully reduced cNOR with NO (11,12) and O 2 (8).
Recently, the crystal structure of a cNOR from Pseudomonas aeruginosa (13), highly homologous (52% sequence identity) to cNOR from P. denitrificans, was determined, and based on this structure and molecular dynamics (MD) simulations, three different proton transfer pathways were proposed, all leading from the periplasmic side of the membrane (see Fig. 1) (13)(14)(15).
When studying the role of individual amino acids for proton transfer, the flow-flash technique has provided extensive information about the HCuOs. Here a single turnover of fully reduced enzyme with O 2 is studied time-resolved so that individual transitions can be resolved. We have previously studied the oxidation of the fully reduced cNOR by O 2 using this technique and found a transition that involves proton-coupled electron transfer from the hemes b and c to the active site, modeled as rate-limited by proton transfer from an internal group (e.g. amino acid, H 2 O molecule, or part of cofactor), crucial for proton transfer to the catalytic site, with a pK a of 6.6 (8). This transition should thus be specifically sensitive to changes in the rate of proton transfer, and in this work we further demonstrate that it shows a kinetic isotope effect (when H 2 O was exchanged for D 2 O as the solvent) of ϳ4 for the maximum rate constant, indicating that the transition is limited by the rate of proton transfer.
Before the crystal structure was known, we had constructed a model of the NorB from P. denitrificans based on the homology to structurally defined HCuOs, and a proton pathway was predicted based on this model, sequence conservation, and biochemical studies (5). This pathway involved residues Glu-135 (Glu-122) and Glu-138 (Glu-125) and was supported by data on P. denitrificans cNOR Glu-122 and Glu-125 variants, which showed effects on catalytic turnover (7,16) as well as proton-coupled electron transfer (9). The crystal structure of the cNOR from P. aeruginosa (13) later showed that the Glu-135 is in fact coordinating a Ca 2ϩ (see Fig. 1), and the Glu-138 stabilizes the loop in which the Glu-135 is located. These residues, although predicted to line the recently proposed proton transfer pathway 1 (Glu-135) and 3 (Glu-135 and Glu-138), thus have structural roles that make interpretations about their roles in proton transfer difficult.
In this work, based on the pathways proposed from the crystal structure and MD simulations, we exchanged individual residues along each of the three proposed pathways for non-protonatable residues. We chose the residues furthest away from the Ca 2ϩ , Fe B , or hemes, because they are less likely to have structural roles. These variants were characterized in terms of their catalytic turnover with both NO and O 2 , their ligand binding properties with CO and O 2 , and finally their proton-coupled electron transfer rates and amplitudes. Our results strongly favor pathway 1 as the only functional proton pathway in cNOR.

EXPERIMENTAL PROCEDURES
Cloning, Expression, and Purification-The pNOREX plasmid with the norB and norC genes was used for cNOR expression (7). Mutations were introduced directly in pNOREX with the use of the QuikChange XL site-directed mutagenesis kit (Stratagene). Correctly mutated pNOREX was transformed to the E. coli JM109 strain together with the pEC86 vector as described (7). E. coli was grown, and cNOR was expressed and purified essentially as described (9) with a few alterations for the cNOR variants and the wild type used for the comparison with the cNOR variants (but not for the experiments with wild type in D 2 O). The altered purification protocol was as follows. Membrane vesicles from a 6-liter cell culture were solubilized in 100 ml of buffer containing 100 mM Tris, pH 7.6, 50 mM NaCl, and 1% (w/v) n-dodecyl-␤-D-maltoside (DDM). The solution was incubated at 4°C with constant stirring for 1 h. Unsolubilized material was removed by centrifugation (30 min, 185,000 ϫ g, 4°C), and the supernatant was filtered over a 0.2-m filter. The filtrate was diluted twice (with 100 mM Tris, pH 7.6, and 50 mM NaCl) to lower the DDM concentration and loaded at 2 ml/min on a 110-ml Q-Sepharose high performance (GE Healthcare) column pre-equilibrated with the same buffer and 0.05% (w/v) DDM. The column was washed with ϳ300 ml of 100 mM Tris, 50 mM NaCl, and 0.05% (w/v) DDM at 5 ml/min. cNOR was eluted from the column in a 475-ml gradient of 200 -500 mM NaCl with 20 mM Tris, pH 7.6, and 0.05% (w/v) DDM at 4 ml/min. 5-ml fractions were collected and diluted 3ϫ in 20 mM Tris, pH 7.6, and 0.05% (w/v) DDM. The absorbance spectra of the collected fractions were analyzed via a dip probe connected to a Cary 50 Bio spectrophotometer (Varian). Fractions with an A 280 nm /A 410 nm of Ͻ2 were collected and concentrated over a 100 kDa cut-off filter (Millipore). 20 mM Tris, pH 7.6, and 0.05% (w/v) DDM were added so that the NaCl concentration was below 50 mM. 25-100-l fractions of 20 -150 M purified protein (A 280 nm /A 410 nm ϳ1) were flash frozen in liquid nitrogen and stored at Ϫ80°C.
The presence of correctly inserted b and c hemes in all the cNOR variants was verified via UV-visible spectra from 260 to 700 nm on a Cary 50 or 400 spectrophotometer (Varian). The cNOR concentration was calculated from ⑀ 550 nm red-ox ϭ 70 mM Ϫ1 cm Ϫ1 .
Multiple Turnover-The reduction rates of the cNOR variants with O 2 and NO were essentially determined as described in (10), but all measurements were conducted at 30°C, and NO measurements were done with a specific NO electrode (World Precision Instruments) and recorded via the LabScribe2 software (World Precision Instruments). There were also small differences in the reaction set-up. For the multiple-turnover measurements with NO, NO-saturated water (2 mM NO) was added in five steps of 5 l (10 M/addition) to a deoxygenated solution of 50 mM HEPES, pH 7.5, 50 mM KCl, 0.05% (w/v) DDM, 30 mM glucose, 20 units/ml catalase, 1 unit/ml glucose oxidase, 500 M N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine, and 20 M cytochrome c (horse heart; Sigma-Aldrich). This resulted in five equal steps in the NO signal. Then 3 mM ascorbate was added (this results in some background signal), followed by 25-100 nM cNOR. Because of substrate inhibition at high [NO], the obtained maximum rate was calculated from the slope at low [NO] (ϳ5 M). At this part of the curve, the ascorbate background is negligible (as verified by a measurement without the addition of cNOR). O 2 turnover was measured in a solution of 50 mM HEPES, pH 7.5, 50 mM KCl, 0.05% (w/v) DDM, 500 M N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine, 20 M cytochrome c, and 3 mM ascorbate with a Clark-type electrode (Hansatech). The maximal rate was obtained directly after cNOR (at ϳ250 nM) addition, and the background rate, recorded for ϳ1 min just before the cNOR addition, was subtracted.
Flash Photolysis and Flow-Flash-Samples of ϳ5 M cNOR were prepared, and measurements were made as described in Ref. 8 on a set-up described in Ref. 17. In short, samples of ϳ5 M cNOR, 10 mM HEPES, pH 7.5, 50 mM KCl, 0.05% (w/v) DDM, 30 mM glucose, 20 units/ml catalase, 0.2 M N-methylphenazinium methosulfate were prepared in a modified Thunberg cuvette. The sample was made anaerobic with N 2 (g), 0.5 unit/ml glucose oxidase was added to remove the remaining oxygen, and 2 mM ascorbate was added to reduce cNOR. The reduced sample was put under 100% (v/v) CO(g) and incubated overnight at 4°C. CO recombination was studied by flash photolysis; the sample was illuminated with a short laser flash (10 ns, 200 mJ, 523 nm, Nd-YAG laser, Quantel), and the kinetic traces were recorded at the indicated wavelength on a digital oscilloscope. The CO concentration was then lowered to ϳ30% (v/v, 70% N 2 (g)) until the average rebinding time constant was ϳ100 s. ϳ50 M dithionite was added to the cNOR sample, and the protein sample was connected to the stopped-flow syringe that was preincubated with 100 mM dithionite and washed with anaerobic water. The other syringe contained an oxygenated buffer with 50 mM HEPES, pH 7.5, 50 mM KCl, and 0.05% (w/v) DDM. To look at pH dependence, the HEPES was exchanged for different buffers at various pH values: MES (pH 6.0 -7.0, HEPES (pH 7-8.5), Tris (pH 8.5), and citric acid (pH Ͻ6). The protein and the buffer samples were mixed in a 1:5 ratio (protein/oxygenated buffer) in a modified stoppedflow apparatus (Applied Photophysics), and after a 200-ms delay, the laser flash was applied to dissociate CO and allow O 2 to bind and initiate the reaction.
Data Handling and Analysis-The time course of the reaction was studied from microseconds to seconds (via two channels, one recording 10 ms and one channel, prefiltered at 30 kHz, recording 2 s) at different wavelengths in the Soret and ␣ regions. At each wavelength, 100,000 data points were collected, and the data set was then reduced to ϳ1000 points by averaging over a progressively increasing number of points. The time-resolved absorbance changes were fitted individually or globally to a model of consecutive irreversible reactions with the software package Pro-K (Applied Photophysics). The pH dependence of the proton-coupled electron transfer phase was fitted with the following equations (8), where k obs represents the obtained rate constant at a certain pH, and k H is the maximum rate at low pH. k H is the ratelimiting internal proton transfer (in Ref. 8 presumed to be from a group, AH (assumed to be in rapid equilibrium with bulk pH), to the active site). ␣ AH is the fraction of protonated AH, determined by its pK a and the pH. For wild type, a small background rate, k 0 , was added to Equation 1 (8). Kinetic

Conservation of Residues in the Predicted Proton Transfer Pathways
In Table 1, we have listed the conservation pattern of the residues involved in the three predicted proton transfer pathways (based on the alignment of 141 cNOR sequences in Ref. 18). Some, but not all, residues of pathway 1 are highly conserved. There are also some highly conserved residues in pathways 2 and 3, but for most of them, we suggest a role that is not related to proton transfer (Table 1), based on the crystal structure (13).

cNOR Variants Made
In pathway 1, we constructed variants for the two amino acids at the entrance of the pathway: K54A C (Lys-53 C in P. aeruginosa) and E58Q C (Glu-57 C ). We also substituted the following aspartate, Asp-185 (Asp-198), for a glutamate (to maintain the negative charge but change the side chain length), asparagine (to maintain the side chain length but remove the charge), or alanine (for both a drastic change in side chain length and the removal of charge).
In pathway 2, the initial glutamate, Glu-145 C (P. aeruginosa; Fig. 1C), is an alanine in the P. denitrificans cNOR  OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42 (Table 1). We therefore exchanged the next polar residue in the path, the equivalent of Gln-415, for a hydrophobic leucine (Q398L). We also changed the following glutamine (the equivalent of Gln-411; Fig. 1C and Table 1) for a hydrophobic methionine (Q394M).

cNOR Uses a Specific Proton Transfer Pathway
In pathway 3, Asn-54 and Asn-60 C were predicted to form a gate that could open and provide connectivity between the bulk water and an internal hydrated cavity (15). The Asn-60 C is not conserved and is an alanine in P. denitrificans (Ala-61 C ) (Fig. 1, A and D, and Table 1). We constructed cNOR variants for the other asparagine: N47F and N47L (Asn-54 in P. aeruginosa cNOR; Fig.  1D). All other PW 3 residues have structural roles (Table 1), and their importance was therefore not analyzed in this study.

Expression, Optical Spectra, and Multiple Turnover
Most cNOR variants along the three proposed pathways could be expressed and resulted in stable protein complexes, except for E78F C (in pathway 2; Table 2). The stable cNOR variants were characterized with respect to their optical spectra (oxidized and reduced) and their catalytic turnover rates with NO and O 2 . All variants showed wild type-like optical spectra (except for N47L, which was therefore not studied further), showing that hemes b and c were integrated normally into the cNOR. All cNOR variants with mutations in pathway 1 were significantly affected in the ability to reduce NO and O 2 ( Table 2), although the mutations were Ͼ7 Å away from both the active site and the Ca 2ϩ . The expressed and stable cNOR variants with mutations in pathway 2 (Q394M and Q398L) or pathway 3 (N47F) reduced both NO and O 2 with rates that were more similar to wild type ( Table 2). N47F showed wild type rates for both NO and O 2 reduction. Q398L reduced NO as wild type but had a slightly slower rate with O 2 (ϳ61%). Q394M reduced NO at ϳ67% and O 2 at ϳ33% of the wild type rate.

Single-turnover O 2 Reduction
Multiple-turnover rates cannot be used to discriminate between changes in proton transfer rates and any other change in the reaction cycle. Therefore, we studied the transitions as the fully reduced cNOR is oxidized by O 2 . In this reaction for wild type (WT) cNOR, O 2 binds to heme b 3 with a time constant of ϳ50 s at 1 mM O 2 (k obs ϳ2 ϫ 10 4 s Ϫ1 ), followed by proton-coupled electron transfer from the hemes b and c to the active site with a time constant of ϳ20 ms at pH 7.5 (k obs ϳ60 s Ϫ1 ; Figs. 2 and 3 and Table 2). These time constants deviate slightly from the ϳ40 s for O 2 binding and ϳ25 ms for the second phase as reported previously (8), probably because of the slightly altered purification conditions. Based on its pH dependence (Fig. 2B), the 20 -25-ms phase was modeled to be rate-limited by proton transfer from an internal group with a pK a of 6.6 (see Ref. 8 and Equations 1 and 2). In this paper, we   The error indicates the range for two measurements for NO multiple turnover. c S.D. value is indicated for O 2 multiple turnover for 5-7 measurements. d The NO multiple turnover activity of the wild type was lower than reported previously (7, 10), possibly because of slight alterations in the purification protocol.
The cNOR variants were prepared in the same way as the wild type, such that the percentage activity reports a valid comparison. e ETPT stands for the proton-coupled electron transfer during single-turnover O 2 reduction. The ETPT in wild type was slightly faster than before (25 ms in Ref. 8).

cNOR Uses a Specific Proton Transfer Pathway
will refer to this phase as the proton-coupled electron transfer (ETPT), but we would like to stress that this term is not intended to imply the order of the reactions.

Kinetic Isotope Effect
To further investigate if the ETPT is indeed limited by proton transfer, we studied its kinetic (deuterium) isotope effect at various pH/pH* values (i.e. reporting the pH meter reading for both H 2 O and D 2 O solutions). The obtained rate constants are plotted in Fig. 2B. The observed rate constants in H 2 O were previously fitted to a pK a of 6.6 Ϯ 0.1 and a k H of 244 Ϯ 7 s Ϫ1 (8). In D 2 O, the corresponding values were pK a * ϭ 6.7 Ϯ 0.1 and k H* ϭ 66 Ϯ 3 s Ϫ1 . The maximum rate constant k H thus has a kinetic isotope effect (ratio of rate constants in H 2 O and D 2 O) of 3.7, indicating that this phase is indeed limited by proton transfer (e.g. see Ref. 19). The small change in pK a when exchanging H 2 O for D 2 O follows the empirical formula (pK H ϭ 0.929 ϫ pK H* ϩ 0.42) determined in Ref. 20.

CO and O 2 Binding Rates
In order to further probe the integrity of the active site in our cNOR variants, we determined the time constant for CO and O 2 binding. O 2 binding was found to be the same as in wild type, with ranging from ϳ40 to ϳ50 s (Fig. 3). Only the D185N and D185A had slightly slower O 2 binding with a of ϳ60 or ϳ65 s, respectively (Fig. 3). CO binding in the D185A and the D185E variants was slower than in wild type, whereas it was essentially unchanged for any of the other variants with normal optical spectra (data not shown). The changes in O 2 and/or CO binding in the Asp-185 variants could indicate an altered environment of the b 3 heme. D185N and D185A were also less well expressed ( Table 2), indicating that the Asp-185 is needed to produce/maintain a stable and functional cNOR.

Proton-coupled Electron Transfer
Because we established that the ETPT is indeed limited by proton transfer, we used it to indicate whether proton transfer was affected in the cNOR variants.
Pathway 1-For the Asp-185 (Asp-198 in P. aeruginosa cNOR) variants, the amplitude for the ETPT is small (430 and 550 nm) or even absent (420 nm), indicating that the reaction is completely inhibited. In the E58Q C (Glu-57 C in P. aeruginosa cNOR) variant, the ETPT rate constant is ϳ30 times slower than in WT at pH 7.5 (with a k obs of ϳ2 s Ϫ1 , of ϳ500 ms). The K54A C (Lys-53 C in P. aeruginosa cNOR) has an ETPT rate constant that is ϳ10 times slower (with a k obs of ϳ4 s Ϫ1 and of ϳ250 ms) than in WT at pH 7.5.  OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 30631 pH Dependence of the Proton-coupled Electron Transfer in Affected cNOR Variants
For the cNOR variants that had slower ETPT rate constants at pH 7.5, we further investigated this reaction at various pH values. However, this could not be done for the Asp-185 variants because their ETPT is small or even absent (and the rates can therefore not be fitted).
In the cNOR variant E58Q C , the reaction displayed severely slower rates in the entire pH range that we tested, with a k H (maximum rate constant) of ϳ50 s Ϫ1 (i.e. ϳ5 times slower than for WT) (Fig. 4). The pK a in E58Q C shifted from 6.6 to ϳ5.8.
The K54A C cNOR variant, however, has a similar k H as wild type (ϳ250 s Ϫ1 ), but the pK a is shifted from 6.6 to 6.4. The ETPT rate constant in the K54A C variant was fitted to a single exponential at all pH values for comparison with wild type, although the fit was better with two exponentials at pH values from 6.5 to 7.5. This could explain why the obtained rate constants at pH 7-7.5 were not fitted well to a single pK a transition (Fig. 4). This behavior can be explained by a model in which at high pH the assumption that the internal protonatable group (AH, Equations 1 and 2) is in rapid equilibrium with the bulk pH is not valid for this cNOR variant (because the entry point to the pathway is altered). Thus, at higher pH, proton diffusion to AH becomes rate-limiting, and this proton diffusion rate (tentatively fitted to ϳ2 ϫ 10 8 M Ϫ1 s Ϫ1 ) is plotted as a red dotted line in Fig. 4. A small background rate k 0 of ϳ1 s Ϫ1 was added to this fit. This altered model could also explain the biphasic nature of the ETPT in this cNOR variant because in the enzyme population with AH protonated (and not in rapid equilibrium with bulk pH), ETPT can occur with the maximum rate k H . With a forced one-exponential fit, we would see an average of the two rate constants (k H and the diffusion-limited rate constant) in the pH range where both populations (protonated/deprotonated AH) contribute significantly. However, we did not pursue a full biphasic fit of the data because of the small number of points that could not be fitted reasonably well with one exponential.

DISCUSSION
The multiple-turnover data with NO and O 2 show severe effects for variants with pathway 1, but not pathway 2 (Q398L, Q394M) or 3 (N47F), residues modified ( Table 2). Furthermore, all cNOR variants that are affected in their NO turnover are also affected in their O 2 turnover ( Table 2). The correlation between O 2 and NO turnover has been observed before (although the extent of the change can differ (7, 10)) and shows the validity of using O 2 as an alternative substrate.
The ET reaction (in this paper called ETPT) during singleturnover reduction of O 2 by fully reduced cNOR was previously suggested to be rate-limited by proton transfer based on the uptake of protons from solution with the same rate constant and the pH dependence of the reaction (8). In this work, we show that the k max (k H ) is decreased, from ϳ250 s Ϫ1 to ϳ70 s Ϫ1 (i.e. by a factor of ϳ4) in D 2 O (Fig. 2), which indicates that the rate of proton transfer indeed limits the overall rate constant for this reaction. We thus have strengthened the basis for using this ETPT as a "reporter" for effects on proton transfer rates in cNOR. The kinetic isotope effect is higher in A-type HCuOs (the F 3 O transition has a kinetic isotope effect of 7 (21)), where there are conformational changes needed in the proton pumping process. In cNOR we do not expect large conformational changes, because cNOR does not pump protons, and no "gating" is required.
The effects on the ETPT in the cNOR variants studied here were qualitatively in agreement with the multiple-turnover data, except for Q394M (PW 2), where the turnover rate with O 2 is slower than wild type, whereas the ETPT is unaffected. Presumably, rereduction is slowed in this cNOR variant but for reasons unknown at this point. We note, however, that Gln-394 is a highly conserved residue (Table 1). For the other variants in PWs 2 and 3 ( Fig. 1), there are no effects on the ETPT (Fig. 3 and Table 2) and no (large) effects on multiple turnover. It could be argued that very few variants were made in PWs 2 and 3, but because we aimed at residues away from the Ca 2ϩ and the hemes without other roles, as suggested by the structure (Table  1), few residues were good candidates (especially for PW 3, which shares parts with PW 1). Even so, the reason why the cNOR variant E78F C (in PW 2) was not expressed and N47L (in PW 3) had altered spectra and CO binding properties is possibly because of their close proximity to heme c (4.9 and 3.5 Å for Glu-78 C and Asn-47, respectively). Both suggested PWs 2 and 3 in the P. aeruginosa cNOR contain one residue that is not conserved in P. denitrificans cNOR (as in many other cNORs; see Table 1 and Ref. 18), also indicating that PWs 2 and 3 are not used for proton transfer.
The variants in PW 1 all affect catalytic turnover (Table 2) and the rate of the ETPT (Fig. 3). None of the modified residues have obvious structural roles as, for example, Ca 2ϩ ligands (Table 1), and all are Ͼ7.5 Å away from the redox centers and the Ca 2ϩ site. Therefore, it is unlikely that all of these residues have structural roles and/or control the heme potentials. The The rates of the ETPT for cNOR wild type and the constructed variants that were affected at pH 7.5 are plotted as a function of pH: wild type (black; circles for the data from Ref. 8 and crosses for the data with the slightly altered purification protocol), K54A C (red triangles), E58Q C (blue diamonds). The WT data (from Ref. 8) were fitted to a pK a of 6.61 Ϯ 0.05 and a k H of 244 Ϯ 7 s Ϫ1 (black line); K54A C data were fitted to a pK a ϭ 6.4 Ϯ 0.1 and a k H ϭ 247 Ϯ 11 s Ϫ1 (red line); and E58Q C data were fitted to a pK a ϭ 5.8 Ϯ 0.1 and a k H ϭ 49 Ϯ 2 s Ϫ1 (blue line). However, data points for K54A C did not follow the fit around pH 7-7.5, possibly because the one-exponential fit is an oversimplification for this mutant. As a comparison, also plotted (as a dotted red line) is the diffusion rate, assuming k diff of ϳ2 ϫ 10 8 M Ϫ1 s Ϫ1 (see "Results" for details). The error bar at pH 7.0 for K54A C indicates that at this pH the range of possible fits is rather large.

cNOR Uses a Specific Proton Transfer Pathway
variants where the Asn-185 (Asn-198 in P. aeruginosa cNOR) was exchanged for a non-protonatable residue (D185A and D185N) retain some residual activity in multiple turnover. In the single-turnover reaction with O 2 , however, only O 2 binding is observed, and there is no ETPT (Fig. 3). This might be due to a limited "window" in which the ETPT reaction can be observed, because on the longer time scale at 420 nm (Fig. 3E), a phase that we attribute to CO rebinding is observed, presumably due to a slow re-equilibration between CO and the O 2 at the active site (9,10). Although D185E had higher multipleturnover rates than D185A and D185N, its rates were still much lower than those of wild type. The effects on the ETPT seemed too severe for the residue to have a "simple" role in proton transfer because an Asp is only a methyl group shorter than a Glu. The chain length might, however, influence the pathway because the Asp-198 could form hydrogen bonds with Lys-53 C (Lys-54 C ), Arg-134 (Arg-121), and various waters (15) (Fig. 5). An alternative explanation is that the aspartate has a more structural role, because the D185N and D185A variants were only expressed at low levels, and ligand binding was also slowed in these variants.
The K54A cNOR variant reaches the same maximum rate at low pH, but it has a shifted pK a and is ϳ10 times slower than wild type at pH 7.5. Thus, the Lys-54 C (Lys-53 C in P. aeruginosa) is important to support rapid proton transfer at the growth pH used for denitrifying conditions (pH 7.5 (22)), where the full denitrification chain is active (23).
In the E58Q C variant (Glu-57 C in P. aeruginosa), even at low pH, the rate constant of the ETPT is still ϳ5 times slower than in WT, and the pH dependence was fitted with a tentative pK a of ϳ5.8. There is thus a severe slowing of the proton transfer rate constants at all pH values, making the effect much more drastic than in K54A C . The Glu-58 C seems thus to be a very important part of the pathway, presumably forming its entry point (Fig. 5). We note that the Lys-54 C is actually more conserved than the Glu-58 C (100% compared with 77%; Table 1 and Ref. 18). Although we do not have an explanation for this at the moment, we do note that the sequences that do not have Glu-58 C cluster together in the alignment and belong to different subgroups of the cNOR family than the subgroup of P. aeruginosa and P. denitrificans (18). The replacing residues are hydrophilic (Gln, His, or Asp), and there are charged residues in the flanking sequence (3 or less residues away) that might take over the role of the Glu-58 C in these cNORs.
The identity of the internal proton donor (AH in Equations 1 and 2, pK a ϭ 6.6 in wild type) remains unknown. There was a large pK a shift (Ͼ3 pH units upshifted) in the ETPT observed in variants with an exchanged Ca 2ϩ ligand (Glu-122 in P. denitrificans cNOR, Glu-135 in P. aeruginosa, also in PW 1) (9). This shift was much larger than the pK a shifts observed with the variants studied in this work and indicates that the donor is located in the vicinity of the Glu-135. However, the coordination of the Ca 2ϩ with multiple groups (one propionate each from the b and b 3 hemes, Tyr-73 C , Gly-71 C , and an H 2 O) gives many possibilities. Because these residues all have structural roles as Ca 2ϩ ligands, it is difficult to study any additional roles. We rule out the Arg-134 (Arg-121 in P. denitrificans) as AH, although it is very close and highly conserved ( Fig. 1B and Table  1) because arginines have been shown to keep their high pK a values (ϳ12.5) even when embedded in protein interiors (24). The heme b propionate and the A propionate of heme b 3 are possible candidates for AH, because one of them could, in principle, transiently dissociate from the Ca 2ϩ and be protonated (Fig. 5). The Asp-198 (Asp-185 in P. denitrificans) is also a possible candidate, being ϳ10 Å from the Glu-135, hydrogenbonded to the Lys-53 C , and giving such severe effects of mutation. However the pK a shift in the variant of the Lys-53 C equivalent (K54A C ) seems too modest (only 0.2 units) for a residue directly hydrogen-bonded to the proton donor. Although not in direct contact with the Glu-135 or the Ca 2ϩ , the D propionate of heme b 3 (Fig. 5) seems like a better candidate for the proton donor for several reasons. First, precisely because it is not coordinated to Ca 2ϩ , it could adopt a higher pK a , possibly around 6.6, and it could still be affected by a change in ligation of the other propionate. Second, the D propionate is better positioned in the path to form a connection "onward" (Fig. 5), toward the reaction intermediate bound at (or between) the irons of heme b 3 and Fe B . MD simulations (15) postulated that protons could move 1) directly from the water cluster around the A and D propionates of heme b 3 to a water cluster around the active site, 2) from the D propionate of heme b 3 via the His-259 to the active site, or 3) from the D propionate of heme b 3 via Thr-330, Ser-277, Glu-280, and Glu-211 to the active site (Fig. 5). It is tempting to suggest that the proton donor in cNOR in the path from the periplasm to the active site became (or originated from, depending on the rooting of the evolutionary tree; see, for example, Ref  The structure is the same as in Fig. 1 (cNOR from P. aeruginosa (Protein Data Bank code 3O0R) (13)) and rendered in the same way (except that for clarity a loop instead of helical representation is used) with the residues in the start of pathway 1 indicated with sticks in cyan (pathway 1) and the ones predicted to be involved in the continuation of the proton pathway to the active site in green. The suggested pathway is indicated with a blue (dotted) arrow. Crystallographic waters in or around the pathway are indicated with blue crosses. The black dotted lines indicate hydrogen bonds. The yellow A and D indicate the heme b 3 propionate A and D, respectively. OCTOBER 18, 2013 • VOLUME 288 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 30633 C-type, where it should be noted that the location of the A propionate of the active site heme corresponds to the location of the D propionate of the b 3 heme in cNOR). Pathway 1 in cNOR is not conserved to the closest O 2 -reducing HCuO, the C-type, where instead parts of the PW 3 are conserved (15). However, the PW 3 residues conserved to C-type HCuOs (Glu-135 (also involved in PW 1), Glu-138, and Arg-57 in P. aeruginosa cNOR) all are involved in defining the Ca 2ϩ site (Table 1). This could explain why they are conserved to C-type HCuOs, which have a Ca 2ϩ bound in a similar manner (28).

cNOR Uses a Specific Proton Transfer Pathway
Taken together, our data presented here strongly suggest that cNORs, at least the one from P. denitrificans, do use a specific pathway for proton transfer from the periplasm into the active site and that this is the suggested pathway 1 (Figs. 1 and 5) and that the other suggested pathways cannot take over the role of pathway 1. We note that a different study, where several residues were mutated in the cNOR from Thermus thermophilus (18), reached the conclusion that there is no preferred pathway in (this) cNOR. They did not investigate PW 3, but they observed effects on turnover rates when changing the residue at the entrance of PW 2, D209E/N/L C , which is an alanine in P. denitrificans (Ala-149 C ) and a glutamate in P. aeruginosa (Glu-145 C ). Their results also differed for residues along pathway 1. Mutating the equivalent of Asp-198 (P. aeruginosa numbering) of PW 1 in the T. thermophilus cNOR to either Glu or Asn had no effect on turnover (although, when changed into a leucine, the cNOR was not assembled). The Glu-57 C is a Gln in T. thermophilus cNOR. Lys-53 C is present in T. thermophilus cNOR, but its role was not investigated. The reason for the differences between the results of our study and the study with T. thermophilus cNOR (18) is not known. It is possible that PW 1 is used in the P. denitrificans cNOR (and P. aeruginosa cNOR) but not in T. thermophilus cNOR. The P. aeruginosa cNOR and the P. denitrificans cNORs are much more closely related to each other (51% identity between the NorB subunits, both belonging to the Proteobacteria phylum) than to the only distantly related T. thermophilus cNOR (belonging to the Thermus phylum and with 38% identity in NorB to both P. aeruginosa and P. denitrificans). There might be larger differences in the proton transfer pathways between subfamilies in cNOR compared with the O 2 -reducing HCuOs because no gating is necessary. It is also possible that effects on the proton transfer rates that we see in the P. denitrificans cNOR for D185N/ D185E would not have been observed in the multiple-turnover experiments with NO in T. thermophilus cNOR, given that the overall turnover rate is orders of magnitude lower in the T. thermophilus enzyme (ϳ0.1 s Ϫ1 compared with ϳ10 s Ϫ1 for the P. denitrificans cNOR used in this study; Table 2).
In principle, cNOR would not need to provide a specific pathway for protons, it could have protons "leak" into the active site from multiple routes. However, the use of a single, specific route can have different reasons. One is the evolutionary relationship to the other heme-copper oxidases. If cNOR evolved from an O 2 -reducing HCuO that had a specific path for the extrusion of the pumped protons, then this pathway could have been "reversed" in cNOR. Also, there is an energetic cost in terms of protein stability involved in creating a polar, waterfilled pathway capable of proton transfer in an otherwise hydro-phobic protein interior. This might minimize the number of proton transfer pathways to those that are absolutely necessary for function.