NO synthase (NOS) enzymes convert l-arginine to NO in two sequential reactions whose rates (kcat1 and kcat2) are both limited by the rate of ferric heme reduction (kr). An enzyme ferric heme–NO complex forms as an immediate product complex and then undergoes either dissociation (at a rate that we denote as kd) to release NO in a productive manner, or reduction (kr) to form a ferrous heme–NO complex that must react with O2 (at a rate that we denote as kox) in a NO dioxygenase reaction that regenerates the ferric enzyme. The interplay of these five kinetic parameters (kcat1, kcat2, kr, kd and kox) determines NOS specific activity, O2 concentration response, and pulsatile versus steady-state NO generation. In the present study, we utilized stopped-flow spectroscopy and single catalytic turnover methods to characterize the individual temperature dependencies of the five kinetic parameters of rat neuronal NOS. We then incorporated the measured kinetic values into computer simulations of the neuronal NOS reaction using a global kinetic model to comprehensively model its temperature-dependent catalytic behaviours. The results obtained provide new mechanistic insights and also reveal that the different temperature dependencies of the five kinetic parameters significantly alter neuronal NOS catalytic behaviours and NO release efficiency as a function of temperature.
dissociation rate of NO from the ferric heme–NO complex of nitric oxide synthase
oxidation rate of the ferrous heme–NO complex of nitric oxide synthase
reduction rate of heme by the reductase (flavoprotein) domain of nitric oxide synthase
neuronal nitric oxide synthase, nNOSoxy oxygenase domain of the neuronal nitric oxide synthase
oxygenase domain of the nitric oxide synthase
nitric oxide synthase
Nitric oxide (NO) is a biological mediator produced in animals by three NO synthase isozymes (NOS; EC 184.108.40.206): inducible NOS (iNOS), neuronal NOS (nNOS) and endothelial NOS (eNOS) [1, 2]. NOS enzymes catalyze a stepwise oxidation of l-arginine (Arg) to form nitric oxide and citrulline [3-5]. In the first reaction, Arg is hydroxylated to form Nω-hydroxy-l-arginine (NOHA) and, in the second reaction, the NOHA intermediate is oxidized to form NO and citrulline (Scheme 1) . Both steps consume one molecule of O2 and utilize NADPH-derived reducing equivalents. NOS proteins comprise an N-terminal oxygenase domain and a C-terminal flavoprotein domain, with a calmodulin (CaM)-binding site connecting the two domains , and each NOS is only active as a homodimer [6, 7]. The oxygenase domain binds Fe–protoporphyrin IX (heme), the substrate Arg and the essential cofactor (6R)-tetrahydrobiopterin (H4B) [8, 9], whereas the flavoprotein domain binds FAD, FMN and NADPH [10-12]. During catalysis, the flavoprotein domain provides electrons to the heme in the NOS oxygenase domain, which enables the heme to bind O2 and initiate oxygen activation in both reactions of NO synthesis [13-15]. Heme reduction also requires that CaM be bound to NOS, and is rate-limiting for the NO biosynthetic steps.
NOS enzymes operate under the constraint of being heme proteins that generate NO, a good heme-binding ligand. Indeed, their newly-made NO binds to the ferric heme many times before it can exit the enzyme . How this intrinsic heme-NO-binding event impacts NOS catalytic cycling is shown in Fig. 1 and has been discussed in detail previously [16-20]. The Arg to NO biosynthetic reactions [FeIII to ferric heme–NO complex (FeIIINO) steps in Fig. 1] involve two sequential oxidation reactions whose rates are both limited by the rate of ferric heme reduction (kr) because all subsequent catalytic steps in either reaction (termed kcat1 and kcat2 in Fig. 1) occur faster than kr. Once the FeIIINO product complex forms in NOS, it can either dissociate and enable the release of NO into the medium (at a rate kd as shown in Fig. 1) or become reduced by the flavoprotein domain (at a rate kr' in Fig. 1; assumed to be equal to kr)  to form the enzyme ferrous heme–NO species (FeIINO), which releases NO extremely slowly [18, 19]. Consequently, the FeIINO species must undergo reaction with O2 at a rate kox (Fig. 1) to regenerate the ferric enzyme and return to an active form. Thus, two cycles exist during steady–state NO synthesis (Fig. 1): NO dissociation from the enzyme (kd) is part of a ‘productive cycle’ that is essential for NOS bioactivity, whereas reduction of the FeIIINO product complex (kr') channels NOS into a ‘futile cycle’ that actually represents a NO dioxygenase activity. The rate constants for each step of NOS catalysis are shown in Fig. S1.
Observed rates for kr, kcat, kd and kox have been measured in NOS enzymes in single catalytic turnover reactions [14, 18, 19, 22-25]. However, because most of the measures have been obtained at only a single temperature (10 °C, or less often at 25 °C), little or nothing is understood regarding the temperature dependencies and thermodynamic properties of the individual steps, as well as how they combine to govern the catalytic behaviour and overall activities of NOS enzymes. This information would improve our understanding NOS catalysis and enable the accurate modelling of its catalytic cycle under physiological temperatures and conditions. In the present study, we poised rat nNOS at different points in the cycle, and utilized stopped-flow spectroscopy to characterize the temperature dependence of five individual kinetic parameters (kr, kcat1, kcat2, kd and kox). The thermodynamic parameters of these individual reactions provide new insights regarding NOS structure–reactivity relationships and provide a better understanding of their physiological relevance.
Results and Discussion
Temperature dependence of heme reduction (kr)
We directly assessed electron transfer (ET) from FMN hydroquinone to heme using stopped-flow spectroscopy under anaerobic conditions. Heme reduction was monitored via the formation of the FeIICO complex at 444 nm. Once ferrous heme is formed by ET from the FMN, the fast and stable binding of CO prevents the accumulation of the ferrous species and makes the ET reaction essentially irreversible. This mimics the normal condition of having O2 present to react with ferrous heme, and insures that the observed rate of heme reduction does not have a significant back reaction component. The reactions were initiated by mixing CaM-bound nNOS with excess NADPH , and heme reduction rates (kr) were measured at 5, 10, 15, 20, 25, 30 and 37 °C. Representative diode array spectra (Fig. 2, upper panel) and the resulting kinetic trace (Fig. 2, upper panel inset) are shown. We fit the kinetic traces to a bi-exponential function, with the initial fast phase representing an absorbance decrease as a result of flavin reduction by NADPH, and the subsequent slower phase representing the absorbance increase as a result of heme reduction in the presence of CO. The observed rates obtained for heme reduction at each temperature are provided in Table 1. The heme reduction rate increased approximately 1.5-fold with every 10 °C rise in temperature. Fitting the experimental data to the linearized form of the Arrhenius equation yielded an activation energy of Ea = 31.5 kJ·mol−1. A linear Eyring plot (Fig. 2, middle panel) of the data yielded values for the enthalpy and entropy of activation for the heme reduction step: ΔH‡ = +29.1 kJ·mol−1; ΔS‡ = −128 J·mol−1·K−1 (Table 2). The derived Gibbs free energy of activation ΔG‡ at 25 °C was calculated as 67.2 kJ·mol−1. The ΔG‡ increases with temperature, which is expected, given the negative entropic contribution. The activation enthalpy and entropy values for heme reduction that we determined for nNOS are similar to values recently reported for human iNOS (36.9 kJ·mol−1 and −89.7 J·mol−1·K−1, respectively), which were derived using a different measurement method (flash photolysis) . This suggests that similar mechanisms may govern heme reduction among NOS isozymes.
Table 1. Measured rates (mean ± SD) at different temperatures
Determined at approximately 202 μm O2, 180 μm O2, 162 μm O2, 145 μm O2, 135 μm O2, 125 μm O2 and 108 μm O2 at 5, 10, 15, 20, 25, 30 and 37 °C, respectively (i.e. half air-saturated concentration at the respective temperature).
4.2 ± 0.3
9.2 ± 0.8
16.4 ± 1.9
1.6 ± 0.2
0.063 ± 0.002
0.12 ± 0.01
5.3 ± 0.2
13. 5 ± 1.3
29 ± 3
3.7 ± 0.3
0.081 ± 0.005
0.27 ± 0.02
6.5 ± 0.4
21 ± 2
52 ± 5
5.9 ± 0.2
0.102 ± 0.002
0.35 ± 0.03
8.1 ± 0.6
31 ± 3
93 ± 4
10.7 ± 0.1
0.130 ± 0.001
0.55 ± 0.05
10.4 ± 0.5
46 ± 3
162 ± 15
20.1 ± 1.5
0.161 ± 0.003
0.82 ± 0.04
12.7 ± 1.1
68 ± 5
278 ± 18
31 ± 1
0.211 ± 0.004
1.12 ± 0.10
17.2 ± 0.6
113 ± 9
572 ± 36
58 ± 6
0.254 ± 0.002
1.40 ± 0.08
Table 2. Thermodynamic parameters for kr, kox, kd, NO synthesis, kcat1 and kcat2. Errors indicate the SD obtained from at least three measurements. Ea values are calculated from an Arrhenius plot
31.5 ± 0.5
29.1 ± 0.5
−128 ± 3
57.1 ± 2.3
54.7 ± 2.3
−29 ± 1
79.5 ± 2.7
77 ± 3
+56 ± 2
83.1 ± 1.7
80.7 ± 1.7
+50 ± 1
31.9 ± 1.0
29.4 ± 1.0
−161 ± 14
54.1 ± 4.5
51.7 ± 4.5
−74 ± 9
Activation parameters of heme reduction have been determined for other heme proteins, such as azurin, cytochrome c oxidase and cd1 nitrite reductase [28-31]. Particularly, activation parameters determined for interprotein ET reactions, such as in cytochrome c-plastocyanin [32-34], indicate modest enthalpies and high negative entropies, similar to the values that we found for nNOS with respect to its putative inter-domain FMN-to-heme ET. The high negative activation entropy determined for nNOS suggests that the rate-limiting elementary step of the heme reduction process is characterized by a relatively ordered transition state.
Because heme reduction in nNOS involves an ET reaction between the FMN and heme prosthetic groups, we analyzed the temperature dependence data of heme reduction in the framework of the semi-classical Marcus theory describing nonadiabatic ET reactions . This approach has been successfully applied to several biological ET systems to shed light on quantitative aspects of the ET step [36, 37]. It is important to note that heme reduction in NOS enzymes is a relatively complex multistep process. A protein conformational change that involves both long- and short-range motions is expected to bring the FMN domain into close proximity to the heme centre in the oxygenase domain of the nitric oxide synthase (NOSoxy) [38, 39]. Mutational and kinetic studies suggest that the FMN domain first undergoes a long range motion to swing away from the ferredoxin NADP-reductase (FNR) domain and then undergoes short-range sampling motions to productively dock on the NOSoxy domain, which is in part guided by complementary charged residues on the domain–domain interface [11, 39-41]. These conformational change(s) precede the actual ET and could conceivably influence the kinetics and thermodynamic behaviour of the heme reduction that we observe. The ET process would be gated if the conformational change is rate-limiting, whereas, if the actual FMN-to-heme ET step is rate-limiting, the process could be either true-ET or coupled-ET. In true-ET, the ET step is rate-limiting and not kinetically influenced by preceding steps, whereas, in coupled-ET, the ET reaction is still rate-limiting but is preceded by a relatively fast and thermodynamically unfavourable non-ET reaction such as conformational change. Although the FMN to heme ET in NOS enzymes has been reported to be conformationally-gated , a detailed study on this aspect is lacking. Application of the Marcus ET formalism to our temperature dependence data provides an opportunity to quantify parameters such as the required reorganization energy, as well as electronic coupling between the FMN electron donor and heme electron acceptor during ET. The rate constant for nonadiabatic ET, kET, is generally described by Equations (1),(2) which respectively, derive from the prediction that rates of ET exhibit Gaussian free-energy dependence, and from the fact that electronic coupling drops exponentially with distance between redox partners:
where λ is the reorganization energy and ΔG° is the thermodynamic driving force determined from the redox potential difference for the redox partners (FMN and heme in this case). The other parameters have their usual meanings: h is Planck's constant; R is the gas constant; and T is absolute temperature. HDA describes the electronic coupling between electron donor (FMN) and electron acceptor (heme). In Equation (2): ko = 1013 s−1 is the characteristic maximum limit for kET when donor and acceptor are in van der Waals' contact (i.e. at ro = 3 Å) and when no activation is involved (i.e. when λ = –ΔG°). The value of parameter β reflects the nature of the intervening medium in mediating the ET, and a value of 1.4 Å−1 is widely used for proteins .
In Fig. 2 (lower panel), the heme reduction data as a function of temperature were fit using Equation (1) to obtain estimates of the reorganization energy (λ) and the electronic coupling element (HDA). The ΔGo value used for the fitting was −0.030 eV (or −2.9 kJ·mol−1). This driving force is determined from the difference of midpoint values of the FMN hydroquinone/semiquinone couple and the heme FeIII/FeIICO couple in nNOS [38, 43-46]. Using this ΔGo value, the nonlinear fitting yielded best-fit values for the reorganization energy λ = 139 kJ·mol−1 (or ~ 1.4 eV) and an electronic coupling factor HDA = 0.2 cm−1. We have varied ΔGo aiming to investigate whether the values determined by fitting are affected by the driving force used. Changes of as much as ± 60 mV in ΔGo do not result in significant changes in the fitted values of λ or HDA. Fitting the same data to Equation (2) gave a similar value for the reorganization energy λ = 135 kJ·mol−1 (Fig. S2, upper panel), which is not significantly different from λ = 139 kJ·mol−1 determined using Equation (1).
The estimate for HDA in nNOS is within the range known for nonadiabatic ET , whereas the estimated reorganization energy (λ = 1.4 eV) lies on the higher end but is still within the range of reorganizational energies reported for most physiological ET reactions (i.e. λ = 0.5–2.3 eV) . The relatively high value of λ in our case reflects a possible coupling of ET to the pre-requisite conformational change needed to bring the FMN domain within a distance required for ET to the heme in the nNOSoxy. Whether an ET reaction is gated or coupled depends on the relative rates of the ET and prerequisite non-ET steps. An estimated value of Keq = 9.0 was reported for the FMN–FNR domain off/on conformational equilibrium for the reduced, CaM-bound nNOS at 10 °C , indicating that, once reduced, the FMN domain is mostly away from the FNR and thus in a conformation that could interact with nNOSoxy for ET. However, the corresponding koff value associated with the FNR–FMN equilibrium under this condition is not yet known and, in our experimental design, the FMN domain must first receive electrons from the FNR to become reduced. Also, the Keq for the corresponding FMN–NOSoxy equilibrium has been studied in the CaM-bound nNOS poised at different redox states. The data indicate a low value (Keq ≤ 0.2) [47, 48] exists in all cases. This suggests that very little FMN–nNOSoxy complex is present at equilibrium and that their interaction is intermittent and/or transient. Given the relatively slow rates for heme reduction that we observe in nNOS, the relatively high corresponding reorganization energy, and the unfavourable on/off thermodynamic equilibrium for the FMN–nNOSoxy interaction, ET from FMN to heme in nNOS is likely coupled to the pre-requisite conformational movement of the FMN domain. Similar ET reactions that are coupled to conformational equilibria are known. One example is the ET reaction from methanol dehydrogenase to the heme in cytochrome c-551i where the ET is rate-limiting but still influenced by a pre-requisite rapid but unfavourable conformational arrangement .
Our fit of the heme reduction versus temperature data to Equation (2) also yielded an estimate for the distance ‘r’ (FMN-to-heme distance) (Fig. S2). The estimated distance at the time of the ET event is r ≈ 13.4 Å. It is important to note that this distance does not change significantly (r = 13.1 Å) when the data are fit with Equation (2) using the λ value found in nonlinear fitting with Equation (1) (λ = 139 kJ·mol−1) (Fig. S2, lower panel). This estimate is in excellent agreement with FMN-to-heme distances of 9–15 Å proposed based on docking models for FMN–NOSoxy complexes [40, 50]. The value is also consistent with the proposed range (12–13 Å) suggested for nNOS based on the kinetics of heme-to-FMN back ET determined by pulse radiolysis  and the recent studies on FMN–heme coupling by pulsed EPR for human iNOS . In this case, a distance of 18.8 Å between the FMN-N5 atom and the heme iron was determined, which could correspond to an edge to edge FMN–heme distance as short as 13.1 Å . Comparable distances have been reported in related flavocytochrome systems. For example, a value of 18.4 Å has been reported for the FMN to heme iron distance in the complex between the heme and FMN-binding domain of cytochrome P450BM-3 .
Temperature dependence of the oxygenation reactions (kcat1 and kcat2)
The kcat1 and kcat2 parameters represent the conversion rates of the heme FeIIO2 species to products in the Arg hydroxylation and NOHA oxidation reactions, respectively. We utilized stopped-flow spectroscopy to compare the nNOS heme transitions that occur during the single turnover reactions, as reported in previous studies [53, 54]. The reactions were initiated by rapid-mixing an oxygen-containing buffer with pre-reduced ferrous nNOS proteins that contained H4B and either Arg or NOHA as substrate. For the Arg reactions, we observed two consecutive heme transitions: FeII → FeIIO2 → FeIII and, in the reaction with NOHA, we observed three transitions: FeII → FeIIO2 → FeIIINO → FeIII. Global analysis of the spectral data according to these models yielded the spectrum of each species (Fig. 3A and 3B, upper panels). The two middle panels of figure 3 show how the concentration of each species changed with time during the reactions. Linear Eyring plots of the observed rates (Fig. 3A and 3B, lower panels) show that both kcat1 and kcat2 increase as the temperature increases (Table 1) but do so according to significantly different ratios, such that the slope of kcat2 is 1.4-fold higher than that of kcat1 (Fig. 3A & 3B, lower panels). Fitting the experimental data to the Arrhenius equation yielded an activation energy Ea = 57.1 kJ·mol−1 for kcat1 and 79.5 kJ·mol−1 for kcat2. Linear Eyring plots give the values for the enthalpy and entropy of activation derived for kcat1: ΔH‡ = +54.7 kJ·mol−1; ΔS‡ = −29 J·mol−1·K−1 (Table 2). The values for the enthalpy and entropy of activation derived for kcat2 are: ΔH‡ = +77 kJ·mol−1; ΔS‡ = +56 J·mol−1·K−1 (Table 2). The thermodynamic parameters derived for kcat1 are distinguished from kcat2 by a negative entropic component. The relative entropic cost observed in the first catalytic step may be attributed to the necessary change in the hydrogen-bonding network expected at the active site upon the transformation of l-arginine to the enzyme bound NOHA .
Temperature dependence of NO dissociation (kd)
A variety of studies [14, 25, 56, 57] indicate that newly-formed NO binds to the NOS ferric heme before it can exit the enzyme. This causes a FeIIINO product complex to form at the end of each catalytic cycle in all NOS enzymes examined to date . In the NOHA single turnover reactions, there was a clear build-up of this immediate FeIIINO product complex (Fig. 3B, upper and middle panels), which allows us to determine the macroscopic off-rate of NO by monitoring its transition to ferric enzyme (kd parameter in Fig. 1). The macroscopic, observed dissociation rate (kd) of the FeIIINO product complex is an important parameter because it impacts NOS enzyme distribution during catalysis and, consequently, steady-state NO synthesis activity [19, 23, 25]. Linear Eyring plots of the observed rates (Fig. 4, upper panel and Table 1) revealed that the kd parameter has a relatively large temperature dependence, increasing by more than three-fold every 10 °C. The thermodynamic parameters of the reaction were derived from the linear Eyring plot and Arrhenius equation, giving the values: Ea = 83.1 kJ·mol−1; ΔH‡ = 80.7 kJ·mol−1; ΔS‡ = 50 J·mol−1·K−1 (Table 2). The activation energy for NO dissociation from FeIII-nNOS (kd) is quite close to the activation energy for citrulline formation by nNOS as reported earlier by Iwanaga (≃103 kJ·mol−1) . Also, the relatively large positive activation entropy that we calculated for the kd in nNOS is similar to values measured for NO dissociation from other ferric heme proteins such as metmyoglobin (ΔH‡ = 78 kJ·mol−1; ΔS‡ = 46 J·mol−1·K−1)  and the camphor-bound cytochrome P450cam (ΔH‡ = 84 kJ·mol−1; ΔS‡ = 41 J·mol−1·K−1) . Overall, the process of NO release from ferric heme in nNOS exhibits similar thermodynamic behaviour to that of other heme proteins [59, 60] as indicated by comparing Eyring plots for NO release data for the different heme proteins across a range of temperatures (Fig. 4, lower panel). Their similar thermodynamic behaviours are indicative of a common mechanism regulating NO dissociation from the ferric heme and its escape into solution.
What is the basis for the relatively large temperature effect on kd? Binding and dissociation of diatomic ligands such as NO, CO and O2 to heme proteins is generally accompanied by conformational changes that sometimes extend far beyond the local heme-binding site [61-63]. This is a consequence of the disparate geometric, electronic and electrostatic requirements imposed by ligand binding and dissociation. The local and global conformational changes and dynamic electrostatic interactions in the binding pocket modulate not only ligand affinities, but also the discrimination between similar diatomic ligands [64, 65]. The large temperature dependence of the observed rate of NO dissociation from ferric heme in nNOS and other heme proteins may reflect the control of NO binding at multiple levels. At the heme level, heme distortions have been shown to be critical for stabilizing the FeIIINO complex, particularly in heme proteins whose physiological functions depend on controlling NO release from FeIII-heme . In NOS proteins, the observed out-of-plane modes in the low frequency regions of Raman spectra of the FeIIINO species (oxygenase domain of the inducible nitric oxide synthase and oxygenase domain of the endothelial nitric oxide synthase) have been assigned to heme distortions and have been shown to be critical in increasing dxy-dπ orbital overlap, which stabilizes the FeIIINO species . At the global protein level, it has been suggested that inter-subunit interactions in the NOS dimer may be important in modulating heme distortion modes , which in turn would affect the heme–NO dissociation rate kd. Also, point mutagenesis studies on NOS enzymes identified amino acids near the opening of the heme pocket that help to kinetically gate release of NO [25, 66, 67]. Thus, the large dependence of kd on the temperature is likely a result of a general temperature effect on both the heme and protein structural fluctuations and dynamics.
Reaction of O2 with the enzyme FeIINO complex (kox)
During NO synthesis a portion of the FeIIINO product complex is reduced by the NOS reductase domain and the resulting FeIINO complex must then react with O2 to return to the catalytic cycle. The rate of this reaction, termed kox, is a kinetic parameter (Fig. 1) that impacts NOS enzyme distribution during catalysis and, consequently, its steady-state behaviour and NO synthesis activity [19, 68-70]. We determined kox values by mixing anaerobic samples of the nNOSoxy FeIINO species with an air-saturated solution in the stopped-flow spectrophotometer at 5, 10, 15, 20, 25, 30 and 37 °C and then monitoring the reactions using a diode array detector. Figure 5 (upper panel) shows representative spectral data collected during the reaction of the nNOSoxy FeIINO complex with air-saturated buffer in the presence of both H4B and l-Arg at 10 °C. The spectral changes are consistent with conversion of the FeIINO complex into the FeIII high-spin form of nNOSoxy. The reaction was a single-step process with no discernible accumulation of intermediate species, as indicated by the several isosbestic points in the spectra (Fig. 5, upper panel) and by the single-exponential decay of the absorbance signal at two different wavelengths (Fig. 5, upper panel inset), which followed the rate of FeIINO complex disappearance (436 nm) and the rate of ferric enzyme formation (393 nm). The spectral and fitting results obtained for reactions run at other temperatures were highly similar (data not shown) and the observed kox rates are listed in Table 1. The corresponding Eyring plot (Fig. 5, lower panel) and Arrhenius equation were used to calculate the thermodynamic parameters: Ea = 31.9 kJ·mol−1; ΔH‡ = 29.4 kJ·mol−1; ΔS‡ = −161 J·mol−1·K−1) (Table 2).
The activation enthalpy for oxidation of the nNOS ferrous heme complex is much lower than the values reported for oxidation of ferrous heme-nitrosyl haemoglobin and myoglobin or for their various mutants (ΔH‡ = 110–120 kJ·mol−1) [61, 71]. Their higher activation enthalpies have been rationalized in terms of a rate-limiting NO dissociation step governing the kinetics of the reversible ligand exchange reaction with O2 that precedes the irreversible oxidation . By contrast, the relatively low ΔH‡ (29.4 kJ·mol−1) that we measured for oxidation of the nNOS ferrous heme-nitrosyl species effectively argues against a NO dissociative mechanism being involved in this case because it is insufficient for breaking the strong heme-FeIINO bond. The relatively low activation barrier for the nNOS reaction supports our recent mechanistic investigation of nNOS and iNOS heme-nitrosyl oxidation reactions, suggesting that it involves a comparatively fast and direct reaction between the ferrous heme-nitrosyl and dioxygen .
Steady-state NO synthesis activity
The steady-state NO synthesis activity is a combined reflection of all the individual steps in Fig. 1; thus, its temperature dependence provides apparent thermodynamic parameters for the global process. Table 1 contains NO synthesis activities that we measured across the temperature range for the present study (5–37 °C). The activity increased by approximately 12-fold over this range. The activity data gave a linear Eyring plot (Fig. 6), from which we determined the activation parameters: Ea = 54.1 kJ·mol−1; ΔH‡ = 51.7 kJ·mol−1; ΔS‡ = −74 J·mol−1·K−1 (Table 2). Our Ea value differs somewhat from an earlier study that reported an activation energy of 79.4 kJ·mol−1 for the steady-state activity of nNOS, which was derived using measures of citrulline formation from [14C]l-Arg .
Implications for nNOS catalytic behaviour
Because the five kinetic parameters investigated in the present study differed with respect to their temperature dependencies, this means that the distribution pattern of nNOS enzyme species during steady-state NO synthesis, and thus nNOS catalytic behaviours, will be temperature-dependent. We utilized the kinetic values that we measured for the five parameters at three different temperatures (10, 25 and 37 °C) in computer simulations of the global kinetic model (Fig. 1)  to explore this concept. The model reports on the five main enzyme species that are present during steady-state NO synthesis, namely the ferric, ferrous, ferrous-O2 (or ferric-superoxy), ferric-NO and ferrous-NO forms (Fig. 1). Any additional kinetic values for associated reactions (i.e. O2 binding) that we needed for the simulations, besides the five kinetic parameters measured in the present study, are described in the 'Experimental procedures'.
At 10 °C, the simulation showed that the fast kr relative to the slower kd and kox values causes the nNOS enzyme to exist predominantly as the FeIINO species during the steady-state [19, 21]. The modelled enzyme distribution indicated that approximately 10% of nNOS is present as ferric enzyme and approximately 86% is present as the FeIINO species (Fig. 7) at 10 °C, with the remaining 4% of nNOS mainly distributed between the FeIIO2 and FeIIINO species. This falls close to estimates reported in literature [16, 19, 21, 72, 73], which range between 67% and 85% FeIINO complex during steady-state catalysis at 10 °C. However, our experimental data reveal that, at higher temperatures, the rate of FeIIINO dissociation (kd) increases significantly faster than the kr and the other kinetic parameters. The simulations indicate how this will alter the steady-state enzyme distributions: At 25 °C, approximately 16% of nNOS is present as ferric enzyme and approximately 81% is present as the FeIINO species (Fig. 7), with the remaining 3% of nNOS mainly distributed between the FeIIO2 and FeIIINO species, whereas, at 37 °C, which is near the physiological temperature of rats (38 °C) [74, 75], only approximately 73% of nNOS is present as the FeIINO species and approximately 24% is present as ferric enzyme. These simulations clearly show that rat nNOS gains in ferric character and has less FeIINO complex build-up during NO synthesis as the temperature increases. Overall, the different thermodynamic behaviours of the individual kinetic steps will cause nNOS to become more and more efficient at NO release as the temperature increases, at the same time as diminishing the enzyme cycling into the futile NO dioxygenase pathway (Fig. 1).
A related consequence of the different temperature dependencies is that the relative contributions of the individual kinetic steps toward limiting the NO synthesis activity, and in determining the catalytic behaviours of the enzyme, will change with temperature. For example, at low temperatures (e.g. 10 °C), the kd and kr have similar magnitudes. This equivalence diminishes the release of NO as a result of approximately half of the FeIIINO product complex being diverted into the futile cycle. Consequently, this increases the impact of kox on the observed enzyme NO synthesis activity and on its apparent oxygen concentration dependence (i.e. apparent KmO2). Specifically, NO synthesis activity becomes much more dependent on the oxygen concentration relationship of the NO dioxygenase reaction described by kox , which displays a linear relationship with O2 concentration across the entire physiological range . However, at higher temperatures, the kd becomes increasingly faster than the kr, which enables a quicker and greater release of NO from the FeIIINO product complex, consequently diminishing enzyme partitioning into the futile cycle. This in turn diminishes the role of kox in limiting the overall NO synthesis activity and diminishes its role in determining the apparent KmO2 for NO synthesis activity, such that the apparent KmO2 becomes more reflective of the oxygen concentration dependence of the biosynthetic reactions, which depend on the ferrous heme affinity for O2 binding. Thus, at higher temperatures (i.e. 37 °C) the contribution of kr in determining the overall NO synthesis activity is increased, whereas the contribution of kox is diminished, with accompanying changes in nNOS enzyme behaviour and the O2 concentration response.
General methods and materials
All reagents and materials were obtained from Sigma-Aldrich (St Louis, MO, USA), Amersham Biosciences (Little Chalfont, UK) or other sources, as described previously [20, 70, 76]. Absorption spectra and steady-state kinetic data were obtained using a Shimadzu UV-2401PC spectrophotometer (Shimadzu Corp., Kyoto, Japan). Stopped-flow experiments were performed using a Hi-Tech Scientific KinetAsyst SF-61DX2 stopped-flow system (Hi-Tech Scientific, Salisbury, UK) equipped with an anaerobic set-up and a rapid scanning photodiode array detector. Data from multiple identical stopped-flow experiments were averaged to improve the signal-to-noise ratio. We did not find much difference in values when individual rates were averaged or the rate was calculated from averaged trace. We calculated SD values from the individual observed rates. The spectral traces were fit according to single or multiple exponential equations using software provided by the manufacturer. The best fit was determined when adding further exponentials did not produce a significant improvement of the fit as judged from the residuals. All plots and some additional curve-fitting were performed using origin, version 8.0 (OriginLab, Northampton, MA, USA). Anaerobic samples were prepared in an air-tight cuvette using repeated cycles of vacuum followed by a positive pressure of catalyst-deoxygenated nitrogen gas.
Expression and purification of wild-type protein
All proteins were purified in the presence of H4B and l-Arg as described previously [20, 26, 70, 76]. The ferrous heme-CO adduct absorbing at 444 nm was used to measure heme protein content with an extinction coefficient of ε444 = 74 mm−1·cm−1 (A444–A500) . The purity of each protein was assessed by SDS/PAGE and spectral analysis.
NO synthesis measurement
Steady-state NO synthesis was determined at 5, 10, 15, 20, 25, 30 and 37 °C using the spectrophotometric oxyhaemoglobin assay, as described previously .
Anaerobic heme reduction measurements
The kinetics of heme reduction were analyzed at 5, 10, 15, 20, 25, 30 and 37 °C, as described previously [20, 26, 70, 76, 77], using a stopped-flow apparatus and diode array detector (Hi-Tech Scientific KinetAsyst SF-61DX2) equipped for anaerobic analysis.
Reaction of FeIINO complexes with oxygen
The kinetics of FeIINO oxidation were analyzed at 5, 10, 15, 20, 25, 30 and 37 °C, as described previously [24, 70], using a stopped-flow apparatus and diode array detector (Hi-Tech Scientific KinetAsyst SF-61DX2) equipped for anaerobic analysis.
Single turnover reactions
Arginine hydroxylation experiments were carried out in a Hi-Tech SF61-DX2 stopped-flow instrument coupled to a diode array detector, as described previously . The reactions were studied at 5, 10, 15, 20, 25, 30 and 37 °C in the presence of 50 μm H4B, 2.5 mm l-arginine, 150 mm NaCl, 10% glycerol, 1 mm dihiothreitol in 40 mm Epps buffer (pH 7.6). nNOSoxy at a concentration of 10 μm was titrated with sodium dithionite and mixed with air-saturated buffer at the respective temperatures ([O2] = ~ 360 μm, 270 μm and 215 μm at 10, 25 and 37 °C, respectively). NOHA oxidation was studied in the same conditions replacing l-arginine by 1 mm NOHA. Sequential spectral data were fitted to A→ B→ C (Arg hydroxylation; FeII→ FeIIO2→FeIII) and A→ B→ C→ D (NOHA oxidation; FeII→ FeIIO2→FeIIINO→FeIII) kinetic models using specfit/32 global analysis software, version 3.0 (Spectrum Software Associates, Marlborough, MA, USA) to obtain the spectra of the different species and the reaction rates. Using specfit/32, we calculated rates for kcat1 (during Arg hydroxylation) and kcat2 (during NOHA oxidation).
Temperature-dependence and thermodynamic analysis
The natural log of the observed rates or rate constants for each averaged set of experimental data were plotted against the reciprocal of the absolute temperature. To calculate the activation energy, the data were then fit to the Arrhenius equation (Equation (3)) using the linear fitting function in origin. In Equation (3), A is the Arrhenius pre-exponential factor and R is the gas constant and Ea is activation energy. Activation enthalpy (ΔH‡) and activation entropy (ΔS‡) were calculated on the basis of Eyring equation (Equation (4)) and Eyring plot [ln(k/T) versus 1/T]. Because the Gibbs free energy of activation (ΔG‡) depends on temperature, it can be calculated for each step at a given temperature and we calculated ΔG‡ at 25 °C using the Equation (5), whereas, ΔS‡ and ΔH‡ are virtually independent of temperature:
where k is the reaction rate constant; T is the absolute temperature; R is the gas constant (8.314472 J mol−1·K−1); kB is the Boltzmann constant (1.3806504 × 10−23 J·K−1); h is Planck's constant (6.626 × 10−34 J·s−1); ΔH‡ is the enthalpy of activation; ΔS‡ is the entropy of activation; ΔG‡ is the Gibbs free energy of activation; and Ea is the energy of activation.
Simulations of nNOS distribution during steady-state NO synthesis
The NO synthesis kinetics were simulated using the global model as implemented in mathcad, version 7.0 [18, 21] or gepasi, version 3.30 . Slight changes in the model were introduced to account for the pterin reduction step, as shown in Fig. 1. Both methods yielded similar results. Simulations assume constant values for [O2] = ~ 360 μm, 270 μm and 215 μm at 10, 25 and 37 °C, respectively, and [NADPH] = 40 μm. This approximation allows to treat all the processes as unimolecular reactions and assimilate the observed rates to apparent rate constants. Thus, for bimolecular reactions k2, k6 and k10 (Table 3), the observed rate is the product of the real rate constant multiplied by the concentration of oxygen. We base our simulations on the observed rates using a correction factor to estimate the value in air-saturated buffer instead of the half air-saturated experimental conditions. Experimentally observed oxidation rates (kox), which we derived at half air-saturated condition, were multiplied by a factor of 1.5 to approximate a full air-saturated condition according to the dependence observed previously . We also multiplied the observed oxygen-binding rate k2 and k6 values by a factor of 2.0 to obtain a full air-saturated condition assuming that the rate is proportional to [O2]. The values for heme reduction (kr), NO dissociation (kd), FeIINO oxidation (kox) and all other rates at different temperatures are shown in Table 3. The percentage of species in the steady-state is calculated after 50 s of simulation, although a steady-state distribution is reached within 5–10 s in all cases.
Table 3. Calculated rates used for the computer simulations. (a) l-Arg-bound enzyme; (b) NOHA-bound enzyme; (b*) ferric enzyme with NOHA and H4B radical bound; for details, see Fig. 1 and Fig. S1. All values are in s−1. For bimolecular reactions (k2, k6, k10), the value used in the calculations is the product of the ‘actual’ value and the concentration of oxygen, which is considered to remain constant during the reaction. For k2 and k6, the values used are the experimental values multiplied by 2 (assumes the rate is proportional to [O2], experimental values are determined in half air-saturated conditions). For k10, the values used are the experimentally determined values × 1.5, as based on the observed oxygen dependence reported in Tejero et al. . k11 values are interpolated from Salerno 
FeIIIa → FeII(a)
FeII(a) (+ O2) → FeIIO2(a)
FeIIO2(a) → FeIII(b*)
FeIII(b*) → FeIII(b)
FeIII(b) → FeII(b)
FeII(b) (+ O2) → FeIIO2(b)
FeIIO2(b) → FeIIINO
FeIIINO → FeIII(a) + NO
FeIIINO → FeIINO
FeIINO (+ O2) → FeIII(a) + NOx
FeIINO → FeII(a) + NO
This work was supported by National Institutes of Health Grants GM51491 and HL58883 to D.J.S. We also acknowledge funding by The National Science Foundation to MB (Grant CHE-0848820).