Nitric Oxide Binding to Prokaryotic Homologs of the Soluble Guanylate Cyclase β1 H-NOX Domain*

The heme cofactor in soluble guanylate cyclase (sGC) is a selective receptor for NO, an important signaling molecule in eukaryotes. The sGC heme domain has been localized to the N-terminal 194 amino acids of the β1 subunit of sGC and is a member of a family of conserved hemoproteins, called the H-NOX family (Heme-Nitric Oxide and/or OXygen-binding domain). Three new members of this family have now been cloned and characterized, two proteins from Legionella pneumophila (L1 H-NOX and L2 H-NOX) and one from Nostoc punctiforme (Np H-NOX). Like sGC, L1 H-NOX forms a 5-coordinate FeII-NO complex. However, both L2 H-NOX and Np H-NOX form temperature-dependent mixtures of 5- and 6-coordinate FeII-NO complexes; at low temperature, they are primarily 6-coordinate, and at high temperature, the equilibrium is shifted toward a 5-coordinate geometry. This equilibrium is fully reversible with temperature in the absence of free NO. This process is analyzed in terms of a thermally labile proximal FeII-His bond and suggests that in both the 5- and 6-coordinate FeII-NO complexes of L2 H-NOX and Np H-NOX, NO is bound in the distal heme pocket of the H-NOX fold. NO dissociation kinetics for L1 H-NOX and L2 H-NOX have been determined and support a model in which NO dissociates from the distal side of the heme in both 5- and 6-coordinate complexes.

The H-NOX (Heme-Nitric Oxide and/or OXygen-binding domain) family of heme proteins has been identified recently, first through sequence analysis (1) and then through biochemical characterization (2). This family was identified based on homology to the heme domain from soluble guanylate cyclase (sGC), 3 the well characterized and conserved eukaryotic nitric oxide receptor (3,4). H-NOX proteins have now been identified in many prokaryotes in addition to the well known eukaryotic sGCs. Several of these prokaryotic H-NOX proteins have now been cloned, expressed, and spectroscopically characterized (2,(5)(6)(7). In addition, the crystal structure of the H-NOX domain from Thermoanaerobacter tengcongensis (Tt H-NOX) has been solved to 1.77 Å resolution (8), providing the first structural data for the H-NOX family and, by homology, for the heme domain in the ␤-subunit of sGC. Taken together, these structural and biochemical results have suggested that some members of this family are able to use a homologous protein fold and an identical heme cofactor to discriminate between NO and O 2 binding.
Although a great deal of information pertaining to the entire H-NOX family has been determined from studying the structure of Tt H-NOX (5,8), specific structural data for an NO complex to any of the H-NOX proteins are still lacking. This is particularly relevant as NO is an important signaling molecule in eukaryotes, and sGC is the only firmly established NO receptor (3). Cytochrome cЈ is a protein that shares ligand binding properties but not sequence or structural homology with sGC. A crystal structure of the Fe II -NO cytochrome cЈ complex shows NO bound to the proximal side of the heme (9), which has led to speculation that the same is true for sGC. Physical studies of NO binding to sGC have been hindered by poor expression systems for this large protein. Thus considerable effort has been spent on shorter constructs of sGC containing its heme-binding domain, such as ␤1-(1-385) (10) and ␤1-   (6). Most of the prokaryotic members of the H-NOX family, however, are stand-alone ϳ200 amino acid proteins that overexpress well in Escherichia coli. These close homologs of the sGC heme domain have proven generally useful in understanding ligand binding in the family.
In this study, we examine three additional prokaryotic H-NOX proteins from Legionella pneumophila (ORF1 and ORF2) and Nostoc punctiforme (Np H-NOX). Unlike all other prokaryotes with predicted H-NOX proteins identified to date, L. pneumophila has two predicted ORFs for H-NOX proteins (L1 H-NOX and L2 H-NOX). Np H-NOX is 39% identical to the H-NOX domain from sGC, and the H-NOX domains L1 H-NOX and L2 H-NOX share 19 and 16% identity with sGC, respectively. The NO binding characteristics of these proteins will be important in determining function in these bacteria. In addition, given the high homology these proteins share with sGC, these results should have implications for the NO-heme complex of sGC.

EXPERIMENTAL PROCEDURES
Materials and General Methods-Unless otherwise noted, all reagents were purchased in their highest available purity and used as received.
Protein Expression and Purification-PCR was used to amplify ZP_421786 from N. punctiforme genomic DNA (ATCC, Manassas, VA) using Expand polymerase (Roche Applied Science). Upstream and downstream primers contained NdeI and NotI restriction sites, respectively. PCR was used to amplify YP_095089 (L1 H-NOX) and AAU28519 (L2 H-NOX) from L. pneumophila genomic DNA (ATCC) using Expand polymerase (Roche Applied Science). Upstream and downstream primers contained NdeI and XhoI restriction sites, respectively. All amplified PCR products were cloned into pET-20b (Novagen) and sequenced (sequencing core; University of California, Berkeley). Mutagenesis was carried out using the QuikChange protocol from Stratagene. Cell culture procedures and purification of Np H-NOX were carried out as described previously for the H-NOX protein from Vibrio cholerae (2). Cell culture procedures of both Legionella H-NOX proteins were carried out as described previously (2). Purification of the Legionella H-NOX proteins took advantage of the C-terminal His 6 tag; proteins were purified by metal affinity (nickel-nitrilotriacetic acid) followed by gel filtration (Superdex 200 HiLoad 26/60).
Sample Preparation-Preparation of the various protein complexes was carried out as published previously (2) with one minor exception, as follows. Rather than generate the Fe II -NO complex with NO gas generated from the head space of a concentrated diethylamine NONOate (Cayman) solution, diethylamine NONOate was added directly to the protein solution, and then excess NO, diethylamine, and diethylamine NONOate were removed from the sample by using a PD10 desalting column. This was to ensure no free NO was in solution, and thus the only NO present in the sample was that bound to the heme in the protein.
Spectroscopy-All electronic spectra were recorded on a Cary 3E spectrophotometer equipped with a Neslab RTE-100 constant temperature bath. For temperatures at 0°C or lower, ethylene glycol (50%) was added to the constant temperature bath and 5% glycerol to the protein sample. Resonance Raman spectra were collected using 406.7 nm excitation from a Kr ϩ laser (Spectra-Physics model 2025). Raman scattering was detected with a cooled, back-illuminated CCD (LN/CCD-1100/PB; Roper Scientific) controlled by an ST-133 controller coupled to a subtractive dispersive double spectrograph. The laser power at the sample was ϳ2 milliwatts. A microspinning sample cell was used to minimize photo-induced degradation. For the temperature dependence studies, the sample temperature was controlled by flowing either cooled (ϳ10°C) or heated (ϳ40°C) N 2 gas over the Raman cell. Samples were equilibrated at the respective temperatures for 30 -60 min prior to data acquisition. Typical data acquisition times ranged from 30 to 60 min, except for the L1 H-NOX NO-complex, which was signal-averaged for 3 h because of a high fluorescence background. Electronic absorption spectra were obtained both before and after the Raman experiments to ensure that sample integrity was maintained. Raman spectra were corrected for wavelength dependence of the spectrometer efficiency, and cyclohexane was used for instrument calibration. The reported frequencies are accurate to Ϯ2 cm Ϫ1 , and the resolution of the spectra is 8 cm Ϫ1 . For each Raman spectrum, the raw data were base-linecorrected, and the buffer background signal was subtracted.
Extinction Coefficient Determination-The extinction coefficients were determined similarly to the methods described previously for Tt H-NOX and Vc H-NOX (2). Specifically, for the 6-coordinate Fe II -NO complex of L2 H-NOX, the electronic spectra of various dilutions of a sample of the aerobic Fe II -NO complex (prepared as described above) at 0°C were recorded. The electronic spectra of dilutions of a sample of horse heart metmyoglobin (⑀ 409 nm ϭ 181 cm Ϫ1 mM Ϫ1 ) were also recorded and used as a standard for heme concentration. The heme content of each sample was determined by HPLC (Hewlett Packard Series II 1090 HPLC with a diode array detector). Each sample (75 l) was applied to a C4 column (250 ϫ 4.6 mm, 5 m; Vydac) that had been equilibrated with 0.1% trifluoroacetic acid. The column was developed with a linear gradient of 0 -75% acetonitrile over 25 min followed by a linear gradient of 75-100% acetonitrile over 5 min. The column was washed and re-equilibrated between runs with a gradient of 100-0% acetonitrile over 3 min followed by 5 min of 100% aqueous phase (0.1% trifluoroacetic acid). The flow rate was 1 ml/min. NO Dissociation Rate-NO dissociation rates were measured as described previously (5). Briefly, Fe II -NO complexes of protein (5 M heme final concentration) diluted in anaerobic 50 mM triethanolamine, 50 mM NaCl, pH 7.5, buffer were rapidly mixed with a saturated carbon monoxide and 30 mM (final concentration) dithionite trap (Na 2 S 2 O 4 ) in the same buffer (anaerobic) (11,12). It has been established previously that CO binding is not rate-limiting in these experiments (11); this was confirmed in experiments using only 30 mM Na 2 S 2 O 4 without CO as a trap. Data were acquired by scanning periodically on a Cary 3E spectrophotometer equipped with a Neslab RTE-100 constant temperature bath set to varying temperatures (0 -70°C). The dissociation of NO from the heme was monitored as the formation of the Fe II -CO complex at 423 nm. Difference spectra were calculated by subtracting the first scan from each subsequent scan. The NO dissociation rate was determined from the increase in absorbance at 423 nm versus time and fit with a single or two parallel exponentials of the form f(x) ϭ A ϫ (1 Ϫ e Ϫkx ). Each experiment was performed a minimum of six times, and the resulting rates were averaged. The dissociation rates measured are independent of CO and dithionite concentration (3, 30, and 300 mM dithionite were tested).
CO, and Fe II -NO complexes at room temperature are shown in Fig. 1 and compared with sGC, Tt H-NOX, and other histidylligated heme proteins in Table 1. Interestingly, although the Fe II -unligated and CO complexes of each of these proteins are similar to sGC and all other H-NOX proteins characterized to date (Table 1), there are some significant differences in the Fe II -NO complexes. Specifically, L1 H-NOX forms a 5-coordinate NO complex with a characteristic Soret absorbance maximum at 398 nm (Fig. 1B) like sGC, but L2 H-NOX (Fig. 1C) and Np H-NOX (Fig. 1A) appear to be composed of a mixture of 5-and 6-coordinate Fe II -NO complexes at 20°C (399 and 416 nm, respectively). L2 H-NOX is a relatively evenly distributed mixture, whereas Np H-NOX appears to be primarily a 6-coordinate complex with a small shoulder corresponding to the 5-coordinate complex. Given the degree of homology of these proteins to sGC (Np H-NOX is 39% identical), which exclusively forms a 5-coordinate Fe II -NO complex, an additional investigation into the NO-binding characteristics of these proteins was carried out.
Resonance Raman Spectroscopy-The resonance Raman spectra of the Fe II -NO complexes of L2 H-NOX, L1 H-NOX, and Np H-NOX are shown in Fig. 2. Table 2 details the assignment of the major heme skeletal modes and compares them to sGC, Tt H-NOX, and other histidyl-ligated heme proteins. The Raman spectra confirm observations made in the electronic absorption spectra that L1 H-NOX forms a 5-coordinate NO complex, whereas L2 H-NOX and Np H-NOX have mixed coordination states, the specifics of which are discussed below. The -electron density marker, 4 , is used to determine the oxidation state of the heme. The spin and coordination state markers, 3 , 2 , and 10 , are sensitive to the core size of the heme macrocycle (13). Typical frequency shifts of 3-15 cm Ϫ1 are often observed upon switching from 5-to 6-coordinate complexes for some of these skeletal markers.
In the low frequency spectrum of L2 H-NOX ( Fig. 2A), two bands at 521 and 550 cm Ϫ1 are assigned to the Fe-N stretching modes ( Fe-N ) for 5-and 6-coordinate NO complexes, respectively, based on their similarity to values reported previously for this vibrational mode (2,14,15). In the high frequency region of L2 H-NOX (Fig. 2D), 3 , 2 , and 10 are primarily positioned at 1502, 1583, and 1633 cm Ϫ1 , respectively, and are characteristic of histidyl-ligated, 6-coordinate, low spin, Fe II -NO heme complexes such as myoglobin and Tt H-NOX (Table 2). However, there are also shoulders that appear at 1508 and 1646 cm Ϫ1 , which correspond well to the marker values typically observed for the 5-coordinate nitrosyl heme complexes, such as sGC (Table 2). Together, these spectra strongly suggest that L2 H-NOX is a mixture of 5-and 6-coordinate complexes under room temperature conditions (ϳ20°C).  Although the signal-to-noise ratio is lower for L1 H-NOX because of a high fluorescence background (Fig. 2, B and E), the main skeletal modes are still discernible. Fig. 2B shows a band at 522 cm Ϫ1 assigned to the Fe-N stretching mode based on similarity to the 5-coordinate ferrous nitrosyl complexes in sGC (525 cm Ϫ1 ) and V. cholerae H-NOX (Vc H-NOX; 523 cm Ϫ1 ). The high frequency region of L1 H-NOX shows 3 , 2 , and 10 at 1507, 1581, and 1643 cm Ϫ1 , respectively (Fig. 2E). These observed vibrational frequencies for L1 H-NOX are consistent with other 5-coordinate Fe II -NO heme complexes (2,14,15) and are clearly shifted from the observed skeletal markers for 6-coordinate nitrosyl complexes (Table 2). Fig. 2, C and F, presents the low and high frequency resonance Raman spectra for the Fe II -NO complex of Np H-NOX. Our results indicate that the nitrosyl complex for Np H-NOX is primarily 6-coordinate, with Fe-N assigned at 559 cm Ϫ1 based on previous work (6, 15) (isotope data not shown); the spin and coordination state markers were observed at 1499, 1580, and 1632 cm Ϫ1 for 3 , 2 , and 10 , respectively. Closer inspection of the high frequency spectra show small, broad shoulder bands for both 3 and 10 at 1506 and 1645 cm Ϫ1 , supporting the appearance of a mixture of 5-and 6-coordinate complexes in the electronic absorption spectra at ϳ20°C.
Temperature Dependence of the 5-and 6-Coordinate NO Complexes-To investigate the mixture of 5-and 6-coordinate complexes, temperature-dependent studies of the Fe II -NO complex of Np H-NOX and L2 H-NOX were carried out. Electronic absorption spectra indicate that both Np H-NOX and L2 H-NOX are in equilibrium between 5-and 6-coordinate complexes at physiologically relevant temperatures, as indicated by the Soret max shift with temperature ( Fig. 3). The temperature was varied between 1 and 45°C in the same sample several times showing that this 5-to 6-coordination behavior was fully reversible. Furthermore, the temperature-dependent coordination states are independent of NO concentration for both H-NOX proteins; no excess NO was added to any of the samples; the only NO in solution was that bound to the heme at the beginning of the experiment. The same results are obtained when the experiment is carried out anaerobically, indicating that O 2 has no effect on this process.
Quantification of the amount of 5-and 6-coordinate complexes present at each temperature requires measurement of the extinction coefficient; however, the only complex that can be isolated to accurately measure the extinction coefficient was L2 H-NOX at 0°C. Thus the extinction coefficient of the 6-coordinate NO complex of L2 H-NOX was determined to be 139 mM Ϫ1 cm Ϫ1 , which was then used to estimate the mixture of each complex at a given temperature. At 40°C, there is approximately a 50% mixture of each coordination state in the Fe II -NO complex of L2 H-NOX. Even at Ϫ12°C, L1 H-NOX shows no evidence of a 6-coordinate NO complex, although it is expected that at a sufficiently low temperature, the NO complex of L1 H-NOX would also convert to the 6-coordinate species.
The temperature-dependent behavior was further investigated for L2 H-NOX between ϳ10 and ϳ40°C by using resonance Raman spectroscopy. The high frequency spectra obtained at ϳ10°C for L2 H-NOX show 3 , 2 , and 10 at 1500, 1586, and 1633 cm Ϫ1 , respectively. In corroboration with the electronic absorption spectra (Fig. 3B), these observed vibrational frequencies show a shift in the equilibrium population toward a 6-coordinate complex with decreasing temperature. The spin state marker, 3 , is clearly split with overlapped bands at 1502 and 1508 cm Ϫ1 upon increasing the temperature to ϳ20°C (Fig. 4, B and E). The vibrational frequency for 3 increases to 1510 cm Ϫ1 when the temperature is raised to ϳ40°C, supporting an equilibrium shift toward the 5-coordinate complex (Fig. 4, C and F). A temperature-dependent   AUGUST 4, 2006 • VOLUME 281 • NUMBER 31

Characterization of NO Binding to H-NOX Domains
shoulder is also observed for 10 at 1646 cm Ϫ1 with increasing temperature (Fig. 4, D-F). This shift in the vibrational frequency for 10 has similarly been observed for 5-and 6-coordinate heme-NO complexes in cytochrome cЈ (16,17) and FixLN (18). Dissociation Rates for NO from H-NOX Domains-NO dissociation rates (k off(NO) ) were measured to further characterize the nature of the 5-and 6-coordinate NO complexes L1 H-NOX, L2 H-NOX, and Np H-NOX. A CO and dithionite trap (11,12,19) for the released NO, consisting of saturating CO and 30 mM dithionite, was employed to minimize recombination of the dissociated NO. The rate of NO dissociation was followed by the formation of the Fe II -CO complex at 423 nm. The experiment was repeated at several different temperatures (0 -40°C) in order to vary the coordination state of the starting complex, and all measured rates were independent of CO and dithionite at all concentrations tested (3-300 mM).
Representative data for L2 H-NOX are shown in Fig. 5, and Table 3 summarizes the data for the H-NOX proteins in this study as well as other Fe II -NO heme proteins. The following reaction can be used to describe the kinetic data at all temperatures. This is based on the published solutions for similar three-state systems (20,21). NO dissociation from the H-NOX domain is assumed to proceed as shown in Reaction 1, where A is the 5-coordinate NO complex; B represents the 6-coordinate NO complex, and C represents the CO-trapped dissociated NO complex. k 1 represents the rate of histidine rebinding, k 2 the rate of histidine dissociation, and k 3 the rate of NO dissociation. The first-order observed rate constant, k m , can be readily derived from Reaction 1 provided that A and B are in equilibrium (21) as shown in Equation 1, If B is the only or the dominant species, then Equation 2 is the result, If A is the only or the dominant species, then the rate constant is as shown in Equation 3, or Equation 4, This same analysis was used to treat all of the data regardless of the coordination state of the starting complex. In the L2 H-NOX system, at low temperature when the starting complex is dominated by the 6-coordinate complex B, only one time constant for NO dissociation is measured, which is k 3 in Equation 2 and is assigned unambiguously as k off . At higher temperatures, there is an ϳ50:50 mixture of 5-and 6-coordinate complexes, and two parallel time constants are measured (called k m1 and k m2 ). k m1 is the same rate seen at lower temperatures (assuming Arrhenius temperature dependence) and is thus assigned as k 3 (k off ); this rate represents the portion of 6-coordinate molecules that are present in the starting mixture as complex B, analogous to the experiments at lower temperatures. The other rate, k m2 , is either k 1 or (k 1 /k 2 ) ϫ k 3 from Equations 3 and 4 and represents the portion of the molecules that are present in the starting mixture as the 5-coordinate complex A. Most importantly, the amplitude of each rate constant represents 50% of the total absorbance change, which correlates well with the known abundances of 5-and 6-coordinate complexes present in solution at 40°C, as estimated using the extinction coefficient of the 6-coordinate complex. Also, by taking into account a doubling of the rate for each 10°C, k 3 (k off ) is exactly the same at all temperatures, strongly supporting the proposed mechanism in Reaction 1. The average rates and amplitudes of k m1 (equivalent to k 3 and reported as k off in Table 3) from six independent experiments are 5.6 ϫ 10 Ϫ4 s Ϫ1 (100%) at 0°C, 10.6 ϫ 10 Ϫ4 s Ϫ1 (100%) at 10°C, and 87.6 ϫ 10 Ϫ4 s Ϫ1 (46%) at 40°C. The average rate and amplitude of k m2 (equivalent to either k 1 or (k 1 /k 2 ) ϫ k 3 ) at 40°C are 14.3 ϫ 10 Ϫ4 s Ϫ1 (54%).
For L1 H-NOX, at all temperatures, two parallel exponents (called k m1 and k m2 ) are needed to fit the data despite the fact that there is only one starting coordination state as assessed by UV-visible and resonance Raman spectroscopy. An F test was used to compare the data fit with one and two exponentials. This test confirms that the probability of the two exponential fit being correct is 100% (p value Ͻ0.0001). If the NO dissociates according to the mechanism in Reaction 1, however, the kinet- ics should fit to a single exponential with a k m , defined by either Equation 3 or 4. The simplest way to explain the requirement for two time constants is to invoke a second species of A that is kinetically distinct, leading to Reaction 2, where A and A* are kinetically distinguishable isomers but that presumably both go through intermediate B before being trapped as C. Reaction 1 is just a subset of Equation 4; fundamentally they are the same kinetic mechanism. This mechanism is based on a published analysis of hydrogen exchange kinetics in proteins in which two kinetically distinct native conformations of a protein are in conformational equi-librium and can both undergo hydrogen exchange through some intermediate species (21). This analysis results in two observed time constants (k m1 and k m2 ), one with a rate constant of k a and the other with a rate constant equal to that of k m , as defined in Equation 1.
NO dissociates from L1 H-NOX with two time constants, defined k m1 and k m2 . One of these, k m1 , is the time constant representing the portion of the starting 5-coordinate complex present at the beginning of the experiment as A that proceeds with a time constant of k m from Equation 1, which should simplify to either Equation 3 or 4. The other, k m2 , represents the 5-coordinate complex present as A* that proceeds with a time constant of k a from Reaction 2. Assuming a doubling of the rate for every 10°C, the same two processes k m1 and k m2 were measured at each temperature. The slower of the two measured rates is assigned as the dissociation rate in Table 3, not because it can be assigned to k 3 in Reaction 2, but because it is the slowest rate in the dissociation process. Temperature dependences of the amplitudes of each process measured are consistent with A and A* being in equilibrium as described in Reaction 2. The average rates and amplitudes of k m1 (reported as k off in Table 3) from four independent experiments are (2.7 Ϯ 0.5) ϫ 10 Ϫ4 s Ϫ1 (88%) at 0°C, (4.7 Ϯ 1.0) ϫ 10 Ϫ4 s Ϫ1 (85%) at 10°C, and (38.5 Ϯ 3.4) ϫ 10 Ϫ4 s Ϫ1 (49%) at 40°C. The average rates and amplitudes of k m2 are (25 Ϯ 11) ϫ 10 Ϫ4 s Ϫ1 (12%) at 0°C, (44 Ϯ 17) ϫ 10 Ϫ4 s Ϫ1 (15%) at 10°C, and (370 Ϯ 150) ϫ 10 Ϫ4 s Ϫ1 (51%) at 40°C.
The Np H-NOX k off(NO) could not be obtained with the CO/dithionite trap, presumably because dithionite reacts with the bound NO complex (based on a dithionite concentration dependence; data not shown). This is the first example of an H-NOX nitrosyl complex that displays this reactivity with dithionite, suggesting this protein may have an altered distal pocket or electronic structure with NO. Therefore, an oxyglobin trap for released NO was employed to determine the NO dissociation rate (11,22). This was also not an effective trap for the NO kinetics of Np H-NOX, instead it seemed the NO dissociation rate was much slower than ambient globin oxidation under all experimental conditions attempted. This again suggests that in comparison with other characterized H-NOX family members, Np H-NOX displays some unique NO ligand binding characteristics.
L2 H-NOX Mutagenesis-Mutation of a distal pocket phenylalanine in L2 H-NOX to tyrosine (L2 F142Y) shifts the equilibrium of the Fe II -NO complex at 20°C exclusively to the 6-coordinate complex (Fig. 1D). As reported previously (5), the F142Y mutation in L2 H-NOX is also responsible for a gain of O 2 -binding function in the L2 H-NOX fold. The NO dissociation constant for L2 F142Y was determined at temperatures varying from 0 to 40°C and fit to Reaction 1 as described above for wild type L2 H-NOX. These data are summarized in Table 3. An exclusively 6-coordinate Fe II -NO complex is observed even at 40°C, as evidenced by electronic absorption spectroscopy and the requirement of only one exponent to fit the k off(NO) data.
The k off(NO) rates for Tt H-NOX and Tt Y140L are also included in Table 3 because they are relevant in the comparison of the effects of distal pocket mutagenesis on the geometry of Fe II -NO complexes in the H-NOX family. The ligand binding properties of Tt H-NOX have been characterized previously (2); both of these proteins have been kinetically studied in the context of O 2 binding in the H-NOX family (5), and the NO dissociation rates at 20°C (both are exclusively 6-coordinate complexes at this temperature) have been reported (5). Here  100%); B, 10°C (k m ϭ k 3 ϭ k off ϭ 10.6 ϫ 10 Ϫ4 s Ϫ1 , 100%); and C, 40°C (k m1 ϭ k 3 ϭ k off ϭ 83.0 ϫ 10 Ϫ4 s Ϫ1 , 46%; k m2 (ϭ either k 1 or (k 1 /k 2 ) ϫ k 3 ) ϭ 12.3 ϫ 10 Ϫ4 s Ϫ1 , 54%) measured by electronic absorption spectroscopy using saturating CO and 30 mM Na 2 S 2 O 4 as a trap for the released NO. Measured rates and amplitudes are indicated in parentheses and were independent of CO and dithionite at all concentrations tested (3-300 mM). Heme concentration was 3 M. For each temperature, the absorbance difference spectrum (the spectrum at time ϭ 0 min is subtracted from the spectrum at each subsequent time point) of the Fe II -CO complex growing over time is shown as well as a plot of the change in absorbance at 423-405 nm (the maximum and the minimum in the difference spectrum) versus time along with the exponential fit of that data. as a trap for the released NO k m1 is assigned as k off (NO) s Ϫ1 either because it could be conclusively assigned as k off (NO) or because it is the slowest rate in the NO dissociation mechanism. d Additional rate needed to fit the data, k m2(NO) s Ϫ1 at 20°C (estimated using a doubling of the rate for every 10°C), k m2 could be k 1 , (k 1 /k 2 ) ϫ k 3 or k a according to Reactions 1 and 2. e The NO complex at temperatures from 0 to 40°C is 6-coordinate, so only one rate, k off (NO) , is measured. f Data could not be measured with this technique. the NO dissociation was followed at temperatures between 0 and 70°C. Tt H-NOX very slowly transitions from a 6-coordinate to a 5-coordinate Fe II -NO complex at 70°C over the course of about 30 min. After more than 60 min at 70°C, however, the Tt Y140L mutant undergoes a transition from a 6-coordinate complex to only about an ϳ50:50 mixture of 5-and 6-coordinate complexes. Furthermore, the k off(NO) results for Tt H-NOX and Tt Y140L (Table 3) indicate that loss of the distal pocket tyrosine slows the dissociation rate, although not significantly.

DISCUSSION
The prokaryotic H-NOXs have greatly advanced our understanding of ligand binding in this family, including providing key insight into the ligand discrimination against O 2 in sGC (5). In this study, we have presented spectroscopic and kinetic data to characterize the NO ligand binding properties for three additional H-NOX family members, L1 H-NOX, L2 H-NOX, and Np H-NOX. L1 H-NOX is very similar to sGC and the previously characterized H-NOX protein from V. cholerae (Vc H-NOX), forming a stable 5-coordinate NO complex. Np H-NOX and L2 H-NOX, however, display some atypical behavior especially with respect to Fe II -NO complexes. Both of these proteins display a physiologically accessible thermal equilibrium between 5-and 6-coordinate Fe II -NO complexes. A temperature dependence on the coordination state of an Fe II -NO complex has been observed in other heme sensor proteins as well (18,23,24).
L. pneumophila is a facultative intracellular pathogen that is ubiquitous in warm fresh water, but this microbe can survive temperatures between 0 and 63°C. At the preferred physiological growth temperature (ϳ37°C), L2 H-NOX is a mixture (ϳ50:50) of the 5-and 6-coordinate Fe II -NO complexes, but the coordination state of NO complex would be primarily 6-coordinate at the lower end of possible physiological growth temperatures and primarily 5-coordinate at the warmer extreme of possible growth temperatures. The preferred growth temperature of cyanobacteria like N. punctiforme may vary from ϳ15 to ϳ40°C; in this range Np H-NOX remains mostly 6-coordinate, although the transition to 5-coordinate occurs at the higher temperatures in this range. L1 H-NOX, like sGC H-NOX and Vc H-NOX, is a 5-coordinate NO complex at all temperatures in the physiological range. Function of the prokaryotic H-NOXs remains to be established, but it is interesting the NO complexes of Np H-NOX and L2 H-NOX would vary from 5-to 6-coordinate over the temperature range of growth.
Distal versus proximal NO coordination in sGC has been the focus of some debate (9) because sGC shares some kinetic and spectroscopic properties with cytochrome cЈ, and the crystal structure of the 5-coordinate Fe II -NO complex of cytochrome cЈ shows that NO binds to the proximal face of the heme in this protein. Based on work with cytochrome cЈ, distal pocket steric crowding has been suggested as the driving force for conversion of a distal pocket-bound 6-coordinate complex into proximal pocket-bound 5-coordinate NO complex (25). Distal pocket steric crowding is not correlated with Fe II -NO complex coordinate number in the H-NOX family, however. The structure of Tt H-NOX indicates that the distal heme pocket is quite packed (8), despite the fact that Tt H-NOX forms a 6-coordinate NO complex. On the other hand, homology modeling of the H-NOX domain of sGC (6) as well as studies using nitrosoalkanes as bulky distal pocket ligands for sGC indicate that the distal pocket of sGC, which forms a 5-coordinate NO complex, is quite accommodating (26).
The (i) temperature dependence of the NO complex coordination number, (ii) kinetics, and (iii) mutational data presented above and discussed below suggest that in both the 5-and 6-coordinate complexes, NO is bound in the distal pocket of the H-NOX fold. The fact that the L2 H-NOX and Np H-NOX proteins are able to reversibly convert between 5-and 6-coordinate NO complexes under aerobic conditions with no additional NO in solution is best explained by a model with NO bound on the distal side of the heme, and the Fe-His bond is broken and re-formed using thermal energy (Fig. 6). It seems unlikely that the NO would dissociate from the distal side of the heme, migrate to the proximal side, displace the axial histidine ligand, and then rebind to the other face of the porphyrin, because this would require the transient formation of a 4-coordinate heme. Photodissociation of CO from a proximal 6-coordinate myoglobin complex shows that CO can migrate to the proximal pocket, though not bound to the iron (27,28).
This temperature-dependent change in coordination state suggests that formation of the Fe-His bond in the H-NOX fold is an exothermic reaction. The equilibrium between 5-and 6-coordinate NO complexes in various H-NOX family members should be observable, although for some the temperature might be quite low. In fact, in sGC a small amount of the 6-coordinate NO complex is observed in EPR experiments at Ϫ196°C (29).
A model in which NO is bound to the distal side of the heme is also most consistent with the NO dissociation kinetics observed in this study. Using temperature to select between a starting 5-or 6-coordinate NO complex in L2 H-NOX, k off(NO) had the expected temperature dependence (doubling for every 10°C) but no dependence on coordination state. Instead, at higher temperatures where the Fe II -NO complex is present as an equilibrium mixture of 5-and 6-coordinate complexes, an additional (parallel), slower exponent is required to fit the data, which is consistent with a population of the sample that proceeds to the final product with a different slow kinetic step, presumably the additional process of Fe-His bond formation to form the 6-coordinate NO complex before NO dissociation. The simplest kinetic model Reaction 1 to explain these observations is consistent with NO bound in the distal pocket, regardless of whether or not the histidine is bound in the proximal pocket in L2 H-NOX.
In L1 H-NOX, Fe-His bond formation is sufficiently exothermic that lowering the temperature failed to generate any 6-coordinate NO complex. Therefore, according to Reaction 1, the kinetics should fit with a single exponential corresponding to the slow kinetic step in the mechanism of NO dissociation. In all cases, however, the kinetics had two time constants. Thus a second 5-coordinate species (A and A*) was required to describe the NO dissociation mechanism (see Reaction 2). In fact, A and A* can only be distinguished kinetically because they are spectroscopically identical; the electronic absorption ( Fig.  1) and resonance Raman (Fig. 2) spectroscopy both indicate only one complex, 5-coordinate, in solution at all temperatures tested. We propose Reaction 2 as the simplest general mechanism for NO dissociation from H-NOX proteins. Importantly, Reaction 1 is a subset of Reaction 2; fundamentally they are the same kinetic mechanism.
A* likely represents a conformational change in the protein that occurs after NO binds and the histidine ligand dissociates, as has been suggested previously based on flash-photolysis studies (30). The fact that the population of this species increases at elevated temperatures (as assessed by the amplitudes of the kinetic constants) supports this interpretation. This conformational change could be an important step in the signal transduction pathway, assuming L1 H-NOX is an NO sensor. To test the possibility that A* could be a 5-coordinate complex in which the dissociated histidine ligand was proto-nated, a pH dependence on the NO dissociation kinetics for L1 H-NOX and L2 H-NOX from pH 6.9 to 8.9 was carried out (data not shown), but the NO dissociation kinetics were found to be independent of pH over this range. It remains a formal possibility that A* is a 5-coordinate complex with NO bound in the proximal heme pocket, as has been proposed for cytochrome cЈ (16,17,25), but based on the discussion above, this seems unlikely here. Fig. 6 illustrates the simplest mechanism for NO dissociation from all members of the H-NOX family that takes into account all the data known at the present. Each step of this mechanism was separately probed by the experiments presented here. NO dissociation from L2 H-NOX at 0 and 10°C proceeds simply as B to C, whereas at 40°C, a population proceeds as B to C, and another population proceeds as A to B to C. NO dissociation from L1 H-NOX, at all temperatures measured, has two popu- FIGURE 6. Schematic representation of the spectroscopy and kinetics of a temperature-dependent Fe II -NO complex in an H-NOX domain. The initial ferrous-unligated complex binds NO leading to either a 5-coordinate (A) or a 6-coordinate (B) complex that can be distinguished by electronic spectroscopy; in addition, kinetic experiments have indicated a second species of the 5-coordinate complex (A*). The dashed arrows are not meant to indicate an NO binding mechanism, they are meant to represent that at the beginning of the NO dissociation experiments examined here, three different NO-bound H-NOX complexes are possible. Electronic absorption spectroscopy as well as kinetic data clearly indicate that species A and B are in thermal equilibrium, and kinetic data suggest that A and A* are also in thermal equilibrium. Using different H-NOX constructs at different temperatures (the specifics of which are indicated in the illustration), the various steps of this mechanism were identified. The dissociation mechanism depicted here and discussed in the text is the simplest mechanism that fits all of the data. lations, one proceeding as A to B to C and another as A* to A to B to C (or A* to B to C, kinetically indistinguishable mechanisms), the latter, starting with A*, being more dominant at higher temperatures. Consistent with this proposed mechanism for the H-NOX family, the experimental NO dissociation data are in good agreement with simulation of the kinetic data, 4 and NO dissociation kinetics from sGC and its H-NOX domain constructs indicate the same mechanism for sGC, 4 emphasizing the similarities in this family of proteins.
Distal pocket mutation affects the equilibrium between 5and 6-coordinate complexes in L2 H-NOX, which is also consistent with the conclusion that NO is bound in the distal pocket in this protein. When a distal pocket tyrosine, capable of H-bonding to the bound ligand (5), is introduced with the L2 F142Y mutation, the equilibrium is shifted to primarily a 6-coordinate Fe II -NO complex. This may be best explained by a model in which the addition of an H-bond weakens the strong trans effect of NO, pulling some electron density out of the heme, leading to strengthening and thus maintenance of the Fe-His bond.
Interestingly, the opposite result was obtained by mutation in the H-NOX protein from Clostridium botulinum (7); replacement of tyrosine with phenylalanine converted the 5-coordinate NO complex to 6-coordinate NO complex. Furthermore, removal of tyrosine from the corresponding site in Tt H-NOX had little effect on the coordination number of the NO complex (5). The explanation for these differences in a family of sequence-related proteins is not obvious. In fact, recent DFT calculations have investigated the contribution of an H-bond (from a phenol group) to the overall energetics of 5-and 6-coordinate heme-NO complexes (32) and found that the distal pocket hydrogen bond has very little effect on the energetics of heme-bound NO.
Further mutation experiments are underway in an attempt to correlate distal pocket functionality with the coordination state of the Fe II -NO complex in the H-NOX family. The structure of an NO-bound H-NOX protein, either 5-or 6-coordinate, or other spectroscopic data yielding quantitative information on Fe-N-O bond angles and Fe-N (NO and histidine) lengths, as well as the rotation of the histidine with respect to the heme, will be crucial in a complete molecular characterization.
Np H-NOX was not amenable to kinetic studies using either dithionite or oxyglobin as a trap for the dissociated NO, which raises interesting questions about the role of this H-NOX protein in N. punctiforme. Are the electronics of the distal pocket of Np H-NOX unique within the H-NOX family? Perhaps this is a redox sensor, and not, as assumed, an NO sensor. Alternatively, perhaps Np H-NOX senses some other ligand. CO may be a candidate for the ferrous oxidation state, and anions, such as Ϫ CN, are options for the ferric oxidation state. Finally, these results may have implications on whether or not breaking the Fe-His bond in sGC is truly the molecular switch for activation of sGC, as has been assumed. As reported here, within the H-NOX family, 5-and 6-coordinate NO complexes occur at physiological temperatures. Thus, by assuming that like sGC, these H-NOX proteins are also NO sensors, perhaps loss of the Fe-His bond is not the structural switch. Consistent with this, CO and YC-1 are well known to fully activate sGC without breaking the Fe-His bond (33,34). Furthermore, there is recent evidence that suggests there may be an additional binding site on sGC, either for NO or a nucleotide, that is proposed to serve as an allosteric site for activation of sGC (19,35).
In summary, three additional members of the H-NOX family of heme proteins, related in sequence and structure to the heme domain from sGC, have been cloned and characterized. Percent sequence identity with sGC, however, is not the best predictor for ligand binding characteristics, as UV-visible and resonance Raman spectroscopy demonstrate that L1 H-NOX domain from L. pneumophila, which shares 19% sequence identity with sGC, is spectroscopically (and kinetically) 4 nearly identical, forming a 5-coordinate, temperature-independent (from Ϫ12 to 40°C) Fe II -NO complex. In contrast, L2 H-NOX from L. pneumophila, which shares 34% identity with L1 H-NOX and 16% identity with sGC, and the H-NOX protein from N. punctiforme, which is 39% identical to sGC, form a temperature-dependent mixture of 5-and 6-coordinate Fe II -NO complexes. At low temperature, they are primarily 6-coordinate, and at high temperature, a shift toward the 5-coordinate geometry is observed. This temperature-dependent process is rationalized in terms of enthalpy, such that lower temperatures favor the more exothermic reaction, in this case Fe II -His bond formation. A kinetic analysis of NO dissociation from both 5-and 6-coordinate complexes of H-NOX family members suggests that each of these proceed according to a mechanism (Fig. 6) in which NO is bound to the distal heme pocket and NO dissociation occurs via the distal pocket 6-coordinate complex. As the heme domain of sGC is also an H-NOX domain, these data suggest that NO is also bound to the distal side of the heme in sGC.