|Home | About | Journals | Submit | Contact Us | Français|
The nitric oxide (NO) produced by inducible Nitric Oxide Synthase (iNOS) up-regulates the expression of heme oxygenase (HO), which in turn produces carbon monoxide (CO) that down-regulates iNOS activity by reducing its expression level or by inhibiting its activity by converting it to an inactive P420 form (iNOSP420). Accordingly, CO has been considered as a potentially important attenuator of inflammation. Despite its importance, the nature of the proximal heme ligand of the iNOSP420 species remains elusive. Here we show that the 221 cm−1 mode of the photoproduct of iNOSP420 does not exhibit any H2O-D2O solvent isotope shift such as that found in the iron-histidine stretching mode of myoglobin, indicating that the proximal ligand of iNOSP420 is not a histidine. The νFe-CO and νC-O data reveal that the proximal heme ligand of iNOSP420 is consistent with a protonated thiol, instead of a thiolate anion. Furthermore, the optical absorption properties of iNOSP420 are similar to those of a neutral thiol-heme model complex, but not myoglobin. Together the data support the scenario that iNOSP420 is inactivated by protonation of the native proximal thiolate ligand to a neutral thiol, instead of by ligand switching to a histidine, as prior studies have suggested.
Nitric Oxide Synthase (NOS) catalyzes the formation of nitric oxide (NO) from L-Arginine (L-Arg) and two molecules of dioxygen in two successive mono-oxygenation reactions. The first reaction produces N-hydroxy-L-Arginine (NOHA) and the second reaction produces L-citrulline and NO.1-4 NOS is a homodimeric heme-containing enzyme. Each monomer is comprised of an oxygenase domain, which binds the heme prosthetic group, the substrate, and the tetrahydrobiopterin (H4B) cofactor, and a reductase domain, which binds FMN, FAD and NADPH.5,6 The electron transfer from the reductase domain to the oxygenase domain, required for the oxygen chemistry, is enabled by the binding of calmodulin in the interface between the two domains.7
In mammals, NO is produced by three isoforms of NOS for diverse physiological functions.1,2 The NO generated by the two constitutive isoforms, endothelial NOS (eNOS) and neuronal NOS (nNOS), is used as a messenger molecule for smooth muscle relaxation and for neurotransmission, respectively. Their activity is regulated by intracellular calcium levels, which controls the electron transfer from the reductase to oxygenase domain via calmodulin binding. The NO produced by the inducible isoform, iNOS, which has calmodulin permanently bound,7 is used for immunological protection. Overexpression of iNOS can cause the depletion of L-Arg and H4B, leading to deleterious effects due to oxidative stress as a result of the release of the O2 from NOS as superoxide via the uncoupled reaction.8-10 The overproduction of NO itself may lead to various diseases characterized by acute and chronic inflammation, such as sepsis,11 atherosclerosis,12 transplant rejection,13,14 asthma,15 inflammatory bowel disease,16 and arthritis.17
As NO production can be a double-edged sword, the activity of iNOS has to be tightly regulated. Accordingly, its expression is controlled by several cellular factors other than calcium-induced calmodulin binding.18,19 In addition, NO produced by iNOS up-regulates the expression of heme oxygenase (HO), which in turn produces carbon monoxide (CO) that can down-regulate iNOS activity by reducing its expression level and by directly inhibiting the enzyme, as illustrated in Scheme 1.20-22 On this basis, CO has been recognized as an important attenuator of inflammation.23-25 Despite its importance, the molecular mechanism underlying the CO-induced inactivation of iNOS remains unclear.
NOSs belong to the P450 family of enzymes, with a negatively charged thiolate ligand, i.e. a deprotonated cysteine, coordinated to the heme iron.26 The binding of CO to iNOS, like other members of the P450 family, produces a species with a Soret band at ~450 nm (iNOSP450),27-29 which can spontaneously convert to an inactive “P420” species (iNOSP420), characterized by a Soret band at ~420 nm.30-33 It has been proposed that the P450→P420 transition in the cytochrome P450 family of enzymes is associated with the replacement of the native thiolate ligand with a histidine residue, on the basis of the following observations. (1) The optical absorption spectra of the P420s are similar to those of hemeproteins with a histidine as the proximal heme ligand, such as myoglobin (Mb).34 (2) A Raman band at 218 cm−1 identified in the 10 ns photoproduct of the P420 form of cytochrome P450 was assigned as an iron-histidine stretching mode based on its similarity to the photoproduct of a low pH form of Mb.35 (3) The equilibrium Raman spectra of the P420 derivatives of NOSs, P450s and chloroperoxidase are similar to that of the low pH form of Mb.36 Although the evidence is compelling, in most of the enzymes, including iNOS, there are no histidine residues present within 10Å of the heme iron; it is not obvious as to why and how CO binding could trigger such a large conformational change in different P450-type enzymes—with diverse sequences and structures—to bring a histidine residue in close proximity to the heme iron in an energetically feasible fashion.
In contrast, instead of ligand-switching, other evidence suggests that the P450→P420 transition is a result of the protonation of the thiolate ligand. (1) When CO-adducts of porphyrin model complexes with a ferrous iron are coordinated by a thiolate ligand, they exhibit P450-type spectra, which convert to P420-type spectra when the thiolate ligand is protonated.26,37 (2) In a cytochrome P450 from M. tuberculosis, the P450→P420 conversion is controlled by pH, with the P420 form preferred at low pH.30
The difficulty in characterizing the molecular properties of P420 lies in the fact that the CO-adducts of neutral thiol- and imidazole-bound hemes display similar optical spectroscopic properties. The spectral similarity between the two types of CO-adduct is surprising, but consistent with the prediction derived from extended-Hückel calculations.38 Along these lines, the CO-adducts of the neutral thiol- and imidazole-bound H93G mutant of Mb exhibit similar MCD spectral patterns and intensities, which are akin to those of the CO-adduct of the wild-type Mb.39
In this study, we used steady-state and time-resolved resonance Raman and optical absorption spectroscopic techniques to determine the identity of the proximal ligand of iNOSP420 by comparing it to horse heart Mb and a CO derivative of a n-propanethiol-coordinated hemin dimethylester model complex. Based on the new data, we conclude that CO binding to the oxygenase domain of iNOS causes the reversible protonation of the native thiolate anion proximal heme ligand to a neutral thiol, instead of switching to histidine as previously proposed.35,36
Hemin dimethylester was purchased from Porphyrin Products (Logan, Utah). The natural abundance CO and the 13C18O isotope were obtained from Tech Air (White Plains, NY) and Icon Isotopes (Summit, NJ), respectively. Lyophilized salt-free horse heart myoglobin and all other chemicals were from Sigma-Aldrich Corp. (St. Louis, MO). All the chemicals were used without further purification, and prepared with deionized water (Millipore). The oxygenase domain of murine iNOS was prepared as described elsewhere.40 Briefly, the iNOSoxy gene with a C-terminal six histidine tag was cloned into a pCWori vector. The protein was expressed in E. coli and purified by using a Ni-NTA affinity column. The purified enzyme in 40 mM pH 7.6 EPPS buffer with 1 mM DTT was stored at 77 K until use. All the protein samples were prepared in 40 mM pH 7.6 EPPS buffer without DTT, except where indicated. To generate the CO-adducts, the protein samples were first purged with CO gas under anaerobic conditions, followed by reduction with a minimal amount of sodium dithionite. The protein concentration used for the equilibrium optical and Raman measurements was 40 μM; that for the photolysis measurements was 150 μM.
The CO-adduct of the neutral thiol-heme model complex was prepared anaerobically by dissolving hemin dimethylester and sodium dithionite in a solvent mixture of water/benzene/n-propanethiol (with a volume ratio of 3:3:1), which was pre-purged with CO. The solution mixture was shaken thoroughly for several minutes until the color of the organic layer, containing the neutral thiol-heme model complex, changed from brown to bright pink (the aqueous layer containing sodium dithionite remained colorless). The CO-adduct of the neutral thiol-heme model complex residing in the organic layer was transferred with a Hamilton gas tight syringe to an argon-purged optical cuvette or Raman rotating cell for absorption or Raman spectroscopic measurements, respectively.
The optical absorption spectra were obtained with a UV2100 spectrophotometer from Shimadzu Scientific Instruments, Inc. (Columbia, MD) with a spectral slit width of 1 nm. To measure the steady-state resonance Raman spectra, the 413.1 nm excitation from a Kr ion laser (Spectra Physics, Mountain View, CA) was focused to a ~30 μm spot on the spinning quartz cell rotating at ~1,000 rpm. The scattered light, collected at a right angle to the incident laser beam, was focused on the 100 μm-wide entrance slit of a 1.25 m Spex spectrometer equipped with a 1200 grooves/mm grating (Horiba Jobin Yvon, Edison, NJ), where it was dispersed and then detected by a liquid nitrogen-cooled CCD detector (Princeton Instruments, Trenton, NJ). A holographic notch filter (Kaiser, Ann Arbor, MI) was used to remove the laser line. The Raman shift was calibrated by using indene and an acetone/ferricyanide mixture for the 300-1700 and 1900-2000 cm−1 spectral windows, respectively. The laser power was kept <5 mW for all measurements to avoid photodissociation of the heme-bound CO. The spectral acquisition times for the νFe-CO and νC-O measurements were ~ 30 and 120 min, respectively.
To obtain the spectrum of the photolyzed species of iNOSP420, a 5 ns laser pulse (532 nm, 10 Hz) from a frequency-doubled Nd:YAG laser (Quantel Brilliant B; Big Sky Laser Technologies, Bozeman, MT) was used to generate a 435.7 nm line from a 1 meter-long hydrogen shifter (Light Age, Inc., Somerset, NJ) pressurized with H2 gas at 150 psi. The 435.7 nm laser pulse was used to photo-dissociate the CO from the protein and at the same time was used as the excitation light source for the Raman measurement. The Raman spectrum was obtained with a 1 m focal length spectrometer (model 1000M, Spex Industries Inc., Edison, NJ) equipped with a CCD detector (Spectrum ONE from Instruments S.A. Inc., Edison, NJ). The averaged laser power and acquisition time were 5-6 mW and 7 hours, respectively. The Mb sample in D2O was prepared by dissolving lyophilized myoglobin in D2O buffer. For iNOS, a 200uL sample of ferric, substrate- and DTT-free iNOS in H2O buffer was exchanged with 20mL of D2O buffer in a Amicon Centrifugal Filter Device (Millipore). Both samples were stored at 4°C for 12 hours before data acquisition.
For the pH-dependent studies, 800 μL of 10 μM iNOS in a 40mM EPPS buffer in the presence of either 2 mM dithiothreitol (DTT) or 2 mM L-Arginine, was purged with CO and reduced with excess dithionite. After the iNOSP450→iNOSP420 conversion reaches the equilibrium state, the sample was titrated with HCl. The conversion from iNOSP450 to iNOSP420 was followed as a function of pH by optical absorption measurements.
As shown in Figure 1, immediately following CO binding to the L-Arg and H4B-free iNOS, the enzyme exhibits a P450-type spectrum with a Soret band at 444 nm, a single visible band at 551 nm. Within a 5-hour time window, it gradually converts to iNOSP420, with a Soret band at 420 nm, visible bands at 539 and 568 nm. Extended incubation (~12 hrs) leads to almost complete conversion to the iNOSP420 species (vide infra). Upon exposure to air, the iNOSP420 species spontaneously oxidizes to the native ferric form (data not shown). As a deoxy species was not observed during the reaction, the reaction appears to be rate-limited by CO-dissociation. The ferric enzyme thus produced can be re-reduced and coordinated by CO to regenerate the active P450 form. The process can be repeated several times without sacrificing the integrity of the enzyme as reported by Abu-Soud et al.41 The data demonstrate that the change in the proximal ligand associated with the iNOSP450→iNOSP420 conversion is a reversible process, and that the native proximal ligand of the enzyme is thermodynamically favored in the ferric state.
The presence of L-Arg or H4B partially inhibits the iNOSP450→iNOSP420 conversion (vide infra), whereas the presence of both L-Arg and H4B completely prevents the transition. It is well-established that the substrate and cofactor-free iNOS exists as a “loose” dimer. The binding of L-Arg or H4B promotes its partial conversion to a “tight” dimer, which exhibits increased resistance towards proteolytic cleavage40 and detergent-induced dissociation of dimer into monomers,42-44 while the presence of both L-Arg and H4B leads to a complete conversion to the “tight” dimer. The conformational stability afforded by the “tight” dimer in L-Arg- or H4B-bound enzyme presumably accounts for the partial inhibition of the iNOSP420 formation. The data indicate that the iNOSP450→iNOSP420 transition is related to the “tight” dimer ↔ “loose” dimer equilibrium.
Like L-Arg, the presence of DTT partially inhibits the iNOSP450→iNOSP420 conversion. DTT has been shown to be a competitive inhibitor of L-Arg binding to nNOS.45,46 The inhibition effect of DTT against iNOSP420 formation can hence be attributed to the occupation of the DTT molecule in the L-Arg binding site, thereby affecting the “tight” dimer ↔ “loose” dimer equilibrium.
In the absence of DTT, L-Arg, and H4B, pH-dependent studies show that, the iNOSP450→iNOSP420 conversion reaches completion independent of the pH (between 6.9 and 8.9); whereas in the presence of L-Arg or DTT the conversion is pH-sensitive. As shown in Figure 2, in the presence of DTT at pH 7.2, ~52% of iNOSP420 was observed, which decreased to ~36% as pH increased to 7.7 in a sigmoidal fashion. The data indicate that the conformational change associated with the iNOSP450→iNOSP420 transition in the presence of DTT is associated with an apparent pKa of ~7.35. In contrast, in the presence of L-Arg at pH 7.2, ~40% of iNOSP420 was observed, which linearly decreased to ~32% as pH increased to 7.7. The linear behavior may reflect the post transition region of the pH-dependent conformational change associated with a lower pKa value in the presence of L-Arg, as compared to DTT, as the percentage change (40-32%) is similar to that observed in the post transition region of the DTT data. Unfortunately, the pH dependent data at pH <7.2, which could test this hypothesis, were not detectable as the protein started to aggregate due to the instability of the protein structure at low pH. In summary, the data indicate that the iNOSP450→iNOSP420 transition is coupled to the “tight” dimer ↔ “loose” dimer equilibrium, and that the equilibrium is sensitive to the pH. P420 Model Analogues. To determine if the iNOSP450→iNOSP420 transition is a result of ligand-switching from a thiolate to a histidine or the protonation of the thiolate ligand, the electronic transition properties of iNOSP420 with respect to a neutral thiol-heme model complex, CO-bound n-propanethiol hemin dimethylester, and CO-bound horse heart Mb were examined. The iNOSP420 spectrum was obtained following a ~12 hr incubation of the L-Arg and H4B-free iNOSP450 sample and compared to that of the neutral thiol-heme model complex. As shown in Figure 3, the optical absorption spectrum of the neutral thiol-heme complex is nearly identical to that of the iNOSP420 complex with absorption maxima at 419, 539, and 568 nm, which are distinct from those of Mb at 422, 541, and 577 nm. As a comparison, the initial iNOSP450 spectrum is also shown with its 444 nm Soret band, broad visible absorption band at 551nm, and a red-shifted UV-component compared to the other spectra. The iNOSP420 spectrum is also distinguishable from that of a negatively charged thiolate model heme complex (formed by mixing CO, a thiolate and the diethylester of Fe-protoporphyrin IX) with a Soret maximum at 449 nm37 and from that of a bis-CO heme model complex (formed by mixing CO with hemin dimethylester) with a Soret maximum at 408 nm as reported47 (data not shown). The optical absorption data are consistent with the assignment of a neutral-thiol as the proximal heme ligand in the iNOSP420 complex as a result of the protonation of the native thiolate ligand.
To further investigate the nature of the proximal ligand of the iNOSP420 species, we examined its resonance Raman spectrum. As shown in Figure 4, the νC-O and νFe-CO modes of iNOSP420 are present at 1951 and 496 cm−1, respectively, in contrast to the 1946 and 491 cm−1 modes observed in native iNOSP450.48 In the difference spectra shown in (b) all the heme modes are canceled out and the remaining positive and negative peaks are associated with the 12C-16O and 13C-18O stretching modes, respectively. As a comparison, the CO-adduct of the neutral thiol-heme model complex was also examined. The Raman spectrum of the model complex shows that the νC-O/νFe-CO modes are at 1971 and 495 cm−1 (Figure 4).
The νC-O and νFe-CO modes typically follow an inverse correlation, reflecting the electrostatic environment of the heme-bound CO due to the resonance between the following two extreme structures of the L-Fe-CO moiety (where L is the proximal heme iron ligand).49
On the other hand, the offset of the inverse correlation line in the νC-O vs νFe-CO plot is determined by the electronic properties of the proximal heme ligand (L). The data from several enzyme systems with thiolate and histidine as proximal ligands are presented in Figure 5.50-52 Hemeproteins with His as a proximal ligand stay on a line lying above the P450 line, constructed by the data from enzymes with a more electron donating thiolate as the proximal ligand. The mammalian NOS (mNOS) line lies between the histidine and the thiolate line, as the proximal thiolate ligand in iNOS accepts an H-bond from a nearby tryptophan residue (see W188 in iNOS shown in Figure 1), which reduces the electron donating capability of the thiolate ligand. Interestingly, the iNOSP450 data point is positioned directly on the mNOS line, while the iNOSP420 data point lies above the mNOS line, consistent with the proposal that its proximal ligand is a protonated thiol, which is expected to donate less electron density to the heme iron as compared to that in the active iNOSP450 species. The data point associated with the neutral thiol-heme model complex is positioned above the thiolate correlation line (Figure 5), also in agreement with the expected weaker electron donation from its neutral thiol proximal ligand. The horizontal displacement of the νC-O of the neutral thiol-heme model complex from the iNOSP420 data in the correlation plot is plausibly a result of (1) the strain imposed by the protein matrix on the proximal ligand in the latter, (2) differences in the orientation of the proximal ligand, or (3) the protonation state of the propionates, as proposed for other heme protein systems.53 These changes can cause horizontal displacement of the νC-O mode while leaving the νFe-CO unaffected.
It is important to note that several heme out-of-plane modes observed in the low frequency region of the spectrum of the native iNOSP450, such as those at 693, 734 and 803 cm−1,54 are absent in the spectrum of the iNOSP420 shown in Figure 4a, indicating that the distorted heme in the native iNOSP450 became planar when it converted to the P420 form. Likewise, the ν2 mode of heme proteins was shown by Czernuszewicz et al to be sensitive to heme distortion based on the studies of nickel octaethylporphyrin model complexes.55 In iNOSP450, it is observed at 1573 cm−1, whereas it is at 1581 cm−1 in iNOSP420,36 consistent with a more planar heme in iNOSP420.
To directly examine the iron-proximal ligand stretching mode, the 5 ns time-resolved resonance Raman spectrum of the photolysis product of iNOSP420 was measured. As shown in Figure 6, the photoproduct of iNOSP420 (in which CO is photolyzed) displays a strong band at 221 cm−1, similar to that of the P420 derivative of a cytochrome P450.35 This band is not present in either the 6-coordinate CO-bound form of iNOSP420 (Figure 4) or in the 5 ns photolysis product of the active iNOSP450 either in the presence or absence of L-Arg and H4B (data not shown). The frequency of this mode is similar to the iron-histidine stretching mode (νFe-His) of myoglobin,56 and is 117 cm−1 lower than the iron-thiolate stretching frequency,57 in line with the expected weaker iron-thiol bond, suggesting that it can be either a νFe-His mode or iron-thiol stretching mode (νFe-SH). To test it, we examined the H2O-D2O solvent isotope effect of the 221 cm mode of iNOSP420, as compared to that of the νFe-His mode of Mb.
Using a simple diatomic oscillator model, the expected shifts for replacing the labile proton with deuterium can be calculated by taking the square root of the ratio of the reduced masses of the deuterated and undeuterated species coordinated to the iron atom (mass = 56). In the case of the imidazole moiety (mass = 67) and the sulfur (mass = 32) of the thiol, the calculated shifts of an Fe-ligand mode at 221 cm−1 are 0.7 and 2 cm−1, respectively. As shown in Figure 6, for Mb, a down-shift in the νFe-His frequency was observed in D2O. The H2O-D2O isotopic shift was calculated to be 1.5 cm−1 based on the method reported previously,58 with the assumption that the νFe-His mode can be described by a Gaussian function.59 The same experiment was performed with iNOSP420, but no isotopic shift was detected for the analogous 221 cm−1 mode. Thus, the data indicate that the 221 cm−1 mode is neither a νFe-His nor a νFe-SH mode; instead it is likely a heme mode that is enhanced when a neutral thiol binds to the heme iron. The absence of a νFe-His mode in the spectrum of the photolyzed iNOSP420 excludes the possibility that the iNOSP450 → iNOSP420 transition is associated with ligand switching of the native thiolate with a histidine. It could be argued that in iNOS, the histidine did not become deuterated, reflecting the lack of a detectable shift. However, the “nearby” histidines (>10Å) are solvent accessible and would likely be readily deuterated. Furthermore, we would expect increased solvent accessibility in the “loose” conformation.
It is noteworthy that we attempted to detect the νFe-SH mode of the photoproduct of the neutral thiol-heme complex; but, upon photolysis, the complex converted to a four-coordinate state, instead of a five-coordinate complex (data not shown). A plausible explanation for why the five-coordinate structure was detectible in iNOS and Mb within 5 ns and not in the model complex is that the protein matrix plays an important role in keeping the proximal ligand close to the heme. In an aqueous study of histidine and CO binding to chlorohemin, there is little ligand recombination between 10 ps and 100 ns,60 and the histidine does not bind until an H2O-Heme-CO complex forms because the histidine has a much lower affinity for unliganded heme than for a CO-coordinated heme.61 In the case of enzymes, however, the proximal ligand is held close to the heme iron, facilitating ligation.
Our data show that, the low frequency mode of the photoproduct of iNOSP420 at ~221 cm−1 does not have an H2O-D2O solvent isotope shift as that found in the genuine νFe-His mode of Mb, indicating that it is neither a νFe-His nor a νFe-SH mode. It confirms that the proximal ligand of iNOSP420 is not a histidine. The νFe-CO and νC-O data reveal that the proximal heme ligand of iNOSP420 is consistent with a protonated thiol, instead of a thiolate anion. In addition, the optical absorption properties of iNOSP420 are similar to a neutral thiol-heme model complex, instead of Mb, indicating that the proximal ligand of iNOSP420 is a protonated thiol.
Thiols in free solution typically exhibit pKa values between 8 and 10. The cysteine sidechain, for example has a pKa of 8.2; but, when incorporated into peptides, the pKa can be as low as 7.4.62 Additionally, when coordinated to a heme iron, the pKa of the cysteine sidechain could be further perturbed, as the positive charge of the ferric iron would stabilize a negatively charged thiolate. For instance, ethanethiol with a pKa of ~10.5 in free solution63 is deprotonated when bound to the H93G cavity mutant of sperm whale Mb in pH 7 aqueous solution.64 Upon reduction of the heme iron to the ferrous state, the pKa of thiolate ligands is increased due to the increase in the electron density on the iron, which increases the basicity of the thiolate anion; as a result, depending on the protein structure, the thiol may either bind to the heme iron as a protonated form (neutral thiol)39 or totally loses its ligation.64
In the native iNOSP450, the proximal thiolate ligand accepts a H-bond from the W188 (Figure 1), which reduces the electron donating capability of the thiolate, thereby stabilizing the ferrous iron-thiolate bond. Our data suggest that the iNOSP450→ iNOSP420 transition is associated with the protonation of the thiolate ligand rather than ligand switching with a histidine. The pH-dependent protonation of the thiolate ligand is concurrent with the breakage of the H-bonding interaction between the W188 and the proximal thiolate ligand. For the L-Arg and H4B-free enzyme, the pKa for the pH-dependent transition in the presence of DTT is ~7.35, which could reflect the pKa of the proximal thiolate ligand.
While the full molecular basis for the iNOSP450→ iNOSP420 transition remains to be further explored, it is noteworthy that iNOSP420 has a more planar heme, with the heme substituents oriented in distinct conformations as compared to the native enzyme. Our data suggest that this conformational change to the heme induced by CO-binding to the substrate- and cofactor-free enzyme modifies the protein matrix, thereby shifting the “tight” dimer↔“loose” dimer equilibrium, thereby disrupting the H-bond between the proximal thiolate ligand and W188, and allowing the entrance of water into the heme pocket to protonate the thiolate ligand to a neutral thiol.
The data demonstrate that the iNOSP450→iNOSP420 transition is controlled by the conformational state of the enzyme. In the substrate and cofactor-bound state, the enzyme is in the “tight” dimer conformation and unable to convert to iNOSP420. With only L-Arg bound, the “tight” dimer ↔“loose” dimer equilibrium shifts towards the “loose” dimer state, it hence is slightly more prone to iNOSP420 formation. With DTT bound in the L-Arg binding site, the equilibrium shifts even more towards the “loose” dimer state, accounting for its higher preference for the iNOSP420 form. In the absence of L-Arg and DTT, the enzyme is most susceptible to iNOSP420 formation at all pH values tested, as the equilibrium completely shifts to the “loose” dimer state.
The importance of the protonation state of the sulfur on the reactivity of thiolate coordinated model complexes has been studied theoretically by DFT calculations and experimentally by sulfur K-edge X-ray absorption spectroscopy recently.65 Likewise, the H-bonding to the thiolate ligand in iNOS, has been recognized to be important in controlling the oxygen chemistry performed by the enzyme and in regulating the NO-linked self-inhibition.48 Here we show that the protonation of the thiolate ligand associated with the iNOSP450→iNOSP420 transition plays a key role in the CO-linked inhibition mechanism of iNOS, underscoring the importance of the proximal control in the NOS family of enzymes. Under conditions of chronic inflammation, such as atherosclerosis, when iNOS is highly expressed, the substrate and cofactor may be depleted. Consequently, the O2 bound to the heme iron can be released as superoxide, which leads to the formation of a hydroxyl radical and other reactive oxygen species, thereby causing oxidative stress. The CO-induced inactivation of iNOS mediated by heme oxygenase (Scheme 1) is hence physiologically imperative for preventing the detrimental uncoupled reaction.
This work is supported by NIH grants GM54806 to D.L.R., F31GM078679 to J.S. and CA53914, GM51491 and HL76491 to D.J.S..