Heme Distortion Modulated by Ligand-Protein Interactions in Inducible Nitric-oxide Synthase*

The catalytic center of nitric-oxide synthase (NOS) consists of a thiolate-coordinated heme macrocycle, a tetrahydrobiopterin (H4B) cofactor, and an l-arginine (l-Arg)/N-hydroxyarginine substrate binding site. To determine how the interplay between the cofactor, the substrates, and the protein matrix housing the heme regulates the enzymatic activity of NOS, the CO-, NO-, and CN--bound adducts of the oxygenase domain of the inducible isoform of NOS (iNOSoxy) were examined with resonance Raman spectroscopy. The Raman data of the CO-bound ferrous protein demonstrated that the presence of l-Arg causes the Fe–C–O moiety to adopt a bent structure because of an H-bonding interaction whereas H4B binding exerts no effect. Similar behavior was found in the CN--bound ferric protein and in the nitric oxide (NO)-bound ferrous protein. In contrast, in the NO-bound ferric complexes, the addition of l-Arg alone does not affect the structural properties of the Fe–N–O moiety, but H4B binding forces it to adopt a bent structure, which is further enhanced by the subsequent addition of l-Arg. The differential interactions between the various heme ligands and the protein matrix in response to l-Arg and/or H4B binding is coupled to heme distortions, as reflected by the development of a variety of out-of-plane heme modes in the low frequency Raman spectra. The extent and symmetry of heme deformation modulated by ligand, substrate, and cofactor binding may provide important control over the catalytic and autoinhibitory properties of the enzyme.

respectively. All three NOS isoforms are dimeric. Each subunit of the dimer contains two domains: a reductase domain that binds FMN, FAD, and NADPH and an oxygenase domain that contains heme and tetrahydrobiopterin (H4B). The electron transfer from the reductase domain to the oxygenase domain, which is essential for the enzymatic activity, is regulated by binding of a calcium-calmodulin complex. When the calciumcalmodulin complex is present, electrons flow from NADPH through FMN and FAD in one subunit to the oxygenase domain of the other subunit (7). The crystal structures of the oxygenase domains from all three isoforms have been determined. They show that the substrate L-Arg binds directly above the heme iron atom, whereas the cofactor H4B binds along the side of the heme. Furthermore, the L-Arg and H4B are linked together through an extended H-bonding network mediated by one of the two propionate groups of the heme (8 -11).
The functional role of H4B in NOS remains an enigma. Recent experimental evidence (12)(13)(14)(15)(16)(17)(18)(19)(20) has demonstrated that H4B is involved in the electron transfer processes in both steps of catalysis. EPR and optical absorption data show that during the hydroxylation of L-Arg, the disappearance of the oxygenbound heme is kinetically and quantitatively coupled to the formation of NOHA and a H4B radical species (15,20), supporting the scenario that H4B serves as an extra electron source. Using rapid freeze-quench EPR and stopped flow optical absorption measurements, it has been demonstrated that in the second step of the catalytic cycle the H4B radical is first formed and then decayed, suggesting that H4B serves as an electron mediator during the reaction (17). Though recent emphasis has been placed on the catalytic role of H4B, experiments have also given indications that it plays an important structural role. Based on the crystal structures of the oxygenase domain of iNOS (iNOS oxy ), Crane et al. (8) concluded that H4B binding resulted in major conformational changes to the protein that are critical for the promotion of subunit assembly into a dimer, the active form of NOS, and the formation of the reductase docking site required for the electron transfer. In addition, biochemical studies of various isoforms of NOS, showed that H4B binding introduces significant changes in protein stability, monomer/dimer equilibrium, proteolytic susceptibility, heme-ligand binding, and substrate binding properties (21)(22)(23)(24)(25)(26). In contrast, based on the crystal structure of the oxygenase domain of eNOS (eNOS oxy ), Raman et al. (11) reported that H4B binding does not produce any conformational changes in the protein, and more importantly the dimeric assembly is retained in the absence of H4B.
The heme iron is coordinated by four pyrrole nitrogen atoms of the porphyrin ring and a proximal thiolate ligand from a cysteine residue. The Fe-S stretching mode of the proximal bond was identified at 338 cm Ϫ1 by Schelvis et al. (27) in the resonance Raman spectrum with near-UV excitation. The Fe-S stretching frequency is lower than that in cytochrome P-450s, indicating a weaker Fe-S bond that may be important for the catalytic function of NOS. Based on resonance Raman and optical absorption spectra of NOS, in the absence of L-Arg and H4B the ferric heme iron is in a six-coordinated low spin electronic configuration with a water or a dithiothreitol (DTT) molecule coordinated to the distal side of the heme. A sixcoordinate low spin to a five-coordinate high spin heme transition was observed upon H4B or L-Arg binding, reflecting the exclusion of the sixth ligand, either water or DTT, from the heme iron (28 -31). This is similar to the low spin to high spin transition observed in the cytochrome P-450 class of proteins, in which a water molecule bound to the heme iron is displaced upon substrate binding (32). It is believed that substrate binding to P450 interferes with ligand binding to the heme iron because of a steric constraint that is imposed by the substrate located directly over the heme iron (33,34). The analogous displacement of water or DTT in NOS induced by L-Arg binding has been attributed to the same origin (35). However, such a steric interaction cannot account for the H4B binding-induced spin transition, considering the fact that the H4B binding site is on the peripheral side of the heme. The origin for the exclusion of the distal ligand upon H4B binding thus has been an open question.
It has been shown that the NO produced at the end of the catalytic reaction remains in the distal pocket and rebinds to the heme iron, thereby inhibiting the enzyme. Although the crystal structures of the oxygenase domains of the three isoforms of NOS are almost identical, the degree of the autoinhibition by NO follows the order nNOS Ͼ iNOS Ͼ eNOS (36,37). Resonance Raman studies of the three NOS isoforms showed that the frequencies of the Fe-CO and C-O modes of CO-bound nNOS were shifted with respect to those of eNOS and iNOS in the presence of L-Arg, plausibly because of a unique binding geometry of L-Arg in nNOS with respect to those in eNOS and iNOS (38). However, ENDOR data of the high spin ferric ligand-free enzyme showed that the binding geometries of L-Arg or NOHA with respect to the heme iron are essentially the same for eNOS, iNOS, and nNOS (39).
Biochemical studies have demonstrated that the catalytic mechanism for the conversion from L-Arg to NOHA is fundamentally different than that from NOHA to citrulline (1,2,40,41). This is supported by the ENDOR data, showing that the hydroxylated nitrogen of NOHA is held 3.8 Å from the Fe, closer than the corresponding guanidino nitrogen of L-Arg (4.05 A) in each of the three isoforms (39). However, Raman data showed that the Fe-C-O-related vibrational modes of the ferrous CO-bound nNOS in the presence of L-Arg are essentially identical to those in the presence of L-NOHA. Furthermore, the O-O stretching frequency of the oxyderivative of nNOS oxy in the presence L-Arg is the same as that in the presence of NOHA (42).
As a first step to reconcile these disparate results, we have examined the influence of L-Arg and/or H4B binding on the ligand-protein interactions in iNOS oxy , by using CO, NO, and CN Ϫ as structural probes for the ferric and ferrous derivatives of iNOS oxy . We found that heme distortion introduced by L-Arg and/or H4B binding plays an important role in modulating the ligand-protein interactions in iNOS oxy . The possible role of this distortion is discussed in the context of the catalytic function of the enzyme.

MATERIALS AND METHODS
L-Arginine, DTT, sodium cyanide, and sodium dithionite were purchased from Sigma. (6R)-5,6,7,8-tetrahydro-L-biopterin was purchased from Alexis Biochemicals (San Diego, CA). The natural abundant gases, N 2 , CO, and NO were obtained from Tech Air (White Plains, NY). The isotopically labeled compounds, 12  The oxygenase domain of the inducible nitric-oxide synthase (iNOS oxy ) was expressed in Escherichia coli and purified in the absence of both L-Arg and H4B as described previously (42). The enzyme was kept in EPPS buffer at pH 7.6 in the presence of 1 mM DTT. Preparations were stored in liquid nitrogen in buffer containing 10% glycerol. Prior to use, the protein was washed three times with EPPS buffer using a centrifugal filtration unit (Ultrafree-15, Biomax-10K NMWL membrane from Millipore). To generate the L-Arg-and/or H4B-bound derivatives, L-Arg and H4B were added to the filtered enzyme in 100 and 3-5 ϫ excess with regard to the heme, respectively. The enzyme was then incubated for ϳ18 h at 4°C. The binding of L-Arg and H4B was confirmed by monitoring changes in the spin and coordination state of the ferric heme with optical absorption spectroscopy. The protein concentration for each sample was ϳ50 M.
The samples used for optical absorption and resonance Raman spectroscopic measurements on the NO, CO, and CN Ϫ derivatives were first purged with N 2 gas in an anaerobic cell. To form the ferric-NO complexes, 400 l of 1 atm of NO was injected into the cell using a Hamilton (Reno, NV) gas-tight syringe. Immediately prior to injection, NO gas was scrubbed by passage through a solution of 10 M NaOH. H4B titrations were performed by injecting different volumes of a N 2 -purged 1 mM H4B stock solution into the anaerobic sample prior to NO injection. To form the ferrous-CO complexes, the purged samples were reduced with sodium dithionite and 400 l of 1 atm of CO was injected into the cell. To form the ferric-CN Ϫ adducts, sodium cyanide solutions were added to the enzyme to a final concentration of 20 mM under anaerobic conditions. Cyanide adducts were kept in anaerobic conditions to minimize the oxidation of H4B, which increased the fluorescence from the samples. Both H4B and cyanide solutions were purged with N 2 gas prior to injection into an anaerobic cell-containing sample.
Optical absorption spectra were taken on a Shimadzu UV2100U spectrophotometer. Resonance Raman spectra were obtained by using 406.7 or 413.1 nm excitation from a Kr ion laser (Spectra Physics, Mountain View, CA) or 441.6 nm excitation from a He-Cd laser (Liconix, Santa Clara, CA). The incident power on the sample was kept under 3 milliwatts, and the sample cell was rotated at ϳ6000 rpm during the spectral acquisition to avoid photodamage. The scattered laser light was collected and focused onto an entrance slit (100 m) of a 1.25-m SPEX spectrophotometer (Jobin Yvon, Edison, NJ) and was then detected using a liquid nitrogen-cooled CCD camera (Roper Scientific, Princeton, NJ). All of the resonance Raman spectra were frequency calibrated by using spectral lines from indene (Sigma), except that for those in the 1800 -2000 cm Ϫ1 spectral region, an acetone/ferricyanide combination was used instead. Cosmic rays artifacts were removed from the spectra by using a routine in the Winspec spectral acquisition software (Roper Scientific). Intensity references were not added to the samples, so the changes that are detected are all relative to the other modes in the spectra. All measurements were made at room temperature. Data were averaged and accumulated for a total integration time of 30 min/spectra for most cases. A longer integration time of 60 min was used to improve the signal to noise ratio for the CN Ϫ adducts and the C-O spectral region of the CO adduct.
The non-planarity of the hemes was analyzed by the normalcoordinate structural decomposition (NSD) program written by J. A. Shelnutt. 2 This program analyzes the heme or porphyrin structure, such as those from the protein data bank, and decomposes any distortions into different symmetry types that may be related directly to vibrational modes. It also gives a mean atomic displacement from the ideal square planar geometry of the porphyrin for the total distortion and for each symmetry type (43).

RESULTS
The optical absorption spectra of various oxidation and ligation states of iNOS oxy are shown in Fig. 1. In the absence of L-Arg and H4B, the Soret transition of the ferric protein is located at 420 nm, which is characteristic of a six-coordinate low spin heme (Fig. 1A). Upon the addition of either L-Arg or H4B, the Soret transition shifts to ϳ400 nm, because of a partial conversion to a five-coordinated high spin heme. When L-Arg and H4B are both present, the Soret transition further shifts to 395 nm, indicating a full conversion to the five-coordinated high spin heme. Binding NO or CN Ϫ to the ferric heme iron causes a red shift of the Soret transition to 439 nm as shown in Fig. 1B, which is typical for a six-coordinated low spin ferric heme with a proximal cysteine axial ligand. On the other hand, the Soret transition of the CO-bound ferrous protein is located at 445 nm. In contrast to the ligand-free enzyme, the Soret transitions of the NO-, CN Ϫ -, and CO-bound complexes were found to be unaffected by the addition of H4B and/or L-Arg (data not shown).
Resonance Raman spectroscopy with Soret excitation has been successfully applied to study structural and functional relationships of heme proteins for several decades. The high frequency region (1000 -1700 cm Ϫ1 ) of the spectrum is very sensitive to the oxidation and coordination states of the heme groups. In particular, the 4 vibrational heme mode in the 1340 -1380-cm Ϫ1 region is very sensitive to the electron density on the heme macrocycle and hence is a good indicator of the oxidation state of the heme iron. The 3 vibrational mode in the 1475-1520-cm Ϫ1 region is sensitive to both the coordination and spin state of the heme iron, whereas the 2 vibrational mode in the 1560 -1590-cm Ϫ1 region is sensitive to the heme spin state. In contrast, in the low frequency region of the spectrum (200 -800 cm Ϫ1 ), the specific axial ligands coordinated to the prosthetic heme group can be identified by detecting iron-ligand stretching and/or bending modes. In addition, when the prosthetic heme group is deformed from the planar structure, several heme out-of-plane modes may be strongly enhanced in this region of the spectrum (44,45). The frequencies and intensities of the Raman lines are further modulated by the protein environment surrounding the heme and, therefore, provide useful structural information for heme proteins.
The Ligand-free Ferric Complex-The high frequency resonance Raman spectra of the ferric derivatives of iNOS oxy are presented in Fig. 2. In the absence of L-Arg and H4B (spectrum a), a typical six-coordinate low spin spectrum was obtained with the 4 and 3 marker lines present at 1372 and 1500 cm Ϫ1 . Upon the addition of L-Arg, the 3 marker line shifts to 1487 cm Ϫ1 (Fig. 2, spectrum b), indicating a conversion to a fivecoordinate high spin complex. Although the optical absorption spectrum showed only partial conversion from the six-coordinate low spin state to the five-coordinate high spin complex (Fig. 1A), only the high spin component was detected in the resonance Raman spectrum, because the spectral lines from the low spin species were not enhanced with 406.7-nm excitation. The low frequency region of the spectrum was not examined in this study. Nonetheless, in a study of eNOS, changes in certain low frequency modes were identified upon the addition of L-Arg and were interpreted as an indication of a protein structural change (27).
The Ferrous CO-bound Complex-The low frequency resonance Raman spectra of the ferrous-CO derivatives of iNOS oxy complexes were obtained in the presence and absence of L-Arg and/or H4B (Fig. 3A). In the absence of L-Arg and H4B, the two lines at 491 and 562 cm Ϫ1 , as shown in the resonance Raman spectrum a, were assigned to the Fe-CO stretching ( Fe-CO ) and Fe-C-O bending (␦ Fe-C-O ) modes, respectively, based on the isotope difference spectrum shown in Fig. 3B. In the difference spectrum, all the heme modes are cancelled out, and the spectral features remaining in the spectrum are associated only with the vibrational modes involving CO. The split of the 491-cm Ϫ1 mode in the difference spectrum is a result of structural inhomogeneity of the Fe-CO moiety as reflected by a broad feature in the original Fe-CO mode. With the same isotope substitution experiment, a C-O stretching mode ( C-O ) was assigned at 1946 cm Ϫ1 as shown in Fig. 3C. These COrelated vibrational modes are consistent with the data reported previously on the full-length enzyme and on other isoforms (38, 42). No significant shifts in the frequencies or changes in spectral shapes were detected in the CO-related vibrational modes upon the addition of H4B (Fig. 3, spectra c). In contrast, the addition of L-Arg caused a shift in the frequency of the Fe-CO mode from 491 to 512 cm Ϫ1 and the ␦ Fe-CO mode from 562 to 569 cm Ϫ1 (spectra b). In addition, the CO mode shifted from 1946 to 1907 cm Ϫ1 . All three of the Fe-C-O-related modes sharpened in the presence of L-Arg, indicating a direct interaction between L-Arg and the heme-bound CO. Similar spectra were observed in the presence of both L-Arg and H4B (spectra d). Table I summarizes the Fe-CO , ␦ Fe-CO , and CO modes of the iNOS oxy complexes examined here and those reported for the other complexes of NOS. The similarity between the oxygenase domain and the full-length enzyme indicates that the reductase domain does not significantly modify the heme environment in the oxygenase domain, and hence, the oxygenase domain serves as a valid model for the native enzyme.
In addition to the changes in the Fe-C-O-related vibrational modes, small changes in the heme modes were also observed. Most noticeable is the increase in the intensity of a heme mode at 693 cm Ϫ1 upon the addition of H4B and/or L-Arg. The mode is strongest in the presence of both L-Arg and H4B, indicating an additive effect of L-Arg and H4B. An enhancement was also observed at 752 and 803 cm Ϫ1 upon the addition of L-Arg and/or H4B, suggesting that they are of similar origin. Analogous spectral changes were also observed in the reported data for the nNOS oxy domain and for the full-length enzymes of all three isoforms (38,42) indicating that related effects occur in all three isoforms.
The Ferric NO-bound Complex-The high frequency resonance Raman spectrum of the ferric NO-bound protein in the absence of H4B and L-Arg is shown in spectrum c of Fig. 2. The 4 and 3 modes were identified at 1372 and 1500 cm Ϫ1 , respectively, indicative of a six-coordinate low spin electronic configuration for the heme iron. The oxidation and coordination state of the ferric NO-bound iNOS oxy are not affected by the addition of L-Arg and/or H4B as evident from the high frequency Raman spectra, which was unchanged from that in Fig. 2, spectrum c (data not shown).
The low frequency resonance Raman spectra (200 -1000 cm Ϫ1 ) of the ferric NO-bound iNOS oxy complexes are shown in Fig. 4A. Fig. 4A, spectrum a is that of the complex in the absence of L-Arg and H4B. The Fe-NO stretching mode ( Fe-NO ), identified at 537 cm Ϫ1 , shifts to 533 cm Ϫ1 upon iso-tope substitution of 14  In contrast, in the presence of H4B, two isotopic sensitive lines were detected at 541 and 550 cm Ϫ1 for the 14 Fig. 4A, spectrum c. Although in the absence of H4B, L-Arg does not affect the heme environment, as no discernable differences were observed in the spectrum upon the addition of L-Arg, in the presence of H4B, L-Arg does affect the ligandrelated vibrational modes. Upon the addition of L-Arg, in the presence of H4B, one single isotopic-sensitive line was observed at 545 cm Ϫ1 that shifted to 537 cm Ϫ1 upon the isotope substitution. The larger isotope shift of 8 cm Ϫ1 with respect to the 4-cm Ϫ1 shift found in the absence of L-Arg and H4B, suggests that the 545-cm Ϫ1 mode originates from a Fe-N-O bending mode (␦ Fe-NO ) instead of a stretching mode ( Fe-NO ). Based on this assignment, we assign the 541-and 550-cm Ϫ1 lines in Fig.  4A, spectrum c to the Fe-NO and ␦ Fe-NO modes, respectively.
It is important to note that similar Fe-N-O stretching and bending modes have been reported by Hu and Kincaid (34,46) for cytochrome P-450 and chloroperoxidase. The authors assigned the low frequency component to the Fe-NO mode and the high frequency component to the ␦ Fe-NO mode. Furthermore, it was found that in P-450 the bending mode was also enhanced in the presence of a substrate just as observed here for the iNOS oxy complex. However, the substrate in P450 binds directly on top of the NO, and the enhancement of the bending mode is accounted for by a direct steric interaction between the substrate and NO, whereas the H4B binding site found in the crystal structure of NOS is remote from the distal ligand binding site. To examine whether there is a second H4B binding site in the distal side of the heme, we titrated H4B into NO-bound iNOS oxy . We found that the binding of H4B to iNOS oxy is stoichiometric with one H4B/protein molecule thereby excluding the possibility of a second binding site for H4B.
In addition to the changes in the Fe-NO stretching and bending modes, the presence of H4B also caused an increase in the relative intensity of the vibrational modes at 685 and 800 cm Ϫ1 and the appearance of new lines at 352, 390, 710, 729, and 746 cm Ϫ1 as shown in Fig. 4A, spectrum c. The absence of any shifts in these lines upon isotope substitution with 15 N 16 O confirms that they are not associated with the ligand moiety; instead, they are assigned to the vibrational modes of the heme as will be discussed later. The addition of L-Arg to the H4B- ϪArg/ϩH4B; d, ϩArg/ϩH4B. All spectra were taken with an excitation wavelength of 442 nm. bound protein complex does not introduce additional changes to these heme modes in contrast to the changes it brings about in the Fe-N-O-related modes. Table II summarizes the Fe-NO and ␦ Fe-N-O modes determined in this work, in comparison to those reported for other complexes of NOS. The data indicate that H4B binding causes the Fe-N-O moiety to be bent thereby enhancing the bending mode. In addition, the degree of bending of the Fe-N-O moiety is increased in the presence of both H4B and L-Arg, although L-Arg alone does not affect the structure of the Fe-N-O moiety suggesting that the H4B binding brings L-Arg closer to the heme iron.
The Ferrous NO-bound Complex-The resonance Raman spectra of the ferrous-NO iNOS oxy complexes were examined in the presence and absence of L-Arg and/or H4B. Unfortunately, in the absence of L-Arg, the ferrous-NO complex is unstable. It forms a five-coordinate NO-bound species in the absence of H4B and undergoes auto-oxidation in the presence of H4B consistent with data reported previously (21,47,48). The resonance Raman spectrum of the NO-bound ferrous derivative in the presence of L-Arg alone is shown in Fig. 5, spectrum a. Upon the addition of H4B, an Fe-NO-related line at 540 cm Ϫ1 is shifted to 550 cm Ϫ1 , indicating that H4B affects the hemebound ligand in a similar fashion as that observed in the ferric-NO complexes (Fig. 4). The inset above spectrum b in Fig.  5 shows the 15 N 16 O-coordinated form of the ferrous derivative of iNOS oxy in the presence of both L-Arg and H4B. Based on this frequency (533 cm Ϫ1 ) and the isotopic difference spectrum, shown in Fig. 5, spectrum c, we assign the mode at 550 cm Ϫ1 to the Fe-NO stretching mode ( Fe-NO ) because of the isotopic shift of 17 cm Ϫ1 . The assignment of this mode is consistent with that reported previously (47,49) for the ferrous-NO derivatives of nNOS as listed in Table III. It is noteworthy that the isotopic shift for the Fe-NO mode in this ferrous NO complex is much greater than that observed for the ferric NO adducts. Similar  isotopic shifts were reported for the NO adducts of the ferrous hemes of cytochrome P-450 and chloroperoxidase (34,46). It was shown by Hu and Kincaid (34) that the large isotopic shift in the ferrous derivative is a consequence of the bent Fe-N-O geometry and the partial mixing of the stretching mode with some bending character. In addition to the changes in the Fe-NO mode, the 692-, 715-, 734-, 752-, and 800-cm Ϫ1 modes were enhanced upon the addition of H4B. Again, they are assigned to the heme modes as was observed in the ferric NO-bound complexes.
The Ferric CN Ϫ -bound Complex-The high frequency Raman spectrum of the ferric-CN Ϫ derivative in the absence of L-Arg and H4B is shown in Fig. 2, spectrum d. The 3 mode was found at 1500 cm Ϫ1 , indicating a six-coordinate low spin species. The addition of L-Arg and/or H4B did not affect the high frequency resonance Raman spectrum of the CN Ϫ -bound derivative, demonstrating that the coordination and spin state remain six-coordinate low spin in the presence of L-Arg and/or H4B (data not shown).
The low frequency resonance Raman spectrum of ferric-CN Ϫ complex in the absence of H4B and L-Arg is shown in Fig. 6, spectrum a. The addition of L-Arg brings about new lines in the 400 -425-cm Ϫ1 spectral region. In the presence of H4B alone, the spectrum is similar to that observed in the absence of L-Arg and H4B. On the other hand, the spectrum obtained in the presence of both L-Arg and H4B is very similar to that obtained with L-Arg alone. Cyanide isotope-substitution experiments revealed ligand contributions in the lines at 402 and 425 cm Ϫ1 . The data did not allow for a clear assignment of these modes because of their dependence on the geometry of the Fe-C-N moiety and the mixing of these modes with other heme vibrational modes, as has been shown in cyanide adducts of P450s (50). In P450s, when the Fe-C-N moiety is linear, the Fe-CN stretching mode is located in the 410 -425-cm Ϫ1 region whereas it is in the 340 -360-cm Ϫ1 region for a bent structure. Furthermore, the Fe-C-N bending mode is in the 385-395cm Ϫ1 region for the linear form and the 420 -440-cm Ϫ1 region for the bent structure. Additional measurements are needed for iNOS oxy to make firm assignments of the Fe-C-N modes. Nonetheless, the data indicate that L-Arg has a significant effect on the structure of the Fe-CN moiety, possibly because of a direct interaction between the substrate and the CN Ϫ moiety as was observed in the ferrous-CO complexes in contrast to behavior of the ferric-NO complexes.
In addition to the CN Ϫ -related modes, heme modes are also changed by the addition of substrate and cofactor. Specifically, the presence of H4B results in a large increase in the relative intensity of the mode at 691 cm Ϫ1 , and a small enhancement in the mode at 713 cm Ϫ1 . The presence of L-Arg also induces similar intensity changes although to a lesser degree.  (45,51). When cytochrome c is unfolded, the porphyrin macrocycle adopts a planar structure with D 4h symmetry. However, when it is folded, the tertiary interactions cause the porphyrin to take on a ruffled structure. Consequently, several out-of-plane heme modes become active, and the low frequency resonance Raman spectrum displays a much more complicated pattern with respect to that of the unfolded protein. We postulate that the changes in the low frequency Raman spectrum of the iNOS oxy complexes induced   by L-Arg and/or H4B binding are a result of a change in the distorted structure of the heme. Many studies have been reported in the past (52-57) on the effects resulting from loss of planarity in porphyrins. The distortion of the porphyrin results in changes in the energies of the iron d orbitals as well as the porphyrin orbitals. As a consequence the electronic properties and redox potentials of the heme protein are changed, and the electron transfer rate with its partner protein is altered. The distortion of the heme in the crystal structures of the NOS isoforms was discussed by Raman et al. (52). It was reported that the binding of H4B did not alter the degree of non-planarity of the heme in the eNOS structure (11). However, as discussed below, based on more recent crystallographic data the addition of H4B does modify the heme planarity in the NObound form of reduced eNOS (58).
The most dramatic changes in the low frequency Raman spectra were observed in the NO derivatives of iNOS oxy induced by H4B binding (Figs. 4 and 5). Because the crystal structures of the NO-bound complexes of iNOS oxy have not been reported, we sought to examine the two structures of the ferrous NO-bound eNOS oxy complexes (1FOO and 1FOP) that are available in the PDB (58). Cursory examination of the structures indicates that the distortion of the heme is significantly greater in the presence of H4B than in the absence of H4B, when L-Arg is present, as shown in Fig. 7, a and b. To quantify the degree of distortion, we applied the normal-coordinate structural decomposition method developed by Shelnutt and co-workers (43). In the NSD method, the heme distortion is broken down into low frequency normal coordinates, including ruffling (B 1u ), saddling (B 2u ), doming (A 2u ), waving (E g ), and pyrrole propellering (A 1u ) deformations as illustrated in the left panel in Fig. 7. With this method, the mean out-of-plane displacement of the atoms in the heme macrocycle associated with each distortion coordinate can be calculated. It should be noted that the degree of heme distortion in the two subunits of the dimer is somewhat different presumably because of intersubunit interactions. For this discussion we use the average value calculated from the two subunits for each complex as listed in Table IV, because in most cases the trend is similar in the two subunits. It is also important to point out that the typical out-of-plane distortion for most heme proteins, such as hemoglobin and myoglobin, is ϳ0 -0.7 Å, and only a few heme protein contain very distorted hemes with a distortion of Ͼ1 Å (43).
We found that the total out-of-plane distortion of the ligandfree ferrous eNOS oxy is 0.97 Å in the presence of L-Arg and in the absence of H4B (Table IV, 1FOL (58)). It is decreased to 0.77 Å upon NO binding (1FOO (58)) indicating a decrease in the heme distortion. Further addition of H4B restores the distorted heme, as reflected by the increase of the total out-ofplane distortion to 1.00 Å (1FOP (58)). These results indicate that the heme distortion is very sensitive to ligand and cofactor binding. They also confirm the increased distortion to the NObound heme upon H4B binding as shown in Fig. 7, a and b. It is important to note that the major contribution to the changes in distortion is saddling with a B 2u symmetry that decreased from 0.65 to 0.42 Å upon the binding of NO, and then increased to 0.68 Å upon the addition of H4B to the NO-bound protein (Table IV).
Although NO binding to the ferrous eNOS oxy in the presence of L-Arg alone reduces the degree of heme distortion, CN Ϫ binding to the ferric iNOS oxy in the presence of both L-Arg and H4B causes the total out-of-plane distortion to increase from 0.83 to 1.08 Å (1NOD (8) versus 1N2N (59)). Furthermore, the major contribution to the changes in distortion is a combination of saddling and doming with B 2u and A 2u symmetries, respectively, as listed in Table IV. It is also interesting to note that the B 2u out-of-plane distortion is greatly diminished in a monomeric form of iNOS oxy that does not bind H4B (1NOS (9)), suggesting that the B 2u distortion may be partially associated with intersubunit interactions in the dimer, although the presence of imidazole in this structure may have influenced the degree of distortion.
Based on the assignments of the vibrational modes in ferrochelatase, which also exhibits a very distorted heme (44), several of the modes in the resonance Raman spectra of the NObound ferric complexes can be tentatively assigned (Fig. 4). In the absence of L-Arg and H4B, shown as the spectrum a in Fig.  4, the lines at 344, 676, and 752 cm Ϫ1 are assigned as 6 , 7 , and 15 , respectively. The weak 685-cm Ϫ1 mode is assigned to an out-of-plane mode, ␥ 15 , with B 2u symmetry. The presence of the weak 685-cm Ϫ1 line in the Fig. 4, spectra a and b thus suggests that the NO-bound ferric heme is slightly saddled in the absence of H4B. In the presence of H4B the new lines at 352, 710, 729, and 746 cm Ϫ1 are assigned to the out-of-plane modes, ␥ 6 , ␥ 11 , ␥ 5 , and ␥ 1 , respectively. The presence of these out-of-plane Raman modes allows for the determination of the symmetry of the distorted heme induced by H4B binding. The ␥ 6 (352 cm Ϫ1 ) and ␥ 5 (729 cm Ϫ1 ) modes are both of A 2u symmetry and are consistent with a doming type of deformation. On the other hand, the ␥ 15 (685 cm Ϫ1 ), ␥ 11 (710 cm Ϫ1 ), and ␥ 1 (746 cm Ϫ1 ) modes have B 2u , B 1u , and A 1u symmetries, respectively, which are consistent with the saddling, ruffling, and propeller deformations, respectively. The presence of these out-of-plane modes with differing symmetry types suggest that the heme is distorted along several coordinates. The similarity between the two spectra shown in Fig. 4A, a and b and that between Fig. 4A, c and d suggests that L-Arg binding does not introduce significant distortion to the NO-bound ferric heme. In addition to heme deformation, the enhancement of the 390-cm Ϫ1 line, which is assigned to a propionate mode, in the spectra c and d, suggests that the orientation of the propionate group with respect to the heme macrocycle is changed upon H4B binding probably because of a direct H-bonding interaction between the propionate and the H4B as indicated in the crystal structures of iNOS.
In the ferrous NO-bound derivative, only the spectra in the presence of L-Arg alone and in the presence of both L-Arg and H4B were obtained, because in the absence of L-Arg the protein is not stable. The H4B binding greatly enhances the 692-cm Ϫ1 line (␥ 15 ) with B 2u symmetry. It also brings about small increases in the 715-(␥ 11 ) and 734-cm Ϫ1 (␥ 5 ) lines with B 1u and A 2u symmetries, respectively. These changes suggest a large change in the saddling (B 2u ) deformation and small changes in the doming (A 2u ) and ruffling (B 1u ) deformations and are consistent with the NSD analysis of the NO-bound ferrous derivative of eNOS oxy in which the addition of H4B in the presence of L-Arg generated a large change in the saddling deformation (0.42-0.68 Å) and small changes in the doming and ruffling coordinates (Table IV).
In the CN Ϫ -bound derivative, the major change induced by the addition of H4B is an increase in the 692-cm Ϫ1 line, which is assigned to the ␥ 15 mode with B 2u symmetry. Again it indicates an increase in the heme deformation along the saddling coordinate. This is consistent with the NSD analysis in which the largest deformation (0.79 Å) in the CN Ϫ derivative of iNOS oxy occurs along the saddling coordinate.
The degree of heme distortion in the CO-bound derivative is smaller than that of the other derivatives of iNOS oxy examined in this work. The addition of L-Arg alone induces some changes to heme distortion as reflected by small enhancement in the 693-, 718-, 752-, and 803-cm Ϫ1 modes (Fig. 3, spectrum b). A similar degree of heme distortion was observed upon H4B binding as shown in Fig. 3, spectrum c. The structural effects imposed by L-Arg and H4B binding appear to be additive, hence a larger degree of heme distortion was observed in the presence of both L-Arg and H4B. As in the other derivatives, the largest change is in the 693-cm Ϫ1 (␥ 15 ) line, thereby suggesting a B 2u saddling deformation. The change at the 718-cm Ϫ1 (␥ 11 ) line suggests a small B 1u ruffling deformation. In the absence of L-Arg and H4B, none of the heme distortion modes are significant, as shown in Fig. 3, spectrum a, suggesting that the heme is in a planar geometry. The substrate and cofactor-induced heme deformation in iNOS oxy is in contrast to that of the nNOS oxy complex. In the CO derivative of nNOS oxy , two lines at 722 and 773 cm Ϫ1 are detected in the absence of the substrate and cofactor (42), which we tentatively assign as the ␥ 11 mode (B 1u ) and the 15 mode (B 1g ), respectively. The presence of the two modes suggests that in the absence of substrate and cofactor, the heme in the CO-bound nNOS oxy is ruffled. Substrate and cofactor binding causes the heme to convert to a different structure as indicated by the disappearance of the ␥ 11 and 15 modes and the increase in intensity of the modes at 752 and 798 cm Ϫ1 . The different behavior in nNOS oxy and iNOS oxy reflects the subtle differences in the structural properties of these two isoforms.
It is noteworthy that the addition of H4B to all of the iNOS oxy derivatives examined here caused an increase in the heme deformation along the saddling coordinate as reflected by the increase in the ␥ 15 mode. This is consistent with the NSD analysis in which a high degree of saddling deformation is observed in all of the derivatives examined, except that in the monomeric derivative (Table IV, 1NOS (9)). Furthermore, because of the differences in the electronic properties of the heme iron and the heme-bound ligand, the degree of heme deformation is in the following order: Fe 3ϩ -NO Ͼ Fe 2ϩ -NO Ͼ Fe 3ϩ -CN Ϫ Ͼ Fe 2ϩ -CO.
Influence of L-Arg and/or H4B on the Structural Properties of Heme-bound Ligands-The presence of L-Arg and/or H4B in iNOS oxy does not only induce heme distortion, it also affects the structural properties of the heme-bound ligands. The stability of an exogenous ligand that coordinates to the sixth position of a heme group depends on the electronic properties of the ligand, the heme iron, and the proximal residue, as well as the environment of the distal binding pocket. Strong field distal ligands, for example CO, NO, CN Ϫ , or imidazole, typically form stable complexes. On the other hand, weak field distal ligands, such as water or DTT, form complexes that are much more labile. In some cases, the binding of a weak ligand to the heme iron requires the stabilization provided by polar residues in the distal pocket. A well known example is aquo-métmyoglobin in which the heme-bound water is stabilized by a distal histidine residue through an H-bond (60). Mutation of the distal histidine to a nonpolar residue destabilizes the water leading to a five-coordinate state. The distal water ligand in metmyoglobin can also be destabilized through the mutation of the proximal histidine ligand to cysteine, because of the alteration in the electronic properties of the proximal ligand (61). In cytochrome-c peroxidase, it is believed that the imidazolate character of the proximal ligand strengthens the proximal ironhistidine bond thus pulling the heme iron out of the porphyrin plane and hindering the coordination of water to the heme iron. In most P450 types of proteins, the substrate-free protein is six-coordinate with a water bound to the distal site. Upon substrate binding to the distal site, the distal water ligand may be displaced resulting in a five-coordinate high spin heme because of unfavorable substrate-ligand steric interactions (62,63). In contrast, in the substrate-free form of chloroperoxidase (PDB code 1CPO (64)), which like P450 has a cysteine proximal ligand, a five-coordinate heme was observed, although there is a water molecule in the distal pocket that is only 3.3 Å away from the heme iron (64).
The addition of L-Arg to the ferric derivative of NOS brings about a conversion from a six-coordinate low spin heme to a five-coordinate high spin heme as demonstrated in Fig. 1, indicating the exclusion of a distal water molecule from the heme iron. In NOS, L-Arg binds directly on top of the ligand binding site; the exclusion of the water is thus attributed to the steric hindrance imposed by L-Arg as that observed in P450 types of proteins. A similar six-coordinate low spin to five-coordinate high spin transition was also observed upon H4B binding, despite the fact that H4B does not directly interact with the heme ligand based on crystallographic data (8,10,11). On the basis of the H4B titration experiment reported here, the possibility of a second binding site for H4B in the distal pocket is ruled out. A direct steric constraint to the distal water because of an allosteric structural transition is also excluded, because the crystal structure of iNOS shows that in the presence of H4B the heme is domed, and the heme iron atom is displaced out of the porphyrin plane in the direction of the proximal thiolate ligand resulting in a very open distal pocket in (PDB code 2NOD) (8). Although a water molecule is present in the distal pocket in this crystal structure, it is 4.28 Å away from the heme iron atom, which is too far to form a covalent bond to the iron. We postulate that the H-bond between the H4B and the heme propionate group causes the distortion of the heme that destabilizes the bonding between the distal water and the heme iron atom leading to the five-coordinate structure. In addition, a local hydrophobic environment does not lead to stabilizing a bound water molecule.
In the CO-derivatives of iNOS oxy , the binding of L-Arg to iNOS oxy causes the Fe-CO mode to sharpen and shift to a higher frequency, regardless of the presence of H4B (Fig. 3,  spectra b and d); in addition, the intensity of the ␦ Fe-C-O bending mode is enhanced significantly. These results can be accounted for by a direct interaction between the L-Arg and the heme-bound CO. The sharpening of the Fe-CO line in the presence of L-Arg suggests a decreased conformational freedom for the Fe-C-O moiety because of the presence of an H-bond between CO and L-Arg. The shift to a higher frequency of the Fe-CO stretching mode and the strengthening of the Fe-C-O bending mode indicates that the interaction with the L-Arg also causes the Fe-C-O moiety to become bent (65). The crystal structures of CO-bound NOS complexes are not available, but in the CO-free structures, the terminal nitrogen of the guanidinium group of L-Arg is located ϳ4 Å away from the heme iron, suggesting that the CO ligand can be stabilized by L-Arg through a hydrogen bond. A direct hydrogen bonding interaction between the CO ligand and the L-Arg is supported by Fourier transform infrared studies of the ferrous-CO derivative of iNOS oxy showing that a 0.8-cm Ϫ1 shift in C-O when the solvent H 2 O was replaced with D 2 O (66). The presence of H4B alone makes negligible changes to the Fe-C-O modes, but small changes to the heme modes are seen. The small changes in the out-of-plane heme modes upon the addition of L-Arg or H4B indicate a slight deformation of the heme. Interestingly, the addition of H4B to the L-Arg-bound protein does not bring about additional changes to the Fe-C-O moiety, whereas the heme distortion is further enhanced suggesting that the changes in heme deformation do not affect the H-bonding interactions between the L-Arg and the heme-bound CO when H4B is present.
The CO-bound ferrous heme iron and the NO-bound ferric heme iron are isoelectronic; in addition, both of them typically bind in a preferentially perpendicular orientation with respect to the porphyrin plane. It was thus anticipated that L-Arg would interact strongly with NO in the NO-bound ferric derivative in a similar fashion as that observed in the CO-bound ferrous derivative. To our surprise, L-Arg had absolutely no effect on the spectrum of the NO-bound complex in the absence of H4B. Unfortunately, because there are no changes in the spectrum, we are unable to determine whether L-Arg binds in a site too far from the NO-bound heme to interact with the NO or whether the L-Arg does not bind at all. In contrast, the binding of H4B alone causes the shift of the Fe-NO mode and the appearance of the ␦ Fe-N-O mode indicating that the Fe-N-O moiety adopts a bent conformation. It is important to note that so far there is no reported case in which the bending mode is present when the Fe-N-O assumes a linear structure that is perpendicular to the heme plane. A direct interaction between H4B and the Fe-N-O moiety is excluded, because there is no evidence that H4B can bind to the distal pocket of NOS. We postulate that the bent Fe-N-O conformation is a result an electronic effect introduced by heme distortion as evident from the enhancement of the heme out-of-plane modes.
Although we are unsure whether L-Arg binds to the ferric NO-bound iNOS oxy in the absence of H4B, the distinct changes in spectrum c in Fig. 4 with respect to spectrum d demonstrates that L-Arg does bind to the protein in the presence of H4B. Upon the addition of L-Arg to the H4B-bound protein, the ␦ Fe-N-O mode is further enhanced and shifted, although the heme out-of-plane modes are unaffected. We postulate that in the presence of H4B the tilt angle of the Fe-N-O moiety is further increased upon the addition of L-Arg as a result of a direct steric or H-bonding interaction imposed by L-Arg. Similar changes, although not as dramatic, were observed in cytochrome P-450 upon the addition of a substrate (34).
To examine whether the difference in ligand-protein interactions in the NO-bound ferric protein and the CO-bound ferrous protein is a result of the differences in the redox state of the heme, we examined a CN Ϫ -bound ferric derivative. It was found that the Fe-C-N moiety is much more sensitive to the binding of L-Arg than to the binding of H4B, similar to the behavior of the ferrous CO-bound derivatives. Thus, we concluded that the ligand-protein interactions are not solely determined by the redox state of the heme iron. We postulate that the presence of L-Arg in the distal pocket of the ferric CN Ϫbound iNOS oxy causes the Fe-C-N moiety to adopt a bent structure, as reflected by the presence of the new modes at 402 and 425 cm Ϫ1 (Fig. 6). Although distinct changes to the heme deformation modes are visible, the Fe-C-N-associated modes are not affected by the binding of H4B alone. However, the binding of H4B the presence of L-Arg causes the enhancement of the mode at 402 cm Ϫ1 with respect to that at 425 cm Ϫ1 , suggesting a more bent structure for the Fe-C-N moiety. This change is associated with further deformation of the heme as evident by the enhancement of the heme out-of-plane modes (Fig. 6, spectrum d with respect to spectrum c). The bent structure of the Fe-CN moiety and the distorted heme is confirmed in the crystal structure of the cyanide complex of iNOS oxy (PDB code 1N2N (59)).
In the NO-bound ferrous iNOS oxy derivative (Fig. 5), a significant change in the Fe-NO mode is also present upon H4B binding in the presence of L-Arg, similar to that observed in the ferric NO-bound derivative (Fig. 4, spectrum b versus d), again consistent with the conclusion that the ligand-protein interactions are not determined by the redox state of the heme iron. The change in the Fe-NO mode is concomitant with a further deformation of the heme as evident by the enhancement of the heme out-of-plane modes in Fig. 5, spectrum b with respect to spectrum a. We postulate that the sensitivity of the Fe-NO mode to H4B is a result of an electronic effect exerted by the distorted heme similar to that observed in the other derivatives of the iNOS oxy complex studied here.
Implications on NOS Physiology-Electron nuclear double resonance (ENDOR) studies of the ferric derivatives of NOS, in the presence of H4B and in the absence of any exogenous ligands, show that the positions of the guanidino nitrogen of L-Arg is 4.1-4.2 Å away from the heme iron regardless of the type of isoform examined (39) consistent with the crystallographic data showing that the distance ranges from 4.0 to 4.4 Å in the various isoforms. This is in contrast to the conclusions drawn by Fan et al. (38) that a clear structural difference was found for nNOS versus the other two isoforms in the presence of L-Arg and H4B based on the resonance Raman studies of the CO-bound ferrous derivatives. In that work, the frequency of the Fe-CO mode of nNOS was found at 503 cm Ϫ1 in the pres-ence of L-Arg in contrast to the 512 cm Ϫ1 found for the other two isoforms as shown in Table I.
To determine whether the disagreement between the EN-DOR and Raman data is a result of the differences in the oxidation state of the heme iron, we compared the ligandrelated modes of the NO-bound ferric and ferrous derivatives of iNOS to that of nNOS as listed in Table II and III. It was found that the ligand-related frequencies were essentially the same for iNOS and nNOS regardless of the oxidation states of the heme iron. Because the distal binding site appears to be the same for the NO derivatives of the two isoforms regardless of their redox state, we postulate that the differences seen in ENDOR and Raman data may be a consequence of a distinct heme distortion or proximal bond strength in nNOS with respect to iNOS (and eNOS) in response to CO coordination. Current experiments are planned to distinguish between these possibilities.
It has been shown that the NO generated from NOS plays an important role in regulating its enzymatic activity by forming a self-inhibitory complex with the heme iron and by influencing the stability of the dimeric interaction (67). At the completion of the catalytic cycle, the NO that is produced in the distal pocket binds geminately to the ferric heme and thereby inhibits the enzyme. Santolini et al. (37) demonstrated that the degree of self-inhibition by NO depends on the following factors: 1) the off-rate of NO from the ferric heme, 2) the ease of reduction of the ferric NO-bound form to the ferrous derivative, and 3) the ease of auto-oxidation from the ferrous form back to the ferric form. Because of isoform-specific rates, the degree of self-inhibition differs substantially from one isoform to another ranging from 70 to 90% in nNOS, to 25% in iNOS, and to a negligible amount in eNOS (36). Whether or not NO binding leads to an extensive inhibition strongly depends on the reduction rate of the NO-bound ferric protein to the ferrous derivative in which the NO off-rate is much slower. Here, we found that the ferrous NO-bound complex of iNOS oxy was not stable without L-Arg. In the absence of both L-Arg and H4B, it converts to a five-coordinate species. In the presence of H4B alone, it spontaneously auto-oxidizes to the ferric protein. In contrast, the auto-oxidation rate of NO-bound ferrous derivatives of nNOS in the presence of H4B alone is much slower (14,32,33). The higher auto-oxidation rate of iNOS oxy with respect to nNOS in the presence of H4B alone may be an important factor that accounts for the lower degree of self-inhibition by NO in iNOS during the enzymatic turnover, because at the end of the catalytic cycle the L-Arg is totally consumed.
In heme proteins with planar heme macrocycles, the reduction of the NO-bound ferric form to the ferrous form happens very rapidly and the NO off rate from the ferrous protein is extremely slow because of the high stability of the reduced state. In those proteins in which NO delivery is physiologically important, such as nitrophorins, the reduction must be inhibited, and this is achieved through heme distortion. Nitrophorins are a family of proteins present in blood-sucking insects that release NO to bring about vasodilation and reduction of blood coagulation (68). It was found that the heme in nitrophorin-4 is highly ruffled, and the NO is bent despite the fact that there are no residues in the distal pocket that can directly interact with the NO. Typically, the low spin configuration of the ferric iron is (d xy ) 2 (d xz ,d yz ) 3 . However, as pointed out by Walker and co-workers (69), ruffling of the heme in nitrophorin-4 is associated with a change in the electronic structure to (d xz ,d yz ) 4 (d xy ) 1 as the unpaired electron in the d xy orbital cannot mix with the porphyrin -system if the heme is planar, whereas upon ruffling of the heme, the orbitals of the porphyrin have in-plane components that can overlap with the d xy orbital. This allows the half-filled d xy orbital of the heme iron to accept additional electron density from the porphyrin orbitals in the ferric oxidation state. This additional electron density in the d xy orbital raises the barrier for the reduction of the iron, because in the ferrous oxidation state the d xy orbital is filled by the electrons originating from the iron. As a consequence, the reduction of the heme iron becomes more difficult.
We postulate that the heme distortion observed in this work serves to regulate the autoinhibition by making the reduction of the NO-bound heme unfavorable in a similar fashion as that observed in nitrophorins. Moreover, possible differences between the rates of reduction in the isoforms may be a consequence of variations in the heme distortion. In addition to reducing the autoinhibition when H4B is present, the control of the heme distortion may also provide an addition safety control for the living cells when H4B is not available. In this case the distortion is reduced, and the enzyme becomes locked in a five-coordinate NO-bound ferrous complex. This may serve to prevent the formation and release of reactive oxygen species, which could have deleterious effects on the cells, because oxygen exposure to the five-coordinate NO-bound ferrous complex will bring about the formation of nitrate and a ferric heme, which is not harmful to the cell.
The ligand-substrate interactions reported here for the various ligation and oxidation states of iNOS oxy also demonstrate the flexibility of the distal pocket. For example, in the absence of H4B, the L-Arg interacted with the CO in the ferrous-CO complex and the cyanide in the ferric-cyanide complex but not the NO in the ferric-NO complex. As each of these ligands typically binds in a preferentially linear conformation, the distinctive ligand-substrate interaction in the NO derivative demonstrates that there is considerable conformational flexibility in the substrate binding site. Physiologically, NOS first converts L-Arg to NOHA and then from NOHA to L-citrulline. It is well established that the catalytic mechanisms are very different for the two steps of the reaction, although both involve the activation of heme-bound dioxygen and the insertion of oxygen into the substrates. In addition, the citrulline product has to be released and not block the substrate binding site. To carry out the two mechanistically distinct reaction steps and release the citrulline, it is important for the enzyme to adopt different conformations in the heme pocket in response to substrate binding. This structural flexibility observed here may serve to ensure the correct substrate stereochemistry required for the differential reactivity for the two reaction steps. Furthermore, the data reported here show that changes in the electrostatic interactions of the heme-bound ligand have an influence on the substrate positioning. During catalysis as the oxygen goes from its initial electronic structure upon binding to peroxo-or ferrylactivated complexes, changes in the substrate stereochemistry may be induced by these differences in the oxygen electronic configuration. Thus, substrate binding flexibility may serve to allow for optimization of the substrate during various steps of the catalytic process.
In summary, the data reported here provide a view of the molecular picture underlying NO production reaction in NOS. We show that interactions between the various heme ligands and the protein matrix in response to L-Arg and/or H4B binding are coupled to differing degrees of heme distortion. The heme distortion affects the electronic as well as the structural properties of the heme-bound ligand that may have an important influence on the substrate positioning as well as distinct oxygen chemistry required for the two catalytic steps in NOS. This novel heme distortion in NOS is also very important for the regulation of the NO autoinhibition to suit the individual physiological needs for each isoform of NOS.