|Home | About | Journals | Submit | Contact Us | Français|
A Ru-diimine wire, [(4,4’,5,5’-tetramethylbipyridine)2Ru(F9bp)]2+ (tmRu-F9bp, where F9bp is 4-methyl-4’-methylperfluorobiphenylbipyridine), binds tightly to the oxidase domain of inducible nitric oxide synthase (iNOSoxy). The binding of tmRu-F9bp is independent of tetrahydrobiopterin, arginine, and imidazole, indicating that the wire resides on the surface of the enzyme, distant from the active-site heme. Photoreduction of an imidazole-bound active-site heme iron in the enzyme-wire conjugate (kET = 2(1) × 107 s-1) is fully seven orders of magnitude faster than the in vivo process.
Nitric oxide synthase (NOS) is a heme monooxygenase that catalyzes the five-electron oxidation of L-arginine and O2 to citrulline and nitric oxide (NO). NOS effects this transformation in two turnovers, producing NG-hydroxy-L-arginine (NHA) as an enzyme-bound intermediate, requiring three electrons from its reductase domain. Each turnover is expected to proceed through a mechanism similar to that of cytochrome P450 (although the two turnovers may utilize a different species for substrate oxidation), central to which are two slow electron transfer (ET) events [1-6]. The first ET event reduces the resting, substrate-bound heme to the ferrous state, which then binds oxygen to create the last observable intermediate (ferric-superoxo) [4, 7]. It is thought that the second ET event, where the electron is supplied by the cofactor tetrahydrobiopterin (BH4), produces one or more high-valent heme species, with substrate oxidation possibly occurring from a ferryl-porphyrin+• intermediate (Compound I) [4, 8]. The sluggishness of the second ET step, however, has so far prevented the characterization of high-valent intermediates in the catalytic cycle in solution [3, 4, 9]. Cryoreduction of the heme domain of ferric-superoxo endothelial NOS at 77 K leads to the formation of a ferric-peroxo species . Annealing at 165 K results in conversion to the product state without the appearance of intermediates. These data suggest that O-O bond cleavage is slower than reaction with substrate.
By employing laser-induced ET to reduce the active-site heme very rapidly, it should be possible to observe high-valent intermediates that follow in the catalytic cycle. Toward this end, we and others have developed photoactive electron tunneling wires to deliver electrons and holes to and from the deeply buried heme active sites in P450cam [10-12] and NOS [13-16]. Importantly, one of the NOS wires, tmRu-F9bp (Chart 1), can potentially probe the catalytic cycle, since it binds tightly and specifically to the oxidase domain of the inducible form of the enzyme (iNOSoxy) in a region that is distant from the active site . Here we demonstrate that an imidazole-ligated heme in tmRu-F9bp:iNOSoxy can be photoreduced several million times faster (kET = 2(1) × 107 s-1) than the physiological ET reaction.
The tmRu-F9bp complex was synthesized as described previously [11, 18, 19]. Tetramethylphenylenediamine (TMPD) was obtained from Aldrich and vacuum-sublimed before use. BH4 (Aldrich) was stored under argon at -20 °C. All other chemicals were used as received from Sigma, JT Baker, Fischer, EM Sciences, and Mallinckrodt. UV-visible absorption spectra were acquired on an Agilent 8453 UV spectrometer. Gel electrophoresis was run on a Phast System (Pharmacia) with 8-25 percent gradient precast agarose gels and SDS buffer strips. Samples were loaded in 4x SDS buffer and stained with Coomassie blue. Samples were run against Precision Plus All-Blue standards (BioRad).
The heme domain of iNOS with a C-terminal His6 tag was overexpressed in E. coli and purified as described previously  with several exceptions. Briefly, expression cells were subjected to two rounds of chemical lysis by pelleting and resuspension in 40 mL of B-PER lysis buffer (protein extraction reagent B, Pierce). The lysis buffer included a cocktail of protease inhibitors (10 μg/mL benzamidine, 5 μg/mL leupeptin, 1 μg/mL each pepstatin, antipain, and chymostatin, and ~500 μM Pefabloc (Roche)) as well as 100 μg/mL DNase, 100 μg/mL RNase, ~500 μg/mL lysozyme, and 20 mM imidazole per liter of cells. The suspension was centrifuged and the supernatant was loaded directly onto a His6 immobilized metalion affinity chromatography column (5 mL Ni2+:HisTrap, Amersham). Once the protein was completely loaded, it was washed with 20 column volumes of 20 mM imidazole in 50 mM NaPi/300 mM NaCl/pH 8. The protein was eluted with 150 mM imidazole and concentrated to ~3 mL in an Amicon Ultra centrifugation device (10,000 MWCO, Millipore). The concentrated sample was then further purified over a size-exclusion column, as described previously . The anion exchange column was omitted when ≥95 percent purity was confirmed by UV-visible spectroscopy and gel electrophoresis. The purified protein was concentrated to ~200 μM, divided into 100 μL aliquots, and stored in 50 percent glycerol at -80 °C.
Small aliquots of iNOSoxy were thawed and exchanged into phosphate buffer (50 mM KPi, 50 mM KCl, pH 7.4) using a PD-10 desalting column (BioRad) immediately before use. The position of the heme Soret maximum (422 nm) confirmed the presence of low-spin, water-bound heme [17, 20]. The heme protein concentration was determined using ε422 = 75 mM-1cm-1 per unit heme . For the inhibitor-bound samples, imidazole (400-500 μM) was added, and binding was confirmed by a Soret shift to 428 nm [17, 20]. For substrate-bound, pterin-free samples, 1 mM arginine was added to dilute (~2-20 μM) iNOSoxy and allowed to incubate at 4 °C for approximately 30 min. In the absence of pterin (BH4), only partial conversion to a high-spin heme (λmax = 398 nm [20, 21]) was observed. For substrate- and pterin-bound samples, fresh BH4 solutions were prepared daily. Phosphate buffer was thoroughly deoxygenated by bubbling with argon for ≥10 min. Solid BH4 was added to the degassed buffer under a counter-flow of argon. Dilute iNOSoxy (~2-20 μM) was deoxygenated by at least 30 evacuation-Ar backfill cycles, taking care to avoid bubbling of the solution. Aliquots of concentrated, deoxygenated pterin and arginine stocks were then added to the protein solution, giving final concentrations of 100 μM BH4 and 1 mM arginine. The solution was incubated for 2 h at 4 °C; binding of BH4 and arginine was confirmed by a Soret shift to 396 nm [22, 23].
For quenching experiments, 1 M ascorbate stock solutions were prepared daily by dissolving ascorbate in thoroughly deoxygenated 1 M KOH. Ascorbate (1 M) and solid TMPD were added to deoxygenated protein solutions under a counter-flow of argon.
Luminescence decay and transient absorption measurements were made as described previously [24-26]. The ~8 ns, 480 nm excitation pulses were produced by a Nd:YAG pumped optical parametric oscillator. Data were collected at 1×109 samples s-1 using a LeCroy digital oscilloscope. Transient absorbance data were converted from intensity to absorbance using the following expression (Eq. 1):
where I is the intensity of light transmitted through the sample excitation volume, and I0 is the average transmitted light intensity during the 200 ns prior to the laser shot. Luminescence decay curves and transient absorbance traces were fit to one, two, or three exponentials using a nonlinear least-squares algorithm (Eq. 2, Igor Pro):
Each experiment was repeated at least three times unless indicated otherwise.
At a given time after excitation, the absorbance observed at a given wavelength (λ) between 400 and 450 nm is (Eq. 3):
Since ascorbate, TMPD, and TMPD+• do not absorb strongly in this region (under the conditions of these experiments, Figure S2), the contributions of these species were neglected. Owing to substantial populations of unbound Ru-complex, the absorbance changes at these wavelengths due to depopulation of RuII and formation of *RuII are large compared to those for FeII formation because [*RuII][FeII]. Moreover, the presence of both free and iNOSoxy-bound wire complicates the transient absorbance kinetics. In fitting these data, we were unable to identify a phase that was distinct from those corresponding to *RuII decay in bound and free wires, and that reliably could be attributed to intraprotein ET from RuI to FeIII (kET).
In order to characterize the RuI to FeIII ET kinetics, we developed a procedure to remove the *RuII contribution from the transient absorbance kinetics. The isosbestic point for low-spin, imidazole-bound FeIII iNOSoxy and the product FeII species occurs at 438 nm. In each experiment, therefore, transient absorbance of the Im-iNOSoxy/wire/quencher system at 438 nm reflects the *RuII bleach and recovery, but contains no contribution from iNOSoxy. Using an experimentally validated *RuII-RuII difference spectrum, we determined scaling factors by which we could multiply the 438 nm transient signals to produce estimates of the *RuII contributions to the observed kinetics at several other wavelengths ([*RuII-RuII]: Δε452/ Δε438 = 1.12, Δε425/ Δε438 = 0.66). The calculated *RuII signals were subtracted from the observed transient kinetics to produce signals corresponding to the time dependence of [FeII]. Transient RuI absorbance was neglected because its Δε values are <10 percent of those for [FeII-FeIII] (Figure S5). The resulting corrected traces were then fit to single exponential functions according to Eq. 2.
We have previously shown that tmRu-F9bp (Chart 1) binds to iNOSoxy independently of substrate and BH4 with a dissociation constant of ~1 μM . Remarkably, this wire binds at a site distant from the active-site channel, as demonstrated by the finding that a known channel-binding wire does not displace tmRu-F9bp from the enzyme . While the precise binding site has not been definitively established, Förster energy transfer measurements indicate that it may be in the hydrophobic pocket thought to be the docking site for the iNOS reductase domain [17, 27]. Experiments with Ru(bpy)32+ show that the photosensitizer alone does not bind to the enzyme (Figure S1), suggesting that the perfluorobiphenyl unit is largely responsible for the strong association of the wire with a hydrophobic iNOSoxy surface region.
In previous work, we described Ru-diimine wires that reduce the heme of cytochrome P450 directly upon photoexcitation . In these experiments the wire termini ligated the iron center, providing an efficient through-bond coupling pathway between the sensitizer and the heme. In contrast, the Ru-wire described here does not directly photoreduce the heme, so we employed a flash/quench method with exogenous reductants to produce FeII . In this experiment, a quencher (Q) reduces the photoexcited sensitizer to create a strongly reducing species (RuI in Scheme 1). In the absence of other electron acceptors, the lifetime of RuI is dependent on the rate of recombination with the oxidized quencher (kr in Scheme 1). Because Q+ and RuI are present at low and equal concentrations, recombination is slow (ms timescale) and heme reduction competes effectively.
Owing to its high solubility in water and lack of spectral interference with heme Soret changes, ascorbate (Asc) is an attractive choice as a quencher for this system. Even at high concentrations (10 mM), however, Asc quenching produces only small yields of FeII (Figures 1 and and2).2). TMPD (Figure S3) is a better quencher than Asc, but has limited solubility in water [29, 30]. Further, TMPD autoxidizes to create a soluble bright blue cation radical in aqueous media . Under conditions necessary for efficient excited-state quenching, the production of the radical rapidly turns the solution dark blue, obscuring small transient changes in the heme spectrum (Supporting Information).
These problems were overcome by employing both quenchers [32, 33]. In a sample containing 10 mM Asc with saturated TMPD, the superior quenching capability of TMPD can be exploited (Figure 1, red dotted trace) while Asc serves to keep the TMPD reduced. With Asc present, TMPD+• does not accumulate, even after 60 min of photoexcitation in the presence of tmRu-F9bp.
Single wavelength transient absorbance measurements with imidazole-bound iNOSoxy in the presence of one equivalent of tmRu-F9bp, 10 mM Asc, and saturated TMPD reveal that photochemically generated RuI disappears with concomitant formation of a new Fe species within 50 ns of excitation at 480 nm (Figure 3). A difference spectrum constructed from the single-wavelength data at 2 μs (Figure 4) shows the bleach of the Im-FeIII Soret absorption at 428 nm and increased absorbance to the red with a difference-spectrum maximum at 445 nm. The rate of decay of RuI and reappearance of RuII is approximately equal to the rate of changes in the Soret region. Given that RuII is reformed, the most likely explanation for spectral changes between 400 and 450 nm is RuI to FeIII ET, which produces a new FeII heme species. Control experiments with Ru(bpy)32+ indicate that the tmRu-F9bp perfluorobiphenyl moiety is required for heme reduction. In the presence of Ru(bpy)32+ and quenchers, transient absorbance traces show only the production of RuI (Figure S4).
In order to estimate the specific rate of FeII formation, *RuII contributions were subtracted from the transient absorbance data as described in Materials and Methods (representative single-wavelength traces are shown in Figure 5). The traces (a minimum of four wavelengths from each of four different experiments completed on different days) were fit to a single exponential function: kET = 2(1)×107 s-1.
This is a remarkably rapid reduction given the estimated Ru-heme distance of 20.2 Å  and the absence of a through-bond pathway to the heme. Given its slim profile, hydrophobicity, and potential to π-stack with aromatic residues, the perfluorobiphenyl moiety of tmRu-F9bp may intercalate into the protein interior, leaving open the possibility of a through-wire hopping mechanism .
In order to determine the nature of the product of electron transfer to the heme, the six-coordinate FeIII-Im species was reduced under equilibrium conditions for comparison with the transient data. Reaction of FeIII-Im with sodium dithionite in a glove box under an inert atmosphere, followed by removal of excess dithionite on a size-exclusion (PD-10) column equilibrated with 10 mM imidazole, produced a species with the absorption spectrum shown in Figure 6.
Reduction of NOS has been extensively studied [9, 23, 34-37]. Six-coordinate ferrous-NO and -CO species have been characterized by several investigators [23, 36, 37]; and, in the absence of arginine and BH4, it has been shown that these six-coordinate species are unstable. Addition of CO (or NO) to five-coordinate FeII causes a red-shift in the Soret band to 444 nm (or 440 nm) . The 444 nm band blue-shifts over time to 421 nm, which suggests that a species analogous to the inactive P420 form of cytochrome P450 is produced. It has been proposed that the axial thiolate is not bound to the heme iron in the 421 nm species [23, 37].
The blue-shift of the iNOSoxy Soret peak upon dithionite reduction (Figure 6, inset) demonstrates that the red-shifted transient FeII species produced by photochemical heme reduction likely has different axial coordination. The steady-state FeII absorption spectrum is in good agreement with that reported for Drosphila melanogaster DHR51, a heme protein believed to possess axial Cys and His ligands . Similar spectra have been reported for FeII forms of mutant cytochrome c and myoglobin engineered to have axial Cys and His ligands [39, 40]. In each of these FeII proteins, the Soret maximum is slightly blue-shifted relative to its position in the FeIII form, indicating the presence of a low-spin FeII heme in which imidazole remains bound but the thiolate ligand has been displaced. Further, five-coordinate ferrous iNOSoxy has been generated, showing a blue-shift of the Soret from the ferric species similar to the spectrum in Figure 6 .
In contrast, the red-shifted Soret band found for photochemically reduced iNOS is analogous to that resulting from cryoreduction of ferric cytochrome P450. Irradiation of six-coordinate low-spin FeIII P450 in a frozen matrix produces a low-spin, presumably six-coordinate FeII product . Annealing at higher temperatures leads to the high-spin FeII product that is observed under equilibrium conditions. We suggest that the transient FeII iNOS species formed by photochemical reduction contains a low-spin FeII heme with axial Cys and imidazole ligands. In our experiment, this species is likely reoxidized by TMPD+• before loss of axial ligation, which would generate the species observed under equilibrium conditions.
We have developed a system in which the heme of inducible nitric oxide synthase can be photoreduced rapidly without interfering with substrate/cofactor binding. Employing flash-quench experiments with a surface-binding Ru-diimine wire in combination with reductive quenchers, we observed ET to the imidazole-bound heme of iNOSoxy fully seven orders of magnitude faster than the natural reduction. This finding represents an important step toward our goal of identifying reactive intermediates in the catalytic cycles of heme monooxygenases.
We thank Michael Marletta for the generous gift of protein reagents and for assistance with the iNOS purification protocol. This work was supported by NIH (DK19038 (HBG); GM070868 (HBG); GM068461 (JRW)), by the Fannie and John Hertz Foundation (ARD), the Ellison Medical Foundation (Senior Scholar Award in Aging to HBG), a NSF graduate research fellowship (CAW), and the Arnold and Mabel Beckman Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.