Insights into Nitrosoalkane Binding to Myoglobin Provided by Crystallography of Wild-type and Distal Pocket Mutant Derivatives

Abstract

Nitrosoalkanes (R-N=O; R = alkyl) are biological intermediates that form from the oxidative metabolism of various amine (RNH₂) drugs or from the reduction of nitroorganics (RNO₂). RNO compounds bind to and inhibit various heme proteins. However, structural information on the resulting Fe–RNO moieties remains limited. We report the preparation of ferrous wild-type and H64A sw Mb^II–RNO derivatives (λ_max 424 nm; R = Me, Et, Pr, iPr) from the reactions of Mb^III–H₂O with dithionite and nitroalkanes. The apparent extent of formation of the wt Mb derivatives followed the order MeNO > EtNO > PrNO > iPrNO, whereas the order was the opposite for the H64A derivatives. Ferricyanide oxidation of the Mb^II–RNO derivatives resulted in the formation of the ferric Mb^III–H₂O precursors with loss of the RNO ligands. X-ray crystal structures of the wt Mb^II–RNO derivatives at 1.76–2.0 Å resoln. revealed N-binding of RNO to Fe and the presence of H-bonding interactions between the nitroso O-atoms and distal pocket His64. The nitroso O-atoms pointed in the general direction of the protein exterior, and the hydrophobic R groups pointed towards the protein interior. X-ray crystal structures for the H64A mutant derivatives were determined at 1.74–1.80 Å resoln. An analysis of the distal pocket amino acid surface landscape provided an explanation for the differences in ligand orientations adopted by the EtNO and PrNO ligands in their wt and H64A structures. Our results provide a good baseline for the structural analysis of RNO binding to heme proteins possessing small distal pockets.

Graphical Abstract

INTRODUCTION

Organic nitrosoalkanes (R-N=O; R = alkyl) are biological intermediates that form during oxidative metabolism of amines (RNH₂), or form from the reductive conversion of nitroorganics (RNO₂) (Figure 1). For instance, the heme protein cytochrome P450 enables the oxidative metabolism of amine-containing drugs such as amphetamine,^1–4 macrolide antibiotics such as erythromycin^5,6 and troleandomycin,⁶ and the antiestrogen tamoxifen.⁷ In these examples, the resulting RNO metabolites bind to the heme center (middle of Figure 1; as determined by spectroscopy) and inhibit protein function.

Figure 1.

Oxidative conversion of amines (left) and reductive conversion of nitroorganics (right) to organic nitroso compounds and subsequent adduct formation with heme.

Reductive activation of RNO₂ to yield heme–RNO derivatives have also been described for some nitroorganics with cytochrome P450,^{8, 9} horse heart (hh)^{10, 11} and sperm whale (sw)¹² myoglobin (Mb), human hemoglobin (Hb),^{10, 13} and other heme proteins;^14–21 these heme–RNO derivatives were characterized by UV-vis spectroscopy. Importantly, it has been reported that human red blood cells possess sufficient reducing capacity for the metabolic conversion of various nitroorganics.²² The binding of the more stable aromatic nitrosoarenes (ArNO) to Hb and Mb have been studied in detail by Gibson²³ and others.^24–26

Published X-ray crystal structural characterizations of heme protein–RNO interactions (R = alkyl or aryl) have to date been limited to only a handful of compounds, namely those of legHb–PhNO,^{27, 28} hh Mb–EtNO,¹¹ H64A sw Mb–nitrosoamphetamine,²⁹ Hb–MeNO,¹³ and Hb–EtNO.¹³ These five structures reveal monodentate N-binding of the RNO groups to the heme centers (Figure 2). To date, the alternate O-binding mode of RNO to heme centers has been limited to synthetic heme models;^30–32 a computational analysis of this binding mode provides a molecular orbital explanation for the relatively weaker Fe^III–RNO interactions in these O-bonded complexes.³² When N-bound, RNO ligands often display π-acid character, withdrawing electron density from the metal centers to which they are bound (i.e., generating Mδ⁺–(RNO)δ- character).^33–36

Figure 2.

Monodentate binding of RNO to heme centers

Of particular interest to us are the observations that the smallest nitroorganics RNO₂ (R = methyl, ethyl, propyl, isopropyl) are utilized extensively in industrial applications and present varied toxicity concerns in the environment.³⁷ In addition, human exposure to these nitroorganics has frequently led to health concerns. For example, accidental ingestions of nitroethane artificial-fingernail removers have led to the onset of methemoglobinemia in children.^{38, 39} Of similar concern, nitrosomethane is known to be generated via metabolic activation of some aziridine-based antitumor drugs and serves to inhibit mitochondrial respiration in living cells.⁴⁰

In this article, we report our UV-vis spectral and X-ray crystal structural investigations on a representative set of wild-type sperm whale (wt sw) Mb–RNO and distal pocket H64A derivatives (left of Figure 3) to assess the role of the His64 residue on complex formation and ligand orientation.

Figure 3.

Sketches of the wt and H64A Mb active sites, and the nitrosoalkanes (boxed area) used in this study.

We also probe the effect of alkyl chain length on complex formation and ligand orientation. Our selection of small RNO₂/RNO ligands are such that they increase sequentially in chain length (Me, Et, Pr) and in ligand volume (e.g., Pr vs. iPr). As indicated, their nitro precursors are utilized heavily in industrial settings and are prone to human exposure. This work complements elegant spectroscopic and crystallographic studies by Olson, Phillips, and coworkers on the related isocyanide derivatives Mb^II–CNR (R = Me, Et, Pr, Bu),^41–45 in which the isocyanide ligand conformations were dependent not only on the identity of the alkyl chain, but also on the crystallographic space group of the wild-type versus native Mb protein.⁴⁵

MATERIALS AND METHODS

Expression and Purification of Wild-Type (wt) and Mutant (H64A) Sperm Whale Myoglobin (sw Mb).

The recombinant wt sw Mb plasmid was a generous gift from Dr. Mario Rivera (then at the University of Kansas, Lawrence, KS); the recombinant wt protein (UniProtKB MB: P02185) contains a D122N mutation. Mutagenesis was performed using the Quick-Change method (Strategene) by Dr. Bing Wang. The resulting plasmid was transformed into E. coli DH5α competent cells and the mutation was confirmed by DNA sequencing.

Wild-type and H64A sperm whale myoglobins were expressed in E. coli BL21 (DE3) cells and purified as described by Springer and Sligar⁴⁶ with a few modifications.^{29, 47, 48} Briefly, cells were resuspended in lysis buffer (50 mM Tris-HCl, 1 mM EDTA, 1 mM PMSF, pH 7.4) and lysed using an Avestin C3 EmulsiFlex homogenizer. All cell lysis and protein purification steps were performed at 4 °C. The lysate was clarified by centrifugation followed by two rounds of ammonium sulfate precipitation. In the first round of precipitation, 60% ammonium sulfate was used to remove most of the unwanted proteins from the supernate, and 95% ammonium sulfate was subsequently used to precipitate the desired sw Mb. The isolated sw Mb was resuspended and dialyzed in low-salt buffer (100 mM potassium phosphate, 40 mM NaCl, pH 6). The resulting sample was loaded onto a CM-52 cellulose (Whatman) cation exchange column and the protein was eluted using a linear salt gradient from 40 mM to 1.0 M NaCl (in 100 mM potassium phosphate, pH 6.0). Once the protein fraction was collected, excess heme was added and the protein/heme mixture was stirred overnight at 4 °C. As the final purification step, the protein was applied to a G75 gel filtration column (Sigma) using the purification buffer (20 mM Tris-HCl, 1 mM EDTA, pH 7.4) and the ensuing Mb samples were sufficiently pure for crystallization and UV-vis spectroscopic studies.

Reactions of the sw Mb Proteins with Nitroalkanes under Reducing Conditions.

UV-vis monitoring of Mb-nitrosoalkane complex formation.

UV-vis spectra were obtained using an HP 8453 spectrometer. Formation of the sw Mb^II–RNO derivatives was monitored by UV-vis spectroscopy using similar reaction conditions as described by Mansuy and coworkers.¹⁰ First, 5 μL of either sw Mb wt or its H64A mutant (both at 30 mg/mL) were added into a cuvette containing 3 mL of 0.1 M phosphate buffer at pH 7.4 (to reach a final concentration of 3 μM), and the spectrum for the metMb was recorded. Sodium dithionite was subsequently added to a final concentration of 20 mM and the spectrum of the resulting ferrous Mb was recorded. After the reduction was complete, as judged by UV-vis spectroscopy, the ligand precursors MeNO₂ (6.44 μL), EtNO₂ (8.5 μL), PrNO₂ or iPrNO₂ (10.8 μL), all half-diluted in MeOH (Sigma, 99.8%), were added into the reaction mixture for a final concentration of 20 mM. The reaction was then monitored spectroscopically at regular time intervals over a 1 h period. The time course for the formation of each Mb^II–RNO complex was determined by tracking the difference between the absorbance at λ 424 nm and the apparent isosbestic point at λ 460 nm. The resulting ΔA₄₂₄–A₄₆₀ values were then plotted as a function of time.

Ligand dissociation upon oxidation with ferricyanide.

In separate experiments, the ferrous Mb^II–RNO derivatives were generated as described above (e.g., by reducing 30 μL of 20 mg/mL sw Mb with excess solid dithionite followed by the addition of 2 μL of each RNO₂ precursor half-diluted in MeOH. Excess dithionite was removed by desalting the reaction mixture using a 5 mL Sephadex G25 column. A portion of the eluent was transferred into a cuvette containing 3 mL of 100 mM potassium phosphate (pH 7.4 or 6.0), and the formation of the Mb^II–RNO complex was confirmed by UV-vis spectroscopy prior to the addition of 1–6 μL of 30 mg/mL potassium ferricyanide.

Preparation of the Crystalline Ferrous sw Mb-nitrosoalkane Derivatives.

Wild-type ferric sw Mb was crystallized using the hanging drop vapor diffusion method as described by Phillips⁴⁹ and Wang⁴⁷ with slight modifications. The wt sw Mb protein (5 μL, 20 mg/mL) was mixed with an equal volume of well solution containing 2.56–3.20 M (NH₄)₂SO₄, 100 mM Tris-HCl, 1 mm EDTA, at pH 7.4 or 9.0. The resulting mixture was allowed to equilibrate at 20 °C as a hanging drop over the well buffer for ca. 8–10 h before seeding with crushed H64A ferric sw Mb^{47, 48} crystals. Hexagonal-shaped crystals formed over a ~5 d period.

Preparation of crystalline wt sw Mb^II–MeNO and –EtNO via the soaking method.

Suitably sized ferric wt sw Mb hexagonal crystals were looped into a 100 μL drop of cryoprotectant solution (3.1 M (NH₄)₂SO₄, 100 mM Tris-HCl, 1 mm EDTA, 10% glycerol, pH 7.4) and covered with light mineral oil to prevent drying out. Subsequently, 2.5 μL of MeNO₂ or EtNO₂ (half diluted in MeOH) were added into the droplet and allowed to soak for 1–2 h. Thereafter, solid dithionite was added into the droplet one grain at a time until a color change from brown to pink was observed under a microscope. The reaction was then monitored every 30 min by extracting and dissolving a single crystal in a 2 μL drop of 100 mM phosphate buffer at pH 7.4 and recording its UV-vis absorption spectra on a Take3 plate using the Synergy HTX multi-mode reader from BioTek^®. At the completion of the reaction (~1 h), the initial Soret band at 408 nm had fully shifted to 424 nm, indicating that the ferrous wt sw Mb^II–MeNO and –EtNO intermediates had formed. The remaining crystals in the droplet were harvested and flash-frozen in liquid nitrogen prior to data collection.

Preparation of crystalline wt sw Mb^II–PrNO via co-crystallization.

The wt sw Mb^II–PrNO complex was formed anaerobically by incubating wt sw Mb (100 μL; 20 mg/mL) in the purification buffer (20 mM Tris-HCl, 1 mM EDTA, pH 7.4) with PrNO₂ (0.5 μL; half diluted in MeOH) for 1 h, followed by the addition of excess solid dithionite for another hour. Once the reaction was complete, as confirmed by UV-vis spectroscopy (5 μL aliquots into 3 mL of 100 mM of phosphate buffer at pH 7.4), the complex was crystallized as described above for ferric wt sw Mb, with the exception that the crystallization was performed anaerobically. Suitable-sized hexagonal crystals grew in ~1 week and were cryoprotected and flash-frozen in liquid nitrogen prior to data collection.

Preparation of the crystalline H64A sw Mb^II–RNO (R = Me, Et, Pr, iPr) derivatives via co-crystallization.

The four H64A sw Mb^II–RNO complexes were prepared similarly. Each complex was generated anaerobically in solution prior to anaerobic co-crystallization using the batch method. First, aliquots of H64A sw Mb (85–100 μL; 60–70 mg/mL) in 20 mM Tris-HCl, 1 mm EDTA, at pH 7.4 were mixed with the respective nitroalkanes RNO₂ (2–5 μL; half diluted in MeOH) for 15–30 min. Then, aliquots of concentrated dithionite were serially added until the solution changed color from brown to pink, at which point the reaction was allowed to proceed for another 30 min. The solution was deemed ready for crystallization once the formation of the λ 424 nm peak (indicative of H64A sw Mb^II–RNO formation) was confirmed by UV-vis spectroscopy (5 μL aliquots into 3 mL of 100 mM of phosphate buffer at pH 7.4). The batch method was used for co-crystallization as described by Phillips⁴⁹ and Wang.⁴⁷ Briefly, an aliquot (10 μL) of the H64A sw Mb^II–RNO solution was mixed with different ratios of crystallization buffer (3.20 M (NH₄)₂SO₄, 100 mM Tris-HCl, 1 mm EDTA, pH 7.4) to obtain a range of final protein concentrations between 2.3–2.6 M (NH₄)₂SO₄. Hexagonal crystals grew after 2–3 d of anaerobic incubation. The crystals were transferred into a droplet of cryosolution (20 μL, 3.1 M (NH₄)₂SO₄, 100 mM Tris-HCl, 1 mm EDTA, 10% glycerol, pH 7.4) and covered with light mineral oil to minimize exposure to air. The H64A sw Mb^II–RNO crystals were looped and mounted directly onto the goniometer for data collection. Notably, H64A sw Mb^II–RNO crystals that were exposed to air returned to the ferric state (indicated by shifting of the Soret band to λ 408 nm) within 24 h, whereas crystals not exposed to air retained the λ 424 nm band for several weeks.

X-ray Diffraction Data Collection, Processing, Structure Solution, and Refinement.

The diffraction data were collected in-house using a Rigaku MicroMax 007HF microfocus X-ray generator equipped with a set of VariMax HF X-ray optics coupled to a Dectris Pilatus 200K silicon pixel detector. The data were collected at 100 K with Cu Kα radiation (λ = 1.54178 Å) with the generator operating at 40 kV/30 mA.

The diffraction data collected were indexed, integrated, and scaled using HKL3000R. The computed sca files were converted into mtz files using scalepack2mtz (CCP4).⁵⁰ Initial phases were calculated by molecular replacement using PHASER MR (CCP4). The model used for molecular replacement was H64A sw Mb (PDB accession code 5ILE, at a resolution of 1.77 Å)⁴⁷ with the heme, waters, and Fe-bound tolyl ligand removed from the structure. All structure refinements were performed using Refmac5 (CCP4) and the models were rebuilt using COOT.⁵¹ The final structures were validated using MolProbity.⁵² Details for each structure are reported in Table S1 in the Supporting Information.

The final structural figures were created in PyMOL.⁵³ The 2F_o–F_c electron density maps were calculated by Fast Fourier Transform (FFT) as contained in the CCP4 software suite. The resulting map files were viewed in PyMOL. All the F_o–F_c electron density maps were generated similarly. First, the ligands were removed from the active site to generate a new pdb file, which was then refined for 5 cycles in Refmac5. The resulting mtz file was input into FFT to generate the map file which was in turn displayed in PyMOL.

wt sw Mb^II–MeNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.4507 to 0.3171. Heme, MeNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Three sulfate ions and two MeNO ligands were added using COOT. The MeNO bound to the heme Fe was modeled at 100% occupancy, while the MeNO ligand in the Xe1 pocket⁵⁴ was modeled at 50% occupancy. The Met0 residue was omitted from the structure due to a lack of electron density. Two conformations were modeled (each with 50% occupancy) for the sidechains Val21 and Glu109. The final model was refined to a resolution of 1.76 Å, with an R factor of 0.1358 and R_free of 0.1719. The final structure has been deposited with the Protein Data Bank (PDB i.d. 6E02).

wt sw Mb^II–EtNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.4471 to 0.3134. Heme, EtNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Five sulfate ions, one EtNO ligand, and a chloride ion were added using COOT. The EtNO ligand bound to the heme Fe was modeled at 100% occupancy. The Met0 residue was omitted from the structure due to a lack of electron density. Two conformations were modeled (each with 50% occupancy) for the sidechain Val21. The final model was refined to a resolution of 1.76 Å, with an R factor of 0.1599 and R_free of 0.1908. The final structure has been deposited with the Protein Data Bank (PDB i.d. 6E03).

wt sw Mb^II–PrNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.3400 to 0.2855. Heme, PrNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Four sulfate ions, one PrNO ligand, and a Na⁺ cation were added using COOT. The PrNO ligand bound to the heme Fe was modeled at 100% occupancy. The Met0 residue was omitted from the structure due to a lack of electron density. Two conformations were modeled (each with 50% occupancy) for the sidechain Lys133. The final model was refined to a resolution of 2.0 Å, with an R factor of 0.1734 and R_free of 0.2052. The final structure has been deposited with the Protein Data Bank (PDB i.d. 6E04).

H64A sw Mb^II–MeNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.4595 to 0.3233. Heme, MeNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. One sulfate ion and one MeNO ligand were added using COOT. The MeNO bound to the Fe heme was modeled at 100% occupancy. The residues Met0, Gln152, and Gly153 were omitted from the structure due to a lack of electron density. Two conformations were modeled (each with 50% occupancy) for the sidechains Ser35, Val68, and Tyr151. The final model was refined to a resolution of 1.75 Å, with an R factor of 0.1540 and R_free of 0.2002. The final structure has been deposited with the Protein Data Bank (PDB i.d 8F9H).

H64A sw Mb^II–EtNO.

During molecular replacement with PHASER MR (CCP4), two Mb molecules per asymmetric unit were identified. Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.3809 to 0.2939. Heme, EtNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Four sulfate ions and two EtNO ligands were added using COOT. The EtNO ligand in the active site of Chain A was modeled at 100% occupancy, while the EtNO ligand in the active site of Chain B was modeled at 60% occupancy (with H₂O bound at the Fe center at 40% occupancy). No residues were omitted due to a lack of electron density in either Chain A or B. Two conformations were modeled (each with 50% occupancy) for the sidechain of Glu59 in Chain A. The final model was refined to a resolution of 1.80 Å, with an R factor of 0.1740 and R_free of 0.2240. The final structure has been deposited with the Protein Data Bank (PDB i.d. 8F9I).

H64A sw Mb^II–PrNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.2896 to 0.2726. Heme, PrNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Four sulfate ions and one PrNO ligand were added using COOT. The PrNO ligand bound to the heme Fe was modeled at 100% occupancy. The Met0 residue was omitted from the structure due to a lack of electron density. The final model was refined to a resolution of 1.75 Å, with an R factor of 0.1602 and R_free of 0.1884. The final structure has been deposited with the Protein Data Bank (PDB i.d. 8F9J).

H64A sw Mb^II–iPrNO.

Ten initial cycles of restrained refinement were run with Refmac5, and the R factor decreased from 0.3040 to 0.2847. Heme, iPrNO, and water molecules were added to the model based on the F_o–F_c electron density maps in the successive refinement cycles. Three sulfate ions and one iPrNO ligand were added using COOT. The iPrNO ligand bound to the heme Fe was modeled at 100% occupancy. The Met0 residue was omitted from the structure due to a lack of electron density. The final model was refined to a resolution of 1.8 Å, with an R factor of 0.1967 and R_free of 0.2298. The final structure has been deposited with the Protein Data Bank (PDB i.d. 8F9N).

RESULTS AND DISCUSSION

Formation and Stability of the Ferrous Mb^II–RNO derivatives.

The nitrosoalkane derivatives of ferrous wild-type sperm whale Mb (wt sw Mb) were prepared by initial dithionite reduction of the ferric Mb precursors followed by the addition of the nitroalkane reagent RNO₂ (eq 1; R = Me, Et, Pr, iPr). Similar methods have been utilized by Mansuy¹⁰ and us⁵⁵ for the preparation of related ferrous horse heart (hh) Mb nitrosoalkane derivatives.

(1)

UV-vis spectroscopy studies were employed to determine the relationship of ligand sterics to the extent of sw Mb^II–RNO complex formation and the impact of the distal pocket His64 residue on the stability of the products.

The observed changes in the UV-vis spectra for the reactions of wt sw Mb are shown in Figure 4. The formation of the ferrous nitrosomethane complex sw Mb^II–MeNO is illustrated in Figure 4A. Upon dithionite reduction of the ferric precursor sw Mb^III–H₂O, the initial Soret band at λ 408 nm shifts to λ 433 nm indicative of the formation of the ferrous deoxy derivative.

Figure 4.

UV-vis spectral characterization of the reduction of ferric wt sw Mb^III-H₂O by dithionite, followed by the reaction of the resulting ferrous wt sw deoxyMb^II with the nitroalkane (RNO₂) precursors to form the respective ferrous wt sw Mb^II–RNO adducts. A) wt sw Mb^II–MeNO, B) wt sw Mb^II–EtNO, C) wt sw Mb^II–PrNO, and D) wt sw Mb^II–iPrNO. Final reaction conditions: 3 μM wt sw Mb, 0.1 M phosphate buffer (pH 7.4), 20 mM dithionite, 20 mM RNO₂.

Addition of nitromethane to the in situ-generated deoxyMb^II derivative results in a further shift of the Soret band to λ 424 nm; this latter band is assigned to the ferrous nitrosomethane product sw Mb^II–MeNO which was also characterized by X-ray crystallography (see later). Shifts in the Q-band region (Figure 4A, inset) are also consistent with the conversion of the ferric precursor to the ferrous Mb^II–MeNO product via the deoxyMb^II intermediate.

Similar spectral shifts were observed for the generation of the other ferrous sw Mb^II–RNO products (R = Et, Pr, iPr) regardless of alkyl chain length as shown in Figs. 4B–D. The extents of formation of the products as a function of alkyl chain length were plotted as a function of time, and the trends are shown in Figure 5. The apparent extents of formation followed the order MeNO > EtNO > PrNO > iPrNO, with the ferrous sw Mb^II–MeNO product reaching maximum formation in a few minutes (<4 min). The Mb^II–EtNO and Mb^II–PrNO products reached maximum formation in ~12 min, whereas the Mb^II–iPrNO product reached its maximum formation in ~25 min (not shown).

Figure 5.

The extent of formation for each wt sw Mb^II–RNO complex as determined by plotting the difference between the absorbances at λ_max 424 and λ 460 nm as a function of time. Absorbance at λ_max 424 nm is indicative of the sw Mb^II–RNO complex and the absorbance at λ 460 nm corresponds to the apparent isosbestic point.

To help assess the influence of the distal pocket His64 residue on the overall sw Mb^II–RNO complex formation and stability, we probed the analogous reactions with the distal pocket H64A mutant. The aerobic reactions of the mutant H64A ferric sw Mb precursors with dithionite and the respective nitroalkanes were performed similarly to those of the wild-type protein described above, and their progress was monitored by UV-vis spectroscopy. The UV-vis spectra of the ferric H64A sw Mb precursor, the reduced deoxyMb^II intermediate, and the respective ferrous H64A sw Mb^II–RNO products are shown in Figure 6. The spectral changes are generally similar to those of the wild-type derivatives although with absorptivity differences and will not be discussed further.

Figure 6.

UV-vis spectral characterization of the reduction of ferric H64A sw Mb^II–H₂O by dithionite, followed by the reaction of ferrous H64A sw deoxyMb^II with the nitroalkane precursors to form the respective ferrous H64A sw Mb^II–RNO adducts. A) H64A sw Mb^II–MeNO, B) H64A sw Mb^II–EtNO, C) H64A sw Mb^II–PrNO, and D) H64A sw Mb^II–iPrNO. Final reaction conditions: 3 μM H64A sw Mb, 0.1 M phosphate buffer (pH 7.4), 20 mM dithionite, 20 mM RNO₂.

We observed, however, that the reactions involving the distal pocket H64A mutant protein proceeded much more rapidly (Figure 7) than those for the wild-type protein that contained the His64 residue. All four complexes (where R = Me, Et, Pr, iPr) reached their respective formation maxima within ~5 min, indicating that ligand sterics for this set of relatively small nitrosoalkanes did not play a significant role in the apparent rates of Mb^II–RNO complex formation in this distal pocket H64A mutant.

Figure 7.

The extent of formation for each ferrous H64A sw Mb^II–RNO complex as determined by plotting the difference between absorbance at λ_max 424 and λ 460 nm as a function of time. The absorbance at λ_max 424 nm is indicative of ferrous H64A sw Mb^II–RNO complex and absorbance at λ 460 nm corresponds to the apparent isosbestic point.

Interestingly, the overall extents of formation of the H64A Mb^II–RNO complexes prepared in this work followed an inverse correlation with increasing alkyl chain size in the order iPrNO > PrNO > EtNO ≈ MeNO, which is opposite to that observed for the wild-type protein (Figure 5). We hypothesize that this is due to the lack of Fe–RNO stabilization by the now-absent His64 H-bonding interaction, allowing easier escape of the smaller (and less hydrophobic) RNO groups from the distal pocket.

Oxidation of the Ferrous Mb^II–RNO derivatives.

Oxidations of the generated ferrous wild-type and mutant H64A Mb^II–RNO products using potassium hexacyanoferrate(III) (aka. ferricyanide)^{56, 57} resulted in spectral changes in the UV-vis region indicative of reformation of their ferric sw Mb^III–H₂O precursors (Figure 8). The oxidation of the ferrous H64A sw Mb^II–RNO complexes back to their ferric precursors was independent of RNO identity and occurred rapidly (~5 min). On the other hand, the qualitative rates of oxidation of the ferrous wt sw Mb^II–RNO complexes back to their ferric precursors were somewhat dependent on RNO identity; the oxidation reactions of the derivatives with larger RNO groups took shorter times to complete (e.g., within a few minutes), whereas those with smaller RNO groups (e.g., MeNO) took a few hours to complete. These spectral observations for the oxidation reactions are not dissimilar to those obtained for other ferrous heme protein-nitrosoalkane/arene complexes.s.

Figure 8.

UV-vis spectra showing conversion of the ferrous wt sw Mb^II–RNO complexes to their ferric wt sw Mb^III–H₂O precursors upon oxidation by ferricyanide. A) wt sw Mb^II–MeNO, B) wt sw Mb^II–EtNO, C) wt sw Mb^II–PrNO, D) wt sw Mb^II–iPrNO. The ferrous wt sw Mb^II–RNO derivatives were prepared as described in the text, followed by the removal of excess dithionite using a desalting column. Afterwards, an aliquot of the sample was placed in a cuvette containing 3 mL of 0.1 M phosphate buffer (pH 6.0), followed by the addition of 3–6 μL of 30 mg/mL potassium ferricyanide.

We have not, to date, been successful at isolating any of the oxidized products Mb^III–RNO. Experimental and computational investigations have determined the molecular orbital origins of the decreased retention of nitrosoarene ligands on ferric heme centers.³²

Crystallographic Studies.

Crystals of the ferrous sw Mb^II–RNO products were obtained either by (i) the soaking method starting with crystals of the ferric Mb^III–H₂O precursor and “soaking (s)” with dithionite and the nitroalkane reagent, or (ii) “cocrystallization (c)” where the reactions were carried out in solution and the product crystallized. The crystal morphologies of the derivatives used in the crystallographic studies for this work are of two main types shown in Figure 9. The crystalline ferrous wt sw Mb^II–MeNO and –EtNO derivatives were obtained by ligand soaking using pre-formed brown hexagonal crystals of ferric sw Mb^III–H₂O indexed with the P6 space group (Figure 9A). The crystal soaking was carried out aerobically, and the reaction was deemed complete once the crystals changed color from brown to violet-pink (Figure 9B). The ferrous wt sw Mb^II–PrNO complex was generated anaerobically in solution and then co-crystallized using the hanging drop diffusion method to yield hexagonal crystals also belonging to the P6 space group (Figure 9B).

Figure 9.

Shapes and colors of the crystals obtained in this study by the soaking (s) and co-crystallization (c) methods. A) A brown hexagonal-shaped crystal of ferric sw Mb^III–H₂O indexed with the P6 space group. B) Representative morphology of the violet-pink, ferrous sw Mb^II–RNO hexagonal-shaped crystals indexed with the P6 space group (H64A Mb^II–iPrNO shown here). C) Representative violet-pink thin plate crystals of ferrous sw Mb^II–RNO obtained by the cocrystallization method and indexed with the P12₁1 space group (H64A Mb^II–MeNO shown here).

Crystals of the mutant H64A sw Mb^II–RNO derivatives were formed anaerobically in solution and then cocrystallized using the batch method. This resulted in violet-pink hexagonal crystals (P6 space group) for the ferrous H64A sw Mb^II–PrNO and Mb^II–iPrNO derivatives, and in violet-pink thin plates (P12₁1 space group) for the ferrous H64A sw Mb^II–MeNO and Mb^II–EtNO products. We observed that crystals of the H64A derivatives were sensitive to air. For example, whereas crystals of H64A sw Mb^II–EtNO retained their violet-pink color for several weeks when kept under a nitrogen atmosphere, they turned brown within a day when exposed to air, suggestive of oxidative degradation. Indeed, the UV-vis spectrum of a crystal of H64A sw Mb^II–EtNO (λ_max 424 nm) kept anaerobically for ~1 month remained unchanged, whereas the UV-vis spectrum of crystals of the same batch but exposed to air for ~1 d showed full conversion to the ferric H64A sw Mb^III–H₂O precursor (λ_max 408 nm). Similar results were obtained for the other H64A sw Mb^II–RNO compounds prepared in this work (Figure S1 in the Supporting Information).

Crystal Structures of the Ferrous Wild-Type Mb^II–RNO (R = Me, Et, Pr) Complexes.

The 1.76 Å-resolution crystal structure of the ferrous wild-type sw Mb^II–MeNO complex is shown in Figure 10A. The F_o–F_c electron density omit map (right of Figure 10A) shows two V-shaped electron density regions corresponding to (a) a MeNO ligand that is N-bound to the heme Fe center and refined at full occupancy, and (b) a MeNO (alternatively the formaldoxime/CH₂=NOH tautomer) molecule in a region referred to as the Xe1 pocket and refined at 50% occupancy. The axial Fe–N(MeNO) distance is 1.9 Å. The C–N=O plane of the bound MeNO ligand is oriented between adjacent porphyrin N(pyrrole) atoms, with a (por)N–Fe–N–O(RNO) torsion angle of 20°. The bound MeNO ligand engages in a H-bond interaction through its nitroso O-atom with the distal pocket His64 residue, with a 2.3 Å distance between the MeN(O) atom and the His64(Nε) atom. The methyl group is oriented towards the hydrophobic interior of the distal pocket, with the closest neighboring residue atom (Cγ² of Val68) at a distance of 3.6 Å from the C-atom of MeNO. Selected heme site geometrical parameters of this and other Mb^II–RNO crystal structures determined in this work are collected in Table S2 in the Supporting Information. A second V-shaped electron density in the F_o–F_c map is assigned to a MeNO molecule, or more likely the formaldoxime CH₂=NOH tautomer,^{58, 59} at 50% occupancy in the Xe1 pocket.

Figure 10.

Final active site models of wt sw Mb^II bound to (A) MeNO, (B) EtNO, and (C) PrNO. The left panels represent the 2F_o–F_c electron density maps (blue mesh) contoured at 1σ. The right panels show the F_o–F_c omit electron density maps (green mesh) contoured at 3σ. Heme-bound ligands are at 100% occupancy. The MeNO/CH₂=NOH molecule located in the Xe1 pocket of the sw Mb^II–MeNO structure is modeled at 50% occupancy.

A comparison of the sw Mb^II–MeNO structure determined here and the horse heart (hh) analog⁶⁰ show no significant differences in ligand coordination within the active site (Figure S2 in the Supporting Information), but with minor positional shifts for the active site Phe43 residue, and significant differences in the folds of the backbones of the GH-loop region (not shown). The imidazole of His64 in the hh structure is located closer to the exterior solvent, and farther from the MeNO ligand, than what is observed in the sw structure; the apparent distance between their His64(N^ε) atoms is measured at 1.1 Å when the structures are aligned along the Cα chain (or 0.8 Å with alignment along the heme pyrrole atoms). This is consistent with the stronger H-bond interaction observed between the MeN(O) and the distal His64 residue in the sw structure.

The 1.76 Å resolution structure of the nitrosoethane derivative, namely sw Mb^II–EtNO is shown in Figure 10B. Similar to the MeNO complex, the ligand is N-bound to the Fe center (Fe–N = 2.1 Å) and was refined with full occupancy. The nitroso moiety is located in a direction between adjacent porphyrin N-atoms, with a (por)N–Fe–N–O(RNO) torsion angle of 37°. The O-atom of the EtNO ligand is H-bonded to the His64 residue, with an EtN(O)…(Nε)His64 distance of 2.1 Å. The ethyl group is similarly oriented, as with the nitrosomethane analog, towards the hydrophobic interior of the protein, with the closest distances between the alkyl chain and the distal pocket resides being between the Ile107 and the terminal C² atom of EtNO (at 3.0 Å), and between Val68 and both the ligand C¹ (at 3.5 Å) and C² (at 3.6 Å) atoms. The alkyl backbone of the EtNO ligand in the sw structure adopts a trans conformation with a ∠C²–C¹–N–O torsion angle of 152°. For comparison, the torsion angle in the hh derivative reported earlier was –41°, and the nitroso O-atom in the hh derivative displayed a weaker H-bond interaction with His64 (EtN(O)…(Nε)His64 distance of 2.7 Å).¹¹ Figure S2 shows the comparison of the sw and hh Mb^II–EtNO structures for both the alignment along the Cα chain and using the heme pyrrole atoms. Similar shifts in the His64 and Phe43 residues were observed as seen in the analogous MeNO-bound structures.

The 2.0 Å resolution structure of the nitrosopropane derivative, namely sw Mb^II–PrNO, is shown in Figure 10C, with the PrNO ligand modeled at full occupancy. In this structure, the PrN(O)…(Nε)His64 H-bonding distance is 2.5 Å, and unlike the structures described above, the NO moiety is closer to eclipsing a heme N-atom, with a (por)N–Fe–N–O(RNO) torsion angle of 20°. The O=N–C¹–C² fragment displays essentially a cis conformation. As observed with the MeNO and EtNO derivatives, the propyl group is also oriented towards the hydrophobic interior of the active site. The closest contacts of the propyl-C¹ atom with distal pocket residues are with Val68(Cγ²) at 3.5 Å, and Val68 (Cδ1) at 3.7 Å. The closest contact of the propyl-C² atom is with Leu29(Cδ²) at 3.4 Å. The closest contacts of the propyl-C³ atom are with Ile107, Phe43, Leu32, and Leu29 at distances of ~3.4–3.7 Å (from their nearest C-atoms). There were no significant changes observed in the conformation of the PrNO ligand over several sample preparations and subsequent crystal structure determinations.

Despite multiple attempts, we were unsuccessful in obtaining suitable crystals of the wt sw Mb^II–iPrNO complex for a crystal structure determination (see later).

Crystal Structures of the Distal Pocket H64A Ferrous sw Mb^II–RNO Nitrosoalkane Complexes and Comparisons with Their Wild-Type Analogues.

As mentioned in the Introduction we sought to determine the relationship between nitrosoalkane sterics and binding orientations in the wild-type protein, and assess the influence of the distal pocket residue His64 on these ligand orientations. Specifically, we were interested in determining if the H-bond provided by the His64 residue in the wild-type protein was the sole factor that controlled the orientations of these nitrosoalkane ligands of varying chain lengths. To do so, we prepared and crystallized the distal pocket H64A mutant protein derivatives and determined their crystal structures. The 2F_o–F_c and F_o–F_c electron density maps and final models of their heme active sites are shown in Figure 11.

Figure 11.

Final active site models of H64A sw Mb^II bound to (A) MeNO, (B) EtNO (Top: Chain A, Bottom: Chain B), (C) PrNO. (D) i-PrNO. The left panels represent the 2F_o–F_c electron density maps (blue mesh) contoured at 1σ. The right panels show the F_o–F_c omit electron density maps (green mesh) contoured at 3σ. Most RNOs are heme bound at 100% occupancy; the exception is EtNO in Chain B, which is bound at 60% occupancy.

The 1.75 Å resolution crystal structure of the H64A sw Mb^II–MeNO complex is shown in Figure 11A. As with the wt derivative, the MeNO ligand, modeled at full occupancy, is N-bound to the Fe center with a Fe–N(MeNO) distance of 1.9 Å. The overall alignment of the H64A and wild-type structures along their Cα chains resulted in an RMSD value of 0.467 Å, with the major structural difference at the tertiary level located in the GH-loop area (not shown). Despite the absence of the His64 ligand, the nitroso O-atom is similarly, as with the wild-type protein, oriented towards the general solvent exterior of the protein; a comparison of the heme–MeNO fragments of the wild-type and H64A derivatives are shown in Figure 12A and Figure S3A. There are three notable differences between the H64A and wild-type Mb^II–MeNO structures. First, in the absence of the His64A ligand in the wild-type protein, the nitroso O-atom is observed to be H-bonded with a fixed H₂O molecule, with a (MeN)O…O(water) distance of 2.7 Å (Figure 11A).

Figure 12.

Comparison of the active sites of wt and H64A sw Mb^II–RNO structures by alignment along the Cα chain. A) Cyan wt sw Mb^II–MeNO, gray H64A sw Mb^II–MeNO; B) magenta wt sw Mb^II–EtNO, yellow wt sw Mb^II–PrNO; C) orange H64A sw Mb^II–EtNO Chain A, green H64A sw Mb^II–EtNO Chain B; D) blue H64A sw Mb^II–PrNO, wheat H64A sw Mb^II–iPrNO. The related alignments by pairwise fitting of the 24-atoms of the heme planes are shown in Figure S3 in the Supporting Information.

Second, the distal pocket Val68 residue in the wild-type derivative adopts a conformation where the Val68-methyl groups point towards the MeNO ligand. On the other hand, the H64A derivative Val68 adopts two conformations. The closest distance between the C-atom of MeNO and the nearest Cγ² of Val68 are 3.6 and 3.2 Å in the wild-type and H64A respectively. The next closest distances between the bound MeNO ligand and distal pocket residues in the H64A derivative are between the C-atom of MeNO and Phe43(Cζ) and Leu29(Cδ²) at 3.8 and 4.0 Å, respectively.

A third notable difference is that we did not observe a second MeNO/CH₂=NOH molecule in the Xe1 pocket as we did in the wild-type structure (c.f. Figure 10A). However, we note that crystals of the wild-type derivative were grown from the soaking method, whereas those for the H64A derivative were grown from co-crystallization, representing two distinct crystallization methods.

Unlike the others, the 1.80 Å resolution crystal structure of the nitrosoethane H64A sw Mb^II–EtNO complex contained two distinct molecules in the asymmetric unit, and we refer to them as Chains A and B (Figure 11B). In Chain A, the N-bound EtNO ligand was modeled at full occupancy. In contrast with that observed for the H64A Mb^II–MeNO derivative described above, the nitroso O-atom does not engage in a H-bond with a fixed water molecule (the closest fixed H₂O molecule is ~7 Å away), and this observation is consistent with the nitroso-(NO) moiety being less clearly directed towards the solvent exterior. In Chain A, the nitrosoethane C²–C¹–N=O backbone displays a torsion angle of –72°. The closest distances between the EtNO ligand and distal pocket residues are between the EtNO terminal C²-atom and Leu29(Cδ²) (at 3.7 Å) and Phe43(Cζ) at 3.8 Å. A comparison of the H64A sw Mb^II–EtNO Chain A structure with that of the wild-type protein, shown in Figure 13 (left) and Figure S3B, reveals differences in the general orientation of the ethyl group of EtNO. In the wild-type structure, the ethyl group is located in a small groove created by residues Val68 and Ile107. In the H64A sw Mb^II–EtNO Chain A structure, the ethyl group points away from Val68 towards a small tunnel located near Phe43. In addition, the orientations of the methyl groups of Val68 differ in these two structures, pointing towards the ligand in the wild-type but away from the ligand in the H64A Chain A structure.

Figure 13.

Left: (A, left) Superimposed models comparing EtNO binding in wt and H64A sw Mb^II–EtNO in Chain A. (B & C, left) Surface representation showing the vicinity of EtNO in (B, left) wt Mb^II–EtNO and (C, left) H64A Mb^II–EtNO in Chain A. Right: (A, right) Superimposed models comparing PrNO binding in wt and H64A sw Mb. Surface representation showing the vicinity of PrNO in (B, right) wt Mb^II–PrNO and (C, right) H64A Mb^II–PrNO. In both panels, the surface of His/Ala64 and Arg45 are omitted for clarity of view. The black mesh represents the F_o–F_c omit electron density maps contoured at 3σ. Each figure is slightly rotated to highlight the different cavities occupied by the alkyl groups. Alignments shown here were made along the Cα chain.

Chain B of the H64A Mb^II–EtNO structure reveals significant differences in EtNO binding to the heme site (Figs. 11B and 12C). The occupancy of the N-bound EtNO ligand refined to 60% occupancy with an Fe–N(EtNO) distance of 1.9 Å, with the additional presence of a H₂O molecule at 40% in this general location with an Fe–OH₂ distance of 2.5 Å. An unexpected observation was that the nitroso O-atom of this relatively small ligand was directed towards the hydrophobic protein interior with no discerned H-bond interactions. The nitrosoethane C²–C¹–N=O torsion angle is 64°, and the ethyl moiety was oriented towards the hydrophilic protein exterior with the closest distance to it being with the C²-atom of the Ala64 residue (at 3.7 Å). X-ray crystal structures of native sw Mb with bound propyl-NC and butyl-NC ligands displayed similar orientations of their hydrophobic alkyl chains towards the exterior solvent.⁴⁵ A similar orientation of a ligand hydrophobic chain towards the protein exterior was documented previously in the related nitrosoamphetamine Mb–AmphNO complex containing a relatively large hydrocarbon chain.²⁹

A comparison of the structures of the nitrosopropane complexes of wild-type (Figure 10C) and H64A Mb^II–PrNO (Figure 11C), with the PrNO ligand modeled at full occupancy in both structures, show differences in ligand conformations (Figure 13 (right), and Figure S3C). In the H64A structure (i.e., in the absence of the His64 residue), the nitroso –NO moiety is significantly rotated in a direction away from the solvent exterior (Figs. 12B and 12D), in a location approximately bisecting adjacent to the porphyrin N²² and N²³ atoms. As shown in Figure 11C, the nitroso O-atom is linked to the solvent exterior by a chain of H-bonded H₂O molecules; the nitrosopropane C²–C¹–N=O and C³–C²–C¹–N torsion angles are 68° and 66°, respectively. The nearest residues to the propyl group are Leu29 and Phe43, with measured distances from Leu29(C³) to PrNO(C³) at 3.6 Å, and from Phe43(Cζ) to PrNO(C²) (at 3.7 Å) and PrNO(C³) at 3.9 Å. The distances from Val68(Cγ²) to the PrNO(C²) and PrNO(C³) atoms are 4.1 Å and 4.4 Å, respectively.

We analyzed the distal pocket structures to ascertain why the nitrosopropane (PrNO) ligand only adopted one orientation in the H64A mutant derivative over several crystallizations, whereas the related nitrosoethane (EtNO) ligand exhibited more than one orientation (i.e., in Chains A and B). Analysis of the protein interior surface within the active site reveals that the propyl group is buried inside a small hydrophobic channel located in the back of the active site (right of Figure 13) in both the H64A and wild-type structures, despite the PrNO ligand displaying different geometric arrangements for the –CNO moieties. In contrast, the ethyl group of the EtNO ligand is too short to be trapped in this channel (left of Figure 13).

Although we were not successful in crystallizing the wild-type sw Mb^II–iPrNO complex despite its formation in solution (Figure 4D), we were successful in crystallizing and obtaining the crystal structure of the H64A Mb^II–iPrNO derivative; its 1.75 Å resolution crystal structure is shown in Figure 11D. The iPrNO ligand was modeled at full occupancy, with the nitroso –NO moiety in a similar orientation away from the solvent exterior as that observed for the PrNO derivative (Figure 12D). The nitroso O-atom is H-bonded to a single fixed H₂O molecule (at 2.6 Å); the nearest other fixed H₂O molecule is 4.3 Å away (not shown). The closest distances of the isopropyl group to distal pocket residues are between the iPrNO(C²) atom and Phe43(Cζ) at 3.4 Å, and with Phe43(Cε¹) at 3.0 Å. The next closest distances are with this iPrNO(C²) atom and Ile107 and Leu29 (at ~3.8 Å).

We then sought to determine the reasons for our inability to crystallize the wild-type Mb^II–iPrNO derivative. In silico docking of the iPrNO ligand into the active site of the sw Mb protein (PDB ID 2MBW), with the nitrosopropane –CNO group in an orientation capable of H-bonding to the His64 residue, revealed steric clashes between the methyl groups of the iPr group and Val68 and Phe43 (see Figure S4 in the Supporting Information). The orientation we selected was also consistent in a heme model derivative (PPDME)Fe(iPrNO)(5-MeIm) that was independently prepared and its structure solved (Figure 14). The nitroso –NO moiety is found in an orientation pointing towards the carboxylate ester groups, similar to what we observed in our wild-type Mb^II–RNO structures described above.

Figure 14.

Molecular structure of the (PPDME)Fe(iPrNO)(5-MeIm) structure. Thermal ellipsoids are drawn at 40% probability. Hydrogen atoms are not shown for clarity.

In this hypothetical wild-type sw Mb^II–iPrNO structure, the distance between iPrNO(C²) and Phe43(Cζ) was calculated at 2.5 Å, and the distance from the iPrNO(C²) and Val68(Cγ¹) was calculated at 2.8 Å; these distances between the iPrNO ligand and distal pocket residues are shorter than those we observed for the other Mb^II–RNO structures reported in this work. A comparison with the structure of the H64A derivative shows that the ligand was able to overcome this problem due to the lack of the H-bonding His64 residue that could limit its orientation. Indeed, as shown in Figure 9D for the H64A Mb^II–iPrNO structure, the –N=O moiety of the iPrNO ligand is rotated away from the solvent (where the His64 residue would be in the wild-type protein), allowing one of the iPr-methyl groups to be angled towards the hydrophobic tunnel located by residue Phe43; the other methyl group is oriented in a general direction towards the solvent exterior of the protein. The net effect is an iPrNO ligand orientation preference that allows a long-lived crystallized H64A Mb^II–iPrNO adduct.

CONCLUSIONS

We report the preparation of a representative set of ferrous wild-type and H64A sw Mb^II–RNO derivatives (R = methyl, ethyl, propyl, isopropyl) from the reactions of the wt and H64A Mb^III–H₂O precursors with dithionite and the respective organic nitroalkanes RNO₂. The Mb^II–RNO derivatives were characterized by UV-vis spectroscopy and display Soret bands at λ_max 424 nm, suggestive of the ligands being bound to Fe via the N-liganded mode. For the wt derivatives, the apparent extent of formation of these derivatives followed the order MeNO > EtNO > PrNO > iPrNO. We observed faster reactions with the H64A mutant, with maximal formation of the Mb^II–RNO derivatives occurring within a 5 min period. However, the extent of formation of the H64A Mb^II–RNO derivatives was observed to be opposite to that of the wild-type, namely MeNO = EtNO < PrNO < iPrNO. We attribute this latter trend to the lack of Fe–RNO stabilization by the now-absent His64 H-bonding interaction, allowing easier escape of the smaller (and less hydrophobic) RNO groups from the distal pocket. Oxidation of the Mb^II–RNO derivatives with ferricyanide resulted in the formation of the ferric Mb^III–H₂O precursors with loss of the RNO ligands, consistent with our earlier rationalization of a weaker interaction of RNO ligands with ferric heme centers.³²

X-ray crystallographic characterization of three of the wt Mb^II–RNO derivatives (R = Me, Et, Pr) revealed both the N-binding modes of the nitrosoalkane ligands to the ferrous centers and the presence of H-bonding interactions between the bound RNO ligands (via the O-atom) with the His64 residue. With this H-bonding interaction, the nitroso O-atoms were pointed in the general direction of the protein exterior, and the hydrophobic R groups were pointed in the general direction of the protein interior. For the H64A mutant derivatives, we were able to crystallize and solve the structures for all four Mb^II–RNO derivatives (R = Me, Et, Pr, iPr). The H64A Mb^II–MeNO and H64A Mb^II–EtNO derivatives retained crystal structures similar to those of the wt structures, with the nitroso O-atoms pointing towards the solvent exterior. Alternate RNO orientations were determined for the EtNO and PrNO derivatives, however, where the nitroso O-atoms were not directly pointing towards the solvent exterior. An analysis of the distal pocket amino acid surface landscape, including identification of hydrophobic channels, helped provide an explanation for the differences in ligand orientations adopted by the EtNO and PrNO ligands in their wt and H64A Mb^II–RNO structures.

Importantly, this current work complements very elegant work by Olson, Phillips, and coworkers on the related alkylisocyanide complexes sw Mb^II–CNR (R = Me, Et, Pr, Bu) whose X-ray crystal structures were determined for both native and wild-type sw Mb.⁴⁵ Comparisons of the crystal structures of our wt sw Mb^II–RNO compounds with those of the wt sw Mb^II–CNR compounds, all crystallizing in the P6 space group, reveal that the hydrophobic alkyl groups point toward the interior of the protein, with only minor differences in the conformations of the distal pocket residues in these structures.

As mentioned in the Introduction, only two other X-ray crystal structures of Mb in complex with nitroalkanes had been reported prior to this work, and these were with horse heart Mb. Further, only a small number of other heme protein–RNO or –ArNO crystal structures have been reported. Our seven Mb–RNO crystal structures reported provide a more systematic structural analysis of the role of distal pocket His64 residue in directing the conformation and orientation of RNO ligands within the Mb active site, and will thus hopefully provide an additional guide for RNO binding to related heme proteins with relatively small distal active-site pockets.