Identification of Protein-Bound Dinitrosyl Iron Complexes by Nuclear Resonance Vibrational Spectroscopy

Among various sites of nitric oxide reactivity in biology,¹ iron-sulfur clusters have emerged as an important class of metallocofactors targeted by NO.² Proteins containing these clusters perform a diverse array of functions from electron transfer to small molecule sensing and substrate binding.³ Consequently, iron-sulfur clusters are potential sites for both NO-derived physiological signal transduction⁴ and toxicity.^2a,5

The primary products of NO reactivity at iron-sulfur clusters are dinitrosyl iron complexes (DNICs).⁶ Identification of DNICs in proteins and tissue samples has traditionally relied upon the diagnostic g = 2.03 EPR signal,⁷ which is characteristic of these S = ½ species. Because iron dinitrosyls can exist in forms other than the paramagnetic mononuclear species (Chart 1), however, EPR characterization is not definitive.⁸ Other means of characterization include ⁵⁷Fe Mössbauer,⁹ UV-vis,¹⁰ Raman,¹¹ and X-ray absorption spectroscopy,¹² but each of these techniques lacks diagnostic spectroscopic markers and cannot easily deconvolute spectra resulting from a mixture of species.

Chart 1.

Examples of sulfur-ligated iron dinitrosyls.

In order to overcome these challenges, we have utilized ⁵⁷Fe nuclear resonance vibrational spectroscopy (NRVS) to characterize the core vibrations of dinitrosyl iron complexes. This technique is selective for motion of the ⁵⁷Fe nucleus, making it an ideal means of probing DNICs in protein environments.¹³ NRVS is not limited by the selection rules of infrared and Raman spectroscopy, ensuring that a complete set of vibrational modes involving motion of the ⁵⁷Fe nucleus is observed. Here we present the first NRVS results for a nitrosylated iron-sulfur cluster protein, as well as complimentary data on several biomimetic model compounds of dinitrosyl iron complexes.

In order to provide spectroscopic benchmarks to identify dinitrosyl iron complexes in proteins, we investigated synthetic analogs comprising the four structures displayed in Chart 1. To provide maximum NRVS signal intensity, the model compounds were prepared from ⁵⁷Fe-enriched Fe(OTf)₂·2CH₃CN (see Supporting Information, SI, for experimental details). The benzene thiolate ligand (Chart 1, R = Ph) was chosen because it facilitates the synthesis and isolation of the three different ⁵⁷Fe-labeled derivatives and because the NRVS spectrum of the related Fe-S cluster, [Fe₄S₄(SPh)₄]²⁻, was recorded previously.¹⁴ Additionally, (Et₄N)[Fe(NO)₂(SPh)₂], [Fe₂(μ-SPh)₂(NO)₄] (Roussin's red ester, RRE), and (Et₄N)[Fe₄S₃(NO)₇] (Roussin's black salt, RBS) were prepared using both ¹⁴NO and ¹⁵NO to help identify normal modes involving the nitrosyl ligands.

A representative NRVS spectrum of one of the model compounds, (Et₄N)[Fe(^14/15NO)₂(SPh)₂], is shown in Figure 1. Of note are the intense peaks above 500 cm⁻¹, which shift to lower energy upon ¹⁵N substitution. The NRVS spectra of [Fe(μ-SPh)₂(^14/15NO)₄] and (Et₄N)[Fe₄S₃(^14/15NO)₇] display similarly shifted peaks (Figure S14), indicating that vibrations in this region most likely originate from the symmetric and asymmetric stretching modes of the N–Fe–N unit.¹¹ In all four model compounds, the peak near 600 cm⁻¹ appears to contain a higher energy shoulder (e.g., Figure 1). We assign this feature to Fermi resonance based on the calculated spectrum for each model compound (see SI) and the disappearance of these peaks upon ¹⁵N substitution (Figure 1). Despite these minor complexities, the region of the spectrum between 500 and 700 cm⁻¹ provides important diagnostic NRVS signatures for each of the iron dinitrosyl complexes displayed in Chart 1 (Figure S20). Furthermore, the vibrations occur in a region of the spectrum that does not overlap with normal modes due to thiolate ligands or the {Fe₄S₄} core, thereby providing clear information about the nature of the Fe-NO species present. EPR spectroscopy fails in this regard because the RRE and RBS compounds are diamagnetic, and ⁵⁷Fe Mössbauer spectroscopy is complicated by the fact that isomer shifts for all four dinitrosyl-containing species are very similar (Figure S25 and Table S2).

Figure 1.

⁵⁷Fe partial vibrational density of states for (Et₄N)[Fe(NO)₂(SPh)₂].

For protein studies, a ferredoxin D14C variant from Pyroccocus furiosus was selected as a model system because of its stability at high concentrations (mM) and relatively solvent exposed Fe₄S₄ cluster (Figure S26).¹⁵ In the native form of the protein, the iron-sulfur cluster is bound by one aspartate (D14) and three cysteinate residues. For the present purposes, a D14C mutant was examined because it provides a more representative model of Fe₄S₄ clusters, which typically have four terminal cysteine thiolate ligands. Moreover, NRVS data on the untreated D14C mutant were already available,¹⁶ providing an opportunity to distinguish vibrations of the unmodified protein from those of the NO reaction product(s) (Figure S28).

Reaction of P. furiosus ferredoxin with the nitric oxide donor PAPA NONOate gave rise to a single g_av = 2.03 EPR signal characteristic of a protein-bound DNIC (Figure S27). EPR spectroscopy provided no evidence for formation of rRRE (g_av = 1.99),¹⁷ and spin quantitation of the DNIC signal demonstrated that this species only accounted for about 5 – 10% of the total protein concentration. Examination of the reaction mixture by NRVS, however, indicated significant formation of a dinitrosyl iron complex (Figure 2 and Figure S28). The spectral features of this species were inconsistent with either the DNIC or the diamagnetic RRE. By comparison with the model compounds, we assign this product as the tetranuclear Roussin's black salt, RBS ([Fe₄S₃(NO)₇]⁻, Chart 1).

Figure 2.

Comparison of the NRVS spectrum of RBS and the spectrum resulting from treatment of ⁵⁷Fe-enriched P. furiosus Fd D14C with PAPA NONOate.

Formation of RBS by nitrosylation of a protein-bound Fe₄S₄ cluster has not been previously reported, although reactions of a variety of synthetic Fe₄S₄ clusters with NO gas generate the black salt.¹⁸ RBS formation may be related to the highly solvent-exposed nature of the Fe₄S₄ cluster in P. furiosus Fd, but further work is required to test this possibility as well as its potential generality. The formation of RBS, however, could help explain the results obtained in reactions of with other Fe₄S₄ cluster proteins with NO, where EPR spin quantitation of protein-bound DNICs accounts for only a fraction of the total iron content.¹⁹

Roussin's black salt is toxic to cells^2a,20 and its formation from Fe-S clusters is a plausible mode of NO-derived toxicity in vivo. Our data provide the first evidence for the direct formation of this species from an iron-sulfur protein. The presence of RBS in addition to DNICs may have important implications for the repair of nitrosylated iron-sulfur clusters,²¹ the mechanism of NO signal transduction,⁴ and other biological consequences of nitric oxide chemistry at iron-sulfur protein cores.²

In conclusion, we provide here the first NRVS spectra of non-heme iron nitrosyl compounds and demonstrate the value of this technique in identifying the products of iron-sulfur cluster nitrosylation. In particular, the method has proved valuable in identifying diamagnetic dinitrosyl iron species that would otherwise be very difficult to observe. Moving forward, we envision extending these studies to identify the products of other biological NO targets and to help define the roles of nitric oxide in modulating the functions non-heme iron proteins.