Covalent Attachment of the Heme to Synechococcus Hemoglobin Alters its Reactivity Toward Nitric Oxide

Abstract

The cyanobacterium Synechococcus sp. PCC 7002 produces a monomeric hemoglobin (GlbN) implicated in the detoxification of reactive nitrogen and oxygen species. GlbN contains a b heme, which can be modified under certain reducing conditions. The modified protein (GlbN-A) has one heme–histidine C–N linkage similar to the C–S linkage of cytochrome c. No clear functional role has been assigned to this modification. Here, optical absorbance and NMR spectroscopies were used to compare the reactivity of GlbN and GlbN-A toward nitric oxide (NO). Both forms of the protein are capable of NO dioxygenase activity and both undergo heme bleaching after multiple NO challenges. GlbN and GlbN-A bind NO in the ferric state and form diamagnetic complexes (Fe^III–NO) that resist reductive nitrosylation to the paramagnetic Fe^II–NO forms. Dithionite reduction of Fe^III–NO GlbN and GlbN-A, however, resulted in distinct outcomes. Whereas GlbN-A rapidly formed the expected Fe^II–NO complex, NO binding to Fe^II GlbN caused immediate heme loss and, remarkably, was followed by slow heme rebinding and HNO (nitrosyl hydride) production. Additionally, combining Fe^III GlbN, ¹⁵N-labeled nitrite, and excess dithionite resulted in the formation of Fe^II–H¹⁵NO GlbN. Dithionite-mediated HNO production was also observed for the related GlbN from Synechocystis sp. PCC 6803. Although ferrous GlbN-A appeared capable of trapping preformed HNO, the histidine–heme post-translational modification extinguished the NO reduction chemistry associated with GlbN. Overall, the results suggest a role for the covalent modification in Fe^II GlbNs: protection from NO-mediated heme loss and prevention of HNO formation.

1. Introduction

Nitric oxide (NO) is a reactive molecule and a potential actor in a variety of cellular processes [1, 2]. As such, its concentration must be carefully regulated under a wide range of physiologic conditions. In plants and many microorganisms, NO can be produced reductively as an intermediate along the nitrate assimilation or denitrification pathways [3, 4]. Other reductive routes as well as oxidative processes [5] can destroy NO. Heme proteins such as hemoglobins (Hbs) are a priori able to switch between source and sink activities and therefore warrant investigation as particularly responsive regulators of free NO concentration. High levels of reactive oxygen and nitrogen species, along with a connection between NO metabolism and release of the potent green-house gas nitrous oxide (N₂O), focus attention on the hemoglobins of abundant photosynthetic unicellular organisms.

The model cyanobacterium Synechococcus sp. PCC 7002 (Synechococcus hereafter) contains a gene (glbN) encoding a hemoglobin (GlbN) of the group 1 truncated lineage (TrHb1) [6]. As purified after recombinant expression in Escherichia coli, apoGlbN binds a b heme (iron-protoporphyrin IX) with two axial histidines, His46 and His70, in both the ferric (Fe^III) (Figure S1A) and ferrous (Fe^II) heme oxidation states [7]. In the presence of exogenous ligands such as CO and CN⁻, His46 (the “distal” heme ligand) can be reversibly displaced from the ferrous or ferric iron (respectively) whereas His70 (the “proximal” heme ligand) remains coordinated [6, 8] (Figure S1A).

Along with ferrous Synechocystis sp. PCC 6803 GlbN (Synechocystis GlbN hereafter), ferrous Synechococcus GlbN is extraordinary among hemoglobins (Hbs) for undergoing a spontaneous and irreversible posttranslational modification (PTM) with the heme group. Specifically, the GlbN PTM consists in the saturation of the 2-vinyl substituent, covalently attaching the heme to His117 Nε2 [9] (Figure S1B). The PTM is facile in that it can be engineered to occur at the 4-vinyl (Leu79His/His117Ala variant of Synechocystis GlbN and Leu75His variant of Chlamydomonas eugametos LI637 CtrHb) and even both the 2- and 4-vinyls (Leu79His variant of Synechocystis GlbN) [10, 11]. Many cyanobacterial TrHb1s have strong sequence identity around a highly conserved His117 and are suspected to undergo the PTM as well. The structural [8] and dynamic [7] properties of wild-type (WT) Synechococcus GlbN and GlbN with PTM (hereafter GlbN-A) are similar. GlbN-A, incapable of heme loss, has increased thermodynamic stability [12], but other consequences of histidine-heme cross-linking have not been investigated.

Mechanistic studies support that the ferrous heme PTM occurs via an electrophilic addition [13, 14]. Starting with apoGlbN reconstituted with ferric heme, the reaction is inhibited if reduction is performed in the presence of CO, presumably because this small ligand binds to ferrous GlbN rapidly and withdraws electron density from the porphyrin π-system through back-bonding [14]. Dioxygen is expected to have a deactivating effect as well through its propensity for superoxide formation (Fe^III−O₂⁻) when bound to ferrous heme [15].

The biological function of GlbN and its modified form has not yet been ascertained. Several studies of the transcriptome of Synechococcus cells have detected constitutive expression of glbN [16–18]. When overexpressed under microoxic conditions, the protein contains covalently attached heme whereas oxic conditions appear to favor unmodified GlbN [16]. Thus, both GlbN and GlbN-A are candidates for physiological relevance, and the PTM may act as a sensor for cellular redox and oxygen status. Elevated levels of reactive oxygen/nitrogen species (ROS/RNS) are detected in a Synechococcus glbN knock-out strain under both aerobic and microoxic growth on nitrate [16]. Although stress from oxygen, temperature, salt, and various medium components did not result in substantial changes to expression levels of glbN, it is of note that expression under these conditions roughly matched that of other transcripts for genes (sodB, katG, msrA, msrB, etc.) encoding proteins known to mitigate ROS/RNS damage [17]. These in vivo observations suggest that GlbN and GlbN-A participate in nitrogen-oxygen chemistry and implicate the heme crosslink in modulating the response.

We have shown previously that ΔglbN Synechococcus cells are more susceptible to nitric oxide (NO) challenge than wild-type cells and that they accumulate higher levels of ROS/RNS [16]. Dedicated NO synthases have not been identified in Synechococcus, but the reduction of nitrate to ammonia likely releases NO adventitiously via enzymes such as nitrate reductase (NarB) [19] as has been observed in plants [20]. In the absence of O₂, ferrous GlbNs are themselves another potential source of NO via their nitrite reductase activity [21]. Released NO can then undergo diffusion-limited combination with superoxide to form the powerful oxidant peroxynitrite (O=NOO⁻), its conjugate acid, and various nitrogen oxides [22–25]. These RNS, like ROS, attack cellular components and require active management. Of particular importance in cyanobacteria is the reaction of NO and NO-derived species with heme proteins, iron-sulfur clusters [26], and free thiols [27, 28], which would interfere with photosynthesis and respiration, among other cellular processes.

In this work, we inspect the NO chemistries of GlbN and address differences in the in vitro reactivity of GlbN and GlbN-A. We show that GlbN and GlbN-A can oxidize NO through nitric oxide dioxygenation, but that both proteins undergo heme bleaching during such activity. We also provide evidence that GlbN, but not GlbN-A, has the remarkable capability of reducing NO to HNO at neutral pH using the common reductant dithionite as the electron source. The differential reactivity allowed us to propose a protective role for the histidine–heme PTM.

2. Materials and Methods

2.1. Protein preparation

Wild-type (WT) Synechococcus GlbN, H117A Synechococcus GlbN, WT Synechocystis GlbN, and WT Chlamydomonas reinhardtii THB1 were produced in Escherichia coli without affinity tag as described previously [6, 29, 30]. All globins are prepared in the ferric state by reconstitution of the purified apoprotein with hemin. Horse skeletal myoglobin was purchased from Sigma. Its apoprotein was prepared by the method of Teale [31].

The (heme-free) diaphorase domain of the C. reinhardtii nitrate reductase (NR) was produced through heterologous expression in E. coli. The protein consists of the C-terminal end of the native NR [32] beginning at residue K509. The corresponding amino acid sequence for this portion of the protein was codon-optimized for bacterial expression and synthesized as double stranded DNA (IDT, Coralville, IA, USA). The synthesized gene was fused at its N-terminus to a sequence encoding a polyhistidine tag and PreScission protease cleavage site within a pET28 expression plasmid. The protein was expressed at 27 °C for 20 h using LB medium supplemented with 50 μg/mL kanamycin and 0.1 mg/mL riboflavin to boost flavin adenine dinucleotide (FAD) incorporation. Under these conditions, expressed soluble protein incorporated the FAD cofactor and no heme. The soluble protein was bound to a Ni-NTA column and eluted with 100 mM imidazole. The protein was dialyzed extensively with 20 mM Tris-HCl pH 8.0, 20 mM NaCl to remove imidazole. 0.1 mg/mL PreScission protease (GE Healthcare) was added prior to dialysis. The cleaved N-terminal tag was eliminated by a subsequent pass through the Ni-NTA column, which also removed the polyhistidine-tagged protease. The concentration of the resulting active protein solution was determined using the free FAD concentration following cofactor release with 0.5% SDS and using an extinction coefficient ε₄₅₀ = 11.3 mM⁻¹ cm⁻¹ [33]. The protein was lyophilized and stored at −80 °C until used.

2.2. Preparation of GlbN-A from GlbN

To prepare GlbN-A from GlbN, a concentrated solution of ferric GlbN (typically 1–2 mM) in 50–250 mM Na/K-phosphate pH 7.0–7.2 was reduced with a 5-fold molar excess of freshly prepared sodium dithionite (DT, Alfa-Aesar). At this pH, the conversion of ferrous GlbN to ferrous GlbN-A is completed in less than 10 s. For storage, the protein was oxidized to the ferric state by addition of excess potassium ferricyanide and purified by passage through a G-25 desalting column.

2.3. Glucose oxidase/D-(+)-glucose/catalase O₂ scavenging system

To generate in vitro microoxic conditions, we used the glucose oxidase/D-(+)-glucose/catalase (GODCAT) O₂ elimination system [34]. For optical absorbance experiments, individual components included (final concentrations): 40 μg/mL bovine catalase (Sigma), 100–200 μg/mL Aspergillus niger glucose oxidase (Sigma), and 0.04% m/v D-glucose. For NMR experiments, the catalase and glucose oxidase concentrations were doubled. At these concentrations the GODCAT system depleted most dissolved O₂ present in a small solution volume in under 5 min as determined using an oxygen electrode.

2.4. Ferredoxin/NADP⁺ and nitrate reductase diaphorase reduction systems

When reduction under oxic conditions was required, two alternative enzymatic systems were used. The first was a ferredoxin-based system [35], which contains glucose 6-phosphate (G6P) as the initial source of electrons, G6P dehydrogenase (G6P DH) to generate NADPH from G6P and NADP⁺, and ferredoxin-NADP⁺ reductase (FNR) to reduce spinach ferredoxin (Fd). Typical optical absorbance experiments included the following components (all from Sigma, final concentrations): 0.1 % m/v G6P, 0.01–0.2 mM NADP⁺, ~20 U Leuconostoc mesenteroides G6P DH, ~0.1 U spinach FNR, 100–500 μg/mL Fd, and 40 μg/mL bovine catalase to eliminate H₂O₂. Order-of-addition control experiments indicated that Fd was the active reductant for GlbNs. In variations of this protocol, some experiments were conducted by omitting the NADP⁺/NADPH recycling system (G6P and G6P DH) and using NADPH (Sigma) instead of NADP⁺. Such optical absorbance experiments included the following components: 25 mU spinach FNR, 0.5 μM Fd, 40 μg/mL catalase, 300 μM NADPH. NADPH alone failed to reduce Fe^III GlbN and GlbN-A. The second reduction system was composed of the heme-free diaphorase domain of C. reinhardtii NR and NADPH as the source of electrons, to circumvent possible long-term damage caused by NO to the Fd/FNR system. Optical experiments included 2 μM NR diaphorase and 300 μM NADPH.

2.5. Optical absorbance spectrophotometry

Absorbance data were acquired at room temperature on a Cary50 UV-vis spectrophotometer. Protein concentration was evaluated with the following coefficients: H117A and WT Synechococcus Fe^III GlbNs [6], ε₄₁₁ = 96 mM⁻¹ cm⁻¹; WT Fe^III GlbN-A, ε₄₀₉ = 87 mM⁻¹ cm⁻¹. Spectra were typically collected over 700–250 nm, with 0.5 s averaging time, and 1 nm step size. For kinetic measurements, this window was narrowed to the desired region and acquisition parameters were 0.1 s averaging time with 1 nm step size. Manual mixing dead times were ~15 s. Unless otherwise noted, optical spectra were collected every 0.5 min for 20 min, and then every 5 min for 2 h.

2.6. NO binding to Fe^III GlbNs monitored by optical absorbance

Stock solutions of the NO donors dipropylenetriamine-NONOate (DPTA-NONOate, Cayman Chemical) or methylamine hexamethylene methylamine-NONOate (MAHMA-NONOate, Cayman Chemical) were prepared in 0.1 M NaOH. The concentration was evaluated after dilution in H₂O using published extinction coefficients (DPTA-NONOate, ε₂₅₂ = 7.9 mM⁻¹ cm⁻¹ [36]; MAHMA-NONOate, ε₂₅₀ = 7.25 mM⁻¹ cm⁻¹ [37]). At pH 13, the NONOates were stable for days to weeks and degradation could be assessed by a decrease in the 250 nm absorbance band. At pH 7, the NONOates degrade by an apparent single exponential process (DPTA-NONOate, t_1/2 ~ 3–4 h, MAHMA-NONOate t_1/2 ~ 3–4 min) to produce two molecules of NO per one molecule donor. In order to produce Fe^III–NO forms for optical absorbance spectroscopy, Fe^III GlbN and GlbN-A (4–10 μM) samples were treated with 0.8–1 mM NONOate.

2.7. NO dioxygenase (NOD) assay

Fe^III GlbN or GlbN-A (20 μM, 50 mM Tris, pH 7.0–7.2) was incubated in O₂–saturated buffer (O₂, Airgas) and in the presence of a Fd reductase system to produce oxy (Fe^II–O₂) complexes. Conversion was followed optically and, when completed, MAHMA-NONOate was added to the specified concentration (typically 1.5-fold molar excess NO on a heme basis) to generate NO and initiate NOD activity. After oxidation, GlbN (or GlbN-A) was re-reduced by the enzyme system and returned to the Fe^II–O₂ form. For multiple NO addition experiments, MAHMA-NONOate was introduced after full recovery to the Fe^II–O₂ form. WT C. reinhardtii THB1 was treated similarly.

2.8. Reduction of Fe^III–NO GlbNs monitored by optical absorbance

Fe^III WT and H117A GlbNs were treated with DPTA- or MAHMA-NONOate to produce the Fe^III–NO (4–10 μM, pH 7.0–7.2) form. The Fe^III–NO GlbN(-A) was then treated with either 2 mM freshly prepared DT or the NR diaphorase reduction system (after incubation with GODCAT) to initiate reduction. As a test for heme loss upon Fe^II–NO formation, an ~ 8-fold molar excess (40–80 μM) of horse skeletal apomyoglobin was incubated with GlbN(-A) prior to the DT reduction step (~15 s). Under these conditions, heme dissociation from GlbN and capture by apomyoglobin (forming apoGlbN and Fe^II–NO myoglobin) can be monitored owing to the different visible spectra of GlbN and myoglobin nitrosyl complexes [38].

2.9. NMR spectroscopy of nitrosyl-GlbNs

NMR data were acquired on Bruker Avance or Avance-II NMR spectrometers operating at 600 MHz, each equipped with a cryoprobe. An unlabeled ~1.1 mM Fe^III GlbN sample (100 mM Na/K phosphate, pH 7.1, GODCAT, 298 K) was treated with four-fold excess MAHMA-NONOate to generate the Fe^III–NO state. A mixture of Fe^III and Fe^III–NO forms was produced and remained relatively stable for ~6 h. Water presaturation ¹H 1-D, ¹H-¹H NOESY (τ_mix = 80 ms), DQF-COSY, and TOCSY (τ_mix = 45 ms) spectra were recorded on the resulting sample in order to assign heme resonances. Because the Fe^III–NO form decayed in less than 12 h, only a preliminary analysis could be achieved. Similarly, an ¹⁵N-labeled ~1.4 mM Fe^III GlbN-A (250 mM phosphate buffer, pH 7.1, GODCAT, 298 K) sample was treated with 5-fold molar excess MAHMA-NONOate to produce a mixture of Fe^III and Fe^III–NO GlbN-A. As above, water-presaturation ¹H 1-D, flip-back WATERGATE ¹H-¹H NOESY, and water-presaturation ¹H-¹H TOCSY spectra were acquired with ¹⁵N-decoupling and used to assign heme resonances. ¹H-¹⁵N HSQC spectra were recorded on a separate ¹⁵N-labeled Fe^III–NO GlbN-A sample (800 μM protein, 5 mM DPTA-NONOate, 250 mM phosphate buffer, pH 7.1, 298 K). After data acquisition, the diamagnetic Fe^III–NO GlbN-A sample was treated with 8 mM DT, under argon (Airgas) atmosphere, and flip-back WATERGATE ¹H-¹⁵N HSQC spectra were recorded on Fe^II–NO GlbN-A.

To generate the nitrosyl hydride complexes (Fe^II–HNO GlbN), two separate procedures were applied. 1) As above for the optical absorbance experiments, Fe^III GlbN was treated with MAHMA-NONOate to produce Fe^III–NO GlbN and subsequently reduced with excess DT. 2) Fe^III GlbN (or H117A GlbN) was incubated in the presence of nitrite or ¹⁵N-labeled nitrite (NO₂⁻ or ¹⁵NO₂⁻) and reduced with excess DT. Here, the nitrite reductase activity of GlbN was exploited to generate NO (or ¹⁵NO) in situ. Procedures 1) and 2) both resulted in a mixture containing the same diamagnetic Fe^II–HNO (or Fe^II–H¹⁵NO) form, which persisted for several hours. Water-presaturation¹H 1-D, NOESY, DQF-COSY, and flip-back WATERGATE ¹H-¹⁵N HSQC spectra were used for heme and HNO assignments. To test for HNO formation in the absence of DT, ¹⁵N-labeled H117A GlbN (~850 μM protein, 100 mM K/Na phosphate buffer, pH 7.1) was combined with ¹⁵NO₂ and the reduction system (5 μM heme-free NR diaphorase domain, 10 mM NADPH). ¹H 1-D and ¹H-¹⁵N HSQC were then collected for several hours. Under those conditions, no ¹H-¹⁵NO signal was detected.

2.10. NMR data processing and analysis

NMR data were processed using Topspin 1.3, Topspin 2.1, or NMRpipe [39]. Spectra were analyzed using Sparky 3 [40]. ¹H chemical shifts were referenced with respect to water (¹H = 4.76 ppm, 298 K). ¹⁵N chemical shifts were referenced against liquid ammonia indirectly using the Ξ ratio [41].

2.11. Nomenclature used for GlbN nitrosyl complexes

In the text, Fe^III–NO or Fe^II–NO⁺ refers to a {FeNO}⁶ complex in Enemark-Feltham notation [42]. Fe^II–NO refers to a {FeNO}⁷ complex and Fe^II–HNO or Fe^II–NO⁻ denotes a {FeNO}⁸ complex.

3. Results

3.1. Ligand binding and electron transfer in bis–histidine GlbNs

Before inspecting GlbN reactions with NO, it is useful to present some common features related to heme hexacoordination. Most processes will require decoordination of the distal histidine to allow for exogenous ligand binding to iron (Equation 1a) and many will involve the loss or the gain of one electron, in conjunction with electron transfer (ET) from or to a redox partner. Two possible ET paths are illustrated with ferric state reduction (Equations 1b and 1c).

$$\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}–\mathrm{His}46\leftrightarrows \mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}+\mathrm{His}46$$ (1a)

$$\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}+{\mathrm{e}}^{-}\to \mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}$$ (1b)

$$\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–\mathrm{His}46+{\mathrm{e}}^{-}\to \mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}–\mathrm{His}46$$ (1c)

Under normal conditions of pH, temperature, and pressure, Equilibrium 1a favors strongly the hexacoordinated state through rapid k_on and slow k_off for His46 (k_on/k_off = K estimated to be ~ 300 for His46 in Synechocystis GlbN-A [43]). It is expected that reaction (1c), which uses the bis–histidine state, outcompetes reaction (1b) if distal ligand recoordination is rapid compared to ET in the pentacoordinated state [44]. Furthermore, water binding to the ferric (but not ferrous) pentacoordinate iron can attenuate reduction kinetics of the non bis–histidine state by increasing the nuclear reorganization energy associated with ET [45].

3.2. O₂ binding and PTM inhibition

Synechococcus GlbN and GlbN-A, along with the closely related Synechocystis protein, undergo a large conformational change upon switching between bis–histidine and exogenous ligand-bound states [8, 46] (Figure S1A). In the cyanide-bound structure (Fe^III–CN^–), which serves as a model for the Fe^II–O₂ state, a distinct array of amino acids interacts with the distal ligand. The conserved Tyr22 OηH is the donor in an H-bond to the cyanide nitrogen; additionally, Gln43 and Gln47 NεH₂ groups are positioned to form a network of H-bonds with bound cyanide and Tyr22 Oη. These same interactions are thought to stabilize activated O₂ and inhibit dissociation or autooxidation [47, 48]. Dioxygen binding and activation to the ferric-superoxide species is shown in Equation 2.

$$\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}+{\text{ O}}_2\leftrightarrows [\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–{\mathrm{O}}_2{}^{-}]$$ (2)

Figure 1 illustrates O₂ binding by GlbN under oxic conditions where [O₂] ˃˃ [GlbN], as pertinent for optical absorbance spectrophotometry of air-equilibrated solutions. The experiment makes use of Fd as an electron shuttle and an enzymatic reduction system to produce GlbN in the Fe^II state. Following mixing of the components, slow Fd-mediated reduction of GlbN is followed by rapid O₂ binding resulting in an apparent Fe^III → Fe^II–O₂ transition (Figure 1A, dashed-line spectrum to solid-line spectrum). Useful absorption maxima are listed in Table 1.

Fig. 1.

(A) Dioxygen binding monitored by optical absorbance spectrophotometry. A ~5.4 μM ferric GlbN sample at pH 7.1 (Fe^III, dashed trace) was incubated with a catalytic ferredoxin (Fd) reduction system. Gradual conversion of Fe^III GlbN to oxy GlbN (Fe^II–O₂) occurred in 10 min (solid traces). (B) Spectral overlay of Fe^II–O₂ GlbN (solid trace) and Fe^II–O₂ GlbN-A (dash-dot trace) prepared with the Fd reduction system.

Table 1.

Absorption maxima of various forms of GlbN

Ligation state	GlbN γ, β, α (nm)	GlbN-A γ, β, α (nm)
His–FeIII–His	411, ~544	408, ~544
His–FeII–O2	411, 550, 584	408, 550, ~585
His–FeIII–NO	424, 538, 574	424, 538, 571
His–FeII–His	–	425, 527, 558
His–FeII–NOa	~416, ~563	~415, 560
His–FeII–HNOa	~417, ~559	–

^aApproximate value because of the presence of other species

In the absence of oxygen and at neutral pH, reduction of Fe^III GlbN to the Fe^II state is followed by PTM [14, 49]. The modification proceeds to completion within a few seconds. Under oxic conditions, however, the Fe^II–O₂ GlbN species (Figure 1B, solid-line spectrum) persists and shows no evidence of conversion to Fe^II–O₂ GlbN-A (Figure 1B, dash-dot spectrum) over 30 min. These observations are consistent with the inhibitory effect of O₂ toward the PTM and the detection of unmodified GlbN in synechococcal cells grown under oxic conditions.

3.3. NOD activity

Because the oxy form of GlbN and GlbN-A could be produced separately, we next compared their ability to catalyze the NO dioxygenation reaction, which is an accepted route for NO destruction by Hbs under oxic conditions [50]. Indeed, NO dioxygenase (NOD) activity in vivo has been proposed for TrHb1s from strains of Mycobacterium [51–53], the ciliate Tetrahymena pyriformis [48], microalgal raphidophytes (Heterosigma akashiwo and Chattonella subsalsa) [54], and the green alga C. reinhardtii [30]. The physiological evidence gathered so far does not rule out a similar role for GlbN and GlbN-A.

Beyond Equation 1a, the NOD reaction is a multi-step process [50] starting with O₂ binding (Equation 2) and proceeding with Equations 3a–b,

$$[\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–{\text{O}}_2{}^{-}]+\text{ NO}\to [\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–{\text{OONO}}^{-}]$$ (3a)

$$[\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–{\text{OONO}}^{-}]\to \mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}+{\text{ NO}}_3{}^{-}$$ (3b)

where process (3b) captures the steps of peroxynitrite isomerization [55]. To undergo NOD turnover, the resulting ferric protein must be re-reduced according to Equation 1b or 1c.

The reaction of GlbNs with NO under oxic conditions was examined with an optical assay developed in prior studies of C. reinhardtii THB1 [30]. Figure 2 presents the α–β region of the spectrum and intensity of the β band in representative experiments. Panel A illustrates the reaction with GlbN. The assay begins with bis–histidine Fe^III GlbN in the presence of the Fd/FNR reduction system but without NADPH (top dashed-line trace). Addition of NADPH results in the formation of Fe^II–O₂ GlbN (top solid trace). Monitoring the absorbance at 550 nm (Figure 2, inset) highlights the appearance of the characteristic β band of Fe^II–O₂ GlbN. Once the conversion to oxy GlbN is complete, the NO releasing agent MAHMA-NONOate is added (upper asterisk in inset). The spectral intensity associated with Fe^II–O₂ GlbN immediately begins to drop. Within 2 to 3 min, the entire sample is converted to the Fe^III bis–histidine state (lower asterisk in inset). This ferric spectrum persists while the release of NO by MAHMA-NONOate and its consumption by regenerated Fe^II–O₂ GlbN continue. Upon depletion of NO, the protein returns to the Fe^II–O₂ GlbN state. These data are consistent with the O₂–binding observations presented above, with no build-up of bis–histidine Fe^II GlbN or production of GlbN-A.

Fig. 2.

NOD assay performed with (A) ~20 μM GlbN or (B) ~20 μM GlbN-A (50 mM Tris, pH 7.2). The Fe^III bis–histidine proteins (most intense of the dashed-line spectra) were incubated with Fd and FNR; 300 μM NADPH was then added to initiate the reactions. (A) Following reduction and dioxygen binding, Fe^II–O₂ GlbN (most intense of the solid-line spectra) was treated with 15 μM of MAHMA-NONOate. Rapid oxidation to Fe^III bis–histidine GlbN followed (middle dashed-line spectrum) and this state persisted for ~6 min during turnover. Return to the Fe^II–O₂ state occurred after NO consumption (middle solid-line spectrum). A second NONOate addition yielded the lowest intensity Fe^III bis–histidine GlbN and Fe^II–O₂ GlbN spectra. The inset presents the response at 550 nm (arrow). Upper asterisks indicate NO donor addition and lower asterisks mark the turnover periods. (B) Data of the same acquired on GlbN-A. Note the different x-axis in the inset.

Following each NO donor addition, the turnover period increased in duration, the extent of Fe^II–O₂ recovery was attenuated (decreasing solid traces), and the ferric state absorbance decreased over the whole spectrum (decreasing dashed-line traces), all indications of GlbN deactivation via heme bleaching. The Fe^III → Fe^II–O₂ transition was also slower, an observation noted for all tested globins and attributed to NO poisoning of the Fd/FNR system. Addition of fresh Fd/FNR to the THB1 control sample (not shown) confirmed the interpretation. Panel B in Figure 2 presents the data for Fe^II–O₂ GlbN-A, which displays faster reduction and O₂ binding but undergoes more bleaching than GlbN on multiple NO additions. The deactivation observed for GlbNs is in contrast with the full recovery of THB1 [30] (Figure S2). Thus, GlbN and GlbN-A appear capable of NOD chemistry and may play such a role under oxic conditions, although heme damage suggests that they are not fit for multiple enzymatic turnovers, unlike THB1.

3.4. NO binding to Fe^III GlbN and GlbN-A

GlbN or GlbN-A may be able to manage endogenous NO concentration through O₂–independent mechanisms. In addition to sequestration, one possibility is reductive nitrosylation (or “autoreduction”) by which the ferric protein binds NO, forms the electrophilic Fe^II–NO⁺ state, and eventually produces NO₂⁻ and ferrous protein under the action of OH⁻ or water [38, 56]. In the presence of excess NO, the ferrous protein is then converted to the Fe^II–NO state. Unlike horse skeletal myoglobin (Figure S3) and several other heme proteins [38, 56], treatment of ferric GlbN (or GlbN-A) with excess NO donor leads to stable {FeNO}⁶ complexes (Figure 3), which were interrogated with NMR spectroscopy (Figures S4 and S5). Narrow ¹H chemical shift dispersion, sharp lines, and heme resonance assignments support that both Fe^III–NO complexes are diamagnetic (S = 0). Table 2 lists heme ¹H chemical shifts of Fe^III–NO GlbN and Fe^III–NO GlbN-A. Because both GlbN and GlbN-A were maintained in the ferric nitrosyl state for ~10 h, it seems unlikely that the NO elimination via the autoreduction process is relevant to their ability to protect Synechococcus from RNS/ROS.

Fig. 3.

Formation of Fe^III–NO GlbN and GlbN-A. Approximately 1.3 mM MAHMA-NONOate was added to a sample of ~9 μM (A) Fe^III bis–histidine GlbN or (B) Fe^III bis–histidine GlbN-A (~100 mM phosphate, pH 7.1, dashed-line spectra). Conversion to the ferric NO adducts (solid-line spectra, acquired ~15 min after NONOate addition) was monitored over time (thin solid-line spectra).

Table 2.

Heme and HNO proton chemical shifts (pH 7.1, 298 K)

	FeIII–NO GlbN-A	FeIII–NO GlbN	FeII–HNO GlbN
1-CH3	3.44	3.40	3.39
2-α	6.81	7.96	7.90
2-βcis,trans	–	6.16, 6.04	5.75, 5.85
2-β-CH3	2.36	–	–
α-meso	9.75	9.80	9.28
3-CH3	3.56	3.53	3.32
4-α	7.80	7.71	7.55
4-βcis,trans	5.94, 6.30	5.95, 6.12	5.61, 5.88
β-meso	9.52	9.36	8.83
5-CH3	2.97	3.03	2.68
γ-meso	10.32	10.30	9.63
8-CH3	3.35	3.33	3.15
δ-meso	9.97	9.96	9.41
HNO	–	–	14.82 14.91 (H117A)

3.5. Reduction of Fe^III–NO GlbN-A to Fe^II–NO GlbN-A

We next examined whether the Fe^III–NO GlbN-A complex could undergo forced reduction to the Fe^II–NO form. Fe^III–NO GlbN-A was prepared as above by treatment of the ferric protein with excess MAHMA-NONOate and monitored optically. Upon saturation with NO, the sample was reacted with excess DT, to make use of a relatively strong reducing agent [57]. Within the dead time of the experiment (~15 s), changes from the Fe^III–NO starting material (Figure 4A, dashed-line spectrum) were observed and suggested that reduction to a stable Fe^II–NO GlbN-A form (Figure 4A, solid-line spectrum) had occurred.

Fig. 4.

Reduction of Fe^III–NO GlbN-A using DT. (A) Dashed-line spectrum: Fe^III–NO GlbN-A obtained by treatment of Fe^III bis–histidine GlbN-A (~9 μM, GODCAT, 100 mM phosphate, pH 7.1) with 0.8 mM MAHMA-NONOate. Solid-line spectrum: Fe^II–NO GlbN-A, 10 min after addition of 2 mM DT; the band at ~558 nm is due to the presence of a small fraction of Fe^II bis–histidine GlbN-A. Thin gray traces: spectra acquired between the 15 s dead time and 10 min (every 30 s) showing that formation of Fe^II–NO GlbN-A occurred almost entirely during the dead time. (B) Overlay of ¹H-¹⁵N HSQC spectra acquired on Fe^III–NO GlbN-A (gray or red) and Fe^II–NO GlbN-A (black). Fe^III–NO GlbN-A (~850 μM GlbN, 250 mM phosphate, pH 7.2, 10% D₂O) was prepared by addition of ~5 mM DPTA-NONOate to the Fe^III bis–histidine protein in the presence of GODCAT. The spectrum of Fe^III–NO GlbN-A was acquired ~5 h following NONOate addition. Fe^II–NO GlbN-A was prepared from the Fe^III–NO GlbN-A sample by addition of 8 mM DT under an argon atmosphere.

An analogous experiment was performed using an ¹⁵N-labeled NMR sample of Fe^III–NO GlbN-A. Figure 4B shows the initial ¹H-¹⁵N HSQC spectrum of Fe^III–NO GlbN-A (gray peaks). Although isoelectronic with Fe^II–CO GlbN-A, Fe^III–NO GlbN-A gave rise to a distinct ¹H-¹⁵N correlation map, suggesting subtle structural differences between the two diamagnetic complexes. Reduction of Fe^III–NO GlbN-A with excess DT resulted in noticeable ¹H-¹⁵N HSQC spectral changes (Figure 4B, black peaks). Despite the chemical shift perturbations, the NMR spectra are quite similar overall, and support that the ferric and ferrous nitrosyl forms share the same fold and mode of hexacoordination (His70–Fe–NO). However, in Fe^II–NO GlbN-A, the unpaired electron causes rapid paramagnetic relaxation for NH nuclei near the iron center and therefore broadens several signals beyond detection (e.g., the tentative His70 backbone amide in Fe^III–NO GlbN-A: ¹H = 5.55 ppm, ¹⁵N = 106.3 ppm, is not observed in Fe^II–NO GlbN-A). These spectra, along with the lack of detectable heme signals in NOESY spectra, support that complete formation of paramagnetic (S = 1/2) Fe^II–NO GlbN-A had occurred.

3.6. The Fe^II–NO GlbN complex and NO reduction to HNO

The reduction behavior of ferric nitrosyl GlbN-A was used as a point of reference for GlbN experiments. As for GlbN-A, a Fe^III–NO GlbN sample was produced by treatment of the ferric protein with excess MAHMA-NONOate and monitored optically. Once saturation was achieved, excess DT was added to initiate reduction. Figure 5A shows representative results of the experiment. GlbN displayed distinct biphasic behavior compared to GlbN-A (Figure 4A). The first (rapid) phase was completed within ~3 min and was characterized by a marked decrease in Soret absorbance occurring concomitantly with a blue-shift (dash-dot spectrum); this initial rapid phase was followed by a slow (t_1/2 ~ 1 h) recovery phase (solid-line spectrum: 417 nm Soret, 559 nm merged α/β). The low Soret intensity and ~390 nm shoulder of the intermediate supported that formation of Fe^II–NO GlbN inhibited ferrous heme crosslinking and instead caused rapid heme dissociation. In agreement, similar biphasic spectral changes were detected upon DT reduction of Fe^III–NO H117A GlbN (a variant incapable of the His117–heme PTM).

Fig. 5.

Reduction of Fe^III–NO GlbN (~10 μM, GODCAT, 100 mM phosphate, pH 7.1) using DT. (A) Dashed-line spectrum: Fe^III–NO GlbN obtained by treatment of Fe^III bis–histidine with 0.8 mM MAHMA-NONOate. Dash-dot spectrum: the mixture obtained ~3 min after DT addition. Solid-line spectrum: the mixture obtained ~200 min after DT reduction. (B) DT reduction of Fe^III–NO GlbN (~5 μM, GODCAT, 100 mM phosphate, pH 7.1) in the presence of ~40 μM apomyoglobin. Dashed-line spectrum: Fe^III–NO GlbN obtained by treatment of Fe^III bis–histidine with 1.6 mM MAHMA-NONOate in the presence of apomyoglobin. The increased baseline absorbance is indicative of Rayleigh scattering. Thin gray spectra: species obtained immediately after DT addition. The intensity increases slightly over a few min. Solid-line spectrum: Fe^II–NO myoglobin prepared by DT reduction of ferric myoglobin in the presence of nitrite.

NO can bind with pM affinity to ferrous heme [58] and is well known for its strong negative σ-trans effect [59]; the latter phenomenon can considerably weaken the coordination bond of the iron ligand on the other side of the heme plane. Indeed, the ~ 390 nm shoulder exhibited by the Fe^II–NO WT and H117A GlbNs intermediate is reminiscent of the blue-shifted Soret band (~ 398 nm) observed for the five-coordinate NO bound form(s) of the heme domain of soluble guanylate cyclase [60]. We examined the plausibility of weakened His70 coordination and lowered heme affinity by performing a heme transfer experiment. Fe^III GlbN was first incubated with an 8-fold molar excess of horse skeletal apomyoglobin; excess NONOate was then added as above (Figure 5A). The spectral changes that followed corresponded to formation of Fe^III–NO GlbN (Figure 5B, dashed-line trace). Upon completion, excess DT was added to initiate reduction. Following a 15-s dead time the distinctive optical signature of Fe^II–NO myoglobin was detected (Figure 5B, thin gray traces) in an indication of rapid heme acquisition. We therefore interpret the first phase observed upon DT reduction of Fe^III–NO GlbN (Figure 5A) as His70 deligation and heme dislodging.

To test the specific role of NO in promoting heme dissociation from ferrous GlbN, two additional heme transfer experiments were performed. In the first, ferric bis–histidine GlbN was reduced by DT in the presence of excess apomyoglobin. Under such conditions, we observed rapid formation of ferrous bis–histidine GlbN-A, and no detectable formation of deoxy Mb (Fe^II Mb) (Figure S6). Thus, uninhibited PTM clearly out-competes heme transfer near neutral pH. More importantly, when the ferrous carbonmonoxy state (Fe^II–CO GlbN) was generated in the presence of excess apomyoglobin (Figure S7), no heme transfer was detected over a period of 50 min. The CO result, in addition to the lack of detectable heme loss in Fe^III–NO GlbN, implicated specifically Fe^II–NO coordination as a causal factor for rapid heme dissociation in GlbN. In contrast to Fe^II–NO GlbN, the spectral features of the Fe^II–NO GlbN-A adduct are consistent with a stable six-coordinate His–Fe–NO complex, which indicates a proximal side of the heme cavity unable to accommodate a decoordinated His70, likely because of the crowding exerted by Met66, Leu73, and Val121.

¹H NMR was used to characterize the nature of the second phase and final product(s) in the reaction of Fe^III–NO GlbN with DT (Figure 5A, dash-dot to solid line transition). In this experiment, a concentrated Fe^III–NO GlbN solution (1.5 mM) was reduced with excess DT and allowed to incubate for ≥ 45 min prior to data acquisition. Unlike with GlbN-A, distinct diamagnetic forms were observed in the resulting mixture, and two ¹H signals (major and minor) reminiscent of nitrosyl hydride (HNO) within Fe^II–HNO myoglobin [61, 62] were detected at ~14.8 ppm (Figure 6A). Production of Fe^II–HNO from Fe^III–NO is summarized in the following reaction:

$$\mathrm{His}70–{\mathrm{Fe}}^{\mathrm{III}}–\text{NO }+{\text{ H}}^{+}+2{\text{e}}^{-}\to \mathrm{His}70–{\mathrm{Fe}}^{\mathrm{II}}–\text{HNO}$$ (4)

Fig. 6.

Downfield region of ¹H NMR spectra of Fe^II–HNO WT GlbN. (A) A 1.5 mM Fe^III bis–histidine WT ¹⁵N GlbN sample (200 mM phosphate, pH 7.1, 10% D₂O, 298 K, GODCAT) was treated with 7.5 mM MAHMA-NONOate. The resulting mixture of Fe^III bis–histidine and Fe^III–NO GlbN was then treated with 7.5 mM DT. After ~ 50 min incubation, two Fe^II–HNO signals are detected (with major:minor intensity ratio ~ 4:1). The total HNO yield was estimated by peak integration to be ~ 23 %. (B) A 400 μM Fe^III bis–histidine WT GlbN sample (70 mM phosphate, pH 8.5 (pre DT), 10% D₂O, 298 K, GODCAT) incubated with 5 mM ¹⁵N-nitrite was treated with 5 mM DT. After a ~ 2 h incubation, two H¹⁵NO doublets, each split by ¹J_NH = 71 Hz, were observed and correspond to Fe^II–H¹⁵NO GlbN (major signals) and Fe^II–H¹⁵NO GlbN-A (minor signals). The total H¹⁵NO yield was estimated by peak integration to be ~ 21 %.

Under anoxic conditions, most Hbs can produce NO via nitrite reductase chemistry [21, 63–68]. This process is essentially the reverse of the autoreduction reaction described above and is initiated in the ferrous (deoxy) state. As an unambiguous test for DT-mediated HNO formation, we took advantage of the nitrite reductase activity of GlbN [21, 65] to produce nitric oxide in situ. Specifically, we incubated unlabeled Fe^III GlbN with ¹⁵N-labeled nitrite (¹⁵NO₂⁻); excess DT was then added to initiate the reaction. Figure 6B shows the relevant region of the ¹H NMR spectrum following a ~ 2-h incubation. Compared with Figure 6A, the major (¹H = 14.82 ppm) and minor (¹H = 14.76 ppm) signals in Figure 6B are both split into doublets with |¹J_NH| ~ 71 Hz; this result is consistent with formation of H¹⁵NO. The magnitude of the ¹H¹⁵NO ¹J_NH–couplings were in good agreement with those determined previously for Fe^II–HNO Hb complexes (|¹J_NH| = 66–72 Hz) [61, 69].

To explore the generality of the NO reduction chemistry, we tested whether H117A GlbN could also produce HNO. Figure 7A displays the ¹H NMR spectrum of H117A GlbN, treated with DT in the presence of ¹⁵NO₂⁻ and allowed to recover for ~ 2 h. This variant is incapable of forming the heme PTM, and as observed for WT GlbN, a diamagnetic species with sharp lines was detected, which is inconsistent with Fe^II–NO H117A GlbN. Unlike the WT protein, however, a single ¹H doublet (|¹J_NH| ~ 71 Hz), centered at ~ 14.9 ppm, was observed (Figure 7B). The doublet indicates that H117A GlbN is also able to produce H¹⁵NO from ¹⁵NO₂⁻. The integrated intensity of the doublet relative to the resolved T80 amide suggested a ~ 36% Fe^II–HNO yield. ¹H-¹⁵N HSQC 2-D spectra revealed the Fe^II–H¹⁵NO ¹⁵N chemical shift to be ~536 ppm (Figure 7C) compared to ~580 ppm reported by others [69].

Fig. 7.

¹H and ¹⁵N NMR evidence for Fe^II–HNO formation in H117A GlbN. (A) The ¹H NMR spectrum of an ~850 μM ¹⁵N H117A GlbN sample (100 mM phosphate, pH 7.1, 10 % D₂O, 298 K, GODCAT) obtained ~2 h after addition of 8 mM DT in the presence of 4 mM ¹⁵N-nitrite. ¹⁵N decoupling (centered at 117 ppm) was applied during ¹H detection. A ¹H doublet at ~14.9 ppm, attributed to H¹⁵NO bound to Fe^II H117A GlbN, was detected with ~ 36% intensity of that of the backbone amide ¹H signals of Thr80. (B) Expansion of (A). The ~14.9 ppm H¹⁵NO signal was split by |¹J_NH| = 71 Hz. ¹⁵N decoupling centered at 550 ppm collapsed the splitting (not shown). (C) Decoupled ¹H-¹⁵N HSQC spectrum of Fe^II–H¹⁵NO H117A GlbN.

The minor peak (¹H = 14.76 ppm) detected in the WT spectra (Figure 6) likely corresponds to HNO trapped by Fe^II GlbN-A formed during the reduction procedure. In agreement with this interpretation, the minor peak was absent when using H117A GlbN (Figure 7). It is also worth reiterating that no HNO and only Fe^II–NO GlbN-A were detected when a pure WT GlbN-A sample was subjected to DT reduction in the presence of NO₂⁻ (and analogously, when Fe^III–NO GlbN-A was reduced with DT).

¹H-¹H NOESY and DQF-COSY spectra collected on Fe^II–HNO WT GlbN allowed for partial heme assignments and confirmed that both heme vinyl groups remained intact (Table 2). Therefore, upon NO or HNO binding, the spontaneous heme PTM was inhibited. Of note, an NOE between ¹HNO and the heme 8-methyl protons provides preliminary orientation information for the distal ligand and facilitates comparison with the detailed studies of HNO-myoglobin in which the HNO proton points toward the 1-methyl group [62].

The ability of GlbN to generate HNO was further examined by using a weaker reducing agent than DT. For this purpose, we turned to the NADPH-dependent diaphorase domain of C. reinhardtii NR, produced with only the FAD cofactor. This avoided the iron-sulfur cluster present in Fd, which could be affected by NO, and the heme present in the holodiaphorase, which could complicate interpretation. NADPH (~300 μM) present in solution was used to reduce the diaphorase flavin; it also allowed for the possibility of a hydride transfer mechanism to produce HNO (Fe^III–NO + H⁻ → Fe^II–HNO) [70].

Optical absorbance spectra monitoring the reactions of Fe^III–NO GlbN and Fe^III–NO GlbN-A with the heme-free diaphorase are shown in Figure 8. Reduction of Fe^III–NO GlbN-A resulted in relatively slow (t_1/2 ~ 5 min) formation of Fe^II–NO GlbN-A; upon addition of DT, only minimal changes were detected (Figure 8A). Addition of the diaphorase domain to Fe^III–NO GlbN caused a slow (t_1/2 ~ 10 min) conversion to a species attributable to Fe^II–NO GlbN (with heme only weakly associated). The spectral changes (decrease in Soret intensity, ~ 390 nm shoulder) were similar to those observed in the first phase following DT treatment (Figure 5A, dashed line to dash-dot line transition). The Fe^II–NO GlbN form could then be converted to Fe^II–HNO GlbN by addition of DT (Figure 8B). Thus, under the tested conditions, it appears that the diaphorase domain is not capable of providing the driving force necessary for Fe^II–HNO GlbN formation. The Fe^II–HNO state was also not detected when using the NADPH/diaphorase system to reduce H117A GlbN in the presence of nitrite (data not shown). These results support that NADPH was unable to reduce Fe^III–NO GlbN to the Fe^II–HNO form via direct hydride transfer.

Fig. 8.

Reduction of Fe^III–NO GlbN-A and GlbN using heme-free diaphorase and NADPH. (A) The Fe^III–NO GlbN-A species (dashed trace) was generated by treatment of Fe^III bis–histidine GlbN-A with excess MAHMA-NONOate. Some residual Fe^III bis–histidine species is present. Addition of NADPH and heme-free diaphorase resulted in the formation of Fe^II–NO GlbN-A (dash-dot line), with minimal spectral change upon addition of DT (solid trace). (B) The Fe^III–NO GlbN species (dashed trace) was slowly converted to Fe^II–NO GlbN (dash-dot trace) by incubation with the heme-free diaphorase and NADPH for 70 min. Addition of DT then resulted in the gradual formation of Fe^II–HNO GlbN (solid trace, ~4 h after DT addition) as evidenced by the increase of the Soret peak to 417 nm.

Interestingly, subsequent optical absorbance experiments supported that the ability for GlbN but not GlbN-A to produce HNO is common to the Synechocystis GlbN/GlbN-A pair (Figure S8 and S9). In an experiment where Fe^III Synechocystis GlbN, excess nitrite, and excess DT were reacted, ¹H NMR spectra were used to positively identify the HNO proton (¹H = 14.80 ppm, data not shown). However, the yield of Fe^II–HNO in Synechocystis GlbN (< 5 %) was considerably lower than that of the Synechococcus proteins (20–35 %). No such HNO signal was observed in the analogous experiment using Synechocystis GlbN-A.

4. Discussion

4.1. NOD activity of GlbN and GlbN-A

The current study supports that both oxy GlbN and oxy GlbN-A are capable of NOD activity. This in itself is not surprising as an increasing number of studies have implicated hemoglobins as a potential detoxifier of biologically produced NO [30, 50, 71]. Hargrove and coworkers have emphasized that single domain Hbs have dioxygenase activity in vitro, but are often limited by the reduction step in vivo [72]. As pointed out previously, the synechococcal cells contain an ample pool of reduced Fd [16], which as shown in our NOD assay can reliably reduce the heme. This would both eliminate the need for a dedicated reductase and link the activity of GlbN to a probable source of NO, the Fd-dependent nitrate reductase (NarB). In addition, the bis–histidine hexacoordination mode observed in GlbN and GlbN-A facilitates ET [49, 73, 74], which is a necessary step for NOD turnover (Equation 1c).

Upon multiple NO exposures, however, GlbN and GlbN-A undergo significant levels of heme damage. The bleaching effect may be due to release of peroxynitrite or nitrogen dioxide radical during the isomerization step [55]. Unlike the proposed role for the heme covalent modifications in mammalian heme peroxidases [75], the histidine–heme crosslink of GlbN does not appear to prevent this inactivation during catalytic activity. As the holoprotein is not resistant to the chemistry it catalyzes, it is unclear whether NOD is the main function of GlbN under oxic conditions. Perhaps the full complement of protective enzymes, including peroxiredoxins, acts to reduce the loss of activity. It is also possible that NOD-mediated damage of GlbN and GlbN-A serves as a signal for RNS/ROS stress in synechococcal cells, for example through tyrosine nitration [76, 77].

4.2. NO binding, heme dissociation, and a stabilization role for the PTM

Figure 3 shows that at high concentration, NO can displace an axial histidine of ferric bis–histidine GlbN and GlbN-A to form a His–Fe^III–NO complex. NMR data support that the proximal histidine remains coordinated while NO binds to the distal site and displaces His46, as CN⁻ does in the ferric state [8]. Unlike many heme proteins [38, 56, 72], Fe^III–NO GlbN and Fe^III–NO GlbN-A show negligible autoreduction to their Fe^II–NO forms over a timescale of several hours. We took advantage of this relative stability to initiate a preliminary NMR analysis of the diamagnetic Fe^III–NO complexes. We were able to assign heme ¹H resonances (Figure S4 and S5, Table 2) and acquire high-quality ¹H-¹⁵N HSQC spectra (Figure 4B). Fe^III–NO GlbNs are therefore promising candidates for future NMR or resonance Raman studies that, for example, interrogate the interactions of bound NO with distal residues involved in the hydrogen bond network present in various TrHb1 complexes [8].

Although both GlbN and GlbN-A show a similar propensity for binding NO in the ferric state, the two proteins show marked differences with respect to their ability to form stable {FeNO}⁷ complexes. Fe^II GlbN-A binds NO like many heme proteins [58, 78, 79]. In contrast, if Fe^II GlbN combines with NO prior to spontaneous histidine–heme crosslinking, the PTM is inhibited and the Fe^II–NO heme group dissociates with an estimated k ≥ 0.1 s⁻¹ (pH 7.1). Partial heme release is demonstrated by the heme-scavenging experiment (Figure 5B) and consistent with the absorbance spectrum of the ferrous nitrosyl GlbN complex formed by reaction with a relatively weak reductant (Figure 8B). This behavior is rationalized with a weakening of the proximal histidine Nε2–Fe coordination bond by the trans bonded NO and the formation of a five-coordinate/six-coordinate mixture. Regardless of the specific determinants, NO-mediated heme loss could potentially interfere with any enzymatic cycle involving ferrous iron, including NOD activity. Thus, we propose that the histidine–heme PTM in GlbNs is critical for stable Fe^II–NO binding.

4.3. NO reduction to HNO

Pentacoordinate ferrous hemoglobins are capable of binding HNO to form Fe^II–HNO adducts [61, 62, 69, 70, 80]. In a detailed NMR study of Fe^II–HNO myoglobin, Sulc and coworkers utilized NOE analysis and porphyrin ring current shifts, along with comparison to the Fe^II–CO form, to propose that an H-bond between the distal histidine (His64) Nε2–H and the HNO oxygen helps to stabilize the adduct [62]. Here, we have presented evidence that the bis–histidine GlbNs from Synechococcus and Synechocystis are also capable of trapping HNO (Figures 6, 7, and S9). Apparently, HNO can out-compete the endogenous distal ligand, His46, to bind the ferrous iron. Unlike Fe^II–HNO myoglobin produced under optimized conditions [61, 62], Fe^II–HNO GlbNs were marginally stable and allowed for only a preliminary NMR analysis. The maximal Fe^II–HNO GlbN yield, attained using the H117A variant, was ~36 %, at the low end of the pentacoordinate globin range (30–100 %, depending on the method used to form the HNO adduct and the specific protein [61, 69]). It is worth noting that HNO exchange between ferrous GlbN and ferrous GlbN-A is likely to occur, as evidenced by the formation of Fe^II–HNO GlbN-A, which is only detected in the presence of GlbN and not when starting with pure GlbN-A (Figure 6). Competition with the distal histidine for the iron may shift coupled equilibria to favor HNO dissociation and subsequent N₂O formation. Additionally, reduction of HNO to hydroxylamine or ammonia [81] and HNO/NO⁻ reaction with NO [82] (and residual O₂) present in solution provide additional outlets that may limit the Fe^II–HNO GlbN yield and lifetime.

The {FeNO}⁸ GlbN complexes have HNO ¹H chemical shifts that cluster around ~14.8 ppm (Figures 6 and 7, Table 2), a value within the range of those observed previously in Hbs (¹H ~ 14.6–15.6 ppm) [61, 69] and upfield from that of [Fe^II–(CN)₅HNO]³⁻ (~20 ppm [83]) because of the porphyrin ring current. The ¹⁵N chemical shift of HNO is upfield from reported values (536 ppm in GlbN, compared to ~580 ppm [69]), likely affected by differences in Fe^II–HNO bonding geometry. The ¹H-¹⁵N J coupling (|¹J_NH| ~ 71 Hz) observed for ¹H¹⁵NO bound to ferrous GlbNs indicates that protonation occurs at the nitrogen and is in agreement with the splittings determined previously in Hbs (|¹J_NH| ~ 66–72 Hz) and [Fe^II–(CN)₅HNO]³⁻ (|¹J_NH| ~ 71 Hz) [61, 69, 83, 84]. In the model complex (octaethylporphyrinato)5-Me-imidazole–Fe^II–HNO (or (OEP)Fe(HNO)(5-MeIm)), the proton chemical shift and ¹H-¹⁵N splitting are slightly different: ¹H = 13.99 ppm and |¹J_NH| = 77 Hz, which may be due to a solvent effect (CDCl₃ versus H₂O) and sample temperature (253 K versus 298 K) [85]. Hydrogen bond interactions along with the heme Fe–N–H bond angle are also expected to influence HNO NMR parameters.

The ionization of free HNO in water occurs with a pK_a ~11.5 and involves a change in spin state (singlet HNO Δ H⁺ + triplet NO⁻) [82, 86]. By analogy with other metal-ligand complexes, it is expected that the pK_a value decreases when HNO is bound to iron. As a complicating factor, it is possible that NO⁻ bound to Fe^II GlbN adopts a singlet ground state [87] (singlet HNO Δ H⁺ + singlet NO⁻), which may increase its basicity above that of free HNO in water. We were able to detect the ¹H signal of HNO bound to WT Fe^II GlbN over a pH range of 7.0 to 9.2. However, the typical Fe^II–HNO yield was ~ 20% at neutral pH (Figure 6) and decreased to ~ 5% at pH 9.2 (data not shown). At basic pH, the decrease in HNO proton intensity could be due to deprotonation (partial formation of Fe^II–NO⁻), enhanced base-catalyzed hydrogen exchange dynamics [84], or decreased formation/trapping efficiency (or any combination of factors). Still, the data provide a tentative apparent pK_a ≥ 8.5 for HNO when bound to Fe^II GlbNs. This lower limit is below that of [Fe^II–(CN)₅HNO]³⁻, which yielded a pK_a ≥ 11 [84] as determined by the absence of pH-dependent changes in the HNO ¹⁷O signal intensity and chemical shift.

The mechanism of NO reduction to HNO is of particular interest. Whereas DT was capable of reducing Fe^III–NO GlbN (or Fe^III GlbN + NO₂⁻) to the Fe^II–HNO form, NADPH and the flavin bound diaphorase domain of NR produced only Fe^II–NO GlbN. Since the SO₂⁻ radical derived from DT homolysis is a one electron donor, the GlbN results suggest a sequential mechanism for HNO formation (Fe^III–NO + e⁻ → Fe^II–NO + e⁻ + H⁺ → Fe^II–HNO, Figure S10), in which the second ET step is coupled to protonation. Reduction of a {FeNO}⁷ complex to the {FeNO}⁸ state by DT has precedents. For example, [Fe^II–(CN)₅HNO]³⁻ can be produced from nitroprusside ion by sequential DT reduction at pH 10 [83]. Additionally, Poole and coworkers have demonstrated that the E. coli flavohemoglobin Hmp is capable of consuming NO (and producing N₂O) under anoxic conditions [88]. Because excess reductant was not required for the reaction, they postulated that NO reduction and NO⁻/HNO release occurred spontaneously (Fe^II–NO → Fe^III + NO⁻) [88].

Our results support that the mechanism of nitric oxide reduction in GlbNs is distinct from that of the dedicated enzyme, cytochrome P450nor. In P450nor, NADH (or borohydride) reduces the Fe^III–NO complex to “intermediate I” (Soret ~ 444 nm), presumably generating a transient Fe^II–HNO species which then undergoes protonation at the HNO oxygen [89–92]. In the presence of excess NO, intermediate I (proposed to be Fe^IV–HNOH) then reacts rapidly to produce N₂O and H₂O [89]. Notably, DT treatment of Fe^III–NO P450nor results only in formation of the Fe^II–NO state [89]. Similarly, the model HNO–heme (OEP)Fe(HNO)(5-MeIm) was produced by borohydride treatment of the {FeNO}⁶ complex (Fe^II–NO⁺ + H⁻ → Fe^II–HNO) [85]. Borohydride was also shown to be capable of reducing nitroprusside ion to [Fe^II–(CN)₅HNO]³⁻ [70]. Indeed, the robustness of hydride transfer as a path to HNO production was demonstrated by Kumar and coworkers [70]: using the HNO scavengers Ni^II tetracyanate and Fe^II N–methyl-D-glucaminedithiocarbamate, they provided evidence that borohydride, but not DT, can reduce free nitrite to HNO (HONO + H⁻ → HNO + OH⁻). The direct one-electron reduction of NO to triplet NO⁻ is highly unfavorable (estimated E°’ ~ −0.8 V vs. NHE [82, 86] at 1 M NO in aqueous solution) even when considering the use of DT as a powerful reducing agent [57]. However, at neutral pH, protonation of triplet NO⁻ to singlet HNO is thermodynamically favored. Bartberger and coworkers demonstrated a pH-dependence to NO reduction in solution and estimated that proton-coupled reduction of NO to HNO at pH 7.2 would increase the E°’ to about −0.5 V vs. NHE [86]. Farmer and coworkers have shown the electrochemical reduction of Fe^II–NO myoglobin occurs at −0.63 V vs. NHE at pH 10 [93]. Near neutral pH and at a concentration of ~1 mM, the effective reduction potential for DT is ~ −0.48 V [57], which is apparently sufficient for reduction of Fe^II–NO GlbN to the Fe^II–HNO state.

The reason that GlbN but not GlbN-A can generate HNO is not obvious, as the two proteins have comparable Fe^III/Fe^II reduction potentials in their bis–histidine forms [49], a similarity that may be expected to extend to the Fe^II–NO⁺/Fe^II–NO and Fe^II–NO/Fe^II–NO⁻ couples. Nevertheless, the differential reactivity points toward a role for GlbN in HNO production, as opposed to, for example, DT reduction of free NO followed by trapping of HNO to the iron center (the latter would be expected to occur indiscriminately in the presence of GlbN and GlbN-A). We speculate that in the low affinity Fe^II–NO state of GlbN, His70 decoordination may alter the electrostatic environment of the iron and facilitate NO reduction to HNO (Figure S10). Additionally, differences in hydrogen bonding could alter the propensity for NO reduction. For example, H-bonding to the proximal nitrogen of Fe^II–NO has been proposed to bias the electron density of the {FeNO}⁷ adduct in such a way as to promote HNO formation [94]. The intermediate Fe^II–NO GlbN complex is of interest for its ability to activate NO for reduction and is accessible for future mechanistic studies using a gentle reductant such as the flavin-containing diaphorase NR domain (Figure 8B).

Interest in HNO arises from its potential biomedical uses [95] and, relevant to cyanobacteria, as a source of N₂O [82]. The slow kinetics of Fe^II–HNO GlbN formation (Figure 5A) suggest that this chemistry is not important in vivo. Furthermore, the low reduction potential required to produce Fe^II–HNO GlbN from Fe^II–NO (E°’ ~ −0.48 V) may be beyond what is accessible physiologically [96]. As mentioned above, bona fide mono-heme NO reductases such as cytochrome P450nor bypass the thermodynamic “dead-end” {FeNO}⁷ state by using a distinct hydride transfer mechanism [89]. Although we did not observe such a reaction in vitro, we cannot rule out that Synechococcus contains a reductase capable of bypassing {FeNO}⁷ GlbN in an analogous manner. Future work will examine the plausibility and potential consequences of GlbN-mediated endogenous HNO production in cyanobacteria.

5. Conclusions

Our exploration of the chemistry of GlbN and GlbN-A has revealed interesting features related to the management of ROS/RNS. Unlike the ferrous deoxy state, in which the spontaneous PTM occurs readily to produce GlbN-A, under excess O₂ conditions, ferrous GlbN rapidly forms an oxy complex and tends to maintain its b heme in the unmodified state. These in vitro results help to explain why GlbN is primarily obtained from Synechococcus cells grown under oxic conditions whereas GlbN-A is recovered from cells grown microoxically. GlbN and GlbN-A both perform the NO dioxygenase reaction; however, heme bleaching during multiple turnovers casts some doubt on this activity being the only one of functional relevance. If NO is able to outcompete O₂ in binding to ferrous GlbN, the Fe^II–NO heme is rapidly dislodged from its cavity and HNO can be formed if a suitable reductant (e.g., DT) is available. Notably, we propose that the mechanism of HNO formation occurs through sequential ET events. NO-mediated heme dissociation and HNO production were also observed for Synechocystis GlbN. In contrast to GlbN, GlbN-A does not convert NO to HNO under similar conditions. In addition, the His117–heme covalent linkage in GlbN-A prevents Fe^II–NO heme loss and may be biologically important to prevent destructive reactions in the cytosol catalyzed by free heme. Whether the differential reactivity between GlbN and GlbN-A toward NO extends to other cyanobacterial GlbNs cannot be predicted at this time, although it is interesting that many such proteins contain a histidine at the analogous position of His117 and therefore may undergo the PTM. The heme-protein crosslink affords the ability to discriminate chemistries (HNO formation, heme loss) on the basis of the availability of a set of inhibitory heme ligands (NO, O₂, CO). Thus, GlbN illustrates a versatile system for the management of NO and other RNS and is a convenient model protein for the study of HNO chemistry.

Abbreviations:

1-D

one-dimensional

2-D

two-dimensional

DH

dehydrogenase

DPTA

dipropylenetriamine

DQF-COSY

double-quantum filtered correlation spectroscopy

DT

dithionite

ET

electron transfer

FAD

flavin adenine dinucleotide

Fd

ferredoxin

FNR

ferredoxin-NADP⁺ reductase

GlbN

Synechococcus sp. PCC 7002 or Synechocystis sp. PCC 6803 hemoglobin

GlbN-A

GlbN with His117–heme covalent attachment

GODCAT

glucose oxidase/D-glucose/catalase

Hb

hemoglobin

HNO

nitrosyl hydride

HSQC

heteronuclear single quantum coherence

MAHMA

methylamine hexamethylene methylamine

Mb

myoglobin

NOD

nitric oxide dioxygenase

NOE

nuclear Overhauser effect

NOESY

nuclear Overhauser effect spectroscopy

NR

nitrate reductase

(OEP)Fe(HNO)(5-MeIm)

(octaethylporphyrinato)5-Me-imidazole–Fe^II–HNO

PTM

posttranslational modification

RNS

reactive nitrogen species

ROS

reactive oxygen species

THB1

Chlamydomonas reinhardtii truncated hemoglobin 1

TOCSY

totally correlated spectroscopy

TrHb1

group 1 truncated hemoglobin

WT

wild-type