Copper nitrite reductase from Sinorhizobium meliloti 2011: Crystal structure and interaction with the physiological versus a nonmetabolically related cupredoxin‐like mediator

Abstract

We report the crystal structure of the copper‐containing nitrite reductase (NirK) from the Gram‐negative bacterium Sinorhizobium meliloti 2011 (Sm), together with complex structural alignment and docking studies with both non‐cognate and the physiologically related pseudoazurins, SmPaz1 and SmPaz2, respectively. S. meliloti is a rhizobacterium used for the formulation of Medicago sativa bionoculants, and SmNirK plays a key role in this symbiosis through the denitrification pathway. The structure of SmNirK, solved at a resolution of 2.5 Å, showed a striking resemblance with the overall structure of the well‐known Class I NirKs composed of two Greek key β‐barrel domains. The activity of SmNirK is ~12% of the activity reported for classical NirKs, which could be attributed to several factors such as subtle structural differences in the secondary proton channel, solvent accessibility of the substrate channel, and that the denitrifying activity has to be finely regulated within the endosymbiont. In vitro kinetics performed in homogenous and heterogeneous media showed that both SmPaz1 and SmPaz2, which are coded in different regions of the genome, donate electrons to SmNirK with similar performance. Even though the energetics of the interprotein electron transfer (ET) process is not favorable with either electron donors, adduct formation mediated by conserved residues allows minimizing the distance between the copper centers involved in the interprotein ET process.

Short abstract

PDB Code(s): 7P2F;

Abbreviations

DMSO

dimethyl sulfoxide

ED

electron donor

ET

electron transfer

NirK

copper‐nitrite reductase

Paz

pseudoazurin

PPC

primary proton channel

SC

substrate channel

SPC

secondary proton channel

1. INTRODUCTION

Rhizobia are a group of microorganisms used for the formulation of bioinoculants of legumes such as Medicago sativa (alfalfa) and Glycine max (soya) seeds. ¹ The symbiotic relationship between these microorganisms and alfalfa and soya crops relies on two finely tuned metabolic pathways: nitrogen fixation and denitrification. ² , ³ Although both metabolic pathways impact plant growth, denitrification may be counterproductive for the environment because it exacerbates greenhouse gas production. ⁴ Through the denitrification pathway, nitrate is converted into atmospheric molecular nitrogen $\left[N{O}_3^{-}\to N{O}_2^{-}\to \mathit{NO}\to N_{2}O\to N_2\left(g\right)\right]$ through several enzymatic steps under anoxic or microaerobic conditions. ⁵ The second step is carried out by nitrite reductases, which are enzymes that catalyze the proton‐coupled one‐electron reduction of nitrite to nitric oxide as $N{O}_2^{-}+2H^{+}+e^{-}\to \mathit{NO}+H_{2}O$. ⁵ Among these enzymes, copper nitrite reductases (NirKs), which are highly distributed within denitrifiers, play a key role.

Most NirKs are homotrimeric two‐domain proteins. ⁶ One of the domains harbors the copper centers, whereas the second is essential to form the macromolecular arrangement. There are two types of copper centers: one copper center of Type 1 (T1Cu), and other of Type 2 (T2Cu). T1Cu is an electron transfer (ET) center involved in intra‐ and interprotein ET processes. Based on the optical absorption properties of this center, NirKs have been classified into green, blue, and greenish‐blue. ⁶ T2Cu is the enzyme active site where the proton‐coupled nitrite reduction takes place. ⁷ , ⁸ T1Cu is connected to T2Cu through the highly conserved Cys–His bridge ⁷ , ⁸ , ⁹ and by the so‐called sensing loop. ¹⁰ The proposed reaction mechanism implies the binding of nitrite to T2Cu and the subsequent nitrite reduction by means of an electron supplied by T1Cu through the Cys–His bridge, in a process controlled by the sensing loop. ¹¹ Two proton channels, called primary and secondary proton channels (PPC and SPC, respectively), conduct the protons required for the reaction to T2Cu, with the most relevant one being SPC. ¹²

NirKs receive electrons from physiological electron donors (EDs) such as small proteins like azurins, pseudoazurins, and cytochromes, which are genetically determined. ⁵ For example, Pseudomona chlororaphis (Pc) and Alcaligenes xyloxidants (Ax) NirKs receive electrons from an azurin, ¹³ , ¹⁴ S. meliloti (Sm) and Alcaligenes faecalis (Af) NirKs from pseudoazurins, ¹⁵ , ¹⁶ and Bradyrhizobium japonicum USDA110 (Bj) and Rhodobacter sphaeroides (Rs) NirKs from Type‐c cytochromes. ¹⁷ , ¹⁸ All NirKs show a well‐conserved region above T1Cu that facilitates the interaction with external physiological EDs. Despite EDs being genetically determined in vivo, promiscuity toward non‐cognate EDs is commonly observed in vitro. ⁶

S. meliloti 2011 is one of the most exploited rhizobia for the formulation of bioinoculants for alfalfa seeds. ¹ Genome mining revealed that this bacterium harbors two pseudoazurin genes, one located at the pSymA megaplasmid together with the SmNirK gene (azu1), and a second (azu2) at the chromosome is clustered with sulfur‐related metabolism genes. ¹⁹ , ²⁰ azu1 and azu2 code for SmPaz1 and SmPaz2, respectively. Despite being co‐transcribed with SorT (a sulfite dehydrogenase) and SorU (cytochrome c), neither SorT nor SorU were able to interact with SmPaz2 in in vitro assays, and SmPaz2 function in the metabolism of S. meliloti remains unclear. ²¹ Since these proteins are located at the periplasm, they can possibly interact even though they are not metabolically related.

We have previously reported the biochemical, spectroscopic, and kinetic characterization of the green‐type SmNirK ²² and its physiological ED, greenish‐blue SmPaz1. ¹⁵ Here, we report the SmNirK crystal structure at 2.5 Å resolution, together with the kinetic (in vitro assays) and molecular properties of the interactions (in silico studies) of SmNirK with both SmPaz1 and SmPaz2. We analyze the adducts in terms of structural alignments with reported NirK–pseudoazurin complex structures and docking studies.

2. RESULTS AND DISCUSSION

2.1. Overall structure

Greenish SmNirK crystals were obtained and diffracted to 2.5 Å (Table S1). The structure was solved through molecular replacement using AfNirK (PDB ID: 1AQ8), ²³ which shares 83% amino acid sequence identity with SmNirK. The ASU contained three subunits related to each other via three‐fold rotational symmetry, constituting the protomer of SmNirK.

The overall homotrimeric structure of SmNirK (Figure 1a) is in agreement with the macromolecular size obtained from SDS‐PAGE and size exclusion chromatography. ²² The domain architecture of each monomer (Figure 1b) resembles that observed within class I NirKs resting as‐isolated structures: AfNirK (PDB ID: 5F7B), ²⁴ Achromobacter cycloclastes NirK (AcNirK; PDB ID: 6GSQ) ²⁵ and A. xylosoxydans NirK (AxNirK; PDB ID: 5ONX) ²⁶ with r.m.s.d. values of 0.38, 0.60, and 0.56 Å (Cα), respectively. Each subunit can be divided into two domains, with both domain I (residues 47–193, N‐terminal) and domain II (residues 210–375, C‐terminal) folded into a Greek key β‐barrel domain (Figure 1b). Sequence alignment of selected NirK representatives classifies SmNirK as class I based on the length of the so‐called linker and tower loops (Figure S1). These loops also clearly differ from those of the thermotolerants Geobacillus kaustophilus (Gk) ²⁷ and Geobacillus thermodenitrificans (Gt) ²⁸ NirKs, and Neisseria gonorrhoeae NirK (NgNirK), ²⁹ with the latter being a typical class II NirK. The linker loop of SmNirK is longer than that found in class II NirKs (Figure S1). Generally, the tower loop of class I NirKs adopts a random coiled structure leading to a four‐turned α‐helix, as observed in SmNirK, which extends toward T1Cu (Figure 1b), while a deletion in the corresponding tower loop in the class II NirKs results in a shortened α‐helix. ²⁷ , ²⁸ , ²⁹ Moreover, the unique extra loop observed in the GkNirK structure ²⁸ is not observed in SmNirK.

FIGURE 1.

SmNirK overall structure. (a) Top and side views of the trimeric structure. The threefold symmetry axis is indicated by a black triangle (Top view) and thick lines (Side view). (b) Domain distribution within each SmNirK subunit. Domain I (blue) harbors the two copper centers and is connected by the linker loop (red) to domain II (green). The Cys–Met loop and the tower loop located in Domain I and Domain II, respectively, play a key role in the enzyme–ED interaction and electron transfer processes. (c) Coordination around each copper center and pathways linking T1Cu and T2Cu. The amino acids in blue belong to the same monomer whereas those in green to a neighbor monomer. Other residues involved in catalysis are depicted with subindexes CAT (catalytic proton channel) and PPC (primary proton channel). Catalytically relevant occluded water molecules are in red

2.2. Type 2 copper center of SmNirK: Geometry, active pocket site, and proton channels

The first coordination sphere of the catalytic T2Cu is composed of three His(N^ε2) and a water molecule (2.01 ± 0.01 Å) in a distorted tetrahedral geometry (Figure 1c). His136 (2.05 ± 0.02 Å) and His171 (2.04 ± 0.03 Å) are provided by domain I, whereas His342 (2.04 ± 0.01 Å) belongs to domain II of an adjacent subunit, as usually observed in NirKs. Like in all NirKs reported so far, with the exception of Thermus scotoductus SA‐01 NirK (TsNirK), ³⁰ the active site is also constituted by the outer first coordination sphere ligands Asp_CAT (Asp134), His_CAT (His291) and Ile/Val_CAT (Ile293). ⁶ , ¹² , ³¹ Like in other class I NirKs, SmNirK has the sequence TRPH_CATL around the essential His_CAT residue (Figure S1).

Two water molecules are conserved, the T2Cu ligand water (W_Ap) and a water molecule bridging Asp_CAT and His_CAT (W_Br) (Figure 1c), which are involved in the catalytic mechanism. ²⁴ , ³² , ³³ Two other residues that play an active role in the catalytic cycle, Glu315 and Thr316, are connected to His_CAT by occluded water molecules (Figure 1c). ²⁴ , ³² His136, hydrogen‐bonded to Glu315, is located at the end of the SmNirK sensing loop, which is thought to transmit information about the T2Cu status for electron shuttling through the Cys–His bridge. ¹⁰ , ³⁴ In the same way, Thr316 is H‐bonded to His_CAT. These residues are catalytically equivalent to Glu279 and Thr280 in AfNirK and conserved amongst class I NirKs (Figure S1). On the other hand, Class II NirKs show the combination of Gln‐Thr/Ser residues and Ser residues at this position, which confers a lower level of activity than that of Thr‐containing NirKs. ²⁹ , ³⁵

Similar to that observed in other two‐domain NirKs, T2Cu is accessible through the substrate channel (SC) (Figure 2a,b). The SC is formed by polar/charged residues of two adjacent subunits H‐bonded to water molecules and also by several hydrophobic residues surrounding the T2Cu site (Figure 2b). ⁶ , ¹² , ³¹

FIGURE 2.

SmNirK proton channels. (a) Primary and secondary proton channel (PPC and SPC, respectively) accessed from the bulk solution. SPC coincides with the substrate channel (SC). SPC/SC (red arrows in bottom view and red circle in side view) and PPC (magenta circles in bottom view) are indicated. Subunits are shown in different colors. (b) SC/SPC mouth architecture in resting as‐isolated structures of rhizobial NirKs. The bulky Phe348 residue in SmNirK defines a narrow mouth compared to that observed in Br ^2DNirK (PDB ID: 6ZAS), which has a less bulky Ala residue in that position. Waters within the channel are indicated as spheres. (c) Comparison of proton channels in SRX structure of SmNirK (blue or green depending on the subunit) with XFEL‐FRIC structure of Br ^2DNirK (cyan). SPC walls architecture in SmNirK: Ile293 (Ile_CAT), Ile336, Ala338, Val340, Leu344, Phe348, Ala353, Ala354, and His355 from subunit A (transparent yellow surface); Asn132, Asp134 (Asp_CAT), Leu142, Gly145, Ala146, Val149, His171, Ala173, Pro175, Gly176, Val178, and Pro179 from subunit B (transparent orange surface). Dashed lines indicate the SPC path. Amino acids related to T2Cu active site are indicated, as well as all the residues involved in the PPC and SPC (SmNirK numbering). Variations on the SPC/SC mouth entry among different NirKs are indicated: Phe348 and Val149 (SmNirK numbering). SmNirK (green) and Br ^2DNirK (cyan) SPC waters are indicated. Occluded water molecules that connect His_CAT with Glu–Thr residues are indicated in red

Nitrite reduction at T2Cu requires the consumption of two protons, which has been intensively investigated in two‐domain NirKs. ¹² , ³¹ , ³³ , ³⁶ Two distinct proton channels named primary (PPC) and secondary (SPC) have been proposed to transport these protons, with SPC identified as the most relevant. ¹² SPC also coincides with the SC. The SmNirK crystal structure shows the two proton channels (Figure 2c), as also observed in other class I NirKs. ⁶ , ¹² , ³¹ The PPC involves the solvent‐exposed His296 (His_PPC) and the T2Cu ligand His136, showing the same architecture observed in class I NirKs. ⁶ , ¹² The sequence portion connecting His_CAT with His_PPC resembles that of class I NirKs (H_CATLI_CATGGH_PPC) and differs from that in class II NirK which lacks His_PPC.

The SPC identified in the AxNirK structure (Asp_CAT‐wat‐wat‐Ala131‐Asn90‐Asn107) is not completely conserved in SmNirK (Figure 2c), as the channel in SmNirK shows a valine residue (Val149) instead of the relevant asparagine (Asn107, AxNirK numbering). ¹² In the highly active green AfNirK and AcNirK, Asn107 is replaced by Glu and Gln, respectively, while an Asp residue was observed in the thermotolerants GkNirK and GtNirK. Hydrophobic residues substituting Asn107 were also observed in the two‐domain rhizobial NirKs from Bradyrhizobium sp. ORS 375 (Br ^2D NirK) and B. japonicum USDA110 (BjNirK), both showing turnover numbers lower than those of AfNirK and AcNirK. ¹⁷ , ³⁷ In line with this observation, SmNirK has a hydrophobic Val residue at this position and, as reported in our previous work, MV‐dependent activity of SmNirK shows only 12% of the AxNirK maximum turnover. ²² This correlation seems consistent as the same position is occupied by hydrophobic Leu residues in both Br ^2D NirK ³⁷ and BjNirK, ¹⁷ which show less than 10% of the AxNirK maximal activity. ¹⁷ , ³⁷ A synchrotron radiation X‐ray (SRX) structure of Br ^2DNirK (PDB ID: 6THF) showed an increased number of waters within the SC compared to our SmNirK SRX structure, conserving the T2Cu‐coordinated water. ³⁷ More recently, it has been suggested that the low level of activity of Br ^2DNirK is due, in part, to increased solvent accessibility and to two strongly T2Cu‐coordinated waters, as observed in the structure obtained by free‐electron laser X‐ray crystallography, which is free from radiation‐induced chemistry (XFEL‐FRIC, PDB ID: 6ZAS). ³⁸ The channel mouth is delimited by a bulky phenylalanine residue in SmNirK (Phe348) (Figure 2b), as observed in AcNirK, AfNirK, and AxNirK structures, but a nonbulky Ala residue occupies that position in Br ^2DNirK. The SC in SmNirK is clearly less populated in waters, like in the channels of AcNirK, AfNirK, and AxNirK. Although SmNirK shows similar structural features in the architecture of the channel compared to the highly active NirKs, and do not have the level of hydration of Br ^2DNirK, other factors might also influence the activity of SmNirK. High‐resolution damage‐free SmNirK structures, as well as mutating the channel residues of the enzyme, will be required to unveil those factors.

2.3. Relationship between structural and optical absorption features of Type 1 copper center of SmNirK, and T1Cu–T2Cu intramolecular electron transfer pathway

T1Cu is located at the top of domain I in the N‐terminal region (Figure 1b). T1Cu is coordinated by two His(N^δ1) residues (His89 and His139), Cys172(S^γ), and Met186(S^δ), forming a distorted tetrahedral geometry as typically observed in NirKs and some pseudoazurins (Figure 1c). Relevant bonding distances and angles, including those observed in others NirKs and some cupredoxins, are shown in Table S2 for comparison. Structural parameters of the T1Cu of SmNirK are similar to those observed in other green‐type NirKs, with the most significant difference being a longer Cu—S(Cys) bond and a shorter Cu—S(Met) bond. It has been suggested that the length of the Cu—S(Met) bond influences the reduction potential of T1Cu E _m values, and the shorter the length of bond, the lower the E _m. This observation is in line with the structural results in SmNirK, which showed a T1Cu E _m (+224 mV) lower than those obtained in two closely related green‐type NirK (AcNirK: +240 mV; AfNirK: +260 mV). ¹⁵ , ³⁹ , ⁴⁰ Substitution of the Met ligand by a Gln, which results in NirK variants with Cu—O (Gln) bond distances shorter than that of the Cu—S(Met) bond, also showed this tendency (+240 mV vs. +173 mV in AxNirK and M144Q variant, respectively), ³⁴ but it cannot be excluded that the E _m shifting might due to the different nature of the ligand. Nevertheless, although the lower E _m of SmNirK seems to be associated with a Cu—S(Met) shortening, it is important to note that the diffraction data was not exempt from radiation damage, which deserves further investigation.

UV–vis spectroscopic properties of T1Cu are inherently related to the geometry of this center. Typical absorption features of T1Cu are mainly due to Cys(S^γ) → Cu charge‐transfer (CT) transitions, namely the S(Cys)π → Cu 3d _x2‐y2 (~600 nm), and S(Cys)σ → Cu 3d _x2‐y2 (~450 nm) transitions. Spectra of green NirKs also present a weaker shoulder at ∼ 400 nm assigned to the Met(S^δ) → Cu CT transition ⁴¹ and a broad band centered at ~750 nm assigned to d–d absorptions. The ε_σ/ε_π ratio of T1Cu determines the color of NirK. A comparison of UV–vis features of T1Cu in SmNirK, as well as those of its physiological ED SmPaz1 ¹⁵ and the as‐purified non‐cognate ED SmPaz2, is shown in Table S3 and Figure S2a. SmNirK had an ε_σ/ε_π ratio of 1.2, in line with a green T1Cu center. The His₂N^δ1–Cu–MetS^δ angle is ~130°, as observed in structures of the classical green AfNirK and AcNirK (Table S2). On the other hand, the blue types Br ^2DNirK and AxNirK showed angles of ~110° in resting structures determined by XFEL‐FRIC and serial femtosecond rotational crystallography (SF‐ROX), respectively. ²⁶ , ³⁸ Additionally, the side‐chain conformation of the Met ligand (Met186) (Figure S2c) is in a “trans (anti)” conformation, in line with that observed in the green‐type AfNirK and AcNirK. ²⁴ , ²⁵ This is in contrast to Br ^2DNirK, which showed a “gauche” conformation similar to AxNirK, both as‐prepared damage‐free structures. ²⁶ , ³⁸ The significance of the Met conformation on the UV–vis spectra of NirKs has not been discussed to the best of our knowledge, but the current information seems to indicate that the conformation of the Met ligand would be important to modulate the electronic properties of the T1Cu center.

The dihedral angle (φ) between the His₁(N^δ1)–Cu–His₂(N^δ1) and Met(S^δ)–Cu–Cys(S^γ) planes was proposed as another factor that influences the UV–vis spectrum of cupredoxin T1Cu centers (Figure S2b). SmNirK showed a φ = 68°. Green NirKs (~66°) have lower φ values than blue NirKs (~74°) in damage‐prone structures. The recently reported damage‐free structures enabled the identification of discrepancies between resting structures of the same protein (AcNirK: 66° in a SRX structure ⁴² and 58° in a SF‐ROX structure ²⁵ ). Thus, the influence of this structural property should be revisited in NirKs.

Other structural features associated with the two links between T1Cu and T2Cu did not show significant differences with that observed in all other two domain NirKs, namely, the Cys–His bridge and the sensing loop that have been proposed to be involved in intramolecular ET reactions either as conduit or triggering electron delivery (Figure 1c).

2.4. Interaction of SmNirK with electron donors

SmNirK receives electrons from its physiological ED SmPaz1 ¹⁵ which is encoded by the azu1 gene located in the denitrification gene cluster within the pSymA megaplasmid. ¹⁹ S. meliloti 2011 contains another cupredoxin‐like ED‐mediator encoded in a sulfur‐metabolism operon located on the chromosome. ²⁰ The azu2 gene (Smc04047) codes for SmPaz2, which, despite being co‐transcribed with a sulfite dehydrogenase gene, was unable to deliver electrons directly to the enzyme. ²¹ In line with a previous characterization reported by Maher and co‐workers, ⁴³ recombinant SmPaz2 is a green colored protein (Table S3 and Figure S2a) with E _m  = 245 ± 5 mV determined by cyclic voltammetry (Figure S3). Despite SmPaz1 and SmPaz2 sharing low sequence identity (46%), the fact that the area surrounding T1Cu is highly conserved, in conjunction with theoretical pI _SmNirK ~ 6.1, pI _SmPaz1 ~ 9.6 and pI _SmPaz2 ~ 8.2, led us to speculate that SmPaz2 may act as an alternative in vitro ED to SmNirK. This hypothesis was confirmed from comparative kinetic assays using either SmPaz2 or SmPaz1 as redox partner of SmNirK under anaerobic conditions, which confirmed that SmPaz2 is indeed able to deliver electrons to SmNirK when nitrite was added (Figure 3a). The oxidation rates of the colorless reduced pseudoazurin is of the same order of magnitude for both mediators (SmPaz1, λ_MAX = 597 nm; SmPaz2, λ_MAX = 590 nm), but the SmNirK‐SmPaz2 couple was ~2 times faster (Figure 3a).

FIGURE 3.

Kinetic assays. (a) Electron donation progress curves of pseudoazurins to SmNirK. Green, SmPaz2 (ε₅₉₀ = 1.51 mM⁻¹ cm⁻¹); blue, SmPaz1 (ε₅₉₇ = 3.17 mM⁻¹ cm⁻¹). Absorbance values were converted to μmol of protein. (b) Currents recorded at different nitrite concentrations by chronoamperometry. The solid line was obtained by least squares analysis assuming a Michaelis–Menten model

We also evaluated the catalytic performance of SmNirK using SmPaz2 as mediator by chronoamperometry following a procedure reported for the SmNirK‐SmPaz1 couple. ¹⁷ The results, analyzed with a Michaelis–Menten model, yielded i _max = 3.79 ± 0.07 μA and K _m ^app = 320 ± 20 μM nitrite, which indicates, in line with the solution kinetics, that SmPaz2 can also act as a suitable in vitro ED to SmNirK (Figure 3b). Thus, both SmPaz1 and SmPaz2 are potentially applicable to the development of nitrite biosensors.

Whether these findings also occur in vivo cannot be concluded with the present data. SmNirK and SmPaz1, as well as SmPaz2, SorT, and SorU are located at the periplasm. Thus, it is reasonable to hypothesize that under certain physiological conditions, both metabolic pathways (sulfite oxidation and denitrification) might coexist. In this context, SmPaz2 could act as an alternative in vivo ED to SmNirK, which should be confirmed by metabolomic studies.

2.5. Intermolecular electron transfer processes involving SmNirK with electron donors

To the best of our knowledge, only two structural NirK‐Paz adducts have been reported. ⁶ , ⁴⁴ These are the NMR structure of AfNirK–AfPaz complex (PDB ID: 2P80) and the X‐ray crystal structure of AxNirK–HdPaz complex (PDB ID: 5B1J). Co‐crystallization of the SmNirK with the respective EDs were unsuccessful. Therefore, the above mentioned adducts were used as models to study the SmNirK–SmPaz1/2 interactions. Since the structure of SmPaz1 is not available, we previously built a structural model of the protein using the Swiss‐Model automated server ⁴⁵ using pseudoazurin from A. cycloclastes (AcPaz) as template (79% identity, PDB ID: 1BQK). We then analyzed the most favorable transient structural arrangements for both SmPaz1 and SmPaz2 to allow T1Cu_Paz → T1Cu _SmNirK ET during catalysis.

Like in all NirKs (Figure S4, Left panel), ⁶ , ⁴⁴ the putative entry point of electrons in SmNirK is surrounded by a set of hydrophobic residues, which are enclosed by a polar region and an outer negatively‐charged region constituted by Asp and Glu residues on the edge of the binding site (Figure 4, left panels). The relevant Trp180 and Tyr239 (Trp138 and Tyr197, AxNirK numbering), surrounded by Asp and Glu residues, ⁴⁶ are also observed in the T1Cu surface area of SmNirK (Figure 4b, left panel). Kinetic studies of variants of both NirK and pseudoazurins have shown that charged residues are involved in the affinity of the protein complex. ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ All these amino acids are well conserved in SmNirK, with the exception of Glu204 (AfNirK numbering) which is replaced by Ala240 in SmNirK (Figure 4b, left panel).

FIGURE 4.

The intermolecular electron transfer process. T1Cu surface area analysis. “Open‐book unfolding” of enzyme–electron donor complex indicating: (a) the conservation of residues on surfaces covering T1Cu centers in both two‐domain NirKs (left) and pseudoazurins (right) and, (b) surface electrostatic analysis depicted by coulombic surface coloring in UCSF Chimera: SmNirK (left); SmPaz1 (homology model constructed by Swiss‐Model) and SmPaz2 (PDB ID: 3TU6) (pseudoazurins, right). The potential contours are shown on a scale from −10.0 k B T (red) to +10.0 k B T (blue). The open‐book unfolding was made after structural alignment based on the X‐ray crystal structure of the AxNirK–HdPaz complex (PDB ID: 5B1J). Amino acids that are essential for the AfNirK–AfPaz interaction, as shown by mutagenesis studies, are indicated. The amino acids that are not preserved compared to the AfNirK–AfPaz complex are indicated in red

The counterpart in SmPaz1 and SmPaz2 is formed by a ring of positively charged residues surrounding the small hydrophobic patch in the T1Cu surface (Figure 4, right panels), as observed in the pseudoazurins from A. faecalis (AfPaz; PDB ID: 1PAZ) and Hyphomicrobium denitrificans (HdPaz; PDB ID: 3EF4; Figure S4, right panel). SmPaz1 shows the corresponding four relevant Lys residues, but Lys77 (AfPaz numbering) is not observed in SmPaz2, which shows a Gln residue in that position (Gln79; Figure 4b, right panel).

2.6. Structural alignment and docking studies of SmNirK‐ED adducts

The structures of the adducts AfNirK–AfPaz (PDB ID: 2P80) ⁴⁴ and AxNirK–HdPaz (PDB ID: 5B1J) ⁶ revealed different alternatives for electron delivery in which the solvent‐exposed His81 ligand to copper is the starting point of the ET pathway in both pseudoazurins. Two coupling pathways were proposed from AfNirK–AfPaz studies, one involving two through‐space jumps, (His81–Met84) _AfPaz–His145 _AfNirK, and the other involving His81 _AfPaz–(Pro139–Pro138–Ala137–Cys136) _AfNirK. ⁴⁴ Alternatively, the X‐ray crystal structure of the AxNirK–HdPaz complex showed a direct interaction between the His81 ligand of HdPaz and the O‐carbonyl of Ala86 of AxNirK, which would imply that the preferred ET pathway depends on the interacting partners. ⁶

The hypothetical arrangement of SmNirK with both EDs based on the AfNirK–AfPaz (d _Cu‐Cu = 15.5 Å) complex structure is shown in Figure 5a. The putative ET pathways in both complexes analyzed by HARLEM algorithm ⁵² are (His83–Met86) _SmPaz2–His181 _SmNirK and His83 _SmPaz2–(Pro175–Pro174–Ala173–Cys172) _SmNirK (d _Cu‐Cu = 14.8 Å). A similar conclusion was drawn for the SmNirK‐SmPaz1 complex (d _Cu‐Cu = 15.1 Å). More details related to both complexes, highlighting the residues involved in protein–protein interaction, are shown in Figure S5a. Trp180 from the Cys–Met loop and Tyr239 from the tower loop (SmNirK numbering) were important for the interaction of AxNirK with the physiological redox partner (Trp138 and Tyr197). ⁴⁶ These residues seem to be involved in hydrophobic contacts with residues near Lys11 of SmPaz2 in the SmNirK‐SmPaz2 complex, though they are not directly involved in the ET process.

FIGURE 5.

The intermolecular electron transfer process: in‐silico models for T1Cu _SmPaz2 → T1Cu _SmNirK ET. Top panel shows a close‐up view of interacting surfaces and the putative ET pathways are indicated in ball‐stick. The jumps are indicated by dashed lines. Amino acids taking part in the ET process are indicated in black. Bottom panel shows a top‐view of the complex. The homotrimeric structure of SmNirK is depicted in yellow. The tower loop and the Cys–Met loop (both indicated in red) take part in the interaction with SmPaz2 (green). (a) Molecular arrangement obtained after structural alignment of SmNirK and SmPaz2 structures on the AfNirK–AfPaz complex structure (PDB ID: 2P80). (b) SmNirK–SmPaz2 complex arrangement obtained using the AxNirK–HdPaz complex (PDB ID: 5B1J) as template. (c) Model obtained after docking performed by ClusPro

The SmNirK–SmPaz2 complex based on AxNirK–HdPaz structure is shown in Figure 5. In this case, the most likely intermolecular ET pathway is His83 _SmPaz2–(Thr128–Leu129–Gln130–His131) _SmNirK (d _Cu‐Cu = 15.8 Å), as proposed from the structure of the AxNirK–HdPaz adduct (d _Cu‐Cu = 15.9 Å). The amino acids interacting between partners are indicated in Figure S5b. Our results also indicated an alternative ET pathway formed by His83 _SmPaz2–(Thr128–Glu83–His131) _SmNirK, in which Trp180 and Tyr239 may be involved in hydrophobic interactions, but involving a region of SmPaz2 different from that observed in the hypothetical model based on AfNirK–AfPaz complex. Similar results were obtained for the interaction with SmPaz1 (d _Cu‐Cu = 16.0 Å).

To obtain further insight on SmNirK–ED interactions, we performed docking studies using the SmNirK structure, the X‐ray structure of SmPaz2 (PDB ID: 3TU6), and the above‐mentioned homology model of SmPaz1. The arrangement of the docked SmNirK–SmPaz2 adduct (Figure 5c) differs from those obtained through structural alignments, with the T1Cu_Paz in a position in‐between those observed in the AfNirK–AfPaz and AxNirK–HdPaz complex structures. Therefore, this is a better structural arrangement characterized by shorter Cu‐Cu distances (d _Cu‐Cu = 14.2 Å and 14.9 Å for SmPaz2 and SmPaz1, respectively) for a more efficient ET between SmNirK and SmPaz2, relative to those obtained from structure alignment. This docking configuration also predicted that the most favorable putative ET pathways should be (His83–Pro82) _SmPaz2–(Met177–His181) _SmNirK for the SmNirK–SmPaz2 adduct (Figure 5c). The same residues are involved in the ET pathway of SmNirK–SmPaz1 complex (not shown), but with a different topology.

2.7. Theoretical ET rates analysis of SmNirK‐ED adducts

Since the putative intermolecular ET reaction in both SmNirK‐SmPaz1/2 is uphill [E _m (T1Cu_NirK) < E _m (T1Cu_Paz)] for both EDs (Table S4), we evaluated the ET rate constants (k _ET) with the equation proposed by Moser et al. for endergonic processes ⁵³ , ⁵⁴

$$\mathit{log}{k}_{\mathit{ET}}^{\mathit{\text{ender}}}=15-0.6d-3.1\frac{{\left(-∆G°+\lambda \right)}^2}{\lambda }-\frac{∆G°}{0.6}$$ (1)

where ΔG° is the driving force in eV, d is the distance between the interacting redox centers in Å, and λ is the reorganization energy. For endergonic ET processes, d is of utmost importance, as a lower distance may compensate unfavorable driving forces (ΔG° > 0). The parameter ΔG° was calculated as $\mathrm{\Delta }G°=-n_e\left[E_m\left(\mathrm{T}1{\mathrm{Cu}}_{\mathit{\text{NirK}}}\right)-E_m\left(\mathrm{T}1{\mathrm{Cu}}_{\mathit{Paz}}\right)\right]$, whereas the d‐values were obtained from the structural models described above. As λ_T1Cu for SmNirK and both SmPaz1/2 are unknown, we used λ_T1Cu = 0.76 eV obtained from Pseudomonas aeruginosa azurin. ⁵⁵ Generally, reorganization energies for metal cofactors included in protein matrices are in the range 0.6–0.9 eV. E _m (mV) and d (Å) parameters for each system, along with the calculated k _ET values, are shown in Table S4. The k _ET‐values obtained with all the models are in the range of 10³–10⁴ s⁻¹, which confirms that the intercenter distance d obtained in all cases may compensate the unfavorable driving force. Remarkably, the k _ET values obtained by docking suggest that SmPaz2, if properly docked, is a more efficient ED to SmNirK, which is in contrast to that obtained by structural alignment. Altogether, this demonstrates that the noncognate SmPaz2 ED is a suitable in vitro redox partner to SmNirK, despite that both proteins are likely not physiologically related.

As observed for all NirK‐Paz complexes, interprotein ET between SmNirK either with SmPaz1 or with SmPaz2 occurs despite the thermodynamic unfavorable driving force (Figure 3a). The fact that the predicted interprotein ET rate (k _ET) is faster than the catalytic turnover (k _cat), not only in SmNirK but also in other NirKs (10³–10⁴ s⁻¹ vs. 10¹–10² s⁻¹), ¹¹ suggests that this might not be the global rate‐limiting event. Thus, catalytic turnover in NirK seems to be determined by other factors such as substrate‐binding rates, proton transfer rates, protein complex formation, protein–protein diffusion steps and structural surface rearrangements, which could considerably slow down the global rate of the process. ⁵⁶ , ⁵⁷

3. CONCLUSIONS

This is the first crystal structure reported for a green NirK from rhizobia, which are microorganisms widely used in agriculture. Particularly, SmNirK plays a key role in symbiosis with roots of alfalfa plants. Although SmNirK shows the main spectroscopic and structural features of green class I NirKs, certain structural differences were observed that are important for catalysis. The principal difference from a structural point of view is in the SPC, which is proposed to be the most relevant proton provider for NirKs to reduce nitrite at T2Cu. This is likely one of the factors that determine the lower activity of SmNirK compared to classical Class I NirKs reported so far. However, it is important to note that efficient regulation of the activity (lower activity in this case) could be a metabolic advantage. Rhizobial nodules are highly regulated in number in the roots of legumes. NirK activity is actively regulated in endosymbiont metabolism, since fine‐tuned denitrification and nitrogen fixation in rhizobacteria results in efficient, active nodules.

The two water molecules in SmNirK that take part in the catalytic mechanism, the water–T2Cu ligand and the bridging water, are conserved like in all resting NirK structures. SmNirK had shorter Cu‐S(Met) and longer Cu‐S(Cys) bonding distances at T1Cu than those determined in other green NirKs, but they do not significantly influence the UV–vis features of the enzyme.

Although SmPaz2 is not related to the denitrification pathway of S. meliloti 2011, this cupredoxin ED was able to deliver electrons to SmNirK with an efficiency similar to that of the physiological ED SmPaz1, which was observed in both homogenous and heterogeneous media. The EDs share low sequence identity, but the areas related to SmNirK interactions are highly conserved. Thus, structural complementary is a key factor for transient adduct formation in NirKs. The ET with both EDs is an energetically uphill process; however, this penalty is overcome with shorter T1Cu_Paz–T1Cu _SmNirK distances, which was confirmed by complex structural alignment and docking studies. Additionally, although the residues involved in the ET pathways are the same, the atoms involved in the process depend on the nature of the adduct.

4. MATERIALS AND METHODS

4.1. Cloning and expression of the proteins

S. meliloti 2011 DNA extraction was performed as previously reported. ²² The S. meliloti 2011 azu2 gene (locus tag smc04047) ²⁰ was amplified using oligonucleotides (Fw: 5′‐CATATGAAGACAAAGATGATGC‐3′; Rv: 5′‐GAGCTCTCAGCCGCCGCTCTC‐3′) which include the NdeI and SacI restriction sites. Amplification was performed using Pfu DNA polymerase (Genbiotech) according to the manufacturer's instructions. PCR was performed with 20 pmol of each primer, 1 ng of DNA, 1 unit of Pfu DNA polymerase, 2 μl of PCR reaction buffer, 0.2 mM dNTPs, 1 mM MgSO₄ and 0.5% vol/vol DMSO in 20 μl reaction volume. The PCR reaction was performed using the following program: 5 min at 94°C, 30 cycles of 1 min at 94°C, 30 s at 55°C, and 1 min at 72°C, and a final elongation at 72°C for 10 min. The azu2 gene was cloned into pJET2.1 blunt (Thermo Fisher) and subcloned into pET22b(+) expression vector (Novagen) to obtain p22SA2 expression constructs. The construct was maintained in Escherichia coli TOP10 cells (Invitrogen) at −80°C. DNA sequences were verified by using the Sanger method. ⁵⁸ The overexpression and purification of SmNirK and SmPaz1 were performed as previously reported. ¹⁵ , ²² Overexpression of the azu2 gene from S. meliloti 2011 was achieved by introducing p22SA2 into E. coli BL21‐Gold (DE3) (Agilent Technologies). The transformed strain was grown aerobically at 37°C overnight with agitation (200 rpm) in Lysogeny broth containing 100 μg ml⁻¹ ampicillin as a starter culture. Expression was performed by inoculating ZYP505 media ⁵⁹ containing 100 μg ml⁻¹ ampicillin with 1% (vol/vol) of starter culture and grown at 20°C for 24 hr (200 rpm). To this high‐density culture, CuSO₄ (0.4 mM) was added and maintained under the same conditions for 1 hr at 100 rpm. Finally, the copper‐fed culture was induced with 250 μM Isopropyl‐β‐d‐1‐thiogalactopyranoside (IPTG), and protein expression allowed to continue for 3 hr (20°C, 100 rpm). Cells were harvested through centrifugation and resuspended in 20 mM Tris–HCl (pH 7) buffer. Expression levels were evaluated using SDS‐PAGE analysis with prestained MRP 2‐105K protein standards (Genbiotech) as molecular mass markers and stained using Coomassie brilliant blue R‐250. ⁶⁰

A cell suspension (0.1 g wet weight ml⁻¹) was disrupted by sonication. The crude extract was recovered by centrifugation at 25,000× g for 1 hr and dialyzed overnight against 20 mM Tris–HCl buffer (pH 7) supplemented with 200 μM CuSO₄ and then centrifuged at 25,000× g for 1 hr. SmPaz2 (pI 7.9) from the crude extract was purified in two chromatographic steps. The crude extract was applied to a cation exchange column (SP Sepharose Fast Flow, 1.6 × 12.5 cm, GE Healthcare) equilibrated in 20 mM Tris–HCl buffer (pH 7) and eluted with 600 ml of 0–500 mM linear gradient of NaCl in equilibration buffer. Deep green fractions containing SmPaz2 were pooled and concentrated by ultrafiltration (10 kDa MWCO). Fractions of 400 μl were loaded onto a Superdex 200 column (1.5 × 42 cm, GE Healthcare) and eluted with 20 mM Tris–HCl buffer (pH 7) containing 200 mM NaCl. The highly pure green SmPaz2 fractions were pooled and concentrated to approximately 20 mg.ml⁻¹ in 20 mM Tris–HCl (pH 7) and stored at −80°C.

4.2. Protein quality methods and biophysical assays

Protein concentration was determined using the Bradford method with bovine serum albumin as standard. ⁶¹ Protein copper content was determined by the biquinoline assay. ³⁰ , ⁶²

Molecular masses of as‐purified proteins were estimated by gel filtration chromatography (Superdex 200 HR 10/30 column, GE Healthcare) connected to an FPLC (Akta Basic, GE Healthcare). The column was equilibrated with 20 mM Tris–HCl buffer (pH 7) containing 200 mM NaCl and calibrated with ferritin (440 kDa), aldolase (158 kDa), conalbumin (75 kDa), carbonic anhydrase (30 kDa) and ribonuclease A (13.7 kDa). Isocratic elution was performed at 0.5 ml min⁻¹ with detection at 280 nm. Absorption spectra were recorded on a Shimadzu UV‐1800 UV–vis spectrophotometer.

4.3. SmNirK crystallization and structure determination

Sitting‐drop vapor‐diffusion screening crystallization trials yielded green SmNirK crystals in several conditions within 1 week at 16°C. Single crystals grew in 2 μl drops consisting of equal volumes of 12 mg ml⁻¹ SmNirK and reservoir solution (0.2 M MgCl₂ 20% wt/vol PEG 3350). X‐ray diffraction data were collected at Diamond Light Source synchrotron (UK) on beamline I04‐1 (0.9282 Å) at 93 K. Data were processed using autoPROC, ⁶³ with indexing and integration using XDS ⁶⁴ and POINTLESS, ⁶⁵ with intensities scaled and merged using AIMLESS ⁶⁶ from the CCP4 suite of programs. ⁶⁷ Molecular replacement was performed using PHASER ⁶⁸ with A. faecalis NirK (AfNirK, PDB ID: 1AQ8) as the search model. Refinement was performed through iterative cycles of manual model building in COOT ⁶⁹ and refinement using REFMAC5. ⁷⁰ Structures were validated using programs within the CCP4 suite. ⁶⁷ Ramachandran distribution gave 97% in the favored region, with 3% in the allowed regions and no outliers. Structure factors and model coordinates have been deposited in the protein data bank under accession number 7P2F.

4.4. Bioinformatics

Multiple sequence alignment was performed with ClustalW ⁷¹ with the aligned sequences illustrated using ESPript. ⁷² Figures of the SmNirK crystal structure were generated using UCSF Chimera. ⁷³ Tunnels and pockets in the SmNirK crystal structure were detected using CASTp. ⁷⁴

Since the X‐ray structure of SmPaz1 is not available, we performed a protein structure homology modeling using the Swiss‐Model automated server. ⁴⁵ The template structure used was A. cycloclastes pseudoazurin (AcPaz, PDB ID: 1BQK) of which the amino acid sequence shows about 79% identity with the mature SmPaz1.

Adduct structural alignments were performed using UCSF Chimera. For these analyzes, both X‐ray crystal ⁶ and NMR structures ⁴⁴ of NirK‐Paz were used as templates.

To investigate the SmNirK–mediator (SmPaz1 or SmPaz2) surface interaction, protein–protein docking was applied by using the ClusPro that performs rigid‐body docking. ⁷⁵ SmNirK subunit B was used as receptor model. The ligand models were the reported crystal structure of SmPaz2 (PDB ID: 3TU6) ⁴³ and the Swiss‐Model structure generated for SmPaz1. The most favorable models for each SmNirK–pseudoazurin complex were evaluated based on the relative positions and orientations of the two docked proteins. Choice of the final models was made on the basis of the lowest energy using the electrostatic favored ClusPro score, the solution with the highest cluster members, and the lowest distance between the T1Cu _SmNirK‐T1Cu_Paz, and further validated by visual inspection of the contacts at the protein–protein interface.

The best ET pathway analysis between the ED metal center and T1Cu _SmNirK center was performed using the pathways module as implemented in HARLEM. ⁵² The nonbonding and hydrogen bonding decays were 1.7 Å⁻¹, and the covalent bond decay was 0.6 Å⁻¹. The choice of the best ET pathway was based on couplings.

4.5. Enzyme‐mediator continuous kinetic assay

In a continuous kinetic assay monitored spectroscopically, the reduced forms of SmPaz1 and SmPaz2 were used as EDs, respectively. A pre‐reaction mixture containing 30 mM MES–CAPS–Tris buffer pH 6.0, 50 μM pseudoazurin, and 50 nM SmNirK in a total volume of 1 ml, was maintained in a septa‐sealed cuvette under gentle argon flow for 30 min. This mixture was reduced with sodium dithionite (ca. 2.5 μl 200 mM dithionite solution in 0.8% NaHCO₃). Once pseudoazurin was stoichiometrically reduced, the reaction was started by the addition of sodium nitrite solution (10 μl of 100 mM). The reoxidation of the ED was followed at λ_max for each mediator (597 nm for SmPaz1, and 590 nm for SmPaz2). A negative control was performed with no enzyme addition. The initial rates ($v_o$) were obtained by nonlinear curve fitting (Equation 2; one/two‐phase exponential association equation) of each anaerobic reoxidation time course data set using the OriginPro 2018 software suite.

$$A\left(t\right)=A_o+A_1\left(1-e^{\raisebox{1ex}{$-t$}\!\left/ \!\raisebox{-1ex}{${\tau }_1$}\right.}\right)+A_2\left(1-e^{\raisebox{1ex}{$-t$}\!\left/ \!\raisebox{-1ex}{${\tau }_2$}\right.}\right);v_o=\frac{A_1}{{\tau }_1}+\frac{A_2}{{\tau }_2}$$ (2)

4.6. Electrochemistry of SmPaz2‐SmNirK

Cyclic voltammetry experiments were performed with a TEQ_04 potentiostat/galvanostat (nanoTeq, Argentina). The data were collected and analyzed using software provided by nanoTeq and OriginPro 2018. A conventional three‐electrode electrochemistry configuration glass cell was used, with a 3 mm diameter polycrystalline gold disk electrode (BASi), a platinum sheet and a saturated Ag/AgCl reference electrode (197 mV vs. SHE at 298 K) as working, auxiliary and reference electrodes, respectively. We proceeded as previously described to determine the reduction potential for SmPaz1. ¹⁵ Briefly, the gold electrode was polished with 1.0, 0.3, and 0.05 μm water alumina slurry (Buehler) and then sonicated in ultrapure water. The polished electrode was subsequently immersed in 1 mM 4,4′‐dithiodipyridine solution (electrode‐protein interaction promoter) for 15 min. Then, 3 μl of 250 μM SmPaz2 solution in 50 mM Tris–HCl buffer (pH 7.2) was deposited on the polished electrode, and a square piece of dialysis membrane (3.5 kDa MWCO) was placed on the top of the electrode. Finally, a rubber ring was fitted around the electrode body, entrapping the enzyme solution at the electrode–membrane interface, forming a uniform thin layer. This electrode assembly placed in the electrolytic solution allows small ions to diffuse through the membrane, and at the same time keeping the protein trapped close to the electrode. ¹⁵ The potential was cycled at different scan rates (1 mV s⁻¹ ≤ ν ≤ 1,000 mV s⁻¹) from 300 to −200 mV versus. Ag/AgCl. The solutions were degassed with pure argon, and the experiment was conducted under anaerobic conditions at room temperature (ca. 298 K).

Chronoamperometry assays to obtain the apparent kinetic parameters of SmNirK‐SmPaz2 couple were performed. For this, the same electrode setup was used but the surface of the gold electrode was incubated with a mixture 1:1 SmNirK:SmPaz2 (250 μM each). The applied potential used was −100 mV versus Ag/AgCl. During the time course, aliquots of 25 μl of a sodium nitrite solution (100 mM) were added to the electrolytic solution to record the faradaic current responses in the range of 50–900 μM nitrite.

CONFLICT OF INTEREST

All authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Cintia Ramírez: Data curation (equal); formal analysis (equal); investigation (equal); methodology (equal); writing – original draft (equal). Carmien Tolmie: Data curation (equal); formal analysis (equal); funding acquisition (equal); investigation (equal); methodology (equal); software (equal); validation (equal); writing – original draft (equal); writing – review and editing (equal). Diederik Opperman: Data curation (equal); funding acquisition (equal); investigation (equal); methodology (equal); project administration (equal); software (equal); validation (equal); writing – original draft (equal); writing – review and editing (equal). Pablo González: Data curation (equal); formal analysis (equal); funding acquisition (equal); investigation (equal); methodology (equal); validation (equal); writing – original draft (equal); writing – review and editing (equal). María Rivas: Funding acquisition (equal); investigation (equal); project administration (equal); resources (equal); writing – original draft (equal); writing – review and editing (equal). Carlos Brondino: Funding acquisition (equal); project administration (equal); resources (equal); writing – original draft (equal); writing – review and editing (equal). FELIX FERRONI: Conceptualization (lead); data curation (lead); formal analysis (lead); funding acquisition (equal); investigation (lead); methodology (lead); resources (equal); software (equal); supervision (lead); validation (lead); visualization (lead); writing – original draft (lead); writing – review and editing (lead).

Supporting information

Appendix S1. Supporting Information.

Ramírez CS, Tolmie C, Opperman DJ, González PJ, Rivas MG, Brondino CD, et al. Copper nitrite reductase from Sinorhizobium meliloti 2011: Crystal structure and interaction with the physiological versus a nonmetabolically related cupredoxin‐like mediator. Protein Science. 2021;30:2310–2323. 10.1002/pro.4195

DATA AVAILABILITY STATEMENT

The atomic coordinates have been deposited in the Protein Data Bank, with the accession code 7P2F. All other data are presented in the main text or Supporting Information and are available upon request.