Structure of heme d₁-free cd₁ nitrite reductase NirS

A key step in anaerobic nitrate respiration is the reduction of nitrite to nitric oxide, which is catalysed by the cd₁ nitrite reductase NirS in, for example, the Gram-negative opportunistic pathogen Pseudomonas aeruginosa. Each subunit of this homodimeric enzyme consists of a cytochrome c domain and an eight-bladed β-propeller that binds the uncommon isobacteriochlorin heme d₁ as an essential part of its active site. Although NirS has been well studied mechanistically and structurally, the focus of previous studies has been on the active heme d₁-bound form. The heme d₁-free form of NirS reported here, which represents a premature state of the reductase, adopts an open conformation with the cytochrome c domains moved away from each other with respect to the active enzyme. Further, the movement of a loop around Trp498 seems to be related to a widening of the propeller, allowing easier access to the heme d₁-binding side. Finally, a possible link between the open conformation of NirS and flagella formation in P. aeruginosa is discussed.

1. Introduction

Denitrification is the stepwise reduction of nitrogen oxides to dinitrogen and is utilized by many bacteria to replace the terminal oxygen-dependent steps of the respiratory chain under anaerobic conditions. The reduction of nitrite to nitric oxide is a highly regulated step, since both of these molecules are toxic to the cell. Many bacteria utilize the cd₁ nitrite reductase NirS to catalyse this step, and the NirS enzymes from Pseudomonas aeruginosa (Pa-NirS) and Paracoccus pantotrophus (Pp-NirS) have been well studied functionally and structurally. Both chains of this homodimeric enzyme possess a c-type cytochrome domain with a covalently attached heme c functioning as an electron-entry point and a C-terminal eight-bladed β-propeller that utilizes the uncommon iso­bacteriochlorin heme d₁ as an active-site cofactor (Fig. 1a; Nurizzo et al., 1997; Williams et al., 1997). The domains interact via an N-terminal arm which, in the case of Pa-NirS, belongs to the opposing subunit of the homodimer (Williams et al., 1997; Nurizzo et al., 1997). Compared with other tetrapyrroles such as heme b, a higher affinity for anionic molecules such as nitrite and a lower affinity for nitric oxide has been observed for the ferrous isobacteriochlorin heme d₁. This is rooted in the unique carbonyl moieties at rings A and B and the replacement of the typical propionate by an acrylate at ring D (Chang et al., 1986; Rinaldo et al., 2011; Fujii et al., 2016). During its reduction, nitrite is coordinated by His327 and His369 on top of the ferrous iron of heme d₁ (Fig. 1b; Cutruzzolà et al., 2001). After dehydration and reduction of the substrate, the tetrapyrrole heme d₁ is then reduced via internal electron transfer to allow efficient replacement of the product nitric oxide by another nitrite anion (Rinaldo et al., 2011). In the case of Pa-NirS, the intramolecular electron transfer is the rate-limiting step and is allosterically controlled by smaller conformational changes within the cytochrome c domain (Farver et al., 2009; Nurizzo et al., 1999, 1998). Interestingly, when Pp-NirS was crystallized under reducing conditions, an ∼60° rotation of the cytochrome c domain around the pseudo-eightfold axis of the β-propeller was observed (Sjögren & Hajdu, 2001). This rotation was also seen in Pa-NirS after the mutation of His327 to alanine (Brown et al., 2001).

Figure 1 (a) Chemical structure of heme d1 with characteristic features highlighted in red. (b) Simplified depiction of the Pa-NirS active site during the reduction of nitrite to nitric oxide.

All structural studies of NirS to date have focused on the heme d₁-bound form of the enzyme and have led to a detailed understanding of the reaction mechanism within the context of denitrification. Recently, however, following the discovery that Pa-NirS forms complexes with the flagella protein FliC and the chaperone DnaK, an enzyme activity-independent scaffolding function of a heme d₁-free form of NirS in flagella formation has been proposed (Borrero-de Acuña et al., 2015). This cofactor-free form is also expected to exist in the course of maturation of NirS, where the enzyme has been demonstrated to transiently interact with NirF and NirN, two proteins that are involved in the biosynthesis of the heme d₁ cofactor (Nicke et al., 2013). We have therefore set out to determine the heme d₁-free structure of NirS in order to gain more insight into these processes.

2. Materials and methods

2.1. Macromolecule production

NirS with bound dihydro-heme d₁ was isolated from P. aeruginosa strain RM361 (nirN::tet) (Kawasaki et al., 1997) after cultivation under anaerobic conditions as described previously (Adamczack et al., 2014; Klünemann et al., 2019). Briefly, purification of NirS with bound dihydroheme d₁, the precursor of heme d₁ that accumulates in the absence of NirN, was achieved by two subsequent ion-exchange chromatographic steps, as first described by Parr et al. (1976). The purified protein was dialysed against 10 mM Tris–HCl pH 8 and loaded onto a Q-Sepharose Fast Flow anion-exchange column. After washing for 4 h with dialysis buffer at a flow rate of 1 ml min⁻¹ the eluted protein lost dihydro-heme d₁, but still contained the covalently attached heme c, as indicated by a UV–Vis absorption band at 410 nm (indicative of heme c) and no absorption at 630 nm (indicative of dihydro-heme d₁) during elution. Interestingly, dihydro-heme d₁ did not seem to be eluted during the washing procedure but remained bound to the resin, and was visible as a greenish coloration of the chromatography column. Prior to crystallization, the protein was subjected to size-exclusion chromatography (SEC; Superdex 200, GE Healthcare) in a buffer consisting of 10 mM Tris–HCl pH 8, 150 mM NaCl.

2.2. Crystallization

Crystallization was performed by sitting-drop vapour diffusion in Intelli-Plates 96-3 (Art Robbins Instruments) at room temperature. Plates were set up using a HoneyBee 961 pipetting robot (Digilab Genomic Solutions), which mixed 200 nl protein solution with 200 nl precipitant solution and provided a reservoir of 60 µl. To obtain structural insight into possible changes in the dihydro-heme d₁-bound form, a screen based on previously published crystallization conditions for NirS (Tegoni et al., 1994; Brown et al., 2001) was performed. After two days, plate-shaped green crystals were observed with phosphate as the precipitant (0.1 M Tris–HCl pH 8, 1.9 M K₂HPO₄). The green colour hinted at the presence of dihydro-heme d₁ in these crystals (Fig. 2). Surprisingly, under conditions containing PEG 4000 or PEG 6000, red rod-shaped or tetragonal crystals started to grow. The red coloration indicated that dihydro-heme d₁ had been lost as a cofactor. As these crystals diffracted X-rays to only 4 Å resolution, sparse-matrix screening with the commercially available crystallization suites JCSG+ (Qiagen) and Morpheus (Molecular Dimensions) was used to identify conditions that were better suited to produce crystals for diffraction experiments. The best-diffracting crystals of heme d₁-free NirS were obtained with 0.1 M disodium hydrogen phosphate, 0.1 M citric acid, 40%(v/v) PEG 300 at pH 4.2 using NirS with dihydro-heme d₁ removed by the chromatographic procedure described above prior to the crystallization experiment. Crystals were flash-cooled in liquid nitrogen after cryoprotection with 10%(v/v) (R,R)-2,3-butanediol.

Figure 2 Photographs of representative crystals grown from NirS isolated with (green frame) or without (red frame) bound dihydro-heme d1. The data sets used in this study were collected from the crystallization conditions marked with an asterisk (*).

2.3. Data collection and processing

For both data sets, 3600 diffraction images, each with an oscillation angle of 0.1°, were collected on beamline P11 at PETRA III, DESY, Hamburg, Germany with a PILATUS 6M fast detector. Images were processed utilizing the autoPROC pipeline (Vonrhein et al., 2011) executing XDS (Kabsch, 2010), POINTLESS (Evans, 2011) and AIMLESS (Evans & Murshudov, 2013). In the case of NirS with bound dihydro-heme d₁ the crystals diffracted anisotropically, which was accounted for using STARANISO (Tickle et al., 2018) within autoPROC. Data-collection and processing statistics are summarized in Table 1.

Table 1
Data collection and processing

Values in parentheses are for the outer shell.

	Heme d1-free NirS	Dihydro-heme d1-bound NirS
Diffraction source	P11, PETRA III	P11, PETRA III
Wavelength (Å)	1.033	1.739
Temperature (K)	100	100
Detector	PILATUS 6M fast	PILATUS 6M fast
Crystal-to-detector distance (mm)	368.1	168.4
Rotation range per image (°)	0.1	0.1
Total rotation range (°)	360	360
Exposure time per image (s)	0.1	0.05
Space group	P4322	P21212
a, b, c (Å)	67.25, 67.25, 277.10	163.76, 91.09, 112.71
α, β, γ (°)	90, 90, 90	90, 90, 90
Resolution range (Å)	48.25–1.86 (1.93–1.86)	112.71–2.38 (2.62–2.38)
Resolution limits along axes (a*, b*, c*)	—	2.89, 2.38, 2.51
Total No. of reflections	1363122 (93274)	625785 (23996)
No. of unique reflections	54709 (5373)	50095 (2505)
Completeness, spherical (%)	99.97 (100)	73.1 (14.7)
Completeness, ellipsoidal (%)	—	92.6 (44.5)
Multiplicity	24.9 (17.4)	12.5 (9.6)
〈I/σ(I)〉	20.3 (1.8)	6.6 (1.6)
CC1/2	0.99 (0.57)	0.99 (0.64)
Rp.i.m.	0.03 (0.32)	0.11 (0.54)
Overall B factor from Wilson plot (Å^2)	26.2	23.6

2.4. Structure solution and refinement

The crystal structure of heme d₁-free NirS was determined by molecular replacement utilizing Phaser (McCoy et al., 2007) and the coordinates of the H327A variant of Pa-NirS (NirS^H327A; PDB entry 1hzu; Brown et al., 2001), which crystallized in the same space group but with unit-cell axes that were approximately 3 Å longer. The structure of NirS with bound dihydro-heme d₁ was phased by Fourier synthesis using a published structure of Pa-NirS with bound heme d₁ as a starting model (PDB entry 1nir; Nurizzo et al., 1997). To remove phase bias, both initial models were subjected to ten cycles of refinement with simulated annealing enabled in phenix.refine (Afonine et al., 2012). The final models were constructed by several cycles of manual optimization in Coot (Emsley et al., 2010) and computational refinement including TLS refinement. Depictions of the final models were prepared with PyMOL (version 1.8; Schrödinger). Refinement statistics are summarized in Table 2.

Table 2
Structure solution and refinement

Values in parentheses are for the outer shell.

	Heme d1-free NirS	Dihydro-heme d1-bound NirS
Resolution range (Å)	48.25–1.86 (1.93–1.86)	79.61–2.38 (2.42–2.38)
Final Rcryst	0.17 (0.23)	0.24 (0.33)
Final Rfree	0.19 (0.27)	0.28 (0.43)
No. of non-H atoms
Protein	4100	8364
Ligand	65	214
Water	255	246
R.m.s. deviations
Bonds (Å)	0.01	0.01
Angles (°)	0.94	1.37
Average B factor (Å^2)	31.0	23.6
Ramachandran plot
Most favoured (%)	96.35	94.24
Allowed (%)	3.65	5.67

3. Results

In previous work with the dihydro-heme d₁ dehydrogenase NirN, we isolated the cd₁ nitrite reductase NirS with bound dihydro-heme d₁, the final intermediate of heme d₁ biosynthesis, which differs from heme d₁ by lacking a double bond in the propionate chain attached to ring D (Klünemann et al., 2019). It has previously been shown that NirS can utilize both cofactors but is less active with dihydro-heme d₁ (Hasegawa et al., 2001; Kawasaki et al., 1997; Adamczack et al., 2014). To determine whether the altered activity is caused by structural changes or by the inherent properties of dihydro-heme d₁, NirS with bound dihydro-heme d₁ was crystallized using the published conditions for heme d₁-bound NirS (Tegoni et al., 1994; Brown et al., 2001). In this complex, anomalous difference density around the iron centres indicates similar occupancies for dihydro-heme d₁ and heme c, with the latter being covalently attached to the polypeptide. Comparison with the previously published heme d₁-bound structure of NirS reveals no significant differences (Nurizzo et al., 1997; C^α r.m.s.d. of 0.38 Å; Figs. 3 and 4). This suggests that the reported differences in the activity of NirS loaded with dihydro-heme d₁ or heme d₁ are likely to be caused by inherent properties of the cofactor.

Figure 3 Cartoon representations of the different conformations observed for the cd1 nitrite reductase NirS from P. aeruginosa (heme d1-bound NirS, PDB entry 1nir, Nurizzo et al., 1997; NirS^H327A, PDB entry 1hzu, Brown et al., 2001) after superposition of the d1 domains. The N-terminal arm is coloured red, the linker blue and the d1 domain grey. The cytochrome c domain and the tetrapyrroles are coloured individually.

Figure 4 Depiction of hydrophilic (a) and hydrophobic (b) residues in the heme d1-binding pocket. Amino acids are shown as sticks (heme d1-free NirS, green) or lines (heme d1-bound NirS, magenta; NirS^H327A, yellow; dihydro-heme d1-bound NirS, cyan). Residues with alternative conformations are marked with an asterisk (*). The position of heme d1 inside the binding pocket is shown as a thin outline. (c) shows the movement of the loops associated with Trp498 narrowing the binding pocket of heme d1-free NirS after binding a d1-type heme

The structure of heme d₁-free NirS possesses a similar dimeric structure and domain organization as mature heme d₁-bound NirS (Nurizzo et al., 1997). Both chains have a C-terminal eight-bladed β-propeller, which has previously been termed the d₁ domain as it is responsible for binding d₁-type tetrapyrroles in NirS and the related proteins NirF and NirN (Fig. 3). In NirS, these domains form homodimers by interaction of the 15th β-strand of each subunit, which is consistent with the SEC data collected during purification (data not shown). The N-terminal domain consists of a cytochrome c fold that is connected via a linker domain to the d₁ domain and harbours a covalently attached heme c. Interestingly, the position of the cytochrome c domain differs from that found in the heme d₁-bound and dihydro-heme d₁-bound structures by a rotation of approximately 60° around the pseudo-eightfold axis of the β-propeller. A similar conformation has previously been observed for Pa-NirS after amino-acid exchange of His327 to alanine or for Pp-NirS crystallized under reducing conditions (Brown et al., 2001; Sjögren & Hajdu, 2001). In both of these structures, as well as in the structure of heme d₁-free NirS reported here, no density is observed for the N-terminal arm, which resides between the cytochrome c domain and the d₁ domain of the opposing subunit in Pa-NirS in the closed conformation, indicating that it is flexible in the open conformation.

An overlay of the cytochrome c domain of heme d₁-free NirS with other Pa-NirS structures reveals a higher similarity to the structure of NirS reduced in crystallo (C^α r.m.s.d. of 0.37 Å) compared with the oxidized form (C^α r.m.s.d. of 1.09 Å) (Nurizzo et al., 1997, 1998). The difference is mostly caused by the position of residues 55–64 and is consistent with an observation previously discussed for NirS^H327A (C^α r.m.s.d. for cytochrome c of 0.35 Å). A similar comparison of the d₁ domain reveals a lower C^α r.m.s.d. with respect to the H327A variant of Pa-NirS (0.75 Å) than to the oxidized or reduced wild-type protein (1.34 and 1.39 Å, respectively). NirS^H327A has been shown to have a wider β-propeller, allowing better solvent access to heme d₁ (Brown et al., 2001). Furthermore, the reported occupancy (0.5) of heme d₁, together with the observed change in coloration of these crystals over time, indicating the loss of heme d₁, suggests that the open conformation allows easy access to the ligand-binding site in general (Brown et al., 2001). Closer inspection of the heme d₁-binding pocket reveals only a few distinct changes related to d₁-type heme binding (Fig. 4). Three arginines that interact with the acetate or propionate side chains of heme d₁ in the heme d₁-bound structure (Arg156, Arg198 and Arg372) adopt different conformations in the heme d₁-free form. Additionally, His327 and His369, which are both involved in substrate binding and catalysis by NirS, have different positions. Interestingly, His182, which acts as the fifth ligand of the iron cation in the centre of heme d₁, shows no conformational changes without bound tetrapyrrole. This stands in contrast to the d₁-type heme-binding β-propellers of the heme d₁-biosynthesis proteins NirN and NirF, in which the corresponding histidine rotates with respect to the ligand-free form to coordinate the iron (Klünemann et al., 2020, 2019). Another distinctive variation in the d₁ domains of heme d₁-free and heme d₁-bound NirS is observed for the loop consisting of residues 497–506, which moves upon heme d₁ binding to enable Trp498 to lock the cofactor inside its binding pocket. Interestingly, the neighbouring loop (residues 473–479) also adopts a slightly different conformation. This loop is connected to the 30th β-strand, which is inside the propeller and moves inwards upon heme d₁ binding, indicating a connection between the widening of the propeller and the movement of Trp498 (Fig. 4c).

4. Discussion

It has long been known that the cd₁ nitrite reductase NirS adopts an open and a closed conformation, and it has been hypothesized that these states are linked to the catalytic cycle (Brown et al., 2001; Sjögren & Hajdu, 2001). The data presented here extend the potential importance of the open conformation by suggesting that it also represents a native but immature form of the enzyme that facilitates binding of the heme d₁ cofactor as the ultimate step of NirS maturation.

The open conformation of NirS may also have additional physiological importance outside the denitrification process. It has recently been found that NirS can function as scaffold protein for flagella formation under denitrifying conditions, even in the absence of heme d₁ (Borrero-de Acuña et al., 2015). Interestingly, the N-terminal arm and the cytochrome c domain of NirS have been shown to be involved in complex formation between NirS, FliC and DnaK in this process (Borrero-de Acuña et al., 2015). Further, kinetic studies with NirS from Pseudomonas stutzeri, which lacks the N-terminal arm, and the mutation of a conserved tyrosine in Pa-NirS and Pp-NirS suggest that the N-terminal arm is not required for the enzymatic activity, despite its interaction with heme d₁ (Cutruzzolà et al., 1997; Gordon et al., 2003; Wilson et al., 2001). Therefore, the conformational change observed in our study can be envisioned to release the N-terminal arm to allow flagella formation, thereby regulating cell motility. Further investigation will be required to confirm that the conformational changes observed in crystallo can also be found in solution and are associated with distinct functions of NirS in vivo.