Structure of heme d 1-free cd 1 nitrite reductase NirS

The crystal structure of a heme d 1-free form of the cd 1 nitrite reductase NirS from Pseudomonas aeruginosa has been determined and provides insight into a premature form of the enzyme.


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 cd 1 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 -propeller that utilizes the uncommon isobacteriochlorin heme d 1 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 d 1 . 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 d 1 (Fig. 1b; Cutruzzolà et al., 2001). After dehydration and reduction of the substrate, the tetrapyrrole heme d 1 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., 1999Nurizzo et al., , 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 -propeller was observed (Sjö gren & Hajdu, 2001). This rotation was also seen in Pa-NirS after the mutation of His327 to alanine .
All structural studies of NirS to date have focused on the heme d 1 -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 d 1 -free form of NirS in flagella formation has been proposed (Borrero-de Acuñ a 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 d 1 cofactor (Nicke et al., 2013). We have therefore set out to determine the heme d 1 -free structure of NirS in order to gain more insight into these processes.

Macromolecule production
NirS with bound dihydro-heme d 1 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;Klü nemann et al., 2019). Briefly, purification of NirS with bound dihydroheme d 1 , the precursor of heme d 1 that accumulates in the absence of NirN, was achieved by two subsequent ion-exchange chromato-graphic 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 À1 the eluted protein lost dihydro-heme d 1 , 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 d 1 ) during elution. Interestingly, dihydro-heme d 1 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.

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 ml. To obtain structural insight into possible changes in the dihydro-heme d 1 -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 K 2 HPO 4 ). The green colour hinted at the presence of dihydroheme d 1 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 d 1 had been lost as a cofactor. As these crystals diffracted X-rays to only 4 Å resolution, sparsematrix 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 d 1 -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 d 1 removed by the chromatographic procedure described above prior to the crystallization experiment. Crystals were flashcooled in liquid nitrogen after cryoprotection with 10%(v/v) (R,R)-2,3-butanediol.

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 dihydroheme d 1 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.

Structure solution and refinement
The crystal structure of heme d 1 -free NirS was determined by molecular replacement utilizing Phaser Photographs of representative crystals grown from NirS isolated with (green frame) or without (red frame) bound dihydro-heme d 1 . The data sets used in this study were collected from the crystallization conditions marked with an asterisk (*).

Results
In previous work with the dihydro-heme d 1 dehydrogenase NirN, we isolated the cd 1 nitrite reductase NirS with bound dihydro-heme d 1 , the final intermediate of heme d 1 biosynthesis, which differs from heme d 1 by lacking a double bond in the propionate chain attached to ring D (Klü nemann et al., 2019). It has previously been shown that NirS can utilize both cofactors but is less active with dihydro-heme d 1 (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 dihydroheme d 1 , NirS with bound dihydro-heme d 1 was crystallized using the published conditions for heme d 1 -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 d 1 and heme c, with the latter being covalently attached to the polypeptide. Comparison with the previously published heme d 1 -bound structure of NirS reveals no significant differences (Nurizzo et al., 1997;C 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 d 1 or heme d 1 are likely to be caused by inherent properties of the cofactor. The structure of heme d 1 -free NirS possesses a similar dimeric structure and domain organization as mature heme d 1 -bound NirS (Nurizzo et al., 1997). Both chains have a C-terminal eight-bladed -propeller, which has previously been termed the d 1 domain as it is responsible for binding d 1 -type tetrapyrroles in NirS and the related proteins NirF and NirN (Fig. 3). In NirS, these domains form homodimers by interaction of the 15th -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 d 1 domain and harbours a covalently attached heme c. Interestingly, the position of the cytochrome c domain differs from that found in the heme d 1 -bound and dihydro-heme d 1 -bound structures by a rotation of approximately 60 around the pseudo-eightfold axis of the -propeller. A similar conformation has previously been observed for Pa-NirS after aminoacid exchange of His327 to alanine or for Pp-NirS crystallized under reducing conditions Sjö gren & Hajdu, 2001). In both of these structures, as well as in the structure of heme d 1 -free NirS reported here, no density is observed for the N-terminal arm, which resides between the cytochrome c domain and the d 1 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 d 1 -free NirS with other Pa-NirS structures reveals a higher similarity to the structure of NirS reduced in crystallo (C r.m.s.d. of 0.37 Å ) compared with the oxidized form (C r.m.s.d. of 1.09 Å ) (Nurizzo et al., 1997(Nurizzo et al., , 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 r.m.s.d. for cytochrome c of 0.35 Å ). A similar comparison of the d 1 domain reveals a lower C 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 -propeller, allowing better solvent access to heme d 1 . Furthermore, the reported occupancy (0.5) of heme d 1 , together with the observed change in coloration of these crystals over time, indicating the loss of heme d 1 , suggests that the open conformation allows easy access to the ligand-binding site in general . Closer inspection of the heme d 1 -binding pocket reveals only a few distinct changes related to d 1 -type heme binding (Fig. 4). Three arginines that interact with the acetate or propionate side chains of heme d 1 in the heme d 1 -bound structure (Arg156, Arg198 and Arg372) adopt different conformations in the heme d 1 -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 d 1 , shows no conformational changes without bound tetrapyrrole. This stands in contrast to the d 1 -type heme-binding -propellers of the heme d 1 -biosynthesis proteins NirN and NirF, in which the corresponding histidine rotates with respect to the ligand-free form to coordinate the iron (Klü nemann et al., 2020(Klü nemann et al., , 2019. Another distinctive variation in the d 1 domains of heme d 1 -free and heme d 1 -bound NirS is observed for the loop consisting of residues 497-506, which moves upon heme d 1 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 -strand, which is inside the propeller and moves inwards upon heme d 1 binding, indicating a connection between the widening of the propeller and the movement of Trp498 (Fig. 4c).

Discussion
It has long been known that the cd 1 nitrite reductase NirS adopts an open and a closed conformation, and it has been hypothesized that these states are linked to the catalytic cycle Sjö gren & 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 d 1 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 d 1 (Borrero-de Acuñ a 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 Acuñ a 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 d 1 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.