NirF is a periplasmic protein that binds d 1 heme as part of its essential role in d 1 heme biogenesis Shilpa Bali 1 , Martin J. Warren 2 and Stuart J. Ferguson 1 1 Department of Biochemistry, University of Oxford, UK 2 Department of Biosciences, University of Kent, Canterbury, UK Introduction Denitrification is a four-step transformation of nitrate to dinitrogen gas by various species of bacteria under anaerobic conditions [1,2]. These four steps are cataly- sed by complex metalloenzymes and involve stepwise conversion of nitrate to nitrite, nitrite to nitric oxide, nitric oxide to nitrous oxide and finally reduction of nitrous oxide to nitrogen. In the denitrification path- way, nitrite reduction is the key step, as it is the point of divergence from assimilatory nitrogen metabolism in which nitrite is reduced to ammonium [2,3]. There are two types of respiratory nitrite reductase involved in denitrification: one is copper-containing nitrite reduc- tase (NirK), which is prevalent in, but not exclusive to, alphaproteobacteria, the other being cytochrome cd 1 (NirS), which prevails in betaproteobacteria [4]. Cytochrome cd 1 nitrite reductase is a homodimeric periplasmic enzyme with each subunit containing a covalently attached c heme and noncovalently attached d 1 heme, bound in a beta-propeller domain, as pros- thetic groups [5,6]. Heme d 1 , which forms the active cen- tre for the one electron reduction of nitrite to nitric oxide, has a unique structure. The structure of this mod- ified heme, a dioxoisobacteriochlorin to be more spe- cific, has been known for more than two decades [7,8], but quite how it is biosynthesized by denitrifying bacte- ria under anaerobic conditions is not understood. Anal- ysis of insertional mutagenesis and complementation work in Pseudomonas aeruginosa, Pseudomonas fluores- cens, Paracoccus denitrificans and Pseudomonas stutzeri have shown that a set of several contiguous genes that always follows the structural gene, nirS, for cytochrome cd 1 , is necessary for the biogenesis of the d 1 cofactor [9–13]. In P. denitrificans and closely related Para- coccus pantotrophus, these genes are cotranscribed as Keywords cytochrome cd 1 ; d 1 heme biosynthesis; denitrification; nitrite reductase; Paracoccus pantotrophus; tetrapyrrole Correspondence S. J. Ferguson, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK Fax: +44 1865 613201 Tel: +44 1865 613299 E-mail: stuart.ferguson@bioch.ox.ac.uk (Received 24 June 2010, revised 27 August 2010, accepted 1 October 2010) doi:10.1111/j.1742-4658.2010.07899.x The cytochrome cd 1 nitrite reductase from Paracoccus pantotrophus catalyses the one electron reduction of nitrite to nitric oxide using two heme cofactors. The site of nitrite reduction is the d 1 heme, which is synthesized under anaer- obic conditions by using nirECFD-LGHJN gene products. In vivo studies with an unmarked deletion strain, DnirF, showed that this gene is essential for cd 1 assembly and consequently for denitrification, which was restored when the DnirF strain was complemented with wild-type, plasmid-borne, nirF. Removal of a signal sequence and deletion of a conserved N-terminal Gly-rich motif from the NirF coded on a plasmid resulted in loss of in vivo NirF activity. We demonstrate here that the product of the nirF gene is a periplasmic protein and, hence, must be involved in a late stage of the cofac- tor biosynthesis. In vitro studies with purified NirF established that it could bind d 1 heme. It is concluded that His41 of NirF, which aligns with His200 of the d 1 heme domain of cd 1 , is essential both for this binding and for the production of d 1 heme; replacement of His41 by Ala, Cys, Lys and Met all gave nonfunctional proteins. Potential functions of NirF are discussed. Abbreviation LB, Luria–Bertani. 4944 FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS an operon in the following order nirECFD-LGHJN.It has been proved that the biosynthesis of the d 1 heme proceeds via a common tetrapyrrole precursor uroporphyrinogen III, which is transformed into pre- corrin-2 by S-adenosyl-L-methionine-dependent methyl transferase, NirE [14,15]. In addition, it has been dem- onstrated that a Paracoccus derivative strain, in which nirN is replaced with a kanamycin resistance cassette, still makes holo-cd 1 , which suggests that this last gene on the operon is dispensable for d 1 heme assembly [15]. Also, the nirC gene encodes a periplasmic c type cyto- chrome that may have an electron transfer role in cyto- chrome cd 1 activity [16] or maturation [15]. Conflicting evidence exists concerning the subcellular location of NirF. NirF from P. pantotrophus shares 54% sequence identity and 72.3% sequence similarity with the NirF from Ps. aeruginosa. However, the pro- tein from Ps. aeruginosa lacks the apparent Sec-depen- dent signal sequence for translocation to the periplasm, which, in contrast, is readily identified in P. pantotro- phus NirF. Information about NirF from the much- studied Ps. aeruginosa has led to the widespread assumption that NirF is cytoplasmic. Accordingly, we wanted to determine the subcellular location of NirF in P. pantotrophus, which produces larger amounts of cd 1 under denitrifying conditions than Ps. aeruginosa. NirF also shares 34% sequence similarity with the beta-pro- peller domain of cd 1 , indicating a scaffolding role for an intermediate of heme d 1 synthesis. Roles for NirF in ferrochelation or dehydrogenase activity have been pro- posed [10,13]. Interestingly, NirF also shows some simi- larity to a cobalamin decarboxylase, CobT [3,13]. The physiological role of NirF would heavily rely on its cel- lular location. For all these reasons we wanted to develop an in vivo system to investigate its role by mak- ing use of the generation and characterization of an unmarked deletion in nirF. The unmarked gene deletion mutant is expected to lose nitrite respiration and the capacity to synthesize this tetrapyrrole derivative. More importantly, an unmarked deletion mutant strain should have denitrification restored on complementa- tion with plasmid-borne nirF and, therefore, should provide an in vivo system to further analyse the physio- logical role of NirF and its variants. Results Construction of the DnirF strain and its in vivo nitrite reductase activity In P. pantotrophus, the operon associated with d 1 heme biosynthesis has many overlapping genes. The nirF gene overlaps four nucleotides with the preceding nirC gene and it is also immediately followed by an overlap- ping ORF for nirD-L. Previous studies in Ps. aerugin- osa demonstrated that a marked mutation in the nirF gene resulted in the formation of inactive nitrite reduc- tase [9]. Similar results were also found for an nirF mutant in P. stutzeri [10], but in this case a polar effect of the mutation was not excluded. The present study utilized an unmarked deletion in nirF where the entire nirF ORF has been deleted from the chromosome. When this unmarked deletion strain of nirF (i.e. DnirF), named SBN11, was grown anaerobically in minimal media supplemented with 20 mm nitrate, it converted the entire available nitrate to nitrite within 10 h of growth and lost its nitrite reductase activity, as shown by no consumption of nitrite to yield any gas- eous products. The extracellular nitrite concentration peaked at 20 mm and remained there even when the cultures had reached the stationary phase. No brown coloration from holo-cytochrome cd 1 or gas evolution from nitrogen production was observed in the SBN11 cultures. Reassuringly, the nitrite reductase activity of the DnirF strain was restored within 10 h of anaerobic growth on nitrate-supplemented minimal media, when it was complemented with a plasmid-borne copy of nirF (Fig. 1). Here, the extracellular nitrite concentra- tion reached a maximum value of 14 mm, followed by a rapid decline. This corresponds to a delay in the expression or activation of functional cytochrome cd 1 , but eventually a complete denitrification pathway was established. As shown by the four independent growth results, the extracellular nitrite concentration was a function of cell density, rather than time, for the DnirF strain expressing plasmid-encoded nirF. In addition, expression of plasmid-encoded strep II-tagged nirF was demonstrated by western blot analysis with alkaline phosphatase-conjugated strep-tactin antibody (Fig. S1). These results confirm the essential role of NirF in d 1 heme assembly. Influence of deletion and replacement of the signal sequence on NirF processing and denitrification activity The derived amino acid sequences of NirF from two different denitrifiers, P. pantotrophus and Ps. aerugin- osa, differ significantly at the N-termini. In P. panto- trophus, nirF encodes a ( 42 kDa) protein that has an N-terminal signal sequence suggestive of a location in the periplasm. On the other hand, in Ps. aeruginosa (PAO1), NirF has no apparent signal sequence and therefore this protein should be located in the cyto- plasm. In order to determine whether export of NirF S. Bali et al. Periplasmic NirF binds d 1 heme FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS 4945 to the periplasm of P. pantotrophus is essential for d 1 heme formation, and thus for the physiological func- tion of NirF, we deleted the presumed signal sequence. We also replaced the putative signal sequence of NirF with the shorter signal sequence of a native periplasmic protein, NirC, to see whether it could still perform its physiological function. The replacement of the signal sequence on the NirF coded for on a plasmid had no effect on nitrite reductase activity as judged by the res- toration of denitrification when this plasmid was used to complement the DnirF strain (Fig. 2). This result also ruled out the need for a specific signal sequence for the function of NirF. On the other hand, a plasmid carrying an nirF gene lacking the native signal sequence failed to restore denitrification upon attempted complementation of the DnirF strain (Fig. 2). A C-terminally strep II-tagged version of NirF was produced anaerobically from a pEG276-based plasmid in the DnirF strain using minimal media supplemented with nitrate as the terminal electron acceptor. When the cells of this derivative strain producing tagged 22 Time (h) 2 6 10 12 22 24 20 18 16 14 12 10 8 6 Nitrate (mM) 4 2 0 GB-17 SBN11 SBN13-1 SBN13-2 Strain SBN13-3 SBN13-4 SBN14 Fig. 1. Restoration of nitrite reduction in a Paracoccus pantotrophus strain carrying an unmarked gene deletion of nirF (DnirF) with plasmid- borne nirF. GB17 is the parental wild-type P. pantotrophus in which nitrite does not accumulate following reduction of added nitrate; SBN11 is the DnirF strain that does not synthesize d 1 heme and hence cannot turnover nitrite to nitric oxide. SBN13 is SBN11 complemented with nirF on pEG276 (four replicas are shown) and SBN14 is a control with the SBN11 strain containing the empty expression vector pEG276 only. Replicas of SBN13 indicate that the concentration of extracellular nitrite is dependent on the cell density at any given time. The code for the times of analysis is shown on the right. 22 SBN20 (ΔnirF + NirF (no signal sequence)) 20 18 16 14 12 10 8 6 Nitrate (mM) 4 2 0 2.2 2 1.8 1.6 1.4 1.2 0.8 0.6 0.4 0.2 0 1 0510 Time (h) 2015 25 SBN28 (ΔnirF + NirF (NirC signal sequence)) 22 2 1.8 1.6 1.4 1.2 0.8 A 600 0.6 0.4 0.2 0 1 20 18 16 14 12 10 8 6 4 2 0 0510 Time (h) 2015 25 30 Fig. 2. A periplasmic targeting sequence is essential for Paracoccus pantotrophus NirF function. Growth plots and time courses of nitrite appearance and disappearance for the SBN11 (DnirF) strain complemented with a plasmid coding for NirF from which the putative periplas- mic targeting sequence had been deleted (to give SBN20). Also shown is the effect of providing the DnirF strain with a plasmid coding for NirF where its native signal sequence had been replaced by the proven periplasmic targeting sequence of NirC (to give SBN28). Cell density was determined at A 600 and is depicted by grey diamonds, whereas the extracellular nitrite concentration was determined using a colorimet- ric method (for more details, see Experimental procedures) and is shown by black triangles. The data shown here are the average of four different experiments. Periplasmic NirF binds d 1 heme S. Bali et al. 4946 FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS NirF were fractionated and run on the SDS ⁄ PAGE for western analysis by using the alkaline phosphatase conjugate of strep-tactin antibody, we found that both membrane and cytoplasmic fractions were free of NirF protein and it was present only in the periplasmic frac- tion (Fig. 3). These results, together with the outcome of the complementation analysis, prove that NirF is a periplasmic protein in P. pantotrophus. Influence of variations of conserved residues on the in vivo activity of NirF Interestingly, like NirN, NirF shares sequence similar- ity with the C-terminal d 1 heme-containing domain of cytochrome cd 1 . Strikingly, the axial ligand of iron in d 1 heme in P. pantotrophus cd 1 (His200) is conserved in NirF (His41); this conservation applies to all other NirF sequences known in the database (Fig. S2). How- ever, the other catalytic site histidines (His 345 and His388) of NirS are not conserved in NirF. Restoration of denitrification upon complementation of the unmarked nirF deletion strain of P. pantotrophus with plasmid-borne nirF provided a good in vivo sys- tem for testing the molecular basis for the NirF activ- ity (Fig. 4). Replacement of the aforementioned His41 with Ala completely abolished the in vivo nitrite reduc- tase activity as seen by the accumulation of large amounts of extracellular nitrite in the DnirF strain complemented with nirF (H41A), when growing under denitrifying conditions (Fig. 5). We were also curious whether denitrification could be rescued to any extent if this His were replaced with some of the other heme iron-binding residues, such as Met, Cys or Lys. All plasmids bearing nirF with this residue changed to any of these three potential heme ligands failed to rescue denitrification in the DnirF strain (Fig. 5). These results indicate that His41 is important for NirF function. It has been reported that NirF shows 21% sequence similarity to the first 100 amino acids of NirE [13]. There is also a highly conserved N-terminal Gly-rich (GXGX 2 GX 7 G) motif in all NirF sequences, which is suggestive of a binding to a nucleotide-containing cofactor. This motif has also been found in several other dehydrogenases involved in tetrapyrrole biosyn- thesis pathways, including CysG A and SirC in the siroheme biogenesis pathway [17]. Furthermore, when a pairwise alignment of NirF was performed with Met8P (a bifunctional dehydrogenase-ferrochelatase from Saccharomyces cerevisiae), we found that the two proteins had 24% sequence similarity. A crystal structure of Met8P has shown that this protein has an aspartate residue (Asp141), which is important for both chelatase and dehydrogenase function [17]; interestingly, this aspartate, Asp129, is also conserved 98 kDa M Wt Insoluble Total cell lysate Periplasm Membrane Cytoplasm kDa M Wt Insoluble Total cell lysate Periplasm Membrane Cytoplasm 62 49 38 28 17 14 98 62 49 38 28 17 14 AB Fig. 3. Distribution of NirF between different cell fractions of P. pantotrophus. (A) Western blot assay with the different fractions of the cells expressing plasmid-borne and strep-tagged NirF from SBN13 strain, using an alkaline phosphatase conjugate of strep-tactin antibody (for more information see Experimental procedures). NirF ( 42 kDa) with a C-terminal strep II tag could be found in the total cell lysate and the periplasmic fraction, but was absent from the membrane and cytoplasmic fractions. (B) The same cell fractions as shown in (A) when sub- jected to SDS ⁄ PAGE analysis and stained with Coomassie blue for proteins. S. Bali et al. Periplasmic NirF binds d 1 heme FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS 4947 in NirF sequences from P. pantotrophus and other den- itrifiers (Fig. S3). When plasmids carrying the variation in NirF of Asp129 to Ala or to Gln were used to complement the DnirF strain, they showed the same phenotype as when complementation was done with wild-type NirF. There were no growth defects and nei- ther of these strains showed a large accumulation of extracellular nitrite during the log phase, when grown on 20 mm nitrate under denitrifying conditions (Fig. 5). This result demonstrated that Asp129 of NirF could not be essential for any function similar to that in Asp141 of Met8P. The idea of NirF being a dehydrogenase is appealing because of the presence of a putative nucleotide-bind- ing motif in the N-terminal of the protein sequence and also because there is a need for oxidation in the d 1 heme biosynthetic pathway, for example, oxidation of C17 propionate to give an acrylate side chain. This type of step would normally require FAD-based chem- istry. Another potential dehydrogenation is NAD ⁄ NADP-dependent oxidation of precorrin-2 to sirohy- drochlorin that might be a shared intermediate in the d 1 heme and cobalamin biosynthesis pathway. Some, but not all, flavoproteins have tightly bound flavin when overexpressed and thus are yellow on extraction. However, no such coloration was observed for the NirF when it was overproduced in either Escherichia coli or in P. pantotrophus under either aerobic or anaerobic conditions. We also did not observe any interaction between the purified NirF and a range of nucleotide-containing cofactors by using a variety of biophysical methods (data not shown). Nonetheless, we still decided to test the effect of the deletion of the entire GXGX 2 GX 7 G motif on the in vivo NirF and nitrite reductase activity. Although deletion of the entire Gly-rich region resulted in an inactive NirF, analysis of variant NirF species with one or more of the individual Gly residues changed to Ala did not result in loss of NirF function. Thus, we conclude that although a significant stretch of the N-terminus is important for the formation of a functional protein, we have no evidence that this functionality relates to the Gly residues; thus, the important residues may lie elsewhere within this N-terminal region. Purification and in vitro characterization of NirF and its variants NirF was recombinantly produced in E. coli and puri- fied from the periplasmic fraction to near homogeneity 22 20 18 16 12 14 10 8 6 4 2 0 0 5 10 20 2515 22 20 18 16 12 14 10 8 6 4 2 0 22 20 18 16 12 14 10 8 6 4 2 0 20 18 16 12 14 10 8 6 4 2 0 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 2.2 1.2 1.7 0.7 0.2 –0.3 2.2 1.2 1.7 0.7 0.2 –0.3 GB17 SBN03 (nirF : :Kan R ) SBN13 (ΔnirF + NirF) SBN11 (ΔnirF ) Nitrite (mM)Nitrite (mM) A 600 A 600 A 600 A 600 Time (h) 0 5 10 20 2515 Time (h) 0 5 10 20 2515 Time (h) 0 5 10 20 2515 Time (h) Fig. 4. Time courses of nitrite accumulation and consumption in Paracoccus pantotro- phus strains. Starter cultures were grown aerobically in LB with shaking before inocu- lation of mineral salt medium containing 20 m M nitrate in a 1% v ⁄ v dilution and appropriate antibiotics. These cultures were incubated without shaking at 37 °C. Growth of wild-type P. pantotrophus GB17 strain is shown in the upper left panel and of the kanamycin insertion mutant nirF strain (SBN03) in the upper right panel. Growth of SBN11 (DnirF) is shown in the bottom left panel and of SBN13 (DnirF) containing pEG276-nirF in the bottom right panel. Cell density was determined at A 600 and is depicted by grey diamonds, whereas the extracellular nitrite concentration was deter- mined using a colorimetric method and is shown by black triangles. The data shown here are the averages of four different experiments. Periplasmic NirF binds d 1 heme S. Bali et al. 4948 FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS in a single affinity chromatography step, with a yield of  3.0 mg protein per litre of culture. All the NirF variants were also produced and purified in a similar manner with a yield ranging from 1.5 to 4.0 mgÆL )1 of culture. SDS ⁄ PAGE gels for all proteins are shown in Fig. S4. Size exclusion chromatography demonstrated that NirF is monomeric. Dynamic light scattering showed that the protein was well folded with a hydro- dynamic radius fitting with the molecular weight of the mature protein. CD of the protein in potassium phos- phate buffer at pH 7.5 displayed a predominantly beta-sheet structure (data not shown). This is consis- tent with the sequence similarity of this protein with the C-terminal beta-propeller domain of NirS (cyto- chrome cd 1 ) that houses d 1 heme. MS confirmed the molecular mass of the protein to be 41.937 kDa, which is expected after processing and cleavage of the signal peptide. Surprisingly, a D129A mutant of NirF failed to give any soluble protein when overexpressed in E. coli, although this variant rescued denitrification when it complemented the Paracoccus DnirF strain under denitrifying conditions. This observation sug- gests that the conserved Asp129 is important for fold- ing of the recombinant protein. SBN23 (ΔnirF + NirF (H41M))SBN19 (ΔnirF + NirF (H41A)) SBN21 (ΔnirF + NirF (H41K)) SBN22 (ΔNirF + NirF (H41C)) SBN25 (ΔnirF + NirF (D129Q))SBN24 (ΔnirF + NirF (D129A)) 22 20 18 16 12 14 10 8 6 4 2 0 0102030 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 Nitrite (mM)Nitrite (m M) Time (h) 22 20 18 16 12 14 10 8 6 4 2 0 0 5 10 2015 25 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 Nitrite (mM) Time (h) 0 0 5 5 10 10 20 20 15 15 25 Time (h) 22 20 18 16 12 14 10 8 6 4 2 0 0 5 10 2015 25 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 Time (h) 22 20 18 16 12 14 10 8 6 4 2 0 0 5 10 2015 25 30 35 40 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 Time (h) 22 20 18 16 12 14 10 8 6 4 2 0 05 1510 20 25 2.2 2 1.8 1.4 1.2 1.6 1 0.8 0.6 0.4 0.2 0 A 600 A 600 A 600 Time (h) Fig. 5. His41 is essential for Paracoc- cus pantotrophus NirF, but Asp129 is dis- pensable. Growth plots and time courses of nitrite appearance and disappearance for P. pantotrophus SBN11 (DnirF) strain com- plemented with plasmid carrying a gene coding for NirF(H41A) (upper left), NirF(H41M) (upper right), NirF(H41K) (middle left), NirF(H41C) (middle right), NirF(D129A) (lower left) and NirF(D129Q) (lower right). Cell density was determined at A 600 and is depicted by grey diamonds, whereas the extracellular nitrite concentration was deter- mined using a colorimetric method and is shown by black triangles. The data shown here are the averages of four different experiments. S. Bali et al. Periplasmic NirF binds d 1 heme FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS 4949 In vitro binding of d 1 heme to NirF As explained above, there are sequence similarities (Fig. S2) between NirF and the d 1 heme-binding domain of cytochrome cd 1 . Therefore, we tested for the binding of d 1 heme to purified NirF. The addition of d 1 heme to the NirF resulted in the appearance of a distinctive visible absorption spectrum (Fig. 6). The d 1 heme peak shifted from 681 to 630 nm. Considering that NirF is colourless and has no absorbance in the UV–visible region, this shift of 50 nm in the spectrum is due to extreme changes in the d 1 heme environment. This binding of d 1 heme with NirF was stoichiometric, i.e. 1 mol of heme was taken up by 1 mol of NirF (concentrations of heme and protein were calculated by using the extinction coefficient mentioned in the experimental section and the standard Bradford assay, respectively). To test whether the binding of d 1 heme to NirF was specific, and thus physiologically signifi- cant, we added heme to NirF and found that there were no shifts in the visible absorption spectrum and thus no interaction. It is already known for other peri- plasmic proteins that the d 1 heme-binding region is very sensitive to proton concentration and prefers a lower pH for d 1 heme addition, consistent with the periplasm probably having a pH lower than 7 [15]. Similarly, the process of d 1 heme addition to NirF was also pH dependent. At relatively high pH values (8 or higher) the spectral change described above for adding d 1 to apo-protein, did not occur; however, when the pH was lowered to neutral pH the uptake of d 1 heme proceeded. As NirF could have at least two other interacting partners in the periplasm for the d 1 heme assembly, namely NirC and NirN, we wanted to test if the com- plex of NirF.d 1 could transfer d 1 heme to NirN, which was recently also shown to bind d 1 heme [15]. This binding would be difficult to judge, as the NirN.d 1 heme complex shows a peak at 627 nm. Unfortunately, we could not observe any significant peak shifts when NirN was added to the NirF.d 1 heme complex in slight molar excess under anaerobic conditions (data not shown). His200 of P. pantotrophus NirS is conserved between NirF and NirS; it is the His residue that in NirS is the proximal axial ligand to the d 1 heme. Replacement of an equivalent His, His41, in NirF by Ala abolished binding of the heme to the protein. Known distal heme-binding residues, such as Met, Lys or Cys [18,19], were substituted for His41 in NirF. No changes in the visible spectra were observed when all three variants, NirF(H41K), NirF(H41M) and NirF(H41C), were added individually to the d 1 heme in slight molar excess. There was no equivalent peak at 630 nm, which was observed for the NirF.d 1 complex. These results, when taken together with in vivo comple- mentation analysis of NirF(H41) variants, suggest that interaction of NirF with d 1 heme is very specific for His41. This His41 residue must play a part in both structural and functional roles of NirF. Discussion On the basis of several criteria, including cell fraction- ation and the consequences of either deleting the putative signal sequence or replacing it by a proven signal sequence from nirC, it can be concluded that NirF is a periplasmic protein in P. pantotrophus. This has an important implication as the only other known d 1 biogenesis proteins with a periplasmic location are NirC and NirN, both of which are not essential for d 1 heme synthesis [15,16]. It follows that, unless there are other unrecognized d 1 biogenesis proteins, then NirF must catalyse the last step(s) in d 1 heme synthesis. The nature of these synthesis steps is conjectural at this stage, but the NirF.d 1 heme complex might reflect a product complex. The failure of the D129A mutation to prevent d 1 heme biogenesis suggests that the activity of NirF cannot be similar to that of Met8P activity where a comparable mutation is inhibitory. A puzzle is that some NirF sequences, notably for two strains of Ps. aeruginosa, PA7 and PAO1, but also that in Magnetospirillum magneticum, do not have any readily recognizable, i.e. N-terminal positive charges, central hydrophobic core (or h-region) of seven to 15 amino acid residues, followed by a peptidase recognized ‘c-region’ [20], signal sequences. These sequences are in 0.21 0.18 0.15 0.12 0.09 0.06 0.03 0 580 600 620 640 660 680 700 720 Wavelength (nm) Absorbance Fig. 6. Visible absorption spectra of oxidized d 1 heme, 0.060 mM, before ( ) and after the addition of NirF ( ) in slight molar excess. The flat trace at the bottom is the visible absorption spectrum of NirF at 0.041 m M. All spectra were taken in 50 mM phosphate buffer, pH 7, at room temperature. Periplasmic NirF binds d 1 heme S. Bali et al. 4950 FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS contrast to many other sequences for NirF proteins where the signal sequence is readily recognizable. It is possible that the function of NirF can be realized in the cytoplasm of Ps. aeruginosa, with the resulting d 1 heme then being translocated to the periplasm. In the case of P. pantotrophus it would be the substrate for NirF that is translocated. In either case the transport process is enigmatic as none of the Nir proteins codes for a transmembrane protein that could be a candidate for moving d 1 heme, or a precursor, across the mem- brane. Alternatively, as suggested by Suzuki et al. [21], NirF in some organisms might be periplasmic but with an N-terminal transmembrane helix anchoring the pro- tein to the membrane. However, our bioinformatics analysis of the N-terminal sequences for NirF for Ps. aeruginosa and M. magneticum does not agree with this suggestion. A function of NirF related to binding nucleotide in a putative Rossman fold now appears unlikely, as the putative Gly of such a fold are not essential. This result also correlates well with the export of NirF via the sec system; a periplasmic pro- tein with a bound nucleotide would be exported via the Tat system in a folded conformation along with the cofactor. Heme d 1 differs from other tetrapyrrole derivatives in that it is a dioxoisobacteriochlorin, as opposed to porphyrin, characterized by the presence of two oxo groups at C3 and C8 and methyl groups at positions C2 and C7 [7,22]. Also, its synthesis is mediated via a separate branch of the tetrapyrrole biosynthetic path- way from uroporphyrinogen III [14,15,23]. Recently we showed that methylation at C2 and C7 is catalysed by NirE to give another tetrapyrrole intermediate pre- corrin-2 [14,15]. Further modifications would include: (a) decarboxylations of the acetate side chain at posi- tions C12 and C18, (b) dehydrogenation of the C17 side chain to give an acrylate moiety, (c) introduction of oxo groups at positions C3 and C8, (d) ferrochela- tion and (f) transport to the periplasm. Not only the enzymes and chemistry of all these steps are unknown, but even the order in which the modifications occur remains mostly unknown. Our result that NirF is a periplasmic enzyme indicates that this protein catalyses the chemistry required for the last stages of d 1 heme biosynthesis. However, defining the substrate for NirF will not be an easy task. Possibilities include the d 1 heme lacking iron and ⁄ or with the side chain satu- rated, but accessing these putative substrates is not trivial. An alternative approach would be to seek accu- mulation of the substrate of NirF in a mutant that lacks NirF; this too is not trivial as the DnirF strain does not accumulate readily detectable amounts of an intermediate of d 1 synthesis. Experimental procedures DNA manipulations DNA manipulations were performed by standard methods. Primers were synthesized by Sigma–Genosys (Haverhill, UK). Amplifications by PCR using KOD DNA polymerase (from Thermococcus kodakaraensis) were according to sup- plier’s instructions (Novagen, now Merck Biosciences, Not- tingham, UK). All constructs generated by PCR were confirmed to be correct by sequencing. All the primers used in this study are shown in Table S1. Construction of bacterial strains Initially, an unmarked deletion was generated in nirF in a wild-type GB17 P. pantotrophus strain. This was performed in a two-step process. First, the 5¢ and 3¢ flanking regions of nirF were cloned and the kan R cassette inserted between them. This was cloned into pJQ200ks (gentamicin-resistant), which is incapable of replication in P. pantotrophus. Chro- mosomal integrants in which double crossover events had replaced the nirF ORF with the kanamycin-resistance cas- sette, but lost the pJQ200ks backbone, were selected as kanamycin-resistant gentamicin-sensitive strains. Correct integration of the cassette was confirmed by PCR screening. Second, the deletion was made unmarked using a con- struct in which the nirF flanks were cloned into the pRVS2 vector (gentamicin resistance), which was modified from pRVS1 [24]. This vector is also incapable of replication in P. pantotrophus. Single crossover events were selected via resistance to streptomycin (P. pantotrophus), gentamicin (pRVS2) and kanamycin (nirF::kan R ). This strain was then screened for a second crossover event in which the kan R cassette was removed via homologous recombination. This strain was selected essentially as described in [24] and iden- tified by the growth of kanamycin-sensitive white colonies in the presence of 200 mgÆmL )1 X-gal (5-bromo-4-chloro-3- indolyl-b-galactoside). Putative strains were confirmed to be correct by PCR screening (D nirF). Full details of the con- struct generation and strategy employed can be found in supporting information (Doc. S1, S2 and S3). Cloning of P. pantotrophus nirF and nirF variants The nirF ORF was amplified from P. pantotrophus genomic DNA using SB5 and SB6, digested with NcoI and XhoI and ligated into NcoI ⁄ XhoI-digested pET22b (for overex- pression in E. coli). A C-terminal strep II tag was intro- duced in the pET22b-based construct by inverse PCR using primers SB45 and SB46, and by self-ligating the purified PCR product after phosphorylation with T4 PNK. The native P. pantotrophus signal sequence of nirF was removed by inverse PCR with SB62 and SB63 to generate a new construct that had the PelB signal sequence in frame with S. Bali et al. Periplasmic NirF binds d 1 heme FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS 4951 downstream nirF, for recombinant production of NirF in E. coli. The internal EcoRI site within the nirF gene was silently mutated using the primers SB30 and SB31 and the product of this PCR was used to amplify EcoRI and Hin- dIII flanked nirF to clone into EcoRI ⁄ HindIII digested pEG276 (for expression in P. pantotrophus strains) [25]. Inverse PCR was used to generate a number of mutations using the following primer combinations on both the pET22b- and pEG276-based clones: H41A – SB82 and SB83, H41K – SB114 and SB115, H41C – SB116 and SB117, H41M – SB118 and SB119, D129A – SB101 and SB102, D129Q – SB103 and SB104. The Gly-rich region in the N-terminus of the nirF was deleted by inverse PCR using primers SB110 and SB111. Similarly, the native signal sequence of the nirF was deleted from the pEG276-based plasmid using the primers SB112 and SB113. A hybrid NirF was made by introduction of the NirC signal sequence in front of NirF by two sequential inverse PCRs using primers SB121, SB122, SB123 and SB124. The mutants generated in this study are detailed in Table 1. Bacterial strains, plasmids and growth conditions The bacterial strains and plasmids used in this study are detailed in Table 1. Paracoccus pantotrophus strains were grown in Luria–Bertani (LB) medium or in a defined mini- mal medium [26] supplemented with 20 mm succinate as a carbon and energy source. Overnight aerobic growth was achieved in 5 mL growth medium in 50 mL universals, which were incubated in a shaker at 250 rpm at 37 °C. Anaerobic growth was conducted in 600 mL growth medium in com- pletely filled bottles, with stationary incubation at 37 °C. For anaerobic growth, cultures were supplemented with 20 mm sodium nitrate. Anaerobic cultures were inoculated with 1% v ⁄ v freshly grown aerobic overnight culture in LB and cell density determined at A 600 . Antibiotic-resistant strains were supplemented with antibiotics at the following concentrations: streptomycin (100 mgÆmL )1 ), kanamycin (50 mgÆmL )1 ), carbenicillin (100 mgÆmL )1 ) and gentamicin (20 mgÆmL )1 ). Growth on solid media used liquid growth medium supplemented with 1.5% bacteriological agar. Analysis of extracellular nitrite Cells were pelleted from 1 mL anaerobic culture via centri- fugation at 14 000 g for 1 min. The nitrite concentration in the medium was estimated colorimetrically using the method in [27]. Fractionation of P. pantotrophus extracts and western blotting Paracoccus pantotrophus strains were grown in 2 L cultures of minimal media supplemented with 20 mm sodium nitrate and 20 mm sodium succinate and harvested at 6000 g for 20 min. Cell pellets were resuspended in 10 mL SET buffer (100 mm Tris ⁄ HCl pH 7.5, 3 mm EDTA and 0.5 m sucrose) to which 1 mgÆmL )1 lysozyme, 75 mg DNaseI and 1 ⁄ 5ofa protease inhibitor tablet were added. This suspension was incubated at 37 °C for 40 min and spun at 26 000 g for 40 min to collect the periplasmic fraction. The pellet from the last step was resuspended in 20 mm Tris ⁄ HCl pH 7.5, and French-pressed three times at 1000 psi. Cell debris and the insoluble fraction were removed by centrifugation at 12 000 g for 30 min. The supernatant was centrifuged at 150 000 g for 2 h to collect the membranes, which were resuspended in 5 mL 20 mm Tris ⁄ HCl pH 7.5 and stored at )80 °C. The supernatant from the 150 000 g step was kept as cytoplasm and stored at )20 °C. Paracoccus pantotrophus strains were grown anaerobically in 50 mL minimal salt medium supplemented with 20 mm sodium nitrate, to an A 600 of  1 before harvesting at 6000 g for 10 min. Pellets were resuspended in BugBuster (Novagen, now Merck) at 0.2 g dry pelletÆmL )1 and incu- bated at room temperature with rocking for 30 min. Ten millilitre samples of lysate or 3 mL membrane extracts con- taining equal protein concentrations (total 30 mg) were run on an SDS ⁄ PAGE gel for analysis. Western blots to detect strep II tags were performed using an alkaline phos- phatase conjugate of strep-tactin antibody (IBA, Go ¨ ttingen, Germany) according to the manufacturer’s instructions. For all SDS ⁄ PAGE, the markers used were SeeBlue Plus 2 (Invitrogen, Paisley, UK). Recombinant production and purification of NirF and its variants Overexpression was performed in the E. coli strain BL21 codonplus (RIPL) (Stratagene, Leicester, UK). All cells expressing protein were grown at 37 °C in 500 mL volumes of LB broth in 2 L flasks from overnight starter cultures to an A 600 of 0.6–0.7 and transferred to 16 °C before induc- tion with 0.2 mm isopropyl thio-b-d-galactoside. After fur- ther incubation for 16 h, the cells from the 2 L culture were harvested and resuspended in 6 mL 50 mm Tris ⁄ HCl pH 7.5, containing a trace amount of DNaseI and protease inhibitor tablet. Periplasmic fractions were obtained by incubating the resuspended cells with 1 mgÆmL )1 polymyxin B sulphate at 4 °C for 45 min and removing the insoluble material by centrifuging at 15 000 g for 40 min. The peri- plasmic fraction was applied to 5 mL of Strep-Tactin- Sepharose (IBA) equilibrated with 50 mm Tris ⁄ HCl, 250 mm NaCl (pH 7.5). The column was washed with six column volumes of 50 mm Tris ⁄ HCl, 250 mm NaCl (pH 7.5) and the protein was eluted with 50 mm Tris ⁄ HCl (pH 7.5), 150 mm NaCl, 2.5 mm desthiobiotin (IBA) according to the manufacturer’s instructions. All the NirF variants were also produced in the same manner. The purity of the samples was checked by running SDS ⁄ PAGE 10% Bis ⁄ Tris NuPAGE gels (Invitrogen). Periplasmic NirF binds d 1 heme S. Bali et al. 4952 FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS MS with the purified NirF For MALDI analysis, the purified protein was desalted on a C18 column. Approximately 10 lm of the purified solution was premixed with the matrix: a-cyano-4-hy- droxycinnamic acid (10 mm in 35% aqueous acetonitrile, 0.1% trifluoroacetic acid) at a 1 : 1 ratio and 1 lLof mixture applied directly to the sample plate. The droplet was air-dried before analysis in the MS. MALDI spectra were obtained in reflectron mode and a nitrogen laser, emitting 337 nm light in a 3 ns pulse, was the ionization source. The accelerating voltage in the ion source was 30 kV. Acknowledgements This work was funded by research grant BBE0229441 from the Biotechnology and Biological Sciences Table 1. Strains and plasmids used in the present study. Strain Genotype and description Reference Escherichia coli DH5a supE44 DlacU169 (f80 lacZDM15) hsdR17 recA1 endA1 gyrA96 thi1 relA1 (general cloning vehicle) Gibco BRL S17-1 Sm r pro r ) m + RP4-2 integrated (Tc::Mu) (Km::Tn7) [28] Paracoccus pantotrophus GB17 wild-type P. pantotrophus strain, Strep R [26] nirF::kan R or SBN3 Chromosomally disrupted copy of nirF This work DnirF or SBN11 Unmarked deletion in nirF This work SBN13 DnirF derivative with pEG276-NirF-strepII This work SBN15 DnirF derivative with pEG276 This work SBN19 DnirF derivative with pEG276-NirF H41A This work SBN20 DnirF derivative with pEG276-NirF (no signal sequence) This work SBN21 DnirF derivative with pEG276-NirF H41K This work SBN22 DnirF derivative with pEG276-NirF H41C This work SBN23 DnirF derivative with pEG276-NirF H41M This work SBN24 DnirF derivative with pEG276-NirF D129A This work SBN25 DnirF derivative with pEG276-NirF D129Q This work SBN26 DnirF derivative with pEG276-NirF (D4-17) This work SBN28 DnirF derivative with pEG276-NirF (nirC signal sequence) This work Plasmids Description Reference pTZ19R Amp R , general cloning vector Fermentas pEG276 Gent R , expression vector [25] pUC4K Amp R , source of kan R cassette Pharmacia pJQ200ks Gent R , suicide vector [29] pRVS1 Strep R , suicide vector [24] pRVS2 Strep R , Gent R , suicide vector This work pJQ200ks-nirF::kan R Contains cassette for nirF disruption This work pRVS2-DnirF Contains cassette for unmarked deletion of nirF This work pEG276-NirF-strepII P. pantotrophus nirF cloned into pEG276 with strep II tag This work pEG276-NirF H41A P. pantotrophus nirF H41A cloned into pEG276 This work pEG276-NirF H41K P. pantotrophus nirF H41K cloned into pEG276 This work pEG276-NirF H41M P. pantotrophus nirF H41M cloned into pEG276 This work pEG276-NirF H41C P. pantotrophus nirF H41C cloned into pEG276 This work pEG276-NirF D129A P. pantotrophus nirF D129A cloned into pEG276 This work pEG276-NirF D129Q P. pantotrophus nirF D129Q cloned into pEG276 This work pEG276-NirF (NirC signal sequence) P. pantotrophus nirF (NirC signal sequence) cloned into pEG276 This work pEG276-NirF (no signal sequence) P. pantotrophus nirF (no signal sequence) cloned into pEG276 This work pEG276-NirF (D4-17) P. pantotrophus nirF (D4-17) cloned into pEG276 where (D4-17) is a deletion of N-terminal GXGX 2 GX 7 G motif This work pET-22b Expression vector, Amp R , Novagen pET-22b-NirF-strepII P. pantotrophus nirF cloned into pEG276 with strep II tag This work pET-22b -NirF H41A P. pantotrophus nirF H41A cloned into pET-22b This work pET-22b -NirF H41K P. pantotrophus nirF H41K cloned into pET-22b This work pET-22b -NirF H41M P. pantotrophus nirF H41M cloned into pET-22b This work pET-22b -NirF H41C P. pantotrophus nirF H41C cloned into pET-22b This work pET-22b -NirF D129A P. pantotrophus nirF D129A cloned into pET-22b This work pET-22b -NirF D129Q P. pantotrophus nirF D129Q cloned into pET-22b This work S. Bali et al. Periplasmic NirF binds d 1 heme FEBS Journal 277 (2010) 4944–4955 ª 2010 The Authors Journal compilation ª 2010 FEBS 4953 [...].. .Periplasmic NirF binds d1 heme S Bali et al Research Council to S J F and M J W Dr Andreas Schlueter and Alfred Phuler at the University of Biele¨ feld are thanked for their kind gift of plasmid pMS255 Dr Katalin di Gleria at The Weatherall Institute of Molecular Medicine, University of Oxford is thanked for her assistance with the MS of the proteins Amy Varney and Christopher Greening are thanked... bacteria Gene 127, 15–21 Periplasmic NirF binds d1 heme Supporting information The following supplementary material is available: Fig S1 Expression of NirF in Paracoccus pantotrophus DnirF strain Fig S2 Multiple sequence alignment of NirF with d1 domain of cd1, NirS Fig S3 Sequence alignment of NirF with Met8p and CysG Fig S4 Purification of recombinant NirF and its variants Doc S1 Construction of nirF: :kanR... Evidence that heme d1 is a 1,3-porphyrindione Biochemistry 25, 8447–8453 8 Matthews JC & Timkovich R (1993) Biosynthetic origins of the carbon skeleton of heme d1 Bioorg Chem 21, 71–82 9 Kawasaki S, Arai H, Igarashi Y & Kodama T (1995) Sequencing and characterisation of the downstream region of the genes encoding nitrite reductase and cytochrome c551 (nirSM) from Pseudomonas aeruginosa: identification of. .. nirF: :kanR disruption cassette Doc S2 Construction of DnirF cassette for generating unmarked nirF deletion and modification of suicidal vector pRVS1 Doc S3 Construction of marked and unmarked deletion in nirF Table S1 Oligonucleotides used in this work This supplementary material can be found in the online version of this article Please note: As a service to our authors and readers, this journal provides... fluorescens deficient in dissimilatory nitrite reduction are also altered in nitric oxide reduction J Bacteriol 174, 2560–2564 12 Deboer APN, Reijnder WNM, Kuenen JG, Stouthamer AH & Vanspanning RJM (1994) Isolation, sequencing and mutational analysis of a gene cluster involved in nitrite reduction in Paracoccus denitrificans Antonie Leeuwenhoek 66, 111–127 13 Kawasaki S, Arai H, Kodama T & Igarashi Y (1997) Gene... required for heme d1 biosynthesis FEBS J 276, 5973–5982 15 Zajicek RS, Bali S, Arnold S, Brindley AA, Warren MJ & Ferguson SJ (2009) d1 haem biogenesis – assessing the roles of three nir gene products FEBS J 276, 6399– 6411 16 Hasegawa N, Arai H & Igarashi Y (2001) Two c-type cytochromes, NirM and NirC, encoded in the nir gene cluster of Pseudomonas aeruginosa act as electron donors for nitrite reductase Biochem... cofactors from bacteria Ciba Found Symp 180, 228–238 24 Van Spanning RJ, Wancell CW, De Boer T, Hazelaar MJ, Anazawa H, Harms N, Oltmann LF & Stouthamer AH (1991) A method for introduction of unmarked mutations in the genome of Paracoccus denitrificans – construction of strains with multiple mutations in the genes encoding periplasmic cytochrome-C550, cytochrome-C551, and cytochrome-C553 J Bacteriol 173, 6962–6970... 1223–1230 17 Schubert HL, Raux E, Brindley AA, Leech HK, Wilson KS, Hill CP & Warren MJ (2002) The structure of Saccharomyces cerevisiae Met8p, a bifunctional dehydrogenase and ferrochelatase EMBO J 21, 2068– 2075 18 Bandi S, Baddam S & Bowler BE (2007) Alkaline conformational transition and gated electron transfer with a Lys 79 -> His variant of iso-1-cytochrome c Biochemistry 46, 10643–10654 19 Gross... modifications indicate differences in axial haem c iron ligation between the related NrfH and NapC families of multihaem c-type cytochromes Biochem J 390, 689–693 20 Von Heijne G & Manoil C (1990) Membrane proteins: from sequence to structure Protein Eng 4, 109–112 21 Suzuki M, Hirai T, Arai H, Ishii M & Igarashi Y (2006) Purification, characterization, and gene cloning of thermophilic cytochrome cd1 nitrite... cluster for dissimilatory nitrite reductase (nir) from Pseudomonas aeruginosa: sequencing and identification of a locus for heme d1 biosynthesis J Bacteriol 179, 235–242 14 Storbeck S, Walther J, Muller J, Parmar V, Schiebel ¨ HM, Kemken D, Dulcks T, Warren MJ & Layer G ¨ (2009) The Pseudomonas aeruginosa nirE gene encodes the S-adenosyl-L-methionine-dependent uroporphyrinogen III methyltransferase required . NirF is a periplasmic protein that binds d 1 heme as part of its essential role in d 1 heme biogenesis Shilpa Bali 1 , Martin J. Warren 2 and Stuart. demonstrated that Asp129 of NirF could not be essential for any function similar to that in Asp141 of Met8P. The idea of NirF being a dehydrogenase is appealing because