Insight into the haem d1 biosynthesis pathway in heliobacteria through bioinformatics analysis

Abstract

Haem d(1) is a unique tetrapyrrole molecule that serves as a prosthetic group of cytochrome cd(1), which reduces nitrite to nitric oxide during the process of denitrification. Very little information is available regarding the biosynthesis of haem d(1). The extreme difficulty in studying the haem d(1) biosynthetic pathway can be partly attributed to the lack of a theoretical basis for experimental investigation. We report here a gene cluster encoding enzymes involved in the biosynthesis of haem d(1) in two heliobacterial species, Heliobacillus mobilis and Heliophilum fasciatum. The gene organization of the cluster is conserved between the two species, and contains a complete set of genes that lead to the biosynthesis of uroporphyrinogen III and genes thought to be involved in the late steps of haem d(1) biosynthesis. Detailed bioinformatics analysis of some of the proteins encoded in the gene cluster revealed important clues to the precise biochemical roles of the proteins in the biosynthesis of haem d(1), as well as the membrane transport and insertion of haem d(1) into an apocytochrome during the maturation of cytochrome cd(1).

Fig. 1

(a) Outline of the putative biosynthetic pathway of haem d₁ in bacteria in which 5-aminolevulinic acid is synthesized through the C-5 pathway. The C-4 pathway, through condensation of succinyl-CoA and glycine, found only in proteobacteria is omitted in this figure. The catalysis of the second half of haem d₁ biosynthesis as well as incorporation of haem d₁ into cytochrome cd₁ are still largely unknown and thus labelled with question marks. (b) Structure of haem d₁. Based on current knowledge, modifications of most of the moieties of uroporphyrinogen III to produce haem d₁ are performed by unknown enzymes (circled and labelled with ?).

Fig. 2

Physical maps of the DNA fragments sequenced from Hb. mobilis and Hp. fasciatum. The arrowed boxes indicate predicted ORFs and direction of transcription. White boxes, genes related to haem biosynthesis and transport; grey boxes, genes presumably irrelevant to haem biosynthesis and transport. Groups of genes predicted to be operons are indicated with brackets.

Fig. 3

Maximum-likelihood tree of the HemA family, showing a recent duplication event that gave rise to HemA1 and HemA2 in heliobacteria. Due to the large size of this sequence family, only a portion of the tree is shown, with filled triangles representing omitted taxa. The numbers on the branches indicate bootstrap values. The scale bar corresponds to 0.1 amino acid substitutions per site.

Fig. 4

(a) Sequence alignment of the CysG^A domain of the CysG^A–HemD fusion protein from Hb. mobilis and Hp. fasciatum (Hm_CysG^A and Hf_CysG^A, respectively) with the CysG^A domain of the CysG protein of S. enterica for which a crystal structure is available (1PJS). Identical sequence matches in the alignment are indicated by ‘*’, strongly similar matches by ‘:’, and weakly similar matches by ‘.’. (b) 3D model of the CysG^A domain for Hb. mobilis based on the above alignment. The position of the bound cofactor SAH (demethylated SAM) is also shown.

Fig. 5

(a) Sequence alignment of CysG^B from the two heliobacterial species with the CysG^B domain of the CysG protein of S. enterica for which a crystal structure is available (1PJS). (b) 3D model of CysG^B for Hb. mobilis based on the above alignment. The bifunctional enzyme has two distinct structural domains, the dehydrogenase domain on the N terminus and the ferrochelatase domain on the C terminus. The bound cofactor NAD for the dehydrogenase domain is also shown.

Fig. 6

(a) Maximum-likelihood tree of the NirJ family, showing ancient divergence of NirJ1 from NirJ2 (indicated by asterisks). The numbers on the branches indicate bootstrap values. The scale bar corresponds to 0.1 amino acid substitutions per site. (b) Sequence alignment of NirJ2 from the two heliobacterial species with MoaA of Staph. aureus for which a crystal structure is available (1TV8). (c) 3D model of NirJ2 for Hb. mobilis based on the above alignment. The positions of the bound cofactors SAM and iron–sulfur centres (Fe₄S₄) are also shown.

Fig. 7

(a) Maximum-likelihood tree of the NirD/L and Lrp family. With the Lrp sequences forming a natural outgroup, the ancient gene duplication event leading to the separation of NirD and NirL is evident. Further gene duplication from the ancestor of either NirD or NirL gave rise to NirG and NirH in Pseudomonas. The number on each branch represents a bootstrap value. The scale bar corresponds to 0.1 amino acid substitutions per site. (b) The result of expression and purification of NirL from Hp. fasciatum using an intein-mediated approach. The protein samples were fractionated in a 12.5 % SDS-polyacrylamide gel stained with Coomassie brilliant blue R-250. Lane 1, clarified cell lysate applied to the chitin-containing affinity column; lane 2, protein sample eluted from the column after in situ protein splicing, showing NirL (17 kDa) being purified to near homogeity; lane 3, protein molecular mass markers with numbers on the right indicating protein size in kDa. (c) The result of DNA mobility shift assay for NirL in a 5 % native polyacrylamide gel stained with SYBR-Gold. Lane 1, nirJ2 promoter DNA only; lane 2, nirJ2 promoter DNA incubated with NirL. The DNA band shift is clearly visible in lane 2, indicating the formation of the DNA–protein complex. (d) Model of NirL binding to DNA. The homology model of NirL was constructed based on an alignment (not shown) with the most closely related Lrp transcription factor from Neisseria meningitidis (Koike et al., 2004; PDB code 1RI7). NirL was modelled in the dimer form based on a dimer unit of the same octameric structure with a double-stranded DNA ligand modelled based on the suggestions of Koike et al. (2004) and Ren et al. (2007). The DNA coordinates were extracted from the structure of Schultz et al. (1991) (PDB code 1CGP).

Fig. 8

(a) Working hypothesis of haem d₁ biosynthesis as well as incorporation of haem d₁ into an apocytochrome to produce cytochrome cd₁, based on the bioinformatics analysis result. The CysG^A domain of the CysG^A–HemD fusion protein is proposed to methylate uroporphyrinogen III at C2 and C7 in two consecutive steps to produce precorrin-2. NirJ is proposed to catalyse the decarboxylation of the acetate sidechains on rings III and IV. CysG^B, which is a bifunctional dehydrogenase and ferrochelatase, is proposed to catalyse the oxidation of the single bond between C15 and C16 to produce a double bond and the insertion of a ferrous iron in porphyrindione d₁ to complete the haem d₁ synthesis. The transport of synthesized haem d₁ and its insertion into an apocytochrome are thought to be mediated by CcsA and Ccs1. (b) Structure of haem d₁ labelled with enzymes proposed to be involved in converting some of the circled moieties. The acrylate formation in ring IV may be catalysed by CysG^B, whereas the conversion of the propionate groups to oxo groups in rings I and II may be catalysed by NirJ, both of which are predicted with a lesser degree of confidence at this stage (labelled with ?).